The DHS evaluates the genetic and epidemiological causes of CVD in individuals with T2DM. Ascertainment, recruitment, and examination have been described in detail . Briefly, siblings with T2DM and without advanced nephropathy were recruited, with unaffected siblings also recruited when possible. T2DM was defined as diabetes developing after 35 years of age, with initial treatment using a combination of, exercise and/or oral agents, not solely insulin, and in the absence of historical evidence of ketoacidosis. Diabetes diagnosis was confirmed upon entrance to the study by measurement of fasting glucose and glycated hemoglobin (HbA1C) testing. The 1,208 European American individuals included in this analysis were from 473 families.
Study protocols were approved by the Institutional Review Board at Wake Forest School of Medicine. Participants provided written informed consent prior to participation. Examinations were conducted in the General Clinical Research Center of the Wake Forest Baptist Medical Center and included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses (blood lipid profile, fasting glucose, HbA1C, and high-sensitivity C-reactive protein (CRP)) 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).
CAC, CarCP, and AACP were measured using fast-gated helical CT scanning, and calcium scores were computed as previously described and reported as Agatston scores [10, 11]. IMT was measured by high-resolution B-mode ultrasonography with a 7.5-MHz transducer and a Biosound Esaote (AU5) ultrasound machine (Biosound Esaote, Inc., Indianapolis, IN) as previously described . Not all measurements were available in all participants.
Vital status was determined from the National Social Security Death Index maintained by the United States Social Security Administration. For those participants confirmed as deceased, length of follow-up was determined from date of the initial study visit to date of death [6, 13]. For deceased participants, copies of death certificates were obtained from relevant county Vital Records Offices to confirm cause of death. For all other participants the length of follow-up was determined from the date of the initial study visit to the December 31, 2012. Causes of death were categorized based on information contained in death certificates as CVD-related (myocardial infarction, 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).
DHS GWAS and imputed data
Genomic DNA was purified from whole-blood samples obtained from subjects using the PUREGENE DNA isolation kit (Gentra Systems., Minneapolis, MN). DNA was quantitated using standardized fluorometric readings on a Hoefer DyNA Quant 200 fluorometer (Hoefer Pharmacia Biotech, Inc., San Francisco, CA). Each sample was diluted to a final concentration of 5 ng/μL.
A GWAS was completed using the Affymetrix Genome-wide Human SNP Array 5.0 (Affymetrix, CA, USA) as reported . Genotype calling was completed using the BRLLM-P algorithm in Genotyping Console v4.0 (Affymetrix). Samples failing to meet an intensity quality control (QC) threshold and those failing to meet a minimum acceptable call rate of 95% were excluded from further analyses (n = 7). An additional 39 samples were included as blind duplicates within the genotyping set to serve as QC samples; the concordance rate for these blind duplicates was 99.0 ± 0.72% (mean ± standard deviation (SD)). Exploratory analyses of genotype data were performed using PLINK v1.07 (http://pngu.mgh.harvard.edu/purcell/plink/) and samples with poor quality genotype calls, gender errors, or unclear/unexpected sibling relationships were excluded from further analysis. Exclusion criteria for single nucleotide polymorphism (SNP) performance included call rate <95% (n = 11,085), Hardy-Weinberg Equilibrium (HWE) p-value <1×10−6 (n = 332), and minor allele frequency (MAF) <0.01 (n = 57,382); 371,951 SNPs were retained for analysis.
Additional genotype data were obtained by imputation from the GWAS. Imputation of 1,000 Genomes Project SNPs was completed using the program IMPUTE2 (http://mathgen.stats.ox.ac.uk/impute/impute_v2.html) and the Phase I v2, cosmopolitan (integrated) reference panel, build 37 . SNPs that were used for imputation were required to have low missingness and show no significant departure from HWE expectations. 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.
Individual SNP genotyping
For those genes implicated in the CHARGE meta-analyses as potential risk loci, exonic variants contained in the National Heart Lung and Blood Institute Grand Opportunity Exome Sequencing Project (NHLBI GO-ESP). Coding SNPs identified as possibly damaging or probably damaging by PolyPhen2  with a minor allele frequency (MAF) less than 0.2 were selected for genotyping. Fourty-four single nucleotide polymorphisms (SNPs) from 17 genes were genotyped in the DHS. Genotyping was performed using the Sequenom Mass ARRAY genotyping system (Sequenom, San Diego, CA) and PCR primers were designed using the Mass ARRAY Assay Design 3.4 Software (Sequenom). An additional 41 quality control (QC) samples were included in the genotyping analysis to serve as blind duplicates. The concordance rate for the blind duplicates was 100%. For all SNPs the minimum acceptable call frequency was 95%. The average call frequency was 97.3 ± 0.009% (mean ± SD). Samples with genotyping efficiency rates <90% were excluded from further analysis. Twenty-eight SNPs from 14 genes were carried forward to analysis after QC. Genotyped SNPs are listed in Additional file 1.
Additional SNPs for those genes implicated in the CHARGE meta-analyses as potential risk loci were also identified as captured by the Illumina® HumanExome BeadChips (Illumina® Inc., San Diego, CA) for which genotype data was available in the DHS. For 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 QC samples; the concordance rate for blind duplicates was 99.9 ± 0.0001% (mean ± SD). Exclusion criteria for SNP performance included call rate <99% (n = 972), monomorphic SNPs (n = 157,754) and Hardy-Weinberg Equilibrium p-value <1×10−6 (n = 26); 88,483 SNPs were retained for analysis. Additional QC of Exome Chip data set was completed to exclude samples with poor quality genotype calls, gender errors, or unclear/unexpected sibling relationships.
Genetic risk scores
Genetic risk scores (GRS) were calculated as previously described . Both unweighted GRS and GRS weighted by SNP effect size were derived for two sets of SNPs previously reported to be associated with CAC or CAC and MI. SNPs included in both GRS are shown in Additional file 2. One set of 12 SNPs had documented effects on CAC (Score 1; 1a = unweighted, 1b = weighted). We created a second GRS from 8 SNPs associated with CAC and MI (Score 2; 2a = unweighted, 2b = weighted). Unweighted scores were derived by adding the number of effect alleles for each SNP for each person. The SNPs were also weighted by their previously reported effect sizes [8, 9]. 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 for a particular SNP, the mean genotype calculated in the DHS for that given SNP was assigned . For all GRS, the effect allele was assigned as the allele associated with an increase in CAC or increased risk for CAC.
All derived GRS (1a, 1b, 2a, and 2b) were tested for association with CAC, CarCP, AACP, IMT, prior history of CVD events, prior history of MI, all-cause mortality, and CVD-cause mortality to evaluate whether the GRS were a measure of genetic contributions to either clinical or subclinical CVD.
Allele frequencies were calculated for a sub-set of unrelated individuals and departure from HWE was calculated from a group of unrelated samples using a chi-squared goodness-of-fit test implemented in PLINK v1.07. Association between the SNP genotypes and CVD measures was examined using variance component methods computed using SOLAR v4.3.1 (Texas Biomedical Research Institute, San Antonio, Tx, USA) which accounted for family structure. Each trait was examined using additive, dominant, and recessive models of inheritance. Most of the associations were observed under the additive model, however, there were associations seen under only the dominant or recessive models. Continuous variables were transformed prior to analysis to approximate conditional normality. Age, gender, T2DM-affection status, and body mass index (BMI) were used as covariates in all single variant association analyses. Additional covariates (e.g. cholesterol medication use, T2DM duration, and smoking) were also tested, but did not meaningfully impact the results. Statistical significance for all single SNP analysis was calculated using the Li & Ji method to determine the effective number of SNPs using SOLAR v. 4.3.1. Statistical significance was set at p < 2.16×10−4. Power calculations for dichotomous traits were run using CaTS (University of Michigan School of Public Health http://www.sph.umich.edu/csg/abecasis/CaTS/). Power calculations for continuous traits were run using Quanto (University of Southern California http://hydra.usc.edu/gxe/).
GRS were considered as continuous variables. Relationships between the GRS and CAC, CarCP, AACP, history of CVD, and history of MI were examined using marginal models with incorporation of generalized estimating equations. These models use a sandwich estimator of the variance under exchangeable correlation in order to account for familial correlation . Relationships between GRS and both all-cause and CVD-cause mortality were examined using Cox proportional hazards models with sandwich-based variance estimation due to the inclusion of related individuals. Associations were adjusted for age, gender, BMI, smoking status (current or prior smoking), hypertension, cholesterol medications, and prior CVD as indicated. All analyses were performed in SAS 9.3 (SAS Institute, Cary, NC) and statistical significance was accepted at p < 0.05.