Clinical and histologic factors associated with discordance between steatosis grade derived from histology vs. MRI-PDFF in NAFLD.

BACKGROUND: Magnetic resonance imaging-proton density fat fraction (MRI-PDFF) is an excellent biomarker for the non-invasive quantification of hepatic steatosis. AIM: To examine clinical and histologic factors associated with discordance between steatosis grade determined by histology and MRI-PDFF in patients with non-alcoholic fatty liver disease (NAFLD) METHODS: We included 728 patients with biopsy-proven NAFLD from UC San Diego (n = 414) and Yokohama City University (n = 314) who underwent MRI-PDFF and liver biopsy. Patients were stratified by steatosis, and matched with MRI-PDFF cut-points for each steatosis grade: 0 (MRI-PDFF < 6.4%), 1 (MRI-PDFF: 6.4%-17.4%), 2 (MRI-PDFF: 17.4%-22.1%), 3 (MRI-PDFF ≥ 22.1%). Primary outcome was major discordance defined as ≥2 steatosis grade difference determined by histology and MRI-PDFF. RESULTS: Mean (±SD) age and BMI were 55.3 (±13.8) years and 29.9 (±4.9) kg/m2 , respectively. The distributions of histology and MRI-PDFF-determined steatosis were 5.5% grade 0 (n = 40), 44.8% 1 (n = 326, 44.8%), 33.9% 2 (n = 247), and 15.8% 3 (n = 115) vs. 23.5% grade 0 (n = 171), 49.7% 1 (n = 362), 12.9% 2 (n = 94), and 13.9% 3 (n = 101). Major discordance rate was 6.6% (n = 48). Most cases with major discordance had greater histology-determined steatosis grade (n = 40, 88.3%), higher serum AST and liver stiffness, and greater likelihood of fibrosis ≥2, ballooning ≥1 and lobular inflammation ≥2 (all p < 0.05). CONCLUSION: Histology overestimates steatosis grade compared to MRI-PDFF. Patients with advanced NASH are likely to be upgraded on steatosis grade by histology. These data have important implications for steatosis estimation and reporting on histology in clinical practice and trials, especially in patients with stage 2 fibrosis.

An optimized hepatocellular carcinoma prediction model for chronic hepatitis B with well-controlled viremia.

BACKGROUND AND AIMS: Hepatocellular carcinoma (HCC) risk in chronic hepatitis B (CHB) substantially decreased in the era of potent antiviral therapy. We developed an optimized HCC risk prediction model for CHB with well-controlled viremia by nucelos(t)ide analogs (NUCs). METHOD: We analysed those who achieved virological response (VR; serum HBV-DNA < 2000 IU/mL on two consecutive assessments) by NUCs. Liver stiffness by transient elastography, ultrasonography and laboratory tests was performed at the time of confirmed VR. Patients with decompensated cirrhosis or HCC at baseline were excluded. Multivariate Cox-regression analysis was used to determine key variables to construct a novel risk-scoring model. RESULTS: Among 1511 patients, 9.5% developed HCC. Cirrhosis on ultrasonography (adjusted HR [aHR] 2.47), age (aHR 1.04), male (aHR 1.90), platelet count <135 000/uL (aHR 1.57), albumin <4.5 g/dL (aHR 1.77) and liver stiffness ≥11 kPa (aHR 6.09) were independently associated with HCC. Using these, CAMPAS model was developed with c-index of 0.874. The predicted and observed HCC probabilities were calibrated with a reliable agreement. Such results were reproduced from internal validation and external validation among the independent cohort (n = 252). The intermediate-risk (CAMPAS model score 75 ~ 161) and high-risk (score >161) groups were more likely to develop HCC compared with the low-risk group (score ≤75) with statistical significances (HRs; 4.43 and 47.693 respectively; both P < .001). CONCLUSION: CAMPAS model derived through comprehensive clinical evaluation of liver disease allowed the more delicate HCC prediction for CHB patients with well-controlled viremia by NUCs.

Early on-treatment change in liver stiffness predicts development of liver-related events in chronic hepatitis B patients receiving antiviral therapy.

BACKGROUNDS/AIMS: Monitoring fibrosis is mandatory for detailed prognostification in patients with chronic liver disease. We developed optimized cut-offs for liver stiffness (LS) values, based on the histological subclassification of cirrhosis, and investigated whether early on-treatment changes in LS values can predict long-term prognosis in patients with hepatitis B virus (HBV)-related advanced liver fibrosis receiving antiviral therapy. METHODS: Between 2005 and 2008, 103 patients with F3 or F4 fibrosis on liver biopsy were enrolled prospectively. Cirrhosis was subclassified into three groups (F4A, F4B and F4C) according to Laennec system. The primary end-point was occurrence of liver-related event (LRE), including decompensation, hepatocellular carcinoma and liver-related death. RESULTS: Suggested LS cut-offs for predicting F4B-FC (vs. F3-F4A) and F4C (vs. F3-F4B) were 11.6 and 18.2 kPa respectively. As proportions of patients with LRE occurrence increased according to histological subclassifications stage F3-4A vs. F4B-4C (7.4% vs. 17.1%) and stage F3-4B vs. F4C (13.8% vs. 18.8%), they also increased according to LS cut-off value of 11.6 kPa (5.9% vs. 23.1%) and 18.2 kPa (9.8% vs. 33.3%) respectively (all P < 0.05). Similarly, according to stratified LS values (<11.6, 11.6-18.2 and ≥18.2 kPa), overall incidence of LREs and each constituent event increased significantly (all P < 0.05). In addition, the observed changes in LS values between baseline and 6 months of follow-up showed significant correlations with LRE development. CONCLUSIONS: Stratified LS values based on Laennec system and dynamic changes in LS values on follow-up may be helpful in assessing risk of LREs in subjects with HBV-related advanced liver fibrosis receiving antiviral therapy.