The role of interleukin-1 gene variation and cognition: A longitudinal study

Background and Purpose. Cognitive decline in older adults and neurodegenerative disorders become increasingly prevalent with age. Causes are likely to be numerous and interact in complex ways. Inflammation has been implicated in these processes and has received growing interest over the last two decades, particularly in regard to Alzheimer's Disease. A body of evidence supports inflammation as an agent of decline and disease and includes investigations involving animal models, studies in adult post-mortem brains, epidemiological studies observing a protective influence of some anti-inflammatory agents, and studies correlating increased serum cytokine levels prior to subsequent cognitive decline and dementia. Taken together, this research provides supporting evidence for the causative nature of brain inflammation. The interleukin-1 (IL1) pro-inflammatory cytokine has been hypothesized as a key orchestrator in mounting a brain immune response that leads to a self-perpetuating cycle of neuronal morbidity and mortality across brain regions. Genetic variations in the interleukin-1 genes have been related to inflammatory diseases and to neurodegenerative disorders, but results are mixed. This dissertation research aims to: (1) develop a method to evaluate the effects of haplotypes on trajectories from quantitative, repeated measures over time; (2) evaluate the association of single SNP and haplotype variation in three IL1 genes with cognitive test performance over time in a population of older adults.;Methods. We create software that extends an existing haplotype association method in a GLM framework (Schaid, Rowland et al. 2002; Lake, Lyon et al. 2003) to mixed effects models. The purpose of this method is to iteratively estimate haplotype phase and regression parameters to achieve maximum power and minimum bias. We simulate longitudinal data and compare the bias, power, coverage and type I error of our method to three other, simpler, methods. We also test all four methods in a small sample of 182 white women from the Women's Health and Aging Study II with variation in the IL1 and the IL6 genes. To evaluate IL1 variants with cognitive test performance over time in a larger sample, this dissertation makes use of 3575 white and 481 African American participants of the Cardiovascular Health Study (CHS). Mixed effects models for single SNPs and the newly developed method for associating haplotypes were performed for longitudinal 3MS and DSST scores.;Results. All haplotype trajectory methods resulted in little bias, acceptable type I error and reasonable power and coverage rates. Our haplotype method yields less bias compared to the simpler methods, especially under a moderate to strong alternative. Using WHAS II data, regression estimates and p-values for particular haplotypes did differ by method, though conclusions about associations did not meaningfully change. Using the CHS data, we found significant SNP associations in the IL1RN gene for whites and in the IL1B gene for AAs. Haplotypes did not appear to increase power in these genes, but did provide a significant signal in whites for the IL1A gene.;Conclusions. Our research suggests that haplotype association analysis in a mixed effects model framework stands to benefit from iterative estimation of haplotype phase and the haplotype effects. The advantages of this method become more profound as the haplotype associations become stronger. This research also suggests that IL1 associations differ by race, provides supporting evidence that interleukin-1 proteins may play an important role in cognitive changes among older adults, and suggests that this pathway is likely the result of a dementing process. IL1 gene variants may exert their influence by increasing the expression of the protein, but may also result from synonymous changes in the coding region.