Liver fibrosis quantified by image morphometry predicts clinical outcomes in patients with non-alcoholic fatty liver disease

Abstract

Background and aims Liver fibrosis predicts adverse clinical outcomes, such as liver-related death (LRD) and hepatocellular carcinoma (HCC) in patients with non-alcoholic fatty liver disease (NAFLD). We aimed to investigate the accuracy of semi-automated quantification of collagen proportionate area (CPA) as an objective new method for predicting clinical outcomes. Method Liver biopsies from patients with NAFLD underwent computerized image morphometry of Sirius Red staining with CPA quantification performed by ImageScope. Clinical outcomes, including total mortality, LRD, and combined liver outcomes (liver decompensation, HCC, or LRD), were determined by medical records and population-based data-linkage. The accuracy of CPA for predicting outcomes was compared with non-invasive fibrosis tests (Hepascore, FIB-4, APRI). Results A total of 295 patients (mean age 50 years) were followed for a median (range) of 9 (0.2–25) years totalling 3253 person-years. Patients with CPA ≥ 10% had significantly higher risks for total death [hazard ratio (HR): 5.0 (1.9–13.2)], LRD [19.0 (2.0–182.0)], and combined liver outcomes [15.6 (3.1–78.6)]. CPA and pathologist fibrosis staging (FS) showed similar accuracy (AUROC) for the prediction of total death (0.68 vs. 0.70), LRD (0.72 vs. 0.77) and combined liver outcomes (0.75 vs. 0.78). Non-invasive serum markers Hepascore, APRI, and FIB-4 reached higher AUROC; however, they were not statistically significant compared to that of CPA except for Hepascore in predicting total mortality (0.86 vs. 0.68, p = 0.009). Conclusion Liver fibrosis quantified by CPA analysis was significantly associated with clinical outcomes including total mortality, LRD, and HCC. CPA achieved similar accuracy in predicting outcomes compared to pathologist fibrosis staging and non-invasive serum markers.