Assessing Maternal Genetic Associations

Publisher Website
Google Scholar
Abstract
In utero exposures, including maternal phenotypes, are potential risk factors for both early-onset and adult-onset diseases. Two alternative study designs use maternal genotypes at polymorphic loci as biomarkers of an offspring's in utero exposure: (1) a traditional case–control study with logistic regression analysis, in which cases, controls, and mothers of both types of subjects are genotyped; and (2) a case–parent triad study with log-linear analysis, in which cases and both parents are genotyped. We used computer simulations to compare the operating characteristics of the log-linear approach to case-parent triad data and the case–control approach for assessing relative risks (RRs) associated with maternal genotypes. For high-risk allele frequencies (chromosomal prevalence; f) between 0.20 and 0.75, both methods allowed for valid, unbiased estimates of maternal RRs. The case–parent triad approach, however, had 43% greater power, on average, than the case–control approach with an equal number of genotypes, and 13% greater power with an equal number of cases. For example, under dominant inheritance, to detect 2-fold maternal RRs with 200 (or 150) cases when allele prevalence is between 0.15 and 0.40, the case-parent triad and equal-genotype case-control designs had, on average, 87% and 62% power, respectively. As f approached 0 or 1, the power of both methods decreased sharply. The greater efficiency of case–parent triads may be due to the inclusion of paternal genotype information, which allows for independent tests of disease association with maternal or offspring genotypes. These results highlight one potential advantage of case–parent triad data in assessing maternal genetics as risk factors for offspring disease. We discuss these findings and other considerations between the 2 methodological approaches.