Causal effects of chemotherapy regimen intensity on survival outcome in osteosarcoma patients through marginal structural Cox models

As patients develop toxic side effects, cancer treatment is adapted over time by either delaying or reducing the dosage of the next chemotherapy cycle. Being at the same time risk factors for mortality and predictors of future exposure levels, toxicities represent time-dependent confounders for the effect of chemotherapy on patient survival. In the presence of confounders, classical survival approaches have limitations in causally interpreting the hazard ratio of the treatment, even when randomized. The Intention-To-Treat approach is widely used in chemotherapy studies, although it is far from representing everyday clinical practice. Marginal Structural Cox Models (Cox MSM) in combination with Inverse Probability of Treatment Weighting (IPTW) are a proper tool to evaluate the causal effects of an exposure on survival outcomes. In this work, using novel definitions of Received Dose Intensity and Multiple Overall Toxicity, suitable IPTW-based techniques and Cox MSM were designed to mimic a randomized trial where chemotherapy joint-exposure is no longer confounded by toxicities. In this pseudo-population, a crude analysis is sufficient to estimate the causal effect of joint-exposure modifications. This paper discusses an innovative and detailed analysis of complex chemotherapy data, with tutorial-like explanations about the difficulties encountered and the novel problem-solving strategies deployed. This work highlights the confounding nature of toxicities and shows the detrimental effect of not considering them in the analyses. To the best of our knowledge, this is the first study combining different methodologies in an innovative way to eliminate the toxicity-treatment-adjustment bias in chemotherapy trial data.