Using bioinformatic tools to identify high-risk variants of uncertain significance in aortopathy genes that increase aortic dissection risk.

Abstract

Variants of uncertain significance (VUS) represent are clinical challenging. We hypothesize that bioinformatic tools can identify VUS that are "high risk." Primary analyses were performed in the Penn Medicine Biobank, which is composed of 43,731 participants, 11,925 of whom carried at least 1 missense VUS in the 11 genes associated with heritable thoracic aortic disease. VUS rare exome variant ensemble learner, AlphaMissense, and minor allele frequency high-risk thresholds were derived using cutpointR. These cutpoints were used to test the association of high-risk VUS with prevalent thoracic aortic disease. In the Penn Medicine Biobank, being heterozygous for a VUS was associated with a modest increased risk of thoracic aortic aneurysm (TAA: OR = 1.14, 95% confidence interval [CI] 1.01 to 1.29, P = .034) but no increased risk of aortic dissection (OR = 1.04, 95% CI 0.55 to 2.00, P = .896). Using cutpointR, we derived are exome variant ensemble learner (>0.649), AlphaMissense (>0.2543), and minor allele frequency (<8.16 × 10^-6) cutpoints that identified 435 high-risk VUS that were robustly associated with prevalent dissection (OR = 7.85, 95% CI 4.73 to 13.03, P < .001), although the association with TAA was attenuated (OR = 2.35, 95% CI 1.62 to 3.42, P < .001). Similar results were observed in 2 independent validation cohorts. High-risk VUS were strongly associated with aortic dissection, suggesting that high-risk thresholds may be applied clinically.