Can the use of blood-based biomarkers in addition to anthropometric indices substantially improve the prediction of visceral fat volume as measured by magnetic resonance imaging?

Abstract

To investigate whether blood-based biomarkers can improve the prediction of visceral fat volume as measured by magnetic resonance imaging (MRI) and thus be used as proxies of visceral adiposity in large-scale epidemiological studies. Whole-body MRI was performed to determine overall and regional body compartments in 542 participants aged 48-80 years (52% men) of the Heidelberg cohort of the European Prospective Investigation into Cancer and Nutrition. Anthropometric measures were taken, and clinical chemistry profiles including 15 routine biomarkers were obtained. Furthermore, nine novel biomarkers of visceral fat were assayed in a discovery sample of 100 participants. Multivariable regression models were calculated to assess associations between anthropometric variables, biomarkers, and visceral fat volume. The proportion of variance in visceral fat volume explained by anthropometric measures was 65.2% in women and 60.8% in men. By using blood-based biomarkers in addition to anthropometric indices, the variance in visceral fat volume explained could be increased by 4.8% in women and 4.0% in men. After backward selection, HbA1c, triglycerides, and adiponectin remained in the final multivariable regression model in women, while in men hsCRP, leukocytes, AST (GOT), GGT, LDL, and adiponectin remained in the final model. In the present study, blood-based biomarkers moderately improved the prediction of visceral fat volume. This finding suggests that the underestimation of true associations between visceral fat and disease outcomes in epidemiological studies remains critical, even when using comprehensive sets of anthropometric and biomarker variables as proxies of visceral adiposity.