Novel Phase I/II Designs for Cytotoxic and Cytostatic Agents, and Combination Treatments

This dissertation develops innovative Bayesian adaptive dose-finding designs for early-phase oncology trials, aiming to improve the identification of optimal biological doses (OBDs) by jointly modeling safety and efficacy. Unlike conventional Phase I designs that rely solely on binary toxicity outcomes to estimate the maximum tolerated dose (MTD), this work focuses on Phase I/II trials using dual continuous endpoints—Normalized Equivalent Toxicity Score (NETS) and continuous efficacy measures—to optimize the risk-benefit profile of investigational therapies. Three novel designs are introduced. First, the EWOUC-NETS design extends the Escalation with Overdose and Underdose Control (EWOUC) framework by incorporating NETS and continuous efficacy outcomes. This model-based approach enhances accuracy in OBD identification for single-agent cytotoxic therapies, using posterior utility functions to guide dose escalation while safeguarding patient safety. Second, for cytostatic agents with non-monotonic dose-efficacy relationships, a model-assisted design—STEIN-NETS—is proposed. By integrating NETS and continuous efficacy, it improves robustness and efficiency in trials where efficacy does not increase linearly with dose. Third, the EWOUC-NETS-COM design generalizes the EWOUC-NETS framework to two-agent combination therapies, addressing the complexity of multidrug regimens. It enables estimation of OBD contours in a multidimensional dose space while maintaining overdose and underdose control. Extensive simulations across varying toxicity-efficacy scenarios demonstrate that the proposed designs consistently outperform existing methods in dose estimation accuracy, patient safety, and therapeutic benefit. Real trial applications and sensitivity analyses further validate their practical utility and robustness. By leveraging all available toxicity and efficacy data, these designs offer a significant advancement in adaptive trial methodology, aligning with regulatory initiatives such as FDA’s Project Optimus. This work provides flexible, efficient, and clinically interpretable tools to improve early-phase oncology drug development.

Author dc:creator

• Jieqi Tu (14949355)

• Biology
• Biostatistics

Statement dc:rights

• In Copyright
• Open Access after 2027-09-01

DOI dc:identifier

https://doi.org/10.25417/uic.30426556.v1

OAI identifier oai:identifier

oai:figshare.com:article/30426556