NEW YORK (GenomeWeb) – Researchers from BGI Diagnostics have published a study tracking the clinical performance of the company's sequencing-based non-invasive prenatal testing in nearly 150,000 pregnancies.
The data, published this week in the journal Ultrasound Obstetrics and Gynecology, demonstrate that the performance metrics of earlier smaller studies of NIPT — such as sensitivity and specificity, false-positives, or false-negatives — largely hold true in the context of a dramatically expanded group of clinical samples.
The study also showed that the test's performance did not differ substantially among higher- and lower-risk women.
Wei Wang, the study's first author, told GenomeWeb in an email this week that the BGI team had several goals in reporting its NIPT performance in what is the largest sample set yet published.
Primarily, Wang wrote, the researchers hoped to confirm measures of the sensitivity and specificity of NIPT that have been demonstrated in earlier, smaller cohorts, and to show that testing performance could be maintained through several years of offering NIPT as a clinical service over many thousands of samples.
"Previous validations have been done in small and well-controlled cohorts … However, no one knows the real clinical performance when it is offered to a huge number of people, and there are concerns about performance deterioration in large clinical practice," Wang wrote.
"In this paper, we showed the first-hand information of NIPT performance in a very large population, and consisten[cy with the] accuracy [shown] in small studies. The results also suggested that high quality of NIPT service can be achieved at NGS-based clinical labs with strict protocols and standards."
Overall, the team reported on 146,958 samples from 508 medical centers in mainlandChinatested successfully between early 2012 and mid-2013 for Trisomy 21, 18, and 13.
Though the company's test measures other chromosomal abnormalities, these results were not included in the study.
Testing during the study period relied on sequencing using the Illumina HiSeq 2000. According to the team's results, BGI's NIPT identified 1,578 trisomy-positive and 145,380 negative samples.
Comparing the NIPT results with confirmation either by follow-up invasive testing, or by tracking patients' eventual pregnancy outcomes, the group then calculated the false positive and false negative rate for each tested trisomy, and the overall sensitivity, specificity, and positive predictive value of NIPT overall.
While there was not full follow-up information for all the samples studied, the team was able to confirm outcomes for about 75 percent of the cohort — 1,066 NIPT-positive cases, and 111,603 NIPT-negative cases. Within this subset, the researchers found that 720 of 781 T21 calls, 167 of the 218 T18 cases, and 22 of the 67 T13 cases were true-positives.
Nine false negatives were also identified in the patient follow-up data, including six T21 and three T18 cases.
Overall, the group calculated the sensitivity of BGI's NIPT as 99.1 percent, 98.2 percent, and 100 percent in the three trisomies, respectively. Specificity was about 99.95 percent in each case.
For T21, the false-positive rate was 0.05 percent, and the positive predictive value was 92.19 percent. For T18, the FPR was also 0.05 percent, and the PPV was 76.6 percent. FPR and PPV for T13 were 0.04 percent and 32.84 percent, respectively.
Because not all the cases tested had follow-up data available to confirm or deny a positive call, the BGI researcher also calculated the range of possible PPV for the cohort as a whole, by assuming two opposite conditions — one in which all the unconfirmed positives and negatives were false-positives or false negatives, and one in which all the unconfirmed results were true positives or true negatives.
This provided a range of PPV for T21 of 65-94 percent, for T18 of 47-85 percent, and for T13 of 12-62 percent.
According to the study authors, most confirmed false positives and false negatives in the study could be traced to noticeable biological factors, such as maternal copy number variation or mosaicism.
Wang said that the group was somewhat surprised to find that low fetal fraction was not a major contributor to NIPT misdiagnoses in their cohort, considering that the importance of fetal fraction has been "overwhelmingly emphasized in the past few years, and low fetal fraction is believed to be a major risk of causing NIPT false results."
In the NIPT false positives and negatives the BGI group studied, all samples had a fetal fraction above the recommended requirements.
Overall, the performance of NIPT for the BGI group reflects closely what other providers have reported in studies of the clinical performance of various sequencing-based strategies.
For example, Natera published a study in the American Journal of Obstetrics and Gynecology that calculated PPV at about 83 percent across all aneuploidies and 90 percent for trisomy 21, in a subset of patients for whom follow-up data was available among a total cohort of 30,000.
Similarly, Ariosa's non-invasive Chromosomal Examination of Trisomy (NEXT) study comparing its Harmony test for trisomy 21 detection with standard combined first semester screening in almost 16,000 women from the general population found the positive predictive value of the Harmony test to be 81 percent and the false positive rate 0.06 percent.
According to Wang, beyond confirming the high sensitivity and specificity of NIPT seen in such earlier studies in a much larger sample set, the BGI team's data also provide a simple demonstration of how far NIPT has taken off in the brief years since it was first introduced and adopted into clinical practice inChina, as well as abroad.
"Since its advent in 2011, in a very short period of time the technology was accepted by most clinicians. … As seen in our study, we managed to collect over 140,000 clinical samples for prospective analysis of NIPT performance. This is unimaginable for previous tests," Wang wrote.
According to the authors, the study results also highlight some unresolved questions about the clinical implementation of NIPT. "On the one hand, the benefits of this test (such as high detection rate, low false positive rate, testing at early pregnancy) are quickly changing people's preference of choosing prenatal screening method, and even the screening strategy in China," Wang wrote.
On the other hand, some problems still remain, such as a lack of standards for genetic counseling and questions about the best way to use this testing in prenatal screening, i.e. either as a primary or a secondary contingent screening method, Wang said.
While there is no doubt about the accuracy of NIPT in testing T21, T18, and T13 in high-risk pregnant women, the method has not been validated as thoroughly as a screening approach for a general population including low-risk pregnancies.
In their study, Wang and the other BGI authors also compared the performance of NIPT among high- and low-risk women in the overall cohort, dividing the group into two subsets based on risk factors, such as age and reproductive history.
Interestingly, there wasn't a significant difference in the performance of NIPT between these two groups, the authors wrote.
"Our data provided so far the largest and most comprehensive evidence of NIPT performance in low-risk pregnant women, which strongly supports the use of NIPT in the general population for screening for T21, T18, and T13," Wang added. "Although the data of low-risk pregnant women came from a selected group [within] the total 140,000[-plus] cases, we believe that it is representative and reflects the truth. Our data, together with several previous studies in the low-risk population, showed a bright future of using NIPT in the general population. We look forward to seeing results from more works of NIPT in the general population."
In March 2014 BGI ceased its NIPT activities after an announcement from the China Food and Drug Administration that genetic testing should be regulated. In July, it then received CFDA approval for its NIPT, NIFTY, to be run on two sequencing systems, BGISEQ-100 and BGISEQ-1000, which are based on Thermo Fisher Scientific's Ion Torrent semiconductor sequencing technology and Complete Genomics' sequencing technology, respectively.
"In the past few months, our test numbers [have] increased rapidly," Wang added. "We have finished a total of 462,000 tests since the launch of our NIPT service, and about 315,000 tests were competed after Aug. 2013."
BGI announced last year that it had entered into an agreement with Poland-based Genomed making its NIFTY testing available in that country. It also signed a similar agreement in October allowing Italian firm Bioscience Genomics to start offering NIPT services based on NIFTY, and another with Spain's NIM Genetics.
The company declined to comment on plans for further dissemination of its test outside of China.