More than 3,500 genetic variations that potentially affect smoking and drinking behaviour have been identified in a study involving almost 3.4 million people with African, American, East Asian and European ancestry.
The findings, published in Nature on 7 December1, highlight how increasing sample size and ethnic diversity improves the power of such genome-screening analyses — called genome-wide association studies (GWAS) — to reveal how various traits are linked to genes, combinations of genes or mutations.
Smoking and drinking are important risk factors for several physical and mental illnesses, including cardiovascular diseases and psychiatric disorders. Although both behaviours are influenced by environmental and social factors, there is evidence that genetics can affect tobacco and alcohol consumption. “We’re at a stage where genetic discoveries are being translated” into clinical applications, says study co-author Dajiang Liu, a statistical geneticist at Penn State College of Medicine in Hershey, Pennsylvania. “If we can forecast someone’s risk of developing nicotine or alcohol dependence using this information, we can intervene early and potentially prevent a lot of deaths.”
Scientists use GWAS to find genetic ties to diseases or behaviours by comparing genetic sequences in large numbers of people. But so far, most of these studies have focused on European populations. Liu and his colleagues constructed a model that incorporated the genomic data of 3,383,199 people, 21% of whom had non-European ancestry.
They identified 3,823 genetic variants that were associated with smoking or drinking behaviours. Thirty-nine of these were linked with the age at which individuals started smoking, 243 with the number of cigarettes smoked per day and 849 with the number of alcoholic drinks consumed per week.
Of the total number of associated variants, 721 were identified only by the multi-ancestry study, and not by an ancestry-naive model that the authors used for comparison. This suggests that large and diverse population samples significantly increase the power of such studies.
The researchers found that the majority of genetic associations for drinking and smoking have similar effects across the different ancestries. “We also find similar heritability estimates [for the traits] across the ancestries … suggesting that generally, the genetic architecture of these behaviours is similar across ancestries,” says Gretchen Saunders, a psychologist at the University of Minnesota, Minneapolis, and co-author of the paper.
However, they also showed that polygenic risk scores — based on a combination of multiple genetic factors — that were specific to the European ancestry group were poor predictors of smoking and drinking behaviours in other ancestry groups. “Even with these big sample sizes, they just do not transfer across populations,” says Saunders.
The similarity across ancestries could be partly because the vast majority of non-European cohorts included in the study live in the United States, and therefore have similar environmental influences — such as public-health policies and the availability of alcohol and nicotine products — says Ananyo Choudhury, a geneticist at the University of the Witwatersrand in Johannesburg, South Africa.
“Epigenetic and environmental factors are really important to turn off and turn on the genes. So maybe it’s because of that reason” that there aren’t many significant differences, adds Şehime Temel, who studies medical genetics at Bursa Uludağ University, Turkey.
The analysis also didn’t include people from Middle Eastern and Indian populations, in which smoking is often prevalent. “Tobacco use is very common [in the Middle East]. There is a huge consumption of the shisha waterpipe,” says Mahmut Ergören, a medical biologist at the Near East University in Nicosia, Cyprus. He adds that including these populations in the analysis would improve its accuracy and help to identify more genetic associations.
The researchers acknowledge that their sample does not capture global diversity in genetic ancestry or geography. “While being the largest and most ancestrally diverse study of smoking and drinking phenotypes so far, it has not covered all populations,” says Liu. “In future phases of the study, we will welcome collaborations from other investigators who have access to additional data sets to further expand our studies.”