Discordant Liver Histology and NAFLD Fibrosis Scores in Morbidly Obese Bariatric Surgery Subjects: 2211

Introduction: Nonalcoholic fatty liver disease (NAFLD) afflicts over 100 million Americans. Liver biopsy is required to phenotype the spectrum of NAFLD into nonalcoholic fatty liver (NAFL) and nonalcoholic steatohepatitis (NASH). Liver biopsy also helps in staging liver fibrosis, an important determinant of liver disease outcomes. NAFLD fibrosis score (NFS, Angulo, 2007 Hepatology) was developed to accurately stratify subjects non-invasively into low and high fibrosis groups. However,the utility of this score in morbidly obese (BMI ≥40 kg/m2) subjects has not been well described. Methods: NAFLD fibrosis score was calculated as previously described by Angulo, 2007, Hepatology. All subjects who had liver biopsy during bariatric surgery, from 2005 to 2014 were included. Relevant clinical data were collected. The fibrosis on liver histology was staged from F0 to F4. Similarly, based on predefined cut offs, non-significant fibrosis (F0-F2) was NFS score < 0.676 and significant fibrosis (F3-F4) was NFS score >0.676. Statistical analyses including McNemar's test, simple Kappa coefficient and agreement plots were performed. Results: Combined data on liver histology and variables required to calculate NFS were available for 182 subjects. Based on liver histology 165 subjects (90.7%) had non-significant (F0-F2) and 17 subjects (9.3 %) had significant fibrosis (F3-F4). In contrast, 54 subjects (29.7 %) were categorized to have non-significant (F0-F2) fibrosis compared to 128 subjects (70.3 %) with significant fibrosis (F3-F4) based on NFS score cut offs. Only 49% subjects with non-significant (F0-F2) and 12% with significant fibrosis (F3-F4) were correctly matched based on NFS score and liver histology. The simple Kappa coefficient of agreement was very low (0.07) and the discordance in agreement was highly significant (p < 0.0001) based on McNemar's test. These findings are illustrated in Figure 1.Figure 1Conclusion: There was a marked discordance between fibrosis on liver histology and non-invasive NAFLD fibrosis score in morbidly obese bariatric surgery population. These findings clearly suggest poor predictability of currently available non-invasive tests. There is a pressing need to further explore other non-invasive methods and models that are readily available, affordable, easily accessible and reproducible to be considered as an alternative to liver histology in the morbidly obese population.