基于代理的新型抗癌药物的快速批准（AA）引起了全球广泛关注。中国国家药品监督管理局（NMPA）也建立了类似的条件批准（CA）计划，以加快批准新药以满足未满足的医疗需求。本横断面研究旨在评估从政策实施到2022年在中国获得CA的抗癌药物的临床试验前批准证据和潜在挑战。利用NMPA的公共数据库，分析了2015年1月1日至2022年12月31日期间获得CA的抗癌药物（初始和补充适应症）。描述了获得CA的抗癌药物的特点。通过荟萃分析定量评估了来自临床试验前的主要疗效终点和安全性，包括有效率（RR），无进展生存期（PFS），总生存期（OS），治疗相关严重不良事件（SAE）和≥3级不良事件（AE）。此外，通过报告的试验级相关分析估计了代理终点与OS之间的相关性。NMPA在2015年至2022年期间批准了72个抗癌适应症（56个新分子实体）的CA。其中34个适应症（47％）也获得了FDA或EMA的批准。72个癌症适应症中，74％（53/72）基于单臂试验设计，而26％（19/72）基于随机对照试验。合并的RR为0.50（95％CI：0.45-0.55，I2 = 96％），不同癌症类型和靶点之间存在显著差异，而PFS的合并风险为0.39（95％CI：0.28-0.53，I2 = 89％），OS的合并风险为0.67（95％CI：0.61-0.73，I2 = 0％）。基于单臂设计的合并治疗相关SAE和≥3级AE分别为15％和25％。在随机对照试验中，CA药物和对照组观察到的合并治疗相关SAE和≥3级AE是可比的。在临床前关键性试验中，代理终点广泛用作主要疗效终点，其中75％（54/72）为RR，10％（7/72）为PFS，4％（3/72）为其他终点。其中，27％（17/63）的代理终点报告了与OS的试验级相关性；三个报告了高相关性（r ≥ 0.85），两个报告了中等相关性（0.70 ≤ r < 0.85），12个报告了低相关性（r < 0.70）。获得CA的大多数新型抗癌药物都是基于单臂试验设计的RR。CA所使用的代理终点与OS之间的治疗效果的相关性报道有限。我们的研究强调，在确认性临床试验中尽量引入OS或基于RCT的生活质量至关重要，以确保患者的临床益处。本研究得到了清华-北京生命科学联合中心（CLS）的博士后奖学金的支持。© 2023作者。

Accelerated approval (AA) of novel anticancer drugs based on surrogacy has attracted considerable concern globally. China National Medical Products Administration (NMPA) also established a similar conditional approval (CA) program to accelerate the approval of novel drugs to address unmet medical needs. This cross-sectional study aimed to evaluate the pre-approval clinical trial evidence and potential challenge of cancer drugs receiving CA in China from policy implementation to 2022.The cancer drugs (initial and supplemental indications) granted CA between January 1, 2015 and December 31, 2022 using the public database of the NMPA were analyzed. The characteristics of the cancer drugs received CA were described. Primary efficacy endpoints and safety derived from the pre-approval clinical trial, including response rates (RR), progression-free survival (PFS), overall survival (OS), treatment-related serious adverse events (SAE) and Grade ≥3 adverse events (AEs) were quantitatively estimated by meta-analysis. Besides, the correlation between the surrogate endpoints and OS was estimated by the reported trial-level correlation analysis.The NMPA approved 72 cancer indications (56 new molecular entities) with CA between 2015 and 2022. 34 indications (47%) were also approved by the FDA or EMA. 74% (53/72) of cancer indications were based on a single-arm trial design while 26% (19/72) for randomized controlled trials. The pooled RR was 0.50 (95% CI: 0.45-0.55, I2 = 96%) with significant differences across cancer types and targets while the pooled hazard risk was 0.39 (95% CI: 0.28-0.53, I2 = 89%) for PFS and 0.67 (95% CI: 0.61-0.73, I2 = 0%) for OS. The pooled treatment-related SAE and Grade ≥3 AEs from single-arm designs resulted in 15% and 25%, respectively. In randomized controlled trials, the pooled treatment-related SAE and Grade ≥3 AEs observed in CA drugs and the control groups were comparable. Surrogate endpoints were widely used as the primary efficacy endpoints in the pre-approval pivotal clinical trials with 75% (54/72) for RR, 10% (7/72) for PFS, and 4% (3/72) for others. Of these, 27% (17/63) of the surrogate endpoints reported a trial-level correlation with OS; three reported high correlation (r ≥ 0.85), two reported moderate correlation (0.70 ≤ r < 0.85) and 12 reported low correlation (r < 0.70).The majority of novel cancer drugs that received CA were based on RR designed for single-arm trials. The reported correlations of treatment effect between the surrogate endpoints and OS used for CA were limited. Our findings highlighted that the introduction of OS or quality of life based on RCT in confirmatory clinical trials as much as feasible was essential to ensure the clinical benefits for patients.This study was supported by postdoctoral fellowship from Tsinghua-Peking Joint Centers for Life Sciences (CLS).© 2023 The Author(s).