ENVIRONMENTAL-FACTORS CAN CONFOUND IDENTIFICATION OF A MAJOR GENE EFFECT - RESULTS FROM A SEGREGATION ANALYSIS OF A SIMULATED POPULATION OFLUNG-CANCER FAMILIES

Proper control of environmental factors can be crucial to the identifi cation of genes that influence susceptibility to a complex trait, espe cially for a trait such as lung cancer, for which the environmental fa ctor (smoking) accounts for a significant etiologic fraction of the di sease. An earlier segregation analysis of 337 Louisiana families, whic h incorporated direct measure of tobacco consumption, provided evidenc e for autosomal codominant inheritance of a major gene that influenced age at onset of lung cancer. Subsequent analyses were performed in wh ich the families were stratified into two subsets based on birth cohor t of the proband; results suggested the presence of heterogeneity that were postulated to reflect the influence of cohort trends in tobacco consumption. To evaluate this hypothesis further, we simulated a popul ation of three-generation pedigrees in which an autosomal dominant mod e of susceptibility to lung cancer was transmitted, but tobacco use va ried across generations corresponding to published trends in smoking. A total of 200,000 individuals in families of various sizes, ages, and cigarette smoking habits were simulated from 1900 to 1980. From this population, 324 families (2,405 individuals) with 380 cases of lung ca ncer were ascertained through 328 lung cancer probands. Complex segreg ation analysis was performed using the REGTL program of S.A.G.E. in wh ich pack-years of tobacco exposure were incorporated directly into the likelihood calculations. Although the no major gene, environmental, a nd Mendelian recessive hypotheses were rejected, both dominant and cod ominant transmission provided a good fit to the data. Thus in a popula tion of simulated families with autosomal dominant susceptibility to l ung cancer, intergenerational differences in tobacco consumption led t o the detection of autosomal codominant transmission as an acceptable hypothesis. These results underscore the potential danger of segregati on analysis of complex traits in which exposure to known environmental influences may differ across generations. (C) 1998 Wiley-Liss, Inc.