Do the ASCO Value Framework and the ESMO Magnitude of Clinical Benefit Scale measure Absolute or Relative clinical benefit?

Abstract

e18378 Background: It is unclear whether the clinical benefit scores from the ASCO or ESMO valuation frameworks were intended to measure absolute or relative survival benefit. To empirically examine the measurement characteristics of these frameworks, we compared their survival efficacy components (ASCO-CBS and ESMO-PMCBGs) with established measures of absolute (median survival difference and restricted mean survival time (RMST) difference) and relative (HRs) survival benefit. Methods: FDA’s Hematology/Oncology Approvals and Safety Notifications pages were reviewed to identify phase III RCTs cited in oncology drug approvals between January 2006 and December 2017. In our primary analysis, ASCO-CBS and ESMO-PMCBGs were calculated for the included trials using the framework defined endpoint (Fr). Sensitivity analyses were conducted by calculating the scores using 1) Fr + tail-of-curve bonus points (ASCO) or long-term plateau adjustments (ESMO) (defined as Fr + TOC), 2) overall survival (OS) data only, and 3) progression-free survival (PFS) data only. For both primary and sensitivity analyses, Spearman correlation coefficients were calculated to examine the relationships between 1) ASCO-CBS/ESMO-PMCBGs and RMST difference, 2) ASCO-CBS/ESMO-PMCBGs and median survival difference, and 3) ASCO-CBS/ESMO-PMCBGs and HRs. Results: 107 RCTs were included. Compared to measures of absolute survival, ESMO-PMCBG showed low-moderate correlations with RMST difference and moderate-high correlations with median survival difference (Rho = 0.44 and 0.64, respectively). ASCO-CBS showed low-moderate correlations with both measures of absolute benefit (Rho = 0.43 and 0.44 for RMST difference and median survival, respectively). Compared to a relative measure of survival (HRs), ESMO-PMCBG revealed a moderate correlation (Rho = 0.47) while ASCO-CBS showed a higher correlation (Rho = 0.76). Conclusions: Both frameworks do not perform solely as absolute measures of survival benefit. We recommend that the frameworks consider incorporating a direct measure of absolute clinical benefit, such as RMST difference, into the survival efficacy components of their algorithms.