Research Article: Associations of coffee genetic risk scores with consumption of coffee, tea and other beverages in the UK Biobank

Date Published: September 29, 2017

Publisher: John Wiley and Sons Inc.

Author(s): Amy E. Taylor, George Davey Smith, Marcus R. Munafò.


To evaluate the utility of coffee‐related genetic variants as proxies for coffee consumption in Mendelian randomization studies, by examining their association with non‐alcoholic beverage consumption (including subtypes of coffee and tea) and a range of socio‐demographic and life‐style factors.

Observational study of the association of genetic risk scores for coffee consumption with different types of non‐alcoholic beverage consumption.

UK general population.

Individuals of European ancestry aged 40–73 years from the UK Biobank between 2006 and 2010 (n = 114 316).

Genetic risk scores were constructed using two, four and eight independent single nucleotide polymorphisms (SNPs) identified in genome‐wide association studies (GWAS) of coffee consumption. Drinks were self‐reported in a baseline questionnaire (all participants) and in detailed 24 dietary recall questionnaires in a subset (n = 48 692).

Genetic risk scores explained up to 0.38, 0.19 and 0.76% of the variance in coffee, tea and combined coffee and tea consumption, respectively. Genetic risk scores demonstrated positive associations with both caffeinated and decaffeinated coffee and tea consumption, and with most subtypes of coffee consumption, but only with standard tea consumption. There was no clear evidence for positive associations with most other non‐alcoholic beverages, but higher genetic risk for coffee consumption was associated with lower daily water consumption. The genetic risk scores were associated with increased alcohol consumption, but not consistently with other socio‐demographic and life‐style factors.

Coffee‐related genetic risk scores could be used as instruments for combined coffee and tea consumption in Mendelian randomization studies. However, associations observed with alcohol consumption require further investigation to determine whether these are due to causal effects of coffee and tea consumption or biological pleiotropy.

Partial Text

Genome‐wide association studies (GWAS) have identified eight independent loci that are associated with coffee consumption at the genome‐wide significance level 1, 2, 3. The two most strongly associated loci identified to date are in or near genes which are involved in caffeine metabolism, namely the cytochrome P450 1A1 and 1A2 (CYP1A1/2) gene region and aryl‐hydrocarbon receptor gene (AHR) 4. CYP1A2 is the enzyme responsible primarily for metabolizing caffeine and AHR affects CYP1A2 activity 4. It is likely that these variants affect coffee consumption through altering rate of caffeine metabolism; there is evidence that the coffee consumption‐increasing alleles of these single nucleotide polymorphisms (SNPs) decrease blood caffeine levels 5, 6. Of note, five of the six additional loci (or close proxies, in or near the following genes: GCKR, ABCG2, MLXIPL, BDNF, EFCAB5) have also been identified in GWAS of other phenotypes, such as body mass index and smoking initiation 3, 7, 8. Combining variants together in an allele score increases the amount of variance in coffee consumption explained 9, 10. However, there is also evidence that coffee‐related SNPs and genetic risk scores associate more broadly with caffeinated beverage consumption 10 and with decaffeinated coffee intake 3. Better characterization of the coffee and caffeine phenotypes that this genetic risk score capture is important for interpreting the results of analyses that use these as proxies or markers of coffee exposure.

We have confirmed associations of genetic risk score for coffee consumption using variants identified in published GWAS 1, 2, 3, with amount of consumption of both coffee and tea in a large sample of older adults of European ancestry from the United Kingdom. Using a more detailed measure of beverage consumption, the 24‐hour dietary recall, we found that the genetic risk score was associated with most subtypes of coffee consumption, and with black tea but not herbal tea consumption. These scores were also associated with decreased consumption of water and total drinks excluding tea and coffee.

A.E.T is in receipt of a GRAND award from Pfizer, which is unrelated to the submitted work.