Identification of Genetic Determinants of Aging Traits

As the population in developed countries ages, the health problems associated with aging will become an increasingly important public health concern. While aging is a complex process that takes place over multiple physiological systems, many aspects of aging are under some amount of genetic control. Identification of the loci underpinning these traits could allow for targeting interventions for the most at-risk individuals in old age, improving both lifespan and quality of life. The Amish represent a uniquely appropriate population for the genetic analysis of aging traits. They live a homogenous lifestyle, with high levels of physical activity throughout their lives. The Amish are descended from a small group of religious separatists from Switzerland, and, by examination of documents and family records, can be placed into a single large pedigree. Gait speed and hand grip strength represent useful measures of physical function in old age. Both are heritable and poor performance on either scale predict incipient mortality. Examination of these traits as continuous measures in the Amish population showed evidence of heritability, but no genome regions showed significant evidence of influence on either trait. The use of gait speed as a continuous trait has the unfortunate side effect of excluding participants too impaired to walk at all. To incorporate these individuals, a dichotomous phenotype of impaired mobility was developed. Linkage analysis of this phe- notype resulted in significant evidence of linkage (LOD \u3e 3) to chromosomes 3, 15, and 22. Not all domains of aging are as easily or objectively measured as gait speed and hand grip strength. Questions of stamina or fine dexterity are better investigated by self-reported responses to questionnaires. I used individual answers to 19 questions from four scales of physical ability to group individuals using latent class analysis. The optimal fit of the model suggested that there are two latent groups within the Amish dataset, and that the major feature separating the two groups was physical endurance. I was unable to identify evidence of heritability for latent class membership and genome-wide screens for linkage and association did not identify any candidate loci. Performing genetic analysis in the Amish requires making assumptions that may not reflect the reality of the dataset. I developed an alternative approach to linkage analysis that incorporates the large number of genotypes now available. I use the GERMLINE algorithm to identify regions shared identical-by-descent (IBD) between individuals, and present a score statistic to quantify the configuration of IBD states within a dataset. This method performs analysis in a fraction of the time, and identifies regions shared IBD with high sensitivity. The specificity of this method depends on the pedigree configuration. I present empirical methods of determining statistical significance for arbitrary pedigrees by gene-dropping and resampling based approaches. I demonstrate the performance of this method in two scenarios. As a proof-of-concept, the model is applied to outbred pedigrees of familial essential tremor, a movement disorder that displays an autosomal dominant pattern of inheritance. My method successfully identified a region shared IBD by all affected pedigree members, and excluded as a false positive another region identified by linkage analysis. Application to a successful aging phenotype successfully replicated a locus on chromosome 6 that was identified by conventional linkage analysis and identified another region missed by linkage. The results I present underscore the importance of using observed data to refine the analytic approaches to genetic analysis of aging traits. Beginning with predefined constructs of aging would miss the importance of endurance in aged individuals in this dataset. Exclusion of individuals from continuous traits would have resulted in missing loci shared by individuals unable to walk. The use of conventional linkage analysis requires the exclusion of the majority of the genotype information, and could result in many false negatives across the genome. Aging is a multifactorial process and any investigation into it must be careful to include all the information present in the data