A novel immune checkpoint-based signature predicts immunotherapy benefit in "driver gene-negative" lung...

Immune checkpoint blockade (ICB) is regarded as a major breakthrough in lung cancer treatment. However, the limited response rates and immune-related adverse events associated with ICB present major challenges in clinical practice. To predict the immunotherapy response, we developed the CCCNP signature comprising 5 immune checkpoint molecules (CD226, CD276, CTLA-4, Nectin-2, and PD-L1) in "driver gene-negative" lung adenocarcinoma (LUAD). Immune checkpoint signatures were identified from large, retrospectively collected tumor cohorts by testing how their immunohistochemical characteristics influence patient survival. The predictive performance of the CCCNP signature was evaluated in a prospective cohort of 34 patients treated with anti-PD-1/PD-L1 therapy. It achieved a numerically higher accuracy of AUC (0.803, 95% CI: 0.661-0.945) compared to clinically used PD-L1, tumor mutational burden (TMB), and microsatellite instability (MSI). When applied to a newly enrolled exploratory cohort with 146 patients, the CCCNP signature could stratify a high-risk subgroup that derived statistically significant clinical benefit from the addition of immunotherapy. Overall, we present a new framework for risk stratification and support precise immunotherapy decision-making in "driver gene-negative" LUAD.