NIPT discordance causes mapped in retrospective 97,743-pregnancy cohort

In a 10-year retrospective cohort study of 97,743 pregnancies, noninvasive prenatal testing produced sharply different levels of diagnostic confirmation for the three common fetal trisomies. The overall positive predictive value, after adding postnatal karyotyping to prenatal diagnostic results, was 71.84%, rising to 84.38% for trisomy 21, or Down syndrome, but falling to 56.00% for trisomy 18, or Edwards syndrome, and 4.76% for trisomy 13, or Patau syndrome. The researchers also found that the causes of discordance differed by direction: false-positive results emerged from several biological and technical sources, whereas the three false-negative cases identified during follow-up were linked principally to confined placental mosaicism.

The study was conducted at Ganzhou Women and Children’s Health Care Hospital in China between April 2015 and December 2024. NIPT was performed on maternal peripheral blood, with cell-free DNA libraries sequenced and aligned to the human reference genome. An aneuploidy was reported when the absolute chromosomal Z-score reached 3 or more. Among 481 NIPT-positive pregnancies, representing 0.49% of the cohort, 377 underwent invasive prenatal diagnosis, usually amniocentesis followed by G-banded karyotyping or chromosomal microarray analysis. The remaining positive cases were followed by telephone six months after the estimated delivery date, while NIPT-negative cases underwent routine prenatal care and postpartum follow-up. The investigators also examined stored plasma from suspected false negatives and obtained placental tissue from selected discordant pregnancies for genomic testing.

Prenatal diagnosis alone confirmed 270 of the 377 positive cases, producing a positive predictive value of 71.62%. Among the 104 positive pregnancies without invasive diagnosis, 27 produced live births; three infants were subsequently found to have trisomy 21 by peripheral-blood karyotyping, while 24 had no abnormalities reported during follow-up. Incorporating those three postnatal confirmations increased the overall positive predictive value to 71.84%. The corrected value was 243 of 288 for trisomy 21, 28 of 50 for trisomy 18 and two of 42 for trisomy 13, with statistically significant differences among the three conditions. Of the 270 confirmed true positives, 262 pregnancies were terminated, four trisomy 21 pregnancies continued, and four twin pregnancies underwent selective fetal reduction and produced normal infants.

The probability of confirmation varied with maternal age, prior serum-screening risk and the magnitude of the NIPT Z-score. For trisomy 21, the corrected positive predictive value was 91.06% among women aged 35 or older and 79.39% among younger women, a statistically significant difference. It was also higher in the high-risk serum-screening group, at 95.24%, than in the intermediate-risk group, at 70.45%, or the low-risk group, at 54.55%. Trisomy 18 showed a similar pattern: all nine cases in the high-risk serum group were confirmed, compared with three of 11 in the non-high-risk group, although the difference by maternal age was not statistically significant. The clearest numerical separation came from Z-score stratification. For trisomy 21, the positive predictive value increased from 21.95% at 3 < Z ≤ 6 to 81.58% at 6 < Z ≤ 10, 98.31% at 10 < Z ≤ 20 and 100% above Z = 20.

Trisomy 18 also showed an association between Z-score magnitude and confirmation, with no true positives among 15 cases in the lowest range, four of 10 in the next range and all 16 cases between 10 and 20. True-positive trisomy 21 cases had a mean Z-score of 17.60, compared with 5.08 among false positives after four cases with maternal trisomy 21 were excluded; for trisomy 18, the corresponding means were 16.78 and 6.63. These observational associations indicate that higher scores were more likely to represent fetal aneuploidy in this cohort, but the study did not establish Z-score cutoffs as stand-alone diagnostic rules. Prenatal diagnosis was offered to NIPT-negative pregnancies when ultrasound found abnormalities, but karyotyping was not routinely performed after all low-risk deliveries.

Mechanistic investigation was possible for only a small subset of the 107 prenatally diagnosed false positives because maternal karyotypes, placental samples or informative follow-up data were frequently unavailable. Of those 107 cases, 17 had sufficient additional data for investigation. Three further false-positive cases without prenatal diagnosis but with follow-up evidence of a vanishing twin were added, producing a 20-case mechanistic sample. Five cases were associated with maternal chromosomal abnormalities. Four women had a full 47,XX,+21 karyotype, and their NIPT trisomy 21 Z-scores were strikingly high, ranging from 72.01 to 80.86, despite a normal fetal karyotype in each case. A fifth woman had mosaic maternal 47,XX,+21 in four of 32 cells and produced an NIPT Z-score of 8.616; a later pregnancy in that patient was confirmed to have fetal trisomy 21 by amniocentesis. Placental testing identified two false-positive cases in which the placenta carried trisomy 18 or trisomy 13 while the fetus had a normal microarray result, plus one case with 10% placental trisomy 21 mosaicism.

Vanishing twins accounted for seven of the 20 investigated false-positive cases and may be more prevalent than routine clinical records suggest. Five produced a discrepancy between an NIPT-predicted male fetus and a female infant at birth; all had low fetal DNA fractions and relatively low Z-scores, and three had a documented history of an early-pregnancy twin demise. The investigators proposed that residual DNA from a demised male fetus could have contributed Y-chromosome reads, distorting both the fetal-fraction calculation and the resulting risk estimate. Two additional vanishing twins had matching predicted and postnatal sex and were identified only through follow-up. This detection method cannot reveal a vanished twin of the same sex as the surviving fetus, and unreported losses may also have been missed, meaning that the study’s estimate of vanishing-twin frequency is not a complete population-level prevalence.

Three false negatives were identified among 97,262 NIPT-negative pregnancies, an observed proportion of 0.003%, although that figure is likely an underestimate because most low-risk infants were not karyotyped. One twin pregnancy and one singleton pregnancy produced infants later confirmed to have mosaic trisomy 18 and full trisomy 21, respectively, despite low initial Z-scores of 0.09 and 0.83. The third case was a singleton pregnancy initially testing negative at a Z-score of −0.20; ultrasound later revealed multiple malformations, and prenatal diagnosis confirmed trisomy 21. Multi-site placental sampling after termination showed striking regional variation in mosaic trisomy 21, from 10% to 30% at maternofetal-side sites and 18% to 21% on the fetal side, rising to 96% at the umbilical-cord insertion. Repeat tests using stored plasma and a new sample remained below the positive threshold, excluding the tested samples as laboratory errors and supporting low-level or regionally distributed placental mosaicism as the likely mechanism.

Several limitations constrain how the findings should be interpreted. The work was retrospective and confined to one hospital, and routine follow-up rather than systematic diagnostic testing was used to assess NIPT-negative pregnancies, so undetected false negatives were likely. Only two trisomy 13 true positives were confirmed, preventing a meaningful Z-score analysis for that condition. Mechanistic conclusions were based on 20 false-positive cases with additional evidence, and only six had placental verification, creating a substantial risk of selection bias. Placental multi-site sampling was also unavailable for some false-negative cases. The study therefore establishes a detailed framework for investigating discordance, particularly the concentration of maternal chromosomal abnormalities, vanishing twins and placental mosaicism among positive cases, but it cannot establish the frequency of each mechanism across all NIPT results or prove that the observed risk stratifications will perform identically in other populations.

Overall, the cohort demonstrates that an NIPT-positive result is not equivalent to a fetal chromosome diagnosis, and the meaning of a discordant result depends strongly on its direction and biological context. Trisomy 21 signals were much more likely to be confirmed than trisomy 13 signals, while Z-score magnitude, maternal age and serum-screening risk provided useful stratification within the studied population. Extreme trisomy 21 Z-scores above 70 prompted suspicion of maternal chromosomal abnormalities, occult vanishing twins supplied a substantial but incompletely measured share of the investigated false positives, and placental mosaicism emerged as the dominant explanation among the small number of false negatives that could be investigated. These findings offer a concrete basis for genetic counseling and further diagnostic evaluation, while also showing why standardized NIPT cannot replace chromosome-level assessment in every pregnancy.

Subject of Research: Causes and diagnostic predictors of discordance between noninvasive prenatal testing and fetal chromosomal status

Article Title: Causes of discordance between noninvasive prenatal testing and fetal chromosomal status: a 10-year cohort study of 97, 743 cases

Article References: Qi, H., Chen, H., Yuan, X., Gan, J., Liu, H., Xie, X., Xie, Q., Chen, J., Shen, S., & Song, L. (2026). Causes of discordance between noninvasive prenatal testing and fetal chromosomal status: a 10-year cohort study of 97, 743 cases. Frontiers in Endocrinology, 17, Article 1945762. https://doi.org/10.3389/fendo.2026.1945762

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1945762

Keywords: noninvasive prenatal testing, fetal aneuploidy, trisomy 21, trisomy 18, trisomy 13, confined placental mosaicism, vanishing twin, maternal chromosomal abnormalities, Z-score, positive predictive value, Causes, discordance