Optimizing Pediatric Dose Finding: A Phase I/II Design Integrating Adult Data

Precision dose selection in pediatric trials is imperative given children’s distinct profiles, yet constrained by ethical barriers and recruitment challenges that could be strategically mitigated through adult trial data integration. To address the limitations of existing methods, we propose the pp-BOIN12 and pp-TITE-BOIN12 designs based on the Power prior framework to identify optimal biological dose (OBD) and address delayed outcomes. These designs adaptively quantify dose-level similarity between adult and pediatric cohorts using Hellinger distance to calibrate real-time borrowing weights. Numerical simulations demonstrate that pp-BOIN12 improves the percentage of correct OBD selection while reducing patient assignments to doses above maximum tolerated dose (MTD). pp-TITE-BOIN12 maintains comparable accuracy with minimal performance degradation while significantly shortening trial duration. This work provides an adaptive and safe method for incorporating adult data into pediatric extrapolation trials.

鉴于儿童群体具有独特的生理特征，儿科临床试验中的精准剂量选择至关重要，但却受制于伦理障碍与招募难题；而通过整合成人试验数据，可从策略层面有效缓解上述问题。为解决现有方法的局限性，本文基于功效先验（Power prior）框架，提出pp-BOIN12与pp-TITE-BOIN12两种设计方案，用于识别最优生物学剂量（optimal biological dose，OBD）并处理延迟结局事件。上述设计通过海林格距离（Hellinger distance）自适应量化成人与儿科队列间的剂量水平相似性，以此校准实时信息借用权重。数值模拟结果表明，pp-BOIN12可提升正确选择最优生物学剂量的比例，同时减少被分配至高于最大耐受剂量（maximum tolerated dose，MTD）的患者例数。pp-TITE-BOIN12则可在保持相当准确性且性能衰减极小的同时，显著缩短试验周期。本研究为在儿科外推试验中整合成人数据提供了一种自适应且安全的方法。