Why the duration of cancer treatment requires a closer look: An empirical analysis of recent FDA approvals.

Abstract

11022 Background: The duration of cancer treatment directly impacts its efficacy, toxicity, and cost. Yet, its systematic evaluation is missing in most clinical trials. In July 2024, the FDA critiqued trial designs for failing to assess the contribution of perioperative (adjuvant and/or neoadjuvant) components but did not address the broader question of optimal treatment duration. Methods: We conducted a comprehensive, retrospective review of FDA anti-tumor drug approvals issued between January 2019, and August 2024. Landmark publications supporting each approval were identified through Google Scholar, PubMed, and DailyMed. Each trial’s duration of treatment design was classified as either indefinite or fixed, with subgroups of fixed perioperative, fixed non-perioperative, and single-dose. We also recorded the planned duration (PDOT) and median duration of treatment patients received (MeDOT). After excluding n=48 trials with altered initial approvals, 216 remained for overall analysis; 35 lacked MeDOT data and were excluded from MeDOT-specific analyses, leaving 169 for those analyses. We performed non-parametric comparisons and LASSO-selected multivariable regression of duration variables to explore predictors such as trial phase, treatment intent, drug class, and cancer type. As this research used publicly available, aggregate data, our protocol was reviewed by a senior team and did not require IRB submission per 45 CFR §46.102(f). Results: 70% of the included trials requiring >1 dose were indefinite. Among fixed, the PDOT – MeDOT difference gap was significantly larger in non-perioperative (median 16.3 months) than in perioperative (0.8 months). The MeDOT distributions for non-perioperative and indefinite trials were right-skewed with medians of 6.4 and 7.6 months respectively. In regression analyses, non-perioperative trials had significantly shorter MeDOTs, often treated hematologic malignancies, used checkpoint inhibitors, or involved front-line palliative intent. Indefinite trials were more frequent for small-molecule and tyrosine kinase inhibitors (TKIs), and fixed TKI trials had notably longer PDOTs (17.6 months). Importantly, none of the 216 approvals were based on trials comparing different treatment durations. Conclusions: While our findings suggest current oncology likely errs toward overtreatment, it is clear that researchers and regulators are not adequately assessing optimal treatment duration. Indefinite duration dominates current trial designs, while non-perioperative shows wide PDOT–MeDOT discrepancies, reflecting uncertainty about when therapeutic benefits plateau and harms increase. Future trials and regulatory directives should prioritize randomized duration comparisons to reduce toxicity and costs while optimizing patient outcomes.