Molecular Diagnosis of 46,XY Disorders of Sex Development: An Efficient Initial Molecular Analysis Using a Custom-Designed Targeted Gene Panel in a Single-Center Study

Abstract

Background

The management of 46,XY disorders of sex development (DSD) is challenging due to genetic heterogeneity and phenotypic variability. This study aimed to characterize the clinical and genetic findings in patients with 46,XY DSD, using a targeted next-generation sequencing (NGS) panel for molecular evaluation.

Methods

A targeted DSD gene panel covering 31 genes was applied in 112 patients with nonsyndromic 46,XY DSD. Forty-six patients had previously tested negative for AR and SRD5A2 by Sanger sequencing. Patients were clinically categorized into disorders of gonadal development, androgen synthesis or action. Variant classification was performed according to the ACMG criteria.

Results

Among the 38 variants detected, 32 were pathogenic or likely pathogenic. Nineteen (50%) variants were novel. A molecular diagnosis was established in 31 (27.7%) patients and inclusion of previously diagnosed cases would have increased the overall diagnostic yield to 43.8%. The HSD17B3 variants were the most common, followed by NR5A1 and LHCGR. In 8 patients, the genetic findings led to reclassification of their clinical diagnosis, particularly in those initially suspected to have a disorder of androgen action.

Conclusion

NGS is a valuable diagnostic tool in the evaluation of 46,XY DSD, offering improved diagnostic yield. For patients without molecular diagnosis, more comprehensive genomic analyses, including noncoding regions, are required.

Introduction

The differential diagnosis of patients with 46,XY disorders of sex development (DSD) is challenging due to the underlying diverse genetic heterogeneity, wide phenotypic spectrum, and poor genotype-phenotype correlation. A precise etiological diagnosis is important for gender assignment, risk assessment of gonadal malignancy, and genetic counseling [1].

The incorporation of genetic tests into clinic, hormonal, and imaging investigations increased the diagnostic accuracy of 46,XY DSD. Prior to the advent of next-generation sequencing (NGS), determining the genetic etiology of 46,XY DSD was feasible in only a limited number of cases [2]. Targeted NGS panels and exome sequencing have increased the molecular detection rate of variants in 46,XY DSD due to their capacity to simultaneously analyze a large number of genes. The studies using NGS for molecular diagnosis have achieved a molecular diagnosis in 30.8%–64.3% of patients with 46,XY DSD [3–9]. In our study, we aim to identify the genetic etiology of patients with 46,XY DSD using a targeted NGS panel that includes 31 known human DSD-associated genes.

Materials and Methods

Selection of Patients

We enrolled 112 patients who were seen at our DSD clinic from 1997 to 2024. All patients have a nonsyndromic DSD and 46,XY karyotype. Before admission to molecular genetic analysis, structural chromosomal abnormalities were excluded. In our laboratory, molecular analysis of patients with 46,XY DSD was performed using Sanger sequencing of the AR and SRD5A2 genes until 2017. In 2017, a targeted NGS panel covering 31 DSD-associated genes was implemented into routine diagnostic practice. Of the 112 patients in our cohort, 46 who tested negative for pathogenic variants in AR and SRD5A2 were included for further analysis using the targeted NGS panel. Before molecular analysis, based on clinical, hormonal, and imaging investigations, we classified patients into three categories: disorders of gonadal development (DGD); disorders of androgen and anti-müllerian hormone synthesis or action (disorder of androgen synthesis [DAS], persistant müllerian duct syndrome, disorders in androgen action [DAA]) [1].

Selection of Candidate Genes

A targeted DSD panel was designed to include 31 genes, related to disorder of gonadal development and differentiation, disorder of androgen and anti-müllerian hormone synthesis or action, by searching online databases and the available literature, including PubMed (http://www.ncbi.nim. nih.gov/pubmed), Online Mendelian Inheritance in Man (OMIM) (http://www.ncbi.nlm.nih.gov/omim), and Human Gene Mutation Database Professional version 2017.2 (HGMD^® 2017.2) (https://my.qiagendigitalinsights.com/) [10].

Sample Preparation, Sequencing, and Analysis

Genomic DNA was extracted from peripheral blood leukocytes using an EZ1 DNA Isolation Kit for Mammalian Blood (Qiagen, Hilden, Germany) following the manufacturer’s instructions. The concentration of all genomic DNA samples was assessed via spectrophotometric measurements (Nanodrop 2000c, Thermo Scientific). For each sample, dilution procedures were performed using MilliQ (18 MΩ) water to achieve a DNA concentration of 1.6–2.0 ng/μL. The final DNA concentrations were adjusted to fall within the required range (1.6–2.0 ng/μL) for next-generation sequencing. Before measuring the diluted DNA samples, a calibration measurement was conducted using the Qubit dsDNA HS Assay Kit.

The custom panel containing the selected genes was designed using Ion Ampliseq designer (Thermo Scientific Inc). Library preparation was performed using a customized Ion Ampliseq Library Kit 2.0 (Life Technologies) and resulted in 96.7% coverage of genes in 315 amplicons. The genes included were as follows: ATF3 (NM_001674.4); AR (NM_000044.6); AMHR2 (NM_020547.3); BNC2 (NM_017637.6); BMP4 (NM_001202.6); CYP11A1 (NM_000781.3); CBX2 (NM_005189.3); CYB5A (NM_001914.4); DMRT1 (NM_021951.3); DHH (NM_021044.4); GATA4 (NM_002052.5); HHAT (NM_018194.6); HSD17B3 (NM_000197.2); HOXA4 (NM_002141.5); HOXB4 (NM_024015.5); HOXB6 (NM_018952.5); LHCGR (NM_000233.4); MAP3K1 (NM_005921.2); MAMLD1 (NM_005491.5); NR5A1 (NM_004959.5); NR0B1 (NM_000475.3); POR (NM_001395413.1); RSPO1 (NM_001038633.4); SRD5A2 (NM_000348.4); STAR (NM_000349.3); SOX9 (NM_000346.4); SOX3 (NM_005634.3); SRY (NM_003140.3); WNT4 (NM_030761.5); WT1 (NM_024426.6); ZFPM2 (NM_012082.4). The amplicon libraries were sequenced on the Ion PGM™ Hi-Q™ View Sequencing Kit (Thermo Fisher Inc.). Ion Torrent PGM (Thermo Fisher Inc.) and Ion S5 (Thermo Fisher Inc.) were used as sequencing platforms. The coding exonic regions of the reference transcripts of these genes, along with 10 bp intronic regions (exon-intron boundaries) and any other noncoding 5′UTR and 3′UTR, were determined as targets. After sequencing, the low-coverage regions (<20) of the SRD5A2, AR, WT1, and MAP3K1 genes were analyzed by Sanger sequencing. The quality assessments of the BAM files transferred to the Torrent Suite Server Software (Thermo Fisher Inc.) were conducted, focusing on regions that were not covered or had insufficient read depth. For analysis, the data were uploaded to the cloud-based Ion Reporter (v.5.6) variant analysis, annotation, and reporting program using the Ion Reporter Uploader. The reads obtained from the raw data were mapped to the GRCh38/hg38 human reference sequence. Sanger sequencing was employed to individually assess targeted regions that were missed by the AmpliSeq primer design or were low in depth coverage. Additionally, all identified variants were validated using custom-designed primers through Sanger sequencing.

Bioinformatics analysis was performed using an in-house pipeline. Sequences with 20X or more read depths in variant analysis were evaluated and retained. Subsequently, variants were further filtered to include only variants with minor allele frequency <0.01 based on gnomAD (v3.1.2), 1,000 Genome Project (https://www.1000genomes.org/) or UCSC (https://genome.ucsc.edu/), RefSeq (https://www.ncbi.nlm.nih.gov/refseq/).

The variants were classified according to the standards and guidelines of the American College of Medical Genetics and Genomics (ACMG) using the Franklin platform [11, 12]. We checked whether the variants had been reported before using dbSNP (http://www.ncbi.nlm.nih.gov/snp) and disease databases, such as HGMD (http://www.hgmd.cf.ac. uk/ac/index.php) and ClinVar (http://www.ncbi.nlm.nih. gov/clinvar). Only the variants classified as pathogenic, likely pathogenic, and variants of uncertain clinical significance (VUS) were evaluated.

The copy number variation analysis was performed in panel using Ion Reporter v5.2 software. In all cases where phenotype-associated variants were not detected, multiplex ligation-dependent probe amplification (MLPA) analysis was performed by using the MLPA probemix SALSA P334-A2 Gonadal Development Disorder kit (MRC Holland, Amsterdam, Holland), which encompasses DMRT1, CYP17A1, SRD5A2, and HSD17B3 genes and SALSA MLPA KIT P185-B1 Intersex kit, which analyses SRY, WNT4, DAX1, SOX9, and NR5A1 genes. The segregation analysis was performed only for family members who consented to novel missense alterations suspected to be causative.

Results

In our cohort, before molecular analysis, 36 out of 112 patients (32.1%) were classified clinically as having a DGD, 30 out of 112 (26.8%) as having a DAS, and 46 out of 112 (41.1%) as having a DAA. The NGS study covered 98.29% of the target regions. The remaining uncovered regions were examined using Sanger sequencing. The average sequence read depth was 511.7 (340–863).

We identified 38 variants in 14 different genes associated with 46,XY DSD, including HSD17B3, LHGCR, AR, SRD5A2, POR, AMHR2, NR5A1, DHH, MAP3K1, WT1, ZFPM2, GATA4, BNC2, and WNT4. Among the 38 identified variants, 32 were classified as pathogenic or likely pathogenic, while six were categorized as VUS. Nineteen of the 38 (50%) variants were novel. Missense variants were the most frequently observed, accounting for 60.5% (23/38), followed by splicing variants [18.5% (7/38)], nonsense variants [10.5% (4/38)], and indel variants [10.5% (4/38)]. Although nearly half of the variants associated with DGD were novel (46.2%), the majority of those identified in DAS and DAA had been previously reported (63.2%). Homozygous variants were the most frequently observed (48.4%), followed by heterozygous (32.2%) and compound heterozygous variants (9.7%). Hemizygous variants were also identified (9.7%). A molecular diagnosis was obtained in 31 out of 112 (27.7%) patients according to the ACMG criteria, including 25 sporadic cases and 6 index patients from six unrelated families. The genetic results of the patients are summarized in Tables 1 and 2.

Table 1.

Genetic results of 46,XY patients with molecular genetic testing

Index	Gene	Zygosity	cDNA/protein	dbSNP/HGMD	ACGM classification (Franklin)	ClinVar classification/ID number	Previously reported
1	HSD17B3	Het	c.139A>G/p.(Met47Val)	rs191153391/CM1616125	P (PM2, PP2, PP5, PM3, PP1)	P/LP (2136797)	Yes [4]
		Het	c.704T>C/p.(Met235Thr)	rs776757787/–	P (PM2, PP3, PM5, PP2, PM3, PP1, PP4)	–	No
2	HSD17B3	Hom	c.607-1G>A/p.(?)	rs730880305/CS941501	P (PM3, PVS1, PM2, PP5)	P (4878)	Yes [13]
3	HSD17B3	Hom	c.182G>A/p.(Gly61Glu)	–/–	P (PP3, PM2, PP2, PP1)	–	No
4	HSD17B3	Hom	c.167C>T/p.(Ala56Val)	–/CM239816	P (PM2, PM5, PP3, PP2, PP4, PP1)	–	Yes [14]
5	HSD17B3	Hom	c.277+4A>T/p.(?)	rs201115371/CS002140	P (PM3, PP1, PM2, PP3, PP5)	P (208587)	Yes [15]
6	HSD17B3	Hom	c.277G>A/p.(Glu93Lys)	rs753360928/CM173149	P (PM2, PP3, PP2, PP5, PP1)	P (2577238)	Yes [16]
7	HSD17B3	Hom	c.639_640insA/p.(Glu214Argfs*4)	–/–	P (PVS1, PM2, PP1)	–	No
8	HSD17B3	Het	c.133C>T/p.(Arg45Trp)	rs139084702/CM1616127	P (PM2, PP2, PP5, PM3, PP1)	P/VUS/LB (708567)	Yes [4]
		Het	c.704T>C/p.(Met235Thr)	rs776757787/-	P (PM2, PP3, PM5, PM3, PP1, PP4)	–	No
9	NR5A1	Het	c.151G>T/p.(Glu51*)	rs775441984/CM1417625	P (PS4, PVS1, PM2)	P (216975)	Yes [17]
10	NR5A1	Het	c.218G>C/p.(Cyst3Ser)	–/CM246864	LP (PM1, PP2, PM2, PM5, PP3)	–	Yes [18]
11	NR5A1	Het	c.251G>A/p.(Arg84His)	rs375469069/CM080459	P (PS4, PS3, PP3, PM2, PM5)	P/LP (641278)	Yes [19]
12	NR5A1	Het	c.247G>T/p.(Val83Leu)	–/CM1817517	P (PM2, PM1, PP2, PP3, PP1, PP4)	–	Yes [18, 20]
13	NR5A1	Het	c.44T>C/p.(Val15Ala)	–	P (PS4, PM1, PP2, PM2, PM5, PP3, PP5)	LP (1691292)	No
14	NR5A1	Het	c.938_939delGCins TT/p.(Arg313Leu)	–	P (PS1, PM5, PM2, PM1, PP2)	–	No
15	NR5A1	Het	c.163T>G/p.(Cys55Gly)	–/CM2020899	LP(PP3, PM2, PM1, PP2	–	Yes [21]
16	LHCGR	Hom	c.1435C>T/p.(Arg479*)	rs757225917/CM163728	P (PS4, PVS1, PM2, PP5)	P (2203074)	Yes [22]
17	LHCGR	Hom	c.537-1G>T/p.(?)	–/CS217708	P (PS4, PVS1, PM2, PP5)	P (996742)	Yes [23]
18	LHCGR	Hom	c.203C>T/p.(Ala68Val)	rs944199288	LP (PM2, PP3, PM3, PP1, PP4)	–	No
19	LHCGR	Hom	c.233+5G>A/p.(?)	–	LP (PM2, PP3, PP1, PM3, PP4)	–	No
20	DHH	Het	c.71G>C/p.(Gly24Ala)	rs778014372	LP (PM2, PP3, PP1, PM3)	–	No
		Het	c.1063C>T/p.(Arg355Cys)	–	LP (PM2, PM3, PP1, PP4)	VUS (3501426)	No
21	DHH	Hom	c.1146G>A/p.(Trp382*)	–	LP (PM2, PVS1, PP1, PM3)	–	No
22	WT1	Het	c.1400G>T/p.(Arg462Leu)	–/CM041493	P (PM1, PP2, PM2, PM5, PP4)	Tier I (strong) (3024186)	Yes [24]
23	WT1	Het	c.1447+4C>T/p.(?)	rs748303899/CS971933	P (PS3, PS2, PM2, PP3, PP5, PP4)	P/LP (3500)	Yes [25]
24	AR	Hem	c.2668G>A/p.(Val890Met)	rs886041133/CM000310	P(PS4, PS3, PM1, PP2, PM2, PM5, PP3, PP5)	P (279690)	Yes [26]
25	AR	Hem	c.1740C>A/p.(Cys580*)	–	P (PVS1, PM2, PP4)	–	No
26	AR	Hem	c.2269A>G/p.(Asn757Asp)	–/CM232323	P (PM1, PP2, PM2, PM5, PP3, PP4)	–	Yes [27]
27	GATA4	Het	c.685A>G/p.(Arg229Gly)	–	LP (PM2, PP3, PP4)	–	No
28	SRD5A2	Hom	c.164T>A/p.(Leu55Gln)	rs121434245/CM920631	P (PM3, PP1, PM2, PM5, PM1, PP5)	P/LP (3339)	Yes [28]
29	SRD5A2	Hom	c.164T>A/p.(Leu55Gln)	rs121434245/CM920631	P (PM3, PP1, PM2, PM5, PM1, PP5)	P/LP (3339)	Yes [28]
30	AMHR2	Hom	c.233-1G>A/p.(?)	rs1360862246/CS185575	P (PVS1, PM2, PM3, PP4)	–	Yes [29]
31	POR	Hom	c.1187_1195dupCCTCGGAGC/p.(Pro396_Gln398dup)	–	P (PM2, PM4, PP1, PP4, PM3)	VUS (3682460)	No

Hem, hemizygous; Het, heterozygous; Hom, homozygous; P, pathogenic; LP, likely pathogenic; PVS, pathogenic very strong; PS, pathogenic strong; PP, pathogenic supporting; PM, pathogenic moderate; VUS, variant of uncertain significance.

Table 2.

Genetic results of patients with variants of uncertain significance (VUS)

Index	Gene	Zygosity	cDNA/protein	dbSNP	ACGM classification (Franklin)	Previously reported
32	MAP3K1	Het	c.1892T>C/p.(Val631Ala)	rs747821377	VUS(PM2)	No
33	ZFPM2	Het	c.286G>T/p.(Asp96Tyr)	rs772042320	VUS(PM2)	No
34	ZFPM2	Het	c.3002A>G/p.(Asn1001Ser)	2013	VUS(PM2)	No
35	BNC2	Het	c.2750_2751delTGinsCT/p.(Leu917Ser)	–	VUS(PM2)	No
36	WNT4	Het	c.86C>G/p.(Ala29Gly)	–	VUS(PM2)	No
37	GATA4	Het	c.1146+4C>T/p.(?)	rs200334160	VUS(PM2)	No

Het, heterozygous; PM, pathogenic moderate.

Clinical classification was consistent with the molecular diagnosis in 23 out of 31 patients. Most patients with molecularly confirmed DAS (14 out of 15) were accurately classified based on clinical and biochemical assessments. Molecular analysis led to the reclassification of 6 patients who had been clinically classified as DAA, revealing variants in genes associated with DGD [including DHH (n = 1), NR5A1 (n = 4)], and persistant müllerian duct syndrome [AMHR2 (n = 1)]. The detailed clinical data of patients harboring variants are presented in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000550371).

The HSD17B3 gene variants were the most common in our cohort. Nine pathogenic/likely pathogenic variants, including three novel variants, were detected in five sporadic and eight cases (3 index patients) from three unrelated families. Except for 1 patient, all patients were from unrelated consanguineous families. Variable phenotypes were observed, including 6 patients who presented with female external genitalia with or without clitoromegaly and inguinal masses and 2 patients with hypospadias and micropenis. The sister and cousin of a female patient, with a homozygous novel missense variant (c.182G>A/p.(Gly61Glu)), were 46,XY and had similar clinical findings. A female patient and her two sisters, having 46,XY harbored a novel homozygous frameshift variant, c.639_640 insA/p.(Glu214Argfs*4), that was predicted to lead to a truncated protein. Two unrelated male patients with hypospadias and micropenis had compound heterozygous missense variants [same novel variant in exon 10, c.704T>C/p.(Met235Thr)] and two different previously reported variants in exon 1 (Table 1).

In our cohort, NR5A1 gene variants were the second highest number of variants. Seven heterozygous pathogenic or likely pathogenic variants were identified, including one nonsense variant, five missense variants, and one indel variant. Two of these variants had not been previously reported. All patients, except one, were raised as males. A female patient presented with clitoromegaly at 4 years 10 months of age, has a known pathogenic nonsense variant, c.151G>T/p.(Glu51*). Her healthy father had this variant heterozygously as 22% mosaic in a peripheral blood sample. A male patient with micropenis and hypospadias has a novel indel variant, c.938_939delGCinsTT/p.(Arg313Leu) in the NR5A1 gene, classified as pathogenic according to the ACMG criteria. His father was heterozygous for this variant. A male patient who carried a novel likely pathogenic missense variant, c.44T>C/p.(Val15Ala), in the NR5A1 gene presented with ambiguous genitalia (Table 1). None of the patients had adrenal insufficiency.

The LHCGR gene variants were the third highest number of variants, including two novel and two known variants (Table 1). All patients except one had female external genitalia with or without clitoromegaly and primary amenorrhea. A female patient carrying a novel homozygous intronic variant, c.233+5G>A, in the LHCGR gene was admitted with a complaint of rapid growth at the age of 12 years, classified as likely pathogenic according to the ACMG criteria. A novel missense likely pathogenic variant, c.203C>T/p.(Ala68Val), in the LHCGR gene was detected in a male patient who presented with micropenis and glandular hypospadias at the neonatal period. The parents of both patients were found to be heterozygous carriers of these variants.

Four novel variants were identified in the DHH gene, including three missenses and one nonsense, classified as pathogenic or likely pathogenic according to the ACMG criteria. Phenotypic variability was observed among patients with DHH gene variants, even for the same variant. A novel homozygous nonsense variant, c.1146G>A/p. (Trp382*), predicted to result in truncated protein was identified in three siblings. The patients exhibited variable phenotypes, ranging from female external genitalia to penoscrotal hypospadias; two were raised as male and one as female. In the second family, two siblings were found to carry compound heterozygous novel missense variants in the DHH gene, c.[71G>C];[1063C>T]/p.[Gly24Ala];[Arg355Cys]. The younger sibling, initially assigned female at birth, was referred to the clinic at 19 days of age with ambiguous genitalia. The older brother, who had undergone hypospadias repair, carried the same variants. Neuropathy has not been detected in any of our patients. In both families, the healthy parents, who were first cousins, were heterozygous carriers of the respective variants (Table 1).

Three patients who were not screened for the AR gene variant previously were found to carrypathogenic/likely pathogenic variants in the AR gene, including two known missense variants and one novel nonsense variant. A female patient with primary amenorrhea was identified with a novel likely pathogenic nonsense variant in the AR gene, c.1740C>A/p.(Cys580*).

A novel likely pathogenic heterozygous variant was identified in the GATA4 gene (Table 1). A male patient, who presented with ambiguous genitalia and bilateral atrophic testicular tissue at 1.5 months of age, was found to have a de novo missense variant, c.685A>G/p.(Arg229Gly), in the GATA4 gene. He also had congenital heart disease, including atrial and ventricular septal defects. His brother had previously died due to congenital heart disease.

The same known homozygous variant in the SRD5A2 gene, c.164T>A/p.(Leu55Gln) was identified in 2 unrelated patients who exhibited distinct clinical phenotypes. The first patient presented at birth with micropenis and penoscrotal hypospadias and was reared as male. The second patient presented at 14 years of age with primary amenorrhea and absence of secondary sexual characteristics (Table 1).

Novel homozygous pathogenic variant in the POR gene, c.1187_1195dupCCTCGGAGC/p.(Pro396_Gln398dup), was detected in a female patient presenting with ambiguous genitalia, bilateral inguinal testes and the absence of a uterus. She exhibited adrenal insufficiency, and no phenotypic features suggestive of Antley-Bixler syndrome were identified. The mother did not experience virilization during pregnancy. The consanguineous parents were found to be heterozygous carriers of the variant.

In our cohort, six novel VUS variants were identified in five genes, including two in ZFPM2 and one each in the MAP3K1, WNT4, GATA4, and BCN2 (Table 2). Two missense variants in the ZFPM2 gene were identified in one sporadic case and two affected siblings. The first patient, who presented in the neonatal period with micropenis, hypospadias, and a left multicystic dysplastic kidney, carried a novel heterozygous missense variant, c.3002A>G/p.(Asn1001Ser). Both his father and brother, who also had hypospadias, were found to carry the same variant. The second patient, who had micropenis and glandular hypospadias, was found to carry a missense variant in the ZFPM2 gene, c.286G>T/p.(Asp96Tyr). Echocardiographic evaluation revealed no cardiac abnormalities in these patients. Neither copy number variation analysis nor MLPA analysis detected any pathogenic deletions or duplications in the targeted genes.

Discussion

The molecular diagnostic yield of the targeted DSD panel covering 31 genes was 27.7% in our cohort. We identified 32 rare pathogenic/likely pathogenic variants in 10 genes, nearly half of which were previously unreported. An additional six novel variants in five genes were classified as VUS due to insufficient current evidence. Genetic analysis enabled the accurate classification of DSD in 8 patients in our cohort. These findings highlight the critical role of genetic analysis in establishing an accurate diagnosis in terms of management, genetic counseling, risk assessment of gonadal malignancy, fertility, and associated features.

Our study demonstrated a relatively lower diagnostic yield compared to previous reports [3–9, 30–36], which can be partly explained by our study design. In our earlier study, Sanger sequencing of the AR and SRD5A2 genes was performed in a subset of 46,XY DSD patients, identifying pathogenic variants in 24% of cases [37]. For the current study, individuals who had previously tested negative for these genes were included alongside new patients who had not undergone prior testing. Consequently, a substantial portion of our cohort had already been prescreened for AR and SRD5A2, and cases with known pathogenic variants were excluded from the NGS panel analysis. This likely contributed to the lower diagnostic yield observed for these genes. Consequently, we observed a lower diagnostic yield for pathogenic variants in the AR (9.7%) and SRD5A2 (6.5%) genes compared to other studies (Table 3). If previously diagnosed cases had been included, our overall diagnostic yield would have reached 43.8%, aligning more closely with previously reported rates. In other NGS studies for 46,XY DSD containing different numbers of genes from different countries, diagnostic yields have been reported between 30.8 and 64.3% and patients with AR and SRD5A2 genes accounted for the greatest percentage of all patients (Table 3).

Table 3.

Summary of results of previous studies using exome sequencing (ES) and targeted panel for 46,XY disorder of sex development

​	Number of patients with 46,XY	Method	Number of genes in panel	Diagnostic yield, %
				overall	AR	SRD5A2
Hughes et al. [3]	73	Panel	30	34.2	28	16
Eggers et al. [4]	278	Panel	1,031	42.2	23.9	12.8
Buonocore et al. [5]a	52	Panel	180	30.8	–	–
Zhang et al. [6]	70	ES	–	64.3	46.7	24.4
Wang et al. [7]	70	Panel	80	42.9	26.7	23.3
Guo et al. [8]	50	Panel	141	32	12.5	68.8
Yu et al. [9]	87	Panel	30	42.5	37.8	32.4
Globa et al. [30]	75	ES	–	43	32.4	2.9
Xie et al. [31]	74	Panel+ES	108	37.8	10.7	7.1
Tang et al. [32]	178	Panel+ES	2,742	35.9	27.1	35.4
Xu et al. [33]	96	Panel	2,742	46.9	24.4	22.2
Kim et al. [34]	37	Panel	67	35.1	38.5	7.7
Dong et al. [35]	13	Panel	219	46.2	33.3	–
Fan et al. [36]	27	Panel	80	33.3	11.1	33.3
This studyb	112	Panel	31	27.7	9.7	6.5

^aAll patients tested for HSD17B3, AR, and SRD5A2.

^bForty-six patients had previously tested for AR and SRD5A2.

We identified three novel HSD17B3 variants, thereby contributing to expanding the variants in the database of the HSD17B3 gene. The frequency of HSD17B3 variants in our study was markedly higher than some previously reported studies [3, 9, 38]. This increased frequency may be explained by the high rate of consanguineous marriages (36.8%) observed in our population. Although 17β-HSD3 deficiency is a rare autosomal recessive disorder, accounting for approximately 0.4% of 46,XY DSD cases in Europe, its prevalence has been reported to be significantly higher in populations with increased rates of consanguinity [38, 39]. In a cohort of Egyptian patients with 46,XY DSD characterized by a parental consanguinity rate of 70%, HSD17B3 mutations were identified in 50% of cases [39].

In our cohort, NR5A1 variants represented the most frequent cause of 46,XY DGD. The phenotypes observed in our 7 patients carrying NR5A1 variants were variable, ranging from clitoromegaly to hypospadias and micropenis. None of the patients had adrenal insufficiency. The phenotypic spectrum associated with NR5A1 variants is notably broad, and the same pathogenic variants may result in strikingly different phenotypes [18–20]. In one of our cases that presented with ambiguous genitalia, an unaffected father carrying the mosaic NR5A1 variant considered the possibility of somatic mosaicism accounting for the lack of phenotypic expression. We identified a novel heterozygous indel NR5A1 variant, c.938_939delGCinsTT/p.(Arg313Leu), in a male patient with hypospadias and micropenis. Notably, a heterozygous missense variant affecting the same position, c.938G>T/p. (Arg313Leu), has been reported previously in 46,XY DSD patient with female external genitalia at birth and spontaneous and progressive virilization around puberty [40]. The arginine at position 313 is located within the ligand-binding domain of NR5A1 and is highly conserved across species. Functional studies have demonstrated that substitutions at this position impair SF-1 transcriptional activity, confirming the functional relevance of this region [41].

Novel homozygous or compound heterozygous variants in the DHH gene were identified in 5 patients from two unrelated families. Interestingly, siblings carrying the same DHH variants within each family exhibited variable phenotypes, ranging from female external genitalia to ambiguous genitalia or hypospadias. No evidence of peripheral neuropathy was detected in any of the patients during follow-up. Previous studies have reported DHH variants in association with 46,XY gonadal dysgenesis, with or without accompanying polyneuropathy [42, 43].

Heterozygous missense variants in the GATA4 gene have been associated with 46,XY DSD, with or without concomitant congenital heart defects [44]. In our cohort, we identified a novel heterozygous missense variant, c.685A>G/p.(Arg229Gly), classified as likely pathogenic according to ACMG criteria and presented with congenital heart defects, including atrial and ventricular septal defects. Notably, his brother had died due to congenital heart disease. This variant was not detected in either parent, supporting its classification as a presumed de novo occurrence. However, the presence of two affected siblings born to phenotypically unaffected parents with no family history of 46,XY DSD raises the possibility of parental germline mosaicism. To further investigate this hypothesis, sequencing of paternal sperm DNA could directly assess the presence of the variant in germ cells. In addition, haplotype analysis using microsatellite markers may help determine the parental origin of the variant and provide indirect support for germline mosaicism.

P450 oxidoreductase deficiency is a rare autosomal recessive form of congenital adrenal hyperplasia. Its clinical presentation is highly variable and may include DSD, adrenal insufficiency, and skeletal malformations resembling Antley-Bixler syndrome [45, 46]. We identified a novel homozygous 9-bp in-frame duplication, p.(Pro396_Gln398dup), in the POR gene in a 46,XY patient who was raised as female. She presented with DSD and adrenal insufficiency was detected; however, skeletal malformations are observed in a substantial proportion of patients with P450 oxidoreductase deficiency, our patient exhibited neither skeletal anomalies nor other dysmorphic features.

Our study has some limitations. The number of genes included in our targeted NGS panel was limited, potentially leading to missed diagnoses in patients with variants in genes not covered. Additionally, functional studies are needed to further assess the pathogenicity and clinical relevance of some identified variants. The major strength of our study is that, to the best of our knowledge, it represents the first large-scale investigation of 46,XY DSD in our country. Determining the prevalence of genetic variants among patients with 46,XY DSD in our population is important for guiding clinical management and offering informed genetic counseling. The identification of novel variants contributes to the expanding genotype-phenotype correlations and enhances our understanding of the genetic heterogeneity underlying 46,XY DSD.

Targeted DSD panels and exome sequencing currently provide genetic diagnosis for only nearly half of DSD patients and mainly capture the protein coding genes. Studies show that some pathogenic variants residing in noncoding regulatory or deep intronic regions can affect function of their target genes [47–49]. Further investigations incorporating advanced genomic technologies for noncoding regulatory elements, copy number variants, deep intronic regions, and epigenetic factors may aid in elucidating the genetic basis of DSD in patients who currently remain undiagnosed.

In conclusion, NGS is a useful diagnostic tool for the evaluation and management of the genetically heterogeneous 46, XY DSD patients, enabling a rapid diagnosis and simplifying hormonal investigation. For the patients whose definitive molecular diagnosis cannot yet be established, more comprehensive genomic analyses, including noncoding regions, should be pursued.

Acknowledgments

We warmly thank the patients and their families for participating in this study.

Statement of Ethics

This study adheres to the ethical principles of the World Medical Association Declaration of Helsinki and is approved by the Istanbul Faculty of Medicine Clinical Research Ethics Committee at Istanbul University under decisions numbered 380079 and 1316113. After thoroughly explaining the study purpose, informed consent was obtained from each patient, their parents, or legal guardians.

Conflict of Interest Statement

The authors have no conflict of interest to declare.

Funding Sources

This research project was supported by the Scientific Research Projects of Istanbul University (BAP Grand No.: TYL-2017–24211).

Author Contributions

S.P., Z.O.U., G.T., A.A., and F.B. conceived and coordinated study. M.Y., Z.Y.A., U.A., S.A., and A.D.A. performed sample collection. Z.O.U., G.T., A.A., and V.K. performed genetic analyses. S.P., Z.O.U., and G.T. manuscript writing. B.K., F.D., and S.B. guided the writing and critically reviewed. All authors approved the final manuscript.

Funding Statement

This research project was supported by the Scientific Research Projects of Istanbul University (BAP Grand No.: TYL-2017–24211).

Data Availability Statement

The datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request.