OBJECTIVES

In oncology, cost-effectiveness is generally sensitive to treatment duration. Unfortunately, treatment durations are usually unavailable for comparators included through indirect comparisons. Progression Free Survival (PFS) is then considered as a proxy of treatment duration with unneglectable consequences on ICERs. The study presented here proposes and compares different methods for approximating this treatment duration curve.

METHODS

Clinical trials in which treatment duration was available were selected. The treatments considered were intended to be administered until progression. The extrapolated time-to-off treatment (TTOT) curve for the new treatment was compared to several curves: the PFS curve of this treatment and the curve obtained after we applied the PFS Hazard Ratio to the TTOT curve of the reference treatment. Extrapolation curves used were those selected for the French Health Technology Assessment.

Several indicators were used to measure the impact of using one or the other of the curves on the results: the rate of patients on treatment at 24, 48 and 72 months, and the Restricted Mean Survival Time (RMST).

RESULTS

Six clinical trials were selected for inclusion in our analysis. Regardless of the approximation method used, the absolute RMST difference between the extrapolated TTOT curve for the new treatment and the proposed approximation curves are small (between 0.05 and 14.41 months when the PFS curve is used and between 1.19 and 5.57 months when the HR of the PFS is applied to the TTOT curve of the reference treatment). The difference in rate of patients on treatment at different time points seems to be higher when PFS curve is used.

CONCLUSIONS

Although the results presented here seems to be in favour of the use of an HR applied to the TTOT curve of the reference treatment, the conclusions are very sensitive to the function used for extrapolation and need to be further explored.