NEW YORK (GenomeWeb) — Though the reported specificity of sequencing-based non invasive prenatal screening tests is generally more than 99 percent, this doesn't mean that false positive rates will be below one percent for all the disorders these tests now target. In fact, depending on the abnormality in question and the gestational age of a fetus, false positive rates could be as much as 50 percent or more, according to recently published research.
The study, published last week in Genetics in Medicine, illustrated this disparity in a series of patients referred to Quest Diagnostics for follow-up testing after receiving positive NIPT results through various providers. According to the study's authors, their work suggests that physicians and patients may not understand just how likely a positive NIPT screen is to be false if they don't understand the difference between specificity and false-positive rate, or positive predictive value.
In addition, the authors suggest that although sequencing-based NIPTs are marketed as screening tests rather than diagnostics and patients are strongly encouraged to follow up all positive results with confirmation by more invasive cytogenetic testing, companies offering the tests should work harder to disclose more clearly the range of false positive rates or positive predictive value for their tests rather than specificity alone.
Charles Strom, the study's senior author and a senior medical director at Quest, told Clinical Sequencing News that he and his coauthors hoped to present real-world evidence demonstrating the discordance that is possible between NIPT and follow-up invasive diagnoses.
While studies of NIPT do disclose positive predictive value — for instance, a study by Illumina and Tufts researchers published earlier this year in the New England Journal of Medicine found positive predictive values for NIPT were 45.5 percent for trisomy 21 and 40 percent for trisomy 18 — patients and physicians may still be misled by reported sensitivity and specificity.
"Companies are saying the tests are 99 percent specific and the lay public and clinicians extrapolate that to mean that if they have a positive test it means they have a 99 percent chance of an affected fetus. That is just not the case," Strom said.
He and his colleagues' brief Genetics in Medicine report compared initial NIPT results in a sequential series of 109 patients with the results of follow-up testing through a Quest Diagnostics cytogenetics lab. Most study subjects had received positive NIPT results but a handful had received negative NIPT and positive ultrasound results.
NIPT providers were listed for only 42 of the cases, and included 20 cases tested by Natera, 13 by Ariosa Diagnostics, eight by Sequenom, and one by Illumina.
Among the cohort, the most common positive NIPT result was trisomy 21, in 41 cases, followed by trisomy 18 in 25 cases, and trisomy 13 in 16 cases. There were also 16 cases of sex chromosome aneuploidy, three cases of trisomy 16, two cases of monosomy 21, and one case each of triploidy and microdeletion of 22q11.2.
Based on the Quest labs' results, only 38 of the 41 NIPT-positive trisomy 21 cases were true positives by cytogenetics. Far fewer of the other aneuploidies — 16 of the 25 trisomy 18 cases, less than half the trisomy 12 cases, and less than 40 percent of the sex chromosome aneuploidy cases — were positive by cytogenetic testing.
According to the study authors, such results are not unexpected statistically, and don't contradict the reported sensitivity and specificity of available sequencing-based NIPT; based on the expected prevalence of some of these aneuploidies at different points in gestation, one would expect positive predictive values far lower than the tests' intrinsic specificity.
For example, the authors wrote, prevalence of trisomies 21, 18, and 13 for a 35-year old woman with a 10-week old fetus are about one in 185, one in 470, and one in 1,500 respectively. Assuming sensitivity and specificity of 99.9 percent, the positive predictive value for each abnormality would be 84, 68, and 40 percent respectively for such a patient — not far from what the Quest group saw in their sample set.
The researchers also combined their results with data shared by other groups at recent scientific meetings and calculated an overall true positive rate for trisomy 21, trisomy 18, trisomy 13, and sex chromosome aneuploidy of 94 percent, 59 percent, 44 percent, and 38 percent respectively.
Important to recognize, Strom said, is that the positive predictive values he and his team found, though much lower than the tests' intrinsic specificities, are still far higher than those of widely used prenatal screening tests that rely on biochemical signatures in pregnant women's serum.
A recent study by University of California, San Francisco and Columbia University researchers sponsored by Ariosa Diagnostics, for example, found that Ariosa's Harmony test outperformed first trimester aneuploidy serum screening according to several different metrics including positive predictive value.
"NIPT is a phenomenal advance over maternal serum screening, which has a positive predictive value in the single digits," Strom said. "It's a much, much better screen. I just don't want people acting on these results as if they are diagnostics, and with this study we really hoped to reinforce this to clinicians."
According to Strom, the issue of variability in positive predictive value may be even more important as NIPT companies move from marketing their tests to high-risk populations, in which prevalence of aneuploidies is relatively high, to lower risk women who have inherently lower likelihood of carrying a fetus with one of these abnormalities.
Tristan Orpin, senior vice president and general manager of Illumina's reproductive and genetic health business told CSN in an email that the company, which markets the Verifi NIPT test, recognizes that "PPV will decrease based on lower prevalence of fetal aneuploidy in the tested population."
"However," he wrote, "… even in a general obstetrical population the PPV of NIPT is at least ten-fold higher than conventional aneuploidy screening methods. While the introduction of NIPT has been a powerful new tool for prenatal care, it is clearly crucial to pair such tests with appropriate patient counseling."
In another email to CSN Sue Gross, the chief medical officer of Natera , which markets the Panorama NIPT test, echoed Orpin's comments.
"We believe that positive predictive value … is important to analyze and discuss to better aid patient counseling," Gross wrote.
A study by Natera researchers of Panorama published this week online ahead of print in the American Journal of Obstetrics and Gynecology found an overall positive predictive value of about 82 percent for the test — much higher than Strom and his colleagues found in their study where they included results from all the NIPT providers. Interestingly, the Natera study also found that PPV was similar between low- and high-risk women.
Though marketers of NIPT tests recommend follow-up testing for all positive results, Strom maintained that in clinical practice it's still likely that a woman might fail to understand what positive NIPT results actually mean in terms of risk.
Even physicians may not understand that a claim of more than 99 percent specificity for such tests does not necessarily translate to a false positive rate below one percent.
Strom said his team hopes that the data they presented offers an independent real-world example of this potential discrepancy, and helps remind the field of the necessity of counseling patients and their doctors that NIPT is fundamentally a screening test, and that its predictive value can be low if the prevalence of a particular abnormality in a particular population is low.
In the study, Strom and his colleagues did not break out false positive rates for the different competing test platforms included in the study, but he said that how these tests compare to each other in terms of positive predictive value will also be an important question moving forward.
An independent registry study recruiting women more widely to compare NIPT and cytogenetic results, which Strom said the Society for Maternal-Fetal Medicine is considering, may help this goal along, as well as provide even better estimates of positive predictive value for NIPT in general.