Diagnostic yield using whole‐genome sequencing and in‐silico panel of 281 genes associated with non‐immune hydrops fetalis in clinical setting

ABSTRACT

Objective

To investigate the diagnostic yield of clinical whole‐genome sequencing (WGS) in prenatally diagnosed non‐immune hydrops fetalis (NIHF).

Methods

This was a retrospective study of 23 fetuses with prenatally diagnosed NIHF, negative for trisomies and copy‐number variants, referred for analysis by WGS with an in‐silico panel of 281 genes associated with hydrops fetalis. Due to identification of a high proportion of causative variants in the HRAS gene in the main cohort, Sanger sequencing of HRAS was performed in a replication cohort, consisting of 24 additional fetuses with NIHF that were negative for trisomies and copy‐number variants and had not undergone WGS.

Results

Of the 23 fetuses in the main cohort, a molecular diagnosis was achieved in 12 (52.2%). Pathogenic or likely pathogenic variants were identified in seven genes: HRAS (n = 5), RIT1 (n = 2), FOXP3 (n = 1), GLB1 (n = 1), MAP2K1 (n = 1), PTPN11 (n = 1) and RASA1 (n = 1). The inheritance pattern of the 12 causative variants was autosomal dominant in 10 cases (HRAS, MAP2K1, PTPN11, RASA1, RIT1), autosomal recessive in one (GLB1) and X‐linked recessive in one (FOXP3). Of the 24 fetuses in the replication cohort, a pathogenic variant in HRAS was identified in one, resulting in an overall frequency of causative HRAS variants of 12.8% (6/47) in our two cohorts.

Conclusions

We demonstrate a diagnostic yield of 52% with clinical WGS in NIHF using an in‐silico panel of 281 genes. However, the high diagnostic yield may be attributed to the small sample size and possible over‐representation of severe phenotypes in the included fetuses. Bearing in mind that chromosomal abnormalities were excluded in our cohorts, a detection rate of up to 75% is possible in prenatally diagnosed NIHF when WGS analysis includes calling of chromosomal aberrations. © 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Short abstract

Linked article: There is a comment on this article by Lin and Li. Click here to view the Correspondence.

This article's abstract has been translated into Spanish and Chinese. Follow the links from the abstract to view the translations.

CONTRIBUTION —

What are the novel findings of this work?

This is the first study to investigate the diagnostic yield of clinical whole‐genome sequencing in prenatally diagnosed non‐immune hydrops fetalis. An incremental diagnostic yield of 52% was demonstrated in cases in which chromosomal aberrations had been excluded.

What are the clinical implications of this work?

Due to the wide range of underlying genetic etiologies of non‐immune hydrops fetalis, a comprehensive approach is essential in the evaluation of affected cases. Whole‐genome sequencing offers a high diagnostic yield by enabling simultaneous detection of different genetic variants across the entire genome.

INTRODUCTION

Hydrops fetalis is defined as excessive fluid accumulation in two or more fetal compartments ¹ , ² . If no maternal red cell alloimmunization is detected, the condition is classified as non‐immune hydrops fetalis (NIHF) ² . NIHF is associated with notable risks, and despite improvements in diagnosis and management, its mortality rate remains high ³ , ⁴ . Identification of the underlying genetic causes of NIHF depends on multiple factors, including the method used and severity of the phenotype. Up to 50% of prenatally diagnosed cases of NIHF are caused by chromosome aberrations, which means that genetic etiology remains undefined in around half of NIHF cases if examined only by chromosome analysis, including array‐based molecular cytogenetic analysis ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ .

Massively parallel sequencing (MPS) is used increasingly in clinical settings. A recent systematic review by Mone et al.¹⁴ demonstrated that exome sequencing afforded an average incremental diagnostic yield of 29% in NIHF over chromosomal microarray analysis (CMA) or karyotyping. Sparks et al.¹⁵ showed that approximately 30% of NIHF cases with an identified monogenic cause belonged to the group of RASopathies, primarily due to variants in PTPN11, causing Noonan syndrome, or in HRAS, causing Costello syndrome. In the systematic review by Mone et al.¹⁴, RASopathies comprised approximately 30% of NIHF cases diagnosed by exome sequencing, most of which were due to PTPN11 variants. Whole‐genome sequencing (WGS) may offer an even more comprehensive method for identifying the underlying cause of NIHF by enabling the simultaneous detection of coding and non‐coding single‐nucleotide variants (SNVs) and insertions and deletions (INDELs), as well as copy‐number variants (CNVs), short tandem repeats (STRs) and balanced chromosome aberrations. To the best of our knowledge, there are no previous studies evaluating prenatal diagnosis of NIHF using WGS in a clinical setting. Hence, the primary aim of this study was to determine the diagnostic yield of clinical WGS in detecting causative variants in prenatally diagnosed NIHF. The secondary aim was to investigate the genetic etiologies identified by WGS.

METHODS

Study population

This was a retrospective study of all prenatally diagnosed cases of NIHF referred for clinical WGS with a hydrops fetalis gene panel at the Department of Clinical Genetics, Karolinska University Hospital, Sweden, between 13 September 2017 and 1 April 2021. The study was approved by the Swedish Ethical Review Authority (reference number: 2020‐05297).

Hydrops fetalis was detected by prenatal ultrasound in all cases but one (Case 15, Table 1). No pathophysiological reason for hydrops was reported by the referring clinicians in any of the cases. Prior to WGS analysis, subjects were analyzed according to our routine clinical protocol, including quantitative fluorescent polymerase chain reaction (QF‐PCR) and in most cases CMA, in order to exclude cases with causative chromosomal aberrations. This resulted in inclusion of 23 fetuses without chromosomal abnormalities in the main cohort of the study (Table 1). Specifically, 20 cases had normal results on both QF‐PCR and CMA, two cases had normal results on QF‐PCR only and one case had normal results on QF‐PCR and karyotype analysis.

Table 1.

Clinical characteristics, pregnancy outcome and chromosomal analysis of 23 prenatally diagnosed cases of non‐immune hydrops fetalis

Case	GA at referral (weeks)	Clinical information	Sample	Pregnancy outcome	Preceding genetic analysis
1	19 + 3	Hydrops fetalis, oligohydramnios and fetal akinesia	Lung tissue	TOP	QF‐PCR: XY; CMA: normal
2	18 + 0	Hydrops fetalis	Lung tissue	Miscarriage	QF‐PCR: XX; CMA: normal
3	18 + 0	Hydrops fetalis	Lung tissue	Miscarriage	QF‐PCR: XX; CMA: normal
4	30 + 0	Hydrops fetalis	CVS	IUFD	QF‐PCR: XY; CMA: normal
5	21 + 0	Hydrops fetalis, suspected hypertelorism, low‐set ears, hepatomegaly, abnormal hands and short extremities	Spleen tissue	TOP	QF‐PCR: XX; CMA: normal
6	20 + 1	Hydrops fetalis, hygroma, pelvic cyst and unilateral renal agenesis	AC	TOP	QF‐PCR: XX; CMA: normal
7	29 + 0	Hydrops fetalis, polyhydramnios and unilateral multicystic dysplastic kidney	AC	IUFD	QF‐PCR: XX; CMA: normal
8	18 + 2	Hydrops fetalis, IUGR, suspected atrial septal defect	Lung tissue	Miscarriage	QF‐PCR: XX; CMA: normal
9	14 + 2	Hydrops fetalis, hygroma	CVS	TOP	QF‐PCR: XY; CMA: normal
10	12 + 2	Hydrops fetalis, hygroma and liver calcifications	Lung tissue	TOP	QF‐PCR: XY; CMA: normal
11	24 + 1	Initially dichorionic diamniotic twin pregnancy: Twin 1 missed miscarriage, Twin 2 hydrops fetalis	CVS	IUFD	QF‐PCR: XX
12	19 + 0	Hydrops fetalis, IUGR, micrognathia, low‐set ears, camptodactyly, lung hypoplasia and liver calcification	Lung tissue	Miscarriage	QF‐PCR: XY; CMA: normal
13	18 + 0	Hydrops fetalis with prior nuchal translucency thickness 4.4 mm	Lung tissue	Miscarriage	QF‐PCR: XX; CMA: normal
14	13 + 0	Hydrops fetalis	CVS	TOP	QF‐PCR: XY; CMA: normal
15	36 + 5	Consanguineous marriage, three previous pathological pregnancies with hydrops fetalis; current pregnancy with IUFD but no sign of hydrops fetalis at time of assessment	Lung tissue	IUFD	QF‐PCR: XY; karyotype analysis (on CVS): XY
16	18 + 0	Hydrops fetalis and IUGR	AC	Miscarriage	QF‐PCR: XY; CMA: normal
17	23 + 0	Resolving hydrops fetalis, micropenis, cardiomegaly and increased nuchal fold	CVS	Live birth	QF‐PCR: XY; CMA: normal
18	30 + 2	Severe hydrops fetalis	AC	Live birth	QF‐PCR: XY; CMA: normal
19	19 + 6	Hydrops fetalis	AC	TOP	QF‐PCR: XX
20	19 + 2	Hydrops fetalis with multiple malformations	Lung tissue	TOP	QF‐PCR: XX; CMA: normal
21	28 + 0	Hydrops fetalis	AC	IUFD	QF‐PCR: XY; CMA: normal
22	13 + 2	Mild hydrops fetalis	CVS	Live birth	QF‐PCR: XY; CMA: normal
23	32 + 4	Hydrops fetalis	AC	IUFD	QF‐PCR: XY; CMA: normal

Clinical information was obtained from referral letters regarding ultrasound findings and autopsy reports, as applicable.

AC, amniocentesis; CMA, chromosomal microarray analysis; CVS, chorionic villus sampling; GA, gestational age; IUFD, intrauterine fetal demise; IUGR, intrauterine growth restriction; QF‐PCR, quantitative fluorescence polymerase chain reaction (targeting chromosomes 13, 18, 21, X and Y); TOP, termination of pregnancy.

Depending on specific sample availability in each pregnancy, fetal samples for DNA extraction were obtained via amniocentesis (n = 7), chorionic villus sampling (n = 6) or sampling of frozen fetal tissue (n = 10). The latter was performed at postmortem examination following miscarriage, intrauterine fetal demise (IUFD; defined as fetal demise > 22 weeks' gestation) or termination of pregnancy. Gestational age at referral ranged from 13 to 37 weeks. Twelve of the 23 pregnancies ended in miscarriage or IUFD, eight pregnancies were terminated and three pregnancies resulted in a live birth.

Due to identification by WGS of a high proportion of causative variants in the HRAS gene in our main cohort, Sanger sequencing of HRAS was performed in a separate replication cohort. This included 24 fetuses with NIHF and normal results on CMA, but for whom WGS analysis had not been requested by the referring clinician.

Whole‐genome sequencing

WGS of genomic DNA extracted from the fetal samples was performed in a collaborating laboratory (Clinical Genomics; Science for Life Laboratory, Solna, Sweden). The detailed process for WGS analysis and variant calling has been described previously ¹⁶ , ¹⁷ . Briefly, extracted DNA was converted to sequencing libraries using a PCR‐free paired‐end protocol: Illumina TruSeq DNA PCR‐free (Illumina Inc., San Diego, CA, USA) for inputs > 1000 ng or Lucigen NxSeq AmpFREE Low DNA (Lucigen Corp., Middleton, WI, USA) for inputs > 100 ng. Sequencing was performed on the Illumina Hiseq X Ten (n = 10) or Illumina NovaSeq 6000 (n = 13) platforms (Illumina Inc.). Bioinformatics analysis was conducted using the Mutation Identification Pipeline (MIP) framework ¹⁸ . FASTQ data were aligned to the human reference genome (hg19) prior to variant calling, which included SNVs, INDELs, structural variants (SVs), STRs and copy‐number identification of the SMN1 and SMN2 genes. Subsequently, variants were annotated and prioritized based on factors such as variant type, mode of inheritance and allele frequency, before being transferred to Scout, a graphical user interface for MPS data and metadata ¹⁷ .

Clinical variant interpretation was performed by filtering the called variants against an in‐silico gene panel comprising 281 genes known to be associated with hydrops fetalis (Table S1). This clinical gene panel is updated regularly (typically 2–4 times per year) based on literature searches and comparisons to publicly available gene lists from other genetic laboratories. Variants were filtered further so that only exonic and predicted splice variants were retained regarding SNVs, INDELs and SVs. The minor allele frequencies of the variants were annotated according to the Genome Aggregation Database (gnomAD) ¹⁹ and a local in‐house database, and variants with minor allele frequencies of > 1% were excluded. Remaining variants were then classified into five categories (benign, likely benign, uncertain significance, likely pathogenic and pathogenic) according to the guidelines of the American College of Medical Genetics and Genomics ²⁰ . Single pathogenic variants in autosomal recessive disease genes were excluded if a compound variant was not detected. Finally, a multidisciplinary team assessed the remaining putatively causative variants based on available prenatal phenotype and pedigree information, which in most cases was extracted from referral letters. Autopsy reports were reviewed in cases of miscarriage/IUFD and pregnancy termination. The primary analyses were performed using proband‐only (fetal without parental) samples. Parental samples were obtained only if needed to classify a variant of uncertain significance or to investigate recurrence risk.

CNV analysis

In addition to the analysis of genetic aberrations included in the hydrops fetalis gene panel, a whole‐genome analysis of large CNVs, corresponding to a CMA, was performed using vcf2cytosure ²¹ . This enables visualization and interpretation of CNVs in CytoSure Interpret Software (Oxford Gene Technology, Begbroke, UK), which is used routinely in our laboratory to analyze CMA data.

HRAS sequencing

All four coding exons and intron–exon boundaries of HRAS (NM_176795.4) were amplified by PCR and sequenced according to standard Sanger methodology (primers and PCR conditions are available on request).

RESULTS

The main study cohort included 23 fetuses without chromosomal aberrations. Clinical WGS analysis using an in‐silico hydrops fetalis gene panel resulted in a molecular diagnosis in 52.2% (12/23) of the studied NIHF cases. A summary of the causative variants detected by WGS is presented in Table 2. The causative variants were located in seven genes, namely HRAS (5/12, 41.7%), RIT1 (2/12, 16.7%), FOXP3 (1/12, 8.3%), GLB1 (1/12, 8.3%), MAP2K1 (1/12, 8.3%), PTPN11 (1/12, 8.3%) and RASA1 (1/12, 8.3%). Of the 12 causative variants, three (25.0%) were novel, comprising one missense variant in HRAS (c.182A>G; p.(Gln61Arg)), one 278‐bp hemizygous deletion in FOXP3 (Figure 1) and one frameshift variant in RASA1. Three variants of uncertain significance were identified in three different genes (PIEZO1, SOS2 and FLT4), but reclassified as likely benign after analysis of parental samples. No SVs or STRs were identified, nor abnormal copy‐number variation in the SMN1 and SMN2 genes. No large CNVs were detected in the vcf2cytosure analysis, which was expected as most of the cases had been analyzed previously with CMA.

Table 2.

Summary of 12 causative variants detected by whole‐genome sequencing in 23 prenatally diagnosed cases of non‐immune hydrops fetalis

Gene (transcript)	Case	Coding change	Zygosity	ACMG class	Diagnosis	Inheritance mode	Parental source
FOXP3 (NM_014009.3)	12	c.(1045‐219)_1104del	Hemi	L‐path	Immunodysregulation, polyendocrinopathy and enteropathy, X‐linked (IPEX) (OMIM #304790)	XLR	Mat
GLB1 (NM_000404.4)	15	c.1077del; p.(Val360Tyrfs*23)	Hom	Path	GM1 gangliosidosis (OMIM #230500, 230 600, 230 650)	AR	Mat/pat
HRAS (NM_176795.4)	5	c.35G>A; p.(Gly12Asp)	Het	Path	Costello syndrome (OMIM #218040)	AD	DN
	10	c.38G>A; p.(Gly13Asp)	Het	Path	Costello syndrome*	AD	DN
	13	c.38G>A; p.(Gly13Asp)	Het	Path	Costello syndrome*	AD	NA
	14	c.38G>A; p.(Gly13Asp)	Het	Path	Costello syndrome*	AD	NA
	3	c.182A>G; p.(Gln61Arg)	Het	L‐path	Costello syndrome*	AD	NA
MAP2K1 (NM_002755)	19	c.158T>C; p.(Phe53Ser)	Het	Path	Cardiofaciocutaneous syndrome 3 (OMIM #615279)	AD	NA
PTPN11 (NM_002834)	21	c.215C>G; p.(Ala72Gly)	Het	Path	Noonan syndrome 1 (OMIM #163950)	AD	NA
RASA1 (NM_002890)	23	c.2299del; p.(Ser767Argfs*7)	Het	Path	Capillary malformation‐ arteriovenous malformation 1 (OMIM #608354)	AD	DN
RIT1 (NM_006912.6)	9	c.229G>A; p.(Ala77Thr)	Het	Path	Noonan syndrome, type 8 (OMIM #615355)	AD	DN
	6	c.270G>A; p.(Met90Ile)	Het	Path	Noonan syndrome, type 8*	AD	DN

^* OMIM as above.

ACMG, American College of Medical Genetics and Genomics; AD, autosomal dominant; AR, autosomal recessive; DN, de novo; hemi, hemizygous; het, heterozygous; hom, homozygous; L‐path, likely pathogenic; mat, maternal; NA, not applicable; pat, paternal; path, pathogenic; XLR, X‐linked recessive.

Figure 1.

Integrative Genomics Viewer (IGV) showing 278‐bp deletion in FOXP3 gene in Case 12. Since FOXP3 is located on X chromosome and Case 12 was male, IGV shows complete lack of reads within the gene, indicating hemizygous deletion.

The inheritance patterns of the causative variants were 10/12 autosomal dominant (HRAS, MAP2K1, PTPN11, RASA1 and RIT1), 1/12 autosomal recessive (GLB1) and 1/12 X‐linked recessive (FOXP3). Parental samples were obtained in seven (58.3%) of the 12 cases with a detected causative variant. Both of the RIT1 variants were de novo, as were the two tested HRAS variants and the RASA1 variant. The homozygous GLB1 variant was inherited from parents who were both heterozygous carriers. The deletion in FOXP3 was detected in a male fetus. The parental samples showed that it was inherited from the mother, which is in line with the mode of inheritance of an X‐linked disease.

In order to investigate further the high number of HRAS variants in our cohort, Sanger sequencing of HRAS was performed on a separate replication cohort of 24 NIHF cases. In one (4.2%) of these 24 samples, a pathogenic variant of HRAS (c.35G>A; p.(Gly12Asp)) was detected. This means that the overall frequency of causative HRAS variants in our two cohorts was 12.8% (6/47).

DISCUSSION

In this study of prenatally diagnosed cases of NIHF prescreened for chromosomal aberrations, clinical WGS using an in‐silico hydrops fetalis gene panel detected a causative variant in 52.2% of fetuses. To the best of our knowledge, this is the first study to evaluate the diagnostic yield of WGS in NIHF. Our findings suggest that WGS could be an effective first‐line test for NIHF and, considering that up to 50% of prenatally diagnosed cases of NIHF are caused by chromosomal aberrations ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , indicate that the diagnostic yield of WGS could be as high as 75%.

The diagnostic yield of WGS in this study is higher than the 29% incremental yield reported in a previous systematic review evaluating prenatal exome sequencing for the diagnosis of NIHF¹⁴. This disparity may be explained by the small sample size of this study and our inclusion of NIHF cases in which the referring clinician requested specifically analysis with the hydrops fetalis gene panel. As WGS is still considered relatively expensive, it is likely to be offered to more severely affected fetuses, potentially creating a selection bias in our cohort. The possible over‐representation of severe cases in our cohort is underlined by the fact that approximately half of the pregnancies ended in miscarriage or IUFD. Possible selection bias may also explain why only one causative variant in HRAS was detected by Sanger sequencing in the 24 additional NIHF cases for which a gene panel had not been requested. Due to the diversity and rarity of genetic etiologies of NIHF, a comprehensive evidence‐based approach is essential to the evaluation of these cases. WGS enables the simultaneous detection of different genetic variants across the entire genome, from single nucleotides to whole chromosomes. Exome sequencing often requires adjunctive CMA in order to detect CNVs. Thus, WGS has the potential to yield a more efficient diagnosis compared with exome sequencing alone. Because this study was performed in a clinical setting, only exonic and predicted splice variants were included regarding SNVs, INDELs and SVs. This, combined with our use of a singleton gene panel rather than a trio analysis, might have reduced the diagnostic yield of WGS by preventing the identification of novel genes and novel variants in non‐coding regions of known genes. Unlike exome sequencing, our approach with WGS, including the analysis of large CNVs, enabled the investigation of not only SNVs and INDELs but also CNVs of different sizes, SVs, STRs and copy‐number identification of the SMN1 and SMN2 genes. However, only one of the detected variants in this study, the 278‐bp deletion in FOXP3, could have been a borderline case for detection by exome sequencing. Further research is needed to determine the precise yield of WGS in cases of NIHF and its clinical applicability considering the time constraints of an ongoing pregnancy.

In this study, WGS identified causative variants in 12/23 analyzed NIHF cases, comprising seven different genes associated with seven different diagnoses (Table 2). Most (5/7, 71.4%) of the detected diagnoses belonged to RASopathies, with a causative variant found in HRAS, MAP2K1, PTPN11, RASA1 or RIT1. This is in line with previous studies on NIHF, albeit with a slightly higher proportion of RASopathies ¹⁴ , ¹⁵ . Interestingly, we found an unexpectedly high incidence of Costello syndrome, as causative variants in HRAS were detected in 5/12 (41.7%) cases receiving a molecular diagnosis.

The most frequently reported germline HRAS variants are de‐novo variants affecting glycine in codon 12 or 13, of which the most common are p.(Gly12Ser) and p.(Gly12Ala) ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ . Accordingly, we found that four of five identified HRAS variants were located in these codons, whereas the fifth was located in codon 61. Germline variants and somatic variants in sporadic tumors share codons 12 and 13 as mutational hotspots in HRAS, while codon 61 has been regarded to be a mutational hotspot only in sporadic tumors ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ . A literature search identified only one reported case of a germline variant located in codon 61 of the HRAS gene in an individual diagnosed with Costello syndrome ³¹ . The fetus presented with nuchal edema, polyhydramnios and signs of fetal akinesia, and exome sequencing detected a likely pathogenic missense variant (c.183G>C; p.(Gln61His)) in codon 61 of HRAS ³¹. This case, together with the variant detected in this study (c.182A>G; p.(Gln61Arg)), suggests that germline variants in codon 61 of the HRAS gene should be taken into account when interpreting results from MPS.

This study highlights the benefits of MPS in prenatal diagnosis. A genetic diagnosis is important for parents and clinicians as it offers an explanation, informs the prognosis, guides prenatal and postnatal care and enables adequate estimation of the recurrence risk in future pregnancies. In this study, 83.3% of identified causative variants were autosomal dominant. Detection of inherited causative variants, i.e. GLB1 and FOXP3 variants, offers families the possibility of prenatal diagnosis or preimplantation genetic testing in future pregnancies.

In summary, in a cohort of 23 prenatally diagnosed cases of NIHF in which chromosomal aberrations had been excluded, a genetic etiology was established in 52% using WGS with an in‐silico hydrops fetalis gene panel. Considering that up to 50% of prenatally diagnosed cases of NIHF are attributed to chromosomal aberrations, a detection rate of up to 75% may be expected when using WGS including an analysis of chromosome aberrations. Most of the causative variants were found in HRAS, highlighting Costello syndrome as an important cause of NIHF.

Supporting information

Table S1 Content of targeted in‐silico hydrops fetalis whole‐genome sequencing gene panel

DATA AVAILABILITY STATEMENT

The data are not publicly available due to the sensitivity of the clinical data.