Joy Tisha, Kennedy Brooke A, Al-Attar Salam, Rutt Brian K, Hegele Robert A
Robarts Research Institute and Schulich School of Medicine and Dentistry, University of Western Ontario, London, Ontario, Canada N6A 5K8.
Metabolism. 2009 Jun;58(6):828-34. doi: 10.1016/j.metabol.2009.03.001.
The objective of the study was to determine correlations between magnetic resonance imaging (MRI) measures of truncal adiposity (trunk fat percentage [TrF %(MRI)], visceral adipose tissue [VAT], and subcutaneous abdominal adipose tissue [SAT]), simple clinical measures (body mass index [BMI], waist circumference [WC], and waist-to-hip ratio [WHR]), and bioelectrical impedance analysis (BIA)-derived measures (total fat percentage [TF %] and TrF %(BIA)) in female patients with familial partial lipodystrophy (FPLD). Our secondary aim was to generate and cross-validate predictive equations for VAT and SAT using these simple clinical and BIA-derived variables. Measures of truncal adiposity were measured using 1.5-T MRI (VAT, SAT, and TrF %(MRI)) and Tanita (Tokyo, Japan) 8-electrode body composition analyzer BC-418 (TrF %(BIA)) in 13 female FPLD patients. Pearson correlation coefficients were determined among the various adiposity parameters (BMI, WC, WHR, SAT, VAT, TrF %(MRI), TrF %(BIA), and TF %). Equations to estimate VAT and SAT were determined among 6 of the 13 FPLD subjects using multilinear regression analysis, and the best equations were then cross-validated in the remaining 7 subjects. Variables entered into the model included age, BMI, WC, WHR, TrF %(BIA), and TF %. The TrF %(MRI) showed moderate correlation (r = 0.647, P = .02) with the TrF %(BIA), but the discrepancy between the 2 variables increased with increasing truncal adiposity. The strongest correlate for TrF %(MRI) was BMI (r = 0.886, P < .0001). Visceral adipose tissue was poorly associated with simple clinical measures of BMI, WC, and WHR, but was inversely correlated with TF %, TrF %(BIA), and SAT. The TF % was the strongest correlate for both SAT and VAT. Thus, the best regression equation for VAT included age, BMI, WC, and TF % (R(2) = 1.0), whereas that for SAT only included TF % (R(2) = 0.75). The corresponding standard error of the estimate for the predictive equations was approximately 0.03 % and 18.5 % of the mean value of VAT and SAT, respectively. In the cross-validation study, differences between predicted and observed values of SAT were larger than those of VAT. We conclude that, among female FPLD patients, (1) no simple clinical anthropometric measure correlates well with VAT, whereas BMI correlates well with SAT; (2) BIA measure of TF % most strongly correlated with both VAT and SAT; and (3) based on the cross-validation study, VAT but not SAT could be more reliably estimated using the regression equations derived.
本研究的目的是确定家族性部分脂肪营养不良（FPLD）女性患者躯干肥胖的磁共振成像（MRI）测量指标（躯干脂肪百分比[TrF %(MRI)]、内脏脂肪组织[VAT]和腹部皮下脂肪组织[SAT]）、简单临床测量指标（体重指数[BMI]、腰围[WC]和腰臀比[WHR]）以及生物电阻抗分析（BIA）衍生指标（总脂肪百分比[TF %]和TrF %(BIA)）之间的相关性。我们的次要目标是使用这些简单的临床和BIA衍生变量生成并交叉验证VAT和SAT的预测方程。使用1.5-T MRI（测量VAT、SAT和TrF %(MRI)）和Tanita（日本东京）8电极人体成分分析仪BC-418（测量TrF %(BIA)）对13例女性FPLD患者的躯干肥胖指标进行测量。确定了各种肥胖参数（BMI、WC、WHR、SAT、VAT、TrF %(MRI)、TrF %(BIA)和TF %）之间的Pearson相关系数。使用多元线性回归分析在13例FPLD受试者中的6例中确定估计VAT和SAT的方程，然后在其余7例受试者中对最佳方程进行交叉验证。纳入模型的变量包括年龄、BMI、WC、WHR、TrF %(BIA)和TF %。TrF %(MRI)与TrF %(BIA)呈中度相关（r = 0.647，P = .02），但随着躯干肥胖程度增加，这两个变量之间的差异增大。TrF %(MRI)的最强相关因素是BMI（r = 0.886，P < .0001）。内脏脂肪组织与BMI、WC和WHR等简单临床测量指标的相关性较差，但与TF %、TrF %(BIA)和SAT呈负相关。TF %是SAT和VAT的最强相关因素。因此，VAT的最佳回归方程包括年龄、BMI、WC和TF %（R(2) = 1.0），而SAT的最佳回归方程仅包括TF %（R(2) = 0.75）。预测方程的相应估计标准误差分别约为VAT和SAT平均值的0.03 %和18.5 %。在交叉验证研究中，SAT预测值与观察值之间的差异大于VAT。我们得出结论，在女性FPLD患者中，（1）没有简单的临床人体测量指标与VAT有良好的相关性，而BMI与SAT有良好的相关性；（2）BIA测量的TF %与VAT和SAT的相关性最强；（3）基于交叉验证研究，使用推导的回归方程可以更可靠地估计VAT而非SAT。
