Accurate Detection of Metabolically Active “Brown” and “White” Adipose Tissues with Computed Tomography

Abstract

Adipose tissues (AT) are highly metabolically active complex endocrine organs and are classified into white (WAT) and brown AT (BAT) with proinflammatory and anti-inflammatory characteristics. The current study investigated the accuracy of computed tomography (CT) to quantitatively detect BAT and WAT based on Hounsfield unit (HU) threshold compared to standardized uptake values (SUVs) of corresponding AT with the use of positron emission tomography (PET). One hundred twenty-four patients who underwent whole-body (18)F-fluorodeoxyglucose ((18)F-FDG) PET/CT were studied. The SUVs and HUs of regions of BAT and WAT were measured and compared in these scans. The receiver operator characteristic (ROC) analysis was used to detect the HU threshold values for the detection of BAT and WAT. The CT HU of BAT, with high (18)F-FDG uptake, was significantly higher compared to WAT (-67.1±8.2 [-10 to -87] versus -93.5±10.1 [-88 to -190], P=.0006). Generalized linear-regression models revealed that mean CT HU of BAT was 26.4 HU higher than that of WAT (P=.001). A strong agreement between CT HU and PET SUV in measuring metabolically active AT (r(2)=0.81, P=.0001) was observed. The ROC curve showed that the optimal cutoff value of CT HU to detect BAT was HU≥-87 with sensitivity of 83.3% and specificity of 100%, and the negative CT HU below -87 is highly specific for WAT. In our study, CT can accurately and quantitatively detect BAT and WAT based on CT HU threshold, which is -10 to -87 forBAT and -88 to -190 for WAT.