OBJECTIVES: Guidelines recommend C+IO as first-line treatment in non-oncogenic NSCLC. METex14 skipping has been identified as a biomarker with distinct characteristics, constituting 3‒4% of the overall NSCLC population. There are no published clinical trial studies available for C+IO in METex14 skipping NSCLC. Various modeling approaches were explored to estimate the relative cost‑effectiveness, given data limitations.

METHODS: Three methods were implemented to compare tepotinib versus C+IO using a partitioned-survival model over a 30-year time horizon: 1) real-world data, specifically in a METex14 skipping population, weighted to the untreated cohort in the tepotinib trial (VISION); 2) published hazard ratios from a C+IO versus chemotherapy trial (KEYNOTE-189) in the wider NSCLC population, applied to real-world chemotherapy data in METex14 skipping NSCLC population; and 3) matching-adjusted indirect comparison adjusting the tepotinib trial to the C+IO KEYNOTE-189 population. For each approach, parametric models were fitted to extrapolate long-term outcomes over the time-horizon. Utilities for each health-state were derived from the tepotinib study, with adverse events and age-associated disutilities extracted from the literature.

RESULTS: Survival and quality-adjusted life-year (QALY) gain in the progression-free health states ranged from 0.53 to 1.74 LYs and 0.33 to 0.94 QALYs. Method 1 estimated the greatest survival benefit for tepotinib versus C+IO of 1.65 years (3.56 vs 1.92), followed by method 3 with 1.42 years (5.06 vs 3.64), whereas method 2 showed a survival decrement of 1.95 years (3.56 vs 5.51). Total incremental QALYs ranged from −0.95 to 0.82.

CONCLUSIONS: All methods demonstrated a progression-free QALY and survival benefit of tepotinib versus C+IO providing evidence of superiority. The greatest differences are driven by post-progression survival which is likely due to the different data sources used, as only method 1 uses the correct METex14 skipping population, whereas methods 2 and 3 rely on the wider NSCLC data, which is a major limitation of these approaches.