分子进化揭示卵巢癌中低患病率的驱动突变、突变协同作用及相关的免疫动态

Ovarian cancer remains significantly underrepresented in genomic studies despite its high mortality rate, highlighting the need to make the most of extant data. Traditional approaches focus on mutation prevalence, potentially overlooking rare but highly oncogenic driver mutations or mutational synergies that provide substantial evolutionary advantages to tumor cells, including those that may shape the cancer-immune landscape. In ovarian cancer, tumor mutational burden (TMB), immune checkpoint molecule expression (LAG-3, ICOS, CTLA-4), and regulatory T cell (Treg) infiltration have been associated with clinical outcomes, yet their relationship to underlying mutational selection pressures remains poorly understood. We applied cancereffectsizeR to quantify the evolutionary benefit of single nucleotide variants (SNVs) in ovarian cancer, analyzing whole-exome, whole-genome, and targeted panel sequencing samples pooled across studies. This approach distinguishes mutational selection from prevalence, enabling us to quantify not only the relative contribution of high-frequency mutations but also to uncover low-frequency yet highly-selected driver mutations. We then investigated epistatic selection patterns to infer mutational ordering and synergistic interactions between mutated genes. Finally, we examined correlations between quantified selection coefficients and immune phenotypes, developing a predictive model to associate patterns of evolutionary selection within individual tumors with clinically relevant immune characteristics. Our analysis revealed numerous low-prevalence driver mutations with significant selective advantages in ovarian cancer, including mutations in BCL10 and PSIP1. Epistatic analysis uncovered mutational synergies implying temporal ordering of mutational acquisition across several gene pairs, with multiple interactions unreported in STRINGdb, such as a positive synergy between mutant EBP and IGSF21. Our predictive model relating cancer effect size to immune phenotypes-including TMB, LAG-3/ICOS/CTLA-4 expression, and Treg infiltration-revealed that somatic mutations under strong positive selection are modestly predictive of the immune landscape in ovarian cancer, suggesting that evolutionary dynamics and immunogenicity are interconnected. These findings provide new insights into ovarian cancer evolution and identify potential biomarkers for immunotherapy response.