Longitudinal trajectories of peripheral blood lymphocyte subsets and PD-1 expression as dynamic prognostic markers in lung cancer

Background Lung cancer is characterized by strong heterogeneity and dynamic evolution. Peripheral immune subsets and checkpoint molecules reflect tumor–immune interactions, but static biomarkers cannot capture temporal changes. This study assessed whether longitudinal trajectories of lymphocyte subsets and PD-1 expression could provide real-time risk stratification for progression-free survival (PFS). Methods This retrospective cohort included 442 lung cancer patients with prospectively collected peripheral blood samples from an institutional biobank. Samples were obtained at baseline and multiple follow-ups. Frequencies of CD4 + T cells, CD8 + T cells, and natural killer (NK) cells, as well as PD-1 levels on CD4 + and CD8 + T cells, were measured by flow cytometry. Kaplan–Meier and time-dependent Cox models evaluated associations with PFS; discrimination was assessed via Harrell’s C-index. Results Mean age was 63.1 ± 8.4 years; 76% were male, 77% had non-small cell lung cancer (NSCLC), and 81% presented at late stage. Progressive disease (PD) occurred in 13.1%. Longitudinal analysis revealed distinct immune alterations in PD patients. In single-immune-variable time-dependent Cox models, time-updated CD8 + /PD-1 was an independent risk factor (HR = 9.40, 95% CI: 1.51–58.7, p = 0.017), while the other immune subsets were not significant in the overall population. Subgroup analyses showed consistent adverse effects of CD8 + /PD-1 in NSCLC (HR = 545.5, p = 0.003), SCLC (HR = 666.0, p = 0.034), early-stage (HR = 2067.7, p = 0.015), and late-stage (HR = 248.8, p = 0.005) patients. Time-updated CD4 + /PD-1 was protective in NSCLC (HR = 0.003, p = 0.009) and late-stage (HR = 0.005, p = 0.012). High baseline CD8 + /PD-1 predicted shorter median PFS (p=0.020). The time-updated model showed good discrimination (C-index 0.694). Conclusion Dynamic CD8 + /PD-1 trajectories provide independent, adverse prognostic value, whereas preserved CD4 + /PD-1 reflects immune competence in specific subgroups. Serial monitoring enables real-time risk assessment and personalized management in lung cancer.