Abstract
The primary pathway of alcohol metabolism involves two main enzymes, alcohol and aldehyde dehydrogenase.
Several genes (ALDH2, ADH1B) that encode these enzymes have variant alleles that alter the rate of metabolism
and result in heightened and protracted exposure to acetaldehyde, a known carcinogen. The variant ALDH2*2
allele is associated with the flushing response (i.e., “Asian glow”) and is found almost exclusively in individuals
of east Asian descent (560+ million people worldwide). Although possession of variant ALDH2 and ADH1B
alleles are protective against heavy drinking and alcohol use disorders, for those who do drink, these variants
also are independently and synergistically associated with striking elevations in risk for several cancers, including
esophageal and head and neck cancers. This cancer risk, however, is not widely known outside of the research
community. The premise of this study is that we can affect early drinking trajectories through personalized
communication about these cancer risks. We specifically target 360 college students of northeast Asian descent,
a high-risk group as college is a time of escalating alcohol consumption. It is also a context in which a risk-
communications intervention, if successful, could be readily scaled-up. We will compare risk communication with
and without personalized genotype feedback by randomizing participants into one of three groups: 1) a group
receiving information about cancer risk associated with alcohol consumption and alcohol metabolism genetic
deficiencies that manifest as flushing, plus personalized flushing (i.e. phenotype) feedback (PHEN), 2) a group
receiving the PHEN information plus personalized genotype feedback on ALDH2 and ADH1B (PHEN+GENE),
or 3) a comparison attention CONTROL group receiving duplicate information from the AlcoholEdu® for College
mandatory online training about alcohol risk already completed by all incoming students. In Aim 1, we will develop
the brief web-based interventions. In Aim 2, we will test our hypotheses that a) informing drinkers of risk will
lower alcohol consumption levels and use of “flush cures”, and b) informing non-drinkers will reduce or delay
drinking onset. We also will test if genotype feedback results in a greater attenuation of drinking than general
risk information and phenotype feedback only, and if the impact is stronger in those at greater risk based on their
phenotype and/or genotypes. Finally, we will test the possible mediating roles of perceived cancer risk, cancer
prevention self-efficacy, and alcohol expectancies as underlying mechanisms of behavior change. Aim 3 will
determine study effect sizes within the latent growth modeling framework we will employ, including for our
mediation analyses. These analyses will establish the feasibility of implementing these interventions and setting
the stage for a larger multisite study. Results of this work may inform intervention/prevention efforts on how to
optimally target personalized feedback protocols for high-risk college samples and scale these at a national level.
