Severity of side effects key to choosing between lithium and quetiapine in bipolar I and II

HOLLYWOOD, FLA. – Because there is no clinically significant difference in bipolar spectrum symptom relief offered by either lithium or quetiapine, which mood stabilizer clinicians choose to prescribe should largely depend on what side effects patients can tolerate.

That’s the conclusion of a panel of experts who presented data from the Bipolar CHOICE (Clinical Health Outcomes Initiative in Comparative Effectiveness for Bipolar Disorder) clinical trial at a meeting of the American Society of Clinical Psychopharmacology, formerly known as the New Clinical Drug Evaluation Unit meeting.

By randomly assigning the study’s 482 participating outpatient adults to either lithium or quetiapine, while allowing clinicians to prescribe adjunctive treatment as necessary; and by using a broader-than-usual set of inclusion criteria, the investigators hoped to provide clinicians with "real world" data on how the respective mood stabilizers perform.

"To my knowledge, this is the first randomized comparative effectiveness study of regular, expert care that looks at lithium versus quetiapine strategies of treatment, both with adjunctive personalized treatment when necessary," study coauthor Dr. Andrew A. Nierenberg told the audience in a well-attended session.

"We tortured the data to get a P value to pop out," said Dr. Nierenberg, who is the director of the Bipolar Clinic and Research Program at Massachusetts General Hospital in Boston. "The bottom line was that the treatment strategies, based on either lithium or quetiapine, were very similar. I think the clinical implication is that the choice of treatment really depends on someone’s tolerability. This may resurrect lithium as something [clinicians] should at least consider, because its use has gone down dramatically in the last 20 years or so."

Side effects such as metabolic or sleep disturbances were slightly less frequent in the lithium group, although the difference was not statistically significant; rates of suicidality and discontinuation of medication due to side effects were also similar across the two groups, according to Dr. David E. Kemp of the psychiatry department at Case Western Reserve University, Cleveland, and a study coauthor who also presented data during the session. Changes from baseline in mood stabilization were also virtually the same across the groups.

‘Rules of engagement’

The study enrolled adults aged 18-68 years who were at least mildly ill with either bipolar disorder I (68%) or II, with a Clinical Global Impression (CGI) score of at least 3, although the average CGI score was 4.5 (standard deviation of plus or minus 0.9). Of the 482 enrolled, 364 completed the study.

The so-called rules of engagement for clinicians from the 11 participating centers were that "they do everything they could to get their patients well," with the caveat that the lithium patient group not receive any antipsychotic, and the quetiapine patient group not receive lithium or any other antipsychotic.

"But they could get anything else that they needed," Dr. Nierenberg said.

At the time of enrollment, subjects were, in the estimation of their treating physician, experiencing symptoms that warranted a change in treatment, and that either lithium or quetiapine would be viable therapeutic options. Patients did not need to be lithium- or quetiapine-naive, but they could not have been treated with either drug in the previous 30 days. If they were already taking a second-generation antipsychotic, participants had to be willing to discontinue that medication and to be randomly assigned to either of the study drugs.

This strict adherence to one mood stabilizer, combined with adjunctive personalized treatments such as antidepressants or benzodiazepines, prescribed at the discretion of the treating clinician, ensured that the study tested the strategy of using either lithium or quetiapine as the base of treatment, Dr. Nierenberg said.

‘Similar’ results

Clinicians also were instructed to use the maximum tolerated dose of the respective mood stabilizer. For the 240 patients in the lithium group, the mean maximum tolerated dose was 1,007.5 mg, with the median dose being a "perfectly reasonable 900 mg," Dr. Nierenberg said. Blood lithium levels were taken at weeks 2, 16, and 24 and also were all found to be "reasonable," with most patients getting to at least 0.6 mEq/L by week 24.

For the 242 members of the quetiapine group, the maximum tolerated dose was 344.9 mg, with a median tolerated dose of 300 mg, which Dr. Nierenberg said was "what you would expect."

The overall result was that regardless of which mood stabilizer was prescribed along with adjunctive personalized treatment, the coprimary outcomes of the benefits and harms ratio from treatment effects over time (the CGI efficacy index), and the number of necessary clinical adjustments (changes in medication excluding titrations according to scale), were virtually parallel: For each primary outcome, the P value for the estimated change in baseline was less than .0001 for both lithium plus adjunctive treatment, and quetiapine plus adjunctive treatment.

Further, Dr. Nierenberg reported the "unexpected" result that at 6 months, about a quarter of each group was observed to be doing well without any adjunctive personalized treatments (23.8% of the lithium group; 27.3% of the quetiapine group; P = .14). "We thought that everybody would be on multiple things," Dr. Nierenberg said.

There was also no real difference in the time to discontinuation. "We were actually surprised that at 6 months, about three-quarters of the patients were still on the study drug, which was pretty good considering that this was a very ill population," Dr. Nierenberg said.

Using a CGI bipolar severity scale rating of up to 2, maintained for at least 8 weeks, the investigators found that overall, 20% were considered doing "really well"; it was also about 20% for the lithium plus adjunctive therapy group and the quetiapine and adjunctive therapy group (P = .14 for all three).

Because the Agency for Healthcare Research and Quality–funded study mandated that researchers end their investigation at precisely 36 months, Dr. Nierenberg said the study sites "randomized the 482 patients to 6 months of treatment within 36 months from start to finish," but he added that the investigators would have preferred to conduct a full 2-year study of participants.

"The good news was that most of the patients actually did improve substantially; the bad news is that while maybe a quarter did really well by the end of 6 months, we don’t know whether if we were to extend the study out further, we would see a difference between the two groups, whether they would continue to improve, or whether they would relapse," he said.

Adverse effects, predictors of response

Because study participants were followed for only 6 months, the long-term adverse effects of many years of exposure to either mood stabilizer, such as renal impairment with lithium or type II diabetes with quetiapine, were not considered, Dr. Kemp said. The adverse-effect profile for lithium generally includes a narrow therapeutic index; nausea, vomiting, and diarrhea; renal impairment; and hypothyroidism. Although Dr. Kemp said that the long-term adverse effects of quetiapine are understudied, known short-term side effects include the potential for sedation or somnolence, and weight gain.

Changes in baseline on the Frequency and Intensity of Side Effects Ratings for the quetiapine plus adjunctive personalized treatment arm were slightly more adverse than for the lithium and additional treatment arm. For the lithium plus adjunctive treatment group, the mean frequency of side effects was –1.41 (–1.78, –1.04 SD); the mean intensity of side effects was –1.48 (–1.80, –1.15 SD); and the mean level of impairment was –1.13 (–1.43, –0.82 SD). For the quetiapine and adjunctive treatment group, the mean frequency measure was –1.08 (–1.45, –0.72 SD); the intensity was –1.12 (–1.44, –0.79 SD); and the impairment was –0.77 (–1.07, –0.47 SD). For all scores in both groups, P was less than .001.

The changes from baseline on the Bipolar Index Severity Scale (BISS) overall for the quetiapine plus adjunctive treatment group was –28.56 (–10.91, –26.21 SD; P less than .0001). For the lithium plus adjunctive treatment group, it was –27.61 (–29.99, –25.24 SD; P less than .0001). The overall difference between the two groups was 0.94 (–2.10, 3.99 SD; P = .54).

Scores on the Longitudinal Interval Follow-up Evaluation Range of Impaired Functioning Tool also slightly favored lithium: –3.74 (–4.29, –3.19 SD) vs. –3.6 (–4.15, –3.07 SD) for quetiapine (P less than .0001 for all).

To the investigators’ surprise, among those with comorbid anxiety, the lithium plus adjunctive treatment group had fewer necessary clinical adjustments per month than did the quetiapine plus adjunctive treatment group, according to Dr. Nierenberg: –0.83 vs. 1.11 (P = .02).

"That seemed counterintuitive, and this difference was only with anxiety, not with any other comorbid psychiatric conditions," Dr. Nierenberg said. He hypothesized that it was possible benzodiazepines were used more frequently and easily with lithium than with quetiapine, but said future analyses would give a clearer answer. "We have a detailed database of every other medication used: when it was started, when it was stopped, and the reason why everything was done."

Not having current anxiety disorder was predictive of a better outcome (odds ratio, 1.81; P = .02), as was employment (OR, 1.67; P = .04). Those with bipolar II disorder responded better to treatment than those with bipolar disorder I, having an OR of response to treatment of nearly 1.8 P = .03). "That was a surprise to us, too," Dr. Nierenberg said.

Just over a quarter (27%) of the study population had metabolic syndrome, according to Dr. Kemp, who said this was not found to influence treatment outcomes.

Nearly half of the overall population was obese (48%), and slightly less than half (44%) had adipose. Prospective analysis indicated that patients with either obesity or adiposity tended to show less improvement in their CGI scores in a statistically significant way, regardless of which mood stabilizer they were on.

"Part of that might be due to overlapping pathophysiology between bipolar disorder and obesity," Dr. Kemp said. "Much of that is centered around inflammation in the central nervous system, as well as alterations in adipokines receptor levels and mitochondrial dysfunction."

"A key issue is to take care of obesity," said panel discussant Dr. Mauricio Tohen, chairman of psychiatry at the University of New Mexico in Albuquerque. "Regardless of the treatment, when there is obesity, the outcome will not be as good."

Efficacy vs. ‘generalizability’

Adjunctive personalized treatment was measured to help minimize the lack of assay sensitivity and internal validity typically inherent in comparative effectiveness trials, while also achieving more broadly applicable results.

"The right study depends on the right question," Dr. Tohen said. "If the question is whether a particular treatment has efficacy and is safe, then of course, we need an efficacy study. The problem with efficacy studies is that they limit the generalizability."

Instead, Dr. Tohen, who disclosed he spent more than a decade evaluating efficacy studies for both Eli Lily and AstraZeneca (makers of quetiapine), praised the metric of the adjunctive personalized treatments, and said a study should answer the question, "Of all the patients in my clinic, who will do better on which treatment? With efficacy studies you cannot answer that question because of the exclusion criteria that need to be taken into account."

In addition, Dr. Tohen said the metric likely would increase in importance under the Affordable Care Act, where efficacy is not the only consideration. "For example, if a patient relapses earlier, it might not be reimbursed, so asking what other outcomes we need to measure in [these kinds of] comparative effectiveness studies is very important."

In the CHOICE study, which had no placebo group, participating physicians were asked to track each patient’s dosage changes, missed doses, new medications added, discontinued medications, and the specific reasons for any of these changes. Changes that were made because of lack of effectiveness or intolerance were measured; however, planned dosage titrations according to normal scales were not considered necessary clinical adjustments but simply the regular course of treatment.

This heterogeneity of adjunctive treatment was seen by the investigators as a way to reflect "real world" practice, although they also noted it could be seen as a limitation to the study.

Another way the study was designed to reflect real-world practice was the inclusion of a broader-than-usual group of participants, and asking them for their feedback at the end of the study period.

"Before we put together these data, we held a stakeholder summit," Dr. Nierenberg said. "We invited patients and their advocates to help us interpret the study. They felt strongly that this was a reasonably positive study, because they thought that people really did get better. You could report these data toward the negative or the positive, but we reported toward the positive because that is what the stakeholders asked us to do." Dr. Tohen said that this kind of feedback in a comparative effectiveness study was another way these kinds of studies could help guide practice in the era of the Affordable Care Act.

As for who was actually admitted to the study, "exclusions were minimal," said Dr. Edward S. Friedman of the psychiatry department at the University of Pittsburgh. Those with a history of drug or alcohol dependence in the previous 30 days, a demonstrated intolerance to either study drug, or severe cardiovascular or renal disease were excluded, as were those with unstable thyroid disease, and pregnant or breastfeeding women.

Overall, the group was multiethnic, although nearly three-quarters were white and 20% were black. Roughly a third of the entire study population was employed, a third unemployed, and the rest were students, retirees, or those on disability.

The overall BISS score at enrollment averaged 56.1 (SD plus or minus 18.8). The average BISS depression score was 37.6 (SD plus or minus 14.0), and the average BISS mania score was 18.5 (SD plus or minus 12.1).

There were many comorbid psychiatric conditions, ranging from panic disorder to agoraphobia, although substance use was highest (61.4%). Current anxiety disorder also was prevalent at 58%.

"The CHOICE sample was similar in age and gender to previous efficacy studies," Dr. Friedman said. "It was more representative of the U.S. population than efficacy studies such as STEP-BD or LITMUS."

The 283 women (59% of the study) were more likely to report having spent a greater percentage of time depressed in the previous year, although the percentage of time spent in manic or hypomanic states was equal across both genders.

"This study was composed of sick individuals who had been sick for a considerable period of time," Dr. Friedman said.

For example, in the CHOICE study, 47% had been hospitalized for their bipolar disorder previously, vs. 43% in the LITMUS study. CHOICE participants had an attempted suicide rate of 36%, which fell between the 41% in the LITMUS trial and the 33.3% in an efficacy study conducted by Dr. Joseph R. Calabrese and his colleagues (J. Clin. Psychiatry 1999;60:79-88).

The average age for the first depressive episode for CHOICE participants was 16.4 years; the first manic episode tended to be at 20 years; and the first mood episode tended to be around 15.5 years. These data were similar to those in the LITMUS study, but differed from data in several efficacy studies such as the one by Dr. Calabrese, where the average age was 31.3 years.

"You might anticipate that in these specialty clinics we would have very few medication-naive patients, and yet we did," Dr. Friedman said during a postpanel audience participation session. "Looking for patients with a low threshold, we brought in patients who were very sick. They were also older, with lots of previous episodes. Maybe we missed younger patients who didn’t have as many previous episodes."

"These are the patients you would see in your practice," Dr. Nierenberg said before asking the audience what they think should be studied next in a comparative effectiveness trial. "We could look at the use of different antipsychotics, the combinations we use, with or without lamotrigine," he said.

"Most of the decisions that are made in medicine are made without evidence. Most of the things that are done are done without evidence. You have combinations that were never studied before or never even used before, and it’s across all medicine." Because medicine needs to be learning more, he said the audience needed to participate. "What are the questions that need to be asked?"

wmcknight@frontlinemedcom.com

On Twitter @whitneymcknight

HOLLYWOOD, FLA. – Because there is no clinically significant difference in bipolar spectrum symptom relief offered by either lithium or quetiapine, which mood stabilizer clinicians choose to prescribe should largely depend on what side effects patients can tolerate.

That’s the conclusion of a panel of experts who presented data from the Bipolar CHOICE (Clinical Health Outcomes Initiative in Comparative Effectiveness for Bipolar Disorder) clinical trial at a meeting of the American Society of Clinical Psychopharmacology, formerly known as the New Clinical Drug Evaluation Unit meeting.

By randomly assigning the study’s 482 participating outpatient adults to either lithium or quetiapine, while allowing clinicians to prescribe adjunctive treatment as necessary; and by using a broader-than-usual set of inclusion criteria, the investigators hoped to provide clinicians with "real world" data on how the respective mood stabilizers perform.

"To my knowledge, this is the first randomized comparative effectiveness study of regular, expert care that looks at lithium versus quetiapine strategies of treatment, both with adjunctive personalized treatment when necessary," study coauthor Dr. Andrew A. Nierenberg told the audience in a well-attended session.

"We tortured the data to get a P value to pop out," said Dr. Nierenberg, who is the director of the Bipolar Clinic and Research Program at Massachusetts General Hospital in Boston. "The bottom line was that the treatment strategies, based on either lithium or quetiapine, were very similar. I think the clinical implication is that the choice of treatment really depends on someone’s tolerability. This may resurrect lithium as something [clinicians] should at least consider, because its use has gone down dramatically in the last 20 years or so."

Side effects such as metabolic or sleep disturbances were slightly less frequent in the lithium group, although the difference was not statistically significant; rates of suicidality and discontinuation of medication due to side effects were also similar across the two groups, according to Dr. David E. Kemp of the psychiatry department at Case Western Reserve University, Cleveland, and a study coauthor who also presented data during the session. Changes from baseline in mood stabilization were also virtually the same across the groups.

‘Rules of engagement’

The study enrolled adults aged 18-68 years who were at least mildly ill with either bipolar disorder I (68%) or II, with a Clinical Global Impression (CGI) score of at least 3, although the average CGI score was 4.5 (standard deviation of plus or minus 0.9). Of the 482 enrolled, 364 completed the study.

The so-called rules of engagement for clinicians from the 11 participating centers were that "they do everything they could to get their patients well," with the caveat that the lithium patient group not receive any antipsychotic, and the quetiapine patient group not receive lithium or any other antipsychotic.

"But they could get anything else that they needed," Dr. Nierenberg said.

At the time of enrollment, subjects were, in the estimation of their treating physician, experiencing symptoms that warranted a change in treatment, and that either lithium or quetiapine would be viable therapeutic options. Patients did not need to be lithium- or quetiapine-naive, but they could not have been treated with either drug in the previous 30 days. If they were already taking a second-generation antipsychotic, participants had to be willing to discontinue that medication and to be randomly assigned to either of the study drugs.

This strict adherence to one mood stabilizer, combined with adjunctive personalized treatments such as antidepressants or benzodiazepines, prescribed at the discretion of the treating clinician, ensured that the study tested the strategy of using either lithium or quetiapine as the base of treatment, Dr. Nierenberg said.

‘Similar’ results

Clinicians also were instructed to use the maximum tolerated dose of the respective mood stabilizer. For the 240 patients in the lithium group, the mean maximum tolerated dose was 1,007.5 mg, with the median dose being a "perfectly reasonable 900 mg," Dr. Nierenberg said. Blood lithium levels were taken at weeks 2, 16, and 24 and also were all found to be "reasonable," with most patients getting to at least 0.6 mEq/L by week 24.

For the 242 members of the quetiapine group, the maximum tolerated dose was 344.9 mg, with a median tolerated dose of 300 mg, which Dr. Nierenberg said was "what you would expect."

The overall result was that regardless of which mood stabilizer was prescribed along with adjunctive personalized treatment, the coprimary outcomes of the benefits and harms ratio from treatment effects over time (the CGI efficacy index), and the number of necessary clinical adjustments (changes in medication excluding titrations according to scale), were virtually parallel: For each primary outcome, the P value for the estimated change in baseline was less than .0001 for both lithium plus adjunctive treatment, and quetiapine plus adjunctive treatment.

Further, Dr. Nierenberg reported the "unexpected" result that at 6 months, about a quarter of each group was observed to be doing well without any adjunctive personalized treatments (23.8% of the lithium group; 27.3% of the quetiapine group; P = .14). "We thought that everybody would be on multiple things," Dr. Nierenberg said.

There was also no real difference in the time to discontinuation. "We were actually surprised that at 6 months, about three-quarters of the patients were still on the study drug, which was pretty good considering that this was a very ill population," Dr. Nierenberg said.

Using a CGI bipolar severity scale rating of up to 2, maintained for at least 8 weeks, the investigators found that overall, 20% were considered doing "really well"; it was also about 20% for the lithium plus adjunctive therapy group and the quetiapine and adjunctive therapy group (P = .14 for all three).

Because the Agency for Healthcare Research and Quality–funded study mandated that researchers end their investigation at precisely 36 months, Dr. Nierenberg said the study sites "randomized the 482 patients to 6 months of treatment within 36 months from start to finish," but he added that the investigators would have preferred to conduct a full 2-year study of participants.

"The good news was that most of the patients actually did improve substantially; the bad news is that while maybe a quarter did really well by the end of 6 months, we don’t know whether if we were to extend the study out further, we would see a difference between the two groups, whether they would continue to improve, or whether they would relapse," he said.

Adverse effects, predictors of response

Because study participants were followed for only 6 months, the long-term adverse effects of many years of exposure to either mood stabilizer, such as renal impairment with lithium or type II diabetes with quetiapine, were not considered, Dr. Kemp said. The adverse-effect profile for lithium generally includes a narrow therapeutic index; nausea, vomiting, and diarrhea; renal impairment; and hypothyroidism. Although Dr. Kemp said that the long-term adverse effects of quetiapine are understudied, known short-term side effects include the potential for sedation or somnolence, and weight gain.

Changes in baseline on the Frequency and Intensity of Side Effects Ratings for the quetiapine plus adjunctive personalized treatment arm were slightly more adverse than for the lithium and additional treatment arm. For the lithium plus adjunctive treatment group, the mean frequency of side effects was –1.41 (–1.78, –1.04 SD); the mean intensity of side effects was –1.48 (–1.80, –1.15 SD); and the mean level of impairment was –1.13 (–1.43, –0.82 SD). For the quetiapine and adjunctive treatment group, the mean frequency measure was –1.08 (–1.45, –0.72 SD); the intensity was –1.12 (–1.44, –0.79 SD); and the impairment was –0.77 (–1.07, –0.47 SD). For all scores in both groups, P was less than .001.

The changes from baseline on the Bipolar Index Severity Scale (BISS) overall for the quetiapine plus adjunctive treatment group was –28.56 (–10.91, –26.21 SD; P less than .0001). For the lithium plus adjunctive treatment group, it was –27.61 (–29.99, –25.24 SD; P less than .0001). The overall difference between the two groups was 0.94 (–2.10, 3.99 SD; P = .54).

Scores on the Longitudinal Interval Follow-up Evaluation Range of Impaired Functioning Tool also slightly favored lithium: –3.74 (–4.29, –3.19 SD) vs. –3.6 (–4.15, –3.07 SD) for quetiapine (P less than .0001 for all).

To the investigators’ surprise, among those with comorbid anxiety, the lithium plus adjunctive treatment group had fewer necessary clinical adjustments per month than did the quetiapine plus adjunctive treatment group, according to Dr. Nierenberg: –0.83 vs. 1.11 (P = .02).

"That seemed counterintuitive, and this difference was only with anxiety, not with any other comorbid psychiatric conditions," Dr. Nierenberg said. He hypothesized that it was possible benzodiazepines were used more frequently and easily with lithium than with quetiapine, but said future analyses would give a clearer answer. "We have a detailed database of every other medication used: when it was started, when it was stopped, and the reason why everything was done."

Not having current anxiety disorder was predictive of a better outcome (odds ratio, 1.81; P = .02), as was employment (OR, 1.67; P = .04). Those with bipolar II disorder responded better to treatment than those with bipolar disorder I, having an OR of response to treatment of nearly 1.8 P = .03). "That was a surprise to us, too," Dr. Nierenberg said.

Just over a quarter (27%) of the study population had metabolic syndrome, according to Dr. Kemp, who said this was not found to influence treatment outcomes.

Nearly half of the overall population was obese (48%), and slightly less than half (44%) had adipose. Prospective analysis indicated that patients with either obesity or adiposity tended to show less improvement in their CGI scores in a statistically significant way, regardless of which mood stabilizer they were on.

"Part of that might be due to overlapping pathophysiology between bipolar disorder and obesity," Dr. Kemp said. "Much of that is centered around inflammation in the central nervous system, as well as alterations in adipokines receptor levels and mitochondrial dysfunction."

"A key issue is to take care of obesity," said panel discussant Dr. Mauricio Tohen, chairman of psychiatry at the University of New Mexico in Albuquerque. "Regardless of the treatment, when there is obesity, the outcome will not be as good."

Efficacy vs. ‘generalizability’

Adjunctive personalized treatment was measured to help minimize the lack of assay sensitivity and internal validity typically inherent in comparative effectiveness trials, while also achieving more broadly applicable results.

"The right study depends on the right question," Dr. Tohen said. "If the question is whether a particular treatment has efficacy and is safe, then of course, we need an efficacy study. The problem with efficacy studies is that they limit the generalizability."

Instead, Dr. Tohen, who disclosed he spent more than a decade evaluating efficacy studies for both Eli Lily and AstraZeneca (makers of quetiapine), praised the metric of the adjunctive personalized treatments, and said a study should answer the question, "Of all the patients in my clinic, who will do better on which treatment? With efficacy studies you cannot answer that question because of the exclusion criteria that need to be taken into account."

In addition, Dr. Tohen said the metric likely would increase in importance under the Affordable Care Act, where efficacy is not the only consideration. "For example, if a patient relapses earlier, it might not be reimbursed, so asking what other outcomes we need to measure in [these kinds of] comparative effectiveness studies is very important."

In the CHOICE study, which had no placebo group, participating physicians were asked to track each patient’s dosage changes, missed doses, new medications added, discontinued medications, and the specific reasons for any of these changes. Changes that were made because of lack of effectiveness or intolerance were measured; however, planned dosage titrations according to normal scales were not considered necessary clinical adjustments but simply the regular course of treatment.

This heterogeneity of adjunctive treatment was seen by the investigators as a way to reflect "real world" practice, although they also noted it could be seen as a limitation to the study.

Another way the study was designed to reflect real-world practice was the inclusion of a broader-than-usual group of participants, and asking them for their feedback at the end of the study period.

"Before we put together these data, we held a stakeholder summit," Dr. Nierenberg said. "We invited patients and their advocates to help us interpret the study. They felt strongly that this was a reasonably positive study, because they thought that people really did get better. You could report these data toward the negative or the positive, but we reported toward the positive because that is what the stakeholders asked us to do." Dr. Tohen said that this kind of feedback in a comparative effectiveness study was another way these kinds of studies could help guide practice in the era of the Affordable Care Act.

As for who was actually admitted to the study, "exclusions were minimal," said Dr. Edward S. Friedman of the psychiatry department at the University of Pittsburgh. Those with a history of drug or alcohol dependence in the previous 30 days, a demonstrated intolerance to either study drug, or severe cardiovascular or renal disease were excluded, as were those with unstable thyroid disease, and pregnant or breastfeeding women.

Overall, the group was multiethnic, although nearly three-quarters were white and 20% were black. Roughly a third of the entire study population was employed, a third unemployed, and the rest were students, retirees, or those on disability.

The overall BISS score at enrollment averaged 56.1 (SD plus or minus 18.8). The average BISS depression score was 37.6 (SD plus or minus 14.0), and the average BISS mania score was 18.5 (SD plus or minus 12.1).

There were many comorbid psychiatric conditions, ranging from panic disorder to agoraphobia, although substance use was highest (61.4%). Current anxiety disorder also was prevalent at 58%.

"The CHOICE sample was similar in age and gender to previous efficacy studies," Dr. Friedman said. "It was more representative of the U.S. population than efficacy studies such as STEP-BD or LITMUS."

The 283 women (59% of the study) were more likely to report having spent a greater percentage of time depressed in the previous year, although the percentage of time spent in manic or hypomanic states was equal across both genders.

"This study was composed of sick individuals who had been sick for a considerable period of time," Dr. Friedman said.

For example, in the CHOICE study, 47% had been hospitalized for their bipolar disorder previously, vs. 43% in the LITMUS study. CHOICE participants had an attempted suicide rate of 36%, which fell between the 41% in the LITMUS trial and the 33.3% in an efficacy study conducted by Dr. Joseph R. Calabrese and his colleagues (J. Clin. Psychiatry 1999;60:79-88).

The average age for the first depressive episode for CHOICE participants was 16.4 years; the first manic episode tended to be at 20 years; and the first mood episode tended to be around 15.5 years. These data were similar to those in the LITMUS study, but differed from data in several efficacy studies such as the one by Dr. Calabrese, where the average age was 31.3 years.

"You might anticipate that in these specialty clinics we would have very few medication-naive patients, and yet we did," Dr. Friedman said during a postpanel audience participation session. "Looking for patients with a low threshold, we brought in patients who were very sick. They were also older, with lots of previous episodes. Maybe we missed younger patients who didn’t have as many previous episodes."

"These are the patients you would see in your practice," Dr. Nierenberg said before asking the audience what they think should be studied next in a comparative effectiveness trial. "We could look at the use of different antipsychotics, the combinations we use, with or without lamotrigine," he said.

"Most of the decisions that are made in medicine are made without evidence. Most of the things that are done are done without evidence. You have combinations that were never studied before or never even used before, and it’s across all medicine." Because medicine needs to be learning more, he said the audience needed to participate. "What are the questions that need to be asked?"

wmcknight@frontlinemedcom.com

On Twitter @whitneymcknight

HOLLYWOOD, FLA. – Because there is no clinically significant difference in bipolar spectrum symptom relief offered by either lithium or quetiapine, which mood stabilizer clinicians choose to prescribe should largely depend on what side effects patients can tolerate.

That’s the conclusion of a panel of experts who presented data from the Bipolar CHOICE (Clinical Health Outcomes Initiative in Comparative Effectiveness for Bipolar Disorder) clinical trial at a meeting of the American Society of Clinical Psychopharmacology, formerly known as the New Clinical Drug Evaluation Unit meeting.

By randomly assigning the study’s 482 participating outpatient adults to either lithium or quetiapine, while allowing clinicians to prescribe adjunctive treatment as necessary; and by using a broader-than-usual set of inclusion criteria, the investigators hoped to provide clinicians with "real world" data on how the respective mood stabilizers perform.

"To my knowledge, this is the first randomized comparative effectiveness study of regular, expert care that looks at lithium versus quetiapine strategies of treatment, both with adjunctive personalized treatment when necessary," study coauthor Dr. Andrew A. Nierenberg told the audience in a well-attended session.

"We tortured the data to get a P value to pop out," said Dr. Nierenberg, who is the director of the Bipolar Clinic and Research Program at Massachusetts General Hospital in Boston. "The bottom line was that the treatment strategies, based on either lithium or quetiapine, were very similar. I think the clinical implication is that the choice of treatment really depends on someone’s tolerability. This may resurrect lithium as something [clinicians] should at least consider, because its use has gone down dramatically in the last 20 years or so."

Side effects such as metabolic or sleep disturbances were slightly less frequent in the lithium group, although the difference was not statistically significant; rates of suicidality and discontinuation of medication due to side effects were also similar across the two groups, according to Dr. David E. Kemp of the psychiatry department at Case Western Reserve University, Cleveland, and a study coauthor who also presented data during the session. Changes from baseline in mood stabilization were also virtually the same across the groups.

‘Rules of engagement’

The study enrolled adults aged 18-68 years who were at least mildly ill with either bipolar disorder I (68%) or II, with a Clinical Global Impression (CGI) score of at least 3, although the average CGI score was 4.5 (standard deviation of plus or minus 0.9). Of the 482 enrolled, 364 completed the study.

The so-called rules of engagement for clinicians from the 11 participating centers were that "they do everything they could to get their patients well," with the caveat that the lithium patient group not receive any antipsychotic, and the quetiapine patient group not receive lithium or any other antipsychotic.

"But they could get anything else that they needed," Dr. Nierenberg said.

At the time of enrollment, subjects were, in the estimation of their treating physician, experiencing symptoms that warranted a change in treatment, and that either lithium or quetiapine would be viable therapeutic options. Patients did not need to be lithium- or quetiapine-naive, but they could not have been treated with either drug in the previous 30 days. If they were already taking a second-generation antipsychotic, participants had to be willing to discontinue that medication and to be randomly assigned to either of the study drugs.

This strict adherence to one mood stabilizer, combined with adjunctive personalized treatments such as antidepressants or benzodiazepines, prescribed at the discretion of the treating clinician, ensured that the study tested the strategy of using either lithium or quetiapine as the base of treatment, Dr. Nierenberg said.

‘Similar’ results

Clinicians also were instructed to use the maximum tolerated dose of the respective mood stabilizer. For the 240 patients in the lithium group, the mean maximum tolerated dose was 1,007.5 mg, with the median dose being a "perfectly reasonable 900 mg," Dr. Nierenberg said. Blood lithium levels were taken at weeks 2, 16, and 24 and also were all found to be "reasonable," with most patients getting to at least 0.6 mEq/L by week 24.

For the 242 members of the quetiapine group, the maximum tolerated dose was 344.9 mg, with a median tolerated dose of 300 mg, which Dr. Nierenberg said was "what you would expect."

The overall result was that regardless of which mood stabilizer was prescribed along with adjunctive personalized treatment, the coprimary outcomes of the benefits and harms ratio from treatment effects over time (the CGI efficacy index), and the number of necessary clinical adjustments (changes in medication excluding titrations according to scale), were virtually parallel: For each primary outcome, the P value for the estimated change in baseline was less than .0001 for both lithium plus adjunctive treatment, and quetiapine plus adjunctive treatment.

Further, Dr. Nierenberg reported the "unexpected" result that at 6 months, about a quarter of each group was observed to be doing well without any adjunctive personalized treatments (23.8% of the lithium group; 27.3% of the quetiapine group; P = .14). "We thought that everybody would be on multiple things," Dr. Nierenberg said.

There was also no real difference in the time to discontinuation. "We were actually surprised that at 6 months, about three-quarters of the patients were still on the study drug, which was pretty good considering that this was a very ill population," Dr. Nierenberg said.

Using a CGI bipolar severity scale rating of up to 2, maintained for at least 8 weeks, the investigators found that overall, 20% were considered doing "really well"; it was also about 20% for the lithium plus adjunctive therapy group and the quetiapine and adjunctive therapy group (P = .14 for all three).

Because the Agency for Healthcare Research and Quality–funded study mandated that researchers end their investigation at precisely 36 months, Dr. Nierenberg said the study sites "randomized the 482 patients to 6 months of treatment within 36 months from start to finish," but he added that the investigators would have preferred to conduct a full 2-year study of participants.

"The good news was that most of the patients actually did improve substantially; the bad news is that while maybe a quarter did really well by the end of 6 months, we don’t know whether if we were to extend the study out further, we would see a difference between the two groups, whether they would continue to improve, or whether they would relapse," he said.

Adverse effects, predictors of response

Because study participants were followed for only 6 months, the long-term adverse effects of many years of exposure to either mood stabilizer, such as renal impairment with lithium or type II diabetes with quetiapine, were not considered, Dr. Kemp said. The adverse-effect profile for lithium generally includes a narrow therapeutic index; nausea, vomiting, and diarrhea; renal impairment; and hypothyroidism. Although Dr. Kemp said that the long-term adverse effects of quetiapine are understudied, known short-term side effects include the potential for sedation or somnolence, and weight gain.

Changes in baseline on the Frequency and Intensity of Side Effects Ratings for the quetiapine plus adjunctive personalized treatment arm were slightly more adverse than for the lithium and additional treatment arm. For the lithium plus adjunctive treatment group, the mean frequency of side effects was –1.41 (–1.78, –1.04 SD); the mean intensity of side effects was –1.48 (–1.80, –1.15 SD); and the mean level of impairment was –1.13 (–1.43, –0.82 SD). For the quetiapine and adjunctive treatment group, the mean frequency measure was –1.08 (–1.45, –0.72 SD); the intensity was –1.12 (–1.44, –0.79 SD); and the impairment was –0.77 (–1.07, –0.47 SD). For all scores in both groups, P was less than .001.

The changes from baseline on the Bipolar Index Severity Scale (BISS) overall for the quetiapine plus adjunctive treatment group was –28.56 (–10.91, –26.21 SD; P less than .0001). For the lithium plus adjunctive treatment group, it was –27.61 (–29.99, –25.24 SD; P less than .0001). The overall difference between the two groups was 0.94 (–2.10, 3.99 SD; P = .54).

Scores on the Longitudinal Interval Follow-up Evaluation Range of Impaired Functioning Tool also slightly favored lithium: –3.74 (–4.29, –3.19 SD) vs. –3.6 (–4.15, –3.07 SD) for quetiapine (P less than .0001 for all).

To the investigators’ surprise, among those with comorbid anxiety, the lithium plus adjunctive treatment group had fewer necessary clinical adjustments per month than did the quetiapine plus adjunctive treatment group, according to Dr. Nierenberg: –0.83 vs. 1.11 (P = .02).

"That seemed counterintuitive, and this difference was only with anxiety, not with any other comorbid psychiatric conditions," Dr. Nierenberg said. He hypothesized that it was possible benzodiazepines were used more frequently and easily with lithium than with quetiapine, but said future analyses would give a clearer answer. "We have a detailed database of every other medication used: when it was started, when it was stopped, and the reason why everything was done."

Not having current anxiety disorder was predictive of a better outcome (odds ratio, 1.81; P = .02), as was employment (OR, 1.67; P = .04). Those with bipolar II disorder responded better to treatment than those with bipolar disorder I, having an OR of response to treatment of nearly 1.8 P = .03). "That was a surprise to us, too," Dr. Nierenberg said.

Just over a quarter (27%) of the study population had metabolic syndrome, according to Dr. Kemp, who said this was not found to influence treatment outcomes.

Nearly half of the overall population was obese (48%), and slightly less than half (44%) had adipose. Prospective analysis indicated that patients with either obesity or adiposity tended to show less improvement in their CGI scores in a statistically significant way, regardless of which mood stabilizer they were on.

"Part of that might be due to overlapping pathophysiology between bipolar disorder and obesity," Dr. Kemp said. "Much of that is centered around inflammation in the central nervous system, as well as alterations in adipokines receptor levels and mitochondrial dysfunction."

"A key issue is to take care of obesity," said panel discussant Dr. Mauricio Tohen, chairman of psychiatry at the University of New Mexico in Albuquerque. "Regardless of the treatment, when there is obesity, the outcome will not be as good."

Efficacy vs. ‘generalizability’

Adjunctive personalized treatment was measured to help minimize the lack of assay sensitivity and internal validity typically inherent in comparative effectiveness trials, while also achieving more broadly applicable results.

"The right study depends on the right question," Dr. Tohen said. "If the question is whether a particular treatment has efficacy and is safe, then of course, we need an efficacy study. The problem with efficacy studies is that they limit the generalizability."

Instead, Dr. Tohen, who disclosed he spent more than a decade evaluating efficacy studies for both Eli Lily and AstraZeneca (makers of quetiapine), praised the metric of the adjunctive personalized treatments, and said a study should answer the question, "Of all the patients in my clinic, who will do better on which treatment? With efficacy studies you cannot answer that question because of the exclusion criteria that need to be taken into account."

In addition, Dr. Tohen said the metric likely would increase in importance under the Affordable Care Act, where efficacy is not the only consideration. "For example, if a patient relapses earlier, it might not be reimbursed, so asking what other outcomes we need to measure in [these kinds of] comparative effectiveness studies is very important."

In the CHOICE study, which had no placebo group, participating physicians were asked to track each patient’s dosage changes, missed doses, new medications added, discontinued medications, and the specific reasons for any of these changes. Changes that were made because of lack of effectiveness or intolerance were measured; however, planned dosage titrations according to normal scales were not considered necessary clinical adjustments but simply the regular course of treatment.

This heterogeneity of adjunctive treatment was seen by the investigators as a way to reflect "real world" practice, although they also noted it could be seen as a limitation to the study.

Another way the study was designed to reflect real-world practice was the inclusion of a broader-than-usual group of participants, and asking them for their feedback at the end of the study period.

"Before we put together these data, we held a stakeholder summit," Dr. Nierenberg said. "We invited patients and their advocates to help us interpret the study. They felt strongly that this was a reasonably positive study, because they thought that people really did get better. You could report these data toward the negative or the positive, but we reported toward the positive because that is what the stakeholders asked us to do." Dr. Tohen said that this kind of feedback in a comparative effectiveness study was another way these kinds of studies could help guide practice in the era of the Affordable Care Act.

As for who was actually admitted to the study, "exclusions were minimal," said Dr. Edward S. Friedman of the psychiatry department at the University of Pittsburgh. Those with a history of drug or alcohol dependence in the previous 30 days, a demonstrated intolerance to either study drug, or severe cardiovascular or renal disease were excluded, as were those with unstable thyroid disease, and pregnant or breastfeeding women.

Overall, the group was multiethnic, although nearly three-quarters were white and 20% were black. Roughly a third of the entire study population was employed, a third unemployed, and the rest were students, retirees, or those on disability.

The overall BISS score at enrollment averaged 56.1 (SD plus or minus 18.8). The average BISS depression score was 37.6 (SD plus or minus 14.0), and the average BISS mania score was 18.5 (SD plus or minus 12.1).

There were many comorbid psychiatric conditions, ranging from panic disorder to agoraphobia, although substance use was highest (61.4%). Current anxiety disorder also was prevalent at 58%.

"The CHOICE sample was similar in age and gender to previous efficacy studies," Dr. Friedman said. "It was more representative of the U.S. population than efficacy studies such as STEP-BD or LITMUS."

The 283 women (59% of the study) were more likely to report having spent a greater percentage of time depressed in the previous year, although the percentage of time spent in manic or hypomanic states was equal across both genders.

"This study was composed of sick individuals who had been sick for a considerable period of time," Dr. Friedman said.

For example, in the CHOICE study, 47% had been hospitalized for their bipolar disorder previously, vs. 43% in the LITMUS study. CHOICE participants had an attempted suicide rate of 36%, which fell between the 41% in the LITMUS trial and the 33.3% in an efficacy study conducted by Dr. Joseph R. Calabrese and his colleagues (J. Clin. Psychiatry 1999;60:79-88).

The average age for the first depressive episode for CHOICE participants was 16.4 years; the first manic episode tended to be at 20 years; and the first mood episode tended to be around 15.5 years. These data were similar to those in the LITMUS study, but differed from data in several efficacy studies such as the one by Dr. Calabrese, where the average age was 31.3 years.

"You might anticipate that in these specialty clinics we would have very few medication-naive patients, and yet we did," Dr. Friedman said during a postpanel audience participation session. "Looking for patients with a low threshold, we brought in patients who were very sick. They were also older, with lots of previous episodes. Maybe we missed younger patients who didn’t have as many previous episodes."

"These are the patients you would see in your practice," Dr. Nierenberg said before asking the audience what they think should be studied next in a comparative effectiveness trial. "We could look at the use of different antipsychotics, the combinations we use, with or without lamotrigine," he said.

"Most of the decisions that are made in medicine are made without evidence. Most of the things that are done are done without evidence. You have combinations that were never studied before or never even used before, and it’s across all medicine." Because medicine needs to be learning more, he said the audience needed to participate. "What are the questions that need to be asked?"

wmcknight@frontlinemedcom.com

On Twitter @whitneymcknight