158 Tobacco Dependence
by gene-environment interactions (Khoury, Beaty, & Cohen,
1993). While no one has yet demonstrated the presence of
speci“c gene-environment interactions on tobacco use in hu-
mans, because of insuf“cient study sample sizes, interactions
have been demonstrated in biometric models of twin similar-
ity for alcohol use (Koopmans, Heath, Neale, & Boomsma,
1997). It has been speculated that if such interactions exist,
they could emerge as potentially powerful determinants of
susceptibility and maintenance of tobacco dependence
(Kendler, 1999; Swan, 1999).
Definition of Phenotypes
In most behavior genetic and genetic epidemiologic studies,
•smokingŽ has been assessed as a static phenotype (i.e., as
if the behavior is a trait that remains constant over time).
(See next section for a review of these studies.) However, a
variety of studies from the developmental, epidemiologic,
psychiatric, and smoking literature suggest that smoking
in general, and the consumption of nicotine on a regular
basis speci“cally, is tremendously more complex than a
simple trait perspective (Swan, 1999b). Not only do reasons
and motivations for smoking vary across individuals, it is
likely that motivations (biological, social, and psychologi-
cal individually and in combination with each other) vary
within an individual across time and situations (Hiatt &
Rimer, 1999; Petraitis, Flay, & Miller, 1995; USDHHS,
1994).
In the “eld of psychiatric genetics, an area fraught with
numerous examples of nonreplication (Kendler, 1999), some
investigators believe that more detailed measures of pheno-
types, relying on actual measurements of behavior, physio-
logical responses, or biological characteristics such as brain
structure from imaging studies, will provide more replicable
associations with genetic markers than have more general
summary measures (Gelernter, 1997; Kendler, 1999). In the
“eld of tobacco use, numerous possibilities exist in which the
relationships of phenotypes to genetic factors may actually
be larger should the full range of phenotypes be explored.
We have developed a classi“cation of possible phenotypes
(referred to as endophenotypes by Kendler, 1999) to organize
phenotype selection for genetic investigations of tobacco use
(see Table 7.2); we make use of this classi“cation in our re-
view of existing genetic studies that follows.
Evidence for a Genetic Basis to a Variety of
Smoking-Related Phenotypes
Extensive literature studying twins reared together and apart
supports the conclusion that genetic in”uences underlie the
initiation and lifetime use of tobacco as well as several indi-
rect measures of tobacco dependence (including amount
smoked and persistence; for reviews, see Heath & Madden,
1995; Sullivan & Kendler, 1999; Swan, 1999a; Swan &
Carmelli, 1997). Pooled analyzes of the results from these
studies lead to the conclusion that 56% of smoking initiation
is attributable to genetic factors (44% to environmental
sources), while 67% of variance in indirect measures of
tobacco dependence can be attributed to genetic factors (33%
to environmental sources; Sullivan & Kendler, 1999). The
vast majority of the studies conducted thus far have examined
genetic in”uences on class I phenotypes.
Several studies have discovered that different genetic and
environmental in”uences play a role at dif ferent stages in the
development of smoking. Heath and Martin (1993) found
that the best-“tting genetic model was one that incorporated
separate but correlated genetic sources of variation for each
phase of the natural history of smoking. At least threeTABLE 7.2 Proposed Tobacco and Nicotine Phenotype
Classification System
Phenotype Characteristics
Class I Relatively crude, broad summary or cross-sectional
measures of smoking behavior, such as ever/never-smoker,
current/former/never-smoker, age at smoking initiation, and
average number of cigarettes smoked per day. The majority
of genetic studies, to date, have involved only this level of
phenotypic description.
Class II Measures or indicators of tobacco dependence, such as the
Fagerström Tolerance Questionnaire and its more recent
variant, the Fagerström Test for Nicotine Dependence,
time to “rst cigarette in the morning, and number of quit
attempts, or psychiatric-type classi“cations based on the
DSM IVsystem.
Class III Longitudinal assessments that emphasize the process of,or
progression towardthe development of regular smoking or
tobacco dependence. Although class III phenotypes can
be •summarizedŽ to create a class I or II phenotype, class III
is differentiated by the nature of the data, whereby class III
retains its longitudinal properties and is presented as a
trajectory. As an illustration, two individuals both might be
classi“ed as ever -smokers, having smoked 100 or more
cigarettes in their life, yet their smoking topography may
differ dramatically„for example, one person might have
taken years to develop into a daily smoker, whereas the
other might have converged rapidly on daily smoking (and
thus have a much steeper •slopeŽ or •trajectoryŽ for the
development of daily smoking).
Class IV Nicotine pharmacokinetic parameters (the effects of the
body on the drug, i.e., nicotine absorption, distribution,
metabolism, and excretion rates; and pharmaco-dynamic
effects (the effects of the drug on the body, including
physiologic responses such as mood alteration, heightened
concentration, and changes in receptor function and
density).