Incidence and risk factors for non-alcoholic steatohepatitis: prospect
http://www.100md.com
《英国医生杂志》
1 Liver Unit, Azienda Ospedaliera Fatebenefratelli e Oftalmico, Corso di Porta Nuova 23, 20121 Milan, Italy, 2 Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, 3 Chemoprevention Unit, European Institute of Oncology, 4 Breast Department, European Institute of Oncology, 5 European Institute of Oncology, 6 Department of Internal Medicine, Azienda Ospedaliera San Paolo, University of Milan, Milan, 7 Department of Pathology, Azienda Ospedaliera San Paolo, 8 Department of Internal Medicine 2a, University of Napoli, Naples, Italy, 9 Liver Unit, Azienda Ospedaliera Sant'Anna, Como, Italy, 10 Casa di Cura Citta' di Bra, Bra, Italy, 11 Ospedale GB Morgagni-L Pierantoni, Forli, Italy, 12 Ospedale Infermi, Rimini, Italy, 13 Gastroenterology Unit, Ospedale San Giacomo, Rome, Italy
Correspondence to: S Bruno savino.bruno@fbf.milano.it
Objective To assess the incidence, cofactors, and excess risk of development of non-alcoholic fatty liver disease, including non-alcoholic steatohepatitis, attributable to tamoxifen in women.
Design Prospective, randomised, double blind, placebo controlled trial.
Setting and participants 5408 healthy women who had had hysterectomies, recruited into the Italian tamoxifen chemoprevention trial from 58 centres in Italy.
Intervention Women were randomly assigned to receive tamoxifen (20 mg daily) or placebo for five years.
Main outcome measure Development of non-alcoholic fatty liver disease in all women with normal baseline liver function who showed at least two elevations of alanine aminotransferase ( 1.5 times upper limit of normal) over a six month period.
Results During follow up, 64 women met the predefined criteria: 12 tested positive for hepatitis C virus, and the remaining 52 were suspected of having developed non-alcoholic fatty liver disease (34 tamoxifen, 18 placebo)—hazard ratio = 2.0 (95% confidence interval 1.1 to 3.5; P = 0.04). In all 52 women ultrasonography confirmed the presence of fatty liver. Other factors associated with the development of non-alcoholic fatty liver disease included overweight (2.4, 1.2 to 4.8), obesity (3.6, 1.7 to 7.6), hypercholesterolaemia (3.4, 1.4 to 7.8), and arterial hypertension (2.0, 1.0 to 3.8). Twenty women had liver biopsies: 15 were diagnosed as having mild to moderate steatohepatitis (12 tamoxifen, 3 placebo), and five had fatty liver alone (1 tamoxifen, 4 placebo). No clinical, biochemical, ultrasonic, or histological signs suggestive of progression to cirrhosis were observed after a median follow up of 8.7 years.
Conclusions Tamoxifen was associated with higher risk of development of non-alcoholic steatohepatitis only in overweight and obese women with features of metabolic syndrome, but the disease, in both the tamoxifen and the placebo group, after 10 years of follow up seems to be indolent.
Tamoxifen is a well known antioestrogen used for the hormone treatment of oestrogen receptor positive breast cancer, and its efficacy in preventing breast cancer in women at high risk has been recently recognised. It is associated with an increased risk of endometrial cancer and other adverse reactions,1 including the development of non-alcoholic fatty liver and non-alcoholic steatohepatitis.2-4 We aimed to evaluate the incidence of, potential risk factors for, and excess risk attributable to tamoxifen in non-alcoholic fatty liver diseases in a large cohort of women who had had hysterectomies and were originally enrolled in the Italian tamoxifen chemoprevention randomised trial.
Methods
The Italian tamoxifen chemoprevention trial included healthy women aged 35 to 70 who had had a total hysterectomy in order to avoid the associated risk of endometrial cancer.5 6 The study started in October 1992, and recruitment ended in 1997. Women were randomly assigned in a double blind manner to receive tamoxifen (20 mg daily) or placebo for five years. Women with severe concurrent illness were excluded from randomisation. At baseline, women had a thorough physical examination and blood testing, and a complete history was taken. Follow up included a physical examination every six months and blood testing and mammography every 12 months.6
Design of the study on incidence of non-alcoholic fatty liver disease
In 1992, in the absence of information about liver toxicity of tamoxifen, we included a surveillance programme in the Italian trial with the aim of identifying and assessing the onset of acute or chronic liver injury.6 We suspected development of non-alcoholic fatty liver disease in all women with normal baseline transaminase concentrations and no history of liver enzyme alterations who developed chronic unexplained hypertransaminasaemia defined as multiple elevations of alanine aminotransferase 1.5 times the normal upper limit over a six month period. We chose this arbitrary criterion to justify the morbidity and to limit the number of false positive liver biopsies in women at normal risk.7 For each woman identified, we collected detailed information on previous liver function tests, hepatitis B or C infection, autoantibodies indicative of autoimmune hepatitis, iron metabolism, alcohol intake, and use of hepatotoxic drugs, in order to exclude recognised causes of liver disease. Women with unexplained hypertransaminasaemia had a ultrasound examination of the liver and were offered liver biopsy for final diagnosis. Both procedures were done by operators blind to the treatment allocation. We invited all women with suspected non-alcoholic fatty liver disease to a repeat physical examination, blood tests, and ultrasonography in March 2004, and those who had had biopsies were invited to a repeat procedure.
Statistical analysis
We used Fisher's exact test to assess statistical differences in the frequency of the outcomes among women in the different study groups. We used the Kaplan-Meier method to assess the cumulative incidence of events reported during follow up and the log-rank test to compare the frequency of events between groups. We assessed the independent effect of multiple factors on the development of events during the trial by using Cox proportional hazards regression. We based all analyses on assigned treatment (intention to treat).
Results
From 1992 to 1997, 5408 women (median age 51 years) who were well and had had hysterectomies were randomised to receive tamoxifen (n = 2708) or placebo (n = 2700) for five years.5 Three hundred and twenty one (5.9%) women had alanine aminotransferase above the normal upper limit at baseline but were considered eligible to enter the trial. Of these, 223 maintained alanine aminotransferase < 1.5 times the normal upper limit during intervention, 52 had a concentration 1.5 times the normal upper limit at a single visit, and 46 had such a concentration on two or more occasions. Of the 5087 women with normal alanine aminotransferase at baseline, 4853 maintained normal concentrations during intervention, 170 had a single concentration 1.5 times the normal upper limit during intervention, and 64 had two or more results 1.5 times the normal upper limit over at least six months (table 1). Of these 64 women, 12 tested positive for hepatitis C virus. The remaining 52 women, who fulfilled our criteria, were invited to have a liver biopsy for final diagnosis. In all cases ultrasonic findings were consistent with the presence of fatty liver.
Table 1 Distribution of women in Italian tamoxifen chemoprevention trial according to alanine aminotransferase concentrations and hepatitis B or C virus profile
Women with high alanine aminotransferase at baseline were heavier, had a higher body mass index, and more often had diabetes than women with normal concentrations at baseline and during follow up (P < 0.0001). A similar proportion of women in these two groups had moderate to severe hypercholesterolaemia or hypertension normalised by treatment. Among women with normal alanine aminotransferase at baseline, development of suspected non-alcoholic fatty liver disease was associated with baseline hypercholesterolaemia (> 6.5 mmol/l) (P = 0.005) and hypertension (P = 0.15) (table 2).
Table 2 Baseline characteristics of women in Italian tamoxifen chemoprevention trial
The mean weight of the 52 women with suspected non-alcoholic fatty liver disease was 71.5 kg at randomisation and 74.4 kg at the first elevation of alanine aminotransferase (P < 0.0001). During the intervention, five women (two placebo, three tamoxifen) developed acute hepatitis (aminotransferase 10 times normal upper limit); four tested positive for hepatitis C, and for one (on placebo) the cause remained unknown.
Liver biopsies
Twenty women (13 tamoxifen, 7 placebo) had liver biopsy: 15 had steatohepatitis (12 tamoxifen, 3 placebo), and 5 had fatty liver (1 tamoxifen, 4 placebo). The positive predictive value for non-alcoholic steatohepatitis was 92% in the tamoxifen group and 43% in the placebo group. The mean weight of the 15 women with steatohepatitis was 70.2 kg at randomisation and 73.5 kg at the time of first elevation of alanine aminotransferase (P = 0.015), whereas the mean weight of the five women with fatty liver remained stable (69.2 kg at randomisation, 69.8 kg at first alanine aminotransferase elevation; P = 0.68). The women who refused biopsy did not differ in any of the considered characteristics from those who agreed to have the procedure (table 2). Six of the 15 women diagnosed as having non-alcoholic steatohepatitis had a mild liver biopsy score according to Brunt,8 and nine had a moderate score; 10 had grade 1 fibrosis, and five had grade 2 fibrosis, but no severe fibrosis occurred.
Three additional women with elevated alanine aminotransferase at baseline had liver biopsy on the advice of their family doctor or local hepatologist. All received tamoxifen and fulfilled our criteria for suspected non-alcoholic fatty liver disease. The first woman was obese, had diabetes, and was diagnosed as having liver cirrhosis. However, her low platelet count (105 000/mm3), prolonged prothrombin time, and aspartate:alanine aminotransferase ratio > 1 before enrolment suggested the presence of pre-existing cirrhosis. Of the two remaining women, one had moderate liver fibrosis and the other had severe fibrosis.
Cumulative incidence of suspected non-alcoholic fatty liver disease
Suspicion of non-alcoholic fatty liver disease was more common in the tamoxifen group (n = 34) than in the placebo group (n = 18) (log rank P = 0.017) (figure). This excess was, however, limited to the first two years of intervention (tamoxifen 21, placebo 7), and no excess was apparent thereafter (tamoxifen 13, placebo 11). The mean interval between randomisation and first elevation of alanine aminotransferase was 23 (range 6-50) months, similar in both study groups.
Cumulative incidence of suspected non-alcoholic fatty liver disease in women with normal alanine aminotransferase at baseline who participated in the Italian tamoxifen chemoprevention trial, according to baseline body mass index
Compared with the placebo group, the cumulative incidence of suspected non-alcoholic fatty liver disease after five years in the tamoxifen group was particularly high (3.8%, 95% confidence interval 1.6% to 6.0%) among obese women (body mass index > 30), moderately high (2.4%, 1.2% to 3.5%) among overweight women (body mass index 25-30), but similar (0.7%, 0.2% to 1.2%) in women of normal weight (body mass index < 25) (figure).
The incidence of suspected non-alcoholic fatty liver disease in women with normal alanine aminotransferase at baseline reached 0.37/100 women/year in the tamoxifen group compared with 0.19/100 in the placebo group. On the basis of the liver biopsy findings, we can reasonably assume that all women in the tamoxifen group and half of the women in the placebo group had steatohepatitis. Therefore, the rate of steatohepatitis would be approximately 0.4/100 women/year in the tamoxifen group compared with 0.1/100 in the placebo group, with an excess rate of steatohepatitis attributed to tamoxifen of 0.3/100 women/year. Similarly, among obese women the rate of steatohepatitis during the first two years of intervention would be 1.13/100 women/year in the tamoxifen group compared with 0.07/100 in the placebo group (table 3).
Table 3 Patient-years of observation and incidence of suspected non-alcoholic fatty liver disease (women with high alanine aminotransferase at baseline excluded) in Italian tamoxifen chemoprevention trial, according to body mass index and treatment group
Prognostic factors for tamoxifen associated suspected non-alcoholic fatty liver disease
Overall, tamoxifen was associated with an increased risk of developing suspected non-alcoholic fatty liver disease (hazard ratio = 2.0, 95% confidence interval 1.1 to 3.5), but the association was restricted to overweight women (2.3, 1.2 to 4.6). Other predictors of non-alcoholic fatty liver disease included overweight (2.4, 1.2 to 4.8), obesity (3.6, 1.7 to 7.6), severe baseline hypercholesterolaemia (> 7.8 mmol/l) (3.4, 1.4 to 7.8), and hypertension (2.0, 1.0 to 3.8) (table 4).
Table 4 Prognostic factors for development of suspected non-alcoholic fatty liver disease in women with normal baseline transaminase concentrations in Italian tamoxifen chemoprevention trial. Values are hazard ratios (95% confidence intervals) obtained from Cox regression model
Among overweight and obese women, tamoxifen and hypertension were independent significant predictors for the development of suspected non-alcoholic fatty liver disease. Among women in the placebo group, diabetes (10.6, 2.9 to 38.6) and severe hypercholesterolaemia (> 7.8 mmol/l) (8.4, 2.5 to 28.3) were associated with an excess risk of developing suspected non-alcoholic fatty liver disease. Among women in the tamoxifen group, only overweight (3.2, 1.3 to 7.9) and obesity (5.4, 2.1 to 14.1) were associated with an excess risk of developing suspected non-alcoholic fatty liver disease (table 4).
Follow up
Women with suspected non-alcoholic fatty liver disease were blindly invited to suspend treatment. During follow up (median 6.9 years, range 2.7-9.9), 22 women had persistently normalised liver enzymes, 14 had fluctuations of transaminases, and 16 maintained high liver enzymes; we found no difference between the tamoxifen and placebo groups. For the 20 women who had biopsies, the mean time between observation of the first elevation of alanine aminotransferase and last visit was 8.7 (range 4.9-10.7) years. In this group, 11 women had features of the metabolic syndrome (arterial hypertension, obesity, diabetes, dyslipidaemia), two had an abnormal oral glucose tolerance test, and 14 had a homoeostasis model assessment test result that suggested insulin resistance; no difference existed between study groups or histological diagnosis of fatty liver or steatohepatitis. No drug treatment for insulin resistance was initiated. During follow up, three of the women who had persistently abnormal alanine aminotransferase had a second liver biopsy after 5-6.5 years. These repeated biopsies showed no histological changes. In March 2004, according to our preestablished clinical, biochemical, and ultrasonic criteria, none of the 52 women had signs of cirrhosis.
Discussion
Fisher B, Costantino JP, Wickerham DL, Redmond CK, Kavanah M, Cronin WM, et al. Tamoxifen for prevention of breast cancer: report of the national surgical adjuvant breast and bowel project P-1 study. J Natl Cancer Inst 1998;90: 1371-88.
Pratt DS, Knox TA, Erban J. Tamoxifen induced steatohepatitis. Ann Intern Med 1995;123: 236.
Van Hoof M, Rahier J, Horsmans Y. Tamoxifen-induced steatohepatitis. Ann Intern Med 1996;124: 855-6.
Oien KA, Moffat D, Curry GW, Dickson J, Habeshaw T, Mills PR, et al. Cirrhosis with steatohepatitis after adjuvant tamoxifen. Lancet 1999;353: 36-7.
Veronesi U, Maisonneuve P, Costa A, Sacchini V, Maltoni C, Robertson C, et al. Prevention of breast cancer with tamoxifen: preliminary findings from the Italian randomised trial among hysterectomised women. Lancet 1998;352: 93-7.
Veronesi U. Prevention of breast cancer with tamoxifen: the Italian study in hysterectomized women. The Breast 1995;4: 267-72.
Van Ness MM, Diehl AM. Is liver biopsy useful in the evaluation of patients with chronically elevated liver enzymes? Ann Intern Med 1989;111: 473-8.
Brunt EM, Janney C, Di Bisceglie A, Neuschwander-Tetri B, Bacon B. Non-alcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999;94: 2467-74.
Yano E, Tagawa K, Yamaoka K, Mori M. Test validity of periodic liver function tests in a population of Japanese male bank employees. J Clin Epidemiol 2001;54: 945-51.
Mofrad P, Contos MJ, Haque M, Sargeant C, Fisher RA, Luketic VA, et al. Clinical and histologic spectrum of non-alcoholic fatty liver disease associated with normal ALT values. Hepatology 2003;37: 1286-92.
Marchesini G, Bugianesi E, Forlani G, Cerrelli F, Lenzi M, Manini R, et al. Non-alcoholic fatty liver, steatohepatitis and metabolic syndrome. Hepatology 2003;37: 917-23.
Matteoni CA, Younossi ZM, Gramlich T, Bhoparai N, Liu YC, McCollough AJ. Nonalcoholic fatty liver disease; a spectrum of clinical and pathological severity. Gastroenterology 1999;116: 1413-9.
Ratziu V, Giral P, Charlotte F, Bruckert E, Thibault V, Theodorou I, et al. Liver fibrosis in overweight patients. Gastroenterology 2000;118: 1117-23.
Lainè F, Bendavid G, Moirand R, Tessier S, Perrin N, Guillygomarc'h A, et al. Prediction of liver fibrosis in patients with features of the metabolic syndrome regardless of alcohol consumption. Hepatology 2004;39: 1639-46.
Nishino M, Hayakawa K, Nakamura Y, Morimoto T, Mukaihara S. Effects of tamoxifen on hepatic fat content and the development of hepatic steatosis in patients with breast cancer: high frequency of involvement and rapid reversal after completion of tamoxifen therapy. Am J Roentgenol 2003;180: 129-34.
Westin J, Nordlinder H, Lagging M, Norkrans G, Wejstal R. Steatosis accelerates fibrosis development over time in hepatitis C virus genotype 3 infected patients. J Hepatol 2002;37: 837-42.
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: 112-7.
Clark JM, Brancati FL, Diehl AM. Non-alcoholic fatty liver disease. Gastroenterology 2002;122: 1649-57.
Ludwig J, Viggiano TR, McGill DB, Oh BJ. Non-alcoholic steatohepatitis: Mayo Clinic experiences with a hitherto unnamed disease. Mayo Clin Proc 1980;55: 434-8.
Silverman JF, O'Brien KF, Long S, Leggett N, Khazanie PG, Pories WJ, et al. Liver pathology in morbidly obese patients with and without diabetes. Am J Gastroenterol 1990;85: 1345-55.
Bugianesi E, Leone N, Vanni E, Marchesini G, Brunello F, Crucci P, et al. Expanding the natural history of nonalcoholic steatohepatitis: from cryptogenic cirrhosis to hepatocellular carcinoma. Gastroenterology 2002;123: 134-40.
Powell EE, Cooksley GE, 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: 74-80.
Adams LA, Keach J, Lindor KD, Angulo P. Time course of fibrosis progression in patients with non alcoholic fatty liver disease. Hepatology 2003;38: 104.
Charlton M, Kasparova P, Weston S, Lindor K, Maor-Kendler Y, Wiesner RH, et al. Frequency of non-alcoholic steatohepatitis as a cause of advanced liver disease. Liver Transpl 2001;7: 608-14.(Savino Bruno, director1, Patrick Maisonn)
Correspondence to: S Bruno savino.bruno@fbf.milano.it
Objective To assess the incidence, cofactors, and excess risk of development of non-alcoholic fatty liver disease, including non-alcoholic steatohepatitis, attributable to tamoxifen in women.
Design Prospective, randomised, double blind, placebo controlled trial.
Setting and participants 5408 healthy women who had had hysterectomies, recruited into the Italian tamoxifen chemoprevention trial from 58 centres in Italy.
Intervention Women were randomly assigned to receive tamoxifen (20 mg daily) or placebo for five years.
Main outcome measure Development of non-alcoholic fatty liver disease in all women with normal baseline liver function who showed at least two elevations of alanine aminotransferase ( 1.5 times upper limit of normal) over a six month period.
Results During follow up, 64 women met the predefined criteria: 12 tested positive for hepatitis C virus, and the remaining 52 were suspected of having developed non-alcoholic fatty liver disease (34 tamoxifen, 18 placebo)—hazard ratio = 2.0 (95% confidence interval 1.1 to 3.5; P = 0.04). In all 52 women ultrasonography confirmed the presence of fatty liver. Other factors associated with the development of non-alcoholic fatty liver disease included overweight (2.4, 1.2 to 4.8), obesity (3.6, 1.7 to 7.6), hypercholesterolaemia (3.4, 1.4 to 7.8), and arterial hypertension (2.0, 1.0 to 3.8). Twenty women had liver biopsies: 15 were diagnosed as having mild to moderate steatohepatitis (12 tamoxifen, 3 placebo), and five had fatty liver alone (1 tamoxifen, 4 placebo). No clinical, biochemical, ultrasonic, or histological signs suggestive of progression to cirrhosis were observed after a median follow up of 8.7 years.
Conclusions Tamoxifen was associated with higher risk of development of non-alcoholic steatohepatitis only in overweight and obese women with features of metabolic syndrome, but the disease, in both the tamoxifen and the placebo group, after 10 years of follow up seems to be indolent.
Tamoxifen is a well known antioestrogen used for the hormone treatment of oestrogen receptor positive breast cancer, and its efficacy in preventing breast cancer in women at high risk has been recently recognised. It is associated with an increased risk of endometrial cancer and other adverse reactions,1 including the development of non-alcoholic fatty liver and non-alcoholic steatohepatitis.2-4 We aimed to evaluate the incidence of, potential risk factors for, and excess risk attributable to tamoxifen in non-alcoholic fatty liver diseases in a large cohort of women who had had hysterectomies and were originally enrolled in the Italian tamoxifen chemoprevention randomised trial.
Methods
The Italian tamoxifen chemoprevention trial included healthy women aged 35 to 70 who had had a total hysterectomy in order to avoid the associated risk of endometrial cancer.5 6 The study started in October 1992, and recruitment ended in 1997. Women were randomly assigned in a double blind manner to receive tamoxifen (20 mg daily) or placebo for five years. Women with severe concurrent illness were excluded from randomisation. At baseline, women had a thorough physical examination and blood testing, and a complete history was taken. Follow up included a physical examination every six months and blood testing and mammography every 12 months.6
Design of the study on incidence of non-alcoholic fatty liver disease
In 1992, in the absence of information about liver toxicity of tamoxifen, we included a surveillance programme in the Italian trial with the aim of identifying and assessing the onset of acute or chronic liver injury.6 We suspected development of non-alcoholic fatty liver disease in all women with normal baseline transaminase concentrations and no history of liver enzyme alterations who developed chronic unexplained hypertransaminasaemia defined as multiple elevations of alanine aminotransferase 1.5 times the normal upper limit over a six month period. We chose this arbitrary criterion to justify the morbidity and to limit the number of false positive liver biopsies in women at normal risk.7 For each woman identified, we collected detailed information on previous liver function tests, hepatitis B or C infection, autoantibodies indicative of autoimmune hepatitis, iron metabolism, alcohol intake, and use of hepatotoxic drugs, in order to exclude recognised causes of liver disease. Women with unexplained hypertransaminasaemia had a ultrasound examination of the liver and were offered liver biopsy for final diagnosis. Both procedures were done by operators blind to the treatment allocation. We invited all women with suspected non-alcoholic fatty liver disease to a repeat physical examination, blood tests, and ultrasonography in March 2004, and those who had had biopsies were invited to a repeat procedure.
Statistical analysis
We used Fisher's exact test to assess statistical differences in the frequency of the outcomes among women in the different study groups. We used the Kaplan-Meier method to assess the cumulative incidence of events reported during follow up and the log-rank test to compare the frequency of events between groups. We assessed the independent effect of multiple factors on the development of events during the trial by using Cox proportional hazards regression. We based all analyses on assigned treatment (intention to treat).
Results
From 1992 to 1997, 5408 women (median age 51 years) who were well and had had hysterectomies were randomised to receive tamoxifen (n = 2708) or placebo (n = 2700) for five years.5 Three hundred and twenty one (5.9%) women had alanine aminotransferase above the normal upper limit at baseline but were considered eligible to enter the trial. Of these, 223 maintained alanine aminotransferase < 1.5 times the normal upper limit during intervention, 52 had a concentration 1.5 times the normal upper limit at a single visit, and 46 had such a concentration on two or more occasions. Of the 5087 women with normal alanine aminotransferase at baseline, 4853 maintained normal concentrations during intervention, 170 had a single concentration 1.5 times the normal upper limit during intervention, and 64 had two or more results 1.5 times the normal upper limit over at least six months (table 1). Of these 64 women, 12 tested positive for hepatitis C virus. The remaining 52 women, who fulfilled our criteria, were invited to have a liver biopsy for final diagnosis. In all cases ultrasonic findings were consistent with the presence of fatty liver.
Table 1 Distribution of women in Italian tamoxifen chemoprevention trial according to alanine aminotransferase concentrations and hepatitis B or C virus profile
Women with high alanine aminotransferase at baseline were heavier, had a higher body mass index, and more often had diabetes than women with normal concentrations at baseline and during follow up (P < 0.0001). A similar proportion of women in these two groups had moderate to severe hypercholesterolaemia or hypertension normalised by treatment. Among women with normal alanine aminotransferase at baseline, development of suspected non-alcoholic fatty liver disease was associated with baseline hypercholesterolaemia (> 6.5 mmol/l) (P = 0.005) and hypertension (P = 0.15) (table 2).
Table 2 Baseline characteristics of women in Italian tamoxifen chemoprevention trial
The mean weight of the 52 women with suspected non-alcoholic fatty liver disease was 71.5 kg at randomisation and 74.4 kg at the first elevation of alanine aminotransferase (P < 0.0001). During the intervention, five women (two placebo, three tamoxifen) developed acute hepatitis (aminotransferase 10 times normal upper limit); four tested positive for hepatitis C, and for one (on placebo) the cause remained unknown.
Liver biopsies
Twenty women (13 tamoxifen, 7 placebo) had liver biopsy: 15 had steatohepatitis (12 tamoxifen, 3 placebo), and 5 had fatty liver (1 tamoxifen, 4 placebo). The positive predictive value for non-alcoholic steatohepatitis was 92% in the tamoxifen group and 43% in the placebo group. The mean weight of the 15 women with steatohepatitis was 70.2 kg at randomisation and 73.5 kg at the time of first elevation of alanine aminotransferase (P = 0.015), whereas the mean weight of the five women with fatty liver remained stable (69.2 kg at randomisation, 69.8 kg at first alanine aminotransferase elevation; P = 0.68). The women who refused biopsy did not differ in any of the considered characteristics from those who agreed to have the procedure (table 2). Six of the 15 women diagnosed as having non-alcoholic steatohepatitis had a mild liver biopsy score according to Brunt,8 and nine had a moderate score; 10 had grade 1 fibrosis, and five had grade 2 fibrosis, but no severe fibrosis occurred.
Three additional women with elevated alanine aminotransferase at baseline had liver biopsy on the advice of their family doctor or local hepatologist. All received tamoxifen and fulfilled our criteria for suspected non-alcoholic fatty liver disease. The first woman was obese, had diabetes, and was diagnosed as having liver cirrhosis. However, her low platelet count (105 000/mm3), prolonged prothrombin time, and aspartate:alanine aminotransferase ratio > 1 before enrolment suggested the presence of pre-existing cirrhosis. Of the two remaining women, one had moderate liver fibrosis and the other had severe fibrosis.
Cumulative incidence of suspected non-alcoholic fatty liver disease
Suspicion of non-alcoholic fatty liver disease was more common in the tamoxifen group (n = 34) than in the placebo group (n = 18) (log rank P = 0.017) (figure). This excess was, however, limited to the first two years of intervention (tamoxifen 21, placebo 7), and no excess was apparent thereafter (tamoxifen 13, placebo 11). The mean interval between randomisation and first elevation of alanine aminotransferase was 23 (range 6-50) months, similar in both study groups.
Cumulative incidence of suspected non-alcoholic fatty liver disease in women with normal alanine aminotransferase at baseline who participated in the Italian tamoxifen chemoprevention trial, according to baseline body mass index
Compared with the placebo group, the cumulative incidence of suspected non-alcoholic fatty liver disease after five years in the tamoxifen group was particularly high (3.8%, 95% confidence interval 1.6% to 6.0%) among obese women (body mass index > 30), moderately high (2.4%, 1.2% to 3.5%) among overweight women (body mass index 25-30), but similar (0.7%, 0.2% to 1.2%) in women of normal weight (body mass index < 25) (figure).
The incidence of suspected non-alcoholic fatty liver disease in women with normal alanine aminotransferase at baseline reached 0.37/100 women/year in the tamoxifen group compared with 0.19/100 in the placebo group. On the basis of the liver biopsy findings, we can reasonably assume that all women in the tamoxifen group and half of the women in the placebo group had steatohepatitis. Therefore, the rate of steatohepatitis would be approximately 0.4/100 women/year in the tamoxifen group compared with 0.1/100 in the placebo group, with an excess rate of steatohepatitis attributed to tamoxifen of 0.3/100 women/year. Similarly, among obese women the rate of steatohepatitis during the first two years of intervention would be 1.13/100 women/year in the tamoxifen group compared with 0.07/100 in the placebo group (table 3).
Table 3 Patient-years of observation and incidence of suspected non-alcoholic fatty liver disease (women with high alanine aminotransferase at baseline excluded) in Italian tamoxifen chemoprevention trial, according to body mass index and treatment group
Prognostic factors for tamoxifen associated suspected non-alcoholic fatty liver disease
Overall, tamoxifen was associated with an increased risk of developing suspected non-alcoholic fatty liver disease (hazard ratio = 2.0, 95% confidence interval 1.1 to 3.5), but the association was restricted to overweight women (2.3, 1.2 to 4.6). Other predictors of non-alcoholic fatty liver disease included overweight (2.4, 1.2 to 4.8), obesity (3.6, 1.7 to 7.6), severe baseline hypercholesterolaemia (> 7.8 mmol/l) (3.4, 1.4 to 7.8), and hypertension (2.0, 1.0 to 3.8) (table 4).
Table 4 Prognostic factors for development of suspected non-alcoholic fatty liver disease in women with normal baseline transaminase concentrations in Italian tamoxifen chemoprevention trial. Values are hazard ratios (95% confidence intervals) obtained from Cox regression model
Among overweight and obese women, tamoxifen and hypertension were independent significant predictors for the development of suspected non-alcoholic fatty liver disease. Among women in the placebo group, diabetes (10.6, 2.9 to 38.6) and severe hypercholesterolaemia (> 7.8 mmol/l) (8.4, 2.5 to 28.3) were associated with an excess risk of developing suspected non-alcoholic fatty liver disease. Among women in the tamoxifen group, only overweight (3.2, 1.3 to 7.9) and obesity (5.4, 2.1 to 14.1) were associated with an excess risk of developing suspected non-alcoholic fatty liver disease (table 4).
Follow up
Women with suspected non-alcoholic fatty liver disease were blindly invited to suspend treatment. During follow up (median 6.9 years, range 2.7-9.9), 22 women had persistently normalised liver enzymes, 14 had fluctuations of transaminases, and 16 maintained high liver enzymes; we found no difference between the tamoxifen and placebo groups. For the 20 women who had biopsies, the mean time between observation of the first elevation of alanine aminotransferase and last visit was 8.7 (range 4.9-10.7) years. In this group, 11 women had features of the metabolic syndrome (arterial hypertension, obesity, diabetes, dyslipidaemia), two had an abnormal oral glucose tolerance test, and 14 had a homoeostasis model assessment test result that suggested insulin resistance; no difference existed between study groups or histological diagnosis of fatty liver or steatohepatitis. No drug treatment for insulin resistance was initiated. During follow up, three of the women who had persistently abnormal alanine aminotransferase had a second liver biopsy after 5-6.5 years. These repeated biopsies showed no histological changes. In March 2004, according to our preestablished clinical, biochemical, and ultrasonic criteria, none of the 52 women had signs of cirrhosis.
Discussion
Fisher B, Costantino JP, Wickerham DL, Redmond CK, Kavanah M, Cronin WM, et al. Tamoxifen for prevention of breast cancer: report of the national surgical adjuvant breast and bowel project P-1 study. J Natl Cancer Inst 1998;90: 1371-88.
Pratt DS, Knox TA, Erban J. Tamoxifen induced steatohepatitis. Ann Intern Med 1995;123: 236.
Van Hoof M, Rahier J, Horsmans Y. Tamoxifen-induced steatohepatitis. Ann Intern Med 1996;124: 855-6.
Oien KA, Moffat D, Curry GW, Dickson J, Habeshaw T, Mills PR, et al. Cirrhosis with steatohepatitis after adjuvant tamoxifen. Lancet 1999;353: 36-7.
Veronesi U, Maisonneuve P, Costa A, Sacchini V, Maltoni C, Robertson C, et al. Prevention of breast cancer with tamoxifen: preliminary findings from the Italian randomised trial among hysterectomised women. Lancet 1998;352: 93-7.
Veronesi U. Prevention of breast cancer with tamoxifen: the Italian study in hysterectomized women. The Breast 1995;4: 267-72.
Van Ness MM, Diehl AM. Is liver biopsy useful in the evaluation of patients with chronically elevated liver enzymes? Ann Intern Med 1989;111: 473-8.
Brunt EM, Janney C, Di Bisceglie A, Neuschwander-Tetri B, Bacon B. Non-alcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999;94: 2467-74.
Yano E, Tagawa K, Yamaoka K, Mori M. Test validity of periodic liver function tests in a population of Japanese male bank employees. J Clin Epidemiol 2001;54: 945-51.
Mofrad P, Contos MJ, Haque M, Sargeant C, Fisher RA, Luketic VA, et al. Clinical and histologic spectrum of non-alcoholic fatty liver disease associated with normal ALT values. Hepatology 2003;37: 1286-92.
Marchesini G, Bugianesi E, Forlani G, Cerrelli F, Lenzi M, Manini R, et al. Non-alcoholic fatty liver, steatohepatitis and metabolic syndrome. Hepatology 2003;37: 917-23.
Matteoni CA, Younossi ZM, Gramlich T, Bhoparai N, Liu YC, McCollough AJ. Nonalcoholic fatty liver disease; a spectrum of clinical and pathological severity. Gastroenterology 1999;116: 1413-9.
Ratziu V, Giral P, Charlotte F, Bruckert E, Thibault V, Theodorou I, et al. Liver fibrosis in overweight patients. Gastroenterology 2000;118: 1117-23.
Lainè F, Bendavid G, Moirand R, Tessier S, Perrin N, Guillygomarc'h A, et al. Prediction of liver fibrosis in patients with features of the metabolic syndrome regardless of alcohol consumption. Hepatology 2004;39: 1639-46.
Nishino M, Hayakawa K, Nakamura Y, Morimoto T, Mukaihara S. Effects of tamoxifen on hepatic fat content and the development of hepatic steatosis in patients with breast cancer: high frequency of involvement and rapid reversal after completion of tamoxifen therapy. Am J Roentgenol 2003;180: 129-34.
Westin J, Nordlinder H, Lagging M, Norkrans G, Wejstal R. Steatosis accelerates fibrosis development over time in hepatitis C virus genotype 3 infected patients. J Hepatol 2002;37: 837-42.
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: 112-7.
Clark JM, Brancati FL, Diehl AM. Non-alcoholic fatty liver disease. Gastroenterology 2002;122: 1649-57.
Ludwig J, Viggiano TR, McGill DB, Oh BJ. Non-alcoholic steatohepatitis: Mayo Clinic experiences with a hitherto unnamed disease. Mayo Clin Proc 1980;55: 434-8.
Silverman JF, O'Brien KF, Long S, Leggett N, Khazanie PG, Pories WJ, et al. Liver pathology in morbidly obese patients with and without diabetes. Am J Gastroenterol 1990;85: 1345-55.
Bugianesi E, Leone N, Vanni E, Marchesini G, Brunello F, Crucci P, et al. Expanding the natural history of nonalcoholic steatohepatitis: from cryptogenic cirrhosis to hepatocellular carcinoma. Gastroenterology 2002;123: 134-40.
Powell EE, Cooksley GE, 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: 74-80.
Adams LA, Keach J, Lindor KD, Angulo P. Time course of fibrosis progression in patients with non alcoholic fatty liver disease. Hepatology 2003;38: 104.
Charlton M, Kasparova P, Weston S, Lindor K, Maor-Kendler Y, Wiesner RH, et al. Frequency of non-alcoholic steatohepatitis as a cause of advanced liver disease. Liver Transpl 2001;7: 608-14.(Savino Bruno, director1, Patrick Maisonn)