Genetic and environment in asthma: the answer of twin studies

Abstract

et al. [1] presentthe results of a large Danish population-based twin studyon asthma. In this study, the heritability in liability to asth-ma, i.e. the proportion of variance due to genetic factors,is estimated to be 0.77 for males and 0.68 for females. Inother studies, the heritability of asthma is estimated to bebetween 0.36 and 0.75 [2–4] (table 1). Thus, the results ofthis Danish study are consistent with those of other twinstudies and add to the body of evidence indicating that thegeneticcontribution toasthma isconsiderable.Inaddition,the results of this study suggest individual specific, un-shared, environmental factors to be important as well. Inthis editorial, the assumptions and methods of twin stu-dies will be assessed, and the role of genetic and envir-onmental factors in asthma reviewed.Genetic studies using the twin-design have four majorassumptions [8]: 1) monozygous (MZ) and dizygous (DZ)twins are samples of the same gene pool; 2) twins are rep-resentative of the general population; 3) self-reported zy-gosity is correct in questionnaire-based studies; and 4) theenvironment for both MZ and DZ twins is similar.The first and second assumptions are valid, providedthat representative or complete samples are taken from thepopulation. The second assumption, the representativity,may not be totally valid because MZ and DZ twins differfrom each other and from singletons with respect to theirintrauterine environment [9]. The shared intrauterine envi-ronment may have an adverse effect on the growth andorgan maturation of the foetus. However, this most likelydoes not influence the development of asthma since theprevalence of asthma is comparable in twins and single-tons [1, 2, 4]. The third assumption has been tested [10,11]. In general, self-reported zygosity questions are ade-quate in 95–98% of cases. Finally, the fourth assumptionof an equal environment may not be valid in the case of as-thma. For instance, it has been shown that MZ twins havemore similar smoking patterns than DZ twins [12]. It isunknown whether or not this higher similarity in MZ twinsis also the case for other environmental factors, such as ex-posure to indoor allergens and viruses. A higher similarityin environment for MZ twins compared to DZ twins maylead to an overestimation of the heritability of asthma.The method for diagnosing asthma is a self-reportedquestionnaire in most large-scale twin studies. Subjects inthese studies are not tested clinically. This method maylead to an under- or overestimation of asthma prevalence[13, 14]. Overestimation could occur, for instance, ifasthma is diagnosed by questions on wheeze. Small chil-dren in particular may wheeze during the course of a res-piratory infection but not have asthma [15]. Therefore,studies on the genetics of asthma are currently directed atmeasurable clinical components of asthma, e.g. airwayhyperresponsiveness (AHR), reversibility and variabilityof airway obstruction. If it were known which of thesecomponents of asthma have a high genetic contribution,these components could then be selected for genetic stud-ies to find genes that regulate these components. As an ex-ample, one twin study of reasonable size on AHR hasbeen published, in which the heritability of AHR to meth-acholine was 0.66 [16]. Clearly, more studies are needed.In general, statistical analyses of twin studies are com-plicated. In the study of twins, phenotypic similarities anddifferences are compared between MZ and DZ twins. MZtwins share 100% of their genetic information and DZtwins share on average 50%. If a trait is influenced bygenetic factors, MZ twins should resemble each other to agreater extent than DZ twins, and the correlations betweenMZ and DZ twin pairs may be used to estimate the rela-tive size of genetic and environmental influences.In biometric modelling, one goes a step further. Sincethese biometric analyses are not frequently presented inpulmonary journals, some assumptions and methods willbe addressed before discussing the results of these analy-ses in the study of asthma. In biometric modelling, aquantitative genetic analysis can be performed with di-chotomous variables (e.g. asthma/not asthma). To permitthese analyses, the first assumption is that disease status isdetermined by an unobserved continuous variable calledthe liability. If the liability falls above a threshold individ-uals are classified as affected. The second assumption isthat the distribution of the liability is normal [17]. Thevariance of the distribution of the liability is composed ofmultiple environmental and genetic influences. The envi-ronmental component can be dissected in influences sharedby both twins and influences not shared. Furthermore, thegenetic component consists of an effect of individual alle-les on the trait (additive effect) or interaction between al-leles at the same locus (dominance effect) [8]. The lastpossible source of genetic effects, i.e. interaction of allelesat different loci (epistasis) cannot be discriminated fromdominant genetic effects in twin studies, which is a lim-itation of this design. Thus, the observed phenotypes P1 oftwin 1 and P2 of twin 2 of a twin pair, will be linear func-tions of the underlying additive genetic influences (A-twin