The Effect of Coffee Consumption on the Development of Hepatocellular Carcinoma in Hepatitis B Virus Endemic Area

Eun Sun Jang; Sook-Hyang Jeong; Sang Hyub Lee; Sung Ho Hwang; So Yeon Ahn; Jaebong Lee; Young Soo Park; Jin Hyoek Hwang; Jin-Wook Kim; Nayoung Kim; Dong Ho Lee; Hyun Young Kim


Liver International. 2013;33(7):1092-1099. 

In This Article

Subjects and Methods


In this study, a total of 1364 subjects were enrolled in health-check examinee (HCE, n = 480), chronic liver disease (CLD, n = 626) and hepatocellular carcinoma (HCC, n = 258) groups. HCE group (control 1) was consisted with subjects who visited the Health Promotion Center of Seoul National University Bundang Hospital and underwent health screening voluntarily, from August 2007 to February 2008. Self-administered questionnaires about smoking, alcohol and coffee drinking habits were distributed to 909 examinees and 500 (55%) agreed to participate to this study. Among them, 20 subjects with chronic liver disease were excluded from HCE group.

Subjects in CLD (control 2) or HCC (case) groups were patients who visited the hepatology clinic in the same study institution during the same period. CLD patients were defined as those with evidences of liver disease including hepatitis B/C virus infection, significant alcohol intake (>60 g/day), autoimmune hepatitis or fatty liver for more than 6 months. The diagnosis of HCC was based on the recommendation of the American Association for the Study of Liver Diseases[11] using serum α-foetoprotein and/or imaging modalities (ultrasonography, computed tomography, magnetic resonance image, or angiography). Histological diagnosis was made in 63 (24.4%) patients with HCC.

Patients co-infected with human immunodeficiency virus infection or those with malignant disease other than HCC were excluded. This study was approved by the Institutional Review Board of Seoul National University Bundang Hospital (IRB No. B-0801/053-001) and written informed consent was obtained from all participants prior to this study.

Questionnaire Survey

We collected information on the demographics, medical history, physical findings and laboratory results of eligible subjects from medical records. At recruitment, a self-reported questionnaire survey was conducted about coffee, alcohol and cigarette consumption habits. The questionnaire used in this study was developed on the basis of available literature and qualitative research including items designed to measure alcohol, smoking and coffee consumption amounts.[10,18,19] We also prepared the questionnaire referred from the verified questionnaire used in the Korean National Health and Nutrition and Examination Survey. To control the response quality of the self-administered questionnaires, an experienced research nurse assisted the participant during answering the questionnaire. All investigators and research nurses have completed a certificated Good Clinical Practice education annually or biannually and our study institution was approved by Association for the Accreditation of Human Research Protection Programs.

Subjects were classified into 'never', 'quit' or 'current' coffee drinkers. To identify those who had quit coffee drinking after the diagnosis of cancer, HCC patients and control subjects were requested to supply the year of quitting coffee. If the subject was a past or current drinker, type and daily amount of coffee and total duration of coffee drinking were recorded. The amount of coffee consumption was asked by a closed-ended question with 4 options (No, <1, 1–2, and ≥3 cups/day). Information about the duration of coffee consumption was obtained through an open question (years). If the subject was a past drinker, he/she was required to answer the questions based on the past habits.

Alcohol drinking status was defined as never, quit or current drinker. Information on the alcohol drinking duration and the daily amount of consumed alcoholic beverages was obtained similarly to those of coffee consumption. We categorized alcohol drinkers according to their mean daily drinking amounts into none, mild (<40 g/day), moderate (40–60 g/day), and heavy (>60 g/day) drinkers.

Smoking histories were classified with the same manner as above. Former or current smokers were asked about their mean cigarette amounts (packs) smoked per day and the number of years having smoked. Participants who smoked ≥0.5 packs/day were considered significant daily smokers.

Physical and Biochemical Measurements

Height and weight of each participant were measured to calculate body mass index (BMI). Obesity was defined as a BMI ≥25 kg/m2 according to the criterion of the World Health Organization and the National Institute of Health for Obesity in Asian populations.[22] The serological status of chronic viral hepatitis was evaluated by assays for HBsAg, hepatitis B surface antibody, HBeAg, hepatitis B e-antibody and anti-hepatitis C virus antibody (anti-HCV). Serum HBV DNA level was measured using the branched DNA detection method (detection range 2000 – 108 HBV genome copies/ml, Versant HBV DNA 3.0 assay, Siemens, Munich, Germany) according to the manufacturer's protocol. All biochemical assays were performed at the laboratory of Seoul National University Bundang Hospital, which complies with the International Federation of Clinical Chemistry and Laboratory Medicine.

Statistical Analysis

Baseline characteristics and distributions of coffee, alcohol drinking and smoking habits among HCE, CLD and HCC groups were compared by chi-squared tests for categorized variables and by one-way analysis of variances (anova) or Kruskal–Wallis test for continuous variables.

To analyse the independent effect of each variable on the risk of HCC compared to healthy control, variables with a P-value of <0.1 in the univariable analyses were further studied using multivariable logistic regression with a stepwise procedure. The risk was measured by odds ratio and its 95% confidence interval (CI).

To estimate the risk of HCC in healthy control group (control 1), multivariable logistic regression model adjusted for age (per year), gender, obesity, past medical history of diabetes mellitus (DM), lifetime smoking amount (≤15 pack*years, >15 pack*years), lifetime alcohol drinking amount (≤150 × 103 g, >150 ×103 g) and lifetime coffee consumption amount (≤20 000 cups, >20 000 cups) was used. For the CLD control (control 2), the viral aetiology of chronic liver disease (none, HCV infection, HBV infection, and HBV/HCV coinfection) with the same variables for control 1 were used in the multivariable model.

To compare the effect of coffee consumption on the HCC risk in CHB patients and in other CLD patients, another multivariable logistic regression model adjusted for age (per year), gender, obesity, lifetime smoking amount (≤15 pack*years, >15 pack*years), lifetime alcohol drinking amount (≤150 × 103 g, >150 × 103 g) and lifetime coffee consumption amount (≤20 000 cups, >20 000 cups) was conducted. For patients infected with HBV, HBeAg status, serum HBV DNA level (≤105 copies/ml, >105 copies/ml) and the history of anti-HBV treatment were added in the above multivariable model.

All statistical analyses were performed using SPSS 17.0 (SPSS Institute, Inc.; Chicago, IL, USA) software. P-values less than 0.05 indicated a statistically significant association.