Correlation Between Nonalcoholic Fatty Liver Disease Categories Defined by the NAFLD Activity Score and Pathologistʼs Diagnosis

Abstract

Purpose: Recently, the nonalcoholic steatohepatitis (NASH) clinical research network designed the nonalcoholic fatty liver disease (NAFLD) activity score (NAS), a semiquantitative scoring system to be used for evaluating histological changes after therapeutic intervention trials. This score is currently being used widely for categorization of NAFLD patients in a host of clinical studies. However, the relationship between the NAS and the overall diagnostic determination has not been validated outside of the NASH clinical research network. Our aim was to examine the relationship between the NAFLD categories derived by the NAS and the pathologist's diagnostic determination. Methods: Our cohort consisted of 103 consecutive patients undergoing a liver biopsy for clinically suspected NAFLD. H&E and Masson's trichrome were evaluated by expert liver pathologists who provided an overall diagnostic interpretation and also reported a NAS for each patient. The NAS were categorized as: normal (NAS=0), steatosis (NAS=1-2), borderline (NAS=3-4), and NASH (NAS ≥5). Agreement between NAS and the overall diagnostic determination was assessed with the kappa (κ) coefficient. Cohen's κ was used for binomial classification of NAS (NASH vs. Not NASH) and a weighted κ coefficient was used otherwise. Results: Mean age of patients was 49.3 (±11.5) years; 49.5% were female and 82.5% were Caucasian. An overall diagnostic determination of NASH was given to 40.8% of patients (Table 1). Observed agreement between the assigned NAS categories and the diagnostic determination was 86% (34% agreement was expected by chance alone) with a κ of 0.80 (95% CI: 0.66, 0.93) (excellent agreement). When NAFLD was dichotomized into NASH (NAS ≥5) and Not NASH (NAS <5), the observed agreement was 90.3% with a κ of 0.79 (95% CI: 0.67, 0.91). Lowering the NAS cutoff for categorization as NASH from 5 to 4 significantly increased the sensitivity and negative predictive value of the NAS score for the diagnosis of NASH (Table 2). Out of the 13 patients with NAS 3 or 4 (borderline category), 4 were diagnosed as steatosis (all had NAS 3), while 9 patients were diagnosed as NASH: 3 with a NAS of 3 and 6 (67%) with a NAS of 4 (P=0.033). Patients with NAS borderline scores (3 or 4) and a diagnostic determination of NASH had significantly higher fibrosis stage than those with a borderline score without a diagnosis of NASH (P=0.014). They also tended to have higher inflammation (P=0.07) and ballooning scores (P=0.083).Table 1: NAFLD Diagnosis: Histologic diagnosis vs. NASTable 2: NAS in the diagnosis of histologic NASHConclusion: Our results demonstrate a high level of agreement between categorization of NAFLD cases using the NAS and the pathologist's diagnostic determination. Use of the NAS will allow greater comparison of biopsy specimens for both therapeutic and diagnostic studies in NAFLD.