The Faustian bargain of genetic association studies: Bigger might not be better, or at least it might not be good enough

A problematic hypothesis … 

Understanding the pathogenesis of human complex diseases, such as allergic inflammation, will require a multidisciplinary approach integrating the expertise of clinicians (who have the critical task of phenotyping study participants according to accurate and standardized criteria), biologists (who need to characterize the pathophysiology of the disease in terms of underlying mechanisms and genetic pathways), geneticists (who need to design studies of adequate power and homogenous ethnicity and must choose target genotypes that capture meaningful variation across a locus), and functional genomicists (who sift through the dozens of polymorphisms currently in the databases to identify functional genetic variants). Working hypotheses formulated out of a similar interactive framework are bound for trouble.

The TH1/TH2 genetic paradigm tested by Maier et al,6 although strongly supported in mouse strains exhibiting polarized patterns of TH cell differentiation and TH cell–dependent phenotypes, demands more nuances in human subjects. Human immunologists know from everyday experience that the contraposition between TH1 and TH2 responses is an oversimplification because human T-cell polarization can be occasionally pronounced but is never complete. Therefore the expectation that genes involved in susceptibility to TH1 inflammation might protect against TH2-dependent high IgE levels or vice versa might have been unrealistic. In fact, the inability of the British study to find genetic evidence in support of a rigid TH1/TH2 dichotomy will provide one more argument in favor of a less mechanic and more articulate view of human immunologic diseases.

… And a large, possibly too large, study population 

According to the more dynamic paradigm that is currently replacing the TH1/TH2 dichotomy, both arms of the T-cell response are modulated and balanced by regulatory populations of T cells18, 19 and, as more recently suggested, even B cells. Of note, the function of these regulatory populations is strongly influenced by environmental factors acting through the innate immune interface.20 In fact, the most updated and functionally oriented version of the hygiene hypothesis attributes the current, rapid, and steep increase in the incidence of both TH1- and TH2-like diseases to the elimination of environmental, pathogen-derived signals that drove the evolution of our immune system and maintained an active regulatory interface.21, 22

The effects of natural variation at this regulatory interface are often tricky and always complex because they are inevitably intertwined with those of the relevant environmental exposures. In polygenic diseases, such as asthma and allergy, the effect of variation on patterns of gene expression is difficult to decipher unless the appropriate exposures are taken into consideration, but what these exposures are is not always clear. A good example is provided by the conflicting results obtained for CD14/−260C>T (also known as CD14/−159), a functional promoter polymorphism.23 The ability of CD14/−260C>T to influence IgE levels and asthma in opposite ways only became evident once the relevant environmental exposures (cat/dog versus stable animals in a European population24 and low versus high endotoxin in subjects of African descent25) were taken into consideration. The striking conclusion reached independently by both these studies was that the same polymorphism can be associated with either disease or protection, depending on the environment to which the subjects are exposed. A genome screen for asthma and bronchial hyperresponsiveness also recently published in the Journal provides another case in point. In 200 Dutch families ascertained through a parent with asthma, strong evidence for linkage was obtained for chromosome 5q but only for the families in which the children were exposed to passive smoking.26 Interestingly, figuring out what could count as a biologically meaningful environment might require some imagination. For instance, recent work strongly suggests that sex could be considered as an environmental factor that can lead to different effects of the same variation in men and women.27, 28 It is not unreasonable to surmise that as-yet undetected gene-environment interactions might confound the picture more often than expected, contributing to the lamented problems of replication frequently encountered by genetic association studies.

Because twin studies on the heritability of total IgE levels have clearly shown 40% of the variance in the phenotype is due to environmental factors,9 it is possible, nay likely, that the current results obtained in the British 1958 Birth Cohort have been affected by the lack of an environmental component to the analysis. Paradoxically, the very size of the study population might have carried along differences in lifestyle, socioeconomic status, and exposure among study participants recruited from different parts of the United Kingdom. Because the association between genetic variations, environment, and polygenic diseases is not always linear and unidirectional,27, 29, 30 lumping together groups of individuals faced with different environmental pressures is likely to drastically dilute the recognizable role of genetic determinants to the point of erasing them. Thus the inclusion of thousands of subjects might have resulted in a Faustian bargain in which sample size failed to compensate for heterogeneity of environment, and the lack of environmental data might have made it impossible to tease out distinct environmental effects on specific genotypes.

If this is true, one might be tempted to conclude that the genetic epidemiology of complex lung diseases is caught in a vicious circle. Genetic dysregulation in allergy and asthma hits multiple steps in key pathways, with small effects. The smaller the effects, the larger a study population needs to be for those effects to become appreciable, but the larger the population, the more difficult it is to control for environmental factors. A solution is to adjust study design to our deepened understanding of the complexity of the interactions underlying disease pathogenesis. Indeed, experience tells us that although large populations recruited without paying attention to environmental determinants might not be terribly informative, targeted recruitment efforts collecting a wealth of relevant environmental data might provide exciting breakthroughs, even when more limited in size.25, 31, 32, 33, 34 In the end, hypothesis-driven genetic epidemiology might be a more effective and more interesting partner for disease-oriented biologic research. Furthermore, dissecting the effect of the environmental context might foster preventive and therapeutic strategies predicated on innocuous surrogates or antagonists of the exposures found to be associated with the phenotypes of interest in susceptible individuals.