Joint modeling of multiple repeated measures and survival data using multidimensional latent trait linear mixed model

Area of Science:

• Biostatistics
• Clinical Trials
• Neurodegenerative Diseases

Background:

• Amyotrophic lateral sclerosis (ALS) causes progressive, multidimensional impairment, making single outcome measures insufficient for tracking disease.
• Clinical trials for ALS require multiple longitudinal outcomes to assess treatment efficacy, complicated by terminal events like death or dropout.
• The time to a terminal event in ALS trials can be influenced by various longitudinal measurements, necessitating sophisticated analytical approaches.

Purpose of the Study:

• To develop a novel joint statistical model for analyzing multidimensional longitudinal outcomes and event time data in Amyotrophic lateral sclerosis (ALS) clinical trials.
• To link multiple longitudinal measures of ALS progression with the time to a terminal event using shared random effects.
• To provide a robust analytical framework for evaluating treatment effects in ALS research.

Main Methods:

• A joint model was developed, integrating a multidimensional latent trait linear mixed model (MLTLMM) for longitudinal data with a proportional hazards model for event time data.
• Shared random effects were employed to connect the longitudinal and event time components of the model.
• Model inference was performed using a Bayesian framework with Markov chain Monte Carlo (MCMC) simulation implemented in the Stan language.

Main Results:

• The proposed joint model effectively analyzes the complex, multidimensional nature of ALS progression.
• Simulation studies demonstrated the model's validity and performance in various scenarios.
• Application to the Ceftriaxone study provided insights into analyzing ALS clinical trial data.

Conclusions:

• The developed joint model offers a powerful tool for analyzing longitudinal and event time data in ALS studies.
• This approach enhances the ability to assess treatment effects by accounting for the multidimensional and progressive nature of ALS.
• The methodology is applicable to other complex diseases with similar data structures in clinical research.