Prediction for steatosis in type-2 diabetes: clinico-biological markers versus 1H-MR spectroscopy

Abstract

The SteatoTest, fatty liver index (FLI) and hepatic steatosis index (HSI) are clinico-biological scores of steatosis validated in general or selected populations. Serum adiponectin (s-adiponectin) and retinol binding protein 4 (s-RBP4) are adipokines that could predict liver steatosis. We investigated whether the Steatotest, FLI, HSI, s-adiponectin and s-RBP4 could be valid predictors of liver steatosis in type-2 diabetic (T2D) patients. We enrolled 220 consecutive T2D patients. Reference standard was 3.0 T (1)H-MR spectroscopy (corrected for T1 and T2 decays). Intraclass correlation coefficients (ICCs), Kappa statistic measures of agreement, receiver operating characteristic (ROC) curves were assessed. Median liver fat content was 91 mg triglyceride/g liver tissue (range: 0-392). ICCs among the Steatotest, FLI, HSI, s-adiponectin, s-RBP4 and spectroscopy were low: 0.384, 0.281, 0.087, -0.297 and 0.048. Agreement between scores and spectroscopy was poor (Kappa range: 0.042-0.281). The areas under the ROC curves were low: 0.674, 0.647, 0.637, 0.616 and 0.540. S-adiponectin and s-RBP4 levels were strongly related to the presence of diabetic nephropathy (P = 0.0037 and P = 0.004; Mann-Whitney). The SteatoTest, FLI, HSI, s-adiponectin, s-RBP4 are not valid predictors of steatosis in T2D patients. Clino-biological markers cannot replace (1)H-MR spectroscopy for the assessment of liver fat in this population. (1) H-MR spectrosopy can reliably estimate the weight fraction of liver steatosis. Type-2 diabetes provides an interesting model for assessing liver steatosis. Clinico-biological markers seem to be invalid predictors for steatosis in type-2 diabetes.