Cambridge, Metabolic and Cardiovascular Disease. (360G-Wellcome-109150_Z_15_Z)

The role of genomic markers of ageing for the risk of metabolic disease 1\. Aim: To systematically identify and study genomic markers of ageing, including telomere length, DNA methylation, and chromosome loss, and investigate their causal roles for morbidity and mortality from ageing-related diseases, with a focus on the type 2 diabetes (T2D) and associated metabolic disorders. My overall aim will be achieved by addressing the following specific objectives: 2\. Objectives: 2.1 To identify determinants and consequences and assess methods for the characterisation of genomic markers of ageing in epidemiological studies. To conduct a systematic literature review of genomic markers of ageing. 2.2 To identify and characterise determinants of genomic ageing. 2.2.1 To identify genetic determinants of genomic markers of ageing. 2.2.2 To identify behavioural and other risk factors of genomic markers of ageing. 2.3 To investigate roles of genomic markers of ageing in T2D. 2.3.1 To study association between genomic markers of aging and incident T2D. 2.3.2 To study causal roles of the genomic markers of ageing in T2D. 2.3.3 To explore pathways through detailed metabolomic characterisation. 3\. Background 3.1 Global ageing and health The proportion of the elderly population is growing rapidly worldwide and is expected that over the next three decades the percentage of world’s population aged over 60 years will change from 12% to 22%.1 Ageing is often associated with increased risk of many common complex diseases, such as T2D, cardiovascular diseases (CVD), Alzheimer’s disease and different types of cancers,2-5 bringing large burdens to the public health. 3.2 Genomic markers of ageing 3.2.1 Telomere length Telomere is a specialized ribonucleoprotein complex that contains double stranded TTAGGG repeats with a guanidine-rich single stranded overhang at the 3’ end.6 Telomere is shortened in each cell cycle of replication, and shorter telomere lengths were more often observed in elderly people,6 with an estimate of the declining rate of 20-40 bp per year based on cross-sectional population studies.7 Large-scale population studies have suggested that shorter telomere length is associated with various types of cancers, including lung adenocarcinoma, colorectal,8 and skin cancers,9 and metabolic diseases.10 3.2.2 DNA methylation DNA methylation is an epigenetic modification that modulates gene expression without changing DNA sequences.11 It introduces methyl groups onto the C5 position of cytosine residues in dinucleotide arrays of cytosines and guanidines.3 DNA methylation is enriched at DNA tandem repeats to maintain genome stability,12 or at or near promoter regions to regulate gene transcription.3 DNA methylation patterns change over time, exhibiting a progressive reduction of global methylation level but rise at particular sites.13 Cross-sectional studies have shown increased variability of DNA methylation in older people,2 leading to higher propensities for ageing-related diseases.14,15 In addition, several specific regions (11 with Bonferroni-corrected p value < 1.15 X 10-8, and 70 with false discovery rate < 0.1) were found to be highly predictive of chronological age.16 3.2.3 Y Chromosome loss Mosaic loss of chromosome Y (LOY) in hematopoietic cells is the most common post-zygotic mutation in men, and the prevalence increases with age.17,18 Few studies have so far assessed its impact on healthy ageing and ageing-related diseases, except recently, LOY has been associated with all-cause mortality (hazard ratio [HR] = 2.13, p = 0.029) and non-haematological cancer-caused mortality (HR = 3.76, p = 0.022) in a cohort study with aged men (70.7-83.6 years old, n = 1141, median follow-up time = 8.7 years)45 and Alzheimer’s disease (OR = 2.08, p = 0.0184) in a case- control study (n = 1611, 37.6% cases).19 4\. Experimental Design and Methods 4.1 Study populations Objectives will be tested using data from several large epidemiological studies. The InterAct study consists of 12,403 middle-aged (mean = 55.6 years old, SD = 7.7) T2D cases, and a centre-stratified random sub-cohort (16,154 individuals, including 778 who developed T2D during follow- up).20,21 The EPIC-Norfolk study recruited a total of 25,639 participants at the first health check, who were then recalled 4 times and followed-up for disease incidence and survival.22,23 The Fenland study covered 12,435 individuals born in 1950-1975 across Fenland in Cambridgeshire, UK, during the last decade in phase 1, while phase 2 is currently ongoing.55 UK Biobank is an open resource, which contains over 500,000 participants that were aged 40-69 years during 2006 to 2010, and followed up thereafter.56 4.2 Identification and assessment of genomic ageing in large–scale population-based studies. To obtain an overview of whether and how genomic ageing contributes to the ageing-related rise of diabetes and related metabolic disorders, I will conduct a systematic literature review in the PubMed and Embase, following guidelines illustrated in the handbook of Cochrane systematic reviews, to search for large-scale population studies on genomic markers of ageing, with a focus on telomere length, DNA methylation, and mosaic LOY. The primary goal is to identify and assess these genomic markers of ageing in metabolic disease, collect and integrate previously published results of their genetic determinants, potential risk factors and involved pathways, which can later be used for meta-analyses with our novel discoveries. 4.3 Identification and characterisation of determinants of genomic ageing. To determine genetic variants that associate with genomic markers of ageing, I will conduct GWAS in several large population data sets using an additive centre/country-stratified multiple regression model adjusted by age and sex. Genomic control and principle components will be included to correct for confounding due to population stratification. SNPs with p values < 5 X 10-8 will be reported as significantly associated with the tested genomic markers of ageing at a genome-wide level. Moreover, conditional analyses may be performed to discover independent signals that are in low linkage disequilibrium with lead SNPs at each associated locus. To identify modifiable behavioural and environmental factors of genomic ageing, such as smoking, diets and physical activity levels, I will first examine the baseline characteristics for assumption assessment of normality and homoscedasticity, and then apply multivariate regression models adjusted for known confounding effects and validate the models by casewise diagnostics and generalization checkings. 4.4 Investigation on the role of genomic markers of ageing in morbidity and mortality from T2D and other ageing-related diseases. Cox regression model will be used to examine time-dependent effects of genomic changes (exposure) on disease (endpoint) hazard. Based on the observational association, I will then investigate the underlying causal relationships by Mendelian randomization (MR). I will apply optimized genetic instruments in two MR models that account for unbalanced pleiotropy. First is the multivariate MR model, in which potential confounding effects are used as covariates in association analyses. Second is the MR-Egger, in which the allele effects on the outcome diseases are regressed against effects on each genomic marker of ageing, weighted by the inverse variances of the allele effects on the outcomes. Moreover, metabolic pathways within the causal trajectories of the genomic markers of aging and the diseases will be explored by detailed metabolomic characterization. Metabolite quantification has been conducted in the Fenland and EPIC Norfolk studies, with targeted Flow Injection Analysis and liquid chromatography tandem mass spectrometry (MS) methods from Biocrates (Innsbruck, Austria), and untargeted ultra-high performance liquid chromatography and gas chromatography MS methods by Metabolon (Durham, USA), respectively. 5\. Strengths and Limitations 5.1 Static versus dynamic measures of genomic markers of ageing Current studies on biological ageing mainly rely on cross-sectional designs, which are incapable of analysing genetic and genomic effects on prospective mortality due to a lack of follow-up, nor could causality be reliably inferred.24 We are currently investigating the feasibility of performing repeated measures of telomere length and Y chromosome loss in samples stored at baseline and at later time points, which would potentiate longitudinal analyses on the genomic markers of ageing. 5.2 Sample size Sample size considerations differ for the objectives of my study. First, to study genomic markers of ageing in incident diseases, the number of new cases matters. Second, to analyse causality using the MR approach, the strength of the genetic instruments and size and availability of GWAS summary statistics are most important. Our research will be based on 4 large-scale population studies, among which InterAct is the largest case-cohort study of T2D, nested in EPIC Europe. UK Biobank concentrates the greatest national health resources with regards to funding, sample size and data comprehensiveness. Notably, not all genomic markers of ageing are available in each data set, which means that power will differ for analyses of different genomic markers of ageing and outcomes, depending on the marker, study design and outcome to be used.