Endpoint Governance and Reporting Transparency in Biomarker-Guided Oncology Trials: A Systematic Review, 2015 to 2024

Biomarker-guided oncology trials frequently prioritize surrogate endpoints to enable efficient development in molecularly defined populations, often before mature overall survival (OS) data are available. We conducted a PRISMA 2020-guided systematic review of prospective interventional biomarker-guided oncology trials published between 2015 and 2024 to evaluate endpoint governance structure (prespecified primary endpoint selection and hierarchy) and endpoint transparency (ability to unambiguously identify prespecified primary endpoint from the full text). Endpoint-opaque trials were retained to quantify opacity but excluded from endpoint-type classification analyses. Among endpoint-transparent trials, we evaluated primary endpoint selection, OS reporting at primary publication, and OS maturity (mature/final vs. immature/interim), and used penalized logistic regression to assess predictors of objective response rate (ORR) primacy and OS immaturity. Among 242 eligible full texts, 166 (68.6%) were endpoint-opaque. Seventy endpoint-transparent trials comprised the analytic cohort: ORR and progression-free survival (PFS) were the prespecified primary endpoints in 23/70 (32.9%) and 33/70 (47.1%) trials, respectively, while OS was the sole primary endpoint in 2/70 (2.9%). OS was reported at primary publication in 62/70 trials (88.6%); among OS-reported trials with classifiable maturity (n=61), OS was mature/final in 30/61 (49.2%) and immature/interim in 31/61 (50.8%). ORR primacy was associated with lower odds of immature OS at primary publication (OR: 0.58; 95% CI: 0.37-0.90). In biomarker-guided oncology, surrogate endpoints commonly serve as decision-defining primary outcomes, while OS is frequently reported but often immature at first publication. Endpoint opacity is prevalent and limits transparent evaluation of evidentiary sufficiency, underscoring the need for explicit endpoint hierarchy reporting.