Here, we developed a novel index for the non-invasive evaluation of steatosis using metabolic stress biomarkers along with NAFLD-related parameters. Compared with previously suggested indices, the MSI-S showed superior predictiveness for hepatic steatosis. Of note, FGF21, as a mitochondrial stress indicator, has high association with fat content and contributed significantly to the high predictive power of this index. These results emphasize the importance of mitochondrial stress in the progression of NAFLD into serious steatosis.
In the present study, quantitative analysis of liver fat content was performed using MRI-PDFF, which is considered the most accurate non-invasive diagnostic modality for steatosis . MRI-PDFF represents the proportion of the mobile proton density attributable to fat composition, providing objective evaluation of the amount of hepatic fat. Even though MRI-PDFF is reasonably precise and is used as reference standards, its high cost, time-consuming processes, and unavailability in many global regions preclude widespread applicability in the general population, especially for screening large cohorts.
Major biomarkers in our index are FGF21, FGF19 and A/L, which have not been used in previous biomarker panels for the prediction of fatty liver. FGF21 is upregulated by physiologic and pathophysiologic stresses. In a pathologic condition such as metabolic syndrome, FGF21 increase to compensate for oxidative stress, ER stress, and mitochondrial dysfunction; thus, higher serum levels reflect uncompensated metabolic stress . In our study, FGF21 correlated with waist circumference, triglycerides, total cholesterol, γ-GT, and the homeostatic model assessment of insulin resistance, which are identified as the main risk factors of NAFLD. FGF21 was one of the strongest independent predictors for the severity of steatosis.
FGF19, mainly expressed in the small intestine, regulates bile acid synthesis and nutrient metabolism. Additionally, FGF19 has proliferative and anti-apoptotic effects on the liver; therefore, low serum FGF19 levels are associated with hepatocyte injury and cell death . Decreased circulating FGF19 levels have been reported in obese and insulin-resistant patients . Consistently, in our study, FGF19 showed negative correlation with steatosis and enhanced the predictiveness of MSI-S, demonstrating the second highest influence on FGF21.
Hypoadiponectinemia and hyperleptinemia are considered common laboratory findings in NAFLD [28, 29]. However, the clinical applicability of the ratio between adiponectin and leptin has not been extensively validated. In our univariate analyses, the A/L ratio had a markedly higher Wald score (42.74) than that of adiponectin (27.39) or leptin (21.57). Furthermore, this ratio was able to discriminate between severe and mild steatosis, whereas adiponectin or leptin alone were not (Supplementary Fig. S1).
Despite significant correlation with steatosis, several factors such as IL-6, were not used in our index (MSI-S). IL-6 is a well-known inflammatory cytokine, mirroring the severity of steatohepatitis . In this study, invasive diagnosis of NASH was not performed; thus, we were unable to evaluate the precise inflammatory status of the NAFLD patients. Instead, serum IL-6 levels had significant correlations with liver fat content (Table 1 and Supplementary Table S5), and increased IL-6 reflected the severity of steatosis (Supplementary Fig. S1). These results suggest the possible contribution of inflammation to the progression of steatosis.
A positive correlation between central obesity and steatosis was detected in this study, compatible with the pathogenic importance of visceral fat accumulation. Systemic inflammation by visceral obesity is thought to accentuate the progression of NAFLD. Prospective long-term studies in patients with simple liver steatosis are required to elucidate the parameters of the NAFL-NASH-cirrhosis progression in this population.
One striking advantage of MSI-S was its effectiveness in reducing unnecessary further investigations including liver biopsy and MR studies. Compared to currently available indices for hepatic steatosis, in MSI-S the populations of patients with ‘indeterminate’ scores were markedly less, as shown in Fig. 4 and Supplementary Table S4. We suggest that MSI-S is an efficient tool for the prediction of steatosis in clinical settings where MR equipment or invasive diagnostic methods are inaccessible.
A limitation of our study was that the validation of steatosis for the general cohort was MR image-based evaluation, not invasive liver biopsy. In fact, it is ethically unfeasible to obtain biopsy samples from healthy cohort subjects. However, as described above, MR-based approaches have advantages in estimating the overall conditions of the liver. Furthermore, we also validated the predictability of MSI-S with liver biopsy-evaluated patients, which was superior to other indices. As another limitation, this study did not encompass an assessment of the wide range of severities of liver steatosis, and this is an opportunity for further validation in larger scale clinical studies.
In conclusion, we suggest that biomarkers based on the pathophysiologic mechanisms of NAFLD could have predictive power in steatosis, inflammation, and fibrosis. Each biomarker may not, by itself, be suitable as an independent predictor, but integrative interpretation with multivariate logistic regression analyses results in a more reliable index, effective for the non-invasive diagnosis of NAFLD progression. Inclusion of mitochondrial stress marker in the algorithm for steatosis markedly enhanced predictive performance, resulting in more accurate and precise index than other existing scoring systems. As the next step, the MSI-S should be validated in an external population and its prognostic utility needs to be confirmed by other prospective cohort studies.