Well designed pivotal clinical trials can provide robust evidence for the market authorisation of new cancer drugs, whereas lower-quality clinical evidence leads to uncertainty about drug benefits and harms. We aimed to investigate the strength of evidence supporting new cancer drug indications approved in China from 2017 to 2021.

In this retrospective analysis, we searched publicly available data from the National Medical Products Administration website to identify pivotal pre-approval efficacy trials supporting all original and supplemental cancer drug indications approved in China from Jan 1, 2017, to Dec 31, 2021. We included small molecules and biologics, and excluded traditional Chinese medicines, prophylactic vaccines, and generic or biosimilar versions of previously approved drugs. We collected trial protocols and publications from ClinicalTrials.gov, PubMed, and the China National Knowledge Infrastructure database. The primary outcome was the strength of the supporting pivotal studies, as measured by study design (randomised or single-arm) and quality. For study quality, we evaluated the ability of single-arm trials to minimise bias, measured via the adoption of external control arm and adjusted confounders, and the risk of bias in randomised controlled trials (RCTs), evaluated with the revised Cochrane tool for risk-of-bias assessment. We also used ratio of hazard ratios (RHR) to quantify differences in effect size in RCTs with different risks of bias.

We found 77 novel cancer drugs for 86 original and 62 supplemental indications that were approved in China during the study dates, based on data from 205 pivotal studies. 44 (30%) indications were supported by single-arm trials only and 104 (70%) were supported by at least one RCT. Of the 54 pivotal single-arm trials with regulatory review documents, six (11%) used aggregated data from earlier trials as external controls, without adjustment for confounders. Of the 128 pivotal RCTs with published results, 47 (37%) were assessed as having some concern and 48 (38%) as having a high risk of bias. Overall, 149 (82%) of 182 pivotal trials that were assessable for quality had limitations in bias control. RCTs with some concern or high risk of bias in the randomisation process had smaller effect sizes (RHR 0·678 [95% CI 0·532–0·864]), and those with some concern or high risk of bias in missing outcome data had larger effect sizes (1·114 [1·004–1·237]), compared with RCTs with low risk of bias in these domains.

Four-fifths of assessable pivotal studies supporting new cancer indication approvals in China from 2017 to 2021 had weaknesses in design, conduct, or reporting that introduced uncertainty to the estimation of treatment effects. To ensure the validity of drug efficacy data and reduce uncertainty, stakeholders should strengthen and implement a high-quality standard on the design, conduct, analysis, and reporting of studies supporting regulatory approval of new therapies.

National Natural Science Foundation of China.