Using Twin Data To Identify Alternative Drug Abuse Phenotypes

1
Using Twin Data To Identify Alternative Drug
Abuse Phenotypes

• Ming T. Tsuang, MD, Ph.D.
• University Professor, University of California
  Director,
• Institute of Behavioral Genomics, Dept. of
  Psychiatry,
• UC San Diego
• Director, Harvard Institute of Psychiatric
  Epidemiology and Genetics, Depts. of Epidemiology
  Psychiatry, Harvard University

2
Background

• Some of the difficulty in identifying genes for
  drug abuse stems from underlying etiologic
  complexity of this phenotype.
• Unlike disorders such as Huntingtons disease,
  drug abuse is presumed to have a multifactorial
  polygenic etiology, in which numerous genes and
  environmental factors all make small
  contributions to the overall risk for the
  illness.

3
Moving Beyond Off the Shelf DSM Phenotypes

• Empirical methods, such as factor analysis, may
  help to identify useful quantitative traits.
• Transitions may reflect different phenotypes with
  different genetic determinants
• Psychiatric disorders that co-occur with
  substance abuse may reflect the same genetic
  vulnerability

4
Previous Relevant Research Using the Vietnam Era
Twin Registry
5
Sample Demographics

• Mean age - 44.6 years (2.8 years)
• Age range - 36 to 55 years
• Ethnicity non-Hispanic white
  90.4 African American 4.9 Hispanic 2.7
  
  Native American 1.3
  Other 0.7

6
Sample Size

• Number of individuals 8,169
• response rate 79.7
• completed pairs MZ - 1,874 DZ - 1,498

7
Vietnam Era Twin Registry

• Consists of 7,375 male-male twin pairs
• Both served on active duty during the Vietnam Era
  (1965-1975)

• Born between 1939 and 1957
• Identified from Dept. of Defense data files by
  computer algorithm

8
Vietnam Era Twin Registry

• Zygosity determined by questions on sibling
  similarity and blood group typing data from
  military records
• All were raised together

9
Using Transitions in Drug Use to Define
Phenotypes for Genetic Research

• Unaffected phenotype should be restricted to
  individuals with the opportunity to become
  abusers who did not become abusers
• Individuals without the opportunity to become
  abusers should be classified phenotype unknown

10
Exposure to Drug
Stages in Drug Usage
Initiation First Use
Phenotype Unknown
Regular Use
Abuse (Problematic Use)
Affected Phenotype
Dependence
Discontinuation of Use
?
11
Conclusions

• There is no single, unitary drug phenotype that
  will encompass all aspects of the relevant
  phenomena.
• There is considerable overlap among the various
  drug phenotypes that could be formulated.

12
Univariate Analyses of Drug Abuse
13
Pairwise Concordance Rates for Drug Abuse
14
Correlations for Drug Abuse
Plt.001
Plt.001
Plt.001
Plt.001
15
Influences on Drug Abuse/Dependence
16
Conclusions from Univariate Analyses

• Genetic factors are the most important
  determinants of heroin addiction.
• Genetic factors may be more important for heroin
  addiction than for addiction to other drugs.
• Results for heroin suggest there may be both
  polygenic effects and a single gene of major
  effect or epistasis.

17

• Question for Multivariate Analyses
• To what extent are determinants of addiction
  shared among all drugs versus unique to each
  individual drug?

18
Vulnerability to Drug Dependence
Heroin/ Opiates 50
Psychedelics 85
Common Vulnerability
Stimulants 77
Sedatives 69
Marijuana 71
19
Determinants of Common Vulnerability to Drug
Dependence
20
Family Environmental Influences on Drug
Dependence
Common Family Environmental Vulnerability
(c215) Heroin/ Opiates 100
(c219) Psychedelics 100
100 Sedatives (c217)
59 Marijuana (c229)
100 Stimulants (c218)
21
Non-Family Environmental Influences on Drug
Dependence
(e252) Psychedelics 71
(e233) Heroin/ Opiates 64
Common Non-Family Environmental Vulnerability
54 Sedatives (e256)
84 Marijuana (e238)
71 Stimulants (e248)
22
Genetic Influences on Drug Dependence
(h254) Heroin/ Opiates 30
Common Genetic Vulnerability
(h226) Psychedelics 100
73 Stimulants (h233)
(h227) Sedatives 81
(h233) Marijuana 67
23
Conclusions from Multivariate Analyses

• The best model to explain the co-occurrence of
  addiction to different drugs is a common latent
  vulnerability.
• The Marijuana Gateway model is a poor fit to
  the data.

24
Conclusions from Multivariate Analyses

• About half of the influences on heroin addiction
  also impart risk for addiction to other illicit
  drugs.
• About half of the influences on heroin addiction
  are unique to heroin (i.e, they dont affect the
  risk for addiction to other drugs).

25
Conclusions from Multivariate Analyses

• Everything about the family environment that
  imparts risk for heroin addiction also imparts
  risk for addiction to all other illicit drugs.
• Only marijuana is affected by family
  environmental factors that dont also influence
  other illicit drugs.

26
Conclusions from Multivariate Analyses

• Some aspects of the non-family environment that
  affect heroin addiction also affect addiction to
  other drugs.
• Some aspects of the non-family environment that
  affect heroin addiction are unique to heroin
  addiction (i.e., dont affect addiction to other
  drugs).

27
Conclusions from Multivariate Analyses

• 70 of the genetic influence on heroin addiction
  is unique to heroin addiction (more than for any
  other drug).
• Heroin addiction is the most heritable
  addiction to an illicit drug (at least given the
  environmental circumstances of our sample).

28
Future Directions for Investigating Phenotypes
for Genetic Research on Drug Abuse
29
Background Significance

• A genetic contribution to the liability toward
  illicit drug use has been firmly established.
• The task is now to identify specific genes that
  influence the liability to substance use
  disorders (SUDs).
• Defining genetically homogenous phenotypes is
  critical to the success of genetic linkage and
  association studies.

30
Alternative Phenotypes

• Heterogeneous nature of phenotypes compounds
  difficulty of identifying genes that impart risk.
• DSM diagnosis of drug abuse may not map neatly
  onto its genetic foundations.

31
Alternative Phenotypes

• May provide a stronger signal in the search for
  underlying risk genes.
• We propose to utilize data from both twin and
  molecular genetic samples to generate and refine
  alternative phenotypic definitions of substance
  use disorders (SUDs).

32
Summary

• There is good evidence for genetic influences on
  SUDs.
• The specific genes that influence SUDs have
  remained elusive.
• A rate-limiting step in finding genes for SUDs
  may be specifying the best phenotypes.