Noninvasive prenatal screening (NIPS) based on cell-free DNA in maternal circulation is used primarily to screen for trisomies 21, 18, and13, that has been shown to outperform conventional screening methods (Norton et al., 2015). In addition, NIPS may reveal other chromosomal abnormalities, such as sex chromosome aneuploidy (SCA) and other autosomal aberrations, by low-coverage whole-genome sequencing of maternal and placental fragments of DNA (Nicolaides et al., 2013; Lefkowitz et al., 2016; van der Meij et al., 2019). Not all positive results detected by NIPS represent fetal aberrations because of certain biological factors, such as confined placental mosaicism (Grati, 2014; Grati, 2016), ‘vanishing’ twins (Grömminger et al., 2014), and maternal genomic contribution (Ji et al., 2018; Zhang et al., 2020). Hence, further prenatal diagnostic testing is required to check NIPS-positive results.

Several frequently used prenatal diagnostic technologies vary in the testing principle, turnaround time, detection range, and cost. Cytogenetic karyotype testing enables the visual detection of chromosomal abnormalities such as full chromosome aneuploidies, polyploidy, mosaicism, and structural abnormalities of more than 5–10 million base pairs (Mb), including balanced and unbalanced rearrangements. Both quantitative fluorescence polymerase chain reaction (QF-PCR) and fluorescence in situ hybridization (FISH) provide a rapid prenatal diagnosis of chromosome aneuploidy, and FISH is a valuable complementary test for detecting mosaicism (Hultén et al., 2003; Su et al., 2015). Chromosomal microarray analysis (CMA) and copy number variation sequencing (CNVseq) assays, as the two major CNV detection techniques, offer a high-resolution approach to diagnose microdeletions or microduplications (Xu et al., 2014; Wang et al., 2020). CMA can be divided into two types: array-based comparative genomic hybridization (aCGH) and single nucleotide polymorphism (SNP) array.

During recent years, NIPS testing has considerably increased. While most studies have focused on the detection range and accuracy of NIPS(Nicolaides et al., 2013; Lefkowitz et al., 2016; Deng et al., 2019; van der Meij et al., 2019), only a few have investigated prenatal diagnostic techniques following positive NIPS results. In this study, we retrospectively analysed the prenatal diagnostic methods following different types of NIPS-positive results. This study, to the best of our knowledge, the first of its kind in southwest China, aims to provide more evidence for the clinical selection of prenatal diagnosis in patients with different types of NIPS-positive results.

According to the NIPS results, patients were divided into three groups: main target disease [T21 (n = 234), T18 (n = 44), and T13 (n = 31)], fetal sex chromosome abnormalities (n = 413), and other chromosomal abnormalities (n = 204) groups. The average maternal age, average gestational age, and true positive rate of each group are summarized in Table 1. In the main target disease group, 240/309 (77.70%) were confirmed as chromosome abnormalities including 233/240 (97.08%) with common trisomy (two of which were combined with other abnormalities) and 8/240 (3.33%) with CNV. In the SCAs group, 140/413 (33.90%) were confirmed as chromosome abnormalities including 115/140 (82.14%) with sex chromosome aneuploidy (one of which was combined with autosome CNV), 26/140 (18.57%) with only CNV and one with structural abnormality. In the other chromosomal abnormalities group, 37/214 (17.29%) were confirmed as chromosome abnormalities including 2/37 (5.41%) with chromosome aneuploidy, 28/37 (75.68%) with CNV and 7/37 (18.92%) with absence of heterozygosity (AOH). The NIPS exception types, diagnostic methods, and results of amniocentesis are presented in Table 2.

With the popularisation of medical knowledge, prenatal testing has been gaining increasing attention. The ACMG emphasises that detailed pre-test counselling and obtaining informed consent of pregnant women are the basis for expanding the scope of NIPS(Gregg et al., 2016). For an increasing number of pregnant women, NIPS is becoming an option to achieve high-resolution, sensitive, and specific detection of common trisomy and additional findings (van Schendel et al., 2015; van der Meij et al., 2019), as confirmed in this study. The PPV of NIPS was 75.40% for the main target disease, 28.09% for SCA, and 11.27% for other abnormalities, which was slightly lower than that predicted by previous studies (Norton et al., 2015; Deng et al., 2019; van der Meij et al., 2019). This might be relate to the fact that only pregnant women undergoing amniocentesis were included in this study.

Karyotyping is the most common validation method for positive NIPS results in many laboratories, followed by CMA, according to previous reports (Dobson et al., 2016; Cherry et al., 2017; Xu et al., 2020). Several problems have been identified in the actual application process. Culturing amniotic fluid cells and G-banding analysis make karyotyping time-consuming and requiring technical expertise; moreover, the long reporting period leads to anxiety in patients. In addition, karyotyping cannot reliably detect genomic segment rearrangements of <5–10 Mb. CMA can detect changes as small as 5–10 kb in size; however, its low throughput and high cost are obvious drawbacks.

Owing to the high PPV of NIPS for the main target disease, a rapid validation result is significant for pregnant women, and it could reduce parental anxiety and improve pregnancy management (Hultén et al., 2003; Ju et al., 2021). QF-PCR can be used to detect common aneuploidies within 24–48 h. When used alone, it helps to reduce parental anxiety and healthcare cost (Hills et al., 2010; de la Paz-Gallardo et al., 2015). Consistent with other clinical research reports (Voglino et al., 2002; Nicolini et al., 2004), the coincidence rate of QF-PCR and karyotyping in the diagnosis of common trisomies in this study was 100%. The results of QF-PCR did help these pregnant women to plan the next steps of pregnancy management more quickly. There has been controversy over whether QF-PCR can replace karyotyping as a stand-alone method in the general pregnant population, because a residual risk exists for clinically significant chromosomal abnormalities associated with normal QF-PCR results. QF-PCR is used in prenatal diagnostics exclusively for the analysis of numerical changes. Herein, the miss rate of QF-PCR tests at the karyotype detection level was 0.32% (1/309)—slightly higher than 0.069–0.12% reported in several studies with a large number of patients (Comas et al., 2010; Hills et al., 2010; Papoulidis et al., 2012); one of the primary causes for this result was the relatively small number of cases in our study. Theoretically, NIPS has a wider screening range and lower omission rate than traditional screening, which should translate to a lower residual risk of clinically significant chromosomal abnormalities associated with normal QF-PCR results in the main target disease group; however, further research is required to confirm this.

In some cases in the main target disease group, additional findings by karyotyping and CNV detection methods provided important information for effective genetic counseling. Robertsonian translocation accounts for approximately 2.31% (5/216) of all Down syndrome cases diagnosed by karyotyping in our study. QF-PCR is unable to distinguish the complete trisomy 21 from the trisomy as a consequence of translocation of chromosome 21, that is unlikely to affect present pregnancy management; however, it may affect the assessment of risk of recurrence. In case 258, considering the QF-PCR result (D13S628 and D13S634 markers both indicate the triallelic patterns) and karyotyping result (46,XX,add (12)(q24)), it suggested that there may be a balanced translocation carrier in fetal parents. Fetal father was identified as a carrier by further karyotyping, which provided an important basis for assessing the risk of offspring with an unbalanced structural abnormality. In a situation like this, karyotyping was a basic technique which cannot be replaced by other prenatal diagnostic techniques. Using CNV detection technology, we found that 6.59% (5/91) had corresponding chromosome microduplication rather than full trisomy, including case 190 and case 323 with ambiguous QF-PCR results. These findings demonstrated that submicroscopic duplications may result in NIPS showing a high-risk of trisomy, and CNV detection methods can aid in the clarification of abnormalities detected by NIPS. Furthermore, CNV detection technology has a 1.10% (1/91) higher yield of pathogenic CNVs than traditional karyotype analysis in women with positive NIPS resut for T21/T18/T13; this is in line with the findings in the general population of pregnant women (Wang et al., 2018). CNV detection techniques increased additional positive results for fetuses with trisomy 21/18/13 suspected by NIPS; however, the majority of additional findings (75%, 6/8) were classified as VUS, which increased the difficulty in genetic counselling and caused anxiety in pregnant women.

Here, the positive rate of the chromosomal aberration in the SCA group detected by karyotyping was lower than that by the CNV detection methods (p = 0.02), thus confirming that the performance of karyotyping is lower than that of CNV detection methods in confirming the SCAs predicted by NIPS. Sund et al. reported additional risks for structural abnormalities of sex chromosomes in fetuses with a noninvasive prenatal screen positive for SCA(Sund et al., 2020). A recent study has shown that the CNV detection method may provide additional findings for women with positive NIPS results for fetal SCA(Xu et al., 2020). In this study, 2.43% (6/247) fetuses in the SCA group were diagnosed with pCNV or likely pCNV (<5 Mb) by CNV detection methods (Table 5). This further highlights that the CNV detection technique can increase the diagnostic yield for women with positive NIPS result for SCA and should be recommended. In the other chromosomal abnormalities group, the detection rates of CNV detection methods and karyotyping were not significantly different. Nevertheless, caution should be exerted when interpreting the lack of statistical significance, as a relatively small number of people chose karyotyping in our study. Based on the assumption that the resolution of karyotyping is on the order of 5–10 Mb, 8.89% (16/180) anomalies diagnosed as <5 Mb by the CNV detection method would be missed by karyotyping in the other chromosomal abnormalities group, including 1.67% (3/180) classified as pathogenic CNV. Furthermore, there were seven (4.83%) cases of AOH identified by CMA, including one (0.69%) that was ultimately diagnosed with uniparental disomy (UPD). AOH provides clues to explore recessive diseases, and UPD for a clinically relevant chromosome (chromosomes 6, 7, 11, 14, 15, or 20) can result in abnormal phenotypes (Shaffer et al., 2001). The corresponding trisomy associated with UPD, which usually presents only in the placenta, can result in an abnormal NIPS Z-scores (Van Opstal et al., 2018). The ACMG statement indicated that any imprinted regions or chromosomes reportedly involved in positive NIPS cases with discordant results should consider undergoing specific UPD analyses (Cherry et al., 2017).

In the past few years, CMA has been widely recommended as the first-tier test for high-risk pregnancies to identify microscopic/submicroscopic CNVs(Chung et al., 2020; Wang et al., 2020). However, its low throughput, high cost, and complicated operation hinder widespread use. CNVseq based on massively parallel sequencing has several beneficial features, such as low cost, high throughput, low demand of DNA (10 ng), and high resolution (0.1 Mb). CNVseq has been demonstrated to be a suitable first-tier diagnostic test for the identification of clinically significant fetal chromosome anomalies, and it has shown similar effectiveness and advantages in the clinical diagnosis of aneuploidy and pCNVs(Ma et al., 2021). CNVseq could even detect some CNVs that were omitted by CMA because of insufficient probes within the corresponding regions and showed a higher sensitivity in detecting low-level mosaicism than CMA (Walker et al., 2019; Wang et al., 2020). Herein, CMA and CNVseq showed no significant differences in the positive rate in the three types of abnormalities indicated by NIPS, which further indicate that CNVseq represents an economic alternative to increase the abnormality detection rate in pregnant women with positive NIPS results. However, CNVseq cannot fully replace CMA. Except for aneuploidy, chromosomal CNV, and mosaicism, CMA based on the SNP platform can identify AOH and UPD; hence, cases with imprinted chromosome abnormalities predicted by NIPS may benefit more from CMA. However, neither CMA nor CNVseq can identify balanced translocations, and neither aCGH array nor CNVseq can identify triploidy. Neither of these chromosomal abnormalities can be detected by NIPS. Thus, karyotyping remains a valuable method for prenatal diagnosis and may provide unexpected yields for NIPS-positive pregnant women.

Additionally, the sex chromosome mosaicism was relatively frequently diagnosed in this study, accounting for 16.13% (20/124) of cases in the group with confirmed sex chromosome abnormalities. The frequency was similar to the rate of 13.74% (18/131) in a review of large case series of NIPS(Xie et al., 2020). Accurate identification of fetal mosaicism has clinical importance in guiding genetic counselling. FISH detection, using targeted locus-specific probes to observe the chromosomes of each cell without cell culture, demonstrates unique advantages in identifying chromosomal mosaicism. FISH can identify true fetal mosaicism by detecting uncultured amniocytes. In addition, FISH can be used to detect placental mosaicism. Confined placental mosaicism is one of the main reasons for the inconsistency between NIPS results and prenatal diagnosis results (Grati, 2016). A reflex FISH test is recommended to confirm chromosome mosaicism when a mosaic pattern is highly suspected (Hultén et al., 2003; Cherry et al., 2017; Wang et al., 2018; Chai et al., 2019). Compared with the FISH results, we found a higher proportion of the 45,X counted by karyotyping. Besides the different number of cells analysed, better growth of the abnormal cell type is likely to be the major contributor. Culture and harvesting of amniotic fluid cells may lead to changes in the mosaic rate of different cell lines and even cause pseudo-mosaicism (Grati, 2014; Chai et al., 2019). Moreover, overgrowth by the X0 cell line might explain why the FISH results for cases 561 and 568 were inconsistent with those of karyotyping. Contrary to the previous two cases, the karyotyping of case 588 revealed mosaicism, whereas all 100 cells analysed by FISH were X0. Maternal cell contamination is recognized as the most likely interpretation. However, the chance of true fetal mosaicism cannot be completely ruled out. Compared with karyotyping, CNV detection techniques can test uncultured amniotic fluid cells, thus reducing the influence of cell culture on the mosaic ratio and restrictions of the cell cycle on prenatal diagnosis. However, owing to fluctuations in the fluorescence signal intensity and background noise, CMA has a low sensitivity to a lower proportion of mosaic patterns than karyotype analysis and CNVseq (Chai et al., 2019; Hao et al., 2020). The CMA result of amniotic fluid cell testing for case 840 was inconsistent with that of karyotyping of neonatal peripheral blood; the false negative result of CMA may be caused by the above reasons or tissue-specific chimerism. Previous studies have confirmed that CNVseq with increased sensitivity for mosaicism can accurately and reproducibly quantitate mosaicism at levels of 20% (Ruttanajit et al., 2016; Wang et al., 2020). However, neither CMA nor CNVseq can recognize fetuses with balanced mosaics made up of different cell lines (such as 47,XXX/45,X) (Markus-Bustani et al., 2012). Therefore, the FISH assay should additionally be performed to derive a more accurate karyotype and more accurately define the levels of sex chromosome mosaicism.

For the majority of pregnant woman with a high-risk of common trisomy predicted by NIPS, QF-PCR stand-alone testing represents an economical and rapid method for prenatal diagnosis. Combined with karyotyping and CNV detection methods, some additional findings can be obtained, which can provide more comprehensive information for genetic counseling. When NIPS shows a positive result for SCA or other abnormalities, CNVseq can be employed as a valid and cost-effective alternative for CMA, which can improve the detection rate of pathogenic or potentially pathogenic chromosomal abnormalities. Considering the peculiarities of sex chromosomal mosaicism, the addition of the FISH assay can help identify true fetal mosaicism.