Body mass index does not impact survival in colorectal cancer: An individual participant data meta-analysis of 9333 patients from seven randomized controlled trials (OCTOPUS Consortium)


Year:

Session type:

Theme:

Lana Lai, Corinna Slawinski, Lee Malcomson, Richard Wilson, Richard Riley, Matthew Sperrin, Andrew G Renehan

Abstract

Background

While obesity is associated with increased cancer incidence, the role of elevated body mass index(BMI) in cancer survival is unclear. Relevant to survivorship strategies, however, results from published meta-analyses of the association between BMI and cancer survival are inconsistent. The approach of using aggregated data fails to take account of between-study heterogeneity in baseline risk and prognostic effects. To account for these limitations, we used an individual participant data(IPD) meta-analysis to assess the association between BMI and overall survival(OS) and disease-free survival(DFS) in patients with colorectal cancer(CRC).

Method

We systematically searched Pubmed, Embase, Cochrane and Web-of-Science from inception to 06/2020 to identify published randomised controlled trials of patients with non-metastatic CRC undergoing curative intent treatment (including surgery, radiotherapy and adjuvant chemotherapy), with derivable BMI at baseline. Primary and secondary outcomes were OS and DFS, respectively, estimated with a one-stage random-effects IPD meta-analysis. We included both linear(Cox regression) and non-linear(Cox regression fitted with a quadratic term) frailty models to explore the BMI-survival relationship, adjusted for the confounders of age, sex, stage and site.

Results

We included 9,333 patients from seven eligible trials who were willing to share IPD. Median BMI ranged from 24.7-26.6kg/m2, while median follow-up ranged from 2.8-7.4years across trials. Linear models suggested no evidence of association between BMI and either OS or DFS, with adjusted hazard ratios(aHRs) per kg/m2 of 0.99(0.98-1.01) and 1.00(0.99-1.01) respectively. In sex-stratified analyses, there was no evidence of association between BMI and OS(aHR: 0.98[0.97-1.00] for males; 1.00[0.98-1.02] for females) and DFS(aHR: 0.99[0.98-1.01] for males; 1.00[0.99-1.02] for females). There was a marginal U-shaped relationship for BMI and survival when allowing for non-linear effects, with lowest mortality risk at 32kg/m2.

Conclusion

Despite the established link between elevated BMI and higher incident CRC risk, our findings suggest that obesity is not associated with an inferior survival outcome in patients with non-metastatic CRC undergoing potentially curative surgery, although potential selection bias cannot be excluded. Weight management strategies in cancer survivorship programmes may still be relevant in improving other outcomes, such as non-cancer deaths.

Impact statement

While obesity is not associated with inferior survival outcome, weight-management strategies may still be relevant in improving non-cancer deaths.