145P Characterising homologous recombination deficiencies in gliomas

Background: Current tumor mutational burden (TMB) testing employs large standardized panels regardless of cancer type, resulting in high costs that restrict clinical adoption.We hypothesized that cancer-specific mutational patterns might allow substantial panel size reduction without compromising clinical accuracy. Methods:Using TCGA-derived mutation data, we developed an optimization algorithm integrating gene mutation frequency, TMB correlation scoring, and cancerspecific weighting based on Cancer Gene Census annotations.Six cancer types were evaluated: breast (BRCA, n=40), prostate (PRAD, n=35), lung adenocarcinoma (LUAD, n=40), head-neck squamous (HNSC, n=35), melanoma (SKCM, n=30), and colorectal (COAD, n=30).Performance was assessed using Pearson correlation, sensitivity, specificity, and ROC analysis against whole-exome TMB reference values.Results: Panel optimization revealed two distinct performance categories.Four cancer types achieved clinical-grade accuracy with 70-gene panels: BRCA (r=0.955,sensitivity 97.3%), PRAD (r=0.940,sensitivity 96.2%), HNSC (r=0.920,sensitivity 94.5%), and LUAD (r=0.869,sensitivity 91.5%).All p-values were <0.001 after Bonferroni correction.In contrast, hypermutated cancers showed limited optimization potential: SKCM peaked at r=0.566 and COAD at r=0.473 even with 200 genes, suggesting that alternative biomarker approaches may be needed.This approach enables 78% cost reduction ($1,800 to $400 per test) for high-performing cancer types, with significant potential impact on healthcare economics. Conclusions:Cancer-specific TMB panels can maintain clinical accuracy while substantially reducing costs for cancers with lower mutational loads.This approach offers an alternative to uniform panel strategies and may help broaden access to immunotherapy biomarker testing.For hypermutated cancers, composite biomarker approaches warrant further investigation.