Marginal structural models with dose-delay joint-exposure for assessing variations to chemotherapy intensity

Area of Science:

• Causal inference
• Biostatistics
• Clinical trial analysis

Background:

• Time-dependent confounders complicate causal effect estimation in longitudinal studies with dynamic treatment adjustments.
• Marginal structural models (MSMs) are effective for analyzing such complex data.
• Chemotherapy regimens, particularly in osteosarcoma, involve intricate, dynamically adjusted treatments influenced by patient toxicity.

Purpose of the Study:

• To propose an innovative marginal structural model (MSM) incorporating a dose-delay joint-exposure model.
• To estimate the causal effect of altering chemotherapy intensity (dosage and delay) on patient outcomes.
• To address the clinical question of whether chemotherapy dosages can be safely reduced.

Main Methods:

• Development of a novel marginal structural model (MSM) with a joint dose-delay exposure model.
• Application of Inverse-Probability-of-Treatment Weighted (IPTW) estimation.
• Utilizing data from a randomized trial of chemotherapy for osteosarcoma, focusing on toxicity-driven treatment modifications.

Main Results:

• The proposed MSM effectively handles the complexity of time-dependent confounding in chemotherapy data.
• Analysis indicates potential feasibility for dosage reduction in certain chemotherapy regimens.
• The model highlights that delaying therapy administration is associated with adverse outcomes and should be avoided.

Conclusions:

• The developed MSM provides a robust framework for causal inference in dynamic treatment regimes.
• Findings support exploring dosage reduction strategies while emphasizing the critical importance of adhering to planned treatment schedules.
• The study underscores the need to consider both dose and delay when evaluating chemotherapy modifications.