Ventricular septal defect (VSD) is a frequent congenital heart disease (CHD), making up 26% of all CHDs (Ferencz et al., 1987). Its incidence rate among live births is 3.5/1000 (Hoffman and Kaplan, 2002). VSD can be due to environmental factors, genetic factors, or can be multifactorial. Genetic factors mainly include aneuploidy, chromosomal rearrangements, copy number variants, and sequence variations (Newman, 1985). Non-genetic causes can be identified in 2% of CHD cases, while 20%–30% of CHD cases can be traced back to genetic causes (Cowan and Ware, 2015). Qiao et al. reported that VSD is a type of CHD most often associated with a genetic cause, and 36.8% of VSDs are associated with genetic factors (Qiao et al., 2021). Although most VSDs are repairable and patients can achieve good long-term prognoses under optimized surgical and medical care conditions, the prognosis is unsatisfactory for some patients suffering from VSDs with related genetic abnormalities (van Nisselrooij et al., 2020; Mone et al., 2021). Therefore, the prenatal definition of genetic abnormalities is very important in the diagnosis of VSD, as it can provide more accurate and appropriate genetic counseling, which can affect the decisions of parents on the continuation/termination of pregnancy, prenatal monitoring, and perinatal care.

Fetal structural abnormalities are indicators for invasive prenatal genetic testing (Fu et al., 2022). Fetuses with structural abnormalities have a higher incidence of aneuploidy, chromosomal rearrangements, and sequence variations (Fu et al., 2018). Conventional karyotype analysis is an effective technique to identify chromosomal rearrangements, with a diagnostic rate ranging from 5.4% to 15.5% (Hanna et al., 1996; Beke et al., 2005). However, G-banding karyotype analysis has low resolution and is time-consuming and laborious. The detection rate of small genomic deletions and duplications increased by up to 10% following the development of array-based molecular cytogenetic techniques, such as CMA, which could not be detected via standard structural malformation fetal karyotype analysis (Hillman et al., 2013; Liao et al., 2014). CMA has high resolution and a short turn around time. In patients with post-natal and prenatal CHDs, it can identify aneuploidy, chromosomal rearrangements, and copy number variations (CNVs). Pathogenic CNVs are detected in 7%–36% of CHD patients (Fu et al., 2018; Wang et al., 2018). For the majority of fetuses with structural abnormalities, the underlying causes of abnormalities are unclear before genetic testing. As a significant advancement in next-generation sequencing (NGS), exome sequencing (ES) is an effective tool for assessing post-natal patients. This detection technology is used for prenatal diagnosis (Best et al., 2018). In addition to improving diagnostic rates, using ES for assessing a large sample size can analyze single nucleotide variations (SNVs)/insertions and deletions (indels) in the gene coding regions and help in the identification of novel pathogenic genes or novel variants in well-known genes in VSD patients (Sifrim et al., 2016; Jin et al., 2017; Fu et al., 2018; Lord et al., 2019; van Nisselrooij et al., 2020). Three extensive studies have shown that ES can provide an increased diagnostic rate of 8.5%–11.6% for fetuses with abnormal ultrasound findings, normal karyotype, and negative CMA results (Lord et al., 2019; Petrovski et al., 2019; Fu et al., 2022). A recent study on prenatal CHDs showed that the diagnostic rate of ES was 20% (6/30) (Westphal et al., 2019).

In the present research, we used CMA and ES to assess the detection efficiency of fetuses with VSD at the chromosomal (aneuploidy), sub-chromosomal (microdeletion/microduplication), and single gene (point variants) levels and evaluated perinatal prognosis to facilitate more accurate genetic counseling in clinical practice.

Ventricular septal defect (VSD) is a very common congenital structural abnormality (Ferencz et al., 1987). Previous studies on the genetic etiology of fetuses with CHDs, including CMA and ES, were conducted with only a few individuals, and diagnostic rates ranged from 2.6% to 23.1% (Yates et al., 2017; Fu et al., 2018; Hu et al., 2018; Lord et al., 2019; Petrovski et al., 2019; Westphal et al., 2019; Sun et al., 2020; Mastromoro et al., 2022). In this study, we described the detection rate of different molecular diagnostic techniques for fetal VSD genetic causes and compared the genetic variation rates of isolated VSD and non-isolated VSD fetuses. This was the largest prenatal study on VSD yet. There were 61 cases (13.03%, 61/468) of chromosomal abnormalities and eight cases (13.7%, 7/51) of sequence variations. In a meta-analysis, Mastromoro et al., 2022. proposed that the incremental CMA diagnostic yield of VSD in fetuses with isolated cardiovascular malformations was the lowest, only 2.64%. The detection rate of our study was 13.03%, higher than that of them, which may be related to sample selection bias. After excluding 33 cases of chromosome aneuploidies, the total incidence of CNVs (28/468) and sequence variations (7/51) was 7.5% (35/468), which was lower than that in a recent study (15.7%) (van Nisselrooij et al., 2020). The reason for the lower diagnosis rate than that in the above study might be that only 51 families with negative CMA results in our center opted for ES, which was considerably lower than the number of cases in the study conducted by Van et al. However, the (likely) pathogenic variants were detected in 13.7% (7/51) of fetal VSD pregnancies with negative karyotype and CMA results, which indicated that ES might have a high additional diagnostic rate for molecular diagnosis of fetal VSD. In our study, we identified six novel variants and expanded variants spectra of VSD-related genes. Furthermore, We summarize the literature on the application of CMA/ES in the prenatal diagnosis of fetal congenital heart disease in Table 5 and Table 6.

We found that the detection rate of total (likely) pathogenic gene variants in the non-isolated VSD group (61/335, 18.2%) was significantly higher than that in the isolated VSD group (7/133, 5.3%) (p < 0.05). In addition, the difference in the aneuploidy detection rates between the two groups was significant (9.3% vs. 1.5%, p = 0.002). The difference in incidence of CNVs and sequence variations between the groups was significantly different, which indicated that VSD was more likely to be related to chromosomal abnormalities and sequence variations when occurring with other structural abnormalities. This finding matched the conclusions of other studies but did not match the findings of Qiao et al., 2021. (van Nisselrooij et al., 2020; Mone et al., 2021). The disagreement with the findings of Qiao et al., 2021. might be because our study was limited to prenatal VSD fetuses, while the study conducted by Qiao et al. included various subtypes of CHD fetuses. These inconsistent findings might also be partially explained by the distribution of CHD subtypes, the proportion of isolated VSDs detected in the current cohort, and the differences in the detection methods used. We suggest that clinicians should consider a wider range of differential diagnoses for non-isolated VSD, including the diagnosis of more multi-system syndromes. These reliable phenotypes can facilitate the analysis of ES and the best practices recommend both a phenotype and genotype driven analysis approach (Austin-Tse et al., 2022; Fu et al., 2022). Considering VSD has a high detection rate (isolated or non-isolated), we suggest invasive prenatal diagnosis and genomic analysis for all ultrasound-detected VSDs. However, the classification of non-isolated and isolated VSD depends on the results of prenatal ultrasound examinations. In the prenatal environment, determining a complete and accurate fetal phenotype is not always possible due to different manifestations of fetal disease, incomplete penetrance, and the presence of ultrasound abnormalities only in late pregnancy. Additionally, some phenotypes, such as intellectual disability, developmental delay, minor malformations, or metabolic abnormalities, cannot be detected by prenatal imaging. These restrictions may increase the complexity and difficulty of prenatal ES analysis (Li et al., 2020).

In this study, we identified six novel variants associated with the prenatal phenotype of VSD for the first time and expanded the variant spectrum of VSD-related genes. Sun et al. conducted a study on prenatal cardiac left-sided lesions and found that 19.7% (13/66) of cases had diagnostic variants, and 10.6% (7/66) of cases had KMT2D variants (Sun et al., 2020). However, in this study, only one fetus had a new truncated variant in the KMT2D gene. KMT2D variants can affect the function of H3K4-ASCOM, which can lead to estrogen receptor-mediated pathway disorders and cause a series of KS (OMIM: 147920) phenotypes, including VSD (Miyake et al., 2013). Considering that KS is not rare, the penetrance of patients with KMT2D pathogenic variants is complete. About 70% of KS patients have congenital heart defects and risk of intellectual disability (Digilio et al., 2017). Therefore, we suggest that the variation of the KMT2D gene should be considered in the prenatal diagnosis of Chinese pregnant women with suspected VSD or other congenital facial abnormalities. The clinical management of KMT2D gene variants should include echocardiography at diagnosis for early diagnosis and treatment. Our study expanded the variant spectrum of the KMT2D gene, and our findings might contribute to genetic counseling.

Some genes and genetic pathways are related to heart development, and the abnormalities of these genes are related to the form and function of congenital heart diseases (Afouda et al., 2008; Aoki et al., 2008; McCulley and Black, 2012). In this study, we detected one variant in the TRIO gene (c.5302C, 181 > T p. Arg1768Trp) in case 5. At present, there is no research has been reported to be associated with VSD or other CHDs. Although it is possible this is a new disease gene, however, a more likely explanation is that this is an incidental finding in this case and is unrelated to the VSD presentation. Two novel variants of the PTPN11 gene, (c.1507G>A, p. Gly503Arg) and (c.836A>G, p. Tyr279Cys), were identified. Both were missense variants located in the protein tyrosine phosphatase, catalytic domain (PTPcs) encoded by the PTPN11 gene. The bioinformatics software predicted that these two variant loci might disrupt wild-type protein structure/function. This variant (c.836A>G, p. Tyr279Cys) was included in the ClinVar database as pathogenic or likely pathogenic (Variation ID: 13328, two stars) and included in HGMD with ID CM021133. Some functional experimental studies showed that this variant could disrupt the enzymatic activity of PTPN11 in vitro and increase the activity of AKT and mTOR in cell culture (Hanna et al., 2006; Kontaridis et al., 2006; Martinelli et al., 2008; Marin et al., 2011; Schramm et al., 2013). This variant (c.1507G>A, p. Gly503Arg) was included in the dbSNP database with the ID rs397507545 but not in the 1000 Genomes and gnomAD databases. The variant locus was included in the ClinVar database as pathogenic (Variation ID: 40559, two stars); it was included in HGMD with ID CM060440. PTPN11 is a cytoplasmic tyrosine phosphatase involved in signaling pathways that are induced by growth factors, cytokines, hormones, and extracellular matrix. It regulates the RAS/mitogen-activated protein kinase (MAPK) pathway. Approximately 50% of Noonan syndrome is caused by functionally acquired missense variants in the PTPN11 gene (Adam et al., 1993). These variants are mainly in the N-SH2 or PTP structural domain, which prevents the inhibition of the N-SH2 structural domain. The PTP structural domain is also activated in the absence of phosphorylated ligand binding and upregulates the Ras/MAPK signaling pathway, leading to the development of Noonan syndrome. The pathogenesis of Noonan syndrome, LEOPARD syndrome, Costello syndrome, and cardiofacial syndrome all involve the Ras/MAPK signaling pathway. Recent studies have also found that many genes are involved in the morphogenesis of the cardiac septum and the development of the cardiac cavity. Our findings suggested that patients with prenatal suspicion of CHDs should undergo routine screening for variants in the PTPN11 gene. However, it is undeniable that the overlapping features observed in RASopathies, the wide range of phenotypes within each trait, and the lack of clinical features with pathological diagnostic value, as well as consensus on specific and routinely used diagnostic criteria, make the prenatal diagnosis and genetic counseling of RASopathies challenging.

It is worth noting that in our study, pregnant women (7/9, 77.8%) with genetic variations classified as VUS chose to terminate the pregnancy. According to the follow-up, the main reasons why pregnant women choose to terminate pregnancy include the uncertainty of fetal prognosis, emotional factors, and family economic factors. This issue triggers our thinking as follows. When prenatal diagnosis results are classified as VUS, clinicians and genetic counselors should pay special attention to whether they are from parents or de novo. Clinicians and genetic counselors should find out as much current information as possible to provide the most adequate and detailed genetic counseling for both couples in order to make the best clinical decision and reduce the rate of blind termination of pregnancy. The post-natal follow-up of children with a prenatal diagnosis of fetal VSD is necessary. We collected data on pregnancy outcomes and postpartum treatment of these live birth VSD cases described in this study. Our findings showed that isolated VSDs had a better prenatal and post-natal prognosis than non-isolated VSDs, and the combination of fetal VSDs with other abnormalities was associated with increased rates of pregnancy termination and post-natal surgery. We also found that the detection rate of total (likely) pathogenic genetic variants was significantly higher in the non-isolated VSD group (61/335, 18.2%) than in the isolated group (7/133, 5.3%) (p < 0.05). Overall, a definitive genetic diagnosis and a good prognosis can help in increasing the confidence of patients to continue their pregnancy, especially those carrying an isolated VSD fetus with a negative genetic test result.

The advantage of this study is that we performed genetic diagnosis in a large cohort of unselected VSD fetuses without sample bias and therefore were able to assess reliable incremental yield. This study also had low heterogeneity. However, our study had several limitations. First, this was a retrospective study with possible recall bias, and obtaining fresh blood from patients for cell or protein chemistry experiments was difficult. Second, we might have overlooked balanced chromosomal rearrangements, low-ratio mosaics, and variants in regulatory elements, intergenic regions, and untranslated regions. Third, in our study, for patients with isolated VSD, there was an ES diagnostic rate of 1/12 (0.08%, 95% CI 0.01–0.35), this is vastly insufficient evidence to determine whether ES is warranted. Larger studies are required to determine the diagnostic yield in this group in the future. Fourth, VUS identified by ES, especially in proband-only samples, might have complicated prenatal counseling and parental decision-making. However, these variations will eventually be characterized and reduced as more studies are conducted. In addition, as this study is a retrospective review of the results of CMA and ES testing performed over the course of 10 years, we cannot rule out the impact of any evolution of technology used during the period of study on the testing results.

In our study, CNVs and gene sequence variations were found in many cases with VSD. We suggest that pregnant women carrying fetuses with VSDs combined with additional associated anomalies, but without chromosomal abnormalities and CNVs, can undergo ES. However, the effectiveness of ES in isolated VSD fetuses needs to be confirmed by larger studies in the future.