We read with great interest the article by Bravi et al.1 The authors report a 41% lower risk of developing hepatocellular carcinoma (HCC) in coffee drinkers compared to never drinkers. They conclude that the inverse relation between coffee drinking and HCC is real, but add that the inference on causality remains open to discussion. Similar conclusions have been made recently about HCC incidence by another meta-analysis2 and about HCC mortality in a Japanese study.3
Providing a causal support to this meta-analysis are several studies that report an inverse association of coffee drinking with serum aminotransferases and gamma-gutamyl transferase levels which are markers of liver injury, and chronic liver disease or liver cirrhosis.4–5 Consistent results from all these studies from populations with different ethnicity and etiologies strongly suggests a protective effect of coffee drinking that may be mediated by one or more compounds in the coffee. Nevertheless, we wish to state that at least some of this association could be due to confounding from factors that were not determined in the epidemiological studies.
Wide variations in the amount of coffee consumed by individuals within a population can be explained in part by its stimulating effects. Some studies offer a biologic basis for caffeine consumption behavior. Persons with genotypes such as adenosine A receptor (ADORA2A) 1083TT are less vulnerable to caffeine dependence.6 A genetic variability of caffeine metabolism or of the main target of caffeine action in the nervous system is associated with habitual caffeine consumption. Other studies suggest that between 35% and 77% variability in caffeine consumption, heavy use, intoxication, and tolerance can be explained by genetic factors.7It is likely that individuals who are fast metabolizers will tolerate more coffee than slow metabolizers. In a study to find a simple test for phenotyping cytochrome P450 1A2 in epidemiological studies using total overnight salivary caffeine assessment (TOSCA) in a group of 131 patients with same caffeine consumption habits, the mean TOSCA in patients with cirrhosis was significantly higher than in controls (P < 0.001).8 Rapid metabolizers (with higher metabolism of caffeine) were more frequent in the group of patients with cirrhosis of metabolic origin (70.8%; P < 0.0001), and the opposite was true for the group of patients with cirrhosis of viral origin, which was composed mostly of poor metabolizers (85.1%; P < 0.001). Furthermore, serum transforming growth factor-beta 1 concentration, a marker of ongoing fibrosis, was high in poor metabolizers.8
Another mechanism for coffee's apparent effect may be through the lowering of hepatic iron reserves, which is a well known proinflammatory hepatic carcinogen. A study of dietary factors modulating iron stores in free-living elderly populations using a sample of The Framingham Heart Study participants revealed that while supplemental iron, dietary vitamin C, and alcohol were positively associated with serum ferritin, coffee intake had a negative association.9
In summary, coffee drinking pattern may be a surrogate marker of a clinical state of enhanced detoxification or clearance of hepatic carcinogens. Studies in this direction are needed before we conclude that coffee drinking will reduce risk of HCC by 40%.