Dr. Hsu and colleagues in their letter raised two issues. The first was related to the validity of the fatty liver index (FLI) and its relationship with insulin resistance (IR). FLI was developed in a general population1 in which the gold-standard technique for the diagnosis of nonalcoholic fatty liver disease (NAFLD), liver biopsy, is neither feasible nor ethically acceptable. The limitations of ultrasonography are well known, but in epidemiology, there is no alternative.1 Although FLI includes parameters associated with IR, this index was developed as a predictor of fatty liver, not IR. We have demonstrated that FLI is strongly related to IR [measured with the insulin sensitivity index (M/I) during the euglycemic-hyperinsulinemic clamp].2 From the FLI formula, this was somehow expected (although not necessarily true). Nonetheless, NAFLD and nonalcoholic steatohepatitis are strongly related to IR, and they are now considered the hepatic manifestations of metabolic syndrome. Indeed, the ability of FLI to predict clinically important outcomes was virtually unmodified after correction for IR.2 In a multivariate model, independent determinants of intima media thickness (IMT) were M/I (P = 0.01), age, systolic blood pressure, low-density lipoprotein cholesterol, and gender (all P < 0.0001, adjusted R2 = 0.27). When FLI was added, it replaced M/I as an independent predictor of IMT (P = 0.0001, adjusted R2 = 0.27). Thus, FLI can be considered a good predictor of early carotid atherosclerosis.
The second issue was related to alanine aminotransferase (ALT) independent of IR. A recent American Association for the Study of Liver Diseases position statement3 cited some evidence that ALT may be a predictor of cardiovascular disease in the general population. However, the role of ALT as a predictor of nonalcoholic steatohepatitis and liver fibrosis in patients with NAFLD is more controversial. In a recent study, ALT was not a predictor of liver histology,4 and in the Dionysos study, most participants had normal liver aminotransferases, despite the presence of fatty liver.1 The case is different for subjects with diabetes and/or cardiovascular diseases, who often have higher liver enzymes. In the Relationship Between Insulin Sensitivity and Cardiovascular Disease (RISC) population, although IMT was correlated with ALT, the correlation coefficient was weak (r = 0.11, P < 0.0001) and much lower than the FLI index (r = 0.30, P < 0.0001).
In conclusion, our data show a moderate association between IR or ALT and cardiovascular disease, whereas FLI is strongly associated with fatty liver and also with early atherosclerosis.