Diagnostic yield and clinical impact of prenatal whole exome sequencing (WES) – four-year single center experience

Introduction

Prenatal whole exome sequencing (WES) represents a significant advancement in prenatal diagnostics as it considerably enhances our ability to establish genetic diagnoses in a prenatal setting [1], 2]. When used subsequent to genome-wide deletion/duplication analysis, WES has been shown to increase diagnostic yield by up to 15 % in fetuses with at least one ultrasound anomaly [3], [4], [5], [6] and even up to 21.8 % in fetuses with multiple anomalies [3]. Given the more limited phenotyping options during pregnancy compared to a postnatal setting, early genetic diagnoses are invaluable, as they aid parents in informed decision-making and are essential for optimal postnatal care planning.

However, implementing prenatal WES in a routine diagnostic setting has its challenges, as reported in our previous paper [7], including [1] the quality and quantity of the starting material [2]; the minimization of the turn-around time (TAT) [3]; the interpretation of variants despite limited phenotypic information and [4] the ethical perspective. Despite these hurdles, prenatal WES is an important evolution in the prenatal diagnostic landscape. This study expands on our previous paper which described the implementation and the first year of prenatal WES at our center [7]. Now we report on the diagnostic yield but in particular on the clinical impact of prenatal WES over a period of more than four years.

Materials and methods

The Center of Medical Genetics in Antwerp, one of eight genetic centers in Belgium, annually processes approximately 500 invasive prenatal samples. These samples are collected from the Antwerp University Hospital and several other hospitals in the region. Indications for invasive testing primarily include a known familial genetic disorder, a positive non-invasive prenatal test (NIPT) result or ultrasound anomalies. Ultrasound anomalies qualifying for invasive genetic testing included the presence of at least one major anomaly or more than one minor anomaly, according to institutional guidelines. At our center, the standard procedure for determining the genetic cause of fetal ultrasound anomalies, regardless of gestational age, involves a step-by-step approach. After extraction of genomic DNA from either amniotic fluid or chorion villi, a quantitative fluorescent PCR (QF-PCR) is conducted to rule out common aneuploidies (trisomies 13, 18, 21, sex chromosome aneuploidies) and triploidy. This step also includes checking for maternal cell contamination and confirming fetal identity by comparison of the fetal and maternal profile. Subsequently, CNV (copy number variant) detection is carried out using single nucleotide polymorphism (SNP) array (CMA) or shallow Whole Genome Sequencing (sWGS), both using the CNV Webstore tool [8] with a 400 kb resolution.

Following negative QF-PCR and CNV analysis, further options in fetuses with at least one ultrasound anomaly include single-gene testing (in case of a high suspicion of a particular disorder) or WES. The decision is made on a case-by-case basis during weekly multidisciplinary meetings.

The guidelines for prenatal WES were developed by the Belgian genetic centers and can be found at the website of the Belgian College of Genetics (www.college-genetics.be (V2021)). Variant classification is performed based on the ACMG guidelines [9] and only pathogenic (class V) and likely pathogenic (class IV) variants with a known effect on gene function and matching the fetal phenotype and the inheritance mode are communicated. Variants of uncertain significance (class III) are in principle not communicated, but exceptions can be made for variants for which further clinical exams (ultrasound, MRI, etc.) are recommended to refine variant classification, possibly leading to a genetic diagnosis (upgrade of the variant to class IV/V). The turn-around-time (TAT) was nationally set at eight weeks for ongoing pregnancies.

Library prep is performed on 50 ng of genomic DNA from uncultured or cultured amniocytes or chorion villi using the Twist Human Core Exome kit (Twist Bioscience, South San Francisco, CA, USA) according to the manufacturer’s instructions on a Hamilton STAR robot (Hamilton, Bonaduz, Switzerland). Twenty-four libraries are pooled equimolarly for sequencing on a NextSeq500 or NextSeq550 instrument with a 2 × 75 bp or 2 × 150 bp flow cell (Illumina, San Diego, CA, USA). WES data are analyzed using an in-house developed pipeline which considers only de novo, X-linked and recessive variants, either in a predefined panel (e.g., in case of a well-defined phenotype like skeletal dysplasia) or exome-wide [7], 10]. Additionally, an AI-driven decision-support software is applied to most samples to complement our pipeline with an independent phenotype-driven analysis based on Human Phenotype Ontology (HPO) terms (MOON, Invitae, San Francisco, CA, USA or Franklin, Genoox). This allows the identification of variants outside the panel (if applied) and of dominantly inherited variants. An independent analysis using single-molecule molecular inversion probes (smMIPs) is performed to detect sample swaps and to verify the family relations within each trio.

Results

In the period March 2020-August 2024, 171 WES were performed on prenatal cases that met the national criteria for exome sequencing in an ongoing pregnancy. Eighty-five of these analyses were performed on DNA extracted from uncultured amniotic cells (49.7 %), 58 from cultured amniotic cells (33.9 %) and 28 from chorion villi (16.4 %).

Since 2021, a steady increase in the number of prenatal WES analyses can be observed (see Table 1). This increase is not related to an increase in the total number of invasive tests, which is stable over this five-year period. Rather, it reflects an increase in the percentage of prenatal samples for which WES is requested (from 7.5 to 16.3 %).

For the majority of cases (158/171 or 91.8 %), an ‘open’ WES was performed, as the ultrasound anomalies did not allow the selection of a predefined gene panel. For the 13 remaining cases, the analysis was restricted to the skeletal dysplasia gene panel (n=10), the cerebral palsy panel (n=1) or a specific gene (n=2) for which no single gene test was available.

The turnaround time (TAT) was defined as the time between the finalization of the CMA/sWGS report and the WES report. If the request for exome sequencing was made any time after finalizing the CMA/sWGS report, the request date was taken as the start date. For cultured amniocytes, the time required for culturing was included in the calculation of the TAT, as to reflect the impact of this option. The general median TAT for all samples combined was 16 days. For uncultured amniocytes or chorion villi samples the median TAT was 15 and 15.5 days, respectively (range 5–62 and 4–26) (Figure 1). With the exception of two cases, all cases were reported within one month. For one of these cases (#37, not in table since no (likely) pathogenic variant was identified), the urgency of the analysis was no longer applicable since the parents did not consider a termination of pregnancy (TOP), but antenatal reporting of the results ensured optimal postnatal care planning. In the second case (#18), the result, a variant of uncertain significance (VUS) in the FLNA gene, did not align with the prenatal phenotype. Since the parents did not consider TOP, regardless of the genetic diagnosis, it was decided to perform more extensive prenatal phenotyping through a fetal MRI hoping it would allow to upgrade the FLNA variant to class IV in case of periventricular nodular heterotopia (PVNH). However, as the fetal MRI failed to show PVNH, a negative WES report was issued after 62 days. A postnatal brain MRI did reveal PVNH, resulting in the variant being upgraded to class IV. The requirement of culturing the amniocytes in case of insufficient yield increased the median TAT to 19 days (range 8–49); four samples had a TAT that surpassed four weeks. There was no significant difference between the TAT for a panel (median 14 days, range 6–25) or ‘open’ WES (median 16 days, range 4–62, p=0.30).

Figure 1:

Turnaround time for (un)cultured AC or CV with median. AC, amniocentesis; CV, chorion villi sampling. Horizontal line represents median, x represents average.

Pathogenic or likely pathogenic variants were identified in 36/171 (21.1 %) cases: 19 de novo variants, 14 autosomal recessive, one autosomal dominant inherited from an affected father and two X-linked dominant variants inherited from an (in hindsight) affected mother (Table 2).

When grouped based on the organ system(s) involved (see also Mellis et al. 2022 [4]), the highest diagnostic yield was obtained in fetuses with skeletal anomalies (4/14 cases, 28.6 %), hydrops (2/7, 28.6 %), multisystem anomalies (24/87, 27.6 %) or central nervous system anomalies (5/28, 17.9 %). In 20 fetuses with a cardiac anomaly, only one diagnosis was obtained. No diagnosis was found in one case with a neoplasm and in 14 cases with isolated severe fetal growth restriction (Figure 2).

Figure 2:

Distribution of cases according to involved organ system(s) and the results of the WES analysis. CNS, central nervous system; FGR, fetal growth restriction.

In the majority of cases the molecular diagnosis was made prenatally. In three cases (#11,#12 and #22) fetal distress or maternal complications led to the obstetric decision to induce a premature delivery, and the WES result was issued in the perinatal period. This eliminated the parental choice for TOP based on the genetic diagnosis.

The prenatal phenotype was consistent with the diagnosis and published prenatal phenotype in all 36 cases. However, in 11 fetuses (cases #1,#7,#8,#9,#13,#18,#20,#21,#26,#30 and #32), the ultrasound abnormalities remained partially unexplained, which might represent unreported or new prenatal presentations of the diagnosed syndrome or could be phenotypic presentations of a second (genetic) cause (Table 2).

Of the 36 cases in which a (likely) pathogenic variant was identified, parents chose to terminate the pregnancy in 21 cases (58.3 %). In three cases (8.3 %), intrauterine death occurred which could be explained by the genetic diagnoses: Greenberg dysplasia (case #17) [11], Noonan syndrome with hydrops (RIT1, case #4) [12] and lethal type multiple pterygium syndrome (case #3) [13], [14], [15], [16]. 11 cases (30.6 %) were brought to term and one case was lost to follow-up (2.8 %). Table 3 provides an overview of the prenatal and postnatal phenotypes of all cases brought to term. For these cases, having a genetic diagnosis was relevant in several ways. In one case (#15), we identified a non-syndromic cause of cleft lip (ARHGAP29 mutation). The exclusion of a syndromic disorder associated with the observed bilateral cleft lip and palate and the fact that it was paternally inherited, gave the parents confidence to proceed with the pregnancy. In 8 cases (72.7 %; cases #10,#11,#12,#18,#21,#26,#29 and #36, details in table 3) the diagnosis facilitated optimal postnatal management, enabling immediate screening and treatment of health problems associated with the diagnosed syndrome. In two cases (18.2 %, cases #22 and #24), the decision to abstain from care was made based on a genetic diagnosis with poor prognoses, i.e. MIRAGE and CHIME syndrome. MIRAGE syndrome is characterized by myelodysplasia, infections, growth restriction, adrenal insufficiency, genital abnormalities and enteropathy. Most patients die in infancy or early childhood, with a median age at death of three years [17], [18], [19], [20]. CHIME syndrome is a rare ectodermal dysplasia syndrome characterized by ocular colobomas, heart defects, ichthyosiform dermatosis, intellectual disability, ear anomalies (e.g. conductive hearing loss) and epilepsy, with a reported poor prognosis [21], 22].

Discussion

This study reports on the experience with prenatal WES over a period of more than four years, analyzing 171 cases and yielding 36 (21.1 %) (likely) diagnoses. This yield is higher than what is generally reported in the literature [3], [4], [5], [6], which can probably be explained by the fact that the majority of the included cases had multiple anomalies. Our results confirm that the diagnostic yield varies greatly depending on the organ system involved; it is highest in fetuses with hydrops and skeletal or multisystem anomalies [3], 4], 6] and lowest in case of cardiac anomalies or isolated fetal growth restriction. While this information can assist clinicians in deciding whether to recommend an exome analysis, it should be kept in mind that the fetal phenotype is subject to change throughout pregnancy, and that prenatal phenotyping has its limitations. In this context, one could advocate for performing prenatal WES in all pregnancies exhibiting ultrasound anomalies, even isolated minor ones, as the selection of candidates for prenatal WES relies on the limited information available from the ultrasound at a specific gestational age [4]. However, given the cost of WES and depending on the healthcare system, information on the diagnostic yield associated with different types of ultrasound anomalies is of great help to determine which pregnancies would benefit the most from WES. One could also argue that exome sequencing in fetuses without ultrasound anomalies is warranted, as a phenotype might develop later in pregnancy or even only postnatally (in case of e.g. neurodevelopmental disorders). Although this has become common practice in some countries [34], we do not foresee this to happen in Belgium in the near future, nor is it currently supported by the International Society for Prenatal Diagnosis [1]. First, there is a risk of a procedure-related miscarriage. Second, there is no such thing as the (genetically) perfectly healthy baby. Even if we screen for all known variants in currently known genes associated with early-onset severe disorders, we will still miss numerous diagnoses since [1] the causal variant can reside in a non-coding or other region inaccessible with exome sequencing [2]; not all genes that have been associated with disease are known [3]; not all types of variants can be detected with the current prenatal diagnostic tests (e.g. inversions or balanced translocations interrupting a gene, copy number variants between ∼50 bp and 400 kb) [35]. Therefore, we deem it more useful to improve the currently available technologies or implement new technologies to increase the diagnostic yield in fetuses with ultrasound anomalies that are likely to be genetic in origin. We do observe a steady increase in the percentage of prenatal samples for which WES is requested in our center (from 7.5 to 16.3 %). This growth is largely attributed to the broader spectrum of indications for prenatal WES.

Prenatal WES is performed to allow informed decision-making for expecting parents. In 21/36 (58.3 %) cases the parents opted for a TOP because of ultrasound anomalies in combination with the genetic diagnosis, while in 11/36 cases the parents and medical team had the opportunity to prepare optimal postnatal care. This corresponds to the results of a similar study in which in 48.9 % of cases with a molecular diagnosis parents decided to terminate the pregnancy [3].

In 2023, we reported on our experience of the first year of prenatal WES in our center [7]. In that paper, we referred to the many hurdles that had to be overcome to offer WES in the prenatal setting including [1] the quality and quantity of the starting material [2]; the short TAT [3]; the interpretation of variants and [4] the ethical perspective. Well into our fifth year of offering prenatal WES, we are in the position to comment on these hurdles [1]. In our hands, DNA extracted directly from amniotic fluid or chorion villi is of high quality and maternal cell contamination is rarely an issue. As the input for our library prep is only 50 ng, the majority (66,1 %) of our exome analyses can be performed on DNA extracted from uncultured cells, reducing the TAT with a couple of days (median 19 vs. 15 days) [2]. With biweekly sequencing runs, the median turnaround time (TAT) is 16 days, significantly shorter than the national threshold of 8 weeks. This enables parents to make timely decisions regarding the continuation or termination of their pregnancy. Even when WES is requested in the third trimester, results are available before birth, facilitating the implementation of appropriate neonatal management [3]. Our variant detection and filtering pipeline has been optimized to minimize the number of variants that need manual curation. This again reflects positively on the TAT and drastically reduces the number of variants of unknown significance and incidental findings that require discussion in a multidisciplinary team meeting. On a total of 171 cases, only 2 ‘hot’ class 3 variants (1.2 %; data not shown) and two incidental findings (1.2 %; data not shown) have been reported, even though most analyses were exome-wide and not panel-based. While variant interpretation remains challenging due to the limited prenatal phenotypic information, our pipeline ensures that clinically relevant variants are accurately identified [4]. From an ethical perspective, it is paramount to balance the benefits of prenatal diagnosis against the potential psychological impact on expectant parents. The discovery of incidental findings, in particular, can add complexity to the counseling. However, with our low number of incidental findings and adequate pre- and post-test counseling, we demonstrate prenatal WES to be beneficial to parents in case of ultrasound anomalies.

Obtaining a diagnosis is not only of relevance in the context of the ongoing pregnancy, but also for future pregnancies. In case of a de novo variant, the recurrence risk is low, but not zero, as low-level parental mosaicism cannot be excluded; an invasive test followed by targeted sequencing of the familial variant can bring reassurance to parents. In case of an inherited variant, the recurrence risk can be as high as 50 % and preimplantation genetic testing can be offered, reducing the risk tremendously. In our cohort, there was a high recurrence risk (25–50 %) in 17/36 (47.2 %) of diagnoses, underscoring the impact of WES on family planning.

In three out of 36 positive cases (8.3 %), the diagnosis would have been missed by using a panel-based approach like the Genomics England Fetal Anomalies panel (2,187 genes; version v5.0), demonstrating that a) it remains difficult to keep panels up to date, as new gene-disease correlations are published on a regular basis and b) the described postnatal phenotype might not warrant the inclusion of a particular gene in the Fetal Anomalies panel, since no fetal phenotype has been described thus far. For the disorders detected in these three cases, there was low or no evidence for a prenatal phenotype, explaining why the genes (COL27A1, SMAD6 and SKIC3) are not included in the Genomics England Fetal Anomalies panel (Supplementary Table S1). By opting for an open WES, we contribute in defining the fetal phenotype of known disorders while, with optimized bioinformatic pipelines, keeping the number of incidental findings and variants of unknown significance low.

The use of Moon, Franklin or any other AI-based software for HPO-based prioritization of variants is complementary to our own pipelines for two reasons [1]; it can prioritize variants in novel genes in case of a panel-based approach and [2] it will return dominantly inherited variants that fit the phenotype but can have a variable presentation. In our cohort of 36 positive cases, three (8.3 %) carried a dominantly inherited variant, which would have been missed in 2/3 cases when using the standard filtering only (for de novo, compound heterozygous, homozygous and hemizygous variants).

The interpretation and classification of variants requires detailed prenatal phenotyping. Achieving this, demands highly trained gynecologists and excellent communication between the genetics and gynecology departments. In our center, all cases are discussed during a weekly multidisciplinary prenatal meeting, ensuring the best diagnostic approach and facilitating the interpretation and discussion of genetic testing results when necessary.

A limitation of our study is that neonates born from pregnancies without a genetic diagnosis are not routinely re-evaluated by a geneticist after birth [1], 2]. The decision to do so is left to the discretion of the parents or pediatrician at the hospital where the baby is born, which is often different from the hospital where pregnancy follow-up and the request for genetic testing took place. This lack of routine re-evaluation prevents us from identifying additional phenotypic features and determining if reanalysis of WES data is necessary. The strength of this study lies in the reporting on the diagnostic yield of a comprehensive, unbiased cohort from a single institution over an extended period of time, enabling a broad and thorough evaluation of the performance of prenatal WES.

Conclusions

Our findings indicate that in one out of five pregnancies with ultrasound anomalies and normal deletion/duplication analysis, a (likely) pathogenic variant explaining the phenotype can be identified. The short turnaround time allows for timely decisions regarding the ongoing pregnancy and neonatal management. In addition, we demonstrated an 8.3 % higher diagnostic yield by performing an ‘open’ WES compared to a gene panel like the Genomics England Fetal Anomalies panel. Furthermore, 47.2 % (17/36) of the identified variants were inherited, highlighting the importance of this analysis for future family planning, as the recurrence risk can be as high as 50 %.