Get access

Gene by environment interaction and asthma


  • S. T. Weiss

    1. Harvard Medical School and Respiratory and Environmental Epidemiology, Channing Laboratory, Brigham and Women's Hospital, Boston, Massachusetts, USA
    Search for more papers by this author

Professor S. T. Weiss Respiratory and Environmental Epidemiology, Channing Laboratory, 181 Longwood Avenue, Boston, MA 02115, USA.


The term interaction has two specific meanings in epidemiology. The first definition is purely a statistical one. Interaction is defined as the coefficient of the product term of two or more risk factors. An alternative definition is a more biologically oriented one in which interaction between two risk factors is defined as their coparticipation in the same causal mechanism of disease development. There are at least five different types of biological gene by environment interaction. It is important to recognize that clinical or epidemiological studies alone are unable to distinguish between these five different types of biological interaction. Asthma is a prototypical complex trait in which gene by environment interaction is important in disease causation. Environmental factors important in asthma include cigarette smoking (active/passive), allergen exposure, viral and bacterial respiratory illness, occupation, and possibly diet. There are two primary study designs that have been utilized for the assessment of gene by environment interaction: a traditional case control study or a nested case control study or a case control study nested within the larger longitudinal cohort. There are three important methodological issues in the assessment of gene by environment interaction. The first is the issue of misspecification of either genotype or environmental exposure or, for that matter, disease status. The second is possible confounding, and the third is the confounding of interaction with dose response. The four most critical factors for determining sample size for gene by environment interaction studies are the odds ratio for the interaction, the prevalence for exposure in the population, the prevalence of the genotype in the population, and the case to control ratio. In general, sample sizes have been too small to detect meaningful gene by environment interactions.