Portal de Periódicos da CAPES

Diagnosing steatohepatitis and predicting liver‐related mortality in patients with NAFLD: Two distinct concepts

2011; Lippincott Williams & Wilkins; Volume: 53; Issue: 6 Linguagem: Inglês

10.1002/hep.24403

1527-3350

Paul Angulo,

Pancreatitis Pathology and Treatment

The diagnosis of nonalcoholic steatohepatitis (NASH) is based on both the presence of certain lesions (i.e., steatosis, inflammation, hepatocyte ballooning, and fibrosis) and the pattern of those lesions within the liver parenchyma in the absence of alcohol abuse. Over the last 2 decades, different criteria have been suggested for scoring and staging the histological lesions, and different definitions of NASH have been used in numerous NASH-related publications. The nonalcoholic fatty liver disease activity score (NAS) system, which has been proposed by the National Institute of Diabetes and Digestive and Kidney Diseases–sponsored NASH Clinical Research Network Pathology Committee, has gained enormous acceptance because of its simplicity and straightforwardness.1 The NAS system provides a numerical score for each histological lesion and allows the grading of steatosis, inflammation, and ballooning. Although fibrosis is not included in the NAS system, this system does contain a separate numerical score for staging fibrosis. Like other systems used to score liver biopsy features for other liver diseases, the NAS system was developed as a means of evaluating the changes in individual histological lesions that can occur over time either spontaneously or because of specific treatments during clinical trials.1 NAFLD, nonalcoholic fatty liver disease; NAS, nonalcoholic fatty liver disease activity score; NASH, nonalcoholic steatohepatitis In 1999, two important studies were published: Brunt et al.2 proposed a scoring system for grading the severity of NASH (mild, moderate, or severe), and Matteoni et al.3 proposed the classification of nonalcoholic fatty liver disease (NAFLD) into four subtypes based on combinations of liver lesions. Subsequently, data from several publications around the world have convincingly demonstrated that patients with confirmed NAFLD have a worse prognosis in comparison with the general population (matched by age and sex),4-7 and the prognosis of patients with NAFLD varies with the severity of the liver injury.8 The original definition proposed by the same investigators in 1999,3 which is based on steatosis plus hepatocyte ballooning, Mallory-Denk bodies, or fibrosis. The definition proposed by Younossi et al. in this publication, which is based on steatosis plus centrilobular ballooning, Mallory-Denk bodies, or fibrosis. An NAS ≥ 5.1 Brunt's definition, which is based on steatosis and lobular inflammation.2 The reported κ statistic (κ = 0.896) indicates almost perfect agreement between Younossi et al.'s current definition of NASH and the original definition; this is not surprising because the two definitions are essentially identical. The agreement between the current definition and an NAS ≥ 5 is moderate (κ = 0.511), and it remains in the moderate range even if the NAS threshold is reduced to 3. The agreement between the current definition and Brunt's definition is only fair (κ = 0.365). The agreement between an NAS ≥ 5 and Brunt's definition is less than fair with a κ value of only 0.178. Although these data are interesting, several issues that may affect their clinical relevance need to be discussed. First, the NAS system is not intended to categorize patients according to their NASH status; instead, it is meant to be used to evaluate the changes in individual histological features, as mentioned previously.1 Second, most experts in the field would agree that the presence of only steatosis and lobular inflammation in a liver biopsy sample (i.e., one of the NASH definitions used in this study) should not be called NASH, at least according to our understanding of the condition. Third, it is unfortunate that the investigators have not included in the agreement analysis another proposed definition of NASH, which separates patients into those with definitive NASH, those with borderline NASH, and those without NASH (according to a liver pathologist) and is based only on the severity and pattern of the liver injury, regardless of the NAS numerical value.1, 10 Finally, it is unclear whether the liver biopsy slides for the 257 patients were read all at once by both pathologists exclusively for this study or the scores of some biopsies were retrieved from older data sets. The second part of Younossi et al.'s study9 included 209 patients who had a median follow-up of approximately 12 years. Sixty-four of these patients (30.6%) died; 18 (8.6%) died because of liver-related causes. In adjusted multivariate Cox regression analyses, NASH according to the original definition with NAFLD subtypes3 and NASH according to the definition in this study correlated significantly with a higher liver-related mortality rate; the hazard ratios were 9.94 (95% confidence interval = 1.28-77.1, P = 0.03) and 4.43 (95% confidence interval = 0.97-20.2, P = 0.05), respectively. NASH according to the other definitions (NAS ≥ 51 and Brunt's criteria2) did not reach statistical significance. Additional multivariate Cox regression models were created to analyze the association of individual histological features with liver-related mortality. Using their own grading system, Younossi et al. report that portal fibrosis grade 3 (which included all patients with bridging fibrosis and cirrhosis) was the only histological lesion independently associated with liver-related mortality (hazard ratio = 5.68, 95% confidence interval = 1.5-21.45). When the same individual histological features were scored according to the NAS system1 and were analyzed by Cox regression analysis, only fibrosis stage 4 (cirrhosis) was independently associated with liver-related mortality (hazard ratio = 5.62, 95% confidence interval = 1.92-6.46). This part of the study on liver-related mortality provides important insights into our understanding of the long-term prognosis of patients with NAFLD. This study suggests that independently of any other histological lesions of NASH, the presence and severity of liver fibrosis dictate liver-related mortality in the long term. This finding agrees with a recent editorial highlighting the importance of liver fibrosis in predicting the long-term prognosis of patients with NAFLD, regardless of the presence and severity of other histological lesions.8 The lack of significance of an NAS ≥ 5 for predicting liver-related mortality can be explained by the fact that the NAS provides a numerical score for only three types of lesions: steatosis, lobular inflammation, and hepatocyte ballooning. Fibrosis is not part of the NAS; thus, roughly similar proportions of patients with fibrosis would be expected among those with an NAS ≥ 5 or an NAS < 5, as reported previously.6, 8 The same can be argued for the lack of significance of Brunt's NASH definition for predicting liver-related mortality; the presence and severity of fibrosis were requirements neither for making the diagnosis of NASH nor for classifying patients into a particular NASH grade.2 In addition, the fact that NASH according to the original definition with NAFLD subtypes3 and NASH according to the definition in this study9 independently correlated with liver-related mortality is simply due to the fact that increased liver fibrosis is included in both definitions of NASH (instead of NASH being diagnosed as an entity); neither hepatocyte ballooning nor Mallory-Denk bodies were associated with liver-related mortality when adjustments were made for the presence and severity of fibrosis in multivariate models. This study by Younossi et al.9 provides additional food for thought. Although the results of their investigation need to be reproduced and validated in larger long-term follow-up studies, the presented data suggest that according to the definition used, NASH may or may not be an indication of increased liver-related mortality. Thus, diagnosing NASH and predicting liver-related morality for patients with NAFLD are two distinct clinicopathological concepts that we should keep in mind when we are counseling patients with NAFLD about their risk for liver-related mortality. If we are going to keep in our thoughts and language the often alleged statement that "NASH is the progressive form of NAFLD," then increased liver fibrosis should be a key histological component in any definition of NASH. Considering all this, when we decide to perform liver biopsy in a patient suspected of having NAFLD and the liver pathologist makes a diagnosis of NASH, we should either ask for details about the meaning of NASH or simply ignore the NASH part of the diagnosis and pay close attention to whether or not the liver biopsy sample shows increased fibrosis and its stage. After all, according to this study by Younossi et al., the presence and severity of fibrosis rather than the diagnosis of NASH dictate liver-related mortality in the long run.