OBJECTIVES : Generally, survival data of clinical trials is extrapolated for HTA purposes based on standard parametric models (SPMs). Alternatively, spline models (SM) have been applied in survival extrapolations submitted to HTA agencies to capture fluctuating hazards observed in a trial. The objective of this study is to assess the performance of SMs versus SPMs in different types of oncology datasets.

METHODS : Immature data of three trials in melanoma, breast cancer (BC) and multiple myeloma (MM) were extrapolated and compared with mature survival data of the same trials. Standard parametric distributions (Weibull, lognormal and loglogistic) and a spline model fitted on instantaneous hazards were compared to corresponding SPMs. The best fitting distributions were selected based on LOOIC. The extrapolation on immature data was compared to mature trial data in terms of ΔMean-Absolute-Deviation (ΔMAD), where a lower ΔMAD implies a better prediction. Restricted mean survival time (RMST) of the mature data was determined to assess the relative size of the ΔMAD.

RESULTS : For the melanoma dataset, where the RMST of the mature dataset was 67.88 months, the ΔMAD was 8.06 months and 23.73 months for the best fitting SM and SPM, respectively. For the BC dataset (RMST=123.42), the ΔMAD was 25.24 months for the best fitting SM and 44.07 months for the best fitting SPM. In the MM dataset (RMST=92.55), the ΔMAD was 8.09 months for the best fitting SM and 7.21 months for the best fitting SPM.

CONCLUSIONS : For two out of three trials, the SMs fitted on the immature data had better predictions for the mature data than the SPMs. This is likely explained by the fluctuating hazards underlying these datasets. For datasets with fluctuating hazards, SMs should be considered in future survival extrapolations.