Coffee and tea consumption and risk of hepatocellular carcinoma in Italy



The role of coffee in the aetiology of hepatocellular carcinoma has raised great interest. In Italy, coffee consumption is high, thus allowing the investigation of the topic over a broad range of consumption. A hospital-based case-control study was conducted in Italy in 1999–2002, including 185 incidents, histologically confirmed cases of hepatocellular carcinoma aged 43–84 years. Controls were 412 subjects admitted to the same hospitals' networks for acute, non-neoplastic diseases unrelated to diet. Coffee and tea consumption were assessed using a validated food-frequency questionnaire. Odds ratios (ORs) and corresponding the 95% confidence intervals (CI) were computed using unconditional multiple logistic regression, adjusting for hepatitis viruses seropositivity, alcohol intake, smoking habits and other potential confounding factors. Compared to people who drunk <14 cups/week of coffee, the risk of hepatocellular carcinoma decreased for increasing levels of consumption (OR = 0.4, 95% CI: 0.2–1.1 for ≥28 cups/week, p for trend = 0.02). In the present study, inverse relations were observed across strata of hepatitis C and, B virus infections and alcohol drinking. No significant association emerged with consumption of decaffeinated coffee (OR = 0.7, 95% CI = 0.2–2.5) or tea (OR = 1.4, 95% CI = 0.8–2.7). The present study supports the hypothesis of a favourable effect of coffee, though not decaffeinated coffee and tea, on the risk on hepatocellular carcinoma. © 2006 Wiley-Liss, Inc.

Hepatocellular carcinoma (HCC) is a highly fatal cancer that, in Italy, accounts for 3.4% of all newly diagnosed cancers among men and for 1.9% among women.1 Likewise, the burden of HCC on total cancer mortality was 4.4% among men and 2.7% among women.2 Incidence and mortality have shown upward trends in most of the developed countries,3, 4, 5 including Italy.3 The major known risk factors are persistent infections with hepatitis C virus (HCV) and hepatitis B virus (HBV),6 but other prevalent medical conditions (i.e., liver cirrhosis) or socio-demographic factors, such as alcohol drinking and tobacco smoking, may play an important role in HCC onset.6, 7

Several studies from Europe, Japan, and United States reported an inverse relation between coffee consumption, γ-glutamyltransferase (GGT)8, 9, 10 and alanine aminotransferase (ALT)10, 11 levels, which are markers of liver injuries. Recently, increasing evidence has emerged on the relation between coffee drinking and risk of cirrhosis12, 13, 14 and HCC.15, 16, 17, 18 A consistent inverse association with coffee consumption against HCC was found in cohort15, 17, 18, 19 and case-control studies.16, 20, 21

A few studies have investigated tea consumption in relation to HCC incidence and mortality,14, 15, 19, 22 and their results were inconclusive.

Italy ranked among the top European countries according to annual pro-capite consumption of coffee (about 6 kg) ( The present paper, based on a case-control study conducted in northern and southern Italy, offered the chance of exploring the relation between coffee and HCC risk in a population whose coffee consumption is highly variable.

Material and methods

Between January 1999 and July 2002, we conducted a case-control study on HCC and lymphomas in the province of Pordenone, in the Northeast of Italy, and the town of Naples, in the South.7 Briefly, cases were patients below 85 years of age with incident HCC, who had not yet received any cancer treatment at study entry. They were admitted at the National Cancer Institute in Aviano, the ‘Santa Maria degli Angeli’ General Hospital in Pordenone, and the ‘Pascale’ National Cancer Institute, plus 4 General Hospitals in Naples. Out of 261 identified HCC cases, 3 subjects refused to participate in the study, 29 HCC cases did not supply a blood sample, and additional 44 were not questioned on dietary habits, thus leaving 185 cases (median age: 66 years; range: 43–84 years) for whom both comprehensive questionnaire information and blood sample were available. Histological or cytological confirmation was available from 78.2% of HCC cases, whereas the remaining HCC cases were diagnosed on ultrasound and/or tomography (12.3%), and elevated α-fetoprotein (9.4%).

The comparison group included patients aged 40–82 years (median age: 65 years) admitted for a wide spectrum of acute conditions to the same hospitals where HCC cases had been interviewed. Specifically excluded from the control group were patients whose hospital admission was due to diseases related to alcohol and tobacco use (e.g., respiratory diseases, peptic ulcer, lung cancer, head and neck cancer, etc.), hepatitis viruses (e.g., hepatitis, cirrhosis, oesophageal varices, etc.), or chronic diseases that might have resulted in substantial lifestyle modifications (e.g., diabetes, cardio- and cerebro-vascular diseases, etc.). However, co-morbidity for such diseases was not an exclusion criterion. Overall, 467 controls were contacted and 462 accepted to participate. Blood samples were available for 431 controls and 412 provided comprehensive information on dietary habits; of these, 27% were admitted to the hospital for trauma; 25% for acute surgical conditions, 24% for non-traumatic orthopaedic diseases, 13% for eye diseases, and 11% for other illnesses. Controls were more often females and were younger than HCC cases as matching was conducted according to the distribution by age and gender of cancer cases in the entire study, which also included lymphomas.23, 24

All study participants signed an informed consent form, in agreement with the requirements of the Ethical Committee of the Aviano National Cancer Institute, approving the study.

Each case and each control provided a 15-ml blood sample the day the interview took place. Sera were screened for antibodies against HCV antiHCV using a third-generation MEIA (AxSYM HCV, version 3.0; Abbott, Wiesbaden, Germany) and for HBV surface antigen (HBsAg) was performed using microparticle enzyme immunoassay (AxSYM HBsAg version 2.0, Abbott Diagnostic Division, Wiesbaden, Germany).7

Trained interviewers administered a structured food frequency-questionnaire (FFQ) to cases and controls during their hospital stay in order to assess the usual diet during the 2 years before diagnosis or hospital admission for the controls. In particular, 3 questions investigated the consumption of coffee and cappuccino, decaffeinated coffee, and tea. The reproducibility and validity of the FFQ were satisfactory,25, 26 in particular for coffee and tea consumption.27 The questionnaire also included information on socio-demographic characteristics, alcohol and tobacco use, and a section on medical history of selected diseases. Drinkers were asked to report any change in alcohol beverage intake, as to compute the maximal lifetime alcohol intake.7 Abstainers were individuals who had abstained from any alcoholic beverage throughout their life.

Adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CI) were calculated by means of unconditional multiple logistic regression models including age (quinquennia), gender, study centre, years of education, HBsAg and/or antiHCV positivity, smoking habits, lifetime maximal alcohol intake, when mentioned.28 Since in our study abstainers from coffee drinking were rare (less the 10% among controls) and could have included people who quitted, subjects drinking <14 cups/week of coffee were considered as the reference category. When analysis was conducted in separate strata, in order to improve the robustness of estimates, the highest categories of coffee consumption were combined (≥21 cups/week).


Table I shows the distribution of HCC cases and controls according to gender, age, study centre, and other selected known risk factors for HCC. Controls were more educated than cases (OR = 0.26, 95% CI: 0.11–0.58). Increased risk of HCC was found for serological evidence of hepatitis C antiHCV and/or hepatitis B (HBsAg) active infection (OR = 30.36, 95% CI: 17.98–51.25) and for high maximal lifetime alcohol intake (OR = 5.17, 95% CI: 1.78–15.06 for ≥35 drinks/week vs. abstainers), but not for tobacco smoking (OR = 1.34, 95% CI: 0.46–3.86 for ≥20 cigarettes/day vs. never smokers).

Table I. Distribution of 185 Cases of Hepatocellular Carcinoma and 412 Controls, Corresponding Odds Ratio and 95% Confidence Intervals for Selected Sociodemographical Variables, Italy, 1999–2002
 Cases n (%)Controls n (%)OR (95% CI)1
  • OR, odds ratios; CI, confidence intervals.

  • 1

    Estimated from unconditional logistic regression adjusted for gender, age, centre, education and hepatitis viruses infection, when appropriate.

  • 2

    Reference category.

 Male149 (80.5)281 (68.2) 
 Female36 (19.5)131 (31.8) 
Age group (years)
 40–5418 (9.7)85 (20.6) 
 55–6456 (30.3)116 (28.2) 
 65–7484 (45.4)147 (35.7) 
 ≥7527 (14.6)64 (15.5) 
Study centre
 Aviano61 (33.0)230 (55.8) 
 Naples124 (67.0)182 (45.2) 
Education (years)
 <7126 (68.1)232 (56.3)12
 7–1145 (24.3)93 (22.6)0.98 (0.54–1.80)
 ≥1214 (7.6)87 (21.2)0.26 (0.11–0.58)
χmath image for trend  8.16; p < 0.01
 AntiHCV− and HBsAg−38 (20.5)365 (88.6)12
 AntiHCV+ and/or HBsAg+147 (79.5)47 (11.4)30.36 (17.98–51.25)
Smoking habits
 Never50 (27.0)135 (32.8)12
 Former67 (36.2)168 (40.8)0.69 (0.34–1.38)
 Current (cigarettes/day)
  <1013 (7.0)31 (7.5)0.63 (0.22–1.86)
  10–1939 (21.1)52 (12.6)1.48 (0.65–3.37)
  ≥2016 (36.7)26 (6.3)1.34 (0.46–3.86)
χmath image for trend  1.66; p = 0.20
Maximal lifetime alcohol intake (drinks/week)
 Abstainers16 (8.7)63 (15.3)12
 <1432 (17.3)116 (28.2)0.90 (0.34–2.36)
 14–3479 (42.7)152 (36.9)1.35 (0.52–3.51)
 ≥3558 (31.4)81 (19.7)5.17 (1.78–15.06)
χmath image for trend  12.79; p < 0.01

Compared to subjects who drunk <14 cups/week of coffee, abstainers were at increased risk of HCC (OR = 2.28, 95% CI: 0.99–5.24) (Table II). A significant trend in risk reduction was seen for increasing levels of coffee consumption (p-value = 0.02), but the ORs were not statistically significant (OR = 0.43, 95% CI: 0.16–1.13 for ≥28 cups/week). Among coffee drinkers, a borderline inverse association was found for an additional daily cup of intake (OR = 0.81, 95% CI: 0.65–1.00, data not shown). Conversely, no association emerged between HCC risk and decaffeinated coffee (OR = 0.72, 95% CI: 0.21–2.50 for drinkers vs. abstainer) or tea (OR = 1.43, 95% CI: 0.76–2.66 for ≥1cup/week vs. abstainers).

Table II. Odds Ratios and Corresponding 95% Confidence Intervals of 185 Cases of Hepatocellular Carcinoma and 412 Controls by Consumption of Coffee, Decaffeinated Coffee and Tea, Italy, 1999–2002
 Cases n (%)Controls n (%)OR (95% CI)1
  •  OR, odds ratios; CI, confidence intervals.

  • 1

    Estimated from unconditional logistic regression adjusted for gender, age, centre, education, smoking habits, maximal lifetime alcohol intake and serological evidence of HCV and/or HBV infection.

  • 2

    Reference category.

Coffee (cups/week)
 Abstainers27 (14.6)41 (10.0)2.28 (0.99–5.24)
 <1467 (36.2)116 (28.2)12
 14–2050 (27.0)104 (25.2)0.54 (0.27–1.07)
 21–2727 (14.6)88 (21.4)0.57 (0.25–1.32)
 ≥2814 (7.6)63 (15.3)0.43 (0.16–1.13)
 χmath image for trend  5.68; p = 0.02
Decaffeinated coffee
 Never174 (94.0)384 (93.2)12
 Ever11 (6.0)28 (6.8)0.72 (0.21–2.50)
Tea (cup/week)
 Abstainers62 (33.5)230 (55.8)12
 <177 (41.6)81 (19.7)1.95 (1.03–3.68)
 ≥146 (24.9)101 (24.5)1.43 (0.76–2.66)
 χmath image for trend  2.00; p = 0.16

The inverse relation between coffee consumption and HCC risk was confirmed among people negative to both antiHCV and HBsAg tests (OR= 0.38, 95% CI = 0.13–1.09, p-value for trend < 0.01), whereas no association was found among subjects with serological evidence of chronic HBC and/or HCV infection (Table III). The ORs were similar across strata of alcohol intake, but a stronger inverse association between coffee and HCC emerged only among those who drunk more than ≥14 drinks/week (OR = 0.43, 95% CI = 0.17–1.09, p-value for trend <0.01). Additional adjustment for tea consumption did not substantially modify the results.

Table III. Odds Ratios and Corresponding 95% Confidence Intervals of 185 Cases of Hepatocellular Carcinoma and 412 Controls by Consumption of Coffee in Separate Strata of Hepatitis Virus Infections and Drinking Habits, Italy, 1999–2002
 Coffee (cups/week)χmath image for trend p-value
  • OR, odds ratios; CI, confidence intervals; Ca, cases; Co, controls.

  • 1

    Estimated from unconditional logistic regression adjusted for gender, age, centre, education, smoking habits, maximal lifetime alcohol intake and serological evidence of HCV and/or HBV infection, when appropriate.

  • 2

    Reference category.

Hepatitis virus
 AntiHCV− and HBsAg−
  OR (95% CI)12.09 (0.72–6.07)120.63 (0.22–1.82)0.38 (0.13–1.09)8.06; p < 0.01
 AntiHCV+ and/or HBsAg+
  OR (95% CI)12.64 (0.59–11.93)120.58 (0.21–1.52)0.84 (0.23–3.01)2.08; p = 0.15
Maximal lifetime alcohol intake (drinks/week)
  OR (95% CI)11.33 (0.30–5.81)120.27 (0.06–1.09)0.64 (0.15–2.63)2.42; p = 0.12
  OR (95% CI)12.70 (0.94–7.76)120.64 (0.28–1.47)0.43 (0.17–1.09)8.90; p < 0.01


The present study contributes to the evidence of an inverse relation between coffee consumption e HCC risk. An inverse association between coffee drinking and HCC risk was firstly suggested by La Vecchia et al.20 in an early study. Two studies from Greece29 and China30 reported no significant relation between coffee drinking and HCC risk, whereas studies from Italy16, 21 reported reduced risk among coffee drinkers. In particular, Gelatti and colleagues16 suggested that coffee consumption reduces the risk of HCC, especially among high-risk subjects (heavy drinkers and people infected with HBV and/or HCV). In our study, the association was stronger among subjects who were not chronically infected with HBV and/or HCV, but no effect modification of alcohol intake emerged. These inconsistencies could be partially due to differences in the study populations. The prevalence of HBV and/or HCV (11.4%) was much higher in our control group, as it averaged the HCV prevalence registered in Naples (13% in general practitioner patients aged 50–65 years, peaking to 20% in people aged over 65 years)31 and in Pordenone (3.2%).32

A Japanese cohort study15 reported about 70% reduction of HCC risk for subjects who drunk ≥5 cups/day. This result is supported by a cohort study on mortality due to HCC18 or liver cirrhosis.13 However, the great variation of coffee consumption across studies makes the comparison difficult.

The favourable effect of coffee drinking on HCC parallels the inverse association with liver cirrhosis12, 33 and chronic liver disease.14 A study from Italy12 reported an inverse association between cirrhosis risk and consumption of coffee, but not for tea and other caffeine-containing beverages. Likewise, a Norwegian study13 reported a 40% reduction of cirrhosis risk for an increase of 2 cups of coffee; the protective effect persisting when alcohol induced cirrhosis was considered. The role of caffeine in cirrhosis onset was investigated by an Italian study,33 which reported a protective effect of caffeine intake from coffee, but not from other beverages. This finding argues against a direct role of caffeine in risk reduction, suggesting that the beneficial effect of coffee could be due to other components of the beverage.

Coffee is rich in antioxidants and other components, which may modify enzyme synthesis. A number of studies suggested that coffee might inhibit the induction of GGT in the liver and protect against liver cell damage due to alcohol.34 Several studies linked coffee consumption to decreased levels of GGT and ALT,8, 9, 10, 11 in particular in high risk subjects.9, 10, 34 However, caffeine metabolism is impaired in patients with liver cirrhosis, who could have been advised to reduce coffee intake in order to avoid caffeine side effects. Thus, confounding due to reduction of coffee use in unhealthy subjects could not be totally ruled out.

Potentially anticarcinogenic effects of the coffee diterpenes cafestol and kahweol have been suggested by studies in animal models and cell culture systems.35, 36 These coffee components may act as blocking agents by modulation of multiple enzymes involved in carcinogen detoxification. They were found to modify the xenobiotic metabolism by means of induction of glutathione-S-transferase (GST) and inhibition of N-acetyltransferase (NAT).37 In particular, a protective effect of cafestol and kahweol against aflatoxin B1-induced genotoxicity was reported in both humans and rats.38

In the present study, abstainers from coffee drinking resulted at A doubled risk for HCC with respect to moderate drinkers (<14 cups/week). However, this result should be considered with caution, since it could be partially due to the inclusion among abstainers of subject who quitted coffee drinking because of early symptoms of cancer or of other liver diseases. We could not evaluate this issue, as the questionnaire did not investigate the history of coffee consumption. Finally, abstaining from coffee drinking could be related to other cofactors or recognized risk factors for HCC.

Recall and selection biases are possible, as in most case-control studies. Awareness of this particular dietary hypothesis in HCC aetiology, however, was limited in the Italian population at the time of the study. Moreover, physicians' recommendation about restriction of coffee drinking in patients with liver diseases could have influenced the results. However, this finding is not supported by previous studies, which showed consistency of results across strata of cirrhosis and hepatitis viruses.15, 16 Likewise, in the present study, inverse relation were observed across strata of hepatitis viruses' infections and alcohol drinking.

An additional potential limitation is the use of hospital controls, which may differ from the general population in relation to dietary habits. However, by study design, great attention was paid in excluding all diagnoses that might have been associated to or had determined special dietary habits in control subjects. The questionnaire was administered to cases and controls by the same interviewers under similar conditions in a hospital setting, thus minimizing information bias. In addition, our findings are strengthened by the nearly complete participation of identified cases and controls. Adjustments for sex, age, centre, education, smoking and drinking habits, and persistent infection with hepatitis viruses were made to address potential confoundings.

Finally, residual confounding due to assessment of alcohol consumption might have occurred in the analyses when stratifying by drinking habits. Hesitation to admit alcohol drinking was seen among self-defined ‘Abstainers’ and overestimation of alcohol intake was reported among heavy drinkers.39 However, the FFQ was valid and reproducible for estimation of alcohol intake in regular drinkers,39 and the classification of drinkers in moderate (<14 drinks/week) and heavy ones (≥14 drinks/week) might have attenuated the bias.

Coffee is widely consumed all over Italy, where this beverage is commonly drunk after meals; indeed, it is one of the most chosen drinks in social gathering. This is to be accounted among the strengths of the present study, as it allowed us to investigate the effect of coffee over a broad range of consumption. An additional strength is the use of a detailed and validated FFQ,25, 26 whose reproducibility was 0.74–0.78 for coffee, and validity was 0.70.27

In conclusion, the findings of the present study supported a favourable role of coffee, but not of decaffeinated coffee and tea, in HCC aetiology.


The authors thank Mrs O. Volpato for study coordination, Ms G. Bessega, L. Zaina., M. Grimaldi, and S. Desicato for their help in data collection, and Ms L Mei for editorial assistance. We are also grateful to Drs. R. Mele, A. Grandi, L. Forner., P. Ascierto, R. Magri, and R. Di Lauro for providing hospital control patients, and to laboratory technicians of the Divisione di Microbiologia e Immunologia, General Hospital of Pordenone, for virological testing.