Abstract 7153: Subclonal mutation load predicts survival and response to immunotherapy in cancers with low to moderate TMB

Abstract Immune checkpoint inhibitors (ICI) have revolutionized cancer treatment, yet current biomarkers like tumor mutation burden (TMB) and mismatch repair deficiency identify responders in only a small fraction of patients, particularly in low-to-moderate TMB cancers. This is critical as the majority of cancer patients (ex. approximately 90% of TCGA) have TMB<10, the FDA-approved threshold for ICI therapy. As such, there is need to identify biomarkers to predict immunotherapy response in these currently overlooked patients.We developed CliPP, a novel statistical machine learning approach that performs subclonal reconstruction through regularized regression. CliPP achieves 100 - 1000 fold faster computation than existing methods while maintaining accuracy in both simulated (n=5, 515) and real patient datasets (n=1, 582 whole-genome sequences). To facilitate broad adoption, we deployed CliPP as a user-friendly web application (CliPP-on-web, companion AACR abstract, Wu et al.) facilitating use by clinicians and researchers without computational expertise. Using CliPP, we analyzed 7, 279 tumors across 32 cancer types from TCGA, 42 samples from two ICI clinical trials in metastatic prostate cancer, and 613 esophageal adenocarcinomas (EAC) from the OCCAMS consortium. Subclonal mutation load (sML) was calculated as the fraction of subclonal mutations, with comprehensive survival analyses performed.High sML showed opposing effects on survival based on TMB: beneficial in 14 cancer types with low-to-moderate TMB (HR=0.06-0.69, P<0.05), but detrimental in 4 high-TMB cancers (HR=1.48-2.70, P<0.05). In two separate metastatic prostate cancer trials (NCT02113657 and NCT02703623) where TMB failed to predict response, high sML predicted favorable outcomes to ipilimumab (NCT02113657: HR=0.18, P=0.004, NCT02703623: HR=0.1, P=0.007) and was associated with increased CD8+ T cell density (P=0.008). Validation in 613 WGS EAC samples with standardized clinical follow-up confirmed improved survival with high sML in moderate-TMB cancers (HR=0.6, P<0.0001), with effects persistent across disease stages.These findings identify sML as a key feature of cancer and reveal an important relationship between evolutionary dynamics and the tumor microenvironment. Furthermore, sML may serve as an orthogonal approach to identify likely responders to ICI in low-to moderate TMB tumors complementing existing biomarkers and potentially broadening the therapeutic reach of checkpoint inhibitors. Citation Format: Matthew Montierth, Yujie Jiang, Kaixian Yu, Shuangxi Ji, Quang Tran, Xiaoqian Liu, Jessica Lal, Shuai Guo, Aaron Wu, Seung Shin, Ruonan Li, Shaolong Cao, Yuxin Tang, Tom Lesluyes, Scott Kopetz, Pavlos Msaouel, Anil K. Sood, Ginny Devonshire, Christopher M. Jones, Jaffer Ajani, Sumit K. Subudhi, Ana Aparicio, Padmanee Sharma, John P. Shen, Marek Kimmel, Jennifer R. Wang, Maxime Tarabichi, Rebecca C. Fitzgerald, Peter Van Loo, Hongtu Zhu, Wenyi Wang. Subclonal mutation load predicts survival and response to immunotherapy in cancers with low to moderate TMB [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7153.