Disorders of sex development: insights from targeted gene sequencing of a large international patient cohort

Abstract

Background: Disorders of sex development (DSD) are congenital conditions in which chromosomal, gonadal, or phenotypic sex is atypical. Clinical management of DSD is often difficult and currently only 13% of patients receive an accurate clinical genetic diagnosis. To address this we have developed a massively parallel sequencing targeted DSD gene panel which allows us to sequence all 64 known diagnostic DSD genes and candidate genes simultaneously.

Results: We analyzed DNA from the largest reported international cohort of patients with DSD (278 patients with 46,XY DSD and 48 with 46,XX DSD). Our targeted gene panel compares favorably with other sequencing platforms. We found a total of 28 diagnostic genes that are implicated in DSD, highlighting the genetic spectrum of this disorder. Sequencing revealed 93 previously unreported DSD gene variants. Overall, we identified a likely genetic diagnosis in 43% of patients with 46,XY DSD. In patients with 46,XY disorders of androgen synthesis and action the genetic diagnosis rate reached 60%. Surprisingly, little difference in diagnostic rate was observed between singletons and trios. In many cases our findings are informative as to the likely cause of the DSD, which will facilitate clinical management.

Conclusions: Our massively parallel sequencing targeted DSD gene panel represents an economical means of improving the genetic diagnostic capability for patients affected by DSD. Implementation of this panel in a large cohort of patients has expanded our understanding of the underlying genetic etiology of DSD. The inclusion of research candidate genes also provides an invaluable resource for future identification of novel genes.

Keywords: Cohort; Disorders of sex development; Genetic diagnosis; Gonad; MPS; Massively parallel sequencing; Mutation; Ovaries; Ovotestes; Targeted gene panel; Testis; Variant.

Fig. 1

Coverage and variant properties of the panel and patient cohort. a The cumulative distribution read coverage across the targeted regions of the HaloPlex panel for 16 evaluation samples. The vertical axis shows the percentage of bases covered with at least the level of coverage specified by the horizontal axis. Although the median coverage is acceptable for all samples, it is notable that 10% of bases are covered at less than 25×, while another 10% of bases are covered at more than 280×. b Coverage depth uniformity of HaloPlex compared to whole genome sequencing (WGS). The cumulative coverage distribution is shown for three samples sequenced by both technologies. HaloPlex is notably less uniform, having a flatter distribution than WGS. c Receiver-operator characteristic (ROC) curve showing sensitivity versus false positive rate (1 − precision) for detecting single nucleotide variants and INDELs smaller than 10 bp, compared to high confidence calls for samples NA12878 and NA12877. Call sets were obtained from the Illumina Platinum Genomes project. A sensitivity of 97 and 95%, respectively, is achieved for a false positive rate smaller than approximately 2% in both cases

Fig. 2

Protein changing variants seen per kilobase sequenced for diagnostic genes. A lower number of variants per kilobase sequenced suggests a higher intolerance to protein altering mutations for the gene, but may also be affected by lower ascertainment in regions that are difficult to sequence. Diagnostic DSD genes are graphed alphabetically; differing colors are used only for clarity. A small number of genes are excluded because they experienced artificially low variant counts due to technical reasons, including poor sequencing performance (CYP21A2, CDKN1C, LHX3), omission from sequencing in some samples (CYB5A), or difficulties in annotating variants accurately (SRD5A2)

Fig. 3

Genetic diagnosis of the DSD cohort. a Proportion of 46,XY DSD patients with a curated variant in a known DSD gene. In 46,XY DSD patients (278 patients), a DSD variant was identified in 57% (159 patients) of the study cohort. This was made up of 76 pathogenic variants and 42 likely pathogenic variants, resulting in a diagnostic rate of 43%. A total of 41 VUS were also found. b In the 46,XX DSD patient cohort (48), only 19% (9) were found to have a variant in a DSD gene, most of which were SRY translocations (8). This resulted in a diagnostic rate of 17%. c Distribution of curated variants in DSD genes among the 46,XY DSD phenotypic categories. Variants in a diagnostic DSD gene found to be pathogenic or likely pathogenic are considered to be a genetic diagnosis. The diagnostic outcome for each of the phenotypic categories is indicated. Disorders of gonadal (testicular) development patients had a total of 21 out of 52 patients with a pathogenic or likely pathogenic DSD variant (40%) and only two patients with a VUS (4%). Of the patients with a suspected disorder of androgen synthesis and action, 22 patients of 37 had a diagnostic variant (60%) and four had a VUS (10%). Of patients in the 46,XY other category (including hypospadias), just 18 out of 56 had a diagnostic variant (32%), with 11 patients having a VUS (19%). Finally, in the broad category 46,XY DSD unknown, which includes 133 patients, 57 had a pathogenic or likely pathogenic (43%) variant, while 24 patients had a VUS (18%). In cases where a patient had variants in multiple genes, the variant with the highest classification (pathogenic > likely pathogenic > VUS) was taken into consideration for this chart

Fig. 4

Reportable DSD variants identified in patients with 46,XY DSD. Variants were identified in 28 of a total of 64 diagnostic DSD genes. The number of previously reported (as disease causing) and unreported changes found in each diagnostic DSD gene as well as the type of change identified (missense or null variants) are shown (all variants can be found in Additional file 1: Table S1). The total number of variants is shown for each gene. The clinical relevance of each variant was checked in ClinVar, HMGD, and OMIM databases and for prior publication in PubMed

Fig. 5

Analysis of the 46,XY DSD cohort: singletons versus trios and patients with a DSD of unknown origin. a, b Singleton or trio analysis of patients with 46,XY DSD. Individuals with 46,XY DSD were either analyzed as a singletons (215 patients) or b trios/duos. The proportion of patients with a DSD variant was higher for singletons than for trios: 68% (128 patients) versus 50% (31 patients). Singletons and trios had a similar genetic diagnostic rate (pathogenic or likely pathogenic variant) at 43 and 41%, respectively. A higher proportion of singletons had a DSD variant classified as VUS (17% of all variants in singleton) compared to trios (8% in trio analyses). c, d Gene variants reveal biological basis of 46,XY DSD. Only limited clinical information was often available for 133 origin unknown patients (c) and 46 hypospadias patients (d). Based on their curated DSD variants, these patients have been assessed on the categories of DSD gene function. In cases where a patient had variants in multiple genes, the variant with the highest classification (pathogenic > likely pathogenic > VUS) was taken into consideration. Variants annotated VUS were also included in this analysis