CLINICAL LIVER, PANCREAS, AND BILIARY TRACT

Transcription

1 GASTROENTEROLOGY 2005;128:24 32 CLINICAL LIVER, PANCREAS, AND BILIARY TRACT Coffee and Caffeine Consumption Reduce the Risk of Elevated Serum Alanine Aminotransferase Activity in the United States CONSTANCE E. RUHL* and JAMES E. EVERHART *Social and Scientific Systems, Inc, Silver Spring; and National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Department of Health and Human Services, Bethesda, Maryland Background & Aims: Based on experimental and epidemiologic studies, we investigated whether coffee and caffeine consumption reduced the risk of elevated alanine aminotransferase (ALT) activity in persons at high risk for liver injury in a national, population-based study. Methods: Participants were 5944 adults in the Third US National Health and Nutrition Examination Survey, , with excessive alcohol consumption, viral hepatitis, iron overload, overweight, or impaired glucose metabolism. Liver injury was indicated by abnormal serum ALT activity (>43 U/L). Results: Elevated ALT activity was found in 8.7% of this high-risk population. In unadjusted analysis, lower ALT activity was associated with increasing consumption of coffee (P.001) and caffeine (P.001). Multivariate logistic regression analyses showed that the risk of elevated ALT activity declined with increasing intake of coffee (P for trend.034) and caffeine (P <.001). Comparing persons who drank more than 2 cups per day with noncoffee drinkers, the odds ratio was.56 (95% confidence interval, ). Comparing persons in the highest caffeine quintile with the lowest, the odds ratio was.31 (95% confidence interval,.16.61). These relationships were consistent across subgroups at risk for liver injury and were relatively unchanged when analyses included the entire population or when limited to persons without impaired liver function or right upper quadrant pain. Fasting insulin concentrations did not mediate the effects. Conclusions: In this large, national, population-based study, among persons at high risk for liver injury, consumption of coffee and especially caffeine was associated with lower risk of elevated ALT activity. Other than restricting ethanol consumption, there is little evidence-based nutritional advice to provide patients with chronic liver disease. With regard to coffee drinking, experimental data have shown potential beneficial effects on the liver of various coffee components, including caffeine, 1 the coffee oils (kahweol and cafestol), 2,3 and aromatic extracts isolated from coffee beans. 4 Epidemiologic surveys have found an association of heavier coffee consumption with lower levels of liver enzymes, mainly -glutamyltransferase A few population-based studies conducted outside of North America found a similar relationship of coffee drinking with serum alanine aminotransferase (ALT) activity, a more specific marker of liver injury than -glutamyltransferase. 7,13,18,19 Furthermore, coffee drinking was inversely related to alcoholic and nonalcoholic liver cirrhosis in case-control studies and to alcoholic cirrhosis in prospective studies. 23,24 The relationship of liver outcomes with caffeine, 10,21 usually considered to be the most biologically active component of coffee, 25 has also received little attention. Research on the effects of coffee or caffeine on the liver in the United States has been limited to Mexican Americans 26 and to a study of -glutamyltransferase in bus drivers. 10 Whether or not either coffee drinking or caffeine consumption may have beneficial effects on the liver in the general US population is unknown. Any potential beneficial effects of coffee or caffeine on the liver might be manifested only in persons at risk for liver injury from some other factor, such as viral hepatitis infection or heavy alcohol consumption. In some studies of liver outcomes, the inverse association was limited to, or stronger in, persons at higher risk for liver injury, such as heavier drinkers, smokers, or the obese. 8,10,13 15,19,20,23 Consequently, studies of the entire general population, most of whom are at low risk for liver injury, may have limited power to detect a potential protective effect. Therefore, our primary analyses concerned the association of coffee and caffeine consumption with liver injury in persons at high risk for the most common causes of chronic liver disease. We first examined whether elevated serum ALT activity was less common with increasing Abbreviations used in this paper: BMI, body mass index; NHANES III, Third National Health and Nutrition Examination Survey by the American Gastroenterological Association /05/$30.00 doi: /j.gastro

2 January 2005 COFFEE, CAFFEINE, AND ELEVATED ALT ACTIVITY 25 coffee and caffeine intake. Additional analyses determined the consistency and robustness of these effects and whether they could be mediated by improved insulin sensitivity. Patients and Methods The Third National Health and Nutrition Examination Survey (NHANES III) was conducted in the United States from 1988 to 1994 by the National Center for Health Statistics of the Centers for Disease Control and Prevention. 27 It consisted of interview, examination, and laboratory data collected from a complex multistage, stratified, clustered probability sample of the civilian, noninstitutionalized population age 2 months and older, with oversampling of the elderly, non-hispanic blacks, and Mexican Americans. The study was approved by the institutional review board of the Centers for Disease Control and Prevention, and all participants provided written consent to participate. Study Sample Of 23,258 sampled persons aged 20 years and older, 16,573 (71%) attended an examination at a mobile examination center. We included all participants at high risk for liver injury based on one or more of the following criteria: 2 alcoholic drinks per day 28 (n 863), positive serum hepatitis B surface antigen and core antibody (n 69), positive serum hepatitis C antibody (n 379), transferrin saturation 50% (n 473), hemoglobin A 1C 6.5% (95th percentile) (n 1185), diagnosed diabetes (n 1163), or both body mass index (BMI) 26.9 kg/m 2 (60th percentile) and waist-to-hip ratio.94 (60th percentile) (n 4021). In this population, levels of BMI and waist-to-hip ratio of at least the 60th percentile were associated with a high prevalence of abnormal ALT activity. 29 Twenty-two percent of participants had more than one risk factor. We excluded persons with type 1 diabetes mellitus (n 22), a current pregnancy (n 69), or missing data on serum ALT activity (n 302) or consumption of caffeine-containing beverages (n 19). The sample for the primary analysis, therefore, consisted of 5944 persons. Variables A venous blood sample was collected and shipped weekly at 20 C. Serum ALT concentration was assayed using a Hitachi 737 Analyzer (Boehringer-Mannheim Diagnostics, Indianapolis, IN). 30 The outcome was defined as a serum ALT level above the upper limit of normal of the NHANES III reference laboratory ( 43 U/L). 27 Participants were asked how often they drank regular coffee with caffeine in the past month. They were not specifically asked about decaffeinated coffee consumption. Coffee consumption (cups per day) was coded as none, 1, 1 2, and 2. Participants were asked about consumption in the past month of regular tea with caffeine as well as regular and diet colas and sodas. Total caffeine intake from beverages (mg/day) was estimated by summing caffeine from regular coffee (136 mg per cup), regular tea (64 mg per cup), and regular and diet colas and sodas (46 mg per bottle or can) and dividing by Total caffeine intake was expressed as quintiles. Persons at high risk for liver injury were identified. Participants were asked how many days of the year they drank alcoholic beverages and the average number of drinks consumed on drinking days during the past 12 months. Doctordiagnosed diabetes was determined by interview. BMI (weight [kg]/height [m 2 ]) and waist and hip circumferences were measured at the examination. 29,32 Hepatitis B surface antigen was measured by radioimmunoassay (Abbott Diagnostics, Chicago, IL) and hepatitis C antibody by enzyme immunoassay (Abbott Diagnostics). Serum iron and total iron-binding capacity were measured by a modified automated AAII-25 colorimetric method using an RFA or RFA 300 Automated Ferrozine Colorimetric Analyzer System (Alpkem, Inc, Clackamas, OR). Serum transferrin saturation was calculated by dividing the serum iron concentration by the serum total iron-binding capacity and multiplying by 100. Hemoglobin A 1C was assayed using high-performance liquid chromatography (Bio- Rad Laboratories, Hercules, CA). Serum albumin and total bilirubin concentrations were measured with a Hitachi 737 Analyzer (Boehringer-Mannheim Diagnostics). Fasting serum insulin concentration was measured by radioimmunoassay (Pharmacia Diagnostics AB, Uppsala, Sweden). Data were collected on the following factors known or believed to be related to liver disease and included as covariates in multivariate analyses: age (years), sex, ethnicity (non-hispanic white, non-hispanic black, Mexican American, other), and cigarette smoking (never, former, less than one pack per day, one or more packs per day). 29 Participants were asked about right upper quadrant pain in the past 12 months. Statistical Analysis All analyses incorporated sample weights, stratification, and clustering using SUDAAN software. 33 We examined the relation of elevated serum ALT activity with coffee and caffeine consumption by first comparing the prevalence of elevated ALT activity among coffee and caffeine intake categories using the Crosstab procedure. To further study these relations, while controlling for the effects of other variables related to abnormal ALT activity, we used multiple logistic regression analysis (logistic procedure). An odds ratio and 95% confidence interval for each category of coffee or caffeine consumption was calculated relative to the lowest category of consumption. Additionally, a test for trend of the odds ratios evaluated the association of elevated ALT activity with coffee categories and caffeine quintiles coded as ordinal variables. Two-way interaction terms were evaluated. Additional analyses were conducted to examine the consistency of relationships of elevated ALT activity with coffee and caffeine consumption within subgroups at risk for liver injury (ie, excessive alcohol intake, viral hepatitis, iron overload, abnormal glucose status, and elevated BMI and waist-to-hip ratio). Multivariate analy-

3 26 RUHL AND EVERHART GASTROENTEROLOGY Vol. 128, No. 1 Table 1. Characteristics of Study Sample by Coffee Consumption Category Characteristic Coffee (cups/day) P a Coffee intake distribution (%) Total sample Sex Men Women Ethnicity Non-Hispanic white Non-Hispanic black Mexican American Other Cigarette smoking Never Former pack per day pack per day Liver injury risk subgroup b Excessive alcohol intake.006 No Yes Viral hepatitis.75 No Yes Elevated transferrin saturation.061 No Yes Abnormal glucose status.018 No Yes High BMI and waist-to-hip ratio.87 No Yes Continuous variables (mean SE) Coffee (cups/day) Serum ALT activity (U/L) Total caffeine from beverages (mg/day) Age (y) a From 2 test for categorical variables or analysis of variance for continuous variables. b High risk defined as 2 alcoholic drinks per day, hepatitis B positive, hepatitis C positive, transferrin saturation 50%, hemoglobin A 1C 6.5% (95th percentile), diagnosed diabetes, or BMI 26.8 kg/m 2 (60th percentile) and waist-to-hip ratio.94 (60th percentile). Subgroups are not mutually exclusive. ses excluded persons with missing values for any factor included in the model. A similar analysis using linear regression was performed with ALT activity as a continuous variable (log base 10 transformed). A P value of.05 was considered statistically significant. Results Coffee consumption ranged from 0 to 20 cups per day. The median intake among persons drinking 0, 1, 1 2, and 2 cups/day was 0,.30, 1.0, and 3.0 cups/day, respectively. Consumption of caffeine from beverages ranged from 0 to 2954 mg/day. The median intake by quintile was 20, 104, 174, 278, and 570 mg/day, respectively. Fifty-one percent of caffeine intake was from coffee. Among this population at high risk for liver injury, elevated ALT activity was found in 8.7% and ALT activity of at least twice that of normal was found in 1.6%. The prevalence of elevated ALT activity among subgroups at risk for liver injury was 8.5% for excessive alcohol intake, 28.9% for viral hepatitis, 10.5% for high transferrin saturation, 8.0% for abnormal glucose status, and 8.8% with high BMI and waist-to-hip ratio. As shown in Tables 1 and 2, mean serum ALT activity decreased with higher consumption of both coffee (P.001) and caffeine (P.001). The prevalence of elevated ALT activity declined with increasing coffee and caffeine consumption (Figure 1). Consistent with the trends in Figure 1, unadjusted logistic regression analyses revealed

4 January 2005 COFFEE, CAFFEINE, AND ELEVATED ALT ACTIVITY 27 Table 2. Characteristics of Study Sample by Caffeine Consumption Quintile Caffeine quintiles (mg/day) Characteristic P a Caffeine intake distribution (%) Total sample Sex Men Women Ethnicity Non-Hispanic white Non-Hispanic black Mexican American Other Cigarette smoking Never Former pack per day pack per day Liver injury risk subgroup b Excessive alcohol intake.004 No Yes Viral hepatitis.34 No Yes Elevated transferrin saturation.42 No Yes Abnormal glucose status.10 No Yes High BMI and waist-to-hip ratio.46 No Yes Continuous variables (mean SE) Total caffeine from beverages (mg/day) Serum ALT activity (U/L) Coffee (cups/day) Age (y) a From 2 test for categorical variables or analysis of variance for continuous variables. b High risk defined as 2 alcoholic drinks per day, hepatitis B positive, hepatitis C positive, transferrin saturation 50%, hemoglobin A 1C 6.5% (95th percentile), diagnosed diabetes, or BMI 26.8 kg/m 2 (60th percentile) and waist-to-hip ratio.94 (60th percentile). Subgroups are not mutually exclusive. a lower risk of elevated ALT activity with increasing intake of coffee and caffeine (Table 3). The relationships of coffee and caffeine consumption with abnormal ALT activity were further examined in individual multivariate models after adjusting for other factors associated with elevated ALT activity (Table 3). Coffee and caffeine intake remained inversely associated with an increased prevalence of elevated ALT activity, although the relationship was stronger with caffeine. Participants in the highest caffeine intake quintile had less than one third the risk of elevated ALT activity compared with those in the lowest quintile (odds ratio,.31; 95% confidence interval,.16.61). Persons who drank more than 2 cups of coffee per day had just over one half the risk of elevated ALT activity compared with noncoffee drinkers, although this relationship no longer reached statistical significance in multivariate analysis (P.058). However, a significant test for trend remained with coffee drinking after adjusting for other factors associated with elevated ALT activity (P.034). This trend was stronger when noncoffee drinkers and drinkers of 1 cup/day were combined into one group (P.009). Addition of quadratic terms to the multivariate logistic regression analyses to test for nonlinear associations with elevated ALT activity did not significantly improve the fit of the model for either coffee (P.11) or caffeine (P.40). No statistically significant interactions were found for either coffee or caffeine intake with other risk factors for liver disease.

5 28 RUHL AND EVERHART GASTROENTEROLOGY Vol. 128, No. 1 Figure 1. Prevalence of elevated ALT activity by caffeine intake quintiles and coffee consumption categories (n 5944). Error bars represent standard errors of percentages. *P.05 compared with lowest quintile for caffeine or 0 cups/day for coffee. To examine whether caffeine explains the association of coffee drinking with elevated ALT activity, coffee and caffeine consumption were included in the same analysis. In unadjusted analysis, a lower prevalence of elevated ALT activity was associated with greater intake of caffeine (P value for trend.029) but was unrelated to coffee consumption independent of caffeine intake (P.88). In multivariate analysis adjusting for demographics and smoking, the inverse association remained with caffeine consumption (P.019), while coffee drinking remained unrelated (P.55). However, because intakes of coffee and caffeine were highly correlated (Spearman correlation coefficient.86), these results should be interpreted with caution. We examined the consistency of the associations of coffee and caffeine consumption with elevated ALT activity across subgroups at risk for liver injury (Table 4). In each subgroup, an inverse relationship was found, although only the association with caffeine among persons with high BMI and waist-to-hip ratio reached statistical significance in multivariate analyses. Caffeine metabolism is decreased in persons with significantly impaired liver function, which, theoretically, could lead to a lower consumption of coffee and other caffeinated beverages. Therefore, we performed a secondary analysis limited to the subgroup with normal concentrations of serum albumin and total bilirubin. We also excluded persons reporting right upper quadrant pain in the past 12 months, in case this might have influenced coffee or caffeine intake. Persons with abnormal liver function tests or right upper quadrant pain (n 973; 622 with decreased liver function, 298 with pain, and 53 with both) consumed less coffee (mean SE, vs cups/day; P.020) and caffeine (mean SE, vs mg/day; P.011) compared with persons with normal liver function tests. However, among 4366 participants with normal liver function tests and no pain, the inverse associations in multivariate analyses were not diminished between elevated ALT activity and consumption of coffee (P value for trend.029) or caffeine (P.001). Table 3. Relationships of Coffee and Caffeine Consumption With Elevated ALT Activity in Persons at High Risk of Liver Injury (N 5944) Unadjusted Multivariate adjusted OR a 95% CI P for trend OR a,b 95% CI P for trend Coffee (cups/day) Caffeine (mg/day) to to to NOTE. High risk defined as 2 alcoholic drinks per day, hepatitis B positive, hepatitis C positive, transferrin saturation 50%, hemoglobin A 1C 6.5% (95th percentile), diagnosed diabetes, or BMI 26.8 kg/m 2 (60th percentile) and waist-to-hip ratio.94 (60th percentile). OR, odds ratio; CI, confidence interval. a From logistic regression analysis. b Adjusted for age, sex, ethnicity, and cigarette smoking.

6 January 2005 COFFEE, CAFFEINE, AND ELEVATED ALT ACTIVITY 29 Table 4. Relationships of Coffee and Caffeine Consumption With Elevated ALT Activity Among Subgroups at Risk of Liver Injury Unadjusted Multivariate adjusted OR a 95% CI P for trend OR a,b 95% CI P for trend Coffee categories Entire high-risk group (N 5944) Excessive alcohol intake (n 863) Viral hepatitis (n 443) Elevated transferrin saturation (n 473) Abnormal glucose status (n 1582) High BMI and waist-to-hip ratio (n 4021) Caffeine quintiles Entire high-risk group Excessive alcohol intake Viral hepatitis Elevated transferrin saturation Abnormal glucose status High BMI and waist-to-hip ratio NOTE. High risk defined as 2 alcoholic drinks per day, hepatitis B and/or C positive, transferrin saturation 50%, diagnosed diabetes and/or hemoglobin A 1C 6.5% (95th percentile), and BMI 26.8 kg/m 2 (60th percentile) and waist-to-hip ratio.94 (60th percentile). Subgroups are not mutually exclusive. OR, odds ratio; CI, confidence interval. a From logistic regression analysis. Odds ratios for the relationship of elevated ALT activity with coffee or caffeine within each liver injury risk factor subgroup. Coffee and caffeine were coded as ordinal variables, so the odds ratio measures the change in risk between each coffee category or caffeine quintile and the next higher category or quintile. b Adjusted for age, sex, ethnicity, and cigarette smoking. To examine the consistency of our results, we performed additional analyses with serum ALT concentration as a continuous variable and with abnormal aspartate aminotransferase or ALT activity as the case definition. With ALT as a continuous variable, the results were similar to our main analysis. In unadjusted analysis, higher ALT activity was associated with lower consumption of both coffee (P value for trend.010) and caffeine (P.004). In multivariate-adjusted analysis, inverse associations remained with both coffee (P.026) and caffeine (P.002). When the outcome was defined as elevated ALT or aspartate aminotransferase activity ( 40 U/L) or both, the results were similar to our main analysis for caffeine, although not as strong for coffee. In unadjusted analysis, there was an association with lower caffeine intake (P value for trend ) and a trend toward a relationship with lower coffee intake (P.077). In multivariate-adjusted analysis, an inverse association remained with caffeine (P.004), while coffee was not significantly related (P.21). As discussed above, our main analysis was among persons at high risk for liver injury. For completeness, we also examined the relationships of coffee and caffeine consumption with elevated ALT activity in the full NHANES III sample (n 15,231), among which the prevalence of elevated ALT activity was 4.1% (and just 1.7% among the subgroup not at high risk). In unadjusted analysis, a lower prevalence of elevated ALT activity was associated with greater intake of both coffee (P value for trend.023) and caffeine (P.003). In multivariate analysis adjusting for demographics, smoking, and liver injury risk factors, the inverse association remained with consumption of caffeine (P.006) but not with coffee (P.073). To investigate whether the protective effects of coffee and caffeine could be mediated by reduced insulin resistance, we investigated the relationship of fasting serum insulin levels with coffee drinking in a subgroup of NHANES III participants (n 6611) examined in the morning after an overnight fast. After adjusting for BMI, body fat distribution, demographics, and other lifestyle factors, coffee consumption was inversely associated with insulin (P.016) and caffeine intake was unrelated (P.21). Of more importance, fasting insulin levels did not influence the association of coffee and caffeine with ALT activity. The inverse relationship of coffee consumption with elevated ALT activity was similar in models without insulin concentration (P value for trend.10) and including insulin concentration (P.12). The association of caffeine with ALT was also similar in models without insulin concentration (P value for trend.048) and including insulin concentration (P.034).

7 30 RUHL AND EVERHART GASTROENTEROLOGY Vol. 128, No. 1 Because transferrin saturation may be affected by fasting status, 30 we also conducted our analysis among the subgroup of the NHANES III population who were examined in the morning after an overnight fast of 9 to 24 hours and who were at high risk for liver injury by our criteria. A total of 2655 fasted participants were at high risk for liver injury. In this smaller sample in multivariate analyses, elevated ALT activity was associated with lower intake of caffeine (P value for trend.018), while the inverse relationship with coffee was not statistically significant (P.17). Discussion In this large, national, population-based study, among persons at high risk for liver injury, consumption of coffee and caffeine was associated with lower risk of abnormal ALT activity. Persons drinking more than 2 cups of coffee per day had approximately one half the risk of noncoffee drinkers, while participants with a caffeine intake in the highest quintile were only one third as likely to have elevated ALT activity as those in the lowest quintile. This result strengthens and extends to the US population the findings of a small number of epidemiologic studies in Italy, in Japan, and among Mexican Americans. 7,13,18,19,26 Furthermore, these findings suggest that the protective relationships of coffee and caffeine occurred regardless of whether participants were at risk for liver injury due to excessive alcohol intake, viral hepatitis, iron overload, abnormal glucose status, or obesity. The protective relationship was stronger with consumption of caffeine than coffee. With both coffee and caffeine included in a multivariate model, coffee was no longer statistically significant. Thus, caffeine may be at least partially responsible for the effects of coffee on the liver. In one animal study, caffeine levels of tea and coffee extracts inversely correlated with liver injury. 1 Caffeine intake from beverages and serum caffeine levels have been associated with lower -glutamyltransferase levels. 10 The relationship of liver enzymes with intake of green tea, a source of caffeine, was inconsistent in Japanese studies 8,13,14,34 and tea consumption was unrelated to liver cirrhosis. 21,23 However, the smaller quantity of caffeine in tea compared with coffee would make a relationship with liver outcomes more difficult to detect. The metabolic effects of caffeine responsible for protection from liver injury are uncertain. Caffeine is sequentially metabolized in the liver into potentially active compounds by cytochrome P450 isoform CYP1A2, N- acetyl-transferase type 2, and xanthine oxidase. 35 Caffeine is an A 1 and A 2 adenosine receptor antagonist, at least acutely, 36 but little is known about the effects of long-term administration on these hepatic receptors. Evidence also suggests that caffeine may have antioxidant effects that could be beneficial if oxidative stress plays a role in liver injury, as is widely believed. 37,38 Caffeine inhibited lipid peroxidation induced in vitro by reactive oxygen species in one study, 39 and antioxidant activity was found even at moderate concentrations in another study. 40 Other potential mechanisms for a beneficial effect of coffee on the liver have been suggested by experimental studies. Antioxidant activities have been attributed to aroma extracts isolated from coffee beans. 4 In animal studies, the diterpenes (nontriglyceride lipid components of coffee oils), cafestol and kahweol, were found to induce synthesis of glutathione, an important mediator against hepatocellular injury. 2,3,41 However, diterpenes are found primarily in boiled coffee 42 and so probably do not fully explain any potential beneficial effects of coffee on the liver in the United States, where filtered coffee is more commonly consumed. Higher fasting serum insulin concentrations were previously found to be associated with elevated ALT activity in NHANES III, 29 and recent studies found that coffee may protect against diabetes. 43,44 Nevertheless, we did not find that fasting insulin levels affected the associations of coffee and caffeine with elevated ALT activity. However, this analysis relied on fasting insulin concentration as an inexact proxy for insulin resistance and was limited to a half-sample of participants who were asked to fast overnight. There are several limitations to the current study. NHANES III is a cross-sectional study; therefore, we are unable to establish a temporal association between coffee and caffeine consumption and abnormal ALT activity. It is possible that the inverse association we found between coffee and caffeine intake and elevated ALT activity was the result of reduced consumption due to impaired caffeine clearance in persons with significant liver injury. 23,45,46 However, this is unlikely in the current study, given that most persons in the US population with elevated ALT activity would not be expected to have the degree of diminished liver function that would result in impaired caffeine clearance. Furthermore, the results were relatively unchanged when analyses included all NHANES III participants or were limited to persons without impaired liver function or right upper quadrant pain. As previously reported, a limitation in using NHANES to study liver injury is reliance on a single

8 January 2005 COFFEE, CAFFEINE, AND ELEVATED ALT ACTIVITY 31 serum ALT level as a marker for liver injury and the possible loss of ALT activity after the brief storage that was necessary in NHANES III. 29,47 Such inaccuracies might have resulted in misclassification of some participants but should not have led to biased results. Using ALT as a continuous variable or including an abnormal aspartate aminotransferase level in the case definition had little effect on the association of lower caffeine consumption with elevated liver enzyme levels, although there was no relationship with coffee drinking with inclusion of aspartate aminotransferase, which is a less specific marker of liver injury than ALT. An additional limitation of the present study was the lack of available information on decaffeinated coffee consumption, brewing method, and variations in cup size and strength. Finally, it is possible that caffeine intake may be a surrogate marker for another more biologically important, but as of yet unidentified, factor. In this large, national, population-based study, among persons at high risk for liver injury, consumption of coffee and especially caffeine was associated with lower risk of abnormal ALT activity. The potential beneficial effects of coffee and caffeine on the liver deserve further investigation. References 1. He P, Noda Y, Sugiyama K. Suppression of lipopolysaccharideinduced liver injury by various types of tea and coffee in D- galactosamine-sensitized rats. Biosci Biotechnol Biochem 2001; 65: Huber WW, Scharf G, Rossmanith W, Prustomersky S, Grasl- Kraupp B, Peter B, Turesky RJ, Schulte-Hermann R. The coffee components kahweol and cafestol induce gamma-glutamylcysteine synthetase, the rate limiting enzyme of chemoprotective glutathione synthesis, in several organs of the rat. Arch Toxicol 2002;75: Scharf G, Prustomersky S, Huber WW. Elevation of glutathione levels by coffee components and its potential mechanisms. Adv Exp Med Biol 2001;500: Lee KG, Mitchell A, Shibamoto T. Antioxidative activities of aroma extracts isolated from natural plants. Biofactors 2000;13: Arnesen E, Huseby NE, Brenn T, Try K. The Tromso Heart Study: distribution of, and determinants for, gamma-glutamyltransferase in a free-living population. Scand J Clin Lab Invest 1986;46: Nilssen O, Forde OH, Brenn T. The Tromso Study. Distribution and population determinants of gamma-glutamyltransferase. Am J Epidemiol 1990;132: Casiglia E, Spolaore P, Ginocchio G, Ambrosio GB. Unexpected effects of coffee consumption on liver enzymes. Eur J Epidemiol 1993;9: Kono S, Shinchi K, Imanishi K, Todoroki I, Hatsuse K. Coffee and serum gamma-glutamyltransferase: a study of self-defense officials in Japan. Am J Epidemiol 1994;139: Nilssen O, Forde OH. Seven-year longitudinal population study of change in gamma-glutamyltransferase: the Tromso Study. Am J Epidemiol 1994;139: Sharp DS, Benowitz NL. Re: Alcohol, smoking, coffee, and cirrhosis and coffee and serum gamma-glutamyltransferase: a study of self-defense officials in Japan. Am J Epidemiol 1995; 141: Pintus F, Mascia P. Distribution and population determinants of gamma-glutamyltransferase in a random sample of Sardinian inhabitants. ATS-SARDEGNA Research Group. Eur J Epidemiol 1996;12: Poikolainen K, Vartiainen E. Determinants of gamma-glutamyltransferase: positive interaction with alcohol and body mass index, negative association with coffee. Am J Epidemiol 1997; 146: Tanaka K, Tokunaga S, Kono S, Tokudome S, Akamatsu T, Moriyama T, Zakouji H. Coffee consumption and decreased serum gamma-glutamyltransferase and aminotransferase activities among male alcohol drinkers. Int J Epidemiol 1998;27: Honjo S, Kono S, Coleman MP, Shinchi K, Sakurai Y, Todoroki I, Umeda T, Wakabayashi K, Imanishi K, Nishikawa H, Ogawa S, Katsurada M, Nakagawa K, Yoshizawa N. Coffee drinking and serum gamma-glutamyltransferase: an extended study of Self- Defense Officials of Japan. Ann Epidemiol 1999;9: Nakanishi N, Nakamura K, Nakajima K, Suzuki K, Tatara K. Coffee consumption and decreased serum gamma-glutamyltransferase: a study of middle-aged Japanese men. Eur J Epidemiol 2000;16: Nakanishi N, Nakamura K, Suzuki K, Tatara K. Lifestyle and the development of increased serum gamma-glutamyltransferase in middle-aged Japanese men. Scand J Clin Lab Invest 2000;60: Nakanishi N, Nakamura K, Suzuki K, Tatara K. Lifestyle and serum gamma-glutamyltransferase: a study of middle-aged Japanese men. Occup Med (Lond) 2000;50: Nakanishi N, Nakamura K, Suzuki K, Tatara K. Effects of coffee consumption against the development of liver dysfunction: a 4-year follow-up study of middle-aged Japanese male office workers. Ind Health 2000;38: Honjo S, Kono S, Coleman MP, Shinchi K, Sakurai Y, Todoroki I, Umeda T, Wakabayashi K, Imanishi K, Nishikawa H, Ogawa S, Katsurada M, Nakagawa K, Yoshizawa N. Coffee consumption and serum aminotransferases in middle-aged Japanese men. J Clin Epidemiol 2001;54: Corrao G, Lepore AR, Torchio P, Valenti M, Galatola G, D Amicis A, Arico S, di Orio F. The effect of drinking coffee and smoking cigarettes on the risk of cirrhosis associated with alcohol consumption. A case-control study. Provincial Group for the Study of Chronic Liver Disease. Eur J Epidemiol 1994;10: Corrao G, Zambon A, Bagnardi V, D Amicis A, Klatsky A. Coffee, caffeine, and the risk of liver cirrhosis. Ann Epidemiol 2001;11: Gallus S, Tavani A, Negri E, La Vecchia C. Does coffee protect against liver cirrhosis? Ann Epidemiol 2002;12: Klatsky AL, Armstrong MA. Alcohol, smoking, coffee, and cirrhosis. Am J Epidemiol 1992;136: Tverdal A, Skurtveit S. Coffee intake and mortality from liver cirrhosis. Ann Epidemiol 2003;13: Spiller MA. The chemical components of coffee. Prog Clin Biol Res 1984;158: Meltzer AA, Everhart JE. Association between diabetes and elevated serum alanine aminotransferase activity among Mexican Americans. Am J Epidemiol 1997;146: Plan and operation of the Third National Health and Nutrition Examination Survey, (DHHS Publication No. [PHS] ). National Center for Health Statistics, Bellentani S, Saccoccio G, Costa G, Tiribelli C, Manenti F, Sodde M, Saveria Croce L, Sasso F, Pozzato G, Cristianini G, Brandi G.

9 32 RUHL AND EVERHART GASTROENTEROLOGY Vol. 128, No. 1 Drinking habits as cofactors of risk for alcohol induced liver damage. The Dionysos Study Group. Gut 1997;41: Ruhl CE, Everhart JE. Determinants of the association of overweight with elevated serum alanine aminotransferase activity in the United States. Gastroenterology 2003;124: Third National Health and Nutrition Examination Survey, , reference manuals and reports. manual for medical technicians and laboratory procedures used for NHANES III. Hyattsville, MD: National Center for Health Statistics, Hernandez-Avila M, Colditz GA, Stampfer MJ, Rosner B, Speizer FE, Willett WC. Caffeine, moderate alcohol intake, and risk of fractures of the hip and forearm in middle-aged women. Am J Clin Nutr 1991;54: National Health and Nutrition Examination Survey III; body measurements (anthropometry). Westat, Research Triangle Institute. SUDAAN user s manual release 7.0. Research Triangle Institute Park, NC: Research Triangle Institute, April Imai K, Nakachi K. Cross sectional study of effects of drinking green tea on cardiovascular and liver diseases. BMJ 1995;310: Vistisen K, Poulsen HE, Loft S. Foreign compound metabolism capacity in man measured from metabolites of dietary caffeine. Carcinogenesis 1992;13: Jacobson KA, von Lubitz DK, Daly JW, Fredholm BB. Adenosine receptor ligands: differences with acute versus chronic treatment. Trends Pharmacol Sci 1996;17: Day CP, James OF. Steatohepatitis: a tale of two hits? Gastroenterology 1998;114: Berson A, De Beco V, Letteron P, Robin MA, Moreau C, El Kahwaji J, Verthier N, Feldmann G, Fromenty B, Pessayre D. Steatohepatitisinducing drugs cause mitochondrial dysfunction and lipid peroxidation in rat hepatocytes. Gastroenterology 1998;114: Devasagayam TP, Kamat JP, Mohan H, Kesavan PC. Caffeine as an antioxidant: inhibition of lipid peroxidation induced by reactive oxygen species. Biochim Biophys Acta 1996;1282: Lee C. Antioxidant ability of caffeine and its metabolites based on the study of oxygen radical absorbing capacity and inhibition of LDL peroxidation. Clin Chim Acta 2000;295: Schilter B, Perrin I, Cavin C, Huggett AC. Placental glutathione S-transferase (GST-P) induction as a potential mechanism for the anti-carcinogenic effect of the coffee-specific components cafestol and kahweol. Carcinogenesis 1996;17: Urgert R, van der Weg G, Kosmeijer-Schuil TG, van de Bovenkamp P, Hovenier R, Katan MB. Levels of the cholesterol-elevating diterpenes cafestol and kahweol in various coffee brews. J Agric Food Chem 1995;43: Salazar-Martinez E, Willett WC, Ascherio A, Manson JE, Leitzmann MF, Stampfer MJ, Hu FB. Coffee consumption and risk for type 2 diabetes mellitus. Ann Intern Med 2004;140: Tuomilehto J, Hu G, Bidel S, Lindstrom J, Jousilahti P. Coffee consumption and risk of type 2 diabetes mellitus among middleaged Finnish men and women. JAMA 2004;291: Hasegawa M, Yamada S, Hirayama C. Fasting plasma caffeine level in cirrhotic patients: relation to plasma levels of catecholamines and renin activity. Hepatology 1989;10: Cheng WS, Murphy TL, Smith MT, Cooksley WG, Halliday JW, Powell LW. Dose-dependent pharmacokinetics of caffeine in humans: relevance as a test of quantitative liver function. Clin Pharmacol Ther 1990;47: Ruhl CE, Everhart JE. Relation of elevated serum alanine aminotransferase activity with iron and antioxidant levels in the United States. Gastroenterology 2003;124: Received April 1, Accepted September 16, Address requests for reprints to: Constance E. Ruhl, MD, PhD, Social and Scientific Systems, Inc, 8757 Georgia Avenue, 12th Floor, Silver Spring, Maryland cruhl@s-3.com; fax: (301) Supported by a contract from the National Institute of Diabetes and Digestive and Kidney Diseases (NO1-DK ).