Clinical features and genetic analysis of androgen receptor gene variants in 30 prepubertal patients with androgen insensitivity syndrome

Abstract

Androgen insensitivity syndrome (AIS; Online Mendelian Inheritance in Man [OMIM; #300068]) is an X-linked recessive disorder caused by pathogenic variants in the androgen receptor (AR) gene located in the Xq11-q13 region. In this retrospective study of 30 patients with AIS, next-generation sequencing identified 24 variants in AR, including 20 missense, 1 nonsense, and 3 splice-site variants. Seven novel variants were detected in 8 patients. Of the 24 variants, 15 were classified as likely pathogenic, 8 as pathogenic, and 1 as of uncertain significance. Variants included 5 de novo and 24 familial cases. These AR variants were predominantly located in the functional domains, with the ligand-binding domain (LBD) harboring 10 variants, the DNA-binding domain (DBD) harboring 5 variants, the N-terminal domain (NTD) harboring 2 variants, and the hinge region (HR) harboring 1 variant. The highest variant detection rate occurred in exon 5 (11/30), followed by exon 3 (9/30). These findings advance our understanding of genotype including 20 missense, 1 nonsense, and 3 splice-site variants through the identification of 7 novel variants.

INTRODUCTION

Disorders of sex development (DSDs) encompass a spectrum of congenital conditions characterized by discordance between external genitalia, gonadal histology, and chromosomal sex. The primary clinical manifestation is ambiguous genitalia, with patients exhibiting mixed male and female phenotypic features; in severe cases, sex assignment at birth becomes clinically indeterminate.1 DSD classification based on chromosomal sex includes three categories: 46,XX DSD, 46,XY DSD, and sex chromosome DSD. Defects in 46,XY DSD can be broadly classified as follows: (1) defects in testicular development; (2) defects in hormonal synthesis or action; and (3) syndromic causes of 46,XY DSD.2 Particularly complex in etiology, 46,XY DSD demonstrates marked phenotypic variability and poses significant diagnostic challenges.3 Of note, androgen insensitivity syndrome (AIS) represents the predominant cause of 46,XY DSD.4

With an estimated incidence of 1:20 000 to 1:99 000 in 46,XY individuals, AIS follows an X-linked recessive inheritance pattern.5,6 This disorder manifests along a clinical continuum from complete AIS (CAIS; phenotypically female) to partial AIS (PAIS) with varying degrees of androgen resistance.5 The phenotypic spectrum ranges from typical female external genitalia to undervirilized males presenting with gynecomastia or impaired spermatogenesis.7 Normal male sexual differentiation is closely related to functional androgen signaling mediated by androgen receptor (AR) function.5 As a ligand-activated nuclear transcription factor, the AR binds testosterone (T) and dihydrotestosterone (DHT) to regulate androgen-responsive genes.8 Pathogenic variants in the AR gene constitute the exclusive molecular etiology of AIS.6 The human AR encodes a 919-amino acid AR protein,9 which comprises three functional domains:10 N-terminal transcriptional activation domain (NTD), central DNA-binding domain (DBD), and C-terminal ligand-binding domain (LBD).11,12 To date, more than 800 AR variants have been reported in patients with AIS.13 These variants are distributed throughout the coding regions and introns of AR and include point variants, insertions, and deletions. Missense variants account for the majority and are predominantly (approximately 70.0%) maternally inherited.14

This study aims to characterize the AR mutational landscape in Chinese patients with AIS to enhance diagnostic precision and clinical management. We conducted a retrospective analysis of genetic and phenotypic data from 30 cases with AIS.

PATIENTS AND METHODS

Patients

A total of 30 unrelated patients with AIS were referred to Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital (Chengdu, China) between May 2016 and December 2021, and their records were reviewed retrospectively and analyzed. These patients were clinically diagnosed with AIS through clinical presentation, physical examination, hormone testing, imaging, karyotype analysis, and genetic testing and were selected as the study population. All patients had a karyotype of 46,XY, sex determining region Y (SRY) gene positive (+). The inclusion criteria for AIS were an external masculinization score (EMS) ≤10 in an under-virilized child, supported by clinical phenotype, family history, hormone levels, and genetic testing (chromosomal and AR), with the exclusion of other genetic DSD.15 This study was approved by the Ethics Committee of the Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital (Approval No. 336 of 2022).

Hormonal analysis

Serum samples were collected retrospectively from patients with AIS to measure hormone levels. The concentrations of T and DHT were detected using chemiluminescence at Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital.

Extraction of genomic DNA

From each patient and family members, 2.0–5.0 ml of venous blood was drawn and placed in EDTA-K2 anticoagulation tubes (Becton, Dickinson and Company, Franklin Lakes, NJ, USA). Then, the Blood Genome Column Small Volume Extraction Kit (Kangwei Century Biotechnology Co., Ltd., Taizhou, China) was used to take about 0.1 ml of specimen blood in EDTA-K2 anticoagulation tubes, and genomic DNA was extracted according to the instructions and stored at −20°C.

Next-generation sequencing of genes related to sexual development

Nextera Rapid Capture Custom Enrichment (Illumina Inc., San Diego, CA, USA) was used to target and capture exons of genes related to disorders of sexual development and their 20 base regions, and high-throughput sequencing libraries were constructed. An Illumina HiSeq 2500 second-generation sequencer (Illumina Inc.) was used to sequence the constructed libraries, and the sequenced data were compared with the Human Genome Reference Database (GRCh38/hg38) after quality control. Screening and annotation of candidate pathogenic variants were performed through the Genome Aggregation Database, the 1000 Genomes Project, Online Mendelian Inheritance in Man, Human Gene Mutation Database, ClinVar database, and the whole exome sequencing database of 1775 normal controls in our laboratory. All variants were prioritized through two important steps: (1) intronic variants and those with a variant frequency >1.0% from the control population database were excluded and (2) software (Mutation Taster [available on https://www.mutationtaster.org/, last accessed on August 3, 2024], SIFT [available on http://sift-dna.org/, last accessed on August 6, 2024], Polyphen2 [available on http://genetics.bwh.harvard.edu/pph2/, last accessed on August 6, 2024], and Human Splicing Finder [available on http://www.umd.be/HSF/, last accessed on August 21, 2024]) were used to predict the possible deleterious effects of each variant on protein structure/function. The pathogenicity of variants was classified according to the “Criteria and Guidelines for the Classification of Genetic variants”15,16 published by the American College of Medical Genetics and Genomics in 2015. Variants were categorized as pathogenic (P), probably pathogenic (LP), variant of uncertain significance (VUS), probably benign, and benign.

Variant validation and inheritance analysis

Candidate pathogenic variants of AR in all cases were verified by Sanger sequencing after next-generation sequencing of genes related to sexual development. Except for case 11, in whom no samples were obtained from the patient’s parents, Sanger sequencing analysis was performed on family members of the remaining patients. Primers were designed using online Primer 3.0 (available on http://primer3.ut.ee/, last accessed on July 10, 2024) based on the AR sequence in GenBank (NM_000044.3). The polymerase chain reaction (PCR) amplification products of the genome where the candidate disease-causing variants are located were purified and then sequenced using the BigDye Sequencing Reaction Kit (Applied Biosystems, Waltham, MA, USA) and detected on an ABI PRISM3130xl Genetic Analyzer instrument (Applied Biosystems). Sequencing results were compared with the AR reference sequence (NM_000044.3).

Analysis of bioinformatics

Protein amino acid conservation among different species was analyzed using the website HomoloGene (available on http://www.ncbi.nim.nih.gov/homologene, last accessed on September 17, 2024). The AlphaFold-predicted wild-type full three-dimensional structure of AR proteins was downloaded from the UniProt database. PyMOL software (available on https://www.schrodinger.com/products/pymol, last accessed on September 17, 2024) was used to generate and display the three-dimensional structure of the proteins. The Dock module in MOE version 2022.02 was used for molecular docking of AR proteins and ligands for the AR new missense variants, including c.967G>A (p.E323K), c.2180G>A (p.R727H), c.2182A>G (p.N728D), and c.2302G>C (p.D768H). Specific binding regions or binding residue sites were selected for docking. The binding conformations were then scored and ranked. The best binding conformations were selected for interaction analysis to predict the binding activity of the protein receptor and ligand for the variant. This process was used to detect the possible effects of amino acid substitutions on protein function. We performed energetic analysis using FoldX (available on http://foldxsuite.crg.eu, last accessed on June 22, 2025) to calculate the effect of AR point variants on enzyme and protein complex binding energy. The mutant binding conformations were ranked by score, which represented the likelihood of ligand–receptor binding. For predicting the change in binding free energy, variant binding conformations were ranked by their calculated scores. The top-ranking variant conformation was selected to represent the reactive state between the two molecules. The protein’s stability was indicated by its negative value, which was indicative of its destabilization.17

RESULTS

Clinical features

All 30 patients had a karyotype of 46,XY. Five of the children had a clinical presentation of CAIS, and the rest had PAIS. Among all 30 patients, 21 (70.0%) children were raised as male, and the other 9 (30.0%) had a female gender assignment. Five patients with clinical CAIS presented at ages ranging from 5 months to 16 years. All had fully feminized genitalia and were reared as female. Twenty-five patients presented with a PAIS phenotype and had varying degrees of male insufficiency, including micropenis, scrotal division, perineal hypospadias, and testicular hypospadias. The median external genitalia score (EGS) was 4.5 (interquartile range [IQR]: 2.0–5.5). Cryptorchidism, all of which was bilateral, was present in 23.3% (7/30) of the children. The abnormal position of the urethral meatus was present in 80.0% (24/30: junction [one case], penile [one case], and perineal [22 cases]) of the patients. The EGS was significantly lower in children who were reared as female (median = 1.0 [IQR: 0.5–3.3]) than in those who were reared as male (median = 4.5 [IQR: 4.0–6.8]). The median T/DHT ratio for all patients was 4.5, ranging from 0.8 to 25.2. The clinical phenotype and hormone levels of these patients are shown in Table 1.

Table 1.

Clinical, phenotypic, and hormonal parameters of the patients with pathogenic variants in the androgen receptor gene

Patient number	Age at first evaluation	Gender of rearing	AIS type	Pathogenic variant	Zygosity	Protein domain	ACMG classification	Clinical phenotype	Inheritance	EGS	LH (mIU ml−1)	FSH (mIU ml−1)	AMH (ng ml−1)	InhB (pg ml−1)	Post hCG stimulation
															T (ng dl−1)	DHT (ng dl−1)	T: DHT
1	2 years	Male	PAIS	c.528C>A (p.S176R)	Hemizygous	NTD	Pathogenic	PH	Maternal	4.5	<0.3	2.5	112.5	277.9	398.7	229.6	1.7
2	10 months	Male	PAIS	c.1789G>A (p.A597T)	Hemizygous	DBD	Pathogenic	PH and BC	Maternal	4.5	3.7	3.4	302.8	314.5	45.5	37.9	1.2
3	6 months	Male	PAIS	c.1823G>A (p.R608Q)	Hemizygous	DBD	Pathogenic	PH	De novo	4.5	2.5	3.5	199.8	274.9	750.5	218.9	3.4
4	3 years	Female	CAIS	c.2182A>G (p.N728D)	Hemizygous	LBD	LP	Complete female external genitalia	Maternal	4.5	1.7	2.0	276.7	392.0	93.7	40.2	2.3
5	1 year and 2 months	Male	PAIS	c.2302G>C (p.D768H)	Hemizygous	LBD	Pathogenic	FG	De novo	2.0	16.8	22.7	352.4	250.9	252.2	26.2	9.6
6	1 year and 5 months	Male	PAIS	c.2522G>A (p.R841H)	Hemizygous	LBD	LP	PH	Maternal	5.5	<0.3	<1.4	229	479.56	713.5	335	2.1
7	9 months	Male	PAIS	c.1789G>A (p.A597T)	hemizygous	DBD	Pathogenic	PH	Maternal	4.5	<0.3	<1.4	99.3	46.0	253.9	116.6	2.1
8	2 years	Male	PAIS	c.2057T>C (p.V686A)	Hemizygous	LBD	LP	PH and BC	Maternal	3.5	0.3	1.4	412.2	316.4	720.1	218.5	3.3
9	11 months	Male	PAIS	c.1823G>A (p.R608Q)	Hemizygous	DBD	Pathogenic	PH and BC	Maternal	5.5	<0.3	3.9	220.5	199.6	296.2	41.8	7.1
10	7 months	Male	PAIS	c.2659A>G (p.M887V)	Hemizygous	LBD	LP	PH	Maternal	2.0	<0.3	<1.4	237.8	211.0	210.4	88.1	2.4
11	16 years	Female	CAIS	c.2242T>C (p.F748L)	Hemizygous	LBD	LP	Complete female external genitalia	NA	1.0	18.0	5.8	47.9	69.5	178.4	31.0	5.8
12	10 years	Male	PAIS	c.2270A>G (p.N757S)	Hemizygous	LBD	LP	PH	Maternal	8.0	1.8	3.4	-	-	713.2	37.6	19.0
13	1 year and 4 months	Male	PAIS	c.2182A>G (p.N728D )	Hemizygous	LBD	LP	PH and MP	Maternal	7.0	0.2	13.4	>23.0	47.9	267.1	107.3	2.5
14	11 years	Female	CAIS	c.2197G>A (p.D733N)	Hemizygous	LBD	LP	Complete female external genitalia	Maternal	1.0	6.9	3.8	-	-	586.7	196.7	3.0
15	3 years	Female	CAIS	c.2494C>T (p.R832*)	Hemizygous	LBD	Pathogenic	Complete female external genitalia	Maternal	0.5	1.8	38.1	-	-	169.4	67.3	2.5
16	5 months	Male	PAIS	c.2174-3C>G	Hemizygous	LBD	LP	PH	De novo	8.0	<0.3	<1.4	-	-	906.2	35.9	25.2
17	3 months	Male	PAIS	c.2318+3A>T	-	LBD	LP	PH	De novo	4.5	0.8	0.7	623.1	508.7	623.0	58.2	10.7
18	1 year and 11 months	Female	PAIS	c.2248A>G (p.M750V)	Hemizygous	LBD	LP	PH	Maternal	0.0	-	-	-	-	-	-	-
19	1 year and 7 months	Male	PAIS	c.1792A>C (p.S598R)	Hemizygous	DBD	LP	PH	Maternal	4.0	<0.3	<1.4	-	-	517.9	32.1	16.1
20	7 months	Male	PAIS	c.1789G>A (p.A597T)	Hemizygous	DBD	Pathogenic	PH	Maternal	4.5	0.5	1.5	15.7	197.4	233.0	18.5	12.6
21	8 years	Male	PAIS	c.1705G>T (p.G569W)	Hemizygous	DBD	Pathogenic	PH	Maternal	1.5	-	-	-	-	-	-	-
22	19 years	Male	PAIS	c.2567G>A (p.R856H)	Hemizygous	LBD	Pathogenic	PH	Maternal	5.5	<0.3	<1.4	-	-	134.0	16.3	8.2
23	3 years	Female	PAIS	c.2581A>T (p.T861S)	Hemizygous	LBD	LP	PH	Maternal	4.0	0.4	2.3	393.2	291.5	514.1	113.9	4.5
24	8 years	Female	PAIS	c.2581A>T (p.T861S)	Hemizygous	LBD	LP	PH	Maternal	2.5	0.4	4.6	371.7	209.7	371.6	91.1	4.1
25	5 months	Male	PAIS	c.1789G>A (p.A597T)	Hemizygous	DBD	Pathogenic	PH	Maternal	4.0	2.5	4.3	114.0	332.3	619.5	34.0	18.2
26	5 months	Female	PAIS	c.1826G>A (p.R609K)	Hemizygous	DBD	Pathogenic	PH	Maternal	2.0	<0.3	<1.39	328.2	280.1	320.7	56.5	5.7
27	7 months	Male	PAIS	c.528C>A (p.S176R)	Hemizygous	NTD	Pathogenic	PH	Maternal	10.5	<0.3	<1.39	280.0	195.4	848.9	78.5	10.8
28	2 years	Male	PAIS	c.967G>A (p.E323K)	Hemizygous	NTD	VUS	PH	Maternal	7.0	<0.3	1.55	166.5	81.1	178.3	38.9	4.6
29	16 years	Male	PAIS	c.1823G>A (p.R608Q)	Hemizygous	DBD	Pathogenic	PH	Maternal	6.5	18.5	29.2	-	-	145.8	186.7	0.8
30	14 years	Female	CAIS	c.2607+5G>A c.2180G>A (p.R727H)	Hemizygous	LBD	LP LP	Complete female external genitalia	De novo Maternal	0.5	11.6	1.2	726.4	-	303.9	20.4	14.9

CAIS: complete androgen insensitivity syndrome; PAIS: partial androgen insensitivity syndrome; Hom: homozygous; LBD: ligand-binding domain; NTD: N-terminal domain; DBD: DNA-binding domain; HR: hinge region; EGS: external genitalia score; LP: likely pathogenic; VUS: variant of uncertain significance; PH: proximal hypospadias; BC: bilateral cryptorchidism; FG: feminized genitalia; MP: micropenis; LH: luteinizing hormone; FSH: follicle-stimulating hormone; AMH: anti-Müllerian hormone; InhB: inhibin B; T: testosterone; DHT: dihydrotestosterone; hCG: human chorionic gonadotropin; NA: not available; AR: androgen receptor; -: not available

Variants in the AR gene

Variations in the AR were detected in all 30 patients with AIS from unrelated families, including 24 variants. Among these cases, there were 5 de novo variants and 24 with maternal inheritance. The AR variants with the highest frequency were c.1789G>A (p.A597T) and c.1823G>A (p.R608Q), which were respectively found in 10.0% (3/30) of patients. The p.N728D variant was found in 6.7% (2/30) of patients. Other variants were only detected in one patient. Twenty-four variants in the AR include 20 missense variants, one nonsense variant, and three splice variants (Figure 1). The identified variants were mainly distributed in various functional regions of the AR, where the LBD region was the hot zone of distribution. The highest detection rate of AR variations in patients with AIS was in exon 5 (11/30), followed by exon 3 (9/30). Among the 30 patients, 10, 5, 2, and 1 had variations in the LBD, DBD, NTD, and hinge regions, respectively. Sanger sequencing validation showed that p.D768H and the 3 splice-site variants were not carried by any of the parents of the patients with these variants, which indicated that they were new variants (Figure 2).

Figure 1.

Diagrammatic representation of the androgen receptor (AR) gene highlighting the identified variants across different exons. NTD: N-terminal domain; DBD: DNA-binding domain; HR: hinge region; LBD: ligand-binding domain.

Figure 2.

Sanger electropherogram of the AR seven novel variants, (a) c.967G>A (p.E323K), (b) c.2174-3C>G, (c) c.2180G>A (p.R727H), (d) c.2182A>G (p.N728D), (e) c.2302G>C (p.D768H), (f) c.2318+3A>T, and (g) c.2607+5G>A, respectively. The position of the AR variant is indicated by a red arrow. AR: androgen receptor.

Analysis of novel variants in the AR gene

Seven novel missense variants were identified in this study. Two unrelated patients had the p.N728D variant (patients 4 and 13; Table 2). Among the novel variants, p.E323K, p.F748L, p.N728D, and p.D768H are missense variants. Three splice-site variants, c.2174-3C>G, c.2318+3A>T, and c.2318+3A>T, were also identified.

Table 2.

Pathogenicity identification of AR novel variants in this study

Patient number	AIS	EGS	Codon	Protein	Type	Exon/functional domain	ACMG classification	Family history	Inheritance
28	PAIS	7	c.967G>A	p.E323K	Missense	Exon1	PM2+PP4=VUS	Mother: heterozygous for mutation	Maternal
16	PAIS	8	c.2174-3C>G	-	Splice variant	IVS4	PM6+PM2+PP3+PP4=LP	Negative	De novo
30	CAIS	0.5	c.2180G>A	p.R727H	Missense	Exon5	PM1+PM2+PM5+PP3+PP4=LP	Mother and sister: heterozygous for mutation	Maternal
4	CAIS	4.5	c.2182A>G	p.N728D	Missense	Exon5	PM1+PM5+PM2+PP3+PP4=LP	Mother and sister: heterozygous for mutation	Maternal
13	PAIS	7	c.2182A>G	p.N728D	Missense	Exon5	PM1+PM5+PM2+PP3+PP4=LP	Brother: affected; and mother: heterozygous for mutation	Maternal
5	PAIS	2	c.2302G>C	p.D768H	Missense	Exon5	PS2+PM1+PM2+PP3=P	Negative	De novo
17	PAIS	4.5	c.2318+3A>T	-	Splice variant	Exon5	PM6+PM1+PM2+PP3+PP4=LP	Negative	De novo
30	CAIS	0.5	c.2607+5G>A	-	Splice variant	Intron7	PM6+PM2+PP3+PP4=LP	Negative	De novo

CAIS: complete androgen insensitivity syndrome; PAIS: partial androgen insensitivity syndrome; PM: pathogenic moderate; PP: pathogenic supporting; PS: pathogenic strong; LP: likely pathogenic; VUS: variant of uncertain significance; ACMG: American College of Medical Genetics and Genomics; IVS: intronic variation site; AR: androgen receptor; EGS: external genitalia score; AIS: androgen insensitivity syndrome; -: not applicable

For four novel missense variants in the AR, the three-dimensional structures of the variant proteins were constructed by PyMOL (Figure 3). Analysis of the protein structure showed that the number of hydrogen bonds around the three variant sites of p.E323K, p.R727H, and p.D768H was decreased compared with wild-type. The p.E323K variant site reduced the interaction with serine at position 324. The p.R727H variant site reduced the interaction with lysine at position 721 and glutamic acid at position 443. The p.N728D variant site showed no significant change in amino acid interactions. The p.D768H variant site reduced the interactions with alanine at position 688, glycine at position 689, and histidine at position 690.

Figure 3.

Three-dimensional predictions for the protein structures of four AR variants. (a) The p.E323K variant (bottom) shows a reduced interaction with Ser324 relative to the wild-type (top). Molecular docking analysis of the p.E323K variant. The model (3D and 2D views) and docking scores indicate that the mutation decreases the binding likelihood of DHT to the AR protein surface, thereby reducing functional activity. (b) Loss of interactions at residues K721 and E443 in the p.R727H variant. The wild-type (top left) and variant (bottom left) structures are shown. Molecular docking analysis of the p.R727H variant. The model (3D and 2D views) and docking scores indicate that the mutation decreases the binding likelihood of DHT to the AR protein surface, thereby reducing functional activity. (c) The p.N728D variant exhibits no significant structural change. Structural comparisons before (top left) and after (bottom left) the variant are shown. Molecular docking of the p.N728D variant. The model (with 3D/2D structural views and docking scores) shows that the variant decreases the binding likelihood of DHT to the AR protein surface, thereby reducing functional activity. (d) Molecular analysis of the p.D768H variant. The variant model (bottom left) shows reduced interactions with the A688-G689-H690 residue triad compared to the wild-type (top left). The p.D768H variant impairs AR-DHT binding. Structural views (3D and 2D) and docking scores demonstrate reduced ligand binding at the protein surface. AR: androgen receptor; DHT: dihydrotestosterone; mol: molecule; rseq: receptor sequence residue; mseq: molecule sequence; K323: 323 lysine; E323: 323 glutamic acid; S324: 324 serine; G322: 322 glycine; E321: 321 glutamic acid; 2D: two dimension; 3D: three dimension.

Molecular docking analysis revealed that variants such as p.E323K, p.R727H, p.N728D, and p.D768H reduced the likelihood of small molecules attaching to the AR protein surface. Consequently, the binding activity of AR and DHT was reduced (Figure 3). Our data showed that variants at sites 727, 728, and 768 are present within the critical binding domain, suggesting that the variants significantly destabilize the domain and affect ligand binding (Table 3). In addition, homology analysis of amino acids in proteins from different species showed that the novel missense variants (except for the p.E323K site) are located in conserved regions of amino acids (Supplementary Figure 1 ^{(160.3KB, tif)} ).

Table 3.

Changing of folding free energy (∆∆G) of androgen receptor variants

Variant	∆∆G (foldX)	Stability
p.E323K	−0.1	Stable
p.R727H	2.1	Destabilized
p.N728D	0.1	Destabilized
p.D768H	1.9	Destabilized

∆∆G (folding free energy change): denotes the free energy difference between a protein’s folded and unfolded states

The three splice-site variants, c.2174-3C>G, c.2318+3A>T, and c.2607+5G>A, are located in introns 4, 5, and 7 of the AR, respectively. Splicing prediction software analysis predicted that these three variants were likely to interfere with splicing of the mRNA, potentially leading to altered protein function. According to the American College of Medical Genetics and Genomics guidelines,16 p.E323K is classified as variant of uncertain significance, and the variants, c.2174-3C>G, p.R727H, p.N728D, p.D768H, c.2318+3A>T, and c.2607+5G>A, are considered potentially pathogenic.

DISCUSSION

In this study, a total of 24 variants in the AR and seven novel variants were found in 30 patients with AIS. Single base substitution resulted in twenty amino acid changes, one nonsense variant, and three splice site variants. Eighteen variants were in the LBD, five in the DBD, two in the NTD, and one in the hinge region. Most of the detected missense variants in the AR were in the LBD region. This is relatively consistent with AR database records, which show that the majority of variants are point mutations located in the LBD. Missense variants in the LBD primarily disrupt AR protein function, causing CAIS through mechanisms such as interference with ligand binding, disruption of dimer formation, or impaired AR-induced transactivation.18 Among the patients with AR variants detected in this study, those with variants located in the LBD had lower EGS scores (median = 3.5) than those with variants in other regions (median = 4.5), indicating a more severe phenotype. This finding also confirms previous findings that LBD variants impair AR function more severely than other domains of the receptor.19 The c.1789G>A (p.A597T) variant was detected in four patients, all of whom were diagnosed with PAIS exhibiting hypospadias, cryptorchidism, and micropenis. All four patients harboring the p.A597T variant were under one year of age and showed no significant differences in their clinical gonadal scores or EGS scores, indicating low phenotypic variability. It has been previously shown that this single amino acid exchange may impair DNA binding of the AR and contribute to PAIS.20

Although the etiologies of a significant proportion of 46,XY DSD cases remain unclear, AIS is one of the most common disorders with a defined molecular etiology. Phenotypic variability associated with AR variants has been noted, including differences in clinical manifestations across populations harboring the same variant.21 However, Shao et al.22 reported no significant differences in the clinical presentation among patients with CAIS with different AR variants. Regarding the uncertainty of the genotype–phenotype relationship, Ledig et al.23 suggested several influencing factors: (1) the coexistence of somatic wild-type and variant alleles resulting from de novo AR variants arising post-zygotically; (2) variations in androgen availability or levels during embryonic sex differentiation; and (3) individual functional differences in coregulators of AR transcriptional activity, which may impact the phenotypic expression of known variants. AR variants causing CAIS or PAIS may also be associated with CAG repeat expansions, as variation in CAG repeat length has been suggested to reduce AR transcriptional activity.6,24 In addition, some AR variants occur in the regulatory or intronic regions, potentially affecting the transcriptional initiation or pre-mRNA splicing, ultimately impairing functional AR synthesis. For example, patient 16 in our study harbored the splice site variant c.2174-3C>G in intron 4. Some DSD phenotypes are indistinguishable from AIS. Therefore, diagnosing AIS requires excluding other etiologies of 46,XY DSD through comprehensive evaluation, including detailed phenotyping, hormonal investigations, and genetic testing.

Our study suggests correlations between specific AR variants and phenotypic abnormalities. However, developing optimal functional bioassays for AR is needed to strengthen and refine these genotype–phenotype correlations. The phenotypic heterogeneity among cases with clinically diagnosed AIS also underscores the need for a comprehensive evaluation of external genitalia development/virilization.25,26 Oligomeric factors, somatic mosaicism, and AR coregulatory factors may also contribute to phenotypic variation.27,28 Somatic cell mosaicism may be present in up to one-third of individual cases.29

Patients suspected of AIS based on clinical and hormonal criteria showed no significant differences in clinical features (e.g., cryptorchidism, hypospadias, stretched penile length, and endogenous estrogen system) or biochemical parameters (e.g., follicle-stimulating hormone, luteinizing hormone, and T levels) between those with or without identified pathogenic AR variants. These findings suggest that other candidate disease-causing genes or factors may contribute to the AIS phenotype.14 In this study, we analyzed the clinical characteristics of 30 patients with AIS. All patients met the clinical and hormonal diagnostic criteria for AIS, and the diagnosis was genetically confirmed in those harboring pathogenic AR variants. Understanding specific AR variants and their residual androgenic effects enables more precise prognoses, which can inform sex assignment decisions for 46,XY individuals with AIS and provide valuable information for genetic counseling of female carriers. Furthermore, correlating the phenotypic expression of patients with specific variants to the location of those variants within the AR can help refine the functional map of AR domains. Given the genetic heterogeneity of AIS, each study investigating AR variants in affected patients provides valuable insights into the functional significance of specific amino acid residues. Because of the low concentrations of steroids in prepubertal children, immunoassay-based methods for hormone detection may exhibit limitations in sensitivity. Moreover, this study assessed only baseline blood steroid levels without the administration of human chorionic gonadotropin stimulation tests. Therefore, liquid chromatography–mass spectrometry should be employed for the measurement of blood hormone levels in future studies to enable more accurate and reliable investigations.

In conclusion, we systematically evaluated seven previously uncharacterized AR variants. We integrated in vivo clinical data (e.g., hormonal profiles and phenotypic severity) to define genotype–phenotype correlations specific to pediatric 46,XY DSD. Over 80.0% of our cohort comprised infants (<1 year old), a population underrepresented in prior AR variant studies. Diagnostic challenges, including those related to the immature hypothalamic–pituitary–gonadal axis, are most pronounced in this age group. While our bioinformatic pipeline successfully identified high-confidence pathogenic variants, the lack of functional validation necessitates caution when interpreting causality. In future studies, comprehensive in vitro experiments should be conducted to provide direct evidence supporting the pathogenicity of the seven novel variants.

AUTHOR CONTRIBUTIONS

JYY conceived and designed the study, performed genetic analysis, and revised this manuscript. JNL performed genetic data analysis, drafted the manuscript, and collected all data. JH and ZXW assisted in clinical data collection. JW collected all clinical data, conducted clinical evaluations, and revised this manuscript. All authors reviewed, edited, read, and approved the final manuscript.

COMPETING INTERESTS

All authors declare no competing interests.

Supplementary Figure 1

The evolutionary conservation of four missense variants (p.E323K, p.R727H, p.N728D, and p.D768H) among different species. Mutated amino acids are indicated by yellow shading. AR: androgen receptor.