The Worldwide Landscape of Variants in the Androgen Receptor Gene Related to Androgen Insensitivity Syndrome: An Integrative Analysis

Abstract

Androgen Insensitivity Syndrome (AIS) is an X-linked genetic disease and it is the most common cause of 46,XY DSD. It is divided into 3 phenotypes: complete (CAIS), partial (PAIS), and mild (MAIS). To analyse the landscape of AR variants in AIS we collected all AR variants reported among AIS in the literature (Pubmed, EMBASE, Medline) and websites (ensemble, HGMD, ClinVar). They were analyzed according to phenotype, exon location, domain, amino acid (aa) conservation, sex assignment, external genitalia virilization (EMS score), molecular and functional studies. Conservation analysis of the AR were performed using CONSURF plataform. To test our hypothesis that non-synonymous AR variants could also impact on splicing, we used both ESEfinder and Human Splicing Finder 3.1. We founded 901 individuals with AIS: CAIS = 565 (62.7%); PAIS = 282 (31.3%); and MAIS = 54 (6%). They had 465 different AR variants: CAIS = 290 (62.3%); PAIS = 135 (29.1%); and MAIS = 40 (8.6%). Among MAIS and PAIS, most variants were at LDB domain (22 out 40 = 55% and 84 out 135 = 62.2%, respectively) whereas they were at NTD domain among CAIS (129 out 290 = 44.5%). Most were missense (81%). However, small indels (11%), nonsense (3%), splicing sites (4%) and large deletions (1%) were all reported. Non-synonymous AR variants accounting for 60%, 96%, and 100% of CAIS, PAIS, and MAIS, respectively. Synonymous AR variants were rarely found (n=3). In 81% only the AR sequencing was performed. The remaining was detected by WES (18%) or WGS (1%). Deep intronic variant was detected in PAIS (n=1) while variants in the 5’UTR of the AR gene in both PAIS and CAIS (n=2). Most AR variants were located at conserved aa (78%), but AR variants at non-conserved aa were more frequently indels (p<.01). Functional studies were found in 38%, mostly showing reduced AR expression. Among PAIS, 48% (n=134) were assigned as male at birth. The median EMS was 5 (95% CI, 5-7) in those assigned as male while it was 3.2 (95% CI, 2-6) in those assigned as female (p<.01). The median of EMS score was lower in variants at NTD domain (2.8, 95% CI, 0-7). We identified 34 AR variants causing more than one AIS phenotype (mostly CAIS and PAIS) and 6 AR variants causing all of three AIS phenotypes. In silico analysis suggests potential to disrupt normal AR splicing in 18 (53%) by creating new acceptor or donor splicing sites (n=11) or exonic splicing signals (n=7). More severe AR variants are related to CAIS. Most AR variants were reported only based on AR sequencing. Therefore, the functional pathogenicity of these variants remains unclear. Further studies including WGS could help to expand the molecular diagnosis of AIS. There is phenotype variability in AIS. So, sex assignment of patients with PAIS cannot be based on a specific identified AR gene mutation. There is potential to alter splicing among non-synonymous AR variants, which could be an explanation for phenotype variability in AIS.