Pembrolizumab and nivolumab are both anti-PD-1 monoclonal antibodies approved for advanced non-small cell lung cancer (NSCLC), yet they differ in PD-1 binding epitope, dosing schedule, and approved indications. Whether their immune mechanisms diverge over time — and whether such differences impact overall survival (OS), pseudo-progression, or treatment discontinuation — has never been examined using longitudinal immune biomarker measurements and time-dependent statistical modeling. This systematic review searched 6 databases (2015–2025), screened 500 records, and included 25 studies for synthesis. Critically, none of the 25 studies employed joint longitudinal-survival models, multi-state analysis, or time-varying covariate frameworks — the primary research question remains entirely unanswered. Conventional random-effects meta-analysis (DerSimonian–Laird) of 3 head-to-head OS studies (n=6,554) yielded a pooled HR of 0.83 (95% CI: 0.70–0.98; p=0.026; I²=45.2%), suggesting a borderline OS advantage for pembrolizumab that is critically dependent on the Xue et al. SEER-Medicare study (weight 82.8%); omitting it renders the result non-significant (p=0.607). Pooled PFS HR: 1.07 (0.76–1.50; p=0.69) — no difference. Pembrolizumab was associated with significantly higher grade ≥3 adverse events: OR 3.44 (95% CI: 1.87–6.32; p<0.001). GRADE certainty: Low for OS/PFS; Moderate for adverse events. Prospective studies with serial immune monitoring and joint longitudinal-survival modeling are urgently required.

Index Terms— NSCLC, pembrolizumab, nivolumab, anti-PD-1, immune checkpoint inhibitor, meta-analysis, hazard ratio, PD-L1, immune trajectory, overall survival.