MRI Assessment of Renal Lipid Deposition and Abnormal Oxygen Metabolism of Type 2 diabetes Mellitus Based on mDixon‐Quant

Background Diabetic nephropathy (DN) is the main cause of end‐stage renal failure. Multiecho Dixon‐based imaging utilizes chemical shift for water–fat separation that may be valuable in detecting changes both fat and oxygen content of the kidney from a single dataset. Purpose To investigate whether multiecho Dixon‐based imaging can assess fat and oxygen metabolism of the kidney in a single breath‐hold acquisition for patients with type 2 diabetes mellitus (DM). Study Type Prospective. Subjects A total of 40 DM patients with laboratory examination of biochemical parameters and 20 age‐ and body mass index ( BMI )‐matched healthy volunteers (controls). Field Strength/Sequence 3D multiecho Dixon gradient‐echo sequence at 3.0 T. Assessment The DM patients were divided into two groups based on urine albumin‐to‐creatinine ratio (ACR): type 2 diabetes mellitus (DM, 20 patients, ACR < 30 mg/g) and diabetic nephropathy (DN, 20 patients, ACR ≥ 30 mg/g). In all subjects, fat fraction (FF) and relaxation rate (R2*) maps were derived from the Dixon‐based imaging dataset, and mean values in manually drawn regions of interest in the cortex and medulla compared among groups. Associations between MRI and biochemical parameters, including β2‐microglobulin, were investigated. Statistical Tests Kruskal–Wallis tests, Spearman correlation analysis, and receiver operating characteristic (ROC) curve analysis. Results FF and R2* values of the renal cortex and medulla were significantly different among the three groups with control group < DM < DN (FF: control, 1.11± 0.30, 1.10 ± 0.39; DM, 1.52 ± 0.32, 1.57 ± 0.35; DN, 1.99 ± 0.66, 2.21 ± 0.59. R2*: Control, 16.88 ± 0.77, 20.70 ± 0.86; DM, 17.94 ± 0.75, 22.10 ± 1.12; DN, 19.20 ± 1.24, 23.63 ± 1.33). The highest correlation between MRI and biochemical parameters was that between cortex R2* and β2‐microglobulin ( r = 0.674). A medulla R2* cutoff of 21.41 seconds −1 resulted in a sensitivity of 80%, a specificity of 85% and achieved the largest area under the ROC curve (AUC) of 0.83 for discriminating DM from the controls. A cortex FF of 1.81% resulted in a sensitivity of 80%, a specificity of 100% and achieved the largest AUC of 0.83 for discriminating DM from DN. Data Conclusion Multiecho Dixon‐based imaging is feasible for noninvasively distinguishing DN, DM and healthy controls by measuring FF and R2* values. Evidence Level 2. Technical Efficacy Stage 2.