ABSTRACT
BACKGROUND/OBJECTIVES: It has been unknown if attenuated insulin sensitivity (Si) in non-alcoholic fatty liver disease (NAFLD) is a cause or a result. We examined the impact of attenuated Si on NAFLD evolution. SUBJECTS/METHODS: We observed 4856 NAFLD- and diabetes-free participants for a mean 2.9 years. Si was indexed by single point insulin sensitivity estimator (SPISE = [600 × HDL-c0.185]/[TG0.2 × BMI1.338]), correlating with 1/HOMA-IR in an independent cohort (n = 1537, Spearman rho = 0.519, P < 0.01). Fatty liver (FL) was diagnosed by ultrasonography and diabetes by fasting plasma glucose (FPG) ≥ 7 mmol/L and/or glycohemoglobin A1c ≥ 6.5%. Multinominal comparison was performed with incident FL (FLw/oDM, n = 486), diabetes (DMw/oFL, n = 171), and FL plus diabetes (FL/diabetes, n = 58) as targets; none of the above (n = 4,138) was the control. SPISE was taken as a predictor with adjustment for covariates. Trajectory of SPISE during the 5 years before development of each condition was also assessed. RESULTS: With SPISE tertile 3 (>10.06) as the reference, tertile 1 (<8.07) was related to incident FLw/oDM and FL/diabetes with OR (95% CI) 3.47 (2.60-4.63) and 1.78 (1.10-2.87), respectively, and tertile 2 (8.07-10.06) related to FLw/oDM with OR (95% CI) 1.38 (1.03-1.85). Low SPISE was not significantly related to incident diabetes. At -5 years, SPISE was 12% (P < 0.05) and 13% (P < 0.01) lower in those developed FLw/oDM and FL/diabetes, respectively, than the control. At year 0, SPISE in the two groups was 18% and 21% lower than the control, respectively (P < 0.01). CONCLUSIONS: Attenuation of Si indexed by SPISE was a risk factor for NAFLD.
