Non alcoholic fatty liver disease and risk of incident diabetes in subjects who are not obese

Background and aims

It is not known whether non alcoholic fatty liver disease (NAFLD) is a risk factor for diabetes in non obese, non centrally-obese subjects. Our aim was to investigate relationships between fatty liver, insulin resistance and a biomarker score for liver fibrosis with incident diabetes at follow up, in subjects who were neither obese nor centrally-obese.

Methods and results

As many as 70,303 subjects with a body mass index (BMI) < 25 kg/m² and without diabetes were followed up for a maximum of 7.9 years. At baseline, fatty liver was identified by liver ultrasound, insulin resistance (IR) by homeostatic model assessment of insulin resistance (HOMA-IR) ≥2.0, and central obesity by waist circumference (waist circumference ≥90 cm (men) and ≥85 cm (women). The Fibrosis-4 (FIB-4 score) was used to estimate extent of liver fibrosis. Cox proportional hazards models adjusted for confounders were used to estimate hazard ratios (aHRs) for incident diabetes. As many as 852 incident cases of diabetes occurred during follow up (median IQR] 3.71 2.03] years). Mean ± SD BMI was 22.8 ± 1.8 and 21.7 ± 2.0 kg/m² in subjects with and without diabetes at follow up. In subjects without central obesity and with fatty liver, aHRs (95% CI) for incident diabetes at follow up were 2.17 (1.56, 3.03) for men, and 2.86 (1.50,5.46) for women. Similar aHRs for incident diabetes occurred with fatty liver, IR and the highest quartile of FIB-4 combined, in men; and there was a non significant trend toward increased risk in women.

Conclusions

In normal weight, non-centrally obese subjects NAFLD is an independent risk factor for incident diabetes.