Comprehensive genetic testing for female and male infertility using next-generation sequencing

Samples of genomic DNA with previously identified variants in the genes interrogated by the sequencing panel were obtained as purified DNA from the Human Genetic Cell Repository (National Institute of General Medical Sciences) at the Coriell Institute for Medical Research (Camden, NJ) or the Indiana Biobank (Indianapolis, IN) or extracted from saliva samples collected by DNAsimple (Philadelphia, PA). All samples submitted to our diagnostics laboratory as saliva were manually extracted using the Qiagen QIAamp Mini Kits following the manufacturer’s instructions. DNA was quantified using Quant-iT Picogreen dsDNA Assay Kit (Life Sciences) and a Varioskan LUX (Thermo Scientific). Supplementary Table S1 describes all samples used in this study, with biorepository numbers, as well as their previous characterization results and rationale for inclusion. All samples presented here are de-identified (HHS 45 CFR part 46.101(b)(4)). IRB approval to handle de-identified samples was obtained through ASPIRE (Santee, CA), protocol IRB-R-003.

A targeted next-generation sequencing panel consisting of 87 genes related to infertility disorders was created (Supplementary Table S2). Genes relevant to disease phenotype were included based on relationships described in Online Mendelian Inheritance in Man (OMIM), Human Phenotype Ontology (HPO), GeneReviews, and primary literature. The panel included all coding exons, splice sites, promoter regions, 5′ untranslated regions (UTRs), and 3′ UTRs for each of the genes. Clinically relevant noncoding (intronic) regions that contained previously described pathogenic variants, as reported in ClinVar NIH database, were also included. The total size of the gene panel is 1,444,982 bp. In addition, selected regions from the Y chromosome (928,649 bp) were included to allow for the detection of Y chromosome microdeletions. Finally, the panel also includes 188 genes that we considered “research” genes, with links to infertility and which we wanted to follow for research purposes (Supplementary Table S3).

All laboratory procedures were performed in a Clinical Laboratory Improvement Amendments (CLIA) laboratory. DNA samples were prepared for sequencing using HyperPlus Library Preparation Kit (Roche, Indianapolis, IN) and sequenced on a NextSeq500 (Illumina, San Diego, CA), following the manufacturer’s instructions. A detailed sequencing protocol is provided in Supplementary Methods section.

All bioinformatics algorithms were implemented within the Elements^™ platform (Phosphorus, New York, NY). FASTQ files were produced from each sequencing run and processed using the germline calling pipeline (version 2.03.01.30066) in DRAGEN (Edico Genome, San Diego, CA). Variants identified by NGS were confirmed by an orthogonal method (microarrays or Sanger sequencing). After confirmation, each variant was classified as pathogenic, likely pathogenic, variant of unknown significance (VUS), likely benign, or benign, following the American College of Medical Genetics (ACMG) guidelines [19].

A custom Affymetrix Axiom array was used to confirm all the reportable variants including SNVs, indels, CNVs, Y chromosome microdeletions, and sex chromosome aneuploidies. Microarrays were processed following the manufacturer’s instructions. To confirm variants not included in the microarray, Sanger sequencing was used. Orthogonal analysis was also performed for cases when there was discrepancy between expected and received NGS results. If both NGS and confirmatory results agreed, the results were counted as concordant. Detailed protocols are described in the Supplementary Methods section.

FMR1 testing was performed with 120 ng genomic DNA per sample and the AmplideX PCR CE FMR1 kit (Asuragen, Austin, TX), following the manufacturer’s instructions. PCR products were separated in a 3500XL capillary electrophoresis system (Applied Biosystems, Foster City, CA) using conditions described in the FMR1 kit manual.

To maximize the clinical impact of our comprehensive infertility genetic test, we focused only on genes with demonstrated impact on infertility phenotype. Genes were classified as “diagnostic” when variants in them were reported to cause infertility across multiple populations, as supported by multiple publications from different laboratories, demonstrating a direct relationship with infertility. Genes were classified as “informative” when variants in them were reported to be associated with infertility, but the causality link has not been unequivocally established. For male infertility, the panel includes Y chromosome microdeletions, CFTR mutations, and sex chromosome aneuploidies [6]. For female infertility, the panel includes sex chromosome aneuploidies and genes in which variants have been associated with recurrent pregnancy loss caused by thrombophilia, primary ovarian insufficiency, polycystic ovary syndrome, and ovarian hyperstimulation syndrome. The gene list, as well as the rationale used for their selection, is shown in Fig. 1 and Supplementary Table S2. Figure 2 describes the laboratory and analysis workflow.

To validate panel performance and determine analytical sensitivity, specificity, and accuracy, we sequenced 24 samples from the 1000 Genomes (1000G) project [20] for which the location of SNVs and indels are known (Supplementary Table S1). Representative sequencing quality control statistics are shown in Supplementary Table S4. NGS results from these validation samples were compared to known 1000G variants (Table 1). Microarray analysis and/or Sanger sequencing was performed to assess the discrepancies of SNVs/indels between results produced by the NGS panel and previously known 1000G data. Analytical sensitivity of the test was > 99% for SNVs and > 91% for indels, and specificity was > 99% for both SNVs and indels. Final accuracy for SNVs and indels was 99.98 and 99.42%, respectively.

To determine clinical sensitivity in the detection of CNVs, sex chromosome aneuploidies and Y chromosome microdeletions, we used 34 samples with 38 known variants (Supplementary Table S1). The set of variants previously known for these samples is limited, so specificity could not be assessed. The analysis correctly detected 3/3 CNVs, 19/19 sex chromosome aneuploidies, and 15/16 Y chromosome microdeletions (Supplementary Tables S5 and S6, Table 1 and Fig. 2). In three cases (NA20435, NA18333, and NA22031), Y chromosome microdeletions of smaller size were found by NGS when compared to previously known data (Supplementary Table S6); however, microarray analysis confirmed the previously reported microdeletion size. In one case (NA20434), NGS data located the microdeletion 1.17 Mbp downstream of the previously known location, and microarray analysis confirmed NGS result. Sample NA12662 has a sex chromosome aneuploidy (duplication of X chromosome) and a Y chromosome microdeletion ((22769319-27097245)x0). The sex chromosome aneuploidy was identified, but the Y chromosome microdeletion was missed. Thus, clinical sensitivity for sex chromosome aneuploidies and CNVs is 100% (22/22), and for Y chromosome microdeletions, it is 93.75% (15/16).

Next, 11 DNA samples with 17 known SNVs/indels were processed (Table 2). Known variants were confirmed in 16/17 cases. Sample NA02795 was reported with a pathogenic variant GALT c.130G>A; p.Val44Met. However, this variant was not identified by NGS and Sanger sequencing did not detect it either; therefore, based on NGS and Sanger sequencing concordance, we did not count this case against the sensitivity. Thus, clinical sensitivity for SNVs/indels by our assay was 100%.

Varied lengths of a thymidine (T)-tract (5, 7, or 9T) are found in front of the splice-acceptor site of intron 8 of the CFTR gene in males with congenital bilateral absence of the vas deferens [21]. The length correlates with the efficiency of exon 9 splicing. This polymorphism is detected clinically by allele-specific multiplex PCR [22]. In order to determine sensitivity of the NGS panel, we analyzed 72 samples from the 1000G project. When 1000G data was used as reference, the NGS panel correctly detected the IVS8 alleles in 67/72 cases, and for the 5 discrepant cases, Sanger sequencing demonstrated that the NGS panel call was the correct one (Supplementary Table S7). Therefore, our assay had 100% sensitivity to detect the CFTR IVS8-5T allele.

CGG triplet expansions are known to cause fragile X syndrome, and alleles in the pre-mutation range have been associated with increased risk for premature ovarian failure [23]. Current NGS protocols based on hybridization enrichment for targeted panels do not accurately quantify the number of repeats. Therefore, to complement the NGS panel with FMR1 analysis, we used an already available protocol based on PCR amplification and capillary electrophoresis detection of CGG repeats [24]. Test sensitivity was determined by analysis of 26 samples harboring different CGG repeat expansions (Supplementary Table S8). We correctly identified all alleles previously classified as “mutation” (> 200 CGG repeats), “pre-mutation” (55 to 199 CGG repeats), and “normal” (< 45 CGG repeats). However, the assay misclassified one allele in the “intermediate” range (between 45 and 54 CGG repeats) as “pre-mutation” (NA20236, 53 vs 55 CGG repeats, Supplementary Table S8). Given that the rate of expansions of intermediate alleles is not well understood, testing at-risk relatives of individuals with an intermediate allele may determine the stability of the allele in the family [25].

Currently, after semen analysis, in cases where severe abnormal semen analysis is observed, genetic testing for variants associated with male infertility is performed with multiple assays. For example, CFTR IVS8-5T is detected by allele-specific multiplex PCR [22], Y chromosome microdeletions are detected by PCR of sequence-tagged sites [26], and sex chromosome aneuploidies are detected by traditional karyotyping or microarrays [27]. NGS can detect all these variants using a single test. In order to determine the economic impact, we performed a cost analysis for the male fertility panel, using average pricing available from reference laboratories performing these assays (Table 3 and Supplementary Table S9). Overall, the savings to providers could be as much as $2723 per case, $3322 if performed by multiple assays vs $599 if performed by NGS, which translates in savings of > 550%. Furthermore, NGS testing has a shorter turnaround time because there is no need to do reflex testing. The only exception may be cases with complex chromosomal rearrangements and mosaics or translocations/inversions that do not result in gain/loss of DNA and are, therefore, not captured by the NGS panel.