摘要
Clinical trial evidence is routinely evaluated for initial drug approvals, yet the benefit of indication extensions remains uncertain. This study evaluates the clinical benefit supporting new cancer drugs' initial and supplemental US Food and Drug Administration (FDA) indication approval.Clinical trial evidence supporting each indication's FDA approval was collected from the Drugs@FDA database between 2003 and 2021. Drug, indication, and clinical trial characteristics are described. Hazard ratios (HRs) for overall survival (OS), progression-free survival (PFS), and relative risk for tumor response were meta-analyzed.Out of 124 FDA-approved drugs, 78 were approved across multiple indications. Out of 374 indications, 141 were approved as combination therapies, 255 for solid cancers, 121 with biomarkers, and 182 as first-line therapy. Approval was mostly supported by open-label (267 [71%]) phase III (238 [64%]) concurrent randomized controlled trials (248 [66%]) with a median of 331 enrolled patients (interquartile range [IQR], 123-665 patients). Across 234 randomized controlled trials with available data, drugs' HRs were 0.73 (95% CI, 0.72 to 0.75; I2 = 29.6%) for OS and 0.57 (95% CI, 0.54 to 0.60; I2 = 90.6%) for PFS, whereas tumor response was 1.38 (95% CI, 1.33 to 1.42; I2 = 80.7%). Novel pharmaceuticals increased patient survival by a median of 2.80 months (IQR, 1.97-4.60 months) for OS and 3.30 months (IQR, 1.50-5.58 months) for PFS. Initial indications more frequently received accelerated approval, supported by single-arm trials for advanced-line monotherapies, than indication extensions. Initial approvals provided a higher PFS (HR, 0.48 v 0.58; P = .002) and tumor response (relative risk, 1.76 v 1.36; P < .001).New cancer drugs substantially reduce the risk of death and tumor progression, yet only marginally extend patient survival. The FDA, physicians, patients, and insurers must evaluate and decide on a drug's safety and efficacy approval, pricing, coverage, and reimbursement on an indication-specific level.