DNA Damage Response Alterations and Immune Checkpoint Blockade Outcomes Across Multiple Cancers
Tian-chi Ma, Wen-heng Guo, De-min Liu, Qian-jie Li, Ren-ci Wang, He-rong Wang, Juan Li, Ya-long He, An-an Yin, Wei LinPURPOSE
DNA damage response (DDR) alterations have been implicated in immunotherapy efficacy, but their pan-cancer predictive value for immune checkpoint blockade (ICB) remains unclear. We aimed to characterize DDR mutational landscapes and evaluate their ability to predict ICB outcomes across multiple cancer types.
MATERIALS AND METHODS
We integrated multiomics data from The Cancer Genome Atlas with independent ICB-treated clinical cohorts. Patients were stratified using DDR mutation patterns identified by unsupervised clustering and machine learning approaches. Survival analyses and multivariable Cox models were used to assess associations with clinical outcomes and independence from tumor mutational burden (TMB). Immune-related features associated with DDR-defined subtypes were evaluated using bioinformatic analyses.
RESULTS
Pan-cancer clustering across 33 The Cancer Genome Atlas tumor types showed limited prognostic value in non-ICB settings. By contrast, DDR mutation–based stratification identified four subtypes that significantly distinguished overall survival in ICB-treated melanoma, non–small cell lung cancer, and gastrointestinal cancer cohorts, but not in clear cell renal cell carcinoma. These DDR subtypes did not predict survival in non–ICB-treated cohorts, indicating treatment-specific predictive relevance. In melanoma, a DDR-defined high-risk subgroup exhibited poor survival independent of TMB, whereas TMB lacked independent predictive value. DDR-defined subgroups were further associated with distinct transcriptional programs related to immune signaling and DNA damage repair.
CONCLUSION
DDR mutational landscapes function as context-dependent biomarkers of ICB efficacy. A DDR-based machine learning model predicts ICB outcomes beyond mutation burden, supporting a translational framework for context-aware biomarker development in precision oncology.