VACS Scientific Meeting

1
Whole-genome association analysis of risk taking
behavior in the context of alcohol abuse in a
cohort of HIV Positive veterans
VACS Scientific Meeting October 15, 2008
Arthur A. Simen, M.D., Ph.D. Assistant
Professor Divisions of Molecular Psychiatry,
Aging Research, and Human Genetics Department of
Psychiatry, Yale University School of Medicine
2
Study Team
Joseph Goulet, Ph.D., Associate Research
Scientist, Internal Medicine and General
Medicine, Yale University and VA-CT Healthcare
System. Amy C. Justice, M.D., Ph.D., Associate
Professor of Internal Medicine, General Medicine,
and Epidemiology and Public Health, Yale
University and VA-CT Healthcare System. John
Krystal, M.D., Robert L. McNeil Jr. Prof
Psychiatry, Yale University and VA-CT Healthcare
System. Shrikant Mane, Ph.D., Director
Affymetrix Resource and Molecular Biology and
Biochemistry Keck Biotech Services, Yale
University. Kristin Mattocks, Ph.D., M.Ph.,
Associate Research Scientist, Internal Medicine
and General Medicine, Yale University and VA-CT
Healthcare System. Arthur A. Simen, M.D., Ph.D.,
Assistant Professor, Department of Psychiatry,
Yale University School of Medicine. Hongyu Zhao,
Ph.D., Professor of Public Health (Biostatistics)
and Genetics Ira V Hiscock Associate Professor
of Epidemiology/Public Health Genetics and
Statistics, Department of Public Health and
Genetics, Yale University.
3
Specific Aims
Aim 1 The goal of this aim is to perform
whole-genome association analysis to identify SNP
variants and haplotypes that are associated with
shared risk for alcohol abuse, risky sexual and
drug-related decision making, and medication
non-compliance in the context of having HIV. A.
We hypothesize that alcohol abuse, risky sexual
and drug-related decision making, and medication
non-compliance show substantial co-morbidity in
part because they share common genetically
determined antecedents in the form of individual
SNP variants and SNP haplotypes that have
multiple (pleiotropic) behavioral effects. B.
We hypothesize that the genetic determinants
identified in Aim 1a will mediate the effects of
excess alcohol consumption and HIV with respect
to exhibiting risky behaviors. Aim 2 The goal
of this aim is to perform whole-genome
association analysis to identify SNP variants and
haplotypes that are associated uniquely with risk
for alcohol abuse in the context of having
HIV. A. We hypothesize that a portion of the
variance in risk for heavy alcohol use in the
context of having HIV is due to genetic variants
that uniquely affect risk for heavy alcohol use.
4
Alcohol use worsens outcomes for HIV positive
individuals

• Alcohol worsens medical outcomes in HIV
  individuals
• Alcohol increases the incidence of risky sexual
  behavior
• Alcohol increases the incidence of risky drug
  taking behavior
• Alcohol increases the incidence of medication
  non-compliance in
• HIV individuals

5
The prefrontal cortex mediates risk taking

• Risk taking is governed in part by prefrontal
  function
• HIV individuals have prefrontal deficits
• Alcohol causes further worsening of prefrontal
  function in HIV individuals

6
Genetic factors

• Genetic risk factors for specific disorders of
  impulse control have been identified
  (including alcoholism)
• Genetic risk factors that confer risk for
  multiple disorders of impulse control have been
  identified
• Genetic factors determine prefrontal function,
  and are strong candidates for impulse control
  disorders

7
Model
This theoretical model suggests that the
correlations between many disparate outcomes
measured by VACS can be understood in terms of a
common antecedent in the form of prefrontal
function. Mapping genes for shared risk will
maximize statistical power and shed light on a
very important public health problem.
8
Preliminary Data

• Multiple correspondence analysis
• 585 VACS subjects
• Multi-way contingency table consisting of binary
  variables
• Current alcohol abuse (AUDIT)
• Current IV drug abuse
• Hepatitis C status (taken as a proxy for IV drug
  abuse)
• Whether the subject drinks before sex (taken as
  indicative of risky sex).
• R/ "ca
• A single dimension accounted for 72.7 of the
  variance in these variables
• Supports the notion that multiple substance use
  and risk behaviors can be meaningfully summarized
  by a single common factor.
• These analyses are preliminary. Principal
  components analysis on the variables of interest
  is in progress.

9
Methods
Composite Phenotype Alcohol use (AUDIT)
instrument Alcohol consequences (Short Inventory
of Problems) Alcohol dependency (Alcohol
Dependency Scale) Tobacco use (Veterans Health
Study questionnaire) Illicit drug use
(DAST-10) Prescription drug abuse (NSDUH) Risky
sexual and drug use (Centers for Disease Control
instrument) Medication adherence (AIDS Clinical
Trails Group instrument) Quantitative trait
First principal component (PC1) of these
data. N500 subjects with highest and N500 with
lowest scores on PC1. Genotyping Illumina Human
1M Duo BeadChips. QC HWE. Gender checks. Drop if
call ratewith low MAFs. Analysis Bonferroni correction.
Genomic control for stratification. Analysis of
individual SNPs and haplotypes.
10
Thank you!
Comments and suggestions appreciated!
Arthur.Simen_at_Yale.edu