Venesection for non-alcoholic fatty liver disease unresponsive to lifestyle counselling a propensity score-adjusted observational study

Transcription

1 Q J Med 2011; 104: doi: /qjmed/hcq170 Advance Access Publication 17 September 2010 Venesection for non-alcoholic fatty liver disease unresponsive to lifestyle counselling a propensity score-adjusted observational study L. VALENTI 1, S. MOSCATIELLO 2, E. VANNI 3, A.L. FRACANZANI 1, E. BUGIANESI 3, S. FARGION 1 and G. MARCHESINI 2 From the 1 Department of Internal Medicine, University of Milan, Fondazione Ospedale Maggiore Policlinico IRCCS, Milan, 2 Unit of Metabolism and Clinical Dietetics, Alma Mater Studiorum University, Bologna and 3 Department of Gastroenterology and Hepatology, University of Turin, Turin, Italy Address correspondence to Prof. G. Marchesini, Unit of Clinical Dietetics, Alma Mater University of Bologna, Policlinico S. Orsola, Via Massarenti, 9, I Bologna, Italy. giulio.marchesini@unibo.it Received 2 April 2010 and in revised form 12 August 2010 Summary Background/Aim: To test the short-term clinical usefulness of venesection associated with lifestyle counselling as against counselling alone on insulin resistance and liver enzymes in subjects with non-alcoholic fatty liver disease (NAFLD), using a propensity score approach. Methods: We carried out a 6- to 8-month observational analysis of 198 NAFLD patients in three Italian referral centres (79 venesection and 119 counselling alone). Insulin resistance was measured by the homeostasis model assessment (HOMA) method. Logistic regression was used to identify factors associated with normal HOMA and normal alanine aminotransferase (ALT) at the end of observation. The results were adjusted for the propensity score to be enrolled in the venesection programme, based on clinical and laboratory data, including common HFE polymorphisms and liver biopsy (available in 161 cases). Introduction Non-alcoholic fatty liver disease (NAFLD) represents the leading cause of raised liver enzymes in the USA and in Western countries, affecting up to one-third Results: After adjustment for propensity and changes in BMI, venesection was significantly associated with normal HOMA [all cases: odds ratio (OR) 3.00; 95% confidence interval (CI) ; cases with histology: OR 2.29; 95% CI ] and ALT within normal limits (all cases: OR 2.56; 95% CI ; cases with histology: OR 2.81; 95% CI ). The results were confirmed in an analysis of 57 pairs matched for propensity, where venesection similarly increased the probability of normal HOMA (OR 3.27; 95% CI ) and normal ALT (OR 5.60; 95% CI ). Similar data were obtained in the subset of cases with normal basal ferritin (<350 ng/ml). Conclusion: Iron depletion by venesection favours the normalization of insulin resistance and raised liver enzymes in non-haemochromatosis patients with NAFLD. of the population. 1,2 The disease, characterized by hepatic insulin resistance, 3,4 is potentially progressive to cirrhosis and ultimately to hepatocarcinoma. 5 7 A diagnosis of NAFLD carries an! The Author Published by Oxford University Press on behalf of the Association of Physicians. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 142 L. Valenti et al. increased risk of both liver-related as well as cardiovascular mortality, in relation to the associated features of the metabolic syndrome, especially in patients with non-alcoholic steatohepatitis (NASH). 3,4,8 Although weight-reducing diets and exercise have proved beneficial, 9,10 a remarkable proportion of patients maintain higher-than-normal liver enzymes and insulin resistance despite recommendations aimed at lifestyle changes and pharmacological therapy with insulin sensitizing agents. 11 The role of altered liver enzymes as markers of disease severity is uncertain. Advanced disease may also be observed in subjects with normal enzymes, 12,13 and high alanine and aspartate aminotransferases (ALT and AST) do not predict disease progression, 14 but treatment-induced histological improvement is systematically associated with reduced aminotransferase levels, which remain a surrogate marker for both patients and physicians. Hyperferritinemia with normal transferrin saturation is observed in a variable proportion of NAFLD cases, depending on the clinical setting. It has been associated with oxidative stress and mild hepatic iron overload, sometimes in the presence of common mutations of the C282Y and H63D polymorphisms of the HFE gene associated with hereditary haemochromatosis. 18 Increased iron stores have been associated with an altered adipokine profile and insulin resistance in humans, 19,20 and increased liver iron may induce hepatic insulin resistance. 18,21 Finally, in Caucasian patients with NAFLD, hyperferritinemia has been associated with progressive liver damage, while the role of HFE mutations in severe liver fibrosis is still controversial In the hypothesis that increased iron stores may play a role in the pathogenesis of insulin resistance and liver damage by increasing oxidative stress, 25 a few studies have evaluated the effect of iron depletion on metabolic parameters. Venesection was well tolerated, 26 and was reported to improve insulin resistance and liver enzymes in a small number of NAFLD patients without iron overload and either impaired 27 or normal 28 glucose tolerance. In a case control study, we have recently reported that iron depletion by venesection improved insulin resistance in patients with NAFLD to a higher extent than lifestyle changes alone. 17 However, the study was non-randomized and was flawed by several biases. First, the venesection group included subjects with more severe disease and iron overload; secondly, histological data were only available in a limited number of cases. In summary, we failed to demonstrate any effect of venesection on liver enzymes. In the present report we tested the short-term clinical usefulness of venesection in a much larger series of NAFLD subjects by comparing subjects entering a venesection programme with patients on continuous treatment by lifestyle changes alone, using a propensity score approach. 29 The analysis confirms that iron depletion improves both serum aminotransferases and insulin resistance, independent of baseline characteristics, including the severity of liver histology, and the weight-reducing effect of nutritional counselling. Methods Study participants We report the data collected in 198 patients with NAFLD observed by three Italian centres in the course of an 18-month period, using a common study protocol. The centres are tertiary referral centres for large areas of the country; therefore patients should not be considered as fully representative of the general NAFLD population. Their clinical and biochemical characteristics are reported in Table 1. We included all cases with bright liver at ultrasonography, associated with chronically elevated or fluctuating aminotransferases and gammaglutamyltransferase (GGT) levels. Hyperferritinemia (>350 ng/ml) was present in 100 cases, transferrin saturation >55% in 16 (all but one with elevated ferritin concentrations). Exclusion criteria were the evidence of liver disease of different aetiology (based on a large screening for hepatitis B and C virus, autoimmune disease, celiac disease, a 1 -antitripsin deficiency, Wilson s disease) and alcohol consumption >20 g/day (after checking with relatives). Subjects with overt diabetes (treated with either hypoglycaemic agents or insulin) and cases with hereditary familial haemochromatosis (HFE) (homozygosity for the C282Y HFE polymorphism) were also excluded from the analysis, as were 13 cases with missing HFE gene analysis. Subjects who normalized their liver enzyme levels within 3 months following counselling for nutrition and physical exercise were also excluded. Anthropometric, clinical and laboratory variables All cases underwent an anthropometric assessment (body mass index, BMI) and an evaluation of the features of the metabolic syndrome. An oral glucose tolerance test was also performed in all cases. Finally, a liver biopsy was available in 161 patients (54 in the venesection group see below 107 in counselling), and 63 cases (18 in venesection and

3 Venesection in NAFLD 143 Table 1 Baseline characteristics of the NAFLD population Venesection (n = 79) Counselling (n = 119) P-value Sex (M/F) 66/13 99/20 >0.999 Age (years) Body mass index (kg/m 2 ) Transferrin saturation (%) <0.001 Ferrritin (ng/ml) 450 [462] 314 [423] Glucose (mg/dl) Insulin (mu/ml) <0.001 HOMA-R (%) Total cholesterol (mg/dl) HDL-cholesterol (mg/dl) <0.001 Triglycerides (mg/dl) ALT (U/l) AST (U/l) Gamma-GT (U/l) 31 [34] 41 [57] HFE genotype class (0/1/2/3) (%) a 53/23/16/8 71/21/8/ DM or IGT at OGTT (%) b 44.3 ( ) 38.7 ( ) NASH at histology (%) c 33.3 ( ) 42.1 ( ) Metabolic syndrome (%) 36.8 ( ) 31.6 ( ) Data are presented as mean SD, as median [interquartile range] or as prevalence (95% confidence interval). a For classification, see text. b Diabetes mellitus or impaired glucose tolerance diagnosed by oral glucose tolerance test. c Based on 161 cases (54 in the venesection arm, 107 in counselling alone). 45 in counselling) were classified as definite NASH according to the recent proposal of the NASH Clinical Research Network. 30 Liver biopsy was usually carried out in the presence of a long-standing elevation of ALT, but also the severity both of the other biochemical tests and of the ultrasonographic assessment was considered. In all cases insulin resistance at baseline and at study end was determined by the homeostasis model assessment (HOMA), according to Matthews et al. 31 In our setting, a HOMA index of insulin resistance (HOMA-R) value >2.7 is indicative of insulin resistance. 32 This value represents the cut-off of the upper quartile of a control population. The upper limit of ALT levels in our laboratories is set at 40 U/l, but a sensitivity analysis was also carried out using the recently-proposed updated definitions of healthy ranges for serum ALT levels (30 IU/l in men and 19 U/l in women). 33 Protocol At first visit, all subjects were given counselling for healthier food intake, eating habits and physical activity. After 3 months, according to physicians judgment, a few of them were also invited to enter an iron depletion programme based on repeated venesection. This visit was considered the starting point of our observation study. The indication for the venesection programme was based on liver tests, iron store parameters, HFE gene polymorphisms and histology data. According to our protocol, all cases with elevated iron indices should be counselled venesection, but the decision was discussed with the patients and the final allocation was also related to motivation, education and job constraints. In the venesection arm, iron removal was carried out to near iron-deficiency by weekly or fortnightly phlebotomy (350 cc of blood, according to baseline haemoglobin values and tolerance), until ferritin was <80 ng/ml in the presence of normal or low transferrin saturation. Maintenance phlebotomy (as many as required) was then instituted to keep iron stores depleted (ferritin <100 ng/ml and transferrin saturation 35%). All patients underwent a follow-up examination 6 8 months after enrolment (3 months after the end of the venesection programme), when anthropometric measurements and liver function tests were repeated. During the observation period, independent of venesection, all subjects attended the Liver Units for educational sessions on lifestyle (either individually or as group sessions, in accordance with in-house procedures) and for regular monitoring of their calorie intake and body weight. The primary outcomes of the study were the effects of venesection on ALT levels and insulin

4 144 L. Valenti et al. resistance. In particular, we aimed to test the prevalence of normal ALT and normal HOMA-R index at the end of a pre-defined period, including a treatment period and a 3- to 4-month maintenance follow-up. The study protocol was submitted to the senior staff committee of individual Departments for approval. The committees are equivalent to Institutional Review Boards and are qualified for the approval of observational analyses. Statistical analysis Sample size According to our previous experience, 40% of cases normalize liver enzymes by appropriate lifestyle changes. To be meaningful for clinical purposes, we considered that venesection could increase the probability of normalizing liver enzymes by 50% (60% of cases). Accordingly, 60 cases per group were needed, considering an a error of 0.05 and a b error of 0.2. This number of cases with liver biopsy could only be obtained by merging three independent databases, derived from Centres operating with a liver biopsy rate of 70 80%. Calculations The three databases were merged in a single database, and the very few missing continuous values (<5% in any column) were replaced by median values. A descriptive analysis was initially carried out on the two groups treated by venesection or by the sole counselling [median and 95% confidence interval (CI) or interquartile range (IQR) for data not following a normal distribution and prevalence of nominal conditions]. Differences between groups were then tested by parametric or nonparametric methods, as appropriate. Multivariate logistic regression analysis was then used to identify factors associated with outcome measures. To adjust data for the likely selection bias of venesection in subjects with different disease severity, HFE gene polymorphisms and/or iron stores, we calculated the estimated propensity score by a separate logistic regression analysis. The propensity score is the probability for a randomly chosen patient to be assigned to a specific treatment, in this case venesection superimposed to nutritional counselling, based on differences in clinical and biochemical variables at baseline, compared to subjects given nutritional counselling alone. It does not only define the conditional probability to be assigned to a specific treatment given the individual characteristics considered by physician but it also includes the probability to accept the proposal suggested by physicians. As it is recommended that covariates be introduced generously into the propensity score model, we included a large number of covariates independently of the significance threshold or other selection criteria. 29,34 36 The following baseline variables were considered: age, gender, BMI, ALT, iron and ferritin levels, transferrin saturation, fasting glucose and HOMA-R, the presence of the metabolic syndrome and HFE mutations. The presence of HFE polymorphisms was transformed into a scoring system with 0 representing wild-type, 1 = heterozygous H63D, 2 = homozygous H63D or heterozygous C282Y, 3 = compound heterozygosity. In a second step, also the presence of definite NASH at histology 30 was added as independent variable. The propensity scores calculated as above were then used to adjust for baseline values in logistics regression analysis in two different models, having normal ALT levels and normal HOMA-R as dependent variables. Furthermore, we checked the validity of our model testing the association between treatment group and target ALT in a subset of matched cases. For this purpose, the whole database was ordered for propensity score, and each subject treated by venesection was pair-matched with a case treated by counselling alone. For matching, only the five cases preceding or following the index case in the list were considered. Using this criterion, we could match 57 cases treated by venesection with corresponding cases treated by counselling alone. Subsequently, the association of venesection with treatment outcomes was reassessed by logistic regression, after adjustment for propensity. Finally, we checked the usefulness of venesection in a post-hoc analysis of cases with normal transferrin saturation (<55%) and normal ferritin levels (<350 ng/ml). Data in the text and in tables are given as mean standard deviation (SD), if not otherwise specified. P-values <0.05 were considered statistically significant. Results On average, our population was largely characterized by overweight (56% of total cases; 58% in the venesection group, 55% in counselling), whereas obesity was present in only 14 and 19% of cases, respectively (P for weight class distribution = 0.608). The venesection group was further characterized by higher transferrin saturation and ferritin levels, higher insulin and insulin resistance (HOMA-R)

5 Venesection in NAFLD 145 and lower HDL-cholesterol, as well as by higher prevalence of HFE polymorphisms. At the end of the study period, the median amount of iron removed by venesection was 1.96 g (IQR: 2.09), and both transferrin saturation and ferritin levels were significantly reduced (Table 2). In addition, venesection was more effective in reducing insulin levels and insulin resistance, without differences in glucose levels, whereas counselling alone promoted a larger weight loss. No differences in mean lipid and liver enzyme levels were observed between groups. At the end of follow-up, 67/79 cases in the venesection group had ALT values <40 U/l (85%) vs. 75/119 in counselling (63%; P = 0.001). The number of cases with HOMA-R <2.7 were 50/79 (63%) and 56/119 (47%), respectively (P = 0.029). The median propensity scores were (IQR, 0.350) and (0.211) in subjects included in the venesection and counselling group, respectively. The corresponding values, recalculated in the subsets with liver biopsy were (0.260) and (0.171), respectively. At univariate analysis (Table 3), age, enzyme levels, the presence of NASH at histology and venesection were associated with ALT within normal limits at the end of the observation period, whereas the correction of insulin resistance was negatively associated with age, BMI, fasting glucose, insulin and HOMA-R, high triglycerides and aminotransferases, and the presence of the metabolic syndrome. The data did not change systematically when the lower cut-offs of ALT were used as the dependent variable (not reported in detail). After adjustment for propensity and changes in BMI, venesection was significantly associated with normal HOMA [all cases: odds ratio (OR) 3.00; 95% CI ; P = 0.017; cases with histology: OR 2.29; 95% CI ; P = 0.031), and with ALT within normal limits (all cases: OR 2.56; 95% CI ; P = 0.007; cases with histology: OR 2.81; 95% CI ; P = 0.015). When the analysis was repeated using the updated definitions of healthy ALT cut-offs of 30 U/l for males and 19 U/l for females, 33 venesection was again systematically associated with ALT normalization (all cases: OR 2.50; 95% CI ; P = 0.015; cases with histology: OR 2.77; 95% CI ; P = 0.018). The results were further confirmed in the analysis of 57 NAFLD pairs, matched for propensity score at baseline with subjects treated by counselling alone (propensity scores: venesection, ; counselling, ). In these subjects, venesection increased by over three times the probability of normal HOMA (OR 3.27; 95% CI ; P = 0.008), and by over five times the probability to have normal ALT at follow-up (OR 5.60; 95% CI ; P < 0.001). In the post-hoc analysis of subjects with normal ferritin and normal transferrin saturation (n = 97, venesection was again associated with a higher probability of normal HOMA (OR 3.42; 95% CI ; P = 0.045) and normal ALT (OR 3.85; 95% CI ; P = 0.014) at follow-up. The beneficial effects on ALT were maintained in subjects where liver biopsy was available (OR 5.20; 95% CI ; P < 0.001). Discussion In the short-term, adding venesection to counselling of healthy food intake and physical activity Table 2 Treatment-induced changes in body mass index and biochemistry Venesection (n = 79) Counselling (n = 119) P-value* Body mass index (kg/m 2 ) Transferrin saturation (%) <0.001 Ferrritin (ng/ml) <0.001 Glucose (mg/dl) Insulin (mu/ml) <0.001 HOMA-R (%) <0.001 HDL-cholesterol (mg/dl) Triglycerides (mg/dl) ALT (U/l) AST (U/l) Gamma-GT (U/l) Data are presented as differences between the end-of-observation values and baseline (mean SD). *Differences between groups, time X treatment ANOVA.

6 146 L. Valenti et al. Table 3 Baseline factors associated with outcomes in the whole series at univariate analysis ALT < 40U/l P-value HOMA <2.7 P-value Male sex 3.46 ( ) ( ) Age (years/10) 0.76 ( ) ( ) <0.001 Body mass index (kg/m 2 ) 1.07 ( ) ( ) <0.001 Transferrin saturation (%) 1.01 ( ) ( ) Ferrritin (ng/ml/50) 1.00 ( ) ( ) Glucose (mg/dl/10) 1.14 ( ) ( ) Insulin (mu/ml) 1.06 ( ) ( ) <0.001 HOMA-R (%) 1.24 ( ) ( ) <0.001 HDL-cholesterol (mg/dl) 0.99 ( ) ( ) Triglycerides (mg/dl/20) 1.00 ( ) ( ) <0.001 ALT (U/l/10) 1.15 ( ) < ( ) AST (U/l/10) 1.27 ( ) ( ) Gamma-GT (U/l/10) 0.99 ( ) ( ) Presence of NASH a 1.13 ( ) ( ) HFE polymorphisms b 0.90 ( ) ( ) Metabolic syndrome 1.26 ( ) ( ) <0.001 Venesection programme 2.28 ( ) ( ) a Analysis limited to the 161 cases with liver biopsy. b For the scoring system, see Methods section. produces a significant improvement on biochemical tests in patients with NAFLD who have not responded to a brief period of counselling alone. Iron depletion was associated with a greater reduction of insulin levels and insulin resistance, and a higher percentage of iron-depleted subjects had normal liver enzymes, compared with lifestyle modifications alone. The benefits of venesection were independent of the baseline characteristics of the patients, including the presence of NASH, as well as of changes in BMI favoured by nutritional counselling. The study expands the results of a much smaller case control study, where venesection had demonstrated a significant effect on insulin resistance, but had unexpectedly failed to improve liver enzymes significantly as compared to counselling alone. 17 The present analysis is based on a totally different approach the use of the propensity score and on a much larger cohort in subjects who did not respond to a 3-month counselling approach. These patients are frequently observed in referral centres for suspected liver disease of metabolic origin and long-standing bright liver at ultrasounds, where subjects with primary NAFLD mix up with subjects with hereditary haemochromatosis. The large majority had chronically-elevated or fluctuating liver enzyme levels, or had biochemical evidence of non-haemochromatosis-related iron accumulation as most patients had increased ferritin in the presence of normal transferrin saturation, particularly in the venesection group, which might generate a bias in outcome assessment. We addressed this bias by means of a propensity score approach at multivariate analysis, a method to adjust groups for a variety of confounders and keep the selection bias to a minimum in this type of study, thus increasing external validity. 29,36 This approach is particularly useful in chronic diseases related to unhealthy behaviours, where the acceptance of any proposed therapy also depends on conditions on patients side, difficult to consider by random allocation and responsible for a higher attrition rate in randomized studies. 37 After adjustment for propensity, venesection was still associated with a much higher probability of treatment effects at follow-up. In clinical practice, the selection of appropriate treatment, both by physicians and patients side, is frequently based on very tiny differences, and the modes of care cannot be easily standardized. We checked the validity of our model by testing the association between treatment group assignment and outcome in a subset of cases matched for propensity score. The data confirmed that iron depletion was associated with a much better outcome also in subjects with baseline histology, possibly reflecting a selected category with more severe clinical picture. This was also the case of patients with ferritin levels and transferrin saturation within normal ranges, where increased iron stores could hardly be suspected.

7 Venesection in NAFLD 147 Overall, our data suggest that iron depletion reduces hepatocellular necrosis, to the same extent as observed in randomized studies of drug treatment with insulin sensitizers, 38 where reduced ALT are also associated with decreased fibrosis in patients with NAFLD. Experimental models indicate that the mechanisms may involve the inhibition of oxidative stress-related damage, decreased apoptosis and an up-regulation of the intracellular pathways promoting survival in hepatocytes. 39 By decreasing hepatic iron stores, venesection is also expected to remove an obstacle to insulin extraction, thereby improving insulin resistance and decreasing the insulin-mediated suppression of hepatic glucose production, a mechanism intimately linked to steatosis in NAFLD. 39 Moreover, in vivo and in vitro experiments have recently related the effects of venesection to an interaction with retinol-binding protein-4 (RBP-4) on insulin-resistance-associated iron overload, 20 and RBP-4 is related to steatosis in a different model of liver injury. 40 All these mechanisms might also be operative in our cohort, where insulin sensitivity was normalized by venesection in a larger proportion of cases, compared with counselling alone. Importantly, given the different biological pathways and the various organs interested by iron accumulation, this therapy might increase the effects of other insulin sensitizing drugs in patients with NASH, such as metformin and glitazones. Venesection might thus represent a valid option both as adjuvant therapy and/or as alternative drug treatment in subjects unresponsive to insulinsensitizers. Given the association of insulin-resistanceassociated hepatic iron overload and NAFLD independently of common mutations of the HFE gene, 41 venesection might be tested in settings where extensive analysis of HFE gene is not available, as there are several patients where the aetiology of iron overload remains largely undefined. Finally, whereas venesection was more effective on liver enzymes, counselling alone produced a larger effect on BMI. Several pilot studies 9,42,43 and a randomized study 44 have shown that lifestyle interventions significantly reduce BMI and improve liver enzymes NAFLD, and recent data have confirmed that weight loss is associated with a better histological outcome. 44 We conclude that an integrated approach might now represent the best available option to NAFLD treatment. The present study has limitations too. First, although all patients were managed in a similar way and the general protocol was agreed a priori, differences in case mix were present in the different centres. The Milan centre, where >70% of cases were observed, is a referral centre for familial haemochromatosis, thus accounting for the very high number of cases with alterations in iron metabolism/stores included in the final analysis. Secondly, the follow-up is quite short, whereas the natural history of NAFLD spans over decades, and no inference can be made on the long-term significance of our results on histological grounds. ALT levels have proven to be associated with disease severity in a few studies, particularly in obesity, where they enter the proposed algorithms for the assessment of advanced fibrosis, 45,46 but in other NAFLD series liver enzymes were independent of histological severity. 12,13 In conclusion, iron depletion by venesection favours the normalization of raised aminotransferases and insulin sensitivity in non-haemochromatosis patients with NAFLD unresponsive to lifestyle counselling, as established by a propensity score-adjusted retrospective analysis of a prospectively-collected large database. The procedure is well tolerated and may be more extensively proposed and investigated in the clinical setting. Conflict of interest: None declared. Funding FIRST Università degli Studi di Milano 2007, 2008 (L.V., S.F.); Ricerca corrente Ospedale Maggiore Policlinico 2006 and 2008 (L.V., S.F.); Centro per lo Studio delle Malattie del Fegato e del Metabolismo (Milan); RFO , Università di Bologna (G.M.). References 1. Bellentani S, Saccoccio G, Masutti F, Croce LS, Brandi G, Sasso F, et al. Prevalence of and risk factors for hepatic steatosis in Northern Italy. Ann Intern Med 2000; 132: Browning JD, Szczepaniak LS, Dobbins R, Nuremberg P, Horton JD, Cohen JC, et al. Prevalence of hepatic steatosis in an urban population in the United States: impact of ethnicity. Hepatology 2004; 40: Marchesini G, Bugianesi E, Forlani G, Cerrelli F, Lenzi M, Manini R, et al. Nonalcoholic fatty liver, steatohepatitis, and the metabolic syndrome. Hepatology 2003; 37: Ekstedt M, Franzen LE, Mathiesen UL, Thorelius L, Holmqvist M, Bodemar G, et al. Long-term follow-up of patients with NAFLD and elevated liver enzymes. Hepatology 2006; 44: Teli MR, James OF, Burt AD, Bennett MK, Day CP. The natural history of nonalcoholic fatty liver: a follow-up study. Hepatology 1995; 22: Matteoni CA, Younossi ZM, Gramlich T, Boparai N, Liu YC, McCullough AJ. Nonalcoholic fatty liver disease: a spectrum of clinical and pathological severity. Gastroenterology 1999; 116:

8 148 L. Valenti et al. 7. Bugianesi E, Leone N, Vanni E, Marchesini G, Brunello F, Carucci P, et al. Expanding the natural history of nonalcoholic steatohepatitis: from cryptogenic cirrhosis to hepatocellular carcinoma. Gastroenterology 2002; 123: Adams LA, Lymp JF, St Sauver J, Sanderson SO, Lindor KD, Feldstein A, et al. The natural history of nonalcoholic fatty liver disease: a population-based cohort study. Gastroenterology 2005; 129: Huang MA, Greenson JK, Chao C, Anderson L, Peterman D, Jacobson J, et al. One-year intense nutritional counseling results in histological improvement in patients with non-alcoholic steatohepatitis: a pilot study. Am J Gastroenterol 2005; 100: Suzuki A, Lindor K, St Sauver J, Lymp J, Mendes F, Muto A, et al. Effect of changes on body weight and lifestyle in nonalcoholic fatty liver disease. J Hepatol 2005; 43: Bugianesi E, Marzocchi R, Villanova N, Marchesini G. Non-alcoholic fatty liver disease/non-alcoholic steatohepatitis (NAFLD/NASH): treatment. Best Pract Res Clin Gastroenterol 2004; 18: Mofrad P, Contos MJ, Haque M, Sargeant C, Fisher RA, Luketic VA, et al. Clinical and histologic spectrum of nonalcoholic fatty liver disease associated with normal ALT values. Hepatology 2003; 37: Fracanzani AL, Valenti L, Bugianesi E, Andreoletti M, Colli A, Vanni E, et al. Risk of severe liver disease in NAFLD with normal aminotransferase levels: a role for insulin resistance and diabetes. Hepatology 2008; 48: Fassio E, Alvarez E, Dominguez N, Landeira G, Longo C. Natural history of nonalcoholic steatohepatitis: a longitudinal study of repeat liver biopsies. Hepatology 2004; 40: Moirand R, Mendler MH, Guillygomarc h A, Brissot P, Deugnier Y. Non-alcoholic steatohepatitis with iron: part of insulin resistance-associated hepatic iron overload? J Hepatol 2000; 33: 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: Valenti L, Fracanzani AL, Dongiovanni P, Bugianesi E, Marchesini G, Manzini P, et al. Iron depletion by phlebotomy improves insulin resistance in patients with nonalcoholic fatty liver disease and hyperferritinemia: evidence from a case-control study. Am J Gastroenterol 2007; 102: Valenti L, Rametta R, Dongiovanni P, Maggioni M, Fracanzani AL, Zappa M, et al. Increased expression and activity of the transcription factor FOXO1 in nonalcoholic steatohepatitis. Diabetes 2008; 57: Fernandez-Real JM, Moreno JM, Lopez-Bermejo A, Chico B, Vendrell J, Ricart W. Circulating soluble transferrin receptor according to glucose tolerance status and insulin sensitivity. Diabetes Care 2007; 30: Fernandez-Real JM, Moreno JM, Ricart W. Circulating RBP4 concentration might reflect insulin resistance-associated iron overload. Diabetes 2008; 57: Ferrannini E. Insulin resistance, iron, and the liver. Lancet 2000; 355: George DK, Goldwurm S, McDonald GA, Cowley LL, Walker NI, Ward PJ, et al. Increased hepatic iron concentration in nonalcoholic steatohepatitis is associated with increased fibrosis. Gastroenterology 1998; 114: Valenti L, Dongiovanni P, Piperno A, Fracanzani AL, Maggioni M, Rametta R, et al. Alpha 1-antitrypsin mutations in NAFLD: high prevalence and association with altered iron metabolism but not with liver damage. Hepatology 2006; 44: Nelson JE, Bhattacharya R, Lindor KD, Chalasani N, Raaka S, Heathcote EJ, et al. HFE C282Y mutations are associated with advanced hepatic fibrosis in Caucasians with nonalcoholic steatohepatitis. Hepatology 2007; 46: Fernandez-Real JM, Lopez-Bermejo A, Ricart W. Cross-talk between iron metabolism and diabetes. Diabetes 2002; 51: Guillygomarc h A, Mendler MH, Moirand R, Laine F, Quentin V, David V, et al. Venesection therapy of insulin resistance-associated hepatic iron overload. J Hepatol 2001; 35: Facchini FS, Hua NW, Stoohs RA. Effect of iron depletion in carbohydrate-intolerant patients with clinical evidence of nonalcoholic fatty liver disease. Gastroenterology 2002; 122: Valenti L, Fracanzani AL, Fargion S. Effect of iron depletion in patients with nonalcoholic fatty liver disease without carbohydrate intolerance. Gastroenterology 2003; 124:866; author reply Rubin DB. The design versus the analysis of observational studies for causal effects: parallels with the design of randomized trials. Stat Med 2007; 26: Kleiner DE, Brunt EM, Van Natta M, Behling C, Contos MJ, Cummings OW, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 2005; 41: Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from plasma fasting glucose and insulin concentrations in man. Diabetologia 1985; 28: Bugianesi E, Pagotto U, Manini R, Vanni E, Gastaldelli A, De Iasio R, et al. Plasma adiponectin in nonalcoholic fatty liver is related to hepatic insulin resistance and hepatic fat content, not to liver disease severity. J Clin Endocrinol Metab 2005; 90: Prati D, Taioli E, Zanella A, Della Torre E, Butelli S, Del Vecchio E, et al. Updated definitions of healthy ranges for serum alanine aminotransferase levels. Ann Intern Med 2002; 137: D Agostino RB Jr. Propensity scores in cardiovascular research. Circulation 2007; 115: Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70: Cavuto S, Bravi F, Grassi MC, Apolone G. Propensity score for the analysis of observational data: an introduction and an illustrative example. Drug Develop Res 2006; 67: Forlani G, Lorusso C, Moscatiello S, Ridolfi V, Melchionda N, Di Domizio S, et al. Are behavioural approaches feasible and effective in the treatment of type 2 diabetes? A propensity score analysis vs. prescriptive diet. Nutr Metab Cardiovasc Dis 2009; 19:

9 Venesection in NAFLD Vuppalanchi R, Chalasani N. Nonalcoholic fatty liver disease and nonalcoholic steatohepatitis: Selected practical issues in their evaluation and management. Hepatology 2009; 49: Rajpathak SN, Crandall JP, Wylie-Rosett J, Kabat GC, Rohan TE, Hu FB. The role of iron in type 2 diabetes in humans. Biochim Biophys Acta 2009; 1790: Petta S, Camma C, Di Marco V, Alessi N, Barbaria F, Cabibi D, et al. Retinol-binding protein 4: a new marker of virus-induced steatosis in patients infected with hepatitis C virus genotype 1. Hepatology 2008; 48: Pelucchi S, Mariani R, Salvioni A, Bonfadini S, Riva A, Bertola F, et al. Novel mutations of the ferroportin gene (SLC40A1): analysis of 56 consecutive patients with unexplained iron overload. Clin Genet 2008; 73: Sreenivasa Baba C, Alexander G, Kalyani B, Pandey R, Rastogi S, Pandey A, et al. Effect of exercise and dietary modification on serum aminotransferase levels in patients with nonalcoholic steatohepatitis. J Gastroenterol Hepatol 2006; 21: St George A, Bauman A, Johnston A, Farrell G, Chey T, George J. Independent effects of physical activity in patients with nonalcoholic fatty liver disease. Hepatology 2009; 50: Promrat K, Kleiner DE, Niemeier HM, Jackvony E, Kearns M, Wands JR, et al. Randomized controlled trial testing the effects of weight loss on nonalcoholic steatohepatitis. Hepatology 2010; 51: Ratziu V, Giral P, Charlotte F, Bruckert E, Thibault V, Theodorou I, et al. Liver fibrosis in overweight patients. Gastroenterology 2000; 118: Dixon JB, Bhathal PS, O Brien PE. Nonalcoholic fatty liver disease: predictors of nonalcoholic steatohepatitis and liver fibrosis in the severely obese. Gastroenterology 2001; 121:

10