A Novel Algorithm Applicable to Cancer Next-Generation Sequencing Panels to Predict Total Tumor Mutation Load and Correlation with Clinical Outcomes in Melanoma.

9071 Background: Next-generation sequencing (NGS) panels of cancer-related genes are increasingly utilized in oncology. We developed an algorithm to estimate total tumor mutation burden in melanoma using genes present in local and commercial NGS panels. The algorithm was validated using independent datasets and analyzed for clinical significance in two cohorts of advanced melanoma patients (pts). Methods: Overlapping genes (n = 170) from two NGS panels were used for algorithm development. Publicly available whole exome sequencing (WES) data (n = 345) from the cutaneous melanoma TCGA was employed to develop a mutation score for each gene in the panel. The summation of the mutation scores yields the Predicted Total Mutation Load (PTML) of the tumor. The algorithm was applied to 3 independent melanoma WES datasets to test the correlation of predicted with actual mutation burden. The PTML was then determined for cohorts of melanoma pts treated with immune-therapy regimens (e.g., adoptive cell transfer (ACT)), and associations with clinical outcomes were determined. Results: The PTML correlated strongly with the actual WES mutation load of each tumor in all 3 tested melanoma WES datasets (Set 1, n = 121, r2 = 0.58; Set 2, n = 64, r2 = 0.84; Set 3, n = 127, r2 = 0.93). For subsequent analyses low PTML was defined as ≤ 100, and high as > 100 based on observed distributions. High PTML predicted both increased PFS (n = 36, p < 0.05) and OS (n = 35, p < 0.05) among pts treated with ACT. Pts with high PTML also had a longer interval from stage III to stage IV (n = 55, p < 0.05) and improved OS from stage IV (n = 62, p < 0.01) in a large cohort of MDACC patients with NGS data. Further testing demonstrated that accurate disease-specific PTML algorithms for other cancers (e.g., lung) are feasible. Conclusions: In this proof of concept study, a novel algorithm applied to NGS panels accurately predicted total mutation burden in cutaneous melanoma pts. High PTML ( > 100 mutations) correlated with improved OS in two cohorts of melanoma pts, including pts treated with immunotherapy. These results support further testing of the PTML algorithm in risk assessment and management of melanoma pts.