Developing and experimental validating a T cell senescence-related gene signature to predict prognosis and immunotherapeutic sensitivity in non-small cell lung cancer.

Background: T cell senescence affects non-small cell lung cancer (NSCLC) by compromising the anti-tumor immune response. However, the prognostic significance of T cell senescence-related genes in NSCLC remains unclear.

Methods: The scRNA-seq data from normal lung and NSCLC tissues, along with co-incubation experiments involving NSCLC cells and T cells, were utilized to identify T cell senescence characteristics. The TCGA-NSCLC dataset was used for training, and 8 independent NSCLC cohorts from GEO were combined for validation. Various machine learning algorithms were employed for feature selection, with multivariate Cox regression used to construct the risk model. Two NSCLC cohorts receiving anti-PD1/PDL1 treatment from GEO were employed to validate the risk model's predictive capability for immunotherapeutic response. Additionally, 10 pairs of paracarcinoma and NSCLC tissues from a local hospital and transfection assays on T cells were used for validation.

Results: T cells in the NSCLC microenvironment displayed increased senescent features (all P < 0.05). SLC2A1, TNS4, and GGTLC1 were integrated into the risk model, which proved to be a significant prognostic predictor in both training (P < 0.001) and validation (P < 0.05) cohorts. The risk signature also demonstrated strong predictive power for immunotherapeutic sensitivity (both AUC > 0.8). Higher CD3⁺SLC2A1⁺ and CD3⁺TNS4⁺ T cell infiltration, along with lower CD3⁺GGTLC1⁺ T cell levels, were observed in NSCLC (all P < 0.05). Moreover, GGTLC1 overexpression suppressed T cell senescence (all P < 0.05).

Conclusion: A T cell senescence-related gene signature has been established to predict prognosis and immunotherapeutic response in NSCLC.