Drug and Alcohol Dependence 84 (2006) 85–92
Comorbidity between alcohol dependence and illicit drug dependence
in adolescents with antisocial behavior and matched controls
Soo Hyun Rheea,b,∗, John K. Hewitta,b, Susan E. Younga, Robin P. Corleya,
Thomas J. Crowleyc, Michael C. Nealed, Michael C. Stallingsa,b
aInstitute for Behavioral Genetics, Campus Box 447, University of Colorado, Boulder, CO 80309, United States
bDepartment of Psychology, Campus Box 345, University of Colorado, Boulder, CO 80309, United States
cDivision of Substance Dependence, Department of Psychiatry, Box C-268-35, University of Colorado Health Sciences Center,
4200 East 9th Avenue, Denver, CO 80262, United States
dDepartments of Psychiatry and Human Genetics, Virginia Institute for Psychiatric and Behavioral Genetics,
Virginia Commonwealth University, P.O. Box 980126, Richmond, VA 23298, United States
Received 28 May 2005; received in revised form 10 December 2005; accepted 14 December 2005
Background: Knowledge regarding the causes of comorbidity among substance use disorders can have significant impact on future research
examining the etiology of these disorders. Unfortunately, the conclusions of past studies examining the comorbidity among substance use disorders
are conflicting; some studies emphasize familial influences common to multiple substances, while others emphasize substance-specific influences.
Discrepancies in results may reflect different analytical approaches or differences in the samples. Here, we examine the causes of comorbidity
between alcohol dependence and illicit drug dependence in adolescents.
Methods: We ascertained a clinical sample of adolescents treated for antisocial behavior and substance use disorders and their siblings and a
matched control sample. A model fitting approach was used to test 13 alternative hypotheses for the causes of comorbidity.
Results: The best supported hypothesis for the comorbidity between alcohol dependence and illicit drug dependence was a model hypothesizing
that comorbid disorders are alternate forms of a single underlying liability. The next best fitting models were two of the correlated liabilities models
(correlated risk factors and reciprocal causation).
Discussion: The results suggest that the best hypotheses explaining the comorbidity between alcohol and illicit drug dependence in adolescents
are that alcohol dependence and illicit drug dependence are manifestations of a single general liability to develop substance dependence or that
there are separate liabilities that are highly correlated.
© 2006 Elsevier Ireland Ltd. All rights reserved.
Keywords: Comorbidity; Alcohol dependence; Illicit drug dependence; Adolescence
Family and twin studies suggest that there are familial and
genetic influences on problems with alcohol (Kaprio et al.,
1987; McGue et al., 1992; Kendler et al., 1997b). Although less
research regarding the etiology of illicit drug abuse and depen-
(Gynther et al., 1995; Tsuang et al., 1996; van den Bree et al.,
1998; Kendler et al., 1997a). Given these findings and the evi-
∗Corresponding author. Tel.: +1 303 492 4631; fax: +1 303 492 8063.
E-mail address: firstname.lastname@example.org (S.H. Rhee).
ily (Kendler et al., 1997a; Bierut et al., 1998; Merikangas et al.,
1998), twin (Grove et al., 1990; Johnson et al., 1996; Pickens et
al., 1995; True et al., 1999; Tsuang et al., 1998), and adoption
studies (Cadoret et al., 1986; Cadoret et al., 1995) have tested
whether there are common familial influences on substance use
Increasing knowledge regarding the causes of comorbidity
among substance use disorders could have significant impact on
future research examining the etiology of these disorders. The
results of studies examining the causes of comorbidity are espe-
cially interesting to geneticists searching for specific genetic
mechanisms underlying risk for substance use disorders. If
0376-8716/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved.
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
studies examining the causes of comorbidity among substance
use disorders suggest the importance of substance-specific
familial influences, the results would recommend the search
for potential quantitative trait loci (QTL) or candidate genes
for problems with specific substances. In contrast, if there are
significant common familial influences on problems with dif-
ferent substances, the results would recommend the search for
QTL or candidate genes that affect the common vulnerability
underlying problems with different substances. Unfortunately,
the conclusions of studies examining the causes of comorbid-
ity among substance use disorders are conflicting; some studies
assert that there are both substance-specific and common (i.e.,
non-substance-specific) familial influences (Bierut et al., 1998;
True et al., 1999; Tsuang et al., 1998), while others empha-
size the importance of substance-specific familial influences
influences (Kendler et al., 1997a; Cadoret et al., 1986; Cadoret
et al., 1995).
Animal studies also have provided somewhat inconsistent
evidence. For example, studies examining behavioral cross-
sensitization and cross-tolerance provide some evidence that
alcohol and illicit drugs act on overlapping neural mecha-
alcohol and nicotine (e.g., Fredriksson et al., 2000), cocaine
(e.g., Itzhak and Martin, 1999), and morphine (e.g., Nestby
et al., 1997), though it was not found with amphetamine
(e.g., Nestby et al., 1997). Further, Lessov and Phillips (2003)
showed that morphine- and cocaine-sensitized mice showed
cross-sensitization to alcohol, whereas alcohol-sensitized mice
did not show cross-sensitization to morphine or cocaine.
Evidence of cross-tolerance also has been reported between
alcohol and several other drugs, including nicotine (e.g.,
Schoedel and Tyndale, 2003), morphine (e.g., Fish et al., 2002),
and gamma-hydroxybutyric acid (GHB; e.g., Gessa et al.,
ences have two important differences. The first is that different
the prevalence of disorders in the relatives of different proband
ting, alternative hypotheses are evaluated by examining the fit
between observed data and expected data given each hypoth-
esized model, and the model yielding the best or most parsi-
monious fit between the observed and expected data is chosen.
Studies using these different methods have provided conflicting
results, with studies examining the comorbidity between alco-
hol and illicit drug problems using family prevalence analyses
finding evidence for substance-specific influences (Merikangas
et al., 1998; Meller et al., 1988) and studies using biometric
model fitting approaches finding evidence for common familial
influences (Kendler et al., 1997a; True et al., 1999).
Another possible reason for the conflicting results is a differ-
ence in the samples examined. Studies that supported common
familial influences on problems with different drugs examined
problems with alcohol and illicit drugs in the general popula-
tion (Kendler et al., 1997a; Cadoret et al., 1986, 1995), whereas
studies that found substance-specific familial influences gener-
ally examined problems with alcohol and illicit drugs in clinical
samples (Merikangas et al., 1998; Meller et al., 1988).
In the present study, we examined the etiology of comorbid-
ity between alcohol dependence and illicit drug dependence in
adolescents treated for severe antisocial behavior and substance
use disorders and a matched control sample. We used the Neale
and Kendler (1995) model fitting approach rather than family
prevalence analyses for two main reasons. First, evidence from
used to test the causes of comorbidity between two disorders
suggest that the Neale and Kendler model fitting approach does
a better job of discriminating the plausible comorbidity models
from the rejectable comorbidity models than family prevalence
analyses. A simulation study examining the validity of family
analyses validly discriminate the alternative forms model from
alternative hypotheses, none of the family prevalence analyses
testing the correlated risk factors model or the three indepen-
dent disorders model were valid (Rhee et al., 2003). In contrast,
the Neale and Kendler model fitting approach discriminated
five classes of models (i.e., the alternate forms model, the ran-
three independent disorders model, and the correlated liabili-
ties models) reliably (Rhee et al., 2004). Second, the Neale and
Kendler model fitting approach enables the examination of 12
other hypotheses explaining the causes of comorbidity in addi-
tion to the model hypothesizing common familial influences as
the cause of comorbidity between two disorders (i.e., the corre-
lated risk factors model). The present study is the first study to
1.1. The Neale and Kendler comorbidity models
All of the Neale and Kendler (1995) comorbidity models
are versions of the continuous liability threshold model (Carter,
1969, 1973) and assume that there is a continuous liability dis-
tribution of multiple genetic and environmental causes for a
disorder. This assumption is a reasonable one for substance use
genetic and environmental influences on substance use disor-
ders (Tsuang et al., 1996; van den Bree et al., 1998). Neale
and Kendler considered a wide range of hypotheses, includ-
ing models that differ in the primary versus secondary disorder,
the number of liability distributions underlying the two disor-
ders (i.e., one, two, or three), and the level at which the two
disorders are related (i.e., at the level of latent liabilities or at
the level of phenotypes). Detailed information regarding the
Neale and Kendler comorbidity models is provided in Neale
and Kendler (1995) and will not be repeated here. Below is
a short description of the 13 Neale and Kendler comorbidity
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
1.1.1. Chance model. The chance model hypothesizes that
comorbidity between two disorders occurs as a result of chance
1.1.2. Alternate forms model. The alternate forms model
hypothesizes that two disorders are comorbid because they are
alternate manifestations of a single liability.
1.1.3. Multiformity models. There are six multiformity mod-
els, which hypothesize that being affected by either disorder
increases the risk for having the other disorder. In the random
multiformity and the extreme multiformity model, individuals
with disorder A or B have an increased risk for having the other
disorder. In random multiformity, the increased risk for comor-
bidity is a probability of p or r, and in extreme multiformity,
individuals must cross a second, higher threshold in order to
have both disorders. In random multiformity of A and extreme
multiformity of A, only individuals with disorder A have the
and extreme multiformity of B, only individuals with disorder
B have the increased risk for comorbidity.
1.1.4. Three independent disorders model. The three indepen-
dent disorders model hypothesizes that comorbidity between
two disorders occurs because the comorbid disorder is a disor-
der that is separate from either disorder occurring alone.
1.1.5. Correlated liabilities models. There are four correlated
liabilities models. The correlated risk factors model hypothe-
sizes that two disorders are comorbid because the risk factors
for the two disorders are correlated. The reciprocal causation
model hypothesizes that comorbidity between two disorders
occurs because A and B cause each other in a feedback loop.
The A causes B model is a sub-model of the reciprocal cau-
sation model that hypothesizes that comorbidity between two
disorders occurs because A causes B. The B causes A model
is a sub-model of the reciprocal causation model that hypothe-
sizes that comorbidity between two disorders occurs because B
In the correlated liabilities models, the causal processes take
place at the liability. In the correlated risk factors model, the
risk factors for the two disorders are correlated, whereas in the
A causes B, B causes A, and reciprocal causation models, the
risk factors are united into a common latent phenotype prior to
causation. The correlated liabilities models are very different
from the multiformity models, which suggest that the increased
risk for the second disorder only occurs if an individual crosses
a threshold on the liability and is affected by a disorder.
Data from a clinical sample and a control sample were ana-
lyzed jointly. The clinical sample consisted of 272 probands
(7.0% female; age 13–20; mean age=16.32; standard devia-
tion of age=1.24) from an adolescent residential or day treat-
ment program for severe antisocial behavior and substance use
disorders and the probands’ 362 siblings (45.9% female; age
11–25; mean age=17.02; standard deviation of age=3.30).
All clinical probands were recruited from consecutive admis-
sions to an unlocked residential or day treatment facility in
the Denver metropolitan area, to which they had been referred
by social service and/or juvenile justice agencies for serious
antisocial behavior and substance use disorders. All probands
in the present study had a full sibling between age 11 and
25. The ethnicity distribution for the clinical probands is
American, 2.9% Native American, 0.4% Asian, and 0.4%
All clinical probands were referred to treatment for severe
substance use disorders and met the criteria for substance abuse
and/or substance dependence for one or more substances. Given
that the present study’s focus is substance dependence, clini-
cal probands who did not surpass the threshold for the pres-
ence of substance dependence and their siblings (30% of the
clinical sample) were excluded from the analyses. The final
sample size of the clinical sample was 191 probands and 253
The control sample consisted of 283 “probands” (8.5%
female; age 12–21; mean age=16.53; standard deviation of
age=1.50) who were matched to the clinical probands by sex,
age (±1 year), ethnicity, and zip code and their 422 siblings
of age=3.38). The control sample was recruited via telephone
queries by a marketing research company. Given that analyses
are restricted to sibling pairs where the sibling is less than 25
years old, there is not a one-to-one correspondence between
the control “probands” and the clinical probands in the current
sample. The ethnicity distribution for the control “probands” is
58.3% non-Hispanic Caucasian, 34.3% Hispanic, 6.0% African
American, and 1.4% Native American.
Written informed assent (from minor participants) or con-
sent (from adult participants or guardians of minor partici-
pants) was obtained after providing the participants with a
complete description of the study. All assent/consent forms and
research protocols were approved by the institutional review
board of the University of Colorado. Diagnostic and Statistical
Manual of Mental Disorders, 4th edition (DSM-IV; American
were assessed by in-person interviews using the Composite
International Diagnostic Interview-Substance Abuse Module
(CIDI-SAM). The CIDI-SAM is a valid and reliable structured
interview (Compton et al., 1996; Cottler et al., 1989; Crowley et
al., 2001; Zanis et al., 1995) that assesses symptoms and diag-
noses of abuse and dependence for alcohol and eight classes
of illicit drugs (i.e., marijuana, opioids, sedative/hypnotics,
inhalants, amphetamines, cocaine, hallucinogens, and phency-
clidine). The CIDI-SAM has been used to assess substance use
disorders in adolescents successfully (Thompson et al., 1996;
Whitmore et al., 1997).
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
Number of participants (%) with DSM-IV substance dependence diagnoses
Clinical sample Control sample
Proband SiblingProband Sibling
Any illicit drug
Total number of participants
We tested 13 alternative comorbidity models to examine
the causes of comorbidity between alcohol dependence and
illicit drug dependence (i.e., dependence on marijuana, opi-
lucinogens, or phencyclidine). Analyses examining dependence
number of participants with individual illicit drug dependence
diagnoses (see Table 1).
toms for all substances increases significantly with age during
adolescence and possible sex differences in the prevalence of
substance dependence (Young et al., 2002). Therefore, rather
than estimating a single threshold for the entire sample, we used
an extension of the Neale and Kendler model fitting approach
vary as a function of his or her age and sex. The formula for an
individual’s threshold is: a threshold intercept parameter+(the
a linear change in threshold with age (which is equivalent to
specifying a cumulative normal function for the probability of
the disorder as a function of age), and different thresholds for
males and females.
Neale and Kendler (1995) quantified the probabilities for the
16 combinations of affected or unaffected status for pairs of
and demonstrated a model-fitting approach to test each comor-
bidity model. In the Neale and Kendler model fitting approach,
the observed data are the number of pairs in each possible com-
both alcohol dependence and illicit drug dependence in siblings
1 and 2, illicit drug dependence only in sibling 1 and alcohol
dependence only in sibling 2, etc.). The observed data are then
compared to the numbers of pairs expected given the pairwise
probabilities for each possible combination of disease state for
the comorbidity model being tested.
tested by analyzing data from the clinical and control samples
in a joint analysis. The inclusion of a clinical sample increases
the power of the analyses while the inclusion of a control sam-
ple enables the estimation of population (unselected) thresholds
rather than fixing them to predetermined estimates. The ascer-
tainment in the clinical sample in the present study is single
ascertainment, where probands have one disorder or both dis-
orders being examined. There were no double-proband fami-
lies, consistent with the probability of being ascertained given
affected status (known as pi; Morton, 1982) being close to zero.
Probands who have neither disorder are not included in the clin-
ical sample; nor are their relatives. Data were analyzed using an
ascertainment correction demonstrated by Rijsdijk et al. (1999),
which corrects for the distortion of the expected frequencies
resulting from this omission. In the correction, each expected
proportion is divided by the sum of the expected proportions so
that the expected proportions sum to one.
Table 1 shows the number of participants with each of the
and the control sample. As expected, the number of participants
with no dependence diagnosis is lowest in the clinical probands
(clinical probands with no dependence diagnosis were removed
from the analyses), intermediate in the siblings of the clinical
is highest in the clinical probands, intermediate in the siblings
of the clinical probands, and lowest in the control sample.
Table 2 shows the model fitting results and the param-
eter estimates for models testing the causes of comorbidity
between alcohol dependence and illicit drug dependence. The
−2 log likelihood (−2lnL) evaluates the discrepancy between
the observed data and the data expected by the model. Mod-
els with lower −2lnL fit the data better, and models with lower
cate a more parsimonious fit, taking model complexity as well
as fit into account.
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
Model fitting results and parameter estimates (i.e., sibling correlation, thresholds, probability of being affected by the disorder, correlation between risk factors,
causal parameters, and ascertainment probability for alcohol dependence only) from Neale and Kendler comorbidity models
Model Model fit Parameters
d.f.AIC AlcoholBothIllicit drug
CH2700.12666 1368.120.24 2.62
AF 2388.59 668 1052.590.37 3.31
RM2383.29 6641055.290.34 4.03
RMA 2385.186651055.18 0.270.29 0.55 0.20
RMB 2392.45665 1062.450.000.29 3.07
EM 2379.66660 1059.660.290.00
EMA 2384.46663 1058.46 0.250.29 0.21
EMB2390.09 663 1064.09 0.00 0.26 3.17
TD2397.00 662 1073.000.28 .273.58
CR2382.78 6641054.78 0.28 3.58
ACB 2385.47665 1055.47 0.330.19 1.0 0.18
BCA 2386.13665 1056.130.120.32 1.0 0.17
RC 2382.70664 1054.700.260.31 0.530.430.17
Note. CH=chance; AF=alternate forms; RM=random multiformity; RMA=random multiformity of A; RMB=random multiformity of B; EM=extreme multi-
formity; EMA=extreme multiformity of A; EMB=extreme multiformity of B; TD=three independent disorders; CR=correlated risk factors; ACB=A causes B;
BCA=B causes A; RC=reciprocal causation; −2lnL =−2 log likelihood; d.f.=degrees of freedom; AIC=Akaike’s Information Criterion; rs=sibling correlation;
t1=intercept parameter for first threshold (the age change in threshold per year and the sex difference in threshold appear in the second and third line, respectively);
t2=intercept parameter for second threshold (the age change in threshold per year and the sex difference in threshold appear in the second and third line, respec-
tively); p=probability of having disorder A or B; r=probability of having disorder A or B; rAB=correlation between influences shared between disorder A and
B; rf=correlation between familial influences shared between disorder A and B; rnf=correlation between nonfamilial influences shared between disorder A and B;
kA=A causes B parameter; kB=B causes A parameter; a=ascertainment probability for alcohol dependence only given ascertainment probability for illicit drug
dependence only equals 1.
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
the lowest AIC) was the alternate forms model. Equal second
tors and reciprocal causation. The results of a simulation study
liabilities models successfully. However, the difference in the
AIC for the alternate forms model (1052.59) and the AICs for
the next best fitting models (1054.70 for the reciprocal causa-
tion model and 1054.78 for the correlated risk factors model) is
small, suggesting that both a model hypothesizing a single lia-
bility distribution for alcohol and drug dependence and a model
hypothesizing two separate liability distributions that are highly
correlated are viable models that cannot be rejected.
The age change in threshold per year (reported in the sec-
ond line under column t1 or t2) suggests that in general, the
age. The sex difference in threshold (reported in the third line
under column t1 or t2) suggests that in general, the prevalence
of alcohol and illicit drug dependence is lower in females than
The ascertainment parameter for alcohol dependence only
ranged from 0.04 to 0.22 when the ascertainment probability
for illicit drug dependence only was fixed to be equal to 1 (see
Table 2). This means that the probability of being treated in
this clinical sample and being ascertained as a participant in
this study is 4.5–25 times higher for children with illicit drug
dependence only than children with alcohol dependence only.
We examined 13 alternative models explaining the causes of
comorbidity in substance dependence in a clinical sample of
adolescents with severe antisocial behavior and substance use
disorders and a matched control sample. The present study is
the first study to test a wide range of hypotheses regarding the
causes of comorbidity between alcohol dependence and illicit
drug dependence. It is also the first study to examine this issue
in adolescents. The best fitting model explaining the comorbid-
the alternate forms model. After the alternate forms model, the
next best fitting models were two of the four correlated liabili-
ties model, the correlated risk factors and reciprocal causation
models, followed by two of the random multiformity models
(random multiformity and random multiformity of A).
Although we could not establish any one model as the single
correct explanation for the comorbidity between alcohol depen-
dence and illicit drug dependence (as the fit of the alternative
models was close), the best-fitting models strongly support the
close overlap between the etiology of alcohol dependence and
illicit drug dependence. The most likely hypotheses are that the
dence exists because dependence symptoms for alcohol and
illicit drugs are alternate manifestations of the same underlying
(i.e., correlated risk factors or reciprocal causation). Although
the alternate forms model and the correlated liabilities model
seem to suggest two different conceptual conclusions (i.e., a
single liability or two separate liabilities), the parameters from
the correlated risk factors model suggest that if there are two
separate liabilities, the correlation between them is high (i.e.,
0.77 for the familial risk factors and 0.78 for the non-familial
Intermediate in likelihood are the two causal models (i.e.,
dependence causes alcohol dependence), and two of the multi-
formity models (random multiformity and random multiformity
of alcohol dependence), which hypothesize that being affected
by either disorder increases the risk for having the other disor-
are the other multiformity models (random multiformity of B,
extreme multiformity, extreme multiformity of A, extreme mul-
tiformity of B), the three independent disorders model (which
hypothesizes a third, independent comorbid disorder), and the
ply as a result of chance).
Our results conflict with previous findings of significant
substance-specific familial influences (Bierut et al., 1998;
Merikangas et al., 1998; True et al., 1999; Tsuang et al., 1998;
Meller et al., 1988). One source of discrepancy is the differ-
ence in analytical methods (i.e., family prevalence analyses or
biometric model fitting). However, the discrepancy between the
present study’s results and previous studies’ support for signifi-
cant substance-specific familial influences cannot be explained
ies using biometric model fitting also concluded that there are
significant substance-specific familial influences (True et al.,
1999; Tsuang et al., 1998). Second, several studies (Merikangas
given the finding that having a family member with disorder A-
only (e.g., alcohol dependence only) does not increase the risk
for disorder B-only (e.g., drug dependence only); simulation
results (Rhee et al., 2003) have shown that the alternate forms
model cannot be the correct model for such results, sampling
There are several additional reasons for the discrepancy
between the present study’s results and previous findings. First,
there are notable differences in the ascertainment of the sample.
problems than participants in the previous studies, who were
recruited from the general population (True et al., 1999; Tsuang
et al., 1998), outpatient treatment centers (Merikangas et al.,
1998), or either inpatient or outpatient treatment centers (Bierut
et al., 1998). Second, the number of female probands and the
number of probands with only a diagnosis of alcohol depen-
dence were small in our sample. Most significantly, probands
in our sample were referred to treatment by social service and
juvenile justice agencies for both antisocial behavior and sub-
had not been referred for treatment of antisocial behavior. In
the present study, liability common to the specific substance
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
dependence diagnoses may reflect liability for antisocial behav-
ior. Therefore, it is possible that probands in our sample may
have a subtype of alcohol or drug dependence representing a
single liability, and that our results may not generalize to sub-
stance use disorders existing without the presence of antisocial
behavior. An important future direction is to test the Neale and
Kendler models in a representative sample; simulation results
from a study examining the validity of the Neale and Kendler
model-fitting approach suggests that such a study will need to
employ a very large sample (Rhee et al., 2004).
cents, whereas previous studies examined substance use disor-
ders in adults. The causes of comorbidity among substance use
disorders may differ in adults and adolescents. The availability
of different substances, the legal consequences of using differ-
ent substances, and the degree of experimentation with different
substances could be very different in adults and adolescents.
For example, the examination of comorbidity between alcohol
dependence and illicit drug dependence is the examination of
two illegal drugs in United States adolescents, but it is an exam-
ination of one legal drug and one illegal drug in United States
adults. Dick et al. (2001) found that the magnitude of genetic
than in rural areas, and they suggested that differences in the
availability of alcohol in the two environments may be one of
two environments. Therefore, it is possible that dependence on
two drugs similar in availability (e.g., alcohol and illicit drugs
in adolescence) is a manifestation of the same underlying lia-
bility while dependence on two drugs dissimilar in availability
(e.g., alcohol and illicit drugs in adulthood) reflects separate
In conclusion, this first examination of a wide range of
hypotheses for the causes of comorbidity among substance use
disorders in adolescents suggests that symptoms of alcohol
dependence and illicit drug dependence are manifestations of
the same underlying liability or two very highly correlated lia-
bilities. These results suggest that the search for specific genetic
influences on substance dependence risk in adolescents should
include the search for QTL or candidate genes influencing the
common vulnerability underlying different substances of abuse.
found significant substance-specific familial influences on sub-
stance use disorders. This discrepancy suggests the possibility
of heterogeneity in the causes of comorbidity between alcohol
clusions may be specific to antisocial substance dependence in
This work was supported by National Institute on Drug
Abuse grants DA-05131, DA-11015, DA-18673, and DA-
13956. Michael C. Neale was supported by NIMH grants MH-
01458 and MH-65322. An earlier version of this paper was
presented at the meeting of the Behavior Genetics Association
on June 26, 2003 in Chicago, IL.
American Psychiatric Association, 1996. Diagnostic and Statistical Manual of
Mental Disorders, 4th ed. American Psychiatric Association, Washington,
Bierut, L.J., Dinwiddie, S.H., Begleiter, H., Crowe, R.R., Hesselbrock, V.,
Nurnberger, J.I., Porjesz, B., Schuckit, M.A., Reich, T., 1998. Familial
transmission of substance dependence: alcohol, marijuana, cocaine, and
habitual smoking. Arch. Gen. Psychiatry 55, 982–988.
Cadoret, R., Troughton, E., O’Gorman, T.W., Heywood, E., 1986. An adop-
tion study of genetic and environmental factors in drug abuse. Arch. Gen.
Psychiatry 43, 1131–1136.
Cadoret, R.J., Yates, W.R., Troughton, E., Woodworth, G., Stewart, M.A.,
1995. Adoption study demonstrating two genetic pathways to drug abuse.
Arch. Gen. Psychiatry 52, 42–52.
Carter, C.O., 1969. Genetics of common disorders. Br. Med. Bull. 25, 52–57.
Carter, C.O., 1973. Multifactorial genetic disease. In: McKusick, V.A., Clair-
borne, R. (Eds.), Medical Genetics. H.P. Publishing, New York, pp.
Compton, W.M., Cottler, L.B., Dorsey, K.B., Spitznagel, E.L., Mager, D.E.,
1996. Comparing assessments of DSM-IV substance dependence disor-
ders using CIDI-SAM and SCAN. Drug Alcohol Depend. 41, 179–187.
Cottler, L.B., Robins, L.N., Helzer, J.E., 1989. The reliability of the CIDI-
SAM: a comprehensive substance abuse interview. Br. J. Addict. 84,
Crowley, T.J., Mikulich, S.K., Ehlers, K.M., Whitmore, E.A., MacDonald,
M.J., 2001. Validity of structured clinical evaluations in adolescents with
conduct and substance problems. J. Am. Acad. Child Adoles. Psychiatry
Dick, D.M., Rose, R.J., Viken, R.J., Kaprio, J., Koskenvuo, M., 2001. Explor-
ing gene–environment interactions: socioregional moderation of alcohol
use. J. Abnorm. Psychol. 110, 625–632.
Fish, E.W., DeBold, J.F., Miczek, K.A., 2002. Repeated alcohol: behavioral
sensitization and alcohol-heightened aggression in mice. Psychopharma-
cology 160, 39–48.
Fredriksson, A., Eriksson, P., Ankarberg, E., Palomo, T., Archer, T., 2000.
Neonatal nicotine administration influences ethanol-induced behaviors.
Alcohol 21, 107–115.
Gessa, G.L., Agabio, R., Mauro, A.M.C., Lobina, C., Pani, M., Reali, R.,
Colombo, G., 2000. Mechanism of the antialcohol effect of gamma-
hydroxybutyric acid. Alcohol 20, 271–276.
Grove, W.M., Eckert, E.D., Heston, L., Bouchard, T.J., Segal, N., Lykken,
D.T., 1990. Heritability of substance abuse and antisocial behavior: a
study of monozygotic twins reared apart. Biol. Psychiatry 27, 1293–
Gynther, L.M., Carey, G., Gottesman, I.I., Vogler, G.P., 1995. A twin study
of non-alcohol substance abuse. Psychiatry Res. 56, 213–220.
Itzhak, Y., Martin, J.L., 1999. Effects of cocaine, nicotine, dizocipline and
alcohol on mice locomotor activity: cocaine-alcohol cross-sensitization
involves upregulation of striatal dopamine transporter binding sites. Brain
Res. 818, 204–211.
Johnson, E.O., van den Bree, M.B.M., Uhl, G.R., Pickens, R.W., 1996. Indica-
tors of genetic and environmental influences in drug abusing individuals.
Drug Alcohol Depend. 41, 17–23.
Kaprio, J., Koskenvuo, M., Langinvainio, H., Romanov, K., Sarna, S., Rose,
R.J., 1987. Genetic influences on use and abuse of alcohol: a study
of 5638 adult Finnish twin brothers. Alcohol Clin. Exp. Res. 11, 349–
Kendler, K.S., Davis, C.G., Kessler, R.C., 1997a. The familial aggregation of
common psychiatric and substance use disorders in the National Comor-
bidity Survey: a family history study. Br. J. Psychiatry 170, 541–548.
Kendler, K.S., Prescott, C.A., Neale, M.C., Pedersen, N.L., 1997b. Temper-
ance board registration for alcohol abuse in a national sample of Swedish
male twins, born 1902 to 1949. Arch. Gen. Psychiatry 54, 178–184.
Kessler, R.C., Crum, R.M., Warner, L.A., Nelson, C.B., Schulenberg, J.,
Anthony, J.C., 1997. Lifetime co-occurrence of DSM-III-R alcohol abuse
and dependence with other psychiatric disorders in the National Comor-
bidity Survey. Arch. Gen. Psychiatry 54, 313–321.
92 Download full-text
S.H. Rhee et al. / Drug and Alcohol Dependence 84 (2006) 85–92
Lessov, C.N., Phillips, T.J., 2003. Cross-sensitization between the locomotor
stimulant effects of ethanol and those of morphine and cocaine in mice.
Alcohol Clin. Exp. Res. 27, 616–627.
McGue, M., Pickens, R.W., Svikis, D.S., 1992. Sex and age effects on the
inheritance of alcohol problems: a twin study. J. Abnorm. Psychol. 101,
Meller, W.H., Rinehart, R., Cadoret, R.J., Troughton, E., 1988. Specific famil-
ial transmission in substance abuse. Int. J. Addict. 23, 1029–1039.
Merikangas, K.R., Stolar, M., Stevens, D.E., Goulet, J., Preisig, M.A., Fenton,
B., Zhang, H., O’Malley, S.S., Rounsaville, B.J., 1998. Familial transmis-
sion of substance use disorders. Arch. Gen. Psychiatry 55, 973–979.
Morton, N.E., 1982. Outline of Genetic Epidemiology. Karger, Basel.
Neale, M.C., Kendler, K.S., 1995. Models of comorbidity for multifactorial
disorders. Am. J. Hum. Genet. 57, 935–953.
Nestby, P., Vanderschuren, L.J., De Vries, T.J., Hogenboom, F., Wardeh, G.,
Mulder, A.H., Schoffelmeer, A.N., 1997. Ethanol, like psychostimulants
and morphine, causes long-lasting hyperreactivity of dopamine and acetyl-
choline neurons of rat nucleus accumbens: possible role in behavioural
sensitization. Psychopharmacology 133, 69–76.
Pickens, R.W., Svikis, D.S., McGue, M., LaBuda, M.C., 1995. Common
genetic mechanisms in alcohol, drug, and mental disorder comorbidity.
Drug Alcohol Depend. 39, 129–138.
Rhee, S.H., Hewitt, J.K., Corley, R.P., Stallings, M.C., 2003. The validity
of analyses testing the etiology of comorbidity between two disorders: a
review of family studies. J. Child Psychol. Psychiatry 44, 612–636.
Rhee, S.H., Hewitt, J.K., Lessem, J.M., Stallings, M.C., Corley, R.P., Neale,
M.C., 2004. The validity of the Neale and Kendler model fitting approach
in examining the etiology of comorbidity. Behav. Genet. 34, 251–265.
Rijsdijk, F.V., Cardno, A.G., Sham, P.C., McGuffin, P., 1999. Statistical
aspects of genetic models of comorbidity of psychotic disorders in
selected twin samples. In: Poster Presented at the Annual Meeting of
the International Society of Psychiatric Genetics, Monterey, CA.
Schoedel, K.A., Tyndale, R.F., 2003. Induction of nicotine-metabolizing
CYP2B1 by ethanol and ethanol-metabolizing CYP2E1 by nicotine: sum-
mary and implications. Biochim. Biophys. Acta 1619, 283–390.
Thompson, L.L., Riggs, P.D., Mikulich, S.K., Crowley, T.J., 1996. Contri-
bution of ADHD symptoms to substance problems and delinquency in
conduct-disordered adolescents. J. Abnorm. Child Psychol. 24, 325–347.
True, W.R., Heath, A.C., Scherrer, J.F., Xian, H., Lin, N., Eisen, S.A., Lyons,
M.J., Goldberg, J., Tsuang, M.T., 1999. Interrelationship of genetic and
environmental influences on conduct disorder and alcohol and marijuana
dependence symptoms. Am. J. Med. Genet. 88, 391–397.
Tsuang, M.T., Lyons, M.J., Eisen, S.A., Goldberg, J., True, W., Lin, N.,
Meyer, J.M., Toomey, R., Faraone, S.V., Eaves, L., 1996. Genetic influ-
ences on DSM-III-R drug abuse and dependence: a study of 3372 twin
pairs. Am. J. Med. Genet. 67, 473–477.
Tsuang, M.T., Lyons, M.J., Meyer, J.M., Doyle, T., Eisen, S.A., Goldberg,
J., True, W., Lin, N., Toomey, R., Eaves, L., 1998. Co-occurrence of
abuse of different drugs in men: the role of drug-specific and shared
vulnerabilities. Arch. Gen. Psychiatry 55, 967–972.
van den Bree, M.B.M., Johnson, E.O., Neale, M.C., Pickens, R.W., 1998.
Genetic and environmental influences on drug use and abuse/dependence
in male and female twins. Drug Alcohol Depend. 52, 231–241.
Whitmore, E.A., Mikulich, S.K., Thompson, L.L., Riggs, P.D., Aarons, G.A.,
Crowley, T.J., 1997. Influences on adolescent substance dependence: con-
duct disorder, depression, attention deficit hyperactivity disorder, and
gender. Drug Alcohol Depend. 47, 87–97.
Young, S.E., Corley, R.P., Stallings, M.C., Rhee, S.H., Crowley, T.J., Hewitt,
J.K., 2002. Substance use, abuse and dependence in adolescence: preva-
lence, symptom profiles and correlates. Drug Alcohol Depend. 68,
Zanis, D.A., McLellan, A.T., Cacciola, J., Vrublevski, A., 1995. DSM-III-
R alcohol dependence criteria in Russian and American men. J. Subst.
Abuse 7, 253–261.