ROC curves are a popular method for displaying sensitivity and specificity of a continuous diagnostic marker, X, for a binary disease variable, D. However, many disease outcomes are time dependent, D(t), and ROC curves that vary as a function of time may be more appropriate. A common example of a time-dependent variable is vital status, where D(t) = 1 if a patient has died prior to time t and zero otherwise. We propose summarizing the discrimination potential of a marker X, measured at baseline (t = 0), by calculating ROC curves for cumulative disease or death incidence by time t, which we denote as ROC(t). A typical complexity with survival data is that observations may be censored. Two ROC curve estimators are proposed that can accommodate censored data. A simple estimator is based on using the Kaplan-Meier estimator for each possible subset X > c. However, this estimator does not guarantee the necessary condition that sensitivity and specificity are monotone in X. An alternative estimator that does guarantee monotonicity is based on a nearest neighbor estimator for the bivariate distribution function of (X, T), where T represents survival time (Akritas, M. J., 1994, Annals of Statistics 22, 1299-1327). We present an example where ROC(t) is used to compare a standard and a modified flow cytometry measurement for predicting survival after detection of breast cancer and an example where the ROC(t) curve displays the impact of modifying eligibility criteria for sample size and power in HIV prevention trials.

Identifying the downstream effects of disease-associated single nucleotide polymorphisms (SNPs) is challenging: the causal gene is often unknown or it is unclear how the SNP affects the causal gene, making it difficult to design experiments that reveal functional consequences. To help overcome this problem, we performed the largest expression quantitative trait locus (eQTL) meta-analysis so far reported in non-transformed peripheral blood samples of 5,311 individuals, with replication in 2,775 individuals. We identified and replicated trans-eQTLs for 233 SNPs (reflecting 103 independent loci) that were previously associated with complex traits at genome-wide significance. Although we did not study specific patient cohorts, we identified trait-associated SNPs that affect multiple trans-genes that are known to be markedly altered in patients: for example, systemic lupus erythematosus (SLE) SNP rs49170141 altered C1QB and five type 1 interferon response genes, both hallmarks of SLE2-4. Subsequent ChIP-seq data analysis on these trans-genes implicated transcription factor IKZF1 as the causal gene at this locus, with DeepSAGE RNA-sequencing revealing that rs4917014 strongly alters 3’ UTR levels of IKZF1. Variants associated with cholesterol metabolism and type 1 diabetes showed similar phenomena, indicating that large-scale eQTL mapping provides insight into the downstream effects of many trait-associated variants.

I present software for analysing complex survey samples in R. The sampling scheme can be explicitly described or represented by replication weights. Variance estimation uses either replication or linearisation.

Blood pressure (BP) is a major cardiovascular disease risk factor. To date, few variants associated with inter-individual BP variation have been identified. A genome-wide association study of systolic (SBP), diastolic BP (DBP), and hypertension in the CHARGE Consortium (n=29,136) identified 13 SNPs for SBP, 20 for DBP, and 10 for hypertension at p <4×10 -7 . The top 10 loci for SBP and DBP were incorporated into a risk score; mean BP and prevalence of hypertension increased in relation to number of risk alleles carried. When 10 CHARGE SNPs for each trait were meta-analyzed jointly with the Global BPgen Consortium (n=34,433), four CHARGE loci attained genome-wide significance (p<5×10 -8 ) for SBP (ATP2B1, CYP17A1, PLEKHA7, SH2B3), six for DBP (ATP2B1, CACNB2, CSK/ULK3, SH2B3, TBX3/TBX5, ULK4), and one for hypertension (ATP2B1). Identifying novel BP genes advances our understanding of BP regulation and highlights potential drug targets for the prevention or treatment of hypertension.High blood pressure affects about one third of adults and contributes to 13.5 million deaths worldwide each year and about half the global risk for stroke and ischemic heart disease. 1,2 Clinical trials, dating back more than forty years, have proven that drug treatment to lower blood pressure dramatically reduces the risk of cardiovascular events in people with hypertension. 3,4 The substantial (30-60 percent) 5 heritability of blood pressure has prompted extensive efforts to identify its genetic underpinnings. The search for genes associated with interindividual variation in blood pressure in the general population has used a variety of complementary approaches, which have yielded relatively few clues. Linkage and candidate gene studies, despite considerable knowledge about pathways that are critical to blood pressure homeostasis, have provided limited consistent evidence of blood pressure quantitative trait loci. 6,7,8 The study of families with rare Mendelian high or low blood pressure syndromes has identified mutations with gain or loss of function in about a dozen renal sodium regulatory genes. 9 Common variants in two renal sodium regulatory genes have been found to be associated with blood pressure in the general population. 10 The vast majority of the genetic contribution to variation in blood pressure, however, remains unexplained.Large-scale genome-wide association studies (GWAS), in which hundreds of thousands of common genetic variants are genotyped and analyzed for disease association, have shown great success in identifying genes associated with common diseases and traits. 11,12 The fact that six GWAS published to date, however, have not identified loci associated with blood pressure or hypertension at p<5×10 -8 , has raised concerns about the utility of this approach for these traits. 13,14,15,16,17,18 If blood pressure variation in the general population is due to multiple variants with small effects, very large study samples are needed to identify them. We established the Cohorts for Heart and Aging Research i...