Predicting total, abdominal, visceral and hepatic adiposity with circulating biomarkers in Caucasian and Japanese American women.

Abstract

BACKGROUND:Characterization of abdominal and intra-abdominal fat requires imaging, and thus is not feasible in large epidemiologic studies. OBJECTIVE:We investigated whether biomarkers may complement anthropometry (body mass index [BMI], waist circumference [WC], and waist-hip ratio [WHR]) in predicting the size of the body fat compartments by analyzing blood biomarkers, including adipocytokines, insulin resistance markers, sex steroid hormones, lipids, liver enzymes and gastro-neuropeptides. METHODS:Fasting levels of 58 blood markers were analyzed in 60 healthy, Caucasian or Japanese American postmenopausal women who underwent anthropometric measurements, dual energy X-ray absorptiometry (DXA), and abdominal magnetic resonance imaging. Total, abdominal, visceral and hepatic adiposity were predicted based on anthropometry and the biomarkers using Random Forest models. RESULTS:Total body fat was well predicted by anthropometry alone (R(2) = 0.85), by the 5 best predictors from the biomarker model alone (leptin, leptin-adiponectin ratio [LAR], free estradiol, plasminogen activator inhibitor-1 [PAI1], alanine transaminase [ALT]; R(2) = 0.69), or by combining these 5 biomarkers with anthropometry (R(2) = 0.91). Abdominal adiposity (DXA trunk-to-periphery fat ratio) was better predicted by combining the two types of predictors (R(2) = 0.58) than by anthropometry alone (R(2) = 0.53) or the 5 best biomarkers alone (25(OH)-vitamin D(3), insulin-like growth factor binding protein-1 [IGFBP1], uric acid, soluble leptin receptor [sLEPR], Coenzyme Q10; R(2) = 0.35). Similarly, visceral fat was slightly better predicted by combining the predictors (R(2) = 0.68) than by anthropometry alone (R(2) = 0.65) or the 5 best biomarker predictors alone (leptin, C-reactive protein [CRP], LAR, lycopene, vitamin D(3); R(2) = 0.58). Percent liver fat was predicted better by the 5 best biomarker predictors (insulin, sex hormone binding globulin [SHBG], LAR, alpha-tocopherol, PAI1; R(2) = 0.42) or by combining the predictors (R(2) = 0.44) than by anthropometry alone (R(2) = 0.29). CONCLUSION:The predictive ability of anthropometry for body fat distribution may be enhanced by measuring a small number of biomarkers. Studies to replicate these data in men and other ethnic groups are warranted.