Abstract 2337: Pan-cancer analysis of sex differences and their associations with ancestry and genomic biomarkers in a large comprehensive genomic profiling dataset

Background: Sex differences in the prevalence of cancer subtypes can inform patient screening, diagnosis, and treatment, but for many disease ontologies (DO), sex disparities have not been fully explored, especially with regards to ancestral and genomic subsets. Furthermore, existing data and reports may lack centralized expert pathology review or comprehensive genomic profiling (CGP) that can enable better characterization of biological and environmental etiologies. Design: The cohort consisted of 276,696 de-identified patient cancers, 45.7% male and 54.3% female, submitted to Foundation Medicine for CGP. Each patient was assigned an ancestry of either African, admixed American, East Asian, European, or South Asian, using a SNP-based machine learning methodology (J. Newberg et al., AACR 2019). Within the dataset 321 DOs originate at a known tissue site of origin common to both sexes. Binomial testing for significance of expected (50%) versus observed ratio of the sexes was performed for each DO, with correction for multiple hypothesis testing. For each DO, Fisher's exact tests were used to identify associations between ancestry and sex and associations between sex and gene-level genomic biomarkers (including the detection of a viral-derived gene) with ≥ 1% prevalence within the DO. Results: Of the 321 DOs, significant sex differences (corrected p-value ≤ 0.05) were seen in 106 DOs, of which 84 were significantly skewed towards males and 22 towards females. Among all DO and ancestry combinations, 14 DO-ancestry pairings showed a statistically significant (corrected p-value ≤ 0.05) skew towards one sex when compared to all other patients in the disease group. Notably, there was an enrichment of lung small cell undifferentiated carcinoma in East Asian men (OR = 2.75, corrected p-value = 0.027) and esophagus squamous cell carcinoma in East Asian men (OR = 2.55, corrected p-value = 0.028). 33 DOs exhibited sex biases in gene alterations (OR ≥ 2 and corrected p ≤ 0.05). Of particular note are a few genes that were highly enriched in a single sex across multiple distinct disease groups: CDKN2A/CDKN2B (bladder, anus, CNS non-glioma, and kidney diseases), TERT (hepato-biliary, kidney, and anus diseases), and ASXL1 (myelomas and plasma neoplasms) for males; ATRX in both lung small cell undifferentiated carcinoma and colon adenocarcinoma for females. Conclusions: Using FMI's large cancer cohort, significant differences in the occurrences of the sexes were observed i