OR31-06 Candidate Gene Variants in a Large Cohort of Women with Primary Ovarian Insufficiency

Abstract

Primary ovarian insufficiency (POI) is highly heritable. The majority of cases have no known cause. We hypothesized that mutations in previously identified genes or genes from the same pathways are the cause of POI in a recessive or dominant manner. Subjects included 294 women diagnosed with POI (amenorrhea with an elevated FSH level). All had a 46XX karyotype, and normal FMR1 repeat number. Subjects were recruited in Boston (n=95), at the NIH and Washington University (n=98), and in Pittsburgh (n=98). Controls included subjects recruited for health in old age and disorders unrelated to reproduction or cancer, and subjects from the 1000 Genomes Project (total n=587). Variants were called using the Sentieon software package (https://www.sentieon.com). Case and control samples were stratified on ethnicity, relatedness and heterozygosity. Peddy and XPAT were used to calculate quality control metrics to detect outlier samples for removal from analysis to create a homogenous dataset. The number of cases (227) and controls (458) was adjusted for downstream analysis. XPAT imposed additional quality filters and removed variants. A second filter removed variants that did not pass a Gnomad filter of <0.001 allele frequency. VAAST was used to determine a composite likelihood ratio (CLR) as the test statistic to represent the aggregate burden of variants of affected individuals in each transcript relative to a set of 458 control genomes. The significance of each transcript’s VAAST CLR score was evaluated by 1 million permutations. We screened exomes for variants in previously identified genes causing POI in humans and those demonstrating infertility in a male or female mouse model. We also used the American College of Medical Genetics and Genomics standards for interpretation of pathogenicity of a variant, with priority on null variants in genes with probability of loss of function intolerance based on the observed vs. expected rate in gnomAD, in vivo or in vitro functional evidence of a damaging effect, significantly increased prevalence compared to controls, i.e. not found in any controls or in fewer than 10 in the gnomAD database if the subject had a matching race/ethnicity. Thirty-four subjects were removed for poor quality exomes and relatedness. Fifty-three subjects had at least one variant in a previously identified POI gene or one in which there was a previously identified functional model. Two subjects carried recessive variants and 30 carried at least one novel heterozygous candidate variant for follow up. Analysis of genetic causes of POI in this large cohort identified candidate causal gene variants in over half of the subjects. The data demonstrate that the genetic architecture is heterogeneous. Although recessive mutations have been identified in consanguineous families, the data suggest that a dominant or oligogenic pattern of inheritance may be important.