The histological course of nonalcoholic fatty liver disease: a longitudinal study of 103 patients with sequential liver biopsies *

Transcription

1 Journal of Hepatology 42 (2005) The histological course of nonalcoholic fatty liver disease: a longitudinal study of 103 patients with sequential liver biopsies * Leon A. Adams 1, Schuyler Sanderson 2, Keith D. Lindor 1, Paul Angulo 1, * 1 Division of Gastroenterology and Hepatology, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, MN 55905, USA 2 Department of Laboratory Medicine and Pathology, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, MN 55905, USA See Editorial, pages Background/Aims: The histological course of nonalcoholic fatty liver disease (NAFLD) remains undescribed. Therefore, we examined the liver histology of NAFLD patients who had undergone sequential liver biopsies. Methods: Data on 103 patients who underwent serial liver biopsies in the absence of effective treatment were reviewed, and biopsies scored in a blind fashion. Results: Mean interval between biopsies was 3.2G3.0 years (range ). Fibrosis stage apparently progressed in 37%, remained stable in 34% and regressed in 29%. Severity of steatosis, inflammation, hepatocyte ballooning and Mallory s hyaline improved significantly. Aminotransferases decreased significantly between biopsies, paralleling improvement in steatosis and inflammatory features but not fibrosis stage. The rate of fibrosis change ranged from K2.05 to 1.7 stages/year. By multivariate analysis, diabetes (PZ0.007) and low initial fibrosis stage (P!0.001) were associated with higher rate of fibrosis progression, as was higher body mass index (PZ0.008) when cirrhotics were excluded. Conclusions: Fibrosis in NAFLD progresses slowly over time with considerable variability in the rate of changes among patients. Changes of aminotransferases do not parallel changes in fibrosis stage. Diabetic patients with elevated BMI and low fibrosis stage are at risk for higher rates of fibrosis progression. q 2004 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Keywords: Fibrosis; Biopsies; Progression; Cirrhosis; Diabetes; Obesity; Inflammation; Ballooning; Mallory s hyaline 1. Introduction Paralleling the rise in incidence of obesity and diabetes, nonalcoholic fatty liver disease (NAFLD) is emerging as one of the most common causes of chronic liver disease [1 3]. The disease is intimately related to insulin resistance and may progress to steatohepatitis (NASH) and cirrhosis with its complications [4 6]. However, it is uncertain what Received 19 July 2004; received in revised form 16 September 2004; accepted 21 September 2004; available online 8 October 2004 * Presented in part during the annual meeting of the American Association for the Study of Liver Diseases, Boston, MA 2003, USA. * Corresponding author. Fax: C address: angulohernandez.paul@mayo.edu (P. Angulo). Abbreviations BMI, body mass index; HDL, high density lipoprotein; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; SD, standard deviation. proportions of patients have progressive disease. In addition, the rate of disease progression or change in liver histological features over time is unknown. Thus, it remains unclear whether some factors predict higher rates of progression. Fibrosis stage is recognized as the most objective indicator of liver damage and is the best prognostic marker for morbidity and mortality in liver disease of various etiologies. Few studies have investigated the natural history of NAFLD by examining fibrosis stage among patients with paired liver biopsies, with the largest series including only 22 patients [7 10] Due to small numbers, conclusions remain limited. We, therefore, sought to evaluate individuals with well-defined NAFLD who had undergone serial liver biopsies during follow-up. Our aims were (1) to determine in a large number of patients, the histological course of NAFLD by analyzing /$30.00 q 2004 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. doi: /j.jhep

2 L.A. Adams et al. / Journal of Hepatology 42 (2005) the change in liver histology over time; and (2) to examine whether routinely determined clinical, laboratory or histological features predicted fibrosis progression. 2. Methods 2.1. Patients One hundred and three patients with NAFLD who had undergone more than one liver biopsy were identified from our master diagnostic index. These patients were seen at the Mayo Clinic Rochester between 1980 and The study was approved by the Mayo Institutional Review Board and all patients gave written informed consent for participation in medical research. The diagnosis of NAFLD was based upon: (1) steatosis involving at least 10% of hepatocytes on biopsy, (2) ethanol consumption of less than 140 g/week, (3) exclusion of patients with evidence of other liver disease using standard clinical, laboratory and histological criteria. Patients were also excluded if they had a secondary cause of NAFLD [1]. One patient had serial biopsies prior to 1990 without hepatitis C serology. He had no hepatitis C risk factors and his histology was not compatible with hepatitis C infection. Patients underwent a complete medical history, physical examination and imaging study. Laboratory parameters included aspartate aminotransferase (AST), alanine aminotransferase (ALT), bilirubin, alkaline phosphatase, albumin, prothrombin time, platelet count, total cholesterol, high density lipoprotein (HDL) cholesterol, triglycerides, glucose, iron studies (serum iron, ferritin, total iron binding capacity and transferrin saturation), hepatitis B and C serology, auto-antibodies, serum gamma globulins, ceruloplasmin, a1-antitrypsin level and phenotype. The homeostatic model assessment (HOMA) was calculated using the formula [11]: IRZ(insulin!glucose)/22.5; and the quantitative insulinsensitivity check index (QUICKI) using the formula [12]: QUICKIZ 1=½logðinsulinÞClogðglucoseÞŠ. The metabolic syndrome was defined using the criteria proposed by the National Cholesterol Education Program (ATP III), i.e., when at least three of the five following features were present [13] hyperglycemia (fasting blood glucose R110 mg/dl, or known diabetes mellitus), hypertension (blood pressure R130/R85 or under treatment), hypertriglyceridemia (R150 mg/dl or under pharmacological treatment), low-hdl cholesterol (!40 mg/dl for males and!50 mg/dl for females), and obesity (BMI R30 kg/m 2 ). As waist circumference was not measured for most of our patients, we substituted a BMI R30 kg/m 2 to define obesity [14]. Of the 103 patients, a repeat liver biopsy was performed as part of their medical follow-up in 26 patients and as part of a clinical trial in 77. These 77 patients included 50 participants in a placebo-controlled trial (27 randomized to placebo and 23 to ursodiol) [15], and 27 participants in a pilot study of clofibrate or ursodiol [16]. Both liver enzymes and histology were unchanged after a year of treatment with clofibrate [16] whereas changes in liver enzymes and histological features were identical among patients treated with ursodiol or placebo for 2 years [15] In addition, patients who received pharmacotherapy as part of a clinical trial, had the same change in histological features and rate of disease progression as those who did not receive pharmacotherapy (Table 1). Hence, as neither clofibrate nor ursodiol affected the liver condition, patients were pooled together for the purpose of this study. The time elapsed from first to last biopsy was not significantly different (PZ0.2) between the 77 participants in a clinical protocol and the 26 patients who had biopsies as part of the standard evaluation Liver histology Liver biopsy specimens were read under coded identification by a single liver pathologist who was unaware of the patient details or biopsy sequence. Biopsies were routinely stained with hematoxylin eosin, and Masson s trichrome. All biopsies were a minimum of 15 mm in length and had an appropriate number of portal tracts to make a confident evaluation of histological features and diagnosis [17]. Histological features were interpreted according to the schema outlined by Brunt et al. [18]. Briefly, steatosis was graded on a 3-point scale: grade 1Zsteatosis involving!33% of hepatocytes, grade 2Z33 66%, grade 3 O66%. Inflammation was graded on a 4-point scale: grade 0Zno or negligible inflammation, grade 1Zmild, grade 2Zmoderate, grade 3Zsevere. Fibrosis was staged on a 5-point scale: stage 0Zno fibrosis, stage 1Zzone 3 perisinusoidal/ perivenular fibrosis, stage 2Zzone 3 and periportal fibrosis, stage 3Zseptal/bridging fibrosis, stage 4Zcirrhosis. In addition, the following histological features were scored: hepatocellular ballooning (0Zabsent, 1Zmild, 2Zmarked); Mallory s hyaline (0Zabsent, 1Zoccasional, 2Zseveral); and hepatocellular iron (0 4Cas per Searle). Severity of lobular inflammation, hepatocellular necrosis, portal tract inflammation, pericellular fibrosis, portal fibrosis, and bridging fibrosis were also recorded and scored as described [18]. NASH was defined as either the presence of steatosis plus mixed lobular inflammation plus hepatocellular ballooning, as proposed during the AASLD single topic conference [19], or the presence of steatosis plus any stage of fibrosis. Steatosis plus either lobular inflammation or ballooning (but not both) was termed steatosis with nonspecific inflammation, whereas steatosis without lobular inflammation, ballooning or fibrosis was termed bland steatosis Statistical analysis Patients were divided into groups according to change in fibrosis stage between biopsies; either progressors (increased in fibrosis stage), stable (no change) and regressors (decreased in fibrosis stage). Fibrosis rate was Table 1 Change in liver histology was not different between patients on drug treatment compared to untreated patients Treated (ursodiol/clofibrate) Untreated (placebo/clinical biopsy) P value N Fibrosis stage Change between biopsies (meangsd) 0.04G G Regressors/stable/progressors (%) 30/36/34 28/32/ Steatosis grade Change between biopsies (meangsd) K0.4G0.8 K0.4G Regressors/stable/progressors (%) 50/38/12 46/42/ Inflammation grade Change between biopsies (meangsd) K0.2G0.7 K0.1G Regressors/stable/progressors (%) 28/60/12 23/62/ Ballooning grade Change between biopsies (meangsd) K0.3G0.6 K0.1G Regressors/stable/progressors (%) 33/63/4 25/61/ The proportion of patients who regressed (42 vs. 35%, PZ0.6), remained stable (23 vs. 38%, PZ0.2), or progressed (35 vs. 27%, PZ0.5) in fibrosis stage was not significantly different between patients biopsied for clinical reasons (nz26) and patients participating in clinical trials (nz77).

3 134 L.A. Adams et al. / Journal of Hepatology 42 (2005) calculated by dividing the difference in fibrosis stage between first and last biopsy, by the time between biopsies in years. Categorical data are presented as number (percentage). Continuous data are presented as meangstandard deviation (SD), and medians (range). Continuous variables were compared using standard parametric methods if normally distributed and nonparametric methods if not normally distributed. Frequency data were compared using chi-squared test or Fisher s exact test where appropriate. The McNemar s test was used to compare paired proportions. Spearman s rank correlation coefficient was used as a measure of association. Predictors of rate of progression were assessed using univariate and multivariate linear regression analysis. A two tailed P value of less than 0.05 was considered statistically significant. Analyses were performed using Systat Results 3.1. Patient characteristics The 103 patients underwent a total of 227 biopsies, from which only the initial and final biopsies were examined. The mean time interval between first and last biopsy was 3.2G3.0 years (range years). Baseline clinical and laboratory features of the patient population are summarized in Table 2. One patient was taking metformin before the initial biopsy and continued on the same dosage until the final biopsy. No patients were taking thiazolidinediones or vitamin E. One patient increased her alcohol consumption to an average gm/day between her two biopsies, although her histological features remained essentially unchanged. Histological features on initial biopsy are shown in Table 3. Ninety-six patients fulfilled criteria for Table 2 Baseline clinical and laboratory features of total cohort (nz103) MeanGSD or n (%) median (range) Age (years) 45G11, 47 (19 65) Sex (female) 65 (63%) Obese 69 (67%) Diabetic 43 (42%) Hypertensive 31 (30%) Hypertriglyceridemia 61 (59%) Low HDL 67 (65%) Metabolic syndrome 55 (53%) Alanine aminotransferase (U/l) 97G56, 82 (21 373) Aspartate aminotransferase (U/l) 75G50, 60 (13 258) AST/ALT ratio 0.9G0.5, 0.8 ( ) Bilirubin (mg/dl) 0.8G0.5, 0.7 ( ) Alkaline phosphatase (U/l) 216G77, 208 (70 626) Albumin (gm/dl) 4.3G0.5, 4.4 ( ) Platelets (!10 9 /l) 211G76, 212 (45 423) Prothrombin (s) 10.9G1.1, 10.8 ( ) Ferritin (mg/l) 280G275, 211 (6 1478) Glucose (mg/dl) 119G40, 102 (78 273) QUICKI a 0.33G0.04, 0.33 ( ) HOMA a 4.23G3.87, 2.93 ( ) Triglyceride (mg/dl) 232 G 233, 179 ( ) HDL cholesterol (mg/dl) 41G11, 40 (12 65) Metabolic syndrome defined as minimum of three of five criteria (obesity, diabetes, hypertension, low HDL, hypertriglyceridemia) [13,14]. a Available in 48 patients. Table 3 Histological features at baseline (nz103) Fibrosis Stage Inflammation Grade Steatosis Grade 0 25 (24%) 0 9 (9%) 1 40 (39%) 1 21 (20%) 1 72 (70%) 2 31 (30%) 2 23 (22%) 2 20 (19%) 3 32 (31%) 3 18 (18%) 3 2 (2%) 4 16 (16%) Ballooning Mallory s hyaline Hepatocellular Iron a 0 8 (8%) 0 62 (60%) 0 77 (85%) 1 86 (83%) 1 32 (31%) 1 7 (8%) 2 9 (9%) 2 9 (9%) 2 6 (7%) Numbers (%) of patients shown. a Iron stains not available for 13 patients. No patients had iron grade three or four. NASH, four had steatosis with nonspecific inflammation and three had bland steatosis Change in fibrosis stage A total of 38 patients (37%) increased in fibrosis stage between first and last biopsy; 35 patients (34%) did not change and 30 (29%) regressed (Table 4). Fourteen patients progressed by two stages or more and four patients progressed three stages or more. Nine patients progressed to cirrhosis, with two of these having no fibrosis on initial biopsy. Two thirds (67%) of early stage patients (stage 0 2) biopsied more than 4 years apart had progressive fibrosis, compared to 40% of patients biopsied within 2 years (Table 5). One quarter (24%) of all patients biopsied more than 4 years apart progressed by two or more stages, compared to 11% of patients biopsied within 4 years. Patients progressing three or more stages had a significantly longer biopsy interval compared to the remaining patients (12G8 vs. 3G2 years, PZ0.003). In particular, two patients progressed from no fibrosis to cirrhosis over 9.2 and 15.4 years each. No clinical or biochemical variables differed significantly between patients who regressed, remained stable or progressed in fibrosis stage (Table 6). Of the baseline histological variables, only fibrosis stage was significantly (inversely) associated with fibrosis progression (PZ0.003). Table 4 Change in fibrosis stage between first and last biopsy (nz103) Initial fibrosis stage Final fibrosis stage F0 F1 F2 F3 F4 F F F F F

4 L.A. Adams et al. / Journal of Hepatology 42 (2005) Table 5 Progression of fibrosis stage according to initial fibrosis stage and time interval between biopsies Biopsy interval!24 Months Months O48 Months Total cohort (nz103) Progressors 11 (34%) 18 (36%) 9 (43%) Stable 13 (41%) 13 (26%) 9 (43%) Regressors 8 (25%) 19 (38%) 3 (14%) Initial stage 0 2 (nz69) Progressors 10 (40%) 16 (50%) 8 (67%) Stable 9 (36%) 8 (25%) 3 (25%) Regressors 6 (24%) 8 (25%) 1 (8%) Initial stage 3 4 (nz34) Progressors 1 (14%) 2 (11%) 1 (12%) Stable 4 (57%) 5 (28%) 6 (67%) Regressors 2 (29%) 11 (61%) 2 (22%) The proportion of NASH patients who progressed was not significantly different from patients without NASH (34.4 vs. 53.8%, PZ0.2). Similar proportions of patients with progressive fibrosis were observed between those with and without progressive inflammation (50 vs. 35%, PZ0.4), progressive steatosis (25 vs. 38%, PZ0.5), and progressive ballooning (36 vs. 44%, PZ0.7) Rate of fibrosis progression To account for different time periods between biopsies, we calculated the rate of fibrosis change by dividing the difference in fibrosis stage between biopsies by the time interval (in years) between biopsies. Rate of fibrosis change varied from K2.05 to 1.70 stages/year with an overall mean rate of 0.02G0.66 stages/year. Seventeen patients progressed at greater than 0.5 stages/ year and 13 patients regressed by more than 0.5 stages/year. If cirrhotics were excluded (as they cannot progress), the rate of fibrosis change was 0.09G0.67 stages/year. By univariate linear regression analysis with rate of fibrosis change as the dependent variable, diabetes (PZ0.01), AST/ALT ratio (PZ0.02), steatosis grade (PZ0.05) and fibrosis stage (PZ0.003) were the only significant variables. These variables were then analyzed in a multivariate model and adjusted for age and BMI. Table 6 Clinical, biochemical and histological features among patients with progressive, stable or regressive fibrosis (nz103) Regressors Stable Progressors P value n (%) 30 (29%) 35 (34%) 38 (37%) Age (years) 46G2 47G2 44G2 0.4 a Sex (female) 20 (66%) 19 (54%) 25 (66%) 0.5 Obese 19 (63%) 22 (65%) 28 (74%) 0.4 Weight gain 14 (47%) 21 (60%) 25 (66%) 0.8 Diabetes 8 (27%) 15 (43%) 20 (53%) 0.1 Development of diabetes 2/22 (9%) 2/20 (10%) 4/19 (21%) 0.5 Hypertension 9 (30%) 10 (29%) 12 (32%) 0.8 Hypertriglyceridemia 20 (67%) 22 (65%) 19 (51%) 0.2 Low HDL 20 (74%) 24 (71%) 23 (66%) 0.6 Metabolic syndrome 16 (53%) 20 (57%) 19 (50%) 0.8 ALT (U/l) 91G57 94G51 104G AST (U/l) 80G71 74G44 71G AST/ALT ratio 1.0G G G Bilirubin (mg/dl) 0.7G G G Alk. Phos. (U/l) 213G79 209G47 226G Albumin (gm/dl) 4.4G G G Platelets (!10 9 /l) 206G67 197G72 228G a Prothrombin (s) 11.0G G G a Ferritin (mg/l) 214G G G Glucose (mg/dl) 117G41 125G45 113G QUICKI 0.32G G G HOMA 3.82G G G Triglyceride (mg/dl) 220G G G HDL (mg/dl) 42G9 40G11 42G a Steatosis grade 1.9G G G Inflammatory grade 1.3G G G Ballooning 1.1G G G Mallory s hyaline 0.6G G G Fibrosis stage 2.5G G G Clinical and laboratory measurements recorded at time of initial biopsy. Weight gain and development of diabetes documented between first and last biopsy. Metabolic syndrome defined as at least three of five criteria (obesity, diabetes, hypertension, low HDL, hypertriglyceridemia). a Continuous variables normally distributed and thus analyzed by ANOVA; remaining data were nonnormally distributed and thus analyzed by Kruskal Wallis test.

5 136 L.A. Adams et al. / Journal of Hepatology 42 (2005) Table 7 Predictors of fibrosis rate by multivariate linear regression analysis Variable Regression coefficient SE of regression coefficient F value P value Total cohort (nz103) AST/ALT ratio K Age Steatosis grade BMI Diabetes Fibrosis stage K !0.001 Excluding cirrhotics (nz87) AST/ALT ratio K Steatosis grade Age BMI Diabetes Fibrosis stage K Ursodiol was not significantly associated with the rate of fibrosis change in any of the models (PO0.4) and its addition as a covariate did not change the significance of the others. By multivariate linear regression analysis, only presence of diabetes and earlier fibrosis stage were significantly associated with a higher rate of fibrosis progression. When cirrhotics were excluded, the same variables (diabetes and fibrosis stage) plus BMI remained significant (Table 7). The addition of ursodiol as a covariate did not change the results of the multivariate analysis, indicating a lack of effect of that drug on fibrosis change. At least three features of the metabolic syndrome were present in 55/103 (53%) patients. However, there was no a significant correlation between rate of fibrosis change and presence of metabolic syndrome (rz0.08, PZ0.4), or rate of fibrosis change and HOMA (rzk0.01, PZ0.9), or QUICKI (rz0.09, PZ0.6). Similarly, rate of fibrosis change was not significantly different between patients with or without NASH (0.014G0.69 vs. 0.19G0.20 stages/year, respectively, PZ0.3) Change in aminotransferase levels There was a significant decrease in ALT (97G56 vs. 74G61 U/l, P!0.001) and AST (75G50 vs. 56G51 U/l, P!0.001) between biopsies. However, comparing patients with progressive fibrosis to those without, there was no significant difference in the magnitude of change in ALT level (K13G63 vs. K29G55 U/l, respectively, PZ0.7) or AST level (K10G63 vs. K24G51 U/l, respectively, PZ0.3). Improvement in ALT correlated with improvement in Mallory s hyaline score (rz0.25, PZ0.02) and overall inflammatory grade (rz0.29, PZ0.006). Improvement in AST correlated with improvement in steatosis grade (rz 0.3, PZ0.004), Mallory s hyaline (rz0.3, PZ0.004), lobular inflammation (rz0.27, PZ0.02) and overall inflammatory grade (rz0.39, P!0.001). Neither change in ALT nor AST correlated with change in fibrosis stage (rz0.07, PZ0.5 and rz0.15, PZ0.15, respectively) Change in other histological parameters Whereas fibrosis tended to progress, there was a significant overall reduction in severity of steatosis, inflammation, ballooning of hepatocytes and Mallory s hyaline, between the first and last biopsy (Fig. 1). Among noncirrhotics, a significant decrease in steatosis (2.0G0.8 vs. 1.6G0.8, P!0.001), and ballooning (1.0G0.4 vs. 0.8G0.4, PZ0.008) occurred between biopsies with a trend for inflammation (1.1G0.6 vs. 1.0G0.5, PZ0.091). Analysis of cirrhotic patients also revealed a significant decrease in steatosis (1.4G0.6 vs. 1.0G0.5, PZ0.008) and ballooning (1.2G0.5 vs. 0.6G0.7, PZ0.01), with a trend for inflammation (1.3G0.7 vs. 1.0G0.6, PZ0.096). Of the three patients with bland steatosis on initial biopsy, two developed NASH and one remained unchanged. None developed fibrosis. All of the four patients with steatosis and nonspecific inflammation developed NASH. Of the 96 original NASH patients, four lost evidence of Fig. 1. Histological features at initial and final biopsy. There was a significant improvement in severity of steatosis, ballooning, Mallory s hyaline and inflammation between first (dark bars) and last (clear bars) liver biopsies (P!0.05 for all comparisons).

6 L.A. Adams et al. / Journal of Hepatology 42 (2005) steatosis, four developed steatosis with nonspecific inflammation and two developed bland steatosis. 4. Discussion In this study, which represents the largest reported series of NAFLD patients with sequential liver biopsies, 37% had progressive fibrosis over a mean period of 3.2 years. The majority (67%) of patients biopsied after a 4-year interval increased in fibrosis stage. The rate of fibrosis change was slow, overall being 0.02G0.66 stages per year (or 0.09G 0.67 stages/year in noncirrhotics). Assuming fibrosis progression occurs at a linear rate, it would take an average of 50 years to advance just one stage (or 11 years in noncirrhotics). However, considerable variability in the rate of fibrosis change was observed, with one in six patients having relatively rapid progression of more than 0.5 stages per year, and four patients progressing from no (stage 0) fibrosis to advanced (stage 3 4) fibrosis over an average interval of only 12 years. A proportion of the variability in rate of fibrosis change was accounted for by diabetes (and BMI when cirrhotics were excluded). Previous studies examining fibrosis change over time, have been limited by small numbers. In addition, patients have generally undergone sequential biopsies due to clinical indications, potentially biasing results towards patients with more severe or atypical disease. The majority of patients in the present study underwent a biopsy at a predetermined interval as part of a clinical protocol, therefore, limiting this type of selection bias. Although nearly half of our patients were taking either ursodeoxycholic acid or clofibrate between biopsies, these agents were ineffective in changing their histological course [15,16]. Despite these trials representing the largest number of patients involved in clinical treatment studies for NASH to date, it may be possible that a real therapeutic effect on histology was missed because of type 2 error. However, we found no evidence that the histological course was different between treated (i.e., ursodiol or clofibrate) and untreated patients, or between participants in clinical trials and those biopsied outside of clinical trials (Table 1). Hence, we believe our series of paired liver biopsies describes the spontaneous histological course that occurs over time in patients with NAFLD. Diabetes mellitus was a strong independent predictor of higher rates of fibrosis progression. This is consistent with cross-sectional studies that have identified presence of diabetes as an indicator of more advanced liver disease in NAFLD [20,21]. In addition, diabetes has been associated with a rapid progression of chronic hepatitis C [22], and recognized as a risk factor for the development of chronic liver disease and liver cancer [23]. Aside from diabetes, BMI was independently associated with higher fibrosis progression among noncirrhotic patients. Thus, higher BMI is an indicator of advanced NAFLD [20,21,24], as well as more rapidly progressive disease. Fibrosis stage on initial biopsy was a strong predictor of fibrosis rate, with lower fibrosis stage associated with a higher rate of fibrosis progression. This may in part be explained by the fact that patients without fibrosis cannot regress, whereas patients with cirrhosis cannot progress in fibrosis stage. However, after excluding cirrhotics (stage 4) and patients without fibrosis (stage 0), lower fibrosis stage remained significantly predictive (PZ0.03) of greater rate of fibrosis progression. Only three patients had bland steatosis and none of them developed fibrosis, which is consistent with other studies reporting a relatively benign course among patients with bland steatosis [25,26]. We observed a significant reduction in steatosis grade over time, including three cirrhotic patients who subsequently lost all evidence of steatosis. This has been observed previously in isolated cases [10,27], emphasizing the importance of considering NAFLD as a differential diagnosis for cryptogenic cirrhosis [6]. Similarly, severity of hepatocyte ballooning, inflammation and Mallory s hyaline decreased over time and did not parallel changes in fibrosis. Furthermore, aminotransferase levels improved regardless of whether or not fibrosis progressed. Thus, improvement in aminotransferase levels appear to indicate improvement in steatosis and inflammation but not fibrosis and, therefore, may provide false reassurance regarding prognosis. This should be kept in mind when interpreting clinical trials lacking histological endpoints, as well as when monitoring patients in a clinical setting. Sampling error on liver biopsy may lead to erroneous grading and staging among patients with NAFLD. Variability of fibrosis stage in paired right and left lobe liver biopsies has been reported to vary by 24 37% [28,29]. Although, we cannot confidently quantify the effect of sampling error on our results, sampling variability is likely to upstage patients as frequently as downstage patients. Therefore, with increasing numbers of biopsies (or power), the noise sampling variability creates becomes less significant and trends and true associations are still observed. Twenty-nine percent of patients had a reduction in fibrosis stage between first and last biopsies. Part of this may be spurious due to sampling error, but we cannot ignore that fibrosis pathogenesis is a dynamic process and can regress with modification of the underlying insult [30]. Regression of fibrosis stage in our patients was not associated with weight loss. Nevertheless, we cannot exclude that between biopsies, some patients may have had modification of risk factors not measured in our study, such as physical exercise, or improved glycemic, lipid or blood pressure control. In summary, fibrosis in NAFLD progresses slowly with considerable variability among patients. Aminotransferases, steatosis and inflammatory features improve or resolve as fibrosis progresses. Among patients with early stage disease, repeat liver biopsy is more likely to reveal progressive

7 138 L.A. Adams et al. / Journal of Hepatology 42 (2005) fibrosis after a four-year interval. Fibrosis progression occurs more rapidly among diabetics, patients with higher BMI and those with lower stages of fibrosis. These patients should be targeted for therapeutic intervention to prevent disease progression. Acknowledgements Leon Adams is supported by a Medical Fellowship from The University of Western Australia. References [1] Angulo P. Nonalcoholic fatty liver disease. N Engl J Med 2002;346: [2] Clark JM, Brancati FL, Diehl AM. The prevalence and etiology of elevated aminotransferase levels in the United States. Am J Gastroenterol 2003;98: [3] Ruhl CE, Everhart JE. Determinants of the association of overweight with elevated serum alanine aminotransferase activity in the United States. Gastroenterology 2003;124: [4] Bugianesi E, Manzini P, D Antico S, Vanni E, Longo F, Leone N, et al. Relative contribution of iron burden, HFE mutations, and insulin resistance to fibrosis in nonalcoholic fatty liver. Hepatology 2004;39: [5] Day CP, James OF. Steatohepatitis: a tale of two hits? Gastroenterology 1998;114: [6] Caldwell SH, Oelsner DH, Iezzoni JC, Hespenheide EE, Battle EH, Driscoll CJ. Cryptogenic cirrhosis: clinical characterization and risk factors for underlying disease. Hepatology 1999;29: [7] Evans CD, Oien KA, MacSween RN, Mills PR. Non-alcoholic steatohepatitis: a common cause of progressive chronic liver injury? J Clin Pathol 2002;55: [8] Lee RG. Nonalcoholic steatohepatitis: a study of 49 patients. Hum Pathol 1989;20: [9] Harrison SA, Torgerson S, Hayashi PH. The natural history of nonalcoholic fatty liver disease: a clinical histopathological study. Am J Gastroenterol 2003;98: [10] Powell EE, Cooksley WG, Hanson R, Searle J, Halliday JW, Powell LW. The natural history of nonalcoholic steatohepatitis: a follow-up study of forty-two patients for up to 21 years. Hepatology 1990;11: [11] Bonora E, Targher G, Alberiche M, Bonadonna RC, Saggiani F, Zenere MB, et al. Homeostasis model assessment closely mirrors the glucose clamp technique in the assessment of insulin sensitivity: studies in subjects with various degrees of glucose tolerance and insulin sensitivity. Diabetes Care 2000;23: [12] Katz A, Nambi SS, Mather K, Baron AD, Follmann DA, Sullivan G, et al. Quantitative insulin sensitivity check index: a simple, accurate method for assessing insulin sensitivity in humans. J Clin Endocrinol Metab 2000;85: [13] Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285: [14] Organisation WH. Report of a WHO consultation: definition of metabolic syndrome in definition, diagnosis and classification of diabetes mellitus. Geneva: World Health Organisation, Department of Noncommunicable Disease Surveillance; [15] Lindor KD, Kowdley KV, Heathcote EJ, Harrison ME, Jorgensen R, Angulo P, et al. Ursodeoxycholic acid for treatment of nonalcoholic steatohepatitis: results of a randomized trial. Hepatology 2004;39: [16] Laurin J, Lindor KD, Crippin JS, Gossard A, Gores GJ, Ludwig J, et al. Ursodeoxycholic acid or clofibrate in the treatment of nonalcohol-induced steatohepatitis: a pilot study. Hepatology 1996;23: [17] Bravo AA, Sheth SG, Chopra S. Liver biopsy. N Engl J Med 2001; 344: [18] Brunt EM, Janney CG, Di Bisceglie AM, Neuschwander-Tetri BA, Bacon BR. Nonalcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999;94: [19] Neuschwander-Tetri BA, Caldwell SH. Nonalcoholic steatohepatitis: summary of an AASLD Single Topic Conference. Hepatology 2003; 37: [20] Angulo P, Keach JC, Batts KP, Lindor KD. Independent predictors of liver fibrosis in patients with nonalcoholic steatohepatitis. Hepatology 1999;30: [21] Wanless IR, Lentz JS. Fatty liver hepatitis (steatohepatitis) and obesity: an autopsy study with analysis of risk factors. Hepatology 1990;12: [22] Ortiz V, Berenguer M, Rayon JM, Carrasco D, Berenguer J. Contribution of obesity to hepatitis C-related fibrosis progression. Am J Gastroenterol 2002;97: [23] El-Serag HB, Tran T, Everhart JE. Diabetes increases the risk of chronic liver disease and hepatocellular carcinoma. Gastroenterology 2004;126: [24] Ratziu V, Giral P, Charlotte F, Bruckert E, Thibault V, Theodorou I, et al. Liver fibrosis in overweight patients. Gastroenterology 2000; 118: [25] Teli MR, James OF, Burt AD, Bennett MK, Day CP. The natural history of nonalcoholic fatty liver: a follow-up study. Hepatology 1995;22: [26] Dam-Larsen S, Franzmann M, Andersen IB, Christoffersen P, Jensen LB, Sorensen TI, et al. Long term prognosis of fatty liver: risk of chronic liver disease and death. Gut 2004;53: [27] Abdelmalek M, Ludwig J, Lindor KD. Two cases from the spectrum of nonalcoholic steatohepatitis. J Clin Gastroenterol 1995;20: [28] Mendez P, Regev A, Molina E, Berho M, Bejarano P, Ladino M, et al. Sampling error and reliability of liver biopsy among patients with nonalcoholic fatty liver disease [Abstract]. Hepatology 2003;38: 673A. [29] Merriman R, Ferrell L, Patti M, Ostroff J, Bagetelos K, Aouizerat B, et al. Histologic correlation of paired right lobe and left lobe liver biopsies in morbidly obese individuals with suspected nonalcoholic fatty liver disease [Abstract]. Hepatology 2003;38:232A. [30] Friedman SL. Liver fibrosis from bench to bedside. J Hepatol 2003; 38:S38 S53.