Tutorial on Conditional Simulations With a Tumor Size-Overall Survival Model to Support Oncology Drug Development

Area of Science:

• Oncology
• Biostatistics
• Drug Development

Background:

• Overall survival (OS) is the gold standard for oncology drug approval, but initial data are often limited.
• Early drug development decisions rely on surrogate endpoints like objective response rate and progression-free survival.
• Tumor size (TS) data, collected early in clinical trials, can be leveraged to predict longer-term OS.

Purpose of the Study:

• To provide a comprehensive tutorial on using Tumor Size-Overall Survival (TS-OS) models with conditional simulations for oncology drug development.
• To demonstrate how to support decision-making by predicting long-term OS from early efficacy and TS data.
• To guide researchers in selecting, applying, and interpreting TS-OS models and simulations for ongoing studies.

Main Methods:

• Utilizing treatment-agnostic TS-OS link functions to connect early tumor size measurements with predicted OS.
• Applying conditional simulations (Bayesian forecasting) to ongoing studies, using interim TS and OS data.
• Detailing the steps for model selection, data application, simulation execution, output generation, and interpretation.

Main Results:

• TS-OS models offer a framework to forecast potential late-stage success based on early efficacy readouts.
• Conditional simulations provide a method to estimate long-term OS outcomes for ongoing oncology trials.
• The tutorial outlines practical steps for implementing these advanced statistical methods in drug development.

Conclusions:

• Conditional simulations with TS-OS models enhance informed decision-making in oncology drug development.
• Leveraging early TS and OS data through these models can optimize resource allocation and trial strategies.
• Accurate interpretation and communication of simulation outputs are crucial for effective decision support.