Body mass index, adjuvant chemotherapy, toxicity, and survival in non-metastatic colorectal cancer: an individual participant data meta-analysis (OCTOPUS)

INTRODUCTION: We aimed to disentangle whether elevated body mass index (BMI) is directly associated with adverse survival in primary colorectal cancer (CRC), or indirectly through treatment-related mechanisms, e.g., dose-capping adjuvant chemotherapy (ACT) and toxicity, using individual participant data meta-analysis (IPD-MA) and causal inference approaches. METHODS: Using BMI from four CRC-ACT randomised trials [OCTOPUS], and two ACT adherence measures (average cumulative relative dose [ACRD]; average relative dose intensity [ARDI]), we performed two-stage random effects IPD-MA, assessing total (TE) and direct effects (DE) (excluding and including mediators, respectively) of pre-defined causal paths, with overall survival (OS) as primary outcome. FINDINGS: BMI increments was a significant ACRD reduction (-1.15% [95% CI -1.92, -0.38]), and ACRD 5% increments were associated with improved OS (HR 0.94 [0.91, 0.96]), implying possible adverse indirect effects; though not large enough to induce an adverse TE of BMI on OS (HR 0.98 [0.90, 1.07]). BMI-ARDI relationships were similar (TE -1.08% [-1.44, -0.72]), but ARDI-OS relationships were inverted (HR 1.05 [1.01, 1.09]). BMI showed no association with grade 3+ toxicity (OR 1.01 (0.91, 1.14)). However, toxicity was associated with worse OS (TE HR 1.37 [1.17, 1.61]), which attenuated on adjusting for ACRD (DE HR 1.20 [1.02, 1.41]), suggesting partial mediation via a significant toxicity-ACRD relationship (-10.37% [-11.77, -8.97]). CONCLUSION: Our study establishes possible adverse indirect effects of obesity on CRC survival, through treatment-selection, supporting full BSA-based ACT dosing.