Diagnostic performance of diabetes biomarkers in non-alcoholic fatty liver disease

Abstract

PurposeNon-alcoholic fatty liver disease (NAFLD) and diabetes often occur together and share physiological pathways; also, NAFLD may be an independent risk factor for diabetes. Whether fasting plasma glucose (FPG), postprandial glucose (PPG) or HbA1c have the same diagnostic accuracy in people with and without NAFLD could inform specific screening recommendations for those with NAFLD. MethodsCross-sectional analysis of the Third National Health and Nutrition Examination Survey (NHANES III). Diabetes was defined as PPG ≥200 mg/dL, FPG ≥126 mg/dL, and HbA1c ≥6.5%. We estimated sensitivity and specific between the six possible pairwise combinations between the three diabetes definitions in people with and without NAFLD. With Poisson regressions, we investigated whether people with NAFLD were more likely to be diagnosed with two biomarkers yet missed with the third one. ResultsThere were 3652 people with mean age 55.6 years and 49.4% were men; 673 (18.4%) people met criteria for NAFLD. In people with NAFLD, compared to NAFLD-free individuals, the specificity was lower in all pairwise comparisons except when PPG was the reference compared with HbA1c. Without strong differences, the sensitivity of FPG was slightly superior to PPG and HbA1c. People with NAFLD were more likely to be diagnosed with FPG and PPG yet missed with HbA1c (Prevalence Ratio = 2.15; P = .020). ConclusionsWhile these three diabetes biomarkers may capture different patients both in people with and without NAFLD, in the NAFLD population FPG appears to have the best sensitivity and there were no strong differences between PPG and HbA1c in terms of specificity. AcknowledgmentsRMC-L is supported by a Wellcome Trust International Training Fellowship (Wellcome Trust 214185/Z/18/Z).