Analysis of the genetic and epidemiological contributors to aging-related traits in the Diabetes Heart Study

Laura Marie Tonks Raffield

A Dissertation Submitted to the Graduate Faculty of

Wake Forest University Graduate School of Arts and Sciences

in Partial Fulfillment of the Requirements

for the Degree of

Doctor of Philosophy

In Molecular Genetics and Genomics

May 2015

Winston-Salem, North Carolina

Committee:

Dr. Donald Bowden (Advisor)

Dr. Timothy Howard (Chair)

Dr. Fang-Chi Hsu

Dr. Barbara Nicklas

Dr. Maggie Ng

Acknowledgements

First, I would like to thank my advisor, Donald Bowden, for giving me the opportunity to work in the Bowden lab. I cannot think of an environment that would have been a better fit for my goals and interests, and I’ve learned a great deal both about human genetics and about science as a profession. I truly appreciate your dedication to mentoring, your advice on analyses, manuscript writing, grants, and your support for my ideas and projects; I couldn’t have asked for a better advisor.

I would also like to thank my committee members for all of their help. Thanks to

Timothy Howard for his help navigating graduate school and leading my committee and for his sense of humor and open door for students. I am so grateful to Fang-Chi Hsu for her perennial kindness and patience with my many, many statistics related questions and for all of her advice…

I have learned so much from you! Thanks to Maggie Ng for patiently answering many questions over the years and for giving me many helpful ideas on how to improve on analyses. Finally, many thanks to Barbara Nicklas for her advice and insights in the field of aging research and for encouraging me to get to know a number of people at the Sticht Center and to attend the

Gerontological Society of America meeting, both of which were amazing learning experiences.

I have been so lucky to be part of the Bowden lab group… thanks to all of you for helping me out countless times and being excellent company, I will miss everyone! I would especially like to thank those who helped me with this dissertation research: Amanda Cox, for her help training me in the lab and her inexhaustible patience and kindness, Nichole Allred, for help with countless questions and many conversations that brightened my day, JJ Xu, for his patience and help with the DHS database, Carrie Smith, for all her hard work on the DHS study, Pam

Hicks, for being an amazing lab manager and awesome company, and Jackie Hellwege, for all her help during graduate school, including helping me edit this dissertation, and for being a great office next door neighbor.

ii

Thanks also to the many investigators at Wake Forest that I have had the privilege to collaborate with… there are too many people to list here, but I would especially like to thank for their help with this dissertation Barry Freedman, for his advice, enthusiasm, and encouragement, and for all his work with the DHS study, and Christina Hugenschmidt, for her help understanding our cognition and neuroimaging data and for organizing journal clubs and other opportunities to help others learn more about the aging brain. Thanks also to everyone in the Genomics Center for their help and support, and to the many individuals who have volunteered to take part in the

Diabetes Heart Study and other studies in the lab; obviously, this research is not possible without you!

I would also like to thank the many individuals at the Medical University of South

Carolina, especially Tammy Nowling and Gary Gilkeson, and at the University of North

Carolina, especially Steve Matson, for their help and support and for encouraging my interest in research and in attending graduate school.

Last but certainly not least, many thanks to my amazing family and friends for their unconditional love and support; I’m so blessed to have you all in my life. Thanks especially to my parents, for always believing in me and being on my side, my siblings, Emily and William, for being so encouraging and making me laugh, and my husband, Taylor, for just generally being the best and helping me more times and in more ways than I can count.

iii

Table of Contents

Page Number List of Tables vi

List of Figures xii

Abbreviations xiii

Abstract xv

Chapter 1: Introduction 1

Chapter 2: Analysis of Coding Variants in C1q/TNF Superfamily Genes in the 15 Diabetes Heart Study and Replication in Additional Cohorts

Chapter 3: Family-Based Linkage Analysis of Cardiometabolic Traits in the 31 Diabetes Heart Study

Chapter 4: Heritability and Genetic Association Analysis of Neuroimaging 60 Measures in the Diabetes Heart Study Published in Neurobiology of Aging

Chapter 5: Impact of HDL genetic risk scores on coronary artery calcified 155 plaque and mortality in individuals with type 2 diabetes from the Diabetes Heart Study Published in Cardiovascular Diabetology

Chapter 6: Analysis of a Cardiovascular Disease Genetic Risk Score in the 187 Diabetes Heart Study Published in Acta Diabetologica

Chapter 7: Predictors of all-cause and cardiovascular disease mortality in type 250 2 diabetes: Diabetes Heart Study Submitted to Diabetology & Metabolic Syndrome

Chapter 8: Cross-sectional analysis of calcium intake for associations with 280 vascular calcification and mortality in individuals with type 2 diabetes from the Diabetes Heart Study Published in American Journal of Clinical Nutrition

Chapter 9: Associations of Coronary Artery Calcified Plaque Density with 304 Mortality and Prior Cardiovascular Disease Events in Type 2 Diabetes: the Diabetes Heart Study

Chapter 10: Associations between Anxiety and Depression Symptoms and 324 Cognitive Testing and Neuroimaging in Type 2 Diabetes

Chapter 11: Associations between Type 2 Diabetes Status and Glycemic 351

iv

Control and Neuroimaging Measures in the Diabetes Heart Study-Mind

Chapter 12: Summary and Conclusions 366

Appendices A.1. Analysis of Apolipoprotein E Polymorphisms and Alzheimer’s 372 Disease Genetic Risk Score in the Diabetes Heart Study

A.2. Longitudinal Analysis of Thoracic Bone Mineral Density in the 384 Diabetes Heart Study

A.3. Association of Magnesium Intake with Vascular Calcification and 390 Mortality in the Diabetes Heart Study

A.4. Additional Genotyped Variants in the Diabetes Heart Study Cohort 398

References 404 Curriculum Vitae 422

v

List of Tables

Page Number Chapter 1 Table 1: Select Age-related Phenotypes available in Diabetes Heart Study and 13 Diabetes Heart Study-Mind Table 2: Baseline Demographic and Clinical Characteristics of Diabetes Heart 14 Study Participants Chapter 2 Table 1: Demographic characteristics of the Diabetes Heart Study population 25 stratified by type 2 diabetes (T2D) affected status. Table 2: Significant single variant association analysis results for Exome Chip 26 coding variants in C1q/TNF superfamily members and their binding partners and receptors. Table 3: Significant gene-based analysis results using the sequence kernel 27 association test (SKAT) for coding variants in C1q/TNF superfamily members and their binding partners and receptors. Table 4: Basic demographic data for genotyped individuals from cohorts 28 included in the Claude Pepper Older Americans Independence Center Biospecimen Repository, including the Diet, Exercise, and Metabolism for Older Women Study (DEMO), the Intensive Diet and Exercise for Arthritis (IDEA) study, the Reconditioning Exercise and COPD Trial II (REACT), and the Lifestyle Interventions and Independence for Elders pilot (LIFE-P). Table 5: Variants genotyped in Claude Pepper Older Americans Independence 29 Center Biospecimen Repository cohorts. Table 6: Meta-analysis of association analysis results from METAL for Claude 30 Pepper Older Americans Independence Center Biospecimen Repository cohorts and the Diabetes Heart Study. Results include only cohorts with a given phenotypic measure, and results are stratified by EA (European American) cohorts only and EA and African American (AA) cohorts combined. Chapter 3 Table 1: Demographic characteristics of the Diabetes Heart Study population 39 stratified by type 2 diabetes (T2D) affected status. Table 2: Top association analysis results (p≤ 1 x 10-5) from the Exome Chip 40 with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait- specific minor allele count ≥ 5 for the 34 cardiometabolic traits available in most Diabetes Heart Study participants. Table 3: Top association analysis results (p≤ 1 x 10-5) from the genome-wide 41 association study array with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait-specific minor allele count ≥ 5 for the 34 cardiometabolic traits available in most Diabetes Heart Study participants. Table 4: Top association analysis results (p≤ 1 x 10-5) from the Exome Chip 42 with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait- specific minor allele count ≥ 5 for the 9 traits available in 520 individuals from the Diabetes Heart Study or less. Table 5: Top association analysis results (p≤ 1 x 10-5) from the genome-wide 43 association study array with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait-specific minor allele count ≥ 5 for the 9 traits available in 520 individuals from the Diabetes Heart Study or less.

vi

Chapter 4 Table 1: Demographic characteristics of the 465 DHS-Mind participants with 78 genotyping data. Table 2: Heritability estimates for MRI imaging variables in related individuals 80 from the Diabetes Heart Study Cohort. Table 3: Genetic association (assuming an additive model of inheritance) 82 between candidate SNPs and neuroimaging measures. Table S1A: Association statistics for the top 50 SNPs associated with TBV. 84 Table S1B: Association statistics for the top 50 SNPs associated with GMV. 87 Table S1C: Association statistics for the top 50 SNPs associated with WMV. 90 Table S1D: Association statistics for the top 50 SNPs associated with GMFA. 93 Table S1E: Association statistics for the top 50 SNPs associated with WMFA. 96 Table S1F: Association statistics for the top 50 SNPs associated with GMMD. 99 Table S1G: Association statistics for the top 50 SNPs associated with WMMD. 102 Table S1H: Association statistics for the top 50 SNPs associated with WMLV. 105 Table S1I: Association statistics for the top 50 SNPs associated with GMCBF. 108 Table S2A: Association statistics for the top 50 SNPs associated with TBV. 111 Table S2B: Association statistics for the top 50 SNPs associated with GMV. 114 Table S2C: Association statistics for the top 50 SNPs associated with WMV. 117 Table S2D: Association statistics for the top 50 SNPs associated with GMFA. 120 Table S2E: Association statistics for the top 50 SNPs associated with WMFA. 123 Table S2F: Association statistics for the top 50 SNPs associated with GMMD. 126 Table S2G: Association statistics for the top 50 SNPs associated with WMMD. 129 Table S2H: Association statistics for the top 50 SNPs associated with WMLV. 132 Table S2I: Association statistics for the top 50 SNPs associated with GMCBF. 135 Table S3A: Association results for the top 25 genes associated with TBV. 138 Table S3B: Association results for the top 25 genes associated with GMV. 139 Table S3C: Association results for the top 25 genes associated with WMV. 140 Table S3D: Association results for the top 25 genes associated with GMFA. 141 Table S3E: Association results for the top 25 genes associated with WMFA. 142 Table S3F: Association results for the top 25 genes associated with GMMD. 143 Table S3G: Association results for the top 25 genes associated with WMMD. 144 Table S3H: Association results for the top 25 genes associated with WMLV. 145 Table S3I: Association results for the top 25 genes associated with GMCBF. 146 Chapter 5 Table 1: Demographic and clinical characteristics of 983 individuals with type 2 173 diabetes from the Diabetes Heart Study. Table 2: HDL-associated SNPs included in the genetic risk scores (GRS). 175 Table 3: Associations between HDL genetic risk scores and HDL, LDL, 177 triglycerides, and coronary artery calcified plaque (CAC). Table 4: Association between HDL genetic risk scores analyzed as a continuous 180 variable and all-cause and CVD-mortality. Additional file 1: Associations between HDL genetic risk scores and HDL, 182 LDL, triglycerides, and coronary artery calcified plaque (CAC), with and without adjustment for diabetic medication use. Additional file 2: Association between HDL genetic risk scores analyzed as a 184 continuous variable and all-cause and CVD-mortality, with and without adjustment for diabetic medication use. Additional file 3: Association between HDL genetic risk score tertiles and all- 186 cause and CVD- mortality using unadjusted proportional hazards regression models and between risk scores and HDL levels using unadjusted marginal

vii models incorporating generalized estimating equations. Chapter 6 Table 1. Demographic characteristics of 983 individuals from 466 families in 200 the Diabetes Heart Study. Table 2: Individual genetic risk scores and their association with the trait of 202 interest in the DHS. Relationships between the risk scores and the traits were examined using marginal models with generalized estimating equations. Models were adjusted for age and sex. Table 3: Associations between composite genetic risk score and prior 204 cardiovascular disease (CVD), prior myocardial infarction (MI), and coronary artery calcification in the DHS were assessed using marginal models with generalized estimating equations. Associations between the composite genetic risk score and all-cause and CVD mortality were assessed using Cox proportional hazards models. Models were adjusted for age and sex. Table 4: Mean coronary artery calcification and prevalence of prior 205 cardiovascular disease (CVD), prior myocardial infarction (MI), and all-cause and CVD mortality for all study participants, participants in the bottom 10% of the composite risk score (GRS) distribution, and participants in the top 10% of the composite GRS distribution. P-value is for the difference between the bottom and top 10% of the composite GRS. Supplementary Table 1: SNPs included in genetic risk scores. 206 Supplementary Table 2: Single SNP association results for SNPs included in 229 genetic risk scores. Supplementary Table 3: Association results for individual trait genetic risk 248 scores. Supplementary Table 4: Association results for all individual trait genetic risk 249 scores analyzed in a single model, with additional adjustment for age and sex. Chapter 7 Table 1: Demographic and clinical characteristics of European American 266 participants with type 2 diabetes, stratified by mortality status. Table 2: Associations of demographic and clinical factors with all-cause and 269 CVD mortality. Table 3: Models selected for all-cause and cardiovascular disease (CVD) 271 mortality in European Americans with type 2 diabetes. Table 4: Average area under the receiver operating characteristic curve (AUC) 272 for the all-cause and CVD mortality models. Table 5: Addition of C-reactive protein and biventricular volume to model 273 selected for all-cause mortality. Table 6: Addition of C-reactive protein and biventricular volume to model 274 selected for cardiovascular disease mortality. Additional File 1: Associations of demographic and clinical factors with all- 275 cause and cardiovascular disease mortality in two randomly selected datasets from European Americans with type 2 diabetes. Additional File 2: Model selected for all-cause mortality using backward 278 elimination, forward selection, and stepwise selection in two randomly selected datasets from European American participants with type 2 diabetes. Additional File 3: Model selected for cardiovascular disease mortality using 279 backward elimination, forward selection, and stepwise selection in two randomly selected datasets from European Americans with type 2 diabetes. Chapter 8 Table 1: Demographic characteristics of Diabetes Heart Study participants with 293

viii type 2 diabetes by daily energy-adjusted total calcium intake quartiles. Table 2: Associations between coronary, carotid, and abdominal aorto-iliac 297 calcification (least squares adjusted mean and 95% confidence interval (CI)) and total, dietary, and supplemental calcium intake. Table 3: Associations between all-cause and CVD mortality (HR and 95% 299 confidence interval (CI)) and total, dietary, and supplemental calcium intake. Supplementary Table 1: Associations between coronary, carotid, and abdominal 301 aorto-iliac calcification (least squares adjusted mean and 95% confidence interval (CI)) and total, dietary, and supplemental calcium intake. Chapter 9 Table 1: Demographic characteristics of European American and African 315 American participants with type 2 diabetes from the Diabetes Heart Study and African American Diabetes Heart Study cohorts with nonzero coronary artery calcification burden. Table 2: Associations with all-cause and cardiovascular disease (CVD) 317 mortality and history of CVD and myocardial infarction (MI) for coronary artery calcification measures analyzed in independent models in European American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures. Table 3: Associations with all-cause and cardiovascular disease (CVD) 319 mortality and history of CVD and myocardial infarction (MI) for density and volume measures analyzed in the same model in European American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures. Table 4: Associations with all-cause and cardiovascular disease (CVD) 320 mortality and history of CVD and myocardial infarction (MI) for coronary artery calcification measures analyzed in independent models in African American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures. Table 5: Associations with all-cause and cardiovascular disease (CVD) 321 mortality and history of CVD and myocardial infarction (MI) for density and volume measures analyzed in the same model in African American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures. Supplementary Table 1: Spearman correlation coefficients for coronary artery 322 calcification measures in European American Diabetes Heart Study participants with type 2 diabetes. Supplementary Table 2- Spearman correlation coefficients for coronary artery 323 calcification measures in African American Diabetes Heart Study participants with type 2 diabetes. Chapter 10 Table 1: Demographic characteristics of Diabetes Heart Study participants with 339 type 2 diabetes stratified by anxiety and depression scores. Table 2: Associations between cognitive testing variables and depression 343 symptoms only, anxiety symptoms only, and anxiety and depression symptoms unadjusted for health status covariates (β value and 95% confidence interval (CI)).

ix

Table 3: Associations between cognitive testing variables and depression 345 symptoms only, anxiety symptoms only, and anxiety and depression symptoms adjusted for health status covariates (β value and 95% confidence interval (CI)). Table 4: Associations between neuroimaging variables and depression 347 symptoms only, anxiety symptoms only, and anxiety and depression symptoms unadjusted for health status covariates (β value and 95% confidence interval (CI)). Table 5: Associations between neuroimaging variables and depression 349 symptoms only, anxiety symptoms only, and anxiety and depression symptoms adjusted for health status covariates (β value and 95% confidence interval (CI)). Chapter 11 Table 1: Demographic characteristics of 790 European American Diabetes 361 Heart Study Mind participants included in the analyses, including 686 participants affected by type 2 diabetes (T2D). Standard deviation (SD), interquartile range (IQR). Table 2: Associations between neuroimaging variables and type 2 diabetes 363 (T2D) affected status in the whole sample (n=790) (β value and standard error (SE)). Analysis was performed using marginal models with generalized estimating equations. Table 3: Associations between neuroimaging variables and diabetes duration, 364 fasting plasma glucose, and glycated hemoglobin (HbA1C) in individuals affected by type 2 diabetes (T2D) (n=686) (β value and standard error (SE)). Analysis was performed using marginal models with generalized estimating equations. Appendix A.1. Table 1: SNPs included in Alzheimer’s disease genetic risk score. 377 Table 2: Association of the Alzheimer’s disease genetic risk score with 378 cognitive testing measures in European American participants from the Diabetes Heart Study-Mind. Table 3: Association of the Alzheimer’s disease genetic risk score with 379 neuroimaging measures in European American participants from the Diabetes Heart Study-Mind. Table 4: Association of rs7412 and rs429358 with cognitive testing measures in 380 Diabetes Heart Study-Mind European American participants. Table 5: Association of rs7412 and rs429358 with cognitive testing measures in 381 Diabetes Heart Study-Mind European American participants. Table 6: Association of rs7412 and rs429358 with cardiovascular disease 382 relevant traits in European American Diabetes Heart Study participants. Table 7: Association of rs7412 and rs429358 with cardiovascular disease 383 relevant traits in African American Diabetes Heart Study participants. Appendix A.2. Table 1: Relationships between clinical and demographic factors and annual 389 percent change in thoracic bone mineral density, analyzed separately in men and women using marginal models with generalized estimating equations. Appendix A.3. Table 1: Mean, minimum, and maximum total energy-adjusted magnesium 394 intake by quartile (mg/day), both for women (n=360) and men (n=360) with type 2 diabetes (T2D) from the Diabetes Heart Study and unaffected women (n=103) and men (n=58) related to these Diabetes Heart Study participants. Table 2: Associations between total energy-adjusted magnesium intake quartiles 395 and coronary artery, carotid artery, and aorto-iliac calcification, both in men and

x women with type 2 diabetes (T2D) and their unaffected siblings. Table 3: Associations between total energy-adjusted magnesium intake quartiles 397 and cardiovascular disease (CVD) and all-cause mortality, both in men and women with type 2 diabetes (T2D) and their unaffected siblings. Models that did not converge are indicated as not applicable (NA). Appendix A.4. Table 1: Additional genotyped single nucleotide polymorphisms (SNPs) in the 399 Diabetes Heart Study.

xi

List of Figures

Page Number Chapter 3 Figure 1: Opposed Manhattan plots including both GWAS and Exome Chip 44 two-point linkage (top, logaritihm of the odds score) and association analysis (bottom, negative logarithm of the p-value) results, for traits including 1A. abdominal aortic calcification, 1B. albumin-to-creatinine ratio, 1C. alkaline phosphatase, 1D. appendicular bone mineral content, 1E. bilirubin, 1F. biventricular volume, 1G. body mass index, 1H. carotid artery calcification, 1I. coronary artery calcification, 1J. diastolic blood pressure, 1K. estimated glomerular filtration rate (extended MDRD equation), 1L. estimated glomerular filtration rate (MDRD equation), 1M. glycated hemoglobin, 1N. high-density lipoprotein cholesterol, 1O. intima media thickness, 1P. low-density lipoprotein cholesterol, 1Q. lumbar bone mineral density, 1R. PR interval, 1S. pulse pressure, 1T. QRS interval, 1U. QT interval, 1V. serum albumin, 1W. serum creatinine, 1X. serum glutamic oxaloacetic transaminase, 1Y. serum glutamic pyruvic transaminase, 1Z. systolic blood pressure, AA. thoracic bone mineral density, BB. total cholesterol, CC. triglycerides, DD. vascular calcification three bed score, EE. waist circumference, FF. whole body bone mineral content, GG. whole body bone mineral density, HH. whole body percent fat. Figure 2: Opposed Manhattan plots including both GWAS and Exome Chip 56 two-point linkage (top, logaritihm of the odds score) and association analysis (bottom, negative logarithm of the p-value) results, for traits including 2A. adiponectin, 2B. insulin, 2C. interleukin 6, 2D. interleukin 8, 2E. leptin, 2F. monocyte chemoattractant protein-1, 2G. pericardial adipose volume, 2H. tumor necrosis factor, 2I. visceral to subcutaneous adipose tissue ratio. Chapter 4 Figures S1A-S1I: Manhattan plots for GWAS associations with (A) TBV, (B) 147 GMV, (C) WMV, (D) GMFA, (E) WMFA, (F) GMMD, (G) WMMD, (H) WMLV, (I) GMCBF. Figures S2A-S2I: Manhattan plots for Exome associations with (A) TBV, (B) 151 GMV, (C) WMV, (D) GMFA, (E) WMFA, (F) GMMD, (G) WMMD, (H) WMLV, (I) GMCBF. Chapter 8 Supplementary Figure 1: Exclusion criteria for analyses of mortality and 303 vascular calcification. Appendix A.2. Figure 1: Distribution of annual percent change in thoracic bone mineral 388 density, with histograms, boxplots, and QQ plots to evaluate normality. A. Distribution of annual percent change in thoracic bone mineral density (measured in mg/cm3) in women (n=92). B. Distribution of annual percent change in thoracic bone mineral density measured in mg/cm3) in men (n= 60).

xii

Abbreviations

3MSE Modified Mini-Mental State Examination AA African American AACP Abdominal aorto-iliac calcified plaque AA-DHS African American Diabetes Heart Study ACR Albumin-to-creatinine ratio APOE Apolipoprotein E BMD Bone mineral density BMI Body mass index CAC Coronary artery calcified plaque CarCP Carotid artery calcified plaque CHARGE Cohorts for Heart and Aging Research in Genomic Epidemiology CI Confidence interval COWA Controlled Oral Word Association Task CSF Cerebrospinal fluid CT Computed tomography CVD Cardiovascular disease DEMO Diet, Exercise, and Metabolism for Older Women Study DHS Diabetes Heart Study DSST Digit Symbol Substitution Task DXA Dual-energy X-ray absorptiometry EA European American eGFR Estimated glomerular filtration rate FLAIR Fluid-attenuated inversion recovery GMCBF Gray matter cerebral blood flow GMFA Gray matter fractional anisotropy GMMD Gray matter mean diffusivity GMV Gray matter volume GRS Genetic risk score GWAS Genome-wide association study HbA1C Glycated hemoglobin HDL High-density lipoprotein cholesterol HR Hazard ratio IBD Identity by descent ICV Intracranial volume IDEA Intensive Diet and Exercise for Arthritis IL1B Interleukin-1 beta IL6 Interleukin 6 IL8 Interleukin 8 IRASFS Insulin Resistance and Atherosclerosis Family Study LDL Low-density lipoprotein cholesterol LIFE-P Lifestyle Interventions and Independence for Elders pilot LOD Logarithm of the odds LST Lesion segmentation toolbox MAF Minor allele frequency MCP-1 Monocyte chemoattractant protein-1 MI Myocardial infarction MNI Montreal Neurologic Imaging MRI Magnetic resonance imaging

xiii

RAVLT Rey Auditory-Verbal Learning Task REACT II Reconditioning Exercise and COPD Trial II SD Standard deviation SKAT Sequence kernel association test SNP Single nucleotide polymorphism SOLAR Sequential Oligogenic Linkage Analysis Routines T2D Type 2 diabetes TBV Total brain volume TNF Tumor necrosis factor US United States WHI Women’s Health Initiative WMFA White matter fractional anisotropy WMLV White matter lesion volume WMMD White matter mean diffusivity WMV White matter volume

xiv

Abstract

Type 2 diabetes (T2D), a disease which affects more than 25% of adults over age 65 in the United States, is characterized by persistent elevations in blood glucose due to lower insulin production and elevated resistance to insulin’s actions in the peripheral tissues. T2D increases risk for a number of age-related comorbidities, including cardiovascular disease and cognitive decline. Risk of these comorbidities is thought to be influenced by both clinical factors, including differences in treatment, lifestyle, and environment, and genetic risk factors. This dissertation explores a number of potential contributors to these comorbidities in the family-based, T2D enriched Diabetes Heart Study cohort.

Both coding variants and common noncoding genetic variants may play a role in age- related comorbidities in patients with T2D. Availability of Illumina HumanExome BeadChips, which include over 240,000 coding variants, allowed analysis of less-commonly studied coding variants in the Diabetes Heart Study, with an initial focus on potential associations of a variety of cardiometabolic traits with variants in the C1q/TNF superfamily of genes. Further global analysis of the Exome Chip data to identify single variants and genes which may be associated with phenotypes of interest, including vascular calcified plaque, adiposity, lipids, inflammatory markers, and MRI derived measures of brain structure, using both two-point linkage and association analysis was also pursued. Genetic risk scores of common, mostly noncoding variants identified in large population-based cohorts were also examined for their cumulative association with age-related comorbidities in a T2D enriched population, in which the effects of these common variants may differ.

Clinical and demographic variables, including medication use, kidney function, subclinical cardiovascular disease measures, diet, and psychosocial factors, are also clearly important contributors to risk of T2D comorbidities and to mortality in T2D patients. Projects included in this dissertation examine which clinical risk factors are independently predictive of

xv mortality, calcium intake’s potential associations with subclinical cardiovascular disease and mortality risk, and relationships between self-reported anxiety and depression and glycemic control and cognitive testing and neuroimaging measures.

The results of these analyses point to roles for less common, coding variants, common, mostly noncoding variants, and clinical and demographic factors in risk for development of aging-related comorbidities in individuals affected by T2D. Further work in the Diabetes Heart

Study and other cohorts of patients with T2D is needed to discover new genetic and epidemiological risk factors and elucidate which are the most important predictors of morbidity and mortality in this patient population.

xvi

Chapter 1 Introduction

Type 2 Diabetes, Cardiovascular Disease, Cognitive Decline, and Aging

Diabetes is a group of chronic metabolic diseases characterized by high levels of blood glucose which arise due to problems with either insulin production in the pancreatic beta cells, insulin’s effects in the peripheral tissues, or both. Type 1diabetes, accounting for approximately

5% of diabetes cases, is characterized by autoimmune destruction of the pancreatic beta cells, ablating pancreatic insulin production and necessitating lifelong use of exogenous insulin in patients. Onset of Type 1 diabetes typically occurs in childhood or adolescence. Type 2 diabetes

(T2D), accounting for 90 to 95% of diabetes cases, is associated with both insulin resistance in peripheral tissues, such as the muscle, liver, and adipose tissue, and impaired insulin secretion from the beta cells. Onset of T2D is typically in adulthood, though it is being diagnosed more frequently in children and adolescents (ADA 2014; CDC 2014). T2D can often be treated with lifestyle changes, such as diet modification, increasing physical activity, and weight loss, and oral medications, but insulin can be used as well if necessary to achieve blood glucose control, with approximately 28.7% of adults with diagnosed diabetes in the US using insulin (CDC 2014).

According to American Diabetes Association standards, diabetes is generally diagnosed using either fasting plasma glucose ≥ 126 mg/dL or glycated hemoglobin (HbA1C) ≥ 6.5%, with additional potential diagnostic criteria available from oral glucose tolerance test results or from a random plasma glucose measure in patients with symptoms of hyperglycemia (ADA 2014). Over

28.9 million adults in the United States (US) are affected by diabetes (CDC 2014). T2D is a complex disease, with risk impacted by genetic, environmental, and lifestyle factors; individuals who are overweight (body mass index (BMI) ≥ 25 kg/m2), physically inactive, dyslipidemic, hypertensive, or have a family history of diabetes are at particularly high risk (ADA 2014).

Prevalence is especially high in the aging population, with 25.9% of American adults over 65 affected by diabetes in 2012 (CDC 2014). The percentage of the US population over 65 years old

1

is steadily increasing, growing 15.1% between 2000 and 2010 as compared to an increase of just

9.7% for the total US population; 40.3 million people were over age 65 years in the US in 2010

(Census 2011), increasing the proportion of the US population impacted by diabetes. Diabetes is the seventh leading cause of death, though this is likely underreported, with only 10-15% of those affected by diabetes listing it as the underlying cause of death (CDC 2014).

As the US population ages, more individuals will be affected by elevated risk of morbidity and mortality due to T2D. T2D is associated with both microvascular and macrovascular complications, increasing risk of heart disease and stroke, kidney failure, limb amputation, and retinopathy and blindness (CDC 2014). Both T2D and normal aging increase the prevalence of a number of comorbidities, including chronic inflammation (Jenny 2012; Donath

2014), dyslipidemia (Carroll, Lacher et al. 2005; Chehade, Gladysz et al. 2013), and declines in kidney function (Glassock and Rule 2012; Afkarian, Sachs et al. 2013). Adjusting for age, sex, and smoking status, individuals with T2D tend to have elevated risk of mortality, both from CVD, as has been frequently reported (Go, Mozaffarian et al. 2014), and from cancer, respiratory disease, and other causes (Gordon-Dseagu, Shelton et al. 2014). Two comorbidities that are of particular interest in the aging T2D affected population are cardiovascular disease (CVD) and cognitive decline.

Mortality risk from CVD is increased two- to four-fold in individuals with T2D, with approximately 68% of T2D affected individuals over age 65 dying of some form of heart disease.

Older adults with diabetes have a reduction of life expectancy free of CVD of approximately eight years versus non-diabetic individuals (Go, Mozaffarian et al. 2014). Recent trials have suggested that improving glycemic control does not reduce CVD events and mortality strikingly for T2D patients, in contrast to the clear impact of glycemic control on microvascular outcomes such as diabetic retinopathy (ADA 2014), increasing interest in better understanding of the etiology of diabetic CVD.

2

As the US population ages, cognitive decline and dementia in late life is an increasing concern; worryingly, T2D has also been shown to accelerate age-related cognitive decline and increase risk of overt dementia, including both Alzheimer’s disease and vascular dementia (Lu,

Lin et al. 2009; Reijmer, van den Berg et al. 2010). Cross-sectional analyses have observed small but consistent decrements in a number of cognitive domains, including executive function, processing speed, and verbal and visual memory, in those with T2D (Palta, Schneider et al. 2014).

Some studies have also found accelerated age-related changes in the brain, as assessed using magnetic resonance imaging (MRI), including increased brain atrophy and microstructural abnormalities in the white matter (Biessels and Reijmer 2014), but this is still an area of active research. In summary, T2D accelerates development of many comorbidities of aging, making a diabetes enriched cohort such as the Diabetes Heart Study (DHS) a unique resource for the study of these comorbidities.

The Diabetes Heart Study Cohort

The Diabetes Heart Study (DHS) is one of the most extensively phenotyped cohorts of

T2D patients available (Table 1), facilitating understanding of the genetics and epidemiology of multiple age-related phenotypes, with a particular emphasis on cardiovascular disease risk. The cohort was recruited from 1998 through 2005 in western North Carolina and includes 1442

European American (EA) and African American (AA) individuals from 564 families, with ~84% of the cohort having overt T2D. Key characteristics of the DHS cohort at time of recruitment are summarized in Table 2. Families were recruited to include at least two T2D affected siblings without advanced renal insufficiency, and in some cases unaffected siblings were also recruited if possible. Characteristics of the cohort, including ascertainment and recruitment, have been previously reviewed (Bowden, Cox et al. 2010). In the DHS study, T2D was defined as diabetes developing after the age of 35 years treated with changes in diet and exercise and/or oral agents and in the absence of initial treatment solely with insulin or historical evidence of ketoacidosis.

Self-reported diabetes status was confirmed by measurement of fasting glucose and HbA1C at the

3

exam visit. The majority of individuals (n= 1221) recruited were of EA descent, and most studies included in this dissertation focus on these participants. Differences in, for example, the distribution of vascular calcification between EA and AA participants prompted the development of the independent African American DHS (AA-DHS) study, recruiting additional unrelated

African American participants with T2D (Bowden, Cox et al. 2010), making the AA-DHS cohort a useful resource for similar future analyses in AA participants.

The initial DHS visit focused on factors which might be associated with CVD risk in individuals with T2D. Vascular calcified plaque, a widely used measure of subclinical CVD, was determined by computed tomography (CT) scan in DHS participants, including assessment of plaque burden in the coronary and carotid arteries and the abdominal aorta. Vascular calcified plaque has been shown in the DHS and other cohorts to be an independent predictor of CVD events and mortality, both in individuals with diabetes and in the general population, (Raggi,

Shaw et al. 2004; Detrano, Guerci et al. 2008; Folsom, Kronmal et al. 2008; Polonsky,

McClelland et al. 2010; Agarwal, Cox et al. 2013; Kramer, Zinman et al. 2013), with individuals affected by diabetes tending to have higher vascular calcified plaque than non-diabetic controls

(Hoff, Quinn et al. 2003). Carotid intima media thickness has also been associated with risk of

CVD events and was assessed using high resolution B-mode carotid ultrasonography (Lorenz,

Markus et al. 2007). Other phenotypic measures available in the DHS cohort, including measures of blood pressure, blood lipid concentrations, electrocardiogram traits, and kidney function, were also selected for their potential associations with CVD risk (Sowers, Epstein et al. 2001; Solano and Goldberg 2006; Giunti, Gruden et al. 2012; Wang, Katzmarzyk et al. 2014). Additional measures with potential roles in CVD risk or that might be related to subclinical CVD burden, such as bone mineral density, inflammatory biomarkers, including for example C-reactive protein and IL6, and adiposity data, including measures of visceral and pericardial adipose tissue volume, were also assessed in a subset of the cohort (Ridker, Rifai et al. 2000; Ridker 2007; Mahabadi,

4

Massaro et al. 2009; Thompson and Towler 2012). Demographic and clinical factors, such as medication use, self-reported history of CVD events, and nutritional intake as assessed using a food frequency questionaire, were also assessed at this baseline visit. Based on follow-up using the Social Security Death Index subsequent to this initial DHS study visit, mortality data is also now available for the DHS cohort over an average of 9.7 years of follow-up.

To better understand cerebrovascular disease and cognitive decline in individuals affected by T2D, individuals from the DHS study were recruited for a follow-up DHS-Mind visit. The

DHS-Mind study has recruited 619 of the original DHS subjects as well as 321 additional individuals for cognitive testing and brain imaging using magnetic resonance imaging (MRI). The study was designed to allow the interactions of cardiovascular disease measures, such as vascular calcified plaque, assessed at the DHS baseline visit with subsequent risk of cognitive decline to be assessed. A cognitive testing battery assessing a number of cognitive domains was administered. Tests administered included the Modified Mini-Mental State Examination (3MSE), a test of global cognitive function often used clinically (Teng and Chui 1987), the Rey Auditory

Verbal Learning Task (RAVLT), a word-list recall task (Lezak, Howieson et al. 2004), the Digit

Symbol Substitution Task (DSST), where participants match numbers and symbols to assess processing speed and working memory (Joy, Kaplan et al. 2004), the Stroop Task, a color-word task to assess executive function (Houx, Jolles et al. 1993; Alvarez and Emory 2006), and tests for Phonemic and Semantic Fluency, both testing verbal fluency (Strauss, Sherman et al. 2006). A number of MRI-based neuroimaging measures were also assessed. Voxel based morphometry was used to assess measures of brain volume, partitioning the MRI images into gray matter, white matter, and cerebrospinal fluid (Matsuda 2013), and diffusion tensor imaging was used to measure the directionality of water molecule diffusion in the brain and assess brain microstructural integrity (Nucifora, Verma et al. 2007). Automated segmentation of white matter lesion volume from fluid-attenuated inversion recovery images was also performed (Maldjian,

Whitlow et al. 2013), as was assessment of cerebral blood flow using pulsed arterial spin

5

labelling methods (Detre, Rao et al. 2012). Cumulatively, these measures allow for determination of factors contributing to variation in cognitive testing performance and neuroimaging measures in the DHS cohort.

Common and Rare Genetic Variants in T2D

Another significant strength of the DHS study is the availability of extensive genetic data. A significant genetic contribution is implicated for many T2D and age-related comorbidities, including cardiovascular disease (CVD) and dementia (Ertekin-Taner 2010; De Jager, Shulman et al. 2012; Kathiresan and Srivastava 2012). These and other comorbidities of T2D are considered to be complex traits, with multiple genetic, environmental, and lifestyle factors all contributing to risk of disease or variation in a given trait (Womack, Jang et al. 2012). Heritability analyses attempt to determine the genetic contribution to phenotypic variation in complex traits by examining trait correlations among related individuals, for example siblings or twins (Manolio,

Collins et al. 2009). While heritability estimates can vary widely depending on the study population, analyses in the DHS as well as many other studies suggest a significant genetic contribution to most T2D related comorbidities, including subclinical cardiovascular disease measures such as vascular calcification and intima-media thickness (Wagenknecht, Bowden et al.

2001; Lange, Bowden et al. 2002; O'Donnell, Chazaro et al. 2002), hypertension measures such as pulse pressure (Hsu, Zaccaro et al. 2005), adiposity measures such as percent fat mass and

BMI (Hsu, Lenchik et al. 2005), neuroimaging measures such as brain volume (Raffield, Cox et al. 2014), and kidney function measures such as glomerular filtration rate and urine albumin-to- creatinine ratio (Langefeld, Beck et al. 2004; Pezzolesi and Krolewski 2013). These significant estimates of heritability increase our interest in genetic analysis of these traits in the DHS.

Earlier studies attempting to identify genetic loci associated with complex traits focused on family-based linkage analysis methods or hypothesis-driven candidate gene approaches. Despite some successes, such as the identification of a common variant in complement factor H which has

6

a large effect on risk of age-related macular degeneration using a combined linkage and association approach (Haines, Hauser et al. 2005), use of these methods has declined. Linkage methods historically had less success for complex as opposed to Mendelian traits, and it can be difficult to refine signals from the large linkage peaks, often derived from sparse linkage panels of single nucleotide polymorphisms (SNPs) or microsatellites, containing hundreds of genes.

Candidate gene approaches have frequent problems with replication and are unlikely to lead to novel insights into disease pathology. Genome-wide association studies (GWAS), where hundreds of thousands of SNPs spanning the genome are examined in an unbiased scan for associated variants, have recently dominated the search for genetic variants associated with complex diseases.

GWAS arrays include mostly noncoding SNPs selected for their ability to capture most of the common variation across the genome, with SNPs included on the array acting as proxies for other

SNPs in linkage disequilibrium which are often inherited together with the typed SNP. These

GWAS allow the identification of common variants which are associated with complex traits

(Pearson and Manolio 2008). In the DHS, data from an Affymetrix® Genome-Wide Human SNP

Array 5.0, from which genotypes for 371,951 SNPs are available after quality control, is available for performing GWAS. Previously completed imputation, where the genotype data of untyped

SNPs can be inferred based on allelic correlations in samples from a reference panel, using this array allows for analysis of a total of ~4.5 million SNPs (Howie, Donnelly et al. 2009). Already existing GWAS studies have identified variants which contribute to T2D-relevant traits such as lipid levels (Willer, Schmidt et al. 2013), pulse pressure (Wain, Verwoert et al. 2011), BMI

(Speliotes, Willer et al. 2010), brain volume (Stein, Medland et al. 2012), and C-reactive protein

(Dehghan, Dupuis et al. 2011). However, with few exceptions, most identified variants have very small effects (OR ~1.1-1.5 at most) (Manolio, Collins et al. 2009), and few studies have been performed specifically in T2D affected populations.

7

While GWAS have had some successes, only a small portion of the heritability for many complex traits, such as T2D, has been elucidated by these studies. Many sources of the “missing heritability” for these complex traits have been proposed, including structural variants, gene-gene or gene-environment interactions, and low frequency coding variants poorly captured by GWAS studies (Manolio, Collins et al. 2009). Coding variants are likely to have larger effect sizes than noncoding variants, and the impacts of coding variants on gene expression and/or protein function are more readily understandable (Kiezun, Garimella et al. 2012). Uncommon coding variants have been shown to contribute to common diseases such as breast cancer and traditional CVD risk factors such as low- and high- density lipoprotein (LDL and HDL) concentrations

(McClellan and King 2010).

Increased interest in analysis of coding variants as a potential source of “missing heritability” for many complex traits has led to the development of new arrays, such as the

Illumina® Infinium HumanExome BeadChips analyzed in the DHS study. These Exome Chips include over 220,000 common and uncommon coding variants and provide a cost-effective approach for analyzing coding variants in a large number of individuals, facilitating the rapid elucidation of which coding variants are likely to contribute to particular phenotypes. Splice variants, promoter SNPs, SNPs previously implicated in GWAS studies, ancestry informative markers, and other SNPs of interest are also included on the Exome Chip array. Coding variants were selected from over 12,000 exome or whole genome sequences from European, African,

Chinese, and Hispanic individuals. After quality control and exclusion of monomorphic SNPs,

88,480 SNPs are available for analysis from the DHS Exome Chip data.

Unfortunately, initial publications from large-scale studies of the association of coding variants with T2D and aging-related phenotypes indicate that variants may have relatively low effect sizes in general and be difficult to detect without the analysis of very large sample sizes.

For example, an early success using Illumina HumanExome Beadchip data was an analysis in

8

8,229 nondiabetic Finnish men for associations with 19 insulin processing and secretion and glycemic traits; associations of low frequency coding variants in GWAS identified loci (SGMS2,

MADD) and newly identified loci (TBC1D30, KANK1, and PAM) with fasting proinsulin and insulinogenic index were discovered, showing the utility of the Illumina Exome Chip technology for analyses of coding variants in a large number of individuals (Huyghe, Jackson et al. 2013).

However, no variant explained more than 1% of trait variance in the population, and the largest effect size, for a low frequency coding variant in TBC1D30, was 0.50 standard deviation units

(Huyghe, Jackson et al. 2013). An Exome Chip analysis of fasting insulin and fasting glucose also identified low frequency variants in novel loci GLP1R and URB2 and additional independent coding variants at G6PC2, although again effect sizes of identified variants were small, with for example the coding variants at G6PC2 explaining only an additional 0.2% of the phenotypic variance in fasting glucose above that explained by the known GWAS SNP (Mahajan,

Sim et al. 2015). Another Exome Chip analysis of glycemic traits additionally identified an intronic ABO variant with modest impacts on fasting glucose (Wessel, Chu et al. 2015). Analysis using Exome Chip data in 56,538 individuals identified only 4 low-frequency variants with impacts on lipid levels; analysis of an Axiom array in 1005 subjects discovered no novel variants for Alzheimer’s disease, and an analysis of an Illumina Exome Chip in 2833 individuals similarly found no significant variants for endometrial cancer risk (Chen, Crous-Bou et al. 2014; Chung,

Kim et al. 2014; Peloso, Auer et al. 2014). Confirmation of some higher impact coding variants recently discovered for T2D and aging-related traits using exome sequencing, for example inactivating mutations in NPC1L1, which reduce coronary heart disease risk (odds ratio (OR)

0.47) or a coding variant in TREM2 which increases risk of Alzheimer’s disease (OR 2.90), required replication in >10,000 individuals due to their low frequency (Jonsson, Stefansson et al.

2013; Stitziel, Won et al. 2014). These early results suggest that large sample sizes or novel approaches (for example, family-based methods) will be needed to discover rare variants for T2D and aging-related phenotypes and that these variants may have effect sizes only modestly greater

9

than GWAS identified variants in most cases. Difficulties in identifying rare variants with substantial effects have renewed interest in alternate analysis methods, such as family-based linkage, which might help prioritize rare variants (Hellwege, Palmer et al. 2014), and gene-based tests of all variants in a locus (Wu, Lee et al. 2011).

Analysis of both the GWAS and Exome Chip data available in the DHS can still add to the literature given the novelty of the DHS cohort, including extensive phenotyping which allows analysis of traits, such as vascular calcification and average fractional anisotropy and mean diffusivity of the brain’s gray and white matter, not available in most cohorts of patients with

T2D. Variants may have a differing impact in a population enriched for T2D, which could aid our search for high impact coding variants and increases interest in analysis of previously identified common variants in a T2D cohort. For example, prior research in the DHS indicated variants in

NOS1AP had a stronger effect on QT interval duration in T2D patients as compared to the general population (Lehtinen, Newton-Cheh et al. 2008). CVD risk SNPs identified in population-based studies do not always associate with risk in individuals with T2D (Farbstein and Levy 2010;

Wang, Peng et al. 2010; Qi, Parast et al. 2011; Qi, Workalemahu et al. 2012), and variants have been described that impact CVD risk only in the presence of diabetes (Farbstein and Levy 2010;

Qi, Qi et al. 2013) or whose effects are modified by factors such as glycemic control (Doria,

Wojcik et al. 2008) and obesity (Bacci, Rizza et al. 2011) in diabetes patients. In this dissertation, we examine both common, mostly noncoding SNPs that have been identified in large GWAS studies and novel, less studied coding variants for their contributions to age-related phenotypes in the DHS.

Hypotheses/Project Objectives

The overall objective of this dissertation research was to identify genetic and epidemiological risk factors associated with phenotypic measures in the DHS cohort, with emphasis on vascular calcification, cognitive and neuroimaging measures, and mortality. We

10

hypothesized that both clinical risk factors and genetic variants, including coding and noncoding

SNPs, contribute significantly to aging and diabetes-related comorbidities in this population. The work presented here includes a number of specific projects which try to address aspects of this overarching hypothesis.

First, data from projects which primarily attempted to identify novel variants associated with age-related traits in the DHS are presented. Chapter 2 describes an analysis focused on coding variants from the candidate C1q/TNF superfamily of genes. Prior research on high impact coding variants in adiponectin (Bowden, An et al. 2010), a C1q family member, prompted us to further examine this superfamily, which includes members with putative roles in diabetes, CVD, inflammation, and other traits relevant to the DHS study. In Chapter 3, an attempt to prioritize coding variants from the DHS Exome Chip data using combined two-point linkage and association analysis approaches for a variety of cardiometabolic traits is described. Chapter 4 focuses specifically on associations of candidate and novel variants from the GWAS and Exome

Chip arrays with several MRI-based neuroimaging measures, with the heritability of these measures in the DHS cohort also assessed.

Next, projects which used genetic risk scores to determine the cumulative impact of common variants identified in prior GWAS studies in the DHS cohort are described. Chapter 5 describes an analysis of risk scores of SNPs previously associated with high-density lipoprotein cholesterol (HDL) concentrations for their potential impacts on vascular calcification and mortality. This approach is extended to the analysis of a composite risk score including variants previously associated with CVD events and multiple CVD risk factors in Chapter 6, assessing associations of the composite risk score with vascular calcification, prior CVD events, and mortality.

Finally, data from projects which primarily focus on clinical and epidemiological risk factors, as opposed to genetic risk factors, which may contribute to variation in aging and diabetes-related outcomes in the DHS, including risk of mortality, are presented. Chapter 7

11

describes an analysis determining which of the CVD risk factors collected at baseline are independently predictive of all-cause and CVD mortality. Chapter 8 focuses specifically on examination of a dietary factor, i.e. calcium intake from diet and supplements previously reported to have impact on CVD for potential associations with vascular calcification and mortality.

Chapter 9 focuses on associations of coronary artery calcified plaque (CAC) density and its associations with CVD events and mortality. Chapters 10 and 11 focus on epidemiological risk factors, specifically self-reported anxiety and depression and measures of glycemic control, as they relate to variation in cognitive testing and neuroimaging measures in DHS-Mind. Together, these analyses seek to expand and extend our understanding of the factors which contribute to variation in health outcomes in individuals with T2D.

12

Table 1: Select Age-related Phenotypes available in Diabetes Heart Study and Diabetes Heart Study-Mind Biomedical domain Clincal Measure or Measurement Method Vascular Disease (Coronary, carotid, and abdominal aorta Computed tomography (CT) calcified plaque) Glucose Control HbA1c, fasting glucose Dyslipidemia Total Cholesterol, HDL, LDL, Triglycerides Albumin/creatinine ratio (ACR), glomerular filtration rate Renal function (GFR) Inflammation Status IL-1β, TNF, IL-6, IL-8, C-reactive protein Adipokines Adiponectin, leptin, resistin Bone density CT, dual energy X-ray absorptiometry (DXA) Adiposity (Visceral and CT Subcutaneous Adipose Tissue) Digit Symbol Substitution Task Modified Mini-Mental State Examination Phonemic Fluency via the Controlled Oral Word Association Cognition Task and Semantic Fluency Stroop Task Rey Auditory-Verbal Learning Task White matter lesion burden, white matter, gray matter and Brain Structure total brain volume, and average fractional anisotropy and mean diffusivity from Magnetic Resonance Imaging (MRI)

13

Table 2: Baseline Demographic and Clinical Characteristics of Diabetes Heart Study Participants T2D-affected (n=1208) T2D-unaffected (n=234) Phenotype Mean (SD) Median (range) Mean (SD) Median (range) Gender (% female) 54% 63% Age (yrs) 61.5 (9.3) 61.4 (34 - 86) 59.3 (10) 59.4 (34 – 83.4) Duration of diabetes (yrs) 10.4 (7.3) 8.0 (1 - 41) BMI (kg/m2) 32.7 (6.9) 31.5(17.1 - 59.8) 29.2 (5.3) 28.3 (16.6 – 44) Total cholesterol (mg/dl) 187.6 (43) 183 (74 - 427) 195.1 (34.1) 196 (104 - 325) HDL (mg/dl) 43.5 (13) 41 (8 - 115) 48.8 (14.3) 47 (23 - 104) LDL (mg/dl) 105.7 (33) 103 (14 - 283) 114.3 (30) 113 (43 - 221) Triglycerides (mg/dl) 200.1 (139) 168(30 - 1310) 161.6 (84.6) 145 (47 - 627) Glycated hemoglobin (%) 7.9 (2) 7.5 (4.3 – 21.8) 5.6 (0.5) 5.6 (3.5 – 6.9) Fasting glucose (mg/dL) 152.2 (62.9) 138 (16 - 568) 95.1 (14.6) 94 (66 - 126) C-reactive Protein (mg/dL) 0.66 (1.02) 0.33 (0 – 12.7) 0.54 (0.81) 0.31 (0.002 – 7.14)

Coronary Artery Calcium 1875 (3391) 479.8 (0 - 50415) 769 (1691) 59 (0 - 11569)

Carotid Artery Calcium 357 (720) 71 (0 - 6122) 159.5 (430) 4.5 (0 - 3271)

Abdominal Aortic Calcium 12118 (0- 94156) 7456 (13450) (0- 72745) (16458)

14

Chapter 2

Analysis of Coding Variants in C1q/TNF Superfamily Genes in the Diabetes Heart Study and Replication in Additional Cohorts

Laura Raffield, Breck Radulovic, Amanda Cox, Daniel Beavers, Michael Berry, Christina Hugenschmidt, Carl Langefeld, Iris Leng, Stephen Messier, Poorva Mudgal, Karin Murphy, Maggie Ng, J. Jeffrey Carr, Barry Freedman, Barbara Nicklas, Donald Bowden

15

Introduction

C1q/TNF superfamily genes have many roles in inflammation and other age-related processes. This superfamily is defined by the homologous C1q globular domain and TNF homology domain, both with a ten β-strand jelly-roll structure. C1q/TNF superfamily members form self- assembling trimers, with some members, such as adiponectin, also forming higher order multimers (Kishore, Gaboriaud et al. 2004; Whitehead, Richards et al. 2006).

Our interest in the C1q/TNF superfamily was prompted by work in which the Bowden laboratory identified coding variants in the C1q family member adiponectin (gene: ADIPOQ).

Adiponectin is an adipokine putatively linked to increased hepatic insulin sensitivity and decreased vascular inflammation, among other effects, and plasma adiponectin levels are decreased in individuals with diabetes and in obese individuals (Whitehead, Richards et al. 2006).

Adiponectin, like many C1q family members, includes a collagen domain, i.e. repeated Gly-X-Y motifs. Mutations within the collagen domain can interfere with protein multimerization (Waki,

Yamauchi et al. 2003; Jungtrakoon, Plengvidhya et al. 2011). Prior work in the Bowden laboratory identified low frequency coding variants in ADIPOQ in the collagen domain which reduce circulating adiponectin protein concentrations by >80% and reduce high molecular weight multimers to <5% of normal in both Hispanic Americans and African Americans. In Hispanics a glycine to arginine substitution at amino acid 45 (G45R) was identified in approximately 1% of the Hispanic population enrolled in the Insulin Resistance Atherosclerosis Family Study

(IRASFS) (Bowden, An et al. 2010), and a similarly high impact variant at amino acid 55 (R55C) was identified in approximately 1% of African Americans (AA) (An, Hanley et al. 2012).

We extended this analysis of adiponectin coding variants to an analysis of all coding variants in C1q/TNF family members included on the Exome Chip, which was available in most

European American participants in the DHS. While little is known about some members of the

16

C1q/TNF superfamily, others have been implicated in an array of age-related phenotypes, increasing our interest in this gene family. For example, TNF has well documented roles in inflammation and is elevated in patients with chronic heart failure, as well as in obese patients

(Aggarwal, Gupta et al. 2012). C1q family member CTRP12 acts as an anti-diabetic adipokine; in mouse models of diabetes and obesity, injection of recombinant CTRP12 results in decreased blood glucose concentrations (Wei, Peterson et al. 2012). C1qTNF3 also can decrease blood glucose in mouse models and suppress gluconeogenic gene expression in the liver, and C1qTNF3 administration was recently shown to improve survival in mice after myocardial infarction, via increased angiogenesis and reduced fibrosis (Peterson, Wei et al. 2010; Yi, Sun et al. 2012). The

TNF family member RANKL stimulates osteoclast mediated bone resorption, which can lead to osteoporosis. Denosamub, a monoclonal antibody that inhibits RANKL, has been approved for treatment of osteoporosis in postmenopausal women (Adler and Gill 2011).

Other C1q/TNF superfamily members may have roles relevant to the cognition and brain structure phenotypes assessed in DHS Mind. C1QL1, C1QL2, C1QL3, and C1QL4 are thought to have roles in the synapses of the brain, binding to brain-specific angiogenesis inhibitor 3

(Bolliger, Martinelli et al. 2011). C1q, which initiates the classical complement cascade, is elevated in the aging brain, and C1q knockout mice exhibited less cognitive decline with aging than wild-type controls (Stephan, Madison et al. 2013).

Here we present association analysis results for 464 polymorphic coding variants from the Exome Chip in 97 C1q/TNF related genes with age-related phenotypes in the DHS, a family- based cohort enriched for type 2 diabetes (T2D). Results of replication analyses for a subset of associated variants in cohorts from the Claude Pepper Older Americans Independence Center

Biospecimen Repository are also presented. These cohorts have collected many phenotypic measures similar to those available in the DHS. For example, the Diet, Exercise, and Metabolism for Older Women Study (DEMO) collected CT measures of visceral and subcutaneous abdominal

17

adipose tissue volume (Nicklas, Wang et al. 2009), and the Intensive Diet and Exercise for

Arthritis (IDEA) study collected DXA measures of bone mineral density (BMD) (Messier,

Legault et al. 2009). The Reconditioning Exercise and COPD Trial II (REACT) and the Lifestyle

Interventions and Independence for Elders pilot (LIFE-P) measured cytokines and inflammatory markers and other relevant measures of lipid levels and bone density (Foy, Wickley et al. 2006;

Katula, Kritchevsky et al. 2007). If SNPs replicate in association analyses performed in these cohorts, this would allow extension of results from the DHS cohort to older cohorts without a focus on recruitment of individuals with T2D.

Methods

Study Population

The DHS recruited siblings with T2D without advanced renal insufficiency from outpatient internal medicine and endocrinology clinics and from the community in the region surrounding Winston-Salem, North Carolina from 1998 through 2005, with unaffected siblings also recruited when possible. The cohort has been described in detail previously (Bowden, Cox et al. 2010). These analyses include only DHS participants with Exome Chip data, in total 1190

European American (EA) individuals from 468 families.

Genotyping

Genotyping for our initial analyses was completed using the Illumina Infinium

HumanExome BeadChip array. Quality control of the DHS Exome Chip data has been previously described (Cox, Ng et al. 2013; Cox, Hugenschmidt et al. 2014). Briefly, individual samples failing to meet a minimum acceptable call rate of 98%, with poor quality genotype calls, with gender errors, or with unclear sibling relationships were excluded, as were unintentionally duplicated samples. SNPs which did not meet a minimum acceptable call rate of 95% were excluded from all analyses. All SNPs were tested for departure from Hardy-Weinberg

18

equilibrium (HWE) with an exact test implemented in PLINK, with only unrelated individuals included in this calculation (Purcell, Neale et al. 2007). SNPs with a p-value of less than 1 x 10-6 for this test of HWE were excluded from further analyses.

After association analyses were performed, a subset of variants for which similar phenotypes were available in additional cohorts from the Claude Pepper Older Americans

Independence Center Biospecimen Repository were genotyped in these cohorts. This set of variants included some more modestly associated variants from the association analysis that did not meet a strict multiple comparisons threshold and are not included in the association analysis results displayed here. Genotyping was pursued using the well validated Sequenom® MassArray

Genotyping platform (Buetow, Edmonson et al. 2001). Samples with a genotyping rate <90% were excluded from association analyses. SNPs with a call rate <95% (rs149032158, rs150126476, rs7412) were also excluded from the results.

Statistical Analysis

In total, 50 CVD- and T2D-related phenotypes were analyzed, including measures of vascular calcified plaque, glucose control, dyslipidemia, inflammation, renal function, bone density, cognitive function, and adiposity. Preliminary analyses were also performed for 8 neuroimaging traits, which have since been subject to further quality control. Phenotypes were transformed as necessary to approximate a normal distribution prior to association analyses.

Single SNP association analyses were performed in the DHS using a variance components based approach in the Sequential Oligogenic Linkage Analysis Routines (SOLAR) software package (Almasy and Blangero 1998) for 464 polymorphic coding variants in 97

C1q/TNF related genes. Gene-based tests using the sequence kernel association test (SKAT) were only performed for the 85 genes containing at least two coding variants (Lee, Teslovich et al.

2013). Associations based on a single copy of the minor allele were excluded for the single

19

variant analyses. Analyses were adjusted for age, sex, and T2D status, and analyses of cognitive tests were also adjusted for educational attainment. All analyses were performed under the additive model. A Bonferroni-corrected p-value cut-off for the single variant analyses of 1.16 x

10-4 (based on 429.502 independent SNPs tested, calculated in SOLAR using the Moskvina &

Schmidt method (Moskvina and Schmidt 2006) to correct for linkage disequilibrium between

SNPs, α=0.05) was used, while for the gene-based analyses in SKAT, a Bonferroni-corrected p- value cut-off of 5.88 x 10-4 was used (based on 85 genes tested, α=0.05).

For the additional variants genotyped in Claude Pepper Older Americans Independence

Center Biospecimen Repository cohorts, single variant association analyses were performed using

PLINK (Purcell, Neale et al. 2007), with association analyses adjusted for age and sex (DEMO included only females, so analyses were age-adjusted only). Association analyses were performed separately for EA and AA participants. Phenotypes were transformed to approximate normality prior to association analysis, and analyses were performed under the additive model. Meta- analysis of results from these in Claude Pepper Older Americans Independence Center cohorts and the DHS was performed using METAL, which weights results by sample size (Willer, Li et al. 2010).

Results

Table 1 presents basic demographic and clinical risk factor information for the 1190 DHS participants with Exome Chip data, stratified by T2D affection status; 83.7% of participants were affected by T2D with average disease duration of 10.5 ± 7.2 years. As might be expected for a cohort recruiting individuals affected by T2D and their family members, obesity and both subclinical cardiovascular disease (CVD) and CVD events are prevalent.

Table 2 displays all single variant association analysis results that met a Bonferroni- corrected p-value cut-off of 1.16 x 10-4. Some of the most interesting results include a variant in

20

cadherin 13 (CDH13, Asn39Ser) associated with abdominal aortic calcification (p= 6.23 x 10-6) and a variant in calreticulin (CALR, Glu381Ala) associated with C-reactive protein (CRP) levels

(p= 1.80 x 10-5). A number of genes also met a Bonferroni-corrected p-value threshold (p=5.88 x

10-4) for gene-based tests using SKAT (Table 3). Again, CDH13 and CALR were highlighted by these gene-based analyses, with 8 variants in CDH13 associated with abdominal aortic calcification (p=2.48 x 10-5) and 2 variants in CALR associated with C-reactive protein (p= 2.70 x

10-5).

Based on these association results, we selected variants for genotyping in Claude Pepper

Older Americans Independence Center Biospecimen Repository cohorts, including IDEA,

REACT, DEMO, and LIFE-P. Basic demographic data for these cohorts is displayed in Table 4.

We genotyped 21 SNPs, with the 18 SNPs genotyped successfully listed in Table 5 with their positions and annotation, with the trait of interest for that given SNP and whether or not that SNP was for this C1q/TNF superfamily project (some SNPs are from other unrelated projects genotyped with these samples). Association analysis was pursued in PLINK, followed by meta- analysis with DHS results from SOLAR using METAL, first in EA cohorts only then in both EA and AA cohorts where available. The results for all variants genotyped for this project, including ones for which no data was available in additional cohorts, are listed in Table 6. Table 6 lists the

SNP of interest, the compellingly associated trait in the DHS, the trait used for meta-analysis across cohorts (not always identical to the original associated trait in the DHS), the total sample size included in the meta-analysis, meta-analysis p-value, and the direction of effect for the allele of interest in each study, with results listed for EA cohorts only and EA and AA cohorts combined. The association of pulse pressure with rs1801106 in the integrin α2 gene (ITGA2,

E534K) was the only association strengthened by meta-analysis (p=0.00099 in the DHS alone, p=0.0004 in all EA cohorts with pulse pressure data).

21

Discussion

This study analyzed coding variants in the C1q/TNF superfamily for potential roles in aging and diabetes relevant biomedical traits in the DHS. This study was prompted by the high impact coding variants in adiponectin, a C1q family member, previously analyzed in the lab. For single variant association analyses, significant associations with mutations in domains important for multimerization, such as those seen in the collagen domain of adiponectin (Bowden, An et al.

2010; An, Hanley et al. 2012), were rare, but several associations were observed in binding partners/receptors of C1q/TNF family members. Several of the same receptors were highlighted in gene-based analyses of all coding variants in a given gene. However, attempts to replicate variants in additional cohorts from the Claude Pepper Older Americans Independence Center

Biospecimen Repository cohorts were generally unsuccessful.

Notable results from the single variant and gene-based analyses include associations of cadherin 13 with abdominal aortic calcification and calreticulin with C-reactive protein.

Adiponectin can bind cadherin 13, which is expressed in endothelial and smooth muscle cells and is associated with atherosclerotic progression(Takeuchi, Adachi et al. 2007). A copy number variant at the cadherin 13 locus has been associated in a small cohort of Taiwanese patients with myocardial infarction risk and hyperlipidemia (Shia, Ku et al. 2011), and the locus has been identified in genome-wide association studies of blood pressure and adiponectin (Org,

Eyheramendy et al. 2009; Chung, Lin et al. 2011). Calreticulin is also thought to bind adiponectin, mediating phagocytosis of apoptotic cells by macrophages and preventing systemic inflammation (Takemura, Ouchi et al. 2007), a function that fits well with a variant which is associated with elevated CRP levels, a marker of systemic inflammation. The eight variants in the cadherin 13 locus were also cumulatively associated with abdominal aortic calcification in analysis using SKAT, and the two variants in calreticulin were also cumulatively associated with

CRP levels. Unfortunately, these variants in calreticulin and cadherin 13 could not be analyzed

22

with similar phenotypes in our replication cohorts and thus were not genotyped. Replication of these uncommon coding variants in additional cohorts would be necessary to validate their roles in CVD- and T2D-related phenotypes.

Few of the variants we selected from our initial analysis for replication were associated with similar phenotypes in additional cohorts. The one exception was the association of pulse pressure with a coding variant in integrin α2, which was strengthened in replication analysis.

Little is known about potential roles for integrin α2 in blood pressure regulation, but in a previous study this variant was associated with differences in platelet adhesion (Kunicki, Williams et al.

2012), and studies of other variants in this gene have found associations with early-onset myocardial infarction and stroke (Carlsson, Santoso et al. 1999; Santoso, Kunicki et al. 1999).

Lack of replication may be due to a number of factors, including differences in cohort ascertainment, with the Claude Pepper Older Americans Independence Center Biospecimen

Repository participants generally being older than the average DHS participant and ascertained based on criteria other than T2D-affected status. There were also differences in the phenotypes upon which the original associations were based and those that could be meta-analyzed between the DHS and the replication cohorts; for example, for rs984274 and rs72807847, the original compelling association was for a CT derived measure of thoracic bone mineral density, but the only comparable phenotype available for analysis in both the DHS and additional cohorts was whole body bone mineral content from DXA. However, it is also likely that many of our initial observations, especially those associations based on only a few individuals, may be spurious, given the now known issues with reproducibility and power in rare variant analyses, especially in small cohorts (Wu, Lee et al. 2011; Kiezun, Garimella et al. 2012; Lee, Teslovich et al. 2013;

Perreault, Legault et al. 2014).

Limitations of this analysis include potentially missing some binding partners/receptors of C1q/TNF family members which were not found in the literature review and using METAL for

23

meta-analysis of AA and EA cohorts. METAL is not ideal for multi-ethnic meta-analysis, unlike the recently developed program MANTRA, for example, which is a Bayesian meta-analysis program that allows for heterogeneity of allelic effects across diverse populations (Morris 2011).

However, using a different program is unlikely to have changed the results, especially given the low number of AA participants. While this analysis of coding variants in C1q/TNF superfamily genes highlighted some interesting loci, larger sample sizes and use of exome sequencing as opposed to Exome Chip data to identify very low frequency or population specific but potentially high impact variants are likely needed in future analyses of this gene family.

24

Table 1: Demographic characteristics of the Diabetes Heart Study participants stratified by type 2 diabetes (T2D) affected status.

T2D-affected (n=996) T2D-unaffected (n=194) Phenotype Mean (SD) Mean (SD) Gender (% female) 51.5% 62.4% Age (yrs) 62.5 (9.1) 60.1 (10.3) Duration of diabetes (yrs) 10.5 (7.2) - Body mass index (kg/m2) 32.3 (6.6) 28.9 (5.1) Glycated hemoglobin (%) 7.6 (1.7) 5.6 (0.5) Fasting glucose (mg/dL) 147.9 (56.1) 93.9 (11.4) Coronary artery calcium (mass score) 1878.9 (3365.3) 639.5 (1568.5) Self-reported history of prior 43.5% 22.5% cardiovascular disease (%)

25

Table 2: Significant single variant association analysis results for Exome Chip coding variants in C1q/TNF superfamily members and their binding partners and receptors in the Diabetes Heart Study (DHS).

Variant Gene Amino Acid Associated DHS β value p-value Change trait Trait- Specific Allele Counts chr11:68153967 LRP5 A400V Interleukin 1 beta 0/2/481 2.04 3.21 x 10-11 chr1:1140855 TNFRSF18 E69K Interleukin 1 beta 0/2/481 1.67 2.41 x 10-9 chr4:94411877 GRID2 T649M Albumin-to- 0/2/1163 5.23 1.82 x 10-6 creatinine ratio chr19:42740863 GSK3A D187D Resistin 0/2/470 -4.50 3.49 x 10-6 chr3:99513830 COL8A1 R362Q White matter 0/2/452 -0.14 5.76 x 10-6 mean diffusivity chr16:82892037 CDH13 N39S Abdominal 1/8/844 77.05 6.23 x 10-6 Aortic Calcification chr6:47277181 TNFRSF21 A23S Interleukin 1 beta 0/7/476 0.80 8.57 x 10-6 chr8:119938850 TNFRSF11B A234T Intima-media 0/2/1090 0.23 9.12 x 10-6 thickness chr13:24242166 TNFRSF19 P262S Plasminogen 0/13/454 -0.93 1.20 x 10-5 activator inhibitor-1 chr19:13054615 CALR E381A C-reactive 0/3/996 0.88 1.80 x 10-5 protein chr4:122591118 ANXA5 Y257Y Interleukin 1 beta 0/2/481 1.48 2.00 x 10-5 chr2:182399097 ITGA4 H961H Total cholesterol 197/541/432 -0.04 2.70 x 10-5 chr1:12175658 TNFRSF8 C273Y Monocyte 0/7/480 0.89 3.30 x 10-5 chemoattractant protein-1 chr2:182350641 ITGA4 V359I Estimated 0/6/1111 0.41 9.40 x 10-5 glomerular filtration rate (extended MDRD equation) chr12:12334271 LRP6 R360H Modified Mini- 0/3/538 -14.92 1.01 x 10-4 Mental State Examination

26

Table 3: Significant gene-based analysis results using the sequence kernel association test (SKAT) for coding variants in C1q/TNF superfamily members and their binding partners and receptors.

Gene p-value # SNPs Trait Top Single Variant Amino Acid Top Single SKAT Association Change Variant Association p- value TNFRSF21 1.28 x 10-6 4 Interleukin 1 6:47277181 A23S 8.57 x 10-6 beta TNFRSF19 1.13 x 10-5 5 Plasminogen 13:24242166 P262S 1.20 x 10-5 activator inhibitor-1 CDH13 2.48 x 10-5 8 Abdominal 16:82892037 N39S 6.23 x 10-6 Aortic Calcification CALR 2.70 x 10-5 2 C-reactive 19:13054615 E381A 1.80 x 10-5 protein GSK3A 2.80 x 10-5 3 Resistin 19:42740863 D187D 3.49 x 10-6

LRP5 1.02 x 10-4 6 Interleukin 1 11:68153967 A400V 3.21 x 10-11 beta C1QA 1.10 x 10-4 4 Plasminogen 1:22964120 P4R 7.31 x 10-8 activator inhibitor-1 EDA2R 2.00 x 10-4 3 Interleukin 1 X:65835841 Y8H 1.33 x 10-4 beta TNFRSF18 2.26 x 10-4 3 Interleukin 1 1:1140855 E69K 2.41 x 10-9 beta TNFRSF13B 7.39 x 10-4 4 Insulin 17:16843084 V220A 2.39 x 10-4

ANXA5 9.27 x 10-4 5 Interleukin 1 4:122591118 Y257Y 2.00 x 10-5 beta

27

Table 4: Basic demographic data for genotyped individuals from cohorts included in the Claude Pepper Older Americans Independence Center Biospecimen Repository, including the Diet, Exercise, and Metabolism for Older Women Study (DEMO), the Intensive Diet and Exercise for Arthritis (IDEA) study, the Reconditioning Exercise and COPD Trial II (REACT), and the Lifestyle Interventions and Independence for Elders pilot (LIFE-P).

Trait DEMO IDEA REACT LIFE-P Number of Participants 110 418 148 361 Age (years, mean ± sd) 58.3 ± 5.4 65.6 ± 6.2 66.5 ± 9.0 76.8 ± 4.3 Sex (% female) 100% 72.5% 44.6% 68.1% Body Mass Index (mean ± sd) 33.2 ± 3.8 33.6 ± 3.7 28.2 ± 6.0 30.2 ± 5.8 % European American 64.6% 80.9% 82.4% 75.4% % African American 34.6% 17.2% 11.5% 18.3% % Other Race/Ethnicity 0.9% 1.9% 6.1% 6.4%

28

Table 5: Variants genotyped in Claude Pepper Older Americans Independence Center Biospecimen Repository cohorts.

SNP Gene Chr Position Amino Acid Trait Candidate Change from another project? rs139380413 COL8A1 3 99513830 R362Q Interleukin 6 No rs138227502 ADIPOQ 3 186571010 R55C Adiponectin Yes rs62625753 ADIPOQ 3 186572026 G90S Adiponectin Yes rs150683134 GRID2 4 94411877 T649M Albumin-to-creatinine ratio No rs41278081 ANXA5 4 122604580 I81T Spine Bone Mineral Density No (DXA) rs1801106 ITGA2 5 52358757 E534K Pulse pressure No rs146679395 TNFRSF21 6 47253884 S182G Tumor necrosis factor-1, No Interleukin 6, Interleukin 8 rs76324416 TNFRSF10C 8 22974278 P172S Thoracic and Lumbar Bone No Mineral Density (CT), Hip Bone Mineral Density (DXA) rs12002324 GRHPR 9 37429744 R170Q Visceral Adipose Tissue Yes rs201320326 LRP5 11 68153967 A400V Interleukin-1 beta No rs143657689 PLA2G4D 15 42364501 E469D Low-density lipoprotein Yes cholesterol rs140199133 SHF 15 45470476 P111L Low-density lipoprotein Yes cholesterol rs247616 CETP 16 56989590 - High-density lipoprotein Yes cholesterol rs72807847 CDH13 16 82892037 N39S Thoracic Bone Mineral No Density and Abdominal Aortic Calcification rs61744697 DNAH9 17 11672607 V2505L Systolic Blood Pressure Yes rs56063729 TNFRSF13B 17 16843084 V220A Fasting Insulin No rs984274 DCC 18 50936935 M1017V Thoracic BMD (CT) No rs143080537 GTSF1L 20 42355316 E7K Adiponectin Yes

29

Table 6: Meta-analysis of association analysis results from METAL for Claude Pepper Older Americans Independence Center Biospecimen Repository cohorts and the Diabetes Heart Study (DHS). Results include only cohorts with a given phenotypic measure, and results are stratified by EA (European American) cohorts only and EA and African American (AA) cohorts combined.

Results-EA Only Results-EA and AA SNP Trait from DHS Analysis Trait Total p- Direction Total p-value Direction Analyzed sample value sample for Meta- size size Analysis rs139380413 Interleukin 6 Interleukin 6 811 0.012 +- 811 0.012 +- rs150683134 Albumin-to-creatinine None- No other cohort had albumin-to-creatinine ratio ratio rs41278081 Spine Bone Mineral Whole Body 1092 0.176 --- 1126 0.102 ?---- Density (DXA) Bone Mineral Content rs1801106 Pulse pressure Pulse 1522 0.0004 +++ 1614 0.0006 ++++- pressure rs146679395 Tumor necrosis factor-1, None- Monomorphic in all other cohorts Interleukin 6, Interleukin 8 rs76324416 Thoracic and Lumbar None- Monomorphic in all other cohorts Bone Mineral Density (CT), Hip Bone Mineral Density (DXA) rs201320326 Interleukin-1 beta None- No other cohort had interleukin-1 beta rs72807847 Thoracic Bone Mineral Whole Body 1092 0.194 ++- 1126 0.238 ?-++- Density (CT) and Bone Abdominal Aortic Mineral Calcification Content rs56063729 Fasting Insulin Fasting 702 0.016 --+ 761 0.078 --++ Insulin rs984274 Thoracic Bone Mineral Whole Body 1092 0.050 --+ 1142 0.039 ----+ Density (CT) Bone Mineral Content

30

Chapter 3

Family-Based Linkage Analysis of Cardiometabolic Traits in the Diabetes Heart Study

Laura M. Raffield, Jacklyn N. Hellwege, Amanda J. Cox, Carl D. Langefeld, Barry I. Freedman, J. Jeffrey Carr, Jianzhao Xu, Nicholette D. Palmer, Donald W. Bowden

31

Introduction

Family-based linkage methods, which assess the cosegregation of a genetic marker and a trait of interest in families, have long been used to successfully identify high impact variants in

Mendelian disease, but efforts to apply these methods to common, complex disease have been less successful. While some notable successes, such as discovery of nucleotide-binding oligomerization domain containing 2 (NOD2) variants associated with Crohn’s disease risk

(Hugot, Chamaillard et al. 2001), were achieved with linkage methods, genome-wide association study (GWAS) methods largely supplanted linkage methods in the search for variants associated with common, complex disease. These GWAS analyses discovered many variants reproducibly associated with risk of common disease; however, most identified variants have small effect sizes and account for little of the heritable component of diseases and quantitative traits (Manolio,

Collins et al. 2009). Current efforts, facilitated by data from next-generation exome sequencing, to identify high effect, low frequency coding variants have renewed interest in linkage methods.

Recent analyses from the Insulin Resistance and Atherosclerosis Family Study (IRASFS) have highlighted several variants using family-based linkage methods, including a novel high impact, low frequency (minor allele frequency (MAF) ~ 1%) coding variant in adiponectin

(ADIPOQ) with adiponectin levels (Bowden, An et al. 2010) and more common variants in cholesteryl ester transfer protein (CETP) and apolipoprotein E (APOE), replicating the previously observed importance of these regions for lipid traits (Hellwege, Palmer et al. 2014). However,

IRASFS has very large, multigenerational families (average family size of 14.2 for African

American participants and 15.7 for Hispanic participants), while the Diabetes Heart Study (DHS) recruited siblings and has much smaller pedigrees. Here, we examine whether two-point linkage analysis, aligned with single variant association analysis, might highlight coding variants and regions of interest from the DHS Illumina Infinium HumanExome BeadChip data (Exome Chip), designed to assay less common coding variants. This is similar to the approach pursued in

32

IRASFS (Hellwege, Palmer et al. 2014). As a supplement, two-point linkage and association analysis was also performed for the mainly noncoding, common variants from an Affymetrix

Genome-Wide Human SNP Array 5.0 (GWAS). 43 traits, including measures of vascular calcified plaque, dyslipidemia, inflammation, renal function, and adiposity, were analysed using both two-point linkage and association analysis.

Methods

Study Population

The DHS recruited siblings concordant for T2D without advanced renal insufficiency from outpatient internal medicine and endocrinology clinics and from the community in the region surrounding Winston-Salem, North Carolina from 1998 through 2005, with siblings unaffected by T2D also recruited when possible. The cohort has been described in detail previously (Bowden, Cox et al. 2010). Two-point linkage and association analyses include only

European American DHS participants with appropriate genotyping data, in total 1190 individuals from 468 families (1318 sibpairs) for the Exome Chip, with 1181 European American individuals included in the supplemental GWAS array analysis.

Genotyping

Quality control of the DHS Exome Chip and GWAS array data has been previously described (Cox, Ng et al. 2013; Cox, Hugenschmidt et al. 2014). Poor quality samples (for example with a low call rate, gender or sibling relationship errors, unintentional duplication) have already been excluded. SNPs with a low call rate (<95%) or a low Hardy-Weinberg equilibrium p-value (<1 x 10-6) were also excluded.

33

Statistical Analysis

Both two-point linkage and association analyses were performed using the Sequential

Oligogenic Linkage Analysis Routines (SOLAR) program (Almasy and Blangero 1998), adjusting for age, sex, and T2D status. Methods used were similar to those used in IRASFS

(Hellwege, Palmer et al. 2014). For a small percentage of variants, identity by descent (IBD) matrices did not calculate correctly, leaving 86,902 variants with linkage and association data for the Exome Chip, out of 88,369 polymorphic variants on chromosomes 1-22, and 344,365 variants with linkage and association data for the GWAS array, out of 364,084 variants on chromosomes

1-22. Linkage results are reported as logarithm of the odds or LOD scores; a LOD score of ≥3.0 is typically considered significant evidence for linkage. For the association analyses, for the Exome

Chip a Bonferroni-corrected p-value threshold is p<5.75 x 10-7 (0.05/86902 total variants), though this is overly stringent as it does not account for the correlation among the variants, and for the

GWAS chip a traditional threshold for genome-wide significance (p<5 x 10-8) was used.

Variables were transformed as necessary to approximate normality prior to association and linkage analysis. We excluded results from non-heritable traits (waist-hip ratio, subcutaneous and visceral adipose tissue volumes, and fasting glucose), traits with high residual kurtosis after transformation (C-reactive protein, plasminogen activator inhibitor-1, resistin, and interleukin 1 beta (IL-1B)), and traits that failed linkage or association analysis, likely due to small sample size

(liver attenuation, liver spleen ratio), from these preliminary results, leaving 43 traits that were successfully analysed using both two-point linkage and association analysis. However, 9 of these traits, mostly cytokines (insulin, adiponectin, tumor necrosis factor (TNF), leptin, interleukin 6

(IL6), interleukin 8 (IL8), monocyte chemoattractant protein-1 (MCP-1), pericardial adipose tissue, visceral to subcutaneous adipose ratio) had greatly reduced sample size (n≤520) and were considered separately.

34

Results

Demographic characteristics of the 1190 European-American individuals included in the

Exome Chip linkage and association analysis are included in Table 1, stratified by T2D affection status. In this mostly T2D-affected cohort, prevalence of obesity, subclinical CVD, and prior

CVD events is high.

Across the 34 traits with data in most DHS participants, 455 LOD scores over 3.0 were observed for the Exome Chip, with 41 LOD scores over 3.0 having an association p-value <0.05;

5285 LOD scores over 3.0 were observed for the GWAS array, with 506 LOD scores over 3.0 also having an association p-value <0.05. Opposed Manhattan plots for these 34 traits including both Exome Chip and GWAS array results, with evidence for linkage at a locus on the top and evidence for association at a locus on the bottom, are displayed in Figure 1. For the Exome Chip, the highest LOD score for these traits was a missense variant in absent in melanoma 1-like

(AIM1L) on chromosome 1 (rs10751735, Ser1294Asn, LOD= 6.98, minor allele frequency

(MAF)= 0.49) with estimated glomerular filtration rate (Modification of Diet in Renal Disease

(MDRD) equation). For the GWAS array, the highest LOD score was located in an intergenic region of chromosome 1 (rs4662124, LOD= 8.56, MAF=0.25) with estimated glomerular filtration rate (MDRD equation). We would note that some regions of linkage, for example on chromosome 3 with coronary artery calcification (rs9812284, LOD= 2.16, MAF=0.44), overlap with prior microsatellite linkage peaks observed in the DHS (Bowden, Rudock et al. 2006); however, direct comparison of results is difficult.

For the 9 traits, mostly cytokines, which were only available in 520 or fewer individuals,

578 LOD scores over 3.0 were observed for the Exome Chip, with 47 LOD scores over 3.0 having an association p-value <0.05; 8609 LOD scores over 3.0 were observed for the GWAS array, with 903 LOD scores over 3.0 also having an association p-value <0.05. Opposed

35

Manhattan plots for these 9 traits including both Exome Chip and GWAS array results, with evidence for linkage at a locus on the top and evidence for association at a locus on the bottom, are displayed in Figure 2. For the Exome Chip, the highest LOD score for these traits was a missense variant in transmembrane protein 99 (TMEM99), located on chromosome 17, with TNF

(rs1044806, Leu95Arg, LOD=8.00, MAF=0.19). For the GWAS array, the highest LOD score was located on chromosome 7 in an intron of maturin, neural progenitor differentiation regulator homolog (MTURN) (rs7780166, LOD=11.42, MAF = 0.497) with TNF.

Combining linkage and association analysis for both the Exome Chip (Table 2) and

GWAS (Table 3) for our 34 top traits highlighted some potentially interesting loci, with the most significantly associated loci (p≤ 1 x 10-5) with a LOD score >1.0 and a trait-specific minor allele count ≥5 displayed. For example, the Exome Chip analyses highlighted variants from the UDP glucuronosyltransferase 1 family, polypeptide A complex locus (UGT1A) gene cluster with evidence for linkage (1.18 ≤ LOD ≤ 1.74) and association (3.30 x 10-20 ≤ p ≤ 4.80 x 10-13) with bilirubin levels. A variant in cadherin 13 (CDH13) (rs72807847, Asn39Ser, MAF= 0.01) with evidence for both association (p= 6.23 x 10-6) and linkage (LOD=2.62) with abdominal aortic calcification was also observed. Results for the 9 traits with reduced sample size are similarly displayed in Table 4 for the Exome Chip and Table 5 for the GWAS array.

Conclusions

Few definitive conclusions can be made from these preliminary, unreplicated linkage results. However, these preliminary results highlight that a combined linkage and association analysis approach may be effective for identifying loci, including loci for late onset diseases such as cardiometabolic traits, in sibpair studies without parental data. This is of particular interest for current analyses of low frequency coding variants, for which power in association analyses is limited even in large cohorts (Bansal, Libiger et al. 2010; Kiezun, Garimella et al. 2012). A few

36

interesting results from these DHS analyses provide support for the potential utility of a combined linkage and association approach. For example, from the Exome Chip results (Tables 2 and 4), the top associations for variants that also have some evidence of linkage (LOD >1.0) are in the

UGT1A gene cluster. Association of variants in the UGT1A gene cluster with bilirubin levels in the DHS and other cohorts has been previously reported (Cox, Ng et al. 2013), but linkage analysis of the Exome Chip provides additional support for the role of this locus. The linked and associated variant in CDH13 (rs72807847) is also of interest due to the functional roles of

CDH13. CDH13 has previously been associated with blood pressure, adiponectin levels, atherosclerotic progression, and risk of myocardial infarction (MI) (Takeuchi, Adachi et al. 2007;

Org, Eyheramendy et al. 2009; Chung, Lin et al. 2011; Shia, Ku et al. 2011), supporting a role for this locus in abdominal aortic calcification.

The linkage and association analyses results for our GWAS chip data provide less support for interesting loci taken independently. None of the top loci with evidence of linkage and association (Tables 3and 5) have readily apparent functional roles. Association analyses are more likely to have already highlighted important loci of interest among these more common variants, for which power is much greater than for rare single variant analyses and many large GWAS analyses have already been conducted (Bansal, Libiger et al. 2010; Kiezun, Garimella et al.

2012). Combining the linkage and association results from the GWAS and Exome Chip arrays

(Figure 1A-HH, Figure 2A-I) may be of more utility to identify regions of interest where the causal variant has not been typed. A recent analysis of the adiponectin locus from our group showed that two-point linkage using SNP array data outperformed association data in identifying the location of a low frequency, high impact variant (rs200573126) when the variant in question was not genotyped (Hellwege, Palmer et al. 2015). Some of the regions with high LOD scores but little evidence of association may contain important variants that were not typed on the Exome

Chip or GWAS arrays. We plan to pursue replication of these preliminary results from the DHS

37

in family studies from the Cohorts for Heart and Aging Research in Genomic Epidemiology

(CHARGE) consortium, which focuses on many phenotypes relevant to T2D, obesity, and CVD

(Psaty, O'Donnell et al. 2009).

38

Table 1: Demographic characteristics of the Diabetes Heart Study population stratified by type 2 diabetes (T2D) affected status.

T2D-affected (n=996) T2D-unaffected (n=194)

Phenotype Mean (SD) or % Mean (SD) or % Gender (% female) 51.5% 62.4% Age (yrs) 62.5 (9.1) 60.1 (10.3) Duration of diabetes (yrs) 10.5 (7.2) - Body mass index (kg/m2) 32.3 (6.6) 28.9 (5.1) Glycated hemoglobin (%) 7.6 (1.7) 5.6 (0.5) Fasting glucose (mg/dL) 147.9 (56.1) 93.9 (11.4) Coronary artery calcification (mass score) 1878.9 (3365.3) 639.5 (1568.5) Self-reported history of prior 43.5% 22.5% cardiovascular disease (%)

39

Table 2: Top association analysis results (p≤ 1 x 10-5) from the Exome Chip with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait-specific minor allele count ≥ 5 for the 34 cardiometabolic traits available in most Diabetes Heart Study (DHS) participants.

Chr Pos Variant Gene Annotation Associated DHS Trait- LOD β value p-value trait Specific Allele Counts 2 234668570 rs887829 UGT1A intronic Bilirubin 107/465/477 1.18 0.07 3.30 x 10-20

2 234673309 rs4148325 UGT1A intronic Bilirubin 105/464/480 1.44 0.07 4.44 x 10-20

2 234672639 rs6742078 UGT1A intronic Bilirubin 105/464/480 1.44 0.07 4.44 x 10-20

1 33237358 rs200947771 KIAA1522 P860S PR interval 0/7/1122 1.12 -0.69 6.57 x 10-17

2 234602202 rs1105879 UGT1A R184S Bilirubin 119/480/450 1.74 0.05 1.21 x 10-14

2 234601669 rs6759892 UGT1A S7A Bilirubin 186/497/366 1.56 0.05 4.80 x 10-13

3 170110147 rs61761943 SKIL M646K PR interval 0/21/1108 1.04 -0.33 2.48 x 10-11

6 159465977 rs1738074 TAGAP 5’ UTR Triglycerides 185/576/409 1.58 -0.11 2.30 x 10-6

11 85961371 rs149126431 EED T50P QT interval 0/5/1140 1.22 -0.19 5.34 x 10-6

16 82892037 rs72807847 CDH13 N39S Abdominal 1/8/844 2.62 77.05 6.23 x 10-6 Aortic Calcification

40

Table 3: Top association analysis results (p≤ 1 x 10-5) from the genome-wide association study array with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait-specific minor allele count ≥ 5 for the 34 cardiometabolic traits available in most Diabetes Heart Study (DHS) participants.

DHS Trait- Associated β Chr Pos Variant Gene Annotation Specific Allele LOD p-value trait value Counts Systolic ONECUT1 15 53172692 rs2440331 intergenic Blood 5/167/1005 1.62 7.08 3.75 x 10-6 Pressure Appendicular CTTNBP2 7 117542907 rs7790040 intergenic Bone Mineral 51/316/472 1.60 -0.80 5.07 x 10-6 Content Serum 4 182775312 rs6840589 intergenic intergenic 0/60/1118 1.00 -0.07 5.43 x 10-6 Creatinine CTNNA3 10 68224205 rs10822834 intronic Bilirubin 1/77/963 1.09 0.08 6.38 x 10-6 Abdominal AFF3 2 100502506 rs17436761 intronic Aortic 0/54/768 1.77 36.41 8.41 x 10-6 Calcification

41

Table 4: Top association analysis results (p≤ 1 x 10-5) from the Exome Chip with some evidence of linkage (logarithm of the odds (LOD) >1.0) and a trait-specific minor allele count ≥ 5 for the 9 traits available in 520 individuals from the Diabetes Heart Study (DHS) or less.

Chr Pos Variant Gene Annotation Associated DHS Trait- LOD β p-value trait Specific Allele value Counts 11 73850653 rs149910292 C2CD3 P235L Tumor 0/6/475 3.13 1.43 1.03 x 10-8 necrosis factor 6 121401996 rs56300302 C6orf170 E1232A Tumor 0/7/474 2.65 1.13 4.99 x 10-6 necrosis factor 6 121452897 rs150710532 C6orf170 D925D Tumor 0/7/474 2.65 1.13 4.99 x 10-6 necrosis factor 8 8235191 rs201219172 SGK223 S243L Interleukin 0/6/480 1.22 1.51 5.43 x 10-6 8

42

Table 5: Top association analysis results (p≤ 1 x 10-5) from the genome-wide association study array with some evidence for linkage (logarithm of the odds (LOD) >1.0) and a trait-specific minor allele count ≥ 5 for the 9 traits available in 520 individuals from the Diabetes Heart Study (DHS) or less.

DHS Trait- Associated Specific β Chr Pos Variant Gene Annotation LOD p-value trait Allele value Counts CEP112 Pericardial 17 63764890 rs7220733 intronic 70/225/207 1.03 0.05 5.57 x 10-6 Adiposity ZNF521 18 22777081 rs4468712 intronic Interleukin 8 88/245/147 3.37 -0.23 7.19 x 10-6

43

Figure 1: Opposed Manhattan plots including both GWAS and Exome Chip two-point linkage (top, logaritihm of the odds score) and association analysis (bottom, negative logarithm of the p- value) results, for traits available in most of the Diabetes Heart Study: 1A. abdominal aortic calcification, 1B. albumin-to-creatinine ratio, 1C. alkaline phosphatase, 1D. appendicular bone mineral content, 1E. bilirubin, 1F. biventricular volume, 1G. body mass index, 1H. carotid artery calcification, 1I. coronary artery calcification, 1J. diastolic blood pressure, 1K. estimated glomerular filtration rate (extended Modification of Diet in Renal Disease (MDRD) equation), 1L. estimated glomerular filtration rate (Modification of Diet in Renal Disease (MDRD) equation), 1M. glycated hemoglobin, 1N. high-density lipoprotein cholesterol, 1O. intima media thickness, 1P. low-density lipoprotein cholesterol, 1Q. lumbar bone mineral density, 1R. PR interval, 1S. pulse pressure, 1T. QRS interval, 1U. QT interval, 1V. serum albumin, 1W. serum creatinine, 1X. serum glutamic oxaloacetic transaminase, 1Y. serum glutamic pyruvic transaminase, 1Z. systolic blood pressure, 1AA. thoracic bone mineral density, 1BB. total cholesterol, 1CC. triglycerides, 1DD. vascular calcification three bed score, 1EE. waist circumference, 1FF. whole body bone mineral content, 1GG. whole body bone mineral density, 1HH. whole body percent fat.

Figure 1A.

44

Figure 1B.

Figure 1C.

Figure 1D.

45

Figure 1E.

Figure 1F.

Figure 1G.

46

Figure 1H.

Figure 1I.

Figure 1J.

47

Figure 1K.

Figure 1L.

Figure 1M.

48

Figure 1N.

Figure 1O.

Figure 1P.

49

Figure 1Q.

Figure 1R.

Figure 1S.

50

Figure 1T.

Figure 1U.

Figure 1V.

51

Figure 1W.

Figure 1X.

Figure 1Y.

52

Figure 1Z.

Figure 1AA.

Figure 1BB.

53

Figure 1CC.

Figure 1DD.

Figure 1EE.

54

Figure 1FF.

Figure 1GG.

Figure 1HH.

55

Figure 2: Opposed Manhattan plots including both GWAS and Exome Chip two-point linkage (top, logaritihm of the odds score) and association analysis (bottom, negative logarithm of the p- value) results, for traits represented by no more than 520 individuals: 2A. adiponectin, 2B. insulin, 2C. interleukin 6, 2D. interleukin 8, 2E. leptin, 2F. monocyte chemoattractant protein-1, 2G. pericardial adipose volume, 2H. tumor necrosis factor, 2I. visceral to subcutaneous adipose tissue ratio.

Figure 2A.

Figure 2B.

56

Figure 2C.

Figure 2D.

Figure 2E.

57

Figure 2F.

Figure 2G.

Figure 2H.

58

Figure 2I.

59

Chapter 4

Heritability and Genetic Association Analysis of Neuroimaging Measures in the Diabetes Heart Study

Laura M Raffield, Amanda J Cox, Christina E Hugenschmidt, Barry I Freedman, Carl D Langefeld, Jeff D Williamson, Fang-Chi Hsu, Joseph A Maldjian, Donald W Bowden

This manuscript was published in the November 2014 issue of Neurobiology of Aging. The reference for this manuscript is as follows: Raffield, L. M., Cox, A.J., Hugenschmidt, C.E., Freedman, B.I., Langefeld, C.D., Williamson, J.D., Hsu, F., Maldjian, J.A., Bowden, D.W. Heritability and genetic association analysis of neuroimaging measures in the Diabetes Heart Study. Neurobiology of Aging, 2014. S0197-4580(14):00724-6. doi: 10.1016/j.neurobiolaging.2014.11.008. Epub 2014 Nov 20. PubMed PMID: 25523635; PubMed Central PMCID: PMC4346514.

60

ABSTRACT

Patients with type 2 diabetes are at increased risk of age-related cognitive decline and dementia. Neuroimaging measures such as white matter lesion volume, brain volume, and fractional anisotropy may reflect the pathogenesis of these cognitive declines, and genetic factors may contribute to variability in these measures. This study examined multiple neuroimaging measures in 465 participants from 238 families with extensive genotype data in the type 2 diabetes enriched Diabetes Heart Study-Mind cohort. Heritability of these phenotypes and their association with candidate single nucleotide polymorphisms (SNPs) and SNP data from genome- and exome-wide arrays was explored. All neuroimaging measures analysed were significantly heritable ( hˆ 2 =0.55-0.99 in unadjusted models). Seventeen candidate SNPs (from 16 genes/regions) associated with neuroimaging phenotypes in prior studies showed no significant evidence of association. A missense variant (rs150706952, A432V) in PLEKHG4B from the exome-wide array was significantly associated with white matter mean diffusivity (p=3.66x10-7) and gray matter mean diffusivity (p=2.14x10-7). This analysis suggests genetic factors contribute to variation in neuroimaging measures in a population enriched for metabolic disease and other associated comorbidities.

KEYWORDS

Magnetic resonance imaging, type 2 diabetes, genetics, heritability

61

1. INTRODUCTION

Prior research has revealed that type 2 diabetes (T2D) accelerates age-related cognitive decline and increases risk of overt dementia (Reijmer, van den Berg et al. 2010). A number of studies have investigated neuroimaging phenotypes using magnetic resonance imaging (MRI) in individuals with T2D as a way of assessing the pathogenesis of these cognitive declines. Prior studies have reported an increased risk of white matter lesions, which are associated with increased risk of cognitive decline and stroke, as well as reduced total brain volume (TBV) in patients with T2D when compared to non-diabetic controls, but these studies have in some cases produced conflicting results, and many are based on relatively limited sample sizes (van Harten, de Leeuw et al. 2006; Jongen and Biessels 2008; Fornage, Debette et al. 2011; Falvey, Rosano et al. 2013; Moran, Phan et al. 2013). T2D has also been associated with reduced white matter fractional anisotropy (WMFA), a measure of the directionality of water molecule diffusion used to assess brain microstructure (Nucifora, Verma et al. 2007; Falvey, Rosano et al. 2013). A number of factors may influence these neuroimaging phenotypes in individuals with T2D, including hypertension (Schmidt, Launer et al. 2004), poor glycemic control (van Elderen, de

Roos et al. 2010), and adiposity (Verstynen 2013); however, few studies have examined the potential influences of genetic risk factors on these neuroimaging measures in individuals with

T2D.

Prior estimates of the heritability of a variety of neuroimaging phenotypes, including TBV, gray and white matter volume (GMV, WMV) (Peper, Brouwer et al. 2007; Blokland, de

Zubicaray et al. 2012), WMFA (Kochunov, Glahn et al. 2010), and total white matter lesion volume (WMLV) (Carmelli, DeCarli et al. 1998; Atwood, Wolf et al. 2004; Turner, Jack et al.

2004), have been high, increasing interest in genetic analysis of these measures. Heritable, quantitative neuroimaging measures are thought to be important endophenotypes for the analysis of genetic contributions to risk of cognitive decline and dementia, increasing power to detect

62

genetic contributions to these complex clinical traits (Gottesman and Gould 2003; Ge, Feng et al.

2012). Multiple variants, including putatively functional coding variants in candidate genes, such as the BDNF V66M and the COMT V158M polymorphisms, have been reported previously to be associated with neuroimaging measures in cohorts not enriched for T2D patients (Honea,

Verchinski et al. 2009; Chiang, Barysheva et al. 2011), though these and other associations have proved difficult to replicate in additional studies(Barnes, Isohanni et al. 2012; Lopez, Bastin et al.

2012). A limited number of genome-wide association studies (GWAS) of neuroimaging measures have also been performed, including analyses of WMFA (Lopez, Bastin et al. 2012), TBV

(Furney, Simmons et al. 2011; Stein, Medland et al. 2012), and white matter lesions (Fornage,

Debette et al. 2011). However, to our knowledge, no prior studies have focused on a cohort enriched for T2D, a high risk population for both cognitive decline and dementia.

The Diabetes Heart Study (DHS) is a family-based study of individuals with T2D designed to assess potential genetic and epidemiological risk factors for cardiovascular disease (CVD) in individuals with T2D. The DHS-Mind ancillary study to DHS performed cognitive testing and neuroimaging on 465 individuals from the original DHS cohort. This cohort provides a unique resource for examining genetic risk factors which may contribute to neuroimaging phenotypes of interest in a cohort enriched for T2D. In this study we evaluated the DHS-Mind neuroimaging dataset for heritability estimation and for associations with the comprehensive genetic data also available in the DHS. This included analysis of candidate SNPs from previously reported MRI- based neuroimaging studies and an exploratory, unbiased GWAS using data from both a traditional genome-wide array, designed to assay common variation across the genome, and an array enriched for exonic variants.

63

2. METHODS

2.1. Study Design and Sample

Participants in the DHS were recruited from outpatient internal medicine and endocrinology clinics and from the community from 1998 through 2005 in western North

Carolina. Siblings concordant for T2D without advanced renal insufficiency were recruited, with additional non-diabetic siblings enrolled whenever possible. Ascertainment and recruitment have been described in detail previously (Wagenknecht, Bowden et al. 2001; Lange, Bowden et al.

2002; Bowden, Rudock et al. 2006; Bowden, Cox et al. 2010). T2D was defined as diabetes developing after the age of 35 years treated with changes in diet and exercise and/or oral agents in the absence of initial treatment solely with insulin and without historical evidence of ketoacidosis.

Diabetes diagnosis was confirmed by measurement of fasting glucose and glycated hemoglobin

(HbA1C) at the exam visit. Extensive measurements of CVD risk factors were obtained during baseline exams from 1998-2006.

The DHS-Mind study is an ancillary study to the DHS initiated in 2008 that performed cognitive testing and neuroimaging to investigate risk factors for cognitive decline in a cohort enriched for T2D. Participants returning from the original DHS investigation were re-examined on average 6.7 ± 1.5 years after their initial visit. Participant examinations were conducted in the

General Clinical Research Center of the Wake Forest Baptist Medical Center. The current analyses are based on a subset of 465 participants (from 238 families) returning from the baseline

DHS exam with neuroimaging phenotypes from the DHS-Mind study visit and available genome- wide SNP genotype data. Subjects were not excluded for Modified Mini-Mental State

Examination (3MSE) scores or other indices of cognitive function indicative of mild cognitive impairment or dementia (Teng and Chui 1987).

64

Study protocols were approved by the Institutional Review Board at Wake Forest School of Medicine, and all study procedures were completed in accordance with the Declaration of

Helsinki. All participants provided written informed consent prior to participation.

2.2. Neuroimaging

Magnetic resonance (MR) image acquisition. MR imaging was performed on a 1.5-T GE

EXCITE HD scanner with twin-speed gradients using a neurovascular head coil (GE Healthcare,

Milwaukee, WI). High-resolution T1 anatomic images were obtained using a 3D volumetric

Inversion Recovery SPGR sequence (TR=7.36 ms; TE=2.02 ms; TI=600 ms; FA=20 degrees; 124 slices, FOV=24 cm, matrix size = 256x256, 1.5 mm slice thickness). Fluid-attenuated inversion recovery (FLAIR) images were acquired in the axial plane (TR=8002 ms; TE=101.29 ms;

TI=2000 ms; FA=90 degrees; FOV=24 cm; matrix size = 256 × 256; 3-mm slice thickness).

Whole brain diffusion tensor imaging (DTI) was performed using echo-planar imaging with 25 directions (TR=16000; TE=84.9; FA=90; b value = 0/1000, FOV = 280 cm, matrix size =

256x256, 3 mm slice thickness). Quantitative cerebral blood flow maps were generated using a

Q2TIPS-FAIR sequence as previously described (Luh, Wong et al. 1999). This sequence generates 60 tag and control image pairs. Imaging parameters are as follows: echo time 28ms,

TI1-800ms, TI1s 1200ms, TI 2000ms, TR 3000ms, receiver bandwidth 62.5 kHz, flip angle 90 degrees, field of view 24 cm (frequency) x 18 cm (phase), an acquisition matrix 64x48 (11 slices,

8 mm thickness, 0mm slice gap), and frequency encoding direction anterior/posterior. A bipolar diffusion gradient with an equivalent b value of 5.25 mm2/sec was added to suppress intra-arterial spins (Yang, Frank et al. 1998).

Image segmentation. Structural T1 images were segmented into gray matter (GM), white matter (WM) and cerebrospinal fluid (CSF), normalized to Montreal Neurologic Imaging (MNI) space, and modulated with the Jacobian determinants (non-linear components only) of the warping procedure to generate volumetric tissue maps using the Dartel high-dimensional warping

65

and the SPM8 (Ashburner and Friston 2000) new segment procedure as implemented in the

VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm.html). Intracranial volume (ICV) (GM + WM

+ CSF), TBV (GM + WM), GMV (GM), and WMV (WM) were determined from the VBM8 automated segmentation procedure which outputs values for native space total GM, WM, and

CSF volumes. The normalized gray matter and white matter segmentation maps (without modulation) were binarized at a probability threshold of 0.5 to create segmentation masks for use in generating the tissue-specific measures of diffusion and cerebral blood flow.

Diffusion tensor processing. Diffusion tensor pre-processing was performed using FSL

(Jenkinson, Beckmann et al. 2012). Eddy current correction of the diffusion tensor images was performed using FSL dti_eddy by normalizing each image to the baseline (B0) image using the mutual information registration algorithm. The diffusion tensor was computed using the Camino software package (www.camino.org.uk). The resulting tensor images were converted to NIfTI symmetric positive orientation using the Diffusion Tensor Imaging ToolKit (DTI-TK)

(http://www.nitrc.org/projects/dtitk). DTI scalar metrics, including FA and mean diffusivity

(MD), were computed using DTI-TK. DTI scalars were normalized to MNI space by coregistering the MD image to the T1 structural data using SPM8, and then combining this transformation matrix with the parameters computed in the VBM8 normalization procedure, allowing derivation of the global mean diffusivity and fractional anisotropy measures analyzed here (WMMD, GMMD, WMFA, GMFA).

Cerebral blood flow processing. Perfusion images were generated using a previously described fully automated data processing pipeline (Maldjian, Laurienti et al. 2008). Quantitative data processing includes data cleaning (removal of individual images with noise spikes or severe motion artifact), realignment (separately for label and control images), spatial smoothing, and calculation of mean cerebral blood flow (CBF) maps; our methods and experience with this technique has been well documented (Deibler, Pollock et al. 2008; Deibler, Pollock et al. 2008;

66

Deibler, Pollock et al. 2008; Maldjian, Laurienti et al. 2008; Pollock, Deibler et al. 2008; Pollock,

Deibler et al. 2008; Pollock, Whitlow et al. 2008; Maldjian, Baer et al. 2009; Pollock, Deibler et al. 2009; Pollock, Tan et al. 2009; Pollock, Whitlow et al. 2009; Tan, Maldjian et al. 2009;

Pollock, Whitlow et al. 2011; McGehee, Pollock et al. 2012; Johnston, Zheng et al. 2013; Watts,

Whitlow et al. 2013). The CBF maps were normalized to MNI space by coregistering to the T1 structural data using SPM8, and then combining this transformation matrix with the parameters computed in the VBM8 normalization procedure, allowing derivation of the gray matter cerebral blood flow (GMCBF) measure analyzed.

White Matter Lesion (WML) Segmentation. WML segmentation was performed using the lesion segmentation toolbox (LST) (Schmidt, Gaser et al. 2012) for SPM8 at a threshold (k) of

0.25. We previously validated the LST for use in the DHS-Mind in a sample of 100 subjects against expert manual segmentation, as well as identifying the optimum threshold in this population (Maldjian, Whitlow et al. 2013). Normalization to MNI space was accomplished by coregistration with the structural T1 and applying the normalization parameters computed in the

VBM8 segmentation procedure. The total WMLV measure used in these analyses was determined by summing the binary lesion maps and multiplying by the voxel volume.

2.3. Genotyping

Genotyping in the DHS has been described in detail previously (Cox, Hugenschmidt et al. 2014). Candidate SNPs previously associated with neuroimaging measures in other cohorts were investigated to determine their relationship with the neuroimaging measures available in the

DHS. Genotype data for individual SNPs was obtained from several genetic datasets available in the DHS derived from (i) the Affymetrix Genome-wide Human SNP Array 5.0 (Affymetrix, CA,

USA) (GWAS set; predominately common variants), (ii) the Illumina Infinium Human Exome

Beadchip v1.0 (Illumina, CA, USA) (Exome set; predominately low-frequency and rare coding variants), and (iii) GWAS Imputed data (Imputed set) imputed from the 1,000 Genomes Project

67

SNPs using IMPUTE2 (http://mathgen.stats.ox.ac.uk/impute/impute_v2.html) and the Phase I v2, cosmopolitan (integrated) reference panel, build 37 (Howie, Donnelly et al. 2009). Genotype data for rs429358, in APOE, was not available from the array-based datasets and was directly genotyped using the MassARRAY SNP Genotyping System as described previously (Buetow,

Edmonson et al. 2001; Cox, Lehtinen et al. 2013).

For the GWAS set, exclusion criteria for SNP performance included a SNP call rate

<95% (n=11,085), Hardy-Weinberg Equilibrium p-value <1x10-6 (n=344), and minor allele frequency <0.01 (n=57,382); 371,951 SNPs were retained for analysis. For the imputed set, SNPs that were used for imputation were required to have low missingness (<5%) and show no significant departure from Hardy-Weinberg expectations (p>1x10-4). Only imputed SNPs with a confidence score >0.90 and information score >0.50 were used, with a total of ~4.5 million SNPs passing imputation quality control. For the Exome set, exclusion criteria for SNP performance included SNP call rate <99% (n=972), monomorphic SNPs (n=157,754) and Hardy-Weinberg

Equilibrium p-value <1x10-6 (n=26); 88,480 SNPs were retained for analysis.

2.4. Statistical Analyses

To determine the contribution of genetic factors to these neuroimaging phenotypes, heritability was estimated in family members using Sequential Oligogenic Linkage Analysis

Routines (SOLAR) version 6.5.8 (Texas Biomedical Research Institute, TX, USA). SOLAR performs a variance components analysis of family data where the total phenotypic variation is partitioned into genetic and non-genetic sources of variation. For both the heritability analysis and genetic association analysis, neuroimaging measures were transformed to approximate the normality assumptions of the analysis if necessary. The natural logarithm of TBV and (total

WMLV+1) and the square root of GMCBF were used. For the heritability analysis, if residual kurtosis remained, an inverse normal transformation was also used. This additional transformation was needed for GMCBF, WMMD, GMMD, and GMFA. The significance of the

68

heritability ( hˆ 2 ) estimates was obtained by likelihood ratio tests. Three models were developed that incorporated an increasing number of covariates to determine the extent that genetic factors contribute to variation in neuroimaging measures independent of other confounding variables.

The first model was an unadjusted model; the second model was adjusted for age and sex; the third model was adjusted for age, sex, and T2D affected status, with additional adjustment for

ICV for all WMLV, GMV, and WMV heritability analyses.

Targeted genetic association analyses were first performed for a set of 17 candidate

SNPs, and genome-wide discovery analyses were next performed using the entire GWAS and

Exome data sets. All single SNP association analyses were performed using variance components methods implemented in SOLAR version 6.4.1 (Texas Biomedical Research Institute, San

Antonio, TX) to account for relatedness between subjects (Almasy and Blangero 1998).

Association was examined assuming an additive model of inheritance. For single variant analyses for both the GWAS set and the Exome set, association results based on a single observation of the rare allele and all associations with X and Y chromosome variants were excluded from the results. Gene-based tests of polymorphic exonic variants from the Exome set were also performed using the sequence kernel association test (SKAT) program with default weights using minor allele frequency. SKAT is a variance components based test that aggregates weighted test statistics for all variants in a gene which is applicable to family data for continuous traits, incorporating a kinship matrix into the models (Chen, Meigs et al. 2013; Lee, Teslovich et al.

2013). Age, sex and T2D affected status were included as covariates in all analyses; ICV was also included as a covariate in analyses of WMLV, GMV, and WMV. For the candidate SNPs statistical significance was accepted at p<0.0029 based on a Bonferroni correction for 17 SNPs tested. For the discovery analyses, genome-wide significance was accepted at the conventional level of p<5x10-8 for analysis of the GWAS set, and significance was accepted at p<5.65x10-7 for

69

single variant analyses and at p<4.84x10-6 for the gene-based analyses for the Exome set based on a Bonferroni correction.

3. RESULTS

The goal of this study was to assess heritability and genetic associations for MRI-based neuroimaging measures in the T2D enriched DHS-Mind study. The clinical characteristics of the

465 individuals of European descent included in the study are summarized in Table 1. Risk factors including hypertension, high body mass and history of prior CVD events are prevalent, as might be expected in a T2D enriched cohort.

Heritability was first calculated for all neuroimaging phenotypes. All neuroimaging phenotypes analysed were significantly heritable ( hˆ 2 =0.55-0.99 in unadjusted models, 1.8 x 10-

15

A total of 17 candidate SNPs (from 16 genes/regions; Table 3) associated with neuroimaging phenotypes in prior studies were selected for targeted association analysis. Top associations from relevant GWAS studies were included in this analysis (Fornage, Debette et al.

2011; Furney, Simmons et al. 2011; Lopez, Bastin et al. 2012), but the focus was on potentially functional coding polymorphisms previously associated with neuroimaging phenotypes (Schmidt,

Schmidt et al. 2000; Schmidt, Schmidt et al. 2001; Kohara, Fujisawa et al. 2003; Godin, Tzourio et al. 2009; Honea, Verchinski et al. 2009; Penke, Muñoz Maniega et al. 2010; Smith, Chebrolu et al. 2010; Braskie, Jahanshad et al. 2011; Chiang, Barysheva et al. 2011; Sprooten, Sussmann et al. 2011; Jahanshad, Kohannim et al. 2012; Kohannim, Jahanshad et al. 2012; Kim, Kim et al.

2013). The selected candidate SNPs were not associated with any of the neuroimaging phenotypes assessed in the DHS-Mind cohort, with no SNP reaching a Bonferroni corrected p-

70

value threshold of p=0.0029. The strongest observed association was between rs903027, an intronic variant in CLVS1, and WMV (p=0.011). Nominal associations were also observed between rs1799945, a missense variant in HFE, and WMV (p=0.035) and rs1042714, a coding variant in the adrenoceptor beta 2 gene (ADRB2), and TBV (p=0.031).

Exploratory genome-wide analyses were performed using genotype data from the GWAS set, which contains mostly common, non-coding variants. No SNPs were associated with any of the neuroimaging phenotypes of interest at a traditional genome-wide significance threshold of

5x10-8; however, there were multiple loci with nominal evidence of association (Figures S1A-

S1I). The top 50 SNPs associated with each of the neuroimaging phenotypes assessed are included in Tables S1A-S1I. The most significant association was for rs10065017 with WMLV

(p=2.06x10-7); this intergenic SNP is in weak linkage disequilibrium (r2 >0.1) with Homo sapiens early B-cell factor 1 (EBF1). This association of the EBF1 locus with WMLV is also supported by an additional 12 SNPs (1.69x10-6

(3.30x10-5

5

5

BTB domain containing 16 (ZBTB16; three SNPs with GMMD (5.67x10-6

5

Additional analyses were also completed using the Exome set, an array-derived set of approximately 88,000 polymorphic variants, most of them uncommon, potentially functional coding variants. Manhattan plots for each neuroimaging phenotype (Figures S2A-S2I) reveal that few variants were associated with the traits of interest at a level consistent with the Exome set

71

array corrected p-value threshold (5.65x10-7, α=0.05). One missense variant (rs150706952,

A432V) in pleckstrin homology domain containing, family G (with RhoGef domain) member 4B

(PLEKHG4B) was significantly associated with WMMD (p=3.66x10-7) and GMMD (p=2.14x10-

7). Other variants with potentially interesting functions were also identified as nominally associated with the neuroimaging phenotypes, for example two missense variants in ZNF224

(rs2068061, M118V, and rs4239529, H162L) associated with WMLV (p=8.71x10-6). The top 50

SNPs from single variant analyses are displayed in Tables S2A-S2I.

Analysis of all genes with two or more polymorphic exonic variants (n=10,341 genes) in the Exome set was also performed using the SKAT program. The top 25 associated genes from

SKAT analyses are displayed in Tables S3A-S3I. Some functionally interesting genes were nominally significant in these analyses, for example MAS-related GPR, member E (MRGPRE) with WMMD (p=7.89x10-6), CCR4-NOT transcription complex, subunit 6 (CNOT6) with

GMCBF (p=2.98x10-5), and triggering receptor expressed on myeloid cells 2 (TREM2) with TBV

(3.91x10-4).

4. DISCUSSION

Given the acceleration of age-related cognitive decline and increased risk of overt dementia previously observed for individuals affected by T2D, determining which individuals with T2D are at higher risk for these comorbidities is of interest (Reijmer, van den Berg et al.

2010). Neuroimaging phenotypes may aid in assessing the pathogenesis and progress of these cognitive declines and may be helpful as endophenotypes for identifying individuals at increased genetic risk (Gottesman and Gould 2003; Ge, Feng et al. 2012). To this end, the current study pursued genetic analysis of multiple neuroimaging phenotypes in the T2D-enriched DHS-Mind sample. All neuroimaging measures assessed were significantly heritable in the DHS-Mind cohort. However, candidate SNPs previously associated with neuroimaging phenotypes were not

72

strongly replicated in the DHS-Mind, but our exploratory analyses of genome-wide genetic data revealed potentially interesting genetic risk loci that may influence traits including TBV, GMV,

WMV, WMLV, GMCBF, and FA and MD of the gray and white matter.

The family-based recruitment strategy for the DHS allowed us to assess the heritability of neuroimaging phenotypes in this cohort. A number of twin studies in the general population have previously calculated very high heritability for TBV (66-97%), with GMV and WMV also found to have estimated heritability greater than 70% (Peper, Brouwer et al. 2007; Blokland, de

Zubicaray et al. 2012). Heritability of WMLV has been estimated to be between 55-71% in twin and family studies (Carmelli, DeCarli et al. 1998; Atwood, Wolf et al. 2004; Turner, Jack et al.

2004). Similarly, heritability estimates for white matter integrity as assessed by FA are generally high, with reported heritability of 52% for global WMFA in a large family study (Kochunov,

Glahn et al. 2010), and high heritability of regional white matter FA also reported (Chiang,

Barysheva et al. 2009). The heritability of white matter MD and gray matter CBF, FA and MD has been less studied; one small twin study pointed to significant genetic influences on MD in specific brain regions during development (Chen, Zhu et al. 2009), and another study found significant heritability for regional differences in MD asymmetry (Jahanshad, Lee et al. 2010), but to our knowledge no large scale studies of the heritability of global measures of MD, GMCBF and GMFA have been performed. However, these measures are correlated with more commonly examined neuroimaging measures, and we had hypothesized that high heritability estimates would be observed for these measures as well.

Consistent with this prior literature, heritability estimates were quite high for all neuroimaging measures assessed, with estimates remaining significant upon adjustment for the potential confounders of age, sex, and T2D affected status. Some heritability estimates, such as the estimate of 1 for TBV in models adjusted for age, sex, and T2D, are most likely inflated due to model instability at high heritability levels; however, this does not impact our essential

73

conclusion that all neuroimaging measures assessed had a significant heritable component.

Metabolic disease and associated comorbidities present in a T2D enriched cohort did not appear to confound the heritable component of these neuroimaging phenotypes in DHS-Mind.

The 17 candidate polymorphisms analysed here were drawn from several studies, including prior GWAS analyses of TBV, white matter hyperintensities and WMFA (Fornage,

Debette et al. 2011; Furney, Simmons et al. 2011; Lopez, Bastin et al. 2012), but our focus was on putatively functional coding polymorphisms (Schmidt, Schmidt et al. 2000; Schmidt, Schmidt et al. 2001; Kohara, Fujisawa et al. 2003; Godin, Tzourio et al. 2009; Honea, Verchinski et al.

2009; Penke, Muñoz Maniega et al. 2010; Smith, Chebrolu et al. 2010; Braskie, Jahanshad et al.

2011; Chiang, Barysheva et al. 2011; Sprooten, Sussmann et al. 2011; Jahanshad, Kohannim et al.

2012; Kohannim, Jahanshad et al. 2012; Kim, Kim et al. 2013). In DHS-Mind, this set of candidate SNPs was not significantly associated with the neuroimaging phenotypes assessed.

Three nominal associations were observed. The strongest was with rs903027, an intronic variant in CLVS1, with WMV (p=0.011); this variant was selected due to its nominal association (p=6 x

10-6) with whole brain volume in a GWAS study of MRI atrophy measures (Furney, Simmons et al. 2011). Nominal association was also observed between a missense variant in HFE (rs1799945,

H63D), selected due to its prior associations with WM FA (Jahanshad, Kohannim et al. 2012), and WMV (p=0.035), and between a missense variant in ADRB2 (rs1042714, E27Q), selected due to previous associations with regional changes in WM FA, as well as differences in cognitive aging (Penke, Muñoz Maniega et al. 2010), and TBV (p=0.031). One candidate locus of note is

APOE; this locus has been associated with an elevated risk of Alzheimer’s disease (AD)

(Schipper 2011), the rate of age-related cognitive decline (De Jager, Shulman et al. 2012), cognition in the general population (Izaks, Gansevoort et al. 2011), and multiple neuroimaging phenotypes. For example, in individuals without cognitive impairment, APOE genotype has been associated with reduced WMFA (Smith, Chebrolu et al. 2010), regional differences in CBF (Kim,

74

Kim et al. 2013), and WMLV (Godin, Tzourio et al. 2009), though some of these associations are based on very small sample sizes. However, in our study, neither of the SNPs which indicate an individual’s APOE haplotype, rs7412 and rs429358, were associated with any of the neuroimaging phenotypes assessed, nor was there association with the APOE haplotype. The small sample sizes of many of the previous genetic association studies, publication bias, variability in neuroimaging methods and phenotypes, variability in cohort ascertainment and confounding by environmental factors may all be contributing to our inability to replicate the associations of these selected candidate polymorphisms with neuroimaging measures in the DHS-

Mind.

Exploratory analysis of genome-wide data from both the GWAS set, containing mostly common, non-coding variants, and the Exome set, containing mostly rare and low frequency coding variants, allowed us to further investigate potential genetic contributions to variation in neuroimaging phenotypes in T2D. While no variants from the GWAS set were associated with our neuroimaging measures at a traditional genome wide significance threshold (p=5x10-8), which is not unexpected given the sample size, nominal associations with several functionally interesting loci were identified. The most significantly associated SNP in these analyses implicated the EBF1 locus as associated with WMLV, an association supported by an additional

12 SNPs in the top 50 association signals for WMLV. The transcription factor EBF1 regulates the commitment of cells to a B cell lineage and B cell function; EBF1 also regulates cell differentiation and neural migration during mouse brain development (Garel, Marín et al. 1999;

Garel, Garcia-Dominguez et al. 2000; Hagman, Ramírez et al. 2012).

Several other loci supported by multiple SNPs in analyses of the GWAS set also have interesting functional roles with potential relevance to our neuroimaging phenotypes. NRG3, associated with TBV, has been implicated in neurodevelopment and in the function of the adult brain, with variants in the NRG3 locus previously associated with developmental delay, cognitive

75

impairment, autism, and schizophrenia(Kao, Wang et al. 2010). CDK8, implicated in GWAS analyses of WMFA and TBV, is an oncogenic transcriptional regulator which is involved in cell cycle progression, WNT/beta-catenin signalling and other important pathways (Galbraith, Donner et al. 2010); expression of CDK8 and its binding partner cyclin C is increased in the astrocytes of

AD patients versus controls (Ueberham, Hessel et al. 2003). ZBTB16, a zinc finger transcription factor expressed in the brain which is protective against glutamate toxicity in vitro in human neuronal cells and is downregulated in experimentally induced stroke (Seidel, Kirsch et al. 2011), was associated with GMMD. HMGCR, associated with GMCBF, is the rate-limiting enzyme for cholesterol synthesis; this locus has known impacts on total cholesterol and LDL cholesterol levels, which are risk factors for atherosclerosis and could thereby influence CBF (Claus, Breteler et al. 1998; Teslovich, Musunuru et al. 2010). Lastly, THSD7A, a protein which may promote endothelial cell migration and tube formation during neuroangiogenesis (Kuo, Wang et al. 2011), was associated with GMFA.

Analysis of the Exome set using both single variant and gene-based analyses also revealed interesting loci. A missense variant (rs150706952, A432V) in the PLEKHG4B locus was associated with WMMD and GMMD in the single variant analysis of the Exome set. Little is known about PLEKHG4B, but it is expressed in the adult and fetal brain (Kikuno, Nagase et al.

1999). A number of genes and variants which did not meet our strict significance thresholds also highlighted loci of potential interest in furthering understanding of brain structure and function.

For example, TREM2, associated with TBV, has putative antiinflammatory functions in the brain, and a missense variant in this gene (rs75932628, R47H) was reported to be compellingly associated with risk of AD, with an OR of 2.90 (Jonsson, Stefansson et al. 2013). MRGPRE, a locus associated with WMMD, is expressed in sensory neurons (Dong, Han et al. 2001) and was implicated in a recent GWAS study of brain lesions in multiple sclerosis (Gourraud, Sdika et al.

2013). Variants in CNOT6, a component of the CCR4-NOT complex which has been implicated

76

in regulation of the differentiation of neural stem cells, were associated with GMCBF (Chen, Ito et al. 2011). ZNF224, which contains two common missense variants associated with WMLV, is a brain-expressed transcriptional repressor with potential roles in regulating carbohydrate oxidation; another polymorphism in this gene (rs3746319, K640E) was previously associated with quantitative measures of global cognitive performance and AD pathology in the Religious

Orders Study (Shulman, Chibnik et al. 2010; Lupo, Cesaro et al. 2011). These findings in the

DHS-Mind support further genetic analysis of these loci in additional cohorts, in particularly other cohorts of patients affected by type 2 diabetes.

5. CONCLUSIONS

In conclusion, the current study observed high heritability of neuroimaging measures such as white matter lesion volume, brain volume, and fractional anisotropy in a T2D enriched cohort with significant metabolic and vascular disease. This study failed to replicate the association of previously identified candidate polymorphisms with neuroimaging measures such as TBV and WMLV. However, exploratory analyses using genome-wide sets of both common, mostly noncoding variants and less common coding variants revealed a number of biologically interesting loci that may be interesting targets for future analysis. While variants identified in these exploratory analyses were not replicated in additional T2D enriched cohorts, these results provide a useful starting point for future research and highlight loci to target in future analyses.

This study in DHS-Mind provided a unique opportunity to examine genetic contributions to neuroimaging measures in a T2D enriched population at elevated risk for cognitive decline and dementia.

77

Table 1: Demographic characteristics of the 465 DHS-Mind participants with genotyping data.

Mean  SD or % Median (range) Demographics Age (years) 67.54  8.92 67.56 (41.25 - 89.21)

Sex (% female) 57.2%

BMI (kg/m2) 30.93  5.87 30.11 (17.56 - 57.68)

% smoking (current or past) 53.9%

Hypertension (%) 88.4%

Self-reported history of prior CVD 29.5%

Type 2 Diabetes Type 2 diabetes affected (%) 74.4%

Diabetes duration (years) 16.37  6.60 14.23 (4.93 - 44.32)

Glucose (mg/dl) 131.97  47.94 119 (40 - 349)

Hemoglobin A1C (%) 7.06  1.33 6.8 (4.9 - 14.8)

Medications Anti-diabetic medication 63.8 %

Cholesterol-lowering medication 52.8 %

Anti-hypertensive medication 72.5 %

Education Less than high school 18.1%

High school 53.3%

Greater than high school 28.6%

Cognition and Neuroimaging Measures Modified Mini Mental State Exam (3MSE) 90.54  6.98 92 (62 - 100)

Gray matter volume (GMV) (cc) 509.8  51.0 507.2 (375- 660)

White matter volume (WMV) (cc) 575.6  70.9 569.7 (400- 815.1)

Total brain volume (TBV) (cc) 1085.4  112.6 1071.5 (775.0- 1447.9)

Intracranial volume (ICV) (cc) 1340.7  139.6 1326.9 (985-1758.8)

78

Fractional anisotropy gray matter (GMFA) 0.206  0.016 0.206 (0.151 - 0.255)

Fractional anisotropy white matter 0.356  0.022 0.357 (0.290 - 0.419) (WMFA)

Mean diffusivity gray matter (GMMD) 1.091  0.094 1.090 (0.800 - 1.376)

Mean diffusivity white matter (WMMD) 0.787  0.051 0.789 (0.630 - 0.951)

Total white matter lesion volume (WMLV) 4.29  6.82 1.62 (0- 59.57) (cc)

Gray matter cerebral blood flow (GMCBF) 47.08  19.35 44.15 (5.25 - 134.93) (mL/100g of tissue/min)

79

Table 2: Heritability estimates for MRI imaging variables in related individuals from the Diabetes Heart Study Cohort.

TBV GMV WMV GMFA WMFA GMMD WMMD WMLV GMCBF

Covariates

Noneϯ

hˆ 2 (SE) 0.82 (0.11) 0.79 (0.12) 0.63 (0.12) 0.57 (0.13) 0.61 (0.11) 0.99 (0.11) 0.96 (0.11) 0.71 (0.12) 0.55 (0.14)

p-value 1.9 x 10-15 2.8 x 10-10 3.7 x 10-9 2.0 x 10-7 1.1 x 10-9 1.8 x 10-14 1.8 x 10-15 9.3 x 10-10 2.5 x 10-5

Age, sexϯ

(SE) 1.00 0.67 (0.12) 0.62 (0.12) 0.49 (0.13) 0.64 (0.11) 0.68 (0.12) 0.84 (0.11) 0.61 (0.13) 0.35 (0.14)

p-value 2.4 x 10-25 2.7 x 10-9 1.3 x 10-8 9.6 x 10-6 1.0 x 10-10 9.1 x 10-9 1.4 x 10-13 2.0 x 10-7 3.2 x 10-3

Age, sex, T2Dϯ

(SE) 1.00 0.66 (0.12) 0.65 (0.12) 0.50 (0.13) 0.64 (0.11) 0.73 (0.12) 0.85 (0.11) 0.61 (0.13) 0.31 (0.14)

p-value 6.8 x 10-25 1.7 x 10-9 3.7 x 10-9 7.2 x 10-6 1.5 x 10-10 9.5 x 10-10 6.7 x 10-14 3.0 x 10-7 0.01

TBV=total brain volume; GMV=gray matter volume; WMV= white matter volume; GMFA= fractional anisotropy gray matter; WMFA= fractional anisotropy white matter; GMMD= mean diffusivity gray matter; WMMD= mean diffusivity white matter; WMLV = total white matter lesion volume; GMCBF= gray matter cerebral blood flow

80

Ϯ WMLV, GMV, and WMV heritability additionally adjusted for intracranial volume (ICV). Note that the natural logarithm of TBV and (total WMLV+1) and the square root of GMCBF were used, with an inverse normal transformation also used for GMCBF, WMMD, GMMD, and GMFA.

81

Table 3: Genetic association (assuming an additive model of inheritance) between candidate SNPs and neuroimaging measures.

DHS Analysis Association p-values (covariates: age, sex, T2D affected status)

Chr Pos SNP Gene Location Source Alleles MAF TBV WMV GMV GMFA WMFA GMMD WMMD WMLVϯ GMCBF Reference (maj/mi n) 1 11856378 rs1801133 MTHFR Exonic Exome G/A 0.333 0.356 0.661 0.840 0.976 0.236 0.328 0.155 0.346 0.823 (Kohara, Fujisawa et al. 2003)

1 48321221 rs946836 TRABD2B Intronic Imputed C/T 0.285 0.371 0.746 0.150 0.472 0.588 0.575 0.646 0.421 0.628 (Lopez, Bastin et al. 2012)

1 156848918 rs6336 NTRK1 Exonic Exome G/A 0.076 0.422 0.318 0.862 0.315 0.113 0.380 0.326 0.923 0.717 (Kohannim, Jahanshad et al. 2012)

1 230845794 rs699 AGT Exonic Exome A/G 0.396 0.594 0.886 0.758 0.239 0.288 0.429 0.414 0.965 0.946 (Schmidt, Schmidt et al. 2001)

1 232144598 rs821616 DISC1 Exonic Exome A/T 0.273 0.701 0.948 0.435 0.158 0.785 0.524 0.626 0.195 0.408 (Sprooten, Sussmann et al. 2011)

3 126993099 rs9871760 C3orf56 Downstr Exome C/A 0.241 0.868 0.823 0.835 0.233 0.567 0.509 0.837 0.224 0.440 (Furney, Simmons eam et al. 2011)

5 148206473 rs1042714 ADRB2 Exonic Exome G/C 0.413 0.031 0.559 0.180 0.937 0.878 0.239 0.336 0.848 0.095 (Penke, Muñoz Maniega et al. 2010)

6 26091179 rs1799945 HFE Exonic Exome C/G 0.169 0.838 0.035 0.630 0.988 0.575 0.728 0.422 0.298 0.773 (Jahanshad, Kohannim et al. 2012)

7 94946084 rs854560 PON1 Exonic Exome A/T 0.370 0.085 0.335 0.559 0.959 0.997 0.563 0.550 0.361 0.257 (Schmidt, Schmidt et al. 2000)

8 27464519 rs11136000 CLU Intronic Exome G/A 0.412 0.119 0.825 0.902 0.132 0.415 0.911 0.962 0.294 0.617 (Braskie, Jahanshad et al. 2011)

8 62409428 rs903027 CLVS1 Intronic Exome A/C 0.169 0.837 0.011 0.164 0.641 0.404 0.555 0.093 0.734 0.536 (Furney, Simmons et al. 2011)

10 17142526 rs6602175 CUBN Intronic Exome A/C 0.269 0.177 0.815 0.229 0.984 0.804 0.735 0.998 0.609 0.889 (Furney, Simmons et al. 2011)

82

11 27679916 rs6265 BDNF Exonic Exome G/A 0.201 0.423 0.745 0.553 0.674 0.636 0.169 0.305 0.575 0.478 (Chiang, Barysheva et al. 2011)

17 73872948 rs1055129 TRIM47 Intronic Exome A/G 0.281 0.386 0.202 0.816 0.415 0.726 0.835 0.318 0.264 0.178 (Fornage, Debette et al. 2011)

19 45411941 rs429358 APOE Exonic Genotyped T/C 0.047 0.842 0.661 0.891 0.831 0.140 0.991 0.355 0.892 0.824 (Godin, Tzourio et al. 2009; Smith, Chebrolu et al. 2010; Kim, Kim et al. 2013)

19 45412079 rs7412 APOE Exonic Exome G/A 0.080 0.430 0.728 0.536 0.169 0.148 0.803 0.672 0.494 0.811 (Godin, Tzourio et al. 2009; Smith, Chebrolu et al. 2010; Kim, Kim et al. 2013)

APOE 0.560 0.518 0.724 0.556 0.097 0.868 0.339 0.926 0.812 (Godin, Tzourio et haplotype al. 2009; Smith, Chebrolu et al. 2010; Kim, Kim et al. 2013)

22 19951271 rs4680 COMT Exonic Exome G/A 0.473 0.853 0.721 0.957 0.488 0.485 0.990 0.551 0.507 0.190 (Honea, Verchinski et al. 2009) TBV=total brain volume; GMV=gray matter volume; WMV= white matter volume; GMFA= fractional anisotropy gray matter; WMFA= fractional anisotropy white matter; GMMD= mean diffusivity gray matter; WMMD= mean diffusivity white matter; WMLV = total white matter lesion volume; GMCBF= gray matter cerebral blood flow

Ϯ WMLV, GMV, WMV analyses additionally adjusted for intracranial volume (ICV).

83

Table S1A: Association statistics for the top 50 SNPs associated with TBV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

4 rs17049741 138582711 PCDH18 intergenic G A 1.000 0.169 1.10E-05 -0.0300 0.4973 461

18 rs11081641 76417960 intergenic intergenic T C 0.993 0.043 1.50E-05 -0.0479 0.0111 458

10 rs12221382 84880520 NRG3 intergenic C A 1.000 0.115 2.10E-05 0.0334 0.1013 461

10 rs7904644 84910240 NRG3 intergenic T C 1.000 0.115 2.10E-05 0.0334 0.1013 461

7 rs4722097 22150888 RAPGEF5 intergenic A C 0.998 0.334 3.30E-05 -0.0218 0.0052 460

10 rs7902171 84789207 NRG3 intergenic G A 0.998 0.122 3.60E-05 0.0307 0.0074 460

4 rs17594509 45307178 PRKRIRP9 intergenic G A 0.991 0.345 4.30E-05 0.0201 0.2379 457

20 rs17000105 53350318 DOK5 intergenic T C 0.972 0.022 5.80E-05 -0.0666 0.0165 448

21 rs2824903 19902476 TMPRSS15 intergenic T A 0.998 0.025 5.90E-05 0.0681 0.0168 460

5 rs4701995 13868537 DNAH5 intronic G A 1.000 0.427 6.30E-05 -0.0197 0.0058 461

10 rs11197306 84860188 NRG3 intergenic C A 0.983 0.125 7.00E-05 0.0269 4.0607 453

4 rs4832855 35930651 ARAP2 intergenic T C 0.998 0.097 7.40E-05 -0.0336 0.0084 460

4 rs6531397 35931559 ARAP2 intergenic G A 1.000 0.097 7.50E-05 -0.0337 0.2408 461

18 rs11877812 67976542 SOCS6 intronic G A 0.972 0.028 8.80E-05 -0.0584 0.0167 448

84

5 rs4443379 68262453 AK128486 intergenic C T 1.000 0.173 9.40E-05 -0.0236 0.0298 461

18 rs656602 8821301 SOGA2 intronic C T 1.000 0.475 9.40E-05 -0.0164 0.7768 461

2 rs4850595 196064399 intergenic intergenic T C 0.991 0.126 9.80E-05 -0.0307 0.0087 457

6 rs12205822 144857747 UTRN intronic G C 1.000 0.355 1.00E-04 -0.0183 1.3051 461

22 rs5995786 40264551 ENTHD1 intronic C T 1.000 0.241 1.17E-04 0.0232 0.0039 461

8 rs2090788 19079894 LOC100128993 intronic C A 0.957 0.382 1.20E-04 -0.0215 0.0057 441

18 rs8182385 70039124 CBLN2 intergenic T G 1.000 0.439 1.25E-04 0.0197 0.0047 461

13 rs1175705 26898489 CDK8 intronic A G 1.000 0.081 1.26E-04 -0.0374 0.0102 461

2 rs336033 38263808 RMDN2 intronic G A 0.998 0.348 1.32E-04 -0.0203 0.0053 460

18 rs7228636 9871426 TXNDC2 intergenic C T 0.996 0.061 1.54E-04 0.0418 0.0110 459

5 rs17677148 4399737 intergenic intergenic G A 1.000 0.310 1.64E-04 -0.0205 0.0054 461

12 rs11067417 115552482 intergenic intergenic T G 1.000 0.091 1.74E-04 -0.0317 2.1503 461

2 rs2942884 101323578 NPAS2 intergenic G A 1.000 0.478 1.77E-04 -0.0187 0.8664 461

12 rs1240271 70289546 MYRFL intronic G A 0.996 0.218 1.97E-04 -0.0227 0.0061 459

3 rs1357544 166457917 CBX1P5 intergenic G A 1.000 0.021 2.08E-04 -0.0652 0.0178 461

8 rs10958128 83666511 intergenic intergenic G C 0.993 0.133 2.11E-04 0.0255 4.6556 458

5 rs11743589 13869398 DNAH5 intronic T C 1.000 0.427 2.16E-04 -0.0160 1.3403 461

6 rs9399478 144851990 UTRN intronic T G 1.000 0.418 2.17E-04 -0.0163 1.5032 461

19 rs6510006 57164930 SMIM17 intronic T C 0.996 0.460 2.18E-04 0.0188 0.0052 459

85

11 rs10766888 21869349 intergenic intergenic G C 1.000 0.358 2.18E-04 -0.0177 1.0646 461

13 rs8002609 26827224 CDK8 intronic T C 0.993 0.081 2.20E-04 0.0366 0.0133 458

10 rs181652 119379087 EMX2 intergenic A G 1.000 0.490 2.21E-04 -0.0188 0.0052 461

15 rs8042307 77804013 HMG20A intergenic T C 1.000 0.357 2.27E-04 0.0171 0.6544 461

22 rs9607271 35923409 RASD2 intergenic C T 0.961 0.433 2.29E-04 0.0196 0.0054 443

10 rs17358365 15328944 FAM171A1 intronic C A 0.998 0.283 2.40E-04 -0.0183 0.8146 460

2 rs1993313 221110780 intergenic intergenic T G 1.000 0.477 2.46E-04 0.0185 0.0050 461

1 rs11122432 232270985 DISC1 intergenic A G 0.991 0.247 2.60E-04 -0.0217 0.0061 457

4 rs348495 45184442 PRKRIRP9 intergenic G A 1.000 0.393 2.63E-04 0.0161 1.2338 461

18 rs7233417 35200180 CELF4 intergenic G A 0.998 0.401 2.63E-04 0.0188 0.0051 460

12 rs1497262 25122814 BCAT1 intergenic T C 0.998 0.122 2.71E-04 0.0294 0.0081 460

4 rs13149932 18804519 intergenic intergenic C A 1.000 0.192 2.77E-04 0.0206 4.4913 461

2 rs12990505 196052944 intergenic intergenic T C 0.996 0.195 2.90E-04 -0.0223 0.2307 459

19 rs10403522 57162101 SMIM17 intronic G A 0.989 0.454 2.93E-04 0.0173 1.3730 456

5 rs1445811 13867462 DNAH5 intronic A G 0.998 0.425 3.08E-04 0.0157 0.5050 460

8 rs10110297 122852102 BC052578 intergenic A G 0.970 0.081 3.11E-04 -0.0336 0.0092 447

6 rs17572402 170651372 FAM120B intronic G A 0.996 0.104 3.16E-04 -0.0299 0.5403 459

86

Table S1B: Association statistics for the top 50 SNPs associated with GMV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, T2D affected status, and intracranial volume. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

9 rs16916336 101236132 GABBR2 intronic C T 0.996 0.047 1.94E-06 19.2638 4.0011 459

2 rs16866742 180330897 ZNF385B intronic A G 0.993 0.099 5.45E-06 12.4685 2.7719 458

5 rs6873702 108015658 FER intergenic T A 1.000 0.149 5.66E-06 10.1063 2.1834 461

2 rs12478234 180334969 ZNF385B intronic G A 0.987 0.099 6.71E-06 -12.2283 2.7693 455

3 rs4600836 18768851 intergenic intergenic T C 1.000 0.163 1.00E-05 -10.0352 2.2371 461

3 rs4626115 18772770 intergenic intergenic C T 1.000 0.163 1.00E-05 10.0352 2.2005 461

18 rs1940436 415401 COLEC12 intronic C T 1.000 0.254 1.00E-05 -8.3782 1.7578 461

10 rs12569589 121872939 SEC23IP intergenic A T 0.948 0.370 1.20E-05 7.3725 1.6380 437

3 rs404552 18652605 SATB1 intergenic G A 0.974 0.336 1.30E-05 -7.7091 1.7270 449

3 rs336641 18617276 SATB1 intergenic A G 0.998 0.317 1.40E-05 7.7982 1.7960 460

21 rs2826000 21452898 intergenic intergenic C T 0.989 0.047 1.60E-05 -16.8080 4.0151 456

4 rs12504498 135742498 intergenic intergenic C T 0.993 0.203 1.60E-05 -8.7957 1.9534 458

3 rs336611 18650507 SATB1 intergenic C T 0.998 0.337 1.70E-05 7.6953 1.8530 460

12 rs10879971 41686097 PDZRN4 intronic T C 1.000 0.347 2.00E-05 7.2876 1.6681 461

87

3 rs3935033 18825074 intergenic intergenic T A 1.000 0.157 2.60E-05 -9.7243 2.1693 461

10 rs986151 23478615 PTF1A intergenic T C 1.000 0.498 2.70E-05 7.0742 1.6103 461

6 rs3777486 131584648 AKAP7 intronic T C 0.993 0.120 2.90E-05 -10.4388 2.4496 458

16 rs12324998 58111073 MMP15 intergenic T G 1.000 0.439 2.90E-05 7.0606 1.6501 461

7 rs17450482 117217987 CFTR intronic T C 0.980 0.077 3.10E-05 11.4422 2.7338 452

21 rs2850030 34854711 TMEM50B intergenic C G 0.991 0.367 3.20E-05 -7.1683 1.8264 457

9 rs16916338 101236304 GABBR2 intronic T G 0.998 0.048 3.30E-05 -16.6730 3.8956 460

11 rs2052690 10664033 MRVI1 intronic G T 1.000 0.284 3.60E-05 -7.1349 1.6922 461

1 rs9426495 4208918 LINC01346 intergenic G A 0.983 0.305 3.80E-05 7.1750 1.7208 453

12 rs7979024 84506012 intergenic intergenic C G 1.000 0.022 4.00E-05 -22.4544 4.9799 461

4 rs4629410 93372969 GRID2 intronic A G 1.000 0.188 4.30E-05 8.7216 2.1899 461

21 rs2027712 21447536 intergenic intergenic G A 1.000 0.039 4.40E-05 17.7740 4.2488 461

21 rs2825998 21446749 intergenic intergenic G A 1.000 0.039 4.40E-05 17.7740 4.2488 461

21 rs2825986 21439905 intergenic intergenic C T 0.998 0.045 4.90E-05 -16.4043 3.8828 460

11 rs4442541 10669172 MRVI1 intronic G A 1.000 0.293 5.50E-05 6.7703 2.1661 461

13 rs683247 30553523 UBL3 intergenic G C 1.000 0.144 5.70E-05 -10.0707 2.5390 461

12 rs4907436 131992746 SFSWAP intergenic G A 1.000 0.022 6.10E-05 20.9847 5.1509 461

12 rs11061573 131983266 SFSWAP intergenic G A 1.000 0.022 6.10E-05 20.9847 5.1509 461

13 rs1328837 109879099 MYO16 intergenic A G 1.000 0.228 6.60E-05 7.7092 1.9358 461

88

19 rs17727145 44874369 ZNF112 intergenic C T 1.000 0.152 6.70E-05 9.4004 2.2971 461

21 rs2826063 21554454 intergenic intergenic G A 1.000 0.036 6.80E-05 18.1030 4.4401 461

4 rs6814134 39191627 WDR19 intronic T C 1.000 0.103 6.80E-05 10.8122 2.7256 461

18 rs8096721 5904873 TMEM200C intergenic G A 0.980 0.283 6.90E-05 -7.2130 1.8141 452

4 rs17268740 92114903 CCSER1 intronic A G 1.000 0.246 7.20E-05 -7.7914 1.9043 461

11 rs2207548 32368744 WT1 intergenic C A 1.000 0.409 7.30E-05 -6.8479 1.7133 461

11 rs622961 94154902 MRE11A intronic G A 0.989 0.394 7.40E-05 6.5781 1.5586 456

4 rs6531698 39263814 WDR19 intronic C T 0.998 0.104 7.60E-05 -10.6719 2.6206 460

10 rs485305 116048045 VWA2 intronic C A 1.000 0.027 7.70E-05 -20.6711 5.3389 461

18 rs482992 2079540 intergenic intergenic T C 0.985 0.477 8.30E-05 6.5858 1.5924 454

16 rs12149466 81429258 GAN intergenic C G 1.000 0.476 8.30E-05 -6.6264 1.7663 461

1 rs1002485 203189488 CHIT1 intronic T C 0.993 0.495 8.40E-05 -6.7061 1.6918 458

8 rs16908346 139156191 FAM135B intronic G A 0.987 0.033 8.50E-05 -17.2100 4.3126 455

1 rs12123330 4227464 LINC01346 intergenic C T 0.996 0.207 8.60E-05 -7.5785 1.8186 459

17 rs17546529 75624399 SEPT9 intergenic G T 1.000 0.175 9.10E-05 8.4406 2.1805 461

21 rs2826077 21568713 intergenic intergenic C T 0.987 0.035 9.20E-05 -17.7369 4.6157 455

7 rs1029558 26588391 SKAP2 intergenic T C 1.000 0.367 9.20E-05 -6.5532 1.6175 461

89

Table S1C: Association statistics for the top 50 SNPs associated with WMV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, T2D affected status, and intracranial volume. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

19 rs10422843 29199789 intergenic intergenic A C 1.000 0.349 2.06E-06 -8.0867 1.7480 461

15 rs532601 53968961 WDR72 intronic T C 1.000 0.013 7.23E-06 -32.7980 7.3477 461

3 rs155268 6514661 GRM7-AS3 intergenic G C 0.996 0.183 1.00E-05 9.0062 1.9381 459

6 rs1932217 83550056 UBE3D intergenic A G 1.000 0.245 1.20E-05 8.2420 1.8709 461

3 rs16830710 179405728 USP13 intronic C T 1.000 0.016 1.30E-05 27.8149 6.1234 461

3 rs13317989 56702236 FAM208A intronic C T 1.000 0.093 1.30E-05 12.3852 2.8222 461

3 rs7612464 56684978 FAM208A intronic G A 0.998 0.092 1.60E-05 -12.3303 2.8853 460

3 rs282525 56630653 CCDC66 intronic C A 1.000 0.092 1.70E-05 -12.2791 2.7217 461

3 rs4681726 56692642 FAM208A intronic A G 1.000 0.092 1.70E-05 12.2791 2.8807 461

5 rs1011220 172689832 NKX2-5 intergenic G T 0.987 0.424 1.70E-05 7.3155 1.7193 455

2 rs11896222 106107078 FHL2 intergenic C G 1.000 0.098 2.10E-05 12.1003 2.7925 461

1 rs17130102 88143908 intergenic intergenic T C 1.000 0.173 2.70E-05 9.2402 2.3043 461

2 rs11689550 178456298 TTC30B intergenic G A 0.996 0.209 2.90E-05 8.5631 2.0174 459

2 rs13004955 178398923 AGPS intronic A T 1.000 0.197 3.00E-05 -8.3363 1.9398 461

90

18 rs2139875 4193891 DLGAP1 intronic T C 0.996 0.270 3.00E-05 -7.5469 1.7625 459

10 rs2394845 73605926 PSAP intronic T G 1.000 0.182 3.10E-05 8.8639 2.1720 461

2 rs4893825 178298687 AGPS intronic C T 0.998 0.196 3.20E-05 -8.3642 2.0722 460

9 rs4265281 17819394 SH3GL2 intergenic G A 0.998 0.186 3.90E-05 -8.2389 1.9509 460

13 rs683247 30553523 UBL3 intergenic G C 1.000 0.144 4.20E-05 10.1753 2.6897 461

16 rs16962035 60423238 LOC729159 intergenic T G 0.998 0.272 4.20E-05 -7.5418 1.7650 460

2 rs6736482 23365209 KLHL29 intergenic A G 1.000 0.377 4.60E-05 6.8153 1.6706 461

1 rs17130101 88143461 intergenic intergenic A G 0.993 0.170 4.70E-05 -8.9856 2.3118 458

12 rs17790300 6685188 CHD4 intronic G A 1.000 0.012 5.30E-05 32.1356 7.8690 461

12 rs11064276 6706982 CHD4 intronic G A 1.000 0.012 5.30E-05 32.1356 7.8690 461

9 rs7032460 80730180 GNAQ intergenic A T 1.000 0.112 5.40E-05 10.0968 2.4883 461

13 rs1330896 78603571 EDNRB intergenic C G 1.000 0.174 5.50E-05 8.8662 2.1427 461

10 rs2688607 75663736 PLAU intergenic G A 0.998 0.208 5.80E-05 -7.8292 1.8535 460

5 rs17618846 35434865 PRLR intergenic A C 1.000 0.080 6.30E-05 -12.2196 3.0272 461

11 rs11236960 76540683 ACER3 intergenic G A 1.000 0.024 6.40E-05 -21.4070 5.3169 461

11 rs7122943 76541678 ACER3 intergenic C G 1.000 0.024 6.40E-05 21.4071 5.2751 461

15 rs16966350 53931936 WDR72 intronic G A 1.000 0.012 6.60E-05 -30.5883 7.6531 461

15 rs6493643 53992347 WDR72 intronic A T 1.000 0.012 6.60E-05 30.5884 7.4074 461

10 rs1567705 5815108 GDI2 intronic T C 0.998 0.112 6.80E-05 -10.3279 2.5879 460

91

8 rs17737611 9854659 MSRA intergenic A G 1.000 0.069 6.90E-05 12.8140 3.3946 461

22 rs5754188 33048216 SYN3 intronic T G 0.998 0.210 6.90E-05 7.7088 1.8850 460

1 rs17693196 181716901 CACNA1E intronic A T 0.998 0.135 7.60E-05 8.7940 2.1581 460

9 rs10512071 80746139 CEP78 intergenic A G 1.000 0.113 7.60E-05 9.9009 2.5546 461

12 rs16940719 110397855 GIT2 intronic C T 0.993 0.024 7.60E-05 20.5077 5.7524 458

16 rs7190134 68381516 PRMT7 intronic G A 1.000 0.098 8.00E-05 -10.9990 2.9234 461

15 rs16971150 79898663 TMED3 intergenic T C 1.000 0.016 8.20E-05 -24.5532 6.1579 461

15 rs2865199 79903372 TMED3 intergenic T C 1.000 0.016 8.20E-05 -24.5532 6.1579 461

2 rs2167902 178302412 AGPS intronic C T 0.976 0.092 8.50E-05 -11.0070 2.8201 450

7 rs9639329 19399831 FERD3L intergenic A G 0.978 0.338 8.60E-05 -6.4229 1.5750 451

3 rs11719707 56691393 FAM208A intronic G A 1.000 0.090 8.90E-05 -11.2898 2.8303 461

21 rs2829934 27205540 APP intergenic C T 1.000 0.034 9.00E-05 18.2609 4.5586 461

3 rs282524 56630469 CCDC66 intronic T G 0.998 0.116 9.10E-05 -9.9586 2.5782 460

7 rs17155256 81238711 HGF intergenic A T 0.998 0.018 9.80E-05 23.8219 6.0193 460

16 rs8058517 68379860 PRMT7 intronic C T 0.998 0.101 9.80E-05 10.7181 2.6778 460

1 rs12562587 181709232 CACNA1E intronic A G 0.972 0.127 1.00E-04 8.7572 2.2775 448

1 rs17459921 88143753 intergenic intergenic C T 1.000 0.110 1.00E-04 -9.8754 2.4369 461

92

Table S1D: Association statistics for the top 50 SNPs associated with GMFA. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

2 rs7596222 39611382 MAP4K3 intronic C T 0.987 0.020 1.77E-06 0.0178 0.0027 451

6 rs10499169 130278909 L3MBTL3 intergenic C T 0.998 0.016 3.55E-06 -0.0192 0.0029 456

18 rs8091628 20858139 TMEM241 intergenic A T 1.000 0.054 4.14E-06 0.0109 0.0020 457

2 rs1485055 139686615 NXPH2 intergenic T G 0.996 0.087 4.56E-06 -0.0082 0.0018 455

18 rs12953383 41244911 intergenic intergenic A G 0.963 0.210 4.67E-06 0.0062 0.0013 440

10 rs1027724 67182256 LOC100421870 intergenic C T 1.000 0.024 4.71E-06 0.0157 0.0032 457

3 rs6804855 168755583 MECOM intergenic T C 0.993 0.169 5.81E-06 -0.0063 0.0013 454

12 rs10858818 89426732 LOC728084 intergenic A C 0.998 0.105 1.00E-05 0.0073 0.0017 456

9 rs10869157 74878937 GDA intergenic A C 0.980 0.481 1.00E-05 0.0049 0.0011 448

2 rs1351924 139708230 NXPH2 intergenic C T 1.000 0.083 1.10E-05 0.0086 0.0019 457

18 rs10502817 41249142 intergenic intergenic C T 1.000 0.221 1.10E-05 0.0058 0.0013 457

9 rs7862279 74858789 GDA intronic T C 0.998 0.477 1.10E-05 -0.0049 0.0011 456

3 rs7647582 168753276 MECOM intergenic G A 1.000 0.170 1.40E-05 -0.0060 0.0014 457

3 rs7631699 168778197 MECOM intergenic G A 0.969 0.201 1.40E-05 -0.0056 0.0013 443

93

5 rs295677 58150803 RAB3C intergenic T C 1.000 0.263 1.80E-05 -0.0053 0.0011 457

2 rs7595102 39410661 CDKL4 intronic T C 1.000 0.020 2.10E-05 -0.0158 0.0037 457

12 rs12311684 89435177 LOC728084 intergenic A G 1.000 0.105 2.30E-05 0.0071 0.0016 457

18 rs1609147 41219407 intergenic intergenic G A 1.000 0.227 2.50E-05 -0.0056 0.0013 457

3 rs838612 143178775 SLC9A9 intronic G C 0.982 0.337 2.50E-05 -0.0049 0.0012 449

18 rs12458486 41253288 intergenic intergenic A G 0.991 0.221 2.60E-05 0.0055 0.0013 453

18 rs8095325 66060991 MTL3P intergenic T C 1.000 0.301 2.70E-05 -0.0049 0.0012 457

18 rs1527433 41216721 intergenic intergenic G T 1.000 0.228 3.10E-05 0.0055 0.0013 457

12 rs17570542 44983452 NELL2 intronic A G 1.000 0.170 3.60E-05 -0.0057 0.0013 457

1 rs12031587 59472053 HSD52 intergenic C T 0.998 0.166 3.70E-05 0.0058 0.0014 456

6 rs1923380 165491551 C6orf118 intergenic A G 0.993 0.302 3.80E-05 0.0048 0.0012 454

7 rs1467345 11783917 THSD7A intronic G A 0.998 0.122 4.30E-05 -0.0066 0.0016 456

6 rs4946672 105991031 PREP intergenic G A 1.000 0.487 4.30E-05 -0.0043 0.0008 457

4 rs6836901 29988423 LOC100130674 intergenic C A 1.000 0.105 4.40E-05 -0.0066 0.0016 457

3 rs16852912 168686676 MECOM intergenic C T 0.963 0.095 4.60E-05 0.0073 0.0018 440

20 rs754285 50103185 NFATC2 intronic C T 1.000 0.143 4.60E-05 0.0059 0.0014 457

4 rs443233 177439156 VEGFC intergenic C T 0.982 0.435 4.90E-05 0.0044 0.0011 449

2 rs1485059 139672713 NXPH2 intergenic T C 0.998 0.079 5.00E-05 -0.0081 0.0020 456

2 rs17036763 47859616 MSH2 intergenic C G 1.000 0.061 5.20E-05 0.0088 0.0019 457

94

12 rs10842688 26338294 SSPN intergenic C A 1.000 0.451 5.20E-05 -0.0040 0.0010 457

12 rs10842689 26338365 SSPN intergenic C T 0.998 0.451 5.30E-05 0.0040 0.0010 456

4 rs17003907 80452832 LINC00989 intronic C G 1.000 0.037 5.60E-05 -0.0106 0.0021 457

12 rs17649306 44982356 NELL2 intronic A G 0.991 0.168 5.70E-05 -0.0055 0.0014 453

18 rs12327228 66063259 MTL3P intergenic C G 1.000 0.297 5.70E-05 0.0047 0.0011 457

18 rs9941444 50315007 DCC intronic G A 0.985 0.032 5.80E-05 -0.0116 0.0029 450

9 rs10781090 74839689 GDA intronic C T 0.954 0.494 5.80E-05 0.0046 0.0011 436

6 rs7769616 7667762 SNRNP48 intergenic T C 1.000 0.233 6.00E-05 -0.0049 0.0012 457

7 rs1465007 11782995 THSD7A intronic C T 1.000 0.121 6.20E-05 0.0065 0.0013 457

6 rs4946666 105975220 PREP intergenic T C 1.000 0.454 6.80E-05 -0.0042 0.0010 457

8 rs10106323 41311222 GOLGA7 intergenic G A 1.000 0.171 7.00E-05 -0.0056 0.0014 457

1 rs4492658 97876523 DPYD intronic G C 0.998 0.304 7.00E-05 0.0046 0.0011 456

2 rs534781 139766782 NXPH2 intergenic T A 1.000 0.081 7.10E-05 -0.0079 0.0020 457

7 rs6460839 11786454 THSD7A intronic G C 0.996 0.121 7.10E-05 -0.0064 0.0016 455

4 rs921312 177433597 VEGFC intergenic A T 0.989 0.440 7.10E-05 0.0043 0.0011 452

15 rs2086329 94657373 MCTP2 intergenic C T 1.000 0.405 7.30E-05 -0.0043 0.0011 457

11 rs510960 133060271 OPCML intronic T C 1.000 0.204 7.60E-05 -0.0053 0.0013 457

95

Table S1E: Association statistics for the top 50 SNPs associated with WMFA. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

9 rs2890109 92742249 GADD45G intergenic C T 1.000 0.112 1.06E-06 0.0109 0.0022 457

21 rs34138687 25880945 intergenic intergenic G A 1.000 0.033 1.07E-06 -0.0182 0.0034 457

5 rs6887681 21826642 CDH12 intronic C T 1.000 0.068 4.03E-06 -0.0122 0.0022 457

9 rs16905942 92716596 GADD45G intergenic G A 1.000 0.128 5.22E-06 -0.0095 0.0021 457

9 rs16905946 92716791 GADD45G intergenic C T 1.000 0.128 5.22E-06 0.0095 0.0021 457

5 rs2174440 26766362 CDH9 intergenic C T 1.000 0.158 5.37E-06 0.0084 0.0018 457

9 rs16906038 92744156 GADD45G intergenic T C 1.000 0.129 5.91E-06 -0.0095 0.0019 457

9 rs2119293 92740770 GADD45G intergenic G T 1.000 0.129 6.42E-06 0.0094 0.0021 457

5 rs2404519 26766426 CDH9 intergenic G A 1.000 0.159 7.37E-06 -0.0083 0.0018 457

16 rs904204 86136699 IRF8 intergenic C G 1.000 0.150 8.97E-06 0.0082 0.0016 457

7 rs10252097 36049833 SEPT7 intergenic G T 1.000 0.012 1.10E-05 0.0240 0.0039 457

7 rs10230752 78554551 MAGI2 intronic T C 1.000 0.019 1.40E-05 -0.0211 0.0049 457

11 rs1783921 128094803 ETS1 intergenic T C 0.998 0.215 2.00E-05 0.0072 0.0017 456

2 rs7596222 39611382 MAP4K3 intronic C T 0.987 0.020 2.20E-05 0.0200 0.0038 451

96

3 rs3805011 4807021 ITPR1 intronic A G 1.000 0.345 4.10E-05 0.0059 0.0014 457

20 rs6077844 10579996 SLX4IP intronic G A 0.998 0.423 4.10E-05 0.0056 0.0014 456

16 rs17678609 83295738 CDH13 intronic T C 0.998 0.222 4.30E-05 -0.0068 0.0017 456

3 rs3891799 189161056 TPRG1 intergenic G A 0.993 0.454 4.60E-05 0.0054 0.0013 454

18 rs8096456 13996751 MC2R intergenic G A 1.000 0.276 5.20E-05 0.0060 0.0015 457

13 rs9523932 93788356 GPC6 intergenic A G 1.000 0.432 5.70E-05 -0.0052 0.0013 457

3 rs7631699 168778197 MECOM intergenic G A 0.969 0.201 5.80E-05 -0.0066 0.0016 443

13 rs1175705 26898489 CDK8 intronic A G 1.000 0.082 5.90E-05 -0.0102 0.0024 457

13 rs160823 93758861 GPC6 intergenic A G 1.000 0.434 6.40E-05 -0.0052 0.0011 457

3 rs4685814 4796784 ITPR1 intronic T A 1.000 0.394 7.20E-05 -0.0055 0.0014 457

13 rs9319287 26839863 CDK8 intronic A G 0.998 0.083 7.30E-05 -0.0100 0.0025 456

6 rs10499169 130278909 L3MBTL3 intergenic C T 0.998 0.016 7.50E-05 -0.0206 0.0039 456

6 rs12189703 40112188 MOCS1 intergenic G T 1.000 0.054 7.60E-05 0.0116 0.0027 457

4 rs17258085 183479255 TENM3 intronic T C 1.000 0.264 7.60E-05 -0.0059 0.0015 457

18 rs12457555 13992052 MC2R intergenic C G 0.998 0.280 7.70E-05 -0.0058 0.0017 456

5 rs6889835 89148875 MIR3660 intergenic A T 0.976 0.127 8.10E-05 -0.0074 0.0018 446

5 rs2036093 83884333 EDIL3 intergenic A G 1.000 0.162 8.10E-05 -0.0067 0.0017 457

21 rs2829156 25910760 intergenic intergenic C A 0.998 0.071 8.40E-05 -0.0099 0.0025 456

9 rs10819780 98826369 LOC158435 intergenic G A 1.000 0.305 8.50E-05 0.0053 0.0014 457

97

3 rs3891798 189161117 TPRG1 intergenic G A 0.993 0.470 8.50E-05 0.0052 0.0013 454

8 rs17732201 96333993 LOC100616530 intronic A G 1.000 0.038 8.70E-05 0.0136 0.0032 457

8 rs4876153 2304334 MYOM2 intergenic A G 0.996 0.299 8.70E-05 0.0061 0.0015 455

10 rs11202366 88931737 FAM35A intronic G A 1.000 0.339 9.10E-05 -0.0059 0.0015 457

2 rs16827800 187248928 ZC3H15 intergenic C T 1.000 0.049 9.30E-05 -0.0123 0.0027 457

10 rs7096107 88847985 GLUD1 intronic C T 0.998 0.350 9.70E-05 0.0059 0.0015 456

5 rs1587071 26410644 intergenic intergenic C T 0.998 0.071 1.02E-04 0.0102 0.0026 456

5 rs17104519 145787415 POU4F3 intergenic C T 1.000 0.032 1.03E-04 -0.0146 0.0037 457

3 rs838612 143178775 SLC9A9 intronic G C 0.982 0.337 1.05E-04 -0.0058 0.0015 449

5 rs2914908 117982046 DTWD2 intergenic C T 1.000 0.451 1.05E-04 -0.0051 0.0014 457

10 rs7096113 88847999 GLUD1 intronic A T 0.998 0.350 1.06E-04 0.0058 0.0015 456

6 rs12194277 154800018 CNKSR3 intronic C T 0.998 0.490 1.09E-04 -0.0052 0.0014 456

2 rs41363353 63072403 EHBP1 intronic C G 0.989 0.170 1.12E-04 0.0067 0.0017 452

5 rs520865 80337457 RASGRF2 intronic G A 0.985 0.437 1.13E-04 -0.0054 0.0014 450

13 rs7322319 26931413 CDK8 intronic A G 1.000 0.084 1.16E-04 -0.0097 0.0025 457

13 rs7326961 26840447 CDK8 intronic C G 1.000 0.084 1.16E-04 -0.0097 0.0023 457

13 rs7997534 26850295 CDK8 intronic C A 1.000 0.084 1.16E-04 0.0097 0.0025 457

98

Table S1F: Association statistics for the top 50 SNPs associated with GMMD. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

8 rs11986940 108783667 RSPO2 intergenic T C 0.998 0.070 8.46E-07 0.0439 0.0088 456

8 rs6981446 108768390 RSPO2 intergenic T C 0.998 0.072 1.31E-06 0.0417 0.0085 456

11 rs11214882 114024199 ZBTB16 intronic C T 0.996 0.107 5.67E-06 0.0332 0.0072 455

2 rs7562947 70758123 TGFA intronic A G 1.000 0.109 1.30E-05 -0.0296 0.0065 457

6 rs3909134 29699253 HLA-F-AS1 intronic A G 0.998 0.019 1.40E-05 0.0650 0.0148 456

6 rs17058460 130498557 SAMD3 intronic T C 1.000 0.077 1.60E-05 -0.0383 0.0088 457

1 rs1758780 182616825 RGS8 intronic A G 0.991 0.019 1.70E-05 0.0822 0.0185 453

6 rs17058461 130498776 SAMD3 intronic C T 1.000 0.078 1.70E-05 0.0379 0.0088 457

5 rs4704298 75514065 SV2C intergenic C T 1.000 0.263 1.70E-05 0.0231 0.0053 457

19 rs17727145 44874369 ZFP112 intergenic C T 1.000 0.150 2.30E-05 -0.0275 0.0065 457

20 rs6068321 51345687 TSHZ2 intergenic T C 1.000 0.075 2.40E-05 -0.0366 0.0085 457

20 rs697702 51368565 TSHZ2 intergenic T A 1.000 0.075 2.40E-05 -0.0366 0.0085 457

8 rs436760 24144598 ADAM28 intergenic T C 0.993 0.307 2.50E-05 0.0205 0.0048 454

21 rs9976680 44200761 PDE9A intergenic C G 0.998 0.117 3.20E-05 0.0277 0.0066 456

99

1 rs4652738 182613666 RGS8 intergenic G A 1.000 0.020 3.80E-05 -0.0774 0.0184 457

5 rs17101236 143303771 HMHB1 intergenic G A 1.000 0.132 3.90E-05 -0.0274 0.0065 457

17 rs11079258 54858579 C17orf67 intergenic C T 0.987 0.226 4.10E-05 0.0216 0.0052 451

5 rs17081201 177845416 COL23A1 intronic T C 1.000 0.023 4.70E-05 -0.0609 0.0149 457

7 rs9648642 9968807 LOC340268 intergenic C T 0.998 0.388 4.70E-05 0.0192 0.0047 456

6 rs2235276 166907173 RPS6KA2 intronic T C 0.998 0.254 5.50E-05 -0.0209 0.0052 456

12 rs7399173 132036428 SFSWAP intergenic G A 1.000 0.071 5.70E-05 0.0325 0.0080 457

12 rs11246833 132057720 SFSWAP intergenic G C 1.000 0.070 6.30E-05 0.0331 0.0082 457

5 rs2577548 128361896 SLC27A6 intronic G A 1.000 0.153 6.80E-05 0.0248 0.0062 457

11 rs10082611 27320108 CCDC34 intergenic G C 0.998 0.381 6.90E-05 -0.0182 0.0046 456

5 rs17275322 9766265 LOC285692 intronic A C 1.000 0.030 7.10E-05 0.0514 0.0125 457

4 rs4997952 178740532 LOC285501 intronic T A 1.000 0.106 7.20E-05 -0.0298 0.0075 457

5 rs12651951 79645963 CRSP8P intergenic G T 0.998 0.043 7.60E-05 0.0447 0.0113 456

4 rs17065243 178744050 LOC285501 intronic T G 0.998 0.106 7.60E-05 -0.0298 0.0075 456

11 rs510358 114064983 ZBTB16 intronic T C 1.000 0.344 7.60E-05 -0.0195 0.0049 457

5 rs257932 128383487 SLC27A6 intergenic A G 1.000 0.179 7.70E-05 -0.0235 0.0058 457

5 rs257933 128383050 SLC27A6 intergenic A G 1.000 0.179 7.70E-05 -0.0235 0.0058 457

5 rs257938 128379649 SLC27A6 intergenic C A 1.000 0.179 7.70E-05 0.0235 0.0059 457

8 rs2013265 24092500 ADAM28 intergenic C T 1.000 0.281 8.00E-05 -0.0203 0.0051 457

100

20 rs241798 51360640 TSHZ2 intergenic A G 0.998 0.075 8.20E-05 0.0334 0.0084 456

11 rs670728 114064733 ZBTB16 intronic T G 0.969 0.334 8.50E-05 -0.0195 0.0049 443

11 rs10767628 27310871 CCDC34 intergenic C G 0.996 0.382 8.90E-05 0.0181 0.0046 455

17 rs4794670 54927829 DGKE intronic A T 1.000 0.231 9.00E-05 0.0214 0.0054 457

4 rs12650193 121900882 NDNF intergenic T C 0.998 0.110 9.10E-05 -0.0283 0.0072 456

5 rs40977 128360191 SLC27A6 intronic C T 0.996 0.177 9.10E-05 -0.0234 0.0059 455

11 rs10082678 27320448 CCDC34 intergenic G A 0.998 0.384 9.10E-05 -0.0179 0.0045 456

8 rs7007223 24092942 ADAM28 intergenic T A 0.991 0.302 9.20E-05 0.0205 0.0052 453

8 rs11991799 24173939 ADAM28 intronic T C 0.993 0.162 9.80E-05 0.0232 0.0059 454

12 rs7398686 132047807 SFSWAP intergenic G A 0.998 0.069 9.90E-05 0.0329 0.0084 456

13 rs7329984 27376741 GPR12 intergenic A G 0.958 0.349 9.90E-05 0.0196 0.0051 438

3 rs10513860 190419541 IL1RAP intergenic G A 1.000 0.069 1.01E-04 0.0351 0.0091 457

18 rs17452662 380831 COLEC12 intronic G C 0.998 0.172 1.02E-04 0.0232 0.0060 456

8 rs2022982 107371924 OXR1 intronic A C 1.000 0.476 1.02E-04 0.0172 0.0044 457

9 rs2520094 136919985 LOC100130548 intronic A T 0.998 0.283 1.03E-04 -0.0208 0.0053 456

2 rs352978 142585905 LRP1B intronic T A 0.993 0.464 1.06E-04 -0.0177 0.0046 454

19 rs2239373 44163033 PLAUR intronic T G 1.000 0.165 1.07E-04 0.0236 0.0061 457

101

Table S1G: Association statistics for the top 50 SNPs associated with WMMD. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Associaton Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

16 rs13330413 12767048 CPPED1 intronic C A 1.000 0.024 1.13E-06 -0.0479 0.0096 457

18 rs1785562 9081185 NDUFV2 intergenic G T 0.976 0.187 3.44E-06 0.0175 0.0037 446

1 rs2923297 105569484 intergenic intergenic C T 1.000 0.024 3.50E-06 -0.0438 0.0087 457

3 rs6788050 190376788 IL1RAP intergenic A G 0.991 0.068 5.27E-06 -0.0259 0.0052 453

3 rs11716597 19854903 EFHB intergenic G T 1.000 0.468 6.27E-06 0.0135 0.0030 457

20 rs3975198 55398116 TFAP2C intergenic G A 0.998 0.409 8.12E-06 -0.0131 0.0033 456

3 rs4858736 19853013 EFHB intergenic G C 0.980 0.468 1.70E-05 -0.0128 0.0029 448

7 rs4718112 64256482 ZNF138 intronic C T 1.000 0.104 1.80E-05 0.0212 0.0048 457

5 rs2526256 128255056 SLC27A6 intergenic T C 1.000 0.430 1.90E-05 -0.0133 0.0031 457

16 rs10514577 83368400 CDH13 intronic G C 1.000 0.417 2.10E-05 0.0126 0.0030 457

5 rs2222123 160390549 ATP10B intergenic C T 1.000 0.158 2.20E-05 0.0167 0.0039 457

5 rs17430099 160381912 ATP10B intergenic G A 1.000 0.159 2.30E-05 -0.0167 0.0039 457

18 rs10775415 10737373 PIEZO2 intronic C T 0.993 0.196 2.90E-05 0.0158 0.9150 454

5 rs3733750 180235503 MGAT1 intronic T C 0.989 0.085 3.10E-05 0.0209 0.0050 452

102

4 rs4697390 23456789 intergenic intergenic T C 0.993 0.144 3.10E-05 0.0176 0.0043 454

18 rs5024299 10741011 PIEZO2 intronic A G 0.998 0.203 3.60E-05 0.0152 0.8975 456

8 rs929685 113304170 CSMD3 intronic G A 0.998 0.441 4.00E-05 0.0120 0.0029 456

16 rs723920 83342389 CDH13 intronic C A 0.998 0.396 4.10E-05 0.0122 0.0030 456

2 rs13011532 70797390 TGFA intergenic G C 0.998 0.367 4.30E-05 0.0121 0.0029 456

3 rs10513860 190419541 IL1RAP intergenic G A 1.000 0.069 4.50E-05 0.0237 0.0057 457

3 rs2292006 170858406 TNIK intronic G T 1.000 0.254 4.60E-05 0.0134 0.0032 457

16 rs375624 24145028 PRKCB intronic A T 1.000 0.321 5.10E-05 -0.0123 0.0029 457

8 rs6472209 66474748 LOC286186 intronic A T 1.000 0.409 5.10E-05 -0.0124 0.9865 457

8 rs11998379 51129075 SNTG1 intronic G A 1.000 0.021 5.30E-05 -0.0392 0.0096 457

3 rs3891799 189161056 TPRG1 intergenic G A 0.993 0.454 5.50E-05 -0.0117 0.0029 454

2 rs4667005 181862579 UBE2E3 intronic A G 0.991 0.278 5.60E-05 0.0127 0.0031 453

7 rs2191277 123794037 TMEM229A intergenic A G 1.000 0.266 5.60E-05 -0.0139 0.0034 457

16 rs13337699 83375743 CDH13 intronic G A 1.000 0.433 5.60E-05 0.0119 0.0030 457

16 rs8057728 76021569 RPL18P13 intergenic G A 1.000 0.382 5.60E-05 0.0124 0.0030 457

16 rs1424189 83330610 CDH13 intronic C T 1.000 0.391 6.10E-05 -0.0120 0.0030 457

8 rs1420859 113318058 CSMD3 intronic T A 0.996 0.442 6.20E-05 0.0119 0.0029 455

1 rs11210886 44071546 PTPRF intronic G A 1.000 0.100 6.30E-05 -0.0200 0.0050 457

4 rs13115147 13903818 LOC152742 intergenic T G 1.000 0.332 6.30E-05 0.0127 0.0032 457

103

16 rs198202 24149041 PRKCB intronic A G 1.000 0.324 6.90E-05 -0.0122 0.0030 457

5 rs7702784 128191568 SLC27A6 intergenic T A 1.000 0.448 7.20E-05 -0.0120 0.0030 457

6 rs7769941 99569709 FAXC intergenic T G 0.998 0.088 7.50E-05 -0.0194 0.0049 456

11 rs763470 10452005 AMPD3 intergenic A G 0.998 0.245 7.50E-05 0.0131 0.0033 456

17 rs2318872 63883342 CEP112 intronic A C 1.000 0.097 7.80E-05 0.0201 0.0050 457

12 rs6487254 22111253 ABCC9 intergenic C T 1.000 0.040 8.00E-05 -0.0289 0.0073 457

16 rs8057050 76021533 RPL18P13 intergenic A C 1.000 0.381 8.20E-05 -0.0121 0.0030 457

3 rs3891798 189161117 TPRG1 intergenic G A 0.993 0.470 8.30E-05 -0.0114 0.0030 454

8 rs6981446 108768390 RSPO2 intergenic T C 0.998 0.072 8.50E-05 0.0216 1.1328 456

8 rs4297024 134138581 TG intronic G A 0.996 0.086 8.70E-05 0.0194 0.0050 455

1 rs17464532 59483091 HSD52 intergenic G C 0.998 0.139 9.30E-05 0.0162 0.0042 456

12 rs10734860 13393194 EMP1 intergenic G A 0.998 0.095 9.40E-05 -0.0203 0.0052 456

8 rs2022982 107371924 OXR1 intronic A C 1.000 0.476 9.80E-05 0.0112 0.0029 457

3 rs6444452 190431641 IL1RAP intergenic C T 1.000 0.071 1.00E-04 -0.0218 0.0054 457

3 rs6791374 190431993 IL1RAP intergenic T C 1.000 0.071 1.00E-04 0.0218 0.0056 457

8 rs1503562 107357180 OXR1 intronic C T 1.000 0.457 1.01E-04 0.0112 0.0029 457

2 rs10194801 181932619 UBE2E3 intergenic T C 0.996 0.366 1.02E-04 0.0114 0.0029 455

104

Table S1H: Association statistics for the top 50 SNPs associated with WMLV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, T2D affected status, and intracranial volume. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

5 rs10065017 157806404 EBF1 intergenic T G 0.995 0.041 2.06E-07 0.7448 0.1420 426

21 rs7275537 33415514 HUNK intergenic T C 0.988 0.214 1.64E-06 -0.3404 0.0700 423

5 rs824848 158086455 EBF1 intergenic A G 0.991 0.409 1.69E-06 0.2827 0.0597 424

5 rs824854 158083398 EBF1 intergenic A G 1.000 0.416 1.99E-06 0.2768 0.0596 428

5 rs2112262 158101763 EBF1 intergenic G A 1.000 0.487 2.69E-06 -0.2775 0.0573 428

5 rs33139 158063159 EBF1 intergenic G A 0.998 0.491 3.34E-06 0.2811 0.0594 427

5 rs10045385 157822886 EBF1 intergenic T C 1.000 0.043 4.22E-06 0.6451 0.1398 428

5 rs17055534 157809993 EBF1 intergenic C T 1.000 0.043 4.22E-06 -0.6451 0.1400 428

5 rs10515769 158096975 EBF1 intergenic G A 1.000 0.499 4.52E-06 0.2739 0.0596 428

5 rs17055538 157812229 EBF1 intergenic C A 0.963 0.058 4.54E-06 0.5729 0.1247 412

5 rs173423 158073131 EBF1 intergenic C G 0.988 0.403 4.98E-06 0.2625 0.0569 423

10 rs887193 119332319 EMX2 intergenic C T 1.000 0.143 7.72E-06 0.3727 0.0828 428

5 rs244654 158074727 EBF1 intergenic C T 0.977 0.379 9.35E-06 0.2666 0.0598 418

19 rs2068061 44610665 ZNF224 Missense M118V G A 0.981 0.148 1.10E-05 0.3584 0.0801 420

105

10 rs7095481 3521369 LOC101927880 intergenic A G 0.991 0.019 1.50E-05 -0.7917 0.1853 424

5 rs824886 158061076 EBF1 intergenic A T 1.000 0.192 1.60E-05 0.3065 0.0721 428

20 rs297702 4385896 ADRA1D intergenic C G 1.000 0.299 2.20E-05 0.2647 0.0625 428

2 rs13013887 148825780 MBD5 intronic A G 1.000 0.367 2.60E-05 0.2502 0.0599 428

14 rs17115605 82043797 SEL1L intergenic A C 1.000 0.069 3.10E-05 -0.4757 0.1135 428

12 rs659964 112130199 ACAD10 intronic G C 1.000 0.182 3.20E-05 -0.3181 0.0766 428

1 rs511187 168695027 DPT intronic A G 1.000 0.262 3.30E-05 -0.2752 0.0644 428

8 rs377057 109142307 RSPO2 intergenic C T 0.991 0.251 3.50E-05 0.2826 0.0697 424

8 rs442355 109128653 RSPO2 intergenic G C 0.998 0.255 3.60E-05 -0.2790 0.0668 427

12 rs609230 112146911 ACAD10 intronic C T 1.000 0.193 3.70E-05 0.3114 0.0758 428

8 rs425932 109142353 RSPO2 intergenic A G 1.000 0.251 3.90E-05 0.2790 0.0670 428

4 rs11727749 11535285 HS3ST1 intergenic C T 1.000 0.102 4.10E-05 0.4013 0.0976 428

20 rs297705 4386321 ADRA1D intergenic G A 0.995 0.292 4.10E-05 -0.2581 0.0622 426

18 rs3893675 24085232 KCTD1 intronic T C 0.991 0.209 4.40E-05 0.2859 0.0697 424

3 rs2280162 63997491 PSMD6-AS2 intronic G A 0.986 0.175 4.50E-05 0.2937 0.0729 422

2 rs13027200 148709653 ORC4 intronic A G 1.000 0.291 4.60E-05 0.2562 0.0632 428

2 rs2890915 148789543 MBD5 intronic C T 0.993 0.295 4.70E-05 0.2568 0.0648 425

2 rs1014064 148612154 ACVR2A intronic A G 0.998 0.290 5.10E-05 0.2564 0.0638 427

2 rs7583759 148842379 MBD5 intronic G C 0.998 0.367 5.10E-05 -0.2416 0.0597 427

106

14 rs12880425 97203041 VRK1 intergenic C T 1.000 0.131 5.20E-05 0.3426 0.0846 428

14 rs8021182 32217812 NUBPL intronic T A 0.981 0.011 5.30E-05 1.0887 0.2687 420

3 rs7652997 178376259 KCNMB2 intronic G A 0.986 0.411 5.40E-05 0.2314 0.0582 422

5 rs12523145 112852568 YTHDC2 intronic A G 1.000 0.377 5.60E-05 -0.2426 0.0602 428

9 rs11143130 74756553 GDA intronic G A 1.000 0.042 5.90E-05 0.5749 0.1438 428

5 rs403334 158096651 EBF1 intergenic G T 0.970 0.204 6.20E-05 0.2673 0.0661 415

4 rs17628329 177184716 ASB5 intronic C T 0.998 0.096 6.60E-05 -0.3777 0.0941 427

6 rs6913021 92229590 MIR4643 intergenic T G 0.967 0.210 6.80E-05 -0.2804 0.0704 414

13 rs4415894 92725832 GPC5 intronic G A 1.000 0.067 6.90E-05 -0.4712 0.1188 428

13 rs983951 84505107 SLITRK1 intergenic G A 0.998 0.159 7.30E-05 0.3085 0.0637 427

2 rs7592576 148839547 MBD5 intronic G A 1.000 0.416 7.30E-05 -0.2319 0.0581 428

16 rs10514579 83359653 CDH13 intronic C G 0.986 0.327 7.60E-05 -0.2417 0.0605 422

13 rs7994558 92743267 GPC5 intronic G A 1.000 0.065 7.70E-05 -0.4722 0.1175 428

16 rs212079 16220126 ABCC1 intronic C T 1.000 0.049 7.80E-05 -0.5349 0.1327 428

2 rs16828270 148872237 MBD5 intronic G A 1.000 0.365 8.10E-05 -0.2352 0.0596 428

10 rs181601 119345838 EMX2 intergenic G A 0.998 0.148 8.40E-05 -0.3268 0.0823 427

8 rs13248658 109307585 EIF3E intergenic C A 1.000 0.256 8.80E-05 -0.2602 0.0666 428

107

Table S1I: Association statistics for the top 50 SNPs associated with GMCBF. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Major Minor Call MAF p-value Beta SE n Location/Impact Rate

2 rs2167565 61476458 USP34 intronic T C 1.000 0.316 5.36E-07 0.4507 0.0891 438

5 rs378147 82309115 TMEM167A intergenic T A 1.000 0.037 5.53E-07 1.0446 0.2028 438

5 rs37547 82296682 TMEM167A intergenic C A 0.995 0.036 1.43E-06 1.0158 0.2016 436

11 rs1075749 15791844 SOX6 intergenic C T 1.000 0.487 7.63E-06 0.4027 0.0864 438

5 rs4383715 82177857 TMEM167A intergenic G A 1.000 0.147 8.93E-06 0.5117 0.1140 438

13 rs4771833 92094008 GPC5 intronic T C 1.000 0.477 1.10E-05 0.3648 0.0836 438

13 rs8001719 99774724 DOCK9 intergenic A G 0.998 0.164 1.20E-05 0.4908 0.1130 437

13 rs928002 67728012 PCDH9 intronic T C 1.000 0.008 1.20E-05 2.0414 0.4441 438

14 rs4905962 100933029 WDR25 intronic C T 0.998 0.382 1.50E-05 -0.3964 0.0912 437

12 rs7979377 7832447 GDF3 intergenic A C 1.000 0.453 1.60E-05 -0.3628 0.0824 438

14 rs8005782 100936793 WDR25 intronic G A 1.000 0.382 1.70E-05 0.3942 0.0938 438

13 rs9560800 92108700 GPC5 intronic C A 0.993 0.475 1.80E-05 0.3571 0.0870 435

1 rs17163852 115913838 NGF intergenic C G 0.998 0.103 1.90E-05 0.6002 0.1396 437

14 rs11160589 100934125 WDR25 intronic G A 1.000 0.384 2.30E-05 0.3885 0.0914 438

108

11 rs1698187 115595946 CADM1 intergenic C A 1.000 0.053 2.40E-05 -0.7986 0.1902 438

20 rs1853092 2304376 TGM3 intronic A G 0.998 0.312 3.20E-05 -0.3874 0.0891 437

13 rs3848052 92077975 GPC5 intronic T C 0.989 0.485 3.20E-05 0.3470 0.0854 433

3 rs4677262 73376013 PDZRN3 intergenic C T 1.000 0.264 3.20E-05 -0.4057 0.0991 438

3 rs966163 73375302 PDZRN3 intergenic G A 1.000 0.264 3.20E-05 0.4057 0.0941 438

5 rs6877840 74611524 HMGCR intergenic T C 0.995 0.064 3.30E-05 0.7369 0.1753 436

13 rs4773630 92094974 GPC5 intronic A G 0.998 0.439 3.40E-05 -0.3451 0.0843 437

8 rs4297024 134138581 TG intronic G A 0.993 0.091 3.70E-05 -0.5832 0.1297 435

14 rs7492607 100945091 WDR25 intronic T C 1.000 0.388 4.10E-05 0.3722 0.0945 438

18 rs8097651 27232011 intergenic intergenic G T 1.000 0.130 4.40E-05 -0.5337 0.1296 438

8 rs7824846 121594384 SNTB1 intronic C G 1.000 0.194 4.60E-05 -0.4311 0.1043 438

5 rs17739025 74603716 HMGCR intergenic C T 0.991 0.059 4.90E-05 -0.7449 0.1725 434

7 rs2392590 38695129 AMPH intergenic G A 1.000 0.050 5.10E-05 0.7568 0.1839 438

7 rs6979339 11512425 THSD7A intronic C T 0.977 0.394 5.20E-05 0.3475 0.1140 428

20 rs2014017 2304755 TGM3 intronic C T 0.954 0.312 5.30E-05 -0.3879 0.0962 418

12 rs12423501 26643355 ITPR2 intronic C T 0.998 0.048 5.90E-05 -0.7126 0.1777 437

5 rs1423526 74606173 HMGCR intergenic T A 1.000 0.059 6.00E-05 0.7306 0.1811 438

5 rs12660028 74605502 HMGCR intergenic A T 1.000 0.063 6.20E-05 -0.7161 0.1798 438

5 rs4072112 74578784 HMGCR intergenic T C 1.000 0.063 6.20E-05 0.7161 0.1772 438

109

8 rs6469936 121594601 SNTB1 intronic A G 0.998 0.195 6.30E-05 -0.4220 0.1044 437

10 rs12414475 23661291 C10orf67 intergenic C T 1.000 0.215 6.40E-05 -0.4206 0.1086 438

3 rs4677261 73374442 PDZRN3 intergenic T C 0.984 0.262 6.60E-05 0.3917 0.0974 431

12 rs1565351 91984257 intergenic intergenic T G 0.998 0.112 7.30E-05 0.5089 0.1224 437

12 rs1214196 128298173 LINC00507 intergenic T C 1.000 0.079 7.70E-05 -0.6012 0.1513 438

10 rs1398024 23665438 C10orf67 intergenic G T 0.998 0.215 7.70E-05 -0.4162 0.1342 437

4 rs3114015 89029887 ABCG2 intronic G C 1.000 0.048 7.80E-05 -0.8159 0.2042 438

1 rs17037551 118221468 FAM46C intergenic T C 0.998 0.023 8.20E-05 -1.0997 0.2760 437

1 rs17037506 118191177 FAM46C intergenic C T 0.995 0.023 8.60E-05 1.1054 0.2882 436

21 rs2839081 47441315 COL6A1 intergenic C T 1.000 0.445 9.10E-05 0.3530 0.0921 438

1 rs2786107 197310777 CRB1 intronic T G 0.998 0.206 9.30E-05 0.3903 0.0981 437

2 rs9309333 61232705 PUS10 intronic G A 1.000 0.242 9.30E-05 0.3914 0.1038 438

5 rs10056109 103164240 intergenic intergenic C T 1.000 0.122 9.30E-05 0.5225 0.1332 438

16 rs723920 83342389 CDH13 intronic C A 0.998 0.397 9.60E-05 -0.3344 0.0879 437

4 rs10014766 113376846 ALPK1 intergenic G C 0.998 0.177 9.60E-05 0.4389 0.1100 437

8 rs10089762 56869085 LYN intronic G A 1.000 0.290 9.60E-05 0.3695 0.0994 438

8 rs2719240 56870048 LYN intronic C G 0.998 0.290 9.90E-05 -0.3691 0.0944 437

110

Table S2A: Association statistics for the top 50 SNPs associated with TBV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

7 exm597095 796593 HEATR2 Missense E478K G A 0.9935 0.0022 4.20E-05 -0.1429 0.0344 458

17 exm1370963 80885156 TBCD Missense M856V A G 1.0000 0.0022 1.07E-04 -0.1570 4.5057 458

8 exm_rs6473383 83669120 intergenic Intergenic - G A 1.0000 0.1354 1.30E-04 0.0289 0.1457 458

19 exm1467106 40430459 FCGBP Missense S495L G A 0.9891 0.0033 1.58E-04 -0.1132 0.0298 458

19 exm1490177 49797664 SLC6A16 Missense R459H G A 1.0000 0.0022 1.59E-04 -0.1526 4.3945 458

22 exm1595105 25246247 SGSM1 Synonymous A101A G A 1.0000 0.0022 1.59E-04 -0.1526 4.3945 458

4 exm_rs12641981 45179883 intergenic Intergenic - G A 1.0000 0.4432 1.68E-04 -0.0177 0.1365 458

4 exm_rs10938397 45182527 intergenic Intergenic - A G 1.0000 0.4432 1.68E-04 -0.0177 0.1365 458

5 exm444604 10683642 DAP Missense R65Q G A 1.0000 0.0022 1.76E-04 -0.1677 1.0816 458

15 exm1177136 75112974 LMAN1L Silent - A G 1.0000 0.0022 1.83E-04 -0.1511 4.5385 458

8 exm728531 144941181 EPPK1 Missense L2081V G C 1.0000 0.2762 2.04E-04 -0.0200 0.3456 458

1 exm15710 12010469 PLOD1 Missense A120S C A 1.0000 0.0906 2.37E-04 0.0317 0.0850 458

19 exm2272826 21837430 ZNF100 Intergenic - A G 1.0000 0.2926 2.94E-04 -0.0200 0.0055 458

111

20 exm1551052 52192512 ZNF217 Missense Q931E G C 1.0000 0.0033 3.12E-04 -0.1420 4.0448 458

1 exm39463 32120423 COL16A1 Missense I1443V A G 1.0000 0.0066 3.46E-04 0.1116 0.0310 458

8 exm726217 144406705 TOP1MT Missense V256I G A 1.0000 0.3090 3.86E-04 0.0183 0.9661 458

13 exm1067677 46629944 CPB2 Missense I310T G A 1.0000 0.3177 4.34E-04 -0.0177 0.1615 458

2 exm184934 32449832 NLRC4 Missense C968F C A 1.0000 0.0033 5.80E-04 0.1287 0.1586 458

13 exm1067460 46541673 ZC3H13 Missense E1429D A T 1.0000 0.3406 5.96E-04 -0.0174 1.4239 458

4 exm384164 3318726 RGS12 Missense M277L T A 1.0000 0.0437 5.99E-04 0.0394 0.8519 458

1 exm10535 7797503 CAMTA1 Missense N1177K C G 1.0000 0.1277 6.72E-04 -0.0245 0.5074 458

17 exm1343072 60042388 MED13 Missense P1608L G A 1.0000 0.0044 7.41E-04 -0.1192 0.0490 458

9 exm761648 95277356 ECM2 Missense R182Q G A 1.0000 0.0404 7.41E-04 -0.0438 3.6234 458

12 exm1002762 51237816 TMPRSS12 Missense A127T G A 1.0000 0.3821 7.48E-04 -0.0159 0.8894 458

1 exm_rs1335532 117100957 CD58 Intronic - A G 1.0000 0.1321 7.61E-04 -0.0239 0.1714 458

5 exm474197 123982685 ZNF608 Missense S1131T G C 1.0000 0.0066 8.03E-04 0.1157 0.0343 458

6 exm542518 36446975 KCTD20 Missense S171T G C 1.0000 0.1965 8.56E-04 0.0206 0.0788 458

3 exm347598 127379313 PODXL2 Missense K148E A G 1.0000 0.0033 8.94E-04 -0.1150 3.7586 458

9 exm2259056 104171523 ZNF189 Synonymous K491K A G 1.0000 0.0786 9.65E-04 -0.0304 0.8338 458

8 exm726892 144654594 C8orf73 Missense H97Q A C 1.0000 0.4782 9.68E-04 -0.0142 1.1288 458

15 exm2260451 59614841 MYO1E Intronic - G A 1.0000 0.1583 1.05E-03 -0.0172 6.0528 458

8 exm729377 144994955 PLEC Missense E3012K G A 1.0000 0.0022 1.13E-03 -0.1640 0.0502 458

112

2 exm_rs10931468 191538562 NAB1 Intronic - C A 1.0000 0.1343 1.19E-03 -0.0219 0.6868 458

19 exm1493619 50730169 MYH14 Missense I274V A G 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

3 exm339688 118623559 IGSF11 Missense A264T G A 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

1 exm72134 86818610 ODF2L Missense L527I T A 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

9 exm758523 87338515 NTRK2 Missense P204H C A 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

9 exm760720 94486078 ROR2 Missense A900T G A 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

11 exm878280 2323396 TSPAN32 Missense R6Q G A 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

11 exm939493 73007871 P2RY6 Missense R103H G A 1.0000 0.0022 1.20E-03 0.1595 0.0505 458

6 exm583937 144852201 UTRN Missense A1974T G A 1.0000 0.1987 1.21E-03 -0.0207 1.0043 458

14 exm1118513 80669318 DIO2 Missense S179F G A 1.0000 0.0033 1.33E-03 0.1600 0.0526 458

6 exm546584 42073031 C6orf132 Missense H873Q C G 1.0000 0.0033 1.33E-03 0.1600 0.0527 458

11 exm888404 7960311 OR10A3 Missense G253S G A 1.0000 0.0033 1.33E-03 0.1600 0.0527 458

11 exm965811 123894518 OR10G9 Missense V267M G A 1.0000 0.0033 1.33E-03 0.1600 0.0527 458

10 exm_rs501120 44753867 CXCL12 Intergenic - A G 1.0000 0.1397 1.34E-03 -0.0227 0.6150 458

11 exm885784 6578792 DNHD1 Missense V2756A A G 1.0000 0.0175 1.34E-03 -0.0613 0.0190 458

11 exm914753 60540907 MS4A15 Missense A57S C A 1.0000 0.0055 1.40E-03 -0.1265 2.2634 458

18 exm2253399 31319825 ASXL3 Synonymous A819A A G 1.0000 0.0066 1.41E-03 -0.1011 0.0662 458

14 exm_rs10498365 42700527 intergenic Intergenic - A G 1.0000 0.4902 1.42E-03 0.0145 0.2020 458

113

Table S2B: Association statistics for the top 50 SNPs associated with GMV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, T2D affected status, and intracranial volume. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

10 exm2267054 121871353 SEC23IP Intergenic - A G 1.0000 0.4880 1.30E-05 -7.2614 1.6439 458

10 exm814452 23482635 PTF1A Missense S263P A G 1.0000 0.4978 2.00E-05 7.2044 1.6806 458

4 exm409479 82096035 PRKG2 Missense M180I G A 1.0000 0.0044 2.20E-05 52.2056 12.3356 458

3 exm_rs13073817 18706858 SATB1 Intergenic - G A 1.0000 0.3428 2.50E-05 -7.5614 1.8073 458

11 exm2267210 94154902 MRE11A Intronic - G A 1.0000 0.3941 6.10E-05 6.6286 1.6348 458

11 exm890669 10673739 MRVI1 Missense I11V G A 1.0000 0.2926 7.00E-05 6.7910 1.6927 458

4 exm425165 139993175 ELF2 Missense P64L G A 1.0000 0.0120 1.10E-04 -32.2494 8.2350 458

11 exm_rs7940646 10669228 MRVI1 Intronic - G A 1.0000 0.2882 1.49E-04 6.5297 1.6454 458

20 exm1545771 44569202 PCIF1 Missense R113Q G A 1.0000 0.0033 1.78E-04 51.3434 13.8123 458

8 exm727162 144671776 EEF1D Missense V159A A G 1.0000 0.0044 1.84E-04 46.5962 12.4448 458

4 exm392314 22820504 GBA3 Missense X149Y T A 1.0000 0.1539 1.86E-04 -8.8288 2.3406 458

11 exm_rs7102710 14265613 SPON1 Intronic - A G 1.0000 0.0797 2.21E-04 -11.3790 3.1086 458

12 exm2267313 4496280 FGF23 Intergenic - G A 1.0000 0.4913 2.21E-04 5.8812 1.5930 458

114

16 exm1201838 1484473 CCDC154 Missense L647P G A 1.0000 0.0950 2.25E-04 10.5391 2.8298 458

3 exm290633 11744520 VGLL4 5' UTR - G A 1.0000 0.0033 2.73E-04 51.1452 14.0808 458

14 exm_rs17094273 97103807 PAPOLA Intergenic - G A 1.0000 0.1223 2.78E-04 9.6966 8.2935 458

17 exm_rs8066857 70696103 SLC39A11 Intronic - G A 1.0000 0.1561 2.82E-04 -8.2707 2.1659 458

10 exm859800 121203194 GRK5 Missense R399H G A 1.0000 0.0022 2.89E-04 59.8831 16.7883 458

12 exm2251025 50218644 NCKAP5L Intronic - A G 1.0000 0.4585 2.90E-04 6.0165 1.6190 458

14 exm1125245 95906321 C14orf49 Missense T668M G A 1.0000 0.3843 2.97E-04 6.3303 1.6700 458

5 exm462968 78135241 ARSB Missense S384N G A 1.0000 0.0590 3.10E-04 -12.4471 3.4349 458

7 exm647640 103202327 RELN Missense V1762I G A 1.0000 0.0033 3.23E-04 50.0208 13.9020 458

10 exm_rs10776612 49735563 ARHGAP22 Intronic - G A 1.0000 0.4487 3.23E-04 -6.1964 1.7145 458

16 exm1255219 70986425 HYDIN Missense K2143E A G 0.9956 0.0340 3.81E-04 -0.0382 0.0108 458

6 exm537448 33037626 HLA-DPA1 Synonymous T46T C G 1.0000 0.1572 3.96E-04 -8.2112 2.7738 458

13 exm1057973 25068808 PARP4 Missense F215Y A T 1.0000 0.1288 3.97E-04 -9.1196 2.5475 458

13 exm1057993 25074490 PARP4 Missense S122N G A 1.0000 0.1288 3.97E-04 -9.1196 2.5475 458

16 exm1223806 20966362 DNAH3 Missense Y3615C A G 1.0000 0.0819 4.12E-04 10.5239 2.9355 458

16 exm1223797 20966273 DNAH3 Missense R3645C G A 1.0000 0.0819 4.12E-04 10.5239 2.9355 458

10 exm857727 116064528 AFAP1L2 Missense L412F G A 1.0000 0.0175 4.61E-04 -21.5906 6.2432 458

5 exm505413 176017639 CDHR2 Missense L1164M C A 1.0000 0.2238 4.61E-04 -7.0308 1.9882 458

4 exm420051 114276880 ANK2 Missense V2369A A G 1.0000 0.1070 4.80E-04 -8.8842 2.5128 458

115

16 exm1217060 10783113 TEKT5 Missense M239T A G 1.0000 0.0055 4.86E-04 -39.6658 11.3290 458

8 exm726892 144654594 C8orf73 Missense H97Q A C 1.0000 0.4782 4.96E-04 -5.7336 1.6341 458

11 exm896873 22646800 FANCF Missense A186V G A 1.0000 0.0055 5.89E-04 37.2924 10.9088 458

14 exm1116107 75563805 NEK9 Missense V724A A G 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

15 exm1148623 40594181 PLCB2 Missense A187S C A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

15 exm1164228 56208983 NEDD4 Missense T16N C A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

7 exm655549 127975992 RBM28 Missense D99H C G 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

12 exm1054064 133277860 PXMP2 Missense A142T G A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

15 exm1186549 89195422 ISG20 Missense G104S G A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

21 exm1575583 45390557 AGPAT3 Synonymous T178T G A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

2 exm177177 25039615 CENPO Missense P226L G A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

21 exm2254660 45390557 AGPAT3 Synonymous T178T G A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

9 exm775588 117400945 C9orf91 Missense P263R G C 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

11 exm901202 40137800 LRRC4C Missense G15S G A 1.0000 0.0022 5.90E-04 61.8401 17.6643 458

16 exm1224465 21190853 TMEM159 Missense E154D C A 1.0000 0.2358 5.94E-04 -6.9466 1.9085 458

2 exm174707 18768318 NT5C1B Missense R81H G A 1.0000 0.0197 6.25E-04 21.6531 6.3903 458

12 exm_rs7953249 121403724 HNF1A Intergenic - A G 1.0000 0.4170 6.48E-04 5.8733 1.6015 458

2 exm2269073 66285871 Intergenic Intergenic - A G 1.0000 0.4913 6.77E-04 -5.7491 1.6482 458

116

Table S2C: Association statistics for the top 50 SNPs associated with WMV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, T2D affected status, and intracranial volume. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

2 exm2265371 178418575 TTC30B Intergenic - G A 1.0000 0.2162 1.50E-05 8.5461 1.9150 458

3 exm289064 9976159 CRELD1 Missense M13V G A 1.0000 0.0033 2.40E-05 -63.1208 14.7968 458

1 exm_rs983332 88132380 Intergenic Intergenic - C A 1.0000 0.1758 3.10E-05 9.2163 2.1952 458

16 exm1265118 84691434 KLHL36 Missense A341T G A 1.0000 0.0033 3.40E-05 -55.6428 13.7194 458

3 exm325013 56658871 FAM208A Missense I998V A G 1.0000 0.0928 4.40E-05 -11.7276 2.9077 458

11 exm944196 77920808 USP35 Missense G636D G A 1.0000 0.0218 4.70E-05 -23.0661 5.7632 458

11 exm909053 56019770 OR5T3 Missense P32L G A 1.0000 0.0033 9.90E-05 -54.3885 13.8931 458

6 exm586358 151152250 PLEKHG1 Missense D668G A G 1.0000 0.0207 1.13E-04 -22.8408 5.9590 458

11 exm_rs236918 117091609 PCSK7 Intronic - G C 1.0000 0.1114 1.53E-04 -9.4378 2.4900 458

22 exm_rs5754188 33048216 SYN3 Intronic - A C 1.0000 0.2118 1.74E-04 7.2742 1.8917 458

11 exm958032 117042377 PAFAH1B2 Missense V151M A G 1.0000 0.1135 1.94E-04 -9.2327 2.4672 458

11 exm_rs7112513 117037361 PAFAH1B2 Intronic - G A 1.0000 0.1135 1.94E-04 -9.2327 2.4672 458

10 exm859800 121203194 GRK5 Missense R399H G A 1.0000 0.0022 2.08E-04 -60.8021 17.8433 458

117

15 exm_rs1568679 37349802 MEIS2 Intronic - A G 1.0000 0.0644 2.17E-04 -11.9816 3.1819 458

22 exm1616623 45255688 ARHGAP8 Missense R319C G A 1.0000 0.0022 2.28E-04 -60.8792 16.5903 458

1 exm91510 145560918 ANKRD35 Missense A259T G A 1.0000 0.0426 2.68E-04 -15.0819 4.1550 458

2 exm261350 210560189 MAP2 Missense M1099V A G 1.0000 0.0317 3.21E-04 16.6388 4.5623 458

11 exm944222 77921348 USP35 Missense R816H G A 1.0000 0.0197 3.33E-04 -21.7540 6.0509 458

16 exm1272683 90075750 DBNDD1 Missense E40D C A 1.0000 0.0033 3.52E-04 -53.8153 14.8447 458

17 exm_rs12150338 1634104 WDR81 Synonymous A74A G A 1.0000 0.0590 3.54E-04 -12.2963 3.5215 458

10 exm842179 94594565 EXOC6 Missense V24I A G 1.0000 0.2467 3.72E-04 -6.8727 1.7735 458

15 exm1155139 43317071 UBR1 Missense I899V A G 1.0000 0.0284 4.06E-04 16.0052 4.5494 458

13 exm1057973 25068808 PARP4 Missense F215Y A T 1.0000 0.1288 4.14E-04 9.0424 2.5917 458

13 exm1057993 25074490 PARP4 Missense S122N G A 1.0000 0.1288 4.14E-04 9.0424 2.5917 458

5 exm498917 156376703 TIMD4 Missense V240A G A 1.0000 0.0764 4.33E-04 11.1949 3.1347 458

7 exm647210 102952123 PMPCB Missense E396D C A 1.0000 0.0764 4.54E-04 11.1356 3.1038 458

18 exm2253515 65193147 LOC643542 Intronic - C A 1.0000 0.4345 4.78E-04 -5.6020 1.6087 458

16 exm1255273 70996023 HYDIN Missense S1935L G A 1.0000 0.0055 4.82E-04 -36.5311 10.5820 458

14 exm2272055 23761094 HOMEZ Intergenic - G A 1.0000 0.3635 4.82E-04 -5.7734 1.6202 458

3 exm324920 56650054 CCDC66 Missense S606P A G 1.0000 0.1092 5.04E-04 -9.4797 2.7050 458

12 exm1036234 110383141 GIT2 Missense A552V G A 1.0000 0.0175 5.45E-04 -21.3835 6.1674 458

12 exm1039385 113553461 RASAL1 Missense R328W G A 1.0000 0.0044 5.66E-04 -42.2405 12.3895 458

118

3 exm337358 111993749 SLC9A10 Missense T203I G A 1.0000 0.0076 5.67E-04 35.1223 10.2989 458

11 exm_rs2075292 116732512 SIK3 Intronic - A C 1.0000 0.1234 5.77E-04 -8.3240 2.3913 458

16 exm1244176 57691387 GPR56 Missense A424T G A 1.0000 0.0033 5.86E-04 46.0171 13.4962 458

1 exm75855 94140470 BCAR3 Missense F6Y T A 1.0000 0.0164 6.04E-04 22.8693 6.6289 458

4 exm395987 40122453 N4BP2 Missense P908S G A 1.0000 0.0055 6.29E-04 -35.4386 10.7627 458

1 exm_rs7538876 17722363 PADI6 Intronic - G A 1.0000 0.3810 6.81E-04 5.9023 1.6459 458

16 exm1224585 21222672 ZP2 Missense G36V C A 1.0000 0.2948 6.93E-04 -6.2971 1.8748 458

9 exm736902 4719161 AK3 Missense I70F T A 1.0000 0.0022 7.12E-04 -55.7156 16.1417 458

1 exm2231567 158979950 IFI16 5' UTR - A G 1.0000 0.1681 7.32E-04 7.1060 2.1224 458

10 exm828515 69881254 MYPN Missense Y20C A G 1.0000 0.0098 7.35E-04 27.5335 8.0044 458

11 exm952045 103158278 DYNC2H1 Missense A3680V G A 1.0000 0.3319 7.39E-04 -6.0319 1.5184 458

11 exm905790 47726274 AGBL2 Missense H136L T A 1.0000 0.0044 7.45E-04 -40.2748 11.9351 458

11 exm877731 1902793 LSP1 Missense Q46L T A 1.0000 0.1004 7.65E-04 9.1690 2.9103 458

20 exm1552752 56735759 C20orf85 Missense I99V A G 1.0000 0.0699 7.66E-04 10.3961 3.1301 458

17 exm1342832 59763347 BRIP1 Missense S919P G A 1.0000 0.3821 7.89E-04 5.5640 1.8502 458

1 exm2252525 62676714 L1TD1 Synonymous T756T A C 0.9978 0.0088 8.07E-04 27.6059 8.3949 457

11 exm966648 124487291 PANX3 Missense K149T A C 1.0000 0.0033 8.24E-04 -44.5772 13.2668 458

7 exm_rs1178979 72856430 BAZ1B Intronic - A G 1.0000 0.1976 8.53E-04 -6.9595 2.0637 458

119

Table S2D: Association statistics for the top 50 SNPs associated with GMFA. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

1 exm90168 144866673 PDE4DIP Missense R1857C G A 1.0000 0.0033 1.48E-06 -0.0441 0.0089 454

22 exm1605352 37457647 KCTD17 Missense R268C G A 1.0000 0.0022 1.30E-05 -0.0464 0.0105 454

19 exm1408876 4354080 MPND Missense R235W G A 1.0000 0.0055 1.80E-05 -0.0325 0.0074 454

9 exm2271088 74853935 GDA Intronic - A G 1.0000 0.4449 2.00E-05 -0.0048 0.0011 454

3 exm2265461 68102663 FAM19A1 Intronic - G A 1.0000 0.4559 4.10E-05 0.0043 0.0010 454

12 exm1051018 129566566 TMEM132D Missense D554G A G 0.9934 0.0055 4.80E-05 0.0257 0.0063 454

14 exm1122116 92583867 NDUFB1 Missense R86W G A 1.0000 0.0022 5.20E-05 0.0428 0.0088 454

15 exm1190227 91424905 FURIN Missense V728I G A 1.0000 0.0044 5.20E-05 -0.0325 0.0080 454

21 exm2272993 17609813 LINC00478 Intronic - A G 1.0000 0.0055 5.30E-05 -0.0305 0.0073 454

17 exm_rs3760318 29247715 ADAP2 Intergenic - G A 1.0000 0.3833 6.10E-05 0.0045 0.0011 454

5 exm489077 140784636 PCDHGA9 Missense T706I G A 1.0000 0.0066 6.40E-05 -0.0209 0.0051 454

22 exm1588462 21384598 SLC7A4 Missense R342H G A 1.0000 0.0066 6.40E-05 -0.0209 0.0051 454

13 exm1056201 21563413 LATS2 Missense R169Q G A 1.0000 0.0033 6.40E-05 -0.0418 0.0102 454

120

15 exm1169144 64970496 ZNF609 Missense T1195M G A 1.0000 0.0033 6.40E-05 -0.0418 0.0102 454

16 exm1196538 460358 DECR2 Missense K151N C A 1.0000 0.0033 6.40E-05 -0.0418 0.0102 454

22 exm1602665 32792192 C22orf28 Missense I287V A G 1.0000 0.0033 6.40E-05 -0.0418 0.0101 454

2 exm171656 9628433 IAH1 Missense S241N G A 1.0000 0.0033 6.40E-05 -0.0418 0.0104 454

3 exm298901 37067468 MLH1 Missense E219A A C 1.0000 0.0033 6.40E-05 -0.0418 0.0102 454

4 exm385449 5399970 STK32B Synonymous H157H G A 1.0000 0.0033 6.40E-05 -0.0418 0.0101 454

5 exm509550 178699945 ADAMTS2 Missense P219S G A 1.0000 0.0033 6.40E-05 -0.0418 0.0096 454

7 exm671549 150656690 KCNH2 Missense R148W G A 1.0000 0.0033 6.40E-05 -0.0418 0.0101 454

9 exm765450 100245383 TDRD7 Missense M889V A G 1.0000 0.0033 6.40E-05 -0.0418 0.0101 454

10 exm842335 94825984 CYP26C1 Missense R378H G A 1.0000 0.0033 6.40E-05 -0.0418 0.0105 454

1 exm22700 17292223 CROCC Missense G1471R G A 1.0000 0.0474 6.60E-05 0.0100 0.0025 454

5 exm_rs7717527 83856911 EDIL3 Intergenic - A G 1.0000 0.1564 7.30E-05 0.0057 0.0014 454

1 exm22640 17281235 CROCC Missense V1110M G A 1.0000 0.0485 7.90E-05 0.0098 0.0025 454

6 exm2257432 105991031 PREP Intergenic - G A 1.0000 0.4868 8.00E-05 -0.0041 0.0010 454

19 exm1415752 7533891 ARHGEF18 Missense A875T G A 1.0000 0.0066 8.40E-05 0.0254 0.0065 454

8 exm688969 22777710 PEBP4 Missense P82L G A 1.0000 0.0066 9.10E-05 -0.0275 0.0069 454

6 exm_rs13194984 26500563 BTN1A1 Intergenic - C A 1.0000 0.1531 9.60E-05 0.0060 0.0015 454

9 exm742545 20881957 KIAA1797 Missense A802V G A 1.0000 0.0077 1.04E-04 -0.0229 0.0060 454

7 exm647709 103234202 RELN Missense G1280E G A 1.0000 0.0077 1.13E-04 -0.0236 0.0060 454

121

1 exm93972 149920922 OTUD7B Missense G396V C A 1.0000 0.0055 1.14E-04 -0.0277 0.0071 454

9 exm794292 136633699 VAV2 Missense I818M C G 1.0000 0.0099 1.18E-04 -0.0218 0.0056 454

19 exm1441201 17305607 MYO9B Missense E1124G A G 1.0000 0.0033 1.25E-04 -0.0337 0.0088 454

20 exm1535473 33330672 NCOA6 Missense A1130T G A 1.0000 0.0033 1.27E-04 -0.0355 0.0091 454

20 exm1536661 33834686 MMP24 Missense R97Q G A 1.0000 0.0033 1.27E-04 -0.0355 0.0093 454

3 exm377362 196771513 DLG1 Missense P887L G A 1.0000 0.0617 1.32E-04 -0.0084 0.0022 454

6 exm528268 30647035 PPP1R18 Missense D581E C A 1.0000 0.0077 1.49E-04 0.0224 0.0053 454

4 exm2256268 108120442 DKK2 Intergenic - A G 1.0000 0.3480 1.52E-04 0.0041 0.0011 454

3 exm340330 119327670 PLA1A Missense R110H G A 1.0000 0.0253 1.55E-04 0.0129 0.0034 454

13 exm_rs7992643 100555038 CLYBL Intergenic - C G 1.0000 0.3910 1.63E-04 0.0042 0.0011 454

4 exm2269876 168965733 ANXA10 Intergenic - C A 1.0000 0.3480 1.70E-04 -0.0043 0.0011 454

9 exm_rs216345 33799370 PRSS3 Intergenic - G A 1.0000 0.3943 2.01E-04 -0.0041 0.0011 454

11 exm956843 113803108 HTR3B Missense S156R A C 0.9978 0.0055 2.14E-04 0.0239 0.0064 454

2 exm_rs16866933 180566678 ZNF385B Intronic - G A 1.0000 0.0793 2.26E-04 -0.0071 0.0019 454

11 exm966648 124487291 PANX3 Missense K149T A C 1.0000 0.0033 2.29E-04 -0.0321 0.0086 454

16 exm2252608 49927513 ZNF423 Intergenic - G A 1.0000 0.4141 2.32E-04 -0.0041 0.0011 454

1 exm2250419 192627431 RGS13 Synonymous R76R G A 1.0000 0.0242 2.42E-04 -0.0120 0.0033 454

6 exm549099 43250558 TTBK1 Missense D694N G A 1.0000 0.0077 2.49E-04 -0.0219 0.0060 454

122

Table S2E: Association statistics for the top 50 SNPs associated with WMFA. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

2 exm212485 96798305 ASTL Missense R204H G A 1.0000 0.0319 2.60E-05 0.0163 0.0038 454

13 exm2267587 26983382 CDK8 intergenic - G A 1.0000 0.0881 3.10E-05 0.0101 0.0025 454

8 exm722865 139163966 FAM135B Missense L918F G A 1.0000 0.0022 5.10E-05 -0.0542 0.0129 454

22 exm1605352 37457647 KCTD17 Missense R268C G A 1.0000 0.0022 5.50E-05 -0.0541 0.0132 454

5 exm_rs7717527 83856911 EDIL3 intergenic - A G 1.0000 0.1564 5.50E-05 0.0074 0.0018 454

12 exm978696 6884585 LAG3 Missense T310S T A 1.0000 0.0044 6.60E-05 -0.0376 0.0092 454

5 exm2270115 117982046 DTWD2 intergenic - G A 1.0000 0.4482 6.90E-05 0.0053 0.0013 454

10 exm_rs7085433 51593354 TIMM23 Intronic - G A 1.0000 0.0892 7.00E-05 0.0091 0.0023 454

10 exm2271415 76180335 ADK Intronic - G A 1.0000 0.3910 9.00E-05 0.0053 0.0014 454

5 exm465494 82834299 VCAN Missense R839H G A 1.0000 0.4064 9.20E-05 0.0054 0.0014 454

5 exm465638 82837631 VCAN Missense D1950Y C A 1.0000 0.4064 9.20E-05 0.0054 0.0014 454

3 exm314478 49294250 CCDC36 Missense D440E C A 1.0000 0.0969 1.18E-04 -0.0087 0.0022 454

3 exm314527 49313978 C3orf62 Missense E110K G A 1.0000 0.0969 1.18E-04 -0.0087 0.0023 454

123

7 exm613083 31692270 CCDC129 Missense D988N G A 1.0000 0.0066 1.52E-04 -0.0299 0.0079 454

3 exm312709 48691316 CELSR3 Missense Q1758R A G 1.0000 0.1134 1.57E-04 -0.0081 0.0021 454

1 exm159723 233353901 PCNXL2 Missense A845V G A 1.0000 0.0330 1.59E-04 -0.0147 0.0039 454

6 exm590388 158510956 SYNJ2 Missense R1181Q G A 1.0000 0.0066 1.81E-04 -0.0343 0.0090 454

19 exm1457459 36232024 IGFLR1 Missense P20S G A 1.0000 0.0033 1.82E-04 -0.0401 0.0106 454

2 exm221835 113520129 CKAP2L Missense L19F G A 1.0000 0.0749 1.99E-04 0.0093 0.0025 454

19 exm1414039 6467910 DENND1C Missense K671E A G 1.0000 0.0297 2.00E-04 -0.0144 0.0035 454

7 exm612999 31618008 CCDC129 Missense H377R G A 1.0000 0.0088 2.01E-04 -0.0254 0.0068 454

12 exm997787 48368541 COL2A1 Missense V1331I G A 1.0000 0.0848 2.52E-04 -0.0089 0.0024 454

12 exm997717 48359676 TMEM106C Missense V103F C A 1.0000 0.0837 2.88E-04 -0.0088 0.0024 454

3 exm312351 48659009 TMEM89 Missense P61T C A 1.0000 0.0628 3.03E-04 -0.0100 0.0028 454

15 exm1175481 74032479 C15orf59 Missense P221A C G 1.0000 0.0066 3.07E-04 -0.0310 0.0085 454

7 exm_rs10270308 105671267 CDHR3 Synonymous D778D A G 1.0000 0.0055 3.09E-04 -0.0329 0.0091 454

3 exm313069 48732480 IP6K2 Missense D140G A G 1.0000 0.1145 3.48E-04 -0.0076 0.0021 454

1 exm51014 43215930 LEPRE1 Missense M549I G A 1.0000 0.0628 3.50E-04 -0.0101 0.0028 454

6 exm583861 144820481 UTRN Missense S1561C C G 1.0000 0.0044 3.90E-04 -0.0328 0.0081 454

3 exm312817 48697654 CELSR3 Missense S805T G C 1.0000 0.1101 4.00E-04 -0.0077 0.0021 454

3 exm315947 49749976 RNF123 Missense R854H G A 1.0000 0.0914 4.21E-04 -0.0082 0.0023 454

2 exm2261016 123252034 intergenic intergenic - A G 1.0000 0.1399 4.52E-04 -0.0068 0.0020 454

124

13 exm2267561 93751957 GPC6 intergenic - G A 1.0000 0.3623 4.63E-04 0.0048 0.0013 454

15 exm1152762 42171462 SPTBN5 Missense V1025I G A 1.0000 0.2478 4.75E-04 -0.0055 0.0016 454

6 exm552656 46661488 TDRD6 Missense N1875D A G 1.0000 0.0110 5.02E-04 -0.0217 0.0063 454

21 exm2234960 35515333 MRPS6 3' UTR - A G 1.0000 0.0617 5.12E-04 -0.0098 0.0028 454

18 exm1384615 43481075 EPG5 Missense A1511V G A 1.0000 0.1498 5.26E-04 0.0067 0.0020 454

12 exm2251099 58158558 CYP27B1 Synonymous L314L G A 1.0000 0.0341 5.37E-04 -0.0124 0.0036 454

22 exm1599932 30953295 GAL3ST1 Missense V29M G A 1.0000 0.3315 5.47E-04 0.0052 0.0014 454

5 exm485116 140308681 PCDHAC1 Missense P735L G A 1.0000 0.0485 5.54E-04 0.0107 0.0031 454

6 exm526155 29556100 OR2H2 Missense P127S G A 1.0000 0.0044 5.57E-04 -0.0354 0.0102 454

1 exm159697 233313594 PCNXL2 Missense A1076E C A 1.0000 0.0341 5.59E-04 -0.0132 0.0038 454

5 exm496382 150431808 TNIP1 Missense Q214K C A 1.0000 0.0033 5.70E-04 0.0378 0.0109 454

2 exm2265201 131037447 CCDC115 intergenic - A G 1.0000 0.3888 5.89E-04 0.0050 0.0015 454

15 exm1146681 34648935 C15orf55 Missense S881I C A 1.0000 0.0826 6.01E-04 0.0079 0.0023 454

15 exm1146708 34649631 C15orf55 Missense R1113H G A 1.0000 0.0826 6.01E-04 0.0079 0.0023 454

15 exm1149480 40914831 CASC5 Missense T816S G C 1.0000 0.0110 6.03E-04 -0.0216 0.0063 454

20 exm1526197 14307024 FLRT3 Missense A377T G A 1.0000 0.0055 6.08E-04 0.0305 0.0089 454

1 exm88255 118483474 WDR3 Missense P234A C G 1.0000 0.0088 6.23E-04 -0.0232 0.0067 454

7 exm616774 43906513 MRPS24 Missense W97R A G 0.9978 0.0585 6.28E-04 -0.0102 0.0030 454

125

Table S2F: Association statistics for the top 50 SNPs associated with GMMD. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

5 exm439781 156902 PLEKHG4B Missense A432V G A 1.0000 0.0044 2.14E-07 -0.1880 0.0352 454

7 exm2270764 79665088 GNAI1 intergenic - A C 1.0000 0.0022 3.39E-06 -0.1959 0.0416 454

15 exm1156601 43827387 PPIP5K1 Missense D1236H C G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

14 exm1113099 73731310 PAPLN Missense V974M G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

14 exm1114043 74411502 FAM161B Missense R217Q G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

14 exm1117222 77580068 KIAA1737 Missense T203A A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

14 exm1120796 91365693 RPS6KA5 Missense H485N C A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

14 exm1129058 102912148 TECPR2 Missense R980T G C 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1144074 29561291 NDNL2 Missense V207I G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1144107 29561866 NDNL2 Missense A15V G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1145379 33261366 FMN1 Missense D623N G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1145712 33954764 RYR3 Missense D1678G A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1146290 34394926 PGBD4 Missense T65I G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

126

15 exm1146568 34640218 C15orf55 Missense P22L A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1149036 40698039 IVD Missense A7V G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1149042 40698116 IVD Missense H33Y G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1149111 40750677 BAHD1 Missense R5Q G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1149418 40913189 CASC5 Missense I243V A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1149570 40917393 CASC5 Missense K1644R A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1150020 41146864 SPINT1 Missense R365H G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1151168 41815516 RPAP1 Missense Q825E C G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1151277 41829316 RPAP1 Missense S3L G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1151832 42108794 MAPKBP1 Missense V511L C A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1156379 43816543 MAP1A Missense Y958H A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1157828 45047125 TRIM69 Missense D12N G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1157929 45335602 SORD Missense V83A A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1158111 45392382 DUOX2 Missense E1017G A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1158601 45442896 DUOX1 Missense I962T A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1159330 48426484 SLC24A5 Missense T111A A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1160017 49048128 CEP152 Missense V1106A A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1163038 52901321 FAM214A Missense T597M G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1163393 54306925 UNC13C Missense G609S A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

127

15 exm1180284 78572759 DNAJA4 Missense D413G A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

16 exm1195447 309524 ITFG3 Missense R104K G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

16 exm1197426 685313 C16orf13 Missense L130F G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

16 exm1216870 10534287 ATF7IP2 Missense P388S G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

16 exm1230253 28856081 TUFM Missense E208K G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

17 exm1301061 18186148 TOP3A Missense E629K G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

1 exm136285 201023671 CACNA1S Missense G1210R G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

18 exm1388979 56203428 ALPK2 Missense R1331W G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

19 exm1397123 757319 C19orf21 Missense R125C G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

19 exm1463891 39230791 CAPN12 Missense G210A C G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

20 exm1536864 33875698 FAM83C Missense R295L C A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

20 exm1542931 42682966 TOX2 Missense D227N G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

15 exm1832844 48722907 FBN1 Missense P2278S G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

3 exm2255927 195516364 MUC4 Missense A696V G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

3 exm2256108 57291038 APPL1 Synonymous P402P A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

3 exm2269613 34833364 intergenic intergenic - C A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

2 exm245696 179410693 TTN Missense I22817T A G 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

2 exm245934 179418346 TTN Missense V20856M G A 1.0000 0.0022 4.03E-06 -0.2476 0.0535 454

128

Table S2G: Association statistics for the top 50 SNPs associated with WMMD. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

5 exm439781 156902 PLEKHG4B Missense A432V G A 1.0000 0.0044 3.66E-07 -0.1215 0.0237 454

4 exm379972 981684 IDUA Missense H82Q C G 1.0000 0.0033 2.24E-06 -0.1207 0.0252 454

17 exm1297354 15848848 ADORA2B Missense L96F G A 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

17 exm1347790 65822320 BPTF Missense D160E C G 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

17 exm1348318 66303666 ARSG Missense A11V G A 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

2 exm199755 70409090 C2orf42 Missense V10L G C 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

2 exm200616 71221876 TEX261 Missense M4I G A 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

2 exm255303 198338605 COQ10B Missense N225S A G 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

3 exm333383 99513830 COL8A1 Missense R362Q G A 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

1 exm68394 74507423 LRRIQ3 Missense I398V A G 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

10 exm835197 75567980 NDST2 Missense S56N G A 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

11 exm879628 3249681 MRGPRE Missense V116I G A 1.0000 0.0022 5.93E-06 -0.1421 0.0312 454

13 exm1075211 96579380 UGGT2 Missense V701G A C 1.0000 0.0022 1.50E-05 0.1256 0.0279 454

129

14 exm1120204 90429694 TDP1 Missense S79T G C 1.0000 0.0022 1.50E-05 0.1256 0.0282 454

22 exm1603378 35730426 TOM1 Missense A378V G A 1.0000 0.0022 1.50E-05 0.1256 0.0285 454

16 exm1235827 31152818 PRSS36 Missense V625I G A 1.0000 0.0121 1.90E-05 -0.0565 0.0132 454

6 exm519336 20488470 E2F3 Missense A376T G A 1.0000 0.0044 3.20E-05 -0.0950 0.0223 454

7 exm663577 141747646 MGAM Missense L854F G A 1.0000 0.0022 3.50E-05 -0.1189 0.0285 454

6 exm525400 28963010 ZNF311 Missense T590K C A 1.0000 0.0022 3.50E-05 -0.1189 0.0285 454

6 exm531711 31627273 C6orf47 Missense R151H G A 1.0000 0.0022 3.50E-05 -0.1189 0.0285 454

7 exm607734 20201458 MACC1 Missense R10W G A 1.0000 0.0022 3.50E-05 -0.1189 0.0285 454

20 exm1540656 37255691 ARHGAP40 Missense G130R G C 1.0000 0.0033 3.90E-05 0.0994 0.0239 454

9 exm739532 13188953 MPDZ Missense S732T T A 1.0000 0.0022 4.20E-05 0.1260 0.0309 454

7 exm2262270 152211627 KMT2C intergenic - C A 1.0000 0.0033 4.90E-05 -0.1065 0.0260 454

3 exm289064 9976159 CRELD1 Missense M13V G A 1.0000 0.0033 6.00E-05 -0.1133 0.0279 454

3 exm334305 100963593 IMPG2 Missense I528V A G 1.0000 0.0033 6.00E-05 -0.1024 0.0232 454

21 exm1576264 45670790 DNMT3L Missense R271Q G A 1.0000 0.0044 6.40E-05 0.0849 0.0211 454

12 exm1053561 133196062 P2RX2 Missense E71K G A 1.0000 0.0033 8.60E-05 -0.0890 0.0232 454

11 exm947908 92624235 FAT3 Missense L4544F G A 1.0000 0.0033 9.20E-05 0.0930 0.0238 454

7 exm623473 64451699 ERV3-1 Missense N569S G A 1.0000 0.0672 1.06E-04 -0.0229 0.0058 454

7 exm623486 64451963 ERV3-1 Missense N481S G A 1.0000 0.0672 1.06E-04 -0.0229 0.0058 454

7 exm623562 64453136 ERV3-1 Missense T90I A G 1.0000 0.0672 1.06E-04 -0.0229 0.0058 454

130

7 exm657942 130041748 CEP41 Missense P206A G C 1.0000 0.0055 1.22E-04 -0.0842 0.0204 454

17 exm1338607 54940127 DGKE Missense Q560R A G 1.0000 0.0055 1.30E-04 -0.0786 0.0201 454

1 exm42103 33957152 ZSCAN20 Missense Y432D C A 1.0000 0.0033 1.34E-04 -0.0909 0.0310 454

4 exm394953 38879949 FAM114A1 Missense G84R G A 1.0000 0.2775 1.64E-04 -0.0123 0.0032 454

4 exm394963 38880046 FAM114A1 Missense L116P A G 1.0000 0.2775 1.64E-04 -0.0123 0.0032 454

15 exm1156601 43827387 PPIP5K1 Missense D1236H C G 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

14 exm1113099 73731310 PAPLN Missense V974M G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

14 exm1114043 74411502 FAM161B Missense R217Q G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

14 exm1117222 77580068 KIAA1737 Missense T203A A G 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

14 exm1120796 91365693 RPS6KA5 Missense H485N C A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

14 exm1129058 102912148 TECPR2 Missense R980T G C 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1144074 29561291 NDNL2 Missense V207I G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1144107 29561866 NDNL2 Missense A15V G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1145379 33261366 FMN1 Missense D623N G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1145712 33954764 RYR3 Missense D1678G A G 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1146290 34394926 PGBD4 Missense T65I G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1146568 34640218 C15orf55 Missense P22L A G 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

15 exm1149036 40698039 IVD Missense A7V G A 1.0000 0.0022 1.83E-04 -0.1384 0.0363 454

131

Table S2H: Association statistics for the top 50 SNPs associated with WMLV. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, T2D affected status, and intracranial volume. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

19 exm1477394 44610665 ZNF224 Missense M118V G A 1.0000 0.1506 8.71E-06 0.3591 0.0802 425

19 exm1477399 44610798 ZNF224 Missense H162L A T 1.0000 0.1506 8.71E-06 0.3591 0.0793 425

16 exm1201371 1395257 BAIAP3 Missense R622C G A 1.0000 0.0035 2.10E-05 1.9057 0.4392 425

1 exm2268943 161983089 OLFML2B Intronic - G A 1.0000 0.3094 2.40E-05 0.2711 0.0638 425

19 exm1396626 624894 POLRMT Missense Q322L T A 1.0000 0.0094 3.30E-05 1.1826 0.2819 425

15 exm1185854 86312294 KLHL25 Missense V250I G A 1.0000 0.1094 3.80E-05 0.3806 0.0827 425

2 exm231661 148716428 ORC4 Missense N78S A G 1.0000 0.2941 1.08E-04 -0.2447 0.0652 425

21 exm1564189 30339234 LTN1 Missense S573G A G 1.0000 0.1082 1.12E-04 0.3493 0.0892 425

12 exm1029765 101680273 UTP20 Missense M167I G A 1.0000 0.0024 1.24E-04 2.1530 0.5483 425

3 exm306137 44879843 KIF15 Missense S1083L G A 1.0000 0.0082 1.34E-04 1.2362 0.3236 425

14 exm1100953 51224417 NIN Missense P1111A C G 1.0000 0.2565 1.48E-04 -0.2629 0.0692 425

4 exm2265744 111055435 ELOVL6 Intronic - G A 1.0000 0.3753 1.63E-04 -0.2196 0.0579 425

3 exm359875 153839806 ARHGEF26 Missense F9I A T 1.0000 0.0035 1.73E-04 1.6605 0.0088 425

132

19 exm1421302 9066390 MUC16 Missense S7019L G A 1.0000 0.0282 2.38E-04 0.6425 0.1725 425

3 exm2265653 195305994 APOD Intronic - A G 1.0000 0.0235 2.40E-04 0.6469 0.1782 425

1 exm2268644 4037665 LOC728716 Intergenic - G A 1.0000 0.4047 2.48E-04 0.2180 0.0589 425

3 exm327706 63967900 ATXN7 Missense K264R A G 1.0000 0.1329 2.53E-04 0.3101 0.0856 425

1 exm109341 156518379 IQGAP3 Missense R663C G A 1.0000 0.2706 2.57E-04 0.2438 0.0698 425

2 exm2255308 28645120 FOSL2 Intergenic - A C 1.0000 0.4965 2.78E-04 0.2108 0.0595 425

1 exm111550 157516861 FCRL5 Missense R60Q G A 1.0000 0.0059 2.95E-04 1.4146 0.3905 425

2 exm238747 167266447 SCN7A Splice variant - A G 1.0000 0.0188 2.98E-04 0.7294 0.2021 425

15 exm2272287 98768182 FAM169B Intergenic - G A 1.0000 0.4812 3.27E-04 0.2092 0.0574 425

16 exm1264226 84228770 ADAD2 Missense G235R G C 1.0000 0.3059 3.46E-04 0.2241 0.0623 425

2 exm2269152 148875643 MBD5 Intronic - A G 1.0000 0.4141 3.46E-04 -0.2115 0.0525 425

1 exm118754 161993192 OLFML2B Missense Y10C A G 1.0000 0.0671 3.57E-04 0.4230 0.1130 425

11 exm917183 61897409 INCENP Missense A137V G A 1.0000 0.0035 3.85E-04 1.7256 0.4811 425

13 exm2251606 98289698 MBNL2 Intergenic - A G 1.0000 0.3282 4.24E-04 -0.2100 0.0594 425

12 exm1025889 91347643 C12orf12 Missense E293K G A 1.0000 0.0447 4.39E-04 -0.4881 0.1396 425

10 exm2271415 76180335 ADK Intronic - G A 1.0000 0.3776 4.56E-04 -0.2080 0.0588 425

18 exm1380280 28604374 DSC3 Missense N239S A G 1.0000 0.0200 4.60E-04 0.7216 0.2049 425

12 exm997647 48240233 VDR Synonymous A303A G A 1.0000 0.0059 4.61E-04 -1.3444 0.3816 425

19 exm1516318 58861808 A1BG Missense C374R A G 1.0000 0.0059 4.70E-04 1.2198 0.3473 425

133

12 exm1023767 80661098 OTOGL Missense S810L G A 1.0000 0.0082 4.74E-04 1.3080 0.1315 425

18 exm1375596 10750150 PIEZO2 Missense H1376Q G C 1.0000 0.0400 5.03E-04 -0.5337 0.1537 425

17 exm1346909 64023624 CEP112 Missense K551E G A 1.0000 0.4541 5.22E-04 0.1954 0.0513 425

21 exm1564180 30339120 LTN1 Missense G611C C A 1.0000 0.1459 5.40E-04 0.2800 0.0785 425

18 exm2260814 7630982 PTPRM Intronic - A G 1.0000 0.1694 5.49E-04 0.2668 0.0767 425

1 exm30955 23763676 ASAP3 Missense A421V G A 1.0000 0.0035 5.86E-04 1.7312 0.4920 425

19 exm1418238 8176947 FBN3 Missense D1292G A G 1.0000 0.0059 6.05E-04 1.2676 0.3672 425

12 exm1038037 112375990 TMEM116 Missense C22G A C 1.0000 0.1965 6.07E-04 -0.2599 0.0749 425

11 exm_rs12280753 116613660 BUD13 Intergenic - G A 1.0000 0.0753 6.18E-04 0.3757 0.1088 425

12 exm2259959 112211833 ALDH2 Intronic - A G 1.0000 0.3506 6.52E-04 -0.2068 0.0606 425

11 exm_rs10767942 32522007 WT1 Intergenic - A G 1.0000 0.1988 6.57E-04 0.2492 0.0700 425

2 exm248742 179605725 TTN Missense I3841V A G 1.0000 0.0141 6.62E-04 0.8155 0.0056 425

1 exm102877 153747981 SLC27A3 Missense A50V G A 1.0000 0.0035 6.80E-04 1.5719 0.4192 425

20 exm1555580 60889385 LAMA5 Missense Q2827K C A 1.0000 0.0177 6.88E-04 -0.7485 0.2180 425

4 exm2261777 91340668 FAM190A Intronic - A C 1.0000 0.1259 7.10E-04 0.2926 0.0853 425

17 exm1329566 42987524 GFAP Missense T426A A G 1.0000 0.2506 7.20E-04 0.2384 0.0714 425

12 exm1001002 49994785 FAM186B Missense T213M G A 1.0000 0.0212 7.32E-04 0.6162 0.1705 425

1 exm_rs17464857 222762709 TAF1A Intronic - A C 1.0000 0.1918 7.52E-04 0.2513 0.0548 425

134

Table S2I: Association statistics for the top 50 SNPs associated with GMCBF. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. Associations based on a single observation of the rare allele have been removed. Effects are reported for the minor allele.

Alleles Association Statistics

Chr SNP Pos Gene Variant Amino Major Minor Call MAF p-value Beta SE n Location/Impact Acid Rate Change

8 exm715710 110457592 PKHD1L1 Missense P1832S G A 1.0000 0.0034 5.78E-07 3.5513 0.7018 435

3 exm342511 121712412 ILDR1 Missense L395S A G 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

7 exm644275 100679024 MUC17 Missense I1443V A G 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

17 exm1285570 6945936 SLC16A11 Missense D189H C G 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

7 exm642839 100468078 TRIP6 Missense E274D C G 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

16 exm1235048 30999108 HSD3B7 Missense T184M G A 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

1 exm134316 197112078 ASPM Missense S435L G A 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

2 exm179997 27355487 PREB Missense R246C G A 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

2 exm244421 178494181 PDE11A Missense S669L G A 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

2 exm256521 201758012 NIF3L1 Missense N160K G C 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

2 exm275493 233393033 CHRND Missense R87H G A 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

6 exm574088 117114050 GPRC6A Missense T679M G A 1.0000 0.0023 2.52E-06 4.2474 0.8810 435

2 exm256047 201436078 SGOL2 Missense A337S C A 1.0000 0.0034 3.16E-06 3.3508 0.7175 435

135

19 exm1453027 33467575 C19orf40 Missense T212M G A 1.0000 0.0034 3.99E-06 3.3435 0.7116 435

2 exm242511 173883491 RAPGEF4 Missense G706R G A 1.0000 0.0034 4.43E-06 3.3149 0.7279 435

19 exm1428210 11408905 TSPAN16 Missense Y53D A C 1.0000 0.0069 6.36E-06 2.4448 0.5819 435

5 exm510951 180001061 CNOT6 Missense H512R A G 1.0000 0.0034 2.00E-05 3.0873 0.7178 435

12 exm976798 6143978 VWF Missense R854Q G A 1.0000 0.0046 2.30E-05 2.6406 0.6183 435

19 exm1459688 37211557 ZNF567 Missense I613T A G 1.0000 0.0057 3.30E-05 2.3200 0.5579 435

3 exm366908 182923984 MCF2L2 Missense R911C G A 1.0000 0.0034 4.20E-05 2.9869 0.7187 435

17 exm1342259 58503639 C17orf64 Missense H91D G C 1.0000 0.0046 4.50E-05 2.5375 0.6097 435

3 exm351757 132427063 NPHP3 Missense N386S A G 1.0000 0.0034 5.10E-05 2.9507 0.7219 435

11 exm914753 60540907 MS4A15 Missense A150S C A 1.0000 0.0057 5.40E-05 2.2473 0.5490 435

2 exm233249 152514514 NEB Missense R2056G A G 1.0000 0.0034 6.60E-05 -2.9235 0.7590 435

2 exm233348 152529193 NEB Missense Y1330C A G 1.0000 0.0034 6.60E-05 -2.9235 0.7590 435

2 exm233387 152536306 NEB Missense K1062E A G 1.0000 0.0034 6.60E-05 -2.9235 0.7590 435

2 exm235908 160761136 LY75 Missense W20R A G 0.9977 0.0035 6.70E-05 -2.9053 0.7583 434

10 exm_rs1398024 23665438 OTUD1 Intergenic - C A 1.0000 0.2149 7.00E-05 0.4210 0.1046 435

16 exm1243235 57093436 NLRC5 Missense K1326N C A 1.0000 0.0276 8.20E-05 1.0505 0.2697 435

10 exm860041 121556976 INPP5F Missense M291T A G 1.0000 0.0023 8.40E-05 3.3426 0.8391 435

14 exm1100159 50558441 C14orf183 Missense S43G A G 1.0000 0.0207 9.10E-05 1.1935 0.3073 435

14 exm1100163 50559288 C14orf183 Missense M25T A G 1.0000 0.0207 9.10E-05 1.1935 0.3073 435

136

18 exm1383650 34850818 CELF4 Missense V337M G A 1.0000 0.0023 9.30E-05 3.3226 0.8412 435

16 exm1201537 1400112 C16orf42 Missense A217V G A 1.0000 0.0034 9.40E-05 2.8096 0.6824 435

7 exm604496 6370734 C7orf70 Missense A18T G A 1.0000 0.0046 9.90E-05 2.4177 0.6303 435

15 exm1153214 42364501 PLA2G4D Missense E469D C G 1.0000 0.0080 1.02E-04 1.8764 0.9723 435

2 exm2265155 61500322 USP34 Intronic - G A 1.0000 0.3724 1.10E-04 -0.3463 0.0895 435

22 exm1615979 44360335 SAMM50 Missense V46I G A 1.0000 0.0092 1.22E-04 1.6554 0.4320 435

6 exm_rs2296343 33626717 ITPR3 Intronic - A G 1.0000 0.3080 1.38E-04 0.3482 0.0912 435

21 exm1573764 43824106 UBASH3A Missense S18G G A 1.0000 0.3379 1.45E-04 -0.3443 0.0894 435

4 exm382312 2172456 POLN Missense S502G A G 1.0000 0.0322 1.62E-04 -0.8684 0.2058 435

3 exm358931 151046506 P2RY13 Missense A113D C A 1.0000 0.0092 1.72E-04 1.6662 0.4422 435

1 exm42103 33957152 ZSCAN20 Missense Y432D C A 1.0000 0.0034 1.74E-04 2.6306 0.6965 435

6 exm2262146 55825851 COL21A1 Intergenic - A G 1.0000 0.0747 1.82E-04 0.5927 0.1572 435

1 exm59946 53532535 PODN Missense A10P G C 1.0000 0.0379 2.09E-04 -0.8583 0.2297 435

1 exm66205 65305287 JAK1 Synonymous D947D G A 1.0000 0.0092 2.11E-04 -1.7721 0.4941 435

15 exm1171256 66615250 DIS3L Missense I435V A G 1.0000 0.0034 2.14E-04 2.7128 0.7321 435

15 exm1171677 66850220 LCTL Synonymous G254G A T 1.0000 0.0034 2.14E-04 2.7128 0.7321 435

14 exm_rs6575793 101032217 BEGAIN Intronic - A G 1.0000 0.4310 2.28E-04 -0.3190 0.1135 435

7 exm_rs7782376 116577124 ST7 Intergenic - G A 1.0000 0.1448 2.36E-04 0.4340 0.1227 435

137

Table S3A: Association results for the top 25 genes associated with TBV.

Gene p-value # SNPs ACOT12 1.47E-04 5 PIWIL3 2.63E-04 7 ITGAM 2.92E-04 5 TREM2 3.91E-04 5 KLHL29 3.94E-04 2 PODXL2 4.17E-04 3 TOP3B 4.49E-04 4 SPATA4 4.49E-04 4 C5orf25 4.63E-04 2 DUSP19 4.85E-04 2 E4F1 5.98E-04 2 OMD 7.31E-04 3 SELO 7.36E-04 8 RALGAPA2 8.10E-04 6 AFMID 9.16E-04 3 BAAT 9.57E-04 5 ZNF518B 9.60E-04 9 P2RY6 1.01E-03 2 NLRC4 1.05E-03 2 ZBTB49 1.13E-03 4 TCIRG1 1.17E-03 5 TMEM234 1.42E-03 2 BACH2 1.47E-03 5 DIO2 1.51E-03 2 RAB40A 1.62E-03 2

138

Table S3B: Association results for the top 25 genes associated with GMV.

Gene p-value # SNPs ASB11 1.13E-05 2 PRKG2 4.59E-05 2 AFAP1L2 1.49E-04 7 PCIF1 2.15E-04 2 ELF2 3.12E-04 3 TSNARE1 3.77E-04 9 PNMA3 5.05E-04 2 TG 6.60E-04 30 PXMP2 6.86E-04 2 ARSB 7.03E-04 4 BNC2 7.14E-04 6 CHPF2 7.85E-04 5 NT5C1B 8.19E-04 5 ETFA 1.14E-03 2 PEX11G 1.27E-03 2 EPDR1 1.49E-03 2 USP53 1.53E-03 4 HIGD1B 1.66E-03 2 TMEM159 1.71E-03 2 TTC5 1.71E-03 3 CRYGC 1.76E-03 2 AGPAT3 1.76E-03 3 TRIML2 1.90E-03 2 VGLL4 1.94E-03 2 ATP6V1G2 2.06E-03 2

139

Table S3C: Association results for the top 25 genes associated with WMV.

Gene p-value # SNPs N4BP2 7.75E-05 12 CRELD1 8.46E-05 4 USP35 1.40E-04 9 OR5T3 2.11E-04 2 WDR81 3.93E-04 7 UBR1 4.73E-04 9 GIT2 7.09E-04 3 FIBCD1 7.41E-04 2 GPHB5 8.89E-04 2 NHSL2 9.98E-04 3 TIMD4 1.00E-03 2 PIK3R6 1.19E-03 2 BCAR3 1.21E-03 5 ANKRD35 1.66E-03 14 WFS1 1.70E-03 16 C11orf41 1.72E-03 5 FAM174A 1.73E-03 2 TSSK1B 1.75E-03 3 MYPN 1.88E-03 10 C9orf64 1.97E-03 3 ABCC11 1.99E-03 11 NADKD1 2.00E-03 2 ATXN2 2.29E-03 3 DDX47 2.32E-03 2 FBXO32 2.32E-03 2

140

Table S3D: Association results for the top 25 genes associated with GMFA.

Gene p-value # SNPs SHROOM4 5.74E-05 2 STK32B 6.27E-05 3 KCNH2 1.22E-04 2 OTUD7B 1.90E-04 2 C22orf28 1.94E-04 2 FURIN 2.12E-04 5 CROCC 2.44E-04 11 MFSD9 2.84E-04 3 VAV2 2.86E-04 4 TDRD7 3.15E-04 7 PCDHGA9 3.16E-04 5 RGS13 3.16E-04 2 C20orf194 3.22E-04 7 IAH1 3.26E-04 2 MPND 3.78E-04 3 SLC7A4 4.00E-04 8 TTBK1 4.12E-04 6 PLA1A 4.18E-04 4 NBAS 4.43E-04 17 MRAP 5.72E-04 2 PPP1R18 5.83E-04 6 HFM1 5.86E-04 5 ICAM3 5.91E-04 4 DCHS2 6.49E-04 41 KCTD17 6.54E-04 2

141

Table S3E: Association results for the top 25 genes associated with WMFA.

Gene p-value # SNPs DDX47 6.06E-05 2 FBXO32 6.06E-05 2 HLA-DRA 6.40E-05 4 CCDC129 1.36E-04 7 ASTL 2.05E-04 6 SEC14L2 2.18E-04 3 PPP1R36 2.53E-04 2 DRP2 2.86E-04 2 KNTC1 4.13E-04 9 KIF19 5.38E-04 7 KCTD17 5.80E-04 2 SYT14 6.32E-04 2 OR6C65 6.90E-04 2 SULF2 6.93E-04 3 CYP27B1 7.32E-04 3 CASC5 8.73E-04 11 LENG1 8.95E-04 3 PRICKLE1 9.04E-04 2 NXPH3 9.21E-04 2 THSD7B 9.80E-04 12 WDR3 9.92E-04 3 LAG3 9.97E-04 4 CHRNA2 1.18E-03 3 PDZD8 1.31E-03 8 EXPH5 1.37E-03 23

142

Table S3F: Association results for the top 25 genes associated with GMMD.

Gene p-value # SNPs KNTC1 6.64E-06 9 HLA-DRA 6.88E-06 4 IVD 8.08E-06 6 DDX47 8.46E-06 2 FBXO32 8.46E-06 2 OBFC1 9.79E-06 4 LOC554223 1.18E-05 3 KIF19 1.24E-05 7 NDNL2 1.24E-05 4 APPL1 1.41E-05 4 MAF1 2.25E-05 2 NEK8 2.36E-05 3 C20orf72 4.60E-05 3 PPP1R36 4.63E-05 2 SLC4A3 4.67E-05 3 DOLK 5.06E-05 2 AKD1 5.38E-05 13 SYTL3 5.38E-05 7 AQR 5.66E-05 4 APCDD1L 1.06E-04 3 TRIM69 1.10E-04 5 ZNF764 1.39E-04 2 FGF1 1.43E-04 2 PLEKHG4B 1.62E-04 15 KRT13 2.11E-04 6

143

Table S3G: Association results for the top 25 genes associated with WMMD.

Gene p-value # SNPs MRGPRE 7.89E-06 3 C2orf42 3.42E-05 4 TOM1 3.77E-05 3 C6orf47 4.05E-05 3 E2F3 4.40E-05 2 ACSL4 5.35E-05 2 IVD 7.21E-05 6 LENG1 9.94E-05 3 IDUA 1.04E-04 7 MACC1 1.38E-04 6 ILDR1 1.49E-04 4 BRAT1 2.22E-04 9 RLF 2.36E-04 9 DRP2 2.75E-04 2 DNMT3L 3.12E-04 4 CEP41 3.22E-04 2 CRELD1 3.55E-04 4 PRSS36 3.90E-04 4 PRF1 4.03E-04 3 ATAD2 4.10E-04 2 GPBP1L1 4.63E-04 2 ZNF311 4.67E-04 4 OBFC1 4.87E-04 4 KRT36 5.40E-04 5 SDCCAG3 5.75E-04 3

144

Table S3H: Association results for the top 25 genes associated with WMLV.

Gene p-value # SNPs POLRMT 3.55E-04 4 INCENP 4.55E-04 3 DSC3 5.84E-04 9 C12orf12 5.85E-04 2 DHDH 6.52E-04 10 EPHX1 7.24E-04 4 VWA5B1 7.62E-04 16 SLC6A5 9.42E-04 6 PIEZO2 1.19E-03 13 FAM186B 1.22E-03 12 C1orf27 1.22E-03 4 FGD6 1.24E-03 7 GIMAP5 1.27E-03 2 FYN 1.55E-03 2 ANGPTL5 1.57E-03 3 GTPBP5 1.81E-03 7 RCL1 1.89E-03 3 HEMK1 1.92E-03 2 ARHGEF26 1.92E-03 4 RRH 1.93E-03 3 MTA1 1.95E-03 3 CDC25A 2.03E-03 2 ZNF16 2.06E-03 2 GPR15 2.13E-03 2 A1BG 2.15E-03 3

145

Table S3I: Association results for the top 25 genes associated with GMCBF.

Gene p-value # SNPs TSPAN16 1.54E-05 3 CNOT6 2.98E-05 2 GNPAT 1.14E-04 3 C7orf70 1.26E-04 2 P2RY13 1.42E-04 3 C14orf183 1.43E-04 4 ILDR1 1.44E-04 4 SAMM50 1.51E-04 3 RAPGEF4 1.56E-04 5 PLA2G4D 2.72E-04 15 C1QTNF1 2.99E-04 2 MS4A15 3.09E-04 4 JAK1 3.15E-04 4 C17orf64 3.25E-04 3 POLN 4.23E-04 10 CNTLN 4.26E-04 13 C4orf40 4.62E-04 2 NIF3L1 5.31E-04 3 ENTPD1 5.86E-04 2 ANKRD7 5.97E-04 5 MRPS26 6.63E-04 2 SLC2A10 6.78E-04 3 ASZ1 6.92E-04 4 RAD54L 7.19E-04 3 C10orf28 7.27E-04 9

146

Figures S1A-S1I Manhattan plots for GWAS associations with (A) TBV, (B) GMV, (C) WMV, (D) GMFA, (E) WMFA, (F) GMMD, (G) WMMD, (H) WMLV, (I) GMCBF. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. WMLV, WMV, and GMV were additionally adjusted for ICV. Associations based on a single observation of the rare allele have been removed.

Figure S1A.

Figure S1B.

147

Figure S1C.

Figure S1D.

Figure S1E.

148

Figure S1F.

Figure S1G.

Figure S1H.

149

Figure S1I.

150

Figures S2A-S2I Manhattan plots for Exome associations with (A) TBV, (B) GMV, (C) WMV, (D) GMFA, (E) WMFA, (F) GMMD, (G) WMMD, (H) WMLV, (I) GMCBF. Association analyses were performed assuming an additive model of inheritance with adjustment for age, sex, and T2D affected status. WMLV, WMV, and GMV were additionally adjusted for ICV. Associations based on a single observation of the rare allele have been removed.

Figure S2A.

Figure S2B.

151

Figure S2C.

Figure S2D.

Figure S2E.

152

Figure S2F.

Figure S2G.

Figure S2H.

153

Figure S2I.

154

Chapter 5

Impact of HDL genetic risk scores on coronary artery calcified plaque and mortality in individuals with type 2 diabetes from the Diabetes Heart Study

Laura M. Raffield, Amanda J. Cox, Fang-Chi Hsu, Maggie C-Y Ng, Carl D Langefeld, J Jeffrey

Carr, Barry I Freedman, Donald W Bowden

This manuscript was published in the June 2013 issue of Cardiovascular Diabetology. The reference for this manuscript is as follows: Raffield, L.M., Cox, A.J., Hsu, F., Ng, M. C. Y., Langefeld, C. D., Carr, J.J., Freedman, B. I., Bowden, D. W. Impact of HDL genetic risk scores on coronary artery calcified plaque and mortality in individuals with type 2 diabetes from the Diabetes Heart Study. Cardiovascular diabetology, 2013. 12: p. 95. doi:10.1186/1475-2840-12- 95. Epub 25 June 2013. PubMed PMID: 23799899; PubMed Central PMCID: PMC3695806.

155

Abstract: Background: Patients with type 2 diabetes (T2D) are at elevated risk for cardiovascular disease (CVD) events and mortality. Recent studies have assessed the impact of genetic variants affecting high-density lipoprotein cholesterol (HDL) concentrations on CVD risk in the general population. This study examined the utility of HDL-associated single nucleotide polymorphisms (SNPs) for CVD risk prediction in European Americans with T2D enrolled in the Diabetes Heart Study (DHS).

Methods:

Genetic risk scores (GRS) of HDL-associated SNPs were constructed and evaluated for potential associations with mortality and with coronary artery calcified atherosclerotic plaque (CAC), a measure of subclinical CVD strongly associated with CVD events and mortality. Two sets of

SNPs were used to construct GRS; while all SNPs were selected primarily for their impacts on

HDL, one set of SNPs had pleiotropic effects on other lipid parameters, while the other set lacked effects on low-density lipoprotein cholesterol (LDL) or triglyceride concentrations.

Results:

The GRS were specifically associated with HDL concentrations (4.90 x 10-7 < p < 0.02) in models adjusted for age, sex, and body mass index (BMI), but were not associated with LDL or triglycerides. Cox proportional hazards regression analysis suggested the HDL-associated GRS had no impact on risk of CVD-mortality (0.48< p < 0.99) in models adjusted for other known

CVD risk factors. However, associations between several of the GRS and CAC were observed

(3.85 x 10-4 < p < 0.03) in models adjusted for other known CVD risk factors.

Conclusions:

The GRS analyzed in this study provide a tool for assessment of HDL-associated SNPs and their impact on CVD risk in T2D. The observed associations between several of the GRS and CAC suggest a potential role for HDL-associated SNPs on subclinical CVD risk in patients with T2D.

156

Keywords:

High-density lipoprotein cholesterol, type 2 diabetes, coronary artery calcified plaque, mortality, genetic risk score

157

Introduction:

Patients with T2D have significantly increased risk for CVD, with mortality rates from heart disease at least two-fold higher than in adults without diabetes (Writing Group,

Lloyd-Jones et al. 2010). In addition, patients with T2D have higher rates of dyslipidemia, an important CVD risk factor. In particular, individuals with T2D tend to have increased triglyceride concentrations and decreased HDL cholesterol concentrations (Haffner and

American Diabetes 2004). An inverse relationship between HDL concentrations and CVD- mortality has long been observed (Gordon, Probstfield et al. 1989); however, it is not clear whether HDL is a significant contributing factor in development of CVD and whether interventions to specifically alter HDL concentrations markedly impact CVD risk.

Voight et al. (Voight, Peloso et al. 2012) used Mendelian randomization to assess the impact of HDL concentrations on myocardial infarction (MI) risk in over 12,000 MI cases and over 40,000 controls. Neither a coding variant in LIPG nor a genetic risk score (GRS) derived from 14 common SNPs, both of which were robustly associated with HDL and not with other lipid parameters, was associated with MI. A one standard deviation increase in HDL cholesterol due to GRS was not associated with a significant change in MI risk, though epidemiological data suggested a change in HDL concentration of this magnitude would be associated with an approximate 38% reduction in risk for MI. These data do not support a major role for HDL-associated SNPs in MI.

Given that CVD accounts for more than 65% of all-cause mortality in individuals with T2D, determining the effects of HDL concentrations on CVD in this high risk group is of particular interest (Writing Group, Lloyd-Jones et al. 2010), as low HDL concentrations have been epidemiologically associated with CVD risk in T2D (Goldbourt, Yaari et al. 1997).

Previous large meta-analyses have not specifically analyzed the effect of HDL associated

158

GRS in clinically relevant, community-based cohorts of patients at high risk for CVD, such as patients with T2D. In the current study we constructed GRS containing SNPs associated solely with HDL, SNPs associated with HDL which have apparent pleiotropic effects on other lipid parameters, and both sets of SNPs for analysis in combined scores. We analyzed these

GRS for associations with mortality and a measure of subclinical CVD burden, coronary artery calcified atherosclerotic plaque (CAC) (Detrano, Guerci et al. 2008; Folsom, Kronmal et al. 2008; Agarwal, Cox et al. 2013), in a cohort of patients with T2D from the Diabetes

Heart Study (DHS).

Methods:

Study Design and Sample

DHS participants were recruited from outpatient internal medicine and endocrinology

clinics and from the community from 1998 through 2005 in western North Carolina. Siblings

concordant for T2D without advanced renal insufficiency were recruited, with additional non-

diabetic siblings enrolled whenever possible. Recruitment was based upon family structure, and

there were no inclusions/exclusions based on evidence of prevalent CVD at the time of

recruitment. Ascertainment and recruitment have been described in detail previously

(Wagenknecht, Bowden et al. 2001; Lange, Bowden et al. 2002; Bowden, Rudock et al. 2006;

Bowden, Cox et al. 2010). T2D was defined as diabetes developing after the age of 35 years

treated with insulin and/or oral agents, in the absence of historical evidence of ketoacidosis.

Diabetes diagnosis was confirmed by measurement of fasting glucose and glycated hemoglobin

(HbA1C) at the exam visit. Analyses completed for the current investigation included 983 self-

described European American individuals with T2D from 466 DHS families.

Study protocols were approved by the Institutional Review Board at Wake Forest School

of Medicine, and all participants provided written informed consent. Participant examinations

159

were conducted in the General Clinical Research Center of the Wake Forest Baptist Medical

Center. Examinations included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses, and spot urine collection. Individuals were considered hypertensive if they were prescribed anti-hypertensive medication or had blood pressure measurements exceeding 140 mmHg (systolic) or 90 mmHg (diastolic). Standard laboratory analyses included fasting glucose,

HbA1C, total cholesterol, HDL, and triglycerides. Low-density lipoprotein cholesterol (LDL) concentration was calculated using the Friedewald equation, and LDL concentrations were considered valid for subjects whose triglycerides were less than 796 mg/dL. CAC was measured using fast-gated helical computed tomography (CT) scanners, and calcium scores were calculated as previously described and reported as an Agatston score (Carr, Crouse et al. 2000; Carr, Nelson et al. 2005).

Vital status was determined for all subjects from the National Social Security Death

Index maintained by the United States Social Security Administration. For participants confirmed as deceased, length of follow-up was determined from the date of initial study visit to date of death. For all other participants, the length of follow-up was determined from the date of the initial study visit to the end of 2011. For deceased participants, copies of death certificates were obtained from relevant county Vital Records Offices to determine cause of death. Cause of death was categorized based on information contained in death certificates as CVD-mortality (MI, congestive heart failure, cardiac arrhythmia, sudden cardiac death, peripheral vascular disease, and stroke) or either cancer, infection, end-stage renal disease, accidental, or other (including obstructive pulmonary disease, pulmonary fibrosis, liver failure and Alzheimer’s dementia).

Association with mortality was assessed for both CVD-mortality and all-cause mortality, i.e. death from any cause.

160

Genotyping

Total genomic DNA was purified from whole blood samples using the PUREGENE

DNA isolation kit (Gentra, Inc., Minneapolis, MN). DNA concentration was quantified using standardized fluorometric readings on a Hoefer DyNA Quant 200 fluorometer (Hoefer Pharmacia

Biotech Inc., San Francisco, CA). Genotype data for specific SNPs was derived from: (i) the

MassARRAY SNP Genotyping System (Sequenom Inc., San Diego, CA) (n= 4 SNPs), (ii) a genome wide association study (GWAS) using the Affymetrix® Genome-Wide Human SNP

Array 5.0 (Affymetrix® Inc., Santa Clara, CA) (n=2 SNPs), (iii) Illumina® HumanExome

BeadChips (Illumina® Inc., San Diego, CA) (n=18 SNPs), and (iv) GWAS imputed data (n=4

SNPs).

Genotyping using the MassARRAY SNP Genotyping System was completed as described previously (Buetow, Edmonson et al. 2001). Primers for PCR amplification and extension reactions were designed using the MassARRAY Assay Design Software (Sequenom).

Samples were diluted to a final concentration of 5 ng/l, and single-base extension reaction products were separated and scored using a matrix-assisted laser desorption ionization/time of flight mass spectrometer. To evaluate genotyping accuracy, 39 quality control samples were included as blind duplicates. The concordance rate for these blind duplicates was 100%.

For the DHS GWAS data, genotype calling was completed using the BRLLM-P algorithm in Genotyping Console v4.0 (Affymetrix). Samples failing to meet an intensity quality control threshold (n=4) were not included for genotype calling and those failing to meet a minimum acceptable call rate of 95% (n=3) were excluded from further analyses. An additional

39 samples were included as blind duplicates within the genotyping set to serve as quality controls; the concordance rate for these blind duplicates was 99.0  0.72% (mean  standard deviation (SD)).

161

For the DHS Exome Chip data, genotype calling was completed using Genome Studio

Software v1.9.4 (Illumina). Samples failing to meet a minimum acceptable call rate of 98% (n=3) were excluded from further analyses. An additional 58 samples were included as blind duplicates within the genotyping set to serve as quality controls; the concordance rate for blind duplicates was 99.9  0.0001% (mean  SD). Additional quality control of GWAS and Exome Chip data sets was completed to exclude samples with poor quality genotype calls, gender errors, or unclear/unexpected sibling relationships.

For SNPs where direct genotyping data was not available, genotype data was obtained from GWAS imputed data. Imputation of 1,000 Genomes Project SNPs was completed using the program IMPUTE2 and the Phase I v2, cosmopolitan (integrated) reference panel, build 37

(Howie, Donnelly et al. 2009). SNPs that were used for imputation were required to have low missingness and show no significant departure from Hardy-Weinberg expectations (p>1 x 10-4).

To maximize the quality of imputation, the samples were not pre-phased. Only imputed SNPs with a confidence score >0.90 and information score >0.50 were used. A total of ~4.5 million

SNPs passed imputation quality control.

For all SNPs used to derive the GRS, the minimum acceptable call rate was 95%; the average SNP call rate was 99.4%  1.2% (mean  SD), and the average sample call rate was

99.4%  1.4%. Allele and genotype frequencies were calculated from unrelated individuals and tested for departures from Hardy-Weinberg equilibrium. No SNPs showed significant departure from Hardy-Weinberg equilibrium (p>0.05). One SNP (rs386000) included by Voight et al.

(Voight, Peloso et al. 2012) in their HDL GRS failed genotyping and was not included in the current analysis.

GRS Calculation

Both unweighted GRS and GRS weighted by SNP effect size were derived for two sets of

SNPs previously reported to be associated with HDL (Voight, Peloso et al. 2012). One set of 14

162

SNPs had documented effects on HDL concentrations only and were used by Voight et al.

(Voight, Peloso et al. 2012) for construction of a GRS. We created GRS from 13 of these SNPS with good quality genotyping data (rs386000 excluded) (Score 1; 1a=unweighted, 1b=weighted).

In addition, Voight et al. (Voight, Peloso et al. 2012) also reported an additional set of 15

SNPs, including a coding variant in LIPG, as associated with HDL concentrations, with some of these SNPs also reported to have pleiotropic effects on LDL cholesterol and triglyceride concentrations. All SNPs were primarily selected for their impact on HDL. Weighted and unweighted GRS were derived from this additional set of 15 SNPs (Score 2; 2a=unweighted,

2b=weighted). SNPs (n=26) from both sets were also combined to derive unweighted and weighted combined GRS; two pairs of SNPs (rs2338104 and rs7134594; rs2271293 and rs16942887) are in strong linkage disequilibrium (r2>0.90), and as such two SNPs (rs7134594 and rs2271293) were excluded from the combined scores. SNPs in the combined GRS were weighted by their effect sizes in mmol/L. The effect sizes used were drawn from the Voight et al. paper or the Global Lipid Genetic Consortium GWAS for Lipids paper that Voight et al. cited

(Teslovich, Musunuru et al. 2010; Voight, Peloso et al. 2012). All derived GRS (1a, 1b, 2a, 2b,

Combined Unweighted, Combined Weighted) were tested for association with HDL, LDL and triglycerides to evaluate whether the GRS were a measure of genetic contributions to either HDL only or to global lipid levels.

For all GRS, the effect allele was assigned as the allele associated with an increase in plasma HDL concentrations, i.e., an increase in GRS can be interpreted as an increase in genetic predisposition for elevated plasma HDL, as seen in the GRS used by Voight et al. (Voight, Peloso et al. 2012). Unweighted scores were derived by adding the number of effect alleles across each

SNP. The SNPs were also weighted by their previously reported effect sizes (Voight, Peloso et al.

2012). For the weighted scores, the number of effect alleles possessed by an individual at a particular SNP locus was multiplied by a weight derived from that SNP’s effect size contribution to the total effect size for all SNPs included in the GRS. For individuals missing genotype data

163

for a particular SNP, the mean genotype calculated in the DHS for that given SNP was assigned

(Fontaine-Bisson, Renström et al. 2010).

Statistical Analysis

For statistical analyses, continuous variables were transformed as necessary to approximate normality. Single SNP association analyses were performed using variance components methods implemented in Sequential Oligogenic Linkage Analysis Routines

(SOLAR) version 6.4.1 (Texas Biomedical Research Institute, San Antonio, TX) to account for relatedness between subjects (Almasy and Blangero 1998). Association was examined assuming an additive model of inheritance. Age and sex were included as covariates in single SNP association analyses for HDL.

GRS were considered as both ordinal (three tertiles: T1, T2, T3 derived from increasing tertile ranges) and continuous variables. Relationships between the GRS and HDL, LDL, triglycerides, CAC, and prior history of CVD and MI were examined using marginal models with generalized estimating equations. The models account for familial correlation using a sandwich estimator of the variance under exchangeable correlation. Relationships between GRS and both all-cause and CVD-mortality were examined using Cox proportional hazards models with sandwich-based variance estimation due to the inclusion of related individuals in this study.

Associations were adjusted for covariates including age, sex, BMI, smoking status (history of current or prior smoking), hypertension, cholesterol medication use, prior CVD, oral T2D medication use, and insulin use as indicated. All analyses were performed in SAS 9.3 (SAS

Institute, Cary, NC). Statistical significance was accepted at p<0.05.

Results:

The goal of this study was to assess the impact of HDL-associated SNPs on CVD risk in subjects from the DHS of families enriched for T2D. We assessed the relationships between GRS based on HDL-associated SNPs and both subclinical CVD and mortality risk. The clinical

164

characteristics of the 983 European American individuals with T2D included in the study are

summarized in Table 1. Subjects exhibited a variety of known CVD risk factors, including

hypertension, high body mass, and dyslipidemia. Subclinical CVD was prevalent, demonstrated

by the large number of subjects who had detectable CAC.

A total of 28 SNPs were genotyped (Table 2), including 13 SNPs solely associated with HDL and 15 SNPs associated with HDL that may also have pleiotropic effects on other lipid parameters (Voight, Peloso et al. 2012). We first performed single SNP association analyses for these 28 SNPs included in our GRS. Four SNPs (rs3764261, rs1800588, rs3890182, rs4759375) were associated (7.00 x 10-5< p <0.02) with HDL concentrations, with the estimated effects in the same direction and of similar magnitude as those previously reported by Voight et al. (Voight, Peloso et al. 2012). The coding variant in LIPG

(rs61755018) reported by Voight et al. was not significantly associated with HDL concentrations in our cohort (Voight, Peloso et al. 2012).

Next, weighted and unweighted GRS were analyzed. One GRS was derived from 13

SNPs (rs386000 excluded) reported to have effects solely on HDL (Score 1; 1a=unweighted,

1b=weighted; Table 2); scores ranged from 8.0 to 21.0 (14.7  2.2, mean  SD) for GRS 1a and from 7.8 to 21.0 (13.4  2.1) for GRS 1b. The second GRS was derived from an additional set of 15 SNPs associated with HDL concentrations, some of which also have pleiotropic effects on LDL cholesterol and triglycerides (Score 2; 2a=unweighted,

2b=weighted; Table 2); scores ranged from 7.0 to 20.0 (13.3  2.0) for GRS 2a and from 5.4 to 17.8 (11.0  2.1) for GRS 2b. Additional GRS were constructed from a combined set of 26

SNPs, with two SNPs not included in the combined scores due to strong linkage disequilibrium with other included SNPs (Combined Unweighted and Combined Weighted

165

scores; Table 2); scores ranged from 17.0 to 36.0 (26.8  2.9) for the Combined Unweighted

GRS and from 12.2 to 31.8 (21.2  3.2) for the Combined Weighted GRS.

An initial analysis of tertiles of GRS 1 and 2 (i.e. upper, middle, and lower third of GRS

distribution) as an ordinal variable was performed to test for a linear trend of association with

plasma HDL concentrations. As GRS 1 and 2 were linearly associated with HDL concentrations

(6.18 x 10-6< p <0.007 in models adjusted for additional CVD risk factors), we next examined the

GRS as continuous variables, including GRS 1 and 2 and the Combined GRS. When analyses

were adjusted for age, sex, and BMI, all weighted and unweighted GRS were associated with

increased HDL (Table 3) but were not associated with LDL cholesterol or triglyceride

concentrations (p>0.05). In addition, significant associations were observed between several of

the GRS and CAC (Table 3). This suggests a potential impact of a genetic predisposition to

increased HDL on reduced risk for subclinical CVD in patients with T2D. Further analyses

indicated that the GRS were not significantly associated with CVD-mortality (Table 4), nor were

any significant associations observed between the GRS and history of prior CVD or history of

prior MI (0.06< p <0.87). The GRS were also not associated with all-cause mortality, with the

exception of the nominal associations of GRS 2b, the weighted GRS which includes some SNPs

with pleiotropic effects on lipid parameters other than HDL, and the Combined Weighted GRS

(Table 4). Analyses were also performed with further adjustment for diabetic medication use;

results were essentially unchanged (Additional files 1 and 2).

In an effort to further quantify these findings, tertiles of GRS 1 and 2 were also

analyzed in reference to the lowest GRS tertile. The highest tertile was consistently associated

with increased HDL concentrations (Additional file 3), with GRS 2b displaying the strongest

association (p=4.31 x 10-5 for comparison of the highest in reference to the lowest GRS

tertile). Further quantification of risk for all-cause mortality suggests individuals in the

166

highest tertile for GRS 2b experience an approximate 40% reduction in risk of mortality

compared to those in the lowest tertile (HR: 0.57; 95% CI: 0.39- 0.83; p=0.003).

Discussion:

The present study evaluated HDL-associated SNPs in patients with T2D using weighted and unweighted GRS. It is important to assess the impact of HDL-associated SNPs in patients with T2D, as decreased HDL concentrations are epidemiologically associated with higher CVD-mortality and are often observed in patients with T2D (Gordon, Probstfield et al.

1989; Haffner and American Diabetes 2004). In addition to conventional lipid parameters, measures of prevalent disease, and mortality, we evaluated associations of these GRS with

CAC. CAC has been shown in the DHS and other cohorts to be a strong independent predictor of CVD events and mortality (Detrano, Guerci et al. 2008; Folsom, Kronmal et al. 2008).

Compared to individuals with low or nonexistent CAC (CAC 0–9), individuals in the DHS with very high CAC (≥ 1,000) experienced >6-fold increased risk for all-cause mortality and

>11-fold increased risk for CVD-mortality (Agarwal, Morgan et al. 2011; Agarwal, Cox et al.

2013). CAC has not previously been examined in studies of the impact of HDL-associated variants on CVD risk. Hence, analysis of our HDL GRS with this subclinical measure of CVD was unique and of clinical relevance.

For the weighted and unweighted GRS assessed, higher GRS values were clearly associated with increased plasma HDL. The GRS were not associated with LDL cholesterol and triglyceride concentrations, allowing assessment of the effects of genetic predisposition to higher or lower HDL without confounding effects from associations with other lipid parameters. No evidence of association between the GRS and CVD-mortality was observed; however, associations between the GRS and CAC were observed. These associations point to

167

a potential reduction in risk for subclinical CVD in patients with T2D with a genetic predisposition to increased HDL.

Prior Analyses of HDL-associated SNPs

Recently, Voight et al. (Voight, Peloso et al. 2012) evaluated two sets of SNPs associated with higher HDL to determine whether HDL-associated SNPs would confer protection from MI in the general population. They concluded that a genetic predisposition to higher HDL did not detectably influence MI risk. However, Voight et al. did not address the association of their GRS in high CVD risk groups such as patients with T2D (Voight, Peloso et al. 2012). Moreover, it is important to test the applicability of results from extremely large meta-analyses in community-based cohorts. This is a necessary step in assessing the value of these GRS for introduction into clinical medicine. We hypothesized that GRS of HDL- associated SNPs could provide a tool for evaluating the effect of HDL concentrations on CVD risk in patients with T2D.

Prior studies in the general population have generally shown little impact of genetic

variants affecting HDL concentrations on CVD risk. In two large Danish studies, variants

associated with a decrease in HDL cholesterol did not increase CVD risk (Frikke-Schmidt,

Nordestgaard et al. 2008; Haase, Tybjærg-Hansen et al. 2012). In another study with 8,473

Caucasian participants, GRS were strongly associated with HDL concentrations, but were not

associated with CVD (Shah, Casas et al. 2012). However, few previous studies have focused on

patients with T2D, so it is unclear whether HDL-associated SNPs impact CVD risk in this group,

a gap that we addressed in this study.

Analysis of GRS for Associations with Lipid Parameters

We did not detect many of the single SNP associations with HDL concentrations reported

by Voight et al. (Voight, Peloso et al. 2012). This is not surprising, as the current study has lower

168

power to detect associations given the modest effect sizes previously reported. However, we observed significant associations between all of the GRS assessed and HDL, allowing us to consider the impact of genetic determinants of HDL concentrations on CVD risk. Some past studies of SNPs associated with HDL in patients with T2D have examined SNPs with pleiotropic effects on other lipid parameters or other traits, such as obesity, which could also influence CVD risk (Porchay-Baldérelli, Péan et al. 2007; Doney, Dannfald et al. 2009; Sharma, Prudente et al.

2011; Chaudhary, Likidlilid et al. 2012). An important strength of this study is that the GRS were not associated with LDL cholesterol and triglyceride concentrations, indicating the GRS are mainly informative for an individual’s HDL concentration, as opposed to more global lipid concentrations.

In our study, consistent with the GRS used by Voight et al., a higher GRS indicates a predisposition to higher HDL (Voight, Peloso et al. 2012). While we refer to all GRS as risk scores by convention, epidemiological data would in fact indicate that higher GRS values should be associated with increased protection from CVD, not increased risk, as higher plasma HDL is protective (Gordon, Probstfield et al. 1989). However, constructing the GRS so that higher GRS values would be associated with decreased HDL would not have changed our results; p-values would remain unchanged and effect estimates would be of the same magnitude in the opposite direction.

Analysis of GRS for Associations with Mortality and Subclinical CVD

Association of the GRS with HDL allowed us to subsequently consider the impact of genetic predisposition to higher or lower HDL on all-cause mortality and CVD-mortality in T2D.

While the HDL GRS were not associated with CVD-mortality, association with all-cause mortality was observed for GRS 2b, the weighted GRS which includes SNPs with pleiotropic effects on LDL cholesterol and triglycerides, and the Combined Weighted GRS. It is possible that these associations are due to the reported pleiotropic effects on LDL and triglycerides of some of the included SNPs, although the GRS overall were not associated with LDL and triglyceride

169

concentrations. Alternatively this observation may be due to impacts of HDL on risk for mortality through pathways other than CVD. A recent study by Qi et al. (Qi, Liang et al. 2012) examined

HDL, LDL, and triglyceride GRS and demonstrated a modest increase in T2D risk with increased

HDL GRS. While the current study consists entirely of T2D affected individuals, the study by Qi et al. (2012) suggests a role for HDL in diabetes pathogenesis, perhaps impacting mortality. Apart from reverse cholesterol transport, the cardinal function of HDL in ameliorating CVD risk, HDL particles also demonstrate anti-inflammatory, antioxidant, and anti-apoptotic effects which may be important mechanisms contributing to risk for mortality (Mahdy Ali, Wonnerth et al. 2012).

Further, while the GRS assessed in this study are associated with HDL concentrations, their impact on HDL particle composition and function is unclear and may further contribute to mechanisms underpinning the observed associations with all-cause mortality.

While we did not observed significant associations between the HDL GRS and CVD- mortality, we did observe interesting associations between the GRS and CAC, a measure of subclinical CVD risk. We observed an association between GRS 1a, the unweighted GRS which contains SNPs which affect HDL concentrations only, and CAC. Likewise associations with CAC were observed for GRS 2a and 2b (SNPs with pleiotropic effects on lipid parameters other than

HDL), as well as the Combined Unweighted and Weighted GRS. This points to a potential role of

HDL-associated SNPs in subclinical CVD risk; it is possible we are not observing the impact of

HDL-associated SNPs on CVD-mortality due to a low number of events, with association with

CAC observed due to increased power for analyses of this continuous trait. This may point to a differing impact of HDL-associated SNPs in individuals with T2D than is seen in most studies of the general population. However, little work on associations between CAC and HDL-associated

SNPs has been done in the general population, so these SNPs may be associated with CAC in the general population as well.

170

Limitations

SNPs included in our genetic risk scores were selected for the strength of their association with HDL concentrations in previous studies (Teslovich, Musunuru et al. 2010;

Voight, Peloso et al. 2012); other SNPs more modestly associated with HDL were not the focus of the current study but may, in some cases, have a greater impact on disease risk (Porchay-

Baldérelli, Péan et al. 2007; Porchay-Baldérelli, Péan et al. 2009). The impact of genetic variants associated with HDL may also be modulated by environmental factors, which was not assessed here(Lamina, Forer et al. 2012; Yin, Wu et al. 2012). The disparate treatments prescribed to individuals in the DHS, including oral T2D medications and insulin, may also impact CVD outcomes; however, inclusion of these covariates in our models did not substantially change our results (Additional files 1 and 2). For analyses of CVD mortality, our power to detect a modest effect (HR~1.1) associated with our GRS was limited (<60% power); however, our power to detect a larger effect (HR~1.2) was high (>95% power). As a result, we cannot exclude a modest impact of our HDL GRS on CVD mortality that was not detected in this study. In contrast, limited power was not a concern for analyses of all-cause mortality (>85% power to detect a HR of ~1.1) due to the increased number of events, nor for analyses of continuous measures such as

CAC and HDL (>85% power for a modest effect size of 0.02).

Conclusions:

In summary, the GRS analyzed in this paper provide a useful tool for assessing the impact of genetic predisposition to higher or lower HDL concentrations on risk for CVD in patients with T2D. As has been observed in studies in the general population, there was no association between the HDL GRS and risk of CVD-mortality. However, associations between some of the GRS and CAC were observed, pointing to a potential role of genetic variants affecting HDL concentrations in risk for subclinical CVD in patients with T2D. Further study of

171

HDL-associated genetic variants will be needed to further clarify whether these variants are important determinants of CVD risk in T2D affected individuals.

Abbreviations:

BMI, Body mass index; CAC, Coronary artery calcified plaque; CT, Computed tomography; CVD, Cardiovascular disease; DHS, Diabetes heart study; HDL, High-density lipoprotein cholesterol; GRS, Genetic risk score; GWAS, Genome wide association study; HbA1C, Glycated hemoglobin; LDL, Low-density lipoprotein cholesterol; MI, Myocardial infarction; SD, Standard deviation; SNP, Single nucleotide polymorphism; SOLAR, Sequential Oligogenic Linkage Analysis Routines; T2D, Type 2 diabetes

Competing interests:

The authors declare no conflicts of interest.

Authors’ contributions:

LMR performed SNP genotyping, perfomed statistical analysis, and prepared the manuscript;

AJC assisted with SNP genotyping, contributed to the statistical analysis, and assisted with the manuscript preparation; FCH contributed to the statistical analysis and reviewed and edited the manuscript; MCYN contributed to the management of the genetic data and reviewed the manuscript; CDL performed the SNP imputation and reviewed the manuscript; JJC was involved in the initial design of the Diabetes Heart Study, contributed to patient ascertainment and clinical evaluation, and reviewed the manuscript; BIF was involved in the initial design of the Diabetes

Heart Study, contributed to patient ascertainment and clinical evaluation, and reviewed and edited the manuscript; DWB leads the Diabetes Heart Study and assisted with the manuscript preparation. All authors read and approved the final manuscript.

Acknowledgements:

This study was supported by R01 HL092301 to D.W.B. The authors thank the Wake Forest

School of Medicine investigators and staff and the participants of the DHS study for their valuable contributions.

172

Table 1: Demographic and clinical characteristics of 983 individuals with type 2 diabetes from the Diabetes Heart Study.

Mean  SD or % Median (range)

Demographic Information

Age (years) 62.5 ± 9.1 63 (34.2-86)

Gender (% female) 51.8%

T2D Duration (years) 10.5 ± 7.2 8 (0-46)

Smoking (current or past) (%) 59.4%

Self-reported history of prior CVD (%) 43.3%

Deceased (%) 23.3%

Deceased (from CVD) (%) 10.3%

Body Composition

Height (cm) 168.7 ± 9.7 168.5 (122.8-202.0)

Weight (kg) 92.2 ± 20.3 89.5 (40.8-209.2)

BMI (kg/m2) 32.3 ± 6.6 31.2 (17.1-58.0)

Medications

Cholesterol Medications (%) 47.8%

Oral T2D Medications (%) 78.8%

Insulin (%) 27.6%

Blood Pressure

Systolic BP (mmHg) 140 ± 18.9 138.5 (94-260)

Diastolic BP (mmHg) 72.7 ± 10.3 72 (36.5-106)

Hypertension (%) 88.9%

Blood Biochemistry

173

Glucose (mg/dL) 148.1 ± 56.2 135 (16-463)

Hemoglobin A1C (%) 7.6 ± 1.7 7.2 (4.3-18.3)

Total Cholesterol (mg/ dL) 184.9 ± 43.7 180.5 (65-427)

HDL cholesterol (mg/dL) 42.2 ± 12 41 (8-98)

LDL cholesterol (mg/dL) 102.8 ± 32.6 100 (12-236)

Triglycerides (mg/dL) 208.8 ± 140.2 175 (30-1310)

Vascular Imaging

Coronary Artery Calcified Plaque (CAC) 1888 ± 3382 478 (0- 50415)

174

Table 2: HDL-associated SNPs included in the genetic risk scores (GRS).

SNPs in Risk Scores 1a and 1b Effect Allele Alleles Effect Frequency Effect (β value) SNP Chr Position Gene(s) Source (Major/Minor) Allele DHS in mmol/L rs1689800 1 182168885 GLUL, ZNF648 Exome Chip A/G A 0.628 0.01 rs13107325 4 103188709 SLC39A8 Exome Chip C/T C 0.925 0.02 rs2293889 8 116599199 TRPS1 Sequenom G/T G 0.554 0.01 rs2923084 11 10388782 AMPD3 Exome Chip A/G A 0.815 0.01 rs7134594 12 110000193 MMAB Exome Chip T/C T 0.520 0.01 rs4759375 12 123796238 SBNO1 Imputed data C/T T 0.092 0.02 rs838880 12 125261593 SCARB1 Exome Chip T/C C 0.337 0.02 rs16942887 16 67928042 PSKH1 Exome Chip G/A A 0.115 0.03 rs881844 17 37810218 STARD3 Imputed data G/C G 0.652 0.01 rs4082919 17 76377482 PGS1 Sequenom G/T T 0.483 0.01 rs7255436 19 8433196 ANGPTL4 Exome Chip A/C A 0.506 0.01 rs737337 19 11347493 DOCK6 Exome Chip T/C T 0.900 0.02 rs181362 22 21932068 UBE2L3 Sequenom C/T C 0.813 0.01 SNPs in Risk Scores 2a and 2b rs4846914 1 230295691 GALNT2 Sequenom A/G A 0.606 0.02 rs17145738 7 72982874 TBL2, BCL7B Exome Chip C/T T 0.124 0.03 rs17482753 8 19832646 LPL Exome Chip G/T T 0.095 0.08 rs17321515 8 126486409 TRIB1 GWAS A/G G 0.465 0.02 rs471364 9 15289578 TTC39B Exome Chip T/C T 0.896 0.03 rs3890182 9 107647655 ABCA1 Exome Chip G/A G 0.890 0.03 rs174547 11 61570783 FADS1 Exome Chip T/C T 0.668 0.03 rs6589566 11 116652423 ZNF259, APOA5 GWAS A/G A 0.924 0.05 rs2338104 12 109895168 KCTD10 Exome Chip G/C G 0.520 0.03 rs1800588 15 58723675 LIPC Exome Chip C/T T 0.207 0.05

175

rs3764261 16 56993324 CETP Exome Chip C/A A 0.331 0.1 rs2271293 16 67902070 NUTF2 Exome Chip G/A A 0.113 0.03 rs61755018 18 47109955 LIPG Imputed data A/G G 0.007 0.14 rs2967605 19 8469738 RAB11B Exome Chip C/T C 0.812 0.05 rs16988929 20 42904315 GDAP1L1 Imputed data C/T T 0.002 0.01 SNPs in Risk Score 1a (unweighted) and Risk Score 1b (weighted) have effects on HDL levels only, while SNPs in Risk Score 2a

(unweighted) and Risk Score 2b (weighted) have effects on HDL levels as well as some pleiotropic effects on LDL and triglyceride levels.

Allele frequencies in the DHS are provided along with the effect size estimates (β value) reported by Voight et al.

176

Table 3: Associations between HDL genetic risk scores and HDL, LDL, triglycerides, and coronary artery calcified plaque (CAC).

Model 1 Model 2 Model 3 β Estimate (CI) p-value β Estimate (CI) p-value β Estimate (CI) p-value HDL Risk Score 1a 0.027 (-0.001- 0.055) 0.057 0.030 (0.005- 0.056) 0.022 0.033 (0.007- 0.058) 0.013

Risk Score 1b 0.033 (0.003- 0.063) 0.031 0.034 (0.006- 0.061) 0.018 0.035 (0.008- 0.063) 0.012

Risk Score 2a 0.046 (0.016- 0.076) 0.003 0.051 (0.023- 0.079) 3.85 x 10-4 0.052 (0.024- 0.080) 3.06 x 10-4

Risk Score 2b 0.061 (0.034- 0.089) 1.42 x 10-5 0.062 (0.036- 0.088) 2.60 x 10-6 0.063 (0.037- 0.088) 1.44 x 10-6

Combined 0.035 (0.015- 0.055) 0.001 0.039 (0.021- 0.058) 3.63 x 10-5 0.040 (0.021- 0.058) 2.34 x 10-5 Unweighted Combined 0.043 (0.025- 0.061) 4.65 x 10-6 0.044 (0.027- 0.061) 4.90 x 10-7 0.044 (0.027- 0.061) 2.75 x 10-7 Weighted

LDL Risk Score 1a 0.317 (-0.678- 1.311) 0.532 0.260 (-0.74- 1.260) 0.611 0.245 (-0.759- 1.248) 0.633

Risk Score 1b 0.243 (-0.782- 1.268) 0.642 0.177 (-0.854- 1.207) 0.737 0.140 (-0.887- 1.167) 0.789

Risk Score 2a -0.823 (-1.978- 0.332) 0.163 -0.838 (-1.987- 0.312) 0.153 -0.805 (-1.955- 0.346) 0.170

Risk Score 2b -0.507 (-1.614- 0.600) 0.369 -0.604 (-1.713- 0.506) 0.286 -0.585 (-1.698- 0.529) 0.303

177

Combined -0.243 (-1.018- 0.533) 0.539 -0.261 (-1.033- 0.512) 0.509 -0.266 (-1.038- 0.507) 0.501 Unweighted

Combined -0.260 (-0.988- 0.468) 0.484 -0.325 (-1.054- 0.404) 0.383 -0.327 (-1.060- 0.406) 0.381 Weighted

Triglycerides Risk Score 1a -0.002 (-0.019- 0.015) 0.853 -0.005 (-0.021- 0.012) 0.591 -0.006 (-0.022- 0.010) 0.469

Risk Score 1b -0.005 (-0.022- 0.013) 0.595 -0.007 (-0.024- 0.010) 0.402 -0.008 (-0.024- 0.009) 0.374

Risk Score 2a -0.013 (-0.034- 0.008) 0.220 -0.016 (-0.036- 0.004) 0.123 -0.015 (-0.036- 0.005) 0.133

Risk Score 2b -0.009 (-0.028- 0.010) 0.370 -0.013 (-0.031- 0.005) 0.162 -0.013 (-0.031- 0.005) 0.163

Combined 0.072 -0.010 (-0.023- 0.004) 0.157 -0.012 (-0.025- 0.001) 0.076 -0.012 (-0.025- 0.001) Unweighted

Combined -0.007 (-0.020- 0.005) 0.246 -0.010 (-0.022- 0.002) 0.092 -0.010 (-0.022- 0.002) 0.096 Weighted Coronary Artery

Calcified Plaque (CAC)

Risk Score 1a -0.079 (-0.154- -0.003) 0.042 -0.065 (-0.128- -0.001) 0.046 -0.068 (-0.128- -0.008) 0.027

178

Risk Score 1b -0.067 (-0.146- 0.012) 0.097 -0.053 (-0.119- 0.014) 0.119 -0.059 (-0.123- 0.004) 0.065

Risk Score 2a -0.072 (-0.154- 0.011) 0.089 -0.063 (-0.134- 0.009) 0.085 -0.082 (-0.149- -0.014) 0.017

Risk Score 2b -0.117 (-0.197- -0.036) 0.005 -0.088 (-0.156- -0.020) 0.011 -0.101 (-0.163- -0.038) 0.002

Combined 0.004 0.001 3.18 x 10-4 -0.083 (-0.140- -0.026) -0.077 (-0.123- -0.031) -0.080 (-0.124- -0.037) Unweighted

Combined 0.002 -0.087 (-0.141- -0.033) -0.069 (-0.114- -0.024) 0.002 -0.075 (-0.117- -0.034) 3.85 x 10-4 Weighted

Analysis was performed using marginal models with generalized estimating equations. Model 1 is unadjusted; Model 2 is adjusted for age, sex, and body mass index (BMI); Model 3 is adjusted for age, sex, BMI, smoking, hypertension, and prior cardiovascular disease.

Associations are reported as the β estimate and its 95% confidence interval (CI).

179

Table 4: Association between HDL genetic risk scores analyzed as a continuous variable and all-cause and CVD-mortality.

Model 1 Model 2 Model 3

All-cause Mortality HR (CI) p-value HR (CI) p-value HR (CI) p-value

Risk Score 1a 0.96 (0.91, 1.02) 0.208 0.98 (0.92, 1.03) 0.373 0.97 (0.92, 1.03) 0.295

Risk Score 1b 0.96 (0.91, 1.02) 0.175 0.97 (0.92, 1.03) 0.314 0.97 (0.92, 1.02) 0.228

Risk Score 2a 0.96 (0.90, 1.03) 0.235 0.97 (0.91, 1.03) 0.302 0.96 (0.90, 1.03) 0.232

Risk Score 2b 0.92 (0.87, 0.98) 0.010 0.93 (0.88, 0.99) 0.020 0.93 (0.88, 0.98) 0.011

Combined Unweighted 0.96 (0.92- 1.01) 0.115 0.97 (0.93- 1.01) 0.170 0.97 (0.93- 1.01) 0.162

Combined Weighted 0.95 (0.91- 0.99) 0.009 0.96 (0.92- 0.99) 0.017 0.95 (0.92- 0.99) 0.011

CVD Mortality

Risk Score 1a 1.00 (0.92, 1.09) 0.958 1.01 (0.93, 1.10) 0.771 1.00 (0.92, 1.09) 0.988

Risk Score 1b 1.00 (0.92, 1.09) 0.936 1.01 (0.93, 1.10) 0.880 0.99 (0.91, 1.08) 0.835

Risk Score 2a 1.03 (0.93, 1.13) 0.621 1.03 (0.94, 1.13) 0.542 1.02 (0.94, 1.12) 0.622

Risk Score 2b 0.96 (0.89, 1.05) 0.407 0.98 (0.90, 1.06) 0.547 0.97 (0.90, 1.05) 0.477

180

Combined Unweighted 1.00 (0.94- 1.07) 0.955 1.01 (0.95- 1.07) 0.826 1.00 (0.94- 1.07) 0.890

Combined Weighted 0.98 (0.92- 1.04) 0.429 0.98 (0.93- 1.04) 0.573 0.98 (0.93- 1.04) 0.502

Analysis was performed using Cox proportional hazards regression. Model 1 is unadjusted; Model 2 is adjusted for age and sex; Model 3 is adjusted for age, sex, body mass index, smoking, hypertension, cholesterol medication use, and prior cardiovascular disease. The hazard ratio (HR) and its 95% confidence interval (CI) are reported.

181

Additional file 1: Associations between HDL genetic risk scores and HDL, LDL, triglycerides, and coronary artery calcified plaque (CAC), with and without adjustment for diabetic medication use.

Model 1 Model 2

β Estimate (CI) p-value β Estimate (CI) p-value

HDL Risk Score 1a 0.033 (0.007- 0.058) 0.013 0.034 (0.008- 0.059) 0.009 Risk Score 1b 0.035 (0.008- 0.063) 0.012 0.036 (0.009- 0.063) 0.010 Risk Score 2a 0.052 (0.024- 0.080) 3.06 x 10-4 0.051 (0.023- 0.079) 4.00 x 10-4 Risk Score 2b 0.063 (0.037- 0.088) 1.44 x 10-6 0.062 (0.036- 0.088) 2.14 x 10-6 Combined Unweighted 0.040 (0.021- 0.058) 2.34 x 10-5 0.040 (0.022- 0.059) 1.49 x 10-5 Combined Weighted 0.044 (0.027- 0.061) 2.75 x 10-7 0.044 (0.027- 0.061) 3.40 x 10-7 LDL Risk Score 1a 0.245 (-0.759- 1.248) 0.633 0.301 (-0.695- 1.297) 0.554 Risk Score 1b 0.140 (-0.887- 1.167) 0.789 0.166 (-0.850- 1.181) 0.749 Risk Score 2a -0.805 (-1.955- 0.346) 0.170 -0.841 (-1.977- 0.296) 0.147 Risk Score 2b -0.585 (-1.698- 0.529) 0.303 -0.612 (-1.719- 0.495) 0.279

Combined Unweighted -0.266 (-1.038- 0.507) 0.501 -0.238 (-1.007- 0.531) 0.545

Combined Weighted -0.327 (-1.060- 0.406) 0.381 -0.334 (-1.063- 0.395) 0.369

Triglycerides

Risk Score 1a -0.006 (-0.022- 0.010) 0.469 -0.007 (-0.023- 0.010) 0.432 Risk Score 1b -0.008 (-0.024- 0.009) 0.374 -0.008 (-0.024- 0.009) 0.357

182

Risk Score 2a -0.015 (-0.036- 0.005) 0.133 -0.016 (-0.036- 0.005) 0.132 Risk Score 2b -0.013 (-0.031- 0.005) 0.163 -0.013 (-0.031- 0.005) 0.158 Combined Unweighted -0.012 (-0.025- 0.001) 0.072 -0.012 (-0.025- 0.001) 0.063

Combined Weighted -0.010 (-0.022- 0.002) 0.096 -0.010 (-0.022- 0.002) 0.090

Coronary Artery Calcified Plaque Risk Score 1a -0.068 (-0.128- -0.008) 0.027 -0.066 (-0.127- -0.006) 0.032 Risk Score 1b -0.059 (-0.123- 0.004) 0.065 -0.058 (-0.122- 0.005) 0.071

Risk Score 2a -0.082 (-0.149- -0.014) 0.017 -0.080 (-0.147- -0.013) 0.019 Risk Score 2b -0.101 (-0.163- -0.038) 0.002 -0.099 (-0.162- -0.037) 0.002 Combined Unweighted -0.080 (-0.124- -0.037) 3.18 x 10-4 -0.079 (-0.123- -0.035) 4.32 x 10-4 Combined Weighted -0.075 (-0.117- -0.034) 3.85 x 10-4 -0.074 (-0.115- -0.032) 4.83 x 10-4

Analysis was performed using marginal models with generalized estimating equations. Model 1 is adjusted for age, sex, body mass index (BMI), smoking, hypertension, and prior cardiovascular disease; Model 2 is adjusted for age, sex, BMI, smoking, hypertension, prior cardiovascular disease, oral T2D medication use, and insulin use. Associations are reported as the β estimate and its 95% confidence interval (CI).

183

Additional file 2: Association between HDL genetic risk scores analyzed as a continuous variable and all-cause and CVD-mortality, with and without adjustment for diabetic medication use.

Model 1 Model 2

All-cause Mortality HR (CI) p-value HR (CI) p-value

Risk Score 1a 0.97 (0.92, 1.03) 0.295 0.97 (0.92, 1.03) 0.332

Risk Score 1b 0.97 (0.92, 1.02) 0.228 0.97 (0.92, 1.02) 0.259

Risk Score 2a 0.96 (0.90, 1.03) 0.232 0.96 (0.91, 1.03) 0.243

Risk Score 2b 0.93 (0.88, 0.98) 0.011 0.93 (0.88, 0.99) 0.014

Combined Unweighted 0.97 (0.93- 1.01) 0.162 0.97 (0.93- 1.01) 0.185

Combined Weighted 0.95 (0.92- 0.99) 0.011 0.96 (0.92- 0.99) 0.016

CVD Mortality

Risk Score 1a 1.00 (0.92, 1.09) 0.988 1.01 (0.92, 1.10) 0.913

Risk Score 1b 0.99 (0.91, 1.08) 0.835 1.00 (0.92, 1.08) 0.903

0.622 0.557 Risk Score 2a 1.02 (0.94, 1.12) 1.03 (0.94, 1.12) 0.477 0.519

184

Risk Score 2b 0.97 (0.90, 1.05) 0.97 (0.90, 1.06)

Combined Unweighted 1.00 (0.94- 1.07) 0.890 1.01 (0.95- 1.07) 0.801

Combined Weighted 0.98 (0.93- 1.04) 0.502 0.98 (0.93- 1.04) 0.560

Analysis was performed using Cox proportional hazards regression. Model 1 is adjusted for age, sex, body mass index (BMI), smoking, hypertension, cholesterol medication use, and prior cardiovascular disease; Model 2 is adjusted for age, sex, BMI, smoking, hypertension, cholesterol medication use, prior cardiovascular disease, oral T2D medication use, and insulin use. The hazard ratio (HR) and its 95% confidence interval (CI) are reported.

185

Additional file 3: Association between HDL genetic risk score tertiles and all-cause and CVD- mortality using unadjusted proportional hazards regression models and between risk scores and HDL levels using unadjusted marginal models incorporating generalized estimating equations.

Hazard ratios (HR) or β estimates (as appropriate) and 95% confidence intervals (CI) are reported relative to the lowest tertile.

HDL All-cause mortality CVD mortality Tertile Tertiles β Estimate (95% CI) p-value HR (95% CI) p-value HR (95% CI) p-value Ranges

T1 ≤ 13 0 1 1

Risk Score 1a T2 14-15 -0.012 (-0.153,0.129) 0.869 1.06 (0.77, 1.47) 0.719 1.21 (0.73, 2.01) 0.451

T3 ≥ 16 0.178 (0.027, 0.328) 0.021 0.98 (0.70, 1.38) 0.908 1.29 (0.76, 2.20) 0.350

T1 ≤ 12 0 1 1

Risk Score 1b T2 13-14 -0.034 (-0.167, 0.100) 0.622 1.12 (0.83, 1.52) 0.449 1.29 (0.83, 2.01) 0.261

T3 ≥ 15 0.171 (0.021, 0.321) 0.025 0.95 (0.68, 1.33) 0.761 1.20 (0.72, 2.00) 0.490

T1 ≤ 12 0 1 1 Risk Score 2a T2 13-14 0.083 (-0.047, 0.213) 0.210 0.81 (0.59, 1.10) 0.172 0.86 (0.55, 1.36) 0.529

T3 ≥ 15 0.213 (0.061, 0.366) 0.006 0.84 (0.60, 1.17) 0.295 1.08 (0.68, 1.72) 0.740

T1 ≤ 10 0 1 1

Risk Score 2b T2 11-12 0.129 (0.003, 0.254) 0.045 0.74 (0.56, 0.98) 0.038 0.81 (0.55, 1.20) 0.292

T3 ≥ 13 0.319 (0.166, 0.471) 4.31 x 10-5 0.57 (0.39 ,0.83) 0.003 0.72 (0.43, 1.19) 0.199

186

Chapter 6

Analysis of a Cardiovascular Disease Genetic Risk Score in the Diabetes Heart Study

Laura M. Raffield, Amanda J. Cox, J. Jeffrey Carr, Barry I. Freedman, Pamela J. Hicks, Carl D.

Langefeld, Fang-Chi Hsu, Donald W. Bowden

This manuscript was published online in February 2015 issue in Acta Diabetologica. The reference for this manuscript is as follows: Raffield, L.M., Cox, A.J., Carr, J.J., Freedman, B. I., Hicks, P.J., Langefeld, C. D., Hsu, F.C., Bowden, D. W. Analysis of a Cardiovascular Disease Genetic Risk Score in the Diabetes Heart Study. Acta Diabetologica, 2015. doi: 10.1007/s00592- 015-0720-5. Epub 21 February 2015. PubMed PMID: 25700702.

187

Abstract

Aims: It remains unclear whether the high cardiovascular disease (CVD) burden in people with type 2 diabetes (T2D) is associated with genetic variants that contribute to CVD in general populations. Recent studies have examined genetic risk scores of single nucleotide polymorphisms (SNPs) identified by genome-wide association studies (GWAS) for their cumulative contribution to CVD-related traits. Most analyses combined SNPs associated with a single phenotypic class, e.g. lipids. In the present analysis, we examined a more comprehensive risk score comprised of SNPs associated with a broad range of CVD risk phenotypes.

Methods: The composite risk score was analyzed for potential associations with subclinical

CVD, self-reported CVD events, and mortality in 983 T2D-affected individuals of European descent from 466 Diabetes Heart Study (DHS) families. Genetic association was examined using marginal models with generalized estimating equations for subclinical CVD and prior CVD events and Cox proportional hazards models with sandwich-based variance estimation for mortality; analyses were adjusted for age and sex.

Results: An increase in genetic risk score was significantly associated with higher levels of coronary artery calcified plaque (p=1.23 x 10-4); however, no significant associations with self- reported myocardial infarction and CVD events and all-cause and CVD mortality were observed.

Conclusions: These results suggest that a genetic risk score of SNPs associated with CVD events and risk factors does not significantly account for CVD risk in the DHS, highlighting the limitations of applying current genetic markers for CVD in individuals with diabetes.

188

Mortality risk from cardiovascular disease (CVD) is increased two to fourfold in individuals with type 2 diabetes (T2D), increasing interest in CVD risk prediction for this population (Go, Mozaffarian et al. 2013). Identifying genetic variants which elevate CVD risk is one strategy for risk prediction. Recent studies have examined genetic risk scores (GRS) comprised of multiple single nucleotide polymorphisms (SNPs) identified by genome-wide association studies (GWAS) for their cumulative impact on CVD-related traits. Most analyses have combined SNPs associated with a single trait class. For example, some studies have examined risk scores of SNPs associated with lipid levels for their impact on CVD events

(Voight, Peloso et al. 2012; Shah, Casas et al. 2013), while other studies have examined SNPs associated with CVD risk for their ability to predict incident events (Paynter, Chasman et al.

2010; Ripatti, Tikkanen et al. 2010) and for association with subclinical CVD (Thanassoulis,

Peloso et al. 2012; van Setten, Isgum et al. 2013). These analyses suggest that GRS may be important tools for CVD risk prediction in the general population. It remains unclear whether the same genetic variants influence CVD burden in individuals with T2D. All CVD-associated SNPs in general populations do not associate with risk in individuals with T2D (Farbstein and Levy

2010; Wang, Peng et al. 2010; Qi, Parast et al. 2011; Qi, Workalemahu et al. 2012), and variants have been described that impact CVD risk only in the presence of diabetes (Farbstein and Levy

2010; Qi, Qi et al. 2013) or whose effects are modified by factors such as glycemic control

(Doria, Wojcik et al. 2008) and obesity (Bacci, Rizza et al. 2011) in diabetes patients. In an effort to gain further insight into the performance of GRS in patients with T2D, we tested a composite

GRS of SNPs associated with CVD events and CVD risk factors.

The Diabetes Heart Study (DHS) is a family-based cohort enriched for T2D (Bowden,

Cox et al. 2010). Prior work in the DHS investigated the associations between SNPs associated with CVD events (Cox, Hsu et al. 2014), coronary artery calcified atherosclerotic plaque (CAC), a measure of subclinical CVD (Adams, Raffield et al. 2014), and high-density lipoprotein (HDL)

189

cholesterol (Raffield, Cox et al. 2013) and CAC, self-reported CVD events, and mortality. In this analysis, we extended these approaches to examine whether a more extensive risk score constructed from SNPs associated with a wide range of CVD risk factors is more strongly associated with measures of CVD risk, including CAC, self-reported CVD, and myocardial infarction (MI), as well as all-cause and CVD mortality, in DHS participants with diabetes.

Methods

Study Design and Sample

The DHS recruited T2D-affected siblings without advanced renal insufficiency from

1998 through 2005 in western North Carolina. T2D was defined as diabetes developing after the age of 35 years treated with changes in diet and exercise and/or oral agents, in the absence of historical evidence of ketoacidosis or initial treatment with insulin. Fasting glucose and glycated hemoglobin (HbA1C) were assessed at the exam visit. Ascertainment and recruitment have been described (Bowden, Cox et al. 2010). Analyses included 983 self-described European American individuals from 466 DHS families, all of whom were affected by diabetes.

Participant examinations were conducted in the General Clinical Research Center of

Wake Forest Baptist Medical Center. Examinations included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses, and spot urine collection. Data on prior MI and CVD events was self-reported by participants and non-adjudicated. Participants reporting history of a

CVD event including MI, angina, or stroke, history of vascular procedures including coronary angioplasty, coronary artery bypass graft, or endarterectomy, or with Q wave abnormalities indicative of prior MI were defined as having prior CVD. Estimated glomerular filtration rate

(eGFR) was computed using the CKD-EPI equation (Levey, Stevens et al. 2009). Low-density lipoprotein (LDL) cholesterol concentration was calculated using the Friedewald equation, and

190

LDL concentrations were considered valid for subjects whose triglycerides were less than 400 mg/dL. CAC was assessed using computed tomography (CT), summing the left main, left anterior descending, circumflex, posterior descending, and right coronary arteries. CT scans were performed on multi-detector CT scanners with cardiac gating in chest scans. CAC scores were measured as previously described and validated (Carr, Crouse et al. 2000; Carr, Nelson et al.

2005). Not all measures were available in all DHS participants.

Mortality was assessed using the National Social Security Death Index. For deceased participants, length of follow-up was determined from the date of initial study visit to date of death. For all other participants the length of follow-up was determined from the date of the initial study visit to December 31, 2013. When possible, copies of death certificates were obtained from county or state Vital Records Offices to determine cause of death. Cause of death was categorized based on death certificates as CVD mortality (MI, congestive heart failure, cardiac arrhythmia, sudden cardiac death, peripheral vascular disease, and stroke) or as mortality from cancer, infection, end-stage renal disease, accidental, or other causes (including obstructive pulmonary disease, pulmonary fibrosis, liver failure and Alzheimer’s disease). Cause of death could not be obtained for 14 participants; these participants were excluded from all analyses of

CVD mortality.

All study protocols were approved by the Institutional Review Board at Wake Forest

School of Medicine, and all participants provided written informed consent.

SNP Selection

SNPs selected for the risk score were genome-wide significant (p-value <5 x 10-8) in large

GWAS studies from the NHGRI GWAS catalog in populations of European descent. For SNPs in linkage disequilibrium (r2 > 0.1) within risk scores for each trait, only one SNP was included.

SNPs included in the risk score are listed in Supplementary Table 1, including SNPs associated

191

with blood pressure traits, body mass index, CAC, coronary artery disease (CAD), C-reactive protein (CRP), electrocardiogram traits, fasting plasma glucose, eGFR, HDL cholesterol, LDL cholesterol, total cholesterol, triglycerides, stroke, and T2D. SNP positions are for the

GRCh37/hg19 reference genome.

Genotyping

Genotyping was completed using an Illumina Infinium HumanExome BeadChip array

(Exome array), an Affymetrix Genome-Wide Human SNP Array 5.0 (GWAS array), imputation of 1000 Genomes project SNPs from this array using IMPUTE2 and the Phase I v2, cosmopolitan

(integrated) reference panel, build 37 (Howie, Donnelly et al. 2009), and direct genotyping using the Sequenom platform (Supplementary Table 1). The array based sources of genotype data and imputation in the DHS have been described previously (Raffield, Cox et al. 2013). 983 European

American individuals with T2D from the DHS had both GWAS array and Exome array data and were included in our GRS analysis. Some SNPs not available from the array-based datasets were directly genotyped using the MassARRAY SNP Genotyping System (Buetow, Edmonson et al.

2001; Cox, Lehtinen et al. 2013). In addition to the 375 SNPs included in our risk scores, an additional 30 SNPs met our criteria for inclusion but were not available.

All SNPs included in the GRS were required to have a minimum call rate of 95% and a

Hardy Weinberg equilibrium p-value>1x10-6. Unweighted GRS for all traits were derived by adding the number of risk alleles across each SNP, including the composite GRS, which was derived by adding the risk alleles across 375 SNPs (Supplementary Table 1). For individuals missing genotype data for a particular SNP, the mean genotype calculated in the DHS for that given SNP was assigned (Fontaine-Bisson, Renström et al. 2010). Scores for the composite GRS ranged from 343 to 421 (382.3  13.4, mean  SD). All risk scores were coded so an increase in risk score would be expected to correlate with an increase in CVD risk.

192

Statistical Analysis

Continuous variables were transformed as necessary to approximate normality prior to the analysis. The natural logarithm of fasting plasma glucose, total cholesterol, (CAC+1), triglyceride levels, CRP, QT interval, QRS interval, and PR interval was employed, and the square root was used for HDL and BMI. All single SNP association analyses were implemented using variance components methods in SOLAR version 6.5.8 (Texas Biomedical Research

Institute, San Antonio, TX) to account for familial relationships (Almasy and Blangero 1998).

Association was examined assuming an additive model of inheritance. Risk score associations were examined using marginal models with generalized estimating equations for subclinical CVD and self-reported CVD events, accounting for familial correlation using a sandwich estimator of the variance under exchangeable correlation. A similar approach was used to test for differences between the top and bottom 10% of the composite GRS distribution. Cox proportional hazards models with sandwich-based variance estimation were used to examine associations with all- cause and CVD mortality. Risk score analyses were performed in SAS 9.3. Single SNP and GRS analyses were adjusted for age and sex. Analysis of the T2D risk score for association with diabetes affected status was not completed, as all individuals included in these analyses were affected by T2D.

Results

The demographic characteristics of the 983 individuals from 466 DHS families included in this GRS analysis are displayed in Table 1. All participants were affected by T2D, with average diabetes duration of 11 ± 7 (mean ± standard deviation (SD)) years. The cohort has a high prevalence of hypertension, obesity, subclinical CVD based on CAC, and prior CVD events.

All-cause mortality was 31.1% over an average follow-up of 10 ± 3 years (mean ± SD).

193

SNPs selected for inclusion in the individual GRS are listed in Supplementary Table 1.

Each SNP’s association with the trait of interest for the risk score for which it was selected was assessed (Supplementary Table 2). The most significant single SNP association was for rs3764261 with HDL cholesterol (p= 5.6 x 10-5). Of the 375 SNPs analysed, only 42 were nominally associated (p<0.05) with the trait of interest in the DHS. However, when the SNPs were analyzed in combination in GRS, most risk scores were associated with the trait of interest

(Table 2); the strongest association was for the 32 SNP triglyceride risk score (p=1.75 x 10-6) with triglyceride levels.

We also assessed the association of both single SNPs and the individual, trait-specific

GRS with all-cause mortality, CVD mortality, self-reported history of MI, self-reported history of

CVD, and CAC. The single SNP association results are shown in Supplementary Table 2; few

SNPs showed evidence of association with these traits in single SNP association analyses, with no SNPs meeting a Bonferroni-adjusted p-value threshold of <1.33 x 10-4 (α=0.05 for 375 SNPs).

Few SNPs were even nominally associated with these phenotypes (p<0.05), with 21 SNPs nominally associated with all-cause mortality, 19 with CVD mortality, 16 with self-reported history of MI, 12 with self-reported history of CVD, and 20 with CAC. The results for the trait- specific GRS are shown in Supplementary Table 3. Most risk scores were not associated with all- cause or CVD mortality, with the exception of nominal associations between the CAD and T2D risk scores and risk of CVD mortality. No individual risk scores were strongly associated with self-reported history of CVD events, but the CAC risk score was nominally associated (p=0.028) with increased odds of prior CVD events. The CAD GRS was associated (p=0.0002) with increased odds of a history of MI; eGFR and stroke risk scores were also nominally associated with MI history. Associations with increased CAC were observed for the CAD (p=0.0001) and triglycerides (p=0.007) GRS, with nominal association with increased CAC also observed for the

CAC risk score.

194

Next, the composite GRS derived from 375 SNPs associated with blood pressure traits,

BMI, CAC, CAD, CRP, electrocardiogram traits, fasting plasma glucose, eGFR, HDL cholesterol, LDL cholesterol, stroke, T2D, total cholesterol, and triglycerides was assessed. An increase in GRS was associated with increased CAC burden (p=1.23 x 10-4), but significant associations with self-reported history of MI and CVD events were not observed (Table 3).

Associations with all-cause or CVD mortality were also not detected (Table 3). The association with CAC is reflected by differences in the mean CAC burden for individuals in the bottom or top

10% of the composite GRS distribution, as compared to the whole sample (Table 4). To further refine these results, a model including each trait-specific GRS as a separate variable in the model was analyzed, adjusted for age and sex, in order to determine which risk scores were driving the observed association with the composite GRS. The association of the combined GRS with increased CAC appeared to be driven mainly by the CAD SNPs (p=0.001), with the electrocardiogram trait SNPs (p=0.042) and the triglyceride SNPs (p=0.040) also nominally associated with increased CAC and contributing to the association (Supplementary Table 4).

Conclusions

This study explored potential associations with a genetic risk score created from SNPs associated with CVD events and CVD risk factors in general populations in a T2D-affected cohort with measures of CAC, self-reported CVD events, and mortality. Not surprisingly, given the prior associations reported in GWAS analyses, risk scores of SNPs selected for each individual CVD risk trait (for example lipid levels, QT interval, and CRP) tended to be associated with that trait of interest (Table 2), indicating that at least some variants identified as associated with CVD risk factors in the general population also impact risk factors in individuals with T2D.

However, while the composite risk score was associated with increased CAC burden, associations with self-reported CVD events and mortality were not observed. There are a number of potential explanations; for example, analysis of a continuous trait like CAC has more power, and self-

195

reported events and CVD mortality have risks of potential misclassification. Given these issues, the genetic association results highlight the limitations of GRS in this high CVD risk group of diabetes affected individuals and raises questions about the translational value of GWAS identified SNPs for risk prediction in community-based cohorts at high CVD risk.

The association of our GRS with CAC shows that the selected CVD-related SNPs are associated with elevated CVD burden in individuals with T2D (p=1.23 x 10-4, β value 0.021 for change in ln transformed CAC per one unit increase in GRS, 95% confidence interval (0.010,

0.031), corresponding to a β value of 0.275 for a one SD change in GRS). This is reflected by a lower mean CAC burden for individuals in the bottom 10% of the composite GRS distribution and higher CAC burden for those in top 10% as compared to the cohort as a whole (Table 4).

CAC is a strong independent predictor of CVD events and mortality in the general population and in T2D (Raggi, Shaw et al. 2004; Detrano, Guerci et al. 2008; Folsom, Kronmal et al. 2008;

Elias-Smale, Proenca et al. 2010; Erbel, Mohlenkamp et al. 2010; Polonsky, McClelland et al.

2010; Agarwal, Morgan et al. 2011; Agarwal, Cox et al. 2013), with individuals affected by diabetes tending to have higher CAC (Hoff, Quinn et al. 2003). However, the association of the composite GRS with CAC seemed to be driven by only a few of the included risk scores; in a model including all of the individual GRS, the only associations were for the CAD, electrocardiogram traits, and triglycerides GRS (Supplementary Table 4), which likely explain much of the association of the composite risk score. A prior analysis of an unweighted risk score in the DHS of 30 SNPs associated with risk of CAD events found a strong association with CAC

(p= 7.34 x 10-5 for unweighted risk score model adjusted for age, sex, and diabetes affected status) (Cox, Hsu et al. 2014), so the association of the CAD risk score included here with CAC

(p=0.0001 (Supplementary Table 3), p=0.001 in model adjusted for other risk scores

(Supplementary Table 4)) is not surprising.

196

The lack of association between our composite risk score and all-cause and CVD mortality and self-report of prior CVD events may be explained by a number of factors. All scores analyzed were unweighted by previously reported SNP effect size; this may weaken associations, but effect size estimates would be difficult to use for weighting in this study as

SNPs within scores were reported in different publications with varying sample sizes, recruitment criteria, and phenotype definitions. The lack of association with prior CVD events may be due to the availability of only self-reported, non-adjudicated event data in the DHS cohort; underreporting or overreporting of events by patients may have weakened potential associations.

Inaccuracies in cause of death data from death certificates are well documented (Coady, Sorlie et al. 2001; Wexelman, Eden et al. 2013); again, under- or overreporting of CVD causes of death by physicians may have weakened associations with CVD mortality. More broadly, GWAS studies to date have identified common variants that account for only a modest percentage of the heritability of most complex traits, with most identified variants having small effect sizes, leading to recent efforts to identify other factors that may contribute to the “missing heritability”

(Manolio, Collins et al. 2009). While our individual risk scores were generally associated with the trait of interest (Table 2), for example BMI, HDL, and CRP levels, the estimated effect sizes were small. Differential effects of genetic variants in individuals affected by T2D may also contribute to these small effect sizes for the GRS, as well as the lack of association with mortality for the composite score. A number of previous studies have found different genetic contributors to CVD risk in diabetes affected individuals as compared to general population cohorts (Farbstein and

Levy 2010; Wang, Peng et al. 2010; Qi, Parast et al. 2011; Qi, Workalemahu et al. 2012; Qi, Qi et al. 2013), with factors such as glycemic control (Doria, Wojcik et al. 2008) and obesity (Bacci,

Rizza et al. 2011) potentially modifying the impact of genetic variants in individuals affected by

T2D. Finally, Mendelian randomization studies have cast doubts that some of the risk factors included (for example HDL (Voight, Peloso et al. 2012) and CRP (Wensley, Gao et al. 2011)) are causally associated with risk of CVD events, which could weaken the association of our

197

composite GRS with CVD risk, though in our analysis results were essentially unchanged when

HDL and CRP associated SNPs were excluded from the composite risk score analysis.

Previous analyses of GRS have found fairly modest predictive power for these scores. A prior analysis of 102 SNPs associated with CVD events and major CVD risk factors, including

LDL, HDL, triglycerides, diabetes, fasting plasma glucose, systolic and diastolic blood pressure, and CRP, in the Framingham Heart Study found a fairly modest association with high CAC

(defined as an Agatston score greater than the 75th percentile in a healthy population) (p=0.002)

(Thanassoulis, Peloso et al. 2012). In analysis of a GRS of 24 CAD SNPs, the risk score was able to explain only 2.4% of Agatston score variance in a cohort of men who were current or former heavy smokers (van Setten, Isgum et al. 2013). A GRS of 101 SNPs associated with CVD or

CVD intermediate phenotypes was modestly associated with increased risk of CVD events in the

Women’s Genome Health Study (HR 1.02 per risk allele, 95% CI (1.00-1.03)), but the GRS was no longer associated after adjustment for traditional CVD risk factors (Paynter, Chasman et al.

2010).

In our analysis of a GRS constructed from 375 SNPs previously associated with CVD events or risk factors, we observed modest association of this risk score with CAC, a marker of subclinical CVD risk, but observed no association with self-reported CVD events or mortality in the DHS cohort. While new GWAS meta-analyses continue to discover new variants, for example for lipid traits (Willer, Schmidt et al. 2013), these variants were able to be discovered only by very large recent meta-analyses due to their small effect sizes, making it unlikely that addition of these variants to our GRS would lead to a significantly stronger association. However, ongoing analyses of lower frequency coding variants and copy number variation may identify higher impact variants that may significantly impact CVD risk in individuals in individuals with T2D.

The current analysis points to the limited associations between currently identified CVD-related common genetic variants and CVD risk, in particular mortality risk, in individuals of European

198

descent affected by T2D. This lack of utility may partly be due to differential impacts of genetic variants in individuals affected by T2D and highlights the need for further analysis of genetic contributors to CVD risk in diabetes affected cohorts.

Acknowledgements

The authors thank the other investigators, the staff, and the participants of the DHS study for their valuable contributions. This study was supported by the National Institutes of Health through R01

HL67348 and R01 HL092301 (to DWB), R01 AR48797 (to JJC), and F31 AG044879 (to LMR).

Conflict of Interest

None.

Statement of Human and Animal Rights

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki

Declaration of 1975, as revised in 2008 (5).

Statement of Informed Consent

Informed consent was obtained from all patients for being included in the study.

199

Table 1. Demographic characteristics of 983 individuals from 466 families in the Diabetes Heart

Study.

Phenotype Mean (SD) or % Median (range)

Demographics

Age (yrs) 62.5 (9.1) 63.0 (34.2 - 86.0)

Gender (% female) 51.8%

BMI (kg/m2) 32.3 (6.6) 31.2 (17.1 – 58.0)

Past smoking 43.1 %

Current smoking 16.3 %

Medications

Cholesterol Medication 47.8%

Insulin 27.6%

Oral Diabetes Medication 78.8%

Hypertension Medication 75.9%

Diabetes Measures

Duration of diabetes (yrs) 11 (7) 8 (0 - 46)

Glycated hemoglobin (%) 7.6 (1.7) 7.2 (4.3 - 18.3)

Glycated hemoglobin (mmol/mol) 60 (19) 55 (23 - 177)

Fasting glucose (mg/dL) 148 (56) 135 (16 - 463)

Lipids

Total Cholesterol (mg/dL) 185 (44) 181 (65 - 427)

HDL Cholesterol (mg/dL) 42 (12) 41(8 - 98)

200

LDL Cholesterol (mg/dL) 103 (33) 100 (12 - 236)

Triglycerides (mg/dL) 209 (140) 175 (30 - 1310)

Cardiovascular Disease Measures

Coronary Artery Calcification 1888 (3382) 478 (0 - 50415)

Self-reported history of prior CVD 43.2%

Self-reported history of myocardial infarction 21.6%

Self-reported history of stroke 10.1%

Systolic BP (mmHg) 140 (19) 139 (94 - 260)

Diastolic BP (mmHg) 73 (10) 72 (37 - 106)

Pulse Pressure (mmHg) 67 (17) 66 (28 - 159)

Estimated Glomerular Filtration Rate 66 (18) 64 (9 - 126) (ml/min/1.73m2)

PR interval (ms) 169 (32) 166 (0 - 420)

QRS interval (ms) 96 (18) 92 (64 - 192)

QT interval (ms) 394 (33) 392 (270 - 564)

C-reactive protein (mg/dL) 0.61 (0.97) 0.30 (0.005 - 12.7)

Mortality

All-cause mortality 31.1%

CVD mortality 14.3%

201

Table 2: Individual genetic risk scores and their association with the trait of interest in the DHS.

Relationships between the risk scores and the traits were examined using marginal models with generalized estimating equations. Models were adjusted for age and sex.

SNP Selection Number of β Standard p-value Trait in DHS Trait SNPs Estimate Error

Blood pressure traits 24 0.244 0.161 0.13 Pulse Pressure

BMI 31 0.023 0.006 3.79 x 10-5 BMI

Coronary Artery Coronary Artery 2 0.178 0.073 0.015 Calcification Calcification

Self-reported

Coronary Artery history of 28 1.10 1.03 2.43 x 10-4 Disease* myocardial

infarction

C-reactive C-reactive protein 18 0.077 0.019 3.32 x 10-5 protein

Electrocardiogram 22 0.003 0.001 5.82 x 10-4 QT interval traits

0.005 0.002 0.016 QRS interval

0.002 0.002 0.226 PR interval

Fasting glucose 16 0.002 0.004 0.604 Fasting glucose

eGFR 23 -0.408 0.167 0.015 eGFR

HDL Cholesterol 40 -0.032 0.007 3.32 x 10-6 HDL

LDL Cholesterol 41 0.794 0.279 0.004 LDL

202

Self-reported Stroke* 5 1.04 1.09 0.643 history of stroke

Type 2 Diabetes 47 - - - Type 2 Diabetes

Total Cholesterol 46 0.006 0.002 6.74 x 10-4 Total Cholesterol

Triglycerides 32 0.026 0.005 1.75 x 10-6 Triglycerides

* Results shown as odds ratio.

203

Table 3: Associations between composite genetic risk score and prior cardiovascular disease

(CVD), prior myocardial infarction (MI), and coronary artery calcification in the DHS were assessed using marginal models with generalized estimating equations. Associations between the composite genetic risk score and all-cause and CVD mortality were assessed using Cox proportional hazards models. Models were adjusted for age and sex.

Trait β Estimate/ 95% Confidence p-value

Odds or Hazard Interval

Ratio

Self-reported history of 1.01 1.00 1.02 0.149 MI *

Self-reported history of 1.00 0.99 1.01 0.481 CVD *

Coronary artery 0.021 0.010 0.031 1.23 x 10-4 calcification Ϯ

All-cause mortality ‡ 1.00 0.99 1.01 0.919

CVD mortality ‡ 1.00 0.99 1.01 0.718

*Calculated using marginal models with generalized estimating equations, odds ratios shown, Ϯ calculated using marginal models with generalized estimating equations, β estimates shown, ‡

Calculated using Cox proportional hazards models, hazard ratios shown

204

Table 4: Mean coronary artery calcification and prevalence of prior cardiovascular disease

(CVD), prior myocardial infarction (MI), and all-cause and CVD mortality for all study participants, participants in the bottom 10% of the composite risk score (GRS) distribution, and participants in the top 10% of the composite GRS distribution. P-value is for the difference between the bottom and top 10% of the composite GRS.

All Bottom 10% Top 10% of p-value

Participants- of Composite Composite Phenotype Mean (SD) or GRS- Mean GRS- Mean

% (SD) or % (SD) or %

Self-reported history of 21.60% 22.68% 29.47% 0.280* MI

Self-reported history of 43.20% 42.11% 46.88% 0.461* CVD

Coronary artery 1888 (3382) 1575 (2355) 2216 (2971) 0.009*Ϯ calcification

All-cause mortality 31.10% 29.59% 25.77% 0.526*

CVD mortality 14.30% 17.02% 7.29% 0.051*

*Calculated using marginal models with generalized estimating equations, Ϯ for ln (coronary artery calcification + 1)

205

Supplementary Table 1: SNPs included in genetic risk scores. Abbreviations: GWAS, genome-wide association study array; Exome, exome array;

MAF, minor allele frequency in the Diabetes Heart Study; HWE, Hardy Weinberg Equilibrium p-value in unrelated individuals from the Diabetes

Heart Study; M/m, Major allele/minor allele; CAC, coronary artery calcification; CAD, coronary artery disease; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate.

Chr Position SNP Gene Source Risk Score Alleles Risk MAF HWE Reference (M/m) Allele

1 11862778 rs17367504 MTHFR Exome Blood pressure traits A/G A 0.18 0.64 1

3 27537909 rs13082711 SLC4A7 Genotyped Blood pressure traits T/C C 0.22 0.90 1

3 47927484 rs319690 MAP4 Imputed Blood pressure traits T/C T 0.31 0.31 1

3 169100886 rs419076 MECOM Imputed Blood pressure traits T/C T 0.49 0.02 1

4 54799245 rs871606 CHIC2 Imputed Blood pressure traits T/C T 0.08 0.73 1

4 81164723 rs1458038 FGF5 Genotyped Blood pressure traits G/A A 0.30 1.00 1

4 103188709 rs13107325 SLC39A8 Exome Blood pressure traits G/A G 0.08 0.34 1

5 32815028 rs1173771 NPR3 Genotyped Blood pressure traits C/T C 0.38 0.22 1

5 157804457 rs9313772 EBF1 GWAS Blood pressure traits C/T C 0.38 1.00 1

6 26107463 rs198846 HFE Exome Blood pressure traits G/A A 0.18 0.04 1

7 106411858 rs17477177 PIK3CG Imputed Blood pressure traits T/C C 0.21 0.88 1

8 120435812 rs2071518 NOV Imputed Blood pressure traits C/T T 0.27 0.06 1

10 18727959 rs12258967 CACNB2 Genotyped Blood pressure traits C/G C 0.30 0.12 1

206

10 63467553 rs4590817 C10orf107 Imputed Blood pressure traits G/C G 0.15 0.35 1

10 95895177 rs9663362 PLCE1 GWAS Blood pressure traits G/C C 0.46 0.85 1

10 104846178 rs11191548 NT5C2 Exome Blood pressure traits A/G A 0.08 0.55 1

11 16902268 rs381815 PLEKHA7 Imputed Blood pressure traits C/T T 0.28 1.00 1

11 130273230 rs11222084 ADAMTS8 GWAS Blood pressure traits A/T T 0.35 0.19 1

12 90060586 rs17249754 ATP2B1 Exome Blood pressure traits G/A G 0.18 1.00 1

12 112007756 rs653178 ATXN2 Exome Blood pressure traits A/G G 0.50 0.31 1

15 75077367 rs1378942 CSK Exome Blood pressure traits A/C C 0.32 0.11 1

17 45013271 rs17608766 GOSR2 Exome Blood pressure traits A/G G 0.15 1.00 1

20 10969030 rs1327235 JAG1 GWAS Blood pressure traits A/G G 0.48 1.00 1

20 57751117 rs6015450 ZNF831 Genotyped Blood pressure traits A/G G 0.12 0.09 1

1 72812440 rs2815752 NEGR1 Exome BMI A/G A 0.38 0.38 2

1 74991644 rs1514175 TNNI3K Exome BMI G/A A 0.42 0.63 2

1 96944797 rs1555543 PTBP2 Exome BMI C/A C 0.40 0.01 2

1 177889480 rs543874 SEC16B Exome BMI A/G G 0.20 0.89 2

2 622827 rs2867125 TMEM18 Exome BMI G/A G 0.15 0.71 2

2 25158008 rs713586 ADCY3 Exome BMI A/G G 0.49 0.64 2

2 59302877 rs887912 FANCL Exome BMI G/A A 0.30 0.19 2

2 142959931 rs2890652 LRP1B Imputed BMI T/C C 0.17 0.48 2

207

3 85884150 rs13078807 CADM2 Exome BMI A/G G 0.21 0.33 2

3 185834499 rs9816226 ETV5 Exome BMI A/T A 0.20 0.30 2

4 45182527 rs10938397 GNPDA2 Exome BMI A/G G 0.44 0.26 2

4 103188709 rs13107325 SLC39A8 Exome BMI G/A A 0.08 0.34 2

5 75015242 rs2112347 HMGCR Exome BMI A/C A 0.36 0.84 2

6 34302869 rs206936 NUDT3 Exome BMI A/G G 0.21 1.00 2

6 50803050 rs987237 TFAP2B Exome BMI A/G G 0.18 0.11 2

9 28414339 rs10968576 LRRN6C Exome BMI A/G G 0.31 0.74 2

11 8604593 rs4929949 STK33 Exome BMI G/A G 0.46 0.51 2

11 27725986 rs10767664 BDNF Exome BMI A/T A 0.20 0.89 2

11 47650993 rs3817334 MTCH2 Exome BMI G/A A 0.43 0.15 2

12 50247468 rs7138803 FAIM2 Exome BMI G/A A 0.37 0.37 2

13 28020180 rs4771122 MTIF3 Exome BMI A/G G 0.23 0.69 2

14 30515112 rs11847697 PRKD1 Exome BMI G/A A 0.05 0.61 2

14 79936964 rs10150332 NRXN3 Imputed BMI T/C C 0.23 0.60 2

15 68086838 rs2241423 MAP2K5 Exome BMI G/A G 0.23 0.11 2

16 19933600 rs12444979 GPRC5B Exome BMI G/A G 0.16 0.86 2

16 28885659 rs7359397 SH2B1 Exome BMI G/A A 0.42 0.70 2

16 53803574 rs1558902 FTO Exome BMI A/T T 0.45 0.19 2

208

18 57839769 rs571312 MC4R Exome BMI C/A A 0.23 1.00 2

19 34309532 rs29941 KCTD15 Exome BMI G/A G 0.31 0.83 2

19 46202172 rs2287019 QPCTL Exome BMI G/A G 0.19 0.17 2

19 47569003 rs3810291 ZC3H4 Exome BMI A/G A 0.32 1.00 2

6 12903957 rs9349379 PHACTR1 Exome CAC A/G G 0.36 0.84 3

9 22125503 rs1333049 CDKN2B Exome CAC C/G G 0.47 0.46 3

1 55496039 rs11206510 PCSK9 Exome CAD A/G A 0.22 1.00 4

1 56962821 rs17114036 PPAP2B Exome CAD A/G A 0.07 1.00 4

1 222823529 rs17465637 MIA3 GWAS CAD C/A C 0.28 0.82 4

1 1098822166 rs599839 SORT1 Exome CAD A/G A 0.24 0.53 4

2 203745885 rs6725887 WDR12 GWAS CAD T/C C 0.14 0.43 4

3 138122122 rs9818870 MRAS GWAS CAD C/T T 0.17 0.73 5

6 12927544 rs12526453 PHACTR1 GWAS CAD C/G C 0.35 1.00 4

6 31184196 rs3869109 HCG27 Exome CAD G/A G 0.42 0.50 6

6 35034800 rs17609940 ANKS1A GWAS CAD G/C G 0.20 1.00 4

6 151252985 rs6922269 MTHFD1L GWAS CAD G/A A 0.29 0.03 7

6 160961137 rs3798220 LPA Exome CAD A/G G 0.02 1.00 4

7 107244545 rs10953541 BCAP29 Exome CAD G/A G 0.23 0.44 8

7 129663496 rs11556924 ZC3HC1 GWAS CAD C/T C 0.39 0.33 4

209

9 22031005 rs7865618 MTAP Exome CAD A/G A 0.43 1.00 9

9 136154168 rs579459 ABO Exome CAD A/G G 0.22 0.28 4

10 30335122 rs2505083 KIAA1462 Exome CAD A/G G 0.42 0.63 8

10 44775824 rs1746048 CXCL12 GWAS CAD C/T C 0.12 0.66 4

10 91002927 rs1412444 LIPA Genotyped CAD G/A A 0.35 0.13 8

10 104719096 rs12413409 CNNM2 GWAS CAD G/A G 0.08 0.77 4

11 103660567 rs974819 PDGFD Exome CAD G/A A 0.28 0.91 8

11 116648917 rs964184 ZNF259 Exome CAD G/C C 0.16 0.73 4

13 110960712 rs4773144 COL4A1 Exome CAD A/G G 0.41 0.44 4

14 100133942 rs2895811 HHIPL1 Exome CAD A/G G 0.42 0.51 4

15 79089111 rs3825807 ADAMTS7 Exome CAD A/G A 0.47 0.26 4

17 2126504 rs216172 SMG6 Exome CAD G/C C 0.38 0.77 4

17 17543722 rs12936587 PEMT Exome CAD G/A G 0.44 0.93 4

17 46988597 rs46522 UBE2Z Exome CAD A/G A 0.46 0.35 4

19 11163601 rs1122608 SMARCA4 Exome CAD C/A C 0.26 0.90 4

1 40064961 rs12037222 PABPC4 Imputed CRP G/A A 0.23 1.00 10

1 66161461 rs4420065 LEPR Exome CRP G/A G 0.40 0.63 10

1 154426264 rs4129267 IL6R Exome CRP G/A G 0.38 1.00 10

1 159678816 rs2794520 CRP Exome CRP G/A G 0.33 0.92 10

210

1 247601595 rs12239046 NLRP3 Exome CRP G/A G 0.37 0.07 10

2 27730940 rs1260326 GCKR Exome CRP G/A A 0.39 0.24 10

2 113841030 rs6734238 IL1F10 Exome CRP A/G G 0.40 1.00 10

5 131839618 rs4705952 IRF1 Exome CRP A/G G 0.23 0.29 10

6 117114025 rs6901250 GPRC6A Exome CRP G/A A 0.33 0.03 10

7 72971231 rs13233571 BCL7B Exome CRP G/A G 0.12 1.00 10

8 9183358 rs9987289 PPP1R3B Exome CRP G/A A 0.11 0.63 10

12 103483094 rs10745954 ASCL1 Exome CRP A/G A 0.49 0.78 10

12 121420807 rs1183910 HNF1A Exome CRP G/A G 0.31 0.45 10

15 60894965 rs340029 RORA Exome CRP A/G A 0.38 0.92 10

16 51158710 rs10521222 SALL1 Exome CRP G/A G 0.06 1.00 10

18 12821593 rs2847281 PTPN2 Exome CRP A/G A 0.41 0.02 10

19 45422946 rs4420638 APOC1 Exome CRP A/G A 0.16 0.31 10

20 43042364 rs1800961 HNF4A Exome CRP G/A G 0.03 1.00 10

1 6279370 rs846111 RNF207 Exome Electrocardiogram traits G/C C 0.27 0.81 11

1 162085685 rs10494366 NOS1AP Exome Electrocardiogram traits A/C C 0.34 0.47 12

1 162210610 rs4657178 NOS1AP Exome Electrocardiogram traits G/A A 0.23 1.00 13

1 169099483 rs10919071 ATP1B1 Exome Electrocardiogram traits A/G A 0.12 1.00 13

3 38593393 rs12053903 SCN5A Exome Electrocardiogram traits A/G A 0.32 1.00 11

211

3 38766675 rs6795970 SCN10A Exome Electrocardiogram traits G/A A 0.38 0.49 14

4 86651464 rs7660702 ARHGAP24 Exome Electrocardiogram traits A/G A 0.32 0.83 12

6 36622900 rs1321311 CDKN1A Exome Electrocardiogram traits C/A A 0.25 0.90 12

6 118653204 rs12210810 PLN Exome Electrocardiogram traits G/C G 0.05 1.00 13

6 118680374 rs11970286 PLN Exome Electrocardiogram traits G/A A 0.46 0.78 13

7 116186241 rs3807989 CAV1 Exome Electrocardiogram traits G/A A 0.42 0.10 12

7 150667210 rs3807375 KCNH2 Exome Electrocardiogram traits G/A A 0.38 0.32 12

11 2484803 rs2074238 KCNQ1 Exome Electrocardiogram traits G/A G 0.07 1.00 11

11 2502319 rs12576239 KCNQ1 Exome Electrocardiogram traits G/A A 0.13 0.84 11

12 114795443 rs3825214 TBX5 Exome Electrocardiogram traits A/G G 0.19 0.17 12

13 48162558 rs2478333 HTR2A Exome Electrocardiogram traits C/A A 0.37 1.00 15

14 23861811 rs365990 MYH6 Genotyped Electrocardiogram traits A/G G 0.38 0.23 12

16 11691753 rs8049607 LITAF Exome Electrocardiogram traits G/A A 0.50 0.64 11

16 58622178 rs7188697 NDRG4 Exome Electrocardiogram traits A/G A 0.24 0.12 13

17 33324382 rs2074518 LIG3 Exome Electrocardiogram traits G/A G 0.42 0.15 11

17 68494992 rs17779747 KCNJ2 Exome Electrocardiogram traits C/A C 0.34 0.68 13

21 35880072 rs727957 KCNE1 Exome Electrocardiogram traits C/A A 0.17 0.19 12

1 214159256 rs340874 PROX1 Exome Fasting glucose G/A G 0.42 0.34 16

2 27741237 rs780094 GCKR Exome Fasting glucose G/A G 0.38 0.27 16

212

2 169763148 rs560887 G6PC2 Exome Fasting glucose G/A G 0.30 0.12 16

3 123065778 rs11708067 ADCY5 Exome Fasting glucose A/G A 0.24 0.61 16

3 170717521 rs11920090 SLC2A2 Exome Fasting glucose T/A T 0.14 0.32 16

7 15064309 rs2191349 DGKB Exome Fasting glucose A/C A 0.44 0.39 16

7 44235668 rs4607517 GCK Exome Fasting glucose G/A A 0.20 0.88 16

8 118185733 rs11558471 SLC30A8 Exome Fasting glucose A/G A 0.31 0.19 16

9 4289050 rs7034200 GLIS3 Exome Fasting glucose C/A A 0.46 0.11 16

10 113042093 rs10885122 ADRA2A Exome Fasting glucose C/A C 0.12 1.00 16

10 114756041 rs4506565 TCF7L2 Exome Fasting glucose A/T T 0.38 0.92 16

11 45873091 rs11605924 CRY2 Exome Fasting glucose C/A A 0.50 0.64 16

11 47336320 rs7944584 MADD Exome Fasting glucose A/T A 0.28 0.49 16

11 61571478 rs174550 FADS1 Exome Fasting glucose A/G A 0.33 0.83 16

11 92708710 rs10830963 MTNR1B Exome Fasting glucose C/G G 0.28 0.08 16

15 62433962 rs11071657 C2CD4B Exome Fasting glucose A/G A 0.38 0.09 16

1 150951477 rs267734 ANXA9 Exome eGFR A/G A 0.20 0.56 17

2 27730940 rs1260326 GCKR Exome eGFR G/A G 0.39 0.24 17

2 73900900 rs10206899 NAT8 Exome eGFR A/G A 0.24 1.00 18

3 141807137 rs347685 TFDP2 Exome eGFR A/C A 0.27 0.56 17

4 77398015 rs10032549 SHROOM3 Imputed eGFR G/A G 0.49 0.44 18

213

5 176817636 rs6420094 SLC34A1 Exome eGFR A/G G 0.35 0.13 17

6 43806609 rs881858 VEGFA Exome eGFR A/G A 0.28 0.82 17

6 160668389 rs2279463 SLC22A2 Exome eGFR A/G G 0.13 0.53 17

7 77416439 rs6465825 TMEM60 Exome eGFR A/G G 0.43 0.07 17

8 23751151 rs10109414 STC1 Exome eGFR G/A A 0.43 0.39 17

9 71434707 rs4744712 PIP5K1B Imputed eGFR C/A A 0.38 0.10 17

10 1156165 rs10794720 WDR37 Exome eGFR G/A A 0.07 0.29 17

11 30760335 rs3925584 MPPED2 Exome eGFR A/G A 0.46 0.40 19

12 349298 rs10774021 SLC6A13 Exome eGFR A/G A 0.37 0.11 17

12 112007756 rs653178 ATXN2 Exome eGFR A/G G 0.50 0.31 17

13 72347696 rs626277 DACH1 Exome eGFR A/C A 0.41 0.77 17

15 45719187 rs17536527 SPATA5L1 Genotyped eGFR G/C C 0.45 0.38 18

15 53946593 rs491567 WDR72 Exome eGFR A/C A 0.25 0.39 17

15 76158983 rs1394125 UBE2Q2 Exome eGFR G/A A 0.35 0.61 17

16 20367690 rs12917707 UMOD Exome eGFR C/A C 0.18 0.88 17

17 59456589 rs9895661 BCAS3 Exome eGFR A/G G 0.15 0.37 17

19 33356891 rs12460876 SLC7A9 Exome eGFR A/G A 0.39 0.20 17

20 23612737 rs911119 CST3 Exome eGFR A/G A 0.21 0.57 17

1 40028180 rs4660293 PABPC4 Exome HDL cholesterol A/G G 0.25 0.90 20

214

1 230295691 rs4846914 GALNT2 Exome HDL cholesterol A/G G 0.40 0.29 20

2 21206183 rs6754295 APOB Exome HDL cholesterol A/C A 0.25 0.80 21

2 165540800 rs12328675 COBLL1 Exome HDL cholesterol A/G A 0.11 0.16 20

2 227100698 rs2972146 IRS1 Exome HDL cholesterol A/C A 0.36 0.84 20

5 53298025 rs6450176 ARL15 Exome HDL cholesterol G/A A 0.25 0.27 20

6 34552797 rs2814944 C6orf106 Exome HDL cholesterol G/A A 0.17 0.62 20

6 139829666 rs605066 CITED2 Exome HDL cholesterol A/G G 0.41 0.92 20

6 161089817 rs1084651 LPA Exome HDL cholesterol G/A A 0.17 0.74 20

7 72982874 rs17145738 MLXIPL GWAS HDL cholesterol C/T C 0.13 1.00 20

7 130433384 rs4731702 KLF14 Exome HDL cholesterol G/A G 0.45 0.16 20

8 9183358 rs9987289 PPP1R3B Exome HDL cholesterol G/A A 0.11 0.63 20

8 19844222 rs12678919 LPL Exome HDL cholesterol A/G A 0.09 0.41 20

8 126490972 rs2954029 TRIB1 Imputed HDL cholesterol A/T A 0.46 0.30 20

9 15289578 rs471364 TTC39B GWAS HDL cholesterol T/C C 0.10 0.48 22

9 107664301 rs1883025 ABCA1 Exome HDL cholesterol G/A A 0.26 1.00 20

11 46743247 rs3136441 F2 Exome HDL cholesterol A/G A 0.13 0.54 20

11 47286290 rs7120118 NR1H3 Exome HDL cholesterol A/G A 0.30 0.50 23

11 48518893 rs7395662 OR4A47 Exome HDL cholesterol G/A G 0.37 0.01 21

11 61569830 rs174546 FADS1 Exome HDL cholesterol G/A A 0.33 0.83 20

215

11 116648917 rs964184 APOA5 Exome HDL cholesterol G/C C 0.16 0.73 20

11 122522375 rs7941030 UBASH3B Exome HDL cholesterol A/G A 0.38 0.24 20

12 20473758 rs7134375 PDE3A Exome HDL cholesterol C/A C 0.43 0.39 20

12 57792580 rs11613352 LRP1 Imputed HDL cholesterol C/T C 0.25 0.12 20

12 109895168 rs2338104 KCTD10 GWAS HDL cholesterol G/C C 0.48 0.52 22

12 124460167 rs4765127 ZNF664 Exome HDL cholesterol C/A C 0.33 0.14 20

15 58683366 rs1532085 LIPC Exome HDL cholesterol G/A G 0.36 0.03 20

15 63396867 rs2652834 LACTB Exome HDL cholesterol G/A A 0.19 0.18 20

16 56993324 rs3764261 CETP Exome HDL cholesterol C/A C 0.32 1.00 20

16 67902070 rs2271293 LCAT GWAS HDL cholesterol G/A G 0.11 0.64 22

16 81534790 rs2925979 CMIP Exome HDL cholesterol G/A A 0.32 0.75 20

17 37813856 rs11869286 STARD3 Exome HDL cholesterol G/C C 0.36 0.42 20

17 66875294 rs4148008 ABCA8 Imputed HDL cholesterol C/G G 0.31 0.83 20

17 76403984 rs4129767 PGS1 Exome HDL cholesterol G/A G 0.49 0.09 20

18 47167214 rs4939883 LIPG GWAS HDL cholesterol C/T T 0.18 0.53 22

18 57849023 rs12967135 MC4R Imputed HDL cholesterol G/A A 0.23 0.90 20

19 8469738 rs2967605 ANGPTL4 Exome HDL cholesterol G/A A 0.19 0.28 22

19 45422946 rs4420638 APOC1 GWAS HDL cholesterol T/C C 0.17 0.74 20

20 43042364 rs1800961 HNF4A Exome HDL cholesterol G/A A 0.03 1.00 20

216

20 44576502 rs7679 PLTP GWAS HDL cholesterol T/C C 0.22 0.67 22

1 25775733 rs12027135 TMEM57 Exome LDL cholesterol A/T A 0.45 0.93 20

1 55496039 rs11206510 PCSK9 Exome LDL cholesterol A/G A 0.22 1.00 22

1 55504650 rs2479409 PCSK9 Exome LDL cholesterol A/G G 0.35 0.18 20

1 63025942 rs2131925 DOCK7 Exome LDL cholesterol A/C A 0.36 1.00 20

1 109817590 rs12740374 CELSR2 Exome LDL cholesterol C/A C 0.24 0.90 22

1 220973563 rs2642442 MOSC1 Genotyped LDL cholesterol T/C C 0.32 0.33 20

1 234858597 rs514230 IRF2BP2 Genotyped LDL cholesterol A/T A 0.48 0.44 20

2 21286057 rs515135 APOB Exome LDL cholesterol G/A G 0.16 0.23 22

2 44065090 rs6756629 ABCG5 Exome LDL cholesterol G/A G 0.07 0.16 21

2 44073881 rs6544713 ABCG8 Exome LDL cholesterol G/A A 0.32 0.59 22

5 74655726 rs3846663 HMGCR GWAS LDL cholesterol C/T T 0.38 1.00 22

5 156398169 rs1501908 TIMD4 GWAS LDL cholesterol C/G C 0.39 0.21 22

6 16127407 rs3757354 IDOL Exome LDL cholesterol G/A G 0.20 0.67 20

6 16197194 rs2142672 MYLIP Exome LDL cholesterol G/A G 0.26 0.19 24

6 26093141 rs1800562 HFE Exome LDL cholesterol G/A G 0.08 1.00 20

6 32412435 rs3177928 HLA Exome LDL cholesterol G/A A 0.13 0.83 20

6 33143948 rs2254287 B3GALT4 Exome LDL cholesterol C/G G 0.41 0.25 25

6 160578860 rs1564348 LPA Exome LDL cholesterol A/G G 0.16 0.49 20

217

7 21607352 rs12670798 DNAH11 Exome LDL cholesterol A/G G 0.22 0.59 20

7 44579180 rs2072183 NPC1L1 Exome LDL cholesterol C/G G 0.24 1.00 20

8 9183358 rs9987289 PPP1R3B Exome LDL cholesterol G/A G 0.11 0.63 20

8 59388565 rs2081687 CYP7A1 Exome LDL cholesterol G/A A 0.36 0.69 20

8 126490972 rs2954029 TRIB1 Imputed LDL cholesterol A/T A 0.46 0.30 20

8 145043543 rs11136341 PLEC1 Exome LDL cholesterol A/G G 0.39 0.14 20

9 136155000 rs635634 ABO Exome LDL cholesterol G/A A 0.20 0.66 20

10 113933886 rs2255141 GPAM Exome LDL cholesterol G/A A 0.29 0.73 20

11 61569830 rs174546 FADS1 Exome LDL cholesterol G/A G 0.33 0.83 20

11 116648917 rs964184 APOA5 Exome LDL cholesterol G/C C 0.16 0.73 20

11 126243952 rs11220462 ST3GAL4 Exome LDL cholesterol G/A A 0.15 0.85 20

12 112072424 rs11065987 BRAP Exome LDL cholesterol A/G A 0.44 0.22 20

12 121388962 rs2650000 HNF1A Exome LDL cholesterol C/A A 0.35 0.54 22

14 24883887 rs8017377 NYNRIN Exome LDL cholesterol G/A A 0.48 0.51 20

16 56993324 rs3764261 CETP Exome LDL cholesterol C/A C 0.32 1.00 20

16 72108093 rs2000999 HPR Exome LDL cholesterol G/A A 0.20 0.24 20

17 45425115 rs7206971 OSBPL7 GWAS LDL cholesterol G/A G 0.48 0.34 20

19 11202306 rs6511720 LDLR Exome LDL cholesterol C/A C 0.14 0.18 20

19 19407718 rs10401969 SUGP1 Exome LDL cholesterol A/G A 0.10 1.00 20

218

19 45422946 rs4420638 APOC1 GWAS LDL cholesterol T/C C 0.17 0.74 20

20 39091487 rs2902940 MAFB Imputed LDL cholesterol A/G A 0.33 0.23 20

20 39228784 rs6102059 MAFB Imputed LDL cholesterol C/T C 0.31 0.91 22

20 39672618 rs6029526 TOP1 Exome LDL cholesterol T/A A 0.47 0.45 20

4 111718067 rs6843082 PITX2 Exome Stroke A/G G 0.21 0.26 26

6 44594159 rs556621 CDC5L Genotyped Stroke C/A A 0.32 1.00 27

7 19049388 rs2107595 HDAC9 Genotyped Stroke C/T T 0.16 1.00 26

12 783484 rs12425791 NINJ2 Exome Stroke G/A A 0.19 0.77 28

16 73068678 rs879324 ZFHX3 GWAS Stroke G/A A 0.18 0.54 26

1 120517959 rs10923931 NOTCH2 GWAS Type 2 Diabetes G/T T 0.10 0.29 29

1 212221342 rs2075423 PROX1 Genotyped Type 2 Diabetes C/A C 0.33 1.00 30

2 43732823 rs7578597 THADA Exome Type 2 Diabetes A/G A 0.08 0.34 29

2 60584819 rs243021 BCL11A Exome Type 2 Diabetes G/A A 0.47 0.23 31

2 161171454 rs7593730 RBMS1 Exome Type 2 Diabetes G/A G 0.21 0.06 32

2 165528876 rs13389219 GRB14 Exome Type 2 Diabetes G/A G 0.38 0.55 30

2 227020653 rs7578326 IRS1 Exome Type 2 Diabetes A/G A 0.34 0.36 30

3 12393125 rs1801282 PPARG GWAS Type 2 Diabetes C/G C 0.10 0.00 30

3 23454790 rs1496653 UBE2E2 GWAS Type 2 Diabetes A/G A 0.19 0.55 30

3 64711904 rs4607103 ADAMTS9 GWAS Type 2 Diabetes C/T C 0.24 0.33 29

219

3 123065778 rs11708067 ADCY5 Exome Type 2 Diabetes A/G A 0.24 0.61 30

3 185511687 rs4402960 IGF2BP2 GWAS Type 2 Diabetes G/T T 0.36 1.00 30

4 6270056 rs4689388 WFS1 Exome Type 2 Diabetes A/G A 0.43 0.25 33

5 55806751 rs459193 ANKRD55 Exome Type 2 Diabetes G/A G 0.25 0.22 30

5 76424949 rs4457053 ZBED3 Exome Type 2 Diabetes A/G G 0.33 0.75 30

6 20688121 rs10440833 CDKAl1 Exome Type 2 Diabetes T/A A 0.29 0.43 30

7 14864807 rs17168486 DGKB Genotyped Type 2 Diabetes C/T T 0.20 0.88 30

7 28180556 rs864745 JAZF1 GWAS Type 2 Diabetes C/T T 0.49 1.00 29

8 41519248 rs516946 ANK1 Exome Type 2 Diabetes G/A G 0.22 0.22 30

8 95960511 rs896854 TP53INP1 Exome Type 2 Diabetes G/A A 0.47 0.78 31

9 22134094 rs10811661 CDKN2B-AS1 GWAS Type 2 Diabetes T/C T 0.15 1.00 30

9 22137685 rs7018475 CDKN2B Genotyped Type 2 Diabetes T/G T 0.28 0.04 34

9 81952128 rs13292136 CHCHD9 Exome Type 2 Diabetes G/A G 0.07 0.73 31

9 84308948 rs2796441 TLE1 Exome Type 2 Diabetes G/A G 0.39 1.00 30

10 12328010 rs12779790 CDC123 Genotyped Type 2 Diabetes A/G G 0.18 0.73 29

10 80942631 rs12571751 ZMIZ1 Exome Type 2 Diabetes A/G A 0.46 0.30 30

10 94465559 rs5015480 HHEX GWAS Type 2 Diabetes C/T C 0.39 0.56 30

10 114758349 rs7903146 TCF7L2 Exome Type 2 Diabetes G/A A 0.36 0.84 30

11 2691471 rs231362 KCNQ1 Genotyped Type 2 Diabetes C/T C 0.45 0.17 31

220

11 2803645 rs163184 KCNQ1 Genotyped Type 2 Diabetes T/G G 0.49 1.00 30

11 17408630 rs5215 KCNJ11 GWAS Type 2 Diabetes T/C C 0.38 0.01 30

11 72433098 rs1552224 ARAP1 Exome Type 2 Diabetes A/C A 0.14 0.43 30

11 92673828 rs1387153 MTNR1B Exome Type 2 Diabetes G/A A 0.30 0.91 31

12 27956150 rs10842994 KLHDC5 GWAS Type 2 Diabetes C/T C 0.17 0.49 30

12 66174894 rs1531343 HMGA2 GWAS Type 2 Diabetes G/C C 0.09 0.76 31

12 71663102 rs7961581 TSPAN8 GWAS Type 2 Diabetes T/C C 0.32 0.58 29

12 121460686 rs7957197 HNF1A Exome Type 2 Diabetes T/A T 0.19 0.55 31

13 80717156 rs1359790 SPRY2 GWAS Type 2 Diabetes G/A G 0.28 0.91 30

15 77832762 rs7177055 HMG20A Exome Type 2 Diabetes A/G A 0.25 0.32 30

15 80432222 rs11634397 ZFAND6 Exome Type 2 Diabetes G/A G 0.33 0.35 31

15 91521337 rs8042680 PRC1 Exome Type 2 Diabetes C/A A 0.30 0.38 31

16 53816275 rs8050136 FTO GWAS Type 2 Diabetes C/A A 0.43 0.51 29

16 73804746 rs7202877 CTRB2 Genotyped Type 2 Diabetes T/G T 0.07 0.76 30

18 7068462 rs8090011 LAMA1 Genotyped Type 2 Diabetes C/G G 0.38 0.06 35

18 57884750 rs12970134 MC4R Exome Type 2 Diabetes G/A A 0.27 0.55 30

19 19407718 rs10401969 CILP2 Exome Type 2 Diabetes A/G G 0.10 1.00 30

23 152899922 rs5945326 DUSP9 Genotyped Type 2 Diabetes A/G A 0.21 0.27 31

1 25768937 rs10903129 TMEM57 Exome Total Cholesterol G/A G 0.44 1.00 21

221

1 55504650 rs2479409 PCSK9 Exome Total Cholesterol A/G G 0.35 0.18 20

1 63118196 rs10889353 DOCK7 GWAS Total Cholesterol A/C A 0.34 0.84 21

1 93009438 rs7515577 EVI5 Exome Total Cholesterol A/C A 0.22 1.00 20

1 109818530 rs646776 CELSR2 Exome Total Cholesterol A/G A 0.24 0.80 21

1 220973563 rs2642442 MOSC1 Genotyped Total Cholesterol T/C C 0.32 0.33 20

1 234858597 rs514230 IRF2BP2 Genotyped Total Cholesterol A/T A 0.48 0.44 20

2 21232195 rs693 APOB GWAS Total Cholesterol G/A A 0.49 0.71 21

2 27730940 rs1260326 GCKR Exome Total Cholesterol G/A A 0.39 0.24 20

2 44065090 rs6756629 ABCG5 Exome Total Cholesterol G/A G 0.07 0.16 21

2 135837906 rs7570971 RAB3GAP1 Exome Total Cholesterol C/A A 0.32 0.24 20

3 12628920 rs2290159 RAF1 Exome Total Cholesterol G/C G 0.19 0.55 20

5 74651084 rs3846662 HMGCR Exome Total Cholesterol A/G G 0.42 0.85 21

5 156390297 rs6882076 TIMD4 Exome Total Cholesterol G/A G 0.39 0.14 20

6 16127407 rs3757354 IDOL Exome Total Cholesterol G/A G 0.20 0.67 20

6 26093141 rs1800562 HFE Exome Total Cholesterol G/A G 0.08 1.00 20

6 32412435 rs3177928 HLA Exome Total Cholesterol G/A A 0.13 0.83 20

6 34546560 rs2814982 C6orf106 Exome Total Cholesterol G/A G 0.10 0.42 20

6 160578860 rs1564348 LPA Exome Total Cholesterol A/G G 0.16 0.49 20

7 21607352 rs12670798 DNAH11 Exome Total Cholesterol A/G G 0.22 0.59 20

222

7 44579180 rs2072183 NPC1L1 Exome Total Cholesterol C/G G 0.24 1.00 20

8 9183358 rs9987289 PPP1R3B Exome Total Cholesterol G/A G 0.11 0.63 20

8 18272881 rs1495741 NAT2 Exome Total Cholesterol A/G G 0.21 0.67 20

8 59388565 rs2081687 CYP7A1 Exome Total Cholesterol G/A A 0.36 0.69 20

8 116648565 rs2737229 TRPS1 Exome Total Cholesterol A/C A 0.32 0.60 20

8 126490972 rs2954029 TRIB1 Imputed Total Cholesterol A/T A 0.46 0.30 20

8 145043543 rs11136341 PLEC1 Exome Total Cholesterol A/G G 0.39 0.14 20

9 107664301 rs1883025 ABCA1 Exome Total Cholesterol G/A G 0.26 1.00 20

9 136155000 rs635634 ABO Exome Total Cholesterol G/A A 0.20 0.66 20

10 113933886 rs2255141 GPAM Exome Total Cholesterol G/A A 0.29 0.73 20

11 18632984 rs10128711 SPTY2D1 Exome Total Cholesterol G/A G 0.24 0.71 20

11 61597212 rs174570 FADS2 Exome Total Cholesterol G/A G 0.13 1.00 21

11 116648917 rs964184 APOA5 Exome Total Cholesterol G/C C 0.16 0.73 20

11 122522375 rs7941030 UBASH3B Exome Total Cholesterol A/G G 0.38 0.24 20

11 126243952 rs11220462 ST3GAL4 Exome Total Cholesterol G/A A 0.15 0.85 20

12 112072424 rs11065987 BRAP Exome Total Cholesterol A/G A 0.44 0.22 20

16 56993324 rs3764261 CETP Exome Total Cholesterol C/A A 0.32 1.00 20

16 72108093 rs2000999 HPR Exome Total Cholesterol G/A A 0.20 0.24 20

17 45425115 rs7206971 OSBPL7 GWAS Total Cholesterol G/A A 0.48 0.34 20

223

19 11210912 rs2228671 LDLR Exome Total Cholesterol G/A G 0.14 0.56 21

19 19789528 rs2304130 NCAN Exome Total Cholesterol A/G A 0.11 0.82 21

19 45395619 rs2075650 TOMM40 Exome Total Cholesterol A/G G 0.12 0.18 21

19 49206417 rs492602 FUT2 Exome Total Cholesterol A/G G 0.49 0.85 20

20 39091487 rs2902940 MAFB Imputed Total Cholesterol A/G A 0.33 0.23 20

20 39672618 rs6029526 TOP1 Exome Total Cholesterol T/A A 0.47 0.45 20

20 43042364 rs1800961 HNF4A Exome Total Cholesterol G/A G 0.03 1.00 20

1 63025942 rs2131925 DOCK7 Exome Triglycerides A/C A 0.36 1.00 20

1 230295691 rs4846914 GALNT2 Exome Triglycerides A/G G 0.40 0.29 20

2 21206183 rs6754295 APOB Exome Triglycerides A/C A 0.25 0.80 21

2 27730940 rs1260326 GCKR Exome Triglycerides G/A A 0.39 0.24 20

2 165513091 rs10195252 COBLL1 Exome Triglycerides A/G A 0.40 0.92 20

2 227100698 rs2972146 IRS1 Exome Triglycerides A/C A 0.36 0.84 20

3 135926622 rs645040 MSL2L1 Exome Triglycerides A/C A 0.22 0.43 20

4 88030261 rs442177 AFF1 Exome Triglycerides A/C A 0.39 1.00 20

5 55861786 rs9686661 ANKRD55 Exome Triglycerides G/A A 0.23 0.60 20

5 156390297 rs6882076 TIMD4 Exome Triglycerides G/A A 0.39 0.14 20

6 31265490 rs2247056 HLA Exome Triglycerides G/A G 0.25 0.71 20

7 72982874 rs17145738 MLXIPL GWAS Triglycerides C/T C 0.13 1.00 20

224

8 11045161 rs7819412 XKR6 GWAS Triglycerides G/A A 0.48 0.46 22

8 18272881 rs1495741 NAT2 Exome Triglycerides A/G G 0.21 0.67 20

8 19844222 rs12678919 LPL Exome Triglycerides A/G A 0.09 0.41 20

8 126486409 rs17321515 TRIB1 GWAS Triglycerides A/G A 0.47 0.71 25

10 65027610 rs10761731 JMJD1C Exome Triglycerides A/T A 0.41 0.85 20

10 94839642 rs2068888 CYP26A1 Exome Triglycerides G/A G 0.44 0.78 20

11 61569830 rs174546 FADS1 Exome Triglycerides G/A A 0.33 0.83 20

11 116652207 rs12286037 ZNF259 GWAS Triglycerides C/T T 0.08 1.00 25

11 116732512 rs2075292 SIK3 Exome Triglycerides A/C C 0.11 0.09 36

12 57792580 rs11613352 LRP1 Imputed Triglycerides C/T C 0.25 0.12 20

12 124460167 rs4765127 ZNF664 Exome Triglycerides C/A C 0.33 0.14 20

15 42683787 rs2412710 CAPN3 Exome Triglycerides G/A A 0.02 1.00 20

15 44245931 rs2929282 FRMD5 Exome Triglycerides A/T T 0.05 1.00 20

15 58683366 rs1532085 LIPC Exome Triglycerides G/A G 0.36 0.03 20

16 30918487 rs11649653 CTF1 GWAS Triglycerides C/G C 0.36 0.36 20

16 56993324 rs3764261 CETP Exome Triglycerides C/A C 0.32 1.00 20

19 19407718 rs10401969 SUGP1 Exome Triglycerides A/G A 0.10 1.00 20

19 45414451 rs439401 APOE Exome Triglycerides G/A G 0.37 0.84 20

20 44576502 rs7679 PLTP GWAS Triglycerides T/C C 0.22 0.67 22

225

22 38546033 rs5756931 PLA2G6 Exome Triglycerides A/G A 0.39 0.84 20

References

1. Wain, L., et al., Genome-wide association study identifies six new loci influencing pulse pressure and mean arterial pressure. Nature genetics, 2011. 43(10): p. 1005-1011.

2. Speliotes, E., et al., Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nature genetics, 2010. 42(11): p. 937-948.

3. O'Donnell, C., et al., Genome-wide association study for coronary artery calcification with follow-up in myocardial infarction. Circulation, 2011. 124(25): p. 2855-2864.

4. Schunkert, H., et al., Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Nat Genet, 2011. 43(4): p. 333-8.

5. Erdmann, J., et al., New susceptibility locus for coronary artery disease on chromosome 3q22.3. Nat Genet, 2009. 41(3): p. 280-2.

6. Davies, R.W., et al., A genome-wide association study for coronary artery disease identifies a novel susceptibility locus in the major histocompatibility complex. Circ Cardiovasc Genet, 2012. 5(2): p. 217-25.

7. Samani, N.J., et al., Genomewide association analysis of coronary artery disease. N Engl J Med, 2007. 357(5): p. 443-53.

8. Coronary Artery Disease Genetics, C., A genome-wide association study in Europeans and South Asians identifies five new loci for coronary artery disease. Nature genetics, 2011. 43(4): p. 339-344.

9. Wild, P.S., et al., A genome-wide association study identifies LIPA as a susceptibility gene for coronary artery disease. Circ Cardiovasc Genet, 2011. 4(4): p. 403-12.

10. Dehghan, A., et al., Meta-analysis of genome-wide association studies in >80 000 subjects identifies multiple loci for C-reactive protein levels. Circulation, 2011. 123(7): p. 731-8.

11. Newton-Cheh, C., et al., Common variants at ten loci influence QT interval duration in the QTGEN Study. Nature genetics, 2009. 41(4): p. 399-406.

226

12. Holm, H., et al., Several common variants modulate heart rate, PR interval and QRS duration. Nat Genet, 2010. 42(2): p. 117-22.

13. Pfeufer, A., et al., Common variants at ten loci modulate the QT interval duration in the QTSCD Study. Nat Genet, 2009. 41(4): p. 407-14.

14. Ritchie, M.D., et al., Genome- and phenome-wide analyses of cardiac conduction identifies markers of arrhythmia risk. Circulation, 2013. 127(13): p. 1377-85.

15. Marroni, F., et al., A genome-wide association scan of RR and QT interval duration in 3 European genetically isolated populations: the EUROSPAN project. Circ Cardiovasc Genet, 2009. 2(4): p. 322-8.

16. Dupuis, J., et al., New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nature genetics, 2010. 42(2): p. 105-116.

17. Köttgen, A., et al., New loci associated with kidney function and chronic kidney disease. Nature genetics, 2010. 42(5): p. 376-384.

18. Tin, A., et al., Using multiple measures for quantitative trait association analyses: application to estimated glomerular filtration rate. J Hum Genet, 2013. 58(7): p. 461-6.

19. Pattaro, C., et al., Genome-wide association and functional follow-up reveals new loci for kidney function. PLoS Genet, 2012. 8(3): p. e1002584.

20. Teslovich, T., et al., Biological, clinical and population relevance of 95 loci for blood lipids. Nature, 2010. 466(7307): p. 707-713.

21. Aulchenko, Y.S., et al., Loci influencing lipid levels and coronary heart disease risk in 16 European population cohorts. Nat Genet, 2009. 41(1): p. 47-55.

22. Kathiresan, S., et al., Common variants at 30 loci contribute to polygenic dyslipidemia. Nature genetics, 2009. 41(1): p. 56-65.

23. Sabatti, C., et al., Genome-wide association analysis of metabolic traits in a birth cohort from a founder population. Nat Genet, 2009. 41(1): p. 35-46.

24. Waterworth, D.M., et al., Genetic variants influencing circulating lipid levels and risk of coronary artery disease. Arterioscler Thromb Vasc Biol, 2010. 30(11): p. 2264-76.

25. Willer, C.J., et al., Newly identified loci that influence lipid concentrations and risk of coronary artery disease. Nat Genet, 2008. 40(2): p. 161- 9.

227

26. Traylor, M., et al., Genetic risk factors for ischaemic stroke and its subtypes (the METASTROKE collaboration): a meta-analysis of genome- wide association studies. Lancet Neurol, 2012. 11(11): p. 951-62.

27. Holliday, E.G., et al., Common variants at 6p21.1 are associated with large artery atherosclerotic stroke. Nat Genet, 2012. 44(10): p. 1147-51.

28. Ikram, M.A., et al., Genomewide association studies of stroke. N Engl J Med, 2009. 360(17): p. 1718-28.

29. Zeggini, E., et al., Meta-analysis of genome-wide association data and large-scale replication identifies additional susceptibility loci for type 2 diabetes. Nature genetics, 2008. 40(5): p. 638-645.

30. Morris, A., et al., Large-scale association analysis provides insights into the genetic architecture and pathophysiology of type 2 diabetes. Nature genetics, 2012. 44(9): p. 981-990.

31. Voight, B.F., et al., Twelve type 2 diabetes susceptibility loci identified through large-scale association analysis. Nat Genet, 2010. 42(7): p. 579-89.

32. Qi, L., et al., Genetic variants at 2q24 are associated with susceptibility to type 2 diabetes. Hum Mol Genet, 2010. 19(13): p. 2706-15.

33. Rung, J., et al., Genetic variant near IRS1 is associated with type 2 diabetes, insulin resistance and hyperinsulinemia. Nat Genet, 2009. 41(10): p. 1110-5.

34. Huang, J., et al., 1000 Genomes-based imputation identifies novel and refined associations for the Wellcome Trust Case Control Consortium phase 1 Data. Eur J Hum Genet, 2012. 20(7): p. 801-5.

35. Perry, J., et al., Stratifying type 2 diabetes cases by BMI identifies genetic risk variants in LAMA1 and enrichment for risk variants in lean compared to obese cases. PLoS Genetics, 2012. 8(5).

36. Kooner, J.S., et al., Genome-wide scan identifies variation in MLXIPL associated with plasma triglycerides. Nat Genet, 2008. 40(2): p. 149- 51.

228

Supplementary Table 2: Single SNP association results for SNPs included in genetic risk scores. Association analyses were adjusted for age and sex. All associations are reported for the risk allele. Abbreviations: CAC, coronary artery calcification; CAD, coronary artery disease; CRP, C- reactive protein; eGFR, estimated glomerular filtration rate; MI, myocardial infarction; OR, odds ratio. For trait of interest analyses, beta values are shown for continuous traits and odds ratios are shown for dichotomous traits.

Self-reported Coronary All-Cause CVD Self-reported Trait of interest History of Artery Mortality Mortality History of MI CVD Calcification Alleles Risk Trait of interest in beta p- p- p- p- p- beta SNP Risk Score p-value OR OR OR OR (M/m) Allele DHS value/OR value value value value value value Blood pressure rs17367504 A/G A Pulse pressure 0.763 0.285 0.426 0.94 0.992 1.00 0.418 0.93 0.158 0.89 0.188 -0.181 traits Blood pressure rs13082711 T/C C Pulse pressure 0.063 -1.625 0.689 0.97 0.801 1.02 0.350 0.93 0.852 1.01 0.747 0.041 traits Blood pressure rs319690 T/C T Pulse pressure 0.916 0.084 0.749 0.98 0.811 0.98 0.097 0.88 0.555 0.96 0.724 0.041 traits Blood pressure rs419076 T/C T Pulse pressure 0.508 -0.470 0.102 1.11 0.311 1.08 0.505 0.96 0.079 1.11 0.325 -0.103 traits Blood pressure rs871606 T/C T Pulse pressure 0.047 2.630 0.764 1.04 0.772 1.04 0.525 1.08 0.983 1.00 0.399 -0.161 traits Blood pressure rs1458038 G/A A Pulse pressure 0.565 -0.468 0.631 0.97 0.865 1.01 0.668 1.03 0.830 0.99 0.615 -0.061 traits Blood pressure rs13107325 G/A G Pulse pressure 0.674 0.585 0.149 0.83 0.762 0.96 0.733 0.96 0.860 1.02 0.939 0.015 traits Blood pressure rs1173771 C/T C Pulse pressure 0.262 0.844 0.749 0.98 0.631 0.96 0.426 1.06 0.359 0.94 0.228 -0.134 traits Blood pressure rs9313772 C/T C Pulse pressure 0.268 0.826 0.257 0.93 0.610 0.96 0.496 0.95 0.690 0.98 0.499 0.074 traits Blood pressure rs198846 G/A A Pulse pressure 0.260 -1.042 0.885 0.99 0.684 0.96 0.031 1.22 0.073 1.15 0.935 -0.011 traits Blood pressure rs17477177 T/C C Pulse pressure 0.294 0.937 0.491 1.06 0.377 0.92 0.911 1.01 0.487 0.95 0.857 0.024 traits Blood pressure rs2071518 C/T T Pulse pressure 0.223 0.980 0.581 0.96 0.889 1.01 0.991 1.00 0.742 1.02 0.345 -0.110 traits Blood pressure rs12258967 C/G C Pulse pressure 0.312 0.797 0.975 1.00 0.918 0.99 0.291 1.08 0.776 0.98 0.628 -0.057 traits

229

Blood pressure rs4590817 G/C G Pulse pressure 0.563 0.577 0.154 0.88 0.438 1.08 0.982 1.00 0.526 0.95 0.346 -0.137 traits Blood pressure rs9663362 G/C C Pulse pressure 0.319 0.723 0.253 0.93 0.671 1.03 0.299 0.93 0.725 1.02 0.260 0.120 traits Blood pressure rs11191548 A/G A Pulse pressure 0.708 -0.491 0.156 0.84 0.788 0.96 0.703 1.05 0.841 1.02 0.395 0.162 traits Blood pressure rs381815 C/T T Pulse pressure 0.051 -1.584 0.070 0.88 0.512 0.95 0.567 1.04 0.174 1.10 0.295 -0.125 traits Blood pressure rs11222084 A/T T Pulse pressure 0.685 0.315 0.189 1.09 0.703 0.97 0.137 1.11 0.584 0.97 0.277 0.125 traits Blood pressure rs17249754 G/A G Pulse pressure 0.144 -1.387 0.849 0.98 0.762 1.03 0.538 1.06 0.041 1.17 0.368 -0.125 traits Blood pressure rs653178 A/G G Pulse pressure 0.365 0.632 0.447 1.05 0.570 0.96 0.095 0.90 0.167 0.92 0.062 -0.192 traits Blood pressure rs1378942 A/C C Pulse pressure 0.075 1.417 0.959 1.00 0.741 0.97 0.107 1.13 0.292 1.07 0.056 -0.224 traits Blood pressure rs17608766 A/G G Pulse pressure 0.406 0.819 0.769 0.98 0.997 1.00 0.311 0.91 0.844 0.98 0.276 0.154 traits Blood pressure rs1327235 A/G G Pulse pressure 0.505 0.473 0.757 1.02 0.566 0.96 0.311 1.07 0.523 1.04 0.879 0.016 traits Blood pressure rs6015450 A/G G Pulse pressure 0.310 1.173 0.001 1.44 0.001 1.57 0.868 1.02 0.729 1.03 0.479 0.121 traits rs2815752 BMI A/G A BMI 0.555 0.015 0.309 0.93 0.172 0.90 0.373 0.94 0.033 0.88 0.335 0.104 rs1514175 BMI G/A A BMI 0.998 -0.0001 0.571 0.96 0.970 1.00 0.824 0.99 0.958 1.00 0.550 -0.063 rs1555543 BMI C/A C BMI 0.343 0.025 0.283 0.93 0.046 0.85 0.041 0.86 0.413 1.05 0.583 -0.060 rs543874 BMI A/G G BMI 0.139 0.047 0.084 0.87 0.150 0.88 0.442 0.94 0.821 0.98 0.728 0.046 rs2867125 BMI G/A G BMI 0.287 -0.039 0.580 0.95 0.867 1.02 0.783 1.03 0.073 1.17 0.648 -0.069 rs713586 BMI A/G G BMI 0.076 0.045 0.329 1.06 0.807 0.98 0.911 1.01 0.261 0.93 0.605 0.054 rs887912 BMI G/A A BMI 0.898 0.004 0.049 1.15 0.720 1.03 0.358 0.93 0.220 0.92 0.169 0.157 rs2890652 BMI T/C C BMI 0.762 -0.011 0.594 0.95 0.577 0.94 0.677 0.96 0.513 0.95 0.790 -0.038

230 rs13078807 BMI A/G G BMI 0.008 0.082 0.861 0.99 0.608 0.95 0.254 0.91 0.189 0.91 0.693 0.050 rs9816226 BMI A/T A BMI 0.035 -0.070 0.133 0.88 0.194 0.88 0.063 1.18 0.252 1.09 0.076 0.247 rs10938397 BMI A/G G BMI 0.716 0.010 0.933 0.99 0.962 1.00 0.633 1.03 0.398 1.05 0.889 0.015 rs13107325 BMI G/A A BMI 0.752 -0.016 0.149 1.20 0.762 1.05 0.733 1.05 0.860 0.98 0.939 -0.015 rs2112347 BMI A/C A BMI 0.968 -0.001 0.111 0.90 0.784 1.02 0.430 0.95 0.950 1.00 0.896 0.014 rs206936 BMI A/G G BMI 0.057 0.059 0.004 0.80 0.297 0.91 0.718 0.97 0.548 1.04 0.506 0.084 rs987237 BMI A/G G BMI 0.368 -0.031 0.621 1.04 0.802 0.97 0.723 0.97 0.820 0.98 0.732 -0.049 rs10968576 BMI A/G G BMI 0.311 0.028 0.360 1.07 0.580 1.05 0.770 1.02 0.170 1.09 0.993 0.001 rs4929949 BMI G/A G BMI 0.461 0.019 0.039 1.14 0.083 1.14 0.937 0.99 0.856 0.99 0.552 -0.063 rs10767664 BMI A/T A BMI 0.081 0.054 0.066 1.15 0.362 1.08 0.923 1.01 0.434 0.94 0.755 -0.039 rs3817334 BMI G/A A BMI 0.059 0.050 0.322 0.94 0.199 0.91 0.920 0.99 0.256 0.93 0.557 0.065 rs7138803 BMI G/A A BMI 0.150 0.037 0.575 1.04 0.497 0.95 0.222 1.09 0.734 0.98 0.348 0.100 rs4771122 BMI A/G G BMI 0.935 -0.002 0.234 0.92 0.264 0.91 0.916 0.99 0.815 0.98 0.570 -0.069 rs11847697 BMI G/A A BMI 0.262 0.069 0.969 0.99 0.793 1.05 0.896 0.98 0.566 1.09 0.895 0.033 rs10150332 BMI T/C C BMI 0.696 -0.012 0.972 1.00 0.677 1.04 0.860 0.99 0.837 1.01 0.879 0.019 rs2241423 BMI G/A G BMI 0.162 0.042 0.704 1.03 0.656 0.96 0.590 1.04 0.846 0.99 0.961 0.006 rs12444979 BMI G/A G BMI 0.505 -0.023 0.270 0.91 0.419 1.08 0.518 1.06 0.654 1.04 0.900 0.017 rs7359397 BMI G/A A BMI 0.056 0.051 0.230 0.92 0.686 0.97 0.044 0.87 0.568 0.96 0.220 0.135 rs1558902 BMI A/T T BMI 0.103 0.041 0.515 0.96 0.739 0.98 0.921 0.99 0.626 0.97 0.586 0.057 rs571312 BMI C/A A BMI 0.148 0.044 0.577 0.96 0.776 1.03 0.898 1.01 0.799 1.02 0.921 -0.012 rs29941 BMI G/A G BMI 0.647 -0.013 0.527 1.05 0.315 1.09 0.167 1.11 0.088 1.12 0.245 -0.134 rs2287019 BMI G/A G BMI 0.234 0.040 0.247 1.10 0.407 1.08 0.684 1.04 0.979 1.00 0.159 0.196 rs3810291 BMI A/G A BMI 0.713 -0.010 0.046 0.87 0.472 1.06 0.246 0.92 0.236 0.93 0.288 0.120 rs9349379 CAC A/G G CAC 0.067 0.202 0.172 0.91 0.285 0.92 0.882 0.99 0.329 0.94 0.067 0.202 rs1333049 CAC C/G G CAC 0.150 0.154 0.935 1.01 0.780 1.02 0.143 0.90 0.058 0.89 0.150 0.154 Self-reported rs11206510 CAD A/G A 0.964 1.00 0.980 1.00 0.205 0.89 0.964 1.00 0.952 1.00 0.410 -0.105 history of MI Self-reported rs17114036 CAD A/G A 0.010 0.72 0.474 0.92 0.729 0.95 0.010 0.72 0.009 0.75 0.516 0.124 history of MI

231

Self-reported rs17465637 CAD C/A C 0.089 0.88 0.133 0.90 0.287 0.92 0.089 0.88 0.518 0.96 0.305 0.120 history of MI Self-reported rs599839 CAD A/G A 0.700 0.97 0.651 1.04 0.357 1.09 0.700 0.97 0.455 0.95 0.042 0.259 history of MI Self-reported rs6725887 CAD T/C C 0.084 0.85 0.062 1.19 0.253 1.13 0.084 0.85 0.796 0.98 0.350 -0.141 history of MI Self-reported rs9818870 CAD C/T T 0.819 1.02 0.853 1.02 0.287 1.12 0.819 1.02 0.890 1.01 0.735 -0.048 history of MI Self-reported rs12526453 CAD C/G C 0.431 0.94 0.909 1.01 0.316 0.92 0.431 0.94 0.160 0.91 0.392 0.097 history of MI Self-reported rs3869109 CAD G/A G 0.924 1.01 0.684 1.03 0.565 1.04 0.924 1.01 0.087 1.11 0.829 0.023 history of MI Self-reported rs17609940 CAD G/C G 0.808 1.02 0.274 0.92 0.030 0.82 0.808 1.02 0.835 1.02 0.274 0.140 history of MI Self-reported rs6922269 CAD G/A A 0.768 1.02 0.058 1.15 0.125 1.14 0.768 1.02 0.982 1.00 0.091 -0.201 history of MI Self-reported rs3798220 CAD A/G G 0.119 0.69 0.763 1.07 0.500 1.22 0.119 0.69 0.103 0.69 0.260 0.453 history of MI Self-reported rs10953541 CAD G/A G 0.814 0.98 0.912 0.99 0.740 1.03 0.814 0.98 0.748 0.98 0.594 0.066 history of MI Self-reported rs11556924 CAD C/T C 0.237 0.92 0.887 0.99 0.761 1.02 0.237 0.92 0.431 0.95 0.125 0.164 history of MI Self-reported rs7865618 CAD A/G A 0.951 1.00 0.142 0.91 0.810 0.98 0.951 1.00 0.159 0.92 0.170 0.144 history of MI Self-reported rs579459 CAD A/G G 0.102 0.87 0.370 0.93 0.255 0.90 0.102 0.87 0.408 0.94 0.229 0.162 history of MI Self-reported rs2505083 CAD A/G G 0.647 1.03 0.334 0.94 0.233 0.91 0.647 1.03 0.214 1.08 0.205 0.135 history of MI Self-reported rs1746048 CAD C/T C 0.810 0.98 0.051 0.82 0.045 0.78 0.810 0.98 0.530 1.06 0.860 0.028 history of MI Self-reported rs1412444 CAD G/A A 0.831 0.98 0.358 1.06 0.408 1.07 0.831 0.98 0.769 0.98 0.971 -0.004 history of MI Self-reported rs12413409 CAD G/A G 0.771 1.04 0.078 0.81 0.522 0.91 0.771 1.04 0.619 1.05 0.426 0.150 history of MI Self-reported rs974819 CAD G/A A 0.317 1.08 0.252 0.92 0.615 0.96 0.317 1.08 0.057 1.14 0.823 -0.026 history of MI Self-reported rs964184 CAD G/C C 0.799 1.02 0.747 0.97 0.375 0.91 0.799 1.02 0.503 1.06 0.835 0.031 history of MI Self-reported rs4773144 CAD A/G G 0.089 0.89 0.007 0.84 0.004 0.81 0.089 0.89 0.021 0.87 0.941 0.008 history of MI Self-reported rs2895811 CAD A/G G 0.163 0.91 0.652 0.97 0.653 0.97 0.163 0.91 0.609 1.03 0.011 0.276 history of MI Self-reported rs3825807 CAD A/G A 0.147 0.91 0.390 1.05 0.808 1.02 0.147 0.91 0.270 0.94 0.076 0.183 history of MI rs216172 CAD G/C C Self-reported 0.203 0.92 0.563 1.04 0.770 1.02 0.203 0.92 0.271 1.07 0.545 0.064

232

history of MI Self-reported rs12936587 CAD G/A G 0.195 0.92 0.537 1.04 0.362 0.93 0.195 0.92 0.495 1.04 0.207 0.133 history of MI Self-reported rs46522 CAD A/G A 0.591 0.96 0.721 0.98 0.640 0.97 0.591 0.96 0.199 0.93 0.465 0.078 history of MI Self-reported rs1122608 CAD C/A C 0.294 0.92 0.191 0.91 0.957 1.00 0.294 0.92 0.468 0.95 0.568 0.066 history of MI rs12037222 CRP G/A A CRP 0.149 0.113 0.311 1.08 0.104 1.16 0.787 1.02 0.342 1.07 0.707 0.048 rs4420065 CRP G/A G CRP 0.078 0.113 0.762 0.98 0.993 1.00 0.147 0.91 0.081 0.90 0.422 0.084 rs4129267 CRP G/A G CRP 0.384 0.056 0.962 1.00 0.254 0.92 0.292 1.07 0.865 1.01 0.701 -0.041 rs2794520 CRP G/A G CRP 0.002 0.207 0.251 0.93 0.935 0.99 0.858 0.99 0.102 0.90 0.202 0.142 rs12239046 CRP G/A G CRP 0.027 0.142 0.619 0.97 0.377 1.07 0.248 0.92 0.750 1.02 0.593 -0.057 rs1260326 CRP G/A A CRP 0.520 0.042 0.672 0.97 0.423 1.06 0.101 1.12 0.605 1.03 0.514 0.069 rs6734238 CRP A/G G CRP 0.127 0.097 0.237 0.93 0.682 0.97 0.847 1.01 0.258 1.07 0.450 0.079 rs4705952 CRP A/G G CRP 0.920 -0.008 0.566 0.96 0.210 0.90 0.683 0.97 0.520 1.05 0.644 0.057 rs6901250 CRP G/A A CRP 0.149 -0.104 0.503 0.95 0.679 0.97 0.115 1.13 0.871 0.99 0.184 -0.155 rs13233571 CRP G/A G CRP 0.084 0.164 0.594 0.95 0.229 0.87 0.315 0.90 0.754 1.03 0.802 0.039 rs9987289 CRP G/A A CRP 0.384 0.092 0.700 0.96 0.792 1.03 0.436 0.92 0.235 0.89 0.750 -0.055 rs10745954 CRP A/G A CRP 0.166 0.091 0.444 0.95 0.176 1.11 0.603 1.04 0.509 1.04 0.550 0.064 rs1183910 CRP G/A G CRP 0.298 0.070 0.657 0.97 0.993 1.00 0.872 1.01 0.869 1.01 0.029 -0.245 rs340029 CRP A/G A CRP 0.886 0.009 0.539 0.96 0.804 0.98 0.886 0.99 0.963 1.00 0.263 0.121 rs10521222 CRP G/A G CRP 0.047 0.277 0.628 1.07 0.665 0.93 0.628 0.93 0.989 1.00 0.807 -0.058 rs2847281 CRP A/G A CRP 0.131 -0.093 0.899 1.01 0.746 1.02 0.993 1.00 0.670 1.03 0.559 -0.060 rs4420638 CRP A/G A CRP 0.005 0.247 0.176 0.89 0.145 0.86 0.121 0.86 0.131 0.88 0.280 0.155 rs1800961 CRP G/A G CRP 0.898 0.024 0.896 1.02 0.186 1.31 0.423 0.85 0.170 0.78 0.186 0.399 Electrocardiogram rs846111 G/C C PR interval 0.464 0.006 0.296 0.93 0.101 0.87 0.688 0.97 0.265 0.93 0.422 0.094 traits Electrocardiogram rs846111 G/C C QRS interval 0.913 -0.001 traits Electrocardiogram rs846111 G/C C QT interval 0.777 0.001 traits

233

Electrocardiogram rs10494366 A/C C PR interval 0.558 -0.005 0.231 0.92 0.133 0.89 0.365 0.94 0.349 0.94 0.774 -0.032 traits Electrocardiogram rs10494366 A/C C QRS interval 0.234 0.010 traits Electrocardiogram rs10494366 A/C C QT interval 0.004 0.012 traits Electrocardiogram rs4657178 G/A A PR interval 0.034 -0.020 0.915 0.99 0.637 1.04 0.351 1.08 0.094 1.13 0.145 0.180 traits Electrocardiogram rs4657178 G/A A QRS interval 0.533 -0.006 traits Electrocardiogram rs4657178 G/A A QT interval 0.277 0.005 traits Electrocardiogram rs10919071 A/G A PR interval 0.136 -0.018 0.506 0.94 0.868 0.98 0.572 0.94 0.964 1.00 0.239 0.189 traits Electrocardiogram rs10919071 A/G A QRS interval 0.661 0.005 traits Electrocardiogram rs10919071 A/G A QT interval 0.116 0.009 traits Electrocardiogram rs12053903 A/G A PR interval 0.184 -0.011 0.479 0.95 0.397 0.93 0.282 1.08 0.304 1.07 0.607 -0.058 traits Electrocardiogram rs12053903 A/G A QRS interval 0.143 -0.012 traits Electrocardiogram rs12053903 A/G A QT interval 0.753 0.001 traits Electrocardiogram rs6795970 G/A A PR interval 0.0002 0.029 0.197 1.09 0.132 1.12 0.655 1.03 0.643 0.97 0.866 0.018 traits Electrocardiogram rs6795970 G/A A QRS interval 0.065 0.014 traits Electrocardiogram rs6795970 G/A A QT interval 0.067 -0.007 traits Electrocardiogram rs7660702 A/G A PR interval 0.630 0.004 0.307 0.93 0.081 0.87 0.635 0.97 0.095 1.11 0.112 -0.180 traits Electrocardiogram rs7660702 A/G A QRS interval 0.355 -0.008 traits Electrocardiogram rs7660702 A/G A QT interval 0.307 -0.004 traits

234

Electrocardiogram rs1321311 C/A A PR interval 0.321 0.009 0.322 1.08 0.451 1.07 0.373 1.07 0.871 1.01 0.705 0.046 traits Electrocardiogram rs1321311 C/A A QRS interval 0.063 0.017 traits Electrocardiogram rs1321311 C/A A QT interval 0.855 0.001 traits Electrocardiogram rs12210810 G/C G PR interval 0.133 0.024 0.036 0.76 0.038 0.71 0.360 0.88 0.254 0.87 0.073 0.386 traits Electrocardiogram rs12210810 G/C G QRS interval 0.086 0.027 traits Electrocardiogram rs12210810 G/C G QT interval 0.013 0.020 traits Electrocardiogram rs11970286 G/A A PR interval 0.836 -0.002 0.685 0.97 0.099 0.89 0.614 1.03 0.166 1.09 0.023 0.238 traits Electrocardiogram rs11970286 G/A A QRS interval 0.049 0.015 traits Electrocardiogram rs11970286 G/A A QT interval 0.008 0.010 traits Electrocardiogram rs3807989 G/A A PR interval 0.009 0.021 0.370 1.06 0.989 1.00 0.988 1.00 0.411 0.95 0.222 0.134 traits Electrocardiogram rs3807989 G/A A QRS interval 0.454 0.006 traits Electrocardiogram rs3807989 G/A A QT interval 0.922 0.0004 traits Electrocardiogram rs3807375 G/A A PR interval 0.853 -0.001 0.884 1.01 0.867 0.99 0.886 1.01 0.477 0.96 0.154 0.152 traits Electrocardiogram rs3807375 G/A A QRS interval 0.674 0.003 traits Electrocardiogram rs3807375 G/A A QT interval 0.078 0.007 traits Electrocardiogram rs2074238 G/A G PR interval 0.588 0.009 0.544 1.08 0.389 1.14 0.858 0.98 0.220 1.16 0.504 -0.144 traits Electrocardiogram rs2074238 G/A G QRS interval 0.140 -0.023 traits Electrocardiogram rs2074238 G/A G QT interval 0.539 0.005 traits

235

Electrocardiogram rs12576239 G/A A PR interval 0.502 -0.008 0.784 1.03 0.827 1.02 0.247 1.12 0.355 1.08 0.906 -0.018 traits Electrocardiogram rs12576239 G/A A QRS interval 0.721 0.004 traits Electrocardiogram rs12576239 G/A A QT interval 0.470 0.004 traits Electrocardiogram rs3825214 A/G G PR interval 0.184 0.014 0.286 0.91 0.024 0.80 0.214 0.90 0.689 0.97 0.412 0.115 traits Electrocardiogram rs3825214 A/G G QRS interval 0.027 0.023 traits Electrocardiogram rs3825214 A/G G QT interval 0.946 -0.0004 traits Electrocardiogram rs2478333 C/A A PR interval 0.929 0.001 0.815 1.02 0.703 1.03 0.824 0.98 0.793 1.02 0.998 0.000 traits Electrocardiogram rs2478333 C/A A QRS interval 0.187 0.011 traits Electrocardiogram rs2478333 C/A A QT interval 0.549 0.002 traits Electrocardiogram rs365990 A/G G PR interval 0.647 -0.004 0.930 0.99 0.592 1.04 0.950 1.00 0.801 1.02 0.564 0.064 traits Electrocardiogram rs365990 A/G G QRS interval 0.465 -0.006 traits Electrocardiogram rs365990 A/G G QT interval 0.297 -0.004 traits Electrocardiogram rs8049607 G/A A PR interval 0.460 -0.006 0.324 0.94 0.592 0.96 0.345 1.07 0.209 1.08 0.109 -0.169 traits Electrocardiogram rs8049607 G/A A QRS interval 0.907 -0.001 traits Electrocardiogram rs8049607 G/A A QT interval 0.101 0.006 traits Electrocardiogram rs7188697 A/G A PR interval 0.269 -0.010 0.726 1.03 0.506 1.06 0.653 0.97 0.382 1.06 0.240 0.142 traits Electrocardiogram rs7188697 A/G A QRS interval 0.604 0.005 traits Electrocardiogram rs7188697 A/G A QT interval 0.512 0.003 traits

236

Electrocardiogram rs2074518 G/A G PR interval 0.419 0.006 0.158 0.92 0.181 0.91 0.603 0.97 0.787 1.02 0.890 0.014 traits Electrocardiogram rs2074518 G/A G QRS interval 0.068 0.014 traits Electrocardiogram rs2074518 G/A G QT interval 0.013 0.010 traits Electrocardiogram rs17779747 C/A C PR interval 0.029 0.018 0.217 0.92 0.519 0.95 0.512 0.95 0.422 0.95 0.968 0.005 traits Electrocardiogram rs17779747 C/A C QRS interval 0.355 0.008 traits Electrocardiogram rs17779747 C/A C QT interval 0.920 0.0004 traits Electrocardiogram rs727957 C/A A PR interval 0.348 -0.010 0.997 1.00 0.746 1.03 0.598 0.95 0.104 0.88 0.342 -0.131 traits Electrocardiogram rs727957 C/A A QRS interval 0.740 0.003 traits Electrocardiogram rs727957 C/A A QT interval 0.202 0.007 traits rs340874 Fasting glucose G/A G Fasting glucose 0.670 -0.007 0.735 0.98 0.904 0.99 0.526 1.04 0.170 1.09 0.750 -0.035 rs780094 Fasting glucose G/A G Fasting glucose 0.339 0.015 0.801 0.98 0.212 0.91 0.041 0.87 0.385 0.95 0.718 -0.039 rs560887 Fasting glucose G/A G Fasting glucose 0.280 -0.018 0.502 0.96 0.827 0.98 0.611 0.96 0.420 0.95 0.774 0.032 rs11708067 Fasting glucose A/G A Fasting glucose 0.983 0.0004 0.027 1.17 0.228 1.11 0.246 1.09 0.130 1.11 0.445 -0.092 rs11920090 Fasting glucose T/A T Fasting glucose 0.730 0.008 0.234 0.89 0.710 0.96 0.828 0.98 0.622 1.05 0.893 -0.021 rs2191349 Fasting glucose A/C A Fasting glucose 0.589 0.009 0.852 1.01 0.844 1.02 0.145 0.90 0.936 1.00 0.536 -0.067 rs4607517 Fasting glucose G/A A Fasting glucose 0.980 0.001 0.974 1.00 0.371 1.09 0.914 0.99 0.871 1.01 0.306 0.138 rs11558471 Fasting glucose A/G A Fasting glucose 0.838 0.004 0.040 1.15 0.079 1.15 0.132 1.12 0.503 1.05 0.566 0.067 rs7034200 Fasting glucose C/A A Fasting glucose 0.930 0.001 0.952 1.00 0.904 0.99 0.080 0.89 0.356 0.95 0.400 -0.086 rs10885122 Fasting glucose C/A C Fasting glucose 0.923 0.002 0.430 0.93 0.178 0.86 0.305 1.11 0.867 0.99 0.239 0.187 rs4506565 Fasting glucose A/T T Fasting glucose 0.460 0.012 0.871 0.99 0.321 0.93 0.749 0.98 0.172 0.92 0.408 0.087 rs11605924 Fasting glucose C/A A Fasting glucose 0.767 -0.005 0.336 1.06 0.658 1.03 0.860 0.99 0.678 0.98 0.334 -0.100 rs7944584 Fasting glucose A/T A Fasting glucose 0.746 0.006 0.229 0.92 0.663 0.96 0.388 1.07 0.778 1.02 0.610 0.060 rs174550 Fasting glucose A/G A Fasting glucose 0.644 0.008 0.316 1.07 0.712 1.03 0.306 1.08 0.717 0.98 0.771 0.033

237

rs10830963 Fasting glucose C/G G Fasting glucose 0.739 -0.006 0.459 1.05 0.069 1.16 0.460 1.06 0.489 1.05 0.654 0.052 rs11071657 Fasting glucose A/G A Fasting glucose 0.976 -0.0005 0.541 1.04 0.819 0.98 0.393 1.06 0.972 1.00 0.185 -0.140 rs267734 eGFR A/G A eGFR 0.309 -0.957 0.843 0.98 0.957 1.01 0.262 0.91 0.313 0.93 0.758 -0.040 rs1260326 eGFR G/A G eGFR 0.398 0.648 0.672 1.03 0.423 0.94 0.101 0.90 0.605 0.97 0.514 -0.069 rs10206899 eGFR A/G A eGFR 0.652 0.408 0.962 1.00 0.831 1.02 0.111 0.88 0.803 0.98 0.012 0.314 rs347685 eGFR A/C A eGFR 0.947 -0.056 0.279 1.08 0.476 1.06 0.909 1.01 0.580 1.04 0.037 0.243 rs10032549 eGFR G/A G eGFR 0.911 0.087 0.202 1.09 0.178 1.11 0.712 1.03 0.788 0.98 0.246 -0.126 rs6420094 eGFR A/G G eGFR 0.549 0.474 0.705 1.03 0.232 1.10 0.651 1.03 0.341 1.06 0.739 -0.037 rs881858 eGFR A/G A eGFR 0.002 -2.560 0.626 0.97 0.213 0.90 0.596 0.96 0.588 0.96 0.133 0.174 rs2279463 eGFR A/G G eGFR 0.092 -1.810 0.958 1.00 0.614 1.05 0.597 1.05 0.076 0.86 0.518 0.095 rs6465825 eGFR A/G G eGFR 0.652 -0.345 0.463 0.95 0.330 1.08 0.877 0.99 0.806 0.99 0.246 0.123 rs10109414 eGFR G/A A eGFR 0.902 -0.094 0.108 0.90 0.118 0.89 0.214 0.92 0.324 0.94 0.168 0.148 rs4744712 eGFR C/A A eGFR 0.581 0.430 0.561 0.96 0.520 0.95 0.553 0.96 0.711 1.02 0.929 0.010 rs10794720 eGFR G/A A eGFR 0.295 -1.576 0.553 0.93 0.565 0.92 0.140 0.82 0.100 0.82 0.512 -0.136 rs3925584 eGFR A/G A eGFR 0.245 -0.895 0.842 1.01 0.345 0.93 0.341 0.94 0.378 0.95 0.448 0.080 rs10774021 eGFR A/G A eGFR 0.773 0.231 0.980 1.00 0.590 0.96 0.724 0.98 0.699 1.02 0.144 0.161 rs653178 eGFR A/G G eGFR 0.311 -0.749 0.447 1.05 0.570 0.96 0.095 0.90 0.167 0.92 0.062 -0.192 rs626277 eGFR A/C A eGFR 0.294 -0.819 0.595 1.04 0.997 1.00 0.623 1.03 0.710 1.02 0.156 -0.154 rs17536527 eGFR G/C C eGFR 0.675 -0.330 0.682 1.03 0.534 0.95 0.015 0.84 0.337 0.94 0.392 -0.094 rs491567 eGFR A/C A eGFR 0.005 -2.528 0.379 1.07 0.389 1.08 0.415 0.94 0.811 0.98 0.201 -0.158 rs1394125 eGFR G/A A eGFR 0.191 1.077 0.444 0.95 0.845 1.02 0.795 0.98 0.663 1.03 0.041 0.232 rs12917707 eGFR C/A C eGFR 0.360 -0.897 0.167 1.12 0.262 1.11 0.448 1.07 0.862 0.99 0.110 -0.217 rs9895661 eGFR A/G G eGFR 1.000 0.001 0.321 0.92 0.567 0.95 0.347 1.09 0.588 1.04 0.555 0.084 rs12460876 eGFR A/G A eGFR 0.242 -0.934 0.544 0.96 0.497 0.95 0.652 1.03 0.141 1.10 0.434 -0.085 rs911119 eGFR A/G A eGFR 0.326 -0.908 0.092 1.14 0.093 1.16 0.301 1.09 0.469 1.05 0.084 -0.219 rs4660293 HDL cholesterol A/G G HDL cholesterol 0.087 0.078 0.481 1.05 0.135 1.14 0.785 1.02 0.361 1.06 0.646 0.056 rs4846914 HDL cholesterol A/G G HDL cholesterol 0.318 -0.041 0.619 1.03 0.799 0.98 0.539 0.96 0.891 1.01 0.070 0.198

238

rs6754295 HDL cholesterol A/C A HDL cholesterol 0.117 -0.071 0.181 0.91 0.409 0.93 0.980 1.00 0.380 0.94 0.725 0.042 rs12328675 HDL cholesterol A/G A HDL cholesterol 0.058 0.114 0.083 1.18 0.043 1.24 0.362 1.09 0.769 1.03 0.748 0.051 rs2972146 HDL cholesterol A/C A HDL cholesterol 0.470 0.030 0.871 0.99 0.314 1.08 0.099 1.12 0.031 1.15 0.733 -0.038 rs6450176 HDL cholesterol G/A A HDL cholesterol 0.467 0.033 0.905 1.01 0.773 1.02 0.302 1.08 0.468 1.05 0.660 -0.052 rs2814944 HDL cholesterol G/A A HDL cholesterol 0.678 0.022 0.874 0.99 0.567 0.95 0.438 0.93 0.502 0.95 0.723 -0.050 rs605066 HDL cholesterol A/G G HDL cholesterol 0.819 -0.009 0.447 0.95 0.877 1.01 0.534 0.96 0.057 0.89 0.807 0.027 rs1084651 HDL cholesterol G/A A HDL cholesterol 0.129 -0.083 0.209 0.90 0.110 0.85 0.511 1.06 0.261 0.91 0.152 0.210 rs17145738 HDL cholesterol C/T C HDL cholesterol 0.666 -0.026 0.728 0.97 0.242 0.87 0.336 0.91 0.633 1.04 0.911 0.017 rs4731702 HDL cholesterol G/A G HDL cholesterol 0.901 -0.005 0.794 0.98 0.160 0.90 0.412 0.95 0.486 0.96 0.438 0.079 rs9987289 HDL cholesterol G/A A HDL cholesterol 0.507 -0.042 0.700 0.96 0.792 1.03 0.436 0.92 0.235 0.89 0.750 -0.055 rs12678919 HDL cholesterol A/G A HDL cholesterol 0.302 -0.072 0.987 1.00 0.830 1.03 0.774 0.97 0.669 1.05 0.039 0.378 rs2954029 HDL cholesterol A/T A HDL cholesterol 0.295 0.042 0.010 1.18 0.172 1.11 0.105 0.89 0.193 0.92 0.832 0.023 rs471364 HDL cholesterol T/C C HDL cholesterol 0.604 -0.033 0.531 0.94 0.570 0.94 0.456 1.09 0.171 0.88 0.522 0.107 rs1883025 HDL cholesterol G/A A HDL cholesterol 0.197 -0.058 0.162 1.11 0.549 1.05 0.172 1.11 0.300 0.93 0.181 -0.159 rs3136441 HDL cholesterol A/G A HDL cholesterol 0.825 0.013 0.122 1.16 0.992 1.00 0.821 0.98 0.157 0.88 0.477 -0.111 rs7120118 HDL cholesterol A/G A HDL cholesterol 0.198 -0.056 0.652 1.03 0.496 0.95 0.422 0.94 0.784 0.98 0.189 -0.151 rs7395662 HDL cholesterol G/A G HDL cholesterol 0.488 -0.029 0.848 0.99 0.744 1.03 0.179 1.10 0.797 0.98 0.805 -0.028 rs174546 HDL cholesterol G/A A HDL cholesterol 0.066 -0.077 0.316 0.93 0.712 0.97 0.306 0.93 0.717 1.02 0.771 -0.033 rs964184 HDL cholesterol G/C C HDL cholesterol 0.172 -0.076 0.747 0.97 0.375 0.91 0.799 1.02 0.503 1.06 0.835 0.031 rs7941030 HDL cholesterol A/G A HDL cholesterol 0.537 -0.025 0.458 1.05 0.900 0.99 0.915 0.99 0.386 1.05 0.426 -0.087 rs7134375 HDL cholesterol C/A C HDL cholesterol 0.042 -0.083 0.677 1.03 0.932 0.99 0.239 0.92 0.568 0.97 0.337 0.105 rs11613352 HDL cholesterol C/T C HDL cholesterol 0.059 -0.085 0.220 1.09 0.257 1.10 0.141 1.12 0.647 0.97 0.476 0.085 rs2338104 HDL cholesterol G/C C HDL cholesterol 0.449 0.031 0.726 1.02 0.563 1.04 0.001 1.25 0.010 1.17 0.178 -0.148 rs4765127 HDL cholesterol C/A C HDL cholesterol 0.562 -0.024 0.157 0.91 0.307 0.92 0.111 0.89 0.695 0.98 0.120 0.170 rs1532085 HDL cholesterol G/A G HDL cholesterol 0.002 -0.128 0.525 1.04 0.900 0.99 0.348 0.94 0.595 0.97 0.490 -0.076 rs2652834 HDL cholesterol G/A A HDL cholesterol 0.015 -0.123 0.951 1.00 0.871 1.02 0.472 0.94 0.603 0.96 0.511 0.089 rs3764261 HDL cholesterol C/A C HDL cholesterol 0.0001 -0.168 0.605 0.97 0.456 0.94 0.073 0.88 0.168 0.92 0.046 0.222

239

rs2271293 HDL cholesterol G/A G HDL cholesterol 0.172 -0.085 0.262 0.89 0.860 0.98 0.477 1.08 0.131 1.15 0.633 -0.079 rs2925979 HDL cholesterol G/A A HDL cholesterol 0.119 -0.066 0.080 1.13 0.150 1.12 0.545 0.96 0.435 0.95 0.406 0.093 rs11869286 HDL cholesterol G/C C HDL cholesterol 0.500 -0.028 0.736 1.02 0.689 0.97 0.595 1.04 0.867 0.99 0.007 0.292 rs4148008 HDL cholesterol C/G G HDL cholesterol 0.123 0.066 0.024 1.17 0.135 1.13 0.681 0.97 0.925 0.99 0.568 -0.065 rs4129767 HDL cholesterol G/A G HDL cholesterol 0.210 -0.050 0.528 0.96 0.879 0.99 0.070 1.13 0.684 1.03 0.743 0.035 rs4939883 HDL cholesterol C/T T HDL cholesterol 0.085 -0.088 0.988 1.00 0.661 1.04 0.996 1.00 0.528 1.05 0.328 -0.133 rs12967135 HDL cholesterol G/A A HDL cholesterol 0.193 -0.062 0.731 0.97 0.831 1.02 0.544 1.05 0.600 1.04 0.470 -0.091 rs2967605 HDL cholesterol G/A A HDL cholesterol 0.207 0.064 0.819 0.98 0.791 1.03 0.425 0.93 0.632 1.04 0.027 0.302 rs4420638 HDL cholesterol T/C C HDL cholesterol 0.072 -0.096 0.222 1.11 0.205 1.14 0.109 1.16 0.124 1.13 0.242 -0.167 rs1800961 HDL cholesterol G/A A HDL cholesterol 0.317 0.116 0.896 0.98 0.186 0.76 0.423 1.18 0.170 1.29 0.186 -0.399 rs7679 HDL cholesterol T/C C HDL cholesterol 0.466 -0.035 0.234 0.91 0.706 0.97 0.890 0.99 0.847 1.01 0.126 0.195 rs12027135 LDL cholesterol A/T A LDL cholesterol 0.919 0.161 0.626 0.97 0.333 1.08 0.663 0.97 0.571 0.97 0.098 0.178 rs11206510 LDL cholesterol A/G A LDL cholesterol 0.696 -0.740 0.980 1.00 0.205 0.89 0.964 1.00 0.952 1.00 0.410 -0.105 rs2479409 LDL cholesterol A/G G LDL cholesterol 0.071 2.911 0.718 0.98 0.266 0.92 0.323 1.07 0.561 1.04 0.873 0.017 rs2131925 LDL cholesterol A/C A LDL cholesterol 0.035 3.473 0.685 1.03 0.859 0.99 0.927 0.99 0.449 0.95 0.254 0.127 rs12740374 LDL cholesterol C/A C LDL cholesterol 0.134 2.792 0.507 1.05 0.245 1.11 0.546 0.95 0.501 0.95 0.045 0.254 rs2642442 LDL cholesterol T/C C LDL cholesterol 0.326 -1.682 0.739 0.98 0.535 1.05 0.049 1.16 0.523 1.04 0.335 0.110 rs514230 LDL cholesterol A/T A LDL cholesterol 0.053 3.024 0.534 0.96 0.719 1.03 0.276 0.93 0.692 0.98 0.648 -0.048 rs515135 LDL cholesterol G/A G LDL cholesterol 0.051 4.006 0.420 0.93 0.447 0.93 0.123 0.87 0.090 0.87 0.268 0.152 rs6756629 LDL cholesterol G/A G LDL cholesterol 0.864 0.526 0.476 1.09 0.427 0.89 0.348 0.88 0.295 0.88 0.789 -0.055 rs6544713 LDL cholesterol G/A A LDL cholesterol 0.051 3.243 0.067 1.13 0.201 1.11 0.471 0.95 0.749 0.98 0.973 0.004 rs3846663 LDL cholesterol C/T T LDL cholesterol 0.018 3.957 0.779 1.02 0.691 0.97 0.237 1.09 0.894 0.99 0.840 -0.023 rs1501908 LDL cholesterol C/G C LDL cholesterol 0.240 -1.936 0.144 1.10 0.256 1.09 0.457 0.95 0.115 0.90 0.128 0.169 rs3757354 LDL cholesterol G/A G LDL cholesterol 0.796 -0.507 0.131 1.13 0.647 1.04 0.714 0.97 0.733 0.97 0.865 -0.022 rs2142672 LDL cholesterol G/A G LDL cholesterol 0.595 0.952 0.117 1.12 0.701 1.03 0.381 1.07 0.183 1.10 0.836 -0.025 rs1800562 LDL cholesterol G/A G LDL cholesterol 0.243 3.478 0.871 1.02 0.023 1.34 0.955 1.01 0.634 0.95 0.401 -0.164 rs3177928 LDL cholesterol G/A A LDL cholesterol 0.654 1.001 0.696 0.97 0.505 1.07 0.678 0.96 0.507 0.94 0.148 -0.217

240

rs2254287 LDL cholesterol C/G G LDL cholesterol 0.940 0.118 0.595 0.97 0.108 0.89 0.921 0.99 0.298 1.07 0.571 0.059 rs1564348 LDL cholesterol A/G G LDL cholesterol 0.090 3.635 0.772 0.98 0.569 1.06 0.754 0.97 0.842 0.98 0.707 0.054 rs12670798 LDL cholesterol A/G G LDL cholesterol 0.092 3.160 0.742 0.98 0.815 1.02 0.882 0.99 0.517 1.05 0.324 0.122 rs2072183 LDL cholesterol C/G G LDL cholesterol 0.488 1.277 0.529 0.95 0.110 0.87 0.806 0.98 0.674 0.97 0.541 0.076 rs9987289 LDL cholesterol G/A G LDL cholesterol 0.031 -5.424 0.700 1.04 0.792 0.97 0.436 1.09 0.235 1.12 0.750 0.055 rs2081687 LDL cholesterol G/A A LDL cholesterol 0.050 3.226 0.407 0.95 0.757 1.02 0.295 0.93 0.149 0.91 0.576 0.061 rs2954029 LDL cholesterol A/T A LDL cholesterol 0.098 -2.665 0.010 1.18 0.172 1.11 0.105 0.89 0.193 0.92 0.832 0.023 rs11136341 LDL cholesterol A/G G LDL cholesterol 0.496 -1.092 0.620 0.97 0.216 0.91 0.911 1.01 0.868 0.99 0.644 0.049 rs635634 LDL cholesterol G/A A LDL cholesterol 0.220 2.524 0.577 0.95 0.202 0.88 0.154 0.88 0.670 0.97 0.388 0.120 rs2255141 LDL cholesterol G/A A LDL cholesterol 0.786 0.475 0.726 0.98 0.878 1.01 0.995 1.00 0.667 1.03 0.196 -0.151 rs174546 LDL cholesterol G/A G LDL cholesterol 0.250 -1.909 0.316 1.07 0.712 1.03 0.306 1.08 0.717 0.98 0.771 0.033 rs964184 LDL cholesterol G/C C LDL cholesterol 0.012 -5.658 0.747 0.97 0.375 0.91 0.799 1.02 0.503 1.06 0.835 0.031 rs11220462 LDL cholesterol G/A A LDL cholesterol 0.967 -0.095 0.707 0.97 0.624 0.95 0.731 1.03 0.363 0.92 0.914 -0.016 rs11065987 LDL cholesterol A/G A LDL cholesterol 0.144 2.230 0.447 0.95 0.705 1.03 0.056 1.13 0.086 1.11 0.085 0.179 rs2650000 LDL cholesterol C/A A LDL cholesterol 0.455 1.220 0.446 1.05 0.847 1.02 0.828 1.02 0.833 0.99 0.054 0.213 rs8017377 LDL cholesterol G/A A LDL cholesterol 0.789 -0.413 0.695 0.98 0.540 0.96 0.626 1.03 0.949 1.00 0.822 -0.024 rs3764261 LDL cholesterol C/A C LDL cholesterol 0.175 2.246 0.605 0.97 0.456 0.94 0.073 0.88 0.168 0.92 0.046 0.222 rs2000999 LDL cholesterol G/A A LDL cholesterol 0.324 -1.925 0.183 1.11 0.023 1.25 0.049 0.85 0.119 0.89 0.566 -0.075 rs7206971 LDL cholesterol G/A G LDL cholesterol 0.206 -2.022 0.030 0.87 0.033 0.85 0.798 1.02 0.296 1.07 0.144 -0.156 rs6511720 LDL cholesterol C/A C LDL cholesterol 0.045 4.335 0.044 0.83 0.339 0.90 0.861 0.98 0.443 1.07 0.482 -0.103 rs10401969 LDL cholesterol A/G A LDL cholesterol 0.743 0.811 0.414 1.09 0.808 0.97 0.562 1.06 0.928 0.99 0.680 0.068 rs4420638 LDL cholesterol T/C C LDL cholesterol 0.042 4.288 0.222 1.11 0.205 1.14 0.109 1.16 0.124 1.13 0.242 -0.167 rs2902940 LDL cholesterol A/G A LDL cholesterol 0.858 -0.295 0.268 1.08 0.224 1.10 0.799 0.98 0.240 0.93 0.462 -0.081 rs6102059 LDL cholesterol C/T C LDL cholesterol 0.544 -1.052 0.401 1.06 0.475 0.94 0.705 0.97 0.716 1.02 0.454 0.088 rs6029526 LDL cholesterol T/A A LDL cholesterol 0.384 1.356 0.373 1.06 0.290 1.08 0.151 1.10 0.288 1.07 0.828 -0.023 Self-reported rs6843082 Stroke A/G G 0.189 1.15 0.934 1.01 0.173 0.88 0.012 1.25 0.082 1.14 0.220 0.158 history of stroke Self-reported rs556621 Stroke C/A A 0.906 0.99 0.843 0.99 0.756 0.98 0.011 1.21 0.358 1.06 0.063 -0.207 history of stroke

241

Self-reported rs2107595 Stroke C/T T 0.647 1.05 0.521 0.94 0.220 0.88 0.609 0.95 0.162 0.89 0.820 -0.034 history of stroke Self-reported rs12425791 Stroke G/A A 0.220 0.88 0.755 0.97 0.160 0.88 0.780 1.02 0.755 1.02 0.287 0.146 history of stroke Self-reported rs879324 Stroke G/A A 0.257 0.89 0.929 1.01 0.394 1.09 0.883 1.01 0.595 1.04 0.706 0.050 history of stroke Type 2 diabetes rs10923931 Type 2 Diabetes G/T T - - 0.819 0.98 0.375 1.12 0.190 1.16 0.302 1.10 0.873 0.027 affected status Type 2 diabetes rs2075423 Type 2 Diabetes C/A C - - 0.560 0.96 0.684 0.97 0.821 0.98 0.557 1.04 0.894 -0.016 affected status Type 2 diabetes rs7578597 Type 2 Diabetes A/G A - - 0.925 0.99 0.333 0.87 0.017 0.73 0.001 0.68 0.155 0.274 affected status Type 2 diabetes rs243021 Type 2 Diabetes G/A A - - 0.705 0.98 0.418 1.06 0.720 0.98 0.884 1.01 0.269 0.113 affected status Type 2 diabetes rs7593730 Type 2 Diabetes G/A G - - 0.468 1.06 0.313 1.10 0.557 1.05 0.762 1.02 0.219 -0.164 affected status Type 2 diabetes rs13389219 Type 2 Diabetes G/A G - - 0.614 1.03 1.000 1.00 0.433 1.06 0.620 0.97 0.131 0.161 affected status Type 2 diabetes rs7578326 Type 2 Diabetes A/G A - - 0.634 1.03 0.244 1.10 0.090 1.13 0.074 1.12 0.881 -0.017 affected status Type 2 diabetes rs1801282 Type 2 Diabetes C/G C - - 0.172 1.15 0.555 1.07 0.177 0.86 0.666 0.96 0.382 -0.142 affected status Type 2 diabetes rs1496653 Type 2 Diabetes A/G A - - 0.127 1.13 0.118 1.15 0.904 1.01 0.212 1.10 0.058 0.247 affected status Type 2 diabetes rs4607103 Type 2 Diabetes C/T C - - 0.712 0.97 0.828 1.02 0.432 1.06 0.303 1.07 0.401 -0.101 affected status Type 2 diabetes rs11708067 Type 2 Diabetes A/G A - - 0.027 1.17 0.228 1.11 0.246 1.09 0.130 1.11 0.445 -0.092 affected status Type 2 diabetes rs4402960 Type 2 Diabetes G/T T - - 0.353 1.07 0.603 0.96 0.064 1.14 0.057 1.13 0.083 -0.198 affected status Type 2 diabetes rs4689388 Type 2 Diabetes A/G A - - 0.155 0.91 0.937 0.99 0.675 0.97 0.045 0.88 0.540 -0.066 affected status Type 2 diabetes rs459193 Type 2 Diabetes G/A G - - 0.049 1.16 0.762 1.03 0.287 0.92 0.285 0.93 0.999 0.000 affected status Type 2 diabetes rs4457053 Type 2 Diabetes A/G G - - 0.415 0.95 0.218 1.11 0.875 1.01 0.962 1.00 0.338 -0.108 affected status Type 2 diabetes rs10440833 Type 2 Diabetes T/A A - - 0.122 1.11 0.234 1.10 0.936 0.99 0.715 1.02 0.386 -0.099 affected status Type 2 diabetes rs17168486 Type 2 Diabetes C/T T - - 0.469 1.06 0.647 1.04 0.383 1.08 0.871 0.99 0.952 -0.008 affected status Type 2 diabetes rs864745 Type 2 Diabetes C/T T - - 0.948 1.00 0.920 1.01 0.989 1.00 0.638 1.03 0.495 -0.073 affected status Type 2 diabetes rs516946 Type 2 Diabetes G/A G - - 0.478 0.95 0.796 1.02 0.903 0.99 0.130 1.12 0.348 -0.120 affected status rs896854 Type 2 Diabetes G/A A Type 2 diabetes - - 0.287 0.93 0.037 0.86 0.077 0.89 0.016 0.87 0.042 0.215

242

affected status Type 2 diabetes rs10811661 Type 2 Diabetes T/C T - - 0.863 1.02 0.949 1.01 1.000 1.00 0.461 1.07 0.726 0.053 affected status Type 2 diabetes rs7018475 Type 2 Diabetes T/G T - - 0.529 0.95 0.134 0.87 0.271 0.92 0.694 0.97 0.755 0.040 affected status Type 2 diabetes rs13292136 Type 2 Diabetes G/A G - - 0.934 1.01 0.872 1.02 0.883 1.02 0.636 1.06 0.315 0.206 affected status Type 2 diabetes rs2796441 Type 2 Diabetes G/A G - - 0.629 1.03 0.983 1.00 0.064 0.88 0.068 0.89 0.732 0.037 affected status Type 2 diabetes rs12779790 Type 2 Diabetes A/G G - - 0.617 1.04 0.551 0.94 0.366 0.92 0.407 0.94 0.380 0.125 affected status Type 2 diabetes rs12571751 Type 2 Diabetes A/G A - - 0.466 0.95 0.243 1.09 0.865 0.99 0.596 1.03 0.686 -0.043 affected status Type 2 diabetes rs5015480 Type 2 Diabetes C/T C - - 0.028 1.15 0.044 1.17 0.846 0.99 0.151 1.09 0.888 0.015 affected status Type 2 diabetes rs7903146 Type 2 Diabetes G/A A - - 0.952 1.00 0.610 0.96 0.879 1.01 0.490 0.96 0.613 0.055 affected status Type 2 diabetes rs231362 Type 2 Diabetes C/T C - - 0.140 0.91 0.788 0.98 0.086 0.89 0.446 0.96 0.708 -0.039 affected status Type 2 diabetes rs163184 Type 2 Diabetes T/G G - - 0.053 1.13 0.059 1.16 0.036 0.86 0.973 1.00 0.456 0.080 affected status Type 2 diabetes rs5215 Type 2 Diabetes T/C C - - 0.756 1.02 0.362 0.93 0.348 0.94 0.363 0.95 0.572 0.060 affected status Type 2 diabetes rs1552224 Type 2 Diabetes A/C A - - 0.644 0.96 0.592 1.06 0.115 1.17 0.200 1.12 0.778 -0.044 affected status Type 2 diabetes rs1387153 Type 2 Diabetes G/A A - - 0.135 1.11 0.001 1.32 0.286 1.08 0.212 1.09 0.927 0.011 affected status Type 2 diabetes rs10842994 Type 2 Diabetes C/T C - - 0.466 1.06 0.803 1.03 0.968 1.00 0.142 0.89 0.193 -0.184 affected status Type 2 diabetes rs1531343 Type 2 Diabetes G/C C - - 0.621 0.95 0.562 1.08 0.307 0.89 0.091 0.84 0.406 0.149 affected status Type 2 diabetes rs7961581 Type 2 Diabetes T/C C - - 0.288 0.93 0.312 1.08 0.399 1.06 0.981 1.00 0.501 -0.074 affected status Type 2 diabetes rs7957197 Type 2 Diabetes T/A T - - 0.044 0.84 0.210 0.88 0.135 0.87 0.523 0.95 0.033 0.292 affected status Type 2 diabetes rs1359790 Type 2 Diabetes G/A G - - 0.349 1.07 0.412 1.07 0.130 1.12 0.152 1.10 0.322 -0.113 affected status Type 2 diabetes rs7177055 Type 2 Diabetes A/G A - - 0.584 0.96 0.658 0.96 0.335 0.93 0.504 0.96 0.607 0.062 affected status Type 2 diabetes rs11634397 Type 2 Diabetes G/A G - - 0.227 0.92 0.016 0.83 0.431 0.95 0.887 1.01 0.730 0.037 affected status Type 2 diabetes rs8042680 Type 2 Diabetes C/A A - - 0.897 0.99 0.951 1.01 0.407 0.94 0.882 1.01 0.549 0.071 affected status Type 2 diabetes rs8050136 Type 2 Diabetes C/A A - - 0.997 1.00 0.704 1.03 0.829 0.99 0.516 0.96 0.516 0.067 affected status

243

Type 2 diabetes rs7202877 Type 2 Diabetes T/G T - - 0.792 0.97 0.780 0.96 0.799 0.97 0.270 0.88 0.596 0.107 affected status Type 2 diabetes rs8090011 Type 2 Diabetes C/G G - - 0.602 1.04 0.901 1.01 0.283 1.08 0.909 1.01 0.754 0.035 affected status Type 2 diabetes rs12970134 Type 2 Diabetes G/A A - - 0.483 0.95 0.499 1.06 0.348 1.07 0.309 1.07 0.391 -0.099 affected status Type 2 diabetes rs10401969 Type 2 Diabetes A/G G - - 0.414 0.92 0.808 1.03 0.562 0.94 0.928 1.01 0.680 -0.068 affected status Type 2 diabetes rs5945326 Type 2 Diabetes A/G A - - 0.904 0.99 0.943 1.01 0.541 1.04 0.571 1.04 0.484 -0.077 affected status rs10903129 Total Cholesterol G/A G Total Cholesterol 0.528 -0.007 0.652 0.97 0.322 1.08 0.687 0.97 0.540 0.96 0.096 0.179 rs2479409 Total Cholesterol A/G G Total Cholesterol 0.015 0.027 0.718 0.98 0.266 0.92 0.323 1.07 0.561 1.04 0.873 0.017 rs10889353 Total Cholesterol A/C A Total Cholesterol 0.006 0.031 0.619 1.03 0.978 1.00 0.812 0.98 0.449 0.95 0.225 0.134 rs7515577 Total Cholesterol A/C A Total Cholesterol 0.260 0.015 0.983 1.00 0.841 0.98 0.484 0.94 0.410 0.94 0.281 -0.137 rs646776 Total Cholesterol A/G A Total Cholesterol 0.332 0.012 0.494 1.05 0.239 1.11 0.555 0.95 0.438 0.95 0.047 0.251 rs2642442 Total Cholesterol T/C C Total Cholesterol 0.444 -0.009 0.739 0.98 0.535 1.05 0.049 1.16 0.523 1.04 0.335 0.110 rs514230 Total Cholesterol A/T A Total Cholesterol 0.399 0.009 0.534 0.96 0.719 1.03 0.276 0.93 0.692 0.98 0.648 -0.048 rs693 Total Cholesterol G/A A Total Cholesterol 0.083 0.018 0.907 1.01 0.512 1.05 0.714 0.98 0.129 0.91 0.346 0.098 rs1260326 Total Cholesterol G/A A Total Cholesterol 0.142 0.016 0.672 0.97 0.423 1.06 0.101 1.12 0.605 1.03 0.514 0.069 rs6756629 Total Cholesterol G/A G Total Cholesterol 0.840 0.004 0.476 1.09 0.427 0.89 0.348 0.88 0.295 0.88 0.789 -0.055 rs7570971 Total Cholesterol C/A A Total Cholesterol 0.096 0.019 0.268 1.08 0.897 0.99 0.347 0.93 0.497 0.96 0.480 0.080 rs2290159 Total Cholesterol G/C G Total Cholesterol 0.189 -0.017 0.344 1.08 0.810 1.02 0.501 1.06 0.375 1.07 0.018 -0.305 rs3846662 Total Cholesterol A/G G Total Cholesterol 0.015 0.027 0.979 1.00 0.748 0.98 0.417 1.06 0.981 1.00 0.689 0.044 rs6882076 Total Cholesterol G/A G Total Cholesterol 0.413 -0.009 0.112 1.11 0.188 1.11 0.505 0.95 0.120 0.91 0.163 0.155 rs3757354 Total Cholesterol G/A G Total Cholesterol 0.948 -0.001 0.131 1.13 0.647 1.04 0.714 0.97 0.733 0.97 0.865 -0.022 rs1800562 Total Cholesterol G/A G Total Cholesterol 0.600 0.011 0.871 1.02 0.023 1.34 0.955 1.01 0.634 0.95 0.401 -0.164 rs3177928 Total Cholesterol G/A A Total Cholesterol 0.897 0.002 0.696 0.97 0.505 1.07 0.678 0.96 0.507 0.94 0.148 -0.217 rs2814982 Total Cholesterol G/A G Total Cholesterol 0.432 -0.014 0.764 0.97 0.479 0.92 0.406 1.09 0.115 1.17 0.461 -0.128 rs1564348 Total Cholesterol A/G G Total Cholesterol 0.253 0.017 0.772 0.98 0.569 1.06 0.754 0.97 0.842 0.98 0.707 0.054 rs12670798 Total Cholesterol A/G G Total Cholesterol 0.551 -0.008 0.742 0.98 0.815 1.02 0.882 0.99 0.517 1.05 0.324 0.122 rs2072183 Total Cholesterol C/G G Total Cholesterol 0.831 -0.003 0.529 0.95 0.110 0.87 0.806 0.98 0.674 0.97 0.541 0.076

244

rs9987289 Total Cholesterol G/A G Total Cholesterol 0.180 -0.023 0.700 1.04 0.792 0.97 0.436 1.09 0.235 1.12 0.750 0.055 rs1495741 Total Cholesterol A/G G Total Cholesterol 0.253 0.015 0.948 0.99 0.112 0.87 0.956 1.00 0.633 0.97 0.260 0.144 rs2081687 Total Cholesterol G/A A Total Cholesterol 0.023 0.025 0.407 0.95 0.757 1.02 0.295 0.93 0.149 0.91 0.576 0.061 rs2737229 Total Cholesterol A/C A Total Cholesterol 0.041 0.024 0.892 0.99 0.914 0.99 0.459 1.06 0.256 1.08 0.079 0.200 rs2954029 Total Cholesterol A/T A Total Cholesterol 0.868 -0.002 0.010 1.18 0.172 1.11 0.105 0.89 0.193 0.92 0.832 0.023 rs11136341 Total Cholesterol A/G G Total Cholesterol 0.059 -0.020 0.620 0.97 0.216 0.91 0.911 1.01 0.868 0.99 0.644 0.049 rs1883025 Total Cholesterol G/A G Total Cholesterol 0.072 0.022 0.162 0.90 0.549 0.95 0.172 0.90 0.300 1.07 0.181 0.159 rs635634 Total Cholesterol G/A A Total Cholesterol 0.206 0.018 0.577 0.95 0.202 0.88 0.154 0.88 0.670 0.97 0.388 0.120 rs2255141 Total Cholesterol G/A A Total Cholesterol 0.923 -0.001 0.726 0.98 0.878 1.01 0.995 1.00 0.667 1.03 0.196 -0.151 rs10128711 Total Cholesterol G/A G Total Cholesterol 0.595 0.006 0.376 1.07 0.798 0.98 0.237 0.91 0.891 1.01 0.469 -0.087 rs174570 Total Cholesterol G/A G Total Cholesterol 0.651 0.007 0.290 1.10 0.511 1.07 0.218 1.13 0.880 1.01 0.470 -0.110 rs964184 Total Cholesterol G/C C Total Cholesterol 0.907 0.002 0.747 0.97 0.375 0.91 0.799 1.02 0.503 1.06 0.835 0.031 rs7941030 Total Cholesterol A/G G Total Cholesterol 0.381 -0.010 0.458 0.95 0.900 1.01 0.915 1.01 0.386 0.95 0.426 0.087 rs11220462 Total Cholesterol G/A A Total Cholesterol 0.685 0.006 0.707 0.97 0.624 0.95 0.731 1.03 0.363 0.92 0.914 -0.016 rs11065987 Total Cholesterol A/G A Total Cholesterol 0.140 0.015 0.447 0.95 0.705 1.03 0.056 1.13 0.086 1.11 0.085 0.179 rs3764261 Total Cholesterol C/A A Total Cholesterol 0.962 0.001 0.605 1.04 0.456 1.06 0.073 1.14 0.168 1.09 0.046 -0.222 rs2000999 Total Cholesterol G/A A Total Cholesterol 0.254 -0.015 0.183 1.11 0.023 1.25 0.049 0.85 0.119 0.89 0.566 -0.075 rs7206971 Total Cholesterol G/A A Total Cholesterol 0.856 0.002 0.030 1.15 0.033 1.18 0.798 0.98 0.296 0.94 0.144 0.156 rs2228671 Total Cholesterol G/A G Total Cholesterol 0.504 0.010 0.162 0.88 0.714 0.96 0.650 0.96 0.954 1.00 0.913 0.016 rs2304130 Total Cholesterol A/G A Total Cholesterol 0.032 0.037 0.347 1.10 0.973 1.00 0.759 0.97 0.216 0.89 0.420 0.138 rs2075650 Total Cholesterol A/G G Total Cholesterol 0.106 0.026 0.836 0.98 0.267 1.14 0.123 1.18 0.315 1.10 0.570 -0.091 rs492602 Total Cholesterol A/G G Total Cholesterol 0.166 -0.015 0.412 1.05 0.034 1.17 0.228 1.08 0.045 1.13 0.753 0.033 rs2902940 Total Cholesterol A/G A Total Cholesterol 0.498 -0.008 0.268 1.08 0.224 1.10 0.799 0.98 0.240 0.93 0.462 -0.081 rs6029526 Total Cholesterol T/A A Total Cholesterol 0.383 0.009 0.373 1.06 0.290 1.08 0.151 1.10 0.288 1.07 0.828 -0.023 rs1800961 Total Cholesterol G/A G Total Cholesterol 0.820 0.007 0.896 1.02 0.186 1.31 0.423 0.85 0.170 0.78 0.186 0.399 rs2131925 Triglycerides A/C A Triglycerides 0.648 -0.013 0.685 1.03 0.859 0.99 0.927 0.99 0.449 0.95 0.254 0.127 rs4846914 Triglycerides A/G G Triglycerides 0.035 0.058 0.619 1.03 0.799 0.98 0.539 0.96 0.891 1.01 0.070 0.198

245

rs6754295 Triglycerides A/C A Triglycerides 0.543 0.018 0.181 0.91 0.409 0.93 0.980 1.00 0.380 0.94 0.725 0.042 rs1260326 Triglycerides G/A A Triglycerides 0.0002 0.097 0.672 0.97 0.423 1.06 0.101 1.12 0.605 1.03 0.514 0.069 rs10195252 Triglycerides A/G A Triglycerides 0.054 0.051 0.417 1.05 0.808 0.98 0.528 1.04 0.670 0.97 0.093 0.177 rs2972146 Triglycerides A/C A Triglycerides 0.123 -0.043 0.871 0.99 0.314 1.08 0.099 1.12 0.031 1.15 0.733 -0.038 rs645040 Triglycerides A/C A Triglycerides 0.423 -0.025 0.226 0.91 0.959 1.00 0.518 0.95 0.641 0.97 0.403 0.105 rs442177 Triglycerides A/C A Triglycerides 0.091 0.045 0.336 1.06 0.255 1.09 0.517 0.96 0.270 1.07 0.729 0.037 rs9686661 Triglycerides G/A A Triglycerides 0.511 0.021 0.490 1.05 0.372 0.93 0.760 0.98 0.775 1.02 0.200 0.160 rs6882076 Triglycerides G/A A Triglycerides 0.257 0.032 0.112 0.90 0.188 0.90 0.505 1.05 0.120 1.10 0.163 -0.155 rs2247056 Triglycerides G/A G Triglycerides 0.751 -0.010 0.170 1.10 0.814 1.02 0.458 0.94 0.273 1.08 0.766 0.036 rs17145738 Triglycerides C/T C Triglycerides 0.181 0.053 0.728 0.97 0.242 0.87 0.336 0.91 0.633 1.04 0.911 0.017 rs7819412 Triglycerides G/A A Triglycerides 0.955 0.001 0.988 1.00 0.299 1.08 0.332 1.07 0.760 1.02 0.348 -0.098 rs1495741 Triglycerides A/G G Triglycerides 0.605 -0.017 0.948 0.99 0.112 0.87 0.956 1.00 0.633 0.97 0.260 0.144 rs12678919 Triglycerides A/G A Triglycerides 0.078 0.082 0.987 1.00 0.830 1.03 0.774 0.97 0.669 1.05 0.039 0.378 rs17321515 Triglycerides A/G A Triglycerides 0.160 0.037 0.035 1.14 0.396 1.06 0.173 0.91 0.152 0.92 0.737 0.035 rs10761731 Triglycerides A/T A Triglycerides 0.529 0.017 0.367 1.06 0.518 1.05 0.005 1.21 0.012 1.16 0.158 -0.148 rs2068888 Triglycerides G/A G Triglycerides 0.029 0.057 0.382 0.95 0.786 0.98 0.998 1.00 0.665 1.03 0.077 0.186 rs174546 Triglycerides G/A A Triglycerides 0.150 0.040 0.316 0.93 0.712 0.97 0.306 0.93 0.717 1.02 0.771 -0.033 rs12286037 Triglycerides C/T T Triglycerides 0.002 0.155 0.198 0.86 0.065 0.79 0.175 0.85 0.254 0.88 0.349 0.183 rs2075292 Triglycerides A/C C Triglycerides 0.008 0.108 0.198 1.14 0.460 1.09 0.654 1.05 0.356 1.09 0.552 -0.096 rs11613352 Triglycerides C/T C Triglycerides 0.645 0.014 0.220 1.09 0.257 1.10 0.141 1.12 0.647 0.97 0.476 0.085 rs4765127 Triglycerides C/A C Triglycerides 0.576 0.015 0.157 0.91 0.307 0.92 0.111 0.89 0.695 0.98 0.120 0.170 rs2412710 Triglycerides G/A A Triglycerides 0.997 -0.0005 0.804 1.06 0.908 0.97 0.164 1.43 0.562 1.13 0.609 0.194 rs2929282 Triglycerides A/T T Triglycerides 0.131 -0.090 0.309 1.16 0.224 1.24 0.528 1.10 0.873 1.02 0.103 0.387 rs1532085 Triglycerides G/A G Triglycerides 0.128 0.042 0.525 1.04 0.900 0.99 0.348 0.94 0.595 0.97 0.490 -0.076 rs11649653 Triglycerides C/G C Triglycerides 0.099 -0.044 0.369 1.06 0.170 1.11 0.706 0.97 0.285 1.07 0.778 -0.030 rs3764261 Triglycerides C/A C Triglycerides 0.961 0.001 0.605 0.97 0.456 0.94 0.073 0.88 0.168 0.92 0.046 0.222 rs10401969 Triglycerides A/G A Triglycerides 0.002 0.127 0.414 1.09 0.808 0.97 0.562 1.06 0.928 0.99 0.680 0.068

246

rs439401 Triglycerides G/A G Triglycerides 0.079 0.048 0.949 1.00 0.400 1.07 0.818 1.02 0.416 1.05 0.151 -0.161 rs7679 Triglycerides T/C C Triglycerides 0.185 0.043 0.234 0.91 0.706 0.97 0.890 0.99 0.847 1.01 0.126 0.195 rs5756931 Triglycerides A/G A Triglycerides 0.913 -0.003 0.879 1.01 0.624 0.96 0.756 1.02 0.628 0.97 0.269 0.118

247

Supplementary Table 3: Association results for individual trait genetic risk scores. Relationships between the risk scores and the traits were examined using marginal models with generalized estimating equations. Models were adjusted for age and sex. Associations are reported for a one unit increase in genetic risk score. Abbreviations: CAC, coronary artery calcification; CAD, coronary artery disease; CI, confidence interval; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; MI, myocardial infarction.

All-Cause Mortality CVD Mortality Self-reported History of MI Self-reported History of CVD Coronary Artery Calcification

Hazard p- Hazard p- Odds p- Odds p- Beta Risk Score 95% CI 95% CI 95% CI 95% CI 95% CI p-value Ratio value Ratio value Ratio value Ratio value Estimate

Blood pressure 1.01 0.97 1.05 0.523 1.00 0.95 1.05 0.983 0.99 0.94 1.04 0.703 0.99 0.94 1.03 0.572 -0.035 -0.086 0.017 0.189 traits

BMI 1.01 0.97 1.05 0.585 1.00 0.95 1.06 0.923 1.04 0.99 1.09 0.137 1.02 0.98 1.06 0.299 0.018 -0.028 0.064 0.435 CAC 1.04 0.93 1.17 0.502 1.04 0.89 1.23 0.595 1.10 0.94 1.29 0.244 1.16 1.02 1.33 0.028 0.178 0.034 0.321 0.015 CAD 1.03 1.00 1.07 0.065 1.07 1.01 1.13 0.020 1.10 1.05 1.16 0.0002 1.03 0.98 1.08 0.216 0.087 0.043 0.131 0.0001 CRP 1.05 0.99 1.10 0.082 1.00 0.93 1.08 0.994 1.00 0.94 1.07 0.895 1.00 0.95 1.06 0.876 0.017 -0.045 0.079 0.590

Electrocardiogr 1.02 0.98 1.06 0.322 1.05 0.99 1.10 0.097 1.00 0.95 1.05 0.926 0.99 0.94 1.03 0.518 0.036 -0.009 0.081 0.118 am traits

Fasting glucose 0.98 0.94 1.02 0.390 0.98 0.92 1.05 0.586 1.00 0.93 1.06 0.892 1.00 0.95 1.06 0.932 -0.006 -0.064 0.051 0.826 eGFR 1.00 0.96 1.03 0.756 0.99 0.95 1.04 0.815 1.05 1.00 1.11 0.043 1.02 0.97 1.06 0.455 0.008 -0.040 0.057 0.733 HDL 0.99 0.96 1.02 0.399 0.99 0.95 1.04 0.709 0.99 0.95 1.04 0.775 1.00 0.97 1.04 0.787 0.029 -0.008 0.066 0.120 cholesterol LDL cholesterol 1.00 0.97 1.02 0.695 1.00 0.96 1.04 1.000 1.01 0.97 1.04 0.750 1.01 0.98 1.05 0.432 0.034 -0.002 0.071 0.064 Stroke 1.04 0.94 1.16 0.436 1.13 0.99 1.30 0.074 0.85 0.74 0.97 0.021 0.94 0.84 1.05 0.284 0.009 -0.105 0.123 0.875 Type 2 Diabetes 0.98 0.96 1.01 0.137 0.95 0.92 0.99 0.008 1.01 0.97 1.05 0.541 1.00 0.96 1.03 0.877 0.006 -0.029 0.041 0.728 Total 0.99 0.96 1.02 0.465 0.97 0.93 1.01 0.102 1.00 0.96 1.04 0.868 1.01 0.98 1.04 0.517 0.034 -0.002 0.070 0.062 Cholesterol Triglycerides 1.00 0.96 1.03 0.828 1.00 0.95 1.05 0.966 1.00 0.95 1.05 0.971 0.98 0.94 1.02 0.237 0.059 0.016 0.102 0.007

248

Supplementary Table 4: Association results for all individual trait genetic risk scores analyzed in a single model, with additional adjustment for age and sex. Relationships were examined using marginal models with generalized estimating equations. Associations are reported for a one unit increase in each genetic risk score, adjusting for covariates and all other individual genetic risk scores. Abbreviations: CAC, coronary artery calcification; CAD, coronary artery disease; CI, confidence interval; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; MI, myocardial infarction.

All-Cause Mortality CVD Mortality Self-reported History of MI Self-reported History of CVD Coronary Artery Calcification

Hazard p- Hazard p- Odds p- Odds p- Beta p- Risk Score 95% CI 95% CI 95% CI 95% CI 95% CI Ratio value Ratio value Ratio value Ratio value Estimate value Blood pressure 1.01 0.97 1.05 0.678 0.99 0.94 1.05 0.745 0.98 0.92 1.04 0.451 0.99 0.94 1.03 0.607 -0.039 -0.089 0.012 0.132 traits BMI 1.01 0.97 1.05 0.628 1.00 0.94 1.06 0.991 1.04 0.99 1.09 0.133 1.02 0.98 1.06 0.265 0.011 -0.035 0.056 0.649 CAC 1.03 0.91 1.16 0.699 1.01 0.85 1.19 0.943 1.06 0.90 1.25 0.479 1.16 1.01 1.33 0.038 0.130 -0.012 0.272 0.074 CAD 1.04 1.00 1.08 0.046 1.08 1.02 1.15 0.011 1.10 1.05 1.16 0.0003 1.02 0.97 1.07 0.400 0.077 0.032 0.122 0.001 CRP 1.05 0.99 1.10 0.087 1.00 0.93 1.09 0.922 1.00 0.93 1.07 0.974 1.00 0.95 1.06 0.894 0.024 -0.037 0.084 0.448 Electrocardiogram 1.03 0.99 1.07 0.193 1.06 1.00 1.12 0.036 1.01 0.95 1.06 0.826 0.99 0.94 1.03 0.547 0.046 0.002 0.091 0.042 traits Fasting glucose 0.99 0.95 1.04 0.678 0.99 0.93 1.06 0.855 0.98 0.91 1.05 0.564 1.00 0.94 1.05 0.860 -0.002 -0.059 0.055 0.952 eGFR 1.00 0.96 1.03 0.832 1.00 0.95 1.05 0.863 1.05 1.00 1.11 0.045 1.02 0.98 1.07 0.355 0.018 -0.031 0.068 0.466 HDL cholesterol 0.98 0.95 1.02 0.281 0.98 0.93 1.03 0.378 0.99 0.94 1.04 0.613 1.01 0.98 1.05 0.484 0.010 -0.030 0.049 0.640 LDL cholesterol 1.00 0.97 1.04 0.820 1.03 0.98 1.09 0.284 1.01 0.96 1.07 0.724 1.01 0.96 1.06 0.683 0.015 -0.030 0.060 0.512 Stroke 1.05 0.95 1.16 0.376 1.15 1.00 1.32 0.050 0.83 0.72 0.96 0.012 0.93 0.83 1.04 0.225 0.001 -0.110 0.111 0.992 Type 2 Diabetes 0.98 0.95 1.01 0.125 0.95 0.91 0.98 0.004 1.01 0.97 1.05 0.679 1.00 0.96 1.03 0.864 -0.003 -0.039 0.032 0.863 Total Cholesterol 0.98 0.95 1.02 0.400 0.94 0.89 0.99 0.030 0.99 0.94 1.04 0.672 1.01 0.96 1.05 0.748 0.019 -0.026 0.063 0.408 Triglycerides 1.00 0.97 1.04 0.861 1.01 0.96 1.07 0.605 1.00 0.95 1.05 0.963 0.97 0.93 1.01 0.124 0.050 0.002 0.097 0.040

249

Chapter 7

Predictors of all-cause and cardiovascular disease mortality in type 2 diabetes: Diabetes Heart Study

Laura M. Raffield, Fang-Chi Hsu, Amanda J. Cox, J. Jeffrey Carr, Barry I. Freedman, Donald W. Bowden

250

Abstract:

Background:

Many studies evaluated the best predictors for cardiovascular disease (CVD) events in individuals with type 2 diabetes (T2D), but few studies examined the factors most strongly associated with mortality in T2D. The Diabetes Heart Study (DHS), an intensively phenotyped family-based cohort enriched for T2D, provided an opportunity to address this question.

Methods:

Associations with mortality were examined in 1022 European Americans affected by

T2D from 476 DHS families. All-cause mortality was 31.2% over an average 9.6 years of follow-up. Cox proportional hazards models with sandwich-based variance estimation were used to evaluate associations between all-cause and CVD mortality and 24 demographic and clinical factors, including coronary artery calcified plaque (CAC), carotid artery intima-media thickness, medications, body mass index, waist hip ratio, lipids, blood pressure, kidney function, QT interval, educational attainment, and glycemic control. Nominally significant factors (p<0.25) from univariate analyses were included in model selection (backward elimination, forward selection, and stepwise selection). Age and sex were included in all models.

Results:

The all-cause mortality model selected from the full DHS sample included age, sex,

CAC, urine albumin:creatinine ratio (UACR), insulin use, current smoking, and

251

educational attainment. The CVD mortality model selected from the full sample included age, sex, CAC, UACR, triglycerides, and history of CVD events. Beyond age, the most significant associations for both mortality models were CAC (2.03 x 10-4 ≤ p ≤ 0.001) and

UACR (1.99 x 10-8 ≤ p ≤ 2.23 x 10-8). To confirm the validity of the main predictors identified with model selection using the full sample, a two-fold cross-validation approach was used, and similar results were observed.

Conclusions:

This analysis highlights important demographic and clinical factors, notably CAC and albuminuria, which predict mortality in the general population of patients with T2D.

Keywords: type 2 diabetes, mortality, coronary artery calcified plaque, urine albumin:creatinine ratio

252

Background:

The Diabetes Heart Study (DHS) is an ongoing family-based cohort study investigating the epidemiology and genetics of cardiovascular disease (CVD) in a population-based sample. Prior analyses in this type 2 diabetes (T2D)-enriched cohort

(Bowden, Cox et al. 2010) have individually examined contributors to all-cause and CVD mortality. Coronary artery calcified atherosclerotic plaque (CAC), a measure of subclinical CVD (Agarwal, Morgan et al. 2011; Agarwal, Cox et al. 2013), C-reactive protein (CRP) (Cox, Agarwal et al. 2012), biventricular volume (Cox, Hugenschmidt et al. 2013), heart rate-corrected electrocardiographic QT interval (Cox, Azeem et al. 2014), serum albumin and creatine, estimated glomerular filtration rate (eGFR), and urine albumin:creatinine ratio (UACR) (Cox, Hsu et al. 2013) were all found to predict mortality in the DHS.

Prior studies have examined risk prediction scores for CVD events in patients with T2D (van Dieren, Beulens et al. 2012; Yeboah, Erbel et al. 2014), which is an important contributor to mortality (Go, Mozaffarian et al. 2013). However, fewer studies have attempted to evaluate the best predictors of all-cause and CVD mortality in T2D

(Wells, Jain et al. 2008; Yang, So et al. 2008; McEwen, Karter et al. 2012; De Cosmo,

Copetti et al. 2013). We performed a comprehensive analysis of which factors were the strongest independent predictors of all-cause and CVD mortality using model selection

(backward elimination, forward selection, and stepwise selection) in European Americans

(EAs) with T2D from the DHS.

253

Methods:

Study Design and Sample

The DHS recruited T2D-affected siblings without advanced renal insufficiency from 1998 through 2005 in western North Carolina. T2D was defined as diabetes developing after the age of 35 years initially treated with changes in diet and exercise and/or oral agents, in the absence of historical evidence of ketoacidosis or initial treatment with insulin. Fasting glucose and glycated hemoglobin (HbA1C) concentrations were assessed at the exam visit. Ascertainment and recruitment criteria have been described (Bowden, Cox et al. 2010). DHS participants were recruited from the general population and broadly reflect the demography of T2D in our community. Importantly, prior evidence of CVD was not an exclusion to participate. All study protocols were approved by the Institutional Review Board at Wake Forest School of Medicine, and all participants provided written informed consent.

Participant examinations were conducted in the General Clinical Research Center of Wake Forest Baptist Medical Center. Examinations included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses, and spot urine collection. Low-density lipoprotein (LDL) cholesterol was calculated using the

Friedewald equation, and LDL measures were considered valid for subjects whose triglycerides were less than 400 mg/dL. Estimated glomerular filtration rate (eGFR) was computed using the CKD-EPI equation (Levey, Stevens et al. 2009). CAC was assessed using computed tomography (CT), summing the left main, left anterior descending,

254

circumflex, posterior descending, and right coronary arteries. CT scans were performed on multi-detector CT scanners with cardiac gating in chest scans. CAC scores were measured as previously described and validated (Carr, Crouse et al. 2000; Carr, Nelson et al. 2005). Data on prior CVD events was self-reported by participants and non- adjudicated.

Mortality was assessed using the National Social Security Death Index. For deceased participants, length of follow-up was determined from the date of initial study visit to date of death. For all other participants the length of follow-up was determined from the date of the initial study visit to December 31, 2013. When possible, copies of death certificates were obtained from county or state Vital Records Offices to determine cause of death. Both all-cause and CVD mortality were analyzed; cause of death was categorized based on death certificates as CVD mortality (myocardial infarction, congestive heart failure, cardiac arrhythmia, sudden cardiac death, peripheral vascular disease, and stroke) or as mortality from other causes. For 18 patients, cause of death information could not be obtained, so these participants were excluded from analyses of

CVD mortality.

Statistical Methods

Cox proportional hazards models with sandwich-based variance estimation (due to the family structure of the DHS) in SAS 9.3 were used to evaluate associations with mortality. For model building, 24 potential predictors of mortality for which data was available for most individuals in the cohort, including CAC, carotid artery intima media thickness (IMT), medications, body mass index (BMI), waist hip ratio (WHR), lipids, blood pressure, kidney function measures, electrocardiographic QT interval, educational

255

attainment, and glycemic control measures, were evaluated in separate models for their association with all-cause mortality and CVD mortality. Variables were transformed prior to analysis; the square root of BMI and high-density lipoprotein (HDL) cholesterol, the square of WHR, and the natural log of glucose, HbA1C, CAC, total cholesterol, triglycerides, UACR, QT interval, diabetes duration, IMT, and mean arterial pressure were used. Nominally significant factors (p<0.25) from univariate analyses were included in model selection performed in SAS 9.3. We confirmed that no factors included in model selection were strongly correlated (r>0.8). Age and sex were included in all models. For backward elimination selection, factors with a p-value < 0.05 for association with mortality were retained in the model. For stepwise selection, factors meeting a threshold of a p-value < 0.25 could enter into the model but required a p-value < 0.05 to be retained. For forward selection, factors meeting a threshold of a p-value < 0.05 could enter into the model. In our analysis, results from the multiple Cox regression models were concordant for forward, backward, and stepwise model selections. Calculations for area under the receiver operating characteristic curve (AUC) were performed in STATA version 12.1.

Results:

All self-described EAs from the DHS with T2D (1022 individuals from 476 families) were included in the analyses. Table 1 displays their demographic and clinical characteristics, presented for the full sample and stratified by mortality status. For the full cohort, mean ± standard deviation (SD) diabetes duration was 10.4 ± 7.2 years.

Prevalence of hypertension, obesity, subclinical CVD based on CAC, and prior CVD

256

events were high. Over a mean ± SD follow-up of 9.6 ± 3.2 years, all-cause mortality was

31.2% and CVD mortality was 14.3%.

The univariate associations of demographic and clinical factors with all–cause and CVD mortality in the full EA T2D-affected cohort are shown in Table 2.

Demographic and clinical factors were selected for model selection based on their associations with all-cause or CVD mortality (p<0.25) (Table 2). Based on these univariate associations, model selection for all-cause mortality included HbA1c, CAC, pulse pressure, HDL, eGFR, UACR, diabetes duration, BMI, high blood pressure medication use, insulin use, current smoking, history of CVD, educational attainment

(less than high school, high school, greater than high school), WHR, mean arterial pressure, and IMT. Model selection for CVD mortality included fasting glucose, HbA1c,

CAC, pulse pressure, HDL, triglycerides, eGFR, UACR, QT interval, diabetes duration,

BMI, high blood pressure medication use, insulin use, history of CVD, educational attainment, WHR, and IMT.

The all-cause mortality model selected from the full sample included age, sex,

CAC, UACR, insulin use, current smoking, and educational attainment (Table 3). Other than age, the most significant associations with all-cause mortality were for CAC (p=2.03 x 10-4, hazard ratio (HR) 1.48 for a standard deviation (SD) change) and UACR (p= 2.23 x 10-8, HR 1.37 for a SD change). The CVD mortality model selected from the full sample included age, sex, CAC, UACR, triglycerides, and history of CVD (Table 3).

Similar to the all-cause mortality analysis, other than age, the most significant associations with CVD mortality were for CAC (p= 0.001, HR 1.71 for a SD change) and

UACR (p= 1.99 x 10-8, HR 1.51 for a SD change).

257

To confirm the validity of the most important predictors identified with model selection using the full DHS sample, a two-fold cross-validation approach was used, where the dataset was randomly split by family into roughly equal halves. The analysis for the full sample was repeated separately in each random dataset. In each random dataset, demographic and clinical factors were selected to include in model building based on their univariate associations with all-cause or CVD mortality (Additional File

1). Models were selected using forward, backward, and stepwise selections for all-cause and CVD mortality in random datasets 1 and 2 (Additional Files 2 and 3). As the associated predictors selected from each dataset differed slightly, a combined model including factors selected in the final model for each dataset was created for both all- cause and CVD mortality. Therefore, from this two-fold cross-validation approach, the all-cause mortality model included age, sex, CAC, UACR, diabetes duration, current smoking, educational attainment, insulin use, and WHR, while the CVD mortality model included age, sex, CAC, UACR, history of CVD events, and diabetes duration. The factors selected are similar to those observed using the full sample, which shows that the results are stable with no evidence of overfitting. This full model was fitted in both random datasets, with a basic model limited to age and sex was also fitted as a comparison to the full model.

Predicted values were then derived in each randomly selected dataset using regression coefficients derived from fitting the model in the opposite dataset, allowing calculation of average AUC for each dataset (Table 4), both for the full model derived using the two-fold cross-validation approach and a basic model including only age and sex. The full model significantly improved prediction of both all-cause and CVD

258

mortality (p<0.05) in both random datasets (Table 4). An estimate of AUC for the whole

EA cohort was also derived by fitting the models derived using the two-fold cross- validation approach in the whole sample, with an AUC of 0.788 for the full all-cause mortality model in the whole sample, compared to 0.699 for a basic model containing only age and sex (p=6.50 x 10-9 for improved prediction using the full model). Similarly, the full CVD mortality model had an AUC of 0.764, compared to 0.652 for a basic model containing only age and sex (p=4.14 x 10-7 for improved prediction using the full model)

(Table 4).

Factors which were predictors of mortality in the DHS in previous analyses, such as CRP (Cox, Agarwal et al. 2012) and biventricular volume (Cox, Hugenschmidt et al.

2013), were not included in the main model selection due to lower sample size (n=848 for

CRP, n= 771 for biventricular volume). When adding biventricular volume to the full all- cause and CVD mortality models selected using a two-fold cross-validation approach, biventricular volume was nominally associated with increased all-cause mortality risk

(HR=1.21 for a SD change, p=0.053) (Table 5) and more strongly associated with CVD mortality risk (HR=1.52 for a SD change, p=0.002) (Table 6). In contrast, CRP was a significant predictor when added to the all-cause mortality model (HR=1.30 for a SD change, p=0.001) (Table 5), but not the CVD mortality model (HR=1.19 for a SD change, p=0.167) (Table 6).

Discussion:

This set of DHS analyses highlights demographic and clinical factors independently predictive of mortality in EAs with T2D. Most prior mortality model

259

analyses in individuals with T2D have included a more limited set of clinical and demographic factors, for example BMI, lipids, medications, glycemic control, and eGFR, compared to those available in the DHS. As such, the literature does not fully address which factors may predict all-cause and CVD mortality by considering novel predictors, particularly CAC, in individuals with T2D. All the clinical and demographic factors included in model selection can be hypothesized to contribute to mortality. Our analysis shows, out of many mortality associated predictors, which ones are the most important to consider in T2D-affected individuals using a statistical model selection approach. While results from model selection for all-cause and CVD mortality differed slightly between model selection using the full cohort and model selection using a two-fold cross- validation approach, the key factors selected, notably CAC and UACR, did not differ.

This demonstrates the validity of model selection results. Other factors selected using the full sample, such as current smoking, insulin use, and educational attainment with all- cause mortality and history of cardiovascular disease with CVD mortality, were also included in the models selected using a two-fold cross-validation approach. This indicates results from the model selection are stable and reveals important factors to consider in future analyses of mortality in community-based cohorts of T2D-affected individuals.

The two factors most consistently associated with all-cause and CVD mortality were CAC and albuminuria. Prior studies in T2D have demonstrated striking associations of UACR with renal end-points, CVD events, CVD mortality, and all-cause mortality

(Ninomiya, Perkovic et al. 2009; Fox, Matsushita et al. 2012; Wada, Haneda et al. 2014).

Prior mortality analyses with shorter follow-up in the full DHS cohort with and without

T2D (83.7% T2D-affected) also found UACR to be associated with all-cause and CVD

260

mortality, independent from CAC and other CVD risk factors such as age, sex, T2D affection status, BMI, current smoking, hypertension, dyslipidemia, renin-angiotensin system blocking medications, and prior CVD (Cox, Hsu et al. 2013). As in prior analyses of individuals with T2D (Ninomiya, Perkovic et al. 2009), UACR was a stronger predictor of CVD mortality than eGFR in our models. UACR is a marker of generalized endothelial dysfunction, more so than kidney disease specifically, which is better reflected by changes in eGFR. UACR was not assessed in some analyses building mortality prediction models in T2D (Wells, Jain et al. 2008; McEwen, Karter et al. 2012), but it was selected as a significant predictor of elevated all-cause mortality in the Hong

Kong Diabetes Registry cohort (Yang, So et al. 2008) and the Gargano Mortality Study

(De Cosmo, Copetti et al. 2013). The incorporation of CAC significantly improves mortality risk prediction, which is expected. CAC is a strong independent predictor of

CVD events and mortality in the general population and in T2D (Raggi, Shaw et al. 2004;

Detrano, Guerci et al. 2008; Folsom, Kronmal et al. 2008; Elias-Smale, Proenca et al.

2010; Erbel, Mohlenkamp et al. 2010; Polonsky, McClelland et al. 2010; Agarwal,

Morgan et al. 2011; Agarwal, Cox et al. 2013; Kramer, Zinman et al. 2013), with individuals affected by diabetes tending to have higher CAC (Hoff, Quinn et al. 2003).

Recent model building using participants with T2D from the Multi-Ethnic Study of

Atherosclerosis (MESA) and the Heinz Nixdorf Recall Study found that using CAC improved risk prediction for incident CVD events, an important contributor to mortality risk in those with T2D, above the Framingham or United Kingdom Prospective Diabetes study risk models which include more conventional risk factors such as age, sex, systolic blood pressure, HbA1c, and lipid levels (Yeboah, Erbel et al. 2014). Notably, carotid

261

IMT, another novel predictor of CVD risk in T2D, was not selected for in these CVD event prediction models in T2D (Yeboah, Erbel et al. 2014), consistent with our finding that IMT was not an independent predictor of mortality in the DHS and previous literature indicating that CAC provides superior risk prediction and reclassification compared to carotid IMT (Yeboah, McClelland et al. 2012; Ferket, van Kempen et al.

2014).

Demographic factors including current smoking, insulin use, and educational attainment were also associated with all-cause mortality in the DHS. Current smoking

(Wells, Jain et al. 2008; McEwen, Karter et al. 2012) and use of insulin (Wells, Jain et al.

2008; Yang, So et al. 2008; McEwen, Karter et al. 2012; De Cosmo, Copetti et al. 2013) have been previously included in prediction models as associated with elevated mortality risk in analyses of individuals with T2D. Lower educational attainment has also been associated with elevated mortality risk in prior assessments of T2D, although this was not always included in final selected models, potentially due to the inclusion of correlated variables such as income that were not used in our analysis (McEwen, Karter et al. 2012).

History of CVD events being associated with CVD mortality is not surprising, given the elevated risk of secondary CVD events in those with established CVD.

Additional factors not included in the primary model building analysis due to reduced sample size may also be important to mortality risk prediction in T2D. Previous analysis in the DHS with shorter mortality follow-up time suggested that biventricular volume significantly increased AUC for prediction of all-cause and CVD mortality versus a model including age, sex, and CAC alone (Cox, Hugenschmidt et al. 2013). This is substantiated by data from this analysis, with biventricular volume improving

262

prediction of CVD mortality (Table 6). Biventricular volume can be derived from the same CT scans used to evaluate CAC, but it appears to be predictive of CVD mortality independent of CAC burden, so this may be an important factor to include in future analyses of mortality in T2D. CRP may also need to be considered in future analyses. In a previously published analysis from the DHS, each one unit increase in log transformed

CRP was associated with a 1.5-fold increase in risk for all-cause mortality (HR 1.54; 95%

CI 1.33–1.77) (Cox, Agarwal et al. 2012). Similar results were observed, though with a slightly attenuated HR, when CRP was added to our full all-cause mortality model from this analysis (Table 5). The lack of association with CVD mortality for CRP may not be surprising, as some studies have cast doubt on the causal role of CRP in CVD events

(Wensley, Gao et al. 2011).

Recent work in the DHS highlighted demographic and clinical factors associated with mortality in individuals with T2D at very high CVD risk based on a CAC score

>1000, with, for example, higher HbA1c, longer diabetes duration, reduced kidney function, reduced use of statins, and higher CRP associated with higher mortality (Cox,

Hsu et al. 2014). We attempted to perform model selection in only T2D affected individuals with CAC >1000 using similar methods to the analysis in the full T2D- affected DHS sample. Models were not perfectly concordant between forward, backward, and stepwise selection, not surprisingly given the reduced sample size (n=371), but selected models all included CAC, UACR, and HbA1c as contributors to elevated risk of all-cause and CVD mortality in this very high risk sample, similar to the results in the full

T2D-affected cohort.

263

Limitations of the current analysis include the lack of a replication cohort with similar mortality data and demographic and clinical factors assessed, necessitating use of a two-fold cross-validation approach. There are, however, few studies with the same breadth of data in a population-based sample of individuals with T2D. The latter, we believe, is an important feature of the DHS since it likely reflects what actually influences mortality in the community, as opposed to in a clinical trial. Previous analyses have been performed in cohorts with different recruitment strategies than the DHS, for example population-based cohorts in Hong Kong (Yang, So et al. 2008) and Italy (De

Cosmo, Copetti et al. 2013), or cohorts extracted from electronic medical record data whose diabetes was not as well characterized as the DHS and with exclusions based on medication use (Wells, Jain et al. 2008). None of these cohorts had data on all the novel measures of CVD risk available in the DHS. Length of follow-up, over nine years on average, was also greater than in previous analyses (Wells, Jain et al. 2008; Yang, So et al. 2008; McEwen, Karter et al. 2012; De Cosmo, Copetti et al. 2013), as was duration of

T2D in study participants (Yang, So et al. 2008; De Cosmo, Copetti et al. 2013). Data for some demographic and clinical factors which predicted all-cause mortality in prior analyses in T2D, such as cancer (Yang, So et al. 2008) or Charlson index, a measure of comorbidity burden (McEwen, Karter et al. 2012), were not available in the DHS, while some factors available in the DHS, notably CAC, were not used in prior model selection, making direct comparisons of mortality models derived in T2D-affected individuals difficult.

Additional analytic limitations include the relatively small number of events, in particular for CVD mortality (14.3% of the cohort), which necessitated use of only two-

264

fold cross validation, as opposed to an analysis strategy splitting the cohort by family into a greater number of small random samples. While these results highlight factors that may be important for mortality risk assessment in the T2D-affected population, causal relationships are uncertain. The current analyses were limited to individuals of EA descent; since levels of CAC differ from those in African Americans and different factors may impact mortality risk in other ethnic groups.

Conclusions:

The present results demonstrate the importance of assessing UACR and CAC in particular for prediction of CVD and all-cause mortality. Further analysis of these predictors of mortality, particularly CAC, in additional T2D-affected cohorts is needed.

265

Table 1: Demographic and clinical characteristics of European American participants with type 2 diabetes, stratified by mortality status.

Full Sample (n=1022) Living (n=703) Deceased (n=319)

Mean (SD) or % Median (range) Mean (SD) or % Median (range) Mean (SD) or % Median (range) Trait 62.43 (9.07) 62.80 (34.21, 60.64 (8.48) 60.54 (34.21, 66.36 (9.09) 67.03 (34.34, Age (years) 85.98) 81.77) 85.98)

51.5% 54.8% 44.2% Female Sex (%) 16.1% 14.0% 20.8% Current Smoking (%) 43.4% 43.6% 43.1% Past Smoking (%) History of 43.2% 56.1% Cardiovascular 37.3% Disease (%) 31.2% 0% All-Cause Mortality 100% Cardiovascular 14.3% 47.8% 0% Disease Mortality (%) Educational 25.6% 20.8% 36.4% Attainment- Less than High School (%) Educational 49.2% 51.4% 44.1% Attainment- High

School (%)

Educational 25.2% 27.8% Attainment- Greater 19.5% than High School (%)

32.38 (6.58) 31.32 (17.10, 32.67 (6.40) 31.61 (17.10, 31.75 (6.93) 30.54 (17.54, Body Mass Index 57.97) 57.97) 56.92) (kg/m2)

266

0.941 (0.082) 0.944 (0.436, 0.937 (0.077) 0.939 (0.648, 0.949 (0.091) 0.955 (0.436, Waist Hip Ratio 1.246) 1.234) 1.246)

147.9 (56.15) 135 (16, 463) 145.6 (51.17) 134 (16, 463) 152.9 (65.60) 140 (46, 436) Glucose (mg/dL) Glycated Hemoglobin 7.60 (1.72) 7.20 (4.30, 18.30) 7.49 (1.61) 7.10 (4.60, 16.40) 7.84 (1.92) 7.60 (4.30, 18.30) (%) Diabetes Duration 10.42 (7.15) 8 (0, 46) 9.32 (6.39) 7 (0, 46) 12.80 (8.09) 11 (1, 41) (years) Coronary Artery 1856 (3335) 1328 (3038) 3048 (3657) Calcified Plaque 449.5 (0, 50415) 256.5 (0, 50415) 1632 (0, 22378) (mass score) 0.683 (0.136) 0.659 (0.449, 0.664 (0.127) 0.639 (0.449, 0.723 (0.144) 0.700 (0.456, Carotid Intima Media 1.569) 1.569) 1.316) Thickness (mm) Total Cholesterol 185.2 (43.42) 181 (65, 427) 184.5 (43.15) 180 (65, 427) 186.9 (44.04) 183 (70, 386) (mg/dL) 42.16 (11.89) 41 (8, 98) 42.46 (11.59) 41 (14, 94) 41.49 (12.52) 40 (8, 98) HDL (mg/dL) 208.9 (138.8) 176 (30, 1310) 206.0 (136.4) 173 (30, 1310) 215.3 (143.9) 183 (30, 1065) Triglycerides (mg/dL) 103.0 (32.6) 100 (12, 236) 102.2 (32.4) 100 (14, 207) 104.9 (33.1) 103 (12, 236) LDL (mg/dL) Pulse Pressure 67.09 (16.93) 65.5 (28, 159) 65.53 (15.85) 64 (28, 124) 70.52 (18.67) 69.5 (31, 159) (mmHg) Mean Arterial 95.05 (11.24) 94 (66.5, 154) 95.19 (10.33) 94 (66.5, 140.3) 94.75 (13.02) 93.17 (67.33, 154) Pressure (mmHg) Estimated 65.76 (18.38) 67.95 (17.86) 65.43 (20.97, 60.96 (18.62) Glomerular Filtration 63.77 (8.91, 126.4) 122.2) 58.43 (8.91, 126.4) Rate (ml/min/1.73m2) Urine 121.5 (530.5) 68.49 (390.8) 239.8 (741.5) Albumin:creatinine 14.29 (0.48, 9449) 11.57 (0.48, 8878) 23.08 (0.78, 9449) Ratio (mg/g) 393.8 (33.14) 392.0 (270.1, 393.2 (32.27) 392.0 (310.1, 395.0 (34.93) 392.0 (270.1, QT Interval (ms) 564.0) 564.0) 526.0)

High Blood Pressure 75.5% 72.0% 83.4%

Medications (%)

267

42.7% 44.8% 37.9% Statin Use (%) Oral Hypoglycemic 78.8% 79.0% 78.4%

Medications (%) 27.3% 22.9% Insulin Use (%) 37.0% 0.611 (1.004) 0.296 (0.005, 0.529 (0.793) 0.801 (1.355) C-reactive Protein 12.73) 0.28(0.005,9.573) 0.353(0.008,12.73) (mg/dL) 384.5 (112.1) 375.6 (156.6, 372.0 (100.7) 361.3 (156.6, 411.8 (129.7) 397.2 (182.6, Biventricular Volume 884.5) 781.8) 884.5) (ml)

268

Table 2: Associations of demographic and clinical factors with all-cause and CVD mortality.

All-cause Mortality CVD mortality Trait HR 95% HR CI p-value n HR 95% HR CI p-value n Age (years) 1.86 1.62 2.13 < 1 x10-16 1022 1.74 1.41 2.13 1.40 x 10-7 1004 Female Sex (%) 0.68 0.54 0.85 8.83 x 10-4 1022 0.58 0.41 0.82 0.002 1004 Current Smoking (%) 1.44 1.11 1.88 0.006 1018 1.04 0.66 1.63 0.866 1000 Past Smoking (%) 0.99 0.79 1.23 0.896 1018 1.13 0.82 1.57 0.452 1000 History of Cardiovascular 1.91 1.52 2.41 3.86 x 10-8 1013 3.25 2.25 4.69 3.21 x 10-10 995 Disease (%)

Educational attainment 0.68 0.58 0.81 1.73 x 10-5 1011 0.74 0.58 0.95 0.02 993 (3 levels)

Body Mass Index (kg/m2) 0.87 0.77 0.99 0.033 1022 0.89 0.74 1.07 0.225 1004 Waist Hip Ratio 1.16 1.02 1.32 0.029 1012 1.2 1 1.45 0.049 994 Glucose (mg/dL) 1.04 0.92 1.17 0.512 1020 1.16 0.98 1.38 0.089 1002 Glycated Hemoglobin (%) 1.09 0.97 1.22 0.169 1016 1.23 1.06 1.43 0.008 998 Diabetes duration (years) 1.54 1.36 1.74 5.62 x 10-12 1005 1.7 1.41 2.04 2.22 x 10-8 987 Coronary Artery Calcified 2.01 1.7 2.37 1.11 x 10-16 968 2.47 1.91 3.21 8.48 x 10-12 952 Plaque (mass score) Carotid Intima Media 1.38 1.25 1.53 5.20 x 10-10 928 1.42 1.24 1.63 911 7.32 x 10-7 Thickness (mm) Total Cholesterol (mg/dL) 1.01 0.91 1.13 0.796 1003 1 0.86 1.17 1 986 HDL (mg/dL) 0.93 0.82 1.05 0.228 1003 0.87 0.72 1.04 0.133 986 Triglycerides (mg/dL) 1.01 0.9 1.14 0.841 1003 1.12 0.95 1.33 0.172 986 LDL (mg/dL) 1.05 0.94 1.17 0.413 928 0.93 0.79 1.1 0.396 912 Pulse Pressure (mmHg) 1.3 1.16 1.46 8.24 x 10-6 1019 1.35 1.13 1.61 7.71 x 10-4 1001 Mean Arterial Pressure 0.9 0.79 1.04 0.149 1019 0.95 0.76 1.2 0.685 1001 (mmHg) Estimated Glomerular 0.63 0.55 0.72 3.60 x 10-11 1021 0.63 0.51 0.78 2.75 x 10-5 1003 Filtration Rate (ml/min/1.73m2)

269

Urine Albumin:creatinine 1.56 1.42 1.72 < 1 x10-16 1000 1.76 1.54 2.01 < 1 x10-16 982 Ratio (mg/g) QT Interval (ms) 1.02 0.91 1.16 0.709 985 1.13 0.93 1.38 0.212 968 High Blood Pressure 1.72 1.28 2.3 2.63 x 10-4 1022 2.28 1.41 3.69 7.86 x 10-4 1004 Medications (%) Statin Use (%) 0.88 0.7 1.11 0.293 1020 1.02 0.72 1.46 0.91 1002 Oral Hypoglycemic 1.04 0.79 1.38 0.769 1022 1.02 0.67 1.54 0.942 1004 Medications (%) Insulin Use (%) 1.71 1.35 2.17 8.51 x 10-6 1022 1.84 1.32 2.58 3.55 x 10-4 1004

Relationships assessed using univariate Cox proportional hazards models. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

270

Table 3: Models selected for all-cause and cardiovascular disease (CVD) mortality in European Americans with type 2 diabetes.

Trait Hazard Ratio 95% Hazard Ratio Confidence Interval p-value All-cause mortality Age 1.72 1.46 2.02 6.46 x 10-11 Female Sex 0.93 0.71 1.21 0.565 Urine Albumin:creatinine Ratio 1.37 1.23 1.53 2.23 x 10-8 Coronary Artery Calcified Plaque 1.48 1.20 1.81 2.03 x 10-4 Current Smoking 1.79 1.30 2.46 3.38 x 10-4 Insulin Use 1.50 1.16 1.93 0.002 Educational Attainment 0.76 0.65 0.91 0.002 CVD mortality Age 1.57 1.18 2.08 0.002 Female Sex 0.94 0.63 1.40 0.756 Urine Albumin:creatinine Ratio 1.51 1.31 1.74 1.99 x 10-8 Coronary Artery Calcified Plaque 1.71 1.23 2.38 0.001 Triglycerides 1.28 1.05 1.56 0.017 History of Cardiovascular Disease 1.59 1.03 2.46 0.036

Models were selected using backward elimination, forward selection, and stepwise selection. Age and sex were forced into all models. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

271

Table 4: Average area under the receiver operating characteristic curve (AUC) for the all-cause and CVD mortality models.

Data Outcome Model AUC 95% Confidence Interval p-value Random 1 All-cause mortality Basic 0.697 (0.644, 0.749) 0.011 All-cause mortality Full 0.759 (0.710, 0.807) CVD mortality Basic 0.658 (0.590, 0.726) 0.002 CVD mortality Full 0.764 (0.705, 0.823)

Random 2 All-cause mortality Basic 0.693 (0.638, 0.747) 0.0001 All-cause mortality Full 0.777 (0.732, 0.823) CVD mortality Basic 0.614 (0.534, 0.694) 0.0003 CVD mortality Full 0.725 (0.661, 0.789)

Full Cohort All-cause mortality Basic 0.699 (0.661, 0.737) 6.50 x 10-9 All-cause mortality Full 0.788 (0.756, 0.821) CVD mortality Basic 0.652 (0.600, 0.704) 4.14 x 10-7 CVD mortality Full 0.764 (0.721, 0.807)

These models were derived using a two-fold cross-validation approach in each randomly selected dataset and in the full cohort. Basic models included age and sex only. Full models for all-cause mortality included age, sex, coronary artery calcified plaque, urine albumin:creatinine ratio, diabetes duration, current smoking, educational attainment, insulin use, and waist hip ratio. Full models for CVD mortality included age, sex, coronary artery calcified plaque, urine albumin:creatinine ratio, history of CVD events, and diabetes duration. P-values for comparing the predictive power of the basic model with the full model are listed.

272

Table 5: Addition of C-reactive protein and biventricular volume to model selected for all-cause mortality.

95% 95% Hazard Hazard Hazard Hazard 95% Hazard Ratio Hazard Ratio Trait Ratio p-value Trait p-value Trait p-value Ratio Ratio Confidence Interval Ratio Confidence Confidence Interval Interval Age 1.67 1.42 1.96 3.68 x 10-10 Age 1.80 1.52 2.13 6.54 x 10-12 Age 1.81 1.53 2.15 9.65 x 10-12 Female Sex 0.92 0.70 1.22 0.581 Female Sex 0.86 0.62 1.19 0.356 Female Sex 1.24 0.85 1.81 0.269 Coronary Coronary Coronary Artery Artery Artery 1.50 1.24 1.81 3.32 x 10-5 1.44 1.17 1.77 0.001 1.45 1.15 1.82 0.001 Calcified Calcified Calcified Plaque Plaque Plaque Urine Urine Urine Albumin: Albumin: Albumin: 1.35 1.21 1.51 1.10 x 10-7 1.35 1.20 1.52 7.85 x 10-7 1.32 1.17 1.49 8.08 x 10-6 creatinine creatinine creatinine Ratio Ratio Ratio Diabetes Diabetes Diabetes 1.10 0.96 1.26 0.173 1.05 0.91 1.21 0.505 Duration 1.09 0.94 1.25 0.261 Duration Duration

Current Current Current 1.70 1.25 2.31 7.38 x 10-4 1.63 1.16 2.28 0.005 2.00 1.42 2.83 8.46 x 10-5 Smoking Smoking Smoking Educational Educational Educational 0.79 0.67 0.93 0.004 0.79 0.66 0.94 0.010 0.78 0.65 0.94 0.008 Attainment Attainment Attainment

Insulin Use 1.40 1.07 1.84 0.015 Insulin Use 1.49 1.11 2.00 0.008 Insulin Use 1.39 1.03 1.86 0.029 Waist Hip Waist Hip Waist Hip 1.01 0.86 1.19 0.913 0.97 0.81 1.17 0.765 1.01 0.86 1.20 0.868 Ratio Ratio Ratio C-reactive Biventricular 1.30 1.11 1.52 0.001 1.21 1.00 1.48 0.053 Protein Volume

The all-cause mortality model was selected using a two-fold cross-validation approach in European American participants with type 2 diabetes. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

273

Table 6: Addition of C-reactive protein and biventricular volume to model selected for cardiovascular disease mortality.

95% Hazard 95% Hazard 95% Hazard Hazard Ratio Hazard Ratio Hazard Ratio Trait p-value Trait p-value Trait p-value Ratio Confidence Ratio Confidence Ratio Confidence Interval Interval Interval Age 1.36 1.08 1.73 0.010 Age 1.63 1.27 2.10 1.20 x 10-4 Age 1.75 1.36 2.27 1.66 x 10-5 Female Sex 0.98 0.68 1.42 0.929 Female Sex 1.02 0.67 1.55 0.942 Female Sex 1.52 0.90 2.55 0.116 Coronary Coronary Coronary Artery 1.67 1.24 2.26 7.18 x 10-4 Artery Calcified 1.60 1.15 2.23 0.005 Artery Calcified 1.40 0.98 2.00 0.067 Calcified Plaque Plaque Plaque Urine Albumin: Urine Albumin: Urine Albumin: 1.52 1.32 1.75 7.59 x 10-9 1.57 1.35 1.83 7.31 x 10-9 1.45 1.24 1.70 3.29 x 10-6 creatinine creatinine Ratio creatinine Ratio Ratio History of History of History of Cardiovascular 1.69 1.11 2.57 0.014 Cardiovascular 1.52 0.95 2.43 0.084 Cardiovascular 1.55 0.98 2.45 0.063 Disease Disease Disease Diabetes Diabetes Diabetes 1.20 0.98 1.46 0.082 1.04 0.84 1.28 0.736 1.09 0.88 1.36 0.435 Duration Duration Duration C-reactive Biventricular 1.19 0.93 1.51 0.167 1.52 1.16 1.98 0.002 protein Volume

The cardiovascular disease mortality model was selected using a two-fold cross-validation approach in European Americans with type 2 diabetes. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

274

Additional File 1: Associations of demographic and clinical factors with all-cause and cardiovascular disease mortality in two randomly selected datasets from European Americans with type 2 diabetes.

Random dataset 1 Random dataset 2

Cardiovascular disease Cardiovascular disease All-cause mortality All-cause mortality mortality mortality Trait Hazard Hazard Hazard Hazard p-value p-value p-value p-value Ratio Ratio Ratio Ratio Age (years) 1.90 7.09 x 10-11 1.96 2.78 x 10-6 1.81 2.44 x 10-9 1.50 0.007 Female Sex (%) 0.76 0.091 0.65 0.057 0.59 0.002 0.50 0.015 Current Smoking 1.20 0.354 0.68 0.277 1.75 0.002 1.57 0.122 (%) Past Smoking 1.02 0.882 1.26 0.317 0.95 0.722 1.00 0.997 (%) History of Cardiovascular 1.78 3.69 x 10-4 3.43 2.50 x 10-6 2.04 4.48 x 10-5 2.94 7.09 x 10-5 Disease (%) Educational Attainment (3 0.71 0.006 0.69 0.029 0.66 5.58 x 10-4 0.84 0.362 levels) Body Mass Index 0.81 0.017 0.78 0.056 0.94 0.491 1.03 0.809 (kg/m2) Waist Hip Ratio 1.23 0.015 1.25 0.055 1.05 0.626 1.11 0.460 Glucose (mg/dL) 1.04 0.680 1.14 0.319 1.04 0.611 1.19 0.117 Glycated 1.06 0.492 1.16 0.163 1.12 0.200 1.32 0.010 Hemoglobin (%)

275

Diabetes 1.55 4.60 x 10-7 1.69 2.79 x 10-5 1.53 2.84 x 10-6 1.70 1.83 x 10-4 Duration (years) Coronary Artery Calcified Plaque 1.82 4.46 x 10-7 2.59 8.60 x 10-8 2.23 1.12 x 10-11 2.38 1.84 x 10-5 (mass score) Carotid Intima Media Thickness 1.42 2.16 x 10-6 1.54 7.72 x 10-6 1.35 1.05 x 10-5 1.33 0.003 (mm)

Total Cholesterol 0.98 0.824 0.99 0.953 1.05 0.508 1.01 0.922 (mg/dL) HDL (mg/dL) 1.00 0.962 0.92 0.479 0.83 0.057 0.80 0.123 Triglycerides 0.96 0.618 1.09 0.457 1.08 0.365 1.16 0.198 (mg/dL) LDL (mg/dL) 1.02 0.783 0.93 0.564 1.07 0.340 0.93 0.532 Pulse pressure 1.37 1.45 x 10-5 1.41 9.17 x 10-4 1.21 0.042 1.26 0.132 (mmHg) Mean Arterial 0.91 0.369 0.92 0.601 0.89 0.240 0.99 0.944 Pressure (mmHg) Estimated Glomerular 0.65 9.16 x 10-6 0.57 1.55 x 10-4 0.61 3.82 x 10-7 0.71 0.032 Filtration Rate (ml/min/1.73m2) Urine Albumin: creatinine Ratio 1.50 1.96 x 10-10 1.73 1.57 x 10-12 1.64 1.33 x 10-10 1.78 2.89 x 10-6 (mg/g) QT interval (ms) 1.00 0.976 1.14 0.285 1.04 0.716 1.10 0.552 High Blood 1.66 0.014 2.11 0.021 1.81 0.004 2.57 0.011

276

Pressure Medications (%) Statin Use (%) 0.87 0.392 1.01 0.961 0.90 0.526 1.03 0.914 Oral Hypoglycemic 0.91 0.590 0.88 0.609 1.28 0.308 1.32 0.464 Medications (%) Insulin Use (%) 1.93 4.04 x 10-5 2.31 1.22 x 10-4 1.46 0.040 1.32 0.340 Associations with all-cause and cardiovascular disease mortality were assessed using univariate Cox proportional hazards models. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

For random dataset 1, coronary artery calcified plaque, pulse pressure, estimated glomerular filtration rate, urine albumin:creatinine ratio, diabetes duration, body mass index, high blood pressure medications, insulin use, history of cardiovascular disease, educational attainment, waist hip ratio, and carotid intima media thickness were included in model selection for all-cause mortality; for cardiovascular disease mortality, glycated hemoglobin, coronary artery calcified plaque, pulse pressure, estimated glomerular filtration rate, urine albumin:creatinine ratio, diabetes duration, body mass index, high blood pressure medications, insulin use, history of cardiovascular disease, educational attainment, waist hip ratio, and carotid intima media thickness were included in model selection.

For random dataset 2, glycated hemoglobin, coronary artery calcified plaque, pulse pressure, HDL, estimated glomerular filtration rate, urine albumin:creatinine ratio, diabetes duration, high blood pressure medications, insulin use, current smoking, history of cardiovascular disease, educational attainment, carotid intima media thickness, and mean arterial pressure were included in model selection for all-cause mortality; for cardiovascular disease mortality, fasting glucose, glycated hemoglobin, coronary artery calcified plaque, pulse pressure, HDL, triglycerides, estimated glomerular filtration rate, urine albumin:creatinine ratio, diabetes duration, high blood pressure medications, current smoking, history of cardiovascular disease, and carotid intima media thickness were included in model selection.

Age and sex were included in all models.

277

Additional File 2: Model selected for all-cause mortality using backward elimination, forward selection, and stepwise selection in two randomly selected datasets from European American participants with type 2 diabetes.

Random dataset 1 Random dataset 2 95% Hazard 95% Hazard Hazard Hazard Ratio Trait Ratio Confidence p-value Trait p-value Ratio Ratio Confidence Interval Interval Age 1.96 1.60 2.40 7.50 x 10-11 Age 1.37 1.07 1.75 0.013 Female Sex Female Sex 0.91 0.65 1.27 0.579 0.97 0.65 1.45 0.893

Urine Coronary Artery Albumin:creatinine 1.39 1.21 1.61 6.68 x 10-6 1.83 1.39 2.40 1.73 x 10-5 Calcified Plaque Ratio Urine Insulin Use 1.65 1.17 2.34 0.005 Albumin:creatinine 1.51 1.26 1.81 1.05 x 10-5

Ratio Waist Hip Ratio 1.22 1.00 1.49 0.054 Diabetes Duration 1.25 1.02 1.53 0.033 Current Smoking 1.67 1.07 2.60 0.024

Educational 0.61 0.48 0.78 1.01 x 10-4 Attainment

Age and sex were forced into all models. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

278

Additional File 3: Model selected for cardiovascular disease mortality using backward elimination, forward selection, and stepwise selection in two randomly selected datasets from European Americans with type 2 diabetes.

Random dataset 1 Random dataset 2 95% Hazard Hazard Hazard 95% Hazard Ratio Trait Ratio Confidence p-value Trait p-value Ratio Ratio Confidence Interval Interval Age 1.73 1.23 2.42 0.002 Age 1.03 0.71 1.50 0.874 Female Sex Female Sex 1.10 0.66 1.84 0.711 0.85 0.46 1.58 0.607

Coronary Coronary Artery Artery 1.64 1.06 2.54 0.027 2.00 1.31 3.03 0.001 Calcified Calcified Plaque Plaque Urine Urine Albumin: Albumin: 1.59 1.35 1.86 1.74 x 10-8 1.59 1.20 2.10 0.001 creatinine creatinine Ratio Ratio History of Diabetes Cardiovascular 1.83 1.04 3.23 0.036 1.50 1.09 2.07 0.013 Duration Disease

Age and sex were forced into all models. Hazards ratios (HRs) are for a one standard deviation change in the predictor (continuous variables) or change in group assignment (dichotomous variables). For medication use HRs, the HRs are for risk of mortality among those individuals using the given medication class.

279

Chapter 8

Cross-sectional analysis of calcium intake for associations with vascular calcification and mortality in individuals with type 2 diabetes from the Diabetes Heart Study

Laura M Raffield, Subhashish Agarwal, Amanda J Cox, Fang-Chi Hsu, J Jeffrey Carr, Barry I

Freedman, Jianzhao Xu, Donald W Bowden, Mara Z Vitolins

This manuscript was published in the October 2014 issue of American Journal of Clinical Nutrition. The reference for this manuscript is as follows: Raffield, L. M., Agarwal, S., Cox A.J., Hsu, F., Carr, J.J., Freedman, B.I., Xu, J., Bowden, D.W., Vitolins, M.Z. Cross-sectional Analysis of Calcium Intake for Effects on Vascular Calcification and Mortality in Individuals with Type 2 Diabetes from the Diabetes Heart Study. American Journal of Clinical Nutrition, 2014. 100(4):1029-35. doi: 10.3945/ajcn.114.090365. Epub 2014 Aug 6. PubMed PMID: 25099552; PubMed Central PMCID: PMC4163793.

280

Abstract

Background:

Use of calcium supplements to prevent declines in bone mineral density and fractures is widespread in the United States, so reports of elevated cardiovascular disease (CVD) risk in users of calcium supplements are a major public health concern. Any elevation in CVD risk with calcium supplement use would be of particular concern in individuals with type 2 diabetes (T2D), given the increased risk of both CVD and fractures observed in this population.

Objective:

In this study, we examined associations between calcium intake from diet and supplements and measures of subclinical CVD (calcified plaque in the coronary artery, carotid artery, and abdominal aorta) and mortality in individuals affected by T2D.

Design:

We performed a cross-sectional analysis in individuals affected by T2D from the family-based

Diabetes Heart Study (DHS) (n=720).

Results:

We observed no significant associations of calcium from diet or supplements with any of our measures of calcified plaque, and no greater mortality risk was observed with increased calcium intake. Calcium supplement use was instead modestly associated with reduced all-cause mortality in women (hazard ratio 0.62, 95% confidence interval (0.42, 0.92), p=0.017).

Conclusions:

Our results do not support a substantial association between calcium intake from diet or supplements and CVD risk in individuals with T2D.

281

Introduction

Patients with type 2 diabetes (T2D) have significantly increased risk for cardiovascular disease (CVD); mortality rates from heart disease are at least two-fold higher in individuals with

T2D, with CVD accounting for as much as 65% of all-cause mortality in T2D patients (Lloyd-

Jones, Adams et al. 2010). Individuals with T2D also have an elevated risk of fracture

(Janghorbani, Dam et al. 2007), though T2D is not thought to be associated with declines in bone mineral density (BMD) (Register, Lenchik et al. 2006; Vestergaard 2007; Ma, Oei et al. 2012).

Calcium supplementation may increase BMD and modestly reduce risk of fracture (Prentice,

Pettinger et al. 2013). Use of calcium supplements is widespread, particularly in postmenopausal women (Gahche, Bailey et al. 2011). However, several recent studies have raised concerns that calcium supplementation may lead to an elevated risk of CVD, in particular higher risk of myocardial infarction (MI) (Bolland, Barber et al. 2008; Bolland, Avenell et al. 2010; Bolland,

Grey et al. 2011; Li, Kaaks et al. 2012; Xiao, Murphy et al. 2013).The elevated risk of both fractures and CVD in individuals with T2D makes potential increases in CVD risk in those taking calcium supplements of particular concern; however, potential associations between calcium intake from diet and supplements and CVD risk have not previously been examined in a cohort of

T2D patients.

It has been suggested that the potential negative impacts of calcium supplements on CVD risk may be mediated by vascular calcification, an important measure of subclinical CVD risk whose regulation has many links to osteogenesis and bone physiology (Bolland, Grey et al. 2011;

Thompson and Towler 2012). However, a recent report found no impact of calcium intake from diet and supplements on coronary artery calcified plaque (CAC) (Samelson, Booth et al. 2012).

Here, we evaluated the relationships between calcium intake from diet and supplements and subclinical CVD as assessed by CAC, specifically focusing on patients with T2D at elevated

282

CVD risk. In addition, the analysis was extended to evaluate potential associations of calcium intake with calcified plaque in multiple vascular beds and with all-cause and CVD mortality.

Subjects and Methods

Study Design and Sample

Diabetes Heart Study (DHS) participants were recruited from 1998 through 2005 from outpatient internal medicine and endocrinology clinics and from the community in western North

Carolina. Siblings concordant for T2D without advanced renal insufficiency were recruited. Ascertainment and recruitment have been described in detail previously (Wagenknecht,

Bowden et al. 2001; Lange, Bowden et al. 2002; Bowden, Rudock et al. 2006; Bowden, Cox et al.

2010). T2D was defined as diabetes developing after the age of 35 years treated with changes in diet and exercise and/or oral agents in the absence of initial treatment solely with insulin and without historical evidence of ketoacidosis. Diabetes diagnosis was confirmed by measurement of fasting glucose and glycated hemoglobin (HbA1C). The analyses described here include all self- described European American individuals with T2D from the DHS with food frequency questionnaire data, in total 720 individuals from 339 DHS families (Supplementary Figure 1).

Study protocols were approved by the Institutional Review Board at Wake Forest School of Medicine, and all participants provided written informed consent. Participant examinations were conducted in the General Clinical Research Center of the Wake Forest Baptist Medical

Center. Examinations included interviews for medical history and health behaviors, anthropometric measures, assessment of resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses, and spot urine collection. Standard laboratory analyses were performed, including assessment of fasting glucose, HbA1C, total cholesterol, HDL, and triglycerides. LDL concentration was calculated using the Friedewald equation; LDL concentrations were considered valid for participants whose triglycerides were less than 796

283

mg/dL (n=4 individuals excluded based on triglyceride levels). Individuals were considered hypertensive if they were prescribed anti-hypertensive medication or had blood pressure measurements exceeding 140 mmHg (systolic) or 90 mmHg (diastolic). Measures of calcified atherosclerotic plaque by computed tomography (CT) included coronary artery calcified plaque

(CAC) as the sum of the left main, left anterior descending, circumflex, posterior descending, and right coronary arteries, carotid artery calcified plaque (CarCP) as the sum of the right and left common, bulb, and internal, and abdominal aorto-iliac (AACP) as the sum of the infrarenal, right, and left common iliac segments. CT scans were performed on multi-detector CT scanners with cardiac gating in chest and helical acquisitions in abdomen. Calcium scores were measured as previously described and validated (Carr, Crouse et al. 2000; Carr, Nelson et al. 2005). Intake of calcium and vitamin D from diet and supplements was determined using a self-administered

Block food frequency questionnaire (Block, Woods et al. 1990).

Mortality was assessed for all DHS participants using the National Social Security Death

Index maintained by the United States Social Security Administration. For deceased participants, length of follow-up was determined from the date of initial study visit to date of death. For all other participants, the length of follow-up was determined from the date of the initial study visit to the end of 2012. Copies of death certificates were obtained from county Vital Records Offices to determine cause of death. Cause of death was categorized based on these death certificates as

CVD mortality (MI, congestive heart failure, cardiac arrhythmia, sudden cardiac death, peripheral vascular disease, and stroke) or as mortality from cancer, infection, end-stage renal disease, accidental, or other causes (including obstructive pulmonary disease, pulmonary fibrosis, liver failure and Alzheimer’s disease). Association with mortality was assessed for both all-cause mortality and CVD mortality.

284

Statistical Analysis

For statistical analyses, continuous variables were transformed as necessary to approximate normality. For our analyses of vascular calcified plaque, we used the natural logarithm of (CAC+1) and (CarCP+1) and the square root of (AACP+10). Dietary calcium and vitamin D intake was adjusted for total energy intake using a residual method as suggested by

Willett et al. (Willett, Howe et al. 1997). Total calcium intake and dietary calcium intake were considered as ordinal (four quartiles) variables, as was calcium intake from supplements (divided into mg intake ranges (0 mg, 1-500 mg, >500 mg) similar to those used by Samelson et al.)

(Samelson, Booth et al. 2012). Quartile 1 for energy-adjusted total calcium intake for women ranged from 216-614.2 mg; quartile 2 ranged from 614.3-851 mg; quartile 3 ranged from 858-

2098 mg; quartile 4 ranged from 2106-5350 mg. Quartile 1 for energy-adjusted dietary calcium intake for women ranged from 216-484 mg; quartile 2 ranged from 487-635 mg; quartile 3 ranged from 638-834 mg; quartile 4 ranged from 837-3090 mg. Quartile 1 for energy-adjusted total calcium intake for men ranged from 280-600 mg; quartile 2 ranged from 601-783 mg; quartile 3 ranged from 784-1039 mg; quartile 4 ranged from 1044-3212 mg. Quartile 1 for energy-adjusted dietary calcium intake for men ranged from 263-531 mg; quartile 2 ranged from 534-681.8 mg; quartile 3 ranged from 682.1-829 mg; quartile 4 ranged from 831-1601 mg.

Relationships between the calcium intake ranges and CAC, CarCP, and AACP were examined using marginal models with generalized estimating equations. The models account for familial correlation using a sandwich estimator of the variance under exchangeable correlation.

Relationships between the calcium intake ranges and both all-cause and CVD mortality were examined using Cox proportional hazards models with sandwich-based variance estimation due to the inclusion of related individuals in this study. Associations were adjusted for covariates including age, total energy intake (kcal), BMI (kg/m2), smoking (never, past, or current), any alcohol consumption, energy-adjusted total vitamin D intake from diet and supplements (IU), use

285

of lipid-lowering medications, calcium intake from supplements (mg), energy-adjusted dietary calcium intake (mg), menopause status, and estrogen use as indicated, similar to the analysis performed by Samelson et al (Samelson, Booth et al. 2012). Individuals with missing data for outcome variables or covariates were excluded from the relevant models (Supplementary Figure

1). Of the 360 women included in these analyses, 18 had missing CAC, 26 had missing CarCP,

100 had missing AACP, 1 had missing smoking status, 1 had missing mortality status, and 30 had missing menopause status. Of the 360 men included in these analyses, 18 had missing CAC, 17 had missing CarCP, 88 had missing AACP, and 1 had missing smoking status. Data was available for all participants for all other outcomes and covariates assessed. All analyses were performed in

SAS 9.3 (SAS Institute, Cary, NC).

Results

The goal of this study was to analyze associations between calcium intake from diet and supplements and CVD risk in participants with T2D from the DHS. We assessed the relationships between calcium intake and both subclinical CVD and mortality risk.

The clinical characteristics of the 720 European American individuals with T2D included in this study stratified by both total calcium intake quartile and by gender are summarized in

Table 1. As would be expected for a cohort affected by T2D, most individuals are overweight or obese, and the cohort has high rates of dyslipidemia and hypertension, a high burden of vascular calcified plaque, and a high rate of prior CVD events. Few measures differed across energy- adjusted total calcium intake quartiles, though age, prevalence of alcohol consumption, and use of lipid lowering medications (mainly statins) did differ significantly in men. These factors were all considered as covariates in fully adjusted models for vascular calcified plaque and mortality. All models were stratified by gender.

286

We first assessed the relationships between calcium intake from diet and supplements and measures of vascular calcified plaque. In minimally adjusted models accounting for age and total energy intake only, neither total energy-adjusted calcium intake nor calcium intake from diet or supplements was significantly associated with any vascular calcified plaque measure (CAC,

CarCP, and AACP) (Supplementary Table 1). Similarly, in models adjusted for age, total energy intake, BMI, smoking, alcohol consumption, energy-adjusted total vitamin D intake from diet and supplements, use of lipid-lowering medications, calcium intake from supplements or dietary calcium intake where appropriate, and, in women, menopause status and estrogen use, no association between total energy-adjusted calcium intake or calcium intake from diet or supplements and any measure of vascular calcified plaque was observed (Table 2).

We also assessed associations between calcium intake and both all-cause and CVD mortality (Table 3). We analyzed both minimally adjusted models, adjusted only for age and total energy intake (Model 1), and more fully adjusted models, additionally adjusted for BMI, smoking, alcohol consumption, energy-adjusted total vitamin D intake from diet and supplements, use of lipid-lowering medications, calcium intake from supplements or dietary calcium intake where appropriate, and, in women, menopause status and estrogen use (Model 2).

For Model 1, we observed a modest protective effect of increased supplemental calcium consumption on all-cause mortality in women (hazard ratio (HR) for all-cause mortality for each increase in calcium supplement use tertile: 0.61, 95% confidence interval (CI): 0.46, 0.83, p=0.001). For CVD mortality, modest protective effects were also observed in both women (HR for CVD mortality: 0.59, 95% CI: 0.37, 0.94, p=0.026) and men (HR for CVD mortality: 0.65,

95% CI: 0.43, 0.98, p=0.042). These associations are attenuated in the more fully adjusted Model

2, with only the association of calcium supplementation with reduced all-cause mortality in women remaining nominally significant (HR for all-cause mortality: 0.62, 95% CI: 0.42, 0.92, p=0.017).

287

Discussion

Recent studies have raised concerns that calcium supplementation may have the unintended negative consequence of increasing CVD risk (Bolland, Barber et al. 2008; Bolland,

Avenell et al. 2010; Bolland, Grey et al. 2011; Li, Kaaks et al. 2012; Xiao, Murphy et al. 2013).

Any increase of CVD risk in patients with T2D would be of particular concern due to their already elevated risk of CVD (Lloyd-Jones, Adams et al. 2010); to our knowledge, this is the first analysis of the potential negative CVD impacts of calcium intake which focused specifically on individuals affected by T2D. CAC, a measure of subclinical CVD, has been shown in the DHS and other cohorts to be a strong independent predictor of CVD events and mortality (Raggi, Shaw et al. 2004; Detrano, Guerci et al. 2008; Folsom, Kronmal et al. 2008; Elias-Smale, Proenca et al.

2010; Erbel, Mohlenkamp et al. 2010; Polonsky, McClelland et al. 2010; Agarwal, Morgan et al.

2011; Agarwal, Cox et al. 2013), with individuals affected by diabetes tending to have higher

CAC (Hoff, Quinn et al. 2003). Statistical power is increased for the analysis of continuous traits such as vascular calcified plaque versus event data, and it has also been proposed that associations of calcium intake with CVD are mediated through vascular calcification (Bolland,

Grey et al. 2011), highlighting the suitability of our vascular calcified plaque measures for assessment of potential negative CVD impacts of calcium intake. We replicated the lack of impact of calcium from diet and supplements on CAC previously observed in a general population cohort (Samelson, Booth et al. 2012) in patients with T2D. Additionally, we did not find any evidence of negative CVD impacts of calcium intake in our analyses of CarCP, AACP, and mortality. CAC is the most commonly assessed form of vascular calcified plaque (Manson,

Allison et al. 2010; Wang, Bolland et al. 2010; Samelson, Booth et al. 2012); however, different risk factors may contribute to calcification in different vascular beds (Wagenknecht, Langefeld et al. 2007; Criqui, Kamineni et al. 2010), highlighting the importance of our additional analysis of

AACP and CarCP.

288

Prior analyses of CAC, notably the analysis in a largely non-diabetic population from the

Framingham Offspring Study, have also shown no significant impact of calcium intake on CAC burden (Samelson, Booth et al. 2012). Similarly, analysis of CAC in 754 Women’s Health

Initiative (WHI) participants (Manson, Allison et al. 2010) and in another small cohort of men

(Wang, Bolland et al. 2010) showed no association with prior randomization to calcium supplement use. Additionally, analyses of the effect of calcium supplementation on CVD events have had mixed results. Analysis of the WHI calcium and vitamin D supplementation trial data by the WHI investigators failed to support the modestly higher CVD risk reported in a meta-analysis using the WHI data by Bolland et al. (Bolland, Grey et al. 2011; Prentice, Pettinger et al. 2013).

A 2010 meta-analysis of several randomized trials also found no significant change in risk of

CVD events for those randomized to calcium supplementation (Wang, Manson et al. 2010).

Initial concerns about negative CVD impacts of calcium supplements were raised by a randomized clinical trial from New Zealand of calcium supplementation (1000 mg/day) which reported a modest increase in MI risk in women taking calcium supplements (Bolland, Barber et al. 2008). This increase in MI risk was replicated in subsequent meta-analyses (Bolland, Avenell et al. 2010; Bolland, Grey et al. 2011). A similar increase in MI risk with calcium supplement use was also seen in a recent prospective study of German men and women (Li, Kaaks et al.

2012), with a study from the US also finding an increased risk of CVD death in men using calcium supplements (Xiao, Murphy et al. 2013). However, these reports did have limitations.

For example, in the initial report by Bolland et al., the reported association was no longer statistically significant when events not self-reported by participants, but instead sourced through a national hospital admissions database and death certificate review, were included in an analysis with verified self-reported events (Bolland, Barber et al. 2008). In both the original trial report

(Bolland, Barber et al. 2008) and the larger subsequent meta-analyses (Bolland, Avenell et al.

2010; Bolland, Grey et al. 2011) by Bolland et al., studies included were originally designed to

289

assess effects of calcium on bone density and fracture, with CVD outcomes considered only in secondary analyses. Further problems with the reports that calcium supplements increase the risk of CVD events have been reviewed previously (Hennekens and Barice 2011; Nordin, Lewis et al.

2011).

In this study we did not observe any negative CVD impacts of differing calcium intake from diet and supplements, in contrast to some previous reports (Bolland, Barber et al. 2008;

Bolland, Avenell et al. 2010; Bolland, Grey et al. 2011; Li, Kaaks et al. 2012; Xiao, Murphy et al.

2013). Instead, calcium supplement use was associated with lower all-cause mortality risk in women. The association may be due to selection bias in terms of which participants in our cross- sectional analysis chose to take calcium supplements. Prior research has shown that individuals taking dietary supplements tend to have higher socioeconomic status, education levels, and physical activity (Foote, Murphy et al. 2003; Harrison, Holt et al. 2004). However, our results were essentially unchanged upon further adjustment for educational attainment (HR for all-cause mortality: 0.63, 95% CI: 0.43, 0.93, p=0.02), suggesting other factors may be responsible.

Positive impacts of calcium supplements in women, notably increasing BMD and preventing fractures, may lead to the observed protective effect (Prentice, Pettinger et al. 2013). Our finding of a slight protective effect for all-cause mortality for women taking calcium supplements is also not unprecedented. A large cross-sectional study found a slightly lower risk of CVD and all-cause mortality in women taking calcium supplements in models adjusted for demographic and health status covariates (Mursu, Robien et al. 2011), suggesting a protective effect independent of socioeconomic status, with a Canadian study also finding reduced mortality risk in women using calcium supplements up to 1000 mg/day (Langsetmo, Berger et al. 2013).

Strengths of the current analysis include the study population with T2D, a group with elevated risk of CVD and fracture, comprehensive data for analysis of vascular calcified plaque in multiple beds, and long-term mortality follow-up of 9.4 ± 5.0 years. Limitations of our cross-

290

sectional analysis include the concurrent measurement of calcium intake from diet and supplements and vascular calcification; if participant’s dietary habits or supplement use had recently changed, impacts may not yet have been observed on vascular calcified plaque.

However, our failure to find higher subsequent mortality risk supports our baseline finding of no association of calcium intake with vascular calcification, as vascular calcified plaque has been shown to be a strong predictor of mortality risk in our cohort (Agarwal, Morgan et al. 2011;

Agarwal, Cox et al. 2013).

In our analysis of calcium intake from diet and supplements in patients with T2D, we did not observe any significant associations between calcium intake and multiple measures of vascular calcified plaque, including CAC, CarCP, and AACP. We also observed no higher risk of all-cause or CVD mortality in individuals taking calcium supplements. In fact, we observed a modest association of increased use of calcium supplements with lower risk of all-cause mortality in women. Our data does not support a significant association between differing calcium intake from diet or supplements and CVD risk in individuals with T2D.

Acknowledgements

LMR perfomed the statistical analysis and wrote the manuscript; SA helped design the research;

AJC compiled the mortality data and contributed to the statistical analysis; FCH contributed to the statistical analysis; JJC was involved in the initial design of the Diabetes Heart Study and contributed to patient ascertainment and clinical evaluation; BIF was involved in the initial design of the Diabetes Heart Study and contributed to patient ascertainment and clinical evaluation; JX contributed to data management and the statistical analysis; DWB leads the Diabetes Heart Study and was involved in the initial design of the study, patient ascertainment, and clinical evaluation;

MZV helped design the research and had primary responsibility for final content. All authors read and edited the manuscript and approved the final version. The authors declare no conflicts of

291

interest in relation to this work. The authors thank the other investigators, the staff, and the participants of the DHS study for their valuable contributions.

292

Table 1: Demographic characteristics of Diabetes Heart Study participants with type 2 diabetes by daily energy-adjusted total calcium intake quartiles. Relationships between calcium intake quartiles and the demographic variables were examined using marginal models with generalized estimating equations.

Energy-adjusted total calcium intake quartiles

Women Men

1 (n=90) 2 (n=90) 3 (n=90) 4 (n=90) p-trend 1 (n=90) 2 (n=90) 3 (n=90) 4 (n=90) p-trend

Median (IQR) 472 (150) 731 (81) 1078 (402) 2816 (441) 482 (127) 692 (96) 926 (153) 1261 (1376) (mg) (n=720)

Range (mg) 216-614 614-851 858-2098 2106-5350 280-600 601-783 784-1039 1044-3212 (n=720)

Age (n=720) 62.1 61.3 57.8 64.4 0.109 64.3 62.8 63.0 60.0 0.048

BMI (kg/m2) 32.4 34.4 36.0 31.1 0.660 30.6 31.2 31.3 31.8 0.241 (n=720)

Past smoker (%) 18.9 37.1 23.3 33.3 0.225 58.4 53.3 67.8 62.2 0.243 (n=718)

Current smoker 22.2 8.9 15.6 11.1 0.121 16.7 20.0 10.0 18.9 0.890 (%) (n=720)

Any alcohol 14.4 13.3 18.9 17.8 0.293 20.0 31.1 26.7 38.9 0.025 consumption (%)

293

(n=720)

Hypertension (%) 92.2 94.4 85.6 86.7 0.087 82.2 90.0 83.3 90.0 0.210 (n=720)

Heart attacks (%) 19.1 12.4 7.8 12.4 0.129 37.1 31.8 31.5 24.4 0.085 (n=713)

Prior Cardiovascular 40.0 30.0 18.9 35.6 0.239 63.3 56.7 55.6 48.9 0.095 Disease Events (%) (n=720)

All-Cause Mortality (%) 30.0 22.2 13.3 16.9 0.014 32.2 27.8 26.7 26.7 0.418 (n=719)

CVD Mortality 11.1 7.8 6.7 7.8 0.446 17.8 14.4 14.4 10.0 0.163 (%) (n=719)

Coronary Artery Calcification Score 1187 995 602 866 0.097 3459 3747 2418 3109 0.327 (n=684)

Carotid Artery Calcification Score 296 223 140 405 0.504 604 502 410 408 0.115 (n=676)

Abdominal Aorto- iliac Calcification 11796 9424 6395 10412 0.288 19490 16406 16860 16996 0.672 Score (n=532)

294

Systolic Blood Pressure (mm Hg) 137.8 141.3 136.8 141.1 0.459 136.9 141.0 137.5 136.8 0.630 (n=717)

Diastolic Blood Pressure (mm Hg) 72.5 71.2 72.7 69.9 0.211 71.4 74.5 73.6 74.2 0.177 (n=717)

Total cholesterol 191.7 201.0 199.9 187.9 0.599 171.2 172.9 180.0 179.3 0.124 (mg/dL) (n=716)

HDL cholesterol 46.0 46.5 46.0 48.4 0.211 37.5 38.3 37.9 39.2 0.193 (mg/dL) (n=716)

LDL cholesterol 107.2 109.7 108.6 103.5 0.526 98.4 95.6 99.9 100.5 0.563 (mg/dL) (n=663)

Triglycerides 202.3 215.8 233.0 197.2 0.985 182.4 203.3 212.5 213.7 0.095 (mg/dL) (n=716)

Hypertension medications (%) 73.3 82.2 73.3 83.3 0.320 74.4 84.4 73.3 76.7 0.817 (n=720)

Lipid-lowering medications (%) 45.6 48.9 38.9 47.8 0.882 61.1 52.2 50.0 40.0 0.005 (n=720)

Statins (%) 43.3 43.3 31.1 44.4 0.707 53.3 50.0 44.4 34.4 0.007 (n=720)

295

Aspirin (%) 54.4 61.1 51.1 66.7 0.252 61.1 62.2 60.0 57.8 0.586 (n=720)

Osteoporosis medications (%) 3.3 6.7 3.3 11.1 0.097 1.1 0.0 0.0 1.1 0.997 (n=720)

Postmenopausal 88.5 92.6 83.1 93.2 0.596 (%) (n=330) Estrogen (%) 23.3 25.6 25.6 28.9 0.457 (n=360)

296

Table 2: Associations between coronary, carotid, and abdominal aorto-iliac calcification (least squares adjusted mean and 95% confidence interval

(CI)) and total, dietary, and supplemental calcium intake. Analysis was performed using marginal models with generalized estimating equations.

Coronary Artery Calcification Carotid Artery Calcification Abdominal Aorto-iliac Calcification

Women Men (n=341) Women (n=303) Men (n=342) Women (n=237) Men (n=272) (n=311)

Energy-adjusted total calcium intake quartile

1 (216-614.2 mg women, 83.24 (67.68, 126.4 (104.5, 280-600 mg men) 5.19 (4.73, 5.66) 6.81 (6.28, 7.34) 3.25 (2.65, 3.85) 4.42 (3.71, 5.13) 98.80) 148.2)

2 (614.3-851 mg women, 78.32 (65.79, 114.9 (100.9, 601-783 mg men) 4.95 (4.43, 5.47) 7.11 (6.72, 7.50) 3.53 (3.01, 4.04) 4.45 (3.93, 4.98) 90.85) 129.0)

3 (858-2098 mg women, 81.03 (66.00, 106.9 (91.18, 784-1039 mg men) 4.80 (4.27, 5.33) 6.91 (6.54, 7.28) 2.59 (2.07, 3.11) 3.88 (3.35, 4.41) 96.07) 122.5)

4 (2106-5350 mg women, 68.69 (57.66, 97.88 (77.28, 1044-3212 mg men) 4.89 (4.40, 5.38) 6.89 (6.37, 7.40) 3.68 (3.21, 4.15) 4.74 (4.11, 5.37) 79.72) 118.5)

p-trend 0.478 0.942 0.436 0.641 0.159 0.137

Energy-adjusted dietary calcium intake quartile

1 (216-484 mg women, 80.54 (62.92, 106.7 (82.38, 263-531 mg men) 5.03 (4.45, 5.61) 6.71 (6.03, 7.39) 3.16 (2.57, 3.74) 4.94 (4.24, 5.65) 98.16) 131.0)

2 (487-635 mg women, 5.14 (4.64, 5.65) 7.04 (6.68, 7.40) 3.29 (2.77, 3.82) 4.31 (3.78, 4.84) 82.61 (69.58, 108.9 (94.34,

297

534-681.8 mg men) 95.63) 123.5)

3 (638-834 mg women, 63.39 (50.32, 108.0 (92.82, 682.1-829 mg men) 4.75 (4.25, 5.24) 6.90 (6.49, 7.30) 3.23 (2.68, 3.77) 4.18 (3.66, 4.71) 76.46) 123.2)

4 (837-3090 mg women, 79.46 (57.86, 124.5 (94.33, 831-1601 mg men) 4.85 (4.19, 5.51) 7.06 (6.20, 7.92) 3.39 (2.75, 4.03) 4.05 (3.16, 4.94) 101.1) 154.6)

p-trend 0.516 0.732 0.750 0.169 0.361 0.755

Supplemental calcium (mg)

0 81.21 (68.58, 120.5 (106.1, 4.96 (4.53, 5.39) 7.04 (6.67, 7.41) 3.54 (3.06, 4.01) 4.43 (3.93, 4.92) 93.83) 134.8)

1-500 83.19 (71.26, 98.06 (75.53, 5.00 (4.47, 5.52) 6.79 (6.30, 7.27) 2.61 (2.06, 3.16) 4.21 (3.48, 4.93) 95.12) 120.6)

>500 68.02 (57.64, 95.58 (70.95, 4.91(4.46, 5.36) 6.68 (6.06, 7.30) 3.42 (2.97, 3.86) 4.64 (3.77, 5.51) 78.41) 120.2)

p-trend 0.845 0.390 0.709 0.646 0.086 0.157

Models were adjusted for age, total energy intake (kcal), BMI (kg/m2), smoking (never, past, or current), any alcohol consumption, energy- adjusted total vitamin D intake from diet and supplements (IU), use of lipid-lowering medications, and, in women, menopause status and estrogen use. Models where dietary calcium intake quartile was the main effect were additionally adjusted for calcium intake from supplements (mg).

Models where tertiles of calcium intake from supplements were the main effect were additionally adjusted for energy-adjusted dietary calcium

(mg). For these analyses, results displayed are for the natural logarithm of (coronary artery calcification+1) and (carotid artery calcification+1) and the square root of (abdominal aorto-iliac calcification +10).

298

Table 3: Associations between all-cause and CVD mortality (HR and 95% confidence interval (CI)) and total, dietary, and supplemental calcium intake. Analysis was performed using Cox proportional hazards regression.

Model 1 Model 2

CVD Mortality All-Cause Mortality CVD Mortality All-Cause Mortality

Women Men Women Men Women Men Women Men (n=359) (n=360) (n=359) (n=360) (n=328) (n=359) (n=328) (n=359)

Energy-adjusted total calcium intake quartile

HR (95% CI) 0.84 (0.59, 0.78 (0.55, 0.71 (0.54, 0.84 (0.64, 0.91 (0.61, 0.78 (0.52, 0.73 (0.53, 0.90 (0.66, 1.20) 1.09) 0.93) 1.09) 1.36) 1.16) 1.01) 1.24)

p-value 0.336 0.146 0.012 0.190 0.658 0.214 0.061 0.528

Energy-adjusted dietary calcium intake quartile

HR (95% CI) 1.79 (0.82, 0.77 (0.43, 0.86 0.80 (0.54, 1.67 (0.66, 0.74 (0.41, 0.80 (0.42, 0.78 (0.53, 3.89) 1.38) (0.50,1.47) 1.19) 4.25) 1.34) 1.51) 1.17)

p-value 0.145 0.380 0.578 0.267 0.281 0.321 0.491 0.228

Supplemental calcium

HR (95% CI) 0.59 (0.37, 0.65 0.61 (0.46, 0.75 (0.54, 0.65 (0.37, 0.51 (0.25, 0.62 (0.42, 0.75 (0.49, 0.94) (0.43,0.98) 0.83) 1.04) 1.13) 1.06) 0.92) 1.17)

299

p-value 0.026 0.042 0.001 0.079 0.127 0.070 0.017 0.210

Calcium intake from supplements was considered as an ordinal variable (divided into mg intake ranges (0 mg, 1-500 mg, >500 mg)). Model 1 is adjusted for age and total energy intake. Model 2 is adjusted for age, total energy intake, BMI (kg/m2), smoking (never, past, or current), any alcohol consumption, energy-adjusted total vitamin D intake from diet and supplements (IU), use of lipid-lowering medications, and, in women, menopause status and estrogen use. For Model 2, if dietary calcium intake quartile was the main effect, the model was additionally adjusted for calcium intake from supplements (mg). For Model 2, if tertiles of calcium intake from supplements were the main effect, the model was additionally adjusted for energy-adjusted dietary calcium (mg).

300

Supplementary Table 1: Associations between coronary, carotid, and abdominal aorto-iliac calcification (least squares adjusted mean and 95% confidence interval (CI)) and total, dietary, and supplemental calcium intake. Analysis was performed using marginal models with generalized estimating equations.

Coronary Artery Carotid Artery Abdominal Aorto-iliac Calcification Calcification Calcification Women Men Women Men Women Men (n=272) (n=342) (n=342) (n=334) (n=343) (n=260) Energy-adjusted total calcium intake quartile 1 (216-614.2 mg women, 280-600 mg men) 4.95 6.83 3.26 4.51 81.85 124.1 (4.45, 5.45) (6.34, 7.33) (2.68, 3.83) (3.90, 5.12) (66.32, 97.39) (105.6, 142.5) 2 (614.3-851 mg women, 601-783 mg men) 4.85 7.11 3.58 4.49 78.19 112.5 (4.35, 5.36) (6.71, 7.51) (3.06, 4.11) (3.98, 5.00) (65.99, 90.39) (98.0, 126.9) 3 (858-2098 mg women, 784-1039 mg men) 4.87 6.93 2.69 3.87 72.66 108.7 (4.35, 5.40) (6.59, 7.26) (2.10, 3.28) (3.36, 4.38) (57.91, 87.42) (92.9, 124.5) 4 (2106-5350 mg women, 1044-3212 mg men) 5.02 6.87 3.63 4.68 73.96 103.4 (4.56, 5.49) (6.31, 7.44) (3.17, 4.08) (4.09, 5.27) (62.65, 85.27) (83.0, 123.8) p-trend 0.770 0.906 0.522 0.953 0.434 0.215 Energy-adjusted dietary calcium intake quartile 1 (216-484 mg women, 263-531 mg men) 4.92 6.74 2.99 5.04 76.86 107.6 (4.37, 5.46) (6.05, 7.44) (2.39, 3.58) (4.31, 5.77) (59.80, 93.91) (81.11, 134.1) 2 (487-635 mg women, 534-681.8 mg men) 4.98 7.05 3.14 4.35 78.66 112.9 (4.44, 5.51) (6.65, 7.45) (2.62, 3.67) (3.81, 4.88) (64.92, 92.41) (96.57, 129.1) 3 (638-834 mg women, 682.1-829 mg men) 4.63 6.88 3.38 4.16 66.19 105.9 (4.17, 5.09) (6.46, 7.29) (2.85, 3.90) (3.63, 4.69) (54.67, 77.71) (90.4, 121.3) 4 (837-3090 mg women, 831-1601 mg men) 5.15 7.06 3.67 3.98 83.24 123.8 (4.46, 5.84) (6.17, 7.96) (2.91, 4.43) (3.06, 4.90) (60.34, 106.1) (90.40, 157.2) p-trend 0.885 0.795 0.223 0.120 0.673 0.929

301

Supplemental calcium (mg) 0 4.82 7.00 3.41 4.47 78.78 116.4 (4.45, 5.18) (6.73, 7.27) (3.02, 3.80) (4.13, 4.81) (68.11, 89.44) (106.1,126.7) 1-500 5.15 6.86 2.91 4.15 79.96 106.4 (4.67, 5.64) (6.52, 7.21) (2.40, 3.41) (3.71, 4.60) (68.42, 91.50) (91.32, 121.55) >500 4.92 6.71 3.43 4.66 70.94 103.5 (4.48,5.35) (6.00, 7.42) (2.99, 3.87) (3.83, 5.49) (60.12, 81.76) (78.13, 128.8) p-trend 0.618 0.397 0.863 0.799 0.362 0.214 Models adjusted for age and total energy intake. For these analyses, results displayed are for the natural logarithm of (coronary artery calcification+1) and (carotid artery calcification+1) and the square root of (abdominal aorto-iliac calcification +10).

302

303

Chapter 9

Associations of Coronary Artery Calcified Plaque Density with Mortality and Prior

Cardiovascular Disease Events in Type 2 Diabetes: the Diabetes Heart Study

Laura M Raffield, Amanda J Cox, Michael H. Criqui, Fang-Chi Hsu, Jianzhao Xu, J Jeffrey Carr,

Barry I Freedman, Donald W Bowden

304

Abstract

Background: Coronary artery calcified plaque (CAC) is strongly predictive of cardiovascular disease (CVD) events and mortality, both in the general population and in individuals with type 2 diabetes (T2D) at high CVD risk. CAC is usually reported as an Agatston score, which is weighted for increased plaque density; recent analyses, however, have questioned the validity of this assumption. The role of CAC density, both independently and with CAC volume, in CVD risk prediction is unclear. We here examine the role of CAC density in individuals with T2D from the family-based Diabetes Heart Study and the African American Diabetes Heart Study.

Methods: CAC density was calculated both as Agatston score divided by area and as mass divided by volume. Associations of these CAC density measures with all-cause and CVD mortality, both independently and in models adjusted for CAC volume, were examined using Cox proportional hazards models in both European American (n=928) and African American (n=595) individuals. Associations of CAC density with prior myocardial infarction and CVD events were also assessed.

Results: In European American participants, CAC density was consistently associated with increased risk of all-cause and CVD mortality (p≤0.003) and higher odds of having a history of

CVD and MI events in individual models (p≤8.32 x 10-8) adjusted for age and sex, with all associations remaining significant when models were additionally adjusted for traditional CVD risk factors. Most associations were no longer significant when models were additionally adjusted for CAC volume, however. Results were similar in African American participants.

Conclusions: Higher CAC density predicts higher risk of mortality and higher odds of a history of

CVD in European and African American individuals with T2D; however, CAC density does not seem to consistently predict CVD risk independent of CAC volume in this population.

305

Introduction

Many studies have found computed tomography (CT) based measures of calcified plaque in the coronary artery (CAC) to be predictive of cardiovascular disease (CVD) events and mortality, independent of more traditional CVD risk factors (Shaw, Raggi et al. 2003; Detrano,

Guerci et al. 2008; Erbel, Mohlenkamp et al. 2010; Polonsky, McClelland et al. 2010). CAC is also a powerful independent risk factor for CVD and mortality in individuals with type 2 diabetes

(T2D) (Agarwal, Morgan et al. 2011; Agarwal, Cox et al. 2013; Kramer, Zinman et al. 2013;

Yeboah, Erbel et al. 2014), with those with T2D having elevated CAC burden compared to unaffected individuals (Hoff, Quinn et al. 2003). This is of particular interest due to the elevated

CVD risk in individuals with T2D, with mortality risk from CVD increased two- to four-fold and approximately 68% of T2D affected individuals over age 65 dying of some form of CVD (Go,

Mozaffarian et al. 2014). Most studies report CAC burden as an Agatston score, but recent studies have questioned whether this is the most appropriate measure (Alluri, Joshi et al. 2015).

Recent work from the Multi-Ethnic Study of Atherosclerosis (MESA) found that increased coronary artery calcium plaque density, in models adjusted for plaque volume, was associated with a decreased risk of coronary heart disease (CHD) events (HR of 0.73, 95% CI,

0.58-0.91 per standard deviation increase) and all cardiovascular (CVD) events, with improved risk prediction for CHD with the inclusion of density. However, in MESA, density alone was not predictive of CHD or CVD events (Criqui, Denenberg et al. 2014). These results were contrary to the assumptions of using Agatston scores to assess CAC, as these scores are weighted for increased density (Agatston, Janowitz et al. 1990; Alluri, Joshi et al. 2015). Further study of the role of plaque density in predicting CVD risk is clearly warranted. Here, we evaluate the predictive power of CAC density measures for mortality and prior CVD events specifically in the high risk population of individuals affected by T2D, including participants of both European

306

American (EA) and African American (AA) descent from the Diabetes Heart Study (DHS) and

African American Diabetes Heart Study (AA-DHS).

Methods

Study Design and Sample

The DHS recruited T2D-affected siblings without advanced renal insufficiency, as well as their unaffected siblings when possible, from outpatient internal medicine and endocrinology clinics at Wake Forest Baptist Medical Center and from the community from 1998 through 2005; this initial DHS cohort included 1221 self- reported EA individuals and 222 self-reported AA participants. Differences in the distribution of vascular calcification between EA and AA participants prompted the development of the independent AA-DHS study, which recruited additional unrelated African American participants with T2D from 2007 to 2010. Recruitment criteria and objectives of the DHS family of studies have been reviewed previously (Bowden,

Cox et al. 2010). T2D was defined in the DHS studies as diabetes developing after the age of 35 years (or after the age of 30 in African Americans) initially treated with changes in diet and exercise and/or oral agents, in the absence of historical evidence of ketoacidosis or initial treatment with insulin. Individuals with prior evidence of CVD were not excluded. Fasting glucose and glycated hemoglobin (HbA1C) concentrations were assessed at the exam visit.

Participant examinations for both the DHS and AA-DHS studies included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses, and spot urine collection. As has been previously described, CAC was assessed using computed tomography (CT), summing the left main, left anterior descending, circumflex, posterior descending, and right coronary arteries (Wagenknecht, Langefeld et al. 2007; Carr, Register et al. 2008; Divers, Palmer et al.

2013). CT scans were performed on multi-detector CT scanners with cardiac gating in chest

307

scans, with a CT slice thickness of 2.5 mm. The protocol for CAC imaging was the same as in two large population-based studies of subclinical cardiovascular disease, MESA and the Coronary

Artery Risk Development in Young Adults (CARDIA) study (Carr, Crouse et al. 2000; Carr,

Nelson et al. 2005; Brown, Kronmal et al. 2008).

Mortality was assessed using the National Social Security Death Index. When possible, copies of death certificates were obtained from county or state Vital Records Offices to determine cause of death. Cause of death was categorized based on death certificates as CVD mortality

(myocardial infarction (MI), congestive heart failure, cardiac arrhythmia, sudden cardiac death, peripheral vascular disease, and stroke) or as mortality from other causes. However, for 16 EA participants and 9 AA participants, cause of death information could not be obtained, so these participants were excluded from analyses of CVD mortality. For deceased participants, length of follow-up was determined from the date of initial study visit to date of death. For all other participants the length of follow-up was determined from the date of the initial study visit to

December 31, 2013.

For this analysis, we included only individuals affected by T2D with nonzero values for the relevant CAC values at the 90 Hounsfield unit (HU) threshold, which is more sensitive than the frequently used 130 HU threshold. This included a total of 928 EA participants from 459 families from the DHS study and 595 AA participants from 507 families from the DHS and AA-

DHS studies.

All study protocols were approved by the Institutional Review Board at Wake Forest

School of Medicine, and all participants provided written informed consent.

Statistical Analysis

Associations were examined for four measures of calcified plaque burden, including volume, Agatston score, and two density measures. One density measure was calculated by

308

dividing mass by volume, with the other density measure calculated as Agatston score by plaque area (calculated as volume divided by a slice thickness of 2.5 mm), similar to the measure analyzed in MESA (Criqui, Denenberg et al. 2014). Plaque measures were transformed to approximate normality prior to these analyses. In EA participants, volume and Agatston scores were ln transformed; in AA participants, volume, Agatston score, and density calculated as mass divided by volume were ln transformed and density calculated as Agatston divided by area was squared. Non-parametric Spearman correlation coefficients were first calculated to determine relationships between these calcified plaque measures.

Associations with self-reported prior cardiovascular events and with all-cause and CVD mortality were assessed. For self-reported cardiovascular disease, we analyzed both self-reported history of MI and a composite measure of any history of cardiovascular disease events, including

MI, angina, or stroke, history of vascular procedures including coronary angioplasty, coronary artery bypass graft, or endarterectomy, or Q wave abnormalities indicative of prior MI using the

Minnesota code. Due to the family structure of the DHS cohort, associations with self-reported

CVD events were examined using marginal models with generalized estimating equations with a sandwich estimator of the variance under exchangeable correlation, and Cox proportional hazards models with sandwich-based variance estimation were used to evaluate associations with mortality. Models were adjusted for covariates as indicated including age, sex, statin use, ln transformed total cholesterol, square root transformed high-density lipoprotein cholesterol

(HDL), systolic blood pressure (SBP), high blood pressure medication use, and current smoking.

All analyses were performed in SAS 9.3.

Results

Demographic characteristics of AA and EA participants in the DHS and AA-DHS included in plaque density analyses are summarized in Table 1. Mean diabetes duration was 10.5

309

± 7.2 years (mean ± standard deviation (SD)) in EA participants and 10.8 ± 7.9 years in AA participants. 44.5% of EA participants and 34.9% of AA participants had a history of CVD, and average BMI was >30 kg/m2 in both groups. While burden of subclinical cardiovascular disease was extensive in both groups, EA participants had higher median plaque burden (359.75) than

AA participants (84), consistent with numerous prior reports (Lee, O'Malley et al. 2003; Bild,

Detrano et al. 2005; Freedman, Hsu et al. 2005). All-cause mortality was greater for EA participants (31.5%) than AA participants (17.7%) as well, likely in part due to longer follow-up time in EA participants. In both EA (Supplementary Table 1) and AA (Supplementary Table 2) participants, the CAC measures are highly correlated, with volume and Agatston score very highly correlated in both groups (r>0.99) and the density measures strikingly, though less strongly, correlating with the each other as well (r>0.73). Volume and Agatston score measures were also correlated with both density measures in both AA and EA participants (r>0.65).

In EA participants (Table 2), all four CAC measures, including both density measures, were associated with increased risk of all-cause and CVD mortality (p≤0.003) and higher odds of having a history of CVD and MI events in individual models (p≤8.32 x 10-8) adjusted for age and sex. Associations were similar when models were additionally adjusted for statin use, total cholesterol, HDL, SBP, high blood pressure medication use, and current smoking. When density and volume measures were included in the same model, volume was still consistently associated with elevated mortality risk and increased odds of history of CVD in all models (p≤4.35 x 10-4 for age and sex adjusted models), but associations with density were less consistent. For density calculated as Agatston score divided by area, there was a nominal trend towards increased density associating with reduced risk of CVD mortality (p=0.063, hazard ratio (HR) 0.72, 95% confidence interval (CI) 0.51, 1.02) in models adjusted for volume, age, and sex, but this association was attenuated (p=0.134) when models were further adjusted for statin use, total cholesterol, HDL, SBP, high blood pressure medication use, and current smoking. However, for

310

density calculated as mass divided by volume, increased density was associated with increased odds of a history of CVD (p=0.001) and increased odds of a history of MI (p= 0.029) in models adjusted for volume, age, and sex, and these associations remained in more fully adjusted models

(Table 3). These significant associations indicate that density may provide some additional predictive power for history of CVD and MI events in EA participants when already adjusting for

CAC volume and traditional risk factors like age, sex, statin use, total cholesterol, HDL, SBP, high blood pressure medication use, and current smoking.

In the smaller sample of AA participants, volume, Agatston score, and density from

Agatston score divided by area were consistently associated with elevated risk of all-cause and

CVD mortality (p≤0.001) and higher odds of having a history of CVD and MI events (p≤6.89 x

10-5) in models adjusted for age and sex, with similar results in more fully adjusted models.

However, density from mass by volume was at best nominally associated (p≥0.044) with elevated

CVD risk in age and sex adjusted models. These associations with density from mass by volume were somewhat strengthened in models adjusted for statin use, total cholesterol, HDL, SBP, high blood pressure medication use, and current smoking, however, with increased density compellingly associated with increased odds of history of CVD (p=7.05 x 10-7) and history of MI

(p=5.44 x 10-7) and more nominally with increased CVD mortality risk (p=0.040) (Table 4).

Similar to the EA participants, in models including both density and volume measures, volume was still consistently associated with elevated mortality risk and increased odds of history of

CVD in all models for the AA participants (p≤ 0.004for age and sex adjusted models), but associations with density were inconsistent. Density calculated as Agatston by area was strongly associated with increased risk of all-cause and CVD mortality (p≤1.55 x 10-15) in age and sex adjusted models, but these associations were attenuated in more fully adjusted models, and no associations with density from mass by volume were observed (Table 5).

311

Conclusions

This analysis from the DHS in both EA and AA participants assessed associations of multiple CAC measures, including plaque volume, Agatston score, and two density measures, with mortality and past CVD events in patients with T2D. We found that both density measures and Agatston score and volume measures were associated with higher risk of all-cause and CVD mortality and higher odds of prior MI and CVD events in EA participants, with broadly similar results in AA participants. In models adjusted for volume, however, few independent associations with density measures remained, with the exception of an association between higher density

(calculated as mass divided by volume) and increased odds of a history of CVD and MI in EA participants and an association between density (calculated as Agatston by area) and increased risk of all-cause and CVD mortality in AA participants. These results do not suggest a consistent association of CAC density with mortality and CVD events independent of volume in patients with T2D but do suggest that increased plaque density is associated with elevated CVD risk in patients with T2D, as is assumed in calculation of Agatston scores.

Our results in the DHS differ from those previously observed in MESA, in which the largest multi-ethnic study of CAC plaque density to date was performed. It is noteworthy that the

DHS is characterized by a high average burden of CAC, with a mean CAC volume of 899.53 in

EA participants and 424.29 in AA participants, as compared to a mean CAC volume of 257.9 for

MESA participants. Density alone was not predictive of CHD or CVD events in MESA, but when models included both CAC volume and CAC density, increased density was associated with a decreased risk of CHD and CVD events, though it is worth noting that this association was weaker in those with high levels of CAC burden (Criqui, Denenberg et al. 2014). In the DHS, density alone was predictive of higher mortality risk and was significantly associated with higher odds of a history of CVD. However, in models adjusted for volume, density was not generally

312

associated with CVD risk for most models, though some associations were observed with density and increased CVD risk, the opposite direction of effect as was observed in MESA.

There are a number of differences between our study and the analysis completed in

MESA. Most notably, our study included only participants with T2D, who tend to have higher

CAC burdens than unaffected controls (Hoff, Quinn et al. 2003), while MESA included only

17.9% T2D-affected participants; however, total sample size for our analysis was significantly smaller. We also stratified our analyses by race/ethnicity; the analysis presented by MESA instead adjusted for race/ethnicity as a covariate and tested for a potential interaction of race/ethnicity with density, which was nonsignificant. We analyzed two measures of density, both an Agatston score by area measure similar to that analyzed in MESA and the more intuitive mass by volume measure, which was not able to be calculated in MESA. We did not have data for incident CVD events, the main outcome in the MESA analysis, with data available in the DHS only for self-reported, non-adjudicated CVD events. Our determination of CVD mortality from death certificate data also has limitations; cause of death from death certificates can be inaccurate, leading to misclassification (Coady, Sorlie et al. 2001; Wexelman, Eden et al. 2013). However, our results for all-cause and CVD mortality are similar, making potential misclassification less of a concern. Analyses in MESA also used the less sensitive 130 HU threshold for CAC assessment, which may contribute to differences in our results. However, in African American participants from the AA-DHS with all CAC measures at both the 130 and 90 HU threshold, results were similar for values from either threshold, with very similar association results for volume,

Agatston score, and both density measures for both mortality and prior CVD events.

The utility of CAC, usually assessed using Agatston or volume measures, in predicting

CVD events and mortality is well-established, both in the general population (Shaw, Raggi et al.

2003; Detrano, Guerci et al. 2008; Erbel, Mohlenkamp et al. 2010; Polonsky, McClelland et al.

2010) and those with T2D (Agarwal, Morgan et al. 2011; Agarwal, Cox et al. 2013; Kramer,

313

Zinman et al. 2013; Yeboah, Erbel et al. 2014). Questions remain, however, about whether the upweighting of Agatston scores for increased CAC density is valid and whether CAC density can add to the predictive power of CAC volume alone. While work from MESA suggests that assessing CAC density from the same CT scan from which CAC volume is derived can improve predictive power for CVD events in the general population, in the T2D-affected DHS population we find that higher CAC density is not consistently associated with risk of mortality or odds of prior CVD events in models adjusted for volume, and the few significant associations of density measures in models adjusted for volume are in the opposite direction of those observed in MESA, with increased density associating with increased, not decreased, CVD risk. These results suggest that the role of CAC density may differ in the high CVD risk population of individuals affected by T2D; further study is needed to determine whether CAC density is independently predictive of

CVD risk and its direction of effect in differing patient populations, with longitudinal analyses of changes in CAC density also an important future goal.

314

Table 1: Demographic characteristics of European American and African American participants with type 2 diabetes from the Diabetes Heart Study and African American Diabetes Heart Study cohorts with nonzero coronary artery calcification burden.

European Americans (n=928) African Americans (n= 595)

Mean (SD) or % Median (range) n Mean (SD) or % Median (range) n Trait 63.31 (34.21, 62.84 (9.01) 928 57.47 (9.3) 58 (35, 86) 595 Age (years) 85.98)

50.4% 928 60.2% 595 Female Sex (%) 16.2% 924 23.7% 592 Current Smoking (%) 43.6% 924 37.0% 592 Past Smoking (%) 22.1% 920 12.1% 586 History of Myocardial Infarction (%) 44.5% 920 34.9% 564 History of Cardiovascular Disease (%) 31.5% 928 17.7% 595 All-Cause Mortality (%) 14.6% 912 8.4% 586 Cardiovascular Disease Mortality (%) 9.49 (3.09) 10.18 (0.3, 15.71) 928 6.09 (3.01) 5.26 (0.35, 15.98) 594 Follow-up Time (years) 32.33 (6.5) 31.3 (17.1, 57.97) 928 35.22 (8.15) 34 (17.06, 77.54) 595 Body Mass Index (kg/m2) 147.79 (56.08) 135 (16, 463) 926 150.97 (65.8) 135 (32, 524) 591 Glucose (mg/dL) 7.54 (1.61) 7.2 (4.6, 16.6) 922 8.2 (2.11) 7.7 (4.4, 21.8) 581 Glycated Hemoglobin (%) 10.52 (7.2) 8 (0, 46) 913 10.75 (7.91) 9 (0, 52) 594 Diabetes Duration (years)

315

1919.99 Coronary Artery Calcified Plaque (mass 520 (0, 50415) 928 985.35 (1983.06) 146 (0.5, 15526.5) 587 score) (3372.59) Coronary Artery Calcified Plaque (volume 899.53 (1326.13) 359.75 (1, 15569) 928 424.29 (768.65) 84 (1, 6103) 595 score) 1033.02 361.25 (0.5, Coronary Artery Calcified Plaque 926 496.27 (961.87) 75.25 (0.5, 7501.5) 594 (Agatston score) (1638.49) 20563) Coronary Artery Calcified Plaque Density 2.22 (0.82) 2.49 (0.63, 4.29) 926 2.08 (1.1) 2.28 (0.42, 18.43) 594 (Agatston/area) Coronary Artery Calcified Plaque Density 1.71 (0.76) 1.75 (0.02, 4.46) 912 1.73 (0.71) 1.7 (0.02, 5.39) 587 (mass/volume) 184.61 (42.97) 180 (65, 391) 911 182.97 (45.07) 178 (81, 428) 586 Total Cholesterol (mg/dL) 42.27 (11.96) 41 (8, 98) 911 48.13 (13.78) 46 (18, 115) 586 HDL (mg/dL) 140.03 (19.05) 138.5 (94, 260) 925 136.98 (19.96) 134 (85, 211) 566 Systolic Blood Pressure (mmHg) 72.38 (10.23) 71.5 (36.5, 106) 925 77.24 (11.6) 77 (48.5, 122) 566 Diastolic Blood Pressure (mmHg) 76.5% 928 83.6% 592 High Blood Pressure Medications (%) 43.6% 926 48.7% 594 Statin Use (%) 79.9% 928 75.8% 594 Oral Hypoglycemic Medications (%) 27.2% 928 41.9% 594 Insulin Use (%)

316

Table 2: Associations with all-cause and cardiovascular disease (CVD) mortality and history of CVD and myocardial infarction (MI) for coronary artery calcification measures analyzed in independent models in European American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures.

Adjusted for age, sex, statin use, total cholesterol, HDL, systolic blood pressure, Adjusted for age and sex. high blood pressure medication use, and current smoking. Hazard 95% Hazard 95% CAC Measure Outcome Ratio/ β Confidence P-value n Ratio/ β Confidence P-value n

Estimate Interval Estimate Interval

All-cause Mortality Volume 1.68 1.39 2.03 8.46 x 10-8 928 1.67 1.37 2.03 3.08 x 10-7 902

All-cause Mortality Agatston score 1.66 1.36 2.01 3.37 x 10-7 926 1.65 1.35 2.02 1.02 x 10-6 901 Density- Mass by All-cause Mortality 1.27 1.10 1.47 0.001 912 1.31 1.12 1.53 6.98 x 10-4 889 volume Density- Agatston All-cause Mortality 1.42 1.21 1.68 2.91 x 10-5 926 1.44 1.21 1.71 3.97 x 10-5 901 by area CVD Mortality Volume 1.99 1.50 2.66 2.37 x 10-6 912 1.93 1.45 2.57 6.58 x 10-6 887

CVD Mortality Agatston score 1.92 1.43 2.57 1.37 x 10-5 910 1.86 1.39 2.49 3.46 x 10-5 886 Density- Mass by CVD Mortality 1.40 1.15 1.70 7.91 x 10-4 897 1.39 1.13 1.70 0.002 875 volume Density- Agatston CVD Mortality 1.40 1.12 1.76 0.003 910 1.38 1.10 1.73 0.005 886 by area History of CVD Volume 0.888 0.678 1.098 <2.22 x 10-16 920 0.787 0.574 1.000 4.49 x 10-13 894

History of CVD Agatston score 0.872 0.664 1.080 2.22 x 10-16 918 0.768 0.557 0.978 9.33 x 10-13 893 Density- Mass by History of CVD 0.554 0.383 0.726 2.45 x 10-10 904 0.498 0.331 0.665 5.22 x 10-9 881 volume

317

Density- Agatston History of CVD 0.679 0.512 0.846 1.55 x 10-15 918 0.599 0.425 0.774 1.82 x 10-11 893 by area History of MI Volume 1.372 1.047 1.697 2.22 x 10-16 920 1.273 0.919 1.628 1.92 x 10-12 894

History of MI Agatston score 1.377 1.033 1.722 4.88 x 10-15 918 1.251 0.882 1.619 2.93 x 10-11 893 Density- Mass by History of MI 0.668 0.424 0.912 8.32 x 10-8 904 0.565 0.323 0.806 4.65 x 10-6 881 volume Density- Agatston History of MI 0.927 0.671 1.182 1.25 x 10-12 918 0.799 0.527 1.070 7.93 x 10-9 893 by area

318

Table 3: Associations with all-cause and cardiovascular disease (CVD) mortality and history of CVD and myocardial infarction (MI) for density and volume measures analyzed in the same model in European American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures.

Adjusted for age, sex, statin use, total cholesterol, HDL, systolic blood pressure, high Adjusted for age and sex blood pressure medication use, current smoking. Hazard 95% Hazard 95% Outcome CAC measure Ratio/ β Confidence P-value n Ratio/ β Confidence P-value n Estimate Interval Estimate Interval All-cause Density- Agatston by area 0.99 0.76 1.28 0.921 926 1.04 0.79 1.36 0.786 901 Mortality Volume 1.69 1.26 2.26 4.35 x 10-4 926 1.61 1.19 2.16 0.002 901 Density- Agatston by area 0.72 0.51 1.02 0.063 910 0.76 0.53 1.09 0.134 886 CVD Mortality Volume 2.62 1.73 3.98 6.14 x 10-6 910 2.41 1.57 3.71 6.37 x 10-5 886 All-cause Density- Mass by volume 1.06 0.92 1.22 0.431 912 1.10 0.94 1.28 0.225 889 Mortality Volume 1.64 1.33 2.03 4.24 x 10-6 912 1.60 1.29 1.99 2.25 x 10-5 889 Density- Mass by volume 1.11 0.91 1.36 0.313 897 1.13 0.92 1.39 0.252 875 CVD Mortality Volume 1.85 1.34 2.55 1.68 x 10-4 897 1.78 1.30 2.44 3.70 x 10-4 875 Density- Agatston by area 0.143 -0.105 0.391 0.258 918 0.126 -0.139 0.392 0.351 893 History of CVD Volume 0.765 0.464 1.067 6.36 x 10-7 918 0.676 0.359 0.992 2.79 x 10-5 893 Density- Agatston by area 0.021 -0.337 0.380 0.9082 918 -0.095 -0.466 0.277 0.618 893 History of MI Volume 1.364 0.926 1.803 1.06 x 10-9 918 1.350 0.878 1.822 2.14 x 10-8 893 Density- Mass by volume 0.253 0.102 0.404 0.001 904 0.251 0.092 0.410 0.002 881 History of CVD Volume 0.772 0.538 1.007 1.12 x 10-10 904 0.656 0.417 0.896 7.45 x 10-8 881 Density- Mass by volume 0.232 0.024 0.440 0.029 904 0.211 0.004 0.417 0.046 881 History of MI Volume 1.238 0.882 1.594 9.76 x 10-12 904 1.141 0.760 1.523 4.63 x 10-9 881

319

Table 4: Associations with all-cause and cardiovascular disease (CVD) mortality and history of CVD and myocardial infarction (MI) for coronary artery calcification measures analyzed in independent models in African American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures.

Adjusted for age, sex, statin use, total cholesterol, HDL, systolic blood pressure, Adjusted for age and sex. high blood pressure medication use, and current smoking. Hazard 95% Hazard 95% CAC Measure Outcome Ratio/ β Confidence P-value n Ratio/ β Confidence P-value n

Estimate Interval Estimate Interval All-cause Volume 1.46 1.17 1.83 0.001 594 1.49 1.15 1.93 0.003 553 Mortality All-cause Agatston score 1.45 1.16 1.82 0.001 593 1.46 1.13 1.89 0.004 552 Mortality All-cause Density- Agatston by area 1.10 1.08 1.13 <1.11 x 10-16 593 1.06 1.02 1.10 0.006 552 Mortality All-cause Density- Mass by volume 1.05 0.91 1.21 0.519 587 1.36 0.95 1.96 0.098 553 Mortality CVD Mortality Volume 2.01 1.36 2.98 4.95 x 10-4 585 2.04 1.28 3.26 0.003 545 CVD Mortality Agatston score 2.05 1.39 3.03 2.85 x 10-4 584 2.05 1.30 3.22 0.002 544 CVD Mortality Density- Agatston by area 1.14 1.11 1.17 <1.11 x 10-16 584 1.12 1.05 1.19 0.001 544 CVD Mortality Density- Mass by volume 1.01 0.84 1.23 0.901 578 1.73 1.02 2.91 0.040 545 History of CVD Volume 0.717 0.495 0.939 2.45 x 10-10 564 0.671 0.426 0.916 8.32 x 10-8 526 History of CVD Agatston score 0.716 0.494 0.937 2.45 x 10-10 563 0.672 0.428 0.916 7.05 x 10-8 525 History of CVD Density- Agatston by area 2.370 1.552 3.188 1.35 x 10-8 563 2.268 1.387 3.148 4.42 x 10-7 525 History of CVD Density- Mass by volume 0.439 0.013 0.865 0.044 557 1.062 0.642 1.482 7.05 x 10-7 526 History of MI Volume 1.236 0.656 1.815 2.92 x 10-5 586 1.372 0.797 1.946 2.87 x 10-6 547 History of MI Agatston score 1.256 0.638 1.874 6.89 x 10-5 585 1.396 0.782 2.011 8.20 x 10-6 546 History of MI Density- Agatston by area 3.824 2.278 5.370 1.23 x 10-6 585 3.764 2.223 5.305 1.67 x 10-6 546 History of MI Density- Mass by volume 0.354 -0.371 1.079 0.339 579 1.572 0.958 2.186 5.44 x 10-7 547

320

Table 5: Associations with all-cause and cardiovascular disease (CVD) mortality and history of CVD and myocardial infarction (MI) for density and volume measures analyzed in the same model in African American participants with type 2 diabetes. Hazard ratios for mortality associations and β estimates for CVD and MI history reported per standard deviation change in coronary artery calcification measures.

Adjusted for age, sex, statin use, total cholesterol, HDL, systolic blood pressure, Adjusted for age and sex high blood pressure medication use, current smoking. Hazard 95% Hazard 95% Outcome CAC measure Ratio/ β Confidence P-value n Ratio/ β Confidence P-value n Estimate Interval Estimate Interval All-cause Volume 1.41 1.12 1.78 0.004 593 1.45 1.10 1.90 0.008 552 Mortality Density- Agatston by area 1.08 1.06 1.10 2.22 x 10-16 593 1.04 1.00 1.08 0.073 552 Volume 1.93 1.29 2.89 0.002 584 1.95 1.21 3.16 0.006 544 CVD Mortality Density- Agatston by area 1.11 1.08 1.14 1.55 x 10-15 584 1.08 1.02 1.15 0.011 544 All-cause Volume 1.45 1.15 1.83 0.002 587 1.73 1.16 2.58 0.007 553 Mortality Density- Mass by volume 0.97 0.86 1.10 0.640 587 0.70 0.38 1.30 0.264 553 Volume 2.05 1.36 3.09 0.001 578 2.58 1.34 4.98 0.005 545 CVD Mortality Density- Mass by volume 0.90 0.78 1.05 0.175 578 0.57 0.23 1.40 0.223 545 Volume 0.564 0.237 0.890 0.001 563 0.478 0.130 0.826 0.007 525 History of CVD Density- Agatston by area 0.769 -0.451 1.988 0.217 563 0.967 -0.304 2.238 0.136 525 Volume 1.058 0.445 1.671 0.001 585 1.229 0.600 1.859 1.28 x 10-4 546 History of MI Density- Agatston by area 0.942 -0.802 2.685 0.290 585 0.823 -0.993 2.639 0.375 546 Volume 0.681 0.439 0.922 3.39 x 10-8 557 0.481 0.153 0.808 0.004 526 History of CVD Density- Mass by volume 0.115 -0.123 0.353 0.344 557 0.476 -0.087 1.039 0.098 526 Volume 1.497 0.941 2.053 1.36 x 10-7 579 1.244 0.622 1.866 8.85 x 10-5 547 History of MI Density- Mass by volume -0.172 -0.430 0.085 0.190 579 0.364 -0.388 1.117 0.342 547

321

Supplementary Table 1: Spearman correlation coefficients for coronary artery calcification measures in European American Diabetes Heart Study participants with type 2 diabetes.

Volume Agatston score Density- Agatston by area Density- Mass by volume

Volume 1.000 0.996 0.817 0.654 Agatston score 0.996 1.000 0.860 0.681

Density- Agatston by area 0.817 0.860 1.000 0.737

Density- Mass by volume 0.654 0.681 0.737 1.000

322

Supplementary Table 2- Spearman correlation coefficients for coronary artery calcification measures in African American Diabetes Heart Study participants with type 2 diabetes.

Volume Agatston score Density- Agatston by area Density- Mass by volume Volume 1.000 0.995 0.790 0.710 Agatston score 0.995 1.000 0.841 0.753 Density- Agatston by area 0.790 0.841 1.000 0.873 Density- Mass by volume 0.710 0.753 0.873 1.000

323

Chapter 10

Associations between Anxiety and Depression Symptoms and Cognitive Testing and Neuroimaging in Type 2 Diabetes

Laura M. Raffield, Gretchen A. Brenes, Amanda J. Cox , Barry I. Freedman, Christina E. Hugenschmidt, Fang-Chi Hsu, Joseph A. Maldjian, Jeff D. Williamson, Donald W. Bowden

324

Abstract

Objective:

Anxiety, depression, accelerated cognitive decline and increased risk of dementia are observed in individuals with type 2 diabetes (T2D). Anxiety and depression may contribute to lower performance on cognitive tests and differences in neuroimaging observed in individuals with

T2D.

Methods:

These relationships were assessed in 655 European Americans with T2D from 504 Diabetes

Heart Study families. Participants completed cognitive testing, brain magnetic resonance imaging, the Brief Symptom Inventory Anxiety subscale, and the Center for Epidemiologic

Studies Depression Scale 10 item measure. We hypothesized that anxiety and depression symptoms would be associated with cognitive performance and neuroimaging measures, with associations expected to be particularly strong in those with comorbid anxiety and depression symptoms.

Results:

In analyses adjusted for age, sex, educational attainment, and use of psychotropic medications, individuals with comorbid anxiety and depression symptoms had lower performance on all cognitive testing measures assessed (p≤0.005). Those with both anxiety and depression also had increased white matter lesion volume (p=0.011), decreased gray matter cerebral blood flow

(p=1.95 x 10-5), decreased gray matter volume (p=0.007), and increased white and gray matter mean diffusivity (p≤0.001). These associations were somewhat attenuated upon further adjustment for health status related covariates.

325

Conclusions:

Significant anxiety and depression symptoms were strongly associated with cognitive performance and brain structure in a European American cohort with T2D.

326

Introduction

Type 2 diabetes (T2D) is associated with accelerated age-related cognitive decline and elevated risk of Alzheimer’s disease and vascular dementia (Lu, Lin et al. 2009; Reijmer, van den

Berg et al. 2010). A number of studies have demonstrated differences in the brain in individuals with T2D, including cerebral atrophy, declines in connectivity, and increased signs of small vessel disease, including infarcts and white matter hyperintensities (Biessels and Reijmer 2014).

Individuals with T2D also have increased incidence of depression (Mezuk, Eaton et al. 2008;

Nouwen, Winkley et al. 2010) and anxiety (Li, Barker et al. 2008), and individuals with depression and anxiety have been shown to have an increased incidence of T2D as well (Mezuk,

Eaton et al. 2008; Farvid, Qi et al. 2014; Vimalananda, Palmer et al. 2014). Symptoms of depression are also associated with increased incidence of dementia and accelerated declines in a number of sensitive cognitive testing measures in adults with T2D (Katon, Lin et al. 2010; Katon,

Lyles et al. 2012; Sullivan, Katon et al. 2013). However, few studies have assessed both anxiety and depression symptoms and their contribution to poorer performance on cognitive testing and differences in neuroimaging measures in individuals with T2D.

These relationships were examined in the Diabetes Heart Study (DHS), a single-site, family-based study enriched for T2D which assessed cognitive performance and brain magnetic resonance imaging (MRI) in the ancillary DHS-Mind study from 2008-2013. Symptoms of anxiety and depression were also assessed using the Brief Symptom Inventory Anxiety subscale and the Center for Epidemiologic Studies Depression Scale 10 item measure. Analyses were conducted in a cohort of 655 European Americans with T2D from 504 families. We hypothesized that anxiety and depression symptoms would associate with poorer performance on cognitive tests and with neuroimaging differences observed in T2D. Several studies reveal that anxiety and

327

depression are often comorbid diagnoses (Kessler, Gruber et al. 2008; Farvid, Qi et al. 2014), and we hypothesized that individuals with comorbid anxiety and depression symptoms would demonstrate poorer performance on cognitive testing and greater differences in neuroimaging than those with depression or anxiety symptoms alone.

Methods and Materials

Study Design and Sample

Participants in the DHS were recruited from outpatient internal medicine and endocrinology clinics and from the community from 1998 through 2005 in western North

Carolina. Siblings affected by T2D without advanced renal insufficiency (serum creatinine concentrations >2.0 mg/dl) were recruited, along with additional non-diabetic siblings. Ascertainment and recruitment have been described in detail (Wagenknecht, Bowden et al. 2001; Lange, Bowden et al. 2002; Bowden, Rudock et al. 2006; Bowden, Cox et al. 2010).

T2D was defined as diabetes developing after the age of 35 years treated with changes in diet and exercise and/or oral agents in the absence of initial treatment solely with insulin and without historical evidence of diabetic ketoacidosis.

The DHS-Mind study is an ancillary study to the DHS conducted between 2008 and

2013. Cognitive testing and neuroimaging were performed to investigate risk factors for cognitive decline in T2D. Examinations included interviews for medical history and health behaviors, anthropometric measures, fasting blood draws, assessment of resting blood pressure, and assessment of anxiety and depression symptoms. Participants were examined in the General

Clinical Research Center of the Wake Forest Baptist Medical Center. All study protocols were approved by the Institutional Review Board at Wake Forest School of Medicine, and all study procedures were completed in accordance with the Declaration of Helsinki. Participants provided written informed consent prior to participation.

328

The current analyses were limited to individuals with T2D, including 337 original DHS participants and 318 new DHS-Mind participants. Original DHS participants were examined on average 6.6 ± 1.4 years after their first study visit. Recruitment criteria in new participants were the same as in the original DHS, barring the requirement that participants have a T2D-affected sibling. Diabetes diagnosis was confirmed for all participants by review of medications and measurement of fasting glucose and glycated hemoglobin (HbA1C) at the exam visit.

Cognitive Testing

The cognitive testing measures assessed in the DHS-Mind study have been described

(Hugenschmidt, Hsu et al. 2013; Cox, Hugenschmidt et al. 2014) . Briefly, they included the

Modified Mini-Mental State Examination (3MSE), a global test of global cognitive function often used clinically (Teng and Chui 1987), the Digit Symbol Substitution Task (DSST), a test where participants match numbers and symbols to assess processing speed and working memory

(Wechsler 1981), the Stroop Task, an assessment of executive function (reported as the response time difference between subtest 2 and subtest 3 with the number of errors from each subtest added to the time scores) (Houx, Jolles et al. 1993), the Rey Auditory-Verbal Learning Task

(RAVLT), where participants are asked to recall word lists (reported as the total words recalled across five trials) (Lezak, Howieson et al. 2004), and tests for Phonemic Fluency (reported as the total words generated for F, A, and S) and Semantic Fluency (reported as the total words generated for the categories kitchen and animals) (Benton, Hamsher et al. 1994; Strauss, Sherman et al. 2006). Subjects were not excluded for 3MSE scores or other indices of cognitive function indicative of mild cognitive impairment or dementia (Teng and Chui 1987). In total, 653 individuals were included for the analyses of 3MSE, 650 individuals for the DSST, 647 individuals for the Stroop Task, 653 individuals for the RAVLT, 652 individuals for Phonemic

Fluency, and 652 individuals for Semantic Fluency. Colorblind individuals were excluded from the Stroop Task. For the 3MSE, DSST, RAVLT, and Phonemic and Semantic Fluency, higher

329

scores indicate better cognitive performance, but for the Stroop Task lower scores indicate better performance.

Neuroimaging

MR image acquisition. MR imaging was performed on a 1.5-T GE EXCITE HD scanner with twin-speed gradients using a neurovascular head coil (GE Healthcare, Milwaukee, WI).

Neuroimaging protocols have already been described in detail (Raffield, Cox et al. 2014).

Briefly, for the volumetric measures, structural T1 images were segmented and native space gray matter volume (GMV), white matter volume (WMV), and intracranial volume (ICV)

(gray matter + white matter + cerebrospinal fluid) were determined from the VBM8 toolbox

(http://dbm.neuro.uni-jena.de/vbm.html) automated segmentation procedure. Diffusion tensor imaging (DTI) scalar metrics, including fractional anisotropy (FA) and mean diffusivity (MD) in the gray and white matter, were computed using the Diffusion Tensor Imaging ToolKit (DTI-TK)

(http://www.nitrc.org/projects/dtitk). Cerebral blood flow (CBF) perfusion images were generated using a previously described fully automated data processing pipeline (Maldjian, Laurienti et al.

2008), allowing derivation of the gray matter cerebral blood flow (GMCBF) measure analyzed.

White matter lesion segmentation was performed using the lesion segmentation toolbox (LST)

(Schmidt, Gaser et al. 2012) for SPM8 at a threshold (k) of 0.25, which has been previously validated in DHS-Mind (Maldjian, Whitlow et al. 2013). The total white matter lesion volume

(WMLV) measure used in these analyses was determined by summing the binary lesion maps and multiplying by the voxel volume.

In total, eight neuroimaging measures were analyzed in this study: GMV, WMV,

WMLV, GMCBF, white matter mean diffusivity (WMMD), gray matter mean diffusivity

(GMMD), white matter fractional anisotropy (WMFA), and gray matter fractional anisotropy

(GMFA). All analyses of GMV, WMV, and WMLV included ICV as a covariate. In total, 577

330

individuals were included for the analyses of GMV, 577 individuals for WMV, 568 individuals for WMLV, 433 individuals for GMCBF, 467 individuals for GMMD, 467 individuals for

WMMD, 568 individuals for GMFA, and 568 individuals for WMFA.

Assessment for Anxiety and Depression

Anxiety was assessed using the Brief Symptom Inventory Anxiety subscale, previously used to reliably assess anxiety symptoms in older individuals (Abu Ruz, Lennie et al. 2010;

Khalil, Hall et al. 2011). Participants scoring >8 on this subtest were classified as having significant anxiety symptoms in these analyses. Depression was assessed using the Center for

Epidemiologic Studies Depression (CES-D) Scale 10 item measure. The CES-D measure (10 or

20 item measure) is thought to have good sensitivity and specificity in adults with diabetes (Roy,

Lloyd et al. 2012) and is able to better discriminate between depression and non-depression, diabetes-related symptoms in individuals with T2D relative to other commonly used screening measures such as the Silverstone Concise Assessment for Depression, the Hospital Anxiety and

Depression Scale, and the Depression in the Medically Ill questionnaire (McHale, Hendrikz et al.

2008). Participants scoring >10 on this measure were classified as having significant depression symptoms for the purposes of this study. A total of 655 European Americans with T2D (from 504 families) who completed the Brief Symptom Inventory Anxiety subscale and CES-D Scale 10 item measure were included in the analyses. Self-report of antianxiety or antidepressant medication use was adjusted for in analyses but was not considered for definition of anxiety/depression symptom groupings. For the purposes of this manuscript, significant self- reported anxiety or depression symptoms from the Brief Symptom Inventory Anxiety subscale or the CES-D will be referred to as anxiety and depression.

331

Statistical Analysis

Continuous variables were transformed as necessary to approximate normality, including analysis of the natural logarithm of the Stroop Task score, the natural logarithm of (WMLV+1), the square root of GMCBF, the natural logarithm of pulse pressure, the natural logarithm of body mass index (BMI), and the natural logarithm of HbA1C. Relationships between anxiety/depression symptom groupings (depression only, anxiety only, both anxiety and depression, neither anxiety nor depression) depression and cognitive testing and neuroimaging were examined using marginal models with generalized estimating equations. The models account for familial correlation using a sandwich estimator of the variance under exchangeable correlation. The overall 3 degrees of freedom (df) test was calculated using a type III test and used to assess whether there were any significant differences between the four anxiety and depression symptom groups. Nominal statistical significance was accepted at p<0.05. For nominally significant associations for the type III test, associations with the cognitive testing or neuroimaging measure were examined for individuals with depression symptoms only, with anxiety symptoms only, and with both anxiety and depression symptoms, comparing these groups to a reference group of individuals without anxiety or depression in a single model. Associations were adjusted for covariates including age, sex, use of antidepressant medications, use of antianxiety medications,

BMI, educational attainment (less than high school, high school, greater than high school),

HbA1C, pulse pressure, prior history of cardiovascular disease (CVD) events, and ICV as indicated. All analyses were performed in SAS 9.3 (SAS Institute, Cary, NC).

Results

Table 1 summarizes the demographic and clinical characteristics of study participants, stratified by anxiety and depression symptom groupings. The mean diabetes duration in the cohort was 15 ± 7.9 years. Most individuals were overweight or obese with a high prevalence of hypertension and prior CVD events, as expected for a T2D-affected cohort.

332

We first assessed the relationships between anxiety and depression symptoms and performance on cognitive testing (Table 2) in models adjusted for age, sex, use of antidepressant and antianxiety medications, and educational attainment. Significant differences in cognitive scores between the anxiety and depression groups were observed for all tests except Phonemic

Fluency (p≤0.006). Individuals with comorbid anxiety and depression displayed the lowest cognitive testing scores as compared to the reference group of individuals without anxiety or depression symptoms; these individuals had significantly poorer performance (p≤0.005).

Individuals with depression alone demonstrated a trend towards poorer performance on all cognitive testing measures assessed except Phonemic Fluency (p≤0.027). Few significant associations, other than a nominal association with poorer performance on the RAVLT (p=0.011), were observed for those with only anxiety symptoms (Table 2). Additional adjustment for health status, including HbA1C, BMI, pulse pressure, and history of CVD, which might mediate the observed associations, attenuated, but did not eliminate, the observed relationships with anxiety and depression symptoms (p≤0.046 for worse performance on all cognitive tests) (Table 3).

Adjustment for the health status covariates strengthened association between anxiety symptoms alone and performance on the RAVLT (p=0.009) and revealed a nominal association with performance on the Stroop Task (p=0.030). Associations with depression symptoms alone were attenuated with adjustment for health status (p≤0.026 for 3MSE, Stroop Task, RAVLT, and

DSST) (Table 3).

Associations between anxiety and depression symptoms and a number of neuroimaging measures, including brain volumes, white matter lesion volume, fractional anisotropy and mean diffusivity measures, and cerebral blood flow, were subsequently assessed to determine whether the observed associations with cognitive performance were reflected on brain MRI. Nominally significant differences between the anxiety and depression groupings were observed for WMMD,

GMMD, WMFA, GMCBF, WMLV, and GMV. Anxiety and depression symptoms were

333

associated with lower WMFA and higher WMMD and GMMD (p≤ 0.001) (Table 4). Anxiety and depression symptoms were also associated with lower GMCBF (p=1.95 x 10-5) and higher

WMLV (p=0.011). These relationships were not observed in individuals with anxiety alone or depression alone. Lower GMV was observed in those with depression alone and those with anxiety and depression symptoms (p≤0.018) (Table 4). In models further adjusted for health status, associations were attenuated but remained significant for comorbid anxiety and depression symptoms with GMMD, WMMD, WMFA, GMCBF, and GMV (p≤0.036) (Table 5).

Discussion

Individuals with T2D have an accelerated rate of mild age-related cognitive decline and increased incidence of dementia. There is increasing interest in factors, for example poorer glycemic control, dyslipidemia, and hypertension (Bryan, Bilello et al. 2014; Williamson, Launer et al. 2014), which may contribute to cognitive and brain anatomic differences in T2D. Diabetic individuals have higher rates of depression and anxiety, which may also contribute to cognitive decline. The present analyses in adults with T2D replicated the previously reported association of depression symptoms with poorer performance on tests of cognitive function but identified stronger associations with cognitive testing performance in individuals with comorbid anxiety and depression symptoms as opposed to depression symptoms alone. Poorer performance on cognitive tests was also reflected by associations with a number of neuroimaging measures, including associations with white matter lesion volume, fractional anisotropy, mean diffusivity, and cerebral blood flow. These results highlight the importance of assessing anxiety and depression symptoms as potential risk factors for cognitive decline in individuals with T2D.

Consistent associations between anxiety and depression symptoms and poorer performance on all cognitive tests were observed, including tests assessing global cognitive function, processing speed, and executive function; associations with depression symptoms alone

334

were also observed for most cognitive measures. A number of studies in general population cohorts and individuals with T2D detected poorer performance on cognitive testing and higher incidence of dementia in those with depression symptoms (Barnes, Alexopoulos et al. 2006;

Katon, Lin et al. 2010; Byers and Yaffe 2011; Katon, Lyles et al. 2012; Sullivan, Katon et al.

2013). The impact of anxiety and comorbid anxiety and depression symptoms on cognitive performance is less clear in T2D; however, a number of studies in older adults found that anxiety symptoms associate with poorer cognitive performance (Wetherell, Reynolds et al. 2002;

Pietrzak, Maruff et al. 2012; Stillman, Rowe et al. 2012). The present results suggest that it is important to assess anxiety and depression symptoms in individuals with T2D. A prior analysis of cognitive testing measures in the Action to Control Cardiovascular Risk in Diabetes–Memory in

Diabetes study found that self-reported depression symptoms were associated with accelerated cognitive decline in individuals with T2D and that this did not appear to be mediated by poor glycemic control, blood pressure, or lipids (Sullivan, Katon et al. 2013). Similarly, we found that even after adjustment for health status, depression and anxiety symptoms were associated with lower performance on cognitive testing and differences in neuroimaging, although many of the associations were attenuated.

Associations of depression and anxiety with poorer cognitive performance may be confounded by differences in motivation and other factors (Austin, Mitchell et al. 2001). As such, we note that the current associations observed with cognitive testing measures were further reflected in concurrently performed neuroimaging studies. Few existing neuroimaging studies of depression and anxiety focused on T2D; however, many of the associations in the present report mirror those in general populations. Increased white matter lesion volume, reflective of increased small vessel disease in the brain (Biessels and Reijmer 2014), was seen in those with comorbid anxiety and depression. Depression has previously been associated with elevated white matter lesion volume independent of other risk factors (e.g., diabetes and CVD) (Taylor, MacFall et al.

335

2005). The associations observed with higher MD and lower FA, generally reflective of decreases in fiber integrity and brain connectivity, suggests cerebral microstructural differences exist in

T2D with anxiety and depression (Thomason and Thompson 2011). A previous analysis of older adults with atherosclerosis found reduced WMFA in those with increased anxiety (Bijanki,

Stillman et al. 2013), and differences in white matter microstructure, as reflected in FA and MD measures, have also been reported in late-life depression (Nobuhara, Okugawa et al. 2006;

Shimony, Sheline et al. 2009). Widespread reductions in GMV have been observed in late life depression (Andreescu, Butters et al. 2008; Sexton, Mackay et al. 2013) as well as anxiety

(Moon, Kim et al. 2014). We saw lower GMV in this T2D-affected cohort in participants with depression alone and comorbid anxiety and depression. Lower GMCBF was also present in individuals with anxiety and depression symptoms; a number of positron emission tomography studies reported associations between depression and reduced cerebral blood flow (Kohn,

Freedman et al. 2007; Dotson, Beason-Held et al. 2009). While neuroimaging associations were somewhat attenuated upon adjustment for health status, residual associations indicate that anxiety and depression may be important contributors to variation in cerebral anatomy in individuals with

T2D independent from established risk factors.

Many factors may mediate links between T2D, anxiety, depression, and cognition and neuroimaging phenotypes. Anxiety and depression are thought to be important risk factors for vascular disease (Scherrer, Chrusciel et al. 2010; Brunner, Shipley et al. 2014), which could be an important mediator of the observed differences in cognition and brain anatomy given the well- established link between vascular disease and risk of cognitive impairment, differences in neuroimaging measures, and dementia (Vidal, Sigurdsson et al. 2010; Warsch and Wright 2010;

Hugenschmidt, Hsu et al. 2013). However, some reports indicate that depression is an independent risk factor for cognitive impairment adjusting for vascular disease (Barnes,

Alexopoulos et al. 2006), and we observed associations between anxiety and depression with

336

cognitive performance and neuroimaging in models adjusted for prior CVD. The potential mediating effects of vascular disease on the observed relationships between anxiety and depression symptoms and cognition and neuroimaging measures was also explored in a subset of the DHS-Mind cohort (n=322) who were affected by T2D and had both anxiety and depression data and measures of coronary artery calcified plaque, a measure of subclinical CVD and a powerful predictor of CVD events and mortality (Raggi, Shaw et al. 2004; Detrano, Guerci et al.

2008; Polonsky, McClelland et al. 2010; Agarwal, Cox et al. 2013). Addition of coronary artery calcification to regression models did not substantively change associations of anxiety and depression with poorer performance on cognitive tests and increased age-related differences in neuroimaging measures in this analysis, suggesting associations of anxiety and depression observed here are not primarily mediated by vascular disease burden.

A number of other factors may also play a role in the links between depression and anxiety and cognition and neuroimaging measures in individuals with T2D. Symptoms of depression are associated with poorer diabetes self-care, including lack of adherence to medications, diet, and exercise (Gonzalez, Safren et al. 2008). Depression is also associated with increased risk of major microvascular and macrovascular complications in T2D (Lin, Rutter et al.

2010). Depression, anxiety, diabetes, and dementia are all associated with systemic inflammation

(Pitsavos, Panagiotakos et al. 2006; Leonard 2007; Stuart and Baune 2012), although the association with depression and anxiety is partially mediated through factors such as BMI, smoking, and physical activity (Duivis, Vogelzangs et al. 2013). Use of certain antidepressant medications has also been linked to weight gain (Serretti and Mandelli 2010), prompting concerns about impacts on T2D risk. However, medications do not appear to mediate the association between depression and metabolic dysregulation (Pyykkonen, Raikkonen et al. 2012). Studies have also linked perturbations in insulin signaling to cognitive impairment and increases in anxiety and depression (Kleinridders, Ferris et al. 2014).

337

Strengths of this study include concurrent assessment of a number of cognitive and neuroimaging phenotypes, the large sample of T2D-affected participants, T2D diagnoses confirmed by assessment of fasting glucose and HbA1C and review of medications, and assessment of both anxiety and depression symptoms, as opposed to only depression. Limitations include the lack of strict correction for multiple comparisons; however, given the consistency of the associations between anxiety and depression, poorer cognitive performance, and increased age-related differences in the brain, we felt that nominal associations were of interest. Self- reported anxiety and depression data are not equivalent to a clinical diagnosis, and we lacked data on dose and duration of antidepressant and antianxiety treatment or participation in psychotherapy, so we were unable to adjust for those factors. Differences in diet and exercise patterns between depressed and non-depressed participants not reflected in measures of glycemic control, BMI, and CVD could also be a source of confounding. MRI data were not available for all participants due to a variety of exclusions (pacemakers, claustrophobia, etc.), and some of these exclusions may be associated with anxiety and depression symptoms. Our cross-sectional analysis is also not able to resolve the direction of causality or potential longitudinal changes for the impact of anxiety and depression on cognitive decline and T2D.

In summary, in a cohort of individuals with T2D, anxiety and depression symptoms were evaluated for associations with cognitive performance and brain imaging. Consistent associations of comorbid anxiety and depression, as well as depression symptoms alone, were detected with poorer cognitive performance; however, associations were strongest for comorbid anxiety and depression, which were also associated with increased white matter lesion volume, decreased gray matter volume and cerebral blood flow, and higher mean diffusivity and lower fractional anisotropy in the gray and white matter . These results highlight the importance of assessing anxiety and depression as contributors to cognitive decline in adults with T2D.

338

Table 1: Demographic characteristics of Diabetes Heart Study participants with type 2 diabetes stratified by anxiety and depression scores.

Neither Depression nor Anxiety Depression Only (n=104) Anxiety Only (n=33) Anxiety and Depression (n=57) (n=461)

Mean  SD or Median Mean  SD or Median (range) Mean  SD or Median (range) Mean  SD or Median

% (range) % % % (range)

Age (years) 66.8  9.7 67 (38 - 93) 62.5  9.8 62 (41 - 84) 66.4  8.0 65 (48 - 83) 61.4  10.2 62 (38 - 81)

Gender (% 47.51 60.58 63.64 52.63 female)

BMI (kg/m2) 32.5  6.7 31 (15 - 59) 34.8  6.3 34 (22 - 52) 33.1  6.1 32 (23 - 48) 35.3  6.5 34 (25 - 55)

Smoking (current or 54.34 51.96 45.45 66.67 past)

Hypertension 88.5 88.46 90.91 82.46

Pulse Pressure 58.7  16.5 57 (18 - 124) 55.5  14.5 56 (25 - 93) 60.5  13.7 60 (35 - 92) 54.9  16.2 54 (17 - 92)

Self-reported history of 33.56 38 43.75 57.69 prior CVD

Diabetes Duration 15.1  8.1 13 (0.4 - 45) 14.7  6.6 14 (2 - 33) 17.6  8.0 17 (4 - 45) 13.3  7.3 13 (2 - 35) (years)

Glucose 138 (57 - 145.0  50.5 135 (40 - 389) 161.2  71.3 145 (56 - 408) 149.5  54.3 131 (60 - 319) 158.5  66.7 (mg/dL) 345)

339

Hemoglobin 7.4  1.4 7 (2 -13) 7.9  1.7 8 (5 - 15) 7.6  1.4 7 (6 - 12) 8.0  1.6 8 (6 - 13) A1C (%)

Anti-diabetic 84.4 83.7 75.8 77.2 Medication*

Cholesterol- lowering 51.6 52.9 54.8 53.9 Medication

Anti- hypertensive 68.1 68.3 75.8 56.1 Medication

Level of educational attainment: 8.9 18.5 15.2 21.1 Less than high school

Level of educational 44.4 51.5 45.5 47.4 attainment: High school

Level of educational attainment: 46.7 30.1 39.4 31.6 Greater than high school

Antidepressan 10.0 12.5 24.2 12.3 t medication

Antianxiety 3.9 5.8 9.1 7.0 medication

340

Either antidepressant 12.6 14.4 27.3 14.0 or antianxiety medication

Digit Symbol Substitution 52  16 52 (14, 106) 49  15 49 (13, 106) 52  15 52 (22, 81) 46  16 45 (16, 89) Task

Stroop 33  18 28 (7, 151) 35  22 29 (5, 147) 35  17 32 (9, 81) 38  17 34 (13, 87)

Rey Auditory- Verbal 41  10 42 (11, 65) 40  10 39 (16, 64) 40  11 37 (22, 60) 39  10 39 (20, 61) Learning Task

Modified Mini-Mental 92  7 93 (62, 100) 91  7 92 (69, 100) 90  7 91 (74, 100) 90  7 91 (71, 100) State Examination

Semantic 32  8 31 (11, 69) 30  9 30(11,58) 31  11 29 (12, 55) 27  7 27 (11, 49) Fluency

Phonemic 31  11 30 (3, 66) 30  11 30 (5, 59) 29  13 31 (7, 55) 26  12 25 (9, 63) Fluency

Gray Matter 0.196 (0.141, 0.201 (0.144, 0.201 (0.155, 0.192 (0.153, Fractional 0.191  0.024 0.196  0.029 0.199  0.025 0.187  0.022 0.255) 0.346) 0.251) 0.229) Anisotropy

White Matter 0.344 (0.261, 0.346 (0.283, 0.336 (0.299, 0.329 (0.251, Fractional 0.342  0.026 0.343  0.027 0.338  0.022 0.33  0.025 0.403) 0.419) 0.378) 0.397) Anisotropy

Gray Matter 537 (401, 526  54 524 (373, 709) 518  54 517 (392, 666) 527  59 531 (407, 635) 528  50 Volume 626)

341

White Matter 559 (461, 576  71 573 (402, 815) 558  71 548 (430, 732) 558  64 556 (461, 712) 576  64 Volume 715)

Gray Matter 1.10 (0.88, 1.08 (0.68, 1.09 (0.92, Mean 1.10  0.09 1.08  0.10 1.07  0.08 1.06 (0.91, 1.24) 1.11  0.10 1.37) 1.38) 1.32) Diffusivity

White Matter 0.80 (0.70, 0.79 (0.66, 0.80 (0.73, Mean 0.80  0.04 0.79  0.05 0.81  0.05 0.80 (0.72, 0.92) 0.81  0.04 0.95) 0.94) 0.91) Diffusivity

Gray Matter 42.0 (2.0, 34.1 (10.2, Cerebral 43.3  15.9 42.1  16.0 41.5 (5.4, 84.9) 43.2  17.4 42.8 (9.1, 87.3) 33.8  13.0 103.3) 64.5) Blood Flow

Total White 1.56 (0, Matter Lesion 4.39  8.32 1.68 (0, 103.3) 4.31  9.02 1.38 (0, 65.6) 5.71  8.01 2.00 (0, 31.98) 6.81  11.83 59.57) Volume

Total 1362 (1028, 1322 (1028, 1341 (1064, 1377 (1081, Intracranial 1368  135 1331  140 1338  149 1377  131 1752) 1627) 1698) 1671) Volume

*either oral hypoglycemic medications or insulin

342

Table 2: Associations between cognitive testing variables and depression symptoms only, anxiety symptoms only, and anxiety and depression symptoms unadjusted for health status covariates (β value and 95% confidence interval (CI)).

Depression Only (n=104) Anxiety Only (n=33) Anxiety and Depression (n=57)

Overall 3 df test β value (95% CI) p-value β value (95% CI) p-value β value (95% CI) p-value

-0.504 -1.36 -4.28 Phonemic Fluency 0.066 0.672 0.478 0.005 (-2.84, 1.83) (-5.13, 2.40) (-7.26, -1.31)

-1.98 -1.09 -4.84 Semantic Fluency 9.12 x 10-5 0.027 0.500 4.66 x 10-7 (-3.73, -0.231) (-4.27, 2.08) (-6.72, -2.96)

Modified Mini- -1.53 -1.56 -2.47 Mental State 0.006 0.020 0.129 0.005 (-2.82, -0.246) (-3.59, 0.458) (-4.17, -0.767) Examination

0.137 0.092 0.230 1.87 x 10-5 Stroop 2.96 x 10-4 0.011 0.238 (0.032, 0.242) (-0.061,0.245) (0.124, 0.335)

Rey Auditory-Verbal -3.53 -3.35 -4.31 1.61 x 10-5 6.50 x 10-4 0.011 1.18 x 10-4 Learning Task (-5.56, -1.50) (-5.93, -0.772) (-6.50, -2.11)

Digit Symbol -4.83 -1.13 -8.08 3.90 x 10-5 6.26 x 10-4 0.605 1.42 x 10-5 Substitution Task (-7.60, -2.06) (-5.42, 3.16) (-11.73, -4.43)

343

Models adjusted for age, gender, antidepressant medication use, antianxiety medication use, and educational attainment (less than high school, high school, greater than high school). Significant anxiety symptoms are defined as a Brief Symptom Inventory Anxiety subscale score > 8; significant depression symptoms are defined as a 10 item Center for Epidemiologic Studies Depression scale score >10.

Reference group: n = 461 individuals with neither depression nor anxiety.

Analysis was performed using marginal models with generalized estimating equations. Results for both an overall 3 degrees of freedom (df) type

III test and for each anxiety/depression grouping as compared to the reference group are displayed.

344

Table 3: Associations between cognitive testing variables and depression symptoms only, anxiety symptoms only, and anxiety and depression symptoms adjusted for health status covariates (β value and 95% confidence interval (CI)).

Depression Only (n=104) Anxiety Only (n=33) Anxiety and Depression (n=57)

Overall 3 df test β value (95% CI) p-value β value (95% CI) p-value β value (95% CI) p-value

0.074 -1.13 -3.43 Phonemic Fluency 0.255 0.953 0.572 0.046 (-2.36, 2.51) (-5.06, 2.80) (-6.80, -0.069) -1.40 -0.640 -4.79 Semantic Fluency 0.001 0.124 0.709 7.12 x 10-6 (-3.18, 0.383) (-4.00, 2.72) (-6.88, -2.70) Modified Mini- -1.49 -1.97 -2.24 Mental State 0.013 0.026 0.060 0.018 (-2.80, -0.176) (-4.02, 0.086) (-4.10, -0.381) Examination 0.114 0.140 0.257 9.42 x 10-6 Stroop 1.61 x 10-4 0.038 0.030 (0.006,0.222) (0.014, 0.266) (0.143, 0.37) Rey Auditory- -3.30 -3.62 -4.20 Verbal Learning 1.16 x 10-4 0.002 0.009 1.04 x 10-3 (-5.35, -1.24) (-6.35, -0.889) (-6.71, -1.69) Task Digit Symbol -4.70 -1.46 -8.09 1.44 x 10-4 7.52 x 10-4 0.518 7.49 x 10-5 Substitution Task (-7.43, -1.96) (-5.87, 2.96) (-12.09, -4.08)

Models adjusted for age, gender, antidepressant medication use, antianxiety medication use, educational attainment (less than high school, high school, greater than high school), HbA1C, BMI, pulse pressure, and history of CVD. Anxiety is defined as a Brief Symptom Inventory Anxiety subscale score > 8; depression is defined as a 10 item Center for Epidemiologic Studies Depression scale score >10.

345

Reference group: n = 461 individuals with neither depression nor anxiety.

Analysis was performed using marginal models with generalized estimating equations. Results for both an overall 3 degrees of freedom (df) type

III test and for each anxiety/depression grouping as compared to the reference group are displayed.

346

Table 4: Associations between neuroimaging variables and depression symptoms only, anxiety symptoms only, and anxiety and depression symptoms unadjusted for health status covariates (β value and 95% confidence interval (CI)).

Depression Only (n=104) Anxiety Only (n=33) Anxiety and Depression (n=57)

Overall 3 df test β value (95% CI) p-value β value (95% CI) p-value β value (95% CI) p-value

White Matter Mean 0.006 0.011 0.017 0.010 0.239 0.095 0.001 Diffusivity (-0.004, 0.015) (-0.002, 0.024) (0.007, 0.027) Gray Matter Mean 0.012 -0.017 0.038 5.9 x 10-4 0.156 0.072 6.33 x 10-5 Diffusivity (-0.005, 0.029) (-0.036,0.002) (0.019, 0.056) White Matter -0.003 -0.004 -0.014 0.003 0.333 0.264 1.23 x 10-4 Fractional Anisotropy (-0.008, 0.003) (-0.012, 0.003) (-0.021, -0.007) Gray Matter Fractional 0.001 0.004 -0.006 0.156 0.646 0.338 0.054 Anisotropy (-0.005, 0.007) (-0.005, 0.013) (-0.012, 0.0001) Gray Matter Cerebral -0.311 -0.245 -0.788 0.002 0.066 0.314 1.95 x 10-5 Blood Flow (-0.643,0.020) (-0.723, 0.232) (-1.15, -0.426) 0.199 0.223 Total White Matter 0.368 0.017 (-0.001, 0.398) 0.051 (-0.05, 0.496) 0.109 0.011 Lesion Volume (0.083, 0.653)

-6.29 6.57 -9.68 Gray Matter Volume 0.004 0.030 0.101 0.007 (-11.96, -0.62) (-1.27, 14.42) (-16.75,-2.61) -0.13 -3.06 -4.52 White Matter Volume 0.717 0.966 0.481 0.319 (-6.09, 5.84) (-11.56, 5.44) (-13.41, 4.38) Models adjusted for age, gender, antidepressant medication use, antianxiety medication use, and educational attainment (less than high school, high school, greater than high school). Analyses of white matter lesion volume, gray matter volume, and white matter volume additionally

347

adjusted for total intracranial volume. Significant anxiety symptoms are defined as a Brief Symptom Inventory Anxiety subscale score > 8; significant depression symptoms are defined as a 10 item Center for Epidemiologic Studies Depression scale score >10.

Reference group: n = 461 individuals with neither depression nor anxiety.

Analysis was performed using marginal models with generalized estimating equations. Results for both an overall 3 degrees of freedom (df) type

III test and for each anxiety/depression grouping as compared to the reference group are displayed.

348

Table 5: Associations between neuroimaging variables and depression symptoms only, anxiety symptoms only, and anxiety and depression symptoms adjusted for health status covariates (β value and 95% confidence interval (CI)).

Depression Only (n=104) Anxiety Only (n=33) Anxiety and Depression (n=57) Overall 3 df test β value (95% CI) p-value β value (95% CI) p-value β value (95% CI) p-value White Matter Mean 0.004 0.013 0.016 0.016 0.444 0.056 2.98 x 10-3 Diffusivity (-0.006, 0.013) (-0.0003, 0.026) (0.005, 0.026) Gray Matter Mean 0.006 -0.018 0.032 0.004 0.459 0.088 6.50 x 10-4 Diffusivity (-0.010, 0.022) (-0.038, 0.003) (0.014, 0.050) White Matter -0.002 -0.005 -0.01 0.057 0.549 0.209 0.009 Fractional Anisotropy (-0.007, 0.004) (-0.012, 0.003) (-0.017, -0.002) Gray Matter 0.002 0.004 -0.005 0.291 0.516 0.336 0.166 Fractional Anisotropy (-0.004, 0.008) (-0.004, 0.013) (-0.011, 0.002) Gray Matter Cerebral -0.190 -0.317 -0.609 0.021 0.265 0.173 0.001 Blood Flow (-0.524, 0.144) (-0.773, 0.139) (-0.976, -0.242) 0.263 Total White Matter 0.194 0.224 0.073 0.066 0.065 0.123 (-0.025, 0.55) Lesion Volume (-0.012, 0.400) (-0.061, 0.509)

-4.65 7.85 -8.33 Gray Matter Volume 0.026 0.104 0.059 0.036 (-10.26, 0.95) (-0.31,16.01) (-16.11, -0.55) 1.26 -3.46 0.23 White Matter Volume 0.847 0.698 0.431 0.963 (-5.09, 7.61) (-12.07, 5.15) (-9.27, 9.72) Models adjusted for age, gender, antidepressant medication use, antianxiety medication use, educational attainment (less than high school, high school, greater than high school), HbA1C, BMI, pulse pressure, and history of CVD. Analyses of white matter lesion volume, gray matter volume, and white matter volume were additionally adjusted for total intracranial volume. Significant anxiety symptoms are defined as a Brief Symptom

Inventory Anxiety subscale score > 8; significant depression symptoms are defined as a 10 item Center for Epidemiologic Studies Depression scale score >10.

349

Reference group: n = 461 individuals with neither depression nor anxiety.

Analysis was performed using marginal models with generalized estimating equations. Results for both an overall 3 degrees of freedom (df) type

III test and for each anxiety/depression grouping as compared to the reference group are displayed.

350

Chapter 11

Associations between Type 2 Diabetes Status and Glycemic Control and Neuroimaging Measures in the Diabetes Heart Study-Mind

Laura M. Raffield, Amanda J. Cox, Barry I. Freedman, Christina E. Hugenschmidt, Fang-Chi Hsu, Joseph A. Maldjian, Donald W. Bowden

351

Abstract

Purpose: To examine the association of type 2 diabetes (T2D) affected status and measures of glycemic control and T2D severity, including fasting plasma glucose, glycated hemoglobin, and diabetes duration, with magenetic resonance imaging (MRI)-derived neuroimaging measures, including gray and white matter volume, white matter lesion volume, diffusion imaging measures, and cerebral blood flow, in the Diabetes Heart Study (DHS) Mind cohort.

Materials and Methods: These relationships were examined using marginal models with generalized estimating equations in 790 participants from 515 families in the DHS Mind.

Analysis of fasting plasma glucose, glycated hemoglobin, and diabetes duration was conducted in the 686 participants with T2D. Models were adjusted for potential confounders, including age, sex, history of cardiovascular disease, smoking, statin use, educational attainment, and blood pressure medication use.

Results: Adjusting for multiple comparisons, T2D affected status was associated with reduced white matter volume (p=1.94 x 10-6), increased white matter lesion volume (p=0.002), increased gray matter mean diffusivity (p=9.67 x 10-4), and decreased fractional anisotropy of the gray and white matter (p≤0.006) in fully adjusted models. However, among the participants affected by

T2D, neither fasting glucose, glycated hemoglobin, nor diabetes duration were associated with the neuroimaging measures assessed (p>0.016 in fully adjusted models).

Conclusion: While T2D affected status was significantly associated with a number of MRI derived neuroimaging measures, differences in glycemic control among individuals with T2D in the DHS Mind study do not appear to significantly contribute to variation in these measures. This suggests the presence or absence of T2D, not fine gradations of glycemic control among those affected, may be more significantly associated with age-related changes in the brain assessed using MRI.

352

Introduction

Type 2 diabetes (T2D) has been associated both with mild, age-related decrements in cognitive testing performance and with increased risk of Alzheimer’s disease and vascular dementia (Lu, Lin et al. 2009; Reijmer, van den Berg et al. 2010; Palta, Schneider et al. 2014).

This is reflected by changes in magnetic resonance imaging (MRI) derived neuroimaging measures in individuals with T2D; in prior studies T2D affected status has been associated with reduced brain volume, reduced white matter fractional anisotropy, and, less consistently, with increased white matter lesion volume (Falvey, Rosano et al. 2013; Moran, Phan et al. 2013;

Biessels and Reijmer 2014), all indicative of accelerated brain aging in T2D affected individuals.

Along with T2D affected status, glycemic control and duration of diabetes may play a role in variability in neuroimaging measures. Poorer glycemic control in both individuals with

T2D (Cukierman-Yaffe, Gerstein et al. 2009; Geijselaers, Sep et al. 2014) and the general population (Yaffe, Blackwell et al. 2006) is associated with lower performance on cognitive testing measures, though a recent systematic review showed that associations of glycemic control measures such as higher fasting glucose and HbA1c with poorer cognitive function tend to be weak, with HbA1c explaining less than 10% of the variation in cognitive function in most T2D cohorts (Geijselaers, Sep et al. 2014). Some analyses have also shown relationships between glycemic control and both cross-sectional brain volumes and progression of brain atrophy (van

Elderen, de Roos et al. 2010; Bryan, Bilello et al. 2014; Geijselaers, Sep et al. 2014), but results are mixed, and data for diffusion tensor imaging and cerebral blood flow measures are limited.

The Diabetes Heart Study Mind (DHS Mind) is a single-site, family-based study which assessed cognitive testing and neuroimaging measures in a population enriched for T2D. Here, we examine associations between neuroimaging measures, including gray and white matter volume, white matter lesion volume, diffusion imaging measures, and gray matter cerebral blood

353

flow, and T2D affected status in all European American participants in DHS Mind with available neuroimaging data, a total of 790 participants from 515 families. We extended these analyses to assess the relationship of fasting plasma glucose (FPG), glycated hemoglobin (HbA1C), and diabetes duration with neuroimaging measures available in 686 individuals with T2D from 506 families.

Methods and Materials

Study Design and Sample

Participants in the DHS were recruited from outpatient internal medicine and endocrinology clinics and from the community from 1998 through 2005 in western North

Carolina. Siblings affected by T2D without advanced renal insufficiency (serum creatinine concentrations >2.0 mg/dl) were recruited, along with additional non-diabetic siblings. Ascertainment and recruitment have been described (Bowden, Cox et al. 2010). T2D was defined as diabetes developing after the age of 35 years treated with changes in diet and exercise and/or oral agents in the absence of initial treatment solely with insulin and without historical evidence of diabetic ketoacidosis. Diabetes diagnosis was confirmed for all participants by review of medications and measurement of fasting glucose and HbA1C at the exam visit.

The DHS Mind study is an ancillary study to the DHS conducted between 2008 and 2013

(Hugenschmidt, Hsu et al. 2013; Cox, Hugenschmidt et al. 2014). Cognitive testing and neuroimaging were performed to investigate risk factors for cognitive decline in T2D.

Examinations included interviews for medical history and health behaviors, anthropometric measures, assessment of resting blood pressure, and fasting blood draws. Neuroimaging was performed in participants from the original DHS study (n=484), examined on average 6.69 ± 1.53 years (mean ± standard deviation(SD)) after their first study visit, and in newly recruited participants (n=306). Recruitment criteria in new participants were the same as in the original

354

DHS, except that siblings were not recruited. However, all new participants were required to have a first-degree relative affected by T2D. Fasting glucose and HbA1C used in the current analyses are from the fasting blood draws at the DHS Mind study visit. Diabetes duration was self-reported by participants from time of diabetes diagnosis to the DHS Mind visit. Participants were examined in the General Clinical Research Center of the Wake Forest Baptist Medical Center. All study protocols were approved by the Institutional Review Board at Wake Forest School of

Medicine, and all study procedures were completed in accordance with the Declaration of

Helsinki. Participants provided written informed consent prior to participation.

Neuroimaging

MR image acquisition. MR imaging was performed on a 1.5-T GE EXCITE HD scanner with twin-speed gradients using a neurovascular head coil (GE Healthcare, Milwaukee, WI) or a similar 3-T scanner. Imaging protocols have been described in detail previously (Raffield, Cox et al. 2014). Briefly, gray matter volume (GMV), white matter volume (WMV), and intracranial volume (ICV) (gray matter + white matter + cerebrospinal fluid) were determined from structural

T1 images using the VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm.html) automated segmentation procedure, which outputs values for native space total gray matter, white matter, and cerebrospinal fluid volumes. Diffusion tensor imaging (DTI) scalar metrics, including fractional anisotropy (FA) and mean diffusivity (MD) in the gray and white matter, were computed using the Diffusion Tensor Imaging ToolKit (DTI-TK)

(http://www.nitrc.org/projects/dtitk). Cerebral blood flow (CBF) perfusion images were generated using a previously described fully automated data processing pipeline (Maldjian, Laurienti et al.

2008), allowing derivation of the gray matter cerebral blood flow (GMCBF) measure analyzed.

White matter lesion segmentation was performed using the lesion segmentation toolbox (LST)

(Schmidt, Gaser et al. 2012) for SPM8 at a threshold (k) of 0.25, which has been previously validated in DHS-Mind (Maldjian, Whitlow et al. 2013). The total white matter lesion volume

355

(WMLV) measure used in these analyses was determined by summing the binary lesion maps and multiplying by the voxel volume.

In total, eight neuroimaging measures were analyzed in this study: GMV, WMV,

WMLV, GMCBF, GMMD, WMMD, GMFA, and WMFA. All analyses of GMV, WMV, and

WMLV included ICV as a covariate. In total, 783 individuals had data for GMV, 783 individuals for WMV, 783 individuals for ICV, 742 individuals for WMLV, 627 individuals for GMCBF,

671 individuals for GMMD, 671 individuals for WMMD, 773 individuals for GMFA, and 773 individuals for WMFA.

Statistical Analysis

Continuous variables were transformed as necessary to approximate normality, including analysis of the natural logarithm of (WMLV+1), the natural logarithm of FPG, the natural logarithm of HbA1C, and the square root of GMCBF. Relationships between T2D affected status, diabetes duration, FPG, and HbA1C and neuroimaging measures were examined using marginal models with generalized estimating equations. The models account for familial correlation using a sandwich estimator of the variance under exchangeable correlation. All individuals had data for

T2D affected status (n=790). Among individuals with T2D (n= 686), 646 individuals had data for

FPG, 646 individuals had data for HbA1c, and 651 individuals had data for diabetes duration. A

Bonferroni correction for multiple comparisons was used, with a threshold of p<0.0063 for

α=0.05 adjusting for the eight traits analysed. Models were adjusted for covariates including age, sex, history of CVD, smoking (never, past, current), statin use, educational attainment (less than high school, high school, greater than high school), high blood pressure medication use, and total intracranial volume as indicated. Covariates chosen were similar to those in a similar recent analysis by the ACCORD-MIND study to facilitate comparison of results between the studies

(Bryan, Bilello et al. 2014). All analyses were performed in SAS 9.3 (SAS Institute, Cary, NC).

356

Results

Table 1 summarizes the demographic and clinical characteristics of study participants, both for the whole sample (n=790), with a T2D prevalence of 86.8%, and for individuals affected by T2D only (n=686). Most individuals were overweight or obese with a high prevalence of hypertension and prior CVD events, as expected for a T2D-enriched cohort. For the individuals affected by T2D, diabetes duration in the cohort was 15.2  7.7 years (mean ± SD (standard deviation)), with a mean FPG of 146.9  54.1 mg/dl and a mean HbA1C of 7.5  1.4 %.

We first analysed the association of T2D affected status with eight neuroimaging measures in the DHS Mind cohort. Out of these eight measures, five (WMV, WMLV, GMFA,

WMFA, GMMD) were significantly associated (p ≤ 0.006) with T2D affected status in models adjusted for age, sex, history of CVD, smoking, statin use, educational attainment, high blood pressure medication use, and, where appropriate, total intracranial volume (Table 2). Information on covariate association statistics is available in Supplementary Table 1.

By contrast, in our analyses of FPG, HbA1C, and diabetes, no significant associations with the eight neuroimaging measures analysed were observed (Table 3). Some nominal associations which did not meet our multiple comparisons corrected significance threshold were observed, with for example higher FPG nominally associated (p=0.036) with lower GMCBF and higher

HbA1C nominally associated with higher gray matter mean diffusivity (p=0.016) in fully adjusted models (age, sex, history of CVD, smoking, statin use, educational attainment, high blood pressure medication use, and total intracranial volume if needed). While these nominally significant findings do indicate a trend towards poorer glycemic control associating with increased age-related changes in the brain, these findings were not statistically significant, in contrast to the signficant associations with T2D affected status for several measures. Information on covariate association statistics is available in Supplementary Table 2.

357

Discussion

In this analysis from the DHS Mind study, the association of T2D affected status and measures of glycemic control with a number of MRI derived neuroimaging traits, including assessments of brain volume, lesion volume, diffusion imaging measures, and cerebral blood flow, was assessed. Significant associations of T2D affected status with brain atrophy, increased white matter lesion volume, and changes in diffusion imaging measures indicative of reduced microstructural integrity were observed. However, among those affected by T2D, HbA1c, FPG, and diabetes duration did not associate with any differences in these neuroimaging measures.

These results suggest that factors other than glycemic control may be driving changes in the brain in T2D affected individuals from DHS Mind.

Comparing individuals affected by T2D to their unaffected siblings in the DHS Mind, the majority of neuroimaging measures assessed showed associations with T2D status suggestive of accelerated brain aging in those affected by T2D. One finding of note is the association of white matter lesion volume with T2D affected status in the DHS Mind cohort (p=0.002 in fully adjusted models). The association of T2D status with brain volume measures, for example the association with total white matter volume observed here (p=1.94 x 10-6 in fully adjusted models), is well established, but analyses of the association of T2D status with white matter lesion volume have been inconsistent (van Harten, de Leeuw et al. 2006; Falvey, Rosano et al. 2013; Moran, Phan et al. 2013; Biessels and Reijmer 2014). The observed association in the DHS Mind remained significant upon adjustment for smoking and use of hypertensive medications; smoking and hypertension, as well as increasing age, are widely thought to be significant contributors to white matter lesions (de Leeuw, de Groot et al. 2002; Longstreth, Arnold et al. 2005). Similar to most previous reports observing differences in white matter lesion volume with T2D affected status, the effect size observed in DHS Mind is modest, especially given the wide variation in white matter lesion volume in older adults (Biessels and Reijmer 2014). We also replicate the

358

previously reported relationship between reduced total white matter fractional anisotropy and

T2D affected status (Falvey, Rosano et al. 2013; Biessels and Reijmer 2014), with associations with diffusion tensor imaging parameters in the gray matter also observed.

Despite these associations with T2D affected status, we did not observe any relationships with measures of diabetes severity and glycemic control among those affected by T2D. A recent analysis from the Action to Control Cardiovascular Risk in Diabetes- Memory in Diabetes

(ACCORD MIND) study in patients with T2D found that increased FPG and longer diabetes duration were nominally associated with decreased gray matter volume, with longer diabetes duration also associated with increased abnormal white matter volume (Bryan, Bilello et al.

2014). We did not replicate these modest associations, perhaps due to differences in sample ascertainment for the DHS Mind, for example the longer mean diabetes duration (15.2 years vs

9.9 years) in the DHS participants. A systemic review of analyses of glycemic control in patients with T2D and brain atrophy and white matter lesions found the current literature to be inconclusive, with some studies finding an association between fasting glucose and brain volume but with HbA1c levels generally not associated with these neuroimaging measures (Geijselaers,

Sep et al. 2014). Other risk factors relevant to T2D which have previously been linked to changes in MRI-derived neuroimaging measures, such as hypertension (Schmidt, Launer et al. 2004), atherosclerosis (Geerlings, Appelman et al. 2010; Vidal, Sigurdsson et al. 2010), and adiposity

(Debette, Beiser et al. 2010; Verstynen 2013), may be contributing more to variation in these neuroimaging measures in the DHS- Mind than diabetes duration or glycemic control.

The current analysis has a number of strengths. While a number of neuroimaging analyses, notably in ACCORD MIND (Bryan, Bilello et al. 2014), have analyzed measures of glycemic control for associations with volumetric measures, such as total brain volume and white matter lesion volume, fewer studies have analysed diffusion tensor imaging and cerebral blood flow measures, an important addition here. ACCORD MIND also does not have non-diabetic

359

controls, making an analysis of the effect of diabetes affected status impossible in this cohort

(Bryan, Bilello et al. 2014). Some prior reports that were able to look at diabetes affected status used a semiquantitative rating scale to assess white matter lesions (van Harten, de Leeuw et al.

2006), as opposed to the quantitative measure of white matter lesion volume (Maldjian, Whitlow et al. 2013) used here, which may increase sensitivity of this measure. The DHS Mind is also a community based cohort, as opposed to a cohort of individuals recruited for a clinical trial like

ACCORD MIND, where there are many more participation exclusions. Thus DHS Mind may better reflect the impact of diabetes in the general population. Limitations of the current analysis include the cross-sectional nature, the relatively small number of non-diabetic individuals for the analysis of T2D affected status, and use of a single measure of HbA1c and fasting glucose.

In the DHS Mind study, T2D affected status is associated with age-related brain atrophy, white matter lesions, and diffusion tensor imaging measures of brain microstructure. However, in patients with T2D, differences in HbA1C, FPG, and diabetes duration do not associate strongly with these neuroimaging measures, suggesting that factors other than differences in glycemic control and diabetes duration may mediate the impact of T2D on brain structure.

360

Table 1: Demographic characteristics of 790 European American Diabetes Heart Study Mind participants included in the analyses, including 686 participants affected by type 2 diabetes (T2D). Standard deviation (SD), interquartile range (IQR).

Whole Sample (n=790) Participants with T2D (n=686) Mean  SD or % Median (IQR) Mean  SD or % Median (IQR) Demographics Age (years) 65.9  9.8 66.3 (59.6, 72.6) 65.8  9.8 66 (59.6, 72.3) Sex (% female) 54.1% 52.3% BMI (kg/m2) 32.2  6.5 31.4 (27.6, 36.1) 32.7  6.4 32 (28.1, 36.9) % smoking (past) 42.1% 42.5% % smoking (current) 11.1% 10.8% Systolic blood pressure (mm Hg) 130.4  17.6 128 (119, 142) 130.5  17.5 128.8 (119, 142) Diastolic blood pressure (mm Hg) 71.6  10.2 71 (64, 79) 71.3  10 71 (64, 78) Self-reported history of prior CVD 32.8% 35.8% Type 2 Diabetes Type 2 diabetes affected (%) 86.8% Diabetes duration (years) 15.2  7.7 13.5 (10.3, 19.2) 15.2  7.7 13.5 (10.3, 19.2) Glucose (mg/dl) 140.1  53.2 127 (105, 160) 146.9  54.1 135 (114, 169)

Hemoglobin A1C (%) 7.3  1.5 6.9 (6.2, 7.9) 7.5  1.4 7.2 (6.5, 8.1) Medications Oral anti-diabetic medication 57.3% 65.6% Insulin use 28.9% 33.1% Statin use 46.6% 48.9% Anti-hypertensive medication 66.1% 68.5% Education Less than high school 14.6% 15.4% High school 46.5% 44.9%

361

Greater than high school 39.0% 39.8% Cognition and Neuroimaging Measures Modified Mini Mental State Exam (3MSE) 91  7 93 (88, 96) 91  7 92 (87, 96) Gray matter volume (GMV) (cc) 521  55 519 (485, 555) 522  54 521 (486, 555) White matter volume (WMV) (cc) 571  70 564 (518, 618) 571  69 565 (518, 618) Intracranial volume (ICV) (cc) 1351  138 1339 (1252, 1447) 1355  136 1345 (1258, 1446) Fractional anisotropy gray matter (GMFA) 0.196  0.024 0.2 (0.176, 0.212) 0.195  0.024 0.199 (0.173, 0.211) Fractional anisotropy white matter (WMFA) 0.346  0.026 0.347 (0.328, 0.364) 0.344  0.027 0.345 (0.326, 0.363) Mean diffusivity gray matter (GMMD) 1.087  0.093 1.083 (1.023, 1.145) 1.09  0.096 1.088 (1.028, 1.149) Mean diffusivity white matter (WMMD) 0.79  0.05 0.789 (0.761, 0.818) 0.791  0.052 0.791 (0.762, 0.822) Total white matter lesion volume (WMLV) (cc) 4.47  8.32 1.46 (0.2, 4.97) 4.78  8.73 1.64 (0.3, 5.23) Gray matter cerebral blood flow (GMCBF) (mL/100g of 44.9  18.6 42.7 (33.5, 54.1) 45.1  19.3 42.6 (33.3, 54.2) tissue/min)

362

Table 2: Associations between neuroimaging variables and type 2 diabetes (T2D) affected status in the whole sample (n=790) (β value and standard error (SE)). Analysis was performed using marginal models with generalized estimating equations.

Model 1 Model 2 Model 3

β value SE p-value β value SE p-value β value SE p-value

White Matter Volume -12.35 2.36 1.70 x 10-7 -11.08 2.36 2.60 x 10-6 -11.55 2.43 1.94 x 10-6

Gray Matter Volume -4.14 2.53 0.102 -3.09 2.52 0.220 -2.47 2.59 0.339

Total White Matter Lesion Volume 0.31 0.08 1.77 x 10-4 0.26 0.08 0.002 0.27 0.09 0.002

Gray Matter Cerebral Blood Flow 0.04 0.13 0.770 0.19 0.12 0.118 0.15 0.12 0.235

Gray Matter Fractional Anisotropy -0.01 0.002 0.001 -0.01 0.002 8.68 x 10-4 -0.01 0.002 1.13 x 10-4

White Matter Fractional Anisotropy -0.005 0.002 0.041 -0.01 0.002 0.021 -0.01 0.002 0.006

Gray Matter Mean Diffusivity 0.03 0.01 3.05 x 10-5 0.02 0.01 7.79 x 10-4 0.02 0.01 9.67 x 10-4

White Matter Mean Diffusivity 0.01 0.004 0.056 0.01 0.004 0.098 0.01 0.004 0.081

Model 1 is adjusted for age. Model 2 is adjusted for age, sex, history of cardiovascular disease, smoking (never, past, current), statin use, and educational attainment (less than high school, high school, greater than high school). Model 3 is adjusted for age, sex, history of cardiovascular disease, smoking (never, past, current), statin use, educational attainment (less than high school, high school, greater than high school), and high blood pressure medication use. All models for white matter lesion volume, gray matter volume, and white matter volume additionally adjusted for total intracranial volume.

363

Table 3: Associations between neuroimaging variables and diabetes duration, fasting plasma glucose, and glycated hemoglobin (HbA1C) in individuals affected by type 2 diabetes (T2D) (n=686) (β value and standard error (SE)). Analysis was performed using marginal models with generalized estimating equations.

Model 1 Model 2 Model 3

Neuroimaging Trait Diabetes Measure β value SE p-value β value SE p-value β value SE p-value

White Matter Volume Diabetes Duration -0.11 0.13 0.403 -0.10 0.14 0.459 -0.13 0.14 0.341

Glucose -5.08 3.07 0.098 -5.89 3.25 0.070 -5.89 3.25 0.070

HbA1C -1.47 5.88 0.802 -2.12 5.87 0.718 -1.31 5.85 0.823

Gray Matter Volume Diabetes Duration -0.21 0.13 0.099 -0.17 0.13 0.184 -0.14 0.13 0.271

Glucose 1.20 2.81 0.670 1.67 2.90 0.565 1.65 2.86 0.565

HbA1C -7.10 5.62 0.206 -6.68 5.64 0.236 -7.46 5.59 0.182

Total White Matter Lesion Volume Diabetes Duration 0.011 0.006 0.041 0.010 0.006 0.075 0.010 0.006 0.067

Glucose 0.078 0.103 0.447 0.062 0.110 0.571 0.061 0.110 0.580

HbA1C 0.327 0.209 0.117 0.325 0.214 0.128 0.318 0.214 0.138

Gray Matter Cerebral Blood Flow Diabetes Duration 0.018 0.009 0.053 0.018 0.009 0.047 0.017 0.009 0.064

Glucose -0.382 0.177 0.030 -0.380 0.181 0.036 -0.377 0.180 0.036

HbA1C -0.576 0.317 0.069 -0.593 0.307 0.054 -0.581 0.306 0.057

Gray Matter Fractional Anisotropy Diabetes Duration 0.0001 0.0001 0.606 0.0001 0.0001 0.580 0.0000 0.0001 0.784

Glucose -0.002 0.003 0.620 -0.002 0.003 0.571 -0.002 0.003 0.577

HbA1C 0.001 0.006 0.914 -0.0001 0.006 0.982 0.001 0.006 0.916

White Matter Fractional Anisotropy Diabetes Duration -0.0001 0.0001 0.460 -0.0001 0.0001 0.463 -0.0001 0.0001 0.359

364

Glucose 0.002 0.003 0.430 0.002 0.003 0.483 0.002 0.003 0.495

HbA1C 0.003 0.006 0.623 0.003 0.006 0.665 0.003 0.006 0.584

Gray Matter Mean Diffusivity Diabetes Duration 0.0007 0.0004 0.095 0.0004 0.0004 0.250 0.0004 0.0004 0.274

Glucose 0.013 0.008 0.096 0.012 0.008 0.103 0.013 0.008 0.097

HbA1C 0.036 0.015 0.015 0.035 0.015 0.019 0.036 0.015 0.016

White Matter Mean Diffusivity Diabetes Duration 0.0002 0.0003 0.425 0 0.0003 0.908 0 0.0003 0.942

Glucose -0.003 0.004 0.570 -0.003 0.005 0.526 -0.003 0.005 0.528

HbA1C 0.0002 0.009 0.984 -0.002 0.009 0.854 -0.002 0.009 0.851

Model 1 is adjusted for age. Model 2 is adjusted for age, sex, history of cardiovascular disease, smoking (never, past, current), statin use, and educational attainment (less than high school, high school, greater than high school). Model 3 is adjusted for age, sex, history of cardiovascular disease, smoking (never, past, current), statin use, educational attainment (less than high school, high school, greater than high school), and high blood pressure medication use. All models for white matter lesion volume, gray matter volume, and white matter volume additionally adjusted for total intracranial volume.

365

Chapter 12 Summary and Conclusions

Individuals affected by T2D are at elevated risk of a number of comorbidities, for example heart disease and stroke, kidney failure, and chronic inflammation (CDC 2014; Donath

2014). Many of these comorbidities are often observed in the aging population, with individuals with T2D experiencing an earlier, elevated risk of morbidity and mortality as compared to a healthy aging population (Go, Mozaffarian et al. 2014; Gordon-Dseagu, Shelton et al. 2014).

However, risk is not equal among all individuals affected by T2D. Prior work in the DHS cohort has noted that even in individuals with very high predicted CVD risk (CAC scores >1000), mortality risk varies; >50% of these individuals have survived with this high CAC burden after an average 8.2 years of follow-up. Factors such as use of cholesterol-lowering medications and better kidney function were associated with increased odds of survival in this high risk group

(Cox, Hsu et al. 2014). In individuals with T2D from the DHS without such a high CAC burden, mortality risk was greatly attenuated (Agarwal, Morgan et al. 2011). Other studies in patients affected by T2D indicate that factors such as endothelial dysfunction and chronic inflammation

(de Jager, Dekker et al. 2006), variation in kidney function (Afkarian, Sachs et al. 2013), and adherence to a Mediterranean diet (Estruch, Ros et al. 2013) may significantly modify risk of

CVD and mortality. A significant literature implicates a genetic component to variation in diabetes related comorbidities (for example, (Wagenknecht, Bowden et al. 2001; Langefeld, Beck et al. 2004; Cho and Sobrin 2014)), but environmental and clinical factors, for example use of drugs like statins (Collins, Armitage et al. 2003) or frequency of physical activity (Tanasescu,

Leitzmann et al. 2003), can also clearly modify risk.

Through the studies in the DHS presented in this dissertation, we sought to gain insight into genetic and epidemiological risk factors which contribute to variation in measures such as vascular calcification, cognitive testing and neuroimaging measures, and mortality in this T2D affected cohort. We had hypothesized that both clinical risk factors and genetic variants,

366

including coding and noncoding SNPs, contribute significantly to aging and diabetes-related comorbidities in this population. The projects described in this work attempt to address this question using a variety of approaches. While we met with some success, our results highlight the need for further research in other cohorts of T2D affected individuals.

Our initial examinations focused on coding variants included on Illumina Infinium

HumanExome Beadchips available in the DHS study. These chips included >240,000 variants, mostly coding variants drawn from large exome sequencing analyses, but only 88,480 variants were available for analysis in the DHS due to the large number of monomorphic variants, most likely due to the rarity and in some cases population specificity of the variants included on the array. As described in Chapter 2, we had difficulty identifying coding variants that had reproducible associations with aging and T2D-related comorbidities; for example, in our initial analysis of variants in C1q/TNF superfamily genes and their binding partners and receptors, few variants were reproducibly associated with similar phenotypes when we attempted replication genotyping in cohorts from the Claude Pepper Older Americans Independence Center

Biospecimen Repository. Though this may be partly due to differences in cohort ascertainment, most coding variants may be too rare for associations with age-related phenotypes to be reliably detected in the DHS cohort. Our study is underpowered to detect modest associations of less common coding variants with our traits of interest; calculations in Quanto, a power calculation program for genetic studies available through USC (http://hydra.usc.edu/gxe), indicate that the

DHS study is underpowered to detect variants with a MAF less than 1%, with statistical power of only around 50% for a variant with a MAF of 1% and an effect size of 0.9 SD, which is larger than the effect sizes normally seen for SNPs. Power is obviously reduced even further for traits which were not assessed in the entire cohort, such as the measures of cognition and brain structure available in the DHS Mind study.

These concerns certainly also apply to our results from the exploratory Exome Chip analyses of neuroimaging measures described in Chapter 4. While we did find one locus which

367

reached Exome Chip-wide significance, a missense variant in PLEKHG4B significantly associated with white matter and gray matter mean diffusivity, this result must be considered preliminary given the small sample size and lack of an appropriate replication cohort. Chapter 3 describes -one way to potentially address these issues with power and reproducibility for rare variants. Combining evidence from both family-based linkage analysis and association analyses may be a way to identify important coding variants, and our preliminary results from the Exome

Chip highlight loci with potential functional relevance, including the UGT1A gene cluster with bilirubin and CDH13 with abdominal aortic calcification. This suggests combining results from linkage and association and focusing on loci with evidence for both may be of use even in studies with small families or sibships, like the DHS. This line of investigation will continue to be pursued through meta-analysis of Exome Chip two-point linkage data across multiple family studies involved in the Cohorts for Heart and Aging Research in Genomic Epidemiology

(CHARGE) consortium, which focuses on many phenotypes relevant to T2D, obesity, and CVD

(Psaty, O'Donnell et al. 2009). Inclusion of Exome Chip association analysis data in meta- analyses through CHARGE and other consortia, for example the Enhancing NeuroImaging

Genetics through Meta-Analysis (ENIGMA) consortium for the neuroimaging measures, may also be pursued (Thompson, Stein et al. 2014).

Apart from these analyses trying to identify novel coding variants, projects in this dissertation also focused on the role of noncoding variants previously identified by GWAS studies in non-T2D affected cohorts in the T2D affected individuals recruited for the DHS. The cumulative effect of GWAS identified variants for a particular trait were interrogated using a genetic risk score (GRS) approach. In Chapter 5, the focus was on SNPs associated with HDL levels: a recent analysis of HDL associated genetic risk variants in the general population had shown that these variants did not associate with risk of myocardial infarction (MI), casting doubt on the causal role of HDL in MI risk (Voight, Peloso et al. 2012). In the DHS, HDL GRS were not associated with risk of CVD mortality but were associated with CAC, suggesting a potential

368

role in CVD risk. We extended this approach in Chapter 6 to analyze GRS for CVD events and risk factors, for example blood pressure traits, body mass index, C-reactive protein, electrocardiogram traits, fasting plasma glucose, eGFR, and lipids, both individually and in aggregate for their impact on past CVD events, all-cause and CVD mortality, and CAC. These risk scores in aggregate were associated with CAC levels, but not with mortality or CVD events, indicating that these risk variants may have only modest impacts on CVD risk in individuals with

T2D. These GRS results from Chapters 5 and 6 suggest that the impacts of CVD risk variants identified in the general population may not be the same in T2D affected individuals

Clearly, there is still a great deal of work needed to better elucidate the genetics of important comorbidities in individuals with T2D. This is of particular importance given the rising prevalence of T2D, with over a quarter of adults over 65 already affected. To better understand the role of common variants in this population, GWAS studies conducted exclusively in T2D samples may reveal important contributors to risk which are not observed in general population cohorts. For example, recent work by Qi et al. revealed a variant related to glutamic acid metabolism that was consistently associated with CVD in individuals with diabetes, but not individuals unaffected by diabetes (Qi, Qi et al. 2013). Our group hopes to leverage the DHS data as well as data from T2D affected individuals in other cohorts from the CHARGE consortium to continue T2D-only GWAS analyses, with coronary artery calcification an important initial target.

A deeper understanding of both genetics and epidemiology are critical to increasing our ability to predict, prevent, or treat diabetes related comordities. To this end, other projects included in this dissertation focused on epidemiological contributors to risk of T2D related comorbidities. One of the recent goals of the DHS study has been to determine contributors to mortality in T2D affected individuals; this work has implicated factors such as CAC (Agarwal,

Morgan et al. 2011; Agarwal, Cox et al. 2013), biventricular volume (Cox, Hugenschmidt et al.

2013), and kidney function (Cox, Hsu et al. 2013). In Chapter 7, we explored the most important independent predictors of all-cause and CVD mortality in the DHS, using mortality data from an

369

average 9.6 years of follow-up. Age and sex were included in all models, but, other than these factors, CAC and urine albumin:creatinine ratio were consistently selected out of 24 potential predictors as the most important contributors to all-cause and CVD mortality risk. In our next analysis, presented in Chapter 8, we focused specifically on calcium intake from diet and supplements, an area of interest given some reports of negative CVD impacts of calcium supplement use in population-based studies (Bolland, Barber et al. 2008; Bolland, Avenell et al.

2010; Bolland, Grey et al. 2011; Li, Kaaks et al. 2012; Xiao, Murphy et al. 2013). In a T2D affected population, we found no impacts of calcium intake from diet and supplements on vascular calcification and in fact a slight reduction in all-cause mortality risk in women using calcium supplements and no impacts in men. Finally, in Chapter 9, we focused on the predictive power of CAC density for CVD risk, as a recent report found that, when already accounting for

CAC volume, CAC density was inversely associated with risk of incident CVD events (Criqui,

Denenberg et al. 2014). In T2D affected individuals from the DHS and AA-DHS studies, CAC density was consistently associated with elevated risk of mortality and increased odds of a history of CVD when analysed independently, but associations with CVD risk were inconsistent in models adjusted for plaque volume.

The last two analyses included in this dissertation were epidemiological assessments of the DHS-Mind dataset, with Chapter 10 describing the associations of self-reported anxiety and depression symptoms with cognitive testing and neuroimaging measures and Chapter 11 describing the associations of T2D affection status and measures of glycemic control with neuroimaging measures. Coincident anxiety and depression symptoms were associated with poorer cognitive performance and changes in a number of neuroimaging measures, similarly,

T2D affection status but not glycemic control associated with a number of neuroimaging measures. Again, the transferability of these results to other T2D-affected populations remains to be determined. The recent completion of a study in the AA-DHS cohort examining similar cognitive testing and neuroimaging measures to those examined in the DHS may allow for similar

370

future analyses, for example analysis of potential impacts of glycemic control on brain structure measures, in an additional T2D affected population

As the US population continues to age, it will become increasingly important to improve prevention and treatment for comorbidities such as CVD and dementia that affect many older adults with T2D. Better understanding of the genetic and epidemiological risk factors which predispose individuals to these and other aging associated disorders could allow for better risk prediction for individuals and more personalized care recommendations, as well as potentially leading to new therapies through the elucidation of which genes and pathways are important for a particular age-related trait.

371

Appendix A.1.

Analysis of Apolipoprotein E Polymorphisms and Alzheimer’s Disease Genetic Risk Score in the

Diabetes Heart Study

Laura M. Raffield, Amanda J Cox, Christina E Hugenschmidt, Barry I Freedman, Pamela J Hicks, Carl D Langefeld, Jeff D Williamson, Fang-Chi Hsu, Joseph A Maldjian, Donald W Bowden

372

Introduction

Genetic variants associated with Alzheimer’s disease risk may be of interest in individuals affected by type 2 diabetes (T2D). Individuals with T2D have an elevated risk of cognitive decline and dementia, and changes in magnetic resonance imaging measures in individuals with T2D have been described (Reijmer, van den Berg et al. 2010; Biessels and

Reijmer 2014). Single nucleotide polymorphisms (SNPs) in the APOE and TOMM40 loci, both of which have previously been associated with Alzheimer’s disease, were found in previous genome-wide association study (GWAS) to be associated with variation in multiple brain regions

(Potkin, Guffanti et al. 2009; Shen, Kim et al. 2010). Candidate gene analyses have also implicated the APOE locus in for example white matter hyperintensity burden (Schilling,

DeStefano et al. 2013), hippocampal and cerebral atrophy, and decreased cerebral blood flow

(Cherbuin, Leach et al. 2007). Alzheimer’s disease and vascular dementia, which is thought to be more strongly associated with diabetes status (Xu, Qiu et al. 2004), are thought to act synergistically in many individuals to produce cognitive dysfunction (Viswanathan, Rocca et al.

2009). We were interested to examine common variants previously identified in large general population cohorts as associated with risk of Alzheimer’s disease for potential associations with cognitive testing and magnetic resonance imaging (MRI) data in Diabetes Heart Study (DHS)

Mind.

To address these questions, we examined the major APOE haplotype defining variants

(rs7412 and rs429358) in the entire European American (EA) Diabetes Heart Study (DHS) Mind and a genetic risk score (GRS) of nine additional common variants previously associated with late onset Alzheimer’s disease in individuals with available Exome Chip and GWAS array data. This analysis was completed in May 2014; minor changes, in particular to cerebral blood flow (CBF) data, have been made to the DHS Mind dataset since this time. We also examined the APOE

373

variants in the DHS Classic cohort and African American (AA) DHS cohort for associations with lipid levels and cardiovascular disease (CVD) and glycemic control related traits.

Methods

Study Population

Recruitment of the DHS cohort and the additional recruitment of the AA DHS cohort have been described (Bowden, Cox et al. 2010). An additional DHS Mind visit with cognitive testing and

MRI has also been performed, including both individuals from the original DHS cohort and newly recruited EA individuals with T2D. Analysis of the Alzheimer’s disease genetic risk score was performed in individuals with both data from the DHS Mind visit and genetic data from both

GWAS and Exome Chip arrays (n=542). Analysis of APOE variants with cognitive testing and neuroimaging measures was performed in the entire DHS Mind cohort in all EA individuals with genotype data, and the analysis of CVD related traits was performed in the entire EA and AA

DHS populations separately.

Genotyping

Genotyping data for the Alzheimer’s disease genetic risk score was drawn both from the

Illumina Exome Chip and from SNPs imputed from an Affymetrix GWAS array available in the

DHS. These sources of genotyping data have been previously described (Cox, Hugenschmidt et al. 2014). Only one SNP from each locus was included (SNPs in strong linkage disequilibrium pruned). Variants were available for all independent loci included in the selected publications

(Hollingworth, Harold et al. 2011; Naj, Jun et al. 2011), though additional variants are now known (Lambert, Ibrahim-Verbaas et al. 2013). Similar to prior Alzheimer’s disease genetic risk score analyses, we excluded the APOE locus from the risk score due to its much greater estimated effect size than the other variants reported to be associated with Alzheimer’s disease risk (De

Jager, Shulman et al. 2012). The nine SNPs included in the risk score are listed in Table 1,

374

including the genotyping platform and coded risk allele. APOE haplotype defining variants

(rs7412 and rs429358) were genotyped using the well-validated Sequenom genotyping platform

(Buetow, Edmonson et al. 2001) or the Exome Chip array.

Statistical Analysis

Unweighted genetic risk scores were derived by adding the number of risk alleles possessed by an individual across nine loci. Relationships between genetic risk scores and traits of interest were examined using SAS 9.3 with marginal models with generalized estimating equations, accounting for familial correlation using a sandwich estimator of the variance under exchangeable correlation. Single SNP association analyses were performed in SOLAR 6.4.1 under the additive model. Cognitive testing models were adjusted for age, gender, T2D affected status, and educational attainment (less than high school, high school, greater than high school).

For neuroimaging measures, models were adjusted for age, gender, and T2D affected status, with models for gray matter volume, white matter volume, and total white matter lesion volume also adjusted for total intracranial volume. For CVD relevant traits, analyses were performed separately in European and African American participants in the DHS and were adjusted for age, sex, and T2D affected status. Note that analyses in African Americans were not adjusted for admixture.

Results

The Alzheimer’s disease genetic risk score was not associated (p>0.05) with any of the cognitive testing (Table 2) or neuroimaging measures (Table 3) available in DHS Mind. For the association analysis for the APOE variants in DHS Mind, no compelling associations were observed for the cognitive testing measures (Table 4), with a nominal association observed between the minor allele of rs429358 and lower scores on the Rey Auditory-Verbal Learning

Task (p=0.042). One significant association for the APOE variants with neuroimaging variables

375

was observed; higher gray matter mean diffusivity was significantly associated (p=0.006) with the minor allele of rs7412, with white matter mean diffusivity also more nominally associated

(p=0.021) (Table 5). Looking at association analyses with CVD relevant traits, including vascular calcification, lipid levels, and HbA1c, in EA participants, both rs7412 (p=0.043) and rs429358

(p=0.006) were associated with LDL, with rs429358 also associated with HDL (p=0.007). In AA participants, rs7412 was also strongly associated with LDL (p=0.0003), though rs429358 was not. rs429358 was also nominally associated with HDL (p=0.03) and triglycerides (p=0.016), with rs7412 associated with abdominal aortic calcification (p=0.005), HbA1c (p=0.002), and total cholesterol (p=0.018).

Conclusions

Genetic variants associated with late-onset Alzheimer’s disease risk do not seem to significantly contribute to variance in cognitive testing and neuroimaging measures in the DHS cohort. This is reflected by both the negative results of the Alzheimer’s disease genetic risk score analysis and analysis of coding variants in APOE. We did observed a number of associations in both AA and EA DHS participants of the APOE variants with lipid traits and CVD relevant measures, not surprisingly given the extensive literature linking these variants with dyslipidemia and CVD risk (Eichner, Dunn et al. 2002; Bennet, Di Angelantonio et al. 2007).

376

Table 1: SNPs included in Alzheimer’s disease genetic risk score.

Risk SNP Source Gene(s) Chromosome Position Alleles Allele rs3818361 Exome CR1 1 207784968 C/T T rs744373 Exome BIN1 2 127894615 C/T C rs9349407 Imputed CD2AP 6 47453378 C/G C rs11767557 Exome EPHA1 7 143109139 C/T T rs1532278 Exome CLU 8 27466315 C/T C rs610932 Exome MS4A6A, MS4A4E 11 59939307 A/C C rs561655 Exome PICALM 11 85800279 A/G A rs3764650 Exome ABCA7 19 1046520 G/T G rs3865444 Exome CD33 19 51727962 G/T G

377

Table 2: Association of the Alzheimer’s disease genetic risk score with cognitive testing measures in European American participants from the Diabetes Heart Study-Mind.

β Estimate Trait β Estimate p-value Standard N Error Modified Mini-Mental State Examination 0.231 0.151 0.127 541

Semantic fluency 0.128 0.182 0.482 503 Phonemic fluency -0.162 0.265 0.540 503 Stroop task 0.010 0.010 0.331 535 Rey Auditory-Verbal Learning Task 0.242 0.184 0.189 540

Digit Symbol Substitution Task 0.396 0.266 0.137 539

378

Table 3: Association of the Alzheimer’s disease genetic risk score with neuroimaging measures in European American participants from the Diabetes Heart Study-Mind.

β Trait β Estimate p-value N Estimate Standard Error White Matter Volume 0.325 0.572 0.570 456 Gray Matter Volume -1.080 0.579 0.062 456 White Matter Fractional Anisotropy -0.0005 0.0005 0.335 452 Gray Matter Fractional Anisotropy 0.939 0 0.0004 452 White Matter Mean Diffusivity 0.0002 0.001 0.846 452 Gray Matter Mean Diffusivity 0.003 0.002 0.151 452 White Matter Lesion Volume 0.020 0.021 0.356 423 Gray Matter Cerebral Blood Flow -0.020 0.034 0.551 433

379

Table 4: Association of rs7412 and rs429358 with cognitive testing measures in Diabetes Heart Study-Mind European American participants.

β Estimate β p- Trait SNP Standard N Estimate value Error Modified Mini-Mental State rs429358 -0.982 0.536 0.067 789 Examination Modified Mini-Mental State rs7412 0.044 0.525 0.933 762 Examination Semantic Fluency rs429358 -0.800 0.658 0.224 771 Semantic Fluency rs7412 0.383 0.643 0.551 742 Phonemic Fluency rs429358 1.430 0.910 0.116 772 Phonemic Fluency rs7412 -0.583 0.884 0.510 743 Stroop Task rs429358 -0.073 0.072 0.309 804 Stroop Task rs7412 0.050 0.070 0.474 776 Digit Symbol Substitution Task rs429358 -0.458 1.104 0.678 808 Digit Symbol Substitution Task rs7412 1.030 1.083 0.342 779 Rey Auditory-Verbal Learning Task rs429358 -1.469 0.723 0.042 810 Rey Auditory-Verbal Learning Task rs7412 0.010 0.704 0.989 781

380

Table 5: Association of rs7412 and rs429358 with cognitive testing measures in Diabetes Heart Study-Mind European American participants.

β β Estimate p- Trait SNP n Estimate Standard value Error Gray Matter Volume rs429358 0.380 2.202 0.863 716 Gray Matter Volume rs7412 -2.519 2.097 0.230 687 White Matter Volume rs429358 -1.389 3.031 0.647 717 White Matter Volume rs7412 -3.179 2.910 0.274 688 Fractional Anisotropy Gray Matter rs429358 -0.205 4.624 0.912 709 Fractional Anisotropy Gray Matter rs7412 0.000 4.475 1.000 680 Fractional Anisotropy White Matter rs429358 0.490 2.294 0.831 709 Fractional Anisotropy White Matter rs7412 0.940 2.181 0.666 680 Mean Diffusivity Gray Matter rs429358 -0.002 0.006 0.767 634 Mean Diffusivity Gray Matter rs7412 0.015 0.006 0.006 614 Mean Diffusivity White Matter rs429358 -0.002 0.004 0.580 633 Mean Diffusivity White Matter rs7412 0.009 0.004 0.021 613 White Matter Lesion Volume rs429358 0.132 0.073 0.072 676 White Matter Lesion Volume rs7412 0.026 0.069 0.713 647 Gray Matter Cerebral Blood Flow rs429358 0.028 0.123 0.819 591 Gray Matter Cerebral Blood Flow rs7412 -0.192 0.118 0.105 565

381

Table 6: Association of rs7412 and rs429358 with cardiovascular disease relevant traits in European American Diabetes Heart Study participants.

β β Estimate p- Trait SNP n Estimate Standard value Error Coronary Artery Calcification rs7412 0.283 0.179 0.115 1056 Coronary Artery Calcification rs429358 -0.171 0.161 0.288 1039 Carotid Artery Calcification rs7412 0.004 0.188 0.982 1050 Carotid Artery Calcification rs429358 -0.168 0.171 0.325 1033 Abdominal Aortic Calcification rs7412 -1.167 5.375 0.828 800 Abdominal Aortic Calcification rs429358 -5.356 4.903 0.275 790 Total Cholesterol rs7412 -0.003 0.017 0.860 1092 Total Cholesterol rs429358 0.017 0.016 0.294 1077 LDL rs7412 -5.132 2.533 0.043 1021 LDL rs429358 6.500 2.370 0.006 1004 HDL rs7412 0.098 0.064 0.128 1092 HDL rs429358 -0.161 0.060 0.007 1077 Triglycerides rs7412 0.066 0.042 0.120 1092 Triglycerides rs429358 -0.024 0.040 0.546 1077

HbA1c rs7412 0.022 0.014 0.126 1101

HbA1c rs429358 -0.020 0.013 0.129 1087

382

Table 7: Association of rs7412 and rs429358 with cardiovascular disease relevant traits in African American Diabetes Heart Study participants.

β Estimate β p- Trait SNP Standard n Estimate value Error Coronary Artery Calcification rs7412 -0.313 0.211 0.138 743 Coronary Artery Calcification rs429358 0.306 0.176 0.083 744 Carotid Artery Calcification rs7412 -0.206 0.184 0.263 741 Carotid Artery Calcification rs429358 0.148 0.154 0.336 742 Abdominal Aortic Calcification rs7412 -10.189 3.644 0.005 707 Abdominal Aortic Calcification rs429358 -0.691 3.077 0.822 708 Total Cholesterol rs7412 -0.042 0.018 0.018 743 Total Cholesterol rs429358 0.015 0.015 0.332 743 LDL rs7412 -9.746 2.704 0.0003 727 LDL rs429358 2.426 2.396 0.312 728 HDL rs7412 0.107 0.073 0.142 743 HDL rs429358 -0.134 0.062 0.030 743 Triglycerides rs7412 0.002 0.042 0.963 743 Triglycerides rs429358 0.085 0.035 0.016 743 HbA1c rs7412 -0.058 0.019 0.002 739 HbA1c rs429358 0.012 0.016 0.467 740

383

Appendix A.2.

Longitudinal Analysis of Thoracic Bone Mineral Density in the Diabetes Heart Study

Laura M. Raffield, J. Jeffrey Carr, Barry I. Freedman, Jianzhao Xu, Donald W. Bowden, Thomas

C. Register, Leon Lenchik

384

Introduction

While type 2 diabetes (T2D) is not thought to be associated with lower bone mineral density (BMD) (Register, Lenchik et al. 2006; Vestergaard 2007; Ma, Oei et al. 2012), individuals affected by T2D have been found to have an elevated risk of fracture (Janghorbani,

Dam et al. 2007), increasing interest in BMD measurements in this population. A recent study looked at change in BMD as assessed by dual-energy X-ray absorptiometry (DXA) and found that women with T2D experienced more rapid bone loss than those without T2D at the hip, spine, and calcaneus, perhaps contributing to fracture risk (Schwartz, Ewing et al. 2013). However, few if any studies have looked at longitudinal measures of BMD assessed by computed tomography

(CT) in patients with T2D. Here, we analyzed longitudinal changes in thoracic BMD in a small cohort with longitudinal measures from the DHS and looked at demographic and clinical factors that might contribute to variation in longitudinal BMD change.

Methods

Study Population

Change in BMD was analyzed in 152 European American individuals from the Diabetes

Heart Study (DHS) with data for thoracic vertebrae 5-7 at two visits. The average BMD for vertebrae 5-7 was used. Note that the thoracic BMD variable that has been used in previous DHS analyses averaged all available vertebrae, not just T5-T7, but limiting the analysis to T5-T7 was necessary so the same measure was assessed at both time points. Average time between visits was

3.79 ± 0.73 years (mean ± standard deviation), with a minimum time between visits of 2.49 years and a maximum time of 5.66 years.

385

Statistical Analysis

We assessed associations of some basic clinical and demographic factors with annual percent change in thoracic BMD (relative to baseline). Associations with annual percent change in thoracic BMD were examined using marginal models with generalized estimating equations, accounting for familial correlation using a sandwich estimator of the variance under exchangeable correlation in SAS 9.3. Men and women were analyzed separately, due to increased prevalence of osteoporosis and differences in bone physiology in women. Relationships with age, T2D affected status, osteoporosis medication use, fasting glucose, glycated hemoglobin (HbA1C), coronary artery calcification, body mass index (BMI), estrogen use (women only), and menopause status

(women only) at baseline were analyzed, as were relationships with annual percent change in

BMI, annual change in coronary artery calcification, and second visit osteoporosis medication use. The natural log of baseline coronary artery calcification, HbA1C, and fasting glucose was analyzed, as was the square root of baseline BMI. β values are reported for a one unit increase in continuous variables and for variable=1/yes for categorical variables.

Results

Few even nominally significant associations with annual percent change in thoracic BMD were observed (Table 1). In men, increased age was nominally associated with a lower percentage change in thoracic BMD (p=0.054) (i.e. a greater decrease in BMD), and higher baseline BMI was associated (p=0.004) with a higher percentage change (i.e. less decrease in BMD). Note that only one man was using osteoporosis medications at the second visit and none at baseline, so the analysis of osteoporosis medications is not very relevant to the men, despite the significant association with baseline medication use. In women, use of osteoporosis medications at baseline was associated with a lower percentage change in thoracic BMD (p=0.027), likely reflecting that

386

women with signs of preexisting osteoporosis were placed on these medications. Associations were not consistent between men and women.

Conclusions

Increases in thoracic BMD were seen in many individuals (Figure 1) from this small DHS cohort. We anticipate that these increases are mostly due to measurement error, as bone mineral density would be expected to decline with age. Other factors, such as changes in BMI, crush fractures, and drugs used to treat osteoporosis may be contributing to these counterintuitive increases in BMD, but given the small sample size and limited changes in BMD over the relatively short follow-up time, these factors would be hard to assess. No extreme outliers seemed to be affecting the analysis (Figure 1), so it was difficult to select scans for further review for potential errors. Few relationships between clinical and demographic factors and change in thoracic BMD were observed in either men or women (Table 1), which we anticipate was due to the limited change in BMD between time points, most likely due to short follow-up time, and the small sample size. We concluded that we were not able to accurately analyze what factors are contributing to changes in CT assessed thoracic BMD in this small sample from the DHS.

387

Figure 1: Distribution of annual percent change in thoracic bone mineral density, with histograms, boxplots, and QQ plots to evaluate normality.

A. Distribution of annual percent change in thoracic bone mineral density (measured in mg/cm3) in women (n=92).

B. Distribution of annual percent change in thoracic bone mineral density measured in mg/cm3) in men (n= 60).

388

Table 1: Relationships between clinical and demographic factors and annual percent change in thoracic bone mineral density, analyzed separately in men and women using marginal models with generalized estimating equations.

Women (n=92) Men (n= 60)

β Standard β Standard Trait p-value n p-value n Estimate Error Estimate Error Age at Baseline 0.032 0.028 0.243 92 -0.068 0.035 0.054 60 Type 2 Diabetes Affected Status at 0.399 0.509 0.433 92 -0.545 0.954 0.568 60 Baseline Baseline BMI 0.141 0.451 0.755 92 2.673 0.924 0.004 60 Annual % Change 0.164 0.108 0.129 92 -0.004 0.049 0.931 60 BMI Baseline HbA1c 0.291 0.962 0.762 92 -1.125 1.525 0.461 60 Baseline Fasting 0.532 0.705 0.451 92 -1.058 1.131 0.350 60 Glucose Baseline Osteoporosis -2.044 0.922 0.027 92 - - - - Medication Use Second Visit Osteoporosis 0.768 1.467 0.601 92 1.453 0.143 <0.0001 60 Medication Use Baseline Coronary Artery -0.053 0.090 0.557 92 -0.045 0.109 0.681 60 Calcification Annual Change Coronary Artery 0.000 0.001 0.870 92 0.000 0.002 0.937 59 Calcification Baseline Estrogen 0.295 0.528 0.576 92 - - - - Use Self-reported Menopause Status 0.158 0.513 0.759 78 - - - - at Baseline

389

Appendix A.3.

Association of Magnesium Intake with Vascular Calcification and Mortality in the Diabetes Heart

Study

Laura M. Raffield, Amanda J. Cox, Fang-Chi Hsu, J. Jeffrey Carr, Barry I. Freedman, Jianzhao

Xu, Donald W. Bowden, Mara Z. Vitolins

390

Introduction

By some estimates, 50-85% of the population of the United States has inadequate magnesium intake (Chaudhary, Sharma et al. 2010). A recent paper from Framingham Heart

Study found a significant inverse relationship between total magnesium intake and coronary artery calcification, with a trend towards lower abdominal aortic calcification in those with high total magnesium intake as well. This analysis was conducted in individuals with no prior cardiovascular disease and included very few individuals with type 2 diabetes (T2D) (Hruby,

O'Donnell et al. 2014). Low blood magnesium is common in individuals with T2D, and some studies have found an association between dietary magnesium intake and subsequent risk of T2D

(Chaudhary, Sharma et al. 2010), making relationships between magnesium and cardiovascular outcomes of particular interest in this population. Using a similar analysis strategy to the recent study from Framingham (Hruby, O'Donnell et al. 2014), relationships between magnesium intake and calcification in three separate vascular beds, including the coronary artery (CAC), carotid artery (CarCP), and abdominal aorto-iliac (AACP) vascular beds, were analysed in both individuals with T2D and their unaffected siblings from the Diabetes Heart Study (DHS). As an extension to this analysis, associations of magnesium with mortality were also assessed.

Methods

Study Population

We analysed magnesium intake in both T2D affected and unaffected individuals with food frequency questionnaire data from the DHS. Intake of magnesium from diet and supplements was determined using a self-administered Block food frequency questionnaire

(Block, Woods et al. 1990). The analysis was stratified by gender, as recommended magnesium intake requirements differ by gender (Chaudhary, Sharma et al. 2010). Recruitment of the cohort

391

and methods for assessing vascular calcification and mortality have previously been described

(Bowden, Cox et al. 2010; Raffield, Agarwal et al. 2014).

Statistical Analysis

For statistical analyses, continuous variables were transformed as necessary to approximate normality. For analyses of vascular calcified plaque, the natural logarithm of

(CAC+1) and (CarCP+1) and the square root of (AACP+10) were used. Dietary magnesium intake and dietary calcium, vitamin D, fiber, and saturated fat intake were adjusted for total energy intake using a residual method as suggested by Willett et al. (Willett, Howe et al. 1997).

Total magnesium intake was considered as an ordinal (four quartiles) variable. Intake for each quartile in mg/day is listed in Table 1. Relationships between magnesium intake and CAC,

CarCP, and AACP were examined using marginal models with generalized estimating equations, and relationships with all-cause and CVD mortality were examined using Cox proportional hazards models with sandwich-based variance estimation in SAS 9.3.

Associations were adjusted for covariates as indicated, including age, total energy intake

(kcal), body mass index (BMI) (kg/m2), smoking (never, former, or current), energy-adjusted total calcium intake from diet and supplements (mg/day), total cholesterol (mg/dL), HDL cholesterol (mg/dL), systolic blood pressure, high blood pressure medication use, lipid-lowering medication use, menopause status, estrogen use, oral diabetes medication use, insulin use, energy- adjusted vitamin D intake from diet and supplements (IU), energy-adjusted fiber intake (g/day), and energy-adjusted saturated fat intake (g/day) as indicated, similar to the analysis performed

Hruby et al. (Hruby, O'Donnell et al. 2014). Use of proton pump inhibitors was also adjusted for, as some studies have implicated these drugs in hypomagnesemia and problems with magnesium absorption (Park, Kim et al. 2014).

392

Results

Magnesium intake was not associated with vascular calcification in individuals with T2D in any of the three vascular beds assessed, with the exception of a nominal association (p=0.029) of higher magnesium intake with lower CAC in an unadjusted model in women (Table 2). Some associations with magnesium intake were observed in unaffected individuals, including association of higher magnesium intake with lower CarCP (p=0.004) and lower AACP (p=0.019) in women in models adjusted for age, total energy intake, and energy-adjusted total calcium intake, but this is based on a very small sample size (Table 2). In individuals with T2D, we also observed no association of magnesium intake with mortality (Table 3). A trend towards higher mortality risk in unaffected men with higher magnesium intake was observed (Table 3), but large confidence intervals and a small sample size make the significance of these results unclear.

We note that we would not overly trust results from Model 5, as there appeared to be model convergence issues for fiber and saturated fat intake and some issues with collinearity.

There were some model convergence issues for Model 4 in the unaffected individuals as well.

Conclusions

Unlike the results observed in a general population sample without CVD from the

Framingham Heart Study, we observed no compelling associations between magnesium intake and vascular calcification in participants with T2D. Some associations of higher magnesium intake with lower vascular calcification were observed in individuals unaffected by T2D, perhaps indicating heterogeneity in effects of magnesium intake due to T2D. However, sample size is very limited for the unaffected individuals, and some models are likely over-adjusted for covariates given the sample size. We feel that our sample is too small to make many firm conclusions, however, and larger samples will be needed to further elucidate potential links between magnesium intake and cardiovascular risk in T2D.

393

Table 1: Mean, minimum, and maximum total energy-adjusted magnesium intake by quartile (mg/day), both for women (n=360) and men (n=360) with type 2 diabetes (T2D) from the Diabetes Heart Study and unaffected women (n=103) and men (n=58) related to these Diabetes Heart Study participants.

Women with T2D (n=360) Men with T2D (n=360) Unaffected Women (n=103 ) Unaffected Men (n=58 ) Mean Minimum Maximum Mean Minimum Maximum Mean Minimum Maximum Mean Minimum Maximum Quartile 1 155.5 86.3 196.0 168.1 92.2 216.1 185.8 142.1 230.6 205.0 169.0 239.5 Quartile 2 236.7 198.2 282.7 253.6 216.3 295.7 272.1 231.3 317.4 283.3 247.1 314.9 Quartile 3 342.0 288.5 392.3 355.1 296.7 405.8 384.6 321.2 438.2 358.5 320.6 391.6 Quartile 4 489.9 392.8 1191.5 489.1 408.5 743.6 489.0 439.8 645.5 459.1 397.5 554.6

394

Table 2: Associations between total energy-adjusted magnesium intake quartiles and coronary artery, carotid artery, and aorto-iliac calcification, both in men and women with type 2 diabetes (T2D) and their unaffected siblings. Women with T2D (n=360) Men with T2D (n=360) Unaffected Women (n=103 ) Unaffected Men (n=58 )

β p- β p- β β Trait 95% CI 95% CI 95% CI p-value 95% CI p-value Estimate value Estimate value Estimate Estimate

Coronary Model artery -0.23 -0.44 -0.02 0.029 -0.14 -0.33 0.04 0.135 -0.18 -0.62 0.26 0.427 0.003 -0.57 0.57 0.991 1 calcification Model 0.07 -0.19 0.33 0.602 0.05 -0.18 0.27 0.696 -0.15 -0.55 0.26 0.477 -0.67 -1.68 0.33 0.190 2 Model 0.06 -0.20 0.32 0.637 0.05 -0.18 0.27 0.697 -0.19 -0.61 0.24 0.389 -0.70 -1.82 0.42 0.222 3 Model 0.07 -0.19 0.33 0.608 0.12 -0.10 0.35 0.286 -0.11 -0.74 0.53 0.742 -2.33 -3.02 -1.64 <0.0001 4 Model -0.39 -0.87 0.10 0.119 0.14 -0.35 0.62 0.583 3.42 1.76 5.08 <0.0001 0.67 -0.69 2.03 0.333 5 Carotid Model artery -0.09 -0.35 0.16 0.483 -0.15 -0.38 0.08 0.200 -0.50 -0.93 -0.06 0.025 0.22 -0.51 0.96 0.549 1 calcification Model -0.07 -0.34 0.20 0.603 -0.18 -0.49 0.12 0.243 -0.70 -1.17 -0.23 0.004 0.19 -0.47 0.84 0.575 2 Model -0.08 -0.34 0.19 0.578 -0.18 -0.49 0.12 0.244 -0.63 -1.09 -0.17 0.007 0.19 -0.43 0.82 0.545 3 Model -0.07 -0.33 0.19 0.592 -0.12 -0.42 0.17 0.405 -1.17 -1.69 -0.65 <.0001 0.43 -0.19 1.05 0.173 4 Model -0.03 -0.49 0.42 0.882 -0.27 -0.95 0.40 0.429 2.25 0.50 3.99 0.012 0.56 -0.66 1.78 0.369 5 Abdominal Model aorto-iliac -4.74 -11.26 1.79 0.155 0.43 -7.48 8.34 0.915 -9.45 -21.45 2.54 0.123 -2.55 -24.62 19.52 0.821 1 calcification Model 1.98 -5.11 9.07 0.584 -0.62 -10.62 9.39 0.904 -13.55 -24.86 -2.24 0.019 -9.31 -31.24 12.62 0.405 2 Model 2.01 -5.08 9.11 0.578 -0.71 -10.77 9.35 0.890 -12.45 -24.28 -0.62 0.039 -8.70 -30.39 13.00 0.432 3 Model 5.03 -2.31 12.37 0.179 2.83 -6.10 11.76 0.534 -12.14 -21.42 -2.86 0.010 -41.43 -45.31 -37.54 <0.0001 4 Model -3.29 -16.23 9.64 0.618 -1.43 -21.69 18.82 0.890 52.58 -2.90 108.06 0.063 -27.49 -58.47 3.49 0.082 5 Model 1: unadjusted

Model 2: adjusted age, total energy intake, energy-adjusted total calcium intake (mg/day)

395

Model 3: adjusted age, total energy intake, energy-adjusted total calcium intake (mg/day), protein pump inhibitor use

Model 4: all covariates in Model 3, body mass index, smoking (never, former, or current), total cholesterol (mg/dL), HDL cholesterol (mg/dL), systolic blood pressure, high blood pressure medication use, lipid-lowering medication use, estrogen use (in women only), menopause status (in women only), oral diabetes medication use (in T2D patients only), insulin use (in T2D patients only)

Model 5: all covariates in Model 4, energy-adjusted vitamin D from diet and supplements (IU), energy-adjusted fiber intake (g/day), energy- adjusted saturated fat intake (g/day)

396

Table 3: Associations between total energy-adjusted magnesium intake quartiles and cardiovascular disease (CVD) and all-cause mortality, both in men and women with type 2 diabetes (T2D) and their unaffected siblings. Models that did not converge are indicated as not applicable (NA).

Women with T2D (n=360) Men with T2D (n=360) Unaffected Women (n=103 ) Unaffected Men (n=58 )

Hazard Hazard Hazard Hazard Trait 95% CI p-value 95% CI p-value 95% CI p-value 95% CI p-value Ratio Ratio Ratio Ratio CVD Model 1 0.85 0.63 1.14 0.270 0.84 0.66 1.06 0.139 0.26 0.06 1.11 0.070 0.66 0.28 1.54 0.333 Mortality Model 2 0.91 0.65 1.27 0.579 1.01 0.76 1.35 0.942 0.58 0.17 2.05 0.402 7.80 2.14 28.45 0.002 Model 3 0.91 0.66 1.27 0.596 1.02 0.76 1.36 0.920 0.51 0.11 2.48 0.408 5.33 1.22 23.35 0.027 Model 4 0.89 0.60 1.32 0.573 0.94 0.69 1.27 0.674 NA NA NA NA 0.40 0.004 41.94 0.698 Model 5 1.07 0.33 3.47 0.916 0.62 0.33 1.19 0.150 NA NA NA NA 0.40 0.004 41.94 0.698

All-cause Model 1 1.00 1.00 1.00 0.472 1.00 1.00 1.00 0.216 0.99 0.98 1.00 0.053 1.00 0.99 1.00 0.205 Mortality Model 2 1.00 1.00 1.00 0.824 1.00 1.00 1.00 0.490 0.99 0.99 1.00 0.165 1.06 1.01 1.11 0.019 Model 3 1.00 1.00 1.00 0.843 1.00 1.00 1.00 0.466 0.99 0.99 1.00 0.212 1.06 1.02 1.09 0.0004 Model 4 1.00 1.00 1.00 0.755 1.00 1.00 1.00 0.371 0.99 0.98 1.00 0.128 2.45 x 1012 0 . 0.955 Model 5 0.00 0.00 5.06 x 105 0.388 17.70 0 1.39 x 109 0.757 NA NA NA NA NA NA NA NA

Model 1: unadjusted

Model 2: adjusted age, total energy intake, energy-adjusted total calcium intake (mg/day)

Model 3: adjusted age, total energy intake, energy-adjusted total calcium intake (mg/day), protein pump inhibitor use

Model 4: all covariates in Model 3, body mass index, smoking (never, former, or current), total cholesterol (mg/dL), HDL cholesterol (mg/dL), systolic blood pressure, high blood pressure medication use, lipid-lowering medication use, estrogen use (in women only), menopause status (in women only), oral diabetes medication use (in T2D patients only), insulin use (in T2D patients only)

Model 5: all covariates in Model 4, energy-adjusted vitamin D from diet and supplements (IU), energy-adjusted fiber intake (g/day), energy- adjusted saturated fat intake (g/day)

397

Appendix A.4.

Additional Genotyped Variants in the Diabetes Heart Study Cohort

398

Table 1: Additional genotyped single nucleotide polymorphisms (SNPs) in the Diabetes Heart Study.

Failed Monomorphic SNP Well rsID Genotyped Cohorts SNP? ? c1p36564213 C1q Well- March 2012 rs199786966 Passed Monomorphic DHS Classic c1p36564970 C1q Well- March 2012 no rsID-A/C Passed Monomorphic DHS Classic c3p161220921 C1q Well- March 2012 rs199985412 Passed Monomorphic DHS Classic c5p159776765 C1q Well- March 2012 rs372169471 Passed Polymorphic DHS Classic chr15:79065470 C1q Well- March 2012 rs200989422 Passed Monomorphic DHS Classic rs10977171 C1q Well- March 2012 rs10977171 Passed Polymorphic DHS Classic rs114477868 C1q Well- March 2012 rs114477868 Passed Monomorphic DHS Classic rs115558158 C1q Well- March 2012 rs115558158 Passed Monomorphic DHS Classic rs137951137 C1q Well- March 2012 rs137951137 Passed Monomorphic DHS Classic rs139171396 C1q Well- March 2012 rs139171396 Passed Monomorphic DHS Classic rs141230621 C1q Well- March 2012 rs141230621 Passed Monomorphic DHS Classic rs142009246 C1q Well- March 2012 rs142009246 Passed Monomorphic DHS Classic rs142463796 C1q Well- March 2012 rs142463796 Passed Monomorphic DHS Classic rs142960593 C1q Well- March 2012 rs142960593 Passed Monomorphic DHS Classic rs143543627 C1q Well- March 2012 rs143543627 Passed Monomorphic DHS Classic rs146980198 C1q Well- March 2012 rs146980198 Passed Monomorphic DHS Classic rs148785195 C1q Well- March 2012 rs148785195 Passed Polymorphic DHS Classic rs148931670 C1q Well- March 2012 rs148931670 Passed Monomorphic DHS Classic rs149507188 C1q Well- March 2012 rs149507188 Passed Monomorphic DHS Classic rs192265997 C1q Well- March 2012 rs192265997 Passed Polymorphic DHS Classic rs77786415 C1q Well- March 2012 rs77786415 Passed Polymorphic DHS Classic c10p88703106 C1q/TNF/HDL Well- June 2012 rs371610334 Failed DHS Classic c12p30879019 C1q/TNF/HDL Well- June 2012 rs370410608 Passed Monomorphic DHS Classic c13p24465488 C1q/TNF/HDL Well- June 2012 No rsID- A/C Passed Monomorphic DHS Classic

399

SNP c1p1178526 C1q/TNF/HDL Well- June 2012 rs372321598 Passed Polymorphic DHS Classic No rsID- A/C c1p1181935 C1q/TNF/HDL Well- June 2012 Failed DHS Classic SNP c1p22974271 C1q/TNF/HDL Well- June 2012 rs370014774 Passed Monomorphic DHS Classic c5p159776339 C1q/TNF/HDL Well- June 2012 rs367699652 Passed Monomorphic DHS Classic c5p159776764 C1q/TNF/HDL Well- June 2012 rs368748511 Passed Monomorphic DHS Classic No rsID- T/C c5p159797599 C1q/TNF/HDL Well- June 2012 Passed Monomorphic DHS Classic SNP c9p117552734 C1q/TNF/HDL Well- June 2012 No rsID Passed Monomorphic DHS Classic c9p117553057 C1q/TNF/HDL Well- June 2012 rs368084272 Passed Monomorphic DHS Classic rs113665219 C1q/TNF/HDL Well- June 2012 rs113665219 Passed Monomorphic DHS Classic rs138283703 C1q/TNF/HDL Well- June 2012 rs138283703 Failed DHS Classic

rs138433713 C1q/TNF/HDL Well- June 2012 rs138433713 Passed Monomorphic DHS Classic rs140629667 C1q/TNF/HDL Well- June 2012 rs140629667 Passed Monomorphic DHS Classic rs140840563 C1q/TNF/HDL Well- June 2012 rs140840563 Passed Monomorphic DHS Classic rs143197428 C1q/TNF/HDL Well- June 2012 rs143197428 Failed DHS Classic

rs143324819 C1q/TNF/HDL Well- June 2012 rs143324819 Passed Monomorphic DHS Classic rs145881139 C1q/TNF/HDL Well- June 2012 rs145881139 Passed Polymorphic DHS Classic rs146372752 C1q/TNF/HDL Well- June 2012 rs146372752 Passed Monomorphic DHS Classic rs146728851 C1q/TNF/HDL Well- June 2012 rs146728851 Failed DHS Classic

rs147185738 C1q/TNF/HDL Well- June 2012 rs147185738 Passed Monomorphic DHS Classic rs147423847 C1q/TNF/HDL Well- June 2012 rs147423847 Passed Monomorphic DHS Classic rs147649645 C1q/TNF/HDL Well- June 2012 rs147649645 Passed Monomorphic DHS Classic rs149058528 C1q/TNF/HDL Well- June 2012 rs149058528 Passed Monomorphic DHS Classic rs150816164 C1q/TNF/HDL Well- June 2012 rs150816164 Passed Monomorphic DHS Classic rs181362 C1q/TNF/HDL Well- June 2012 rs181362 Passed Polymorphic DHS Classic rs2293889 C1q/TNF/HDL Well- June 2012 rs2293889 Passed Polymorphic DHS Classic

400

rs2967605 C1q/TNF/HDL Well- June 2012 rs2967605 Failed DHS Classic

rs4082919 C1q/TNF/HDL Well- June 2012 rs4082919 Passed Polymorphic DHS Classic rs4846914 C1q/TNF/HDL Well- June 2012 rs4846914 Passed Polymorphic DHS Classic ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS ASB18RH422 rs200718582 Passed Polymorphic 2013 Mind chr13p3359117 ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs200063460 Passed Polymorphic 3 2013 Mind exm_rs1051873 ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs10518733 Failed 3 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS exm_rs491567 rs491567 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS exm510137 exm510137 Failed 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs1007211 rs1007211 Failed 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs10513788 rs10513788 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs11096990 rs11096990 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs115511178 rs115511178 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs11556510 rs11556510 Passed Monomorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs139596791 rs139596791 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs1421001 rs1421001 Failed 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs144124124 rs144124124 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs16983214 rs16983214 Failed 2013 Mind rs16988409 ApoE/Cognition Well1- February rs16988409 Passed Polymorphic DHS Classic, AA DHS, DHS

401

2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs2062162 rs2062162 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs2258689 rs2258689 Failed 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs35214987 rs35214987 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs3733280 rs3733280 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs3741489 rs3741489 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs429358 rs429358 Failed 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs4600514 rs4600514 Failed 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs61753666 rs61753666 Passed Polymorphic 2013 Mind ApoE/Cognition Well1- February DHS Classic, AA DHS, DHS rs7412 rs7412 Passed Polymorphic 2013 Mind DHS Classic, AA DHS, DHS exm755967 ApoE Well 2- March 2013 rs2282683 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs10499169 ApoE Well 2- March 2013 rs10499169 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs11088671 ApoE Well 2- March 2013 rs11088671 Failed Mind DHS Classic, AA DHS, DHS rs11214882 ApoE Well 2- March 2013 rs11214882 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs143332484 ApoE Well 2- March 2013 rs143332484 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs147377709 ApoE Well 2- March 2013 rs147377709 Failed Mind DHS Classic, AA DHS, DHS rs160277 ApoE Well 2- March 2013 rs160277 Passed Polymorphic Mind

402

DHS Classic, AA DHS, DHS rs16906038 ApoE Well 2- March 2013 rs16906038 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs17132289 ApoE Well 2- March 2013 rs17132289 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs188703 ApoE Well 2- March 2013 rs188703 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs2890109 ApoE Well 2- March 2013 rs2890109 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs34138687 ApoE Well 2- March 2013 rs34138687 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs41272737 ApoE Well 2- March 2013 rs41272737 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs429358 ApoE Well 2- March 2013 rs429358 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs4873803 ApoE Well 2- March 2013 rs4873803 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs4946672 ApoE Well 2- March 2013 rs4946672 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs683247 ApoE Well 2- March 2013 rs683247 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs6835064 ApoE Well 2- March 2013 rs6835064 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs6864869 ApoE Well 2- March 2013 rs6864869 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs76807842 ApoE Well 2- March 2013 rs76807842 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs7717527 ApoE Well 2- March 2013 rs7717527 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs822176 ApoE Well 2- March 2013 rs822176 Passed Polymorphic Mind DHS Classic, AA DHS, DHS rs9812284 ApoE Well 2- March 2013 rs9812284 Passed Polymorphic Mind

403

References

Abu Ruz, M. E., T. A. Lennie, et al. (2010). "Evidence that the brief symptom inventory can be used to measure anxiety quickly and reliably in patients hospitalized for acute myocardial infarction." J Cardiovasc Nurs 25(2): 117-123. ADA (2014). "Standards of medical care in diabetes--2014." Diabetes care 37 Suppl 1: S14-80. Adams, J. N., L. M. Raffield, et al. (2014). "Analysis of common and coding variants with cardiovascular disease in the Diabetes Heart Study." Cardiovasc Diabetol 13: 77. Adler, R. and R. Gill (2011). "Clinical utility of denosumab for treatment of bone loss in men and women." Clinical interventions in aging 6: 119-124. Afkarian, M., M. C. Sachs, et al. (2013). "Kidney disease and increased mortality risk in type 2 diabetes." J Am Soc Nephrol 24(2): 302-308. Agarwal, S., A. Cox, et al. (2013). "Coronary Calcium Score Predicts Cardiovascular Mortality in Diabetes: Diabetes Heart Study." Diabetes care 36(4): 972-977. Agarwal, S., T. Morgan, et al. (2011). "Coronary calcium score and prediction of all-cause mortality in diabetes: the diabetes heart study." Diabetes care 34(5): 1219-1224. Agatston, A. S., W. R. Janowitz, et al. (1990). "Quantification of coronary artery calcium using ultrafast computed tomography." J Am Coll Cardiol 15(4): 827-832. Aggarwal, B., S. Gupta, et al. (2012). "Historical perspectives on tumor necrosis factor and its superfamily: 25 years later, a golden journey." Blood 119(3): 651-665. Alluri, K., P. H. Joshi, et al. (2015). "Scoring of coronary artery calcium scans: History, assumptions, current limitations, and future directions." Atherosclerosis 239(1): 109-117. Almasy, L. and J. Blangero (1998). "Multipoint quantitative-trait linkage analysis in general pedigrees." American journal of human genetics 62(5): 1198-1211. Alvarez, J. A. and E. Emory (2006). "Executive function and the frontal lobes: a meta-analytic review." Neuropsychol Rev 16(1): 17-42. An, S., A. Hanley, et al. (2012). "Association between ADIPOQ SNPs with plasma adiponectin and glucose homeostasis and adiposity phenotypes in the IRAS Family Study." Molecular genetics and metabolism 107(4): 721-728. Andreescu, C., M. A. Butters, et al. (2008). "Gray matter changes in late life depression--a structural MRI analysis." Neuropsychopharmacology 33(11): 2566-2572. Ashburner, J. and K. J. Friston (2000). "Voxel-based morphometry--the methods." Neuroimage 11(6 Pt 1): 805-821. Atwood, L., P. Wolf, et al. (2004). "Genetic variation in white matter hyperintensity volume in the Framingham Study." Stroke; a journal of cerebral circulation 35(7): 1609-1613. Austin, M. P., P. Mitchell, et al. (2001). "Cognitive deficits in depression: possible implications for functional neuropathology." Br J Psychiatry 178: 200-206. Bacci, S., S. Rizza, et al. (2011). "The ENPP1 Q121 variant predicts major cardiovascular events in high-risk individuals: evidence for interaction with obesity in diabetic patients." Diabetes 60(3): 1000-1007. Bansal, V., O. Libiger, et al. (2010). "Statistical analysis strategies for association studies involving rare variants." Nature reviews. Genetics 11(11): 773-785. Barnes, A., M. Isohanni, et al. (2012). "No association of COMT (Val158Met) genotype with brain structure differences between men and women." PLoS ONE 7(3): e33964. Barnes, D. E., G. S. Alexopoulos, et al. (2006). "Depressive symptoms, vascular disease, and mild cognitive impairment: findings from the Cardiovascular Health Study." Arch Gen Psychiatry 63(3): 273-279. Bennet, A. M., E. Di Angelantonio, et al. (2007). "Association of apolipoprotein E genotypes with lipid levels and coronary risk." JAMA 298(11): 1300-1311. Benton, A., K. Hamsher, et al. (1994). Multilingual Aphasia Examination. Iowa City, AJA Associates.

404

Biessels, G. J. and Y. D. Reijmer (2014). "Brain Changes Underlying Cognitive Dysfunction in Diabetes: What Can We Learn From MRI?" Diabetes 63. Bijanki, K. R., A. N. Stillman, et al. (2013). "White matter fractional anisotropy is inversely related to anxious symptoms in older adults with atherosclerosis." Int J Geriatr Psychiatry 28(10): 1069-1076. Bild, D. E., R. Detrano, et al. (2005). "Ethnic differences in coronary calcification: the Multi- Ethnic Study of Atherosclerosis (MESA)." Circulation 111(10): 1313-1320. Block, G., M. Woods, et al. (1990). "Validation of a self-administered diet history questionnaire using multiple diet records." Journal of clinical epidemiology 43(12): 1327-1335. Blokland, G., G. de Zubicaray, et al. (2012). "Genetic and environmental influences on neuroimaging phenotypes: a meta-analytical perspective on twin imaging studies." Twin research and human genetics : the official journal of the International Society for Twin Studies 15(3): 351-371. Bolland, M., P. Barber, et al. (2008). "Vascular events in healthy older women receiving calcium supplementation: randomised controlled trial." BMJ (Clinical research ed.) 336(7638): 262-266. Bolland, M., A. Grey, et al. (2011). "Calcium supplements with or without vitamin D and risk of cardiovascular events: reanalysis of the Women's Health Initiative limited access dataset and meta-analysis." BMJ (Clinical research ed.) 342. Bolland, M. J., A. Avenell, et al. (2010). "Effect of calcium supplements on risk of myocardial infarction and cardiovascular events: meta-analysis." BMJ 341. Bolliger, M., D. Martinelli, et al. (2011). "The cell-adhesion G protein-coupled receptor BAI3 is a high-affinity receptor for C1q-like proteins." Proceedings of the National Academy of Sciences of the United States of America 108(6): 2534-2539. Bowden, D., S. An, et al. (2010). "Molecular basis of a linkage peak: exome sequencing and family-based analysis identify a rare genetic variant in the ADIPOQ gene in the IRAS Family Study." Human molecular genetics 19(20): 4112-4120. Bowden, D., A. Cox, et al. (2010). "Review of the Diabetes Heart Study (DHS) family of studies: a comprehensively examined sample for genetic and epidemiological studies of type 2 diabetes and its complications." The review of diabetic studies : RDS 7(3): 188-201. Bowden, D., M. Rudock, et al. (2006). "Coincident linkage of type 2 diabetes, metabolic syndrome, and measures of cardiovascular disease in a genome scan of the diabetes heart study." Diabetes 55(7): 1985-1994. Braskie, M., N. Jahanshad, et al. (2011). "Common Alzheimer's disease risk variant within the CLU gene affects white matter microstructure in young adults." The Journal of neuroscience : the official journal of the Society for Neuroscience 31(18): 6764-6770. Brown, E. R., R. A. Kronmal, et al. (2008). "Coronary calcium coverage score: determination, correlates, and predictive accuracy in the Multi-Ethnic Study of Atherosclerosis." Radiology 247(3): 669-675. Brunner, E. J., M. J. Shipley, et al. (2014). "Depressive disorder, coronary heart disease, and stroke: dose-response and reverse causation effects in the Whitehall II cohort study." Eur J Prev Cardiol 21(3): 340-346. Bryan, R. N., M. Bilello, et al. (2014). "Effect of Diabetes on Brain Structure: The Action to Control Cardiovascular Risk in Diabetes MR Imaging Baseline Data." Radiology 272(1): 210-216. Buetow, K. H., M. Edmonson, et al. (2001). "High-throughput development and characterization of a genomewide collection of gene-based single nucleotide polymorphism markers by chip-based matrix-assisted laser desorption/ionization time-of-flight mass spectrometry." Proc Natl Acad Sci U S A 98(2): 581-584. Byers, A. L. and K. Yaffe (2011). "Depression and risk of developing dementia." Nat Rev Neurol 7(6): 323-331.

405

Carlsson, L. E., S. Santoso, et al. (1999). "The alpha2 gene coding sequence T807/A873 of the platelet collagen receptor integrin alpha2beta1 might be a genetic risk factor for the development of stroke in younger patients." Blood 93(11): 3583-3586. Carmelli, D., C. DeCarli, et al. (1998). "Evidence for genetic variance in white matter hyperintensity volume in normal elderly male twins." Stroke; a journal of cerebral circulation 29(6): 1177-1181. Carr, J., T. Register, et al. (2008). "Calcified atherosclerotic plaque and bone mineral density in type 2 diabetes: the diabetes heart study." Bone 42(1): 43-52. Carr, J. J., J. R. Crouse, 3rd, et al. (2000). "Evaluation of subsecond gated helical CT for quantification of coronary artery calcium and comparison with electron beam CT." AJR Am J Roentgenol 174(4): 915-921. Carr, J. J., J. C. Nelson, et al. (2005). "Calcified coronary artery plaque measurement with cardiac CT in population-based studies: standardized protocol of Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in Young Adults (CARDIA) study." Radiology 234(1): 35-43. Carroll, M. D., D. A. Lacher, et al. (2005). "Trends in serum lipids and lipoproteins of adults, 1960-2002." JAMA 294(14): 1773-1781. CDC (2014). National Diabetes Statistics Report, 2014. Census, U. (2011). "The Older Population: 2010." Retrieved June 4, 2012, from http://www.census.gov/prod/cen2010/briefs/c2010br-09.pdf. Chaudhary, D. P., R. Sharma, et al. (2010). "Implications of magnesium deficiency in type 2 diabetes: a review." Biol Trace Elem Res 134(2): 119-129. Chaudhary, R., A. Likidlilid, et al. (2012). "Apolipoprotein E gene polymorphism: effects on plasma lipids and risk of type 2 diabetes and coronary artery disease." Cardiovascular diabetology 11: 36. Chehade, J. M., M. Gladysz, et al. (2013). "Dyslipidemia in type 2 diabetes: prevalence, pathophysiology, and management." Drugs 73(4): 327-339. Chen, C., K. Ito, et al. (2011). "Distinct expression patterns of the subunits of the CCR4-NOT deadenylase complex during neural development." Biochemical and biophysical research communications 411(2): 360-364. Chen, H., J. Meigs, et al. (2013). "Sequence kernel association test for quantitative traits in family samples." Genetic Epidemiology 37(2): 196-204. Chen, M. M., M. Crous-Bou, et al. (2014). "Exome-wide association study of endometrial cancer in a multiethnic population." PLoS ONE 9(5): e97045. Chen, Y., H. Zhu, et al. (2009). "Mapping growth patterns and genetic influences on early brain development in twins." Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer- Assisted Intervention 12(Pt 2): 232-239. Cherbuin, N., L. S. Leach, et al. (2007). "Neuroimaging and APOE genotype: a systematic qualitative review." Dement Geriatr Cogn Disord 24(5): 348-362. Chiang, M.-C., M. Barysheva, et al. (2009). "Genetics of brain fiber architecture and intellectual performance." The Journal of neuroscience : the official journal of the Society for Neuroscience 29(7): 2212-2224. Chiang, M.-C., M. Barysheva, et al. (2011). "BDNF gene effects on brain circuitry replicated in 455 twins." NeuroImage 55(2): 448-454. Cho, H. and L. Sobrin (2014). "Genetics of diabetic retinopathy." Curr Diab Rep 14(8): 515. Chung, C.-M., T.-H. Lin, et al. (2011). "A genome-wide association study reveals a quantitative trait locus of adiponectin on CDH13 that predicts cardiometabolic outcomes." Diabetes 60(9): 2417-2423. Chung, S. J., M. J. Kim, et al. (2014). "Exome array study did not identify novel variants in Alzheimer's disease." Neurobiol Aging 35(8): 1958 e1913-1954.

406

Claus, J., M. Breteler, et al. (1998). "Regional cerebral blood flow and cerebrovascular risk factors in the elderly population." Neurobiology of aging 19(1): 57-64. Coady, S. A., P. D. Sorlie, et al. (2001). "Validation of death certificate diagnosis for coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study." J Clin Epidemiol 54(1): 40-50. Collins, R., J. Armitage, et al. (2003). "MRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 5963 people with diabetes: a randomised placebo-controlled trial." Lancet 361(9374): 2005-2016. Cox, A., S. Agarwal, et al. (2012). "C-reactive protein concentration predicts mortality in type 2 diabetes: the Diabetes Heart Study." Diabetic medicine : a journal of the British Diabetic Association 29(6): 767-770. Cox, A., F.-C. Hsu, et al. (2014). "Genetic risk score associations with cardiovascular disease and mortality in the diabetes heart study." Diabetes care 37(4): 1157-1164. Cox, A., C. Hugenschmidt, et al. (2013). "Usefulness of Biventricular Volume as a Predictor of Mortality in Patients With Diabetes Mellitus (from the Diabetes Heart Study)." The American journal of cardiology 111(8): 1152-1158. Cox, A., A. Lehtinen, et al. (2013). "Polymorphisms in the Selenoprotein S gene and subclinical cardiovascular disease in the Diabetes Heart Study." Acta Diabetologica 50(3): 391-399. Cox, A. J., A. Azeem, et al. (2014). "Heart rate-corrected QT interval is an independent predictor of all-cause and cardiovascular mortality in individuals with type 2 diabetes: the Diabetes Heart Study." Diabetes care 37(5): 1454-1461. Cox, A. J., F. C. Hsu, et al. (2013). "Glomerular filtration rate and albuminuria predict mortality independently from coronary artery calcified plaque in the Diabetes Heart Study." Cardiovasc Diabetol 12: 68. Cox, A. J., F. C. Hsu, et al. (2014). "Contributors to mortality in high-risk diabetic patients in the Diabetes Heart Study." Diabetes care 37(10): 2798-2803. Cox, A. J., C. E. Hugenschmidt, et al. (2014). "Heritability and genetic association analysis of cognition in the Diabetes Heart Study." Neurobiol Aging 35(8): 1958.e1953-1958.e1912. Cox, A. J., M. C. Ng, et al. (2013). "Association of SNPs in the UGT1A gene cluster with total bilirubin and mortality in the Diabetes Heart Study." Atherosclerosis 229(1): 155-160. Criqui, M., A. Kamineni, et al. (2010). "Risk factor differences for aortic versus coronary calcified atherosclerosis: the multiethnic study of atherosclerosis." Arteriosclerosis, thrombosis, and vascular biology 30(11): 2289-2296. Criqui, M. H., J. O. Denenberg, et al. (2014). "Calcium density of coronary artery plaque and risk of incident cardiovascular events." JAMA 311(3): 271-278. Cukierman-Yaffe, T., H. C. Gerstein, et al. (2009). "Relationship between baseline glycemic control and cognitive function in individuals with type 2 diabetes and other cardiovascular risk factors: the action to control cardiovascular risk in diabetes-memory in diabetes (ACCORD-MIND) trial." Diabetes care 32(2): 221-226. De Cosmo, S., M. Copetti, et al. (2013). "Development and Validation of a Predicting Model of All-Cause Mortality in Patients With Type 2 Diabetes." Diabetes care 36(9): 2830-2835. de Jager, J., J. M. Dekker, et al. (2006). "Endothelial dysfunction and low-grade inflammation explain much of the excess cardiovascular mortality in individuals with type 2 diabetes: the Hoorn Study." Arterioscler Thromb Vasc Biol 26(5): 1086-1093. De Jager, P., J. Shulman, et al. (2012). "A genome-wide scan for common variants affecting the rate of age-related cognitive decline." Neurobiology of Aging 33(5): 10170-10115. de Leeuw, F. E., J. C. de Groot, et al. (2002). "Hypertension and cerebral white matter lesions in a prospective cohort study." Brain 125(Pt 4): 765-772. Debette, S., A. Beiser, et al. (2010). "Visceral fat is associated with lower brain volume in healthy middle-aged adults." Ann Neurol 68(2): 136-144.

407

Dehghan, A., J. Dupuis, et al. (2011). "Meta-analysis of genome-wide association studies in >80 000 subjects identifies multiple loci for C-reactive protein levels." Circulation 123(7): 731-738. Deibler, A. R., J. M. Pollock, et al. (2008). "Arterial spin-labeling in routine clinical practice, part 1: technique and artifacts." AJNR Am J Neuroradiol 29(7): 1228-1234. Deibler, A. R., J. M. Pollock, et al. (2008). "Arterial spin-labeling in routine clinical practice, part 2: hypoperfusion patterns." AJNR Am J Neuroradiol 29(7): 1235-1241. Deibler, A. R., J. M. Pollock, et al. (2008). "Arterial spin-labeling in routine clinical practice, part 3: hyperperfusion patterns." AJNR Am J Neuroradiol 29(8): 1428-1435. Detrano, R., A. Guerci, et al. (2008). "Coronary calcium as a predictor of coronary events in four racial or ethnic groups." The New England journal of medicine 358(13): 1336-1345. Detre, J. A., H. Rao, et al. (2012). "Applications of arterial spin labeled MRI in the brain." J Magn Reson Imaging 35(5): 1026-1037. Divers, J., N. D. Palmer, et al. (2013). "Admixture mapping of coronary artery calcified plaque in African Americans with type 2 diabetes mellitus." Circ Cardiovasc Genet 6(1): 97-105. Donath, M. Y. (2014). "Targeting inflammation in the treatment of type 2 diabetes: time to start." Nat Rev Drug Discov 13(6): 465-476. Doney, A. S. F., J. Dannfald, et al. (2009). "The FTO Gene Is Associated With an Atherogenic Lipid Profile and Myocardial Infarction in Patients With Type 2 Diabetes: A Genetics of Diabetes Audit and Research Study in Tayside Scotland (Go-DARTS) Study." Circulation: Cardiovascular Genetics 2(3): 255-259. Dong, X., S. Han, et al. (2001). "A diverse family of GPCRs expressed in specific subsets of nociceptive sensory neurons." Cell 106(5): 619-632. Doria, A., J. Wojcik, et al. (2008). "Interaction between poor glycemic control and 9p21 locus on risk of coronary artery disease in type 2 diabetes." JAMA 300(20): 2389-2397. Dotson, V. M., L. Beason-Held, et al. (2009). "Longitudinal study of chronic depressive symptoms and regional cerebral blood flow in older men and women." Int J Geriatr Psychiatry 24(8): 809-819. Duivis, H. E., N. Vogelzangs, et al. (2013). "Differential association of somatic and cognitive symptoms of depression and anxiety with inflammation: findings from the Netherlands Study of Depression and Anxiety (NESDA)." Psychoneuroendocrinology 38(9): 1573- 1585. Eichner, J. E., S. T. Dunn, et al. (2002). "Apolipoprotein E polymorphism and cardiovascular disease: a HuGE review." Am J Epidemiol 155(6): 487-495. Elias-Smale, S. E., R. V. Proenca, et al. (2010). "Coronary calcium score improves classification of coronary heart disease risk in the elderly: the Rotterdam study." J Am Coll Cardiol 56(17): 1407-1414. Erbel, R., S. Mohlenkamp, et al. (2010). "Coronary risk stratification, discrimination, and reclassification improvement based on quantification of subclinical coronary atherosclerosis: the Heinz Nixdorf Recall study." J Am Coll Cardiol 56(17): 1397-1406. Ertekin-Taner, N. (2010). "Genetics of Alzheimer disease in the pre- and post-GWAS era." Alzheimer's research & therapy 2(1): 3. Estruch, R., E. Ros, et al. (2013). "Primary prevention of cardiovascular disease with a Mediterranean diet." N Engl J Med 368(14): 1279-1290. Falvey, C., C. Rosano, et al. (2013). "Macro- and microstructural magnetic resonance imaging indices associated with diabetes among community-dwelling older adults." Diabetes care 36(3): 677-682. Farbstein, D. and A. P. Levy (2010). "The genetics of vascular complications in diabetes mellitus." Cardiol Clin 28(3): 477-496. Farvid, M., L. Qi, et al. (2014). "Phobic anxiety symptom scores and incidence of type 2 diabetes in US men and women." Brain Behav Immun 36: 176-182.

408

Ferket, B. S., B. J. van Kempen, et al. (2014). "Predictive value of updating Framingham risk scores with novel risk markers in the U.S. general population." PLoS ONE 9(2): e88312. Folsom, A., R. Kronmal, et al. (2008). "Coronary artery calcification compared with carotid intima-media thickness in the prediction of cardiovascular disease incidence: the Multi- Ethnic Study of Atherosclerosis (MESA)." Archives of internal medicine 168(12): 1333- 1339. Fontaine-Bisson, B., F. Renström, et al. (2010). "Evaluating the discriminative power of multi- trait genetic risk scores for type 2 diabetes in a northern Swedish population." Diabetologia 53(10): 2155-2162. Foote, J., S. Murphy, et al. (2003). "Factors associated with dietary supplement use among healthy adults of five ethnicities: the Multiethnic Cohort Study." American Journal of Epidemiology 157(10): 888-897. Fornage, M., S. Debette, et al. (2011). "Genome-wide association studies of cerebral white matter lesion burden: the CHARGE consortium." Annals of neurology 69(6): 928-939. Fox, C. S., K. Matsushita, et al. (2012). "Associations of kidney disease measures with mortality and end-stage renal disease in individuals with and without diabetes: a meta-analysis." Lancet 380(9854): 1662-1673. Foy, C., K. Wickley, et al. (2006). "The Reconditioning Exercise and Chronic Obstructive Pulmonary Disease Trial II (REACT II): rationale and study design for a clinical trial of physical activity among individuals with chronic obstructive pulmonary disease." Contemporary clinical trials 27(2): 135-146. Freedman, B. I., F. C. Hsu, et al. (2005). "The impact of ethnicity and sex on subclinical cardiovascular disease: the Diabetes Heart Study." Diabetologia 48(12): 2511-2518. Frikke-Schmidt, R., B. Nordestgaard, et al. (2008). "Association of loss-of-function mutations in the ABCA1 gene with high-density lipoprotein cholesterol levels and risk of ischemic heart disease." JAMA : the journal of the American Medical Association 299(21): 2524- 2532. Furney, S., A. Simmons, et al. (2011). "Genome-wide association with MRI atrophy measures as a quantitative trait locus for Alzheimer's disease." Molecular psychiatry 16(11): 1130- 1138. Gahche, J., R. Bailey, et al. (2011). "Dietary supplement use among U.S. adults has increased since NHANES III (1988-1994)." NCHS data brief(61): 1-8. Galbraith, M., A. Donner, et al. (2010). "CDK8: a positive regulator of transcription." Transcription 1(1): 4-12. Garel, S., M. Garcia-Dominguez, et al. (2000). "Control of the migratory pathway of facial branchiomotor neurones." Development (Cambridge, England) 127(24): 5297-5307. Garel, S., F. Marín, et al. (1999). "Ebf1 controls early cell differentiation in the embryonic striatum." Development (Cambridge, England) 126(23): 5285-5294. Ge, T., J. Feng, et al. (2012). "Increasing power for voxel-wise genome-wide association studies: the random field theory, least square kernel machines and fast permutation procedures." NeuroImage 63(2): 858-873. Geerlings, M., A. Appelman, et al. (2010). "Brain volumes and cerebrovascular lesions on MRI in patients with atherosclerotic disease. The SMART-MR study." Atherosclerosis 210(1): 130-136. Geijselaers, S. L., S. J. Sep, et al. (2014). "Glucose regulation, cognition, and brain MRI in type 2 diabetes: a systematic review." Lancet Diabetes Endocrinol. Giunti, S., G. Gruden, et al. (2012). "Increased QT interval dispersion predicts 15-year cardiovascular mortality in type 2 diabetic subjects: the population-based Casale Monferrato Study." Diabetes care 35(3): 581-583. Glassock, R. J. and A. D. Rule (2012). "The implications of anatomical and functional changes of the aging kidney: with an emphasis on the glomeruli." Kidney Int 82(3): 270-277.

409

Go, A. S., D. Mozaffarian, et al. (2014). "Heart disease and stroke statistics--2014 update: a report from the American Heart Association." Circulation 129(3): e28-e292. Go, A. S., D. Mozaffarian, et al. (2013). "Heart disease and stroke statistics--2013 update: a report from the American Heart Association." Circulation 127(1): e6-e245. Godin, O., C. Tzourio, et al. (2009). "Apolipoprotein E genotype is related to progression of white matter lesion load." Stroke; a journal of cerebral circulation 40(10): 3186-3190. Goldbourt, U., S. Yaari, et al. (1997). "Isolated low HDL cholesterol as a risk factor for coronary heart disease mortality. A 21-year follow-up of 8000 men." Arteriosclerosis, thrombosis, and vascular biology 17(1): 107-113. Gonzalez, J., S. Safren, et al. (2008). "Symptoms of depression prospectively predict poorer self- care in patients with Type 2 diabetes." Diabetic medicine : a journal of the British Diabetic Association 25(9): 1102-1107. Gordon-Dseagu, V. L., N. Shelton, et al. (2014). "Diabetes mellitus and mortality from all-causes, cancer, cardiovascular and respiratory disease: Evidence from the Health Survey for England and Scottish Health Survey cohorts." J Diabetes Complications 28(6): 791-797. Gordon, D., J. Probstfield, et al. (1989). "High-density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies." Circulation 79(1): 8-15. Gottesman, I. I. and T. D. Gould (2003). "The endophenotype concept in psychiatry: etymology and strategic intentions." American Journal of Psychiatry 160(4): 636-645. Gourraud, P.-A., M. Sdika, et al. (2013). "A genome-wide association study of brain lesion distribution in multiple sclerosis." Brain : a journal of neurology 136(Pt 4): 1012-1024. Haase, C., A. Tybjærg-Hansen, et al. (2012). "LCAT, HDL cholesterol and ischemic cardiovascular disease: a Mendelian randomization study of HDL cholesterol in 54,500 individuals." The Journal of clinical endocrinology and metabolism 97(2): E248-256. Haffner, S. and A. American Diabetes (2004). "Dyslipidemia management in adults with diabetes." Diabetes care 27 Suppl 1: S68-71. Hagman, J., J. Ramírez, et al. (2012). "B lymphocyte lineage specification, commitment and epigenetic control of transcription by early B cell factor 1." Current topics in microbiology and immunology 356: 17-38. Haines, J. L., M. A. Hauser, et al. (2005). "Complement factor H variant increases the risk of age- related macular degeneration." Science 308(5720): 419-421. Harrison, R., D. Holt, et al. (2004). "Are those in need taking dietary supplements? A survey of 21 923 adults." The British journal of nutrition 91(4): 617-623. Hellwege, J. N., N. D. Palmer, et al. (2015). "Empirical characteristics of family-based linkage to a complex trait: the ADIPOQ region and adiponectin levels." Hum Genet 134(2): 203- 213. Hellwege, J. N., N. D. Palmer, et al. (2014). "Genome-wide family-based linkage analysis of exome chip variants and cardiometabolic risk." Genet Epidemiol 38(4): 345-352. Hennekens, C. and E. Barice (2011). "Calcium supplements and risk of myocardial infarction: a hypothesis formulated but not yet adequately tested." The American Journal of Medicine 124(12): 1097-1098. Hoff, J., L. Quinn, et al. (2003). "The prevalence of coronary artery calcium among diabetic individuals without known coronary artery disease." Journal of the American College of Cardiology 41(6): 1008-1012. Hollingworth, P., D. Harold, et al. (2011). "Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP are associated with Alzheimer's disease." Nature genetics 43(5): 429-435. Honea, R., B. Verchinski, et al. (2009). "Impact of interacting functional variants in COMT on regional gray matter volume in human brain." NeuroImage 45(1): 44-51. Houx, P. J., J. Jolles, et al. (1993). "Stroop interference: aging effects assessed with the Stroop Color-Word Test." Exp Aging Res 19(3): 209-224.

410

Howie, B., P. Donnelly, et al. (2009). "A flexible and accurate genotype imputation method for the next generation of genome-wide association studies." PLoS Genetics 5(6). Hruby, A., C. J. O'Donnell, et al. (2014). "Magnesium intake is inversely associated with coronary artery calcification: the Framingham Heart Study." JACC Cardiovasc Imaging 7(1): 59-69. Hsu, F.-C., L. Lenchik, et al. (2005). "Heritability of body composition measured by DXA in the diabetes heart study." Obesity research 13(2): 312-319. Hsu, F. C., D. J. Zaccaro, et al. (2005). "The impact of pedigree structure on heritability estimates for pulse pressure in three studies." Hum Hered 60(2): 63-72. Hugenschmidt, C., F.-C. Hsu, et al. (2013). "The influence of subclinical cardiovascular disease and related risk factors on cognition in type 2 diabetes mellitus: The DHS-Mind study." Journal of diabetes and its complications 27(5): 422-428. Hugot, J. P., M. Chamaillard, et al. (2001). "Association of NOD2 leucine-rich repeat variants with susceptibility to Crohn's disease." Nature 411(6837): 599-603. Huyghe, J. R., A. U. Jackson, et al. (2013). "Exome array analysis identifies novel loci and low- frequency variants for insulin processing and secretion." Nat Genet 45(2): 197-201. Izaks, G., R. Gansevoort, et al. (2011). "The association of APOE genotype with cognitive function in persons aged 35 years or older." PLoS ONE 6(11): e27415. Jahanshad, N., O. Kohannim, et al. (2012). "Brain structure in healthy adults is related to serum transferrin and the H63D polymorphism in the HFE gene." Proceedings of the National Academy of Sciences 109(14): E851-859. Jahanshad, N., A. Lee, et al. (2010). "Genetic influences on brain asymmetry: a DTI study of 374 twins and siblings." NeuroImage 52(2): 455-469. Janghorbani, M., R. M. V. Dam, et al. (2007). "Systematic Review of Type 1 and Type 2 Diabetes Mellitus and Risk of Fracture." American Journal of Epidemiology 166. Jenkinson, M., C. Beckmann, et al. (2012). "FSL." NeuroImage 62(2): 782-790. Jenny, N. S. (2012). "Inflammation in aging: cause, effect, or both?" Discov Med 13(73): 451- 460. Johnston, M. E., Z. Zheng, et al. (2013). "Cerebral blood flow quantification in swine using pseudo-continuous arterial spin labeling." J Magn Reson Imaging 38(5): 1111-1118. Jongen, C. and G. Biessels (2008). "Structural brain imaging in diabetes: a methodological perspective." European journal of pharmacology 585(1): 208-218. Jonsson, T., H. Stefansson, et al. (2013). "Variant of TREM2 associated with the risk of Alzheimer's disease." The New England journal of medicine 368(2): 107-116. Joy, S., E. Kaplan, et al. (2004). "Speed and memory in the WAIS-III Digit Symbol--Coding subtest across the adult lifespan." Arch Clin Neuropsychol 19(6): 759-767. Jungtrakoon, P., N. Plengvidhya, et al. (2011). "Novel adiponectin variants identified in type 2 diabetic patients reveal multimerization and secretion defects." PLoS ONE 6(10). Kao, W.-T., Y. Wang, et al. (2010). "Common genetic variation in Neuregulin 3 (NRG3) influences risk for schizophrenia and impacts NRG3 expression in human brain." Proceedings of the National Academy of Sciences of the United States of America 107(35): 15619-15624. Kathiresan, S. and D. Srivastava (2012). "Genetics of human cardiovascular disease." Cell 148(6): 1242-1257. Katon, W., E. Lin, et al. (2010). "Comorbid depression is associated with an increased risk of dementia diagnosis in patients with diabetes: a prospective cohort study." Journal of general internal medicine 25(5): 423-429. Katon, W., C. R. Lyles, et al. (2012). "Association of depression with increased risk of dementia in patients with type 2 diabetes: the Diabetes and Aging Study." Archives of general psychiatry 69(4): 410-417.

411

Katula, J., S. Kritchevsky, et al. (2007). "Lifestyle Interventions and Independence for Elders pilot study: recruitment and baseline characteristics." Journal of the American Geriatrics Society 55(5): 674-683. Kessler, R. C., M. Gruber, et al. (2008). "Co-morbid major depression and generalized anxiety disorders in the National Comorbidity Survey follow-up." Psychol Med 38(3): 365-374. Khalil, A. A., L. A. Hall, et al. (2011). "The psychometric properties of the Brief Symptom Inventory depression and anxiety subscales in patients with heart failure and with or without renal dysfunction." Arch Psychiatr Nurs 25(6): 419-429. Kiezun, A., K. Garimella, et al. (2012). "Exome sequencing and the genetic basis of complex traits." Nature genetics 44(6): 623-630. Kikuno, R., T. Nagase, et al. (1999). "Prediction of the Coding Sequences of Unidentified Human Genes. XIV. The Complete Sequences of 100 New cDNA Clones from Brain Which Code for Large Proteins in vitro." DNA Research 6(3): 197-205. Kim, S., M. Kim, et al. (2013). "Regional cerebral perfusion in patients with Alzheimer's disease and mild cognitive impairment: effect of APOE epsilon4 allele." Neuroradiology 55(1): 25-34. Kishore, U., C. Gaboriaud, et al. (2004). "C1q and tumor necrosis factor superfamily: modularity and versatility." Trends in immunology 25(10): 551-561. Kleinridders, A., H. A. Ferris, et al. (2014). "Insulin Action in Brain Regulates Systemic Metabolism and Brain Function." Diabetes 63(7): 2232-2243. Kochunov, P., D. Glahn, et al. (2010). "Genetics of microstructure of cerebral white matter using diffusion tensor imaging." NeuroImage 53(3): 1109-1116. Kohannim, O., N. Jahanshad, et al. (2012). "Predicting white matter integrity from multiple common genetic variants." Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology 37(9): 2012-2019. Kohara, K., M. Fujisawa, et al. (2003). "MTHFR gene polymorphism as a risk factor for silent brain infarcts and white matter lesions in the Japanese general population: The NILS- LSA Study." Stroke; a journal of cerebral circulation 34(5): 1130-1135. Kohn, Y., N. Freedman, et al. (2007). "99mTc-HMPAO SPECT study of cerebral perfusion after treatment with medication and electroconvulsive therapy in major depression." J Nucl Med 48(8): 1273-1278. Kramer, C. K., B. Zinman, et al. (2013). "Coronary artery calcium score prediction of all cause mortality and cardiovascular events in people with type 2 diabetes: systematic review and meta-analysis." BMJ 346: f1654. Kunicki, T. J., S. A. Williams, et al. (2012). "Platelet adhesion to decorin but not collagen I correlates with the integrin alpha2 dimorphism E534K, the basis of the human platelet alloantigen (HPA)-5 system." Haematologica 97(5): 692-695. Kuo, M.-W., C.-H. Wang, et al. (2011). "Soluble THSD7A is an N-glycoprotein that promotes endothelial cell migration and tube formation in angiogenesis." PLoS ONE 6(12). Lambert, J. C., C. A. Ibrahim-Verbaas, et al. (2013). "Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer's disease." Nat Genet 45(12): 1452- 1458. Lamina, C., L. Forer, et al. (2012). "Evaluation of gene-obesity interaction effects on cholesterol levels: a genetic predisposition score on HDL-cholesterol is modified by obesity." Atherosclerosis 225(2): 363-369. Lange, L., D. Bowden, et al. (2002). "Heritability of carotid artery intima-medial thickness in type 2 diabetes." Stroke; a journal of cerebral circulation 33(7): 1876-1881. Langefeld, C. D., S. R. Beck, et al. (2004). "Heritability of GFR and albuminuria in Caucasians with type 2 diabetes mellitus." Am J Kidney Dis 43(5): 796-800.

412

Langsetmo, L., C. Berger, et al. (2013). "Calcium and vitamin D intake and mortality: results from the Canadian Multicentre Osteoporosis Study (CaMos)." The Journal of clinical endocrinology and metabolism 98(7): 3010-3018. Lee, S., T. Teslovich, et al. (2013). "General Framework for Meta-analysis of Rare Variants in Sequencing Association Studies." American journal of human genetics 93(1): 42-53. Lee, T. C., P. G. O'Malley, et al. (2003). "The prevalence and severity of coronary artery calcification on coronary artery computed tomography in black and white subjects." J Am Coll Cardiol 41(1): 39-44. Lehtinen, A., C. Newton-Cheh, et al. (2008). "Association of NOS1AP genetic variants with QT interval duration in families from the Diabetes Heart Study." Diabetes 57(4): 1108-1114. Leonard, B. (2007). "Inflammation, depression and dementia: are they connected?" Neurochemical research 32(10): 1749-1756. Levey, A. S., L. A. Stevens, et al. (2009). "A new equation to estimate glomerular filtration rate." Ann Intern Med 150(9): 604-612. Lezak, M., D. Howieson, et al. (2004). Neuropsychological Assessment. New York, Oxford University Press. Li, C., L. Barker, et al. (2008). "Diabetes and anxiety in US adults: findings from the 2006 Behavioral Risk Factor Surveillance System." Diabet Med 25(7): 878-881. Li, K., R. Kaaks, et al. (2012). "Associations of dietary calcium intake and calcium supplementation with myocardial infarction and stroke risk and overall cardiovascular mortality in the Heidelberg cohort of the European Prospective Investigation into Cancer and Nutrition study (EPIC-Heidelberg)." Heart (British Cardiac Society) 98(12): 920- 925. Lin, E., C. Rutter, et al. (2010). "Depression and advanced complications of diabetes: a prospective cohort study." Diabetes care 33(2): 264-269. Lloyd-Jones, D., R. Adams, et al. (2010). "Heart disease and stroke statistics--2010 update: a report from the American Heart Association." Circulation 121(7): e46-e215. Longstreth, W. T., Jr., A. M. Arnold, et al. (2005). "Incidence, manifestations, and predictors of worsening white matter on serial cranial magnetic resonance imaging in the elderly: the Cardiovascular Health Study." Stroke 36(1): 56-61. Lopez, L., M. Bastin, et al. (2012). "A genome-wide search for genetic influences and biological pathways related to the brain's white matter integrity." Neurobiology of aging 33(8): 18470-18414. Lorenz, M. W., H. S. Markus, et al. (2007). "Prediction of clinical cardiovascular events with carotid intima-media thickness: a systematic review and meta-analysis." Circulation 115(4): 459-467. Lu, F.-P., K.-P. Lin, et al. (2009). "Diabetes and the risk of multi-system aging phenotypes: a systematic review and meta-analysis." PLoS ONE 4(1). Luh, W. M., E. C. Wong, et al. (1999). "QUIPSS II with thin-slice TI1 periodic saturation: a method for improving accuracy of quantitative perfusion imaging using pulsed arterial spin labeling." Magn Reson Med 41(6): 1246-1254. Lupo, A., E. Cesaro, et al. (2011). "ZNF224: Structure and role of a multifunctional KRAB-ZFP protein." The international journal of biochemistry & cell biology 43(4): 470-473. Ma, L., L. Oei, et al. (2012). "Association between bone mineral density and type 2 diabetes mellitus: a meta-analysis of observational studies." Eur J Epidemiol 27(5): 319-332. Mahabadi, A. A., J. M. Massaro, et al. (2009). "Association of pericardial fat, intrathoracic fat, and visceral abdominal fat with cardiovascular disease burden: the Framingham Heart Study." Eur Heart J 30(7): 850-856. Mahajan, A., X. Sim, et al. (2015). "Identification and Functional Characterization of G6PC2 Coding Variants Influencing Glycemic Traits Define an Effector Transcript at the G6PC2-ABCB11 Locus." PLoS Genet 11(1): e1004876.

413

Mahdy Ali, K., A. Wonnerth, et al. (2012). "Cardiovascular disease risk reduction by raising HDL cholesterol--current therapies and future opportunities." British journal of pharmacology 167(6): 1177-1194. Maldjian, J., C. Whitlow, et al. (2013). "Automated white matter total lesion volume segmentation in diabetes." AJNR. American journal of neuroradiology 34(12): 2265- 2270. Maldjian, J. A., A. H. Baer, et al. (2009). "Fully automated processing of fMRI data in SPM: from MRI scanner to PACS." Neuroinformatics 7(1): 57-72. Maldjian, J. A., P. J. Laurienti, et al. (2008). "Clinical implementation of spin-tag perfusion magnetic resonance imaging." J Comput Assist Tomogr 32(3): 403-406. Manolio, T., F. Collins, et al. (2009). "Finding the missing heritability of complex diseases." Nature 461(7265): 747-753. Manson, J., M. Allison, et al. (2010). "Calcium/vitamin D supplementation and coronary artery calcification in the Women's Health Initiative." Menopause (New York, N.Y.) 17(4): 683-691. Matsuda, H. (2013). "Voxel-based Morphometry of Brain MRI in Normal Aging and Alzheimer's Disease." Aging and Disease. McClellan, J. and M.-C. King (2010). "Genetic heterogeneity in human disease." Cell 141(2): 210-217. McEwen, L. N., A. J. Karter, et al. (2012). "Predictors of mortality over 8 years in type 2 diabetic patients: Translating Research Into Action for Diabetes (TRIAD)." Diabetes care 35(6): 1301-1309. McGehee, B. E., J. M. Pollock, et al. (2012). "Brain perfusion imaging: How does it work and what should I use?" J Magn Reson Imaging 36(6): 1257-1272. McHale, M., J. Hendrikz, et al. (2008). "Screening for depression in patients with diabetes mellitus." Psychosomatic medicine 70(8): 869-874. Messier, S., C. Legault, et al. (2009). "The Intensive Diet and Exercise for Arthritis (IDEA) trial: design and rationale." BMC musculoskeletal disorders 10: 93. Mezuk, B., W. Eaton, et al. (2008). "Depression and type 2 diabetes over the lifespan: a meta- analysis." Diabetes care 31(12): 2383-2390. Moon, C. M., G. W. Kim, et al. (2014). "Whole-brain gray matter volume abnormalities in patients with generalized anxiety disorder: voxel-based morphometry." Neuroreport 25(3): 184-189. Moran, C., T. Phan, et al. (2013). "Brain Atrophy in Type 2 Diabetes: Regional distribution and influence on cognition." Diabetes care 36(12): 4036-4042. Morris, A. (2011). "Transethnic meta-analysis of genomewide association studies." Genetic Epidemiology 35(8): 809-822. Moskvina, V. and K. Schmidt (2006). "Individual SNP allele reconstruction from informative markers selected by a non-linear Gauss-type algorithm." Human heredity 62(2): 97-106. Mursu, J., K. Robien, et al. (2011). "Dietary supplements and mortality rate in older women: the Iowa Women's Health Study." Archives of internal medicine 171(18): 1625-1633. Naj, A., G. Jun, et al. (2011). "Common variants at MS4A4/MS4A6E, CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer's disease." Nature genetics 43(5): 436- 441. Nicklas, B. J., X. Wang, et al. (2009). "Effect of exercise intensity on abdominal fat loss during calorie restriction in overweight and obese postmenopausal women: a randomized, controlled trial." American Journal of Clinical Nutrition 89. Ninomiya, T., V. Perkovic, et al. (2009). "Albuminuria and kidney function independently predict cardiovascular and renal outcomes in diabetes." J Am Soc Nephrol 20(8): 1813-1821.

414

Nobuhara, K., G. Okugawa, et al. (2006). "Frontal white matter anisotropy and symptom severity of late-life depression: a magnetic resonance diffusion tensor imaging study." J Neurol Neurosurg Psychiatry 77(1): 120-122. Nordin, B., J. Lewis, et al. (2011). "The calcium scare--what would Austin Bradford Hill have thought?" Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA 22(12): 3073-3077. Nouwen, A., K. Winkley, et al. (2010). "Type 2 diabetes mellitus as a risk factor for the onset of depression: a systematic review and meta-analysis." Diabetologia 53(12): 2480-2486. Nucifora, P., R. Verma, et al. (2007). "Diffusion-tensor MR imaging and tractography: exploring brain microstructure and connectivity." Radiology 245(2): 367-384. O'Donnell, C. J., I. Chazaro, et al. (2002). "Evidence for heritability of abdominal aortic calcific deposits in the Framingham Heart Study." Circulation 106(3): 337-341. Org, E., S. Eyheramendy, et al. (2009). "Genome-wide scan identifies CDH13 as a novel susceptibility locus contributing to blood pressure determination in two European populations." Human molecular genetics 18(12): 2288-2296. Palta, P., A. L. Schneider, et al. (2014). "Magnitude of cognitive dysfunction in adults with type 2 diabetes: a meta-analysis of six cognitive domains and the most frequently reported neuropsychological tests within domains." J Int Neuropsychol Soc 20(3): 278-291. Park, C. H., E. H. Kim, et al. (2014). "The association between the use of proton pump inhibitors and the risk of hypomagnesemia: a systematic review and meta-analysis." PLoS ONE 9(11): e112558. Paynter, N., D. Chasman, et al. (2010). "Association between a literature-based genetic risk score and cardiovascular events in women." JAMA : the journal of the American Medical Association 303(7): 631-637. Pearson, T. A. and T. A. Manolio (2008). "How to interpret a genome-wide association study." JAMA 299(11): 1335-1344. Peloso, G. M., P. L. Auer, et al. (2014). "Association of low-frequency and rare coding-sequence variants with blood lipids and coronary heart disease in 56,000 whites and blacks." Am J Hum Genet 94(2): 223-232. Penke, L., S. Muñoz Maniega, et al. (2010). "White matter integrity in the splenium of the corpus callosum is related to successful cognitive aging and partly mediates the protective effect of an ancestral polymorphism in ADRB2." Behavior genetics 40(2): 146-156. Peper, J., R. Brouwer, et al. (2007). "Genetic influences on human brain structure: a review of brain imaging studies in twins." Human brain mapping 28(6): 464-473. Perreault, L. P., M. A. Legault, et al. (2014). "Comparison of genotype clustering tools with rare variants." BMC Bioinformatics 15: 52. Peterson, J., Z. Wei, et al. (2010). "C1q/TNF-related protein-3 (CTRP3), a novel adipokine that regulates hepatic glucose output." The Journal of biological chemistry 285(51): 39691- 39701. Pezzolesi, M. G. and A. S. Krolewski (2013). "The genetic risk of kidney disease in type 2 diabetes." Med Clin North Am 97(1): 91-107. Pietrzak, R. H., P. Maruff, et al. (2012). "Mild worry symptoms predict decline in learning and memory in healthy older adults: a 2-year prospective cohort study." Am J Geriatr Psychiatry 20(3): 266-275. Pitsavos, C., D. B. Panagiotakos, et al. (2006). "Anxiety in relation to inflammation and coagulation markers, among healthy adults: the ATTICA study." Atherosclerosis 185(2): 320-326. Pollock, J. M., A. R. Deibler, et al. (2008). "Migraine associated cerebral hyperperfusion with arterial spin-labeled MR imaging." AJNR Am J Neuroradiol 29(8): 1494-1497.

415

Pollock, J. M., A. R. Deibler, et al. (2008). "Arterial spin-labeled magnetic resonance imaging in hyperperfused seizure focus: a case report." J Comput Assist Tomogr 32(2): 291-292. Pollock, J. M., A. R. Deibler, et al. (2009). "Hypercapnia-induced cerebral hyperperfusion: an underrecognized clinical entity." AJNR Am J Neuroradiol 30(2): 378-385. Pollock, J. M., H. Tan, et al. (2009). "Arterial spin-labeled MR perfusion imaging: clinical applications." Magn Reson Imaging Clin N Am 17(2): 315-338. Pollock, J. M., C. T. Whitlow, et al. (2008). "Anoxic injury-associated cerebral hyperperfusion identified with arterial spin-labeled MR imaging." AJNR Am J Neuroradiol 29(7): 1302- 1307. Pollock, J. M., C. T. Whitlow, et al. (2011). "Response of arteriovenous malformations to gamma knife therapy evaluated with pulsed arterial spin-labeling MRI perfusion." AJR Am J Roentgenol 196(1): 15-22. Pollock, J. M., C. T. Whitlow, et al. (2009). "Pulsed arterial spin-labeled MR imaging evaluation of tuberous sclerosis." AJNR Am J Neuroradiol 30(4): 815-820. Polonsky, T. S., R. L. McClelland, et al. (2010). "Coronary artery calcium score and risk classification for coronary heart disease prediction." JAMA 303(16): 1610-1616. Porchay-Baldérelli, I., F. Péan, et al. (2007). "The CETP TaqIB polymorphism is associated with the risk of sudden death in type 2 diabetic patients." Diabetes care 30(11): 2863-2867. Porchay-Baldérelli, I., F. Péan, et al. (2009). "Relationships between common polymorphisms of adenosine triphosphate–binding cassette transporter A1 and high-density lipoprotein cholesterol and coronary heart disease in a population with type 2 diabetes mellitus." Metabolism 58: 74-79. Potkin, S., G. Guffanti, et al. (2009). "Hippocampal atrophy as a quantitative trait in a genome- wide association study identifying novel susceptibility genes for Alzheimer's disease." PLoS ONE 4(8). Prentice, R., M. Pettinger, et al. (2013). "Health risks and benefits from calcium and vitamin D supplementation: Women's Health Initiative clinical trial and cohort study." Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA 24(2): 567-580. Psaty, B. M., C. J. O'Donnell, et al. (2009). "Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium: Design of prospective meta-analyses of genome- wide association studies from 5 cohorts." Circ Cardiovasc Genet 2(1): 73-80. Purcell, S., B. Neale, et al. (2007). "PLINK: a tool set for whole-genome association and population-based linkage analyses." American journal of human genetics 81(3): 559-575. Pyykkonen, A. J., K. Raikkonen, et al. (2012). "Association between depressive symptoms and metabolic syndrome is not explained by antidepressant medication: results from the PPP- Botnia Study." Ann Med 44(3): 279-288. Qi, L., L. Parast, et al. (2011). "Genetic susceptibility to coronary heart disease in type 2 diabetes: 3 independent studies." J Am Coll Cardiol 58(25): 2675-2682. Qi, L., Q. Qi, et al. (2013). "Association between a genetic variant related to glutamic acid metabolism and coronary heart disease in individuals with type 2 diabetes." JAMA 310(8): 821-828. Qi, Q., L. Liang, et al. (2012). "Genetic predisposition to dyslipidemia and type 2 diabetes risk in two prospective cohorts." Diabetes 61(3): 745-752. Qi, Q., T. Workalemahu, et al. (2012). "Genetic variants, plasma lipoprotein(a) levels, and risk of cardiovascular morbidity and mortality among two prospective cohorts of type 2 diabetes." Eur Heart J 33(3): 325-334. Raffield, L., A. Cox, et al. (2013). "Impact of HDL genetic risk scores on coronary artery calcified plaque and mortality in individuals with type 2 diabetes from the Diabetes Heart Study." Cardiovascular diabetology 12: 95.

416

Raffield, L. M., S. Agarwal, et al. (2014). "Cross-sectional analysis of calcium intake for associations with vascular calcification and mortality in individuals with type 2 diabetes from the Diabetes Heart Study." Am J Clin Nutr 100(4): 1029-1035. Raffield, L. M., A. J. Cox, et al. (2014). "Heritability and genetic association analysis of neuroimaging measures in the Diabetes Heart Study." Neurobiol Aging. Raggi, P., L. J. Shaw, et al. (2004). "Prognostic value of coronary artery calcium screening in subjects with and without diabetes." J Am Coll Cardiol 43(9): 1663-1669. Register, T. C., L. Lenchik, et al. (2006). "Type 2 diabetes is not independently associated with spinal trabecular volumetric bone mineral density measured by QCT in the Diabetes Heart Study." Bone 39. Reijmer, Y., E. van den Berg, et al. (2010). "Cognitive dysfunction in patients with type 2 diabetes." Diabetes/metabolism research and reviews 26(7): 507-519. Ridker, P. M. (2007). "C-reactive protein and the prediction of cardiovascular events among those at intermediate risk: moving an inflammatory hypothesis toward consensus." J Am Coll Cardiol 49(21): 2129-2138. Ridker, P. M., N. Rifai, et al. (2000). "Plasma concentration of interleukin-6 and the risk of future myocardial infarction among apparently healthy men." Circulation 101(15): 1767-1772. Ripatti, S., E. Tikkanen, et al. (2010). "A multilocus genetic risk score for coronary heart disease: case-control and prospective cohort analyses." Lancet 376(9750): 1393-1400. Roy, T., C. Lloyd, et al. (2012). "Screening tools used for measuring depression among people with Type 1 and Type 2 diabetes: a systematic review." Diabetic medicine : a journal of the British Diabetic Association 29(2): 164-175. Samelson, E., S. Booth, et al. (2012). "Calcium intake is not associated with increased coronary artery calcification: the Framingham Study." The American journal of clinical nutrition 96(6): 1274-1280. Santoso, S., T. J. Kunicki, et al. (1999). "Association of the platelet glycoprotein Ia C807T gene polymorphism with nonfatal myocardial infarction in younger patients." Blood 93(8): 2449-2453. Scherrer, J. F., T. Chrusciel, et al. (2010). "Anxiety disorders increase risk for incident myocardial infarction in depressed and nondepressed Veterans Administration patients." Am Heart J 159(5): 772-779. Schilling, S., A. L. DeStefano, et al. (2013). "APOE genotype and MRI markers of cerebrovascular disease: systematic review and meta-analysis." Neurology 81(3): 292- 300. Schipper, H. (2011). "Apolipoprotein E: implications for AD neurobiology, epidemiology and risk assessment." Neurobiology of Aging 32(5): 778-790. Schmidt, P., C. Gaser, et al. (2012). "An automated tool for detection of FLAIR-hyperintense white-matter lesions in Multiple Sclerosis." NeuroImage 59(4): 3774-3783. Schmidt, R., L. Launer, et al. (2004). "Magnetic resonance imaging of the brain in diabetes: the Cardiovascular Determinants of Dementia (CASCADE) Study." Diabetes 53(3): 687- 692. Schmidt, R., H. Schmidt, et al. (2000). "MRI cerebral white matter lesions and paraoxonase PON1 polymorphisms : three-year follow-up of the austrian stroke prevention study." Arteriosclerosis, thrombosis, and vascular biology 20(7): 1811-1816. Schmidt, R., H. Schmidt, et al. (2001). "Angiotensinogen polymorphism M235T, carotid atherosclerosis, and small-vessel disease-related cerebral abnormalities." Hypertension 38(1): 110-115. Schwartz, A. V., S. K. Ewing, et al. (2013). "Diabetes and change in bone mineral density at the hip, calcaneus, spine, and radius in older women." Front Endocrinol (Lausanne) 4: 62.

417

Seidel, K., S. Kirsch, et al. (2011). "The promyelocytic leukemia zinc finger (PLZF) protein exerts neuroprotective effects in neuronal cells and is dysregulated in experimental stroke." Brain pathology (Zurich, Switzerland) 21(1): 31-43. Serretti, A. and L. Mandelli (2010). "Antidepressants and body weight: a comprehensive review and meta-analysis." J Clin Psychiatry 71(10): 1259-1272. Sexton, C. E., C. E. Mackay, et al. (2013). "A systematic review and meta-analysis of magnetic resonance imaging studies in late-life depression." Am J Geriatr Psychiatry 21(2): 184- 195. Shah, S., J. P. Casas, et al. (2013). "Influence of common genetic variation on blood lipid levels, cardiovascular risk, and coronary events in two British prospective cohort studies." Eur Heart J 34(13): 972-981. Shah, S., J. P. Casas, et al. (2012). "Influence of common genetic variation on blood lipid levels, cardiovascular risk, and coronary events in two British prospective cohort studies." European Heart Journal 34(13): 972-981. Sharma, R., S. Prudente, et al. (2011). "The type 2 diabetes and insulin-resistance locus near IRS1 is a determinant of HDL cholesterol and triglycerides levels among diabetic subjects." Atherosclerosis 216(1): 157-160. Shaw, L. J., P. Raggi, et al. (2003). "Prognostic value of cardiac risk factors and coronary artery calcium screening for all-cause mortality." Radiology 228(3): 826-833. Shen, L., S. Kim, et al. (2010). "Whole genome association study of brain-wide imaging phenotypes for identifying quantitative trait loci in MCI and AD: A study of the ADNI cohort." NeuroImage 53(3): 1051-1063. Shia, W.-C., T.-H. Ku, et al. (2011). "Genetic copy number variants in myocardial infarction patients with hyperlipidemia." BMC genomics 12 Suppl 3. Shimony, J. S., Y. I. Sheline, et al. (2009). "Diffuse microstructural abnormalities of normal- appearing white matter in late life depression: a diffusion tensor imaging study." Biol Psychiatry 66(3): 245-252. Shulman, J., L. Chibnik, et al. (2010). "Intermediate phenotypes identify divergent pathways to Alzheimer's disease." PLoS ONE 5(6). Smith, C., H. Chebrolu, et al. (2010). "White matter diffusion alterations in normal women at risk of Alzheimer's disease." Neurobiology of Aging 31(7): 1122-1131. Solano, M. P. and R. B. Goldberg (2006). "Management of dyslipidemia in diabetes." Cardiol Rev 14(3): 125-135. Sowers, J. R., M. Epstein, et al. (2001). "Diabetes, hypertension, and cardiovascular disease: an update." Hypertension 37(4): 1053-1059. Speliotes, E., C. Willer, et al. (2010). "Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index." Nature genetics 42(11): 937-948. Sprooten, E., J. Sussmann, et al. (2011). "Association of white matter integrity with genetic variation in an exonic DISC1 SNP." Molecular psychiatry 16(7): 9. Stein, J., S. Medland, et al. (2012). "Identification of common variants associated with human hippocampal and intracranial volumes." Nature genetics 44(5): 552-561. Stephan, A., D. Madison, et al. (2013). "A Dramatic Increase of C1q Protein in the CNS during Normal Aging." The Journal of neuroscience : the official journal of the Society for Neuroscience 33(33): 13460-13474. Stillman, A. N., K. C. Rowe, et al. (2012). "Anxious symptoms and cognitive function in non- demented older adults: an inverse relationship." Int J Geriatr Psychiatry 27(8): 792-798. Stitziel, N. O., H. H. Won, et al. (2014). "Inactivating mutations in NPC1L1 and protection from coronary heart disease." N Engl J Med 371(22): 2072-2082. Strauss, E., E. Sherman, et al. (2006). A compendium of neuropsychological tests: administration, norms and commentary New York, Oxford Univeristy Press.

418

Stuart, M. J. and B. T. Baune (2012). "Depression and type 2 diabetes: inflammatory mechanisms of a psychoneuroendocrine co-morbidity." Neurosci Biobehav Rev 36(1): 658-676. Sullivan, M., W. Katon, et al. (2013). "Association of depression with accelerated cognitive decline among patients with type 2 diabetes in the ACCORD-MIND trial." JAMA Psychiatry 70(10): 1041-1047. Takemura, Y., N. Ouchi, et al. (2007). "Adiponectin modulates inflammatory reactions via calreticulin receptor-dependent clearance of early apoptotic bodies." The Journal of clinical investigation 117(2): 375-386. Takeuchi, T., Y. Adachi, et al. (2007). "Adiponectin receptors, with special focus on the role of the third receptor, T-cadherin, in vascular disease." Medical molecular morphology 40(3): 115-120. Tan, H., J. A. Maldjian, et al. (2009). "A fast, effective filtering method for improving clinical pulsed arterial spin labeling MRI." J Magn Reson Imaging 29(5): 1134-1139. Tanasescu, M., M. F. Leitzmann, et al. (2003). "Physical activity in relation to cardiovascular disease and total mortality among men with type 2 diabetes." Circulation 107(19): 2435- 2439. Taylor, W. D., J. R. MacFall, et al. (2005). "Greater MRI lesion volumes in elderly depressed subjects than in control subjects." Psychiatry Res 139(1): 1-7. Teng, E. L. and H. C. Chui (1987). "The Modified Mini-Mental State (3MS) examination." J Clin Psychiatry 48(8): 314-318. Teslovich, T., K. Musunuru, et al. (2010). "Biological, clinical and population relevance of 95 loci for blood lipids." Nature 466(7307): 707-713. Thanassoulis, G., G. M. Peloso, et al. (2012). "A genetic risk score is associated with incident cardiovascular disease and coronary artery calcium: the Framingham Heart Study." Circ Cardiovasc Genet 5(1): 113-121. Thomason, M. and P. Thompson (2011). "Diffusion imaging, white matter, and psychopathology." Annual review of clinical psychology 7: 63-85. Thompson, B. and D. Towler (2012). "Arterial calcification and bone physiology: role of the bone-vascular axis." Nature reviews. Endocrinology 8(9): 529-543. Thompson, P. M., J. L. Stein, et al. (2014). "The ENIGMA Consortium: large-scale collaborative analyses of neuroimaging and genetic data." Brain Imaging Behav 8(2): 153-182. Turner, S., C. Jack, et al. (2004). "Heritability of leukoaraiosis in hypertensive sibships." Hypertension 43(2): 483-487. Ueberham, U., A. Hessel, et al. (2003). "Cyclin C expression is involved in the pathogenesis of Alzheimer's disease." Neurobiology of aging 24(3): 427-435. van Dieren, S., J. W. Beulens, et al. (2012). "Prediction models for the risk of cardiovascular disease in patients with type 2 diabetes: a systematic review." Heart 98(5): 360-369. van Elderen, S., A. de Roos, et al. (2010). "Progression of brain atrophy and cognitive decline in diabetes mellitus: a 3-year follow-up." Neurology 75(11): 997-1002. van Harten, B., F.-E. de Leeuw, et al. (2006). "Brain imaging in patients with diabetes: a systematic review." Diabetes care 29(11): 2539-2548. van Setten, J., I. Isgum, et al. (2013). "Genome-wide association study of coronary and aortic calcification implicates risk loci for coronary artery disease and myocardial infarction." Atherosclerosis 228(2): 400-405. Verstynen, T., Weinstein, A., Erickson, K., Sheu, L., Marsland, A., Gianaro, P. (2013). "Competing physiological pathways link individual differences in weight and abdominal adiposity to white matter microstructure." NeuroImage 79: 129-137. Vestergaard, P. (2007). "Discrepancies in bone mineral density and fracture risk in patients with type 1 and type 2 diabetes--a meta-analysis." Osteoporos Int 18(4): 427-444. Vidal, J. S., S. Sigurdsson, et al. (2010). "Coronary artery calcium, brain function and structure: the AGES-Reykjavik Study." Stroke 41(5): 891-897.

419

Vimalananda, V. G., J. R. Palmer, et al. (2014). "Depressive Symptoms, Antidepressant Use, and the Incidence of Diabetes in the Black Women's Health Study." Diabetes care 37(8): 2211-2217. Viswanathan, A., W. Rocca, et al. (2009). "Vascular risk factors and dementia: how to move forward?" Neurology 72(4): 368-374. Voight, B., G. Peloso, et al. (2012). "Plasma HDL cholesterol and risk of myocardial infarction: a mendelian randomisation study." Lancet 380(9841): 572-580. Wada, T., M. Haneda, et al. (2014). "Clinical impact of albuminuria and glomerular filtration rate on renal and cardiovascular events, and all-cause mortality in Japanese patients with type 2 diabetes." Clin Exp Nephrol 18(4): 613-620. Wagenknecht, L., D. Bowden, et al. (2001). "Familial aggregation of coronary artery calcium in families with type 2 diabetes." Diabetes 50(4): 861-866. Wagenknecht, L., C. Langefeld, et al. (2007). "A comparison of risk factors for calcified atherosclerotic plaque in the coronary, carotid, and abdominal aortic arteries: the diabetes heart study." American Journal of Epidemiology 166(3): 340-347. Wain, L., G. Verwoert, et al. (2011). "Genome-wide association study identifies six new loci influencing pulse pressure and mean arterial pressure." Nature genetics 43(10): 1005- 1011. Waki, H., T. Yamauchi, et al. (2003). "Impaired multimerization of human adiponectin mutants associated with diabetes. Molecular structure and multimer formation of adiponectin." The Journal of biological chemistry 278(41): 40352-40363. Wang, L., J. Manson, et al. (2010). "Systematic review: Vitamin D and calcium supplementation in prevention of cardiovascular events." Annals of internal medicine 152(5): 315-323. Wang, T., M. Bolland, et al. (2010). "Relationships between vascular calcification, calcium metabolism, bone density, and fractures." Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research 25(12): 2777- 2785. Wang, W., W. Peng, et al. (2010). "Chromosome 9p21.3 polymorphism in a Chinese Han population is associated with angiographic coronary plaque progression in non-diabetic but not in type 2 diabetic patients." Cardiovasc Diabetol 9: 33. Wang, Y., P. T. Katzmarzyk, et al. (2014). "Kidney function and the risk of cardiovascular disease in patients with type 2 diabetes." Kidney Int 85(5): 1192-1199. Warsch, J. and C. Wright (2010). "The aging mind: vascular health in normal cognitive aging." Journal of the American Geriatrics Society 58 Suppl 2: 24. Watts, J. M., C. T. Whitlow, et al. (2013). "Clinical applications of arterial spin labeling." NMR Biomed 26(8): 892-900. Wechsler, D. (1981). Manual for the Wechlser Adult Intelligence Scale-Revised. New York, Pssychological Corporation. Wei, Z., J. Peterson, et al. (2012). "C1q/TNF-related protein-12 (CTRP12), a novel adipokine that improves insulin sensitivity and glycemic control in mouse models of obesity and diabetes." The Journal of biological chemistry 287(13): 10301-10315. Wells, B. J., A. Jain, et al. (2008). "Predicting 6-Year Mortality Risk in Patients With Type 2 Diabetes." Diabetes care 31(12): 2301-2306. Wensley, F., P. Gao, et al. (2011). "Association between C reactive protein and coronary heart disease: mendelian randomisation analysis based on individual participant data." BMJ 342: d548. Wessel, J., A. Y. Chu, et al. (2015). "Low-frequency and rare exome chip variants associate with fasting glucose and type 2 diabetes susceptibility." Nat Commun 6: 5897. Wetherell, J. L., C. A. Reynolds, et al. (2002). "Anxiety, cognitive performance, and cognitive decline in normal aging." J Gerontol B Psychol Sci Soc Sci 57(3): P246-255.

420

Wexelman, B. A., E. Eden, et al. (2013). "Survey of New York City resident physicians on cause- of-death reporting, 2010." Prev Chronic Dis 10: E76. Whitehead, J., A. Richards, et al. (2006). "Adiponectin--a key adipokine in the metabolic syndrome." Diabetes, obesity & metabolism 8(3): 264-280. Willer, C., Y. Li, et al. (2010). "METAL: fast and efficient meta-analysis of genomewide association scans." Bioinformatics (Oxford, England) 26(17): 2190-2191. Willer, C. J., E. M. Schmidt, et al. (2013). "Discovery and refinement of loci associated with lipid levels." Nat Genet 45(11): 1274-1283. Willett, W., G. Howe, et al. (1997). "Adjustment for total energy intake in epidemiologic studies." The American journal of clinical nutrition 65(suppl): 1220S-1228S. Williamson, J. D., L. J. Launer, et al. (2014). "Cognitive function and brain structure in persons with type 2 diabetes mellitus after intensive lowering of blood pressure and lipid levels: a randomized clinical trial." JAMA Intern Med 174(3): 324-333. Womack, J. E., H. J. Jang, et al. (2012). "Genomics of complex traits." Ann N Y Acad Sci 1271: 33-36. Writing Group, M., D. Lloyd-Jones, et al. (2010). "Heart disease and stroke statistics--2010 update: a report from the American Heart Association." Circulation 121(7): e46-e215. Wu, M., S. Lee, et al. (2011). "Rare-variant association testing for sequencing data with the sequence kernel association test." American journal of human genetics 89(1): 82-93. Xiao, Q., R. Murphy, et al. (2013). "Dietary and supplemental calcium intake and cardiovascular disease mortality: the National Institutes of Health-AARP diet and health study." JAMA internal medicine 173(8): 639-646. Xu, W., C. Qiu, et al. (2004). "Diabetes mellitus and risk of dementia in the Kungsholmen project: a 6-year follow-up study." Neurology 63(7): 1181-1186. Yaffe, K., T. Blackwell, et al. (2006). "Glycosylated hemoglobin level and development of mild cognitive impairment or dementia in older women." The journal of nutrition, health & aging 10(4): 292-295. Yang, X., W. Y. So, et al. (2008). "Development and validation of an all-cause mortality risk score in type 2 diabetes." Arch Intern Med 168(5): 451-457. Yang, Y., J. A. Frank, et al. (1998). "Multislice imaging of quantitative cerebral perfusion with pulsed arterial spin labeling." Magn Reson Med 39(5): 825-832. Yeboah, J., R. Erbel, et al. (2014). "Development of a new diabetes risk prediction tool for incident coronary heart disease events: the Multi-Ethnic Study of Atherosclerosis and the Heinz Nixdorf Recall Study." Atherosclerosis 236(2): 411-417. Yeboah, J., R. McClelland, et al. (2012). "Comparison of novel risk markers for improvement in cardiovascular risk assessment in intermediate-risk individuals." JAMA : the journal of the American Medical Association 308(8): 788-795. Yi, W., Y. Sun, et al. (2012). "C1q/Tumor Necrosis Factor-Related Protein-3, a Newly Identified Adipokine, Is a Novel Antiapoptotic, Proangiogenic, and Cardioprotective Molecule in the Ischemic Mouse Heart." Circulation 125. Yin, R.-X., D.-F. Wu, et al. (2012). "Several genetic polymorphisms interact with overweight/obesity to influence serum lipid levels." Cardiovascular diabetology 11: 123.

421

Laura Marie Tonks Raffield

Center for Genomics and Personalized Medicine Research 1 Medical Center Boulevard Winston-Salem, NC 27157 (336) 713-7513 [email protected]

EDUCATION

Wake Forest University, Winston-Salem, NC, August 2011-Present Defense Date: May 12, 2015 Doctor of Philosophy, Molecular Genetics and Genomics Dissertation: “Analysis of the genetic and epidemiological contributors to aging-related traits in the Diabetes Heart Study” Faculty Advisor: Dr. Donald W. Bowden GPA: 4.0 / 4.0

University of North Carolina, Chapel Hill, NC, August 2007-May 2011 Bachelor of Science in Biology with Highest Distinction, Highest Honors in Biology, Minor in Chemistry Thesis: “Very Short Patch Repair in E. coli: Roles for MutS and MutL?” Faculty Advisor: Dr. Steven W. Matson GPA: 3.9 / 4.0

FELLOWSHIPS

Recipient NRSA grant 1F31AG044879-01 from National Institute on Aging (May 2013-2015)  Funds tuition and stipend expenses

RESEARCH EXPERIENCE

Graduate Student, Center for Genomics and Personalized Medicine Research, Wake Forest University, 2011 – Present Faculty Advisor: Dr. Donald W. Bowden  Performed dissertation research analysing coding and noncoding variants and epidemiological factors for association with a number of aging related traits, including vascular calcification and cognitive testing and neuroimaging measures, in the Diabetes Heart Study (DHS) and the DHS-Mind  Initiated collaborations with other Wake Forest University investigators, including analyses in DHS and DHS-Mind of potential cardiovascular impacts of calcium intake from diet and supplements and the influence of anxiety and depression symptoms on neuroimaging and cognitive testing measures  Genotyped SNPs, for example APOE variants and variants in genes from the C1q/TNF superfamily, in the DHS, DHS-Mind Study, and African American DHS  Coordinated analyses with collaborators at Wake Forest and other institutions

422

 Assisted with linkage and association analyses in the Insulin Resistance and Atherosclerosis Family Study  First year rotation in the lab of Dr. Tom Register, a DHS collaborator, on an adipose tissue immunohistochemistry project

Undergraduate Research Student, Biology Department, University of North Carolina, 2009-2011 Faculty Advisor: Dr. Steven Matson  Purified Vsr endonuclease and performed endonuclease assays for a thesis project elucidating the role of MutL and MutS in the VSP repair pathway in E. coli

Summer Undergraduate Research Program Intern/Lab Assistant, Rheumatology Department, Medical University of South Carolina, 2008-2010 Faculty Advisor: Dr. Tammy Nowling  Analyzed anti-telomere antibodies and vitamin D deficiency in systemic lupus erythematosus patients  Assisted with general lab work: cloning, cell culture, HeLa transfections, preparation of mouse cells for flow cytometry, other assays with primary T cells

REFEREED PUBLICATIONS

Raffield, L.M., Cox, A. J., Carr, J. J., Freedman, B. I., Hicks, P.J., Hsu, F., Langefeld, C. D., Bowden, D. W. Analysis of a Cardiovascular Disease Genetic Risk Score in the Diabetes Heart Study. Acta Diabetologica, 2015.

Murea, M., Hsu, F.C., Adams, J.A., Cox, A.J., Hugenschmidt, C.E., Xu, J., Raffield, L.M., Whitlow, C.T., Maldjian, J.A., Bowden, D.W., Freedman, B.I. Structural and functional assessment of the brain in mild-moderate kidney disease: Diabetes Heart Study-MIND. Nephrology Dialysis Transplantation, 2015. 0: 1–8.

Raffield, L. M., Cox, A.J., Hugenschmidt, C.E., Freedman, B.I., Langefeld, C.D., Williamson, J.D., Hsu, F., Maldjian, J.A., Bowden, D.W. Heritability and genetic association analysis of neuroimaging measures in the Diabetes Heart Study. Neurobiology of Aging, 2014. S0197- 4580(14):00724-6.

Raffield, L. M., Agarwal, S., Cox A.J., Hsu, F., Carr, J.J., Freedman, B.I., Xu, J., Bowden, D.W., Vitolins, M.Z. Cross-sectional Analysis of Calcium Intake for Effects on Vascular Calcification and Mortality in Individuals with Type 2 Diabetes from the Diabetes Heart Study. American Journal of Clinical Nutrition, 2014. 100(4):1029-35.

Cox, A.J., Hugenschmidt, C.E., Raffield, L.M., Langefeld, C.D., Freedman, B. I., Williamson, J.D., Hsu, F., Bowden, D. W. Heritability and genetic association analysis of cognition in the Diabetes Heart Study. Neurobiology of Aging, 2014. 35(8):1958.e3-1958.e12.

Hellwege, J.N., Palmer, N.D., Raffield, L.M., Ng, M. C., Hawkins, G.A., Long, J., Lorenzo, C., Norris, J.M., Ida Chen Y.D., Speliotes E.K., Rotter, J.I., Langefeld, C. D., Wagenknecht, L.E., Bowden, D.W. Genome-Wide Family-Based Linkage Analysis of Exome Chip Variants and Cardiometabolic Risk. Genetic Epidemiology, 2014. 38(4):345-52.

423

Adams, J.N., Raffield, L.M., Freedman, B. I., Langefeld, C. D., Ng, M. C. Y., Carr, J.J., Cox, A.J., Bowden, D. W. Analysis of Common and Coding Variants with Cardiovascular Disease in the Diabetes Heart Study. Cardiovascular Diabetology, 2014. 13: p. 77.

Freedman, B.I., Langefeld, C.D., Lu, L., Palmer, N.D., Smith, S.C., Bagwell, B.M., Hicks, P.J., Xu, J., Wagenknecht, L.E., Raffield, L.M., Register, T.C., Carr, J.J., Bowden, D.W., Divers, J. APOL1 associations with nephropathy, atherosclerosis, and all-cause mortality in African Americans with type 2 diabetes. Kidney International, 2014.

Raffield, L.M., Cox, A.J., Hsu, F., Ng, M. C. Y., Langefeld, C. D., Carr, J.J., Freedman, B. I., Bowden, D. W. Impact of HDL genetic risk scores on coronary artery calcified plaque and mortality in individuals with type 2 diabetes from the Diabetes Heart Study. Cardiovascular diabetology, 2013. 12: p. 95.

Hoffecker, B.M.*, Raffield, L.M.*, Kamen, D.L., Nowling, T.K. Systemic lupus erythematosus and vitamin D deficiency are associated with shorter telomere length among African Americans: a case-control study. PLOS ONE, 2013. 8(5). *contributed equally to this work

MANUSCRIPTS IN PREPARATION

Raffield, L.M., Hsu, F.C., Cox, A.J., Carr, J.J., Freedman, B.I., Bowden, D. W. Predictors of all- cause and cardiovascular disease mortality in type 2 diabetes: Diabetes Heart Study. Submitted to Diabetology & Metabolic Syndrome.

Hsu, F.C., Raffield, L.M., Hugenschmidt, C.E., Cox, A.J., Xu, J., Carr, J.J., Freedman, B.I., Maldjian, J.A., Williamson, J.D., Bowden, D.W. Relationships between cognitive performance, neuroimaging, and vascular disease: the DHS-Mind Study. Submitted to Neuroepidemiology.

Haws, B., Wuertzer, S., Fitch, M.T., Raffield, L.M., Lenchik, L., Miller, A.N. Criteria for Level 1 and Level 2 Trauma Codes: Are Pelvic Ring Injuries Undertriaged? Submitted to The Journal of Emergency Medicine.

PRESENTATIONS

Raffield, L.M.*, Masterson Creber, R*. Invited ESPO Professional Development Webinar Series Presentation: Show Me the Money! Grant Writing for Emerging Scholars and Professionals. Presented February 20, 2015. *contributed equally to this work

Bleyer, A.J., Kmoch, S., Langefeld, C.D., Raffield, L.M., Lu, L., Bowden, D.W., Freedman, B.I. Uromodulin (UMOD) Gene Polymorphism Shows Different Associations with Chronic Kidney Disease in Euro-American versus African-American Individuals with Diabetes Mellitus; (Abstract FR-PO205). Presented at the American Society of Nephrology Kidney Week, November 14, 2014 in Philadelphia, PA.

Raffield, L.M., Brenes, G.A., Freedman, B.I., Hugenschmidt, C.E., Hsu, F., Maldjian, J.A., Williamson, J.D., Bowden, D.W. Association of Self-reported Anxiety and Depression with Cognitive Testing and Neuroimaging Measures in Individuals with Type 2 Diabetes from the

424

Diabetes Heart Study; (Abstract 121). Presented at the Gerontological Society of America Annual Scientific Meeting, November 8, 2014 in Washington, DC.

Raffield, L. M., Hellwege, J. N., Cox, A. J., Langefeld, C. D., Carr, J. J., Freedman, B. I., Palmer, N. D., Bowden, D. W. Family-Based Linkage Analysis of Coding Variants with Cardiometabolic Traits in the Diabetes Heart Study; (Abstract 1086T). Presented at the 64th Annual Meeting of The American Society of Human Genetics, October 21, 2014 in San Diego, CA.

Hellwege, J. N., Raffield, L. M., Palmer, N. D., Cox, A. J., Norris, J. M., Lorenzo, C., Chen, Y.- D. I., Rotter, J. I., Langefeld, C. D., Freedman, B. I., Bowden, D. W. Simple Linkage-Based Methods to Identify Cardiometabolic Risk in Families; (Abstract 1093S). Presented at the 64th Annual Meeting of The American Society of Human Genetics, October 19, 2014 in San Diego, CA.

Adams, J.N., Raffield, L.M., Cox, A.J., Barton, E.T., Langefeld, CD., Freedman, BI., Ng, MCY., Palmer, ND., Bowden, DW. Analysis of Haptoglobin Duplication with Type 2 Diabetes and Diabetic End Stage Kidney Disease; (Abstract 838S). Presented at the 64th Annual Meeting of The American Society of Human Genetics, October 19, 2014 in San Diego, CA.

Raffield, L.M., Cox, A. J., Carr, J. J., Freedman, B. I., Hsu, F., Langefeld, C. D., Bowden, D. W. Analysis of a Cardiovascular Disease Genetic Risk Score in the Diabetes Heart Study [abstract]. Diabetes. 2014; 63(suppl 1A). Presented at the 74th Scientific Sessions of The American Diabetes Association, June 15, 2014 in San Francisco, CA.

Raffield, L. M., Cox, A. J., Langefeld, C. D., Ng, M. C. Y., Carr, J. J., Freedman, B. I., Bowden, D. W. Analysis of Coding Variants in C1q/TNF Superfamily Genes in the Diabetes Heart Study; (Abstract # 2122T). Presented at the 63rd Annual Meeting of The American Society of Human Genetics, October 24, 2013 in Boston, MA.

Adams, J. N., Raffield, L. M., Freedman, B. I., Langefeld, C. D., Ng, M. C. Y., Carr, J. J., Cox, A. J., Bowden, D. W. Analysis of Common and Coding Variants with Cardiovascular Disease in the Diabetes Heart Study; (Abstract # 2118W). Presented at the 63rd Annual Meeting of The American Society of Human Genetics, October 23, 2013 in Boston, MA.

Hellwege, J. N., Palmer, N. D., Raffield, L. M., Ng, M. C. Y., Hawkins, G. A., Long, J., Lorenzo, C., Norris, J. M., Rotter, J. I., Langefeld, C. D., Wagenknecht, L. E., Bowden, D. W. Genome-Wide Family-Based Linkage Analysis of Coding Variants and Cardiometabolic Risk; (Abstract # 978W) Presented at the 63rd Annual Meeting of The American Society of Human Genetics, October 23, 2013 in Boston, MA.

Tonks, L.M., Matson S.W. “Very Short Patch Repair in E. coli: Roles for MutS and MutL?,” Spring 2011, University of North Carolina at Chapel Hill, Biology Department Undergraduate Research Honors Symposium

Tonks, L.M. “Anti-telomere antibody levels and vitamin D deficiency in systemic lupus erythematosus patients,” Fall 2010, Medical University of South Carolina, Student Research Day

SELECTED HONORS AND AWARDS

Travel Fellowship, UAB Short Course on Statistical Genetics and Genomics (Summer 2014)

425

 Received NIGMS funded travel fellowship award to cover registration and travel expenses for the course

University of North Carolina Public Service Scholar (Spring 2011)  Participated in over 250 hours of work with Habitat for Humanity, Relay for Life, Kidzu Children's Museum, World Micro-Market, and additional organizations

Phi Beta Kappa (Spring 2010)

University of North Carolina Honors Program (2007- 2011)

Chancellor's Carolina Scholarship (2007-2011)  Full academic scholarship to University of North Carolina

National Merit Scholarship (2007-2011) TEACHING EXPERIENCE

Summer Student Mentor, Center for Genomics and Personalized Medicine Research, Wake Forest University, 2013

 Worked with a local high school student to complete genotyping and data analysis in the Diabetes Heart Study

Teaching Assistant, Scholars Program, University of North Carolina, 2010-2011  Assisted instructor in planning a course: Reimagining the American Landscape  Led course discussion section, assisted with grading

SERVICE

Volunteer, American Society of Human Genetics DNA Day Essay Contest Judge, 2014, 2015

Departmental Representative, Wake Forest Graduate Student Association, 2014

Volunteer, Brain Awareness Club, 2014

Coach and Volunteer, Special Olympics of North Carolina, 2007-2011, 2013

Co-Chair, Honors College Student Executive Board, University of North Carolina, 2010-2011  Coordinated the Academic, Service, Alumni, and Admissions committees  Helped plan evaluation sessions for the Honors Program Strategic Plan  Organized a new campus activities fair for Honors first year students

Service Committee Co-Chair, Honors College Student Executive Board, University of North Carolina, 2008-2009

 Recruited and coordinated members of the service committee  Organized information session for Honors first year students about community service

426

opportunities  Organized groups for events including Habitat for Humanity build days, volunteering with the NC Botanical Garden and Carolina Tiger Rescue, food packaging with Million Meals

COMPUTER AND STATISTICAL ANALYSIS SKILLS

Regularly use SAS, SOLAR, PLINK, and R for data management and analysis

Participant in course on Statistical Genetics and Genomics at the University of Alabama at Birmingham

Completed statistical coursework from Wake Forest Division of Public Health Sciences

 Applied Linear Models (Spring 2013)  Introduction to Statistics (Fall 2012)

PROFESSIONAL MEMBERSHIPS

American Society of Human Genetics, 2013- 2015

Gerontological Society of America, 2014- 2015

427