There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Moonil Kang |
| Institution: | Boston University School of Medicine |
| Department: | Department of Medicine (Biomedical Genetics) |
| Country: | |
| Proposed Analysis: | Background Pleiotropy occurs when a genetic locus or gene region affects multiple phenotypes and is a widespread phenomenon in human complex diseases and traits. There is increasing evidence of pleiotropic effects in psychiatric disorders and neurodegenerative-related traits, such as general cognitive functions, Alzheimer’s disease (AD), and dementia. A large-scale genome-wide association study (GWAS) meta-analysis of brain disorders suggested robust evidence that both psychiatric disorders and neurodegenerative diseases had genetic correlations with cognitive and personality measures. Furthermore, a meta-analysis of general cognitive functions reported its pleiotropic effects on major psychiatric disorders were age-dependent. Several loci of cognitive functions, such as SLC39A1, CTNND2, MAPT, WNT3, CRHR1, KANSL1, and NSF, were also implicated in AD. Recent consortium-based GWASs, on the other hand, identified numerous variants associated with brain structures and provided evidence of its genetic correlations with cognitive functions. Collectively, these findings suggest the possibility of pleiotropy for cognitive functions and AD-related brain changes. Research Hypothesis We hypothesize that a particular single-nucleotide polymorphism (SNP) or aggregated genetic variants in a gene region influence cognitive functioning in multiple domains and AD-related changes in multiple brain regions. Compared to previous pleiotropy analyses of these traits, we plan to evaluate more detailed sub-domains of cognitive functions and structural brain changes measured by magnetic resonance imaging (MRI). We will also focus our pleiotropy analyses on traits that are modestly or moderately correlated with each other. We expect to detect novel associations of genetic variants or loci with multiple cognitive domains or brain MRI measures considered as joint outcomes that would not be identified when analyzed as single outcomes. Analysis Plan Study Design: We will conduct a discovery genome-wide pleiotropy analysis using variables derived from neuropsychological (NP) tests and brain MRI measures obtained from more than 5,000 non-Hispanic white (NHW) participants of the Framingham Heart Study (FHS). We will attempt to replicate top findings from the FHS dataset in a large sample of subjects ascertained from the NIH-funded Alzheimer’s Disease Research Centers (ADRCs). We intend to analyze GWAS datasets generated and assembled by the Alzheimer’s Disease Genetics Consortium (ADGC) and phenotypic datasets from the same subjects curated by the National Alzheimer’s Coordinating Center (NACC). NACC and FHS Datasets: More than 11,100 NACC participants had received the uniform dataset (UDS) version 2 (V2) NP tests at least once. The available test scores are as follows: mini-mental state examination (MMSE), logical memory immediate recall, logical memory delayed recall, digit span forward, digit span backward, Boston naming test (BNT), and digit symbol. About 4,700 NACC participants, in addition, had received UDS V3 NP tests at least once. The followed tests are available: Montreal cognitive assessment (MoCA), craft story immediate recall, craft story delayed recall, number span forward test, number span backward test, multilingual naming test (MINT), copy of Benson figure, and delayed drawing of Benson figure. Test scores of animals, vegetables, trail making A, and trail making B are available in both NP batteries. Identical or similar NP tests were administered to FHS subjects. There are about 6,900 NACC subjects with at least one available structural MRI measure. These datasets include a wide range of measures of gross brain and regional volumes, as well as gray matter and white matter. Identical or similar MRI measures are also available for FHS subjects. Analytical Approach: Models analyzing individual traits and pleiotropy with paired outcomes, including an NP test score and a brain MRI measure, will be adjusted for age, sex, NP battery, education, and the first five principal components (PCs) in each statistical model. As the NACC dataset includes several visits at different time points, we will average repeated measures that change over time, for example, age, NP test scores, and brain MRI measures. We also plan to consider the time difference between each test and MRI measure by subtracting the mean age across visits from the age at MRI and using the value as a covariate in the analysis. We will adopt a weighted regression model to account for the different numbers of observations from different individuals. We will use the highest level of education (under high school degree, high school degree, some college, and over college graduate groups) and pre-existed PCs for NHWs. The prevalence of mild cognitive impairment (MCI), AD, and other dementias are much higher in the NACC than in the FHS: MCI (20.0% vs. 14.6%) and AD or other dementias (23.4% vs. 8.8%). Therefore, in sensitivity analyses, we will down-weight subjects with MCI, AD, and other dementias to adjust for the prevalence differences. Briefly, the analytical approaches in both discovery and replication datasets will comprise three stages. First, we will reduce the number of pairs of variables selected from within and between cognitive domains and brain MRI measures based on genetic and phenotypic correlations estimated using SOLAR-Eclipse (sequential oligogenic linkage analysis routines) software. We intend to focus on modestly or moderately correlated traits to gain higher power to detect pleiotropy in those variables. We will also carefully consider the known functional and structural relationships between NP tests and brain MRI measures to reduce the number of tests. Next, we perform a GWAS for each pair of traits and generate summary statistics of each trait using GENESIS (genetic estimation and inference in structured samples) R package. Finally, we will conduct pleiotropy analyses using PLACO (pleiotropic analysis under composite null hypothesis) package to identify associations for each pair of modestly or moderately correlated traits. Compared to conventional cross-phenotype association tests, PLACO focuses on SNPs or genetic variants associated with multiple traits. It adopts the following null and alternative hypotheses: 1) H0: a tested variant is associated with at most one trait, 2) HA: a tested variant is associated with both traits. Statistical Power: We believe that the NACC sample for each NP test and brain MRI measure is large enough to provide sufficient power to identify pleiotropy using PLACO, especially when considering that the NACC dataset will be used for replication. In extensive simulations with differing conditions at a stringent significance level, PLACO had shown dramatically improved power to detect pleiotropy with a controlled type I error rate than alternative methods typically used. In the simulation using two independent case-control studies (nstudy1=2,000, nstudy2=2,000), PLACO showed improved power to find pleiotropy compared to other existing methods. In another simulation using larger populations (nstudy1=8,000, nstudy2=2,000), PLACO also provided overwhelming power in a pleiotropy analysis compared to the other approaches. Indeed, PLACO is a powerful method in detecting pleiotropy, regardless of sample sizes, overlapping samples, genetic effect sizes, and directions. Considering the sample sizes of the FHS (n>5,000) and NACC (n>11,000) datasets are much larger than those used in the simulation studies and dichotomizing a continuous variable reduces the statistical power, PLACO will provide enough power to detect pleiotropy in cognitive domains and brain MRI measures in our study. |
| Additional Investigators |
