Skip to Main Content

An official website of the United States government

Principal Investigator
Yingxi Chen
M.D., Ph.D., M.P.H.(Res)
National Cancer Institute
Position Title
Postdoctoral fellow
About this CDAS Project
PLCO (Learn more about this study)
Project ID
Initial CDAS Request Approval
Nov 16, 2023
A Mendelian Randomization study of alcohol intake and health effects in the PLCO study
Background: Heavy alcohol intake is an important cause of mortality and morbidity, although the effect of light or moderate alcohol use is complex. Evidence from observational studies suggests a J-shaped association where light to moderate drinking is related to a lower risk of mortality, cardiovascular disease, and certain types of cancer. However, due to the nature of traditional epidemiological design, it has been hypothesized that the observed beneficial effects may be due to residual confounding from lifestyle, socioeconomic status, and behavioral factors that corelated with modest alcohol intake. A Mendelian Randomization (MR) approach is intended to improve causal inference in traditional observational studies in conjunction with the growing availability of large-scale genomic databases because they are less susceptible to confounding but have its own limitations. One recent MR analysis used a nonlinear genetic instrument to analyze genetic data from the UK Biobank and reported a consistently risk-increasing relationship between cardiovascular risk and alcohol consumption at all levels. Light alcohol intake reportedly was associated with minimal increases in cardiovascular risk, whereas heavier intake showed exponential increases in risk. We propose to use the PLCO cohort linked to the newly released genotyping data to explore the association between alcohol intake and risk of cause-specific mortality and incident cancer. We will compare self-reported alcohol intake with linear and nonlinear MR to elucidate the impact of low-dose alcohol intake in a US population.

Methods: We propose to use linkage of GWAS data to questionnaire information, cancer registry, Medicare, and mortality data from 110,000 PLCO participants to examine the effects of alcohol intake on cause-specific mortality and incident cancer across levels of intake. The exposures will include self-reported and genetically predicted alcohol intake. Genetic instruments of alcohol intake will be constructed using single-nucleotide polymorphisms (SNPs) associated with alcohol intake identified in external data. We will first remove SNPs independently related to confounding factors, including smoking, coffee intake, body mass index, physical activity, vegetable intake, red meat intake, and overall health rating to create refined nonpleiotropic genetic instruments. The main outcome of interest will be cause-specific mortality, incident cancer. PLCO data will also be linked to Medicare data to assess the association between alcohol and incident events of liver disease and myocardial infarction. Analysis will be conducted using the inverse variance–weighted, weighted median, MR–pleiotropy residual sum and outlier, traditional MR methods, and nonlinear MR methods.

Expected outcomes: We expect genetic evidence to support a causal link between alcohol intake and health outcomes, though effects may vary across level of intake and by types of health outcomes. We expect detailed description of health effect in relation to light and moderate alcohol intake and a comparison of the utility of self-reported versus MR approaches.

Conclusion: This large-scale nonlinear MR analysis of alcohol use and health outcomes will provide important insights contributing to public health recommendations.

Using linkage of GWAS data to questionnaire information, cancer registry, Medicare, and mortality data from 110,000 individuals enrolled in the PLCO, our overall objective is to assess the effect of alcohol intake in relation to health outcomes of interest, and to evaluate the direction and relative magnitude of risk associated with different amounts of alcohol intake.

Specific aim 1: To evaluate all-cause mortality in relation to alcohol intake at all levels. Using nonlinear MR analysis of linked GWAS data to questionnaire and mortality data, we will examine the association with a specific focus on the effects of light to moderate alcohol intake.

Specific aim 2: To evaluate the association between alcohol intake and incident cancer. Using nonlinear MR analysis of linked GWAS data to questionnaire and registry data, we will estimate the effect of alcohol intake at all levels in relation to overall incident cancer and alcohol-related cancers.

Specific aim 3: To evaluate and compare the estimates using self-reported and genetic predicted alcohol intake. Using genetic instruments of alcohol intake identified in external data, we will estimate genetic predicted alcohol intake in the study population. We will then compare the results to self-reported alcohol intake.


Yukiko Yano, Metabolic Epidemiology Branch, DCEG, NCI
Christian Abnet, Metabolic Epidemiology Branch, DCEG, NCI
Haoyu Zhang, Biostatistics Branch, DCEG, NCI
Wen-Yi Huang, Metabolic Epidemiology Branch
Sonja Berndt, Occupational Environmental Epidemiology Branch, DCEG, NCI