From 2012 onwards, NIPT for fetal aneuploidies was broadly implemented for detecting common autosomal trisomies and SCAs owing to the advantages associated with it such as non-invasiveness, zero risk for the unborn baby, capability to acquire diagnostic hints as early as the 10th week of gestation onwards, immediate results within as early as two weeks, as well as high sensitivity (99.3% for T21, 97.4% for T18, and 97.4% for T13) and specificity (pooled specificity was 99.9% for all three trisomies) [17, 18]. However, this approach identifies only 75 to 85% of clinically relevant aneuploidies [19]. Therefore, additional screening based on identifying RATs and MMS is necessary. Here, we assessed a series of 626 NIPT-positive cases with low genomic coverage and detected a broad range of aneuploidy classes, namely the common autosomal trisomies, SCAs, RATs, and MMS. The PPV of T21 (81.23%) observed using our platform in the present study was within the range of values reported in published literature (between 80% and 90%) [20]. The PPVs of T18 and T13, presented as the main positive results, were 37.93% and 18.42%, respectively, which were slightly lower than those reported by previous studies on the same platform [21]. The PPVs of SCAs, RATs, and MMS, presented as the additional positive results, were 48.83%, 18.37%, and 41.68%, respectively, which were slightly higher than those reported by previous studies on the same platform [22]. PPV obtained via NIPT, excluding that of T21, is known to have large variation associated with prior risk factors, such as maternal age, and the individual trisomies [23, 24]. NIPT results are affected by an insufficient or absent fetal fraction, fetoplacental mosaicism, the presence of a vanishing twin, maternal mosaicism, maternal CNVs, and maternal malignancy leading to false positives that are discordant with results obtained by other methods [25, 26]. Moreover, technical factors such as testing procedures, sequencing algorithms and depths, as well as Z-scores may also be important in terms of their effect on NIPT results [20]. This makes the fluctuation of the PPV of NIPT in different study populations a common occurrence. In our research, we found RATs to have a PPV of 18.37%, which was similar to that of T13 presented as the main positive results, and could therefore act as an extension of NIPT screening. For MMS, there was a higher PPV than that of T18 shown as positive results, but most of the CNVs were identified as hereditary and of unknown significance. Disclosure of these results to pregnant women did not provide them any substantial help with pregnancy-related decisions and had a negative psychological impact on them. Therefore, for cases of MMS suggested by NIPT results, it is recommended to only present the diagnoses to pregnant women if the CNVs are in genomic regions that have definite associations with certain syndromes or present them after pathogenicity has been identified.
Discordant results associated with NIPT often occur during screening and diagnosis. Reports of discordant results focused on the causes of false positives and false negatives detected by NIPT highlight the impact of confined placental mosaicism and true fetal mosaicism on NIPT-based analyses [25]. In this study, our analysis of the discordance between the positive results of NIPT and IPD showed 53 discordant cases (which accounted for 9.35% of the total cases). As expected, NIPT was ineffective in identifying balanced structural rearrangements due to the limitations of sequencing depth and fragment read length of NGS. The assessment of the cases in our analysis confirmed the importance of testing by IPD in addition to NIPT, which could not accurately determine the abnormality as being on one or two chromosomes, being a trisomy or monosomy, euploidy or mosaic, or a trisomy or partial abnormality. Among cases of the “one-to-one” type of discordance, we found that NIPT suggested trisomy/monosomy for 21 cases where IPD results indicated mosaicism. This accounted for the largest proportion of discordance that was observed between NIPT and IPD results. NIPT uses ccfDNA fragments that originate from the cytotrophoblast cells of the chorionic villi in the placenta to detect fetal trisomies; however, the karyotype of the cytotrophoblast cells does not always represent the fetal chromosome constitution [27]. Our observations show that in some cases that were diagnosed with very low rates of mosaicism confirmed by multiple detection methods, pregnant women choosing to continue the pregnancy had fetuses that developed well after birth (Supplementary Table S1, Case 439). Therefore, it is advised that pregnant women getting a positive NIPT result should not hastily be driven to a negative attitude, and should actively undergo a follow-up consultation to determine the proportion of fetal mosaicism by means of IPD, and only then make decisions regarding the continuation or termination of pregnancy. Some unbalanced structural rearrangements involving other chromosomes and trisomies that were not identified by NIPT and sSMC detection were discovered by accident in our research. It must be taken into account that NIPT, which is based on second-generation sequencing technologies, is not sensitive to some DNA fragments with a high average content of guanine and cytosine bases. Therefore, NIPT cannot be regarded as a diagnostic tool for conclusive diagnoses, and positive NIPT results must further be assessed by invasive prenatal diagnostic approaches.
G-banded karyotyping, which has limited resolution (5 ~ 10 Mb), is a common diagnostic technique and the gold standard for the diagnosis of chromosomal disorders. It can detect chromosomal aneuploidy or polyploidy, large deletions/duplications in chromosomes, and balanced chromosomal rearrangement. Other commonly used prenatal diagnostic techniques, namely CMA and CNV-seq, can be used to analyze aneuploidy as well as microdeletion and microduplication (≥ 100 kB) [15, 28]. In our study, a total of 43 discordant cases were found in the chromosomal analysis of 308 patients performed by means of both karyotyping and CMA/CNVseq. Four instances of sex chromosome mosaicism were detected by karyotyping, which were not indicated by CMA. For cases of sex chromosome abnormality indicated by NIPT, karyotyping was seen to be more effective than CMA in confirming true positive detection of sex chromosome mosaicism. Additionally, two cases of autosomal mosaicism were detected by karyotyping, and not suggested by CMA; while one case of autosomal mosaicism were detected by CMA, and not suggested by karyotyping.
Karyotyping and CMA each have certain advantages and disadvantages for their use in the detection of autosomal mosaicism. Although karyotyping requires cell culture, it can detect mosaics of different types, including those of a very low proportion. However, multiple factors, such as aberration of the primary amniotic cells themselves and cell aberration resulting from in vitro culturing, may lead to pseudomosaicism, loss or increase of the abnormal cell line resulting in a change in the proportion of mosaic cells, or even to missing detection of autosomal mosaicism [29]. Conversely, CMA can only stably detect mosaicism in cells with larger proportions (> 30%) of it and can detect the genome of the amniotic fluid directly, thus being capable of reflecting the proportion of true mosaicism in the sample. Additionally, CMA has unique advantages in detecting CNVs and ROH, which cannot be detected by karyotyping. Our results show that in comparison with CNVs detected by karyotyping, 10 more cases of pathogenic CNVs were detected by CMA, which thus had an improved diagnostic rate of 5.18% compared to that of karyotyping. In addition, for NIPT positive samples showing normal karyotypes, a total of 3.11% ROH was detected by SNP-based microarrays. The presence of large fragments of ROH in the fetus is associated with the risk of uniparental disomy (UPD), which is the result of the successful rescue of cells from aneuploidy to euploidy after fertilization of germ cells. A UPD diagnosis should be considered when NIPT suggests trisomy, especially on chromosomes 6, 7, 11, 14, 15, and 20 [30]. Thus, it can be seen that a single detection method can easily lead to misdiagnosis. Therefore, the combination of karyotyping with CMA seems to be preferable for obtaining accurate diagnoses of chromosomal abnormalities.
At the later stages of follow-up, most women with fetuses diagnosed with autosomal trisomies had terminated pregnancy, excluding one case of T21 with a low rate of mosaicism. SCAs are the most frequent chromosomal abnormalities encountered in NIPT. In true positive cases, the overall termination of pregnancy rate was 22.7% (5/22) for Triple X syndrome and 36.36% (8/22) for 47, XYY syndrome, which was significantly lower than those of other chromosomal syndromes. The prevalence of Triple X syndrome and 47, XYY among newborns is high, Triple X: 11 per 100,000 females, and 47, XYY: 18 per 100,000 males respectively [31]. Although an increased risk of psychosocial problems or psychiatric disorders (such as autism) during childhood has been associated with the 47, XYY syndrome, long-term, unbiased follow-up studies have concluded that Triple X syndrome and 47, XYY syndrome, do not cause postnatal development disorders, children with these conditions have IQs in the normal range despite physical abnormalities being occasionally observed [32]. The acceptance of fetuses with SCAs tends to be affected by many factors, such as social and cultural background, disease type, genetic counseling methods, and the economic status of the family. In China, an increasing number of people are accepting children with Triple X syndrome and 47, XYY syndrome. Therefore the exclusion of Triple X syndrome and 47, XYY syndrome from the NIPT process is expected in the near future. Moreover, the true or false positive nature of ultrasound findings is also an important factor in determining the decision to continue a pregnancy.