Medical Grand Rounds

Much has improved in renal transplantation over the past 20 years. The focus has shifted to using stronger immunotherapy rather than trying to minimize it. There has been increasing recognition of infection and ways to prevent and treat it. Induction therapy now has greater emphasis so that maintenance therapy can be eased, with the aim of reducing long-term toxicity. Perhaps the biggest change is the practice of screening for donor-specific antibodies at the time of transplant so that predictable problems can be prevented or better handled if they occur. Such advances have helped patients directly and by extending the life of their transplanted organs.

LONGER SURVIVAL

As early as the 1990s, it was recognized that kidney transplant offers a survival advantage for patients with end-stage renal disease over maintenance on dialysis.¹ Although the risk of death is higher immediately after transplant, within a few months it becomes much lower than for patients on dialysis. Survival varies according to the health of the patient and the quality of the transplanted organ.

In general, patients who obtain the greatest benefit from transplants in terms of years of life gained are those with diabetes, especially those who are younger. Those ages 20 to 39 live about 8 years on dialysis vs 25 years after transplant.

CONTRAINDICATIONS TO TRANSPLANT

There are multiple contraindications to a solitary kidney transplant (Table 1), including smoking. Most transplant centers require that smokers quit before transplant. Long-standing smokers almost double their risk of a cardiac event after transplant and double their rate of malignancy. Active smoking at the time of transplant is associated with twice the risk of death by 10 years after transplant compared with that of nonsmokers.² Cotinine testing can detect whether a patient is an active smoker.

WAITING-LIST CONSIDERATIONS

Organs are scarce

The number of patients on the kidney waiting list has increased rapidly in the last few decades, while the number of transplants performed each year has remained about the same. In 2016, about 100,000 patients were on the list, but only about 19,000 transplants were performed.³ Wait times, especially for deceased-donor organs, have increased to about 6 years, varying by blood type and geographic region.

Waiting-list placement

Placement on the waiting list for a deceased-donor kidney transplant occurs when a patient has an estimated glomerular filtration rate (GFR) of 20 mL/min/1.73 m² or less, although referral to the list can be made earlier. Early listing remains advantageous, as total time on the list will be counted before starting dialysis. “Preemptive transplant” means the patient had no dialysis before transplant; this applies to about 10% of transplant recipients. These patients tend to fare the best and are usually recipients of a living-donor organ.

Most patients do not receive a transplant until the GFR is less than 15 mL/min/1.73 m².

Since 2014, wait time has been measured from the beginning of dialysis rather than the date of waiting-list placement in patients who are listed after starting dialysis therapy. This approach is more fair but sometimes introduces problems. A patient who did not previously know about the list may suddenly jump to the head of the line after 10 years of dialysis, by which time comorbidities associated with long-term dialysis make the patient less likely to gain as much benefit from a transplant as people lower on the list. Time on dialysis, or “dialysis vintage,” predicts patient and kidney survival after transplant, with reduced survival associated with increasing time on dialysis.⁴

Shorter wait for a suboptimal kidney

The aging population has increased the number of older patients being listed for transplant, presenting multiple challenges. Patients age 65 or older have a 50% chance of dying before they receive a transplant during a 5-year wait.

A patient may shorten the wait by joining the list for a suboptimal organ. All deceased-donor organs are given a Kidney Donor Profile Index score, which predicts the longevity of an organ after transplant. The score is determined by donor age, kidney function based on the serum creatinine at the time of death, and other donor factors.

A kidney with a score higher than 85% is likely to function longer than only 15% of available kidneys. Patients who receive a kidney with that score have a longer period of risk of death soon after transplant and a slightly higher risk of death in the long term than patients who receive a healthier kidney, although on average they still do better than patients on dialysis.⁵

Older patients should be encouraged to sign up for both the regular waiting list and the suboptimal kidney waiting list to reduce the risk of dying before they get a kidney.

Many advantages

Living-donor organ transplant is associated with a better survival rate than deceased-donor organ transplant, and the advantage becomes greater over time. At 1 year, patient survival is more than 90% in both groups, but by 5 years about 80% of patients with a living-donor organ are still alive vs only about 65% of patients with a deceased-donor organ.

The waiting time for a living-donor transplant may be only weeks to months, rather than years. Because increasing time on dialysis predicts worse patient and graft survival after transplant, the shorter wait time is a big advantage. In addition, because the donor and recipient are typically in adjacent operating rooms, the organ sustains less ischemic damage. In general, the kidney quality is better from healthy donors, resulting in superior function early on and longer graft survival by an average of 4 years. If the living donor is related to the recipient, human leukocyte antigen matching also tends to be better and predicts better outcomes.

Special challenges

Opting for a living-donor organ also entails special challenges. In addition to the ethical issues surrounding living-donor organ donation, an appropriate donor must be found. Donors must be highly motivated and pass physical, laboratory, and psychological evaluations.

For older patients, if the donor is a spouse or close friend, he or she is also likely to be older, making the organ less viable than one from a younger person. Even an adult child may not be an ideal donor if there is a family propensity to kidney disease, such as diabetic nephropathy. No test is available to determine the risk for future diabetes, but it is known to run in families.

POTENT IMMUNOSUPPRESSION

Induction therapy

Induction therapy with antithymocyte globulin or basiliximab provides intense immunosuppression to prevent acute rejection during the early posttransplant period.

Antithymocyte globulin is a potent agent that contains antibodies directed at T cells, B cells, neutrophils, platelets, adhesion molecules, and complement. It binds T cells and removes them from circulation by opsonization in splenic and lymphoid tissue. The immunosuppressive effect is sustained for at least 2 to 3 months after a series of injections (dosage 1.5 mg/kg/day, usually for 4 to 10 doses). Antithymocyte globulin is also used to treat acute rejection, especially high-grade rejection for which steroid therapy is likely to be insufficient.

Basiliximab consists of antibodies to the interleukin 2 (IL-2) receptor of T cells. Binding to T cells prevents their activation rather than removing them from circulation. The drug prevents rejection, with 30% relative reduction in early studies compared with placebo. However, it is ineffective in reversing established rejection. Dosage is 20 mg at day 0 and day 4, which provides receptor saturation for 30 to 45 days.

Basiliximab is also sometimes used off-label for patients who need to discontinue a calcineurin inhibitor (ie, tacrolimus or cyclosporine). In such cases, normal therapy is put on hold while basiliximab is given for 1 or 2 doses. Case series have been reported for this use, particularly for patients with a heart and liver transplant who develop acute kidney injury while hospitalized.^6,7

Antithymocyte globulin is more effective but also more risky. Brennan et al⁸ randomized 278 transplant recipients to either antithymocyte globulin or basiliximab. Patients in the antithymocyte globulin group had a 16% rejection rate vs 26% in the basiliximab group.

Antithymocyte globulin therapy is associated with multiple adverse effects, including fever and chills, pulmonary edema, and long-standing immunosuppressive effects such as increased risk of lymphoma and cytomegalovirus (CMV) infection. Basiliximab side-effect profiles are similar to those of placebo.

Maintenance therapy

The calcineurin inhibitors cyclosporine and tacrolimus remain the standard of care in kidney transplant despite multiple drug interactions and side effects that include renal toxicity and fibrosis. Cyclosporine and tacrolimus both bind intracellular immunophilins and thereby prevent transcription of IL-2 and production of T cells. The drugs work similarly but have different binding sites. Cyclosporine has largely been replaced by tacrolimus because its reliability of dosing and higher potency are associated with lower rejection rates.

Tacrolimus is typically given twice daily (1–6 mg/dose). Twelve-hour trough levels are followed (target: 8–12 ng/mL early on, then 5–8 ng/mL after 3 months posttransplant). Side effects include hypertension and hypercholesterolemia, but less so than with cyclosporine. On the other hand, hyperglycemia tends to be worse with tacrolimus than with cyclosporine, and combining tacrolimus with steroids frequently leads to diabetes. Tacrolimus can also cause acute and chronic renal failure, especially at high drug levels, as well as neurotoxicity, tremors, and hair loss.

Cyclosporine, tacrolimus, and sirolimus (not a calcineurin inhibitor) are metabolized through the same cytochrome P450 pathway (CYP3A4), so they have common drug interactions (Table 2).

Mycophenolate mofetil is typically used as an adjunct therapy (500–1,000 mg twice daily). It is also used for other kidney diseases before transplant, including lupus nephritis. Transplanted kidney rejection rates with mycophenolate mofetil with steroids are about 40%, so the drug is not potent enough to be used without a calcineurin inhibitor.

Side effects include gastrointestinal toxicity in up to 20% of patients, and leukopenia, which is associated with viral infections.

CORONARY ARTERY DISEASE IS COMMON WITH DIALYSIS

Coronary artery disease is highly associated with end-stage kidney disease and occurs in as many as 85% of older patients with diabetes on dialysis. Although patients with end-stage kidney disease tend to have more numerous and severe atherosclerotic lesions compared with the general population, justifying aggressive management, cardiac care tends to be conservative in patients on dialysis.⁹

Death from acute myocardial infarction occurs in about 20% to 30% of patients on dialysis vs about 2% of patients with normal renal function. Five years after myocardial infarction, survival is only about 30% in patients on dialysis.⁹

There are many explanations for excess coronary artery disease in patients on dialysis. In addition to the traditional cardiovascular risk factors of diabetes, hypertension, and preexisting coronary artery disease, patients are in a proinflammatory uremic state and have high levels of phosphorus and fibroblast growth factor 23 that contribute to vascular calcification. Almost all patients have high homocysteine levels and hemodynamic instability, particularly if they are on hemodialysis.

Pretransplant evaluation for heart disease

Patients on the kidney transplant waiting list are screened aggressively for heart disease. A history of myocardial infarction usually results in removal from the list. All patients have an initial electrocardiogram and echocardiogram. Thallium or echocardiographic stress testing is used for patients who are age 50 and older, have diabetes, or have had dialysis for many years. Patients with evidence of ischemia undergo catheterization.

Patients are also screened with computed tomography before transplant. Because the kidney is typically anastomosed to the iliac artery and vein, heavy calcification of the iliac artery can make the procedure too difficult to perform.

Kasiske et al¹⁰ analyzed data from more than 50,000 patients from the US Renal Data System and found that, for about the first year after transplant, patients who underwent kidney transplant were more likely to have a myocardial infarction than those on dialysis. After that, they fared better than patients who remained on dialysis. Those with a living-donor transplant were less likely at all times to have a myocardial infarction than those with a deceased-donor transplant. By 3 years after transplant, the relative risk of having a myocardial infarction was 0.89 for deceased-donor organ recipients and 0.69 for living-donor recipients compared with patients on the waiting list.¹⁰

INFECTIOUS COMPLICATIONS IN KIDNEY RECIPIENTS

Kidney recipients are prone to many common and uncommon infections (Table 3). All potential recipients are tested pretransplant for hepatitis B, hepatitis C, human immunodeficiency virus, syphilis, and tuberculosis. A positive result does not necessarily rule out transplant.

The following viral serology tests are also done before transplant:

Epstein-Barr virus (antibodies are positive in about 90% of adults)

CMV (about 70% of adults are seropositive)

Varicella zoster (seronegative patients should be given live-attenuated varicella vaccine).

Risk of transmission of these viruses relates to the serostatus of the donor and recipient before transplant. If a donor is positive for viral antibodies but the recipient is not (a so-called “mismatch”), risk is higher after transplant.

Hepatitis C

Patients with hepatitis C fare better if they get a transplant than if they remain on dialysis, although their posttransplant course is worse compared with transplant patients who do not have hepatitis. Some patients develop accelerated liver disease after kidney transplant. Hepatitis C-related kidney disease—membranous proliferative glomerulonephritis—also occurs, as do comorbidities such as diabetes.

Careful evaluation is warranted before transplant, including liver imaging, alpha-fetoprotein testing, and liver biopsy to evaluate for hepatocellular carcinoma. A patient with advanced fibrosis or cirrhosis may not be a candidate for kidney transplant alone but could possibly receive a combined kidney and liver transplant.

There is a need to determine the best time to treat hepatitis C infection. Patients with advanced liver disease or hepatitis C-related kidney disease would likely benefit from early treatment. However, delaying treatment could shorten the wait time for a deceased-donor organ positive for hepatitis C.  Transplant candidates with active hepatitis C are uniquely considered to accept hepatitis C-positive kidneys, which are often discarded, and may only wait weeks for such a transplant. The shortened kidney survival associated with a hepatitis C-positive kidney may no longer be true with the new antiviral hepatitis C therapy, which has been shown to be effective post-transplant.

Hepatitis B

No cure is available for hepatitis B infection, but it can be well controlled with antiviral therapy. Patients with hepatitis B infection may be candidates for transplant, but they should be stable on antiviral therapy (lamivudine, entecavir, or tenofovir) to eliminate the viral load before transplant, and therapy should be continued afterward. Liver imaging, alpha-fetoprotein levels, and biopsy are recommended for evaluation. All hepatitis B- negative patients should be vaccinated before transplant.

Organs from living or deceased donors that test positive for hepatitis B core antibody, indicating prior exposure, can be considered for transplant in a patient who tests positive for hepatitis B surface antibody, indicating successful vaccination or prior exposure in the recipient. But donors must have negative surface antigen and polymerase chain reaction (PCR) tests that indicate no active hepatitis B infection.

Cytomegalovirus

CMV typically does not appear until prophylactic therapy is stopped. Classic symptoms are fever, leukopenia, and diarrhea. Infection can involve any organ, and patients may present with hepatitis, pancreatitis or, less commonly, pneumonitis.

Patients who are negative for CMV before transplant and receive a donor-positive organ are at the highest risk. Patients who are CMV IgG-positive are considered to be at intermediate risk, regardless of the donor status. Patients who are negative for CMV and receive a donor-negative organ are at the lowest risk and do not need prophylaxis with valganciclovir.

CMV infection is diagnosed by PCR testing of the blood or immunostaining in tissue biopsy. Occasionally, blood testing is negative in the face of tissue-based disease.

BK virus

BK is a polyoma virus and a common virus associated with kidney transplant. Viremia is seen in about 18% of patients, whereas actual kidney disease associated with a higher level of virus is seen in fewer than 10% of patients. Most people are exposed to BK virus, often in childhood, and it can remain indolent in the bladder and uroepithelium.

Patients can develop BK nephropathy after exposure to transplant immunosuppression.¹¹ Posttransplant monitoring protocols typically include PCR testing for BK virus at 1, 3, 6, and 12 months. No agent has been identified to specifically treat BK virus. The general strategy is to minimize immunosuppressive therapy by reducing or eliminating mycophenolate mofetil. Fortunately, BK virus does not tend to recur, and patients can have a low-level viremia (< 10,000 copies/mL) persisting over months or even years but often without clinical consequences.

The appearance of BK virus on biopsy can mimic acute rejection. Before BK viral nephropathy was a recognized entity, patients would have been diagnosed with acute rejection and may have been put on high-dose steroids, which would have worsened the BK infection.

Posttransplant lymphoproliferative disorder

Posttransplant lymphoproliferative disorder is most often associated with Epstein-Barr virus and usually involves a large, diffuse B-cell lymphoma. Burkitt lymphoma and plasma cell neoplasms also can occur less commonly.

The condition is about 30 times more common in patients after transplant than in the general population, and it is the third most common malignancy in transplant patients after skin and cervical cancers. About 80% of the cases occur early after transplant, within the first year.

Patients typically have a marked elevation in viral load of Epstein-Barr virus, although a negative viral load does not rule it out. A patient who is serologically negative for Epstein-Barr virus receiving a donor-positive kidney is at highest risk; this situation is most often seen in the pediatric population. Potent induction therapies (eg, antilymphocyte antibody therapy) are also associated with posttransplant lymphoproliferative disorder.

Patients typically present with fever of unknown origin with no localizing signs or symptoms. Mass lesions can be challenging to find; positron emission tomography may be helpful. The culprit is usually a focal mass, ulcer (especially in the gastrointestinal tract), or infiltrate (commonly localized to the allograft). Multifocal or disseminated disease can also occur, including lymphoma or with central nervous system, gastrointestinal, or pulmonary involvement.

Biopsy of the affected site is required for histopathology and Epstein-Barr virus markers. PCR blood testing is often positive for Epstein-Barr virus.

Typical antiviral therapy does not eliminate Epstein-Barr virus. In early polyclonal viral proliferation, the first goal is to reduce immunosuppressive therapy. Rituximab alone may also help in polymorphic cases. With disease that is clearly monomorphic and has transformed to a true malignancy, cytotoxic chemotherapy is also required. “R-CHOP,” a combination therapy consisting of rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone, is usually used. Radiation therapy may help in some cases.

Cryptococcal infection

Previously seen in patients with acquired immune deficiency syndrome, cryptococcal infection is now most commonly encountered in patients with solid-organ transplants. Vilchez et al¹² found a 1% incidence in a series of more than 5,000 patients who had received an organ transplant.

Immunosuppression likely conveys risk, but because cryptococcal infection is acquired, environmental exposure also plays a role. It tends to appear more than 6 months after transplant, indicating that its cause is a primary infection by spore inhalation rather than by reactivation or transmission from the donor organ.¹³ Bird exposure is a risk factor for cryptococcal infection. One case identified the same strain of Cryptococcus in a kidney transplant recipient and the family’s pet cockatoo.¹⁴

Cryptococcal infection typically starts as pneumonia, which may be subclinical. The infection can then disseminate, with meningitis presenting with headache and mental status changes being the most concerning complication. The death rate is about 50% in most series of patients with meningitis. Skin and soft-tissue manifestations may also occur in 10% to 15% of cases and can be nodular, ulcerative, or cellulitic.

More than 75% of fungal infections requiring hospitalization in US patients who have undergone transplant are attributed to either Candida, Aspergillus, or Cryptococcus species.¹⁵ Risk of fungal infection is increased with diabetes, duration of pretransplant dialysis, tacrolimus therapy, or rejection treatment.

Much has improved in renal transplantation over the past 20 years. The focus has shifted to using stronger immunotherapy rather than trying to minimize it. There has been increasing recognition of infection and ways to prevent and treat it. Induction therapy now has greater emphasis so that maintenance therapy can be eased, with the aim of reducing long-term toxicity. Perhaps the biggest change is the practice of screening for donor-specific antibodies at the time of transplant so that predictable problems can be prevented or better handled if they occur. Such advances have helped patients directly and by extending the life of their transplanted organs.

LONGER SURVIVAL

As early as the 1990s, it was recognized that kidney transplant offers a survival advantage for patients with end-stage renal disease over maintenance on dialysis.¹ Although the risk of death is higher immediately after transplant, within a few months it becomes much lower than for patients on dialysis. Survival varies according to the health of the patient and the quality of the transplanted organ.

In general, patients who obtain the greatest benefit from transplants in terms of years of life gained are those with diabetes, especially those who are younger. Those ages 20 to 39 live about 8 years on dialysis vs 25 years after transplant.

CONTRAINDICATIONS TO TRANSPLANT

There are multiple contraindications to a solitary kidney transplant (Table 1), including smoking. Most transplant centers require that smokers quit before transplant. Long-standing smokers almost double their risk of a cardiac event after transplant and double their rate of malignancy. Active smoking at the time of transplant is associated with twice the risk of death by 10 years after transplant compared with that of nonsmokers.² Cotinine testing can detect whether a patient is an active smoker.

WAITING-LIST CONSIDERATIONS

Organs are scarce

The number of patients on the kidney waiting list has increased rapidly in the last few decades, while the number of transplants performed each year has remained about the same. In 2016, about 100,000 patients were on the list, but only about 19,000 transplants were performed.³ Wait times, especially for deceased-donor organs, have increased to about 6 years, varying by blood type and geographic region.

Waiting-list placement

Placement on the waiting list for a deceased-donor kidney transplant occurs when a patient has an estimated glomerular filtration rate (GFR) of 20 mL/min/1.73 m² or less, although referral to the list can be made earlier. Early listing remains advantageous, as total time on the list will be counted before starting dialysis. “Preemptive transplant” means the patient had no dialysis before transplant; this applies to about 10% of transplant recipients. These patients tend to fare the best and are usually recipients of a living-donor organ.

Most patients do not receive a transplant until the GFR is less than 15 mL/min/1.73 m².

Since 2014, wait time has been measured from the beginning of dialysis rather than the date of waiting-list placement in patients who are listed after starting dialysis therapy. This approach is more fair but sometimes introduces problems. A patient who did not previously know about the list may suddenly jump to the head of the line after 10 years of dialysis, by which time comorbidities associated with long-term dialysis make the patient less likely to gain as much benefit from a transplant as people lower on the list. Time on dialysis, or “dialysis vintage,” predicts patient and kidney survival after transplant, with reduced survival associated with increasing time on dialysis.⁴

Shorter wait for a suboptimal kidney

The aging population has increased the number of older patients being listed for transplant, presenting multiple challenges. Patients age 65 or older have a 50% chance of dying before they receive a transplant during a 5-year wait.

A patient may shorten the wait by joining the list for a suboptimal organ. All deceased-donor organs are given a Kidney Donor Profile Index score, which predicts the longevity of an organ after transplant. The score is determined by donor age, kidney function based on the serum creatinine at the time of death, and other donor factors.

A kidney with a score higher than 85% is likely to function longer than only 15% of available kidneys. Patients who receive a kidney with that score have a longer period of risk of death soon after transplant and a slightly higher risk of death in the long term than patients who receive a healthier kidney, although on average they still do better than patients on dialysis.⁵

Older patients should be encouraged to sign up for both the regular waiting list and the suboptimal kidney waiting list to reduce the risk of dying before they get a kidney.

Many advantages

Living-donor organ transplant is associated with a better survival rate than deceased-donor organ transplant, and the advantage becomes greater over time. At 1 year, patient survival is more than 90% in both groups, but by 5 years about 80% of patients with a living-donor organ are still alive vs only about 65% of patients with a deceased-donor organ.

The waiting time for a living-donor transplant may be only weeks to months, rather than years. Because increasing time on dialysis predicts worse patient and graft survival after transplant, the shorter wait time is a big advantage. In addition, because the donor and recipient are typically in adjacent operating rooms, the organ sustains less ischemic damage. In general, the kidney quality is better from healthy donors, resulting in superior function early on and longer graft survival by an average of 4 years. If the living donor is related to the recipient, human leukocyte antigen matching also tends to be better and predicts better outcomes.

Special challenges

Opting for a living-donor organ also entails special challenges. In addition to the ethical issues surrounding living-donor organ donation, an appropriate donor must be found. Donors must be highly motivated and pass physical, laboratory, and psychological evaluations.

For older patients, if the donor is a spouse or close friend, he or she is also likely to be older, making the organ less viable than one from a younger person. Even an adult child may not be an ideal donor if there is a family propensity to kidney disease, such as diabetic nephropathy. No test is available to determine the risk for future diabetes, but it is known to run in families.

POTENT IMMUNOSUPPRESSION

Induction therapy

Induction therapy with antithymocyte globulin or basiliximab provides intense immunosuppression to prevent acute rejection during the early posttransplant period.

Antithymocyte globulin is a potent agent that contains antibodies directed at T cells, B cells, neutrophils, platelets, adhesion molecules, and complement. It binds T cells and removes them from circulation by opsonization in splenic and lymphoid tissue. The immunosuppressive effect is sustained for at least 2 to 3 months after a series of injections (dosage 1.5 mg/kg/day, usually for 4 to 10 doses). Antithymocyte globulin is also used to treat acute rejection, especially high-grade rejection for which steroid therapy is likely to be insufficient.

Basiliximab consists of antibodies to the interleukin 2 (IL-2) receptor of T cells. Binding to T cells prevents their activation rather than removing them from circulation. The drug prevents rejection, with 30% relative reduction in early studies compared with placebo. However, it is ineffective in reversing established rejection. Dosage is 20 mg at day 0 and day 4, which provides receptor saturation for 30 to 45 days.

Basiliximab is also sometimes used off-label for patients who need to discontinue a calcineurin inhibitor (ie, tacrolimus or cyclosporine). In such cases, normal therapy is put on hold while basiliximab is given for 1 or 2 doses. Case series have been reported for this use, particularly for patients with a heart and liver transplant who develop acute kidney injury while hospitalized.^6,7

Antithymocyte globulin is more effective but also more risky. Brennan et al⁸ randomized 278 transplant recipients to either antithymocyte globulin or basiliximab. Patients in the antithymocyte globulin group had a 16% rejection rate vs 26% in the basiliximab group.

Antithymocyte globulin therapy is associated with multiple adverse effects, including fever and chills, pulmonary edema, and long-standing immunosuppressive effects such as increased risk of lymphoma and cytomegalovirus (CMV) infection. Basiliximab side-effect profiles are similar to those of placebo.

Maintenance therapy

The calcineurin inhibitors cyclosporine and tacrolimus remain the standard of care in kidney transplant despite multiple drug interactions and side effects that include renal toxicity and fibrosis. Cyclosporine and tacrolimus both bind intracellular immunophilins and thereby prevent transcription of IL-2 and production of T cells. The drugs work similarly but have different binding sites. Cyclosporine has largely been replaced by tacrolimus because its reliability of dosing and higher potency are associated with lower rejection rates.

Tacrolimus is typically given twice daily (1–6 mg/dose). Twelve-hour trough levels are followed (target: 8–12 ng/mL early on, then 5–8 ng/mL after 3 months posttransplant). Side effects include hypertension and hypercholesterolemia, but less so than with cyclosporine. On the other hand, hyperglycemia tends to be worse with tacrolimus than with cyclosporine, and combining tacrolimus with steroids frequently leads to diabetes. Tacrolimus can also cause acute and chronic renal failure, especially at high drug levels, as well as neurotoxicity, tremors, and hair loss.

Cyclosporine, tacrolimus, and sirolimus (not a calcineurin inhibitor) are metabolized through the same cytochrome P450 pathway (CYP3A4), so they have common drug interactions (Table 2).

Mycophenolate mofetil is typically used as an adjunct therapy (500–1,000 mg twice daily). It is also used for other kidney diseases before transplant, including lupus nephritis. Transplanted kidney rejection rates with mycophenolate mofetil with steroids are about 40%, so the drug is not potent enough to be used without a calcineurin inhibitor.

Side effects include gastrointestinal toxicity in up to 20% of patients, and leukopenia, which is associated with viral infections.

CORONARY ARTERY DISEASE IS COMMON WITH DIALYSIS

Coronary artery disease is highly associated with end-stage kidney disease and occurs in as many as 85% of older patients with diabetes on dialysis. Although patients with end-stage kidney disease tend to have more numerous and severe atherosclerotic lesions compared with the general population, justifying aggressive management, cardiac care tends to be conservative in patients on dialysis.⁹

Death from acute myocardial infarction occurs in about 20% to 30% of patients on dialysis vs about 2% of patients with normal renal function. Five years after myocardial infarction, survival is only about 30% in patients on dialysis.⁹

There are many explanations for excess coronary artery disease in patients on dialysis. In addition to the traditional cardiovascular risk factors of diabetes, hypertension, and preexisting coronary artery disease, patients are in a proinflammatory uremic state and have high levels of phosphorus and fibroblast growth factor 23 that contribute to vascular calcification. Almost all patients have high homocysteine levels and hemodynamic instability, particularly if they are on hemodialysis.

Pretransplant evaluation for heart disease

Patients on the kidney transplant waiting list are screened aggressively for heart disease. A history of myocardial infarction usually results in removal from the list. All patients have an initial electrocardiogram and echocardiogram. Thallium or echocardiographic stress testing is used for patients who are age 50 and older, have diabetes, or have had dialysis for many years. Patients with evidence of ischemia undergo catheterization.

Patients are also screened with computed tomography before transplant. Because the kidney is typically anastomosed to the iliac artery and vein, heavy calcification of the iliac artery can make the procedure too difficult to perform.

Kasiske et al¹⁰ analyzed data from more than 50,000 patients from the US Renal Data System and found that, for about the first year after transplant, patients who underwent kidney transplant were more likely to have a myocardial infarction than those on dialysis. After that, they fared better than patients who remained on dialysis. Those with a living-donor transplant were less likely at all times to have a myocardial infarction than those with a deceased-donor transplant. By 3 years after transplant, the relative risk of having a myocardial infarction was 0.89 for deceased-donor organ recipients and 0.69 for living-donor recipients compared with patients on the waiting list.¹⁰

INFECTIOUS COMPLICATIONS IN KIDNEY RECIPIENTS

Kidney recipients are prone to many common and uncommon infections (Table 3). All potential recipients are tested pretransplant for hepatitis B, hepatitis C, human immunodeficiency virus, syphilis, and tuberculosis. A positive result does not necessarily rule out transplant.

The following viral serology tests are also done before transplant:

Epstein-Barr virus (antibodies are positive in about 90% of adults)

CMV (about 70% of adults are seropositive)

Varicella zoster (seronegative patients should be given live-attenuated varicella vaccine).

Risk of transmission of these viruses relates to the serostatus of the donor and recipient before transplant. If a donor is positive for viral antibodies but the recipient is not (a so-called “mismatch”), risk is higher after transplant.

Hepatitis C

Patients with hepatitis C fare better if they get a transplant than if they remain on dialysis, although their posttransplant course is worse compared with transplant patients who do not have hepatitis. Some patients develop accelerated liver disease after kidney transplant. Hepatitis C-related kidney disease—membranous proliferative glomerulonephritis—also occurs, as do comorbidities such as diabetes.

Careful evaluation is warranted before transplant, including liver imaging, alpha-fetoprotein testing, and liver biopsy to evaluate for hepatocellular carcinoma. A patient with advanced fibrosis or cirrhosis may not be a candidate for kidney transplant alone but could possibly receive a combined kidney and liver transplant.

There is a need to determine the best time to treat hepatitis C infection. Patients with advanced liver disease or hepatitis C-related kidney disease would likely benefit from early treatment. However, delaying treatment could shorten the wait time for a deceased-donor organ positive for hepatitis C.  Transplant candidates with active hepatitis C are uniquely considered to accept hepatitis C-positive kidneys, which are often discarded, and may only wait weeks for such a transplant. The shortened kidney survival associated with a hepatitis C-positive kidney may no longer be true with the new antiviral hepatitis C therapy, which has been shown to be effective post-transplant.

Hepatitis B

No cure is available for hepatitis B infection, but it can be well controlled with antiviral therapy. Patients with hepatitis B infection may be candidates for transplant, but they should be stable on antiviral therapy (lamivudine, entecavir, or tenofovir) to eliminate the viral load before transplant, and therapy should be continued afterward. Liver imaging, alpha-fetoprotein levels, and biopsy are recommended for evaluation. All hepatitis B- negative patients should be vaccinated before transplant.

Organs from living or deceased donors that test positive for hepatitis B core antibody, indicating prior exposure, can be considered for transplant in a patient who tests positive for hepatitis B surface antibody, indicating successful vaccination or prior exposure in the recipient. But donors must have negative surface antigen and polymerase chain reaction (PCR) tests that indicate no active hepatitis B infection.

Cytomegalovirus

CMV typically does not appear until prophylactic therapy is stopped. Classic symptoms are fever, leukopenia, and diarrhea. Infection can involve any organ, and patients may present with hepatitis, pancreatitis or, less commonly, pneumonitis.

Patients who are negative for CMV before transplant and receive a donor-positive organ are at the highest risk. Patients who are CMV IgG-positive are considered to be at intermediate risk, regardless of the donor status. Patients who are negative for CMV and receive a donor-negative organ are at the lowest risk and do not need prophylaxis with valganciclovir.

CMV infection is diagnosed by PCR testing of the blood or immunostaining in tissue biopsy. Occasionally, blood testing is negative in the face of tissue-based disease.

BK virus

BK is a polyoma virus and a common virus associated with kidney transplant. Viremia is seen in about 18% of patients, whereas actual kidney disease associated with a higher level of virus is seen in fewer than 10% of patients. Most people are exposed to BK virus, often in childhood, and it can remain indolent in the bladder and uroepithelium.

Patients can develop BK nephropathy after exposure to transplant immunosuppression.¹¹ Posttransplant monitoring protocols typically include PCR testing for BK virus at 1, 3, 6, and 12 months. No agent has been identified to specifically treat BK virus. The general strategy is to minimize immunosuppressive therapy by reducing or eliminating mycophenolate mofetil. Fortunately, BK virus does not tend to recur, and patients can have a low-level viremia (< 10,000 copies/mL) persisting over months or even years but often without clinical consequences.

The appearance of BK virus on biopsy can mimic acute rejection. Before BK viral nephropathy was a recognized entity, patients would have been diagnosed with acute rejection and may have been put on high-dose steroids, which would have worsened the BK infection.

Posttransplant lymphoproliferative disorder

Posttransplant lymphoproliferative disorder is most often associated with Epstein-Barr virus and usually involves a large, diffuse B-cell lymphoma. Burkitt lymphoma and plasma cell neoplasms also can occur less commonly.

The condition is about 30 times more common in patients after transplant than in the general population, and it is the third most common malignancy in transplant patients after skin and cervical cancers. About 80% of the cases occur early after transplant, within the first year.

Patients typically have a marked elevation in viral load of Epstein-Barr virus, although a negative viral load does not rule it out. A patient who is serologically negative for Epstein-Barr virus receiving a donor-positive kidney is at highest risk; this situation is most often seen in the pediatric population. Potent induction therapies (eg, antilymphocyte antibody therapy) are also associated with posttransplant lymphoproliferative disorder.

Patients typically present with fever of unknown origin with no localizing signs or symptoms. Mass lesions can be challenging to find; positron emission tomography may be helpful. The culprit is usually a focal mass, ulcer (especially in the gastrointestinal tract), or infiltrate (commonly localized to the allograft). Multifocal or disseminated disease can also occur, including lymphoma or with central nervous system, gastrointestinal, or pulmonary involvement.

Biopsy of the affected site is required for histopathology and Epstein-Barr virus markers. PCR blood testing is often positive for Epstein-Barr virus.

Typical antiviral therapy does not eliminate Epstein-Barr virus. In early polyclonal viral proliferation, the first goal is to reduce immunosuppressive therapy. Rituximab alone may also help in polymorphic cases. With disease that is clearly monomorphic and has transformed to a true malignancy, cytotoxic chemotherapy is also required. “R-CHOP,” a combination therapy consisting of rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone, is usually used. Radiation therapy may help in some cases.

Cryptococcal infection

Previously seen in patients with acquired immune deficiency syndrome, cryptococcal infection is now most commonly encountered in patients with solid-organ transplants. Vilchez et al¹² found a 1% incidence in a series of more than 5,000 patients who had received an organ transplant.

Immunosuppression likely conveys risk, but because cryptococcal infection is acquired, environmental exposure also plays a role. It tends to appear more than 6 months after transplant, indicating that its cause is a primary infection by spore inhalation rather than by reactivation or transmission from the donor organ.¹³ Bird exposure is a risk factor for cryptococcal infection. One case identified the same strain of Cryptococcus in a kidney transplant recipient and the family’s pet cockatoo.¹⁴

Cryptococcal infection typically starts as pneumonia, which may be subclinical. The infection can then disseminate, with meningitis presenting with headache and mental status changes being the most concerning complication. The death rate is about 50% in most series of patients with meningitis. Skin and soft-tissue manifestations may also occur in 10% to 15% of cases and can be nodular, ulcerative, or cellulitic.

More than 75% of fungal infections requiring hospitalization in US patients who have undergone transplant are attributed to either Candida, Aspergillus, or Cryptococcus species.¹⁵ Risk of fungal infection is increased with diabetes, duration of pretransplant dialysis, tacrolimus therapy, or rejection treatment.

Much has improved in renal transplantation over the past 20 years. The focus has shifted to using stronger immunotherapy rather than trying to minimize it. There has been increasing recognition of infection and ways to prevent and treat it. Induction therapy now has greater emphasis so that maintenance therapy can be eased, with the aim of reducing long-term toxicity. Perhaps the biggest change is the practice of screening for donor-specific antibodies at the time of transplant so that predictable problems can be prevented or better handled if they occur. Such advances have helped patients directly and by extending the life of their transplanted organs.

LONGER SURVIVAL

As early as the 1990s, it was recognized that kidney transplant offers a survival advantage for patients with end-stage renal disease over maintenance on dialysis.¹ Although the risk of death is higher immediately after transplant, within a few months it becomes much lower than for patients on dialysis. Survival varies according to the health of the patient and the quality of the transplanted organ.

In general, patients who obtain the greatest benefit from transplants in terms of years of life gained are those with diabetes, especially those who are younger. Those ages 20 to 39 live about 8 years on dialysis vs 25 years after transplant.

CONTRAINDICATIONS TO TRANSPLANT

There are multiple contraindications to a solitary kidney transplant (Table 1), including smoking. Most transplant centers require that smokers quit before transplant. Long-standing smokers almost double their risk of a cardiac event after transplant and double their rate of malignancy. Active smoking at the time of transplant is associated with twice the risk of death by 10 years after transplant compared with that of nonsmokers.² Cotinine testing can detect whether a patient is an active smoker.

WAITING-LIST CONSIDERATIONS

Organs are scarce

The number of patients on the kidney waiting list has increased rapidly in the last few decades, while the number of transplants performed each year has remained about the same. In 2016, about 100,000 patients were on the list, but only about 19,000 transplants were performed.³ Wait times, especially for deceased-donor organs, have increased to about 6 years, varying by blood type and geographic region.

Waiting-list placement

Placement on the waiting list for a deceased-donor kidney transplant occurs when a patient has an estimated glomerular filtration rate (GFR) of 20 mL/min/1.73 m² or less, although referral to the list can be made earlier. Early listing remains advantageous, as total time on the list will be counted before starting dialysis. “Preemptive transplant” means the patient had no dialysis before transplant; this applies to about 10% of transplant recipients. These patients tend to fare the best and are usually recipients of a living-donor organ.

Most patients do not receive a transplant until the GFR is less than 15 mL/min/1.73 m².

Since 2014, wait time has been measured from the beginning of dialysis rather than the date of waiting-list placement in patients who are listed after starting dialysis therapy. This approach is more fair but sometimes introduces problems. A patient who did not previously know about the list may suddenly jump to the head of the line after 10 years of dialysis, by which time comorbidities associated with long-term dialysis make the patient less likely to gain as much benefit from a transplant as people lower on the list. Time on dialysis, or “dialysis vintage,” predicts patient and kidney survival after transplant, with reduced survival associated with increasing time on dialysis.⁴

Shorter wait for a suboptimal kidney

The aging population has increased the number of older patients being listed for transplant, presenting multiple challenges. Patients age 65 or older have a 50% chance of dying before they receive a transplant during a 5-year wait.

A patient may shorten the wait by joining the list for a suboptimal organ. All deceased-donor organs are given a Kidney Donor Profile Index score, which predicts the longevity of an organ after transplant. The score is determined by donor age, kidney function based on the serum creatinine at the time of death, and other donor factors.

A kidney with a score higher than 85% is likely to function longer than only 15% of available kidneys. Patients who receive a kidney with that score have a longer period of risk of death soon after transplant and a slightly higher risk of death in the long term than patients who receive a healthier kidney, although on average they still do better than patients on dialysis.⁵

Older patients should be encouraged to sign up for both the regular waiting list and the suboptimal kidney waiting list to reduce the risk of dying before they get a kidney.

Many advantages

Living-donor organ transplant is associated with a better survival rate than deceased-donor organ transplant, and the advantage becomes greater over time. At 1 year, patient survival is more than 90% in both groups, but by 5 years about 80% of patients with a living-donor organ are still alive vs only about 65% of patients with a deceased-donor organ.

The waiting time for a living-donor transplant may be only weeks to months, rather than years. Because increasing time on dialysis predicts worse patient and graft survival after transplant, the shorter wait time is a big advantage. In addition, because the donor and recipient are typically in adjacent operating rooms, the organ sustains less ischemic damage. In general, the kidney quality is better from healthy donors, resulting in superior function early on and longer graft survival by an average of 4 years. If the living donor is related to the recipient, human leukocyte antigen matching also tends to be better and predicts better outcomes.

Special challenges

Opting for a living-donor organ also entails special challenges. In addition to the ethical issues surrounding living-donor organ donation, an appropriate donor must be found. Donors must be highly motivated and pass physical, laboratory, and psychological evaluations.

For older patients, if the donor is a spouse or close friend, he or she is also likely to be older, making the organ less viable than one from a younger person. Even an adult child may not be an ideal donor if there is a family propensity to kidney disease, such as diabetic nephropathy. No test is available to determine the risk for future diabetes, but it is known to run in families.

POTENT IMMUNOSUPPRESSION

Induction therapy

Induction therapy with antithymocyte globulin or basiliximab provides intense immunosuppression to prevent acute rejection during the early posttransplant period.

Antithymocyte globulin is a potent agent that contains antibodies directed at T cells, B cells, neutrophils, platelets, adhesion molecules, and complement. It binds T cells and removes them from circulation by opsonization in splenic and lymphoid tissue. The immunosuppressive effect is sustained for at least 2 to 3 months after a series of injections (dosage 1.5 mg/kg/day, usually for 4 to 10 doses). Antithymocyte globulin is also used to treat acute rejection, especially high-grade rejection for which steroid therapy is likely to be insufficient.

Basiliximab consists of antibodies to the interleukin 2 (IL-2) receptor of T cells. Binding to T cells prevents their activation rather than removing them from circulation. The drug prevents rejection, with 30% relative reduction in early studies compared with placebo. However, it is ineffective in reversing established rejection. Dosage is 20 mg at day 0 and day 4, which provides receptor saturation for 30 to 45 days.

Basiliximab is also sometimes used off-label for patients who need to discontinue a calcineurin inhibitor (ie, tacrolimus or cyclosporine). In such cases, normal therapy is put on hold while basiliximab is given for 1 or 2 doses. Case series have been reported for this use, particularly for patients with a heart and liver transplant who develop acute kidney injury while hospitalized.^6,7

Antithymocyte globulin is more effective but also more risky. Brennan et al⁸ randomized 278 transplant recipients to either antithymocyte globulin or basiliximab. Patients in the antithymocyte globulin group had a 16% rejection rate vs 26% in the basiliximab group.

Antithymocyte globulin therapy is associated with multiple adverse effects, including fever and chills, pulmonary edema, and long-standing immunosuppressive effects such as increased risk of lymphoma and cytomegalovirus (CMV) infection. Basiliximab side-effect profiles are similar to those of placebo.

Maintenance therapy

The calcineurin inhibitors cyclosporine and tacrolimus remain the standard of care in kidney transplant despite multiple drug interactions and side effects that include renal toxicity and fibrosis. Cyclosporine and tacrolimus both bind intracellular immunophilins and thereby prevent transcription of IL-2 and production of T cells. The drugs work similarly but have different binding sites. Cyclosporine has largely been replaced by tacrolimus because its reliability of dosing and higher potency are associated with lower rejection rates.

Tacrolimus is typically given twice daily (1–6 mg/dose). Twelve-hour trough levels are followed (target: 8–12 ng/mL early on, then 5–8 ng/mL after 3 months posttransplant). Side effects include hypertension and hypercholesterolemia, but less so than with cyclosporine. On the other hand, hyperglycemia tends to be worse with tacrolimus than with cyclosporine, and combining tacrolimus with steroids frequently leads to diabetes. Tacrolimus can also cause acute and chronic renal failure, especially at high drug levels, as well as neurotoxicity, tremors, and hair loss.

Cyclosporine, tacrolimus, and sirolimus (not a calcineurin inhibitor) are metabolized through the same cytochrome P450 pathway (CYP3A4), so they have common drug interactions (Table 2).

Mycophenolate mofetil is typically used as an adjunct therapy (500–1,000 mg twice daily). It is also used for other kidney diseases before transplant, including lupus nephritis. Transplanted kidney rejection rates with mycophenolate mofetil with steroids are about 40%, so the drug is not potent enough to be used without a calcineurin inhibitor.

Side effects include gastrointestinal toxicity in up to 20% of patients, and leukopenia, which is associated with viral infections.

CORONARY ARTERY DISEASE IS COMMON WITH DIALYSIS

Coronary artery disease is highly associated with end-stage kidney disease and occurs in as many as 85% of older patients with diabetes on dialysis. Although patients with end-stage kidney disease tend to have more numerous and severe atherosclerotic lesions compared with the general population, justifying aggressive management, cardiac care tends to be conservative in patients on dialysis.⁹

Death from acute myocardial infarction occurs in about 20% to 30% of patients on dialysis vs about 2% of patients with normal renal function. Five years after myocardial infarction, survival is only about 30% in patients on dialysis.⁹

There are many explanations for excess coronary artery disease in patients on dialysis. In addition to the traditional cardiovascular risk factors of diabetes, hypertension, and preexisting coronary artery disease, patients are in a proinflammatory uremic state and have high levels of phosphorus and fibroblast growth factor 23 that contribute to vascular calcification. Almost all patients have high homocysteine levels and hemodynamic instability, particularly if they are on hemodialysis.

Pretransplant evaluation for heart disease

Patients on the kidney transplant waiting list are screened aggressively for heart disease. A history of myocardial infarction usually results in removal from the list. All patients have an initial electrocardiogram and echocardiogram. Thallium or echocardiographic stress testing is used for patients who are age 50 and older, have diabetes, or have had dialysis for many years. Patients with evidence of ischemia undergo catheterization.

Patients are also screened with computed tomography before transplant. Because the kidney is typically anastomosed to the iliac artery and vein, heavy calcification of the iliac artery can make the procedure too difficult to perform.

Kasiske et al¹⁰ analyzed data from more than 50,000 patients from the US Renal Data System and found that, for about the first year after transplant, patients who underwent kidney transplant were more likely to have a myocardial infarction than those on dialysis. After that, they fared better than patients who remained on dialysis. Those with a living-donor transplant were less likely at all times to have a myocardial infarction than those with a deceased-donor transplant. By 3 years after transplant, the relative risk of having a myocardial infarction was 0.89 for deceased-donor organ recipients and 0.69 for living-donor recipients compared with patients on the waiting list.¹⁰

INFECTIOUS COMPLICATIONS IN KIDNEY RECIPIENTS

Kidney recipients are prone to many common and uncommon infections (Table 3). All potential recipients are tested pretransplant for hepatitis B, hepatitis C, human immunodeficiency virus, syphilis, and tuberculosis. A positive result does not necessarily rule out transplant.

The following viral serology tests are also done before transplant:

Epstein-Barr virus (antibodies are positive in about 90% of adults)

CMV (about 70% of adults are seropositive)

Varicella zoster (seronegative patients should be given live-attenuated varicella vaccine).

Risk of transmission of these viruses relates to the serostatus of the donor and recipient before transplant. If a donor is positive for viral antibodies but the recipient is not (a so-called “mismatch”), risk is higher after transplant.

Hepatitis C

Patients with hepatitis C fare better if they get a transplant than if they remain on dialysis, although their posttransplant course is worse compared with transplant patients who do not have hepatitis. Some patients develop accelerated liver disease after kidney transplant. Hepatitis C-related kidney disease—membranous proliferative glomerulonephritis—also occurs, as do comorbidities such as diabetes.

Careful evaluation is warranted before transplant, including liver imaging, alpha-fetoprotein testing, and liver biopsy to evaluate for hepatocellular carcinoma. A patient with advanced fibrosis or cirrhosis may not be a candidate for kidney transplant alone but could possibly receive a combined kidney and liver transplant.

There is a need to determine the best time to treat hepatitis C infection. Patients with advanced liver disease or hepatitis C-related kidney disease would likely benefit from early treatment. However, delaying treatment could shorten the wait time for a deceased-donor organ positive for hepatitis C.  Transplant candidates with active hepatitis C are uniquely considered to accept hepatitis C-positive kidneys, which are often discarded, and may only wait weeks for such a transplant. The shortened kidney survival associated with a hepatitis C-positive kidney may no longer be true with the new antiviral hepatitis C therapy, which has been shown to be effective post-transplant.

Hepatitis B

No cure is available for hepatitis B infection, but it can be well controlled with antiviral therapy. Patients with hepatitis B infection may be candidates for transplant, but they should be stable on antiviral therapy (lamivudine, entecavir, or tenofovir) to eliminate the viral load before transplant, and therapy should be continued afterward. Liver imaging, alpha-fetoprotein levels, and biopsy are recommended for evaluation. All hepatitis B- negative patients should be vaccinated before transplant.

Organs from living or deceased donors that test positive for hepatitis B core antibody, indicating prior exposure, can be considered for transplant in a patient who tests positive for hepatitis B surface antibody, indicating successful vaccination or prior exposure in the recipient. But donors must have negative surface antigen and polymerase chain reaction (PCR) tests that indicate no active hepatitis B infection.

Cytomegalovirus

CMV typically does not appear until prophylactic therapy is stopped. Classic symptoms are fever, leukopenia, and diarrhea. Infection can involve any organ, and patients may present with hepatitis, pancreatitis or, less commonly, pneumonitis.

Patients who are negative for CMV before transplant and receive a donor-positive organ are at the highest risk. Patients who are CMV IgG-positive are considered to be at intermediate risk, regardless of the donor status. Patients who are negative for CMV and receive a donor-negative organ are at the lowest risk and do not need prophylaxis with valganciclovir.

CMV infection is diagnosed by PCR testing of the blood or immunostaining in tissue biopsy. Occasionally, blood testing is negative in the face of tissue-based disease.

BK virus

BK is a polyoma virus and a common virus associated with kidney transplant. Viremia is seen in about 18% of patients, whereas actual kidney disease associated with a higher level of virus is seen in fewer than 10% of patients. Most people are exposed to BK virus, often in childhood, and it can remain indolent in the bladder and uroepithelium.

Patients can develop BK nephropathy after exposure to transplant immunosuppression.¹¹ Posttransplant monitoring protocols typically include PCR testing for BK virus at 1, 3, 6, and 12 months. No agent has been identified to specifically treat BK virus. The general strategy is to minimize immunosuppressive therapy by reducing or eliminating mycophenolate mofetil. Fortunately, BK virus does not tend to recur, and patients can have a low-level viremia (< 10,000 copies/mL) persisting over months or even years but often without clinical consequences.

The appearance of BK virus on biopsy can mimic acute rejection. Before BK viral nephropathy was a recognized entity, patients would have been diagnosed with acute rejection and may have been put on high-dose steroids, which would have worsened the BK infection.

Posttransplant lymphoproliferative disorder

Posttransplant lymphoproliferative disorder is most often associated with Epstein-Barr virus and usually involves a large, diffuse B-cell lymphoma. Burkitt lymphoma and plasma cell neoplasms also can occur less commonly.

The condition is about 30 times more common in patients after transplant than in the general population, and it is the third most common malignancy in transplant patients after skin and cervical cancers. About 80% of the cases occur early after transplant, within the first year.

Patients typically have a marked elevation in viral load of Epstein-Barr virus, although a negative viral load does not rule it out. A patient who is serologically negative for Epstein-Barr virus receiving a donor-positive kidney is at highest risk; this situation is most often seen in the pediatric population. Potent induction therapies (eg, antilymphocyte antibody therapy) are also associated with posttransplant lymphoproliferative disorder.

Patients typically present with fever of unknown origin with no localizing signs or symptoms. Mass lesions can be challenging to find; positron emission tomography may be helpful. The culprit is usually a focal mass, ulcer (especially in the gastrointestinal tract), or infiltrate (commonly localized to the allograft). Multifocal or disseminated disease can also occur, including lymphoma or with central nervous system, gastrointestinal, or pulmonary involvement.

Biopsy of the affected site is required for histopathology and Epstein-Barr virus markers. PCR blood testing is often positive for Epstein-Barr virus.

Typical antiviral therapy does not eliminate Epstein-Barr virus. In early polyclonal viral proliferation, the first goal is to reduce immunosuppressive therapy. Rituximab alone may also help in polymorphic cases. With disease that is clearly monomorphic and has transformed to a true malignancy, cytotoxic chemotherapy is also required. “R-CHOP,” a combination therapy consisting of rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone, is usually used. Radiation therapy may help in some cases.

Cryptococcal infection

Previously seen in patients with acquired immune deficiency syndrome, cryptococcal infection is now most commonly encountered in patients with solid-organ transplants. Vilchez et al¹² found a 1% incidence in a series of more than 5,000 patients who had received an organ transplant.

Immunosuppression likely conveys risk, but because cryptococcal infection is acquired, environmental exposure also plays a role. It tends to appear more than 6 months after transplant, indicating that its cause is a primary infection by spore inhalation rather than by reactivation or transmission from the donor organ.¹³ Bird exposure is a risk factor for cryptococcal infection. One case identified the same strain of Cryptococcus in a kidney transplant recipient and the family’s pet cockatoo.¹⁴

Cryptococcal infection typically starts as pneumonia, which may be subclinical. The infection can then disseminate, with meningitis presenting with headache and mental status changes being the most concerning complication. The death rate is about 50% in most series of patients with meningitis. Skin and soft-tissue manifestations may also occur in 10% to 15% of cases and can be nodular, ulcerative, or cellulitic.

More than 75% of fungal infections requiring hospitalization in US patients who have undergone transplant are attributed to either Candida, Aspergillus, or Cryptococcus species.¹⁵ Risk of fungal infection is increased with diabetes, duration of pretransplant dialysis, tacrolimus therapy, or rejection treatment.

Multiple randomized controlled trials have compared percutaneous coronary intervention (PCI) vs optimal medical therapy for patients with chronic stable angina. All have consistently shown that PCI does not reduce the risk of death or even myocardial infarction (MI) but that it may relieve angina temporarily. Nevertheless, PCI is still commonly performed for patients with stable coronary disease, often in the absence of angina, and patients mistakenly believe the procedure is life-saving. Cardiologists may not be aware of patients’ misperceptions, or worse, may encourage them. In either case, if patients do not understand the benefits of the procedure, they cannot give informed consent.

See related editorial

This article reviews the pathophysiology of coronary artery disease, evidence from clinical trials of the value of PCI for chronic stable angina, patient and physician perceptions of PCI, and ways to promote patient-centered, shared decision-making.

CLINICAL CASE: EXERTIONAL ANGINA

While climbing 4 flights of stairs, a 55-year-old man noticed tightness in his chest, which lasted for 5 minutes and resolved spontaneously. Several weeks later, when visiting his primary care physician, he mentioned the episode. He had had no symptoms in the interim, but the physician ordered an exercise stress test.

Six minutes into a standard Bruce protocol, the patient experienced the same chest tightness, accompanied by 1-mm ST-segment depressions in leads II, III, and aVF. He was then referred to a cardiologist, who recommended catheterization.

Catheterization demonstrated a 95% stenosis of the right coronary artery with nonsignificant stenoses of the left anterior descending and circumflex arteries. A drug-eluting stent was placed in the right coronary artery, with no residual stenosis.

Did this intervention likely prevent an MI and perhaps save the man’s life?

HOW MYOCARDIAL INFARCTION HAPPENS

Understanding the pathogenesis of MI is critical to having realistic expectations of the benefits of stent placement.

Doctors often describe coronary atherosclerosis as a plumbing problem, where deposits of cholesterol and fat build up in arterial walls, clogging the pipes and eventually causing a heart attack. This analogy, which has been around since the 1950s, is easy to for patients to grasp and has been popularized in the press and internalized by the public—as one patient with a 95% stenosis put it, “I was 95% dead.” In that model, angioplasty and stenting can resolve the blockage and “fix” the problem, much as a plumber can clear your pipes with a Roto-Rooter.

Despite the visual appeal of this model,¹ it doesn’t accurately convey what we know about the pathophysiology of coronary artery disease. Instead of a gradual buildup of fatty deposits, low-density lipoprotein cholesterol particles infiltrate arterial walls and trigger an inflammatory reaction as they are engulfed by macrophages, leading to a cascade of cytokines and recruitment of more inflammatory cells.² This immune response can eventually cause the rupture of the plaque’s fibrous cap, triggering thrombosis and infarction, often at a site of insignificant stenosis.

In this new model, coronary artery disease is primarily a problem of inflammation distributed throughout the vasculature, rather than a mechanical problem localized to the site of a significant stenosis.

Significant stenosis does not equal unstable plaque

Not all plaques are equally likely to rupture. Stable plaques tend to be long-standing and calcified, with a thick fibrous cap. A stable plaque causing a 95% stenosis may cause symptoms with exertion, but it is unlikely to cause infarction.³ Conversely, rupture-prone plaques may cause little stenosis, but a large and dangerous plaque may be lurking beneath the thin fibrous cap.

Relying on angiography can be misleading. Treating all significant stenoses improves blood flow, but does not reduce the risk of infarction, because infarction most often occurs in areas where the lumen is not obstructed. A plaque causing only 30% stenosis can suddenly rupture, causing thrombosis and complete occlusion.

The current model explains why PCI is no better than optimal medical therapy (ie, risk factor modification, antiplatelet therapy with aspirin, and a statin). Diet, exercise, smoking cessation, and statins target inflammatory processes and lower low-density lipoprotein cholesterol levels, while aspirin prevents platelet aggregation, among other likely actions.

The model also explains why coronary artery bypass grafting reduces the risk of MI and death in patients with left main or 3-vessel disease. A patient with generalized coronary artery disease has multiple lesions, many of which do not cause significant stenoses. PCI corrects only a single stenosis, whereas coronary artery bypass grafting circumvents all the vulnerable plaques in a vessel.

THE LANDMARK COURAGE TRIAL

Published in 2007, the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial⁴ randomized more than 2,000 patients to receive either optimal medical therapy plus PCI or optimal medical therapy alone. The primary outcome was a composite of death from any cause and nonfatal MI. Patients were followed for at least 3 years, and some for as long as 7 years.

There was an initial small upward spike in the primary outcome in the PCI arm due to periprocedural events. By 5 years, the outcomes of the 2 arms converged and then stayed the same for up to 15 years.⁵ The authors concluded that PCI conferred no benefit over optimal medical therapy in the risk of death or MI.

Some doctors dismiss the study because of its stringent entry criteria—of 35,539 patients assessed, only 3,071 met the eligibility criteria. However, the entry criteria were meant to identify patients most likely to benefit from PCI. Many patients who undergo PCI today would not have qualified for the study because they lack objective evidence of ischemia.⁶ To enroll, patients needed a proximal stenosis of at least 70% and objective evidence of ischemia or a coronary stenosis of more than 80% and classic angina. Exclusion criteria disqualified few patients: Canadian Cardiovascular Society class IV angina (ie, angina evoked from minimal activity or at rest); a markedly positive stress test (substantial ST-segment depression or hypotension during stage I of the Bruce protocol); refractory heart failure or cardiogenic shock; an ejection fraction of less than 30%; revascularization within the past 6 months; and coronary anatomy unsuitable for PCI.

Although COURAGE was hailed as a landmark trial, it largely supported the results of previous studies. A meta-analysis of PCI vs optimal medical therapy published in 2005 found no significant differences in death, cardiac death, MI, or nonfatal MI.⁷ MI was actually slightly more common in the PCI group due to the increased risk of MI during the periprocedural period.

Nor has the evidence from COURAGE discouraged additional studies of the same topic. Despite consistent findings that fit with our understanding of coronary disease as inflammation, we continue to conduct studies aimed at addressing significant stenosis, as if that was the problem. Thus, there have been studies of angioplasty alone, followed by studies of bare-metal stents and then drug-eluting stents.

In 2009, Trikalinos et al published a review of 61 randomized controlled trials comprising more than 25,000 patients with stable coronary disease and comparing medical therapy and angioplasty in its various forms over the previous 20 years.⁸ In all direct and indirect comparisons of PCI and medical therapy, there were no improvements in rates of death or MI.

Even so, the studies continue. The most recent “improvement” was the addition of fractional flow reserve, which served as the inclusion criterion for the Fractional Flow Reserve versus Angiography for Multivessel Evaluation 2 (FAME 2) trial.⁹ In that study, patients with at least 1 stenosis with a fractional flow reserve less than 0.80 were randomized to PCI plus medical therapy or to medical therapy alone. The primary end point was a composite of death from any cause, MI, and urgent revascularization. Unfortunately, the study was stopped early when the primary end point was met due to a reduction in the need for urgent revascularization. There was no reduction in the rate of MI (hazard ratio 1.05, 95% confidence interval 0.51–2.19).

The reduction in urgent revascularization has also been shown consistently in past studies, but this is the weakest outcome measure because it does not equate to a reduction in the rate of MI. There is no demonstrable harm to putting off stent placement, even in functionally significant arteries, and most patients do not require a stent, even in the future.

In summary, the primary benefit of getting a stent now is a reduced likelihood of needing one later.

PCI MAY IMPROVE ANGINA FASTER

Another important finding of the COURAGE trial was that PCI improved symptoms more than optimal medical therapy.¹⁰ This is not surprising, because angina is often a direct result of a significant stenosis. What was unexpected was that even after PCI, most patients were not symptom-free. At 1 month, significantly more PCI patients were angina-free (42%) than were medical patients (33%). This translates into an absolute risk reduction of 9% or a number needed to treat of 11 to prevent 1 case of angina.

Patients in both groups improved over time, and after 3 years, the difference between the 2 groups was no longer significant: 59% in the PCI group vs 56% in the medical therapy group were angina-free.

A more recent study has raised the possibility that the improvement in angina with PCI is primarily a placebo effect. Researchers in the United Kingdom randomized patients with stable angina and at least a 70% stenosis of one vessel to either PCI or sham PCI, in which they threaded the catheter but did not deploy the stent.¹¹ All patients received aggressive antianginal therapy before the procedure. At 6 weeks, there was improvement in angina in both groups, but no statistically significant difference between them in either exercise time or angina. Approximately half the patients in each group improved by at least 1 grade on the Canadian Cardiovascular Society angina classification, and more than 20% improved 2 grades.

This finding is not without precedent. Ligation of the internal mammary arteries, a popular treatment for angina in the 1950s, often provided dramatic relief of symptoms, until it was proven to be no better than a sham operation.^12,13 More recently, a placebo-controlled trial of percutaneous laser myocardial revascularization also failed to show improvement over a sham treatment, despite promising results from a phase 1 trial.¹⁴ Together, these studies emphasize the subjective nature of angina as an outcome and call into question the routine use of PCI to relieve it.

PCI ENTAILS RISK

PCI entails a small but not inconsequential risk. During the procedure, 2% of patients develop bleeding or blood vessel damage, and another 1% die or have an MI or a stroke. In the first year after stent placement, 3% of patients have a bleeding event from the antiplatelet therapy needed for the stent, and an additional 2% develop a clot in the stent that leads to MI.¹⁵

INFORMED CONSENT IS CRITICAL

As demonstrated above, for patients with stable angina, the only evidence-based benefit of PCI over optimal medical therapy is that symptoms may respond faster. At the same time, there are costs and risks associated with the procedure. Because symptoms are subjective, patients should play a key role in deciding whether PCI is appropriate for them.

The American Medical Association states that a physician providing any treatment or procedure should disclose and discuss with patients the risks and benefits. Unfortunately, a substantial body of evidence demonstrates that this is not occurring in practice.

Patients and cardiologists have conflicting beliefs about PCI

Studies over the past 20 years demonstrate that patients with chronic stable angina consistently overestimate the benefits of PCI, with 71% to 88% believing that it will reduce their chance of death.^16–19 Patients also understand that PCI can relieve their symptoms, though no study seems to have assessed the perceived magnitude of this benefit.

In contrast, when cardiologists were asked about the benefits their patients could expect from PCI, only 20% said that it would reduce mortality and 25% said it would prevent MI.¹⁸ These are still surprisingly high percentages, since the study was conducted after the COURAGE trial.

Nevertheless, these differences in perception show that cardiologists fail to successfully communicate the benefits of the procedure to their patients. Without complete information, patients cannot make informed decisions.

If PCI cannot improve hard outcomes like MI or death in stable coronary disease, why do cardiologists continue to perform it so frequently?

Soon after the COURAGE trial, Lin et al conducted focus groups with cardiologists to find out.²⁰ Some said that they doubted the clinical trial evidence, given the reduction in the cardiac mortality rate over the past 30 years. Others remarked that their overriding goal is to stamp out ischemia, and that once a lesion is found by catheterization, one must proceed with PCI. This has been termed the “oculostenotic reflex,” ie, the interventionist sees coronary artery disease and immediately places a stent.

Atreya et al found objective evidence of this practice.²¹ In a 2016 study of 207 patients with obstructive lesions amenable to PCI, the only factors associated with medical management were those that increased the risk of the procedure: age, chronic kidney disease, distal location of the lesion, and type C lesions (the most difficult ones to treat by PCI). More important, evidence of ischemia, presence of angina, and being on optimal medical therapy or maximal antianginal therapy were not associated with PCI.

When surveyed, cardiologists offered reasons similar to those identified by Lin et al, including a positive stress test (70%) and significant myocardium at risk (50%).¹⁸ Optimal medical therapy failure was cited less often (40%). Over 30% identified relief of chest pain for patients who were not prescribed optimal medical therapy. Another 30% said that patient anxiety contributed to their decision, but patients who reported anxiety were not more likely to get PCI than those who did not.

True informed consent rarely occurs

Surveys of patients and recordings of doctor visits suggest that doctors often discuss the risks of the procedure but rarely accurately describe the benefits or mention alternative treatments, including optimal medical therapy.

Fowler et al²² surveyed 472 Medicare patients who had undergone PCI in the past year about their consent discussion, particularly regarding alternative options. Only 6% of patients recalled discussing medication as a serious option with their doctor.

In 2 published studies,^23,24 we analyzed recorded conversations between doctors and patients in which angiography and PCI were discussed.

In a qualitative assessment of how cardiologists presented the rationale for PCI to patients,²³ we observed that cardiologists gave an accurate presentation of the benefits in only 5% of cases. In 13% of the conversations the benefits were explicitly overstated (eg, “If you don’t do it [angiogram/PCI], what could happen? Well, you could…have a heart attack involving that area which can lead to a sudden death”). In another 35% of cases, physicians offered an implicit overstatement of the benefit by saying they could “fix” the problem (eg, “So that’s where we start thinking, Well maybe we better try to fix that [blockage]”), without specifically stating that fixing the problem would offer any benefit. Patients were left to fill in the blanks. Conversations frequently focused on the rationale for performing PCI (eg, ischemia on a stress test) and a description of the procedure, rather than on the risks and benefits.

In a quantitative study of the same data set, we assessed how often physicians addressed the 7 elements of informed decision-making as defined by Braddock et al.²⁴

• Explaining the patient’s role in decision-making (ie, that the patient has a choice to make) was present in half of the conversations. Sometimes a doctor would simply say, “The next step is to perform catheterization.”
• Discussion of clinical issues (eg, having a blockage, stress test results) was performed in almost every case, demonstrating physicians’ comfort with that element.
• Discussing treatment alternatives occurred in only 1 in 4 conversations. This was more frequent than previously reported, and appeared most often when patients expressed hesitancy about proceeding to PCI.
• Discussing pros and cons of the alternatives was done in 42%.
• Uncertainty about the procedure (eg, that it might not relieve the angina) was expressed in only 10% of conversations.
• Assessment of patient understanding was done in 65% of cases. This included even minimal efforts (eg, “Do you have any questions?”). More advanced methods such as teach-back were never used.
• Exploration of patient preferences (eg, asking patients which treatment they prefer, or attempting to understand how angina affects a patient’s life) the final element, occurred in 73% of conversations.

Only 3% of the conversations contained all 7 elements. Even using a more relaxed definition of 3 critical elements (ie, discussing clinical issues, treatment alternatives, and pros and cons), only 13% of conversations included them all.

Discussion affects decisions

Informed decision-making is not only important because of its ethical implications. Offering patients more information was associated with their choosing not to have PCI. The probability of a patient undergoing PCI was negatively associated with 3 specific elements of informed decision-making. Patients were less likely to choose PCI if the patient’s role in decision-making was discussed (61% vs 86% chose PCI, P < .03); if alternatives were discussed (31% vs 89% chose PCI, P < .01); and if uncertainties were discussed (17% vs 80% chose PCI, P < .01).

There was also a linear relationship between the total number of elements discussed and the probability of choosing PCI: it ranged from 100% of patients choosing PCI when just 1 element was present to 3% of patients choosing PCI when all 7 elements were present. The relationship is not entirely causal, since doctors were more likely to talk about alternatives and risks if patients hesitated and raised questions. Cautious patients received more information.

From these observational studies, we know that physicians do not generally communicate the benefits of PCI, and patients make incorrect assumptions about the benefits they can expect. We know that those who receive more information are less likely to choose PCI, but what would happen if patients were randomly assigned to receive complete information?

We conducted an online survey of more than 1,000 participants over age 50 who had never undergone PCI, asking them to imagine visiting a cardiologist after having a positive stress test for stable chest pain.²⁵ Three intervention groups read different scenarios couched as information provided by their cardiologist:

• The “standard care” group received no specific information about the effects of PCI on the risk of myocardial infarction
• The “specific information” group was specifically told that PCI does not reduce the risk of myocardial infarction
• The “explanatory information” group was told how medications work and why PCI does not reduce the risk of myocardial infarction.

All 3 groups received information about the risks of PCI, its role in reducing angina, and the risks and benefits of optimal medical therapy.

After reading their scenario, all participants completed an identical questionnaire, which asked if they would opt for PCI, medical therapy, or both. Overall, 55% chose PCI, ranging from 70% in the standard care group to 46% in the group given explanatory information. Rates in the specific-information and explanatory-information groups were not statistically different from each other, but both were significantly different from that in the standard-care group. Interestingly, the more information patients were given about PCI, the more likely they were to choose optimal medical therapy.

After reading the scenario, participants were also asked if PCI would “prevent a heart attack.” Of those who received standard care, 71% endorsed that belief, which is remarkably similar to studies of real patients who have received standard care. In contrast, only 39% of those given specific information and 31% given explanatory information held that belief. Moreover, the belief that PCI prevented MI was the strongest predictor of choosing PCI (odds ratio 5.82, 95% confidence interval 4.13–8.26).²⁵

Interestingly, 52% of the standard care group falsely remembered that the doctor had told them that PCI would prevent an MI, even though the doctor said nothing about it one way or the other. It appears that participants were projecting their own beliefs onto the encounter. This highlights the importance of providing full information to patients who are considering this procedure.

TOWARD SHARED DECISION-MAKING

Shared decision-making is a process in which physicians enter into a partnership with a patient, offer information, elicit the patient’s preferences, and then come to a decision in concert with the patient.

Although many decisions can and should involve elements of shared decision-making, the decision to proceed with PCI for stable angina is particularly well-suited to shared decision-making. This is because the benefit of PCI depends on the value a patient attaches to being free of angina sooner. Since there is no difference in the risk of MI or death, the patient must decide if the risks of the procedure and the inconvenience of taking dual antiplatelet therapy are worth the benefit of improving symptoms faster. Presumably, patients who have more severe symptoms or experienced side effects from antianginal therapy would be more likely to choose PCI.

Despite having substantial experience educating patients, most physicians are unfamiliar with the process of shared decision-making. In particular, the process of eliciting preferences is often overlooked.

To address this issue, researchers at the Mayo Clinic developed a decision aid that compares PCI plus optimal medical therapy vs optimal medical therapy alone in an easily understandable information card.¹⁵ On one side, the 2 options are clearly stated, with the magnitude of symptom improvement over time graphically illustrated and the statement, “NO DIFFERENCE in heart attack or death,” prominently displayed. The back of the card discusses the risks of each option in easily understood tables.

The decision aid was compared with standard care in a randomized trial involving patients who were referred for catheterization and possible PCI.²⁶ The decision aid improved patients’ overall knowledge about PCI. In particular, 60% of those who used the decision aid knew that PCI did not prevent death or MI vs 40% of usual-care patients—results similar to those of the online experiment.

Interestingly, the decision about whether to undergo PCI did not differ significantly between the 2 groups, although there was a trend toward more patients in the decision-aid group choosing medical therapy alone (53%) vs the standard-care patients (39%).

To understand why the decision aid did not make more of a difference, the investigators performed qualitative interviews of the cardiologists in the study.²⁷ One theme was the timing of the intervention. Patients using the decision aid had already been referred for catheterization, and some felt the process should have occurred earlier. Engaging in shared decision-making with a general cardiologist before referral could help to improve the quality of patient decisions.

Cardiologists also noted the difficulty in changing their work flow to incorporate the decision aid. Although some embraced the idea of shared decision-making, others were concerned that many patients could not participate, and there was confusion about the difference between an educational tool, which could be used by a patient alone, and a decision aid, which is meant to generate discussion between the doctor and patient. Some expressed interest in using the tool in the future.

These findings serve to emphasize that providing information alone is not enough. If the physician does not “buy in” to the idea of shared decision-making, it will not occur.

PRACTICE IMPLICATIONS

Based on the pathophysiology of coronary artery disease and the results of multiple randomized controlled trials, it is evident that PCI does not prevent heart attacks in patients with chronic stable angina. However, most patients who undergo PCI are unaware of this and therefore do not truly give informed consent. In the absence of explicit information to the contrary, most patients with stable angina assume that PCI prevents MI and thus are biased toward choosing PCI.

Even minimal amounts of explicit information can partially overcome that bias and influence decision-making. In particular, explaining why PCI does not prevent MI was the most effective means of overcoming the bias.

To this end, shared decision aids may help physicians to engage in shared decision-making. Shared decision-making is most likely to occur if physicians are trained in the concept of shared decision-making, are committed to practicing it, and can fit it into their work flow. Ideally, this would occur in the office of a general cardiologist before referral for PCI.

For those practicing in accountable-care organizations, Medicare has recently introduced the shared decision-making model for 6 preference-sensitive conditions, including stable ischemic heart disease. Participants in this program will have the opportunity to receive payments for shared decision-making services and to share in any savings that result from reduced use of resources. Use of these tools holds the promise for providing more patient-centered care at lower cost.

Multiple randomized controlled trials have compared percutaneous coronary intervention (PCI) vs optimal medical therapy for patients with chronic stable angina. All have consistently shown that PCI does not reduce the risk of death or even myocardial infarction (MI) but that it may relieve angina temporarily. Nevertheless, PCI is still commonly performed for patients with stable coronary disease, often in the absence of angina, and patients mistakenly believe the procedure is life-saving. Cardiologists may not be aware of patients’ misperceptions, or worse, may encourage them. In either case, if patients do not understand the benefits of the procedure, they cannot give informed consent.

See related editorial

This article reviews the pathophysiology of coronary artery disease, evidence from clinical trials of the value of PCI for chronic stable angina, patient and physician perceptions of PCI, and ways to promote patient-centered, shared decision-making.

CLINICAL CASE: EXERTIONAL ANGINA

While climbing 4 flights of stairs, a 55-year-old man noticed tightness in his chest, which lasted for 5 minutes and resolved spontaneously. Several weeks later, when visiting his primary care physician, he mentioned the episode. He had had no symptoms in the interim, but the physician ordered an exercise stress test.

Six minutes into a standard Bruce protocol, the patient experienced the same chest tightness, accompanied by 1-mm ST-segment depressions in leads II, III, and aVF. He was then referred to a cardiologist, who recommended catheterization.

Catheterization demonstrated a 95% stenosis of the right coronary artery with nonsignificant stenoses of the left anterior descending and circumflex arteries. A drug-eluting stent was placed in the right coronary artery, with no residual stenosis.

Did this intervention likely prevent an MI and perhaps save the man’s life?

HOW MYOCARDIAL INFARCTION HAPPENS

Understanding the pathogenesis of MI is critical to having realistic expectations of the benefits of stent placement.

Doctors often describe coronary atherosclerosis as a plumbing problem, where deposits of cholesterol and fat build up in arterial walls, clogging the pipes and eventually causing a heart attack. This analogy, which has been around since the 1950s, is easy to for patients to grasp and has been popularized in the press and internalized by the public—as one patient with a 95% stenosis put it, “I was 95% dead.” In that model, angioplasty and stenting can resolve the blockage and “fix” the problem, much as a plumber can clear your pipes with a Roto-Rooter.

Despite the visual appeal of this model,¹ it doesn’t accurately convey what we know about the pathophysiology of coronary artery disease. Instead of a gradual buildup of fatty deposits, low-density lipoprotein cholesterol particles infiltrate arterial walls and trigger an inflammatory reaction as they are engulfed by macrophages, leading to a cascade of cytokines and recruitment of more inflammatory cells.² This immune response can eventually cause the rupture of the plaque’s fibrous cap, triggering thrombosis and infarction, often at a site of insignificant stenosis.

In this new model, coronary artery disease is primarily a problem of inflammation distributed throughout the vasculature, rather than a mechanical problem localized to the site of a significant stenosis.

Significant stenosis does not equal unstable plaque

Not all plaques are equally likely to rupture. Stable plaques tend to be long-standing and calcified, with a thick fibrous cap. A stable plaque causing a 95% stenosis may cause symptoms with exertion, but it is unlikely to cause infarction.³ Conversely, rupture-prone plaques may cause little stenosis, but a large and dangerous plaque may be lurking beneath the thin fibrous cap.

Relying on angiography can be misleading. Treating all significant stenoses improves blood flow, but does not reduce the risk of infarction, because infarction most often occurs in areas where the lumen is not obstructed. A plaque causing only 30% stenosis can suddenly rupture, causing thrombosis and complete occlusion.

The current model explains why PCI is no better than optimal medical therapy (ie, risk factor modification, antiplatelet therapy with aspirin, and a statin). Diet, exercise, smoking cessation, and statins target inflammatory processes and lower low-density lipoprotein cholesterol levels, while aspirin prevents platelet aggregation, among other likely actions.

The model also explains why coronary artery bypass grafting reduces the risk of MI and death in patients with left main or 3-vessel disease. A patient with generalized coronary artery disease has multiple lesions, many of which do not cause significant stenoses. PCI corrects only a single stenosis, whereas coronary artery bypass grafting circumvents all the vulnerable plaques in a vessel.

THE LANDMARK COURAGE TRIAL

Published in 2007, the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial⁴ randomized more than 2,000 patients to receive either optimal medical therapy plus PCI or optimal medical therapy alone. The primary outcome was a composite of death from any cause and nonfatal MI. Patients were followed for at least 3 years, and some for as long as 7 years.

There was an initial small upward spike in the primary outcome in the PCI arm due to periprocedural events. By 5 years, the outcomes of the 2 arms converged and then stayed the same for up to 15 years.⁵ The authors concluded that PCI conferred no benefit over optimal medical therapy in the risk of death or MI.

Some doctors dismiss the study because of its stringent entry criteria—of 35,539 patients assessed, only 3,071 met the eligibility criteria. However, the entry criteria were meant to identify patients most likely to benefit from PCI. Many patients who undergo PCI today would not have qualified for the study because they lack objective evidence of ischemia.⁶ To enroll, patients needed a proximal stenosis of at least 70% and objective evidence of ischemia or a coronary stenosis of more than 80% and classic angina. Exclusion criteria disqualified few patients: Canadian Cardiovascular Society class IV angina (ie, angina evoked from minimal activity or at rest); a markedly positive stress test (substantial ST-segment depression or hypotension during stage I of the Bruce protocol); refractory heart failure or cardiogenic shock; an ejection fraction of less than 30%; revascularization within the past 6 months; and coronary anatomy unsuitable for PCI.

Although COURAGE was hailed as a landmark trial, it largely supported the results of previous studies. A meta-analysis of PCI vs optimal medical therapy published in 2005 found no significant differences in death, cardiac death, MI, or nonfatal MI.⁷ MI was actually slightly more common in the PCI group due to the increased risk of MI during the periprocedural period.

Nor has the evidence from COURAGE discouraged additional studies of the same topic. Despite consistent findings that fit with our understanding of coronary disease as inflammation, we continue to conduct studies aimed at addressing significant stenosis, as if that was the problem. Thus, there have been studies of angioplasty alone, followed by studies of bare-metal stents and then drug-eluting stents.

In 2009, Trikalinos et al published a review of 61 randomized controlled trials comprising more than 25,000 patients with stable coronary disease and comparing medical therapy and angioplasty in its various forms over the previous 20 years.⁸ In all direct and indirect comparisons of PCI and medical therapy, there were no improvements in rates of death or MI.

Even so, the studies continue. The most recent “improvement” was the addition of fractional flow reserve, which served as the inclusion criterion for the Fractional Flow Reserve versus Angiography for Multivessel Evaluation 2 (FAME 2) trial.⁹ In that study, patients with at least 1 stenosis with a fractional flow reserve less than 0.80 were randomized to PCI plus medical therapy or to medical therapy alone. The primary end point was a composite of death from any cause, MI, and urgent revascularization. Unfortunately, the study was stopped early when the primary end point was met due to a reduction in the need for urgent revascularization. There was no reduction in the rate of MI (hazard ratio 1.05, 95% confidence interval 0.51–2.19).

The reduction in urgent revascularization has also been shown consistently in past studies, but this is the weakest outcome measure because it does not equate to a reduction in the rate of MI. There is no demonstrable harm to putting off stent placement, even in functionally significant arteries, and most patients do not require a stent, even in the future.

In summary, the primary benefit of getting a stent now is a reduced likelihood of needing one later.

PCI MAY IMPROVE ANGINA FASTER

Another important finding of the COURAGE trial was that PCI improved symptoms more than optimal medical therapy.¹⁰ This is not surprising, because angina is often a direct result of a significant stenosis. What was unexpected was that even after PCI, most patients were not symptom-free. At 1 month, significantly more PCI patients were angina-free (42%) than were medical patients (33%). This translates into an absolute risk reduction of 9% or a number needed to treat of 11 to prevent 1 case of angina.

Patients in both groups improved over time, and after 3 years, the difference between the 2 groups was no longer significant: 59% in the PCI group vs 56% in the medical therapy group were angina-free.

A more recent study has raised the possibility that the improvement in angina with PCI is primarily a placebo effect. Researchers in the United Kingdom randomized patients with stable angina and at least a 70% stenosis of one vessel to either PCI or sham PCI, in which they threaded the catheter but did not deploy the stent.¹¹ All patients received aggressive antianginal therapy before the procedure. At 6 weeks, there was improvement in angina in both groups, but no statistically significant difference between them in either exercise time or angina. Approximately half the patients in each group improved by at least 1 grade on the Canadian Cardiovascular Society angina classification, and more than 20% improved 2 grades.

This finding is not without precedent. Ligation of the internal mammary arteries, a popular treatment for angina in the 1950s, often provided dramatic relief of symptoms, until it was proven to be no better than a sham operation.^12,13 More recently, a placebo-controlled trial of percutaneous laser myocardial revascularization also failed to show improvement over a sham treatment, despite promising results from a phase 1 trial.¹⁴ Together, these studies emphasize the subjective nature of angina as an outcome and call into question the routine use of PCI to relieve it.

PCI ENTAILS RISK

PCI entails a small but not inconsequential risk. During the procedure, 2% of patients develop bleeding or blood vessel damage, and another 1% die or have an MI or a stroke. In the first year after stent placement, 3% of patients have a bleeding event from the antiplatelet therapy needed for the stent, and an additional 2% develop a clot in the stent that leads to MI.¹⁵

INFORMED CONSENT IS CRITICAL

As demonstrated above, for patients with stable angina, the only evidence-based benefit of PCI over optimal medical therapy is that symptoms may respond faster. At the same time, there are costs and risks associated with the procedure. Because symptoms are subjective, patients should play a key role in deciding whether PCI is appropriate for them.

The American Medical Association states that a physician providing any treatment or procedure should disclose and discuss with patients the risks and benefits. Unfortunately, a substantial body of evidence demonstrates that this is not occurring in practice.

Patients and cardiologists have conflicting beliefs about PCI

Studies over the past 20 years demonstrate that patients with chronic stable angina consistently overestimate the benefits of PCI, with 71% to 88% believing that it will reduce their chance of death.^16–19 Patients also understand that PCI can relieve their symptoms, though no study seems to have assessed the perceived magnitude of this benefit.

In contrast, when cardiologists were asked about the benefits their patients could expect from PCI, only 20% said that it would reduce mortality and 25% said it would prevent MI.¹⁸ These are still surprisingly high percentages, since the study was conducted after the COURAGE trial.

Nevertheless, these differences in perception show that cardiologists fail to successfully communicate the benefits of the procedure to their patients. Without complete information, patients cannot make informed decisions.

If PCI cannot improve hard outcomes like MI or death in stable coronary disease, why do cardiologists continue to perform it so frequently?

Soon after the COURAGE trial, Lin et al conducted focus groups with cardiologists to find out.²⁰ Some said that they doubted the clinical trial evidence, given the reduction in the cardiac mortality rate over the past 30 years. Others remarked that their overriding goal is to stamp out ischemia, and that once a lesion is found by catheterization, one must proceed with PCI. This has been termed the “oculostenotic reflex,” ie, the interventionist sees coronary artery disease and immediately places a stent.

Atreya et al found objective evidence of this practice.²¹ In a 2016 study of 207 patients with obstructive lesions amenable to PCI, the only factors associated with medical management were those that increased the risk of the procedure: age, chronic kidney disease, distal location of the lesion, and type C lesions (the most difficult ones to treat by PCI). More important, evidence of ischemia, presence of angina, and being on optimal medical therapy or maximal antianginal therapy were not associated with PCI.

When surveyed, cardiologists offered reasons similar to those identified by Lin et al, including a positive stress test (70%) and significant myocardium at risk (50%).¹⁸ Optimal medical therapy failure was cited less often (40%). Over 30% identified relief of chest pain for patients who were not prescribed optimal medical therapy. Another 30% said that patient anxiety contributed to their decision, but patients who reported anxiety were not more likely to get PCI than those who did not.

True informed consent rarely occurs

Surveys of patients and recordings of doctor visits suggest that doctors often discuss the risks of the procedure but rarely accurately describe the benefits or mention alternative treatments, including optimal medical therapy.

Fowler et al²² surveyed 472 Medicare patients who had undergone PCI in the past year about their consent discussion, particularly regarding alternative options. Only 6% of patients recalled discussing medication as a serious option with their doctor.

In 2 published studies,^23,24 we analyzed recorded conversations between doctors and patients in which angiography and PCI were discussed.

In a qualitative assessment of how cardiologists presented the rationale for PCI to patients,²³ we observed that cardiologists gave an accurate presentation of the benefits in only 5% of cases. In 13% of the conversations the benefits were explicitly overstated (eg, “If you don’t do it [angiogram/PCI], what could happen? Well, you could…have a heart attack involving that area which can lead to a sudden death”). In another 35% of cases, physicians offered an implicit overstatement of the benefit by saying they could “fix” the problem (eg, “So that’s where we start thinking, Well maybe we better try to fix that [blockage]”), without specifically stating that fixing the problem would offer any benefit. Patients were left to fill in the blanks. Conversations frequently focused on the rationale for performing PCI (eg, ischemia on a stress test) and a description of the procedure, rather than on the risks and benefits.

In a quantitative study of the same data set, we assessed how often physicians addressed the 7 elements of informed decision-making as defined by Braddock et al.²⁴

• Explaining the patient’s role in decision-making (ie, that the patient has a choice to make) was present in half of the conversations. Sometimes a doctor would simply say, “The next step is to perform catheterization.”
• Discussion of clinical issues (eg, having a blockage, stress test results) was performed in almost every case, demonstrating physicians’ comfort with that element.
• Discussing treatment alternatives occurred in only 1 in 4 conversations. This was more frequent than previously reported, and appeared most often when patients expressed hesitancy about proceeding to PCI.
• Discussing pros and cons of the alternatives was done in 42%.
• Uncertainty about the procedure (eg, that it might not relieve the angina) was expressed in only 10% of conversations.
• Assessment of patient understanding was done in 65% of cases. This included even minimal efforts (eg, “Do you have any questions?”). More advanced methods such as teach-back were never used.
• Exploration of patient preferences (eg, asking patients which treatment they prefer, or attempting to understand how angina affects a patient’s life) the final element, occurred in 73% of conversations.

Only 3% of the conversations contained all 7 elements. Even using a more relaxed definition of 3 critical elements (ie, discussing clinical issues, treatment alternatives, and pros and cons), only 13% of conversations included them all.

Discussion affects decisions

Informed decision-making is not only important because of its ethical implications. Offering patients more information was associated with their choosing not to have PCI. The probability of a patient undergoing PCI was negatively associated with 3 specific elements of informed decision-making. Patients were less likely to choose PCI if the patient’s role in decision-making was discussed (61% vs 86% chose PCI, P < .03); if alternatives were discussed (31% vs 89% chose PCI, P < .01); and if uncertainties were discussed (17% vs 80% chose PCI, P < .01).

There was also a linear relationship between the total number of elements discussed and the probability of choosing PCI: it ranged from 100% of patients choosing PCI when just 1 element was present to 3% of patients choosing PCI when all 7 elements were present. The relationship is not entirely causal, since doctors were more likely to talk about alternatives and risks if patients hesitated and raised questions. Cautious patients received more information.

From these observational studies, we know that physicians do not generally communicate the benefits of PCI, and patients make incorrect assumptions about the benefits they can expect. We know that those who receive more information are less likely to choose PCI, but what would happen if patients were randomly assigned to receive complete information?

We conducted an online survey of more than 1,000 participants over age 50 who had never undergone PCI, asking them to imagine visiting a cardiologist after having a positive stress test for stable chest pain.²⁵ Three intervention groups read different scenarios couched as information provided by their cardiologist:

• The “standard care” group received no specific information about the effects of PCI on the risk of myocardial infarction
• The “specific information” group was specifically told that PCI does not reduce the risk of myocardial infarction
• The “explanatory information” group was told how medications work and why PCI does not reduce the risk of myocardial infarction.

All 3 groups received information about the risks of PCI, its role in reducing angina, and the risks and benefits of optimal medical therapy.

After reading their scenario, all participants completed an identical questionnaire, which asked if they would opt for PCI, medical therapy, or both. Overall, 55% chose PCI, ranging from 70% in the standard care group to 46% in the group given explanatory information. Rates in the specific-information and explanatory-information groups were not statistically different from each other, but both were significantly different from that in the standard-care group. Interestingly, the more information patients were given about PCI, the more likely they were to choose optimal medical therapy.

After reading the scenario, participants were also asked if PCI would “prevent a heart attack.” Of those who received standard care, 71% endorsed that belief, which is remarkably similar to studies of real patients who have received standard care. In contrast, only 39% of those given specific information and 31% given explanatory information held that belief. Moreover, the belief that PCI prevented MI was the strongest predictor of choosing PCI (odds ratio 5.82, 95% confidence interval 4.13–8.26).²⁵

Interestingly, 52% of the standard care group falsely remembered that the doctor had told them that PCI would prevent an MI, even though the doctor said nothing about it one way or the other. It appears that participants were projecting their own beliefs onto the encounter. This highlights the importance of providing full information to patients who are considering this procedure.

TOWARD SHARED DECISION-MAKING

Shared decision-making is a process in which physicians enter into a partnership with a patient, offer information, elicit the patient’s preferences, and then come to a decision in concert with the patient.

Although many decisions can and should involve elements of shared decision-making, the decision to proceed with PCI for stable angina is particularly well-suited to shared decision-making. This is because the benefit of PCI depends on the value a patient attaches to being free of angina sooner. Since there is no difference in the risk of MI or death, the patient must decide if the risks of the procedure and the inconvenience of taking dual antiplatelet therapy are worth the benefit of improving symptoms faster. Presumably, patients who have more severe symptoms or experienced side effects from antianginal therapy would be more likely to choose PCI.

Despite having substantial experience educating patients, most physicians are unfamiliar with the process of shared decision-making. In particular, the process of eliciting preferences is often overlooked.

To address this issue, researchers at the Mayo Clinic developed a decision aid that compares PCI plus optimal medical therapy vs optimal medical therapy alone in an easily understandable information card.¹⁵ On one side, the 2 options are clearly stated, with the magnitude of symptom improvement over time graphically illustrated and the statement, “NO DIFFERENCE in heart attack or death,” prominently displayed. The back of the card discusses the risks of each option in easily understood tables.

The decision aid was compared with standard care in a randomized trial involving patients who were referred for catheterization and possible PCI.²⁶ The decision aid improved patients’ overall knowledge about PCI. In particular, 60% of those who used the decision aid knew that PCI did not prevent death or MI vs 40% of usual-care patients—results similar to those of the online experiment.

Interestingly, the decision about whether to undergo PCI did not differ significantly between the 2 groups, although there was a trend toward more patients in the decision-aid group choosing medical therapy alone (53%) vs the standard-care patients (39%).

To understand why the decision aid did not make more of a difference, the investigators performed qualitative interviews of the cardiologists in the study.²⁷ One theme was the timing of the intervention. Patients using the decision aid had already been referred for catheterization, and some felt the process should have occurred earlier. Engaging in shared decision-making with a general cardiologist before referral could help to improve the quality of patient decisions.

Cardiologists also noted the difficulty in changing their work flow to incorporate the decision aid. Although some embraced the idea of shared decision-making, others were concerned that many patients could not participate, and there was confusion about the difference between an educational tool, which could be used by a patient alone, and a decision aid, which is meant to generate discussion between the doctor and patient. Some expressed interest in using the tool in the future.

These findings serve to emphasize that providing information alone is not enough. If the physician does not “buy in” to the idea of shared decision-making, it will not occur.

PRACTICE IMPLICATIONS

Based on the pathophysiology of coronary artery disease and the results of multiple randomized controlled trials, it is evident that PCI does not prevent heart attacks in patients with chronic stable angina. However, most patients who undergo PCI are unaware of this and therefore do not truly give informed consent. In the absence of explicit information to the contrary, most patients with stable angina assume that PCI prevents MI and thus are biased toward choosing PCI.

Even minimal amounts of explicit information can partially overcome that bias and influence decision-making. In particular, explaining why PCI does not prevent MI was the most effective means of overcoming the bias.

To this end, shared decision aids may help physicians to engage in shared decision-making. Shared decision-making is most likely to occur if physicians are trained in the concept of shared decision-making, are committed to practicing it, and can fit it into their work flow. Ideally, this would occur in the office of a general cardiologist before referral for PCI.

For those practicing in accountable-care organizations, Medicare has recently introduced the shared decision-making model for 6 preference-sensitive conditions, including stable ischemic heart disease. Participants in this program will have the opportunity to receive payments for shared decision-making services and to share in any savings that result from reduced use of resources. Use of these tools holds the promise for providing more patient-centered care at lower cost.

Multiple randomized controlled trials have compared percutaneous coronary intervention (PCI) vs optimal medical therapy for patients with chronic stable angina. All have consistently shown that PCI does not reduce the risk of death or even myocardial infarction (MI) but that it may relieve angina temporarily. Nevertheless, PCI is still commonly performed for patients with stable coronary disease, often in the absence of angina, and patients mistakenly believe the procedure is life-saving. Cardiologists may not be aware of patients’ misperceptions, or worse, may encourage them. In either case, if patients do not understand the benefits of the procedure, they cannot give informed consent.

See related editorial

This article reviews the pathophysiology of coronary artery disease, evidence from clinical trials of the value of PCI for chronic stable angina, patient and physician perceptions of PCI, and ways to promote patient-centered, shared decision-making.

CLINICAL CASE: EXERTIONAL ANGINA

While climbing 4 flights of stairs, a 55-year-old man noticed tightness in his chest, which lasted for 5 minutes and resolved spontaneously. Several weeks later, when visiting his primary care physician, he mentioned the episode. He had had no symptoms in the interim, but the physician ordered an exercise stress test.

Six minutes into a standard Bruce protocol, the patient experienced the same chest tightness, accompanied by 1-mm ST-segment depressions in leads II, III, and aVF. He was then referred to a cardiologist, who recommended catheterization.

Catheterization demonstrated a 95% stenosis of the right coronary artery with nonsignificant stenoses of the left anterior descending and circumflex arteries. A drug-eluting stent was placed in the right coronary artery, with no residual stenosis.

Did this intervention likely prevent an MI and perhaps save the man’s life?

HOW MYOCARDIAL INFARCTION HAPPENS

Understanding the pathogenesis of MI is critical to having realistic expectations of the benefits of stent placement.

Doctors often describe coronary atherosclerosis as a plumbing problem, where deposits of cholesterol and fat build up in arterial walls, clogging the pipes and eventually causing a heart attack. This analogy, which has been around since the 1950s, is easy to for patients to grasp and has been popularized in the press and internalized by the public—as one patient with a 95% stenosis put it, “I was 95% dead.” In that model, angioplasty and stenting can resolve the blockage and “fix” the problem, much as a plumber can clear your pipes with a Roto-Rooter.

Despite the visual appeal of this model,¹ it doesn’t accurately convey what we know about the pathophysiology of coronary artery disease. Instead of a gradual buildup of fatty deposits, low-density lipoprotein cholesterol particles infiltrate arterial walls and trigger an inflammatory reaction as they are engulfed by macrophages, leading to a cascade of cytokines and recruitment of more inflammatory cells.² This immune response can eventually cause the rupture of the plaque’s fibrous cap, triggering thrombosis and infarction, often at a site of insignificant stenosis.

In this new model, coronary artery disease is primarily a problem of inflammation distributed throughout the vasculature, rather than a mechanical problem localized to the site of a significant stenosis.

Significant stenosis does not equal unstable plaque

Not all plaques are equally likely to rupture. Stable plaques tend to be long-standing and calcified, with a thick fibrous cap. A stable plaque causing a 95% stenosis may cause symptoms with exertion, but it is unlikely to cause infarction.³ Conversely, rupture-prone plaques may cause little stenosis, but a large and dangerous plaque may be lurking beneath the thin fibrous cap.

Relying on angiography can be misleading. Treating all significant stenoses improves blood flow, but does not reduce the risk of infarction, because infarction most often occurs in areas where the lumen is not obstructed. A plaque causing only 30% stenosis can suddenly rupture, causing thrombosis and complete occlusion.

The current model explains why PCI is no better than optimal medical therapy (ie, risk factor modification, antiplatelet therapy with aspirin, and a statin). Diet, exercise, smoking cessation, and statins target inflammatory processes and lower low-density lipoprotein cholesterol levels, while aspirin prevents platelet aggregation, among other likely actions.

The model also explains why coronary artery bypass grafting reduces the risk of MI and death in patients with left main or 3-vessel disease. A patient with generalized coronary artery disease has multiple lesions, many of which do not cause significant stenoses. PCI corrects only a single stenosis, whereas coronary artery bypass grafting circumvents all the vulnerable plaques in a vessel.

THE LANDMARK COURAGE TRIAL

Published in 2007, the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial⁴ randomized more than 2,000 patients to receive either optimal medical therapy plus PCI or optimal medical therapy alone. The primary outcome was a composite of death from any cause and nonfatal MI. Patients were followed for at least 3 years, and some for as long as 7 years.

There was an initial small upward spike in the primary outcome in the PCI arm due to periprocedural events. By 5 years, the outcomes of the 2 arms converged and then stayed the same for up to 15 years.⁵ The authors concluded that PCI conferred no benefit over optimal medical therapy in the risk of death or MI.

Some doctors dismiss the study because of its stringent entry criteria—of 35,539 patients assessed, only 3,071 met the eligibility criteria. However, the entry criteria were meant to identify patients most likely to benefit from PCI. Many patients who undergo PCI today would not have qualified for the study because they lack objective evidence of ischemia.⁶ To enroll, patients needed a proximal stenosis of at least 70% and objective evidence of ischemia or a coronary stenosis of more than 80% and classic angina. Exclusion criteria disqualified few patients: Canadian Cardiovascular Society class IV angina (ie, angina evoked from minimal activity or at rest); a markedly positive stress test (substantial ST-segment depression or hypotension during stage I of the Bruce protocol); refractory heart failure or cardiogenic shock; an ejection fraction of less than 30%; revascularization within the past 6 months; and coronary anatomy unsuitable for PCI.

Although COURAGE was hailed as a landmark trial, it largely supported the results of previous studies. A meta-analysis of PCI vs optimal medical therapy published in 2005 found no significant differences in death, cardiac death, MI, or nonfatal MI.⁷ MI was actually slightly more common in the PCI group due to the increased risk of MI during the periprocedural period.

Nor has the evidence from COURAGE discouraged additional studies of the same topic. Despite consistent findings that fit with our understanding of coronary disease as inflammation, we continue to conduct studies aimed at addressing significant stenosis, as if that was the problem. Thus, there have been studies of angioplasty alone, followed by studies of bare-metal stents and then drug-eluting stents.

In 2009, Trikalinos et al published a review of 61 randomized controlled trials comprising more than 25,000 patients with stable coronary disease and comparing medical therapy and angioplasty in its various forms over the previous 20 years.⁸ In all direct and indirect comparisons of PCI and medical therapy, there were no improvements in rates of death or MI.

Even so, the studies continue. The most recent “improvement” was the addition of fractional flow reserve, which served as the inclusion criterion for the Fractional Flow Reserve versus Angiography for Multivessel Evaluation 2 (FAME 2) trial.⁹ In that study, patients with at least 1 stenosis with a fractional flow reserve less than 0.80 were randomized to PCI plus medical therapy or to medical therapy alone. The primary end point was a composite of death from any cause, MI, and urgent revascularization. Unfortunately, the study was stopped early when the primary end point was met due to a reduction in the need for urgent revascularization. There was no reduction in the rate of MI (hazard ratio 1.05, 95% confidence interval 0.51–2.19).

The reduction in urgent revascularization has also been shown consistently in past studies, but this is the weakest outcome measure because it does not equate to a reduction in the rate of MI. There is no demonstrable harm to putting off stent placement, even in functionally significant arteries, and most patients do not require a stent, even in the future.

In summary, the primary benefit of getting a stent now is a reduced likelihood of needing one later.

PCI MAY IMPROVE ANGINA FASTER

Another important finding of the COURAGE trial was that PCI improved symptoms more than optimal medical therapy.¹⁰ This is not surprising, because angina is often a direct result of a significant stenosis. What was unexpected was that even after PCI, most patients were not symptom-free. At 1 month, significantly more PCI patients were angina-free (42%) than were medical patients (33%). This translates into an absolute risk reduction of 9% or a number needed to treat of 11 to prevent 1 case of angina.

Patients in both groups improved over time, and after 3 years, the difference between the 2 groups was no longer significant: 59% in the PCI group vs 56% in the medical therapy group were angina-free.

A more recent study has raised the possibility that the improvement in angina with PCI is primarily a placebo effect. Researchers in the United Kingdom randomized patients with stable angina and at least a 70% stenosis of one vessel to either PCI or sham PCI, in which they threaded the catheter but did not deploy the stent.¹¹ All patients received aggressive antianginal therapy before the procedure. At 6 weeks, there was improvement in angina in both groups, but no statistically significant difference between them in either exercise time or angina. Approximately half the patients in each group improved by at least 1 grade on the Canadian Cardiovascular Society angina classification, and more than 20% improved 2 grades.

This finding is not without precedent. Ligation of the internal mammary arteries, a popular treatment for angina in the 1950s, often provided dramatic relief of symptoms, until it was proven to be no better than a sham operation.^12,13 More recently, a placebo-controlled trial of percutaneous laser myocardial revascularization also failed to show improvement over a sham treatment, despite promising results from a phase 1 trial.¹⁴ Together, these studies emphasize the subjective nature of angina as an outcome and call into question the routine use of PCI to relieve it.

PCI ENTAILS RISK

PCI entails a small but not inconsequential risk. During the procedure, 2% of patients develop bleeding or blood vessel damage, and another 1% die or have an MI or a stroke. In the first year after stent placement, 3% of patients have a bleeding event from the antiplatelet therapy needed for the stent, and an additional 2% develop a clot in the stent that leads to MI.¹⁵

INFORMED CONSENT IS CRITICAL

As demonstrated above, for patients with stable angina, the only evidence-based benefit of PCI over optimal medical therapy is that symptoms may respond faster. At the same time, there are costs and risks associated with the procedure. Because symptoms are subjective, patients should play a key role in deciding whether PCI is appropriate for them.

The American Medical Association states that a physician providing any treatment or procedure should disclose and discuss with patients the risks and benefits. Unfortunately, a substantial body of evidence demonstrates that this is not occurring in practice.

Patients and cardiologists have conflicting beliefs about PCI

Studies over the past 20 years demonstrate that patients with chronic stable angina consistently overestimate the benefits of PCI, with 71% to 88% believing that it will reduce their chance of death.^16–19 Patients also understand that PCI can relieve their symptoms, though no study seems to have assessed the perceived magnitude of this benefit.

In contrast, when cardiologists were asked about the benefits their patients could expect from PCI, only 20% said that it would reduce mortality and 25% said it would prevent MI.¹⁸ These are still surprisingly high percentages, since the study was conducted after the COURAGE trial.

Nevertheless, these differences in perception show that cardiologists fail to successfully communicate the benefits of the procedure to their patients. Without complete information, patients cannot make informed decisions.

If PCI cannot improve hard outcomes like MI or death in stable coronary disease, why do cardiologists continue to perform it so frequently?

Soon after the COURAGE trial, Lin et al conducted focus groups with cardiologists to find out.²⁰ Some said that they doubted the clinical trial evidence, given the reduction in the cardiac mortality rate over the past 30 years. Others remarked that their overriding goal is to stamp out ischemia, and that once a lesion is found by catheterization, one must proceed with PCI. This has been termed the “oculostenotic reflex,” ie, the interventionist sees coronary artery disease and immediately places a stent.

Atreya et al found objective evidence of this practice.²¹ In a 2016 study of 207 patients with obstructive lesions amenable to PCI, the only factors associated with medical management were those that increased the risk of the procedure: age, chronic kidney disease, distal location of the lesion, and type C lesions (the most difficult ones to treat by PCI). More important, evidence of ischemia, presence of angina, and being on optimal medical therapy or maximal antianginal therapy were not associated with PCI.

When surveyed, cardiologists offered reasons similar to those identified by Lin et al, including a positive stress test (70%) and significant myocardium at risk (50%).¹⁸ Optimal medical therapy failure was cited less often (40%). Over 30% identified relief of chest pain for patients who were not prescribed optimal medical therapy. Another 30% said that patient anxiety contributed to their decision, but patients who reported anxiety were not more likely to get PCI than those who did not.

True informed consent rarely occurs

Surveys of patients and recordings of doctor visits suggest that doctors often discuss the risks of the procedure but rarely accurately describe the benefits or mention alternative treatments, including optimal medical therapy.

Fowler et al²² surveyed 472 Medicare patients who had undergone PCI in the past year about their consent discussion, particularly regarding alternative options. Only 6% of patients recalled discussing medication as a serious option with their doctor.

In 2 published studies,^23,24 we analyzed recorded conversations between doctors and patients in which angiography and PCI were discussed.

In a qualitative assessment of how cardiologists presented the rationale for PCI to patients,²³ we observed that cardiologists gave an accurate presentation of the benefits in only 5% of cases. In 13% of the conversations the benefits were explicitly overstated (eg, “If you don’t do it [angiogram/PCI], what could happen? Well, you could…have a heart attack involving that area which can lead to a sudden death”). In another 35% of cases, physicians offered an implicit overstatement of the benefit by saying they could “fix” the problem (eg, “So that’s where we start thinking, Well maybe we better try to fix that [blockage]”), without specifically stating that fixing the problem would offer any benefit. Patients were left to fill in the blanks. Conversations frequently focused on the rationale for performing PCI (eg, ischemia on a stress test) and a description of the procedure, rather than on the risks and benefits.

In a quantitative study of the same data set, we assessed how often physicians addressed the 7 elements of informed decision-making as defined by Braddock et al.²⁴

• Explaining the patient’s role in decision-making (ie, that the patient has a choice to make) was present in half of the conversations. Sometimes a doctor would simply say, “The next step is to perform catheterization.”
• Discussion of clinical issues (eg, having a blockage, stress test results) was performed in almost every case, demonstrating physicians’ comfort with that element.
• Discussing treatment alternatives occurred in only 1 in 4 conversations. This was more frequent than previously reported, and appeared most often when patients expressed hesitancy about proceeding to PCI.
• Discussing pros and cons of the alternatives was done in 42%.
• Uncertainty about the procedure (eg, that it might not relieve the angina) was expressed in only 10% of conversations.
• Assessment of patient understanding was done in 65% of cases. This included even minimal efforts (eg, “Do you have any questions?”). More advanced methods such as teach-back were never used.
• Exploration of patient preferences (eg, asking patients which treatment they prefer, or attempting to understand how angina affects a patient’s life) the final element, occurred in 73% of conversations.

Only 3% of the conversations contained all 7 elements. Even using a more relaxed definition of 3 critical elements (ie, discussing clinical issues, treatment alternatives, and pros and cons), only 13% of conversations included them all.

Discussion affects decisions

Informed decision-making is not only important because of its ethical implications. Offering patients more information was associated with their choosing not to have PCI. The probability of a patient undergoing PCI was negatively associated with 3 specific elements of informed decision-making. Patients were less likely to choose PCI if the patient’s role in decision-making was discussed (61% vs 86% chose PCI, P < .03); if alternatives were discussed (31% vs 89% chose PCI, P < .01); and if uncertainties were discussed (17% vs 80% chose PCI, P < .01).

There was also a linear relationship between the total number of elements discussed and the probability of choosing PCI: it ranged from 100% of patients choosing PCI when just 1 element was present to 3% of patients choosing PCI when all 7 elements were present. The relationship is not entirely causal, since doctors were more likely to talk about alternatives and risks if patients hesitated and raised questions. Cautious patients received more information.

From these observational studies, we know that physicians do not generally communicate the benefits of PCI, and patients make incorrect assumptions about the benefits they can expect. We know that those who receive more information are less likely to choose PCI, but what would happen if patients were randomly assigned to receive complete information?

We conducted an online survey of more than 1,000 participants over age 50 who had never undergone PCI, asking them to imagine visiting a cardiologist after having a positive stress test for stable chest pain.²⁵ Three intervention groups read different scenarios couched as information provided by their cardiologist:

• The “standard care” group received no specific information about the effects of PCI on the risk of myocardial infarction
• The “specific information” group was specifically told that PCI does not reduce the risk of myocardial infarction
• The “explanatory information” group was told how medications work and why PCI does not reduce the risk of myocardial infarction.

All 3 groups received information about the risks of PCI, its role in reducing angina, and the risks and benefits of optimal medical therapy.

After reading their scenario, all participants completed an identical questionnaire, which asked if they would opt for PCI, medical therapy, or both. Overall, 55% chose PCI, ranging from 70% in the standard care group to 46% in the group given explanatory information. Rates in the specific-information and explanatory-information groups were not statistically different from each other, but both were significantly different from that in the standard-care group. Interestingly, the more information patients were given about PCI, the more likely they were to choose optimal medical therapy.

After reading the scenario, participants were also asked if PCI would “prevent a heart attack.” Of those who received standard care, 71% endorsed that belief, which is remarkably similar to studies of real patients who have received standard care. In contrast, only 39% of those given specific information and 31% given explanatory information held that belief. Moreover, the belief that PCI prevented MI was the strongest predictor of choosing PCI (odds ratio 5.82, 95% confidence interval 4.13–8.26).²⁵

Interestingly, 52% of the standard care group falsely remembered that the doctor had told them that PCI would prevent an MI, even though the doctor said nothing about it one way or the other. It appears that participants were projecting their own beliefs onto the encounter. This highlights the importance of providing full information to patients who are considering this procedure.

TOWARD SHARED DECISION-MAKING

Shared decision-making is a process in which physicians enter into a partnership with a patient, offer information, elicit the patient’s preferences, and then come to a decision in concert with the patient.

Although many decisions can and should involve elements of shared decision-making, the decision to proceed with PCI for stable angina is particularly well-suited to shared decision-making. This is because the benefit of PCI depends on the value a patient attaches to being free of angina sooner. Since there is no difference in the risk of MI or death, the patient must decide if the risks of the procedure and the inconvenience of taking dual antiplatelet therapy are worth the benefit of improving symptoms faster. Presumably, patients who have more severe symptoms or experienced side effects from antianginal therapy would be more likely to choose PCI.

Despite having substantial experience educating patients, most physicians are unfamiliar with the process of shared decision-making. In particular, the process of eliciting preferences is often overlooked.

To address this issue, researchers at the Mayo Clinic developed a decision aid that compares PCI plus optimal medical therapy vs optimal medical therapy alone in an easily understandable information card.¹⁵ On one side, the 2 options are clearly stated, with the magnitude of symptom improvement over time graphically illustrated and the statement, “NO DIFFERENCE in heart attack or death,” prominently displayed. The back of the card discusses the risks of each option in easily understood tables.

The decision aid was compared with standard care in a randomized trial involving patients who were referred for catheterization and possible PCI.²⁶ The decision aid improved patients’ overall knowledge about PCI. In particular, 60% of those who used the decision aid knew that PCI did not prevent death or MI vs 40% of usual-care patients—results similar to those of the online experiment.

Interestingly, the decision about whether to undergo PCI did not differ significantly between the 2 groups, although there was a trend toward more patients in the decision-aid group choosing medical therapy alone (53%) vs the standard-care patients (39%).

To understand why the decision aid did not make more of a difference, the investigators performed qualitative interviews of the cardiologists in the study.²⁷ One theme was the timing of the intervention. Patients using the decision aid had already been referred for catheterization, and some felt the process should have occurred earlier. Engaging in shared decision-making with a general cardiologist before referral could help to improve the quality of patient decisions.

Cardiologists also noted the difficulty in changing their work flow to incorporate the decision aid. Although some embraced the idea of shared decision-making, others were concerned that many patients could not participate, and there was confusion about the difference between an educational tool, which could be used by a patient alone, and a decision aid, which is meant to generate discussion between the doctor and patient. Some expressed interest in using the tool in the future.

These findings serve to emphasize that providing information alone is not enough. If the physician does not “buy in” to the idea of shared decision-making, it will not occur.

PRACTICE IMPLICATIONS

Based on the pathophysiology of coronary artery disease and the results of multiple randomized controlled trials, it is evident that PCI does not prevent heart attacks in patients with chronic stable angina. However, most patients who undergo PCI are unaware of this and therefore do not truly give informed consent. In the absence of explicit information to the contrary, most patients with stable angina assume that PCI prevents MI and thus are biased toward choosing PCI.

Even minimal amounts of explicit information can partially overcome that bias and influence decision-making. In particular, explaining why PCI does not prevent MI was the most effective means of overcoming the bias.

To this end, shared decision aids may help physicians to engage in shared decision-making. Shared decision-making is most likely to occur if physicians are trained in the concept of shared decision-making, are committed to practicing it, and can fit it into their work flow. Ideally, this would occur in the office of a general cardiologist before referral for PCI.

For those practicing in accountable-care organizations, Medicare has recently introduced the shared decision-making model for 6 preference-sensitive conditions, including stable ischemic heart disease. Participants in this program will have the opportunity to receive payments for shared decision-making services and to share in any savings that result from reduced use of resources. Use of these tools holds the promise for providing more patient-centered care at lower cost.

This article reviews recommendations and evidence concerning current anticoagulant management for venous thromboembolism and perioperative care, with an emphasis on individualizing treatment for real-world patients.

TREATING ACUTE VENOUS THROMBOEMBOLISM

Case 1: Deep vein thrombosis in an otherwise healthy man

A 40-year-old man presents with 7 days of progressive right leg swelling. He has no antecedent risk factors for deep vein thrombosis or other medical problems. Venous ultrasonography reveals an iliofemoral deep vein thrombosis. How should he be managed?

• Outpatient treatment with low-molecular-weight heparin for 4 to 6 days plus warfarin
• Outpatient treatment with a direct oral anticoagulant, ie, apixaban, dabigatran (which requires 4 to 6 days of initial treatment with low-molecular-weight heparin), or rivaroxaban
• Catheter-directed thrombolysis followed by low-molecular-weight heparin, then warfarin or a direct oral anticoagulant
• Inpatient intravenous heparin for 7 to 10 days, then warfarin or a direct oral anticoagulant

All of these are acceptable for managing acute venous thromboembolism, but the clinician’s role is to identify which treatment is most appropriate for an individual patient.

Deep vein thrombosis is not a single condition

Multiple guidelines exist to help decide on a management strategy. Those of the American College of Chest Physicians (ACCP)¹ are used most often.

That said, guidelines are established for “average” patients, so it is important to look beyond guidelines and individualize management. Venous thromboembolism is not a single entity; it has a myriad of clinical presentations that could call for different treatments. Most patients have submassive deep vein thrombosis or pulmonary embolism, which is not limb-threatening nor associated with hemodynamic instability. It can also differ in terms of etiology and can be unprovoked (or idiopathic), cancer-related, catheter-associated, or provoked by surgery or immobility.

Deep vein thrombosis has a wide spectrum of presentations. It can involve the veins of the calf only, or it can involve the femoral and iliac veins and other locations including the splanchnic veins, the cerebral sinuses, and upper extremities. Pulmonary embolism can be massive (defined as being associated with hemodynamic instability or impending respiratory failure) or submassive. Similarly, patients differ in terms of baseline medical conditions, mobility, and lifestyle. Anticoagulant management decisions should take all these factors into account.

Consider clot location

Our patient with iliofemoral deep vein thrombosis is best managed differently than a more typical patient with less extensive thrombosis that would involve the popliteal or femoral vein segments, or both. A clot that involves the iliac vein is more likely to lead to postthrombotic chronic pain and swelling as the lack of venous outflow bypass channels to circumvent the clot location creates higher venous pressure within the affected leg. Therefore, for our patient, catheter-directed thrombolysis is an option that should be considered.

Catheter-directed thrombolysis trials

According to the “open-vein hypothesis,” quickly eliminating the thrombus and restoring unobstructed venous flow may mitigate the risk not only of recurrent thrombosis, but also of postthrombotic syndrome, which is often not given much consideration acutely but can cause significant, life-altering chronic disability.

The “valve-integrity hypothesis” is also important; it considers whether lytic therapy may help prevent damage to such valves in an attempt to mitigate the amount of venous hypertension.

Thus, catheter-directed thrombolysis offers theoretical benefits, and recent trials have assessed it against standard anticoagulation treatments.

The CaVenT trial (Catheter-Directed Venous Thrombolysis),² conducted in Norway, randomized 209 patients with midfemoral to iliac deep vein thrombosis to conventional treatment (anticoagulation alone) or anticoagulation plus catheter-directed thrombolysis. At 2 years, postthrombotic syndrome had occurred in 41% of the catheter-directed thrombolysis group compared with 56% of the conventional treatment group (P = .047). At 5 years, the difference widened to 43% vs 71% (P < .01, number needed to treat = 4).³ Despite the superiority of lytic therapy, the incidence of postthrombotic syndrome remained high in patients who received this treatment. 

The ATTRACT trial (Acute Venous Thrombosis: Thrombus Removal With Adjunctive Catheter-Directed Thrombolysis),⁴ a US multicenter, open-label, assessor-blind study, randomized 698 patients with femoral or more-proximal deep vein thrombosis to either standard care (anticoagulant therapy and graduated elastic compression stockings) or standard care plus catheter-directed thrombolysis. In preliminary results presented at the Society of Interventional Radiology meeting in March 2017, although no difference was found in the primary outcome (postthrombotic syndrome at 24 months), catheter-directed thrombolysis for iliofemoral deep vein thrombosis led to a 25% reduction in moderate to severe postthrombotic syndrome.

Although it is too early to draw conclusions before publication of the ATTRACT study, the preliminary results highlight the need to individualize treatment and to be selective about using catheter-directed thrombolysis. The trials provide reassurance that catheter-directed lysis is a reasonable and safe intervention when performed by physicians experienced in the procedure. The risk of major bleeding appears to be low (about 2%) and that for intracranial hemorrhage even lower (< 0.5%).

Catheter-directed thrombolysis is appropriate in some cases

The 2016 ACCP guidelines¹ recommend anticoagulant therapy alone over catheter-directed thrombolysis for patients with acute proximal deep vein thrombosis of the leg. However, it is a grade 2C (weak) recommendation.

They provide no specific recommendation as to the clinical indications for catheter-directed thrombolysis, but identify patients who would be most likely to benefit, ie, those who have: 

• Iliofemoral deep vein thrombosis
• Symptoms for less than 14 days
• Good functional status
• Life expectancy of more than 1 year
• Low risk of bleeding.

Our patient satisfies these criteria, suggesting that catheter-directed thrombolysis is a reasonable option for him. 

Timing is important. Catheter-directed lysis is more likely to be beneficial if used before fibrin deposits form and stiffen the venous valves, causing irreversible damage that leads to postthrombotic syndrome. 

The availability of direct oral anticoagulants has generated interest in defining their therapeutic role in patients with venous thromboembolism.

In a meta-analysis⁵ of major trials comparing direct oral anticoagulants and vitamin K antagonists such as warfarin, no significant difference was found for the risk of recurrent venous thromboembolism or venous thromboembolism-related deaths. However, fewer patients experienced major bleeding with direct oral anticoagulants (relative risk 0.61, P = .002). Although significant, the absolute risk reduction was small; the incidence of major bleeding was 1.1% with direct oral anticoagulants vs 1.8% with vitamin K antagonists.

The main advantage of direct oral anticoagulants is greater convenience for the patient.

WHICH PATIENTS ON WARFARIN NEED BRIDGING PREOPERATIVELY?

Many patients still take warfarin, particularly those with atrial fibrillation, a mechanical heart valve, or venous thromboembolism. In many countries, warfarin remains the dominant anticoagulant for stroke prevention. Whether these patients need heparin during the period of perioperative warfarin interruption is a frequently encountered scenario that, until recently, was controversial. Recent studies have helped to inform the need for heparin bridging in many of these patients.

Case 2: An elderly woman on warfarin facing cancer surgery

A 75-year-old woman weighing 65 kg is scheduled for elective colon resection for incidentally found colon cancer. She is taking warfarin for atrial fibrillation. She also has hypertension and diabetes and had a transient ischemic attack 10 years ago.

One doctor told her she needs to be assessed for heparin bridging, but another told her she does not need bridging.

The default management should be not to bridge patients who have atrial fibrillation, but to consider bridging in selected patients, such as those with recent stroke or transient ischemic attack or a prior thromboembolic event during warfarin interruption. However, decisions about bridging should not be made on the basis of the CHADS₂ score alone. For the patient described here, I would recommend not bridging.

Complex factors contribute to stroke risk

Stroke risk for patients with atrial fibrillation can be quickly estimated with the CHADS₂ score, based on: 

• Congestive heart failure (1 point)
• Hypertension (1 point)
• Age at least 75 (1 point)
• Diabetes (1 point)
• Stroke or transient ischemic attack (2 points).

Our patient has a score of 5, corresponding to an annual adjusted stroke risk of 12.5%. Whether her transient ischemic attack of 10 years ago is comparable in significance to a recent stroke is debatable and highlights a weakness of clinical prediction rules. Moreover, such prediction scores were developed to estimate the long-term risk of stroke if anticoagulants are not given, and they have not been assessed in a perioperative setting where there is short-term interruption of anticoagulants. Also, the perioperative milieu is associated with additional factors not captured in these clinical prediction rules that may affect the risk of stroke.

Thus, the risk of perioperative stroke likely involves the interplay of multiple factors, including the type of surgery the patient is undergoing. Some factors may be mitigated:

• Rebound hypercoagulability after stopping an oral anticoagulant can be prevented by intraoperative blood pressure and volume control
• Elevated biochemical factors (eg, D-dimer, B-type natriuretic peptide, troponin) may be lowered with perioperative aspirin therapy
• Lipid and genetic factors may be mitigated with perioperative statin use.

Can heparin bridging also mitigate the risk?

Bridging in patients with atrial fibrillation

Most patients who are taking warfarin are doing so because of atrial fibrillation, so most evidence about perioperative bridging was developed in such patients.

The BRIDGE trial (Bridging Anticoagulation in Patients Who Require Temporary Interruption of Warfarin Therapy for an Elective Invasive Procedure or Surgery)⁶ was the first randomized controlled trial to compare a bridging and no-bridging strategy for patients with atrial fibrillation who required warfarin interruption for elective surgery. Nearly 2,000 patients were given either low-molecular-weight heparin or placebo starting 3 days before until 24 hours before a procedure, and then for 5 to 10 days afterwards. For all patients, warfarin was stopped 5 days before the procedure and was resumed within 24 hours afterwards.

A no-bridging strategy was noninferior to bridging: the risk of perioperative arterial thromboembolism was 0.4% without bridging vs 0.3% with bridging (P = .01 for noninferiority). In addition, a no-bridging strategy conferred a lower risk of major bleeding than bridging: 1.3% vs 3.2% (relative risk 0.41, P = .005 for superiority).

Although the difference in absolute bleeding risk was small, bleeding rates were lower than those seen outside of clinical trials, as the bridging protocol used in BRIDGE was designed to minimize the risk of bleeding. Also, although only 5% of patients had a CHADS₂ score of 5 or 6, such patients are infrequent in clinical practice, and BRIDGE did include a considerable proportion (17%) of patients with a prior stroke or transient ischemic attack who would be considered at high risk.

Other evidence about heparin bridging is derived from observational studies, more than 10 of which have been conducted. In general, they have found that not bridging is associated with low rates of arterial thromboembolism (< 0.5%) and that bridging is associated with high rates of major bleeding (4%–7%).^7–12

Bridging in patients with a mechanical heart valve

Warfarin is the only anticoagulant option for patients who have a mechanical heart valve. No randomized controlled trials have evaluated the benefits of perioperative bridging vs no bridging in this setting.

Observational (cohort) studies suggest that the risk of perioperative arterial thromboembolism is similar with or without bridging anticoagulation, although most patients studied were bridged and those not bridged were considered at low risk (eg, with a bileaflet aortic valve and no additional risk factors).¹³ However, without stronger evidence from randomized controlled trials, bridging should be the default management for patients with a mechanical heart valve. In our practice, we bridge most patients who have a mechanical heart valve unless they are considered to be at low risk, such as those who have a bileaflet aortic valve.

Even less evidence is available for periprocedural management of patients who have a history of venous thromboembolism. No randomized controlled trials exist evaluating bridging vs no bridging. In 1 cohort study in which more than 90% of patients had had thromboembolism more than 3 months before the procedure, the rate of recurrent venous thromboembolism without bridging was less than 0.5%.¹⁴

It is reasonable to bridge patients who need anticoagulant interruption within 3 months of diagnosis of a deep vein thrombosis or pulmonary embolism, and to consider using a temporary inferior vena cava filter for patients who have had a clot who need treatment interruption during the initial 3 to 4 weeks after diagnosis.

Practice guidelines: Perioperative anticoagulation

Guidance for preoperative and postoperative bridging for patients taking warfarin is summarized in Table 2.

CARDIAC PROCEDURES

For patients facing a procedure to implant an implantable cardioverter-defibrillator (ICD) or pacemaker, a procedure-specific concern is the avoidance of pocket hematoma.

Patients on warfarin: Do not bridge

The BRUISE CONTROL-1 trial (Bridge or Continue Coumadin for Device Surgery Randomized Controlled Trial)¹⁹ randomized patients undergoing pacemaker or ICD implantation to either continued anticoagulation therapy and not bridging (ie, continued warfarin so long as the international normalized ratio was < 3) vs conventional bridging treatment (ie, stopping warfarin and bridging with low-molecular-weight heparin). A clinically significant device-pocket hematoma occurred in 3.5% of the continued-warfarin group vs 16.0% in the heparin-bridging group (P < .001). Thromboembolic complications were rare, and rates did not differ between the 2 groups.

Results of the BRUISE CONTROL-1 trial serve as a caution to at least not be too aggressive with bridging. The study design involved resuming heparin 24 hours after surgery, which is perhaps more aggressive than standard practice. In our practice, we wait at least 24 hours to reinstate heparin after minor surgery, and 48 to 72 hours after surgery with higher bleeding risk.

These results are perhaps not surprising if one considers how carefully surgeons try to control bleeding during surgery for patients taking anticoagulants. For patients who are not on an anticoagulant, small bleeding may be less of a concern during a procedure. When high doses of heparin are introduced soon after surgery, small concerns during surgery may become big problems afterward.

Based on these results, it is reasonable to undertake device implantation without interruption of a vitamin K antagonist such as warfarin.

Patients on direct oral anticoagulants: The jury is still out

The similar BRUISE CONTROL-2 trial is currently under way, comparing interruption vs continuation of dabigatran for patients undergoing cardiac device surgery.

In Europe, surgeons are less concerned than those in the United States about operating while a patient is on anticoagulant therapy. But the safety of this practice is not backed by strong evidence.

Direct oral anticoagulants: Consider pharmacokinetics

Direct oral anticoagulants are potent and fast-acting, with a peak effect 1 to 3 hours after intake. This rapid anticoagulant action is similar to that of bridging with low-molecular-weight heparin, and caution is needed when administering direct oral anticoagulants, especially after major surgery or surgery with a high bleeding risk.

Frost et al²⁰ compared the pharmacokinetics of apixaban (with twice-daily dosing) and rivaroxaban (once-daily dosing) and found that peak anticoagulant activity is faster and higher with rivaroxaban. This is important, because many patients will take their anticoagulant first thing in the morning. Consequently, if patients require any kind of procedure (including dental), they should skip the morning dose of the direct oral anticoagulant to avoid having the procedure done during the peak anticoagulant effect, and they should either not take that day’s dose or defer the dose until the evening after the procedure. 

MANAGING SURGERY FOR PATIENTS ON A DIRECT ORAL ANTICOAGULANT

Case 3: An elderly woman on apixaban facing surgery

Let us imagine that our previous patient takes apixaban instead of warfarin. She is 75 years old, has atrial fibrillation, and is about to undergo elective colon resection for cancer. One doctor advises her to simply stop apixaban for 2 days, while another says she should go off apixaban for 5 days and will need bridging. Which plan is best?

In the perioperative setting, our goal is to interrupt patients’ anticoagulant therapy for the shortest time that results in no residual anticoagulant effect at the time of the procedure. 

They further recommend that if the risk of venous thromboembolism is high, low-molecular-weight heparin bridging should be done while stopping the direct oral anticoagulant, with the heparin discontinued 24 hours before the procedure. This recommendation seems counterintuitive, as it is advising replacing a short-acting anticoagulant with low-molecular-weight heparin, another short-acting anticoagulant.

The guidelines committee was unable to provide strength and grading of their recommendations, as too few well-designed studies are available to support them. The doctor in case 3 who advised stopping apixaban for 5 days and bridging is following the guidelines, but without much evidence to support this strategy.

There are no randomized, controlled trials of bridging vs no bridging in patients taking direct oral anticoagulants. Substudies exist of patients taking these drugs for atrial fibrillation who had treatment interrupted for procedures, but the studies did not randomize bridging vs no bridging, nor were bridging regimens standardized. Three of the four atrial fibrillation trials had a blinded design (warfarin vs direct oral anticoagulants), making perioperative management difficult, as physicians did not know the pharmacokinetics of the drugs their patients were taking.^22–24

We used the database from the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial²² to evaluate bridging in patients taking either warfarin or dabigatran. With an open-label study design (the blinding was only for the 110 mg and 150 mg dabigatran doses), clinicians were aware of whether patients were receiving warfarin or dabigatran, thereby facilitating perioperative management. Among dabigatran-treated patients, those who were bridged had significantly more major bleeding than those not bridged (6.5% vs 1.8%, P < .001), with no difference between the groups for stroke or systemic embolism. Although it is not a randomized controlled trial, it does provide evidence that bridging may not be advisable for patients taking a direct oral anticoagulant.

The 2017 American College of Cardiology guidelines²⁵ conclude that parenteral bridging is not indicated for direct oral anticoagulants. Although this is not based on strong evidence, the guidance appears reasonable according to the evidence at hand.

The 2017 American Heart Association Guidelines¹⁶ recommend a somewhat complex approach based on periprocedural bleeding risk and thromboembolic risk.

How long to interrupt direct oral anticoagulants?

Evidence for this approach comes from a prospective cohort study²⁷ of 541 patients being treated with dabigatran who were having an elective surgery or invasive procedure. Patients received standard perioperative management, with the timing of the last dabigatran dose before the procedure (24 hours, 48 hours, or 96 hours) based on the bleeding risk of surgery and the patient’s creatinine clearance. Dabigatran was resumed 24 to 72 hours after the procedure. No heparin bridging was done. Patients were followed for up to 30 days postoperatively. The results were favorable with few complications: one transient ischemic attack (0.2%), 10 major bleeding episodes (1.8%), and 28 minor bleeding episodes (5.2%).

A subgroup of 181 patients in this study²⁸ had a plasma sample drawn just before surgery, allowing the investigators to assess the level of coagulation factors after dabigatran interruption. Results were as follows:

• 93% had a normal prothrombin time 
• 80% had a normal activated partial thromboplastin time
• 33% had a normal thrombin time
• 81% had a normal dilute thrombin time.

The dilute thrombin time is considered the most reliable test of the anticoagulant effect of dabigatran but is not widely available. The activated partial thromboplastin time can provide a more widely used coagulation test to assess (in a less precise manner) whether there is an anticoagulant effect of dabigatran present, and more sensitive activated partial thromboplastin time assays can be used to better detect any residual dabigatran effect.

Dabigatran levels were also measured. Although 66% of patients had low drug levels just before surgery, the others still had substantial dabigatran on board. The fact that bleeding event rates were so low in this study despite the presence of dabigatran in many patients raises the question of whether having some drug on board is a good predictor of bleeding risk.

An interruption protocol with a longer interruption interval—12 to 14 hours longer than in the previous study (3 days for high-bleed risk procedures, 2 days for low-bleed risk procedures)—brought the activated partial thromboplastin time and dilute thrombin time to normal levels for 100% of patients with the protocol for high-bleeding-risk surgery. This study was based on small numbers and its interruption strategy needs further investigation.²⁹

Case 3 continued

The PAUSE study (NCT02228798), a multicenter, prospective cohort study, is designed to establish a safe, standardized protocol for the perioperative management of patients with atrial fibrillation taking dabigatran, rivaroxaban, or apixaban and will include 3,300 patients.

PATIENTS WITH A CORONARY STENT WHO NEED SURGERY

Case 4: A woman with a stent facing surgery

A 70-year-old woman needs breast cancer resection. She has coronary artery disease and had a drug-eluting stent placed 5 months ago after elective cardiac catheterization. She also has hypertension, obesity, and type 2 diabetes. Her medications include an angiotensin II receptor blocker, hydrochlorothiazide, insulin, and an oral hypoglycemic. She is also taking aspirin 81 mg daily and ticagrelor (a P2Y₁₂ receptor antagonist) 90 mg twice daily.

Her cardiologist is concerned that stopping antiplatelet therapy could trigger acute stent thrombosis, which has a 50% or higher mortality rate.

Should she stop taking aspirin before surgery? What about the ticagrelor?

Evidence concerning aspirin during surgery comes from Perioperative Ischemic Evaluation 2 (POISE-2), a double-blind, randomized controlled trial.³⁰ Patients who had known cardiovascular disease or risk factors for cardiovascular disease and were about to undergo noncardiac surgery were stratified according to whether they had been taking aspirin before the study (patients taking aspirin within 72 hours of the surgery were excluded from randomization). Participants in each group were randomized to take either aspirin or placebo just before surgery. The primary outcome was the combined rate of death or nonfatal myocardial infarction 30 days after randomization.

The study found no differences in the primary end point between the two groups. However, major bleeding occurred significantly more often in the aspirin group (4.6% vs 3.8%, hazard ratio 1.2, 95% confidence interval 1.0–1.5).

Moreover, only 4% of the patients in this trial had a cardiac stent. The trial excluded patients who had had a bare-metal stent placed within 6 weeks or a drug-eluting stent placed within 1 year, so it does not help us answer whether aspirin should be stopped for our current patient.

Is surgery safe for patients with stents?

The safety of undergoing surgery with a stent was investigated in a large US Veterans Administration retrospective cohort study.³¹ More than 20,000 patients with stents who underwent noncardiac surgery within 2 years of stent placement were compared with a control group of more than 41,000 patients with stents who did not undergo surgery. Patients were matched by stent type and cardiac risk factors at the time of stent placement.

The risk of an adverse cardiac event in both the surgical and nonsurgical cohorts was highest in the initial 6 weeks after stent placement and plateaued 6 months after stent placement, when the risk difference between the surgical and nonsurgical groups leveled off to 1%.

The risk of a major adverse cardiac event postoperatively was much more dependent on the timing of stent placement in complex and inpatient surgeries. For outpatient surgeries, the risk of a major cardiac event was very low and the timing of stent placement did not matter.

A Danish observational study³² compared more than 4,000 patients with drug-eluting stents having surgery to more than 20,000 matched controls without coronary heart disease having similar surgery. The risk of myocardial infarction or cardiac death was much higher for patients undergoing surgery within 1 month after drug-eluting stent placement compared with controls without heart disease and patients with stent placement longer than 1 month before surgery.

Our practice is to continue aspirin for surgery in patients with coronary stents regardless of the timing of placement. Although there is a small increased risk of bleeding, this must be balanced against thrombotic risk. We typically stop clopidogrel 5 to 7 days before surgery and ticagrelor 3 to 5 days before surgery. We may decide to give platelets before very-high-risk surgery (eg, intracranial, spinal) if there is a decision to continue both antiplatelet drugs—for example, in a patient who recently received a drug-eluting stent (ie, within 3 months). It is essential to involve the cardiologist and surgeon in these decisions.

BOTTOM LINE

This article reviews recommendations and evidence concerning current anticoagulant management for venous thromboembolism and perioperative care, with an emphasis on individualizing treatment for real-world patients.

TREATING ACUTE VENOUS THROMBOEMBOLISM

Case 1: Deep vein thrombosis in an otherwise healthy man

A 40-year-old man presents with 7 days of progressive right leg swelling. He has no antecedent risk factors for deep vein thrombosis or other medical problems. Venous ultrasonography reveals an iliofemoral deep vein thrombosis. How should he be managed?

• Outpatient treatment with low-molecular-weight heparin for 4 to 6 days plus warfarin
• Outpatient treatment with a direct oral anticoagulant, ie, apixaban, dabigatran (which requires 4 to 6 days of initial treatment with low-molecular-weight heparin), or rivaroxaban
• Catheter-directed thrombolysis followed by low-molecular-weight heparin, then warfarin or a direct oral anticoagulant
• Inpatient intravenous heparin for 7 to 10 days, then warfarin or a direct oral anticoagulant

All of these are acceptable for managing acute venous thromboembolism, but the clinician’s role is to identify which treatment is most appropriate for an individual patient.

Deep vein thrombosis is not a single condition

Multiple guidelines exist to help decide on a management strategy. Those of the American College of Chest Physicians (ACCP)¹ are used most often.

That said, guidelines are established for “average” patients, so it is important to look beyond guidelines and individualize management. Venous thromboembolism is not a single entity; it has a myriad of clinical presentations that could call for different treatments. Most patients have submassive deep vein thrombosis or pulmonary embolism, which is not limb-threatening nor associated with hemodynamic instability. It can also differ in terms of etiology and can be unprovoked (or idiopathic), cancer-related, catheter-associated, or provoked by surgery or immobility.

Deep vein thrombosis has a wide spectrum of presentations. It can involve the veins of the calf only, or it can involve the femoral and iliac veins and other locations including the splanchnic veins, the cerebral sinuses, and upper extremities. Pulmonary embolism can be massive (defined as being associated with hemodynamic instability or impending respiratory failure) or submassive. Similarly, patients differ in terms of baseline medical conditions, mobility, and lifestyle. Anticoagulant management decisions should take all these factors into account.

Consider clot location

Our patient with iliofemoral deep vein thrombosis is best managed differently than a more typical patient with less extensive thrombosis that would involve the popliteal or femoral vein segments, or both. A clot that involves the iliac vein is more likely to lead to postthrombotic chronic pain and swelling as the lack of venous outflow bypass channels to circumvent the clot location creates higher venous pressure within the affected leg. Therefore, for our patient, catheter-directed thrombolysis is an option that should be considered.

Catheter-directed thrombolysis trials

According to the “open-vein hypothesis,” quickly eliminating the thrombus and restoring unobstructed venous flow may mitigate the risk not only of recurrent thrombosis, but also of postthrombotic syndrome, which is often not given much consideration acutely but can cause significant, life-altering chronic disability.

The “valve-integrity hypothesis” is also important; it considers whether lytic therapy may help prevent damage to such valves in an attempt to mitigate the amount of venous hypertension.

Thus, catheter-directed thrombolysis offers theoretical benefits, and recent trials have assessed it against standard anticoagulation treatments.

The CaVenT trial (Catheter-Directed Venous Thrombolysis),² conducted in Norway, randomized 209 patients with midfemoral to iliac deep vein thrombosis to conventional treatment (anticoagulation alone) or anticoagulation plus catheter-directed thrombolysis. At 2 years, postthrombotic syndrome had occurred in 41% of the catheter-directed thrombolysis group compared with 56% of the conventional treatment group (P = .047). At 5 years, the difference widened to 43% vs 71% (P < .01, number needed to treat = 4).³ Despite the superiority of lytic therapy, the incidence of postthrombotic syndrome remained high in patients who received this treatment. 

The ATTRACT trial (Acute Venous Thrombosis: Thrombus Removal With Adjunctive Catheter-Directed Thrombolysis),⁴ a US multicenter, open-label, assessor-blind study, randomized 698 patients with femoral or more-proximal deep vein thrombosis to either standard care (anticoagulant therapy and graduated elastic compression stockings) or standard care plus catheter-directed thrombolysis. In preliminary results presented at the Society of Interventional Radiology meeting in March 2017, although no difference was found in the primary outcome (postthrombotic syndrome at 24 months), catheter-directed thrombolysis for iliofemoral deep vein thrombosis led to a 25% reduction in moderate to severe postthrombotic syndrome.

Although it is too early to draw conclusions before publication of the ATTRACT study, the preliminary results highlight the need to individualize treatment and to be selective about using catheter-directed thrombolysis. The trials provide reassurance that catheter-directed lysis is a reasonable and safe intervention when performed by physicians experienced in the procedure. The risk of major bleeding appears to be low (about 2%) and that for intracranial hemorrhage even lower (< 0.5%).

Catheter-directed thrombolysis is appropriate in some cases

The 2016 ACCP guidelines¹ recommend anticoagulant therapy alone over catheter-directed thrombolysis for patients with acute proximal deep vein thrombosis of the leg. However, it is a grade 2C (weak) recommendation.

They provide no specific recommendation as to the clinical indications for catheter-directed thrombolysis, but identify patients who would be most likely to benefit, ie, those who have: 

• Iliofemoral deep vein thrombosis
• Symptoms for less than 14 days
• Good functional status
• Life expectancy of more than 1 year
• Low risk of bleeding.

Our patient satisfies these criteria, suggesting that catheter-directed thrombolysis is a reasonable option for him. 

Timing is important. Catheter-directed lysis is more likely to be beneficial if used before fibrin deposits form and stiffen the venous valves, causing irreversible damage that leads to postthrombotic syndrome. 

The availability of direct oral anticoagulants has generated interest in defining their therapeutic role in patients with venous thromboembolism.

In a meta-analysis⁵ of major trials comparing direct oral anticoagulants and vitamin K antagonists such as warfarin, no significant difference was found for the risk of recurrent venous thromboembolism or venous thromboembolism-related deaths. However, fewer patients experienced major bleeding with direct oral anticoagulants (relative risk 0.61, P = .002). Although significant, the absolute risk reduction was small; the incidence of major bleeding was 1.1% with direct oral anticoagulants vs 1.8% with vitamin K antagonists.

The main advantage of direct oral anticoagulants is greater convenience for the patient.

WHICH PATIENTS ON WARFARIN NEED BRIDGING PREOPERATIVELY?

Many patients still take warfarin, particularly those with atrial fibrillation, a mechanical heart valve, or venous thromboembolism. In many countries, warfarin remains the dominant anticoagulant for stroke prevention. Whether these patients need heparin during the period of perioperative warfarin interruption is a frequently encountered scenario that, until recently, was controversial. Recent studies have helped to inform the need for heparin bridging in many of these patients.

Case 2: An elderly woman on warfarin facing cancer surgery

A 75-year-old woman weighing 65 kg is scheduled for elective colon resection for incidentally found colon cancer. She is taking warfarin for atrial fibrillation. She also has hypertension and diabetes and had a transient ischemic attack 10 years ago.

One doctor told her she needs to be assessed for heparin bridging, but another told her she does not need bridging.

The default management should be not to bridge patients who have atrial fibrillation, but to consider bridging in selected patients, such as those with recent stroke or transient ischemic attack or a prior thromboembolic event during warfarin interruption. However, decisions about bridging should not be made on the basis of the CHADS₂ score alone. For the patient described here, I would recommend not bridging.

Complex factors contribute to stroke risk

Stroke risk for patients with atrial fibrillation can be quickly estimated with the CHADS₂ score, based on: 

• Congestive heart failure (1 point)
• Hypertension (1 point)
• Age at least 75 (1 point)
• Diabetes (1 point)
• Stroke or transient ischemic attack (2 points).

Our patient has a score of 5, corresponding to an annual adjusted stroke risk of 12.5%. Whether her transient ischemic attack of 10 years ago is comparable in significance to a recent stroke is debatable and highlights a weakness of clinical prediction rules. Moreover, such prediction scores were developed to estimate the long-term risk of stroke if anticoagulants are not given, and they have not been assessed in a perioperative setting where there is short-term interruption of anticoagulants. Also, the perioperative milieu is associated with additional factors not captured in these clinical prediction rules that may affect the risk of stroke.

Thus, the risk of perioperative stroke likely involves the interplay of multiple factors, including the type of surgery the patient is undergoing. Some factors may be mitigated:

• Rebound hypercoagulability after stopping an oral anticoagulant can be prevented by intraoperative blood pressure and volume control
• Elevated biochemical factors (eg, D-dimer, B-type natriuretic peptide, troponin) may be lowered with perioperative aspirin therapy
• Lipid and genetic factors may be mitigated with perioperative statin use.

Can heparin bridging also mitigate the risk?

Bridging in patients with atrial fibrillation

Most patients who are taking warfarin are doing so because of atrial fibrillation, so most evidence about perioperative bridging was developed in such patients.

The BRIDGE trial (Bridging Anticoagulation in Patients Who Require Temporary Interruption of Warfarin Therapy for an Elective Invasive Procedure or Surgery)⁶ was the first randomized controlled trial to compare a bridging and no-bridging strategy for patients with atrial fibrillation who required warfarin interruption for elective surgery. Nearly 2,000 patients were given either low-molecular-weight heparin or placebo starting 3 days before until 24 hours before a procedure, and then for 5 to 10 days afterwards. For all patients, warfarin was stopped 5 days before the procedure and was resumed within 24 hours afterwards.

A no-bridging strategy was noninferior to bridging: the risk of perioperative arterial thromboembolism was 0.4% without bridging vs 0.3% with bridging (P = .01 for noninferiority). In addition, a no-bridging strategy conferred a lower risk of major bleeding than bridging: 1.3% vs 3.2% (relative risk 0.41, P = .005 for superiority).

Although the difference in absolute bleeding risk was small, bleeding rates were lower than those seen outside of clinical trials, as the bridging protocol used in BRIDGE was designed to minimize the risk of bleeding. Also, although only 5% of patients had a CHADS₂ score of 5 or 6, such patients are infrequent in clinical practice, and BRIDGE did include a considerable proportion (17%) of patients with a prior stroke or transient ischemic attack who would be considered at high risk.

Other evidence about heparin bridging is derived from observational studies, more than 10 of which have been conducted. In general, they have found that not bridging is associated with low rates of arterial thromboembolism (< 0.5%) and that bridging is associated with high rates of major bleeding (4%–7%).^7–12

Bridging in patients with a mechanical heart valve

Warfarin is the only anticoagulant option for patients who have a mechanical heart valve. No randomized controlled trials have evaluated the benefits of perioperative bridging vs no bridging in this setting.

Observational (cohort) studies suggest that the risk of perioperative arterial thromboembolism is similar with or without bridging anticoagulation, although most patients studied were bridged and those not bridged were considered at low risk (eg, with a bileaflet aortic valve and no additional risk factors).¹³ However, without stronger evidence from randomized controlled trials, bridging should be the default management for patients with a mechanical heart valve. In our practice, we bridge most patients who have a mechanical heart valve unless they are considered to be at low risk, such as those who have a bileaflet aortic valve.

Even less evidence is available for periprocedural management of patients who have a history of venous thromboembolism. No randomized controlled trials exist evaluating bridging vs no bridging. In 1 cohort study in which more than 90% of patients had had thromboembolism more than 3 months before the procedure, the rate of recurrent venous thromboembolism without bridging was less than 0.5%.¹⁴

It is reasonable to bridge patients who need anticoagulant interruption within 3 months of diagnosis of a deep vein thrombosis or pulmonary embolism, and to consider using a temporary inferior vena cava filter for patients who have had a clot who need treatment interruption during the initial 3 to 4 weeks after diagnosis.

Practice guidelines: Perioperative anticoagulation

Guidance for preoperative and postoperative bridging for patients taking warfarin is summarized in Table 2.

CARDIAC PROCEDURES

For patients facing a procedure to implant an implantable cardioverter-defibrillator (ICD) or pacemaker, a procedure-specific concern is the avoidance of pocket hematoma.

Patients on warfarin: Do not bridge

The BRUISE CONTROL-1 trial (Bridge or Continue Coumadin for Device Surgery Randomized Controlled Trial)¹⁹ randomized patients undergoing pacemaker or ICD implantation to either continued anticoagulation therapy and not bridging (ie, continued warfarin so long as the international normalized ratio was < 3) vs conventional bridging treatment (ie, stopping warfarin and bridging with low-molecular-weight heparin). A clinically significant device-pocket hematoma occurred in 3.5% of the continued-warfarin group vs 16.0% in the heparin-bridging group (P < .001). Thromboembolic complications were rare, and rates did not differ between the 2 groups.

Results of the BRUISE CONTROL-1 trial serve as a caution to at least not be too aggressive with bridging. The study design involved resuming heparin 24 hours after surgery, which is perhaps more aggressive than standard practice. In our practice, we wait at least 24 hours to reinstate heparin after minor surgery, and 48 to 72 hours after surgery with higher bleeding risk.

These results are perhaps not surprising if one considers how carefully surgeons try to control bleeding during surgery for patients taking anticoagulants. For patients who are not on an anticoagulant, small bleeding may be less of a concern during a procedure. When high doses of heparin are introduced soon after surgery, small concerns during surgery may become big problems afterward.

Based on these results, it is reasonable to undertake device implantation without interruption of a vitamin K antagonist such as warfarin.

Patients on direct oral anticoagulants: The jury is still out

The similar BRUISE CONTROL-2 trial is currently under way, comparing interruption vs continuation of dabigatran for patients undergoing cardiac device surgery.

In Europe, surgeons are less concerned than those in the United States about operating while a patient is on anticoagulant therapy. But the safety of this practice is not backed by strong evidence.

Direct oral anticoagulants: Consider pharmacokinetics

Direct oral anticoagulants are potent and fast-acting, with a peak effect 1 to 3 hours after intake. This rapid anticoagulant action is similar to that of bridging with low-molecular-weight heparin, and caution is needed when administering direct oral anticoagulants, especially after major surgery or surgery with a high bleeding risk.

Frost et al²⁰ compared the pharmacokinetics of apixaban (with twice-daily dosing) and rivaroxaban (once-daily dosing) and found that peak anticoagulant activity is faster and higher with rivaroxaban. This is important, because many patients will take their anticoagulant first thing in the morning. Consequently, if patients require any kind of procedure (including dental), they should skip the morning dose of the direct oral anticoagulant to avoid having the procedure done during the peak anticoagulant effect, and they should either not take that day’s dose or defer the dose until the evening after the procedure. 

MANAGING SURGERY FOR PATIENTS ON A DIRECT ORAL ANTICOAGULANT

Case 3: An elderly woman on apixaban facing surgery

Let us imagine that our previous patient takes apixaban instead of warfarin. She is 75 years old, has atrial fibrillation, and is about to undergo elective colon resection for cancer. One doctor advises her to simply stop apixaban for 2 days, while another says she should go off apixaban for 5 days and will need bridging. Which plan is best?

In the perioperative setting, our goal is to interrupt patients’ anticoagulant therapy for the shortest time that results in no residual anticoagulant effect at the time of the procedure. 

They further recommend that if the risk of venous thromboembolism is high, low-molecular-weight heparin bridging should be done while stopping the direct oral anticoagulant, with the heparin discontinued 24 hours before the procedure. This recommendation seems counterintuitive, as it is advising replacing a short-acting anticoagulant with low-molecular-weight heparin, another short-acting anticoagulant.

The guidelines committee was unable to provide strength and grading of their recommendations, as too few well-designed studies are available to support them. The doctor in case 3 who advised stopping apixaban for 5 days and bridging is following the guidelines, but without much evidence to support this strategy.

There are no randomized, controlled trials of bridging vs no bridging in patients taking direct oral anticoagulants. Substudies exist of patients taking these drugs for atrial fibrillation who had treatment interrupted for procedures, but the studies did not randomize bridging vs no bridging, nor were bridging regimens standardized. Three of the four atrial fibrillation trials had a blinded design (warfarin vs direct oral anticoagulants), making perioperative management difficult, as physicians did not know the pharmacokinetics of the drugs their patients were taking.^22–24

We used the database from the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial²² to evaluate bridging in patients taking either warfarin or dabigatran. With an open-label study design (the blinding was only for the 110 mg and 150 mg dabigatran doses), clinicians were aware of whether patients were receiving warfarin or dabigatran, thereby facilitating perioperative management. Among dabigatran-treated patients, those who were bridged had significantly more major bleeding than those not bridged (6.5% vs 1.8%, P < .001), with no difference between the groups for stroke or systemic embolism. Although it is not a randomized controlled trial, it does provide evidence that bridging may not be advisable for patients taking a direct oral anticoagulant.

The 2017 American College of Cardiology guidelines²⁵ conclude that parenteral bridging is not indicated for direct oral anticoagulants. Although this is not based on strong evidence, the guidance appears reasonable according to the evidence at hand.

The 2017 American Heart Association Guidelines¹⁶ recommend a somewhat complex approach based on periprocedural bleeding risk and thromboembolic risk.

How long to interrupt direct oral anticoagulants?

Evidence for this approach comes from a prospective cohort study²⁷ of 541 patients being treated with dabigatran who were having an elective surgery or invasive procedure. Patients received standard perioperative management, with the timing of the last dabigatran dose before the procedure (24 hours, 48 hours, or 96 hours) based on the bleeding risk of surgery and the patient’s creatinine clearance. Dabigatran was resumed 24 to 72 hours after the procedure. No heparin bridging was done. Patients were followed for up to 30 days postoperatively. The results were favorable with few complications: one transient ischemic attack (0.2%), 10 major bleeding episodes (1.8%), and 28 minor bleeding episodes (5.2%).

A subgroup of 181 patients in this study²⁸ had a plasma sample drawn just before surgery, allowing the investigators to assess the level of coagulation factors after dabigatran interruption. Results were as follows:

• 93% had a normal prothrombin time 
• 80% had a normal activated partial thromboplastin time
• 33% had a normal thrombin time
• 81% had a normal dilute thrombin time.

The dilute thrombin time is considered the most reliable test of the anticoagulant effect of dabigatran but is not widely available. The activated partial thromboplastin time can provide a more widely used coagulation test to assess (in a less precise manner) whether there is an anticoagulant effect of dabigatran present, and more sensitive activated partial thromboplastin time assays can be used to better detect any residual dabigatran effect.

Dabigatran levels were also measured. Although 66% of patients had low drug levels just before surgery, the others still had substantial dabigatran on board. The fact that bleeding event rates were so low in this study despite the presence of dabigatran in many patients raises the question of whether having some drug on board is a good predictor of bleeding risk.

An interruption protocol with a longer interruption interval—12 to 14 hours longer than in the previous study (3 days for high-bleed risk procedures, 2 days for low-bleed risk procedures)—brought the activated partial thromboplastin time and dilute thrombin time to normal levels for 100% of patients with the protocol for high-bleeding-risk surgery. This study was based on small numbers and its interruption strategy needs further investigation.²⁹

Case 3 continued

The PAUSE study (NCT02228798), a multicenter, prospective cohort study, is designed to establish a safe, standardized protocol for the perioperative management of patients with atrial fibrillation taking dabigatran, rivaroxaban, or apixaban and will include 3,300 patients.

PATIENTS WITH A CORONARY STENT WHO NEED SURGERY

Case 4: A woman with a stent facing surgery

A 70-year-old woman needs breast cancer resection. She has coronary artery disease and had a drug-eluting stent placed 5 months ago after elective cardiac catheterization. She also has hypertension, obesity, and type 2 diabetes. Her medications include an angiotensin II receptor blocker, hydrochlorothiazide, insulin, and an oral hypoglycemic. She is also taking aspirin 81 mg daily and ticagrelor (a P2Y₁₂ receptor antagonist) 90 mg twice daily.

Her cardiologist is concerned that stopping antiplatelet therapy could trigger acute stent thrombosis, which has a 50% or higher mortality rate.

Should she stop taking aspirin before surgery? What about the ticagrelor?

Evidence concerning aspirin during surgery comes from Perioperative Ischemic Evaluation 2 (POISE-2), a double-blind, randomized controlled trial.³⁰ Patients who had known cardiovascular disease or risk factors for cardiovascular disease and were about to undergo noncardiac surgery were stratified according to whether they had been taking aspirin before the study (patients taking aspirin within 72 hours of the surgery were excluded from randomization). Participants in each group were randomized to take either aspirin or placebo just before surgery. The primary outcome was the combined rate of death or nonfatal myocardial infarction 30 days after randomization.

The study found no differences in the primary end point between the two groups. However, major bleeding occurred significantly more often in the aspirin group (4.6% vs 3.8%, hazard ratio 1.2, 95% confidence interval 1.0–1.5).

Moreover, only 4% of the patients in this trial had a cardiac stent. The trial excluded patients who had had a bare-metal stent placed within 6 weeks or a drug-eluting stent placed within 1 year, so it does not help us answer whether aspirin should be stopped for our current patient.

Is surgery safe for patients with stents?

The safety of undergoing surgery with a stent was investigated in a large US Veterans Administration retrospective cohort study.³¹ More than 20,000 patients with stents who underwent noncardiac surgery within 2 years of stent placement were compared with a control group of more than 41,000 patients with stents who did not undergo surgery. Patients were matched by stent type and cardiac risk factors at the time of stent placement.

The risk of an adverse cardiac event in both the surgical and nonsurgical cohorts was highest in the initial 6 weeks after stent placement and plateaued 6 months after stent placement, when the risk difference between the surgical and nonsurgical groups leveled off to 1%.

The risk of a major adverse cardiac event postoperatively was much more dependent on the timing of stent placement in complex and inpatient surgeries. For outpatient surgeries, the risk of a major cardiac event was very low and the timing of stent placement did not matter.

A Danish observational study³² compared more than 4,000 patients with drug-eluting stents having surgery to more than 20,000 matched controls without coronary heart disease having similar surgery. The risk of myocardial infarction or cardiac death was much higher for patients undergoing surgery within 1 month after drug-eluting stent placement compared with controls without heart disease and patients with stent placement longer than 1 month before surgery.

Our practice is to continue aspirin for surgery in patients with coronary stents regardless of the timing of placement. Although there is a small increased risk of bleeding, this must be balanced against thrombotic risk. We typically stop clopidogrel 5 to 7 days before surgery and ticagrelor 3 to 5 days before surgery. We may decide to give platelets before very-high-risk surgery (eg, intracranial, spinal) if there is a decision to continue both antiplatelet drugs—for example, in a patient who recently received a drug-eluting stent (ie, within 3 months). It is essential to involve the cardiologist and surgeon in these decisions.

BOTTOM LINE

This article reviews recommendations and evidence concerning current anticoagulant management for venous thromboembolism and perioperative care, with an emphasis on individualizing treatment for real-world patients.

TREATING ACUTE VENOUS THROMBOEMBOLISM

Case 1: Deep vein thrombosis in an otherwise healthy man

A 40-year-old man presents with 7 days of progressive right leg swelling. He has no antecedent risk factors for deep vein thrombosis or other medical problems. Venous ultrasonography reveals an iliofemoral deep vein thrombosis. How should he be managed?

• Outpatient treatment with low-molecular-weight heparin for 4 to 6 days plus warfarin
• Outpatient treatment with a direct oral anticoagulant, ie, apixaban, dabigatran (which requires 4 to 6 days of initial treatment with low-molecular-weight heparin), or rivaroxaban
• Catheter-directed thrombolysis followed by low-molecular-weight heparin, then warfarin or a direct oral anticoagulant
• Inpatient intravenous heparin for 7 to 10 days, then warfarin or a direct oral anticoagulant

All of these are acceptable for managing acute venous thromboembolism, but the clinician’s role is to identify which treatment is most appropriate for an individual patient.

Deep vein thrombosis is not a single condition

Multiple guidelines exist to help decide on a management strategy. Those of the American College of Chest Physicians (ACCP)¹ are used most often.

That said, guidelines are established for “average” patients, so it is important to look beyond guidelines and individualize management. Venous thromboembolism is not a single entity; it has a myriad of clinical presentations that could call for different treatments. Most patients have submassive deep vein thrombosis or pulmonary embolism, which is not limb-threatening nor associated with hemodynamic instability. It can also differ in terms of etiology and can be unprovoked (or idiopathic), cancer-related, catheter-associated, or provoked by surgery or immobility.

Deep vein thrombosis has a wide spectrum of presentations. It can involve the veins of the calf only, or it can involve the femoral and iliac veins and other locations including the splanchnic veins, the cerebral sinuses, and upper extremities. Pulmonary embolism can be massive (defined as being associated with hemodynamic instability or impending respiratory failure) or submassive. Similarly, patients differ in terms of baseline medical conditions, mobility, and lifestyle. Anticoagulant management decisions should take all these factors into account.

Consider clot location

Our patient with iliofemoral deep vein thrombosis is best managed differently than a more typical patient with less extensive thrombosis that would involve the popliteal or femoral vein segments, or both. A clot that involves the iliac vein is more likely to lead to postthrombotic chronic pain and swelling as the lack of venous outflow bypass channels to circumvent the clot location creates higher venous pressure within the affected leg. Therefore, for our patient, catheter-directed thrombolysis is an option that should be considered.

Catheter-directed thrombolysis trials

According to the “open-vein hypothesis,” quickly eliminating the thrombus and restoring unobstructed venous flow may mitigate the risk not only of recurrent thrombosis, but also of postthrombotic syndrome, which is often not given much consideration acutely but can cause significant, life-altering chronic disability.

The “valve-integrity hypothesis” is also important; it considers whether lytic therapy may help prevent damage to such valves in an attempt to mitigate the amount of venous hypertension.

Thus, catheter-directed thrombolysis offers theoretical benefits, and recent trials have assessed it against standard anticoagulation treatments.

The CaVenT trial (Catheter-Directed Venous Thrombolysis),² conducted in Norway, randomized 209 patients with midfemoral to iliac deep vein thrombosis to conventional treatment (anticoagulation alone) or anticoagulation plus catheter-directed thrombolysis. At 2 years, postthrombotic syndrome had occurred in 41% of the catheter-directed thrombolysis group compared with 56% of the conventional treatment group (P = .047). At 5 years, the difference widened to 43% vs 71% (P < .01, number needed to treat = 4).³ Despite the superiority of lytic therapy, the incidence of postthrombotic syndrome remained high in patients who received this treatment. 

The ATTRACT trial (Acute Venous Thrombosis: Thrombus Removal With Adjunctive Catheter-Directed Thrombolysis),⁴ a US multicenter, open-label, assessor-blind study, randomized 698 patients with femoral or more-proximal deep vein thrombosis to either standard care (anticoagulant therapy and graduated elastic compression stockings) or standard care plus catheter-directed thrombolysis. In preliminary results presented at the Society of Interventional Radiology meeting in March 2017, although no difference was found in the primary outcome (postthrombotic syndrome at 24 months), catheter-directed thrombolysis for iliofemoral deep vein thrombosis led to a 25% reduction in moderate to severe postthrombotic syndrome.

Although it is too early to draw conclusions before publication of the ATTRACT study, the preliminary results highlight the need to individualize treatment and to be selective about using catheter-directed thrombolysis. The trials provide reassurance that catheter-directed lysis is a reasonable and safe intervention when performed by physicians experienced in the procedure. The risk of major bleeding appears to be low (about 2%) and that for intracranial hemorrhage even lower (< 0.5%).

Catheter-directed thrombolysis is appropriate in some cases

The 2016 ACCP guidelines¹ recommend anticoagulant therapy alone over catheter-directed thrombolysis for patients with acute proximal deep vein thrombosis of the leg. However, it is a grade 2C (weak) recommendation.

They provide no specific recommendation as to the clinical indications for catheter-directed thrombolysis, but identify patients who would be most likely to benefit, ie, those who have: 

• Iliofemoral deep vein thrombosis
• Symptoms for less than 14 days
• Good functional status
• Life expectancy of more than 1 year
• Low risk of bleeding.

Our patient satisfies these criteria, suggesting that catheter-directed thrombolysis is a reasonable option for him. 

Timing is important. Catheter-directed lysis is more likely to be beneficial if used before fibrin deposits form and stiffen the venous valves, causing irreversible damage that leads to postthrombotic syndrome. 

The availability of direct oral anticoagulants has generated interest in defining their therapeutic role in patients with venous thromboembolism.

In a meta-analysis⁵ of major trials comparing direct oral anticoagulants and vitamin K antagonists such as warfarin, no significant difference was found for the risk of recurrent venous thromboembolism or venous thromboembolism-related deaths. However, fewer patients experienced major bleeding with direct oral anticoagulants (relative risk 0.61, P = .002). Although significant, the absolute risk reduction was small; the incidence of major bleeding was 1.1% with direct oral anticoagulants vs 1.8% with vitamin K antagonists.

The main advantage of direct oral anticoagulants is greater convenience for the patient.

WHICH PATIENTS ON WARFARIN NEED BRIDGING PREOPERATIVELY?

Many patients still take warfarin, particularly those with atrial fibrillation, a mechanical heart valve, or venous thromboembolism. In many countries, warfarin remains the dominant anticoagulant for stroke prevention. Whether these patients need heparin during the period of perioperative warfarin interruption is a frequently encountered scenario that, until recently, was controversial. Recent studies have helped to inform the need for heparin bridging in many of these patients.

Case 2: An elderly woman on warfarin facing cancer surgery

A 75-year-old woman weighing 65 kg is scheduled for elective colon resection for incidentally found colon cancer. She is taking warfarin for atrial fibrillation. She also has hypertension and diabetes and had a transient ischemic attack 10 years ago.

One doctor told her she needs to be assessed for heparin bridging, but another told her she does not need bridging.

The default management should be not to bridge patients who have atrial fibrillation, but to consider bridging in selected patients, such as those with recent stroke or transient ischemic attack or a prior thromboembolic event during warfarin interruption. However, decisions about bridging should not be made on the basis of the CHADS₂ score alone. For the patient described here, I would recommend not bridging.

Complex factors contribute to stroke risk

Stroke risk for patients with atrial fibrillation can be quickly estimated with the CHADS₂ score, based on: 

• Congestive heart failure (1 point)
• Hypertension (1 point)
• Age at least 75 (1 point)
• Diabetes (1 point)
• Stroke or transient ischemic attack (2 points).

Our patient has a score of 5, corresponding to an annual adjusted stroke risk of 12.5%. Whether her transient ischemic attack of 10 years ago is comparable in significance to a recent stroke is debatable and highlights a weakness of clinical prediction rules. Moreover, such prediction scores were developed to estimate the long-term risk of stroke if anticoagulants are not given, and they have not been assessed in a perioperative setting where there is short-term interruption of anticoagulants. Also, the perioperative milieu is associated with additional factors not captured in these clinical prediction rules that may affect the risk of stroke.

Thus, the risk of perioperative stroke likely involves the interplay of multiple factors, including the type of surgery the patient is undergoing. Some factors may be mitigated:

• Rebound hypercoagulability after stopping an oral anticoagulant can be prevented by intraoperative blood pressure and volume control
• Elevated biochemical factors (eg, D-dimer, B-type natriuretic peptide, troponin) may be lowered with perioperative aspirin therapy
• Lipid and genetic factors may be mitigated with perioperative statin use.

Can heparin bridging also mitigate the risk?

Bridging in patients with atrial fibrillation

Most patients who are taking warfarin are doing so because of atrial fibrillation, so most evidence about perioperative bridging was developed in such patients.

The BRIDGE trial (Bridging Anticoagulation in Patients Who Require Temporary Interruption of Warfarin Therapy for an Elective Invasive Procedure or Surgery)⁶ was the first randomized controlled trial to compare a bridging and no-bridging strategy for patients with atrial fibrillation who required warfarin interruption for elective surgery. Nearly 2,000 patients were given either low-molecular-weight heparin or placebo starting 3 days before until 24 hours before a procedure, and then for 5 to 10 days afterwards. For all patients, warfarin was stopped 5 days before the procedure and was resumed within 24 hours afterwards.

A no-bridging strategy was noninferior to bridging: the risk of perioperative arterial thromboembolism was 0.4% without bridging vs 0.3% with bridging (P = .01 for noninferiority). In addition, a no-bridging strategy conferred a lower risk of major bleeding than bridging: 1.3% vs 3.2% (relative risk 0.41, P = .005 for superiority).

Although the difference in absolute bleeding risk was small, bleeding rates were lower than those seen outside of clinical trials, as the bridging protocol used in BRIDGE was designed to minimize the risk of bleeding. Also, although only 5% of patients had a CHADS₂ score of 5 or 6, such patients are infrequent in clinical practice, and BRIDGE did include a considerable proportion (17%) of patients with a prior stroke or transient ischemic attack who would be considered at high risk.

Other evidence about heparin bridging is derived from observational studies, more than 10 of which have been conducted. In general, they have found that not bridging is associated with low rates of arterial thromboembolism (< 0.5%) and that bridging is associated with high rates of major bleeding (4%–7%).^7–12

Bridging in patients with a mechanical heart valve

Warfarin is the only anticoagulant option for patients who have a mechanical heart valve. No randomized controlled trials have evaluated the benefits of perioperative bridging vs no bridging in this setting.

Observational (cohort) studies suggest that the risk of perioperative arterial thromboembolism is similar with or without bridging anticoagulation, although most patients studied were bridged and those not bridged were considered at low risk (eg, with a bileaflet aortic valve and no additional risk factors).¹³ However, without stronger evidence from randomized controlled trials, bridging should be the default management for patients with a mechanical heart valve. In our practice, we bridge most patients who have a mechanical heart valve unless they are considered to be at low risk, such as those who have a bileaflet aortic valve.

Even less evidence is available for periprocedural management of patients who have a history of venous thromboembolism. No randomized controlled trials exist evaluating bridging vs no bridging. In 1 cohort study in which more than 90% of patients had had thromboembolism more than 3 months before the procedure, the rate of recurrent venous thromboembolism without bridging was less than 0.5%.¹⁴

It is reasonable to bridge patients who need anticoagulant interruption within 3 months of diagnosis of a deep vein thrombosis or pulmonary embolism, and to consider using a temporary inferior vena cava filter for patients who have had a clot who need treatment interruption during the initial 3 to 4 weeks after diagnosis.

Practice guidelines: Perioperative anticoagulation

Guidance for preoperative and postoperative bridging for patients taking warfarin is summarized in Table 2.

CARDIAC PROCEDURES

For patients facing a procedure to implant an implantable cardioverter-defibrillator (ICD) or pacemaker, a procedure-specific concern is the avoidance of pocket hematoma.

Patients on warfarin: Do not bridge

The BRUISE CONTROL-1 trial (Bridge or Continue Coumadin for Device Surgery Randomized Controlled Trial)¹⁹ randomized patients undergoing pacemaker or ICD implantation to either continued anticoagulation therapy and not bridging (ie, continued warfarin so long as the international normalized ratio was < 3) vs conventional bridging treatment (ie, stopping warfarin and bridging with low-molecular-weight heparin). A clinically significant device-pocket hematoma occurred in 3.5% of the continued-warfarin group vs 16.0% in the heparin-bridging group (P < .001). Thromboembolic complications were rare, and rates did not differ between the 2 groups.

Results of the BRUISE CONTROL-1 trial serve as a caution to at least not be too aggressive with bridging. The study design involved resuming heparin 24 hours after surgery, which is perhaps more aggressive than standard practice. In our practice, we wait at least 24 hours to reinstate heparin after minor surgery, and 48 to 72 hours after surgery with higher bleeding risk.

These results are perhaps not surprising if one considers how carefully surgeons try to control bleeding during surgery for patients taking anticoagulants. For patients who are not on an anticoagulant, small bleeding may be less of a concern during a procedure. When high doses of heparin are introduced soon after surgery, small concerns during surgery may become big problems afterward.

Based on these results, it is reasonable to undertake device implantation without interruption of a vitamin K antagonist such as warfarin.

Patients on direct oral anticoagulants: The jury is still out

The similar BRUISE CONTROL-2 trial is currently under way, comparing interruption vs continuation of dabigatran for patients undergoing cardiac device surgery.

In Europe, surgeons are less concerned than those in the United States about operating while a patient is on anticoagulant therapy. But the safety of this practice is not backed by strong evidence.

Direct oral anticoagulants: Consider pharmacokinetics

Direct oral anticoagulants are potent and fast-acting, with a peak effect 1 to 3 hours after intake. This rapid anticoagulant action is similar to that of bridging with low-molecular-weight heparin, and caution is needed when administering direct oral anticoagulants, especially after major surgery or surgery with a high bleeding risk.

Frost et al²⁰ compared the pharmacokinetics of apixaban (with twice-daily dosing) and rivaroxaban (once-daily dosing) and found that peak anticoagulant activity is faster and higher with rivaroxaban. This is important, because many patients will take their anticoagulant first thing in the morning. Consequently, if patients require any kind of procedure (including dental), they should skip the morning dose of the direct oral anticoagulant to avoid having the procedure done during the peak anticoagulant effect, and they should either not take that day’s dose or defer the dose until the evening after the procedure. 

MANAGING SURGERY FOR PATIENTS ON A DIRECT ORAL ANTICOAGULANT

Case 3: An elderly woman on apixaban facing surgery

Let us imagine that our previous patient takes apixaban instead of warfarin. She is 75 years old, has atrial fibrillation, and is about to undergo elective colon resection for cancer. One doctor advises her to simply stop apixaban for 2 days, while another says she should go off apixaban for 5 days and will need bridging. Which plan is best?

In the perioperative setting, our goal is to interrupt patients’ anticoagulant therapy for the shortest time that results in no residual anticoagulant effect at the time of the procedure. 

They further recommend that if the risk of venous thromboembolism is high, low-molecular-weight heparin bridging should be done while stopping the direct oral anticoagulant, with the heparin discontinued 24 hours before the procedure. This recommendation seems counterintuitive, as it is advising replacing a short-acting anticoagulant with low-molecular-weight heparin, another short-acting anticoagulant.

The guidelines committee was unable to provide strength and grading of their recommendations, as too few well-designed studies are available to support them. The doctor in case 3 who advised stopping apixaban for 5 days and bridging is following the guidelines, but without much evidence to support this strategy.

There are no randomized, controlled trials of bridging vs no bridging in patients taking direct oral anticoagulants. Substudies exist of patients taking these drugs for atrial fibrillation who had treatment interrupted for procedures, but the studies did not randomize bridging vs no bridging, nor were bridging regimens standardized. Three of the four atrial fibrillation trials had a blinded design (warfarin vs direct oral anticoagulants), making perioperative management difficult, as physicians did not know the pharmacokinetics of the drugs their patients were taking.^22–24

We used the database from the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial²² to evaluate bridging in patients taking either warfarin or dabigatran. With an open-label study design (the blinding was only for the 110 mg and 150 mg dabigatran doses), clinicians were aware of whether patients were receiving warfarin or dabigatran, thereby facilitating perioperative management. Among dabigatran-treated patients, those who were bridged had significantly more major bleeding than those not bridged (6.5% vs 1.8%, P < .001), with no difference between the groups for stroke or systemic embolism. Although it is not a randomized controlled trial, it does provide evidence that bridging may not be advisable for patients taking a direct oral anticoagulant.

The 2017 American College of Cardiology guidelines²⁵ conclude that parenteral bridging is not indicated for direct oral anticoagulants. Although this is not based on strong evidence, the guidance appears reasonable according to the evidence at hand.

The 2017 American Heart Association Guidelines¹⁶ recommend a somewhat complex approach based on periprocedural bleeding risk and thromboembolic risk.

How long to interrupt direct oral anticoagulants?

Evidence for this approach comes from a prospective cohort study²⁷ of 541 patients being treated with dabigatran who were having an elective surgery or invasive procedure. Patients received standard perioperative management, with the timing of the last dabigatran dose before the procedure (24 hours, 48 hours, or 96 hours) based on the bleeding risk of surgery and the patient’s creatinine clearance. Dabigatran was resumed 24 to 72 hours after the procedure. No heparin bridging was done. Patients were followed for up to 30 days postoperatively. The results were favorable with few complications: one transient ischemic attack (0.2%), 10 major bleeding episodes (1.8%), and 28 minor bleeding episodes (5.2%).

A subgroup of 181 patients in this study²⁸ had a plasma sample drawn just before surgery, allowing the investigators to assess the level of coagulation factors after dabigatran interruption. Results were as follows:

• 93% had a normal prothrombin time 
• 80% had a normal activated partial thromboplastin time
• 33% had a normal thrombin time
• 81% had a normal dilute thrombin time.

The dilute thrombin time is considered the most reliable test of the anticoagulant effect of dabigatran but is not widely available. The activated partial thromboplastin time can provide a more widely used coagulation test to assess (in a less precise manner) whether there is an anticoagulant effect of dabigatran present, and more sensitive activated partial thromboplastin time assays can be used to better detect any residual dabigatran effect.

Dabigatran levels were also measured. Although 66% of patients had low drug levels just before surgery, the others still had substantial dabigatran on board. The fact that bleeding event rates were so low in this study despite the presence of dabigatran in many patients raises the question of whether having some drug on board is a good predictor of bleeding risk.

An interruption protocol with a longer interruption interval—12 to 14 hours longer than in the previous study (3 days for high-bleed risk procedures, 2 days for low-bleed risk procedures)—brought the activated partial thromboplastin time and dilute thrombin time to normal levels for 100% of patients with the protocol for high-bleeding-risk surgery. This study was based on small numbers and its interruption strategy needs further investigation.²⁹

Case 3 continued

The PAUSE study (NCT02228798), a multicenter, prospective cohort study, is designed to establish a safe, standardized protocol for the perioperative management of patients with atrial fibrillation taking dabigatran, rivaroxaban, or apixaban and will include 3,300 patients.

PATIENTS WITH A CORONARY STENT WHO NEED SURGERY

Case 4: A woman with a stent facing surgery

A 70-year-old woman needs breast cancer resection. She has coronary artery disease and had a drug-eluting stent placed 5 months ago after elective cardiac catheterization. She also has hypertension, obesity, and type 2 diabetes. Her medications include an angiotensin II receptor blocker, hydrochlorothiazide, insulin, and an oral hypoglycemic. She is also taking aspirin 81 mg daily and ticagrelor (a P2Y₁₂ receptor antagonist) 90 mg twice daily.

Her cardiologist is concerned that stopping antiplatelet therapy could trigger acute stent thrombosis, which has a 50% or higher mortality rate.

Should she stop taking aspirin before surgery? What about the ticagrelor?

Evidence concerning aspirin during surgery comes from Perioperative Ischemic Evaluation 2 (POISE-2), a double-blind, randomized controlled trial.³⁰ Patients who had known cardiovascular disease or risk factors for cardiovascular disease and were about to undergo noncardiac surgery were stratified according to whether they had been taking aspirin before the study (patients taking aspirin within 72 hours of the surgery were excluded from randomization). Participants in each group were randomized to take either aspirin or placebo just before surgery. The primary outcome was the combined rate of death or nonfatal myocardial infarction 30 days after randomization.

The study found no differences in the primary end point between the two groups. However, major bleeding occurred significantly more often in the aspirin group (4.6% vs 3.8%, hazard ratio 1.2, 95% confidence interval 1.0–1.5).

Moreover, only 4% of the patients in this trial had a cardiac stent. The trial excluded patients who had had a bare-metal stent placed within 6 weeks or a drug-eluting stent placed within 1 year, so it does not help us answer whether aspirin should be stopped for our current patient.

Is surgery safe for patients with stents?

The safety of undergoing surgery with a stent was investigated in a large US Veterans Administration retrospective cohort study.³¹ More than 20,000 patients with stents who underwent noncardiac surgery within 2 years of stent placement were compared with a control group of more than 41,000 patients with stents who did not undergo surgery. Patients were matched by stent type and cardiac risk factors at the time of stent placement.

The risk of an adverse cardiac event in both the surgical and nonsurgical cohorts was highest in the initial 6 weeks after stent placement and plateaued 6 months after stent placement, when the risk difference between the surgical and nonsurgical groups leveled off to 1%.

The risk of a major adverse cardiac event postoperatively was much more dependent on the timing of stent placement in complex and inpatient surgeries. For outpatient surgeries, the risk of a major cardiac event was very low and the timing of stent placement did not matter.

A Danish observational study³² compared more than 4,000 patients with drug-eluting stents having surgery to more than 20,000 matched controls without coronary heart disease having similar surgery. The risk of myocardial infarction or cardiac death was much higher for patients undergoing surgery within 1 month after drug-eluting stent placement compared with controls without heart disease and patients with stent placement longer than 1 month before surgery.

Our practice is to continue aspirin for surgery in patients with coronary stents regardless of the timing of placement. Although there is a small increased risk of bleeding, this must be balanced against thrombotic risk. We typically stop clopidogrel 5 to 7 days before surgery and ticagrelor 3 to 5 days before surgery. We may decide to give platelets before very-high-risk surgery (eg, intracranial, spinal) if there is a decision to continue both antiplatelet drugs—for example, in a patient who recently received a drug-eluting stent (ie, within 3 months). It is essential to involve the cardiologist and surgeon in these decisions.

BOTTOM LINE

Since 2008, the US Food and Drug Administration (FDA) has required new diabetes drugs to demonstrate cardiovascular safety, resulting in large and lengthy clinical trials. Under the new regulations, several dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and glucagon-like peptide-1 (GLP-1) receptor agonists have demonstrated cardiovascular safety, with some demonstrating superior cardiovascular efficacy. In 2016, the SGLT-2 inhibitor empagliflozin became the first (and as of this writing, the only) diabetes drug approved by the FDA for a clinical outcome indication, ie, to reduce the risk of cardiovascular death.

DIABETES DRUG DEVELOPMENT

Changing priorities

The International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) was formed in 1990 as a collaborative effort across global regulatory agencies and coordinated by the World Health Organization to universalize criteria for drug development. The ICH standards for type 2 diabetes drug development included the following requirements for patient exposure to investigational products to satisfy new drug application requirements:

• 1,500 individuals total (including single-dose exposure)
• 300–600 patients for 6 months
• 100 patients for 1 year.

Thus, just 250 patient-years of exposure were needed for approval of a drug that patients might take for decades. These standards were unlikely to reveal rare, serious complications and had no ability to assess clinical outcomes efficacy for either microvascular or macrovascular disease complications.

Since 1995, when metformin was approved in the United States, a new class of antihyperglycemic medication has been approved about once every 2 years, so that by 2008, 12 classes of medications had become available for the treatment of type 2 diabetes. This extraordinary rate of drug development has now yielded more classes of medications to treat type 2 diabetes than we presently have for the treatment of hypertension.

This proliferation of new treatments resolved much of the pressure of the unmet medical need, over a period of increasing awareness of the cardiovascular complications of type 2 diabetes, along with numerous examples of adverse cardiovascular effects observed with some of the drugs. In this context, the FDA (and in parallel the European Medicines Agency) made paradigm-shifting changes in the requirements for the development of new type 2 diabetes drugs, requiring large-scale randomized clinical outcome data to assess cardiovascular safety of the new drugs. In December 2008, the FDA published a Guidance for Industry,¹ recommending that sponsors of new drugs for type 2 diabetes demonstrate that therapy would not only improve glucose control, but also that it would, at a minimum, not result in an unacceptable increase in cardiovascular risk.¹ To better assess new diabetes drugs, the requirement for patient-years of exposure to the studied drug was increased by over 60-fold from 250 patient-years to more than 15,000.

INCRETIN MODULATORS

The incretin system, a regulator of postprandial glucose metabolism, is an attractive target for glycemic control, as it promotes early satiety and lowers blood glucose.

After a meal, endocrine cells in the distal small intestine secrete the incretin hormones GLP-1 and gastric inhibitory polypeptide (GIP), among others, which reduce gastric motility, stimulate the pancreas to augment glucose-appropriate insulin secretion, and decrease postprandial glucagon release. GLP-1 also interacts with the satiety center of the hypothalamus, suppressing appetite. GLP-1 and GIP are rapidly inactivated by the circulating protease DPP-4. Injectable formulations of GLP-1 receptor agonists that are resistant to DPP-4 degradation have been developed.

Ten incretin modulators are now available in the United States. The 4 available DPP-4 inhibitors are all once-daily oral medications, and the 6 GLP-1 receptor agonists are all injectable (Table 1).

Small studies in humans and animals suggest that DPP-4 inhibitors and GLP-1-receptor agonists may have multiple favorable effects on the cardiovascular system independent of their glycemic effects. These include reducing myocardial infarct size,^2–5 improving endothelial function,⁶ reducing inflammation and oxidative stress,⁷ reducing atherosclerotic plaque volume,⁸ improving left ventricular function, ^9,10 and lowering triglyceride levels.¹¹ However, large clinical trials are needed to determine clinical effectiveness.

DPP-4 INHIBITORS: NOT INFERIOR TO PLACEBO 

Saxagliptin

Saxagliptin, a DPP-4 inhibitor, was found in a meta-analysis of phase 2B and early phase 3 trial data involving almost 5,000 patients to be associated with a dramatic 56% relative risk reduction in cardiovascular death, heart attack, and stroke. However, this analysis was limited by the extremely low number of events to analyze, with only 41 total patients with cardiovascular events in that dataset.¹²

The SAVOR-TIMI 53 trial¹³ subsequently compared saxagliptin and placebo in a randomized, double-blind trial conducted in 26 countries with nearly 16,500 patients with type 2 diabetes. All patients continued their conventional diabetes treatment at the discretion of their physicians.

During an average follow-up of 2 years, 1,222 events of cardiovascular death, myocardial infarction, or stroke occurred. No significant difference in event rates was found between the saxagliptin and placebo groups. This did not demonstrate the expected cardiovascular benefit based on prior meta-analysis of phase 2B and phase 3 data presented above, but saxagliptin did not increase cardiovascular risk and was the first diabetes drug to earn this distinction of robustly statistically proven cardiovascular safety.

Further analysis of the SAVOR-TIMI 53 trial data revealed a 27% increased relative risk of heart failure hospitalization with saxagliptin compared with placebo.¹⁴ Although the risk was statistically significant, the absolute difference in heart failure incidence between the drug and placebo groups was only 0.7% (3.5% vs 2.8%, respectively). As the average follow-up in the trial was 2 years, the absolute incremental risk of heart failure seen with saxagliptin is 0.35% annually—almost identical in magnitude to the increased heart failure risk with pioglitazone. The increased risk of heart failure was seen within the first 6 months of the trial and persisted throughout the trial, indicating an increased up-front risk of heart failure.

Alogliptin

The EXAMINE trial¹⁵ compared the DPP-4 inhibitor alogliptin and placebo in 5,380 patients with type 2 diabetes who had had a recent acute coronary event.¹⁵ Over the 30 months of the trial, more than 600 primary outcome events of cardiovascular death, myocardial infarction, or stroke occurred, with no significant difference between drug and placebo groups with established nominal statistical noninferiority. A numerically higher incidence of heart failure was noted in patients who received alogliptin than with placebo, but the difference was not statistically significant.¹⁶ However, this study was not powered to detect such an increased risk. In patients entering the trial with no history of heart failure, the risk of hospitalization for heart failure was 76% higher in the alogliptin group than in the placebo group, with a nominally significant P value less than .05 in this subgroup.

These analyses led the FDA in 2016 to mandate label warnings for saxagliptin and alogliptin regarding the increased risk of heart failure.¹⁷

Sitagliptin

The TECOS trial¹⁸ tested the DPP-4 inhibitor sitagliptin and, unlike the SAVOR or EXAMINE trials, included hospitalization for unstable angina in the composite end point. Nearly 15,000 patients with type 2 diabetes and established cardiovascular disease were enrolled, and almost 2,500 events occurred. No significant difference was found between the 2 groups.

In a series of analyses prospectively planned, sitagliptin was not associated with an increased risk of hospitalization for heart failure.¹⁹ But despite these robust analyses demonstrating no incremental heart failure risk with sitagliptin, in August 2017, the US product label for sitagliptin was modified to include a warning that other DPP-4 inhibitors have been associated with heart failure and to suggest caution. The label for linagliptin had the same FDA-required changes, with no data yet available from outcomes trials with linagliptin.

GLP-1 RECEPTOR AGONISTS

Lixisenatide: Noninferior to placebo

The ELIXA trial²⁰ assessed the cardiovascular safety of the GLP-1 receptor agonist lixisenatide in patients with type 2 diabetes who recently had an acute coronary event. The study enrolled 6,068 patients from 49 countries, and nearly 1,000 events (cardiovascular death, myocardial infarction, stroke, or unstable angina) occurred during the median 25 months of the study. Results showed lixisenatide did not increase or decrease cardiovascular events or adverse events when compared with placebo.

Liraglutide: Evidence of benefit

The LEADER trial²¹ randomized 9,340 patients with or at increased risk for cardiovascular disease to receive the injectable GLP-1 receptor agonist liraglutide or placebo. After a median of 3.8 years of follow-up, liraglutide use was associated with a statistically significant 13% relative reduction in major adverse cardiovascular events, mostly driven by a 22% reduction in cardiovascular death.

Semaglutide: Evidence of benefit

The SUSTAIN-6 trial²² found a statistically significant 26% relative risk reduction in cardiovascular outcomes comparing once-weekly semaglutide (an injectable GLP-1 receptor agonist) and placebo in 3,297 patients with type 2 diabetes and established cardiovascular disease, chronic kidney disease, or risk factors for cardiovascular disease. The significant reduction in the incidence of nonfatal stroke with semaglutide was the main driver of the observed benefit.

Taspoglutide: Development halted

Taspoglutide was a candidate GLP-1 receptor agonist that underwent clinical trials for cardiovascular outcomes planned to involve about 8,000 patients. The trials were stopped early and drug development was halted after about 600 patient-years of exposure because of antibody formation in about half of patients exposed to taspoglutide, with anaphylactoid reactions and anaphylaxis reported.²³

SGLT-2 INHIBITORS

The renal glomeruli filter about 180 g of glucose every day in normal adults; nearly all of it is reabsorbed by SGLT-2 in the proximal tubules, so that very little glucose is excreted in the urine.^24–26 The benign condition hereditary glucosuria occurs due to loss-of-function mutations in the gene for SGLT-2. Individuals with this condition rarely if ever develop type 2 diabetes or obesity, and this observation led pharmaceutical researchers to probe SGLT-2 as a therapeutic target.

Inhibitors of SGLT-2 block glucose reabsorption in the renal proximal tubules and lead to glucosuria. Patients treated with an SGLT-2 inhibitor have lower serum glucose levels and lose weight. Inhibitors also reduce sodium reabsorption via SGLT-2 and lead to increased sodium excretion and decreased blood pressure.²⁷

Three SGLT-2 antagonists are available in the United States: canagliflozin, dapagliflozin, and empagliflozin (Table 1). Ertugliflozin is currently in a phase 3B trial, and cardiovascular outcomes trials are in the planning phase for sotagliflozin, a dual SGLT-1/SGLT-2 inhibitor with SGLT-1 localized to the gastrointestinal tract.²⁸

Empaglifozin: Evidence of benefit

The EMPA-REG OUTCOME trial²⁹ randomized more than 7,200 patients with type 2 diabetes and atherosclerotic vascular disease to receive the SGLT-2 inhibitor empagliflozin or placebo as once-daily tablets, with both groups receiving off-study treatment for glycemic control at the discretion of their own care providers. Two doses of empagliflozin were evaluated in the trial (10 and 25 mg per day), with the 2 dosing groups pooled for all analyses as prospectively planned.

Patients taking empagliflozin had a 14% relative risk reduction of the composite outcome (cardiovascular death, myocardial infarction, and stroke) vs placebo, with no difference in effect between the 2 randomized doses. The improvement in the composite outcome was seen early in the empagliflozin group and persisted for the 4 years of the study.

This was the first trial of newly developed diabetes drugs that showed a statistically significant reduction in cardiovascular risk. The study revealed a 38% relative risk reduction in cardiovascular death in the treatment group. The risk reduction occurred early in the trial and improved throughout the duration of the study. This is a dramatic finding, unequaled even in trials of drugs that specifically target cardiovascular disease. Both doses of empagliflozin studied provided similar benefit over placebo, reinforcing the validity of the findings. Interestingly, in the empagliflozin group, there was a 35% relative risk reduction in heart failure hospitalizations.

Canaglifozin: Evidence of benefit

The CANVAS Program consisted of two sister trials, CANVAS and CANVAS-R, and examined the safety and efficacy of canagliflozin.³⁰ More than 10,000 participants with type 2 diabetes and atherosclerotic disease or at increased risk of cardiovascular disease were randomized to receive canagliflozin or placebo. Canagliflozin led to a 14% relative risk reduction in the composite outcome of cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke, but there was a statistically significant doubling in the incidence of amputations. Unlike empagliflozin, canagliflozin did not demonstrate a significant reduction in death from cardiovascular causes, suggesting that this may not be a class effect of SGLT-2 inhibitors. As with empagliflozin, canagliflozin led to a 33% relative risk reduction in heart failure hospitalizations.

Cardiovascular benefits independent of glucose-lowering

The cardiovascular benefits of empagliflozin in EMPA-REG OUTCOME and canagliflozin in CANVAS were observed early, suggesting that the mechanism may be due to the direct effects on the cardiovascular system rather than glycemic modification.

Improved glycemic control with the SGLT-2 inhibitor was seen early in both studies, but with the trials designed for glycemic equipoise encouraging open-label therapy targeting hemoglobin A_1c to standard-of-care targets in both groups, the contrast in hemoglobin A_1c between groups diminished throughout the trial after its first assessment. Although hemoglobin A_1c levels in the SGLT-2 inhibitor groups decreased in the first 12 weeks, they increased over time nearly to the level seen in the placebo group. The adjusted mean hemoglobin A_1c level in the placebo groups remained near 8.0% throughout the studies, a target consistent with guidelines from the American Diabetes Association and the European Association for the Study of Diabetes³¹ for the high-risk populations recruited and enrolled.

Blood pressure reduction and weight loss do not explain cardiovascular benefits

SGLT-2 inhibitors lower blood pressure independent of their diuretic effects. In the EMPA-REG OUTCOME trial, the adjusted mean systolic blood pressure was 3 to 4 mm Hg lower in the treatment groups than in the placebo group throughout the trial.²⁹ This level of blood pressure lowering translates to an estimated 10% to 12% relative risk reduction for major adverse cardiovascular events, including heart failure. Although the risk reduction from blood pressure lowering is not insignificant, it does not explain the 38% reduction in cardiovascular deaths seen in the trial. Canagliflozin led to a similar 4-mm Hg reduction in systolic pressure compared with the placebo group.³⁰

Weight loss was seen with both empagliflozin and canagliflozin but was not dramatic and is unlikely to account for the described cardiovascular benefits.

Theories of cardiovascular benefit

Several mechanisms have been proposed to help explain the observed cardiovascular benefits of SGLT-2 inhibitors.³²

Ketone-body elevation. Ferrannini et al³³ found that the blood concentration of the ketone-body beta-hydroxybutyrate is about twice as high in patients with type 2 diabetes in the fasting state who are chronically taking empagliflozin as in  patients not receiving the drug. Beta-hydroxybutyrate levels peak after a meal and then return to baseline over several hours before rising again during the fasting period. Although the ketone elevation is not nearly as extreme as in diabetic ketoacidosis (about a 1,000-fold increase), the observed increase may reduce myocardial oxygen demand, as beta-hydroxybutyrate is among the most efficient metabolic substrates for the myocardium.       

Red blood cell expansion. Perhaps a more likely explanation of the cardiovascular benefit seen with SGLT-2 inhibitor therapy is the increase in hemoglobin and hematocrit levels. At first attributed to hemoconcentration secondary to diuresis, this has been disproven by a number of studies. The EMPA-REG OUTCOME trial²⁹ found that within 12 weeks of exposure to empagliflozin, hematocrit levels rose nearly 4% absolutely compared with the levels in the placebo group. This increase is equivalent to transfusing a unit of red blood cells, favorably affecting myocardial oxygen supply.

Reduction in glomerular hypertension. The kidneys regulate glomerular filtration in a process involving the macula densa, an area of specialized cells in the juxtaglomerular apparatus in the loop of Henle that responds to sodium concentration in the urine. Normally, SGLT-2 receptors upstream from the loop of Henle reabsorb sodium and glucose into the bloodstream, reducing sodium delivery to the macula densa, which senses this as a low-volume state. The macula densa cells respond by releasing factors that dilate afferent arterioles and increase glomerular filtration. People with diabetes have more glucose to reabsorb and therefore also reabsorb more sodium, leading to glomerular hypertension.

SGLT-2 inhibitors block both glucose and sodium reuptake at SGLT-2 receptors, normalizing the response at the macula densa, restoring a normal glomerular filtration rate, and alleviating glomerular hypertension. As the kidney perceives a more normal volume status, renin-angiotensin-aldosterone stimulation is attenuated and sympathetic nervous system activity improves.^27,34 If this model of SGLT-2 inhibitor effects on the kidney is correct, these drugs have similar effects as angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), mineralocorticoid antagonists, and beta-blockers combined.

Empagliflozin³⁵ and canagliflozin³⁰ both reduced the rate of progression of kidney dysfunction and led to fewer clinically relevant renal events compared with placebo. Treatment and placebo groups also received standard care, so many patients were treated with renin-angiotensin-aldosterone system inhibitors and with good blood pressure control, making the finding that SGLT-2 inhibitors had a significant beneficial effect even more dramatic. Beneficial effects on markers of kidney function were seen early on, suggesting a more favorable hemodynamic effect on the kidney rather than improved glycemic control attenuating microvascular disease.

Empagliflozin approved to reduce clinical events

In December 2016, the FDA approved the indication for empagliflozin to reduce the risk of cardiovascular death in patients with type 2 diabetes,³⁶ the first-ever clinical outcome indication for a type 2 diabetes medication. The European Society of Cardiology guidelines now include empagliflozin as preferred therapy for type 2 diabetes, recommending it to prevent the onset of heart failure and prolong life.³⁷ This recommendation goes beyond the evidence from the EMPA-REG OUTCOME trial on which it is based, as the trial only studied patients with known atherosclerotic vascular disease.

The 2016 European Guidelines on cardiovascular disease prevention also recommend that an SGLT-2 inhibitor be considered early for patients with type 2 diabetes and cardiovascular disease to reduce cardiovascular and total mortality.³⁸ The American Diabetes Association in their 2017 guidelines also endorse empagliflozin for treating patients with type 2 diabetes and cardiovascular disease.³⁹ The fact that the American Diabetes Association recommendation is not based on glycemic control, in line with the product-labeled indication, is a major shift in the association’s guidance.

Cautions with SGLT-2 inhibitors

• Use SGLT-2 inhibitors in patients with low blood pressure with caution, and with increased blood pressure monitoring just following initiation.
• Consider modifying antihypertensive drugs in patients with labile blood pressure.
• Consider stopping or reducing background diuretics when starting an SGLT-2 inhibitor, and reassess volume status after 1 to 2 weeks.
• For patients on insulin, sulfonylureas, or both, consider decreasing dosages when starting an SGLT-2 inhibitor, and reassess glycemic control periodically.
• Counsel patients about urinary hygiene. Although bacterial urinary tract infections have not emerged as a problem, fungal genital infections have, particularly in women and uncircumcised men.
• Consider SGLT-2 inhibitors to be “sick-day” medications. Patients with diabetes must adjust their diabetes medications if their oral intake is reduced for a day or more, such as while sick or fasting. SGLT-2 inhibitors should not be taken on these days. Cases of diabetic ketoacidosis have arisen in patients who reduced oral intake while continuing their SGLT-2 inhibitor.

OTHER DRUGS WITH DEVELOPMENT HALTED

Aleglitazar, a peroxisome proliferator-activated receptor agonist taken orally once daily, raised high expectations when it was found in early studies to lower serum triglycerides and raise high-density lipoprotein cholesterol levels in addition to lowering blood glucose. However, a phase 3 trial in more than 7,000 patients was terminated after a median follow up of 2 years because of increased rates of heart failure, worsened kidney function, bone fractures, and gastrointestinal bleeding.⁴⁰ Development of this drug was stopped.

Fasiglifam, a G-protein-coupled receptor 40 agonist, was tested in a cardiovascular clinical outcomes trial. Compared with placebo, fasiglifam reduced hemoglobin A_1c  levels with low risk of hypoglycemia.⁴¹ However,  safety concerns about increased liver enzyme levels led to the cessation of the drug’s development.⁴²

HOW WILL THIS AFFECT DIABETES MANAGEMENT?

Metformin is still the most commonly prescribed drug for type 2 diabetes but has only marginal evidence for its cardiovascular benefits and may not be the first-line therapy for the management of diabetes in the future. In the EMPA REG OUTCOME, LEADER, and SUSTAIN-6 trials, the novel diabetes medications were given to patients who were already treated with available therapies, often including metformin. Treatment with empagliflozin, liraglutide, and semaglutide may be indicated for patients with diabetes and atherosclerotic vascular disease as first-line therapies in the future.

SGLT-2 inhibitor therapy can cost about $500 per month, and GLP-1 inhibitors are only slightly less expensive. The cost may be prohibitive for many patients. As more evidence, guidelines, and FDA criteria support the use of these novel diabetes drugs, third-party payers and pharmaceutical companies may be motivated to lower costs to help reach more patients who can benefit from these therapies.

Since 2008, the US Food and Drug Administration (FDA) has required new diabetes drugs to demonstrate cardiovascular safety, resulting in large and lengthy clinical trials. Under the new regulations, several dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and glucagon-like peptide-1 (GLP-1) receptor agonists have demonstrated cardiovascular safety, with some demonstrating superior cardiovascular efficacy. In 2016, the SGLT-2 inhibitor empagliflozin became the first (and as of this writing, the only) diabetes drug approved by the FDA for a clinical outcome indication, ie, to reduce the risk of cardiovascular death.

DIABETES DRUG DEVELOPMENT

Changing priorities

The International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) was formed in 1990 as a collaborative effort across global regulatory agencies and coordinated by the World Health Organization to universalize criteria for drug development. The ICH standards for type 2 diabetes drug development included the following requirements for patient exposure to investigational products to satisfy new drug application requirements:

• 1,500 individuals total (including single-dose exposure)
• 300–600 patients for 6 months
• 100 patients for 1 year.

Thus, just 250 patient-years of exposure were needed for approval of a drug that patients might take for decades. These standards were unlikely to reveal rare, serious complications and had no ability to assess clinical outcomes efficacy for either microvascular or macrovascular disease complications.

Since 1995, when metformin was approved in the United States, a new class of antihyperglycemic medication has been approved about once every 2 years, so that by 2008, 12 classes of medications had become available for the treatment of type 2 diabetes. This extraordinary rate of drug development has now yielded more classes of medications to treat type 2 diabetes than we presently have for the treatment of hypertension.

This proliferation of new treatments resolved much of the pressure of the unmet medical need, over a period of increasing awareness of the cardiovascular complications of type 2 diabetes, along with numerous examples of adverse cardiovascular effects observed with some of the drugs. In this context, the FDA (and in parallel the European Medicines Agency) made paradigm-shifting changes in the requirements for the development of new type 2 diabetes drugs, requiring large-scale randomized clinical outcome data to assess cardiovascular safety of the new drugs. In December 2008, the FDA published a Guidance for Industry,¹ recommending that sponsors of new drugs for type 2 diabetes demonstrate that therapy would not only improve glucose control, but also that it would, at a minimum, not result in an unacceptable increase in cardiovascular risk.¹ To better assess new diabetes drugs, the requirement for patient-years of exposure to the studied drug was increased by over 60-fold from 250 patient-years to more than 15,000.

INCRETIN MODULATORS

The incretin system, a regulator of postprandial glucose metabolism, is an attractive target for glycemic control, as it promotes early satiety and lowers blood glucose.

After a meal, endocrine cells in the distal small intestine secrete the incretin hormones GLP-1 and gastric inhibitory polypeptide (GIP), among others, which reduce gastric motility, stimulate the pancreas to augment glucose-appropriate insulin secretion, and decrease postprandial glucagon release. GLP-1 also interacts with the satiety center of the hypothalamus, suppressing appetite. GLP-1 and GIP are rapidly inactivated by the circulating protease DPP-4. Injectable formulations of GLP-1 receptor agonists that are resistant to DPP-4 degradation have been developed.

Ten incretin modulators are now available in the United States. The 4 available DPP-4 inhibitors are all once-daily oral medications, and the 6 GLP-1 receptor agonists are all injectable (Table 1).

Small studies in humans and animals suggest that DPP-4 inhibitors and GLP-1-receptor agonists may have multiple favorable effects on the cardiovascular system independent of their glycemic effects. These include reducing myocardial infarct size,^2–5 improving endothelial function,⁶ reducing inflammation and oxidative stress,⁷ reducing atherosclerotic plaque volume,⁸ improving left ventricular function, ^9,10 and lowering triglyceride levels.¹¹ However, large clinical trials are needed to determine clinical effectiveness.

DPP-4 INHIBITORS: NOT INFERIOR TO PLACEBO 

Saxagliptin

Saxagliptin, a DPP-4 inhibitor, was found in a meta-analysis of phase 2B and early phase 3 trial data involving almost 5,000 patients to be associated with a dramatic 56% relative risk reduction in cardiovascular death, heart attack, and stroke. However, this analysis was limited by the extremely low number of events to analyze, with only 41 total patients with cardiovascular events in that dataset.¹²

The SAVOR-TIMI 53 trial¹³ subsequently compared saxagliptin and placebo in a randomized, double-blind trial conducted in 26 countries with nearly 16,500 patients with type 2 diabetes. All patients continued their conventional diabetes treatment at the discretion of their physicians.

During an average follow-up of 2 years, 1,222 events of cardiovascular death, myocardial infarction, or stroke occurred. No significant difference in event rates was found between the saxagliptin and placebo groups. This did not demonstrate the expected cardiovascular benefit based on prior meta-analysis of phase 2B and phase 3 data presented above, but saxagliptin did not increase cardiovascular risk and was the first diabetes drug to earn this distinction of robustly statistically proven cardiovascular safety.

Further analysis of the SAVOR-TIMI 53 trial data revealed a 27% increased relative risk of heart failure hospitalization with saxagliptin compared with placebo.¹⁴ Although the risk was statistically significant, the absolute difference in heart failure incidence between the drug and placebo groups was only 0.7% (3.5% vs 2.8%, respectively). As the average follow-up in the trial was 2 years, the absolute incremental risk of heart failure seen with saxagliptin is 0.35% annually—almost identical in magnitude to the increased heart failure risk with pioglitazone. The increased risk of heart failure was seen within the first 6 months of the trial and persisted throughout the trial, indicating an increased up-front risk of heart failure.

Alogliptin

The EXAMINE trial¹⁵ compared the DPP-4 inhibitor alogliptin and placebo in 5,380 patients with type 2 diabetes who had had a recent acute coronary event.¹⁵ Over the 30 months of the trial, more than 600 primary outcome events of cardiovascular death, myocardial infarction, or stroke occurred, with no significant difference between drug and placebo groups with established nominal statistical noninferiority. A numerically higher incidence of heart failure was noted in patients who received alogliptin than with placebo, but the difference was not statistically significant.¹⁶ However, this study was not powered to detect such an increased risk. In patients entering the trial with no history of heart failure, the risk of hospitalization for heart failure was 76% higher in the alogliptin group than in the placebo group, with a nominally significant P value less than .05 in this subgroup.

These analyses led the FDA in 2016 to mandate label warnings for saxagliptin and alogliptin regarding the increased risk of heart failure.¹⁷

Sitagliptin

The TECOS trial¹⁸ tested the DPP-4 inhibitor sitagliptin and, unlike the SAVOR or EXAMINE trials, included hospitalization for unstable angina in the composite end point. Nearly 15,000 patients with type 2 diabetes and established cardiovascular disease were enrolled, and almost 2,500 events occurred. No significant difference was found between the 2 groups.

In a series of analyses prospectively planned, sitagliptin was not associated with an increased risk of hospitalization for heart failure.¹⁹ But despite these robust analyses demonstrating no incremental heart failure risk with sitagliptin, in August 2017, the US product label for sitagliptin was modified to include a warning that other DPP-4 inhibitors have been associated with heart failure and to suggest caution. The label for linagliptin had the same FDA-required changes, with no data yet available from outcomes trials with linagliptin.

GLP-1 RECEPTOR AGONISTS

Lixisenatide: Noninferior to placebo

The ELIXA trial²⁰ assessed the cardiovascular safety of the GLP-1 receptor agonist lixisenatide in patients with type 2 diabetes who recently had an acute coronary event. The study enrolled 6,068 patients from 49 countries, and nearly 1,000 events (cardiovascular death, myocardial infarction, stroke, or unstable angina) occurred during the median 25 months of the study. Results showed lixisenatide did not increase or decrease cardiovascular events or adverse events when compared with placebo.

Liraglutide: Evidence of benefit

The LEADER trial²¹ randomized 9,340 patients with or at increased risk for cardiovascular disease to receive the injectable GLP-1 receptor agonist liraglutide or placebo. After a median of 3.8 years of follow-up, liraglutide use was associated with a statistically significant 13% relative reduction in major adverse cardiovascular events, mostly driven by a 22% reduction in cardiovascular death.

Semaglutide: Evidence of benefit

The SUSTAIN-6 trial²² found a statistically significant 26% relative risk reduction in cardiovascular outcomes comparing once-weekly semaglutide (an injectable GLP-1 receptor agonist) and placebo in 3,297 patients with type 2 diabetes and established cardiovascular disease, chronic kidney disease, or risk factors for cardiovascular disease. The significant reduction in the incidence of nonfatal stroke with semaglutide was the main driver of the observed benefit.

Taspoglutide: Development halted

Taspoglutide was a candidate GLP-1 receptor agonist that underwent clinical trials for cardiovascular outcomes planned to involve about 8,000 patients. The trials were stopped early and drug development was halted after about 600 patient-years of exposure because of antibody formation in about half of patients exposed to taspoglutide, with anaphylactoid reactions and anaphylaxis reported.²³

SGLT-2 INHIBITORS

The renal glomeruli filter about 180 g of glucose every day in normal adults; nearly all of it is reabsorbed by SGLT-2 in the proximal tubules, so that very little glucose is excreted in the urine.^24–26 The benign condition hereditary glucosuria occurs due to loss-of-function mutations in the gene for SGLT-2. Individuals with this condition rarely if ever develop type 2 diabetes or obesity, and this observation led pharmaceutical researchers to probe SGLT-2 as a therapeutic target.

Inhibitors of SGLT-2 block glucose reabsorption in the renal proximal tubules and lead to glucosuria. Patients treated with an SGLT-2 inhibitor have lower serum glucose levels and lose weight. Inhibitors also reduce sodium reabsorption via SGLT-2 and lead to increased sodium excretion and decreased blood pressure.²⁷

Three SGLT-2 antagonists are available in the United States: canagliflozin, dapagliflozin, and empagliflozin (Table 1). Ertugliflozin is currently in a phase 3B trial, and cardiovascular outcomes trials are in the planning phase for sotagliflozin, a dual SGLT-1/SGLT-2 inhibitor with SGLT-1 localized to the gastrointestinal tract.²⁸

Empaglifozin: Evidence of benefit

The EMPA-REG OUTCOME trial²⁹ randomized more than 7,200 patients with type 2 diabetes and atherosclerotic vascular disease to receive the SGLT-2 inhibitor empagliflozin or placebo as once-daily tablets, with both groups receiving off-study treatment for glycemic control at the discretion of their own care providers. Two doses of empagliflozin were evaluated in the trial (10 and 25 mg per day), with the 2 dosing groups pooled for all analyses as prospectively planned.

Patients taking empagliflozin had a 14% relative risk reduction of the composite outcome (cardiovascular death, myocardial infarction, and stroke) vs placebo, with no difference in effect between the 2 randomized doses. The improvement in the composite outcome was seen early in the empagliflozin group and persisted for the 4 years of the study.

This was the first trial of newly developed diabetes drugs that showed a statistically significant reduction in cardiovascular risk. The study revealed a 38% relative risk reduction in cardiovascular death in the treatment group. The risk reduction occurred early in the trial and improved throughout the duration of the study. This is a dramatic finding, unequaled even in trials of drugs that specifically target cardiovascular disease. Both doses of empagliflozin studied provided similar benefit over placebo, reinforcing the validity of the findings. Interestingly, in the empagliflozin group, there was a 35% relative risk reduction in heart failure hospitalizations.

Canaglifozin: Evidence of benefit

The CANVAS Program consisted of two sister trials, CANVAS and CANVAS-R, and examined the safety and efficacy of canagliflozin.³⁰ More than 10,000 participants with type 2 diabetes and atherosclerotic disease or at increased risk of cardiovascular disease were randomized to receive canagliflozin or placebo. Canagliflozin led to a 14% relative risk reduction in the composite outcome of cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke, but there was a statistically significant doubling in the incidence of amputations. Unlike empagliflozin, canagliflozin did not demonstrate a significant reduction in death from cardiovascular causes, suggesting that this may not be a class effect of SGLT-2 inhibitors. As with empagliflozin, canagliflozin led to a 33% relative risk reduction in heart failure hospitalizations.

Cardiovascular benefits independent of glucose-lowering

The cardiovascular benefits of empagliflozin in EMPA-REG OUTCOME and canagliflozin in CANVAS were observed early, suggesting that the mechanism may be due to the direct effects on the cardiovascular system rather than glycemic modification.

Improved glycemic control with the SGLT-2 inhibitor was seen early in both studies, but with the trials designed for glycemic equipoise encouraging open-label therapy targeting hemoglobin A_1c to standard-of-care targets in both groups, the contrast in hemoglobin A_1c between groups diminished throughout the trial after its first assessment. Although hemoglobin A_1c levels in the SGLT-2 inhibitor groups decreased in the first 12 weeks, they increased over time nearly to the level seen in the placebo group. The adjusted mean hemoglobin A_1c level in the placebo groups remained near 8.0% throughout the studies, a target consistent with guidelines from the American Diabetes Association and the European Association for the Study of Diabetes³¹ for the high-risk populations recruited and enrolled.

Blood pressure reduction and weight loss do not explain cardiovascular benefits

SGLT-2 inhibitors lower blood pressure independent of their diuretic effects. In the EMPA-REG OUTCOME trial, the adjusted mean systolic blood pressure was 3 to 4 mm Hg lower in the treatment groups than in the placebo group throughout the trial.²⁹ This level of blood pressure lowering translates to an estimated 10% to 12% relative risk reduction for major adverse cardiovascular events, including heart failure. Although the risk reduction from blood pressure lowering is not insignificant, it does not explain the 38% reduction in cardiovascular deaths seen in the trial. Canagliflozin led to a similar 4-mm Hg reduction in systolic pressure compared with the placebo group.³⁰

Weight loss was seen with both empagliflozin and canagliflozin but was not dramatic and is unlikely to account for the described cardiovascular benefits.

Theories of cardiovascular benefit

Several mechanisms have been proposed to help explain the observed cardiovascular benefits of SGLT-2 inhibitors.³²

Ketone-body elevation. Ferrannini et al³³ found that the blood concentration of the ketone-body beta-hydroxybutyrate is about twice as high in patients with type 2 diabetes in the fasting state who are chronically taking empagliflozin as in  patients not receiving the drug. Beta-hydroxybutyrate levels peak after a meal and then return to baseline over several hours before rising again during the fasting period. Although the ketone elevation is not nearly as extreme as in diabetic ketoacidosis (about a 1,000-fold increase), the observed increase may reduce myocardial oxygen demand, as beta-hydroxybutyrate is among the most efficient metabolic substrates for the myocardium.       

Red blood cell expansion. Perhaps a more likely explanation of the cardiovascular benefit seen with SGLT-2 inhibitor therapy is the increase in hemoglobin and hematocrit levels. At first attributed to hemoconcentration secondary to diuresis, this has been disproven by a number of studies. The EMPA-REG OUTCOME trial²⁹ found that within 12 weeks of exposure to empagliflozin, hematocrit levels rose nearly 4% absolutely compared with the levels in the placebo group. This increase is equivalent to transfusing a unit of red blood cells, favorably affecting myocardial oxygen supply.

Reduction in glomerular hypertension. The kidneys regulate glomerular filtration in a process involving the macula densa, an area of specialized cells in the juxtaglomerular apparatus in the loop of Henle that responds to sodium concentration in the urine. Normally, SGLT-2 receptors upstream from the loop of Henle reabsorb sodium and glucose into the bloodstream, reducing sodium delivery to the macula densa, which senses this as a low-volume state. The macula densa cells respond by releasing factors that dilate afferent arterioles and increase glomerular filtration. People with diabetes have more glucose to reabsorb and therefore also reabsorb more sodium, leading to glomerular hypertension.

SGLT-2 inhibitors block both glucose and sodium reuptake at SGLT-2 receptors, normalizing the response at the macula densa, restoring a normal glomerular filtration rate, and alleviating glomerular hypertension. As the kidney perceives a more normal volume status, renin-angiotensin-aldosterone stimulation is attenuated and sympathetic nervous system activity improves.^27,34 If this model of SGLT-2 inhibitor effects on the kidney is correct, these drugs have similar effects as angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), mineralocorticoid antagonists, and beta-blockers combined.

Empagliflozin³⁵ and canagliflozin³⁰ both reduced the rate of progression of kidney dysfunction and led to fewer clinically relevant renal events compared with placebo. Treatment and placebo groups also received standard care, so many patients were treated with renin-angiotensin-aldosterone system inhibitors and with good blood pressure control, making the finding that SGLT-2 inhibitors had a significant beneficial effect even more dramatic. Beneficial effects on markers of kidney function were seen early on, suggesting a more favorable hemodynamic effect on the kidney rather than improved glycemic control attenuating microvascular disease.

Empagliflozin approved to reduce clinical events

In December 2016, the FDA approved the indication for empagliflozin to reduce the risk of cardiovascular death in patients with type 2 diabetes,³⁶ the first-ever clinical outcome indication for a type 2 diabetes medication. The European Society of Cardiology guidelines now include empagliflozin as preferred therapy for type 2 diabetes, recommending it to prevent the onset of heart failure and prolong life.³⁷ This recommendation goes beyond the evidence from the EMPA-REG OUTCOME trial on which it is based, as the trial only studied patients with known atherosclerotic vascular disease.

The 2016 European Guidelines on cardiovascular disease prevention also recommend that an SGLT-2 inhibitor be considered early for patients with type 2 diabetes and cardiovascular disease to reduce cardiovascular and total mortality.³⁸ The American Diabetes Association in their 2017 guidelines also endorse empagliflozin for treating patients with type 2 diabetes and cardiovascular disease.³⁹ The fact that the American Diabetes Association recommendation is not based on glycemic control, in line with the product-labeled indication, is a major shift in the association’s guidance.

Cautions with SGLT-2 inhibitors

• Use SGLT-2 inhibitors in patients with low blood pressure with caution, and with increased blood pressure monitoring just following initiation.
• Consider modifying antihypertensive drugs in patients with labile blood pressure.
• Consider stopping or reducing background diuretics when starting an SGLT-2 inhibitor, and reassess volume status after 1 to 2 weeks.
• For patients on insulin, sulfonylureas, or both, consider decreasing dosages when starting an SGLT-2 inhibitor, and reassess glycemic control periodically.
• Counsel patients about urinary hygiene. Although bacterial urinary tract infections have not emerged as a problem, fungal genital infections have, particularly in women and uncircumcised men.
• Consider SGLT-2 inhibitors to be “sick-day” medications. Patients with diabetes must adjust their diabetes medications if their oral intake is reduced for a day or more, such as while sick or fasting. SGLT-2 inhibitors should not be taken on these days. Cases of diabetic ketoacidosis have arisen in patients who reduced oral intake while continuing their SGLT-2 inhibitor.

OTHER DRUGS WITH DEVELOPMENT HALTED

Aleglitazar, a peroxisome proliferator-activated receptor agonist taken orally once daily, raised high expectations when it was found in early studies to lower serum triglycerides and raise high-density lipoprotein cholesterol levels in addition to lowering blood glucose. However, a phase 3 trial in more than 7,000 patients was terminated after a median follow up of 2 years because of increased rates of heart failure, worsened kidney function, bone fractures, and gastrointestinal bleeding.⁴⁰ Development of this drug was stopped.

Fasiglifam, a G-protein-coupled receptor 40 agonist, was tested in a cardiovascular clinical outcomes trial. Compared with placebo, fasiglifam reduced hemoglobin A_1c  levels with low risk of hypoglycemia.⁴¹ However,  safety concerns about increased liver enzyme levels led to the cessation of the drug’s development.⁴²

HOW WILL THIS AFFECT DIABETES MANAGEMENT?

Metformin is still the most commonly prescribed drug for type 2 diabetes but has only marginal evidence for its cardiovascular benefits and may not be the first-line therapy for the management of diabetes in the future. In the EMPA REG OUTCOME, LEADER, and SUSTAIN-6 trials, the novel diabetes medications were given to patients who were already treated with available therapies, often including metformin. Treatment with empagliflozin, liraglutide, and semaglutide may be indicated for patients with diabetes and atherosclerotic vascular disease as first-line therapies in the future.

SGLT-2 inhibitor therapy can cost about $500 per month, and GLP-1 inhibitors are only slightly less expensive. The cost may be prohibitive for many patients. As more evidence, guidelines, and FDA criteria support the use of these novel diabetes drugs, third-party payers and pharmaceutical companies may be motivated to lower costs to help reach more patients who can benefit from these therapies.

Since 2008, the US Food and Drug Administration (FDA) has required new diabetes drugs to demonstrate cardiovascular safety, resulting in large and lengthy clinical trials. Under the new regulations, several dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and glucagon-like peptide-1 (GLP-1) receptor agonists have demonstrated cardiovascular safety, with some demonstrating superior cardiovascular efficacy. In 2016, the SGLT-2 inhibitor empagliflozin became the first (and as of this writing, the only) diabetes drug approved by the FDA for a clinical outcome indication, ie, to reduce the risk of cardiovascular death.

DIABETES DRUG DEVELOPMENT

Changing priorities

The International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) was formed in 1990 as a collaborative effort across global regulatory agencies and coordinated by the World Health Organization to universalize criteria for drug development. The ICH standards for type 2 diabetes drug development included the following requirements for patient exposure to investigational products to satisfy new drug application requirements:

• 1,500 individuals total (including single-dose exposure)
• 300–600 patients for 6 months
• 100 patients for 1 year.

Thus, just 250 patient-years of exposure were needed for approval of a drug that patients might take for decades. These standards were unlikely to reveal rare, serious complications and had no ability to assess clinical outcomes efficacy for either microvascular or macrovascular disease complications.

Since 1995, when metformin was approved in the United States, a new class of antihyperglycemic medication has been approved about once every 2 years, so that by 2008, 12 classes of medications had become available for the treatment of type 2 diabetes. This extraordinary rate of drug development has now yielded more classes of medications to treat type 2 diabetes than we presently have for the treatment of hypertension.

This proliferation of new treatments resolved much of the pressure of the unmet medical need, over a period of increasing awareness of the cardiovascular complications of type 2 diabetes, along with numerous examples of adverse cardiovascular effects observed with some of the drugs. In this context, the FDA (and in parallel the European Medicines Agency) made paradigm-shifting changes in the requirements for the development of new type 2 diabetes drugs, requiring large-scale randomized clinical outcome data to assess cardiovascular safety of the new drugs. In December 2008, the FDA published a Guidance for Industry,¹ recommending that sponsors of new drugs for type 2 diabetes demonstrate that therapy would not only improve glucose control, but also that it would, at a minimum, not result in an unacceptable increase in cardiovascular risk.¹ To better assess new diabetes drugs, the requirement for patient-years of exposure to the studied drug was increased by over 60-fold from 250 patient-years to more than 15,000.

INCRETIN MODULATORS

The incretin system, a regulator of postprandial glucose metabolism, is an attractive target for glycemic control, as it promotes early satiety and lowers blood glucose.

After a meal, endocrine cells in the distal small intestine secrete the incretin hormones GLP-1 and gastric inhibitory polypeptide (GIP), among others, which reduce gastric motility, stimulate the pancreas to augment glucose-appropriate insulin secretion, and decrease postprandial glucagon release. GLP-1 also interacts with the satiety center of the hypothalamus, suppressing appetite. GLP-1 and GIP are rapidly inactivated by the circulating protease DPP-4. Injectable formulations of GLP-1 receptor agonists that are resistant to DPP-4 degradation have been developed.

Ten incretin modulators are now available in the United States. The 4 available DPP-4 inhibitors are all once-daily oral medications, and the 6 GLP-1 receptor agonists are all injectable (Table 1).

Small studies in humans and animals suggest that DPP-4 inhibitors and GLP-1-receptor agonists may have multiple favorable effects on the cardiovascular system independent of their glycemic effects. These include reducing myocardial infarct size,^2–5 improving endothelial function,⁶ reducing inflammation and oxidative stress,⁷ reducing atherosclerotic plaque volume,⁸ improving left ventricular function, ^9,10 and lowering triglyceride levels.¹¹ However, large clinical trials are needed to determine clinical effectiveness.

DPP-4 INHIBITORS: NOT INFERIOR TO PLACEBO 

Saxagliptin

Saxagliptin, a DPP-4 inhibitor, was found in a meta-analysis of phase 2B and early phase 3 trial data involving almost 5,000 patients to be associated with a dramatic 56% relative risk reduction in cardiovascular death, heart attack, and stroke. However, this analysis was limited by the extremely low number of events to analyze, with only 41 total patients with cardiovascular events in that dataset.¹²

The SAVOR-TIMI 53 trial¹³ subsequently compared saxagliptin and placebo in a randomized, double-blind trial conducted in 26 countries with nearly 16,500 patients with type 2 diabetes. All patients continued their conventional diabetes treatment at the discretion of their physicians.

During an average follow-up of 2 years, 1,222 events of cardiovascular death, myocardial infarction, or stroke occurred. No significant difference in event rates was found between the saxagliptin and placebo groups. This did not demonstrate the expected cardiovascular benefit based on prior meta-analysis of phase 2B and phase 3 data presented above, but saxagliptin did not increase cardiovascular risk and was the first diabetes drug to earn this distinction of robustly statistically proven cardiovascular safety.

Further analysis of the SAVOR-TIMI 53 trial data revealed a 27% increased relative risk of heart failure hospitalization with saxagliptin compared with placebo.¹⁴ Although the risk was statistically significant, the absolute difference in heart failure incidence between the drug and placebo groups was only 0.7% (3.5% vs 2.8%, respectively). As the average follow-up in the trial was 2 years, the absolute incremental risk of heart failure seen with saxagliptin is 0.35% annually—almost identical in magnitude to the increased heart failure risk with pioglitazone. The increased risk of heart failure was seen within the first 6 months of the trial and persisted throughout the trial, indicating an increased up-front risk of heart failure.

Alogliptin

The EXAMINE trial¹⁵ compared the DPP-4 inhibitor alogliptin and placebo in 5,380 patients with type 2 diabetes who had had a recent acute coronary event.¹⁵ Over the 30 months of the trial, more than 600 primary outcome events of cardiovascular death, myocardial infarction, or stroke occurred, with no significant difference between drug and placebo groups with established nominal statistical noninferiority. A numerically higher incidence of heart failure was noted in patients who received alogliptin than with placebo, but the difference was not statistically significant.¹⁶ However, this study was not powered to detect such an increased risk. In patients entering the trial with no history of heart failure, the risk of hospitalization for heart failure was 76% higher in the alogliptin group than in the placebo group, with a nominally significant P value less than .05 in this subgroup.

These analyses led the FDA in 2016 to mandate label warnings for saxagliptin and alogliptin regarding the increased risk of heart failure.¹⁷

Sitagliptin

The TECOS trial¹⁸ tested the DPP-4 inhibitor sitagliptin and, unlike the SAVOR or EXAMINE trials, included hospitalization for unstable angina in the composite end point. Nearly 15,000 patients with type 2 diabetes and established cardiovascular disease were enrolled, and almost 2,500 events occurred. No significant difference was found between the 2 groups.

In a series of analyses prospectively planned, sitagliptin was not associated with an increased risk of hospitalization for heart failure.¹⁹ But despite these robust analyses demonstrating no incremental heart failure risk with sitagliptin, in August 2017, the US product label for sitagliptin was modified to include a warning that other DPP-4 inhibitors have been associated with heart failure and to suggest caution. The label for linagliptin had the same FDA-required changes, with no data yet available from outcomes trials with linagliptin.

GLP-1 RECEPTOR AGONISTS

Lixisenatide: Noninferior to placebo

The ELIXA trial²⁰ assessed the cardiovascular safety of the GLP-1 receptor agonist lixisenatide in patients with type 2 diabetes who recently had an acute coronary event. The study enrolled 6,068 patients from 49 countries, and nearly 1,000 events (cardiovascular death, myocardial infarction, stroke, or unstable angina) occurred during the median 25 months of the study. Results showed lixisenatide did not increase or decrease cardiovascular events or adverse events when compared with placebo.

Liraglutide: Evidence of benefit

The LEADER trial²¹ randomized 9,340 patients with or at increased risk for cardiovascular disease to receive the injectable GLP-1 receptor agonist liraglutide or placebo. After a median of 3.8 years of follow-up, liraglutide use was associated with a statistically significant 13% relative reduction in major adverse cardiovascular events, mostly driven by a 22% reduction in cardiovascular death.

Semaglutide: Evidence of benefit

The SUSTAIN-6 trial²² found a statistically significant 26% relative risk reduction in cardiovascular outcomes comparing once-weekly semaglutide (an injectable GLP-1 receptor agonist) and placebo in 3,297 patients with type 2 diabetes and established cardiovascular disease, chronic kidney disease, or risk factors for cardiovascular disease. The significant reduction in the incidence of nonfatal stroke with semaglutide was the main driver of the observed benefit.

Taspoglutide: Development halted

Taspoglutide was a candidate GLP-1 receptor agonist that underwent clinical trials for cardiovascular outcomes planned to involve about 8,000 patients. The trials were stopped early and drug development was halted after about 600 patient-years of exposure because of antibody formation in about half of patients exposed to taspoglutide, with anaphylactoid reactions and anaphylaxis reported.²³

SGLT-2 INHIBITORS

The renal glomeruli filter about 180 g of glucose every day in normal adults; nearly all of it is reabsorbed by SGLT-2 in the proximal tubules, so that very little glucose is excreted in the urine.^24–26 The benign condition hereditary glucosuria occurs due to loss-of-function mutations in the gene for SGLT-2. Individuals with this condition rarely if ever develop type 2 diabetes or obesity, and this observation led pharmaceutical researchers to probe SGLT-2 as a therapeutic target.

Inhibitors of SGLT-2 block glucose reabsorption in the renal proximal tubules and lead to glucosuria. Patients treated with an SGLT-2 inhibitor have lower serum glucose levels and lose weight. Inhibitors also reduce sodium reabsorption via SGLT-2 and lead to increased sodium excretion and decreased blood pressure.²⁷

Three SGLT-2 antagonists are available in the United States: canagliflozin, dapagliflozin, and empagliflozin (Table 1). Ertugliflozin is currently in a phase 3B trial, and cardiovascular outcomes trials are in the planning phase for sotagliflozin, a dual SGLT-1/SGLT-2 inhibitor with SGLT-1 localized to the gastrointestinal tract.²⁸

Empaglifozin: Evidence of benefit

The EMPA-REG OUTCOME trial²⁹ randomized more than 7,200 patients with type 2 diabetes and atherosclerotic vascular disease to receive the SGLT-2 inhibitor empagliflozin or placebo as once-daily tablets, with both groups receiving off-study treatment for glycemic control at the discretion of their own care providers. Two doses of empagliflozin were evaluated in the trial (10 and 25 mg per day), with the 2 dosing groups pooled for all analyses as prospectively planned.

Patients taking empagliflozin had a 14% relative risk reduction of the composite outcome (cardiovascular death, myocardial infarction, and stroke) vs placebo, with no difference in effect between the 2 randomized doses. The improvement in the composite outcome was seen early in the empagliflozin group and persisted for the 4 years of the study.

This was the first trial of newly developed diabetes drugs that showed a statistically significant reduction in cardiovascular risk. The study revealed a 38% relative risk reduction in cardiovascular death in the treatment group. The risk reduction occurred early in the trial and improved throughout the duration of the study. This is a dramatic finding, unequaled even in trials of drugs that specifically target cardiovascular disease. Both doses of empagliflozin studied provided similar benefit over placebo, reinforcing the validity of the findings. Interestingly, in the empagliflozin group, there was a 35% relative risk reduction in heart failure hospitalizations.

Canaglifozin: Evidence of benefit

The CANVAS Program consisted of two sister trials, CANVAS and CANVAS-R, and examined the safety and efficacy of canagliflozin.³⁰ More than 10,000 participants with type 2 diabetes and atherosclerotic disease or at increased risk of cardiovascular disease were randomized to receive canagliflozin or placebo. Canagliflozin led to a 14% relative risk reduction in the composite outcome of cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke, but there was a statistically significant doubling in the incidence of amputations. Unlike empagliflozin, canagliflozin did not demonstrate a significant reduction in death from cardiovascular causes, suggesting that this may not be a class effect of SGLT-2 inhibitors. As with empagliflozin, canagliflozin led to a 33% relative risk reduction in heart failure hospitalizations.

Cardiovascular benefits independent of glucose-lowering

The cardiovascular benefits of empagliflozin in EMPA-REG OUTCOME and canagliflozin in CANVAS were observed early, suggesting that the mechanism may be due to the direct effects on the cardiovascular system rather than glycemic modification.

Improved glycemic control with the SGLT-2 inhibitor was seen early in both studies, but with the trials designed for glycemic equipoise encouraging open-label therapy targeting hemoglobin A_1c to standard-of-care targets in both groups, the contrast in hemoglobin A_1c between groups diminished throughout the trial after its first assessment. Although hemoglobin A_1c levels in the SGLT-2 inhibitor groups decreased in the first 12 weeks, they increased over time nearly to the level seen in the placebo group. The adjusted mean hemoglobin A_1c level in the placebo groups remained near 8.0% throughout the studies, a target consistent with guidelines from the American Diabetes Association and the European Association for the Study of Diabetes³¹ for the high-risk populations recruited and enrolled.

Blood pressure reduction and weight loss do not explain cardiovascular benefits

SGLT-2 inhibitors lower blood pressure independent of their diuretic effects. In the EMPA-REG OUTCOME trial, the adjusted mean systolic blood pressure was 3 to 4 mm Hg lower in the treatment groups than in the placebo group throughout the trial.²⁹ This level of blood pressure lowering translates to an estimated 10% to 12% relative risk reduction for major adverse cardiovascular events, including heart failure. Although the risk reduction from blood pressure lowering is not insignificant, it does not explain the 38% reduction in cardiovascular deaths seen in the trial. Canagliflozin led to a similar 4-mm Hg reduction in systolic pressure compared with the placebo group.³⁰

Weight loss was seen with both empagliflozin and canagliflozin but was not dramatic and is unlikely to account for the described cardiovascular benefits.

Theories of cardiovascular benefit

Several mechanisms have been proposed to help explain the observed cardiovascular benefits of SGLT-2 inhibitors.³²

Ketone-body elevation. Ferrannini et al³³ found that the blood concentration of the ketone-body beta-hydroxybutyrate is about twice as high in patients with type 2 diabetes in the fasting state who are chronically taking empagliflozin as in  patients not receiving the drug. Beta-hydroxybutyrate levels peak after a meal and then return to baseline over several hours before rising again during the fasting period. Although the ketone elevation is not nearly as extreme as in diabetic ketoacidosis (about a 1,000-fold increase), the observed increase may reduce myocardial oxygen demand, as beta-hydroxybutyrate is among the most efficient metabolic substrates for the myocardium.       

Red blood cell expansion. Perhaps a more likely explanation of the cardiovascular benefit seen with SGLT-2 inhibitor therapy is the increase in hemoglobin and hematocrit levels. At first attributed to hemoconcentration secondary to diuresis, this has been disproven by a number of studies. The EMPA-REG OUTCOME trial²⁹ found that within 12 weeks of exposure to empagliflozin, hematocrit levels rose nearly 4% absolutely compared with the levels in the placebo group. This increase is equivalent to transfusing a unit of red blood cells, favorably affecting myocardial oxygen supply.

Reduction in glomerular hypertension. The kidneys regulate glomerular filtration in a process involving the macula densa, an area of specialized cells in the juxtaglomerular apparatus in the loop of Henle that responds to sodium concentration in the urine. Normally, SGLT-2 receptors upstream from the loop of Henle reabsorb sodium and glucose into the bloodstream, reducing sodium delivery to the macula densa, which senses this as a low-volume state. The macula densa cells respond by releasing factors that dilate afferent arterioles and increase glomerular filtration. People with diabetes have more glucose to reabsorb and therefore also reabsorb more sodium, leading to glomerular hypertension.

SGLT-2 inhibitors block both glucose and sodium reuptake at SGLT-2 receptors, normalizing the response at the macula densa, restoring a normal glomerular filtration rate, and alleviating glomerular hypertension. As the kidney perceives a more normal volume status, renin-angiotensin-aldosterone stimulation is attenuated and sympathetic nervous system activity improves.^27,34 If this model of SGLT-2 inhibitor effects on the kidney is correct, these drugs have similar effects as angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), mineralocorticoid antagonists, and beta-blockers combined.

Empagliflozin³⁵ and canagliflozin³⁰ both reduced the rate of progression of kidney dysfunction and led to fewer clinically relevant renal events compared with placebo. Treatment and placebo groups also received standard care, so many patients were treated with renin-angiotensin-aldosterone system inhibitors and with good blood pressure control, making the finding that SGLT-2 inhibitors had a significant beneficial effect even more dramatic. Beneficial effects on markers of kidney function were seen early on, suggesting a more favorable hemodynamic effect on the kidney rather than improved glycemic control attenuating microvascular disease.

Empagliflozin approved to reduce clinical events

In December 2016, the FDA approved the indication for empagliflozin to reduce the risk of cardiovascular death in patients with type 2 diabetes,³⁶ the first-ever clinical outcome indication for a type 2 diabetes medication. The European Society of Cardiology guidelines now include empagliflozin as preferred therapy for type 2 diabetes, recommending it to prevent the onset of heart failure and prolong life.³⁷ This recommendation goes beyond the evidence from the EMPA-REG OUTCOME trial on which it is based, as the trial only studied patients with known atherosclerotic vascular disease.

The 2016 European Guidelines on cardiovascular disease prevention also recommend that an SGLT-2 inhibitor be considered early for patients with type 2 diabetes and cardiovascular disease to reduce cardiovascular and total mortality.³⁸ The American Diabetes Association in their 2017 guidelines also endorse empagliflozin for treating patients with type 2 diabetes and cardiovascular disease.³⁹ The fact that the American Diabetes Association recommendation is not based on glycemic control, in line with the product-labeled indication, is a major shift in the association’s guidance.

Cautions with SGLT-2 inhibitors

• Use SGLT-2 inhibitors in patients with low blood pressure with caution, and with increased blood pressure monitoring just following initiation.
• Consider modifying antihypertensive drugs in patients with labile blood pressure.
• Consider stopping or reducing background diuretics when starting an SGLT-2 inhibitor, and reassess volume status after 1 to 2 weeks.
• For patients on insulin, sulfonylureas, or both, consider decreasing dosages when starting an SGLT-2 inhibitor, and reassess glycemic control periodically.
• Counsel patients about urinary hygiene. Although bacterial urinary tract infections have not emerged as a problem, fungal genital infections have, particularly in women and uncircumcised men.
• Consider SGLT-2 inhibitors to be “sick-day” medications. Patients with diabetes must adjust their diabetes medications if their oral intake is reduced for a day or more, such as while sick or fasting. SGLT-2 inhibitors should not be taken on these days. Cases of diabetic ketoacidosis have arisen in patients who reduced oral intake while continuing their SGLT-2 inhibitor.

OTHER DRUGS WITH DEVELOPMENT HALTED

Aleglitazar, a peroxisome proliferator-activated receptor agonist taken orally once daily, raised high expectations when it was found in early studies to lower serum triglycerides and raise high-density lipoprotein cholesterol levels in addition to lowering blood glucose. However, a phase 3 trial in more than 7,000 patients was terminated after a median follow up of 2 years because of increased rates of heart failure, worsened kidney function, bone fractures, and gastrointestinal bleeding.⁴⁰ Development of this drug was stopped.

Fasiglifam, a G-protein-coupled receptor 40 agonist, was tested in a cardiovascular clinical outcomes trial. Compared with placebo, fasiglifam reduced hemoglobin A_1c  levels with low risk of hypoglycemia.⁴¹ However,  safety concerns about increased liver enzyme levels led to the cessation of the drug’s development.⁴²

HOW WILL THIS AFFECT DIABETES MANAGEMENT?

Metformin is still the most commonly prescribed drug for type 2 diabetes but has only marginal evidence for its cardiovascular benefits and may not be the first-line therapy for the management of diabetes in the future. In the EMPA REG OUTCOME, LEADER, and SUSTAIN-6 trials, the novel diabetes medications were given to patients who were already treated with available therapies, often including metformin. Treatment with empagliflozin, liraglutide, and semaglutide may be indicated for patients with diabetes and atherosclerotic vascular disease as first-line therapies in the future.

SGLT-2 inhibitor therapy can cost about $500 per month, and GLP-1 inhibitors are only slightly less expensive. The cost may be prohibitive for many patients. As more evidence, guidelines, and FDA criteria support the use of these novel diabetes drugs, third-party payers and pharmaceutical companies may be motivated to lower costs to help reach more patients who can benefit from these therapies.

Antibiotic stewardship has always been a good idea. Now it is also required by the Joint Commission and the Center for Medicare and Medicaid Services (CMS). This article reviews the state of antibiotic use in the United States and efforts to improve antibiotic stewardship in practice. 

ANTIBIOTICS ARE DIFFERENT FROM OTHER DRUGS

Their efficacy wanes over time. Antibiotics are the only medications that become less useful over time even if used correctly. Although other types of drugs are continuously being improved, the old ones work as well today as they did when they first came out. But antibiotics that were in use 50 years ago are no longer as effective.

They are a shared resource. Antibiotics are regularly used by many specialties to deliver routine and advanced medical care. Surgeries, transplantation, and immunosuppressive therapy would be unsafe without antibiotics to treat infections. Some patients awaiting lung transplant are not considered good candidates if they have evidence of colonization by antibiotic-resistant organisms.

Individual use may harm others. Even people who are not exposed to an antibiotic can suffer the consequences of how others use them.

In a retrospective cohort study, Freedberg et al¹ analyzed the risk of hospitalized patients developing Clostridium difficile infection and found that the risk was higher if the previous occupant of the bed had received antibiotics. The putative mechanism is that a patient receiving antibiotics develops altered gut flora, leading to C difficile spores released into the environment and not eradicated by normal cleaning. The next patient using the bed is then exposed and infected.

ANTIBIOTIC USE IS HIGH

The US Centers for Disease Control (CDC) monitors antibiotic prescriptions throughout the United States. In the outpatient setting, enough antibiotics are prescribed nationwide for 5 out of every 6 people to get 1 course of antibiotics annually (835 prescriptions per 1,000 people). Rates vary widely among states, with the lowest rate in Alaska (501 prescriptions per 1,000 people) and the highest in West Virginia (1,285 prescriptions per 1,000 people).² In comparison, Scandinavian countries prescribe about 400 courses per 1,000 people, about 20% less than our lowest-prescribing state.³

Antibiotics are probably the most frequently prescribed drugs in US hospitals. Data from 2006 to 2012 showed that 55% of hospitalized patients received at least 1 dose of an antibiotic and that overall about 75% of all hospital days involved an antibiotic.⁴ Rates did not vary by hospital size, but nonteaching hospitals tended to use antibiotics more than teaching hospitals. Antibiotic use is much more common in intensive care units than in hospital wards (1,092 and 720 days of antibiotic treatment per 1,000 patient-days, respectively).

Although overall antibiotic use did not change significantly over the years of the survey, use patterns did: fluoroquinolone use dropped by 20%, possibly reflecting rising resistance or increased attention to associated side effects (although fluoroquinolones remain the most widely prescribed inpatient antibiotic class), and use of first-generation cephalosporins fell by 7%. A cause for concern is that the use of broad-spectrum and “last-resort” antibiotics increased: carbapenem use by 37%, vancomycin use by 32%, beta-lactam/beta-lactamase inhibitor use by 26%, and third- and fourth-generation cephalosporin use by 12%.⁴

About one-third of use is unnecessary

Many studies have tried to measure the extent of inappropriate or unnecessary antibiotic use. The results have been remarkably consistent at 20% to 40% for both inpatient and outpatient studies. One study of hospitalized patients not in the intensive care unit found that 30% of 1,941 days of prescribed antimicrobial therapy were unnecessary, mostly because patients received antibiotics for longer than needed or because antibiotics were used to treat noninfectious syndromes or colonizing microorganisms.⁵  

ANTIBIOTIC EXPOSURE HAS NEGATIVE CONSEQUENCES

Any exposure to a medication involves the potential for side effects; this is true for antibiotics whether or not their use is appropriate. An estimated 140,000 visits to emergency departments occur annually for adverse reactions to antibiotics.⁶ In hospitalized patients, these reactions can be severe, including renal and bone marrow toxicity. As with any medications, the risks and benefits of antibiotic therapy must be weighed patient by patient.

Disturbance of gut microbiome

Antibiotics’ disruptive effects on normal gut flora are becoming better understood and are even believed to increase the risk of obesity and asthma.^7,8 

Animal models provide evidence that altered flora is associated with sepsis, which is attributed to the gut microbiome’s role in containing dissemination of bacteria in the body.⁹ An ecological study provides further evidence. Baggs et al¹⁰ retrospectively studied more than 9 million patients discharged without sepsis from 473 US hospitals, of whom 0.6% were readmitted for sepsis within 90 days. Exposure to a broad-spectrum antibiotic was associated with a 50% increased risk of readmission within 90 days of discharge because of sepsis (odds ratio 1.50, 95% confidence interval 1.47–1.53).

Increase of C difficile infections

Antibiotics exert selective pressure, killing susceptible bacteria and allowing resistant bacteria to thrive.

The risk of C difficile infection is 7 to 10 times higher than at baseline for 1 month after antibiotic use and 3 times higher than baseline in the 2 months after that.¹¹ Multiple studies have found that stewardship efforts to reduce antibiotic use have resulted in fewer C difficile infections.

A nationwide effort in England over the past decade to reduce C difficile infections has resulted in 50% less use of fluoroquinolones and third-generation cephalosporins in patients over age 65. During that time, the incidence of C difficile infection in that age group fell by about 70%, with concomitant reductions in mortality and colectomy associated with infection. No increase in rates of hospital admissions, infection complications, or death were observed.^12–14

The primary goal of antibiotic stewardship is better patient care. The goal is not reduced antibiotic use or cost savings, although these could be viewed as favorable side effects. Sometimes, better patient care involves using more antibiotics: eg, a patient with presumed sepsis should be started quickly on broad-spectrum antibiotics, an action that also falls under antibiotic stewardship. The focus for stewardship efforts should be on optimizing appropriate use, ie, promoting the use of the right agent at the correct dosage and for the proper duration.

Stewardship improves clinical outcomes

Antibiotic stewardship is important not only to society but to individual patients.

Singh et al¹⁵ randomized patients suspected of having ventilator-associated pneumonia (but with a low likelihood of pneumonia) to either a 3-day course of ciprofloxacin or standard care (antibiotics for 10 to 21 days, with the drug and duration chosen by the treating physician). After 3 days, the patients in the experimental group were reevaluated, and antibiotics were stopped if the likelihood of pneumonia was still deemed low. In patients who received only the short course of antibiotics, mean length of stay in the intensive care unit was 9 days and the risk of acquiring an antibiotic-resistant superinfection during hospitalization was 14%, compared with a 15-day length of stay and 38% risk of antibiotic-resistant superinfection in patients in the standard treatment group.

Fishman¹⁶ reported a study at a single hospital that randomized patients to either receive standard care according to physician choice or be treated according to an antibiotic stewardship program. Patients in the antibiotic stewardship group were almost 3 times more likely than controls to receive appropriate therapy according to guidelines. More important, the antibiotic stewardship patients were almost twice as likely to be cured of their infection and were more than 80% less likely to have treatment failure.

DEVELOPING EFFECTIVE ANTIBIOTIC STEWARDSHIP PROGRAMS

A good model for improving antibiotic use is a recent nationwide program designed to reduce central line-associated bloodstream infections.¹⁷ Rates of these infections have dropped by about 50% over the past 5 years. The program included:

• Research to better understand the problem and how to fight it
• Well-defined programs and interventions
• Education to implement interventions, eg, deploying teams to teach better techniques of inserting and maintaining central lines
• A strong national measurement system (the CDC’s National Healthcare Safety Network) to track infections.

What constitutes an antibiotic stewardship program?

The CDC examined successful stewardship programs in a variety of hospital types, including large academic hospitals and smaller hospitals, and identified 7 common core elements that could serve as general principles that were common to successful antibiotic stewardship programs¹⁸:

• Leadership commitment from administration
• A single leader responsible for outcomes
• A single pharmacy leader
• Tracking of antibiotic use
• Regular reporting of antibiotic use and resistance
• Educating providers on use and resistance
• Specific improvement interventions.

Stewardship is harder in some settings

In reply to a CDC survey in 2014, 41% of more than 4,000 hospitals reported that they had antibiotic stewardship programs with all 7 core elements. The single element that predicted whether a complete program was in place was leadership support.¹⁹ The following year, 48% of respondents reported that they had a complete program in place. Percentages varied among states, with highs in Utah (77%) and California (70%) and lows in North Dakota (12%) and Vermont (7%). Large hospitals and major teaching hospitals were more likely to have a program with all 7 elements: 31% of hospitals with 50 or fewer beds had a complete program vs 66% of hospitals with at least 200 beds.²⁰

Short-stay, critical-access hospitals pose a special challenge, as only 26% reported having all core elements.^19,20 These facilities have fewer than 25 beds, and many patient stays are less than 3 days. Some do not employ full-time pharmacists or full-time clinicians. The CDC is collaborating with the American Hospital Association and the Pew Charitable Trusts to focus efforts on helping these hospitals, which requires a more flexible approach. About 100 critical-access hospitals nationwide have reported implementing all of the core elements and can serve as models for the others.

MEASURING IMPROVEMENT

The CDC has adopted a 3-pronged approach to measuring improvements in hospital antibiotic use:

• Estimate national aggregate antibiotic use described above
• Acquire information on antibiotic use at facility, practice, and provider levels
• Assess appropriate antibiotic use.

In hospitals, the CDC has concentrated on facility-level measurement. Hospitals need a system to track their own use and compare it with that of similar facilities. The CDC’s monitoring program, the Antibiotic Use Option of the National Healthcare Safety Network, captures electronic data on antibiotic use in a facility, enabling monitoring of use in each unit. Data can also be aggregated at regional, state, and national levels. This information can be used to develop benchmarks for antibiotic use, so that similar hospitals can be compared.

What is the ‘right’ amount of antibiotic use? Enter SAAR

Creating benchmarks for antibiotic use poses a number of challenges compared with most other areas in healthcare. Most public health measures are binary—eg, people either get an infection, a vaccination, or a smoking cessation intervention or not—and the direction of progress is clear. Antibiotics are different: not everybody needs them, but some people do. Usage should be reduced, but by exactly how much is unclear and varies between hospitals. In addition, being an outlier does not necessarily indicate a problem: a hospital unit for organ transplants will have high rates of antibiotic use, which is likely appropriate.

The CDC has taken initial steps to develop a risk-adjusted benchmark measure for hospital antibiotic use, the Standardized Antimicrobial Administration Ratio (SAAR). It compares a hospital’s observed antibiotic use with a calculation of predicted use based on its facility characteristics. Although still at an early stage, SAAR has been released and has been endorsed by the National Quality Forum. About 200 hospitals are submitting data to the CDC and collaborating with the CDC to evaluate the SAAR’s utility in driving improved antibiotic use.

Problems in measuring appropriate use

Measuring appropriate antibiotic use is easier in the outpatient setting, where detailed data have been collected for many years.

Fleming-Dutra et al²¹ compared medications prescribed during outpatient visits and the diagnoses coded for the visits. They found that about 13% of all outpatient visits resulted in an antibiotic prescription, 30% of which had no listed diagnosis that would justify an antibiotic (eg, viral upper respiratory infection). This kind of information provides a target for stewardship programs.

It is more difficult to conduct such a study in a hospital setting. Simply comparing discharge diagnoses to antibiotics prescribed is not useful: often antibiotics are started presumptively on admission for a patient with signs and symptoms of an infection, then stopped if the diagnosis does not warrant antibiotics, which is a reasonable strategy.

Also, many times, a patient with asymptomatic bacteriuria, which does not warrant antibiotics, is misdiagnosed as having a urinary tract infection, which does. So simply looking at the discharge code may not reveal whether therapy was appropriate.

Some studies have provided useful information. Fridkin et al²² studied 36 hospitals for the use of vancomycin, which is an especially good candidate drug for study because guidelines exist for appropriate use. Data were collected only from patients given vancomycin for more than 3 days, which should have eliminated empiric use of the drug and included only pathogen-driven therapy. Cases where therapy was for skin and soft-tissue infections were excluded because cultures are not usually obtained for these cases. Of patients given vancomycin, 9% had no diagnostic culture obtained at antibiotic initiation, 22% had diagnostic culture but results showed no gram-positive bacterial growth, and 5% had culture results revealing only oxacillin-susceptible Staphylococcus aureus. In 36% of cases, opportunities existed for improved prescribing.

Such data could be collected from the electronic medical record, and the CDC is focusing efforts in this direction.

NATIONAL ACTIVITIES IN ANTIBIOTIC STEWARDSHIP

In 2014, the White House launched a national strategy to combat antibiotic resistance,²³ followed by an action plan in 2015.²⁴ As a result, new investments have been made to improve antibiotic use, including funding for state health departments to begin stewardship efforts and to expand public awareness of the problems of antibiotic overuse. Research efforts are also being funded to improve implementation of existing stewardship practices and to develop new ones.

CMS is also exploring how to drive improved antibiotic use. In October 2016, it started requiring all US nursing homes to have antibiotic stewardship programs, and a similar requirement for hospitals has been proposed.

The Joint Commission issued a standard requiring that all their accredited facilities, starting with hospitals, have an antibiotic stewardship program by January 2017. This standard requires implementation of all the CDC’s core elements.

PROVEN INTERVENTIONS

Focusing on key interventions that are likely to be effective and well received by providers is a useful strategy for antibiotic stewardship efforts. A number of such interventions have been supported by research.

Postprescription antibiotic reviews or antibiotic ‘time-outs’

Antibiotics are often started empirically to treat hospitalized patients suspected of having an infection. The need for the antibiotic should be assessed a few days later, when culture results and more clinical information are available.  

Elligsen et al²⁵ evaluated the effects of providing a formal review and suggestions for antimicrobial optimization to critical care teams of 3 intensive care units in a single hospital after 3 and 10 days of antibiotic therapy. Mean monthly antibiotic use decreased from 644 days of therapy per 1,000 patient-days in the preintervention period to 503 days of therapy per 1,000 patient-days (P < .0001). C difficile infections were reduced from 11 cases to 6. Overall gram-negative susceptibility to meropenem increased in the critical care units. 

Targeting specific infections

Some infections are especially important to target with improvement efforts.

In 2011, Magill et al²⁶  conducted 1-day prevalence surveys in 183 hospitals in 10 states to examine patterns of antibiotic use. They found that lower respiratory tract infections and urinary tract infections accounted for more than half of all antibiotic use (35% and 22%, respectively), making them good candidates for improved use.

Community-acquired pneumonia can be targeted at multiple fronts.  One study showed that almost 30% of patients diagnosed with community-acquired pneumonia in the emergency department did not actually have pneumonia.²⁷  Duration of antibiotic therapy could also be targeted. Guidelines recommend that most patients with uncomplicated community-acquired pneumonia receive 5 to 7 days of antibiotic therapy. Avdic et al²⁸ performed a simple intervention involving education and feedback to teams in 1 hospital regarding antibiotic choice and duration. This resulted in reducing the duration of therapy for community-acquired pneumonia from a median of 10 to 7 days.

Asymptomatic bacteriuria is often misdiagnosed as a urinary tract infection and treated unnecessarily.^29–31

Trautner et al³² addressed this problem by targeting urine cultures rather than antibiotics, using a simple algorithm: if a patient did not have symptoms of urinary tract infection (fever, acute hematuria, delirium, rigors, flank pain, pelvic discomfort, urgency, frequency, dysuria, suprapubic pain), a urine culture was not recommended. If a patient did have symptoms but a problem other than urinary tract infection was deemed likely, evaluation of other sources of infection was recommended. Use of the algorithm resulted in fewer urine cultures and less antibiotic overtreatment of asymptomatic bacteriuria. Reductions persisted after the intervention ended. 

Antibiotic time-out at hospital discharge

Another study evaluated an intervention that required a pharmacist consultation for the critical care team when a patient was to be discharged with intravenous antibiotics (most often for pneumonia). In 28% of cases, chart review revealed that the infection had been completely treated at the time of discharge, so further antibiotic treatment was not indicated. No patients who avoided antibiotics at discharge were readmitted or subsequently visited the emergency department.³³

Targeting outpatient settings

A number of studies have evaluated simple interventions to improve outpatient antibiotic prescribing. Meeker et al³⁴ had providers place a poster in their examination rooms with a picture of the physician and a signed letter committing to the appropriate use of antibiotics. Inappropriate antibiotic use decreased 20% in the intervention group vs controls (P = .02). 

In a subsequent study,³⁵ the same group required providers to include a justification note in the electronic medical record every time an antibiotic was prescribed for an indication when guidelines do not recommend one. Inappropriate prescribing dropped from 23% to 5% (P < .001).

Another intervention in this study³⁵ provided physicians with periodic feedback according to whether their therapy was concordant with guidelines. They received an email with a subject line of either “You are a top performer” or “You are not a top performer.” The contents of the email provided data on how many antibiotic prescriptions they wrote for conditions that did not warrant them and how their prescribing habits compared with those of their top-performing peers. Mean inappropriate antibiotic prescribing fell from 20% to 4%.³⁵

This is a critical time for antibiotic stewardship efforts in the United States. The need has never been more urgent and, fortunately, the opportunities have never been more abundant. Requirements for stewardship programs will drive implementation, but hospitals will need support and guidance to help ensure that stewardship programs are as effective as possible. Ultimately, improving antibiotic use will require collaboration among all stakeholders. CDC is eager to partner with providers and others in their efforts to improve antibiotic use.

Antibiotic stewardship has always been a good idea. Now it is also required by the Joint Commission and the Center for Medicare and Medicaid Services (CMS). This article reviews the state of antibiotic use in the United States and efforts to improve antibiotic stewardship in practice. 

ANTIBIOTICS ARE DIFFERENT FROM OTHER DRUGS

Their efficacy wanes over time. Antibiotics are the only medications that become less useful over time even if used correctly. Although other types of drugs are continuously being improved, the old ones work as well today as they did when they first came out. But antibiotics that were in use 50 years ago are no longer as effective.

They are a shared resource. Antibiotics are regularly used by many specialties to deliver routine and advanced medical care. Surgeries, transplantation, and immunosuppressive therapy would be unsafe without antibiotics to treat infections. Some patients awaiting lung transplant are not considered good candidates if they have evidence of colonization by antibiotic-resistant organisms.

Individual use may harm others. Even people who are not exposed to an antibiotic can suffer the consequences of how others use them.

In a retrospective cohort study, Freedberg et al¹ analyzed the risk of hospitalized patients developing Clostridium difficile infection and found that the risk was higher if the previous occupant of the bed had received antibiotics. The putative mechanism is that a patient receiving antibiotics develops altered gut flora, leading to C difficile spores released into the environment and not eradicated by normal cleaning. The next patient using the bed is then exposed and infected.

ANTIBIOTIC USE IS HIGH

The US Centers for Disease Control (CDC) monitors antibiotic prescriptions throughout the United States. In the outpatient setting, enough antibiotics are prescribed nationwide for 5 out of every 6 people to get 1 course of antibiotics annually (835 prescriptions per 1,000 people). Rates vary widely among states, with the lowest rate in Alaska (501 prescriptions per 1,000 people) and the highest in West Virginia (1,285 prescriptions per 1,000 people).² In comparison, Scandinavian countries prescribe about 400 courses per 1,000 people, about 20% less than our lowest-prescribing state.³

Antibiotics are probably the most frequently prescribed drugs in US hospitals. Data from 2006 to 2012 showed that 55% of hospitalized patients received at least 1 dose of an antibiotic and that overall about 75% of all hospital days involved an antibiotic.⁴ Rates did not vary by hospital size, but nonteaching hospitals tended to use antibiotics more than teaching hospitals. Antibiotic use is much more common in intensive care units than in hospital wards (1,092 and 720 days of antibiotic treatment per 1,000 patient-days, respectively).

Although overall antibiotic use did not change significantly over the years of the survey, use patterns did: fluoroquinolone use dropped by 20%, possibly reflecting rising resistance or increased attention to associated side effects (although fluoroquinolones remain the most widely prescribed inpatient antibiotic class), and use of first-generation cephalosporins fell by 7%. A cause for concern is that the use of broad-spectrum and “last-resort” antibiotics increased: carbapenem use by 37%, vancomycin use by 32%, beta-lactam/beta-lactamase inhibitor use by 26%, and third- and fourth-generation cephalosporin use by 12%.⁴

About one-third of use is unnecessary

Many studies have tried to measure the extent of inappropriate or unnecessary antibiotic use. The results have been remarkably consistent at 20% to 40% for both inpatient and outpatient studies. One study of hospitalized patients not in the intensive care unit found that 30% of 1,941 days of prescribed antimicrobial therapy were unnecessary, mostly because patients received antibiotics for longer than needed or because antibiotics were used to treat noninfectious syndromes or colonizing microorganisms.⁵  

ANTIBIOTIC EXPOSURE HAS NEGATIVE CONSEQUENCES

Any exposure to a medication involves the potential for side effects; this is true for antibiotics whether or not their use is appropriate. An estimated 140,000 visits to emergency departments occur annually for adverse reactions to antibiotics.⁶ In hospitalized patients, these reactions can be severe, including renal and bone marrow toxicity. As with any medications, the risks and benefits of antibiotic therapy must be weighed patient by patient.

Disturbance of gut microbiome

Antibiotics’ disruptive effects on normal gut flora are becoming better understood and are even believed to increase the risk of obesity and asthma.^7,8 

Animal models provide evidence that altered flora is associated with sepsis, which is attributed to the gut microbiome’s role in containing dissemination of bacteria in the body.⁹ An ecological study provides further evidence. Baggs et al¹⁰ retrospectively studied more than 9 million patients discharged without sepsis from 473 US hospitals, of whom 0.6% were readmitted for sepsis within 90 days. Exposure to a broad-spectrum antibiotic was associated with a 50% increased risk of readmission within 90 days of discharge because of sepsis (odds ratio 1.50, 95% confidence interval 1.47–1.53).

Increase of C difficile infections

Antibiotics exert selective pressure, killing susceptible bacteria and allowing resistant bacteria to thrive.

The risk of C difficile infection is 7 to 10 times higher than at baseline for 1 month after antibiotic use and 3 times higher than baseline in the 2 months after that.¹¹ Multiple studies have found that stewardship efforts to reduce antibiotic use have resulted in fewer C difficile infections.

A nationwide effort in England over the past decade to reduce C difficile infections has resulted in 50% less use of fluoroquinolones and third-generation cephalosporins in patients over age 65. During that time, the incidence of C difficile infection in that age group fell by about 70%, with concomitant reductions in mortality and colectomy associated with infection. No increase in rates of hospital admissions, infection complications, or death were observed.^12–14

The primary goal of antibiotic stewardship is better patient care. The goal is not reduced antibiotic use or cost savings, although these could be viewed as favorable side effects. Sometimes, better patient care involves using more antibiotics: eg, a patient with presumed sepsis should be started quickly on broad-spectrum antibiotics, an action that also falls under antibiotic stewardship. The focus for stewardship efforts should be on optimizing appropriate use, ie, promoting the use of the right agent at the correct dosage and for the proper duration.

Stewardship improves clinical outcomes

Antibiotic stewardship is important not only to society but to individual patients.

Singh et al¹⁵ randomized patients suspected of having ventilator-associated pneumonia (but with a low likelihood of pneumonia) to either a 3-day course of ciprofloxacin or standard care (antibiotics for 10 to 21 days, with the drug and duration chosen by the treating physician). After 3 days, the patients in the experimental group were reevaluated, and antibiotics were stopped if the likelihood of pneumonia was still deemed low. In patients who received only the short course of antibiotics, mean length of stay in the intensive care unit was 9 days and the risk of acquiring an antibiotic-resistant superinfection during hospitalization was 14%, compared with a 15-day length of stay and 38% risk of antibiotic-resistant superinfection in patients in the standard treatment group.

Fishman¹⁶ reported a study at a single hospital that randomized patients to either receive standard care according to physician choice or be treated according to an antibiotic stewardship program. Patients in the antibiotic stewardship group were almost 3 times more likely than controls to receive appropriate therapy according to guidelines. More important, the antibiotic stewardship patients were almost twice as likely to be cured of their infection and were more than 80% less likely to have treatment failure.

DEVELOPING EFFECTIVE ANTIBIOTIC STEWARDSHIP PROGRAMS

A good model for improving antibiotic use is a recent nationwide program designed to reduce central line-associated bloodstream infections.¹⁷ Rates of these infections have dropped by about 50% over the past 5 years. The program included:

• Research to better understand the problem and how to fight it
• Well-defined programs and interventions
• Education to implement interventions, eg, deploying teams to teach better techniques of inserting and maintaining central lines
• A strong national measurement system (the CDC’s National Healthcare Safety Network) to track infections.

What constitutes an antibiotic stewardship program?

The CDC examined successful stewardship programs in a variety of hospital types, including large academic hospitals and smaller hospitals, and identified 7 common core elements that could serve as general principles that were common to successful antibiotic stewardship programs¹⁸:

• Leadership commitment from administration
• A single leader responsible for outcomes
• A single pharmacy leader
• Tracking of antibiotic use
• Regular reporting of antibiotic use and resistance
• Educating providers on use and resistance
• Specific improvement interventions.

Stewardship is harder in some settings

In reply to a CDC survey in 2014, 41% of more than 4,000 hospitals reported that they had antibiotic stewardship programs with all 7 core elements. The single element that predicted whether a complete program was in place was leadership support.¹⁹ The following year, 48% of respondents reported that they had a complete program in place. Percentages varied among states, with highs in Utah (77%) and California (70%) and lows in North Dakota (12%) and Vermont (7%). Large hospitals and major teaching hospitals were more likely to have a program with all 7 elements: 31% of hospitals with 50 or fewer beds had a complete program vs 66% of hospitals with at least 200 beds.²⁰

Short-stay, critical-access hospitals pose a special challenge, as only 26% reported having all core elements.^19,20 These facilities have fewer than 25 beds, and many patient stays are less than 3 days. Some do not employ full-time pharmacists or full-time clinicians. The CDC is collaborating with the American Hospital Association and the Pew Charitable Trusts to focus efforts on helping these hospitals, which requires a more flexible approach. About 100 critical-access hospitals nationwide have reported implementing all of the core elements and can serve as models for the others.

MEASURING IMPROVEMENT

The CDC has adopted a 3-pronged approach to measuring improvements in hospital antibiotic use:

• Estimate national aggregate antibiotic use described above
• Acquire information on antibiotic use at facility, practice, and provider levels
• Assess appropriate antibiotic use.

In hospitals, the CDC has concentrated on facility-level measurement. Hospitals need a system to track their own use and compare it with that of similar facilities. The CDC’s monitoring program, the Antibiotic Use Option of the National Healthcare Safety Network, captures electronic data on antibiotic use in a facility, enabling monitoring of use in each unit. Data can also be aggregated at regional, state, and national levels. This information can be used to develop benchmarks for antibiotic use, so that similar hospitals can be compared.

What is the ‘right’ amount of antibiotic use? Enter SAAR

Creating benchmarks for antibiotic use poses a number of challenges compared with most other areas in healthcare. Most public health measures are binary—eg, people either get an infection, a vaccination, or a smoking cessation intervention or not—and the direction of progress is clear. Antibiotics are different: not everybody needs them, but some people do. Usage should be reduced, but by exactly how much is unclear and varies between hospitals. In addition, being an outlier does not necessarily indicate a problem: a hospital unit for organ transplants will have high rates of antibiotic use, which is likely appropriate.

The CDC has taken initial steps to develop a risk-adjusted benchmark measure for hospital antibiotic use, the Standardized Antimicrobial Administration Ratio (SAAR). It compares a hospital’s observed antibiotic use with a calculation of predicted use based on its facility characteristics. Although still at an early stage, SAAR has been released and has been endorsed by the National Quality Forum. About 200 hospitals are submitting data to the CDC and collaborating with the CDC to evaluate the SAAR’s utility in driving improved antibiotic use.

Problems in measuring appropriate use

Measuring appropriate antibiotic use is easier in the outpatient setting, where detailed data have been collected for many years.

Fleming-Dutra et al²¹ compared medications prescribed during outpatient visits and the diagnoses coded for the visits. They found that about 13% of all outpatient visits resulted in an antibiotic prescription, 30% of which had no listed diagnosis that would justify an antibiotic (eg, viral upper respiratory infection). This kind of information provides a target for stewardship programs.

It is more difficult to conduct such a study in a hospital setting. Simply comparing discharge diagnoses to antibiotics prescribed is not useful: often antibiotics are started presumptively on admission for a patient with signs and symptoms of an infection, then stopped if the diagnosis does not warrant antibiotics, which is a reasonable strategy.

Also, many times, a patient with asymptomatic bacteriuria, which does not warrant antibiotics, is misdiagnosed as having a urinary tract infection, which does. So simply looking at the discharge code may not reveal whether therapy was appropriate.

Some studies have provided useful information. Fridkin et al²² studied 36 hospitals for the use of vancomycin, which is an especially good candidate drug for study because guidelines exist for appropriate use. Data were collected only from patients given vancomycin for more than 3 days, which should have eliminated empiric use of the drug and included only pathogen-driven therapy. Cases where therapy was for skin and soft-tissue infections were excluded because cultures are not usually obtained for these cases. Of patients given vancomycin, 9% had no diagnostic culture obtained at antibiotic initiation, 22% had diagnostic culture but results showed no gram-positive bacterial growth, and 5% had culture results revealing only oxacillin-susceptible Staphylococcus aureus. In 36% of cases, opportunities existed for improved prescribing.

Such data could be collected from the electronic medical record, and the CDC is focusing efforts in this direction.

NATIONAL ACTIVITIES IN ANTIBIOTIC STEWARDSHIP

In 2014, the White House launched a national strategy to combat antibiotic resistance,²³ followed by an action plan in 2015.²⁴ As a result, new investments have been made to improve antibiotic use, including funding for state health departments to begin stewardship efforts and to expand public awareness of the problems of antibiotic overuse. Research efforts are also being funded to improve implementation of existing stewardship practices and to develop new ones.

CMS is also exploring how to drive improved antibiotic use. In October 2016, it started requiring all US nursing homes to have antibiotic stewardship programs, and a similar requirement for hospitals has been proposed.

The Joint Commission issued a standard requiring that all their accredited facilities, starting with hospitals, have an antibiotic stewardship program by January 2017. This standard requires implementation of all the CDC’s core elements.

PROVEN INTERVENTIONS

Focusing on key interventions that are likely to be effective and well received by providers is a useful strategy for antibiotic stewardship efforts. A number of such interventions have been supported by research.

Postprescription antibiotic reviews or antibiotic ‘time-outs’

Antibiotics are often started empirically to treat hospitalized patients suspected of having an infection. The need for the antibiotic should be assessed a few days later, when culture results and more clinical information are available.  

Elligsen et al²⁵ evaluated the effects of providing a formal review and suggestions for antimicrobial optimization to critical care teams of 3 intensive care units in a single hospital after 3 and 10 days of antibiotic therapy. Mean monthly antibiotic use decreased from 644 days of therapy per 1,000 patient-days in the preintervention period to 503 days of therapy per 1,000 patient-days (P < .0001). C difficile infections were reduced from 11 cases to 6. Overall gram-negative susceptibility to meropenem increased in the critical care units. 

Targeting specific infections

Some infections are especially important to target with improvement efforts.

In 2011, Magill et al²⁶  conducted 1-day prevalence surveys in 183 hospitals in 10 states to examine patterns of antibiotic use. They found that lower respiratory tract infections and urinary tract infections accounted for more than half of all antibiotic use (35% and 22%, respectively), making them good candidates for improved use.

Community-acquired pneumonia can be targeted at multiple fronts.  One study showed that almost 30% of patients diagnosed with community-acquired pneumonia in the emergency department did not actually have pneumonia.²⁷  Duration of antibiotic therapy could also be targeted. Guidelines recommend that most patients with uncomplicated community-acquired pneumonia receive 5 to 7 days of antibiotic therapy. Avdic et al²⁸ performed a simple intervention involving education and feedback to teams in 1 hospital regarding antibiotic choice and duration. This resulted in reducing the duration of therapy for community-acquired pneumonia from a median of 10 to 7 days.

Asymptomatic bacteriuria is often misdiagnosed as a urinary tract infection and treated unnecessarily.^29–31

Trautner et al³² addressed this problem by targeting urine cultures rather than antibiotics, using a simple algorithm: if a patient did not have symptoms of urinary tract infection (fever, acute hematuria, delirium, rigors, flank pain, pelvic discomfort, urgency, frequency, dysuria, suprapubic pain), a urine culture was not recommended. If a patient did have symptoms but a problem other than urinary tract infection was deemed likely, evaluation of other sources of infection was recommended. Use of the algorithm resulted in fewer urine cultures and less antibiotic overtreatment of asymptomatic bacteriuria. Reductions persisted after the intervention ended. 

Antibiotic time-out at hospital discharge

Another study evaluated an intervention that required a pharmacist consultation for the critical care team when a patient was to be discharged with intravenous antibiotics (most often for pneumonia). In 28% of cases, chart review revealed that the infection had been completely treated at the time of discharge, so further antibiotic treatment was not indicated. No patients who avoided antibiotics at discharge were readmitted or subsequently visited the emergency department.³³

Targeting outpatient settings

A number of studies have evaluated simple interventions to improve outpatient antibiotic prescribing. Meeker et al³⁴ had providers place a poster in their examination rooms with a picture of the physician and a signed letter committing to the appropriate use of antibiotics. Inappropriate antibiotic use decreased 20% in the intervention group vs controls (P = .02). 

In a subsequent study,³⁵ the same group required providers to include a justification note in the electronic medical record every time an antibiotic was prescribed for an indication when guidelines do not recommend one. Inappropriate prescribing dropped from 23% to 5% (P < .001).

Another intervention in this study³⁵ provided physicians with periodic feedback according to whether their therapy was concordant with guidelines. They received an email with a subject line of either “You are a top performer” or “You are not a top performer.” The contents of the email provided data on how many antibiotic prescriptions they wrote for conditions that did not warrant them and how their prescribing habits compared with those of their top-performing peers. Mean inappropriate antibiotic prescribing fell from 20% to 4%.³⁵

This is a critical time for antibiotic stewardship efforts in the United States. The need has never been more urgent and, fortunately, the opportunities have never been more abundant. Requirements for stewardship programs will drive implementation, but hospitals will need support and guidance to help ensure that stewardship programs are as effective as possible. Ultimately, improving antibiotic use will require collaboration among all stakeholders. CDC is eager to partner with providers and others in their efforts to improve antibiotic use.

Antibiotic stewardship has always been a good idea. Now it is also required by the Joint Commission and the Center for Medicare and Medicaid Services (CMS). This article reviews the state of antibiotic use in the United States and efforts to improve antibiotic stewardship in practice. 

ANTIBIOTICS ARE DIFFERENT FROM OTHER DRUGS

Their efficacy wanes over time. Antibiotics are the only medications that become less useful over time even if used correctly. Although other types of drugs are continuously being improved, the old ones work as well today as they did when they first came out. But antibiotics that were in use 50 years ago are no longer as effective.

They are a shared resource. Antibiotics are regularly used by many specialties to deliver routine and advanced medical care. Surgeries, transplantation, and immunosuppressive therapy would be unsafe without antibiotics to treat infections. Some patients awaiting lung transplant are not considered good candidates if they have evidence of colonization by antibiotic-resistant organisms.

Individual use may harm others. Even people who are not exposed to an antibiotic can suffer the consequences of how others use them.

In a retrospective cohort study, Freedberg et al¹ analyzed the risk of hospitalized patients developing Clostridium difficile infection and found that the risk was higher if the previous occupant of the bed had received antibiotics. The putative mechanism is that a patient receiving antibiotics develops altered gut flora, leading to C difficile spores released into the environment and not eradicated by normal cleaning. The next patient using the bed is then exposed and infected.

ANTIBIOTIC USE IS HIGH

The US Centers for Disease Control (CDC) monitors antibiotic prescriptions throughout the United States. In the outpatient setting, enough antibiotics are prescribed nationwide for 5 out of every 6 people to get 1 course of antibiotics annually (835 prescriptions per 1,000 people). Rates vary widely among states, with the lowest rate in Alaska (501 prescriptions per 1,000 people) and the highest in West Virginia (1,285 prescriptions per 1,000 people).² In comparison, Scandinavian countries prescribe about 400 courses per 1,000 people, about 20% less than our lowest-prescribing state.³

Antibiotics are probably the most frequently prescribed drugs in US hospitals. Data from 2006 to 2012 showed that 55% of hospitalized patients received at least 1 dose of an antibiotic and that overall about 75% of all hospital days involved an antibiotic.⁴ Rates did not vary by hospital size, but nonteaching hospitals tended to use antibiotics more than teaching hospitals. Antibiotic use is much more common in intensive care units than in hospital wards (1,092 and 720 days of antibiotic treatment per 1,000 patient-days, respectively).

Although overall antibiotic use did not change significantly over the years of the survey, use patterns did: fluoroquinolone use dropped by 20%, possibly reflecting rising resistance or increased attention to associated side effects (although fluoroquinolones remain the most widely prescribed inpatient antibiotic class), and use of first-generation cephalosporins fell by 7%. A cause for concern is that the use of broad-spectrum and “last-resort” antibiotics increased: carbapenem use by 37%, vancomycin use by 32%, beta-lactam/beta-lactamase inhibitor use by 26%, and third- and fourth-generation cephalosporin use by 12%.⁴

About one-third of use is unnecessary

Many studies have tried to measure the extent of inappropriate or unnecessary antibiotic use. The results have been remarkably consistent at 20% to 40% for both inpatient and outpatient studies. One study of hospitalized patients not in the intensive care unit found that 30% of 1,941 days of prescribed antimicrobial therapy were unnecessary, mostly because patients received antibiotics for longer than needed or because antibiotics were used to treat noninfectious syndromes or colonizing microorganisms.⁵  

ANTIBIOTIC EXPOSURE HAS NEGATIVE CONSEQUENCES

Any exposure to a medication involves the potential for side effects; this is true for antibiotics whether or not their use is appropriate. An estimated 140,000 visits to emergency departments occur annually for adverse reactions to antibiotics.⁶ In hospitalized patients, these reactions can be severe, including renal and bone marrow toxicity. As with any medications, the risks and benefits of antibiotic therapy must be weighed patient by patient.

Disturbance of gut microbiome

Antibiotics’ disruptive effects on normal gut flora are becoming better understood and are even believed to increase the risk of obesity and asthma.^7,8 

Animal models provide evidence that altered flora is associated with sepsis, which is attributed to the gut microbiome’s role in containing dissemination of bacteria in the body.⁹ An ecological study provides further evidence. Baggs et al¹⁰ retrospectively studied more than 9 million patients discharged without sepsis from 473 US hospitals, of whom 0.6% were readmitted for sepsis within 90 days. Exposure to a broad-spectrum antibiotic was associated with a 50% increased risk of readmission within 90 days of discharge because of sepsis (odds ratio 1.50, 95% confidence interval 1.47–1.53).

Increase of C difficile infections

Antibiotics exert selective pressure, killing susceptible bacteria and allowing resistant bacteria to thrive.

The risk of C difficile infection is 7 to 10 times higher than at baseline for 1 month after antibiotic use and 3 times higher than baseline in the 2 months after that.¹¹ Multiple studies have found that stewardship efforts to reduce antibiotic use have resulted in fewer C difficile infections.

A nationwide effort in England over the past decade to reduce C difficile infections has resulted in 50% less use of fluoroquinolones and third-generation cephalosporins in patients over age 65. During that time, the incidence of C difficile infection in that age group fell by about 70%, with concomitant reductions in mortality and colectomy associated with infection. No increase in rates of hospital admissions, infection complications, or death were observed.^12–14

The primary goal of antibiotic stewardship is better patient care. The goal is not reduced antibiotic use or cost savings, although these could be viewed as favorable side effects. Sometimes, better patient care involves using more antibiotics: eg, a patient with presumed sepsis should be started quickly on broad-spectrum antibiotics, an action that also falls under antibiotic stewardship. The focus for stewardship efforts should be on optimizing appropriate use, ie, promoting the use of the right agent at the correct dosage and for the proper duration.

Stewardship improves clinical outcomes

Antibiotic stewardship is important not only to society but to individual patients.

Singh et al¹⁵ randomized patients suspected of having ventilator-associated pneumonia (but with a low likelihood of pneumonia) to either a 3-day course of ciprofloxacin or standard care (antibiotics for 10 to 21 days, with the drug and duration chosen by the treating physician). After 3 days, the patients in the experimental group were reevaluated, and antibiotics were stopped if the likelihood of pneumonia was still deemed low. In patients who received only the short course of antibiotics, mean length of stay in the intensive care unit was 9 days and the risk of acquiring an antibiotic-resistant superinfection during hospitalization was 14%, compared with a 15-day length of stay and 38% risk of antibiotic-resistant superinfection in patients in the standard treatment group.

Fishman¹⁶ reported a study at a single hospital that randomized patients to either receive standard care according to physician choice or be treated according to an antibiotic stewardship program. Patients in the antibiotic stewardship group were almost 3 times more likely than controls to receive appropriate therapy according to guidelines. More important, the antibiotic stewardship patients were almost twice as likely to be cured of their infection and were more than 80% less likely to have treatment failure.

DEVELOPING EFFECTIVE ANTIBIOTIC STEWARDSHIP PROGRAMS

A good model for improving antibiotic use is a recent nationwide program designed to reduce central line-associated bloodstream infections.¹⁷ Rates of these infections have dropped by about 50% over the past 5 years. The program included:

• Research to better understand the problem and how to fight it
• Well-defined programs and interventions
• Education to implement interventions, eg, deploying teams to teach better techniques of inserting and maintaining central lines
• A strong national measurement system (the CDC’s National Healthcare Safety Network) to track infections.

What constitutes an antibiotic stewardship program?

The CDC examined successful stewardship programs in a variety of hospital types, including large academic hospitals and smaller hospitals, and identified 7 common core elements that could serve as general principles that were common to successful antibiotic stewardship programs¹⁸:

• Leadership commitment from administration
• A single leader responsible for outcomes
• A single pharmacy leader
• Tracking of antibiotic use
• Regular reporting of antibiotic use and resistance
• Educating providers on use and resistance
• Specific improvement interventions.

Stewardship is harder in some settings

In reply to a CDC survey in 2014, 41% of more than 4,000 hospitals reported that they had antibiotic stewardship programs with all 7 core elements. The single element that predicted whether a complete program was in place was leadership support.¹⁹ The following year, 48% of respondents reported that they had a complete program in place. Percentages varied among states, with highs in Utah (77%) and California (70%) and lows in North Dakota (12%) and Vermont (7%). Large hospitals and major teaching hospitals were more likely to have a program with all 7 elements: 31% of hospitals with 50 or fewer beds had a complete program vs 66% of hospitals with at least 200 beds.²⁰

Short-stay, critical-access hospitals pose a special challenge, as only 26% reported having all core elements.^19,20 These facilities have fewer than 25 beds, and many patient stays are less than 3 days. Some do not employ full-time pharmacists or full-time clinicians. The CDC is collaborating with the American Hospital Association and the Pew Charitable Trusts to focus efforts on helping these hospitals, which requires a more flexible approach. About 100 critical-access hospitals nationwide have reported implementing all of the core elements and can serve as models for the others.

MEASURING IMPROVEMENT

The CDC has adopted a 3-pronged approach to measuring improvements in hospital antibiotic use:

• Estimate national aggregate antibiotic use described above
• Acquire information on antibiotic use at facility, practice, and provider levels
• Assess appropriate antibiotic use.

In hospitals, the CDC has concentrated on facility-level measurement. Hospitals need a system to track their own use and compare it with that of similar facilities. The CDC’s monitoring program, the Antibiotic Use Option of the National Healthcare Safety Network, captures electronic data on antibiotic use in a facility, enabling monitoring of use in each unit. Data can also be aggregated at regional, state, and national levels. This information can be used to develop benchmarks for antibiotic use, so that similar hospitals can be compared.

What is the ‘right’ amount of antibiotic use? Enter SAAR

Creating benchmarks for antibiotic use poses a number of challenges compared with most other areas in healthcare. Most public health measures are binary—eg, people either get an infection, a vaccination, or a smoking cessation intervention or not—and the direction of progress is clear. Antibiotics are different: not everybody needs them, but some people do. Usage should be reduced, but by exactly how much is unclear and varies between hospitals. In addition, being an outlier does not necessarily indicate a problem: a hospital unit for organ transplants will have high rates of antibiotic use, which is likely appropriate.

The CDC has taken initial steps to develop a risk-adjusted benchmark measure for hospital antibiotic use, the Standardized Antimicrobial Administration Ratio (SAAR). It compares a hospital’s observed antibiotic use with a calculation of predicted use based on its facility characteristics. Although still at an early stage, SAAR has been released and has been endorsed by the National Quality Forum. About 200 hospitals are submitting data to the CDC and collaborating with the CDC to evaluate the SAAR’s utility in driving improved antibiotic use.

Problems in measuring appropriate use

Measuring appropriate antibiotic use is easier in the outpatient setting, where detailed data have been collected for many years.

Fleming-Dutra et al²¹ compared medications prescribed during outpatient visits and the diagnoses coded for the visits. They found that about 13% of all outpatient visits resulted in an antibiotic prescription, 30% of which had no listed diagnosis that would justify an antibiotic (eg, viral upper respiratory infection). This kind of information provides a target for stewardship programs.

It is more difficult to conduct such a study in a hospital setting. Simply comparing discharge diagnoses to antibiotics prescribed is not useful: often antibiotics are started presumptively on admission for a patient with signs and symptoms of an infection, then stopped if the diagnosis does not warrant antibiotics, which is a reasonable strategy.

Also, many times, a patient with asymptomatic bacteriuria, which does not warrant antibiotics, is misdiagnosed as having a urinary tract infection, which does. So simply looking at the discharge code may not reveal whether therapy was appropriate.

Some studies have provided useful information. Fridkin et al²² studied 36 hospitals for the use of vancomycin, which is an especially good candidate drug for study because guidelines exist for appropriate use. Data were collected only from patients given vancomycin for more than 3 days, which should have eliminated empiric use of the drug and included only pathogen-driven therapy. Cases where therapy was for skin and soft-tissue infections were excluded because cultures are not usually obtained for these cases. Of patients given vancomycin, 9% had no diagnostic culture obtained at antibiotic initiation, 22% had diagnostic culture but results showed no gram-positive bacterial growth, and 5% had culture results revealing only oxacillin-susceptible Staphylococcus aureus. In 36% of cases, opportunities existed for improved prescribing.

Such data could be collected from the electronic medical record, and the CDC is focusing efforts in this direction.

NATIONAL ACTIVITIES IN ANTIBIOTIC STEWARDSHIP

In 2014, the White House launched a national strategy to combat antibiotic resistance,²³ followed by an action plan in 2015.²⁴ As a result, new investments have been made to improve antibiotic use, including funding for state health departments to begin stewardship efforts and to expand public awareness of the problems of antibiotic overuse. Research efforts are also being funded to improve implementation of existing stewardship practices and to develop new ones.

CMS is also exploring how to drive improved antibiotic use. In October 2016, it started requiring all US nursing homes to have antibiotic stewardship programs, and a similar requirement for hospitals has been proposed.

The Joint Commission issued a standard requiring that all their accredited facilities, starting with hospitals, have an antibiotic stewardship program by January 2017. This standard requires implementation of all the CDC’s core elements.

PROVEN INTERVENTIONS

Focusing on key interventions that are likely to be effective and well received by providers is a useful strategy for antibiotic stewardship efforts. A number of such interventions have been supported by research.

Postprescription antibiotic reviews or antibiotic ‘time-outs’

Antibiotics are often started empirically to treat hospitalized patients suspected of having an infection. The need for the antibiotic should be assessed a few days later, when culture results and more clinical information are available.  

Elligsen et al²⁵ evaluated the effects of providing a formal review and suggestions for antimicrobial optimization to critical care teams of 3 intensive care units in a single hospital after 3 and 10 days of antibiotic therapy. Mean monthly antibiotic use decreased from 644 days of therapy per 1,000 patient-days in the preintervention period to 503 days of therapy per 1,000 patient-days (P < .0001). C difficile infections were reduced from 11 cases to 6. Overall gram-negative susceptibility to meropenem increased in the critical care units. 

Targeting specific infections

Some infections are especially important to target with improvement efforts.

In 2011, Magill et al²⁶  conducted 1-day prevalence surveys in 183 hospitals in 10 states to examine patterns of antibiotic use. They found that lower respiratory tract infections and urinary tract infections accounted for more than half of all antibiotic use (35% and 22%, respectively), making them good candidates for improved use.

Community-acquired pneumonia can be targeted at multiple fronts.  One study showed that almost 30% of patients diagnosed with community-acquired pneumonia in the emergency department did not actually have pneumonia.²⁷  Duration of antibiotic therapy could also be targeted. Guidelines recommend that most patients with uncomplicated community-acquired pneumonia receive 5 to 7 days of antibiotic therapy. Avdic et al²⁸ performed a simple intervention involving education and feedback to teams in 1 hospital regarding antibiotic choice and duration. This resulted in reducing the duration of therapy for community-acquired pneumonia from a median of 10 to 7 days.

Asymptomatic bacteriuria is often misdiagnosed as a urinary tract infection and treated unnecessarily.^29–31

Trautner et al³² addressed this problem by targeting urine cultures rather than antibiotics, using a simple algorithm: if a patient did not have symptoms of urinary tract infection (fever, acute hematuria, delirium, rigors, flank pain, pelvic discomfort, urgency, frequency, dysuria, suprapubic pain), a urine culture was not recommended. If a patient did have symptoms but a problem other than urinary tract infection was deemed likely, evaluation of other sources of infection was recommended. Use of the algorithm resulted in fewer urine cultures and less antibiotic overtreatment of asymptomatic bacteriuria. Reductions persisted after the intervention ended. 

Antibiotic time-out at hospital discharge

Another study evaluated an intervention that required a pharmacist consultation for the critical care team when a patient was to be discharged with intravenous antibiotics (most often for pneumonia). In 28% of cases, chart review revealed that the infection had been completely treated at the time of discharge, so further antibiotic treatment was not indicated. No patients who avoided antibiotics at discharge were readmitted or subsequently visited the emergency department.³³

Targeting outpatient settings

A number of studies have evaluated simple interventions to improve outpatient antibiotic prescribing. Meeker et al³⁴ had providers place a poster in their examination rooms with a picture of the physician and a signed letter committing to the appropriate use of antibiotics. Inappropriate antibiotic use decreased 20% in the intervention group vs controls (P = .02). 

In a subsequent study,³⁵ the same group required providers to include a justification note in the electronic medical record every time an antibiotic was prescribed for an indication when guidelines do not recommend one. Inappropriate prescribing dropped from 23% to 5% (P < .001).

Another intervention in this study³⁵ provided physicians with periodic feedback according to whether their therapy was concordant with guidelines. They received an email with a subject line of either “You are a top performer” or “You are not a top performer.” The contents of the email provided data on how many antibiotic prescriptions they wrote for conditions that did not warrant them and how their prescribing habits compared with those of their top-performing peers. Mean inappropriate antibiotic prescribing fell from 20% to 4%.³⁵

This is a critical time for antibiotic stewardship efforts in the United States. The need has never been more urgent and, fortunately, the opportunities have never been more abundant. Requirements for stewardship programs will drive implementation, but hospitals will need support and guidance to help ensure that stewardship programs are as effective as possible. Ultimately, improving antibiotic use will require collaboration among all stakeholders. CDC is eager to partner with providers and others in their efforts to improve antibiotic use.

A 43-year-old man presented to a community hospital with acute chest pain and shortness of breath and was diagnosed with anterior ST-elevation myocardial infarction. He was a smoker with a history of alcohol abuse, hypertension, and hyperlipidemia, and in the past he had undergone percutaneous coronary interventions to the right coronary artery and the first obtuse marginal artery.

Angiography showed total occlusion in the left anterior descending artery, 90% stenosis in the right coronary artery, and mild disease in the left circumflex artery. A drug-eluting stent was placed in the left anterior descending artery, resulting in good blood flow. 

However, his left ventricle continued to have severe dysfunction. An intra-aortic balloon pump was inserted. Afterward, computed tomography showed subsegmental pulmonary embolism with congestion. His mean arterial pressure was 60 mm Hg (normal 70–110), central venous pressure 12 mm Hg (3–8), pulmonary artery pressure 38/26 mm Hg (15–30/4–12), pulmonary capillary wedge pressure 24 mm Hg (2–15), and cardiac index 1.4 L/min (2.5–4).

The patient was started on dobutamine and norepinephrine and transferred to Cleveland Clinic on day 2. Over the next day, he had runs of ventricular tachycardia, for which he was given amiodarone and lidocaine. His urine output was low, and his serum creatinine was elevated at 1.65 mg/dL (baseline 1.2, normal 0.5–1.5). Liver function tests were also elevated, with aspartate aminotransferase at 115 U/L(14–40) and alanine aminotransferase at 187 U/L (10–54).

Poor oxygenation was evident: his arterial partial pressure of oxygen was 64 mm Hg (normal 75–100). He was intubated and given 100% oxygen with positive end-expiratory pressure of 12 cm H₂O.

Echocardiography showed a left ventricular ejection fraction of 15% (normal 55%–70%) and mild right ventricular dysfunction.

ECMO and then Impella placement

On his third hospital day, a venoarterial extracorporeal membrane oxygenation (ECMO) device was placed peripherally (Figure 1).

His hemodynamic variables stabilized, and he was weaned off dobutamine and norepinephrine. Results of liver function tests normalized, his urinary output increased, and his serum creatinine dropped to a normal 1.0 mg/dL. However, a chest radiograph showed pulmonary congestion, and echocardiography now showed severe left ventricular dysfunction.

On hospital day 5, the patient underwent surgical placement of an Impella 5.0 device (Abiomed, Danvers, MA) through the right axillary artery in an effort to improve his pulmonary edema. The ECMO device was removed. Placement of a venovenous ECMO device was deemed unnecessary when oxygenation improved with the Impella.

Three days after Impella placement, radiography showed improved edema with some remaining pleural effusion.

ACUTE CARDIOGENIC SHOCK

Cardiogenic shock remains a challenging clinical problem: patients with it are among the sickest in the hospital, and many of them die. ECMO was once the only therapy available and is still widely used. However, it is a 2-edged sword; complications such as bleeding, infection, and thrombosis are almost inevitable if it is used for long. Importantly, patients are usually kept intubated and bedridden.

In recent years, new devices have become available that are easier to place (some in the catheterization laboratory or even at the bedside) and allow safer bridging to recovery, transplant, or other therapies.

This case illustrates the natural history of cardiogenic shock and the preferred clinical approach: ie, ongoing evaluation that permits rapid response to evolving challenges.

In general, acute cardiogenic shock occurs within 24 to 48 hours after the initial insult, so even if a procedure succeeds, the patient may develop progressive hypotension and organ dysfunction. Reduced cardiac output causes a downward spiral with multiple systemic and inflammatory processes as well as increased nitric oxide synthesis, leading to progressive decline and eventual end-organ dysfunction.

Continuously evaluate

The cardiac team should continuously assess the acuity and severity of a patient’s condition, with the goals of maintaining end-organ perfusion and identifying the source of problems. Refractory cardiogenic shock, with tissue hypoperfusion despite vasoactive medications and treatment of the underlying cause, is associated with in-hospital mortality rates ranging from 30% to 50%.^1,2 The rates have actually increased over the past decade, as sicker patients are being treated.

When a patient presents with cardiogenic shock, we first try a series of vasoactive drugs and usually an intra-aortic balloon pump (Figure 2). We then tailor treatment depending on etiology. For example, a patient may have viral myocarditis and may even require a biopsy.

If cardiogenic shock is refractory, mechanical circulatory support devices can be a short-term bridge to either recovery or a new decision. A multidisciplinary team should be consulted to consider transplant, a long-term device, or palliative care. Sometimes a case requires “bridging to a bridge,” with several devices used short-term in turn.

Prognostic factors in cardiogenic shock

Several tools help predict outcome in a severely ill patient. End-organ function, indicated by blood lactate levels and estimated glomerular filtration rate, is perhaps the most informative and should be monitored serially.

CardShock³ is a simple scoring system based on age, mental status at presentation, laboratory values, and medical history. Patients receive 1 point for each of the following factors:

• Age > 75
• Confusion at presentation
• Previous myocardial infarction or coronary artery bypass grafting
• Acute coronary syndrome etiology
• Left ventricular ejection fraction < 40%
• Blood lactate level between 2 and 4 mmol/L, inclusively (2 points for lactate levels > 4 mmol/L)
• Estimated glomerular filtration rate between 30 and 60 mL/min/1.73 m², inclusively (2 points if < 30 mL/min/1.73 m²). 

Thus, scores range from 0 (best) to 9 (worst). A score of 0 to 3 points was associated with a 9% risk of death in the hospital, a score of 4 or 5 with a risk of 36%, and a score of 6 through 9 with a risk of 77%.³

The Survival After Veno-arterial ECMO (SAVE) score (www.save-score.com) is a prediction tool derived from a large international ECMO registry.⁴ It is based on patient age, diagnosis, and indicators of end-organ dysfunction. Scores range from –35 (worst) to +7 (best).

The mortality rate associated with postcardiotomy cardiogenic shock increases with the amount of inotropic support provided. In a 1996–1999 case series of patients who underwent open-heart surgery,⁵ the hospital mortality rate was 40% in those who received 2 inotropes in high doses and 80% in those who received 3. A strategy of early implementation of mechanical support is critical.

Selection criteria for destination therapy

Deciding whether a patient should receive a long-term device is frequently a challenge. The decision often must be based on limited information about not only the medical indications but also psychosocial factors that influence long-term success.

The Centers for Medicare and Medicaid Services have established criteria for candidates for left ventricular assist devices (LVADs) as destination therapy.⁶ Contraindications established for heart transplant should also be considered (Table 1).

CASE REVISITED

Several factors argued against LVAD placement in our patient. He had no health insurance and had been off medications. He smoked and said he consumed 3 hard liquor drinks per week. His Stanford Integrated Psychosocial Assessment for Transplantation score was 30 (minimally acceptable). He had hypoxia with subsegmental pulmonary edema, a strong contraindication to immediate transplant.

On the other hand, he had only mild right ventricular dysfunction. His CardShock score was 4 (intermediate risk, based on lactate 1.5 mmol/L and estimated glomerular filtration rate 52 mL/min/1.73 m²). His SAVE score was –9 (class IV), which overall is associated with a 30% risk of death (low enough to consider treatment).

During the patient’s time on temporary support, the team had the opportunity to better understand him and assess his family support and his ability to handle a permanent device. His surviving the acute course bolstered the team’s confidence that he could enjoy long-term survival with destination therapy.

CATHETERIZATION LABORATORY DEVICE CAPABILITIES

Although most implantation procedures are done in the operating room, they are often done in the catheterization laboratory because patients undergoing catheterization may not be stable enough for transfer, or an emergency intervention may be required during the night. Catheterization interventionists are also an important part of the team to help determine the best approach for long-term therapy.

The catheterization laboratory has multiple acute intervention options. Usually, decisions must be made quickly. In general, patients needing mechanical support are managed as follows:

• Those who need circulation support and oxygenation receive ECMO
• Those who need circulation support alone because of mechanical issues (eg, myocardial infarction) are considered for an intra-aortic balloon pump, Impella, or TandemHeart pump (Cardiac Assist, Pittsburgh, PA).

Factors that guide the selection of a temporary pump include:

• Left ventricular function
• Right ventricular function
• Aortic valve stenosis (some devices cannot be inserted through critical aortic stenosis)
• Aortic regurgitation (can affect some devices)
• Peripheral artery disease (some devices are large and must be placed percutaneously).

CHOOSING AMONG PERCUTANEOUS DEVICES

Circulatory support in cardiogenic shock improves outcomes, and devices play an important role in supporting high-risk procedures. The goal is not necessarily to use the device throughout the hospital stay. Acute stabilization is most important initially; a more considered decision about long-term therapy can be made when more is known about the patient.

Patient selection is the most important component of success. However, randomized data to support outcomes with the various devices are sparse and complicated by the critically ill state of the patient population.

SHORT-TERM CIRCULATORY SUPPORT: ECMO, IMPELLA, TANDEMHEART

A menu of options is available for temporary mechanical support. Options differ by their degree of circulatory support and ease of insertion (Table 2).

ECMO: A fast option with many advantages

ECMO has evolved and now can be placed quickly. A remote diagnostic platform such as CardioHub permits management at the bedside, in the medical unit, or in the cardiac intensive care unit.⁷

ECMO has several advantages. It can be used during cardiopulmonary bypass, it provides oxygenation, it is the only option in the setting of lung injury, it can be placed peripherally (without thoracotomy), and it is the only percutaneous option for biventricular support.

ECMO also has significant disadvantages

ECMO is a good device for acute resuscitation of a patient in shock, as it offers quick placement and resuscitation. But it is falling out of favor because of significant disadvantages.

Its major drawback is that it provides no left ventricular unloading. Although in a very unstable patient ECMO can stabilize end organs and restore their function, the lack of left ventricular unloading and reduced ventricular work threaten the myocardium. It creates extremely high afterload; therefore, in a left ventricle with poor function, wall tension and myocardial oxygen demand increase. Multiple studies have shown that coronary perfusion worsens, especially if the patient is cannulated peripherally. Because relative cerebral hypoxia occurs in many situations, it is imperative to check blood saturations at multiple sites to determine if perfusion is adequate everywhere.    

Ineffective left ventricular unloading with venoarterial ECMO is managed in several ways. Sometimes left ventricular distention is slight and the effects are subtle. Left ventricular distention causing pulmonary edema can be addressed with:

• Inotropes (in moderate doses)
• Anticoagulation to prevent left ventricular thrombus formation
• An intra-aortic balloon pump. Most patients on ECMO already have an intra-aortic balloon pump in place, and it should be left in to provide additional support. For those who do not have one, it should be placed via the contralateral femoral artery.

If problems persist despite these measures, apical cannulation or left ventricular septostomy can be performed.

Outcomes with ECMO have been disappointing. Studies show that whether ECMO was indicated for cardiac failure or for respiratory failure, survival is only about 25% at 5 years. Analyzing data only for arteriovenous ECMO, survival was 48% in bridged patients and 41% in patients who were weaned.

The Extracorporeal Life Support Organization Registry, in their international summary from 2010, found that 34% of cardiac patients on ECMO survived to discharge or transfer. Most of these patients had cardiogenic shock from acute myocardial infarction. Outcomes are so poor because of complications endemic to ECMO, eg, dialysis-dependent renal failure (about 40%) and neurologic complications (about 30%), often involving ischemic or hemorrhagic stroke.

Limb and pump complications were also significant in the past. These have been reduced with the new reperfusion cannula and the Quadrox oxygenator.  

Complications unique to ECMO should be understood and anticipated so that they can be avoided. Better tools are available, ie, Impella and TandemHeart.

Left-sided Impella: A longer-term temporary support

ECMO is a temporary fix that is usually used only for a few days. If longer support is needed, axillary placement of an Impella should be used as a bridge to recovery, transplant, or a durable LVAD.

The Impella device (Figure 3) is a miniature rotary blood pump increasingly used to treat cardiogenic shock. It is inserted retrograde across the aortic valve to provide short-term ventricular support. Most devices are approved by the US Food and Drug Administration (FDA) for less than 7 days of use, but we have experience using them up to 30 days. They are very hemocompatible, involving minimal hemolysis. Axillary placement allows early extubation and ambulation and is more stable than groin placement.

Several models are available: the 2.5 and 3.5 L/min devices can be placed percutaneously, while the 5 L/min model must be surgically placed in the axillary or groin region. Heparin is required with their use. They can replace ECMO. A right ventricular assist device (RVAD), Impella RP, is also available.

Physiologic impact of the Impella

The Impella fully unloads the left ventricle, reducing myocardial oxygen demand and increasing myocardial blood flow. It reduces end-diastolic volume and pressure, the mechanical work of the heart, and wall tension. Microvascular resistance is reduced, allowing increased coronary flow. Cardiac output and power are increased by multiple means.^8–11

The RECOVER 1 trial evaluated the 5L Impella placed after cardiac surgery. The cardiac index increased in all the patients, and the systemic vascular resistance and wedge pressure decreased.¹²

Unloading the ventricle is critical. Meyns  and colleagues¹³ found a fivefold reduction in infarct size from baseline in a left anterior descending occlusion model in pigs after off-loading the ventricle.

Impella has the advantage of simple percutaneous insertion (the 2.5 and CP models). It also tests right ventricular tolerance: if the right ventricle is doing well, one can predict with high certainty that it will tolerate an LVAD (eg, HeartWare, HeartMate 2 (Pleasanton, CA), or HeartMate 3 when available).

Disadvantages include that it provides only left ventricular support, although a right ventricular device can be inserted for dual support. Placement requires fluoroscopic or echocardiographic guidance.

TandemHeart requires septal puncture

The TandemHeart is approved for short-term and biventricular use. It consists of an extracorporeal centrifugal pump that withdraws blood from the left atrium via a trans-septal cannula placed through the femoral vein (Figure 4) and returns it to one or both femoral arteries. The blood is pumped at up to 5 L/min.

It is designed to reduce the pulmonary capillary wedge pressure, ventricular work, and myocardial oxygen demand and increase cardiac output and mean arterial pressure. It has the advantages of percutaneous placement and the ability to provide biventricular support with 2 devices. It can be used for up to 3 weeks. It can easily be converted to ECMO by either splicing in an oxygenator or adding another cannula.

Although the TandemHeart provides significant support, it is no longer often used. A 21F venous cannula must be passed to the left atrium by trans-septal puncture, which requires advanced skill and must be done in the catheterization laboratory. Insertion can take too much time and cause bleeding in patients taking an anticoagulant. Insertion usually destroys the septum, and removal requires a complete patch of the entire septum. Systemic anticoagulation is required. Other disadvantages are risks of hemolysis, limb ischemia, and infection with longer support times.

The CentriMag (Levitronix LLC; Framingham, MA) is an improved device that requires only 1 cannula instead of 2 to cover both areas.

DEVICES FOR RIGHT-SIDED SUPPORT

Most early devices were designed for left-sided support. The right heart, especially in failure, has been more difficult to manage. Previously the only option for a patient with right ventricular failure was venoarterial ECMO. This is more support than needed for a patient with isolated right ventricular failure and involves the risk of multiple complications from the device.

With more options available for the right heart (Table 3), we can choose the most appropriate device according to the underlying cause of right heart failure (eg, right ventricular infarct, pulmonary hypertension), the likelihood of recovery, and the expected time to recovery.

The ideal RVAD would be easy to implant, maintain, and remove. It would allow for chest closure and patient ambulation. It would be durable and biocompatible, so that it could remain implanted for months if necessary. It would cause little blood trauma, have the capability for adding an oxygenator for pulmonary support, and be cost-effective.

Although no single system has all these qualities, each available device fulfills certain combinations of these criteria, so the best one can be selected for each patient’s needs.

ECMO Rotaflow centrifugal pump: Fast, simple, inexpensive

A recent improvement to ECMO is the Rotaflow centrifugal pump (Maquet, Wayne, NJ), which is connected by sewing an 8-mm graft onto the pulmonary artery and placing a venous cannula in the femoral vein. If the patient is not bleeding, the chest can then be closed. This creates a fast, simple, and inexpensive temporary RVAD system. When the patient is ready to be weaned, the outflow graft can be disconnected at the bedside without reopening the chest.

The disadvantage is that the Rotaflow system contains a sapphire bearing. Although it is magnetically coupled, it generates heat and is a nidus for thrombus formation, which can lead to pump failure and embolization. This system can be used for patients who are expected to need support for less than 5 to 7 days. Beyond this duration, the incidence of complications increases. 

CentriMag Ventricular Assist System offers right, left, or bilateral support

The CentriMag Ventricular Assist System is a fully magnetically levitated pump containing no bearings or seals, and with the same technology as is found in many of the durable devices such as HeartMate 3. It is coupled with a reusable motor and is easy to use.

CentriMag offers versatility, allowing for right, left, or bilateral ventricular support. An oxygenator can be added for pulmonary edema and additional support. It is the most biocompatible device and is FDA-approved for use for 4 weeks, although it has been used successfully for much longer. It allows for chest closure and ambulation. It is especially important as a bridge to transplant. The main disadvantage is that insertion and removal require sternotomy.

Impella RP: One size does not fit all

The Impella RP (Figure 5) has an 11F catheter diameter, 23F pump, and a maximum flow rate of more than 4 L/minute. It has a unique 3-dimensional cannula design based on computed tomography 3-dimensional reconstructions from hundreds of patients.

The device is biocompatible and can be used for support for more than 7 days, although most patients require only 3 or 4 days. There is almost no priming volume, so there is no hemodilution.

The disadvantages are that it is more challenging to place than other devices, and some patients cannot use it because the cannula does not fit. It also does not provide pulmonary support. Finally, it is the most expensive of the 3 right-sided devices.

CASE REVISITED

The patient described at the beginning of this article was extubated on day 12 but was then reintubated. On day 20, a tracheotomy tube was placed. By day 24, he had improved so little that his family signed a “do-not-resuscitate–comfort-care-arrest” order (ie, if the patient’s heart or breathing stops, only comfort care is to be provided).

But slowly he got better, and the Impella was removed on day 30. Afterward, serum creatinine and liver function tests began rising again, requiring dobutamine for heart support.

On day 34, his family reversed the do-not-resuscitate order, and he was reevaluated for an LVAD as destination therapy. At this point, echocardiography showed a left ventricular ejection fraction of 10%, normal right ventricular function, with a normal heartbeat and valves. On day 47, a HeartMate II LVAD was placed.

On postoperative day 18, he was transferred out of the intensive care unit, then discharged to an acute rehabilitation facility 8 days later (hospital day 73). He was subsequently discharged.

At a recent follow-up appointment, the patient said that he was feeling “pretty good” and walked with no shortness of breath.

A 43-year-old man presented to a community hospital with acute chest pain and shortness of breath and was diagnosed with anterior ST-elevation myocardial infarction. He was a smoker with a history of alcohol abuse, hypertension, and hyperlipidemia, and in the past he had undergone percutaneous coronary interventions to the right coronary artery and the first obtuse marginal artery.

Angiography showed total occlusion in the left anterior descending artery, 90% stenosis in the right coronary artery, and mild disease in the left circumflex artery. A drug-eluting stent was placed in the left anterior descending artery, resulting in good blood flow. 

However, his left ventricle continued to have severe dysfunction. An intra-aortic balloon pump was inserted. Afterward, computed tomography showed subsegmental pulmonary embolism with congestion. His mean arterial pressure was 60 mm Hg (normal 70–110), central venous pressure 12 mm Hg (3–8), pulmonary artery pressure 38/26 mm Hg (15–30/4–12), pulmonary capillary wedge pressure 24 mm Hg (2–15), and cardiac index 1.4 L/min (2.5–4).

The patient was started on dobutamine and norepinephrine and transferred to Cleveland Clinic on day 2. Over the next day, he had runs of ventricular tachycardia, for which he was given amiodarone and lidocaine. His urine output was low, and his serum creatinine was elevated at 1.65 mg/dL (baseline 1.2, normal 0.5–1.5). Liver function tests were also elevated, with aspartate aminotransferase at 115 U/L(14–40) and alanine aminotransferase at 187 U/L (10–54).

Poor oxygenation was evident: his arterial partial pressure of oxygen was 64 mm Hg (normal 75–100). He was intubated and given 100% oxygen with positive end-expiratory pressure of 12 cm H₂O.

Echocardiography showed a left ventricular ejection fraction of 15% (normal 55%–70%) and mild right ventricular dysfunction.

ECMO and then Impella placement

On his third hospital day, a venoarterial extracorporeal membrane oxygenation (ECMO) device was placed peripherally (Figure 1).

His hemodynamic variables stabilized, and he was weaned off dobutamine and norepinephrine. Results of liver function tests normalized, his urinary output increased, and his serum creatinine dropped to a normal 1.0 mg/dL. However, a chest radiograph showed pulmonary congestion, and echocardiography now showed severe left ventricular dysfunction.

On hospital day 5, the patient underwent surgical placement of an Impella 5.0 device (Abiomed, Danvers, MA) through the right axillary artery in an effort to improve his pulmonary edema. The ECMO device was removed. Placement of a venovenous ECMO device was deemed unnecessary when oxygenation improved with the Impella.

Three days after Impella placement, radiography showed improved edema with some remaining pleural effusion.

ACUTE CARDIOGENIC SHOCK

Cardiogenic shock remains a challenging clinical problem: patients with it are among the sickest in the hospital, and many of them die. ECMO was once the only therapy available and is still widely used. However, it is a 2-edged sword; complications such as bleeding, infection, and thrombosis are almost inevitable if it is used for long. Importantly, patients are usually kept intubated and bedridden.

In recent years, new devices have become available that are easier to place (some in the catheterization laboratory or even at the bedside) and allow safer bridging to recovery, transplant, or other therapies.

This case illustrates the natural history of cardiogenic shock and the preferred clinical approach: ie, ongoing evaluation that permits rapid response to evolving challenges.

In general, acute cardiogenic shock occurs within 24 to 48 hours after the initial insult, so even if a procedure succeeds, the patient may develop progressive hypotension and organ dysfunction. Reduced cardiac output causes a downward spiral with multiple systemic and inflammatory processes as well as increased nitric oxide synthesis, leading to progressive decline and eventual end-organ dysfunction.

Continuously evaluate

The cardiac team should continuously assess the acuity and severity of a patient’s condition, with the goals of maintaining end-organ perfusion and identifying the source of problems. Refractory cardiogenic shock, with tissue hypoperfusion despite vasoactive medications and treatment of the underlying cause, is associated with in-hospital mortality rates ranging from 30% to 50%.^1,2 The rates have actually increased over the past decade, as sicker patients are being treated.

When a patient presents with cardiogenic shock, we first try a series of vasoactive drugs and usually an intra-aortic balloon pump (Figure 2). We then tailor treatment depending on etiology. For example, a patient may have viral myocarditis and may even require a biopsy.

If cardiogenic shock is refractory, mechanical circulatory support devices can be a short-term bridge to either recovery or a new decision. A multidisciplinary team should be consulted to consider transplant, a long-term device, or palliative care. Sometimes a case requires “bridging to a bridge,” with several devices used short-term in turn.

Prognostic factors in cardiogenic shock

Several tools help predict outcome in a severely ill patient. End-organ function, indicated by blood lactate levels and estimated glomerular filtration rate, is perhaps the most informative and should be monitored serially.

CardShock³ is a simple scoring system based on age, mental status at presentation, laboratory values, and medical history. Patients receive 1 point for each of the following factors:

• Age > 75
• Confusion at presentation
• Previous myocardial infarction or coronary artery bypass grafting
• Acute coronary syndrome etiology
• Left ventricular ejection fraction < 40%
• Blood lactate level between 2 and 4 mmol/L, inclusively (2 points for lactate levels > 4 mmol/L)
• Estimated glomerular filtration rate between 30 and 60 mL/min/1.73 m², inclusively (2 points if < 30 mL/min/1.73 m²). 

Thus, scores range from 0 (best) to 9 (worst). A score of 0 to 3 points was associated with a 9% risk of death in the hospital, a score of 4 or 5 with a risk of 36%, and a score of 6 through 9 with a risk of 77%.³

The Survival After Veno-arterial ECMO (SAVE) score (www.save-score.com) is a prediction tool derived from a large international ECMO registry.⁴ It is based on patient age, diagnosis, and indicators of end-organ dysfunction. Scores range from –35 (worst) to +7 (best).

The mortality rate associated with postcardiotomy cardiogenic shock increases with the amount of inotropic support provided. In a 1996–1999 case series of patients who underwent open-heart surgery,⁵ the hospital mortality rate was 40% in those who received 2 inotropes in high doses and 80% in those who received 3. A strategy of early implementation of mechanical support is critical.

Selection criteria for destination therapy

Deciding whether a patient should receive a long-term device is frequently a challenge. The decision often must be based on limited information about not only the medical indications but also psychosocial factors that influence long-term success.

The Centers for Medicare and Medicaid Services have established criteria for candidates for left ventricular assist devices (LVADs) as destination therapy.⁶ Contraindications established for heart transplant should also be considered (Table 1).

CASE REVISITED

Several factors argued against LVAD placement in our patient. He had no health insurance and had been off medications. He smoked and said he consumed 3 hard liquor drinks per week. His Stanford Integrated Psychosocial Assessment for Transplantation score was 30 (minimally acceptable). He had hypoxia with subsegmental pulmonary edema, a strong contraindication to immediate transplant.

On the other hand, he had only mild right ventricular dysfunction. His CardShock score was 4 (intermediate risk, based on lactate 1.5 mmol/L and estimated glomerular filtration rate 52 mL/min/1.73 m²). His SAVE score was –9 (class IV), which overall is associated with a 30% risk of death (low enough to consider treatment).

During the patient’s time on temporary support, the team had the opportunity to better understand him and assess his family support and his ability to handle a permanent device. His surviving the acute course bolstered the team’s confidence that he could enjoy long-term survival with destination therapy.

CATHETERIZATION LABORATORY DEVICE CAPABILITIES

Although most implantation procedures are done in the operating room, they are often done in the catheterization laboratory because patients undergoing catheterization may not be stable enough for transfer, or an emergency intervention may be required during the night. Catheterization interventionists are also an important part of the team to help determine the best approach for long-term therapy.

The catheterization laboratory has multiple acute intervention options. Usually, decisions must be made quickly. In general, patients needing mechanical support are managed as follows:

• Those who need circulation support and oxygenation receive ECMO
• Those who need circulation support alone because of mechanical issues (eg, myocardial infarction) are considered for an intra-aortic balloon pump, Impella, or TandemHeart pump (Cardiac Assist, Pittsburgh, PA).

Factors that guide the selection of a temporary pump include:

• Left ventricular function
• Right ventricular function
• Aortic valve stenosis (some devices cannot be inserted through critical aortic stenosis)
• Aortic regurgitation (can affect some devices)
• Peripheral artery disease (some devices are large and must be placed percutaneously).

CHOOSING AMONG PERCUTANEOUS DEVICES

Circulatory support in cardiogenic shock improves outcomes, and devices play an important role in supporting high-risk procedures. The goal is not necessarily to use the device throughout the hospital stay. Acute stabilization is most important initially; a more considered decision about long-term therapy can be made when more is known about the patient.

Patient selection is the most important component of success. However, randomized data to support outcomes with the various devices are sparse and complicated by the critically ill state of the patient population.

SHORT-TERM CIRCULATORY SUPPORT: ECMO, IMPELLA, TANDEMHEART

A menu of options is available for temporary mechanical support. Options differ by their degree of circulatory support and ease of insertion (Table 2).

ECMO: A fast option with many advantages

ECMO has evolved and now can be placed quickly. A remote diagnostic platform such as CardioHub permits management at the bedside, in the medical unit, or in the cardiac intensive care unit.⁷

ECMO has several advantages. It can be used during cardiopulmonary bypass, it provides oxygenation, it is the only option in the setting of lung injury, it can be placed peripherally (without thoracotomy), and it is the only percutaneous option for biventricular support.

ECMO also has significant disadvantages

ECMO is a good device for acute resuscitation of a patient in shock, as it offers quick placement and resuscitation. But it is falling out of favor because of significant disadvantages.

Its major drawback is that it provides no left ventricular unloading. Although in a very unstable patient ECMO can stabilize end organs and restore their function, the lack of left ventricular unloading and reduced ventricular work threaten the myocardium. It creates extremely high afterload; therefore, in a left ventricle with poor function, wall tension and myocardial oxygen demand increase. Multiple studies have shown that coronary perfusion worsens, especially if the patient is cannulated peripherally. Because relative cerebral hypoxia occurs in many situations, it is imperative to check blood saturations at multiple sites to determine if perfusion is adequate everywhere.    

Ineffective left ventricular unloading with venoarterial ECMO is managed in several ways. Sometimes left ventricular distention is slight and the effects are subtle. Left ventricular distention causing pulmonary edema can be addressed with:

• Inotropes (in moderate doses)
• Anticoagulation to prevent left ventricular thrombus formation
• An intra-aortic balloon pump. Most patients on ECMO already have an intra-aortic balloon pump in place, and it should be left in to provide additional support. For those who do not have one, it should be placed via the contralateral femoral artery.

If problems persist despite these measures, apical cannulation or left ventricular septostomy can be performed.

Outcomes with ECMO have been disappointing. Studies show that whether ECMO was indicated for cardiac failure or for respiratory failure, survival is only about 25% at 5 years. Analyzing data only for arteriovenous ECMO, survival was 48% in bridged patients and 41% in patients who were weaned.

The Extracorporeal Life Support Organization Registry, in their international summary from 2010, found that 34% of cardiac patients on ECMO survived to discharge or transfer. Most of these patients had cardiogenic shock from acute myocardial infarction. Outcomes are so poor because of complications endemic to ECMO, eg, dialysis-dependent renal failure (about 40%) and neurologic complications (about 30%), often involving ischemic or hemorrhagic stroke.

Limb and pump complications were also significant in the past. These have been reduced with the new reperfusion cannula and the Quadrox oxygenator.  

Complications unique to ECMO should be understood and anticipated so that they can be avoided. Better tools are available, ie, Impella and TandemHeart.

Left-sided Impella: A longer-term temporary support

ECMO is a temporary fix that is usually used only for a few days. If longer support is needed, axillary placement of an Impella should be used as a bridge to recovery, transplant, or a durable LVAD.

The Impella device (Figure 3) is a miniature rotary blood pump increasingly used to treat cardiogenic shock. It is inserted retrograde across the aortic valve to provide short-term ventricular support. Most devices are approved by the US Food and Drug Administration (FDA) for less than 7 days of use, but we have experience using them up to 30 days. They are very hemocompatible, involving minimal hemolysis. Axillary placement allows early extubation and ambulation and is more stable than groin placement.

Several models are available: the 2.5 and 3.5 L/min devices can be placed percutaneously, while the 5 L/min model must be surgically placed in the axillary or groin region. Heparin is required with their use. They can replace ECMO. A right ventricular assist device (RVAD), Impella RP, is also available.

Physiologic impact of the Impella

The Impella fully unloads the left ventricle, reducing myocardial oxygen demand and increasing myocardial blood flow. It reduces end-diastolic volume and pressure, the mechanical work of the heart, and wall tension. Microvascular resistance is reduced, allowing increased coronary flow. Cardiac output and power are increased by multiple means.^8–11

The RECOVER 1 trial evaluated the 5L Impella placed after cardiac surgery. The cardiac index increased in all the patients, and the systemic vascular resistance and wedge pressure decreased.¹²

Unloading the ventricle is critical. Meyns  and colleagues¹³ found a fivefold reduction in infarct size from baseline in a left anterior descending occlusion model in pigs after off-loading the ventricle.

Impella has the advantage of simple percutaneous insertion (the 2.5 and CP models). It also tests right ventricular tolerance: if the right ventricle is doing well, one can predict with high certainty that it will tolerate an LVAD (eg, HeartWare, HeartMate 2 (Pleasanton, CA), or HeartMate 3 when available).

Disadvantages include that it provides only left ventricular support, although a right ventricular device can be inserted for dual support. Placement requires fluoroscopic or echocardiographic guidance.

TandemHeart requires septal puncture

The TandemHeart is approved for short-term and biventricular use. It consists of an extracorporeal centrifugal pump that withdraws blood from the left atrium via a trans-septal cannula placed through the femoral vein (Figure 4) and returns it to one or both femoral arteries. The blood is pumped at up to 5 L/min.

It is designed to reduce the pulmonary capillary wedge pressure, ventricular work, and myocardial oxygen demand and increase cardiac output and mean arterial pressure. It has the advantages of percutaneous placement and the ability to provide biventricular support with 2 devices. It can be used for up to 3 weeks. It can easily be converted to ECMO by either splicing in an oxygenator or adding another cannula.

Although the TandemHeart provides significant support, it is no longer often used. A 21F venous cannula must be passed to the left atrium by trans-septal puncture, which requires advanced skill and must be done in the catheterization laboratory. Insertion can take too much time and cause bleeding in patients taking an anticoagulant. Insertion usually destroys the septum, and removal requires a complete patch of the entire septum. Systemic anticoagulation is required. Other disadvantages are risks of hemolysis, limb ischemia, and infection with longer support times.

The CentriMag (Levitronix LLC; Framingham, MA) is an improved device that requires only 1 cannula instead of 2 to cover both areas.

DEVICES FOR RIGHT-SIDED SUPPORT

Most early devices were designed for left-sided support. The right heart, especially in failure, has been more difficult to manage. Previously the only option for a patient with right ventricular failure was venoarterial ECMO. This is more support than needed for a patient with isolated right ventricular failure and involves the risk of multiple complications from the device.

With more options available for the right heart (Table 3), we can choose the most appropriate device according to the underlying cause of right heart failure (eg, right ventricular infarct, pulmonary hypertension), the likelihood of recovery, and the expected time to recovery.

The ideal RVAD would be easy to implant, maintain, and remove. It would allow for chest closure and patient ambulation. It would be durable and biocompatible, so that it could remain implanted for months if necessary. It would cause little blood trauma, have the capability for adding an oxygenator for pulmonary support, and be cost-effective.

Although no single system has all these qualities, each available device fulfills certain combinations of these criteria, so the best one can be selected for each patient’s needs.

ECMO Rotaflow centrifugal pump: Fast, simple, inexpensive

A recent improvement to ECMO is the Rotaflow centrifugal pump (Maquet, Wayne, NJ), which is connected by sewing an 8-mm graft onto the pulmonary artery and placing a venous cannula in the femoral vein. If the patient is not bleeding, the chest can then be closed. This creates a fast, simple, and inexpensive temporary RVAD system. When the patient is ready to be weaned, the outflow graft can be disconnected at the bedside without reopening the chest.

The disadvantage is that the Rotaflow system contains a sapphire bearing. Although it is magnetically coupled, it generates heat and is a nidus for thrombus formation, which can lead to pump failure and embolization. This system can be used for patients who are expected to need support for less than 5 to 7 days. Beyond this duration, the incidence of complications increases. 

CentriMag Ventricular Assist System offers right, left, or bilateral support

The CentriMag Ventricular Assist System is a fully magnetically levitated pump containing no bearings or seals, and with the same technology as is found in many of the durable devices such as HeartMate 3. It is coupled with a reusable motor and is easy to use.

CentriMag offers versatility, allowing for right, left, or bilateral ventricular support. An oxygenator can be added for pulmonary edema and additional support. It is the most biocompatible device and is FDA-approved for use for 4 weeks, although it has been used successfully for much longer. It allows for chest closure and ambulation. It is especially important as a bridge to transplant. The main disadvantage is that insertion and removal require sternotomy.

Impella RP: One size does not fit all

The Impella RP (Figure 5) has an 11F catheter diameter, 23F pump, and a maximum flow rate of more than 4 L/minute. It has a unique 3-dimensional cannula design based on computed tomography 3-dimensional reconstructions from hundreds of patients.

The device is biocompatible and can be used for support for more than 7 days, although most patients require only 3 or 4 days. There is almost no priming volume, so there is no hemodilution.

The disadvantages are that it is more challenging to place than other devices, and some patients cannot use it because the cannula does not fit. It also does not provide pulmonary support. Finally, it is the most expensive of the 3 right-sided devices.

CASE REVISITED

The patient described at the beginning of this article was extubated on day 12 but was then reintubated. On day 20, a tracheotomy tube was placed. By day 24, he had improved so little that his family signed a “do-not-resuscitate–comfort-care-arrest” order (ie, if the patient’s heart or breathing stops, only comfort care is to be provided).

But slowly he got better, and the Impella was removed on day 30. Afterward, serum creatinine and liver function tests began rising again, requiring dobutamine for heart support.

On day 34, his family reversed the do-not-resuscitate order, and he was reevaluated for an LVAD as destination therapy. At this point, echocardiography showed a left ventricular ejection fraction of 10%, normal right ventricular function, with a normal heartbeat and valves. On day 47, a HeartMate II LVAD was placed.

On postoperative day 18, he was transferred out of the intensive care unit, then discharged to an acute rehabilitation facility 8 days later (hospital day 73). He was subsequently discharged.

At a recent follow-up appointment, the patient said that he was feeling “pretty good” and walked with no shortness of breath.

A 43-year-old man presented to a community hospital with acute chest pain and shortness of breath and was diagnosed with anterior ST-elevation myocardial infarction. He was a smoker with a history of alcohol abuse, hypertension, and hyperlipidemia, and in the past he had undergone percutaneous coronary interventions to the right coronary artery and the first obtuse marginal artery.

Angiography showed total occlusion in the left anterior descending artery, 90% stenosis in the right coronary artery, and mild disease in the left circumflex artery. A drug-eluting stent was placed in the left anterior descending artery, resulting in good blood flow. 

However, his left ventricle continued to have severe dysfunction. An intra-aortic balloon pump was inserted. Afterward, computed tomography showed subsegmental pulmonary embolism with congestion. His mean arterial pressure was 60 mm Hg (normal 70–110), central venous pressure 12 mm Hg (3–8), pulmonary artery pressure 38/26 mm Hg (15–30/4–12), pulmonary capillary wedge pressure 24 mm Hg (2–15), and cardiac index 1.4 L/min (2.5–4).

The patient was started on dobutamine and norepinephrine and transferred to Cleveland Clinic on day 2. Over the next day, he had runs of ventricular tachycardia, for which he was given amiodarone and lidocaine. His urine output was low, and his serum creatinine was elevated at 1.65 mg/dL (baseline 1.2, normal 0.5–1.5). Liver function tests were also elevated, with aspartate aminotransferase at 115 U/L(14–40) and alanine aminotransferase at 187 U/L (10–54).

Poor oxygenation was evident: his arterial partial pressure of oxygen was 64 mm Hg (normal 75–100). He was intubated and given 100% oxygen with positive end-expiratory pressure of 12 cm H₂O.

Echocardiography showed a left ventricular ejection fraction of 15% (normal 55%–70%) and mild right ventricular dysfunction.

ECMO and then Impella placement

On his third hospital day, a venoarterial extracorporeal membrane oxygenation (ECMO) device was placed peripherally (Figure 1).

His hemodynamic variables stabilized, and he was weaned off dobutamine and norepinephrine. Results of liver function tests normalized, his urinary output increased, and his serum creatinine dropped to a normal 1.0 mg/dL. However, a chest radiograph showed pulmonary congestion, and echocardiography now showed severe left ventricular dysfunction.

On hospital day 5, the patient underwent surgical placement of an Impella 5.0 device (Abiomed, Danvers, MA) through the right axillary artery in an effort to improve his pulmonary edema. The ECMO device was removed. Placement of a venovenous ECMO device was deemed unnecessary when oxygenation improved with the Impella.

Three days after Impella placement, radiography showed improved edema with some remaining pleural effusion.

ACUTE CARDIOGENIC SHOCK

Cardiogenic shock remains a challenging clinical problem: patients with it are among the sickest in the hospital, and many of them die. ECMO was once the only therapy available and is still widely used. However, it is a 2-edged sword; complications such as bleeding, infection, and thrombosis are almost inevitable if it is used for long. Importantly, patients are usually kept intubated and bedridden.

In recent years, new devices have become available that are easier to place (some in the catheterization laboratory or even at the bedside) and allow safer bridging to recovery, transplant, or other therapies.

This case illustrates the natural history of cardiogenic shock and the preferred clinical approach: ie, ongoing evaluation that permits rapid response to evolving challenges.

In general, acute cardiogenic shock occurs within 24 to 48 hours after the initial insult, so even if a procedure succeeds, the patient may develop progressive hypotension and organ dysfunction. Reduced cardiac output causes a downward spiral with multiple systemic and inflammatory processes as well as increased nitric oxide synthesis, leading to progressive decline and eventual end-organ dysfunction.

Continuously evaluate

The cardiac team should continuously assess the acuity and severity of a patient’s condition, with the goals of maintaining end-organ perfusion and identifying the source of problems. Refractory cardiogenic shock, with tissue hypoperfusion despite vasoactive medications and treatment of the underlying cause, is associated with in-hospital mortality rates ranging from 30% to 50%.^1,2 The rates have actually increased over the past decade, as sicker patients are being treated.

When a patient presents with cardiogenic shock, we first try a series of vasoactive drugs and usually an intra-aortic balloon pump (Figure 2). We then tailor treatment depending on etiology. For example, a patient may have viral myocarditis and may even require a biopsy.

If cardiogenic shock is refractory, mechanical circulatory support devices can be a short-term bridge to either recovery or a new decision. A multidisciplinary team should be consulted to consider transplant, a long-term device, or palliative care. Sometimes a case requires “bridging to a bridge,” with several devices used short-term in turn.

Prognostic factors in cardiogenic shock

Several tools help predict outcome in a severely ill patient. End-organ function, indicated by blood lactate levels and estimated glomerular filtration rate, is perhaps the most informative and should be monitored serially.

CardShock³ is a simple scoring system based on age, mental status at presentation, laboratory values, and medical history. Patients receive 1 point for each of the following factors:

• Age > 75
• Confusion at presentation
• Previous myocardial infarction or coronary artery bypass grafting
• Acute coronary syndrome etiology
• Left ventricular ejection fraction < 40%
• Blood lactate level between 2 and 4 mmol/L, inclusively (2 points for lactate levels > 4 mmol/L)
• Estimated glomerular filtration rate between 30 and 60 mL/min/1.73 m², inclusively (2 points if < 30 mL/min/1.73 m²). 

Thus, scores range from 0 (best) to 9 (worst). A score of 0 to 3 points was associated with a 9% risk of death in the hospital, a score of 4 or 5 with a risk of 36%, and a score of 6 through 9 with a risk of 77%.³

The Survival After Veno-arterial ECMO (SAVE) score (www.save-score.com) is a prediction tool derived from a large international ECMO registry.⁴ It is based on patient age, diagnosis, and indicators of end-organ dysfunction. Scores range from –35 (worst) to +7 (best).

The mortality rate associated with postcardiotomy cardiogenic shock increases with the amount of inotropic support provided. In a 1996–1999 case series of patients who underwent open-heart surgery,⁵ the hospital mortality rate was 40% in those who received 2 inotropes in high doses and 80% in those who received 3. A strategy of early implementation of mechanical support is critical.

Selection criteria for destination therapy

Deciding whether a patient should receive a long-term device is frequently a challenge. The decision often must be based on limited information about not only the medical indications but also psychosocial factors that influence long-term success.

The Centers for Medicare and Medicaid Services have established criteria for candidates for left ventricular assist devices (LVADs) as destination therapy.⁶ Contraindications established for heart transplant should also be considered (Table 1).

CASE REVISITED

Several factors argued against LVAD placement in our patient. He had no health insurance and had been off medications. He smoked and said he consumed 3 hard liquor drinks per week. His Stanford Integrated Psychosocial Assessment for Transplantation score was 30 (minimally acceptable). He had hypoxia with subsegmental pulmonary edema, a strong contraindication to immediate transplant.

On the other hand, he had only mild right ventricular dysfunction. His CardShock score was 4 (intermediate risk, based on lactate 1.5 mmol/L and estimated glomerular filtration rate 52 mL/min/1.73 m²). His SAVE score was –9 (class IV), which overall is associated with a 30% risk of death (low enough to consider treatment).

During the patient’s time on temporary support, the team had the opportunity to better understand him and assess his family support and his ability to handle a permanent device. His surviving the acute course bolstered the team’s confidence that he could enjoy long-term survival with destination therapy.

CATHETERIZATION LABORATORY DEVICE CAPABILITIES

Although most implantation procedures are done in the operating room, they are often done in the catheterization laboratory because patients undergoing catheterization may not be stable enough for transfer, or an emergency intervention may be required during the night. Catheterization interventionists are also an important part of the team to help determine the best approach for long-term therapy.

The catheterization laboratory has multiple acute intervention options. Usually, decisions must be made quickly. In general, patients needing mechanical support are managed as follows:

• Those who need circulation support and oxygenation receive ECMO
• Those who need circulation support alone because of mechanical issues (eg, myocardial infarction) are considered for an intra-aortic balloon pump, Impella, or TandemHeart pump (Cardiac Assist, Pittsburgh, PA).

Factors that guide the selection of a temporary pump include:

• Left ventricular function
• Right ventricular function
• Aortic valve stenosis (some devices cannot be inserted through critical aortic stenosis)
• Aortic regurgitation (can affect some devices)
• Peripheral artery disease (some devices are large and must be placed percutaneously).

CHOOSING AMONG PERCUTANEOUS DEVICES

Circulatory support in cardiogenic shock improves outcomes, and devices play an important role in supporting high-risk procedures. The goal is not necessarily to use the device throughout the hospital stay. Acute stabilization is most important initially; a more considered decision about long-term therapy can be made when more is known about the patient.

Patient selection is the most important component of success. However, randomized data to support outcomes with the various devices are sparse and complicated by the critically ill state of the patient population.

SHORT-TERM CIRCULATORY SUPPORT: ECMO, IMPELLA, TANDEMHEART

A menu of options is available for temporary mechanical support. Options differ by their degree of circulatory support and ease of insertion (Table 2).

ECMO: A fast option with many advantages

ECMO has evolved and now can be placed quickly. A remote diagnostic platform such as CardioHub permits management at the bedside, in the medical unit, or in the cardiac intensive care unit.⁷

ECMO has several advantages. It can be used during cardiopulmonary bypass, it provides oxygenation, it is the only option in the setting of lung injury, it can be placed peripherally (without thoracotomy), and it is the only percutaneous option for biventricular support.

ECMO also has significant disadvantages

ECMO is a good device for acute resuscitation of a patient in shock, as it offers quick placement and resuscitation. But it is falling out of favor because of significant disadvantages.

Its major drawback is that it provides no left ventricular unloading. Although in a very unstable patient ECMO can stabilize end organs and restore their function, the lack of left ventricular unloading and reduced ventricular work threaten the myocardium. It creates extremely high afterload; therefore, in a left ventricle with poor function, wall tension and myocardial oxygen demand increase. Multiple studies have shown that coronary perfusion worsens, especially if the patient is cannulated peripherally. Because relative cerebral hypoxia occurs in many situations, it is imperative to check blood saturations at multiple sites to determine if perfusion is adequate everywhere.    

Ineffective left ventricular unloading with venoarterial ECMO is managed in several ways. Sometimes left ventricular distention is slight and the effects are subtle. Left ventricular distention causing pulmonary edema can be addressed with:

• Inotropes (in moderate doses)
• Anticoagulation to prevent left ventricular thrombus formation
• An intra-aortic balloon pump. Most patients on ECMO already have an intra-aortic balloon pump in place, and it should be left in to provide additional support. For those who do not have one, it should be placed via the contralateral femoral artery.

If problems persist despite these measures, apical cannulation or left ventricular septostomy can be performed.

Outcomes with ECMO have been disappointing. Studies show that whether ECMO was indicated for cardiac failure or for respiratory failure, survival is only about 25% at 5 years. Analyzing data only for arteriovenous ECMO, survival was 48% in bridged patients and 41% in patients who were weaned.

The Extracorporeal Life Support Organization Registry, in their international summary from 2010, found that 34% of cardiac patients on ECMO survived to discharge or transfer. Most of these patients had cardiogenic shock from acute myocardial infarction. Outcomes are so poor because of complications endemic to ECMO, eg, dialysis-dependent renal failure (about 40%) and neurologic complications (about 30%), often involving ischemic or hemorrhagic stroke.

Limb and pump complications were also significant in the past. These have been reduced with the new reperfusion cannula and the Quadrox oxygenator.  

Complications unique to ECMO should be understood and anticipated so that they can be avoided. Better tools are available, ie, Impella and TandemHeart.

Left-sided Impella: A longer-term temporary support

ECMO is a temporary fix that is usually used only for a few days. If longer support is needed, axillary placement of an Impella should be used as a bridge to recovery, transplant, or a durable LVAD.

The Impella device (Figure 3) is a miniature rotary blood pump increasingly used to treat cardiogenic shock. It is inserted retrograde across the aortic valve to provide short-term ventricular support. Most devices are approved by the US Food and Drug Administration (FDA) for less than 7 days of use, but we have experience using them up to 30 days. They are very hemocompatible, involving minimal hemolysis. Axillary placement allows early extubation and ambulation and is more stable than groin placement.

Several models are available: the 2.5 and 3.5 L/min devices can be placed percutaneously, while the 5 L/min model must be surgically placed in the axillary or groin region. Heparin is required with their use. They can replace ECMO. A right ventricular assist device (RVAD), Impella RP, is also available.

Physiologic impact of the Impella

The Impella fully unloads the left ventricle, reducing myocardial oxygen demand and increasing myocardial blood flow. It reduces end-diastolic volume and pressure, the mechanical work of the heart, and wall tension. Microvascular resistance is reduced, allowing increased coronary flow. Cardiac output and power are increased by multiple means.^8–11

The RECOVER 1 trial evaluated the 5L Impella placed after cardiac surgery. The cardiac index increased in all the patients, and the systemic vascular resistance and wedge pressure decreased.¹²

Unloading the ventricle is critical. Meyns  and colleagues¹³ found a fivefold reduction in infarct size from baseline in a left anterior descending occlusion model in pigs after off-loading the ventricle.

Impella has the advantage of simple percutaneous insertion (the 2.5 and CP models). It also tests right ventricular tolerance: if the right ventricle is doing well, one can predict with high certainty that it will tolerate an LVAD (eg, HeartWare, HeartMate 2 (Pleasanton, CA), or HeartMate 3 when available).

Disadvantages include that it provides only left ventricular support, although a right ventricular device can be inserted for dual support. Placement requires fluoroscopic or echocardiographic guidance.

TandemHeart requires septal puncture

The TandemHeart is approved for short-term and biventricular use. It consists of an extracorporeal centrifugal pump that withdraws blood from the left atrium via a trans-septal cannula placed through the femoral vein (Figure 4) and returns it to one or both femoral arteries. The blood is pumped at up to 5 L/min.

It is designed to reduce the pulmonary capillary wedge pressure, ventricular work, and myocardial oxygen demand and increase cardiac output and mean arterial pressure. It has the advantages of percutaneous placement and the ability to provide biventricular support with 2 devices. It can be used for up to 3 weeks. It can easily be converted to ECMO by either splicing in an oxygenator or adding another cannula.

Although the TandemHeart provides significant support, it is no longer often used. A 21F venous cannula must be passed to the left atrium by trans-septal puncture, which requires advanced skill and must be done in the catheterization laboratory. Insertion can take too much time and cause bleeding in patients taking an anticoagulant. Insertion usually destroys the septum, and removal requires a complete patch of the entire septum. Systemic anticoagulation is required. Other disadvantages are risks of hemolysis, limb ischemia, and infection with longer support times.

The CentriMag (Levitronix LLC; Framingham, MA) is an improved device that requires only 1 cannula instead of 2 to cover both areas.

DEVICES FOR RIGHT-SIDED SUPPORT

Most early devices were designed for left-sided support. The right heart, especially in failure, has been more difficult to manage. Previously the only option for a patient with right ventricular failure was venoarterial ECMO. This is more support than needed for a patient with isolated right ventricular failure and involves the risk of multiple complications from the device.

With more options available for the right heart (Table 3), we can choose the most appropriate device according to the underlying cause of right heart failure (eg, right ventricular infarct, pulmonary hypertension), the likelihood of recovery, and the expected time to recovery.

The ideal RVAD would be easy to implant, maintain, and remove. It would allow for chest closure and patient ambulation. It would be durable and biocompatible, so that it could remain implanted for months if necessary. It would cause little blood trauma, have the capability for adding an oxygenator for pulmonary support, and be cost-effective.

Although no single system has all these qualities, each available device fulfills certain combinations of these criteria, so the best one can be selected for each patient’s needs.

ECMO Rotaflow centrifugal pump: Fast, simple, inexpensive

A recent improvement to ECMO is the Rotaflow centrifugal pump (Maquet, Wayne, NJ), which is connected by sewing an 8-mm graft onto the pulmonary artery and placing a venous cannula in the femoral vein. If the patient is not bleeding, the chest can then be closed. This creates a fast, simple, and inexpensive temporary RVAD system. When the patient is ready to be weaned, the outflow graft can be disconnected at the bedside without reopening the chest.

The disadvantage is that the Rotaflow system contains a sapphire bearing. Although it is magnetically coupled, it generates heat and is a nidus for thrombus formation, which can lead to pump failure and embolization. This system can be used for patients who are expected to need support for less than 5 to 7 days. Beyond this duration, the incidence of complications increases. 

CentriMag Ventricular Assist System offers right, left, or bilateral support

The CentriMag Ventricular Assist System is a fully magnetically levitated pump containing no bearings or seals, and with the same technology as is found in many of the durable devices such as HeartMate 3. It is coupled with a reusable motor and is easy to use.

CentriMag offers versatility, allowing for right, left, or bilateral ventricular support. An oxygenator can be added for pulmonary edema and additional support. It is the most biocompatible device and is FDA-approved for use for 4 weeks, although it has been used successfully for much longer. It allows for chest closure and ambulation. It is especially important as a bridge to transplant. The main disadvantage is that insertion and removal require sternotomy.

Impella RP: One size does not fit all

The Impella RP (Figure 5) has an 11F catheter diameter, 23F pump, and a maximum flow rate of more than 4 L/minute. It has a unique 3-dimensional cannula design based on computed tomography 3-dimensional reconstructions from hundreds of patients.

The device is biocompatible and can be used for support for more than 7 days, although most patients require only 3 or 4 days. There is almost no priming volume, so there is no hemodilution.

The disadvantages are that it is more challenging to place than other devices, and some patients cannot use it because the cannula does not fit. It also does not provide pulmonary support. Finally, it is the most expensive of the 3 right-sided devices.

CASE REVISITED

The patient described at the beginning of this article was extubated on day 12 but was then reintubated. On day 20, a tracheotomy tube was placed. By day 24, he had improved so little that his family signed a “do-not-resuscitate–comfort-care-arrest” order (ie, if the patient’s heart or breathing stops, only comfort care is to be provided).

But slowly he got better, and the Impella was removed on day 30. Afterward, serum creatinine and liver function tests began rising again, requiring dobutamine for heart support.

On day 34, his family reversed the do-not-resuscitate order, and he was reevaluated for an LVAD as destination therapy. At this point, echocardiography showed a left ventricular ejection fraction of 10%, normal right ventricular function, with a normal heartbeat and valves. On day 47, a HeartMate II LVAD was placed.

On postoperative day 18, he was transferred out of the intensive care unit, then discharged to an acute rehabilitation facility 8 days later (hospital day 73). He was subsequently discharged.

At a recent follow-up appointment, the patient said that he was feeling “pretty good” and walked with no shortness of breath.