Abstract
BACKGROUND: Cytotoxic chemotherapy can modulate antitumor immunity, yet its impact on the peripheral T-cell receptor (TCR) repertoire in non-targetable advanced NSCLC remains poorly characterized. We prospectively investigated chemotherapy-induced TCRβ repertoire dynamics and their prognostic relevance. METHODS: Patients with recurrent unresectable locally advanced or stage IV NSCLC without actionable mutations received first-line platinum-based chemotherapy (no immunotherapy) at Ramathibodi Hospital (2021-2024). Peripheral blood was collected at baseline (T1), chemotherapy completion (T2), and confirmed disease progression (T3). TCRβ sequencing (Ion Torrent™ Oncomine™ TCR Beta-LR) was performed on samples rarefied to >1.5 million reads. Shannon diversity, Pielou evenness, TCR convergence frequency, unique clone counts, and principal component analysis (PCA) of clonal frequencies were analyzed. Multiple comparisons were Benjamini-Hochberg corrected (significance: p<0.05, BH-adj. q<0.05). RESULTS: Of 42 enrolled patients, 34 were T1-evaluable and 15 at T2; longitudinal attrition was driven primarily by pre-T2 death (74% vs. 27%; p=0.014). Disease control was achieved in 11/15 T2-evaluable patients (73%). Disease control patients trended higher TCR convergence (0.0040 vs. 0.0023) and significantly smaller decline in unique clone counts (-16% vs. -41%; Wilcoxon p=0.020, q=0.030). PCA revealed compact repertoire clustering in disease control versus wide PC2 dispersion in progressive disease (PD) patients (p=0.030, q=0.030). Among 4 patients reaching T3 (all PD), post-chemotherapy rises in Shannon diversity and convergence reversed at T3, consistent with immune attrition. Overall survival was significantly longer in disease control patients (log-rank p=0.036). CONCLUSION: Stable clonal convergence and preserved clone counts associate with disease control and survival in non-targetable advanced NSCLC receiving chemotherapy, supporting peripheral TCRβ profiling as an exploratory biomarker warranting prospective validation.