Molecular analysis of individuals with suspected 46,XY differences of sex development in a homogenous and understudied population

Abstract

Differences of sex development (DSD) are a group of rare congenital conditions defined by atypical chromosomal, gonadal, and/or hormonal sex. Despite advances in massively parallel sequencing (MPS), more than half of DSD cases have an unknown genetic aetiology. We recruited and analysed 21 individuals with 46,XY DSD from the Greater Middle East population using chromosomal microarray and whole exome sequencing. Participants had DSD ranging from micropenis to anorchia (absence of testes) with extra-genital features reported in four individuals (19%). Using a combination of microarray and WES, a genetic diagnosis (variants curated as likely pathogenic or pathogenic) was identified in 12/21 (57%) individuals. Microarray analysis showed two DSD participants with extra genital features had chromosomal abnormalities (48,XXXY and mosaic Y chromosomal rearrangement). Microarray also indicated a high degree of consanguinity, with extensive long contiguous stretches of homozygosity (LCSH) (>3% of the genome) in 6/21 (28.6%) individuals, all of whom received a genetic diagnosis. WES analysis revealed variants in the NR5A1 (three individuals), SRD5A2 (three individuals), TALDO1 (one individual) and AR (two individuals) genes. This includes the novel frameshift variant, c.1309del (p.Leu437Cysfs*59), in NR5A1. This study contributes to the characterisation of clinical features and molecular findings in individuals with DSD in this understudied and homogenous population and highlights the challenges with DSD diagnosis in the region. The genetic diagnoses identified may contribute to improved patient care and management.

Graphical Abstract

Graphical Abstract.

Introduction

Differences of Sex Development (DSD) describe a group of congenital conditions characterised by atypical chromosomal, gonadal, or phenotypic sex [1]. DSD phenotypes are diverse, ranging from subtle variations with minimal clinical consequences to more pronounced, severe differences that affect gonadal function and fertility. DSD can be classified as 46,XY DSD, 46,XX DSD or sex chromosomal DSD, each with further subclassification [1]. 46,XY DSD includes 46,XY gonadal dysgenesis, disorder of anti-Müllerian hormone (AMH) synthesis or action, and disorders of androgen synthesis or action [1]. The incidence rate of DSD varies from 1.5/100000 newborns (for 46,XY gonadal dysgenesis) to 460/100000 newborns (hypospadias) [2]. Due to the variable clinical presentations, determining the cause of DSD is challenging, as both genetic [3] and environmental influences have been suggested [4, 5]. These include exposure to endocrine-disrupting compounds, such as pesticides, which may interfere with endogenous hormone signalling [6, 7].

Technological advances, especially the use of massive parallel sequencing (MPS), have been helpful in the diagnosis of DSD [8, 9]. Establishing a genetic diagnosis can guide clinical care and decision making [10], and improve psychosocial well-being by providing a rational explanation to individuals with DSD and their families [10, 11]. However, like for other congenital conditions, the diagnostic rate using MPS rarely exceeds 50% [8, 9].

Although MPS is currently the most common technology used to find a genetic diagnosis in DSD cases [9], there are barriers to its application in DSD and other genetic conditions. Firstly, it is not commonly available in Low- and Middle-Income Countries (LMIC) [11, 12], although there are efforts to improve the use of MPS in these countries to investigate rare monogenic disorders [13–16]. Additionally, identifying and validating genes that cause DSD has also proven challenging. Although some forms of DSD are caused by variations in a single gene [3], MPS has revealed the potential contribution of oligogenicity in DSD [17, 18], which further complicates analysis and diagnosis.

In this study, we report the genetic analysis of a 46,XY DSD cohort from Iraq using both chromosomal microarray and WES. Iraq is part of the Greater Middle East (GME) region. A defining genetic feature of the GME population is the presence of long stretches of homozygosity, caused by the prevalence of endogamy—with rates of consanguinity ranging from 20% to 90% of all marriages [19]. Consanguinity increases the chance of inheriting identical copies of a deleterious recessive allele [20]. As several DSD are caused by recessive genetic variants, this high rate of consanguinity may increase the incidence rates of both 46,XX [21] and 46,XY DSD [22].

In this study, we have analysed 21 individuals with 46,XY DSD using a combination of microarray and WES, achieving an overall diagnostic rate of 57% (52.6% with WES). Microarray identified two individuals with aberrant chromosomes or aneuploidy, and WES detected variants in diagnostic genes in 11 individuals.

Results

Participants

A cohort of 21 individuals (Table 1) and 67 family members were recruited and consented prior to their participation. The age at examination ranged from 5 days old to 12 years old. Six out of 21 individuals (28.6%) (Individuals 1, 3, 4, 5, 11, and 19) were reported to have been born into consanguineous marriages, with the parents of the proband noted to be first cousins in all six families. Twelve families were reported to be non-consanguineous, and consanguinity information was unavailable for three families (Individuals 7, 13, and 14) (Table 1). Extra-genital features were reported in 4/21 individuals (19%), including congenital heart disease (Individual 12), enlarged spleen (Individual 13), and short stature (Individuals 1 and 5) (Table 1). All 21 patients had some hormonal testing (Supplementary Table 1). In two individuals (Individuals 1 and 17), hormonal testing was limited to AMH measurement (Supplementary Table 1), and reference values were not always available. Elevated LH levels were most common and found in 8/21 (38%) individuals, indicating hypergonadotropic hypogonadism [23]. Individuals 4 and 5 underwent hCG stimulation testing, which resulted in elevated Testosterone with no change in Dihydrotestosterone (DHT) levels (Table 1).

Table 1.

Clinical phenotype and microarray results for individuals with DSD recruited in this study.

ID	Age at examination	Karyotype	Consanguinity	Microarray results		Clinical Presentation
				Genomic Imbalance found	LCSH (>2 Mb)	Extra-genital features	External genitalia	Gonads	Pelvic Internal Organs (Ultrasound/Laparoscopy)/Other Examinations	Hormone levels
1	4 yo	N/A	Parents are cousins	arr[GRCh38] Yp11.2(2786292_10220039)×1 ~ 2,Yq11.21q12(10622985_26671489)×0 ~ 1,Yq12(26790468_57227415)×0	Yes (8.7% of genome)	Short, height and weight below the 3rd percentile.	Hooded foreskin, partially fused labioscrotal swellings, perineal urethral opening	Non palpable intra-abdominal gonads.	No uterus detected within the pelvic cavity. Testes not detected within the inguinal, pelvic, or abdominal cavity. Mullerian duct abnormality with large hydrocolpos and small uterus, a well-defined 3 cm cyst was detected in the right ovary. Small left ovary. No other internal organ abnormality.
2	N/A	48,XXXY	No		No	No	Perineal urethral opening; Size of genital tubercle 15 mm	Size of gonads (3 ml)
3	3mo	46,XY	Parents are 1st maternal cousins	No genomic imbalance	Yes (3.2% of genome)	No	Bifid scrotum, perineal urethral opening, genital tubercle: 20 mm	Size of gonads (3 ml)	N/A
4	N/A	46,XY	Parents are 1st cousins	No genomic imbalance	Yes (4.5% of genome)	No	EMS: 3, female appearing genitalia, bifid scrotum, perineal urethral opening	Palpable gonads in the labioscrotal fold, ultrasound: testes in the upper inguinal canal, size of gonads (2–3 ml)	No uterus or ovaries	Before HCG test = T:0.131 ng/ml, DHT: 0.037 ng/ml; After HCG test = T:3.913 ng/ml; DHT: 0.146 ng/ml
5	12 yo	46,XY	Parents are 1st paternal cousins	No genomic imbalance	Yes	Short stature, ichthyosis, gingival recession, long mobile teeth	Small genital tubercle, absent vagina, labioscrotal swellings: Not fused; Labioscrotal folds: Labia minora; perineal urethral opening.	Testes palpable in the labioscrotal fold. Size of gonads (3 ml)	N/A
6	1 mo	46,XY	No	No genomic imbalance	No	No	Completely fused labioscrotal swellings. Labia minora. Vaginal opening is separated from the urethra. Perineal urethral opening.	Non palpable testes	A thin linear hypoechoic band is seen in the bladder and rectum of about 5.5 cm in length, representing the vagina, and the presence of an underdeveloped uterus. Histopathology: Sections of testis show adequate numbers of seminiferous tubules lined by a single layer of Sertoli cells, normal cytology, and devoid of germ cell population with thickened basement membranes. The Interstitial tissue contains high number of Leydig cells.	LH: 31 miU/ml, FSH: 89.7 miU/ml
7	7 mo	46,XY	N/A	No genomic imbalance	No	No	Partially fused labioscrotal swelling. Size of genital tubercle 6 mm, perineal urethral opening. After treatment, 3 monthly doses of 25 mcg testosterone, penile length 2.5 cm.	Gonads in the labioscrotal fold, size of the gonads 3 ml.	N/A	LH: 2.7 miU/ml
8	45 do	46,XY	No	No genomic imbalance	No	No	Fused labioscrotal swelling, hooded foreskin, perineal urethral opening	Intra-abdominal unpalpable gonads	No Mullerian remnant	LH: 4.16 miU/ml
9	5 do	46,XY	No	No genomic imbalance	Yes (0.3% of genome)	No	Bifid scrotum, perineal urethral opening, hooded foreskin, genital tubercle: 20 mm	Testis in the inguinal canal. Size of gonads (2 ml)	N/A	T: 2.134 ng/ml, DHT: 0.057 ng/ml, LH: 9 miU/ml, FSH: 1.89 miU/ml
10	3 mo	46,XY	No	No genomic imbalance	No	No	Labioscrotal swellings not fused (labia majora), hooded foreskin, genital tubercle: 20 mm, perineal urethral opening	Size of gonads (2 ml), bilateral inguinal hernia	N/A	LH: 5.21 miU/ml
11	8 yo	46,XY	Parents are 1st paternal cousins	No genomic imbalance	Yes (9% of genome)	No	Small penis, at 6 yo penile length 3.5 cm	Undescended testes in the upper inguinal canal	N/A
12	13 mo	46,XY	No	No genomic imbalance	Yes (1% of genome)	Congenital heart disease, small fine features, polydactyly	Micropenis, bilateral undescended testes in the inguinal canal. Penile length first seen about 0.5 cm	N/A	No uterus or ovaries
13	1 yo 8 mo	46,XY	N/A	No genomic imbalance	Yes (7% of genome)	Severely enlarged spleen size (11 cm). Homozygous, no focal lesion, small patent ductus arteriosus	Microgenitalia, genital tubercle 20 mm	Both testes are small and firm, the right side has a hydrocele, both are in the scrotum. Right testis 1 ml, Left: Difficult to palpate.	N/A	LH:15.2 miU/ml, FSH: 112.7 miU/ml
14	18 do	46,XY	N/A	arr[GRCh38] Xp22.31(8439917_8641364)×2	Yes (0.7% of genome)	No	Bifid scrotum, perineal urethral opening, genital tubercle: 27 mm. Hooded foreskin	Gonads in the labioscrotal fold. Normal testicles in the suprascrotal region. No ovaries or uterus. Size of gonads (3 ml)	N/A	LH: 2.39 miU/ml
15	20 do	N/A	No	No genomic imbalance	No	No	Perineal urethral opening, penile length 1 cm, positive rugae on the scrotum	Palpable testes in both scrotal folds	Female internal genitalia not detected
16	10 do	46,XY	No	No genomic imbalance	Yes (0.3% of genome)	No	Vaginal opening is not visible distinctively from the urethra. Striation is present, no pigmentation. Labioscrotal swelling is completely fused. Genital tubercle: 1.5 cm	Testes 3 ml in scrotal sac	No uterus or ovaries	LH: 17.1 miU/ml
17	2 yo	46,XY	No	No genomic imbalance	Yes (0.2% of genome)	No	Penile length 4 cm, well-formed scrotal sacs	Unpalpable absent testes, bilateral testes not seen, but a well-defined hypoechoic solid nodule, most likely atrophied testes.	N/A
18	2 yo	46,XY	No	No genomic imbalance	No	No	Penile length (1 cm), bifid scrotum	Palpable testes in both scrotal fold, testes 2 ml	N/A
19	1 yo 3 mo	46,XY	N/A	No genomic imbalance	No	No	Completely fused labioscrotal swellings. Urethral opening tip of the penis.	Testes very small and impalpable. The right testis could not be found.	Bilateral undescended testes are high in the inguinal canal.
20	10 mo	46,XY	Parents are 1st maternal cousins	No genomic imbalance	No	No	Bifid scrotum, perineal urethral opening, genital tubercle: 35 mm after operation and treatment.	Testes palpable in the labioscrotal fold. Size of gonads (3 ml)	N/A
21	6 yo	46,XY	No	No genomic imbalance	No	No	Poorly developed scrotum, testes not palpable, penile length 3.2 cm, urethral opening is tip of the penis	No testes in the scrotum or inguinal canal, nor intra-abdominal. Normal urinary bladder.	Laparoscopy: No testes found bilaterally. The morphological features of the biopsy are in keeping with testicular regression syndrome.	LH:1.6 miU/ml

LCSH = Long Contiguous Stretches of Homozygosity, N/A = Data not available

Microarray

Microarray analysis was performed in all individuals to determine sex chromosome complement and detect genomic imbalances. Nineteen out of 21 (81%) individuals had a normal 46,XY karyotype (Table 1). Individual 1 has a mosaic structural rearrangement of the Y chromosome. The abnormal Y chromosome consists of 1.4 copies of the Yp11.2 region, 0.6 copies of the Yq11.21q12 region, and a complete loss of the Yq12-qter region. Structural rearrangements of the Y chromosome can result in chromosomal instability, sometimes causing loss of the Y chromosome in some cells and leading to a wide spectrum of phenotypes including atypical genitalia [24]. Individual 2 has a 48,XXXY karyotype. Six individuals (Individuals 1, 3, 4, 5, 11, and 13) have extensive LCSH across more than 3% of their genomes (Table 1). One family reported to be consanguineous, however, no LCSH regions were found in the proband (Individual 19). Individual 14 has a chromosome Xp22.31 duplication (arr[GRCh38] Xp22.31 (8439917_8641364)×2). The Xp22.31 duplication was identified in a region containing the ANOS1 gene, in which loss-of-function variants have been associated with Kallman Syndrome [25, 26] with typical clinical features including hypogonadotropic hypogonadism. Hormonal analysis results (Supplementary Table 1) indicated mini puberty in Individual 14. The effect of duplication of the ANOS1 gene or this genomic region, and whether it can cause 46,XY DSD are currently unknown, hence, this CNV identified in Individual 14 was classified as a Variant of Uncertain Significance (VUS).

Whole exome sequencing

Nineteen individuals with a normal 46,XY karyotype were then analysed using WES. In total, we have sequenced 61 individuals (19 probands and 42 family members). This includes one singleton, ten trios, five quartets, and two quintets, with one family of four with two affected siblings (Individuals 9 and 10). Rare variants with potential diagnostic relevance to DSD were then analysed and curated. No CNVs or InDels in known DSD genes were detected. Ten out of 19 individuals carried rare, high-quality single-nucleotide variants (SNVs) (52.6%) relevant to DSD, including six missense, one truncating, and two frameshift variants. These variants fall within the steroid 5-alpha reductase 2 (SRD5A2), nuclear receptor subfamily 5, group A, member 1 (NR5A1), androgen receptor (AR), gonadotropin-releasing hormone 1 (GNRH1), BRCA1-interacting protein C-terminal helicase 1 (BRIP1) (Fig. 1A), and Transaldolase 1 (TALDO1) genes. All genes are classified as green genes in PanelApp-AUS, with SRD5A2, NR5A1, AR, and GNRH1 in the DSD gene panel and BRIP1 and TALDO1 in the Mendeliome gene panel. Nine variants were classified as likely pathogenic (1 variant) and pathogenic (8 variants), which we considered diagnostic (Table 2). These are detailed below.

Figure 1.

A. Locations of the genetic variants found in their respective genes (SRD5A2, NR5A1, AR, GNRH1, BRIP1, and TALDO1). The novel NR5A1 p.Leu437Cysfs*59 variant is underlined and it is the third frameshift variant resulting in an elongated NR5A1 protein. The previously reported p.Arg427Alafs*140 and p.Leu437Thrfs*57 are indicated in italics. B. The NR5A1 p.Leu437Cysfs*59 variant alters the final 25 amino acids of the NR5A1 protein (left) and replaces them with a new and elongated 59-amino-acid C-terminal section (right). TM = transmembrane domain, DBD = DNA binding-domain, FtzF1 = Fushi-tarazu factor 1 box, AF2 = Activation functional domain 2, GnRH = Gonadotropin releasing hormone domain, DEAD_2 = Helicase DEAD-like domain, Helicase_C_2 = Helicase C-terminal domain, TAL_FSA = Transladolase/Fructose-6-phoosphate aldolase domain.

Table 2.

Variants identified in participants with 46,XY DSD.

ID	Gene	Genomic change	Reference transcript	cDNA change	Protein change	rs number	Zygosity	Inheritance	gnomAD (v4.1.1) freq	Iranome freq	ACMG classification
3	SRD5A2	chr2:g.31529419G>A	NM_000348.4	c.586G>A	p.(Gly196Ser)	rs121434250	Hom	Biparental	0.00008	0.000416	P
4	SRD5A2	chr2:g.2:31526224G>A	NM_000348.4	c.737G>A	p.(Arg246Gln)	rs9332967	Hom	N/A	0.0001	Absent	P
5	SRD5A2	chr2:g.31580801G>A	NM_000348.4	c.100G>A	p.(Gly34Arg)	rs782032018	Hom	Biparental	0.00001	Absent	P
6	NR5A1	chr9:g.124491109_124491112del	NM_004959.5	c.1106_1109del	p.(Val369Alafs*12)	rs2538674045	Het	De Novo	Absent	Absent	P
7	NR5A1	chr9:g.124500709G>A	NM_004959.5	c.251G>A	p.(Arg84His)	rs375469069	Het	Paternal	0.00000062	Absent	P
8	NR5A1	chr9:g.124482834del	NM_004959.5	c.1309del	p.(Leu437Cysfs*59)	Absent	Het	De novo	Absent	Absent	VUS
9	AR	chrX:g.67643345G>T	NM_000044.6	c.1706G>T	p.(Gly569Val)	Absent	Hem	Maternal	Absent	Absent	P
10	AR	chrX:g.67643345G>T	NM_000044.6	c.1706G>T	p.(Gly569Val)	Absent	Hem	Maternal	Absent	Absent	P
11	GNRH1	chr8:g.25423244del	NM_000825.3	c.87delA	p.(Leu30Cysfs*12)	rs587777859	Hom	Biparental	0.00002	Absent	P
12	BRIP1	chr17:g.61716051C>T	NM_032043.3	c.2392C>T	p.(Arg798*)	rs137852986	Hom	Biparental	0.0002	0.000416	P
13	TALDO1	chr11:g.763450G>T	NM_006755.2	c.568G>T	p.(Val190Phe)	rs1025695642	Hom	Biparental	Absent	Absent	LP

Freq = Frequency, P: Pathogenic, LP: Likely Pathogenic, VUS: Variant of Uncertain Significance, N/A: No data available; Hom: Homozygote; Het: Heterozygote; Hem: Hemizygote

Variants in steroid 5-alpha-reductase 2 (SRD5A2)

Three participants (Individuals 3, 4, 5) are homozygous for previously reported, missense variants in the SRD5A2 gene (c.586G>A (p.Gly196Ser); c.737G>A (p.Arg246Gln); and c.100G>A (p.Gly34Arg) respectively) (Fig. 1A). The p.Gly34Arg and p.Gly196Ser variants have been previously shown to reduce SRD5A2 enzyme activity to 1.7% and 8.3% of WT, respectively [27]. The p.Arg246Gln variant is thought to decrease the enzyme’s affinity to its coenzyme, nicotinamide adenine dinucleotide phosphate, reducing its activity [28]. Two variants (c.100G>A (p.Gly34Arg) and c.586G>A (p.Gly196Ser)) are inherited from heterozygous carrier parents. Genetic testing was not performed on the parents of Individual 4 and thus we could not confirm the inheritance of the p.Arg246Gln variant. All three of these participants had a phenotype consistent with 5-alpha reductase deficiency (OMIM #607306), including a perineal urethral opening and a bifid scrotum, as noted in Individuals 3 and 4, and evidence of deficient 5-alpha reductase activity (Individual 4) (Table 1). One participant (Case #4) had elevated Testosterone levels after a Human Chorionic Gonadotrophin (HCG) stimulation test (0.131 ng/ml to 3.913 ng/ml) without DHT elevation (0.0372 ng/ml to 0.1455 ng/ml) (Table 1). The consanguineous marriages reported in two families (Individuals 3 and 4) likely contribute to the inheritance of SRD5A2 variants, given the homozygosity and a history of DSD in more distant family members was reported -a maternal aunt of Individual 3 was reported to have ambiguous genitalia and was raised as a female with signs of virilisation, and Individual 4 has a maternal uncle with micropenis. However, genetic testing was not performed on these individuals. In the highly consanguineous and homogenous population of Saudi Arabia, it has been reported that there is a high rate of SRD5A2 variants [22]. The SRD5A2 variants identified are all within an LCSH region of chromosome 2 (Individual 3, chr2:7,654,805-34,125,567; Individual 4, chr2:30,255,762-35,123,825; Individual 5, chr2:12,992,852-90,202,954) (Supplementary Table 2). These SRD5A2 variants were curated as pathogenic in all cases and were considered diagnostic findings.

Variants in nuclear receptor subfamily 5, group a, member 1 (NR5A1)

Three variants were also detected in the NR5A1 gene in three participants (Individuals 6, 7, and 8) (Fig. 1A), who presented with a range of clinical features (Table 2). These include a completely fused labioscrotal swelling with unpalpable testes (Individual 6), an intra-abdominal unpalpable gonad (Individual 7), gonads in the labioscrotal fold (Individual 8), elevated LH (Individual 6: 31 miU/ml; Individual 7: 2.7 miU/ml; Individual 8: 4.16 miU/ml) (Table 1). A de novo frameshift variant (c.1106_1109del (p.Val369Alafs*12)), predicted to result in a truncated NR5A1 protein that escapes nonsense-mediated decay, was found in Individual 6 (Table 2). This has been previously reported (rs2538674045) and is considered pathogenic. Additionally, a previously reported heterozygous missense variant (c.251G>A (p.Arg84His), rs375469069) identified in Individual 7 is paternally inherited and classified as a pathogenic variant (Table 2). This variant has been previously shown to impair the transcriptional activity of NR5A1 on the luteinizing hormone β-subunit promoter [29] and affect NR5A1 nuclear localisation [30]. Incomplete penetrance and variable expressivity are commonly reported with variants in NR5A1, with variants even showing variable phenotypes within families [8, 18]. Additionally, some reports have suggested that oligogenecity plays a role in the broad clinical phenotype observed in individuals with NR5A1-related DSD [18]. Thus, we cannot rule out additional variants contributing to the DSD phenotype in Individual 7 and reanalysis may be warranted as our understanding of NR5A1-associated DSD becomes available.

Finally, we also identified a novel de novo heterozygous frameshift variant in NR5A1 (c.1309del (p.Leu437Cysfs*59)) in Individual 8 (Fig. 1A and B, Table 2). This individual was noted to have intra-abdominal gonads, perineal urethral opening, and elevated LH (4.16 miU/ml) (Table 1). The frameshift is caused by a single base-pair (G) deletion at c.1309, which is predicted to produce an elongated NR5A1 protein (+33 AA) (Fig. 1). This frameshift and elongation are unlikely to cause NMD, however, it is likely to affect both NR5A1 ligand binding and C-terminal domains and affect the interaction between NR5A1 and other downstream cofactors [31]. This variant is absent from the gnomAD and Iranome population databases and is predicted to be pathogenic by the PROVEAN and MutationTaster prediction algorithms. Due to the absence of functional genomics data, we have classified this variant as of uncertain significance (VUS) and have not included this finding in the genetic diagnosis, although variants in this gene are a well-known cause of DSD, and the patient’s clinical presentation is consistent with the range of phenotypes caused by NR5A1 variants.

Variants in androgen receptor (AR)

Two affected siblings with suspected 46,XY partial Androgen insensitivity syndrome (PAIS, OMIM# 300068) carried the same maternally inherited hemizygous missense variant in the AR gene, c.1706G>T (p.Gly569Val) ((Fig. 1A and Table 2). Individual 9 has testes in the inguinal canal with low DHT levels (0.006 ng/ml). Individual 10 has a bilateral inguinal hernia; however, no DHT hormonal data were available for this individual. Both siblings have perineal urethral openings. (Table 1). The variant has previously been reported in the AR mutation database [32], however, it is absent from both gnomAD (v4.1.1) [33] and Iranome population databases (https://iranome.com/). This variant is classified as pathogenic.

A homozygous variant in gonadotropin-releasing hormone 1 (GNRH1)

Individual 11 is homozygous for a missense variant in GNRH1 (c.87delA (p.Leu30Cysfs*12)) ((Fig. 1 and Table 2). This is a previously reported disease-causing variant [34] (rs587777859). This frameshift variant is rare in gnomAD (MAF = 0.00002) and observed as heterozygous, with no homozygotes reported. This variant is also absent in Iranome and found to be inherited from both parents. Parents were noted to be consanguineous, first-degree paternal cousins (Table 1), and extensive LCSH affecting multiple chromosomes (chromosomes 4, 6, 7, 8, 9, 12, 17, and 21) was identified. The variant is located within an LCSH region of chromosome 8 (chr8:22,284,498-31,413,154) (Supplementary Table 2). Homozygous GNRH1 variants are known to cause hypogonadotropic hypogonadism [34–37] characterised by defective synthesis and secretion of pituitary hormones (LH and FSH) from the hypothalamus [38]. Given the limited hormonal data available for Individual 11, the impact of this variant on LH and FSH production in this individual is unclear, and further hormonal testing has been recommended.

Variants in individuals with extragenital features

Individual 12 has a homozygous BRIP1 nonsense variant (c.2392C>T (p.Arg798*)), which is inherited from heterozygous parents ((Fig. 1A and Table 2). This 46,XY individual was reported to have undescended testes and micropenis as well as extra genital phenotypes, including congenital heart disease and polydactyly (Table 1). Variants in BRIP1 are causative for Fanconi Anaemia (OMIM #227650), and the phenotype observed in this individual is consistent with this disease, including genital, skeletal and cardiac malformations. No consanguinity was reported in this family, however, microarray data show that LCSH is found on chromosomes 2, 12, and 17, representing about 1% of the genome of this individual (Supplementary Table 2). The variant lies within the LCSH region of chromosome 17 (chr17:59,404,407-62,802,140). The p.Arg798* is a previously reported truncating variant [39–41] and is classified as pathogenic. Previous functional studies have found that this variant causes nonsense-mediated decay (NMD), resulting in the loss of BRIP1 expression [39] and consequently disrupts double-stranded DNA break repair [42]. Normal gonadal development may depend on effective repair of double-stranded DNA breaks as variants in other genes such as BRCA2 have been implicated in ovarian dysgenesis in humans, with animal models showing a loss of gonads in both sexes [43]. A homozygous TALDO1 missense variant (c.568G>T (p.Val190Phe) in Individual 13 ((Fig. 1 and Table 2) was identified. This variant was found to be inherited from both heterozygous parents and within a region showing loss of heterozygosity on chromosome 11 (chr11:198,986-3,128,167) (Supplementary Table 2). Variants in TALDO1 cause Transaldolase Deficiency (OMIM #606003), affecting the pentose phosphate pathway (PPP), which produces ribose 5-phosphate and nicotinamide adenine dinucleotide phosphate (NADPH) (reviewed in [44]). PPP is present in multiple tissues, and disruptions in this pathway may affect the liver, lung, testis, ovaries, and adrenal cortex [45], therefore, Transaldolase deficiency has a wide clinical phenotype including short stature, hepatosplenomegaly, cardiac abnormality, and gonadal dysfunction (hypergonadotropic hypogonadism) [46]. Disrupted NADPH production may result in abnormal steroid hormone production and lead to the 46,XY DSD phenotype observed [47]. This is consistent with the clinical phenotype of Individual 13, who presented with elevated LH and FSH, splenomegaly, micropenis, and a cardiac defect in the form of a small patent ductus arteriosus (Table 1). The patient was also reported to have atypical genital development, such as micropenis and small, firm testes. The c.568G>T (p.Val190Phe) variant was previously reported in ClinVar (ClinVar ID 2436937) and classified as a VUS. The suggested diagnostic workflow to confirm the pathogenicity of this variant is to measure urinary polyols, followed by levels of C-7 sugars [46]. However, these tests were unavailable in the region and could not be performed. Manual curation using the ACMG guidelines, however, classified this variant as likely pathogenic (PM2, PM1, PP3), and we conclude that this variant, previously reported in ClinVar, is likely causative of the patient’s phenotype.

Undiagnosed individuals and candidate gene variants

Following both microarray and three phases of WES variant analysis (DSD gene panel, Mendeliome gene panel, and variant-focused analysis), eight individuals still had no genetic diagnosis. We have also used the identified LCSH regions in Individuals 14, 16, and 17 (Supplementary Table 2) to guide our variant analysis, however, no strong candidate genes or variants were found within these regions. Several strong candidate/research variants were instead identified outside the reported LCSH regions in Individuals 14, 15, 17, 19, and 20 are listed in Supplementary Table 3. This includes theterozygous variants in two known DSD genes, SRD5A2 (c.271T>C (p.Tyr91His) and HSD17B3 (c.761_762del (p.Glu254Valfs*10). However, as variants in these genes typically manifest clinically in the homozygous state, they did not reach the diagnostic threshold. We also noted several 46,XY DSD candidate genes. Two missense variants (c.1391G>A (p. p.Arg464Gln) and c.920C>G (p.Ala307Gly)) were found in the candidate DSD gene, PLXNA3 (Plexin A3), in Individuals 15 and 20. This gene is important in the development and function of the hypothalamic–pituitary-gonadal axis, and loss-of-function variants in PLXNA3 and SEMA3F have been associated with hypogonadotropic hypogonadism due to their roles in GnRH secretion and signalling [48]. However, a hypogonadotropic hypogonadism diagnosis is inconsistent with the clinical presentations in Individuals 15 and 20, whose gonadotropin levels were at normal levels and highly elevated, respectively (Supplementary Table 1), so these variants remain uncertain until further hormonal analysis can be carried out. A compound heterozygous variant in ROBO1 was also identified in Individual 14 (c.1802A>G (p.His601Arg) and c.3907C>T (p.Arg1303Trp)). ROBO1 (Roundabout guidance receptor 1) encodes a receptor for SLIT proteins, and SLIT/ROBO signalling may play an important role in testosterone synthesis [49] in the Leydig cells of the testes [50]. The gene has not yet been validated as a causative gene for 46,XY DSD, however, its role in the regulation of steroid hormones may warrant further investigation.

Discussion

In this study, we described the molecular analysis of 21 individuals with suspected 46,XY DSD from Iraq. This study is one of only a few DSD studies focusing on populations in the Greater Middle East [51–53]. Previous studies have noted the challenges of DSD research in this setting including cultural, social, religious sensitivities, as well as limited professional experience [53]. This highlights the need for collaboration between institutions in resource-limited settings and those with more established, well-equipped laboratories and expertise to improve patient care. Using a combination of microarray and WES, a diagnostic yield of 57% was achieved, with 52.6% of diagnoses made by WES in individuals with 46,XY DSD following a negative microarray. Few DSD cohort studies have reported a genetic diagnosis rate greater than 50%, with the exception of studies from China (64.3%) [54] and Brazil (59.3%) [55] achieving this rate using a combination of clinical/biochemical approaches. Other studies from Ukraine [56], China [57], and Algeria [58] have achieved a diagnostic rate of 46.5%, 41.9%, and 49.6%, respectively. Indeed, our overall diagnostic rate is slightly higher than these studies, as we included syndromic 46,XY DSD, and recruitment is targeted to individuals whose 46,XY DSD phenotype was more likely to be genetically driven rather than influenced by environmental factors. In addition to the diagnoses we found we also identified potential novel gene variants in diagnostic DSD genes (NR5A1 and SRD5A2) but because we followed stringent variant curation guidelines in the end there not enough evidence to classify them as causative. It is possible, especially for those individuals with a single SRD5A2 variant, that a second variant is present in the non-coding regions that we miss in WES. Indeed, noncoding variants in genes DSD genes such as SOX9 [59] and NR5A1 [60] have been identified in individuals with DSD. Application of whole-genome sequencing (WGS) may help address this diagnostic gap in these undiagnosed individuals.

As reviewed by Jiali et al. [61], the most common causative genetic variants for 46,XY DSD affect the NR5A1, AR, and SRD5A2 genes. This was also the case in the cohort analysed in this study, with variants in these three genes found in 8/11 individuals with genetic diagnoses. A novel heterozygous frameshift variant, c.1309del (p.Leu437Cysfs*59), was detected in the NR5A1 gene of an individual with intra-abdominal gonads, perineal urethral opening, and no Müllerian remnant. Nearly 300 variants have been reported in NR5A1 [18, 62] and this novel frameshift variant, which is located in the C-terminal domain, is the third to be described as causing elongation of the NR5A1 protein. NR5A1 frameshift variants causing elongated NR5A1 proteins (p.Leu437Thrfs*57 and p.Arg427Alafs*140) were previously reported [29, 63] affecting the AF-2 domain and L11–12 loop structure (Fig. 1). The p.Leu437Thrfs*57 variant was reported in an individual raised as female with no uterus and remnants of the fallopian tube [63] and the p.Arg427Alafs*140 variant was found in a patient with ambiguous genitalia, absent Müllerian structure and low testosterone levels [29]. The 46,XY DSD phenotypes observed in these individuals are comparable to the phenotype of the individual harbouring the c.1309del p.Leu437Cysfs*59 frameshift variant in our study [29, 63]. However, how the elongated protein affects normal NR5A1 function remains unclear, and functional studies are now required to confirm the pathogenicity of this variant. The inclusion of individuals with syndromic 46,XY DSD revealed variants in BRIP1, GNRH1, and TALDO1 genes. The truncating BRIP1 variant, p.Arg798*, was first reported in 2005 in Fanconi anaemia [40] and shown to cause a loss of BRIP1 expression, which impairs double-stranded DNA break repair [39, 40]. Fanconi anaemia is rare and characterised by congenital malformations affecting the skin, eyes, ears, cardiac, and genital development due to endocrine dysfunction [64]. The same truncating variant was previously reported in an individual of Libyan descent with Fanconi anaemia [65]. Indeed, the p.Arg798* variant is reported in 40% of all Fanconi anaemia patients in Saudi Arabia, compared to only 2% of patients in Europe [41].

We also found a homozygous TALDO1 variant. TALDO1 variants are known causes of transaldolase deficiency, which affects the pentose phosphate pathway [66]. Several cases of transaldolase deficiency have been reported in patients of Turkish [67], Arabic [68], and Emirati [69] origins and the individual in our study adds to the scarce knowledge of this rare cause of hypergonadotropic hypogonadism in the GME population. Currently, there is no effective treatment for transaldolase deficiency, however, patients might benefit from liver transplantation, although complications may occur [70]. Therapeutic options primarily aim to increase glutathione production using N-acetylcysteine, which is effective in mice [71] and well-tolerated over a 6-month period in patients [72]. Patients may also require growth hormone treatment, as short stature and bone abnormalities have been reported [46]. Early recognition of this condition is crucial to managing any future complications.

The high rate of consanguinity in the GME population (100× higher than that in the Western European or American population [73]) contributes to a unique LCSH signature, which may increase the risk of recessive conditions caused by inheriting identical copies of the same allele [20, 73]. This high rate of endogamy also increases the likelihood of finding the same germline variant in homozygous form in affected relatives, as observed in a large-scale genomic study in Qatar [74] and Tunisia [75]. The discovery of novel candidate genes that are causative of human disorders can also be achieved with fewer participants from a more homogeneous population. Studies from both Iran [76] and Saudi Arabia [77] have found more novel genes associated with neurodevelopmental disorders and intellectual disability using fewer individuals than a similar study in the United Kingdom [78]. Indeed, the recurrence of a pathogenic variant is lower in a genetically heterogeneous population as these deleterious variants are naturally removed [79]. Therefore, the GME population provides unique insight into the study of rare loss-of-function variants, including those that can cause DSD. A recent review noted the consanguinity rate ranges from 24.3% to 71.4% in the Iraqi population [80], and our study found evidence of consanguinity in 28.6% of the probands analysed. As not all families reported consanguineous marriages, microarray analysis is valuable for identifying regions of LCSH, which can help establish parental relationships and guide the investigation and diagnosis of rare genetic disorders [81, 82]. Here, using WES, all five individuals with extensive LCSH (Individuals 3, 4, 5, 11, and 13) have genetic diagnoses, with all identified variants homozygous. Four of these individuals were reported to be from consanguineous marriages, while the consanguinity status of one family with extensive LCSH was unknown. However, microarray analysis provided evidence of consanguinity through the extensive LCSH identified in the proband. These results show an increased risk of inheriting DSD-causative autosomal recessive variants in the analysed Iraqi population, which has a high rate of consanguinity. In those without a genetic diagnosis, we used genomic regions with loss of heterozygosity to guide our variant analysis search. However, we found no variants with sufficient evidence to explain the 46,XY DSD phenotype in these individuals. Clinical interpretation of variants can be challenging, and identifying LCSH regions with clinical significance may help establish a diagnosis.

Materials and methods

Participants and phenotyping

Participants and their families were recruited by clinicians from Iraq between 2021 and 2023. The inclusion criteria for this study were individuals with suspected 46,XY DSD based on initial examination by the clinician. Participants were assigned an ID number for de-identification purposes. Local clinicians and laboratories performed clinical phenotyping, investigations of internal organs, and hormonal analyses according to local clinical practice. Ethical approval was provided by the Human Research Ethics Committee at the Royal Children’s Hospital, Melbourne, Australia (HREC22073) and all participants signed informed consent.

Genetic analysis

DNA was extracted from peripheral EDTA blood at the Victorian Clinical Genetics Services (VCGS). DNA was quantified using the Qubit dsDNA Broad Range kit (Thermofisher) prior to microarray and WES analyses.

Microarray

Microarray to assess karyotype and Copy Number Variations (CNVs) was undertaken by the VCGS Cytogenetics Services, Melbourne, Australia using 300 ng of genomic DNA with the Illumina Infinium GSA-24 v3.0 (Illumina) utilising 650 000 SNP markers with a resolution of 0.20 Mb. Long Contiguous Stretches of Homozygosity (LCSH) greater than 2 Mb were reported.

Whole exome sequencing

WES was undertaken by the VCGS Clinical Genetics Services, Melbourne, Australia using Twist exome 2.0 (Twist Biosciences) on either Illumina NextSeq2000 or NovaSeq6000 (Illumina) with 80× coverage. Library preparation, sequencing, and analysis were according to the manufacturer’s instructions. Variants are identified using the Murdoch Children’s Research Institute’s (MCRI) bioinformatics pipeline, Cpipe [83] and structural variants inferred from short read sequencing data were analysed using CXGo [84]. Variant filtering and identification were carried out using Broad Institute’s SeqR (https://seqr.broadinstitute.org) [85] in three phases. The first phase of the analysis was performed using the Differences of Sex Development gene list, and a second phase analysis was done using the Mendeliome gene list from PanelApp-AUS (https://panelapp-aus.org), an online curated platform which uses a traffic light system to classify genes in a panel [86]. Green indicates genes with significant clinical evidence, amber indicates genes with insufficient evidence for clinical use, and red indicates genes with low clinical evidence. In these two phases using the PanelApp gene filters, we include variants found in genes with green, amber, and red ratings. We filtered for high-quality variants (Q > 20, allele balance > 25) with minor allele frequency (MAF) < 0.001 to account for rare variants. MAF frequency used the population database, gnoMAD v4.1.0 ((https://gnomad.broadinstitute.org/). The third phase of analysis was done using a variant-centric analysis focusing on high-quality variants (Q > 20, allele balance > 25) in any gene and any inheritance with a MAF < 0.001 likely leading to loss of function (nonsense, splice site, frameshift) or predicted moderate-high impact (missense, frameshift, in-frame insertions/deletions, nonsense or splice site) variants. The third phase of the analysis also uses a less stringent MAF threshold of < 0.01 to account for potentially pathogenic variants that may be present at a higher frequency in genetically isolated populations [87] and with particular focus on variants found within LCSH regions. Identified variants were classified based on ACMG guidelines [88] and reviewed in consultation with experienced variant curators from VCGS. To account for the MAF frequency that better reflects the GME population, Iranome (https://iranome.com/) database was used to curate the variant. The Iranome database was established through the sequencing 800 individuals from eight major ethnic groups in Iran, which represents the second-largest population in the Middle East, and may also provide a useful population reference or neighbouring regions such as Iraq [89].

Conclusion

In conclusion, our study highlights the utility of molecular cytogenetic techniques and WES in providing genetic diagnoses to individuals with DSD within an understudied population. We achieved a diagnostic rate of 57% and identified a strong VUS in NR5A1, which may be reclassified as likely pathogenic/pathogenic in the future with additional functional evidence. In recent years, WES has become more accessible and relatively more affordable, and it remains a valuable tool for obtaining a genetic diagnosis in individuals with DSD, which is crucial for the timely clinical management of these patients. However, it is important to recognise the limitations of WES, which may be overcome through the use of more comprehensive genomic approaches, such as WGS.

Author contributions

Firman Prathama Idris (Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft, Writing—review & editing), Tara Hussein Tayeb (Conceptualization, Data curation, Formal analysis, Investigation, Project administration, Resources, Validation), Gorjana Robevska (Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Writing—review & editing), Khalid Hama Sharef (Formal analysis, Investigation, Resources), Dalya Bikhtiyar Jalal (Formal analysis, Investigation, Resources), Jocelyn van den Bergen (Data curation, Formal analysis, Methodology, Validation, Writing—review & editing), Gabby Atlas (Formal analysis, Investigation, Validation), Katrina Bell (Data curation, Investigation, Software), Tiong Yang Tan (Formal analysis, Investigation, Methodology, Supervision, Validation, Writing—original draft, Writing—review & editing), Andrew Sinclair (Funding acquisition, Project administration, Resources, Supervision, Writing—original draft, Writing—review & editing), and Katie L Ayers (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing—original draft, Writing—review & editing)

Conflicts of interest

The authors declare no competing interests.

Funding

AS was supported by a grant from the National Health and Medical Research Council (GNT2025619), which also supported this work. KA was supported by Cybec Foundation and Rebecca L. Cooper Foundation. FI was supported by the Australia Awards Scholarship.