We collected data on clinical trials registered in the Chinese Clinical Trial Registry and Information Disclosure Platform (Chinadrugtrials.org.cn) as of December 31, 2021. This platform, established by the former China Food and Drug Administration (CFDA, now the National Medical Products Administration) on September 6, 2013, mandates that all investigational new drugs authorized by the CFDA to register and disclose their information, including the Clinical Trial Registration (CTR) number, drug name, drug type, indication, trial design, and other study identifiers (Name/Acronym).

We identified adult phase-I clinical trials of anticancer therapeutic drugs registered between September 6, 2013 and December 31, 2021. Non-phase-I trials, including phase-I/II trials, post-marketing studies, pharmacokinetic trials, supportive care trials (e.g., pain management and granulocyte colony-stimulating factor), bioequivalence studies, trials of traditional Chinese medicines, and studies involving pediatric participants were excluded.

Since the Chinese Clinical Trial Registry and Information Disclosure Platform contains only basic information on clinical trials at the time of registration and does not require the disclosure of trial results, we conducted searches in external databases to identify published clinical trial results. Given that phase-I trial results are sometimes disseminated in the form of conference abstracts, we also included a search for conference abstracts as a supplementary source of data.

We searched PubMed for published results and Google Scholar for conference abstracts, using CTR, NCT, and ChiCTR numbers. Additionally, we searched the China National Knowledge Infrastructure (CNKI) for results published in Chinese, using keywords such as drug name, indication, and trial design. The cutoff date for retrieving clinical trial results was December 31, 2023.

To identify the NCT number, we searched the study title or acronym in clinicaltrials.gov, the US National Library of Medicine database of clinical trials. For the ChiCTR number, we utilized keywords including drug name, indication, trial phase, trial name, and study design in the Chinese Clinical Trial Registry (chictr.org.cn) which is established by West China Hospital, Sichuan University.

Data extracted from Chinadrugtrials.org.cn included drug name, drug type, indication, registration date, and data pertaining to trial design, such as the inclusion of patients’ performance score (PS) scores. Additionally, we extracted indicators such as response rate (RR) and incidence of adverse events from the latest publications and conference abstracts.

The outcomes of this study measured the benefits of phase-I trials, specifically RR and disease control rate (DCR) [8, 13, 14]. Established efficacy endpoint measures included the Response Evaluation Criteria in Solid Tumors (RECIST) for solid tumors, while various researcher-specific guidelines, such as the 2014 Lugano Criteria for lymphomas and International Workshop Criteria, were used for hematological malignancies.

Risk was assessed by the incidence of adverse events, specifically the incidence of grade-3/4 adverse events, overall adverse event incidence, and treatment-related deaths [9, 10, 15]. We extracted data for these three indicators, which are commonly evaluated using the Common Terminology Criteria for Adverse Events (CTCAE) developed by the National Cancer Institute (NCI). The version of CTCAE used depends on the protocol of the clinical trial.

Descriptive statistics were employed to characterize the trials, summarizing RR, DCR, and incidence of grade-3/4 adverse events at both the trial level and median [IQR]. Due to the infrequent occurrence of treatment-related deaths, the indicator was aggregated as total counts and proportions within each trial subgroup. Furthermore, the non-parametric Kruskal–Wallis test and Wilcoxon test were computed to assess differences in probability distributions between groups, given the non-normality of the data (Supplementary e-Fig. 1and e-Fig. 2). Spearman correlation analysis was conducted to explore the relationship between RR and the incidence of grade-3/4 adverse events. All statistical analyses were performed using SAS 9.4 (level 1M7), with a two-tailed p-value of less than 0.05 deemed statistically significant.