Estimation of delay time in survival data with delayed treatment effect

Area of Science:

• Biostatistics
• Clinical Trials
• Immuno-oncology

Background:

• Delayed treatment effects are common in immuno-oncology, necessitating accurate estimation for effective therapy.
• Understanding treatment delay patterns is vital for characterizing comparative effects and optimizing therapeutic strategies.

Purpose of the Study:

• To develop statistical methods for estimating fixed and random delay times in clinical trials.
• To evaluate methods for linking pre- and post-delay hazard ratios to average hazard ratios.
• To propose a robust semiparametric joint survival model for analyzing delay and event times.

Main Methods:

• Maximum likelihood estimation for fixed delay times.
• Evaluation of functions linking pre- and post-delay hazard ratios.
• Development of a semiparametric joint survival model for random delay times, assuming a Beta distribution.
• Extension of the model for subgroup-specific delay time estimation.

Main Results:

• The proposed maximum likelihood estimator for fixed delay time was evaluated via simulation.
• Functions linking hazard ratios were assessed.
• The semiparametric joint survival model effectively estimated mean delay time and post-delay hazard ratio in simulations.
• The model demonstrated robustness in a colon cancer clinical trial data application.

Conclusions:

• Statistical models were successfully developed and validated for estimating delayed treatment effects in clinical trials.
• The proposed methods provide valuable tools for analyzing immuno-oncology data and informing treatment strategies.
• The joint survival model is robust and applicable to real-world clinical trial data, including subgroup analyses.