Somatic mutation at the glycophorin A locus in human newborns: Statistical considerations and the effects of in utero exposures.

摘要

Epidemiological data suggest that somatic cell genetic alterations occurring during embryonic/fetal development have important disease implications. This study is focused on two aspects of the relationship between embryonic/fetal somatic mutation at the glycophorin A (GPA) locus and maternal exposures/lifestyle factors during pregnancy. First, an examination of several statistical analysis methods of GPA somatic mutation data was used to determine which approach yields the most accurate and reliable results. Second, application of the most accurate and robust statistical method was used to determine the contribution of maternal exposures/lifestyle factors to in utero somatic mutation measured in human newborns.;The GPA assay uses immunolabelling and flow cytometry to enumerate two types of rare variant erythrocytes occurring in individuals heterozygous at the GPA locus, &Empty; /N allele-loss variants and N/N allele-loss with duplication variants. The assay is sensitive to a wide range of mutational events and has been used in response to a variety of exposures.;A comparison of statistical analysis approaches found that analysis methods for overdispersed count data, i.e. negative binomial regression, yielded the most reliable and accurate results for GPA assay variant cell frequency (Vf) data.;Four maternal exposures, found to be independently associated with newborn GPA Vf, were combined into a single measure of third trimester risk that included active tobacco smoking, and consumption of alcohol, charbroiled meats and caffeine. Maternal subjects were placed into risk groups based upon the number of exposures they reported having, with those in the lowest risk group reporting none of the exposures, those in the medium risk group reporting any one or two of the exposures, and those in the high risk group reporting any three or four of the exposures. Third trimester risk group, maternal age and race were used in a multivariable negative binomial regression model to determine the relative risk (RR) associated with each factor. Maternal age (RR = 0.97, p = 0.008) and third trimester risk group (medium risk: RR = 1.87, p 0.001; high risk: RR = 2.80, p 0.001) were associated with newborn &Empty; /N Vf. Maternal age (RR = 0.97, p = 0.008), race (RR = 0.50, p 0.001), and third trimester risk group (high risk: RR = 3.81, p 0.001) were associated with newborn N/N Vf .