Prenatal detection of chromosomal abnormalities and copy number variants in fetuses with congenital gastrointestinal obstruction

Congenital gastrointestinal obstruction (CGIO) mainly refers to the stenosis or atresia of any part from the esophagus to the anus and is one of the most common surgical causes in the neonatal period, with an incidence of 1 in every 2000 newborns [1]. Esophageal stenosis/atresia represents a life-threatening condition in the upper gastrointestinal tract, and duodenal stenosis/atresia and jejunoileal stenosis/atresia represent the major causes of CGIO, with an incidence ranging from 1.3 to 2.8 out of 10 000 live births in the lower gastrointestinal tract [2, 3]. Colonic stenosis/atresia, the less frequent form of lower gastrointestinal tract atresia, and anorectal malformation are also included in lower gastrointestinal tract atresia. CGIO is associated with complex embryological and genetic factors. The 4–8th week of gestation in the embryonic period is known as the developmental origin of CGIO. For esophageal stenosis/atresia, a failure of invagination of the lateral trachea-esophageal grooves is proposed. Duodenal stenosis/atresia is believed to result from failure of bowel recanalization following a temporary solid stage, while the probable cause of jejunoileal stenosis/atresia and colonic stenosis/atresia may be a late mesenteric vascular accident. An impaired process in the urorectal septum might lead to anorectal malformation [4‐6].

In addition to embryological factors, the concept of genetic factors as an etiology of CGIO has been generally accepted [7, 8]. The Prevention Network in the United States shows that gastrointestinal malformations are often correlated with trisomy 13, 18, 21 and Turner syndrome, in which the associations between duodenal stenosis/atresia and trisomy 21, esophageal stenosis/atresia and trisomy 18 are especially prominent [9]. With the wide application of high-resolution chromosome analysis technology in prenatal diagnosis, increasing evidence has shown that pathogenic copy number variants (pCNVs) account for a certain proportion of fetuses with ultrasound abnormalities. In recent years, chromosomal microarray analysis (CMA) has become a mature clinical high-resolution chromosome analysis technique for detecting submicroscopic chromosomal imbalances. However, the high cost and low throughput of CMA restrict its application as a routine detection method for prenatal diagnosis. With the development of next-generation sequencing (NGS) technology, NGS-based copy number variation sequencing (CNV-seq) technology has gradually developed into a high-throughput, high-resolution, short turn-around time, and low-cost detection method [10], and it has been utilized in most prenatal diagnoses as a viable alternative methodology to CMA [11, 12]. A few studies have shown that DO is also related to CNVs, such as 4q22.3 deletion [13] and 13q deletion [14, 15]. Nevertheless, genetic investigations of CGIO have mainly focused on aneuploidy, and the focus has been on DO.

Therefore, the objective of this study was to evaluate the risk of chromosome aberrations (including numeric and structural aberrations) by karyotyping in different types of CGIO. A second objective was to assess the risk of abnormal CNVs detected by CNV-seq in fetuses with different types of CGIO to provide better prenatal counseling and clinical management.

Congenital gastrointestinal obstruction is one of the most frequent anomalies second to central nervous system anomalies in fetuses [17]. The causes of gastrointestinal atresia are complex, and chromosomal abnormalities constitute an important factor in its pathogenesis. Studies published to date have mainly focused on the relationship between aneuploid and CGIO. Up to 44% of fetuses with duodenal stenosis/atresia are associated with trisomy 21 [18‐20] and may be higher if other chromosomal abnormalities are included. In our research, even if chromosome structure abnormalities were included, the rate of abnormal karyotypes (including trisomy 21) in cases with duodenal atresia was only 25%, which was lower than that in previous studies. Becky et al. reported that only 15% (4/27) of cases with DA were diagnosed with trisomy 21, at a lower rate; the same was true in the study by Zhang et al., at a detection rate of 5.9% (3/51). Combined with our research, we believe that the association between duodenal atresia and trisomy 21 should be reconsidered because of the wide use of noninvasive prenatal testing (NIPT) and improvements in ultrasonographic techniques in the first trimester. In addition, we found that the prevalence of chromosomal anomalies in cases of isolated CGIO (10.8%) was lower than that with complicated CGIO (33.8%), consistent with that reported in previous studies [21], indicating that it is necessary to strengthen the observation of other structures or soft markers when CGIO is encountered.

Although both CMA and CNV-seq provide high-resolution analysis for accurate and reliable diagnosis of clinically significant CNVs, CNV-seq does have some potential advantages over CMA, including high throughput sample analysis, lower gDNA input detection, and shorter turnaround time. In the past few years, several large studies published the incremental yield of CMA over karyotype in fetuses with ultrasonographic structure anomalies [22‐24], but few studies were conducted on the relationship between CNV-seq and karyotype. In 2018, a large prospective study that performed CNV-seq on 3429 amniotic fluid samples in fetuses with a low risk of CNV abnormalities showed that the incremental yield of CNV-seq over karyotype was approximately 1% and first proposed that CNV-seq could be considered the first-tier diagnostic technique for detecting pCNVs [11]. Moreover, it was suggested that the frequency of pCNVs in fetuses with ultrasonographic abnormalities should be further refined by the organ system involved and the number of anomalies observed [25]. Xia et al. reported that the detection rates of pCNV and VOUS were 4.55% and 9.09% in digestive disorders, respectively [26]. In a recent study performed by Zhang et al. [15], pCNVs were identified in 5 of 48 fetuses with DO at a detection rate of 10.4%. In our study, we observed different types of CGIO caused by stenosis or atresia. CNV-seq revealed an incremental diagnostic value of 9.7% over conventional karyotyping in fetuses with CGIO, and the genomic alterations differed between isolated CGIO and complicated CGIO. The incremental diagnostic yield of CNV-seq in complicated CGIOs (20%) was higher than that in isolated CGIOs (4.8%), and the highest prevalence of pCNVs/likely pCNVs was found in the duodenal stenosis/atresia group (17.5%), followed by the anorectal malformation group (15.4%). These data were inconsistent with the reports by Bishop et al. [13] in which a pathogenic microdeletion was found only in isolated duodenal atresia and by Zhang et al. [15] in which no significant difference in pCNVs was observed between the isolated group and complicated group. However, we believe that the divergence might be due to the relatively low number of fetuses included in those studies.

The 13q deletion, resulting in Waardenburg syndrome type 4A (OMIM 277580), was the most frequently identified CNV in our study (cases 1 and 2). This finding was consistent with a finding in a previous study [15]. CNV-seq revealed a 9.76-MB deletion in 13q21.33q34 and an 8.99-MB deletion in 13q22.31q33.3. According to the AMC guidelines, they were both classified as pathogenic CNVs. The ClinGen database shows that the fragment contains the EDNRB gene, which is a signaling molecule and key component of the endothelin pathway and plays an important role in the migration of enteric nervous system (ENS) precursors within the gut [27]. Several studies reported that mutations of EDNRB were identified in Hirschsprung disease (HSCR, absence of enteric neurons in distal portions of the gut) and Waardenburg syndrome type 4A (WS4A, pigmentation defects and deafness due to altered development of melanocytes) [27‐30]. Notably, abnormal migration of neural crest cells in conjunction with destruction of blood vessels may be the possible pathogenesis of duodenal atresia. Meanwhile, the relationship between 13q deletion and duodenal atresia has been described in several case reports [15, 31, 32]. In summary, we consider that haploinsufficiency of EDNRB might be a candidate gene to produce the phenotype of duodenal stenosis/atresia in a fetus with 13q deletion syndrome.

Another variant that occurred with high frequency was 22q11.2 microduplication, which was present in one case of anorectal malformation (case 3) and one case of duodenal atresia (case 4). The 22q11.2 duplication is associated with mild but highly variable phenotypes, ranging from normal to developmental delay, growth restriction, hypotonia, and intellectual disability [33]. Due to incomplete penetrance and variable phenotypes, this CNV can cause parental anxiety about the future health and development of their children. Although the effects of these potentially pathogenic abnormalities do not appear after birth or for a long time after birth, the limited information on 22q11.2 duplication is not trivial. Providing parents with this information may not only help parents better understand abnormalities that may occur in their children but also help families take better care of them and seek therapeutic intervention earlier.

Case 5 was a fetus with a 9.84-Mb 10q26.13q26.3 deletion (de novo) and a 0.56-Mb 4q24 deletion (maternal) with duodenal atresia and persistent left superior vena cava on prenatal ultrasound. The deletion of 10q26 includes the EBF3 gene, of which low expression has been linked to hypotonia, ataxia, and delayed development syndrome (OMIM 617330), resulting in various congenital abnormalities, including microcephaly, growth retardation, intellectual disability, craniofacial dysmorphism, micropenis, cryptorchidism, etc. [34‐37]. Maruyama et al. reported a patient with duodenal atresia combined with partial monosomy 10q with partial trisomy 11q, showing that the synergistic effects of partial monosomy 10q and partial trisomy 11q on the phenotype might be related to the development of duodenal atresia [38]. The deletion of 4q24 encompasses the TBCK and AIMP1 genes, and loss of them may be responsible for psychomotor retardation. Bishop et al. [13] reported a case of isolated duodenal atresia who had a likely pathogenic microdeletion of chromosome 4q22.3. Therefore, the literature, combined with our current results, suggests that 10q26 and 4q24 microdeletions might be additional genetic etiologies of duodenal stenosis/atresia.

Case 7, a fetus with duodenal atresia, had a 4.82-Mb deletion in chromosome 2p24.3p24.2. This segment contains the genes MYCN and NBAS. MYCN is significantly associated with Feingold syndrome-1 (FGLDS1), which is characterized by esophageal and duodenal atresia, microcephaly, limb malformation, and mental retardation. Vertebral anomalies, cardiac malformations, and deafness have also been reported in a minority of patients [39]. NBAS is associated with infantile liver failure syndrome, short stature and optic nerve atrophy [40, 41]. Although diagnosed with a normal karyotype, TOP was chosen for this patient due to pathogenic CNV. For the deletion of 20p12.2 and the duplication of 4q in our reports, there are no previous studies about their association with congenital gastrointestinal deformities and were considered to be incidental findings by us. In addition, two VOUS were detected in one case with duodenal atresia and one case with anorectal malformation accompanied by other structural abnormalities. The fetuses were both delivered at term, showing no obvious phenotypic abnormality up to the time the article was written.

The mortality rate of CGIO is usually low. Most neonates with these conditions can be relieved by surgery, and the overall outcome is good; however, the premise is that there are no genetic abnormalities in prenatal diagnosis. Submicroscopic chromosome aberrations should be considered in addition to aneuploidies. In our study, no pathogenic CNVs were found in esophageal stenosis/atresia, jejunoileum stenosis/atresia or colonic intestine/stenosis. Rare inherited and de novo CNVs were identified in fetuses with esophageal atresia. A multicenter study detected 375 patients with esophageal atresia and found that 2.7% had pathogenic CNVs, expanding the genetic scope of esophageal atresia [42]. Therefore, the correlation between esophageal atresia and CNVs still requires a large number of case studies and targeted genotype–phenotype analysis. Approximately 55% of infants born with esophageal atresia have other anomalies or birth defects, 10% of infants have a nonrandom VACTERL syndrome, and 1% of infants also have CHARGE syndrome, so fetuses with esophageal atresia require detailed imaging and genetic testing to assess the risk [43‐46]. Research has suggested that the occurrence of jejunoileum stenosis/atresia is related to race and maternal age, and the correlation between them and chromosomal abnormalities is lower than that of duodenal atresia and esophageal atresia [47]. In 2015, a number of European research centers retrospectively analyzed 423 cases of jejunoileum stenosis/atresia and found that the prevalence of chromosomal abnormalities was 3.8% and that of aneuploidy was only 0.3% [21]. These findings demonstrated that the incidence of chromosomal abnormalities of jejunoileum stenosis/atresia was relatively low. The mechanism of colonic atresia is generally accepted to be vascular injury. This may be the reason for its low incidence of chromosomal abnormalities. Meanwhile, the weak connection between chromosomal abnormalities and jejunoileum stenosis/atresia and colonic intestine/stenosis observed in our study provides information for clinical prenatal counseling and decision-making.

There are several limitations in our study. First, as a retrospective study, there were limited numbers of some categories of CGIO, and less than half of the fetuses underwent CNV-seq. Therefore, the pCNV and likely pCNV rates of different subgroups may change if the sample size was expanded. Second, not all fetal CGIOs were confirmed, and some misdiagnosed cases were included in our study. We should accept that these factors might influence the rates of chromosomal abnormalities to a certain extent. Finally, not all women with CGIO underwent amniocentesis or cordocentesis, which might be relevant from the perspective of selection bias. Some results found in our study also differed from previous reports. The abnormal genetic results found in our study may not represent the majority of CGIO abnormalities but can also expand the genetic scope of CGIO to some extent. In the future, more cases need to be collected, likely in cooperation with multiple centers, to conduct further statistical analysis and to generate a more complete summary of the findings.

In summary, our study showed that the risk of pathogenic chromosomal abnormalities and CNVs was increased in the complicated CGIO group compared to that in the isolated CGIO group, especially with DO and anorectal malformation. CNC-seq was recommended to detect submicroscopic chromosomal aberrations for DO and anorectal malformation with a normal karyotype, as the information derived can provide additional clinically relevant information. The relationship between genotypes and phenotypes needs to be explored in the future to facilitate prenatal diagnosis of fetal CGIO and provide new clues into their etiologies.