Aim The association between sugar‐sweetened beverages and metabolic disorders has been well studied. However, it has not been determined whether fasting serum fructose is associated with metabolic dysfunction‐associated fatty liver disease (MAFLD). Methods Participants were enrolled from 2011 to 2012 in Shanghai. Fasting serum fructose concentration was measured with a validated liquid chromatography–tandem mass spectrometry method. Results A total of 954 participants without diabetes were included. They were followed for an average of 3.5 years. A total of 320 (33.5%) participants had MAFLD at baseline. With the increase in fasting serum fructose level by quartile, the MAFLD prevalence was increased by 27.0%, 25.0%, 37.4%, and 44.5%, respectively (p < 0.001). Each SD increase in fasting serum fructose level was associated with a 60% increased risk of MAFLD (odds ratio 1.60; 95% confidence interval [CI], 1.36–1.88; p < 0.001). Fasting serum fructose levels were more closely associated with four components of MAFLD (hepatic steatosis, prediabetes, insulin resistance, and low high‐density lipoprotein). We built a diagnostic model named the fructose fat index (FFI). The area under the receiver operating characteristic curve of the FFI was 0.879 (95% CI, 0.850–0.908) in the derivation cohort and 0.827 (95% CI, 0.776–0.878) in the validation cohort. Subsequent prospective studies found that the incidence risk of MAFLD was 2.26 times higher in the high‐fructose group than in the low‐fructose group among female participants (95% CI, 1.46–3.49; p < 0.001). Conclusion Fasting serum fructose concentration, which mostly reflects endogenous fructose, was associated with a higher risk of MAFLD. The FFI derived from fasting serum fructose could be used to predict MAFLD.

Background &amp; aimsThe effect of change in non-alcoholic fatty liver disease (NAFLD) status on incident diabetes has not been well studied. We aimed to investigate the association of NAFLD development and remission with the risk of incident diabetes during a median of 3.5-year follow-up.MethodsA total of 2690 participants without diabetes were recruited in 2011-2012 and assessed for incident diabetes in 2014. Abdominal ultrasonography was used to determine the change of NAFLD. 75 g oral glucose tolerance test (OGTT) was performed to determine diabetes. NAFLD severity was assessed using Gholam’s model. The odds ratios (ORs) for incident diabetes were estimated by logistic regression models.ResultsNAFLD was developed in 580 (33.2%) participants and NAFLD remission occurred in 150 (15.9%) participants during a median of 3.5-year follow-up. A total of 484 participants developed diabetes during follow-up, including 170 (14.6%) in consistent non-NAFLD group, 111 (19.1%) in NAFLD developed group, 19 (12.7%) in NAFLD remission group, and 184 (23.2%) in sustained NAFLD group. The development of NAFLD increased the risk of incident diabetes by 43% (OR, 1.43; 95%CI, 1.10-1.86) after adjustment for multiple confounders. Compared with sustained NAFLD group, remission of NAFLD reduced the risk of incident diabetes by 52% (OR, 0.48; 95%CI, 0.29-0.80). The effect of NAFLD alteration on incident diabetes was not changed after adjustment for body mass index or waist circumference, change of body mass index or waist circumference. In NAFLD remission group, participants with non-alcoholic steatohepatitis (NASH) at baseline were more likely to develop diabetes (OR, 3.03; 95%CI, 1.01-9.12).ConclusionsNAFLD development increases the risk of incident diabetes, whereas NAFLD remission reduces the risk of incident diabetes. Moreover, presence of NASH at baseline could attenuate the protective effect of NAFLD remission on incident diabetes. Our study suggests that early intervention of NAFLD and maintenance of non-NAFLD are important for prevention of diabetes.