In the study of a non-diabetic, healthy Japanese population, we found that higher FIB-4 index is a risk for a subsequent decrease in insulin secretion. This implies that FIB-4 index may represent a predictor for the risk of developing DM in non-diabetic Japanese people. This finding is not consistent with those of previously published studies, in which NAFLD was shown to increase insulin resistance, with a compensatory increase in insulin secretion21,22, which is eventually followed by the development of glucose intolerance or DM when this compensatory mechanism fails5,6,7, while relationships of FIB-4 index with HOMA-β but not with HOMA-R were recently reported in a population-based cross-sectional study of Japanese23. Although we have no precise explanation for this inconsistency, differences in study samples may be responsible. The participants in the present study were non-diabetic, recruited from a health promotion study, and, thus, appear to be very healthy, as their BMI, HOMA-β, and HOMA-R were 22.6, 134.3, and 0.90, respectively (Table 1). This may explain why HOMA-R changed only modestly during the 3-year period of the study (Fig. 1 and Table 3) and, therefore, the relationship between FIB-4 index and the change in HOMA-R might not be properly evaluated. Nevertheless, because HOMA-β decreased significantly during the study period, the identified association between FIB-4 index and HOMA-β is likely to be more reliable. Furthermore, as previously described, Asian people show lower insulin secretion and resistance than other ethnicities13, and therefore the decreases in insulin secretion might be involved more substantially than an increase in insulin resistance in the pathogenesis of abnormal glucose metabolism, at least in the non-diabetic, non-obese, healthy Japanese population.
Although associations between NAFLD and incident DM have been frequently reported6,7, associations between the FIB-4 index and incident DM or abnormal glucose metabolism have rarely been reported. A longitudinal study of patients with biopsy-proven NAFLD showed that FIB-4 index did not predict the development of DM24. Furthermore, a longitudinal study of non-obese individuals also showed that FIB-4 index alone did not predict incident DM after adjustment for multiple factors25. NAFLD comprises two principal components, steatosis and fibrosis, and the FIB-4 index describes the severity of hepatic fibrosis, but not steatosis. Fat accumulation in the liver is an important risk factor for insulin resistance, and therefore DM21,26, which may explain why FIB-4 index has not been shown to be associated with incident DM previously. Although steatosis can be estimated by several indices such as Fatty liver index and Zhejiang University index27,28, we did not evaluated such indices here, and, thus, the above mentioned issue needs to be examined with such indices in the future.
Here, we have shown an association between FIB-4 index and a development of decreased insulin secretion, rather than with incident DM per se, as the number of subjects in the longitudinal study (n = 439) appeared to be small for such analyses (incidence of diabetes during the 3-year period was 9). Nevertheless, although not significant (p = 0.088), the subjects who had incident diabetes during the 3-year period had higher FIB-4 index at the baseline (1.46 ± 0.31) than the others (1.22 ± 0.59). As described previously, because the main feature of the pathogenesis of DM that is linked to NAFLD is insulin resistance, the impact of a decrease in insulin secretion on incident DM may not be substantial, and therefore may be missed in the studies with small sample size and/or of short duration. Nevertheless, the lack of association between FIB-4 index and the change in insulin resistance identified during the 3-year study period is, at least in part, in accordance with previous findings. To our knowledge, no previous study has aimed to determine the relationship between FIB-4 index and the change in insulin secretion or a decreased insulin secretion, which makes the present findings novel, although further larger and longer studies should be conducted in other, especially non-Asian, populations.
Although we here reported that the at-risk group defined based on the cut-off value, 1.592, of the FIB-4 index is a risk for a subsequent decrease in insulin secretion, the value itself seems to be tentative, since the clinical relevance of the value was not evaluated in this study. However, during the 3-year period of the study, the numbers of participants reached the endpoint in the at-risk and the not at-risk groups were 16.7% and 8.4%, respectively, and therefore the value may have, at least, some clinical relevance. Anyway, such values should be evaluated in consideration with clinical relevance in the future with much larger sample of general Japanese population with longer follow-up period.
The mechanisms underpinning the association between high FIB-4 index and a decreased insulin secretion are uncertain if insulin resistance is not the key mediator. High FIB-4 index may indicate that the patient has had NAFLD for longer, and has therefore been exposed to metabolic or other insults that affect insulin secretion for a longer period of time. For example, increased free fatty acids and their related compounds such as ceramides and diacylglycerols often observed in NAFLD can impair insulin secretion (known as lipotoxicity)7,29,30. Dietary energy restriction causes a reduction in the fat content of both in the liver and the pancreases, which can normalize β-cell function and hepatic insulin sensitivity in diabetic subjects31,32. Therefore, extent of steatosis of the liver or the severity of NAFLD may, at least in part, mirror the effects of pancreatic steatosis.
The present study had several strengths and limitations. With regard to its strengths, we studied a sample from the general population and accounted for multiple factors that could have confounded the statistical analyses. Furthermore, we conducted a longitudinal study, as well as a cross-sectional study, and could therefore evaluate the relationship between FIB-4 index and a subsequent development of decreased insulin secretion. Moreover, we excluded diabetic participants and those with FBG concentrations ≦ 63 mg/dl to permit more precise evaluation of the HOMA indices. Therefore, the results obtained should accurately reflect the relationship between FIB-4 index and HOMA indices. However, these exclusions, together with the recruitment of participants from a health promotion study, rather than from a population attending health checks, may also represent a limitation. Indeed, female proportion is substantially high (61.0%). Namely, the participants may have been more invested in their health than the general population, and therefore may not accurately represent the general population. In addition, the presence of NAFLD was not evaluated using any methods other than FIB-4 index, which is not a marker of liver steatosis, but of liver fibrosis alone17,33. Therefore, we evaluated the severity of NAFLD, but not its presence per se. Furthermore, although FIB-4 index has been developed to evaluate the liver fibrosis in the chronic liver disease including viral hepatitis and NAFLD and well validated by comparing it with findings of liver biopsy, it still remains unclear whether FIB-4 index represents liver fibrosis in the subjects without definite chronic liver disease, and therefore using FIB-4 index in the study with such healthier subjects may not be adequate. However, FIB-4 index was significantly correlated with γGTP in this study population (β = 0.090, p = 0.036, after adjustment with age and gender), and, thus, can be, at least, applied for this study. Alternatively, the observed association between higher FIB-4 index and a development of decreased insulin secretion may not be via liver fibrosis per se. Moreover, presence and/or incidence of any other liver diseases were not considered, as any markers other than AST, ALT, and γ-GTP activities or markers of viral hepatitis and autoimmune liver diseases were not measured, and therefore studies with such considerations are warranted. Besides, as described, since we did not determine the presence of NAFLD per se by any methods for the diagnosis such as ultrasonography and/or transient elastography, the results obtained cannot be regarded in association with NAFLD per se. Therefore, without such precise diagnosis of NAFLD, no mechanisms involved in the observed association can be precisely evaluated. However, the study is to evaluate association between FIB-4 index, not NFLD per se, and a subsequent development of decreased insulin secretion or DM, and, thus, the results per se seem to be, at least, reliable. Finally, the 3-year follow-up period may be short for this kind of longitudinal study. However, since the number of individuals who attended consecutive appointments becomes smaller if we chose longer follow-up period, we did not choose longer follow-up period to have sufficient statistical power. Nevertheless, in the 3-year follow-up period, significant difference between the groups was found, and, thus, the follow-up period seemed to be sufficient, at least, for the analysis.
In conclusion, FIB-4 index may represent a useful predictor of a subsequent development of decreased insulin secretion or DM, at least in a non-diabetic, apparently healthy Japanese population.