Comparison of treatment effect from randomised controlled phase II trials and subsequent phase III trials using identical regimens in the same treatment setting

PURPOSE:

We aimed to determine whether treatment effect size differed between randomised controlled phase II trials and subsequentphase III trials and examine potential predictor of positive phase III trials.

METHODS:

We searched MEDLINE for randomised controlled phase II studies published from January 2006 to December 2015. Matched phase III trials that investigated same intervention in the same setting of the same cancer were identified through Web of Science, ClinicalTrials.gov and conference proceedings. For each pair of phase II and phase III trials, we extracted hazard ratios (HRs) with 95% confidence intervals (CIs) for both overall survival (OS) and progression-free survival (PFS) and evaluated the differences by ratio of HRs (rHRs): the HR for phase II trial to that for phase III trial. A summary rHR was obtained through a random-effect meta-analysis. Univariable analyses were conducted to identify predictors of positive phase III trials.

RESULTS:

We identified 57 pairs of phase II and phase III trials. Compared with phase III trials, treatment effect sizes of PFS were, on average, 26% larger in phase II trials (rHR = 0.74, P < 0.001, 95% CI: 0.68-0.80). Treatment effect sizes of OS were 27% greater in phase II trials than in phase III trials (rHR = 0.73, P < 0.001, 95% CI: 0.66-0.79). Fifteen (26.3%) phase III trials were positive, and the only predictor of positive phase III trials was positive phase II trials CONCLUSION: Treatment effects in randomised controlled phase IItrials were greater than those in matched phase III trials. Caution must be taken when interpreting promising results from randomisedcontrolled phase II trials.