In Reply: We agree with Dr Stefan and colleagues that the epidemiological evidence for an association between hepatic steatosis and metabolic disease is both robust and intriguing; however, the causal nature of this relationship remains poorly defined. Therefore, we tested the association between baseline liver fat (measured by magnetic resonance spectroscopy) and the incidence of prediabetes and diabetes over 7 years of follow-up in our obese cohort.
When included in the final multivariable models, percentage of liver fat was not associated with incident diabetes in the overall cohort (odds ratio [OR] per 1-SD increase in liver fat, 1.3; 95% CI, 0.8-1.9) or the composite of prediabetes or diabetes among those with normal fasting glucose at baseline (OR, 1.1; 95% CI, 0.8-1.4). Additionally, the χ2 value for liver fat was weaker than for visceral adiposity (1.7 vs 10.8).