Predictability of Severe Adverse Events in Phase 3 Trials from Safety Information on Phase 1 Trials in Oncology Drug Development

Safety information obtained in phase 1 trials is becoming increasingly important with the recent changes in the development strategy of oncology drugs, including the skipping of phase 2 or 3 trials before regulatory approval. This study aimed to investigate the predictive factors for severe adverse events (AEs) in phase 3 trials based on phase 1 trial data.

The data on phase 1 and phase 3 trials applied for their marketing approval of the newly approved anticancer drugs in Japan were used for analysis. Regression analyses were performed to investigate factors related to the predictability of the occurrence of severe AEs in phase 3 trials based on phase 1 trial data.

Thirty-two drugs (80 phase 1 trials and 40 phase 3 trials) were selected for the analyses. As a result of multivariate regression analyses, immune therapy agents (P = 0.009) and a pair of monotherapy in the phase 1 trials and combination therapy in the phase 3 trials (P = 0.017) were associated with low predictability of severe AEs in the phase 3 trials; signal inhibitor agents (P = 0.002) and large number of subjects in phase 1 trials (P = 0.008) were associated with high predictability. A significant relationship between the actual number of subjects in phase 1 trials and the predictability of severe AEs was observed when trials for immune checkpoint inhibitors were excluded (P < 0.001).

These results should be effectively utilized for the strategic design of early-stage oncology drug development.