AlphaMissense (AI algorithm) correctly predicts PTCH1 genomic variants that underlie Gorlin syndrome

Summary: Abstract Body: Gorlin syndrome (GS), or basal cell nevus syndrome, is a rare autosomal dominant disease leading to development of numerous BCCs. While GS is commonly driven by variants of tumor suppressor gene PTCH1, exact variants are diverse and not well-characterized. Our study aimed to evaluate the functional impact of PTCH1 variants associated with GS. Using sequencing data from a prior phase 3 clinical trial of Hedgehog pathway inhibitor patidegib, we applied AlphaMissense, an AI-based pathogenicity prediction tool developed by Google DeepMind (2023), to predict pathogenicity of PTCH1 missense variants. Of 131 patients with genomic testing, 118 had PTCH1 variants. These comprised 30 missense, 22 premature termination codons (PTCs), 38 frameshifts leading to PTCs, 10 splice sites, 3 introns, 13 deletions, and 2 duplications. AlphaMissense accurately predicted pathogenicity in 95% of missense variants with established classification on ClinVar (N=20/21). After filtering duplicates, there were 7 missense variants of unknown significance (VUS), and AlphaMissense classified all 7 novel variants as “Likely pathogenic”, with pathogenicity scores ranging from 0.860 to 0.999 (scores range from 0-1, higher values indicate higher likelihood of pathogenicity). The scores showed high correlation with other mutation predictors, including PolyPhen-2 (0.984 using Pearson correlation, p-value <0.0001) and MutPred2 (0.890, p-value=0.007). These 7 subjects had many BCCs and other clinical symptoms of GS, confirming the classification of pathogenic variant. AlphaMissense enabled reclassification of novel VUS as pathogenic in patients with clinical GS, highlighting its potential for analyzing and interpreting VUS. While PTCH1 variants are identified in 85% of GS patients, many VUS may currently be misclassified. An enhanced understanding of PTCH1 variants may improve diagnostic and treatment strategies for GS patients. Amy Xiong¹, Ryan Kern¹, Qianqian Wang², Philip Beachy², Jean Y. Tang¹ 1. Department of Dermatology, Stanford University School of Medicine, Stanford, CA, United States. 2. Department of Developmental Biology, Stanford University School of Medicine, Stanford, CA, United States. Genetic Disease, Gene Regulation, Gene Therapy & Epigenetics