| Abstract: | BACKGROUND: Non-alcoholic fatty liver disease (NAFLD) is the commonest hepatic disease in many parts of the World, with particularly high prevalence in patients with type 2 diabetes (T2DM). However, a good screening test for NAFLD in T2DM has not been established. Insulin resistance (IR) has been associated with NAFLD, and homeostatic model assessment of insulin resistance (HOMA-IR), a good proxy for IR, may represent an affordable predictive test which could be easily applied in routine clinical practice. We aimed to evaluate the diagnostic accuracy of HOMA-IR for NAFLD in T2DM and sought to estimate an optimal cut-off value for discriminating NAFLD from non-NAFLD cases. METHODS: We conducted a retrospective analysis of 56 well-controlled patients with T2DM (HbAc1<7%, on oral anti-diabetic and/or glucagon-like peptide-1 agonist treatment), who had at least one glucose and insulin level determined, and at least one hepatic imaging test (ultrasonography or computed tomography scanning). RESULTS: The prevalence of NAFLD was 73.2% (95% CI: 59.7-84.2) in our population. An association between HOMA-IR and NAFLD was found (OR 1.5; 95% CI: 1.03-2.1; p=0.033), independently of transaminases, fat percentage, BMI and triglyceride levels. The AUROC curve of HOMA-IR for identifying NAFLD was 80.7% (95% CI: 68.9-92.5). A value of HOMA-IR of 4.5 was estimated to be an optimal threshold for discriminating NAFLD from non-NAFLD cases. CONCLUSION: HOMA-IR is independently associated with the presence of NAFLD in adults with T2DM, and might potentially be applied in clinical practice as a screen for this condition. |
