Mixture models of serum iron measures in population screening for hemochromatosis and iron overload

Abstract

Homozygosity for the C282Y mutation of the hemochromatosis gene on chromosome 6p (HFE) is a common genetic trait that increases susceptibility to iron overload. The authors describe and apply methodology developed for the analysis of phenotypic and genotypic data from 46,136 non-Hispanic Caucasians, a subset of the multi-ethnic cohort enrolled in the Hemochromatosis and Iron Overload Screening (HEIRS) Study. For analysis of the distribution of transferrin saturation (TS), mixtures of normal distributions were considered and the expectation-maximization (EM) algorithm was applied for parameter estimation. Maximized log-likelihoods were compared, and significance was assessed by resampling. Sensitivity, specificity, and predictive values from the modeled subpopulations were compared with the actual observed genotypes for C282Y and H63D mutations in the HFE gene. A strong association between HFE genotype and TS subpopulations was found in these data collected from different geographic regions, confirming the external validity of the statistical approach when applied to population-based data. It was concluded that mixture modeling of phenotypic data may provide a clinical guide for screening with gender-specific thresholds to identify potential samples for genetic testing. © 2006 Mosby, Inc. All rights reserved.