Abstract
INTRODUCTION: Complex underlying risk functions associated with immuno-oncology treatments have led to exploration of different methods (parametric survival, spline, landmark, and cure-fraction models) to estimate long-term survival outcomes. The objective of this study was to examine differences in estimated short- and long-term survival in previously treated metastatic Merkel cell carcinoma (mMCC) patients receiving avelumab, when using alternative extrapolation approaches. METHODS: Efficacy data from the phase 2 JAVELIN Merkel 200 trial (part A) with at least 12 months of follow-up were analyzed. Standard parametric survival analyses and analyses of overall survival (OS) as a function of surrogate outcomes comprised of response (landmark analyses) and progression-free survival plus post-progression survival (PFS + PPS) were used to project OS. Overall survival throughout lifetime was projected and compared with the observed OS data with at least 24 months of follow-up. RESULTS: Estimated OS from all three approaches provided a good fit to the observed OS curve from at least 12 months of follow-up. However, performance compared with OS data from at least 24 months showed that the landmark approach followed by PFS + PPS provided a better fit to the data as compared to standard parametric analysis. Mean life expectancy estimated with avelumab was 2.48 years with best-fitting parametric function (a log-normal distribution), 3.15 years with the landmark approach, and 3.54 years with PFS + PPS. CONCLUSION: Although standard parametric survival analysis may provide a good fit to short-term survival, it appears to underestimate the long-term survival benefits associated with avelumab in mMCC. Extrapolations based on surrogate outcomes of response or progression predict OS outcomes from longer follow-up better and appear to provide more clinically plausible projections. FUNDING: EMD Serono Inc, Rockland, MA, a business of Merck KGaA, Darmstadt, Germany.