To NIPT or not to NIPT?
A lay-person’s inquiry into the accuracy of non-invasive prenatal screening for sex chromosome aneuploidies.
Abstract: This article reviews the state of Non-Invasive Prenatal Testing (NIPT) for a particular Sex Chromosome Aneuploidy: Turner Syndrome. Much of this material was gathered after we received a positive NIPT screen for Turner Syndrome in our first trimester. We found that, when it comes to communicating the limitations of the NIPT for sex chromosome aneuploidies, the medical community and the businesses that serve it, have done a poor job of educating their customers. This is an attempt to dispel some of the mystery, and to present some facts that we wish we had known before we decided to take the NIPT.
As medical care in the developed world becomes increasingly sophisticated, pregnant women are being offered a long list of tests. Some of these tests — such as the first-trimester ultrasound, and the second-trimester anatomical scan — are well-known; expectant parents eagerly look forward to them. The first trimester ultrasound, performed between the 11th and 14th weeks of pregnancy, is when most mothers see the developing fetus for the first time. In the second trimester ultrasound, the doctor can tell if you have a boy or a girl. These scans allow doctors to observe key organs of the developing fetus and to screen for abnormalities. For instance, in the first trimester ultrasound, a high nuchal translucency (NT) measurement, which refers to the width of the baby’s neck fold, suggests the possibility of Down Syndrome.
Doctors are careful to mention, however, that an NT measurement performed over an ultrasound is merely a screening test and not a diagnostic test. If a higher-than-normal NT measurement is found, the mother is offered the option of invasive tests to conclusively confirm or reject the hypothesis of a chromosomal abnormality. There are two main kinds of invasive procedures: Chorionic Villus Sampling (CVS) which extracts and analyzes material from the mother’s placenta, and Amniocentesis, which extracts and analyzes the amniotic fluid surrounding the fetus in the womb. In both CVS and amnio, doctors are looking for the genetic composition of the extracted material. An extra chromosome 21, for example, confirms the existence of Down Syndrome. These tests are significantly more accurate than ultrasound screenings, and are treated as diagnostic all over the world. Unfortunately, going through a diagnostic procedure has drawbacks: Though the procedures are short, some patients fear the prospect of having a needle inserted into their womb. Moreover, these procedures carry a small risk of miscarriage, and not all parents are willing to submit to them.
The NIPT, A new non-invasive testing method for trisomies
Over the past decade, a convenient, non-invasive test has emerged as a possible alternative. It has several names, the most common ones being Non-Invasive Prenatal Testing (NIPT), and Cell-free Fetal DNA Testing (CfF-DNA). The new test is based on a genetic examination of a blood sample taken from a pregnant woman. The procedure is painless and indistinguishable from a conventional blood draw: a couple of vials of blood are extracted from the mother’s arm, and as such, it poses no risk to the baby. The NIPT has, justifiably, been touted as a convenient way to reduce the number of invasive procedures. For example, if an expectant mother has a normal ultrasound, and screens negative for certain genetic conditions on the NIPT, her doctor is unlikely to prescribe an invasive procedure.
Today, all the major NIPT companies offer tests for trisomy-21 (Down Syndrome), trisomy-13 (Patau Syndrome) and trisomy-18 (Edwards Syndrome). The accuracy of these tests depends upon the condition being tested, the maternal age and the gestational age when the tests are carried out. This information is made available on the companies’ websites and is provided to patients by their care providers. Crucially, unlike CVS or Amnio, doctors consider the NIPT a screening test, not a diagnostic procedure. If a pregnant woman receives a positive NIPT screen for a genetic condition, such as Down Syndrome, she has two options: Undergo CVS or Amnio, or wait until the end of the pregnancy and have the baby examined soon after delivery. If a blood sample is to be collected after delivery, it is sometimes recommended that, rather than extracting blood from the baby, the labs use blood from the umbilical cord.
The methods used to analyze the blood samples are proprietary: The intellectual property is owned by a handful of companies — Illumina, Ariosa Diagnostics, Natera, and Sequenom being the dominant names— that have, in turn, licensed their methods to a growing number of smaller labs. A blood sample is taken from the pregnant woman, shipped to one of these labs, and analyzed within about five business days. Hospitals appear to have contracts with one or two labs, and genetic counsellors talk to expectant parents about the option to take the NIPT toward the end of their first trimester. The counsellors are required to clarify that the NIPT is not diagnostic. They also provide the parents-to-be with flyers and information about the NIPT lab with which their hospital has a contract.
Extending the NIPT to Sex Chromosome Aneuploidies
Exact dates are difficult to track down, but until about 2012, the NIPT screens were available for trisomy-21, trisomy-18 and trisomy-13. After 2012, some companies enhanced their NIPT offering by including an option to screen for abnormalities in the sex chromosomes (referred to as Sex Chromosome Aneuploidies or SCAs). These include Turner Syndrome, Kleinfelter Syndrome, Triple X, and XYY Syndrome. Nowadays, there is an additional option to screen for micro-deletions in some chromosomes.
How reliable are these new tests for SCAs? Unfortunately, unlike the NIPT for trisomies — for which we have more data about the test’s accuracy — we know very little about the accuracy of the NIPT for SCAs. There are research papers and meta-analyses on the accuracy of the NIPT for SCAs, but the results are not always presented in a manner accessible to the general public. To make matters worse, advertising around the new NIPT offerings remains ambiguous, if not outrightly misleading: Sequenom’s MaterniT21 test, for example, is said to provide “highly accurate answers to important questions”, and that mothers can “rest assured in five easy steps.” In its marketing material, Illumina promises, that its Verifi prenatal screening tests use proprietary technology that is “Easy, Fast, Safe and, Accurate”. The ads do not specify whether these favorable qualifiers (especially “Accurate”) apply only to the screening of trisomies (for which more comprehensive information is available) or whether they extend to SCAs (for which accuracy numbers have still not been established).
Turner Syndrome
Before proceeding further, it is appropriate to describe Turner Syndrome in a little more detail. Most people have 46 chromosomes, composed of 22 pairs of autosomes and one pair of sex chromosomes. The two sex chromosomes are designated (X, X) in females and (X, Y) in males. Fetuses with Turner syndrome are all women with one significant difference: They have only a single X chromosome. Because they lack a second X chromosome, they are likely to exhibit one or more unusual conditions, including lower-than-normal height, hypoplastic left-heart syndrome (which affects blood flow through the heart), coarctation of the aorta (which is a narrowing of the main blood vessel that carries blood out of the heart), a single horse-shoe shaped kidney, a low hairline at the back of the head, ovarian dysgenesis, and some others. Some patients develop problems with their vision and hearing. The expected life span of infants born with Turner Syndrome is slightly lower than normal, primarily because they are more likely to develop heart or kidney problems. Developmentally, women with Turner Syndrome are considered normal, though they might score slightly lower on mathematical and abstract thinking ability. Women who have Turner Syndrome in all their cells are unable to conceive children through sex.
Many of these conditions are not life-threatening and can be mitigated by modern medicine. The height deficiency is addressed by administering Human Growth Hormone (HGH) injections to a Turner child. The narrowing of the aorta can be surgically corrected within the first few weeks of life. Not all patients exhibit all these symptoms, especially if they have mosaic Turner Syndrome, i.e., some of their cells are missing an X chromosome (designated as 45X) while others have the usual pair of X chromosomes (designated as 46XX). The absence of the second X chromosome in the Turner cells is thought to be due to a random error during conception. The probability of this error is considered to be unrelated to maternal age, and is not known to affect subsequent pregnancies.
It is not always obvious that a girl has Turner Syndrome. Indeed, in the past, many women did not know that they had Turner Syndrome until they asked for a checkup because of delayed puberty or because they failed to menstruate. In the impersonal language of medical research, the prenatal prognosis for Turner Syndrome is grim while the postnatal prognosis is good. What this means is that, a fetus with Turner Syndrome is very likely going to die in the womb and result in a miscarriage (or a missed miscarriage) before the end of the second trimester. However, if the baby survives through the pregnancy, then her chances of having a normal life are much better than, for example, a patient with Down Syndrome. Even with the relative mildness of Turner Syndrome in a live newborn, however, abortions of Turner fetuses (diagnosed via amniocentesis) were relatively common in the 1980s and 1990s (source1, source2).
In-utero, one of the most common telltale signs of Turner Syndrome is a fluid-filled sac, called a cystic hygroma, near the neck of the fetus. Doctors can usually spot a hygroma in the first trimester ultrasound. Though the presence of a hygroma does not confirm Turner Syndrome, doctors advise further testing to be absolutely sure. In mild Turner cases, the hygroma resolves or becomes smaller over the term of the pregnancy, leaving behind a fold of skin around the newborn girl’s neck. In severe cases, the hygroma continues to grow and develops into a fatal condition called “immune hydrops”, which claims the baby’s life in the womb.
The NIPT for Turner Syndrome
The NIPT for Turner Syndrome is offered as an additional test that is optionally bundled with the NIPT for trisomies. It is also performed on a blood sample similarly extracted from the mother’s arm. The NIPT companies advertise that this test is very accurate, a claim that is propagated to prospective parents by genetic counselors at maternity clinics. In some states, genetic counselors are not required to disclose whether they are being paid by the NIPT companies to do their work. In our own experience, we did not doubt that genetic counsellors at hospitals were acting in good faith, in describing these new tests to us, and pointing us to the companies’ websites for more information. However, as we realized after visiting the websites of the prenatal testing labs, it is easy to come away with misguided notions of the accuracy of the NIPT for SCAs.
This is partially because many people, even those working in the medical field, don’t quite understand what it means for a test to be “accurate”. The Verifi screening test for Monosomy X (Turner Syndrome), for example, promises a sensitivity of 95% and a specificity of 99%. Sensitivity is defined as the probability that the test will detect the presence of Turner Syndrome, while specificity is the probability that the test will correctly identify blood samples without Turner Syndrome. When faced with these numbers, it is natural for most people (even some medical practitioners) to think that this is a very accurate test. Indeed, there are stories on message boards, in which care providers have told patients that a positive NIPT screen for Turner Syndrome meant a 95% chance that their baby girl had Turner. Naturally, this verdict causes a lot of anguish.
As it turns out, a test should not be judged only by its sensitivity and specificity values. It is necessary to examine the positive and negative predictive values. Simply put, the Positive Predictive Value (PPV) is the likelihood that a patient who screened positive for Turner Syndrome, really does have a fetus with the disease. The Negative Predictive Value (NPV) is the likelihood that a patient who screened negative for Turner Syndrome, really does not have a fetus with the disease. We calculate the PPV and NPV as shown in the diagram below. Here, TP is the number of True Positive results observed after testing a large number of patients, FP is the number of False Positive screens, and so on. Note how specificity and sensitivity formulaē use values along the columns, while the PPV and NPV formulaē use values along the rows.
Another way to calculate the PPV for a screening test — which explicitly shows the relationship between sensitivity/specificity and PPV/NPV — is to apply Bayes Rule, a statistical technique attributed to Thomas Bayes, an 18th century minister and philosopher. In the equations below, the notation “|” should be read as “given that”:
Thus, to report the PPV, it is not sufficient to just know the sensitivity and specificity. It is also necessary to know the prevalence of the disease in the tested population. To know the PPV of the NIPT screen, it is necessary to know what fraction of all screens were true positives and what fraction were false positives. To know whether a test was a true positive or a false positive, you would have to follow up with the expectant parents after they have had an invasive (diagnostic) procedure, or after their baby has been delivered, or in more unfortunate circumstances, after the baby has miscarried.
If you look at the PPV formula closely, you will notice that, if the disorder is very rare, i.e., if the prevalence is very small, then the positive predictive value can be low, even when the specificity and sensitivity are in the high 90’s. The labs that administer the NIPT are aware of the importance of PPVs and NPVs; Illumina, for example, provides a table containing the PPV and NPV values of the NIPT for trisomies at various stages of a pregnancy. For Turner Syndrome, however, Illumina does not provide the PPV and NPV.
Unfortunately, the messaging on many labs’ websites does not explicitly clarify that, from the patient’s point of view, the PPV and NPV are much more important than the raw sensitivity and specificity values. Even when all the tests have very high sensitivity and specificity, the PPV’s for sex chromosome abnormalities can be very different from the ones published for the trisomies. The lack of PPV values creates a blind spot: When prospective parents are asked whether they want to be tested for sex chromosome abnormalities (in addition to the more mature test for the trisomies), they have no quantifiable basis on which to say yes or no. From the point of view of the patient, the NIPT procedure for SCAs like Turner Syndrome looks exactly like the screening for trisomies. You could be forgiven for assuming that, like the NIPT for trisomies, experts have quantified the accuracy of the new tests for SCAs.
What, if anything, do we know about the PPV of Turner Syndrome?
There is an important difference when it comes to computing the PPV for the trisomies versus those for Turner Syndrome: The medical community seems to agree on the prevalence of the trisomies in the population. In the case of Turner Syndrome, however, prevalence information is more elusive. This might well be because (unlike babies with trisomy 13, 18 and 21) Turner babies might not be diagnosed at birth, and might discover their condition later in life. Without the correct prevalence information, it is difficult to predict the accuracy of a test (See the PPV formula above).
The quest for leads on the prevalence of Turner Syndrome takes us farther back, to research papers written several years before NIPTs began to be offered. These papers are quite tough to parse for non-specialist readers, or even for nurse practitioners and genetic counselors who advise pregnant couples. Instead, many genetic counsellors and care providers direct prospective parents to a handy PPV calculator developed by the Perinatal Quality Foundation. On their website, you choose the condition you are interested in, say Monosomy-X (Turner Syndrome). You are also asked to choose the maternal age. Older pregnant women are more likely to have babies with Down Syndrome, but no such correlation is known for Turner Syndrome. Here is what the page looks like:
The sensitivity and specificity numbers are taken from the cited paper by Gil et al. (2015). If you press the “Calculate” button, the Bayes formula (described above) is applied and you are shown a PPV of about 40.9%. Compare that with the 90.3% sensitivity value and the 99.77% specificity value. Given the sensitivity and specificity, most people would be hard pressed to conclude that their NIPT screen was less reliable than a coin toss. (For a more intuitive explanation of why this is the case, without resorting to formulas, see Katherine Hobson’s explanation of how accurate tests can be mostly wrong.)
The situation gets murkier as we dig a little deeper. The prevalence value used in the above calculation is 1 in 568 cases at 16 weeks of gestation. Where does this value come from? Answering this question took a bit of detective work. Unless this is a rare coincidence, we found that this prevalence value could be computed from the results reported in a 1996 paper by Gravholt et al., though it is never explicitly reported. The paper analyzes Turner Syndrome tests performed over several years in Denmark. There, the syndrome was diagnosed either via an ultrasound, or amniocentesis or chorionic villus sampling (CVS). Particularly, amongst 40,357 amniocentesis procedures, 71 fetuses were diagnosed with Turner Syndrome, leading to a prevalence of 71/40357 or 1 in 568.41. If the Perinatal Quality Foundation has a different reason or source for using a prevalence of 1 in 568 in their calculations, we have been unable to track it down.
While other findings from that paper are very illuminating, the key point here is that prevalence is a statistical estimate. In particular, the prevalence of 1 in 568 fetuses at 16 weeks gestational age, actually means that the test subjects underwent amniocentesis when they were at an average of 16 weeks — not exactly 16 weeks — pregnant. Efforts at interpreting this prevalence are complicated by the fact that most fetuses with Turner Syndrome die in the womb before the end of the second trimester, with week 13 being the most lethal period. Thus, if you had received a positive NIPT screen for Turner Syndrome in week 11, but are now in the 29th week of your pregnancy, it might mean that your baby is one of the lucky survivors, or that your NIPT result might be a false positive. Which is more likely?
Some researchers peg the prevalence of Turner Syndrome at birth to 1 in every 2000–2500 live births. If instead of 1 in 568, you substitute a prevalence of 1 in 2000 in the probability formula that we gave above, the positive predictive value drops from 40.9% to 16.41%. We received a positive NIPT screen for Turner syndrome in week 11. A month later, still racked by worry, and neck-deep in the statistical morass surrounding the NIPT, we tried to retain a semblance of perspective by telling ourselves this: The longer our baby stays alive and healthy in the womb, the less likely it is that our positive Turner screen is a true positive. Still, we had to prepare for the true positive result, to make sure that immediate post-delivery care would be available if needed.
Whose responsibility should it be to read, understand and interpret these statistics? Should it be the task of the care provider, or the genetic counselor? Or, should this be left to the NIPT lab that sells the test for a profit?
Two perspectives from the scientific literature
We decided to examine scientific publications that have measured the sensitivity, specificity and predictive values of the NIPT. We observed that many of these publications report a conflict of interest: The experimental studies are sponsored by one or more of the commercial labs offering these tests. In many instances, one or more of the authors are either employed by the labs, or are researchers who have received funding from the labs. We do not intend to cast doubts on the experiments and statistical analyses presented in these papers. Indeed, it is noteworthy that the conflict of interest is explicitly reported. Still, we are left to wonder whether the implicit bias involved in proving the efficacy of a test colors the interpretation, reporting and marketing of the results.
Consider, for example, the work of a study group formed to evaluate the accuracy of massively parallel sequencing of genetic material from the mother’s blood. The group’s name is suitably aspirational, if a little contrived: MELISSA, an acronym for “MatErnal BLood IS Source to Accurately diagnose fetal aneuploidy”. A 2012 paper authored by some of the participants of this study reports high sensitivity and specificity values and advocates the use of cell-free DNA testing for screening of sex chromosome aneuploidies. The authors caution against the low accuracy of the test for rare disorders, but provide no guidance about positive predictive values for those rare disorders (likely because the PPV values cannot be reliably determined). They envision a future in which the NIPT will graduate from being a screening test to a full-fledged diagnostic test, eventually replacing invasive procedures. Several of the paper’s authors were either employed by Verinata Health, or received honoraria for their work from Verinata Health. Verinata was a prenatal testing company that designed the Verifi NIPT test, before it was acquired by the genetic testing company Illumina in 2013, for $450 million.
Compare the conclusions of the MELISSA study with a meta-analysis performed by another group of authors. Notably, the sensitivity and specificity values from this meta-analysis were used by the Perinatal Quality Foundation to develop the helpful PPV calculator that we saw earlier. This group of authors was funded by the Fetal Medicine Foundation, a registered charity. After examining several experimental NIPT studies conducted on various chromosomal abnormalities, the authors conclude: “It may be inappropriate to offer pregnant women screening for sex chromosome aneuploidies by cfDNA testing just because it is feasible.” One reason for their conclusion is that tests for sex chromosome abnormalities like Turner Syndrome have high false positive rates — another way of calling out low PPVs. A second reason, particularly in the case of Turner Syndrome is that, amongst fetuses that survive to birth, the severity is generally mild, compared to the trisomies.
Confusing Reportage from the NIPT Labs
In many parts of the United States, Europe, China and Australia, more and more expectant parents are being steered toward the NIPT. An increasing number of them are being screened for SCAs like Turner Syndrome, while information flow from medical practitioners to the patients remains inadequate.
The labs are not always transparent about the way in which the results are reported to patients. For example, the Verifi test generates one of 3 possible reports for trisomies: “No Aneuploidy Detected”, “Aneuploidy Suspected”, “Aneuploidy Detected”, where “Detected” is an expression of stronger confidence than “Suspected”. For SCA’s, however, Verifi reports only two kinds of results: “No Aneuploidy Detected”, and “Aneuploidy Detected”. Why is this the case? Illumina, the company that now offers the test, does not clarify this distinction in the FAQ about how sex chromosome screens are reported. We would never have discovered this difference in reporting, if not for an inadvertent search result that led us to a research project at Emory University’s Genetics Laboratory.
While writing this article, we discovered, via a rather complicated Google search, that Illumina does mention the difference in reporting, just not directly to their pregnant patients: In Illumina’s marketing materials is a presentation prepared for the flipbook that genetic counselors use when they talk to pregnant couples about the NIPT. The slides clarify the difference in the way results are reported for trisomies and SCAs. There is absolutely no reason why this clarification should not be present in the FAQ, which is the first place many patients go to for clarification.
During our phone call with a very courteous genetic counsellor from Counsyl — the lab that conducted our NIPT — we asked why the reporting was different for trisomies and SCA’s. She said she didn’t know. As it turns out, Counsyl licenses its NIPT technology from Illumina. Compared to the information available on Illumina’s website, the informational content on Counsyl’s website remains pitifully thin, with very little detail about how the tests are carried out, how results are reported, and why they might result in false positives or false negatives.
Why might the NIPT produce a false positive result?
The blood drawn from the mother consists of the mother’s blood cells and a small amount of blood cells from the mother’s placenta. The percentage of placental blood cells in the sample is called the fetal fraction (although, technically, the placental blood cells are not from the fetus. This difference is crucial.). A blood sample is usually considered unfit for NIPT if the fetal fraction is below 8%. In such cases, a second blood draw is requested. However, even if the fetal fraction is sufficient (say 10%), there are several confounding factors that may result in a false positive.
One reason is Confined Placental Mosaicism, which means that the cells of the placenta may have a different genetic composition than those of the fetus. If the cells in the blood sample contain placental cells that lack an X Chromosome, then the NIPT might report a positive screen for Turner Syndrome, even though the fetus has the normal consignment of two X chromosomes.
Another reason for a false positive is a possible “vanishing twin” — a phenomenon in which the pregnancy started out with twins, but one of the twins did not survive. The dead twin’s tissue can be absorbed into the placenta, and its genetic material can find its way into the mother’s bloodstream. Given that chromosomal abnormalities are one of the likely reasons for fetal death, it is theoretically possible that the NIPT picked up a signal from the deceased twin and claimed a positive result for Turner Syndrome, even when the surviving twin does not have the condition.
Yet another cause for a false positive NIPT screen is maternal mosaicism — a condition in which some of the mother’s cells can drop an X chromosome, turning from a normal 46XX to a 45X. Studies show that X chromosome loss tends to be highly correlated with age, increasing as women gets older. Maternal mosaicism could be a culprit in a small number of false positive NIPT screens for Turner Syndrome (source1, source2).
The failure rate of the NIPT varies depending on the algorithm and genetic testing methodology used by the testing company. Sequenom’s MaterniT21 test and Verinata’s Verifi test use a method called Massively Parallel Shotgun Sequencing (MPSS), which cannot reliably differentiate between maternal and fetal DNA. Amusingly, in 2014, doctors at Boston Maternal and Fetal Medicine were able to fool the MaterniT21 and Verifi tests by providing samples from two women who were not pregnant. Both tests duly predicted that the subjects were going to have (healthy) baby girls.
On the other hand, the Panorama test, developed by Natera, uses a method based on Single Nucleotide Polymorphisms (SNPs, pronounced “snips”), which does allow for more reliable differentiation into maternal and fetal cells. Some experimental studies suggest that this might reduce the rate of false positives. To the best of our knowledge, the Panorama test is the only test we found that does report a PPV for Turner Syndrome, which is taken from a published study performed on 17,885 women. (See a sample Panorama test report.) In case you were holding your breath, the value is 50%. Again, you might as well toss a coin.
[Footnote: Some other kinds of maternal mosaicism might contribute to a false negative NIPT screen for other sex chromosome aneuploidies. This is a potentially serious flaw: A false positive screen can be followed up with amniocentesis and thus discovered, but a false negative screen might never receive such a follow up. This is important especially for those SCAs that exhibit no overt symptoms during pregnancy (source). Fortunately, false negatives with the NIPT are exceedingly rare.]
Other options to screen for Turner Syndrome:
Some parents who screen positive for Turner Syndrome via the NIPT, decline invasive testing because of the risk — however small — of miscarriage, or because of fear of the big needle. In such cases, it is recommended that the pregnant mother undergo a Level-2 ultrasound and a Fetal Echo Cardiogram (ECG). The Level 2 ultrasound looks at the baby’s organs, trying to pick up tell-tale stigmata of Turner Syndrome, such as a horse-shoe kidney or a cystic hygroma. The Fetal ECG takes a detailed look at the baby’s heart, listening to the rhythm of the beats, graphing the blood flow through the chambers, examining the valves between the chambers, and looking for possible narrowing of the aorta. Unlike the invasive procedures, these are extended ultrasound sessions with no known risks to the baby.
To a perinatologist, the Level 2 ultrasound and Fetal Echo provide a more comprehensive picture than the usual ultrasound, and more visual detail than the uninformative binary result provided by the NIPT. However, they are still considered screenings only, i.e., doctors cannot absolutely rule out the existence of Turner Syndrome even if none of the stigmata of Turner Syndrome are visible in the Fetal ECG or in the Level Ultrasound. Thus, if the mother rejects amniocentesis and CVS, the only way to be certain about the NIPT screen is to analyze the baby’s blood (or the blood from the umbilical chord) immediately after delivery.
The lack of regulatory oversight
The value proposition of the NIPT is that it reduces the number of invasive procedures performed on pregnant women. This is indeed true: A woman with a normal first trimester ultrasound and a negative NIPT for all the trisomies and all the SCAs is unlikely to be steered toward CVS or amniocentesis. False negatives with the NIPT are extremely rare. The problems are with false positives, i.e., when the NIPT screens positive for one of these conditions, even though they don’t exist in the fetus.
The NIPT labs have good reasons to worry that the results of these tests could be used by parents to terminate pregnancies. Stanford University has reported 3 cases in which women chose to terminate their pregnancy based on a positive NIPT screen for Down Syndrome, only to discover later that the fetus was healthy.
To prevent abortions of healthy fetuses, doctors, and genetic counsellors are instructed to make patients aware that these tests are not diagnostic. If you screen positive for Turner Syndrome, your doctor will suggest an invasive procedure to get confirmation before any decision is made about aborting the fetus. Designating the NIPT as a screen (as opposed to a diagnostic test) is an implicit admission of its unreliability, however the healthcare system goes no further. There seems to be no effort invested in describing (or requiring NIPT labs to describe) to the public just how unreliable the NIPT currently is. Even the more mature test for Down Syndrome is not without controversy, though there is at least some information available about the predictive values of that test. Other than the binary “detected” / “not detected” result, the Turner Syndrome test from most labs provides no information at all.
In this climate of uncertainty, a multi-million dollar business is thriving, without the need for regulatory approval. In the US, for example, the NIPT is exempt from regulation by the FDA. A typical NIPT test for trisomy-21, trisomy-13 and trisomy-18 costs $250–300 after health insurance. Adding a NIPT test for the sex chromosome aneuploidies (Turner Syndrome, Kleinfelter Syndrome, XXX, XYY) incurs additional costs. While the NIPT for trisomies might be covered by health insurance, the NIPT for sex chromosome testing might not.
Meanwhile, companies offering NIPT screening are engaged in intense legal competition over intellectual property, a scenario reminiscent of suing and counter-suing between companies manufacturing mobile phones a few years ago (see diagram). Some of the lawsuits appear to have culminated in licensing agreements. Nevertheless, the litigiousness and competition interfere with scientific progress. Instead of sharing information about their performance with the medical community, the companies may decide to keep the data to themselves, using it “to improve their products or to gain a competitive advantage” (source).
The toll on expectant parents
As we have seen, the positive predictive value of the NIPT screen for Turner Syndrome can be low, even though the test has very high sensitivity and specificity. In the absence of predictive values, women who screen positive for Turner Syndrome are faced with an unpleasant trade-off: Should I undergo an invasive procedure to have a conclusive answer, knowing that the procedure carries a small risk of miscarriage? Or, can I endure the uncertainty and wait until the end of the pregnancy to test blood from my baby’s umbilical cord? The lack of transparency from the NIPT companies makes it very difficult for a pregnant woman trying to resolve this tradeoff.
Upon receiving a positive screen for Turner Syndrome in an NIPT test performed by Counsyl, we contacted the lab for some clarification and sought more informative statistics regarding their screen. The genetic counselor at Counsyl, could offer us no specific information about our tests (other than “Aneuploidy Detected”, which was already printed in the report). When we asked what false positive rates they were observing, the counsellor told us that she would follow up with us. This she did, however, the email did not contain answers to our questions. Instead, we were told that a new policy at Counsyl meant that these issues could not be addressed via email, and could only be discussed on the phone by her manager. At that point, already very anxious about the result, frustrated with Counsyl’s opaque procedures, and busy in scheduling follow-up clinical procedures, we dropped the ball on further communications with Counsyl. A year later, we wonder if, had we persisted and talked to the Counsyl manager, we might have received more illuminating details.
More often, we have asked ourselves why Counsyl was not publishing their predictive values for Turner Syndrome. Are the tests are not reliable enough? Are they privately aware that there simply isn’t enough data (yet) to quantify the accuracy of this test? Does their contract with Illumina forbid them from publishing these values? Is Illumina concerned that revealing these numbers will make them look worse than the competition? Does discussing customer issues on the phone give them plausible deniability, where email might not?
On online forums such as BabyCenter, some mothers confide that they decided to undergo amniocentesis or CVS because it is too difficult to live with the uncertainty, or because they might want to abort the fetus if the positive screen is confirmed. A few couples decide that, since they want to have the baby anyway, they will wait it out, somehow soldiering through anxiety for two more trimesters and conclusively testing the baby after delivery. One cannot pretend that it is easy to make any of these decisions, or to live with their consequences.
A more empathetic and ethical healthcare system than the one we have, would question the wisdom of marketing the NIPT for sex chromosome aneuploidies like Turner Syndrome. A better regulatory system would require the labs to publish the predictive values that they have observed over more than 4 years of administering the new NIPT for SCAs. The labs themselves stand to benefit by following up with mothers after their delivery: It would give them an opportunity to collect incidence data on a statistically significant population and to determine whether their screen was a false positive or not. To this day, we remain puzzled that neither the hospital that directed us to Counsyl, nor Counsyl itself, have followed up to ask whether we had a Turner baby or not. How else can they hope to determine the PPV? As Turner Syndrome is relatively rare, it makes sense to sample as many fetuses as possible, in order to get a reliable estimate of the PPV.
Admittedly, some of these changes are easier hypothesized than implemented. At the very least, it would help if genetic counselors were better educated about the limitations of the NIPT for sex chromosome aneuploidies compared to the more mature NIPT for trisomies. Equipped with knowledge of these limitations, uncertainties, and other screening options, pregnant couples might make better decisions about whether the NIPT makes sense for their situation. As Megan Allyse of the Mayo Clinic recently noted, “Women are not saying, I don’t want the test. They’re saying: I wish I’d known (that the false positive rate was so high). I wish somebody had explained this to me.”
Epilogue: Our positive NIPT screen changed the tone of our pregnancy. Over the sleepless nights that ensued, we tried to fill the statistical gaps while hoping that our pregnancy would go to term. Unwilling to risk any harm to the baby, we declined invasive procedures. The wait for certainty would last seven months. Four days after delivery, we found out, via cord blood karyotyping, that our NIPT screen had been a false positive.
