OBJECTIVES: Mature OS data are required for HTAs in oncology, particularly in reassessments following periods of managed access. Randomised controlled trial (RCT) OS data, particularly in the placebo arm, are often confounded by complex post-progression crossover and subsequent treatment, resulting in substantial censoring in early-line settings. Drawing reliable conclusions from comparative results is difficult in advanced ovarian cancer where survival can be protracted with numerous subsequent lines of therapy. This case study explored alternative modelling approaches to resolve identified limitations and provide plausible survival estimates using the National Institute for Health and Care Excellence (NICE) reappraisal of niraparib in relapsed platinum-sensitive ovarian cancer.

METHODS: Alternative methods to the evaluation of RCT placebo OS were explored, including PFS benefit to OS benefit ratio, real-world OS data, and external RCT OS data adjusted to the relevant population and used as an external comparator arm. The latter was explored as a whole and as part of a pair-wise fit option, where a ‘tail’ from a selected timepoint in a Kaplan–Meier niraparib RCT OS curve was modelled. Methods were compared and assessed for strength of evidence and tested for clinical plausibility.

RESULTS: Multiple approaches were submitted to NICE and appraised by the Evidence Review Group and the Committee. The approach most favoured by the NICE Committee utilised external RCT OS data, adjusted to the relevant population using a matching-adjusted indirect comparison. Using UK-based RWE OS routine surveillance (RS) data was criticised due to differences in the niraparib and RS RWE populations.

CONCLUSIONS: These OS modelling approaches informed NICE decision-making during reassessment of niraparib while ensuring continued patient access. This example supports consideration of modelling approaches beyond extrapolation of RCT OS; alternative approaches are likely to be required in future reassessments impacted by complex post-progression crossover and censoring to minimise uncertainty and ensure continued patient access.