Steatohepatitis/Metabolic Liver Disease
Article first published online: 22 DEC 2011
Copyright © 2011 American Association for the Study of Liver Diseases
Volume 55, Issue 2, pages 429–436, February 2012
How to Cite
Molloy, J. W., Calcagno, C. J., Williams, C. D., Jones, F. J., Torres, D. M. and Harrison, S. A. (2012), Association of coffee and caffeine consumption with fatty liver disease, nonalcoholic steatohepatitis, and degree of hepatic fibrosis. Hepatology, 55: 429–436. doi: 10.1002/hep.24731
Potential conflict of interest: Dr. Harrison consults for Amylin
The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the views of the U.S. Department of Army, U.S. Department of Air Force, or the U.S. Department of Defense.
- Issue published online: 27 JAN 2012
- Article first published online: 22 DEC 2011
- Accepted manuscript online: 10 OCT 2011 11:04AM EST
- Manuscript Accepted: 18 SEP 2011
- Manuscript Received: 11 MAY 2011
Coffee caffeine consumption (CC) is associated with reduced hepatic fibrosis in patients with chronic liver diseases, such as hepatitis C. The association of CC with nonalcoholic fatty liver disease (NAFLD) has not been established. The aim of this study was to correlate CC with the prevalence and severity of NAFLD. Patients involved in a previously published NAFLD prevalence study, as well as additional NASH patients identified in the Brooke Army Medical Center Hepatology clinic, were queried about their caffeine intake. A validated questionnaire for CC was utilized to assess for a relationship between caffeine and four groups: ultrasound negative (controls), bland steatosis/not-NASH, NASH stage 0-1, and NASH stage 2-4. A total of 306 patients responded to the CC questionnaire. Average milligrams of total caffeine/coffee CC per day in controls, bland steatosis/not-NASH, NASH stage 0-1, and NASH stage 2-4 were 307/228, 229/160, 351/255, and 252/152, respectively. When comparing patients with bland steatosis/not-NASH to those with NASH stage 0-1, there was a significant difference in CC between the two groups (P = 0.005). Additionally, when comparing patients with NASH stage 0-1 to those with NASH stage 2-4, there was a significant difference in coffee CC (P = 0.016). Spearman's rank correlation analysis further supported a negative relationship between coffee CC and hepatic fibrosis (r = −0.215; P = 0.035). Conclusion: Coffee CC is associated with a significant reduction in risk of fibrosis among NASH patients. (Hepatology 2012)
An association between coffee consumption and a decreased risk of liver disease was established almost 20 years ago. Initially, the relationship was noted specifically with alcoholic liver disease.1 Subsequently, coffee was found to correlate with a decreased risk of liver-associated enzyme elevations, hospitalizations and mortality in alcoholic cirrhotics, and cirrhosis in general.2-7 Despite these associations, no histopathologic correlate was made until the last few years when research tied coffee consumption to decreased rates of clinical and pathologic progression of liver fibrosis in hepatitis C virus (HCV) patients.8, 9 This news is significant, because any modality that decreases the progression of fibrosis in chronic liver disease, especially if it confers few adverse effects, has the potential to improve morbidity and mortality. Mitigating fibrosis progression in chronic liver disease would potentially help prevent the associated complications of cirrhosis and hepatocellular carcinoma (HCC).10
Nonalcoholic fatty liver disease (NAFLD) prevalence has been increasing over the past 20 years in accord with increasing rates of diabetes, obesity, and metabolic syndrome.11 It is likely that NAFLD is the leading cause of chronic liver disease in the United States and has been associated with lower survival rates than the general population.12-16 No data exist on the relationship between coffee/caffeine intake and risk of NAFLD, nonalcoholic steatohepatitis (NASH), and, specifically, the level of fibrosis in NASH patients. The aim of the current study was to correlate histopathologic severity of NAFLD/NASH with reported coffee and caffeine intake to determine whether a relationship exists, specifically looking at the degree of fibrosis in NASH patients.
Patients and Methods
Patients included in the present study were identified from medical records of the Brooke Army Medical Center Hepatology clinic (Fort Sam Houston, TX). One source was a recently published prospective NAFLD prevalence study that screened asymptomatic patients without known liver disease using right upper quadrant ultrasound (US).17 In this study, 400 patients 18-70 years in age were enrolled after obtaining informed consent. Those with negative US served as controls and those with steatosis went on to have liver biopsies. Based on liver biopsy results, patients were categorized as follows: bland steatosis/not-NASH, NASH with stage 0-1 fibrosis, and NASH stage 2-4 fibrosis, utilizing the Brunt system for grading and staging of steatohepatitis.18 Additional patients were identified and enrolled by virtue of a previous histopathologic diagnosis of NASH from percutaneous liver biopsy in the hepatology clinic. A single expert hepatopathologist reviewed all liver biopsies. In accord with ethical standards, a research protocol was approved by the Institutional Review Board of Brooke Army Medical Center.
A validated caffeine questionnaire was utilized to assess for a relationship between caffeine consumption (CC) and the four groups identified in the previously published study.9 These four groups included US-negative controls, bland steatosis/not-NASH, NASH stage 0-1 fibrosis, and NASH stage 2-4 fibrosis. This questionnaire was modified to reflect the source of caffeine and the quantity consumed at the time of the biopsy. After informed consent was obtained, patients were interviewed by telephone and were asked to quantify the amount and type of caffeine-containing products they consumed. Caffeine-containing products included in the questionnaire are reported in Table 1. Frequency was stratified as never, 1-3 per month, 1 per week, 2-4 per week, 5-6 per week, 1 per day, 2-3 per day, 4-5 per day, or 6 or more per day. Patients were asked to report their typical daily consumption at the time of the biopsy. Caffeine content per unit used in our calculations was previously published (Table 1).19
|Caffeine Content (mg)|
|Caffeinated cola (1 can)||46|
|Regular coffee (8 oz cup)||137|
|Decaffeinated coffee (8 oz cup)||3|
|Back tea (8 oz cup)||47|
|Green tea (8 oz cup)||30|
|Chinese (oolong) tea (8 oz cup)||30|
|Cocoa (8 oz cup)||6|
|Caffeine-fortified drinks (1 can)||71|
|Candy chocolate bar (1 oz)||7|
|Caffeine pills (1 pill)||200|
Total caffeine and total coffee caffeine intake was calculated for each individual using the caffeine content estimate information (Table 1) and the results of the questionnaire. Individual data were grouped by biopsy result as negative US, bland steatosis/not-NASH, NASH stage 0-1 fibrosis, and NASH stage 2-4 fibrosis. Mean values are reported. The null hypothesis is that there is no relationship between estimated CC and biopsy results. The alternative hypothesis is that there will be a higher estimated CC in subjects with the diagnosis of NASH. The hypothesis was tested with a one-way analysis of variance (ANOVA), followed by independent sample t tests and Spearman's rank correlations. Additionally, a Kruskall-Wallis nonparametric ANOVA was used to evaluate differences in caffeine and coffee caffeine intake with respect to body mass index (BMI). A multiple logistic regression was performed, comparing those patients with NASH (stage 0-1 and stage 2-4) to those without NASH (i.e., controls and bland steatosis). Variables analyzed in this logistic regression included gender, BMI, race, age, total caffeine intake in mg/day, and total regular coffee caffeine intake in mg/day. Spearman's rank correlation, followed by independent sample t tests, were performed to observe relationships between all types of caffeine intake and NASH versus non-NASH groups, as well as low fibrosis (stage 0-1) versus high fibrosis (stage 2-4). All data were interpreted using SPSS version 16.0 software (SPSS, Inc., Chicago, IL).
Caffeine questionnaires were administered from March 2010 to March 2011. A total of 306 patients were contacted and completed the caffeine questionnaire. Of these, 150 were female; mean age was 53.8 years; 192 were Caucasian, 34 were African American, 67 were Hispanic, and 13 were “other race”; mean BMI was 30.6 kg/m2 (Table 2). The patients were divided into groups by the findings on liver US and biopsy. Descriptions of the four groups, with respect to number of patients, gender, BMI, race, and age, are shown in Table 2. Histologic details of all NASH patients, including stage of fibrosis, grade of inflammation, ballooning, steatosis, and NAS score, are shown in Table 3.
|Total||U/S Negative||Bland Steatosis||NASH Stage 0-1||NASH Stage 2-4|
|* P < 0.05|
|Mean Age (yrs)||53.8||53.8||56.2*||50.9*||53.5|
|* P < 0.05|
|Mean BMI (kg/m2)||30.6||27.7*||31.6*||33.6*||33.4*|
|* P < 0.05 between US neg and all other groups|
|Stage of fibrosis|
|Grade of inflammation|
Estimated daily consumption of caffeine from food and beverages ranged from none to 1,104 mg/day and averaged 288 mg/day. Estimated daily consumption of coffee caffeine ranged from none to 822 mg/day. Furthermore, 100 patients reported drinking no regular coffee. Of all caffeine consumed, 71.5% came from regular coffee. In NASH patients with no to early fibrosis (group 3) 57.5% of caffeine came from regular coffee, whereas in NASH patients with advanced fibrosis (group 4), only 35.9% of caffeine consumed came from regular coffee (P = 0.041).
Caffeine Intake and Severity of Liver Disease.
Mean caffeine intake for each group was as follows: group 1: 307.1 mg/day; group 2: 229.2 mg/day; group 3: 351.3 mg/day; and group 4: 252.7 mg/day (Table 4; Fig. 1). On ANOVA, there was a statistically significant difference in caffeine intake between diagnosis groups (P = 0.024). On the Mann-Whitney rank-sum test, group 1 was significantly different from group 2 (P < 0.032). Additionally, group 2 was significantly different from group 3 (P = 0.005). Testing demonstrated no statistically significant difference between groups 3 and 4.
|Total caffeine per day in mg*||1 = negative ultrasound||126||307.1‡||0.0||1,104.0|
|2 = bland steatosis||83||229.2§||0.0||825.0|
|3 = NASH stage 0-1||61||351.3||7.0||1,040.5|
|4 = NASH stage 2-4||36||252.7||6.0||1,089.5|
|Regular coffee caffeine per day in mg†||1 = negative ultrasound||126||228.0||0.0||822.0|
|2 = bland steatosis||83||160.5||0.0||822.0|
|3 = NASH stage 0-1||61||255.9∥||0.0||822.0|
|4 = NASH stage 2-4||36||152.9||0.0||822.0|
Coffee Intake and Severity of Liver Disease.
Mean coffee caffeine intake for each group was as follows: group 1: 227.8 mg/day; group 2: 160.3 mg/day; group 3: 255.7 mg/day; and group 4: 152.7 mg/day (Table 4; Fig. 1). A statistically significant difference in coffee caffeine intake between diagnosis groups (P = 0.011) was found. Further analysis demonstrated a statistically significant difference in coffee caffeine intake between groups 3 and 4 (P = 0.016). After Bonferroni's correction for multiple comparisons, there were no other significant differences in coffee consumption between groups. NASH patients were then grouped by stage of fibrosis and compared with respect to mean, median, and range of coffee CC and total caffeine per day in milligrams (Table 5). The distribution of data could then be demonstrated in a scatter plot of CC against the stages of fibrosis (Fig. 2).
|Stage of Fibrosis||Coffee Caffeine Per Day (mg)||Total Caffeine Per Day (mg)|
|1||Mean||255.89 (coffee cup equivalents per day = 1.87)||348.21|
|2||Mean||170.30 (coffee cup equivalents per day = 1.24)||258.69|
|3||Mean||122.00 (coffee cup equivalents per day = 0.89)||242.21|
Differences in Demographics Between Groups.
On a one-way ANOVA, there was a statistically significant difference between groups in gender, age, and BMI (P < 0.001). On the post-hoc Student-Newman-Keuls test, there was a statistically significant difference in gender between groups 1 and 3 (P < 0.05). A significant difference in age between groups 2 and 3 (P < 0.05), and a significant difference in BMI between group 1 and all other groups (P < 0.05), was also found. No statistically significant difference in race by group (P > 0.05) was identified. However, there was a statistically significant difference by race in both coffee CC (P = 0.010) and total CC (P = 0.001). When the data were split to evaluate Caucasians versus non-Caucasians, we were able to perform a nonparametric Mann-Whitney rank-sum test. This evaluation found a statistically significant difference with respect to both regular coffee CC (P = 0.005) and total CC (P ≤ 0.001). On subgroup analysis of patients stratified by normal weight (BMI, <25 mg/kg2), overweight (BMI, 25-30 mg/kg2), or obese (BMI, >30 mg/kg2), ANOVA testing did not find a significant difference in coffee or caffeine intake based on BMI.
Logistic Regression Analysis of Probability of NASH.
A multiple logistic regression was performed, looking at variables of gender, BMI, race, age, total caffeine, and total coffee intake with respect to risk of NASH versus not-NASH. Cohorts 1 and 2 were combined into a single group labeled not-NASH, and cohorts 3 and 4 were combined into a group of patients labeled NASH. Only the variables, age (P < 0.001), gender (P = 0.022), and BMI (P < 0.001), remained statistically significant in predicting a patient's status as NASH or not-NASH. The strongest probability involved an increased risk of NASH with increasing BMI. Older age had a negative relationship with respect to risk of NASH.
Determination of the Correlation Between CC and coffee CC With Respect to Alanine Aminotransferase, Hemoglobin A1C, Homeostasis Model Assessment of Insulin Resistance, and Histologic Data.
Correlation analysis was performed to compare caffeine and coffee CC with chemical and histologic parameters (Table 6). The data demonstrate a weakly positive relationship between total daily caffeine and alanine aminotransferase (ALT) (r = 0.173; P = 0.021). Again noted is the significant negative correlation between coffee CC and stage of fibrosis (r = −0.215; P = 0.035). No other significant correlations were found on this comparison. Analysis was then performed to compare patients with insulin resistance, defined as a homeostasis model assessment of insulin resistance (HOMA-IR) >3 or diabetes defined as a hemoglobin (Hgb) A1C test >6.5 versus caffeine, chemical, and histologic data in patients with biopsy data. The descriptive numbers of patients with insulin resistance and diabetes are shown in Table 7. There was a weak positive relationship between diabetes versus Nonalcoholic Fatty Liver Disease Activity Score (NAS), grade of inflammation, and steatosis. There was no relationship between diabetes/insulin resistance versus fibrosis, caffeine intake, or coffee intake (Table 8).
|Spearman's rho||ALT||A1C||HOMA-IR||Stage Fibrosis|
|Regular coffee caffeine per day||Correlation coefficient||0.118||−0.097||0.026||−0.215|
|Total caffeine per day||Correlation coefficient||0.173||−0.059||0.009||−0.161|
|Spearman's rho||Grade Inflammation||NAS||Ballooning||Steatosis|
|Regular coffee caffeine per day||Correlation coefficient||−0.134||−0.004||−0.024||−0.054|
|Total caffeine per day||Correlation coefficient||−0.147||0.002||−0.027||−0.081|
|Number of patients with HOMA-IR >2.5||110|
|Number of patients with HOMA-IR >3||94|
|Number of patients with Hgb A1C >6.5||29|
|Number of patients with Hgb A1C >6.5, but HOMA-IR ≤2.5||4|
|Regular Coffee Caffeine Per Day||Total Caffeine Per Day||ALT||Stage Fibrosis|
|HOMA-IR >2.5||Correlation coefficient||0.065||0.050||0.331||0.069|
|HOMA-IR >3||Correlation coefficient||0.011||−0.003||0.331||0.054|
|A1C >6.5||Correlation coefficient||−0.109||−0.112||−0.016||0.151|
|Spearman's rho||Grade Inflammation||NAS||Ballooning||Steatosis|
|HOMA-IR >2.5||Correlation coefficient||0.128||0.119||0.033||0.023|
|HOMA-IR >3||Correlation coefficient||0.113||0.092||0.151||0.065|
|A1C >6.5||Correlation coefficient||0.213||0.245||0.086||0.226|
Determination of the Correlation Between Gender, Age, and BMI With CC, Coffee CC, NASH, and Level of Fibrosis in NASH Patients.
A strong correlation between coffee intake and total CC was found (r = 0.866; P = 0.00). There was a weak correlation between male gender and total daily caffeine intake (r = 0.149; P = 0.009). Also notable was an inverse relationship between BMI and daily coffee intake (r = −0.144; P = 0.047). When looking at NASH versus not-NASH patients, a significant correlation existed with respect to BMI (r = 0.343; P = 0.00), gender (r = 0.194; P = 0.001), and inversely to age (r = −0.164; P = 0.004). However, none of these three risk factors correlated with level of fibrosis when NASH patients were divided back into groups 3 and 4. These three demographic risk factors seem to play a predictive role with respect to NASH versus not-NASH, but not with respect to fibrosis severity in NASH patients.
Evaluation of Caffeine From Sources Other Than Coffee.
Spearman's correlations were performed with follow-on t tests, looking at the other types of caffeine evaluated on the questionnaire. No caffeine type was correlated with risk of NASH versus not-NASH. When evaluating for NASH stage 0-1 versus NASH stage 2-4, three significant correlations were found. Again, coffee was found to have an inverse correlation with NASH stage 2-4 (r = −0.245; P = 0.015). Chocolate candy and caffeine pills appeared to have a positive correlation to NASH stage 2-4 (r = 0.310, P = 0.002; r = 0.282, P = 0.005; respectively). However, on t tests, only coffee remained significantly correlated to low risk of fibrosis (P = 0.044).
The aim of this study was to determine whether a correlation exists between NAFLD, specifically the level of fibrosis in NASH patients, and caffeine/coffee intake. Despite the known association of caffeine/coffee and liver disease, a pathologic correlation to liver fibrosis in fatty liver disease had not previously been established. In this cross-sectional study of patients with NAFLD, an inverse relationship between regular coffee CC and hepatic fibrosis was observed, with a statistically significant difference in caffeine and coffee intake observed between US-negative controls, bland steatosis/not-NASH patients, NASH stage 0-1 fibrosis patients, and NASH stage 2-4 fibrosis patients. Importantly, on multiple statistical analyses, the trend of increased coffee caffeine intake and decreased fibrosis remained statistically significant (Figs. 1 and 2; Tables 4 and 6).
It is not clear, from our data, what amount of coffee confers the greatest decreased risk of fibrosis. Further prospective study is required to determine the optimal quantity of caffeine or coffee intake. It is interesting that patients with bland steatosis/not-NASH, as well as the control group, drank less coffee than those patients with NASH stage 0-1 fibrosis. It may be that coffee is only beneficial to those NAFLD patients with a propensity for fibrosis (i.e., NASH patients). It has been previously proposed that any potential beneficial effects of caffeine or coffee on liver disease might be manifested in persons at risk for liver injury from other factors, such as NASH.7 Accordingly, most research on this relationship has been on higher risk populations, such as those with HCV or alcoholic liver disease.
The different rates of NASH versus non-NASH as a function of total caffeine intake versus coffee intake warrant mention. In further review of our data, total caffeine was not significantly correlated with risk of NASH versus not-NASH. However, coffee continued to show a negative correlation with NASH stage 2-4 versus NASH stage 0-1 on this same analysis. This idea is supported by the finding that there was a stepwise decrease in coffee consumption as fibrosis increased (Fig. 2; Table 5). This would suggest that other properties of coffee beyond caffeine may affect disease progression in NASH patients.
A recent abstract evaluated whether antioxidant activity resulting from caffeine/coffee affected degree of steatosis in NAFLD patients.20 The investigators concluded that increased intake of coffee correlated with decreased risk of steatosis, but that levels of antioxidant activity were not significantly affected, suggesting that coffee's effects may be more than strictly related to caffeine's antioxidant behaviors. Additionally, a rat model of fatty liver disease showed that animals with a high-fat diet that were given decaffeinated coffee demonstrated lower levels of hepatic fat and collagen, reduced liver oxidative stress through glutathione metabolism, and liver inflammation.21 This again suggests the importance of coffee, not caffeine, in preventing the progression of NASH.
When looking at demographic factor effects on the data, there was, as expected, a direct correlation between increasing BMI and NAFLD/NASH. However, when all patients were pooled by BMI, analysis did not find a statistically significant difference in caffeine or coffee intake based on BMI. Additionally, gender, age, and BMI all seemed to correlate with NASH versus not-NASH, but did not have a correlation with higher or lower levels of fibrosis within the NASH group. There was a negative correlation between age and risk of NASH consistent with the concept that the development of NASH versus NAFLD may be related to a genetic tendency, although this is speculation.
Diabetes is a risk factor for fibrosis in fatty liver disease. In this study, we compared histopathologic findings with coffee CC data from our NASH patients. On this evaluation, the strongest correlation found was, again, the inverse relationship between coffee CC and hepatic fibrosis (Table 6). As expected, there was a positive, albeit weak, relationship between diabetes, when compared to NAS, grade of inflammation, and steatosis. There was no statistically significant relationship noted between diabetes/insulin resistance and degree of fibrosis or caffeine/coffee intake in this study. However, there was a trend toward a positive relationship between diabetes and fibrosis, as would be expected.
This study was not prospective and thus did not follow the effects of differences in fibrosis on clinical outcomes over time. It will be critical to determine whether preventing fibrosis could lower rates of clinical decompensation and HCC in patients on caffeine/regular coffee. However, this initial histopathologic correlate with coffee in NASH patients suggests that perhaps coffee may have some antifibrogenic mechanism that could prevent future disease progression. This suggestion is further corroborated by a recent abstract that showed that habitual coffee consumption appears to be protective against fibrosis progression in NAFLD.22 This study was consistent with the results shown here, that coffee specifically was important, rather than caffeine intake from other sources.
Historically, there has been little nutritional advice to give to patients with chronic liver disease. In fatty liver disease, some have advocated for the use of vitamin E and omega-3 fatty acids, but additional recommendations would be welcomed.23-25 Coffee has shown benefits in liver disease for unclear reasons, with some proposing that aromatic extracts isolated from coffee beans may be beneficial.26 Additionally, coffee oils, called dipterenes, have been evaluated. The dipterenes, kahweol and cafestol, have been shown to have an effect on liver enzymes. Dipterenes also have beneficial effects on glutathione metabolism, but have been shown to increase cholesterol levels, a finding that may not be beneficial in fatty liver disease. Regardless, the idea of dipterene effects on liver health may be moot by virtue of the fact that most of these fats are filtered out in traditional American coffee, and, therefore, our patient population was not exposed to them.27-29 Perhaps, it is the antioxidant effects of caffeine that have the greatest effect, because antioxidant activity has been found at moderate concentrations of caffeine, in some studies.30, 31 In our study, there did seem to be a significant difference in total caffeine intake, when comparing NAFLD and NASH stage 0-1; however, in multiple logistic regression analysis, this significance did not bear out. Again, this suggests that the most significant results of this study were because of coffee. Finally, recent research has shown that coffee induces UDP glucuronosyltransferases, which may cause the protective, antioxidant effects in liver disease independent of caffeine, cafestol, or kahweol.32 A better understanding of the specific components of coffee, and their effects on metabolic activities in the liver, is required.
This study was limited by a lack of blinding of the interviewers to the cohorts and the retrospective nature of this cross-sectional style of research. The lack of blinding was mitigated by having different questioners assigned to the different cohorts of patients. Regardless, the data are corroborated by other studies showing a similar effect of coffee intake with respect to decreased fibrosis risk in other disease states and, recently, in NAFLD.2, 6, 9, 22
In conclusion, this study is the first demonstration of a histopathologic correlation between fatty liver disease and estimated coffee intake. Results demonstrate that in the cohort of patients with NASH, increased intake of coffee confers a significantly decreased risk of advanced fibrosis. Moderate coffee consumption may be a benign adjunct to the comprehensive management of patients with NASH. Knowing the data of beneficial effects of coffee intake on other liver diseases, future prospective research should be designed to follow patients for clinical outcomes with respect to administered coffee intake.
The authors thank John A. Ward, Ph.D., for his statistical analysis and proofreading/editing of this article.
- 18Nonalcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999; 94: 2467-2474., , , , .Direct Link:
- 20High coffee intake is associated to lower grade of hepatic steatosis. The role of peripheral antioxidant activity. Abstract 1149, International Liver Congress, Berlin, Germany, March 30-April 3, 2011., , , , , , et al.
- 22Coffee consumption increases hepatic expression of cytochrome P450s and significantly reduces liver fibrosis in patients with NAFLD. Abstract 776, International Liver Congress, Berlin, Germany, March 30-April 3, 2011., , , , , , et al.
- 23Efficacy of Vitamin E and Vitamin C in the treatment of nonalcoholic steatohepatitis. Pract Gastroenterol 2010; 5: 20-26., .