Psychological Medicine, 1999, 29, 1069-1081. Printed in the United Kingdom
©
1999
Cambridge University Press
Genetic differences in alcohol sensitivity and the
inheritance
of
alcoholism risk
A. C.
HEATH,l
P.
A. F.
MADDEN,
K.
K.
BUCHOLZ,
S.
H.
DINWIDDIE,
W.
S.
SLUTSKE,
L. J.
BIERUT,
J.
W.
ROHRBAUGH,
D. J.
STATHAM,
M.
P.
DUNNE,
J.
B.
WHITFIELD
AND
N.
G.
MAR
TIN
From
the Department
of
Psychiatry, Washington University School
of
Medicine, St. Louis and Department
of
Psychology, University
of
Missouri, Columbia, MO, and Department
of
Psychiatry, Chicago Medical School,
North Chicago, IL,
USA
..
and Department
of
Epidemiology and Population Health, Queensland Institute
of
Medical Research and Department
of
Epidemiology, Queensland University
of
Technology, Brisbane,
Queensland, and Department
of
Clinical Biochemistry, Royal Prince Alfred Hospital, Sydney, New South
Wales, Australia
ABSTRACT
Background. Substantial evidence exists for
an
important
genetic contribution to alcohol dependence
risk
in
women
and
men.
It
has been suggested
that
genetically determined differences in alcohol
sensitivity may represent one pathway by which
an
increase in alcohol dependence risk occurs.
Methods. Telephone interview follow-up
data
were obtained
on
twins from male, female
and
unlike-
sex twin pairs who had participated
in
an
alcohol challenge study in 1979-81, as well as other pairs
from the same Australian twin panel surveyed by mail in 1980-82.
Results.
At
follow-up, alcohol challenge men did
not
differ from
other
male twins from the same
age
cohort
on
measures
of
lifetime psychopathology
or
drinking habits;
but
alcohol challenge
women were
on
average heavier drinkers
than
other women. A composite alcohol sensitivity measure,
combining subjective intoxication
and
increase in body-sway after alcohol challenge in 1979-81,
exhibited high heritability (60 %). Parental alcoholism history was weakly associated with decreased
alcohol sensitivity in women,
but
not
after adjustment for baseline drinking history,
or
in men. High
alcohol sensitivity in men was associated with substantially reduced alcohol dependence risk
(OR
= 0'05,
95
%
CI
0'01-0'39).
Furthermore,
significantly decreased (i.e. low) alcohol sensitivity
was observed in non-alcoholic males whose
MZ
co-twin had a history
of
alcohol dependence,
compared
to
other non-alcoholics. These associations remained significant in conservative analyses
that
controlled for respondents' alcohol consumption levels
and
alcohol problems in 1979-81.
Conclusions. Men (but
not
women)
at
increased genetic risk
of
alcohol dependence (assessed by
MZ
co-twin's history
of
alcohol dependence) exhibited reduced alcohol sensitivity. Associations with
parental alcoholism were inconsistent.
INTRODUCTION
Twin
and
adoption studies provide compelling
evidence for
an
important
genetic contribution
to alcohol dependence risk (AlcD) in populations
of
predominantly European ancestry
both
in
men (reviewed in McGue, 1994; Heath, 1995;
Heath
et
al. 1997
a)
and
in women (Heath, 1995;
Heath
et
al. 1997 b). Emerging positive findings
I Address for correspondence: Dr Andrew
C.
Heath.
40
North
Kingshighway.
Suite.
I.
St.
Louis, MO 63108. USA.
in linkage studies
of
alcohol dependence in men
and
women (Long
et
al. 1998; Reich
et
al. 1998),
and
genetic association studies (e,g. Whitfield
et
al. 1998) also suggest genetic influences. Con-
siderably less progress has been made in
delineating the mechanisms by which genetic
differences between individuals ultimately give
rise to differences in AlcD risk.
History
of
major depression, which
is
known
to exhibit substantial co-morbidity with alcohol
dependence (Regier
et
al. 1990; Kessler
et
al.
1069
1070
A.
C.
Heath and others
1996)
has been found to
be
significantly but only
moderately genetically correlated with alcohol-
ism risk (Kend1er
et al.
1993,
1995).
Childhood
conduct problems,
or
adult antisocial behaviour,
which exhibit even stronger associations with
alcohol dependence (Regier
et
al. 1990; Kessler
et al.
1996)
have been found to be significantly
genetically correlated with alcohol dependence
in some (Grove
et al. 1990; Pickens et al. 1995;
Slutske et al.
1998)
but not all (True
et
al. 1999)
twin studies that have examined this issue, but
findings from adoption studies have generally
been negative (Schulsinger, 1972; Goodwin
et
al. 1973; Crowe, 1974; Cadoret et al. 1985,
1987),
perhaps because
of
the reduced power
of
detecting genetic correlations in the adoption
compared with the twin design. It does not
appear that history
of
depression and conduct
problems entirely accounts for genetic influences
on alcoholism risk. In analyses
of
alcoholism
data from a large interview survey
of
the
Australian twin panel, the alcoholism history
of
the respondent's MZ co-twin remained a power-
ful
predictor even when history
of
depression,
childhood conduct disorder, and various per-
sonality and sociodemographic measures were
controlled for, implying that much
of
the genetic
influence on alcoholism risk could not be
explained
by
co-morbid psychopathology or
personality differences (Heath
et
al.
1997
b).
Known polymorphisms at the alcohol dehydro-
genase ADH2 and ADH3 loci were found to
account for a relatively small proportion
of
the
total variance
in
alcohol problem and alcohol
consumption measures
in
a subsample from this
same panel (Whitfield
et al.
1998),
suggesting
that other important sources
of
genetic variation
remain to be detected.
Schuckit has argued that innate differences in
alcohol sensitivity, operationalized as differences
in
response to a challenge dose
of
alcohol
(including subjective intoxication (Schuckit,
1984a); static ataxia (Schuckit,
1985)
and
hormonal measures (Schuckit,
1984b» may
be
predictors
of
increased alcoholism risk. Con-
sistent with this hypothesis, adult offspring
of
alcoholics have generally been found to
give
lower ratings
of
intoxication· after alcohol
challenge than controls
(Pollock,
1992)
and have
usually (Lipscomb
et
al.
1979; Hegedus et af.
1984;
Lex
et al. 1988; McCaul et al.
1991)
but
not always (Behar
et
af.
1983; O'Malley &
Maisto, 1985; Nagoshi & Wilson, 1987; Bauer
& Hesselbrock,
1993)
been found to differ in
amount
of
increase in body-sway, though the
direction
of
effect has not always been consistent
across studies
of
body-sway. Also consistent
with this hypothesis, long-term follow-up
of
male alcohol challenge participants has found
increased rates
of
alcohol dependence among
those with low initial alcohol sensitivity
(Schuckit &
Smith,
1996).
Direct evidence from
twin and adoption studies that these measures
of
alcohol response are
in
part under genetic control
is
somewhat meager, though results from the
Australian Alcohol Challenge Twin
Study
(AACTS) (Martin et al. 1985a, b) confirm
genetic influences on post-alcohol increase in
body-sway (Martin
et
al. 1985b; Heath &
Martin,
1992)
and subjective intoxication rating
(Neale & Martin, 1989; Heath & Martin,
1992).
To our knowledge, the question
of
whether such
differences in alcohol sensitivity do indeed
mediate genetic influences on alcohol depen-
dence risk has not previously been directly
examined using the design than can provide the
most powerful test
of
this hypothesis, the
classical twin design. In addition, much
of
the
existing literature on alcohol challenge per-
formance has been gathered only in male
subjects. Here
we
address the question
of
whether differences in alcohol sensitivity are
genetically correlated with AlcD risk,
in
both
women and men, using data from a follow-up
interview survey
of
the Australian twin panel
(Heath
et al.
1997
b), that included twins who
participated
in
AACTS. Unusually, for such an
early study,
AACTS included approximately
equal numbers
of
men and women. Because
information about alcohol dehydrogenase geno-
types (ADH2 and ADH3) are also available for
this sample (Whitfield
et al.
1998),
we
also assess
whether polymorphisms at these loci contribute
significantly to differences
in
alcohol sensitivity.
METHOD
Sample
Participants were twins from a volunteer adult
twin register, formed in 1978-9 and maintained
by
the Australian National Health and Medical
Research Council (NH&MRC). Almost all were
of
European ancestry. In 1979-81,
206
young
adult twin pairs born 1944-63 had participated
Alcohol sensitivity and inheritance
of
alcoholism risk
1071
in an alcohol challenge study (Martin et al.
1985a, b),
including
45
monozygotic (MZ)
fe-
male,
43
MZ
male,
42
DZ
female,
37
DZ
male
and
39
DZ
unlike-sex pairs. Approximately
equal numbers
of
women
(N
= 213) and men
(N
=
199)
were tested. Additional data on this
sample were obtained in a mailed questionnaire
survey
of
the entire adult twin panel conducted
in
1980-82, with responses from
132
complete
pairs and
16
single twins from participants in the
alcohol challenge study (Jardine
et
al.
1984;
Kend1er et
al.
1986),
and in a telephone interview
follow-up in 1992-3 (Heath
et al.
1997
b). At
interview follow-up, data were obtained from
187
women (87,4% response rate) and
162
men
(81·4
%). An additional
five
subjects were de-
ceased, seven subjects were overseas and could
not be reached, and
28
subjects were not
contacted because they either could not be
located or had previously requested that they
not be contacted for further research studies.
Excluding these cases, the cooperation rate
in
the interview follow-up was
95·4
% for women
and
90·1
% in men.
In addition to participants in the alcohol
challenge study, an additional
3676
complete
twin pairs and
551
single twins from the
NH&MRC twin panel, born 1893-1964,
responded to the
1980-2 questionnaire survey.
In the 1992-3 survey, follow-up interviews were
completed with
3659
of
these women and
1879
men (respectively
88·3
% and
82·5
%
of
those
eligible for follow-up (Heath
et
al.
1997
b».
Comparisons
of
the subset
of
these subjects
born 1944-63
(2628
women,
1584
men) with the
alcohol challenge (AC) participants were used to
identify correlates
of
willingness to participate
in
the alcohol challenge study, and thus provide
a check on the generalizability
of
findings from
the AC sample.
Assessments
Participants in the Australian Alcohol Challenge
Twin Study (AACTS) received baseline ques-
tionnaire assessments
of
drinking history, per-
sonality and sociodemographic variables, and
baseline and post-alcohol
(0'75 g ethanol/kg
body weight) assessments
of
subjective intoxi-
cation (rated on a
10-point scale where
I
= completely sober,
10
= most drunk the
respondent has ever been, and rescaled
by
dividing by
10),
static ataxia assessed both with
eyes open and
eyes
closed, and other measures
of
psychomotor coordination (Martin et
al.
1985b; Neale & Martin, 1989; Heath & Martin,
1992).
Guided by the work
of
Schuckit & Smith
(1996), analyses
of
associations with alcohol
dependence risk focused on subjective ratings
of
intoxication and static ataxia measures, using
data from the first round
of
post-alcohol testing,
beginning
20
min after dosing. To quantify
change in body-sway, regression residuals were
estimated from a regression equation predicting
post-alcohol body-sway from pre-alcohol sway
scores (Nagoshi & Wilson,
1987),
and a quad-
ratic power transformation
was
used to reduce
skewness and kurtosis.
As
presented here, low
scores indicate a smaller increase in body-sway
after alcohol challenge (i.e. low sensitivity to
alcohol). Principal component analysis
of
the
subjective intoxication rating and static ataxia
residual scores was used to generate 'alcohol
sensitivity' component scores. Loadings on the
first principal component
of
the subjective
intoxication and
eyes
open and eyes closed
ataxia measures were
0'40,
0·64
and
0·65
in
women, and 0'49,
0·61
and
0·62
in men. The first
principal component accounted for
61
%
of
the
variance in these variables in both women and
men.
Included
in
the AACTS baseline questionnaire
were limited drinking history questions that
included measures
of
heavy drinking (number
of
drinks per typical drinking occasion, scaled
as
(0)
1-2;
(I)
3-5;
(2)
6-8; and
(3)
~
9),
excessive
drinking (drinking more than the respondent
felt was good for him/her), morning drinking,
frequency
of
being drunk, and frequency of
being hungover (these two latter with response
categories never, sometimes or often). Responses
to these
five
items were summed to yield a
baseline problem drinking score with value
ranging from
0-15. Quantity and frequency
measures
of
typical
weekly
alcohol consumption
were obtained, from which total weekly alcohol
consumption
in
standard drinks
was
computed,
and log-transformed. In this early study no
attempt was made at baseline to assess family
history
of
alcoholism, nor to exclude subjects
reporting a personal history
of
alcohol problems.
Follow-up assessments
in
1992-3 included
global ratings
of
maternal and paternal alcohol
problems (Slutske
el
al.
1996),
as
well
as DSM-
III-R assessments
of
history
of
AlcD, major
1072
A.
C.
Heath and others
depressive disorder
(MDD)
(DSM-IV diagnoses
were derived for this variable), panic disorder,
social phobia
and
adolescent conduct disorder
(CD). These diagnostic assessments were adap-
ted for telephone administration from the Semi-
Structured Assessment for the Genetics
of
Alcoholism (SSAGA) (Bucholz et
al.
1994). A
positive parental history
of
alcoholism was
inferred
if
at
least one twin reported parental
alcohol problems. In addition, recent (12-month)
quantity and frequency measures
of
alcohol use,
as well as 12-month
and
lifetime estimates
of
the
respondent's heaviest consumption in a single
day, were obtained. All consumption measures
were assessed in standard drinks.
As described elsewhere,
ADH2
and ADH3
genotypes for 369 participants in the alcohol
challenge study (176 male,
193
female) were
typed either
at
the time
of
the interview follow-
up,
or
at
the time
of
a separate follow-up study
of
alcohol challenge participants (Whitfield et
al.
1998): genotypes for
ADH2
and ADH3 were
determined using
DNA
extracted from white
blood cells, using the polymerase chain reaction
and
restriction digestion, followed by electro-
phoresis
of
the polymerase chain reaction (PCR)
products (von Wartburg
et
al.
1988; Xu et
al.
1988). The only genotypes observed in the
sample were
ADH2
*1/*1
and
*1/*2,andADH3
*1/*1, *1/*2 and *2/*2.
All participants in the original alcohol chal-
lenge study gave written informed consent.
Implied consent procedures were followed with
the
1981
questionnaire survey, a cover letter
explaining the research
and
that
participation
was voluntary. All participants in the telephone
interview survey gave verbal consent to par-
ticipate in the research, after the elements
of
informed consent had been reviewed with them
verbally.
Statistical analyses
Rates
of
psychopathology and heavy alcohol
use were compared between the alcohol chal-
lenge and comparison samples by chi-square
test.
For
these comparisons no adjustment was
made for non-independence
of
observations on
twin pairs, since in this case the overestimation
of
statistical significance would be conservative,
i.e. would cause us to suspect sampling biases
where none existed. To adjust for known
differences at baseline between the alcohol
challenge and comparison samples on socio-
demographic
and
baseline alcohol consumption
measures, results
of
unweighted analyses were
compared to analyses using sampling weights to
adjust for this volunteer bias effect (Rosenbaum
& Rubin, 1983; Heath et
al.
1998). Twin-pair
covariances were computed for alcohol sen-
sitivity scores, and models allowing for additive
and non-additive genetic
and
non-shared en-
vironmental influences were fitted to these data
by maximum-likelihood using standard model-
fitting methods (Neale
& Cardon, 1992; Neale,
1998).
Associations between alcohol sensitivity
scores and parental history
of
alcohol depen-
dence, respondent's history
of
alcohol depen-
dence, and co-twin's. history
of
alcohol de-
pendence, were assessed using the SAS
(1990)
GLM
procedure. Post-hoc analyses predicting
lifetime alcohol dependence as a function
of
low
alcohol sensitivity (
<
25
%ile
of
the sex-specific
distribution
of
alcohol sensitivity scores)
or
high
alcohol sensitivity
(>
75%ile) or ADH2
or
ADH3 genotypes, were conducted using logistic
regression analysis. Most analyses are reported
with and without adjustment for baseline
covariates (baseline weekly alcohol consumption
and alcohol problem measures): since differences
in alcohol sensitivity may contribute to
differences in alcohol dependence risk via effects
on levels
of
alcohol consumption
or
excessive
drinking, covariate adjustment must be con-
sidered a conservative procedure, which may
cause us to underestimate the significance
of
findings.
For
critical analyses
of
associations
with parental
or
own and co-twin's alcoholism
history, bootstrapping (Efron
& Tibshirani,
1986)
- randomly resampling the observed data
with replacement, using the twin pair as the unit
for resampling - was used to obtain empirical
estimates
of
the standard errors
of
parameters,
adjusted for the non-independence
of
obser-
vations on twin pairs (i.e. clustered sampling). In
each case,
3000 bootstraps were run per analysis.
Logistic regression analyses were conducted
using STAT A (Stata Corp.,
1997)
to obtain
95
% confidence intervals that were corrected
for clustered sampling.
Other significance tests.
which were not significant even without ad-
justment for correlated observations on twin
pairs, are reported using the unadjusted test
statistic, since correction for non-independence
would only make this statistic even
less
signifi-
Alcohol sensitivity and inheritance
of
alcoholism risk
1073
cant. Where significant chi-squares were
obtained,
we
used the conservative adjustment
of
dividing the chi-square by one-half, equivalent
to assuming
that
observations on twin pairs
were perfectly correlated. Finally, bivariate
genetic models (e.g. Neale
& Cardon, 1992;
Kendler
et
al.
1993)
were fitted to alcohol
dependence symptom count and alcohol sen-
sitivity score data, to quantify the magnitude
of
the genetic correlation between sensitivity score
and dependence symptom count.
For
this final
analyses, models were fitted to raw data by
maximum likelihood, using all available data on
alcohol dependence symptom count from male
and female twins born 1944-63, plus alcohol
sensitivity data from male twins only (since no
association was observed in female twins).
RESULTS
GeneralizabiIity
of
sample
Lifetime history
of
psychopathology, reported
family history
of
alcohol problems, and current
alcohol use
of
the AC and comparison samples,
as assessed
at
interview follow-up, are sum-
marized in Table
1.
While male AC participants
appeared to be representative, with respect to
these measures,
of
all males from the same birth
years on the Australian twin panel, this was not
true for women. Among women, the lifetime
prevalence
of
DSM-III-R
AlcD was substan-
tially higher among those who had volunteered
for the AC study than among those who had
not. There were also modestly increased rates
of
conduct disorder (P = 0'03) and
of
social phobia
(P
=
0·02)
among the AC women.
At
the time
of
follow-up, AC women remained heavier drinkers
on measures
of
high current frequency
of
alcohol
use, number
of
drinks per drinking occasion,
and maximum I-day consumption in the pre-
ceding year, and lifetime maximum I-day con-
sumption was also higher. Male AC participants
did not differ significantly from the comparison
sample, though there was a trend for lower
lifetime rates
of
psychopathology among male
AC participants. Reported rates
of
parental
alcohol problems did not differ significantly
in
either gender, nor when data were pooled across
gender (not shown).
Despite the over representation
of
heavy
drinkers among women AC participants, there
was a wide range
of
levels
of
exposure to
alcohol, as can
be
seen from Fig. I. Fig. 1 (a)
summarizes the distribution
of
average weekly
Table
1.
History
of
psychopathology and alcohol use at interview follow-up
of
the alcohol
challenge and comparison samples (all differences are non-significant unless otherwise indicated)
Women
Men
Alcohol Comparison Alcohol Comparison
challenge sample challenge
sample
(N=
182-190)
(N
= 2452-2628)
(N
= 158-182)
(N
= 1421-1584)
%
% % %
Psychopathology/family history
Childhood conduct disorder
5-4
2'7*
18·0
19·3
Alcohol dependence
15·3
7'0***
24·1
27·5
Major depression (DSM·IV)
26·2
22·2
12-4
16·9
Panic disorder
5·4
4·4
1·9
2-4
Social phobia
4·9 2')*
),2
2-4
Paternal alcohol problems
26·9
23·5 25·3
21'6
Maternal alcohol problems
7·9
5·6
5·0
4·2
Any parental alcohol problems
30·0
27·0
28·3
24·3
Alcohol use
lifetime
maximum consumption/24 h
28·0
:W'I*
20·8
17'5
(>
10
drinks in women; >
25
drinks in men)
Current maximum consumption/24 h
20·9 l3-3**
27·2
26·3
(>
6 drinks in women; >
10
drinks in men)
Frequency
of
alcohol use
;.:
3 days/week
31·7
20-2*··
41'4
36·5
Drinks per drinking occasion
13·2
11'3***
34·6 29·9
(>
3 drinks)
Diagnostic measures are based on
DSM·III·R
criteria.
and
alcohol-use measures are for preceding
12
months. unless otherwise noted.
Sample sizes for family history assessments include co-twin reports in cases where a subject was lost to follow-up.
* P < 0'05;
***
P < 0·00 I for comparisons between alcohol challenge and comparison samples.
