Few studies have assessed the differences in outcomes between phase II versus III landmark studies in gastrointestinal (GI) malignancies. The primary aim of this study was to determine differences in overall survival (OS) and overall response rate (ORR) between phase III studies and their preceeding phase II studies and to assess for differences in reported grade 3 and 4 toxicities.
A PubMed search was performed and reports of phase III randomized clinical trials (RCTs) evaluating systemic therapies in GI malignancies, published between 2008 and 2018 in JAMA, NEJM, JCO, Ann Oncol, Lancet, and Lancet Oncol were included. Preceeding phase II studies were identified by searching in PubMed and Medline using author and group names from the phase III studies. Phase II studies could be published in any journal. Hazard ratios (HRs) for OS, ORR, and incidence of grade 3/4 adverse events (AEs) were collected and compared between available phase III and phase II pairs via meta-analysis.
Out of 788 phase III studies identified, 91 met our inclusion criteria. Corresponding phase II studies were found in 43 cases (47%). The most studied tumour sites were colorectal, gastric and pancreas cancer. Fifty-five (60.4%) phase III studies were entirely funded by industry and 44 (80%) focused on metastatic disease. In the localized disease setting, only 5 (21.7%) studies were preceeded by a phase II trial, while in the metastatic setting this occurred in 38 (55.9%) studies. Comparisons between phase III and their phase II trials, showed no difference in ORR (p = 0.191) and a hazard ratio increase in phase III OS (p = 0.028). Additionally, we observed no difference in AEs (p = 0.203), where only 52 (57%) of the phase III studies clearly stated the overall incidence of grade 3/4 toxicities.
Lack of preceeding phase II studies in 48 (52.7%) phase III studies published in high impact journals suggests there may be insufficient background data for the performance of pivotal trials. The reported ORRs and grade 3/4 toxicities were similar between phase III and II studies, however, phase III studies show an increment in HRs for OS of 19.7% (2.0-40.5 95%CI) compared to phase II studies.
Has not received any funding.
All authors have declared no conflicts of interest.