Association of body mass index and computed tomography-derived body composition with postoperative cardiovascular events: a secondary analysis of two prospective cohorts.

Abstract

Body mass index does not distinguish adiposity from muscle. We examined associations of BMI and CT-derived body composition phenotypes with postoperative cardiovascular events. This secondary analysis included patients aged ≥45 yr undergoing major noncardiac surgery. In the overall cohort, BMI associations with postoperative cardiovascular events were examined using multivariable logistic regression with restricted cubic splines. In patients with preoperative CT, body composition phenotypes were derived using k-means clustering. Associations between CT-derived body composition phenotypes and postoperative cardiovascular events were examined after adjustment for BMI category and prespecified covariates; model performance was compared before and after adding phenotype to the BMI category model. The overall cohort included 6320 patients. Lower BMI was associated with higher postoperative cardiovascular risk, with evidence of non-linearity (P<0.001; P for non-linearity=0.001) and heterogeneity by surgical type (P for interaction=0.005). The CT sub-cohort included 2399 patients. Three CT-derived phenotypes were identified: low muscle/low adiposity (Cluster 1), adiposity predominance (Cluster 2), and muscle predominance (Cluster 3). After adjustment for BMI category and prespecified covariates, Cluster 3 was associated with lower odds than Cluster 1 (adjusted odds ratio=0.59, 95% confidence interval=0.40-0.88; P=0.010). Adding body composition phenotype to the BMI category model improved overall model fit, but gains in discrimination and reclassification were modest. Lower BMI was associated with higher postoperative cardiovascular risk, with heterogeneity by surgical type. When preoperative CT is available, CT-derived body composition phenotyping may provide complementary information beyond BMI, although added predictive value appears modest.