First-Line Pembrolizumab in Metastatic Non-Small Cell Lung Cancer: A Clinical Text-mining Analysis of Effectiveness, Costs, and Trial Eligibility

Background KEYNOTE-024 (KN024) positioned first-line pembrolizumab monotherapy as the standard for metastatic non-small cell lung cancer (NSCLC) with PD-L1 ≥50%, but eligibility criteria may limit generalizability to routine care. We aimed to evaluate real-world eligibility rates and compare patient characteristics, effectiveness, and costs between trial-eligible and ineligible patients, and against the KN024 trial population. Methods We conducted a retrospective cohort study in two Dutch hospitals (2017–2025). Adults with metastatic NSCLC receiving first-line pembrolizumab monotherapy were classified by KN024 eligibility (major ineligibility factors: ECOG > 2, active brain metastases, PD-L1 <50%, autoimmune disease). Overall survival (OS) and progression-free survival (PFS) were estimated using Kaplan–Meier and adjusted Cox models. Sixty-month cumulative mean costs (CMC), stratified by pembrolizumab and other hospital expenditures, were calculated using inverse probability of censoring weighting. Results 324 patients were included in the study, of whom less than half were KN024-eligible (47%). Among trial-eligible patients, median OS was shorter than in KN024 (22.3 vs 26.3 months). Ineligible patients showed a pattern towards shorter OS (16.7 vs. 22.3 months) despite similar PFS (9.0 vs. 10.2 months) compared to eligible patients, while healthcare expenditures at 60 months were comparable. Conclusions The findings reflect a significant difference between characteristics of trial populations and patients in clinical practice. While on-treatment disease control and healthcare costs are similar regardless of eligibility, ineligible patients experience shorter OS, likely due to reduced post-progression survival. The combination of comparable costs with divergent survival outcomes between subgroups warrants dedicated cost-effectiveness analyses incorporating an appropriate comparator and quality-of-life data.