Polygenic Risk Score for the Non-Invasive Prediction of Surgical Endometriosis [OP04-3D

Abstract

INTRODUCTION: Endometriosis commonly causes infertility, dysmenorrhea, dyspareunia, dyschezia, chronic pelvic pain, and other gynecologic symptoms. Unfortunately, history, physical exam, pelvic ultrasound, and even the “gold standard” of laparoscopic visualization/histology have limitations with regard to diagnostic sensitivity and specificity. The objective of this study is development of a predictive DNA test for endometriosis. METHODS: Using IRB approved protocols, endometriosis associated markers were discovered using genome wide association studies (GWAS) involving 5,571 women with surgically confirmed endometriosis and 21,971 controls and whole exome sequencing (WES) of 2,904 affected women. Allele frequencies for WES markers were compared to published data [55,860 non-Finnish European in gnomAD database (Broad Institute)]. An additive prediction model was created using 71 GWAS markers combined into a single risk term using logistic regression and 550 low-frequency exome variants individually weighted using observed odds ratios. The predictive algorithm was validated using independent cohorts (1,458 Caucasian endometriosis patients and 758 population controls). RESULTS: The mean polygenic risk score was 1.2 in population controls and 2.9 in endometriosis patients (p<10-16). The calculated Area Under the Curve (AUC) of predictor was excellent (0.88), even though endometriosis was not excluded in population controls (10% may have endometriosis). CONCLUSION: Predictive DNA testing may help to reduce the under-diagnosis and delayed diagnosis documented for endometriosis today. Blood, saliva, or other cell/tissue samples can be tested using this assay. DNA tests also have the potential to provide insights regarding molecular pathways involved, likely treatment responses, malignant potential, and other disease characteristics unavailable from histology.