TTP and RR have been suggested as potential surrogate endpoints for OS in advanced HCC. Here, we report correlation between OS and TTP/RR using simulated trials based on patient-level data from the phase 3 SHARP and AP trials in patients with advanced HCC randomized to sorafenib (SOR) or placebo (PBO).
A bootstrap approach was applied to simulate 10,000 trials of patients with advanced HCC from SHARP (N = 602; NCT00105443) and AP (N = 226; NCT00492752). RR was measured by investigator assessment per RECIST with SHARP–BCLC amendments (Reig et al. Semin Liver Dis 2014) prior to crossover of PBO subjects to SOR treatment. Pearson correlation was calculated between estimated median OS (mOS) and estimated median TTP (mTTP)/estimated RR for the SOR and PBO arms separately. Pearson correlation of log-rank test statistics comparing SOR and PBO were calculated for OS and TTP; Cochran–Mantel–Haenszel test statistic comparing the 2 arms for RR was also evaluated.
The mean of mOS, mTTP, and RR from simulated trials was similar to that reported in SHARP and AP (Table). The correlation between mOS and mTTP was 0.270 for SOR and 0.218 for PBO in SHARP, and 0.315 for SOR and 0.258 for PBO in AP; the correlation of log-rank test statistics comparing SOR and PBO was 0.387 in SHARP and 0.581 in AP. In SHARP, the correlation between mOS and RR was 0.174 for SOR and 0.051 for PBO; the correlation of test statistics comparing SOR and PBO was 0.156. In AP, the correlation between mOS and RR was 0.138 for SOR and 0.099 for PBO; the correlation of test statistics comparing SOR and PBO was 0.211.Table:
|Trial||Treatment||mOS, days||Simulated data Mean (SD) of mOS, days||mTTP, days||Simulated data Mean (SD) of mTTP, days||RR||Simulated data Mean (SD) of RR|
|SHARP||Sorafenib||327||329 (26.5)||119||118 (10.1)||0.057||0.057 (0.013)|
|Placebo||243||245 (21.7)||82||83 (2.4)||0.023||0.023 (0.009)|
|AP||Sorafenib||198||198 (18.8)||84||85 (6.4)||0.033||0.033 (0.014)|
|Placebo||127||129 (14.5)||42||42 (2.5)||0.013||0.013 (0.013)|
m, median; SD, standard deviation
The simulated data were representative of patient population data in the SHARP and AP trials for mOS, mTTP, and RR. Our analysis showed a weak correlation between OS and TTP/RR in these trials, suggesting that TTP and RR by RECIST are not reliable surrogate endpoints for OS in patient with advanced HCC.
Clinical trial identification
Legal entity responsible for the study
L. Huang, M. Shan, G. Meinhardt, K. Nakajima: Employment and stock ownership: Bayer. Y. De Sanctis: Employment and stock ownership: Bayer. J. Bruix: Research/Education Grant: Daiichi Sankyo, ArQule, Bayer, Sirtex. Honoraria: Gilead, AbbVie, Kowa, Bayer, BTG, ArQule, Terumo, Sirtex, BMS, Eisai, BI, Novartis, OSI, Roche, Onxeo. Advisory Board: Bayer, Kowa, BTG, ArQule, Terumo, Sirtex, BMS, Eisai, Novartis, OSI, Roche, Onxeo. Consulting: Gilead, AbbVie, Kowa, Bayer, BTG, ArQule, Terumo, Sirtex, BMS, BI, Kowa, Novartis, OSI, Roche, Onxeo, Daiichi Sankyo, Abbot, Glaxo, Eli Lilly. J. Llovet: Research/education grant: Bayer, Blueprint, Boehringer Ingelheim, Incyte Advisory board attendance: Bayer, Eisai, BMS, Eli Lilly Consultancy: Bayer, BMS, Blueprint, Eisai, Eli Lilly, Celsion, Boehringer Ingelheim. A-L. Cheng: Honoraria, advisory board attendance, consulting: ONO Pharmaceuticals, BMS, Merck, MSD, Novartis, Bayer.