Next-generation sequencing is increasingly used in prenatal diagnosis. Targeted gene panels and exome sequencing are both available, but the comparative diagnostic yields of these approaches are not known.
We compared the diagnostic yield of exome sequencing with the simulated application of commercial targeted gene panels in a large cohort of fetuses with nonimmune hydrops fetalis.
This was a secondary analysis of a cohort study of exome sequencing for nonimmune hydrops fetalis, in which recruitment, exome sequencing, and phenotype-driven variant analysis were completed in 127 pregnancies with features of nonimmune hydrops fetalis. An Internet search was performed to identify commercial laboratories that offer targeted gene panels for the prenatal evaluation of nonimmune hydrops fetalis or for specific disorders associated with nonimmune hydrops fetalis using the terms “non-immune hydrops fetalis,” “fetal non-immune hydrops,” “hydrops,” “cystic hygroma,” “lysosomal storage disease,” “metabolic disorder,” “inborn error of metabolism,” “RASopathy,” and “Noonan.” Our primary outcome was the proportion of all genetic variants identified through exome sequencing that would have been identified if a targeted gene panel had instead been used. The secondary outcomes were the proportion of genetic variants that would have been identified by type of targeted gene panel (general nonimmune hydrops fetalis, RASopathy, or metabolic) and the percent of variants of uncertain significance that would have been identified on the panels, assuming 100% analytical sensitivity and specificity of panels for variants in the included genes.
Exome sequencing identified a pathogenic or likely pathogenic variant in 37 of 127 cases (29%) in a total of 29 genes. A variant of uncertain significance, strongly suspected to be associated with the phenotype, was identified in another 12 cases (9%). We identified 7 laboratories that offer 10 relevant targeted gene panels; 6 are described as RASopathy panels, 3 as nonimmune hydrops fetalis panels, and 1 as a metabolic panel. The median number of genes included on each of these panels is 22, ranging from 11 to 148. Had a nonimmune hydrops fetalis targeted gene panel been used instead of exome sequencing, 13 to 15 of the 29 genes (45%–52%) identified in our nonimmune hydrops fetalis cohort would have been sequenced, and 19 to 24 of the pathogenic variants (51%–62%) would have been detected. The yield was predicted to be the lowest with the metabolic panel (11%) and the highest with the largest nonimmune hydrops fetalis panel (62%). The largest nonimmune hydrops fetalis targeted gene panel would have had a diagnostic yield of 18% compared with 29% with exome sequencing. The exome sequencing platform used provided 30× or more coverage for all of the exons on the commercial targeted gene panels, supporting our assumption of 100% analytical sensitivity for exome sequencing.
The broader coverage of exome sequencing for genetically heterogeneous disorders, such as nonimmune hydrops fetalis, made it a superior alternative to targeted gene panel testing.
Next-generation sequencing is increasingly used for the evaluation of fetal structural anomalies. Historically, molecular testing for Mendelian disorders involved the analysis of a single gene, and molecular genetic diagnoses were rarely made before birth, in the absence of a family history. With the advent of next-generation sequencing and the capability to test simultaneously for a large number of genes either as part of a targeted gene panel or with exome sequencing, these methods are increasingly being applied to establish genetic diagnoses in the prenatal setting.
Why was this study conducted?
Nonimmune hydrops is a heterogeneous condition that can manifest in the setting of a broad array of genetic disorders. Targeted gene panels and exome sequencing are both options for the evaluation of affected fetuses, and it is not known how the diagnostic yield differs.
In a cohort of 127 fetuses with nonimmune hydrops, we determined that the use of available targeted gene panels would have detected a pathogenic or likely pathogenic variant in 11% to 62% of the 37 cases that received a genetic diagnosis with exome sequencing or 3% to 18% of the total cases.
What does this add to what is known?
The diagnostic yield of exome sequencing in nonimmune hydrops has been reported to be 29%; the use of targeted gene panels instead of exome sequencing will diagnose substantially fewer cases.
Clinical sequencing in the prenatal period often involves choosing between a targeted gene panel that targets a group of selected genes associated with a similar phenotype and exome sequencing to examine a broader array of genes. Focusing on a restricted set of genes enables greater depth of coverage and therefore can potentially provide greater analytical sensitivity and specificity, particularly for challenging variants, such as small (exon-level) deletions or duplications. In contrast, exome sequencing involves sequencing of all known protein-coding regions of genes that make up 1% to 2% of the entire genome; this approach is often applied to clinical disorders with a broad differential diagnosis. , In addition, providers often raise concerns that exome sequencing may be more likely to report variants of uncertain significance that may be difficult to interpret, particularly in a prenatal setting, although data supporting these concerns are limited. , Furthermore, whether targeted gene panels or exome sequencing is used in a given case is often decided on the basis of insurance coverage or cost considerations. There are limited data comparing the clinical benefits and diagnostic yield of targeted gene panels vs exome sequencing for pediatric populations, and there is a lack of data comparing these approaches for fetal anomalies. , Importantly, there is a lack of data comparing each of these approaches by phenotype, because some phenotypes are associated with a wider differential diagnosis than others.
Nonimmune hydrops fetalis (NIHF) is a complex disorder caused by a broad range of genetic diseases that may manifest with abnormal fetal fluid collections early or late in gestation. This condition affects 1 in 1700 to 3000 pregnancies and is associated with a high risk of stillbirth, preterm birth, and neonatal complications or death. Although cases resulting from aneuploidy can be diagnosed with karyotype or chromosomal microarray analysis, the etiology of most cases remains uncertain after standard evaluation. , As the range of single-gene disorders associated with NIHF has been increasingly recognized, and these are not detected with karyotype or chromosomal microarray, genomic sequencing is more often employed for euploid cases. To date, several laboratories offer targeted gene panels for the evaluation of NIHF; the included genes vary greatly across panels and may include those associated with RASopathies, inborn errors of metabolism, and other categories of disorders. Importantly, the diagnostic yield using these targeted gene panels for unexplained NIHF cases remains unclear. Exome sequencing has been used to assess a large cohort of pregnancies affected with NIHF and identified a causative gene variant in 29% of the cases. ,
Our goal was to compare the diagnostic yield of targeted gene panels and exome sequencing in unexplained NIHF. We performed a secondary analysis of a large cohort that underwent exome sequencing for NIHF, to determine the predicted diagnostic yield, had targeted gene panels been used. We hypothesized that exome sequencing would identify many additional single-gene disorders beyond those detected through targeted gene panels. Given the importance of cost considerations when choosing a testing strategy, we also collected data on the costs of the included targeted gene panels and of prenatal exome sequencing.
Study design and participants
This was a secondary analysis of a cohort of prenatally diagnosed NIHF cases that underwent exome sequencing. The findings of the primary study have been published previously. The cohort included cases with abnormal fetal effusions, including 1 or more of increased nuchal translucency (NT) of ≥3.5 mm, cystic hygroma, pleural effusion, pericardial effusion, ascites, or skin edema. This range in phenotypes was included, as the literature supporting the traditional criteria of ≥2 abnormal fluid collections for a diagnosis of NIHF, is lacking. Furthermore, many genetic disorders associated with abnormal fetal effusions can present early in pregnancy with increased NT or cystic hygroma or later in pregnancy with NIHF as traditionally defined. , , Eligible patients had a nondiagnostic karyotype or chromosomal microarray analysis.
Details regarding the exome sequencing are provided in the previous report, but briefly, trio exome sequencing using DNA from prenatal diagnosis samples was performed in most cases. The University of California, San Francisco (UCSF) Genomic Medicine Laboratory performed exome sequencing with the Illumina HiSeq 2500 (Illumina, Inc, San Diego, CA) or Illumina NovaSeq 6000 sequencing system (Illumina, Inc). Mean sample exome coverage was 80× for the HiSeq and 148× for NovaSeq. Variant Call Format files were uploaded for variant filtering into Ingenuity Variant Analysis (Qiagen, Hilden, Germany) or Moon (Diploid, Leuven, Belgium; and Invitae, San Francisco, CA), clinical informatics experts manually curated the variants, and a multidisciplinary review of curated variants in the context of phenotypic features was performed for each case. Genetic variants were classified according to recommendations of the American College of Medical Genetics and Genomics (ACMG) and the Association for Medical Pathology. In situations where the gene-disease relationship was high but the ACMG criteria for pathogenicity were not met for the specific variant and there was evidence to support a strong potential for clinical significance, the laboratory reported as a variant of uncertain significance (VUS).
We identified commercial laboratories that provide targeted gene panel testing for the prenatal evaluation of NIHF. These laboratories were identified through a general Internet search and query of the Concert Genetics search engine, using terms, including “non-immune hydrops fetalis,” “hydrops fetalis,” “fetal non-immune hydrops,” “hydrops,” “cystic hygroma,” “nuchal translucency,” “lysosomal storage disease,” “metabolic disorder,” “inborn error of metabolism,” “RASopathy,” and “Noonan.” We included panels that test for genes associated with NIHF, including genes causative of disorders known to be associated with NIHF, such as RASopathies, lymphedema disorders, and lysosomal storage diseases. The genes included on each targeted panel were identified on each laboratory’s website. Some laboratories offer >1 relevant panel, for example, a general NIHF panel and a more specific RASopathy panel. In such cases, both targeted gene panels were analyzed and reported separately.
The primary outcome was the proportion of all pathogenic or likely pathogenic genetic variants identified through exome sequencing that would have been identified if a targeted gene panel had instead been used. Secondary outcomes were the hypothetical proportions of genetic variants that would have been identified by type of targeted gene panel (general NIHF, RASopathy, or metabolic), percent of VUS detected by exome that would have been identified on the panels, and proportion of variants that would have been identified through panels for isolated NIHF cases compared with those with additional structural anomalies. These calculations were done assuming 100% analytical sensitivity and specificity. , Genetic variants identified by exome were classified as pathogenic or likely pathogenic by ACMG criteria and as a VUS when ACMG criteria for pathogenicity were not met, but the multidisciplinary review determined the variant to be suspicious and likely to be associated with the phenotype.
To assess the ability of exome sequencing to detect variants identified through targeted gene panels, we determined the coverage of exome sequencing for all genes on the panels, including for genes not identified in any cases in our exome sequencing cohort. To compare costs, we contacted the laboratories that provide targeted gene panels or prenatal exome sequencing and collected data on the costs of each of these tests.
Primary and secondary outcomes were reported as proportions. Statistical analyses were performed in Excel. Approval was obtained through the UCSF Institutional Review Board (IRB) for the primary study; as this secondary analysis used publicly available information from commercial laboratories about targeted gene panels, additional IRB approval was not necessary.
The cohort is described in Table 1 . Of the 127 cases, most fluid collections were in 2 or more cavities (77 [61%]), whereas 21 cases (17%) had a single fetal effusion, such as isolated ascites, and 29 cases (23%) presented with early enlarged NT or cystic hygroma (of which 15 were isolated without other anomalies or additional abnormal fluid effusions). Overall, 64 cases (50%) had a concurrent structural anomaly.
|Median maternal age (IQR), y||32 (29–35)|
|Nulliparous (%)||45 (57/127)|
|Median gestational age at diagnosis of NIHF (range), wk||20.0 (13.4–24.6)|
|Any concurrent anomaly (%)||50 (64/127)|
|Maternal race and ethnicity (%)|
|Hispanic or Latina||9 (12/127)|
|Type of abnormal fetal effusion (%)|
|Early onset (increased NT or cystic hygroma)||23 (29/127)|
|Single abnormal fetal effusion||17 (21/127)|
|Traditionally defined NIHF with ≥2 abnormal effusions||61 (77/127)|
Exome sequencing identified a pathogenic or likely pathogenic variant in 37 of 127 cases (29%). Overall, 29 genes were represented, including 6 for RASopathies, 4 for musculoskeletal disorders, 3 for inborn errors of metabolism, 3 for lymphedema disorders, 3 for neurodevelopmental disorders, 3 for cardiovascular disorders, 2 for hematologic disorders, 2 for immunologic disorders, and 1 each for renal, ciliopathy, overgrowth, and CHARGE (Coloboma of the eye, Heart defects, Atresia of the choanae, Retardation of Growth and development, and Ear abnormalities and deafness) syndrome. Moreover, 4 genes were implicated multiple times, including 4 cases with variants in PTPN11 , 3 cases with HRAS , 3 cases with PIEZO1 , and 2 cases with GUSB . Among the 37 pathogenic or likely pathogenic variants, 16 (43%) were identified in cases with isolated NIHF and 21 (57%) in cases with concurrent structural anomalies. Overall, 9 (24%) presented early with cystic hygroma or increased NT, 2 (5%) with a later single abnormal fetal effusion, and 26 (70%) with fluid effusions in 2 or more cavities.
We identified 7 laboratories that offer 10 relevant targeted gene panels; 6 were described as RASopathy panels, 3 as NIHF panels, and 1 as a metabolic panel. The median number of genes on the RASopathy panels was 19 (11–23). The largest NIHF panel (laboratory 3) was updated to include additional genes after the publication of our primary analysis ; the median number of genes on NIHF panels was 87 (66–128) before this update and 87 (66–148) afterward. The 1 metabolic panel included 51 genes ( Table 2 ). Overall, the targeted gene panels included 169 unique genes, none of the 169 genes were included on all panels, and 57 genes were represented only on a single-gene panel ( Supplemental Table ).
|Laboratory||Targeted gene panel description||Number of genes included||Disorders a||Total genes vs UCSF exome (n=29 genes), n (%)||Total detection vs UCSF exome (n=37 variants), n (%)|
|Laboratory 1||NIHF||87||RASopathies, skeletal dysplasias, metabolic disorders, arthrogryposes, multiple congenital anomaly syndromes||15 (52)||23 (62)|
|RASopathy||23||RASopathies||6 (21)||11 (30)|
|Laboratory 2||Fetal hydrops||66||RASopathies||13 (45)||21 (57)|
|RASopathy||19||RASopathies||6 (21)||11 (30)|
|Laboratory 3||NIHF||Before update b : 128||RASopathies, skeletal dysplasias, metabolic disorders, congenital anemias, arthrogryposes, multiple congenital anomaly syndromes||15 (52)||23 (62)|
|After update b : 148||29 (100)||37 (100)|
|RASopathy||20||RASopathies||6 (21)||11 (30)|
|Laboratory 4||Metabolic NIHF||51||Metabolic disorders only; cases not associated with malformations||3 (10)||4 (11)|
|Laboratory 5||Prenatal Noonan syndrome||19||RASopathies||6 (21)||11 (30)|
|Laboratory 6||Prenatal Noonan spectrum disorders||11||RASopathies||6 (21)||11 (30)|
|Laboratory 7||Noonan spectrum disorders||16||RASopathies||6 (21)||11 (30)|