Meta-analysis Confirms Correlation of Coffee Consumption with Reduced Liver Cancer Risk; No Significant Effect on Biliary Tract Cancers

Coffee beverages from coffee (Coffea arabica, Rubiaceae) beans can offer many health benefits to the liver, with anti-cancer effects exerted by caffeine, and antioxidant and anti-inflammatory effects by phenolic components. Coffee diterpenes may have anti-cancer effects. The liver benefits from coffee’s contributions to lipid metabolism regulation. Systematic reviews have found coffee consumption associated with lower rates of steatosis, non-alcoholic steatohepatitis, chronic liver disease, and cirrhosis. Chronic viral hepatitis may progress more slowly in coffee users. Both caffeinated and decaffeinated (decaf) coffee affect the biliary tract (BT), inhibiting gallstone formation and increasing gallbladder contraction; in patients with gallstones, coffee may cause pain. Previous meta-analyses of the correlation between coffee drinking and liver cancer (LC) risk lack data on BT cancer (BTC)* risk, dose-response, or potential confounding factors. The authors performed a meta-analysis including these parameters, following Meta-Analysis of Observational Studies in Epidemiology (MOOSE) protocols.

A search of electronic databases from inception to March 2017 yielded 9248 English-language references. Of these, 7953 were excluded based on the title. Of 1295 screened, 1260 were excluded based on the abstract. Of 35 evaluated in full, 17 did not meet inclusion criteria, leaving 18 (five on BTC and 13 on LC) for analysis. They were all prospective or case-control studies reporting hazard ratios (HRs), risk ratios (RRs), or odds ratios (ORs), with a minimum of three levels of coffee consumption. Study quality was assessed via the Newcastle-Ottawa scale; heterogeneity by the Q test and I2 statistic; sensitivity by the leave-one-out method; and publication bias by funnel plot examination. HRs/ORs/RRs for each type of cancer were pooled to calculate summary RRs with 95% confidence intervals (CIs) for highest vs. lowest coffee intake. Median or mean intake in each category was assigned to the corresponding OR/HR/RR for each study. A two-stage random effects dose-response analysis was performed to calculate study-specific linear and non-linear trends across categories of coffee intake.

Of the five studies reporting on BTC, one pooled nine cohort studies; two studies included a total of three prospective cohorts, and two were case-control studies. They represented 1,375,626 subjects and 726 cases. Most studies reported sex-specific data and adjusted for cigarette smoking; other factors, e.g., type of coffee used and hepatitis status, were sometimes reported. Three were conducted in the United States; one in Europe; and one, with two large long-term cohorts, in Japan. Pooled RR for highest vs. lowest intake was 0.83 (95% CI, 0.64-1.08; P=0.58), with no evidence of heterogeneity (I2=0%) and no publication bias found. For cohort studies only, pooled RR=0.84 (95% CI, 0.61-1.15; P=0.27; I2=22%); for case-control studies, RR=0.74 (95% CI, 0.34-1.63; P=0.82; I2=0%). No subgroup analyses were undertaken due to the small number of BTC studies.

Of LC reports, seven used six prospective cohorts and one multicenter study; one, a pooling study of nine prospective cohorts; and five, case-control studies. They included 2,105,104 subjects and 4227 cases. Two were conducted in the United States; six in Europe; and five in Asia. Pooled RR of LC for highest vs. lowest category of coffee intake was 0.52 (95% CI, 0.42-0.63; P=0.02), with moderate heterogeneity (I2=44%) and no publication bias observed. For prospective cohort studies, pooled RR=0.53 (95% CI, 0.41-0.69; P=0.03; I2=46%); for case-control studies, RR=0.48 (95% CI, 0.33-0.70; P=0.08; I2=47%). Subgroup analyses were performed to account for effects of study design, geographic location, gender, smoking status, coffee type (caffeinated vs. decaf), and hepatitis status. From a table, it appears that 12 studies reported on men and women separately; 11, tobacco (Nicotiana tabacum, Solanaceae) use status; and four, hepatitis status—none of these significantly affected results. Another table shows that three datasets covered caffeinated coffee and four, decaf. It is not possible to tell how many studies yielded these data. A significant decrease in LC risk is noted for caffeinated coffee (RR=0.65; 95% CI, 0.49-0.86) vs. decaf (RR=0.85; 95% CI, 0.63-1.14). Less risk of LC was seen in studies conducted in Europe and Asia compared to those in the United States.

Dose-response analysis showed no significant association between BTC (three studies included) and coffee intake. A non-significant lower risk was found at low levels of intake (≤2 cups/d), but no further benefit or increased risk with higher intake. Seven LC studies yielded evidence of a linear association between coffee consumption and risk (Pfor nonlinearity=0.954). Compared to no coffee, 1 cup/d resulted in pooled RR=0.82 (95% CI, 0.70-0.98). The highest intake charted, 7 cups/d, showed pooled RR=0.27 (95% CI, 0.17-0.43). Associations were similar for men and women, but data for women had more heterogeneity (Pheterogeneity=0.692). Coffee may be a valuable functional food for liver health. In vitro and in vivo studies report evidence regarding molecular targets of coffee compounds and biological rationales for its effects, backing up epidemiological studies.

* Gallbladder and extrahepatic/intrahepatic bile duct cancers, with similar etiology, are grouped as BTC.


Godos J, Micek A, Marranzano M, Salomone F, Del Rio D, Ray S. Coffee consumption and risk of biliary tract cancers and liver cancer: a dose-response meta-analysis of prospective cohort studies. Nutrients. August 28, 2017;9(9):950. doi: 10.3390/nu9090950.