NAFLD is one of the most prevalent liver diseases, especially in Western countries, making it one of the biggest health care challenges [3]. Despite its prevalence, noninvasive tests are currently lacking, while the present gold standard, liver biopsy, is rarely performed [6]. Based on the study of Coilly et al., MMP9 and FABP4 levels appeared to be promising candidates as serum markers for NAFLD [8]. The biobank offered the unique opportunity to simultaneously assess tissues, as well as serum markers. Histological examinations of liver tissue were performed to accurately diagnose NAFL/NASH and ELISAs for MMP9 and FABP4 levels in serum samples. Significantly elevated circulating MMP9 levels in patients with NASH based on the NAS were detected. In agreement with this, the NAS and MMP9 levels showed a positive correlation. Higher fibrosis scores were not associated with higher MMP9 levels.
Comparisons with other studies and what does the current work add to the existing knowledge
Previously, two reports showed conflicting data for MMP9 levels in NAFLD-related liver fibrosis. Goyale et al. reported a decrease in circulating MMP9 levels in patients with more advanced fibrosis; however, this report lacked liver histology; therefore, it provides little evidence of MMP9 levels in patients with NASH [16]. Yilmaz et al. and Eren et al. did not observe a correlation between MMP9 levels and fibrosis scores [17]. In agreement with the later report, this study also did not find an association between MMP9 levels and fibrosis scores. However, only five patients had fibrosis scores of three or four. Therefore, the evidence remains limited for more advanced fibrosis stages.
In line with the present data, Yilmaz and Eren reported increased MMP9 levels in patients with NAFLD. They also performed liver biopsies but were likewise unable to distinguish patients with NAFL and NASH based solely on circulating MMP9 levels [17]. Nevertheless, data of this study provide evidence that MMP9 levels remain an attractive biomarker in NAFLD/NASH and that its combination with other markers is promising.
Elevated serum FABP4 levels in patients with obesity were previously reported [23]. However, data on circulating FABP4 levels in patients with NASH are scarce. Suh and colleagues reported increased serum FABP4 levels in patients with NAFLD. However, the included patients had normal weight or were overweight. Additionally, they did not perform liver biopsy [18]. Similar findings were reported recently, but this study also did not include liver biopsy [24]. This study shows that serum FABP4 levels were not increased in patients with NASH or fibrosis. It also did not detect a correlation between FABP4 levels and the NAS. Several possible reasons exist for this lack of correlation. One factor could be the source of FABP4. FABP4 is mainly expressed within adipose tissue. In obesity, FABP4 expression increases in the liver and decreases in the adipose tissue [23]. However, the increased expression in the liver might not contribute enough to the circulating levels of FABP4. Therefore, FABP4 is not the ideal candidate for noninvasive NASH diagnosis in patients with obesity.
This study was the first study, which investigated both MMP9 and FABP4 as markers for NASH. Furthermore, it investigated whether other routinely measured clinical parameters might improve the value of MMP9 to predict NAS. Hence, other clinical parameters were correlated with the NAS and chose the most promising candidates. Several markers associated with NAS, such as ALT, AST, HbA1c and MMP9 were identified. The developed models reached a fair discrimination with good AUC. Among all the different methods, the neural network fit best with the dataset, with an AUC of 83% and an accuracy of 88%. The AUC and accuracy of the neural network model showed that this test has the potential to be the starting point for future research. Other promising candidates to include in future analysis are cytokeratin-18 fragments (CK18) and soluble Fas [25]. Both are markers of apoptosis and could be a useful addition to MMP9, which is a marker of tissue remodeling [26].
Strengths and limitations
The present study has several strengths and is the first to investigate the combination of circulating MMP9 and FABP4 concentrations as noninvasive tests for NASH. The strengths of this study include the paired biopsy and serum samples, as well as the robust and innovative model development.
Nevertheless, this study suffers from some limitations. The number of patients was relatively small, with only 84 patients included. However, even with a larger number of patients, it is not expected to find differing results. Additionally, an external validation of our findings is needed. Since patients with diabetes had higher MMP9 levels and higher NAS values, this study could not safely rule out that presence of diabetes influences MMP9 concentrations. However, MMP9 levels and HbA1c concentrations did not correlate with each other, therefore it is unlikely that diabetes severity influences MMP9 concentrations. Furthermore, all patients in this study underwent bariatric surgery. Hence, it is only possible to draw conclusions for patients with obesity.