IS IT POSSIBLE TO DERIVE A RELIABLE ESTIMATE OF HUMAN VISCERAL AND SUBCUTANEOUS ABDOMINAL ADIPOSE-TISSUE FROM SIMPLE ANTHROPOMETRIC MEASUREMENTS

The aim of the study was to generate equations predicting visceral (VA T) and subcutaneous (SAT) abdominal adipose tissue (AT) from simple an thropometric measurements. Magnetic resonance imaging (MRI) was used t o measure VAT and SAT cross-sectional areas at the level of L4 in 49 s ubjects (19 men and 30 women) with a large range of age and body mass index (BMI). BMI, waist and hip circumferences, waist to hip ratio (WH R), subscapular and paraumbilical skinfolds tie. ''simple'' anthropome tric measurements), total body fat content by the isotope-dilution met hod, and abdominal sagittal diameter by MRI tie, ''nonsimple'' anthrop ometric measurements) were also measured. Equations to estimate VAT an d SAT from age and simple anthropometric measurements tie, excluding t otal body fat and abdominal sagittal diameter) were developed. These e quations were then used in 24 subjects (nine men and 15 women) to cros s-validate them. The best regression equations, including waist circum ference in men and waist circumference and age in women, explained 56% and 68% of VAT variability, respectively. The corresponding standard error of the estimate (SEE) int men was approximately 40% and in women approximately 37% of the mean value of VAT measured by MRI. The best regression equations developed to predict SAT had a higher explained v ariability (similar to 87% in both men and women) and a lower SEE (<20 % of the mean values of SAT measured by MRI). In men, the equation inc luded BMI and hip circumference, and in women, BMI and age. The inclus ion of a higher number of simple anthropometric parameters in the pred ictive models neither significantly increased the explained variabilit y of VAT or SAT nor significantly decreased the SEE of VAT or SAT. Als o, inclusion in the multiple regression analysis of total body fat con tent and abdominal sagittal diameter did not improve prediction. In th e cross-validation study, differences between predicted and observed v alues of VAT were large, with a tendency to overestimation in both men and women. In contrast, differences between predicted and observed va lues of SAT were small. We suggest that SAT but not VAT can be estimat ed from age and simple anthropometric measurements. Direct methods (MR I, computed tomography [CT], car other options) should be used for ass essment of VAT. Copyright (C) 1995 by W.B. Saunders Company