Developments and issues in prenatal screening and diagnosis

Prenatal assessments for major chromosomal abnormalities have, like a pendulum, swung over the last 50 years between advancements in screening tests and diagnostic procedures.

In the 1960s, screening for advanced maternal age gave way to diagnostic amniocentesis. Maternal serum alpha-fetoprotein screening for neural tube defects came on the scene in the 1970s, and Down syndrome screening and chorionic villus sampling (CVS) followed in the 1980s. Better ultrasound screening markers were used in combination with biochemistry to advance first-trimester screening in the 1990s and 2000s, leading to a significant decline in diagnostic procedures. Now, free fetal DNA measurement, known as noninvasive prenatal screening, or NIPS, has entered the scene. This development, along with advances in the accuracy of diagnostic lab testing through microarray analysis, will soon lead the pendulum to swing back toward more definitive diagnostic procedures that require either CVS or amniocentesis.

Screening tests provide us with odds adjustments, not definitive answers, and are meant for everyone. Diagnostic tests are meant to give us definitive answers, may have risks, and therefore have traditionally been done on "at-risk" patients. Fundamentally, a screening test gives us an impression, while a diagnostic test gives us harsh reality. As always, there will be trade-offs. No approach is perfect, and no one size fits everyone.

Risks beyond Down syndrome

During the prenatal period, patients will often say, "I’m concerned about having a baby with Down syndrome." What they really mean is that they’re concerned about having a baby with a serious problem, and Down syndrome is the name they know.

Serious problems – a Mendelian disorder, a multifactorial disorder, or a major chromosomal abnormality – affect 2%-3% of all births. Less serious chromosomal abnormalities affect 5%-6% of births. Although advanced maternal age is no longer the sole criterion for deciding who should be offered diagnostic testing, age still is a principal factor for risk determination.

At age 35, the chance of having a baby with Down syndrome is 1 in 380, but the chance of having any chromosomal abnormality detectable by karyotype is 1 in 190. For a 30-year-old, the chance of having a baby with any chromosomal abnormality is 1 in 380, and for a 40-year-old, the risk is 1 in 65.

With the first-trimester screening approach that combines maternal serum free beta-human chorionic gonadotropin (free beta-HCG) and pregnancy-associated plasma protein A (PAPP-A) with fetal nuchal translucency measurement, we are able to detect upward of 85% of fetuses with Down syndrome, or trisomy 21. Yet the disorder is only one of a large number of chromosomal abnormalities observed.

In a recent single-center study of more than 20,000 first-trimester screenings, 5.6% were positive for Down syndrome risk. Of those who subsequently had an amniocentesis or CVS, we found 4% had an abnormal karyotype. Interestingly, 40% of the time the abnormality was not Down syndrome, but another chromosomal abnormality. Similar analyses for trisomies 13 and 18 – the other major abnormalities targeted in first-trimester combined screening – yielded similar statistics (Prenat. Diagn. 2013;33:251-6).

All told, of the screen-positive pregnancies found to have an abnormal karyotype, at least 30% had chromosomal abnormalities outside of those for which they were screen positive. Such findings speak to the limitations of screening as opposed to diagnostic testing, and have implications for patient counseling. Patients should be counseled about the possibility of all chromosome abnormalities – not just Down syndrome.

The NIPS rollout

We have known for well over 100 years that fetal cells cross the placental barrier in small numbers, driving the development of what’s currently known as NIPS. The future of NIPS actually lies in an ever-expanding number of disorders, and will eventually end with sequencing the entire genome.

There are two main methods by which NIPS is done. The original and predominant method uses massive parallel shotgun sequencing, known as next-generation sequencing. This method involves whole-genome amplification and collects enormous amounts of information. Investigators are now attempting to direct amplification at the subchromosome level, mimicking some of what microarray analysis can do.

The second approach uses selected probes, or targeted sequencing, to focus on those sections of DNA that are of interest. Although this method may be cheaper in the short run, one drawback is that new probes will need to be created for each new disorder.

Initially, investigators attempted to isolate nucleated fetal cells from the maternal blood and use them for aneuploidy detection. However, a National Institutes of Health–funded fetal-cell isolation study that ended in 2002 reported disappointing results: Fetal-cell isolation methods had low sensitivity and other technological shortcomings. Subsequently, a number of companies attempted to replicate and improve the work, also without much success.

Concurrent with efforts to use fetal-cell isolation to perform NIPS was the discovery, in the late 1990s by Dr. Dennis Lo, upon whose work next-generation sequencing for NIPS is based, of the presence of circulating cell-free fetal DNA in maternal plasma. The concentration of cell-free fetal DNA may be as much as 5%-8% of the total circulating cell-free DNA in maternal plasma, making free fetal DNA a promising source of fetal genetic material for noninvasive prenatal investigation.

The first high-quality trials on the use of free fetal DNA measurement for detection of trisomy 21 were published in the fall of 2011, and demonstrated up to a 99% detection rate for Down syndrome (less for trisomies 18 and 13). Companies subsequently began to manufacture free fetal DNA tests as off-label products, The tests were initially designated as "noninvasive prenatal diagnosis" tools, but experts around the world objected to the "diagnosis" label, and the terminology shifted to "noninvasive prenatal testing" and finally to "noninvasive prenatal screening"– a designation that I believe accurately reflects its current role.

The uptake in utilization of NIPS has been faster than anyone could have predicted: In just over 2 years, several hundred thousand screening tests have been performed. With 98% to 99% sensitivity, NIPS is an excellent screening test for Down syndrome. However, 99% sensitivity does not equate to 99% positive predictive value, that is, the risk that a patient with a positive screen actually has Down syndrome is much lower.

The published studies of NIPS cite false positive rates of 0.2%-1.0%, but this rate will increase as more disorders are screened. Positive predictive value is directly proportional to the underlying risk. For example, a test with 99% sensitivity and 99% specificity (1% false positive rate) means that a 26-year-old woman who has a positive free fetal DNA test actually has as low as an 11% chance of having a baby with Down syndrome. In older women, who have a higher incidence of having a baby with Down syndrome, a positive NIPS result will have a higher positive predictive value of giving birth to a baby with Down syndrome. However, at the current time, NIPS is an excellent screening test for Down syndrome, but it is not ready for universal primary screening in younger women.

Falsely reassuring are the hyped marketing claims in the United States that NIPS "replaces amnio" and eliminates the need for a nuchal translucency (NT) test. When clinicians abandon performing or referring for a high-quality NT measurement, they can miss or significantly delay the diagnosis of twins/zygosity, growth abnormalities and placentation, cardiac defects, and numerous other anomalies. Similarly, when patients who otherwise would have opted to have CVS or amniocentesis forego having the procedure and have NIPS performed instead, they may regret this decision.

The danger that overreliance on screening tests may paradoxically increase the number of births of babies with otherwise detectable problems was raised in a study led by the late Dr. George Henry, an obstetrician-geneticist in Denver. Curious about declining rates of diagnostic testing in his own practice, Dr. Henry examined trends in his state and found that while the utilization of diagnostic tests had plummeted by 70% over about 15 years, the birth rate of babies born with Down syndrome during this time had doubled among mothers over age 35, and stayed the same among mothers under age 35.

A sociologist-researcher examined the trend that Dr. Henry identified and found that the rise in Down syndrome births was not due to an increase in women electing to keep these pregnancies, but to the fact that the abnormality was not detected in the first place (Fetal Diagn. Ther. 2008;23:308-15). The same risk of screening tests replacing definitive diagnosis exists today with the uptake of NIPS.

The impact of microarray

Microarray technologies have been developed over the past decade. The National Institutes of Health study on chromosomal microarray versus karyotyping, published a year ago, is a game changer. The microarray is analogous to a 15-fold magnifying glass on the karyotype. While the smallest piece of a chromosome that can be evaluated by karyotype analysis is about 5 million base pairs, microarray analysis zooms in on about 200,000 base pairs, allowing us to see small genomic deletions and duplications (copy number variants) that we’ve never seen before.

The trial looked at upward of 4,000 women undergoing CVS or amniocentesis at one of 29 centers. Each diagnostic sample was split in two, with standard karyotyping performed on one portion and chromosomal microarray on the other (N. Engl. J. Med. 2012;367:2175-84).

There were several significant findings: One is that almost one-third of patients have a copy variation now known to be benign. Another is that 2.5% of women who had ultrasound-identified anomalies and normal karyotypes had microdeletion/duplications on the microarray that were clearly associated with a known clinical problem. Moreover, another 3.2% had gains and losses of potential clinical significance. As such, close to 6% of women with ultrasound-identified anomalies and a normal karyotype had clinically relevant copy number variants that only the microarray could find.

When microarray analysis was performed in women whose only indication for prenatal diagnosis was advanced age or an abnormal result on Down syndrome screening, as opposed to an ultrasound-detected anomaly, 0.5%, or 1 in 200, were found to have a pathogenic abnormality that otherwise would have been missed. This is significant: Previously, with karyotyping, we quoted patients a minimum of 1/500 risk of finding a clinically relevant chromosomal anomaly even if the combined report suggested much lower risks of Down syndrome. Now, with microarray, that risk is 1/200. Other studies already published or about to be published show this same level of risk determined by microarray.

With any new technological advance, we get our numerators of the problem before our denominators. The first cases published are those in which clinical or laboratory findings are associated with an abnormality. Only then do researchers go back and look at cases with those findings to test these associations – and only then are the markers sometimes found not to be associated. It will take a number of years to acquire a sizable database on microarray-detected copy number variants. In the end, microarray may help us to explain many of the approximately 1% of serious problems that we have been unable to diagnose until now.

Current decision-making

Ultimately, the future lies with routine, complete genomic sequencing that provides a detailed view of the fetal genome. This is likely to be about 7-10 years away, and the main question on the table is whether it will be performed invasively or with a maternal blood sample.

Today, when a 35-year-old woman comes into my office early in her pregnancy, I will tell her that the risk of having a baby with a chromosomal abnormality is 1 in 190. If she wants to know more, I will explain that the second-trimester quad screen will detect 60% of Down syndrome cases, that the first-trimester combined screen can identify 85% or more, and that the free fetal DNA test will get closer to 99%.

Despite the significant advances in screening, all of these options are still the fundamental equivalent of a Gallup poll. If the patient wants a definitive answer, she will need either an amniocentesis or a CVS, which, in experienced hands, have been shown through an increasing number of studies to be of equal risk. Because there is no such thing as "no" risk when it comes to prenatal diagnostic testing, the question that each patient must answer is, "Where do you want to put that risk – in the test or gambling on the outcome?"

We also have a great cultural divide in the United States. Some people want to know everything, others want to know nothing. Our affluent patients are getting older. Our poorer patients are getting younger. Some people will pay whatever it takes to get the answers they want, while others can or will not pay a dime beyond what their insurance will cover.

There is no one algorithm that can handle these two extremes of patients. Right now, many programs around the country have seen a diminishing number of patients having diagnostic testing – a phenomenon I believe is the result of a false sense of confidence, of people being lulled by the faulty argument that screening protocols can find Down syndrome, so what else is there to know?

Ultimately, a model for younger women would start with "contingent screening," in which patients could start with first-trimester combined screening and move straight to CVS if found to be at "high risk," and end testing if found to be at "low risk." Women who fall in the middle could undergo either free fetal DNA analysis or CVS, and we are developing methods to improve the mathematical processing of data to improve the sensitivity and specificity of all screening programs.

In my program, CVS and microarray are now offered to all patients regardless of age, as the risk that a microarray will find a significant abnormality is at least 1/200, which is the risk we have been quoting to 35-year-old patients for nearly 50 years.

Dr. Evans is president of the Fetal Medicine Foundation and International Fetal Medicine and Surgery Society Foundation, and professor of obstetrics and gynecology at the Icahn School of Medicine at Mount Sinai, New York. Dr. Evans disclosed that he is a consultant for PerkinElmer, a genetics company based in Waltham, Mass.

Prenatal assessments for major chromosomal abnormalities have, like a pendulum, swung over the last 50 years between advancements in screening tests and diagnostic procedures.

In the 1960s, screening for advanced maternal age gave way to diagnostic amniocentesis. Maternal serum alpha-fetoprotein screening for neural tube defects came on the scene in the 1970s, and Down syndrome screening and chorionic villus sampling (CVS) followed in the 1980s. Better ultrasound screening markers were used in combination with biochemistry to advance first-trimester screening in the 1990s and 2000s, leading to a significant decline in diagnostic procedures. Now, free fetal DNA measurement, known as noninvasive prenatal screening, or NIPS, has entered the scene. This development, along with advances in the accuracy of diagnostic lab testing through microarray analysis, will soon lead the pendulum to swing back toward more definitive diagnostic procedures that require either CVS or amniocentesis.

Screening tests provide us with odds adjustments, not definitive answers, and are meant for everyone. Diagnostic tests are meant to give us definitive answers, may have risks, and therefore have traditionally been done on "at-risk" patients. Fundamentally, a screening test gives us an impression, while a diagnostic test gives us harsh reality. As always, there will be trade-offs. No approach is perfect, and no one size fits everyone.

Risks beyond Down syndrome

During the prenatal period, patients will often say, "I’m concerned about having a baby with Down syndrome." What they really mean is that they’re concerned about having a baby with a serious problem, and Down syndrome is the name they know.

Serious problems – a Mendelian disorder, a multifactorial disorder, or a major chromosomal abnormality – affect 2%-3% of all births. Less serious chromosomal abnormalities affect 5%-6% of births. Although advanced maternal age is no longer the sole criterion for deciding who should be offered diagnostic testing, age still is a principal factor for risk determination.

At age 35, the chance of having a baby with Down syndrome is 1 in 380, but the chance of having any chromosomal abnormality detectable by karyotype is 1 in 190. For a 30-year-old, the chance of having a baby with any chromosomal abnormality is 1 in 380, and for a 40-year-old, the risk is 1 in 65.

With the first-trimester screening approach that combines maternal serum free beta-human chorionic gonadotropin (free beta-HCG) and pregnancy-associated plasma protein A (PAPP-A) with fetal nuchal translucency measurement, we are able to detect upward of 85% of fetuses with Down syndrome, or trisomy 21. Yet the disorder is only one of a large number of chromosomal abnormalities observed.

In a recent single-center study of more than 20,000 first-trimester screenings, 5.6% were positive for Down syndrome risk. Of those who subsequently had an amniocentesis or CVS, we found 4% had an abnormal karyotype. Interestingly, 40% of the time the abnormality was not Down syndrome, but another chromosomal abnormality. Similar analyses for trisomies 13 and 18 – the other major abnormalities targeted in first-trimester combined screening – yielded similar statistics (Prenat. Diagn. 2013;33:251-6).

All told, of the screen-positive pregnancies found to have an abnormal karyotype, at least 30% had chromosomal abnormalities outside of those for which they were screen positive. Such findings speak to the limitations of screening as opposed to diagnostic testing, and have implications for patient counseling. Patients should be counseled about the possibility of all chromosome abnormalities – not just Down syndrome.

The NIPS rollout

We have known for well over 100 years that fetal cells cross the placental barrier in small numbers, driving the development of what’s currently known as NIPS. The future of NIPS actually lies in an ever-expanding number of disorders, and will eventually end with sequencing the entire genome.

There are two main methods by which NIPS is done. The original and predominant method uses massive parallel shotgun sequencing, known as next-generation sequencing. This method involves whole-genome amplification and collects enormous amounts of information. Investigators are now attempting to direct amplification at the subchromosome level, mimicking some of what microarray analysis can do.

The second approach uses selected probes, or targeted sequencing, to focus on those sections of DNA that are of interest. Although this method may be cheaper in the short run, one drawback is that new probes will need to be created for each new disorder.

Initially, investigators attempted to isolate nucleated fetal cells from the maternal blood and use them for aneuploidy detection. However, a National Institutes of Health–funded fetal-cell isolation study that ended in 2002 reported disappointing results: Fetal-cell isolation methods had low sensitivity and other technological shortcomings. Subsequently, a number of companies attempted to replicate and improve the work, also without much success.

Concurrent with efforts to use fetal-cell isolation to perform NIPS was the discovery, in the late 1990s by Dr. Dennis Lo, upon whose work next-generation sequencing for NIPS is based, of the presence of circulating cell-free fetal DNA in maternal plasma. The concentration of cell-free fetal DNA may be as much as 5%-8% of the total circulating cell-free DNA in maternal plasma, making free fetal DNA a promising source of fetal genetic material for noninvasive prenatal investigation.

The first high-quality trials on the use of free fetal DNA measurement for detection of trisomy 21 were published in the fall of 2011, and demonstrated up to a 99% detection rate for Down syndrome (less for trisomies 18 and 13). Companies subsequently began to manufacture free fetal DNA tests as off-label products, The tests were initially designated as "noninvasive prenatal diagnosis" tools, but experts around the world objected to the "diagnosis" label, and the terminology shifted to "noninvasive prenatal testing" and finally to "noninvasive prenatal screening"– a designation that I believe accurately reflects its current role.

The uptake in utilization of NIPS has been faster than anyone could have predicted: In just over 2 years, several hundred thousand screening tests have been performed. With 98% to 99% sensitivity, NIPS is an excellent screening test for Down syndrome. However, 99% sensitivity does not equate to 99% positive predictive value, that is, the risk that a patient with a positive screen actually has Down syndrome is much lower.

The published studies of NIPS cite false positive rates of 0.2%-1.0%, but this rate will increase as more disorders are screened. Positive predictive value is directly proportional to the underlying risk. For example, a test with 99% sensitivity and 99% specificity (1% false positive rate) means that a 26-year-old woman who has a positive free fetal DNA test actually has as low as an 11% chance of having a baby with Down syndrome. In older women, who have a higher incidence of having a baby with Down syndrome, a positive NIPS result will have a higher positive predictive value of giving birth to a baby with Down syndrome. However, at the current time, NIPS is an excellent screening test for Down syndrome, but it is not ready for universal primary screening in younger women.

Falsely reassuring are the hyped marketing claims in the United States that NIPS "replaces amnio" and eliminates the need for a nuchal translucency (NT) test. When clinicians abandon performing or referring for a high-quality NT measurement, they can miss or significantly delay the diagnosis of twins/zygosity, growth abnormalities and placentation, cardiac defects, and numerous other anomalies. Similarly, when patients who otherwise would have opted to have CVS or amniocentesis forego having the procedure and have NIPS performed instead, they may regret this decision.

The danger that overreliance on screening tests may paradoxically increase the number of births of babies with otherwise detectable problems was raised in a study led by the late Dr. George Henry, an obstetrician-geneticist in Denver. Curious about declining rates of diagnostic testing in his own practice, Dr. Henry examined trends in his state and found that while the utilization of diagnostic tests had plummeted by 70% over about 15 years, the birth rate of babies born with Down syndrome during this time had doubled among mothers over age 35, and stayed the same among mothers under age 35.

A sociologist-researcher examined the trend that Dr. Henry identified and found that the rise in Down syndrome births was not due to an increase in women electing to keep these pregnancies, but to the fact that the abnormality was not detected in the first place (Fetal Diagn. Ther. 2008;23:308-15). The same risk of screening tests replacing definitive diagnosis exists today with the uptake of NIPS.

The impact of microarray

Microarray technologies have been developed over the past decade. The National Institutes of Health study on chromosomal microarray versus karyotyping, published a year ago, is a game changer. The microarray is analogous to a 15-fold magnifying glass on the karyotype. While the smallest piece of a chromosome that can be evaluated by karyotype analysis is about 5 million base pairs, microarray analysis zooms in on about 200,000 base pairs, allowing us to see small genomic deletions and duplications (copy number variants) that we’ve never seen before.

The trial looked at upward of 4,000 women undergoing CVS or amniocentesis at one of 29 centers. Each diagnostic sample was split in two, with standard karyotyping performed on one portion and chromosomal microarray on the other (N. Engl. J. Med. 2012;367:2175-84).

There were several significant findings: One is that almost one-third of patients have a copy variation now known to be benign. Another is that 2.5% of women who had ultrasound-identified anomalies and normal karyotypes had microdeletion/duplications on the microarray that were clearly associated with a known clinical problem. Moreover, another 3.2% had gains and losses of potential clinical significance. As such, close to 6% of women with ultrasound-identified anomalies and a normal karyotype had clinically relevant copy number variants that only the microarray could find.

When microarray analysis was performed in women whose only indication for prenatal diagnosis was advanced age or an abnormal result on Down syndrome screening, as opposed to an ultrasound-detected anomaly, 0.5%, or 1 in 200, were found to have a pathogenic abnormality that otherwise would have been missed. This is significant: Previously, with karyotyping, we quoted patients a minimum of 1/500 risk of finding a clinically relevant chromosomal anomaly even if the combined report suggested much lower risks of Down syndrome. Now, with microarray, that risk is 1/200. Other studies already published or about to be published show this same level of risk determined by microarray.

With any new technological advance, we get our numerators of the problem before our denominators. The first cases published are those in which clinical or laboratory findings are associated with an abnormality. Only then do researchers go back and look at cases with those findings to test these associations – and only then are the markers sometimes found not to be associated. It will take a number of years to acquire a sizable database on microarray-detected copy number variants. In the end, microarray may help us to explain many of the approximately 1% of serious problems that we have been unable to diagnose until now.

Current decision-making

Ultimately, the future lies with routine, complete genomic sequencing that provides a detailed view of the fetal genome. This is likely to be about 7-10 years away, and the main question on the table is whether it will be performed invasively or with a maternal blood sample.

Today, when a 35-year-old woman comes into my office early in her pregnancy, I will tell her that the risk of having a baby with a chromosomal abnormality is 1 in 190. If she wants to know more, I will explain that the second-trimester quad screen will detect 60% of Down syndrome cases, that the first-trimester combined screen can identify 85% or more, and that the free fetal DNA test will get closer to 99%.

Despite the significant advances in screening, all of these options are still the fundamental equivalent of a Gallup poll. If the patient wants a definitive answer, she will need either an amniocentesis or a CVS, which, in experienced hands, have been shown through an increasing number of studies to be of equal risk. Because there is no such thing as "no" risk when it comes to prenatal diagnostic testing, the question that each patient must answer is, "Where do you want to put that risk – in the test or gambling on the outcome?"

We also have a great cultural divide in the United States. Some people want to know everything, others want to know nothing. Our affluent patients are getting older. Our poorer patients are getting younger. Some people will pay whatever it takes to get the answers they want, while others can or will not pay a dime beyond what their insurance will cover.

There is no one algorithm that can handle these two extremes of patients. Right now, many programs around the country have seen a diminishing number of patients having diagnostic testing – a phenomenon I believe is the result of a false sense of confidence, of people being lulled by the faulty argument that screening protocols can find Down syndrome, so what else is there to know?

Ultimately, a model for younger women would start with "contingent screening," in which patients could start with first-trimester combined screening and move straight to CVS if found to be at "high risk," and end testing if found to be at "low risk." Women who fall in the middle could undergo either free fetal DNA analysis or CVS, and we are developing methods to improve the mathematical processing of data to improve the sensitivity and specificity of all screening programs.

In my program, CVS and microarray are now offered to all patients regardless of age, as the risk that a microarray will find a significant abnormality is at least 1/200, which is the risk we have been quoting to 35-year-old patients for nearly 50 years.

Dr. Evans is president of the Fetal Medicine Foundation and International Fetal Medicine and Surgery Society Foundation, and professor of obstetrics and gynecology at the Icahn School of Medicine at Mount Sinai, New York. Dr. Evans disclosed that he is a consultant for PerkinElmer, a genetics company based in Waltham, Mass.

Prenatal assessments for major chromosomal abnormalities have, like a pendulum, swung over the last 50 years between advancements in screening tests and diagnostic procedures.

In the 1960s, screening for advanced maternal age gave way to diagnostic amniocentesis. Maternal serum alpha-fetoprotein screening for neural tube defects came on the scene in the 1970s, and Down syndrome screening and chorionic villus sampling (CVS) followed in the 1980s. Better ultrasound screening markers were used in combination with biochemistry to advance first-trimester screening in the 1990s and 2000s, leading to a significant decline in diagnostic procedures. Now, free fetal DNA measurement, known as noninvasive prenatal screening, or NIPS, has entered the scene. This development, along with advances in the accuracy of diagnostic lab testing through microarray analysis, will soon lead the pendulum to swing back toward more definitive diagnostic procedures that require either CVS or amniocentesis.

Screening tests provide us with odds adjustments, not definitive answers, and are meant for everyone. Diagnostic tests are meant to give us definitive answers, may have risks, and therefore have traditionally been done on "at-risk" patients. Fundamentally, a screening test gives us an impression, while a diagnostic test gives us harsh reality. As always, there will be trade-offs. No approach is perfect, and no one size fits everyone.

Risks beyond Down syndrome

During the prenatal period, patients will often say, "I’m concerned about having a baby with Down syndrome." What they really mean is that they’re concerned about having a baby with a serious problem, and Down syndrome is the name they know.

Serious problems – a Mendelian disorder, a multifactorial disorder, or a major chromosomal abnormality – affect 2%-3% of all births. Less serious chromosomal abnormalities affect 5%-6% of births. Although advanced maternal age is no longer the sole criterion for deciding who should be offered diagnostic testing, age still is a principal factor for risk determination.

At age 35, the chance of having a baby with Down syndrome is 1 in 380, but the chance of having any chromosomal abnormality detectable by karyotype is 1 in 190. For a 30-year-old, the chance of having a baby with any chromosomal abnormality is 1 in 380, and for a 40-year-old, the risk is 1 in 65.

With the first-trimester screening approach that combines maternal serum free beta-human chorionic gonadotropin (free beta-HCG) and pregnancy-associated plasma protein A (PAPP-A) with fetal nuchal translucency measurement, we are able to detect upward of 85% of fetuses with Down syndrome, or trisomy 21. Yet the disorder is only one of a large number of chromosomal abnormalities observed.

In a recent single-center study of more than 20,000 first-trimester screenings, 5.6% were positive for Down syndrome risk. Of those who subsequently had an amniocentesis or CVS, we found 4% had an abnormal karyotype. Interestingly, 40% of the time the abnormality was not Down syndrome, but another chromosomal abnormality. Similar analyses for trisomies 13 and 18 – the other major abnormalities targeted in first-trimester combined screening – yielded similar statistics (Prenat. Diagn. 2013;33:251-6).

All told, of the screen-positive pregnancies found to have an abnormal karyotype, at least 30% had chromosomal abnormalities outside of those for which they were screen positive. Such findings speak to the limitations of screening as opposed to diagnostic testing, and have implications for patient counseling. Patients should be counseled about the possibility of all chromosome abnormalities – not just Down syndrome.

The NIPS rollout

We have known for well over 100 years that fetal cells cross the placental barrier in small numbers, driving the development of what’s currently known as NIPS. The future of NIPS actually lies in an ever-expanding number of disorders, and will eventually end with sequencing the entire genome.

There are two main methods by which NIPS is done. The original and predominant method uses massive parallel shotgun sequencing, known as next-generation sequencing. This method involves whole-genome amplification and collects enormous amounts of information. Investigators are now attempting to direct amplification at the subchromosome level, mimicking some of what microarray analysis can do.

The second approach uses selected probes, or targeted sequencing, to focus on those sections of DNA that are of interest. Although this method may be cheaper in the short run, one drawback is that new probes will need to be created for each new disorder.

Initially, investigators attempted to isolate nucleated fetal cells from the maternal blood and use them for aneuploidy detection. However, a National Institutes of Health–funded fetal-cell isolation study that ended in 2002 reported disappointing results: Fetal-cell isolation methods had low sensitivity and other technological shortcomings. Subsequently, a number of companies attempted to replicate and improve the work, also without much success.

Concurrent with efforts to use fetal-cell isolation to perform NIPS was the discovery, in the late 1990s by Dr. Dennis Lo, upon whose work next-generation sequencing for NIPS is based, of the presence of circulating cell-free fetal DNA in maternal plasma. The concentration of cell-free fetal DNA may be as much as 5%-8% of the total circulating cell-free DNA in maternal plasma, making free fetal DNA a promising source of fetal genetic material for noninvasive prenatal investigation.

The first high-quality trials on the use of free fetal DNA measurement for detection of trisomy 21 were published in the fall of 2011, and demonstrated up to a 99% detection rate for Down syndrome (less for trisomies 18 and 13). Companies subsequently began to manufacture free fetal DNA tests as off-label products, The tests were initially designated as "noninvasive prenatal diagnosis" tools, but experts around the world objected to the "diagnosis" label, and the terminology shifted to "noninvasive prenatal testing" and finally to "noninvasive prenatal screening"– a designation that I believe accurately reflects its current role.

The uptake in utilization of NIPS has been faster than anyone could have predicted: In just over 2 years, several hundred thousand screening tests have been performed. With 98% to 99% sensitivity, NIPS is an excellent screening test for Down syndrome. However, 99% sensitivity does not equate to 99% positive predictive value, that is, the risk that a patient with a positive screen actually has Down syndrome is much lower.

The published studies of NIPS cite false positive rates of 0.2%-1.0%, but this rate will increase as more disorders are screened. Positive predictive value is directly proportional to the underlying risk. For example, a test with 99% sensitivity and 99% specificity (1% false positive rate) means that a 26-year-old woman who has a positive free fetal DNA test actually has as low as an 11% chance of having a baby with Down syndrome. In older women, who have a higher incidence of having a baby with Down syndrome, a positive NIPS result will have a higher positive predictive value of giving birth to a baby with Down syndrome. However, at the current time, NIPS is an excellent screening test for Down syndrome, but it is not ready for universal primary screening in younger women.

Falsely reassuring are the hyped marketing claims in the United States that NIPS "replaces amnio" and eliminates the need for a nuchal translucency (NT) test. When clinicians abandon performing or referring for a high-quality NT measurement, they can miss or significantly delay the diagnosis of twins/zygosity, growth abnormalities and placentation, cardiac defects, and numerous other anomalies. Similarly, when patients who otherwise would have opted to have CVS or amniocentesis forego having the procedure and have NIPS performed instead, they may regret this decision.

The danger that overreliance on screening tests may paradoxically increase the number of births of babies with otherwise detectable problems was raised in a study led by the late Dr. George Henry, an obstetrician-geneticist in Denver. Curious about declining rates of diagnostic testing in his own practice, Dr. Henry examined trends in his state and found that while the utilization of diagnostic tests had plummeted by 70% over about 15 years, the birth rate of babies born with Down syndrome during this time had doubled among mothers over age 35, and stayed the same among mothers under age 35.

A sociologist-researcher examined the trend that Dr. Henry identified and found that the rise in Down syndrome births was not due to an increase in women electing to keep these pregnancies, but to the fact that the abnormality was not detected in the first place (Fetal Diagn. Ther. 2008;23:308-15). The same risk of screening tests replacing definitive diagnosis exists today with the uptake of NIPS.

The impact of microarray

Microarray technologies have been developed over the past decade. The National Institutes of Health study on chromosomal microarray versus karyotyping, published a year ago, is a game changer. The microarray is analogous to a 15-fold magnifying glass on the karyotype. While the smallest piece of a chromosome that can be evaluated by karyotype analysis is about 5 million base pairs, microarray analysis zooms in on about 200,000 base pairs, allowing us to see small genomic deletions and duplications (copy number variants) that we’ve never seen before.

The trial looked at upward of 4,000 women undergoing CVS or amniocentesis at one of 29 centers. Each diagnostic sample was split in two, with standard karyotyping performed on one portion and chromosomal microarray on the other (N. Engl. J. Med. 2012;367:2175-84).

There were several significant findings: One is that almost one-third of patients have a copy variation now known to be benign. Another is that 2.5% of women who had ultrasound-identified anomalies and normal karyotypes had microdeletion/duplications on the microarray that were clearly associated with a known clinical problem. Moreover, another 3.2% had gains and losses of potential clinical significance. As such, close to 6% of women with ultrasound-identified anomalies and a normal karyotype had clinically relevant copy number variants that only the microarray could find.

When microarray analysis was performed in women whose only indication for prenatal diagnosis was advanced age or an abnormal result on Down syndrome screening, as opposed to an ultrasound-detected anomaly, 0.5%, or 1 in 200, were found to have a pathogenic abnormality that otherwise would have been missed. This is significant: Previously, with karyotyping, we quoted patients a minimum of 1/500 risk of finding a clinically relevant chromosomal anomaly even if the combined report suggested much lower risks of Down syndrome. Now, with microarray, that risk is 1/200. Other studies already published or about to be published show this same level of risk determined by microarray.

With any new technological advance, we get our numerators of the problem before our denominators. The first cases published are those in which clinical or laboratory findings are associated with an abnormality. Only then do researchers go back and look at cases with those findings to test these associations – and only then are the markers sometimes found not to be associated. It will take a number of years to acquire a sizable database on microarray-detected copy number variants. In the end, microarray may help us to explain many of the approximately 1% of serious problems that we have been unable to diagnose until now.

Current decision-making

Ultimately, the future lies with routine, complete genomic sequencing that provides a detailed view of the fetal genome. This is likely to be about 7-10 years away, and the main question on the table is whether it will be performed invasively or with a maternal blood sample.

Today, when a 35-year-old woman comes into my office early in her pregnancy, I will tell her that the risk of having a baby with a chromosomal abnormality is 1 in 190. If she wants to know more, I will explain that the second-trimester quad screen will detect 60% of Down syndrome cases, that the first-trimester combined screen can identify 85% or more, and that the free fetal DNA test will get closer to 99%.

Despite the significant advances in screening, all of these options are still the fundamental equivalent of a Gallup poll. If the patient wants a definitive answer, she will need either an amniocentesis or a CVS, which, in experienced hands, have been shown through an increasing number of studies to be of equal risk. Because there is no such thing as "no" risk when it comes to prenatal diagnostic testing, the question that each patient must answer is, "Where do you want to put that risk – in the test or gambling on the outcome?"

We also have a great cultural divide in the United States. Some people want to know everything, others want to know nothing. Our affluent patients are getting older. Our poorer patients are getting younger. Some people will pay whatever it takes to get the answers they want, while others can or will not pay a dime beyond what their insurance will cover.

There is no one algorithm that can handle these two extremes of patients. Right now, many programs around the country have seen a diminishing number of patients having diagnostic testing – a phenomenon I believe is the result of a false sense of confidence, of people being lulled by the faulty argument that screening protocols can find Down syndrome, so what else is there to know?

Ultimately, a model for younger women would start with "contingent screening," in which patients could start with first-trimester combined screening and move straight to CVS if found to be at "high risk," and end testing if found to be at "low risk." Women who fall in the middle could undergo either free fetal DNA analysis or CVS, and we are developing methods to improve the mathematical processing of data to improve the sensitivity and specificity of all screening programs.

In my program, CVS and microarray are now offered to all patients regardless of age, as the risk that a microarray will find a significant abnormality is at least 1/200, which is the risk we have been quoting to 35-year-old patients for nearly 50 years.

Dr. Evans is president of the Fetal Medicine Foundation and International Fetal Medicine and Surgery Society Foundation, and professor of obstetrics and gynecology at the Icahn School of Medicine at Mount Sinai, New York. Dr. Evans disclosed that he is a consultant for PerkinElmer, a genetics company based in Waltham, Mass.