Lange, Ethan M.

Abstract RATIONALE: Chronic Obstructive Pulmonary Disease (COPD) is a leading cause of mortality worldwide. We previously published the polygenic transcriptome risk score (PTRS) which uses the cumulative effect of predicted gene expression to construct genetic predictors for complex diseases, and we demonstrated the value of PTRS for COPD built on Genotype-Tissue Expression (GTEx) Lung tissue with significantly improved cross-ancestry portability. However, the postmortem collection and lack of lung disease status on GTEx samples likely affect results. We hypothesized that the performance of PTRS will be improved by using disease-relevant expression quantitative trait loci (eQTLs) to construct the score. Here, we aimed to improve performance of PTRS by leveraging disease-relevant eQTLs. METHODS: We constructed two transcriptome prediction models, Elastic Net (EN) and Prediction Using Models Informed by Chromatin conformation and Epigenomics (PUMICE), using TOPMed lung RNA-seq data from the Lung Tissue Research Consortium (LTRC), which included COPD patients undergoing lung surgery. We compared their performance with two published models built on RNA-seq from GTEx Lung tissue, GTEx-EN and GTEx-PUMICE. We first integrated each of four transcriptome models with multi-ancestry GWAS of FEV1/FVC ratio (Shrine et al. 2023) to generate transcriptome-wide association study (TWAS) results, which produced trait-associated genes. Pathway analysis was then conducted on significant TWAS genes. Finally, the PTRS was computed on TWAS genes that were included in the top 20 significantly enriched pathways. We tested performance of PTRS in race-stratified analysis for moderate-to-severe and severe COPD in COPDGene. RESULTS: Our models produced more significant TWAS genes (FDR<0.05) and had more overlaps with genes identified for FEV1/FVC ratio from Shrine et al. 2023 compared to two GTEx models (Fig. A). Additionally, our models had significantly higher AUC for predicting both moderate-to-severe and severe COPD in race-stratified analysis. The mean AUC from LTRC-EN for predicting severe COPD was 0.57 and 0.55 respectively for EUR and AFA, which was significantly higher than mean AUC of 0.54 and 0.51 from GTEx-EN (Delong p-value=8.15x10-10 for EUR and 1.42x10-3 for AFA) (Fig. B). Similarly, the PTRS derived from our models produced higher Odds Ratios (OR) per standard deviation. The mean OR from LTRC-PUMICE for moderate-to-severe COPD was 1.25 and 1.18 respectively for EUR and AFA, while it was 1.17 and 1.12 from GTEx-PUMICE (Fig. C). CONCLUSIONS: Our study demonstrates the value of leveraging disease-relevant RNA-seq to construct transcriptome prediction models and their improvement on identifying lung function genes and predicting COPD across ancestries.

P10 Coronary artery calcification (CAC), quantified by electron beam computed tomography, provides a measure of extent of coronary atherosclerosis in asymptomatic and symptomatic individuals. It is well recognized that coronary artery disease and CAC are complex processes that are influenced by many genetic and environmental factors. Several aspects of CAC quantity make genetic studies of it particularly challenging. These include the non-normality of CAC frequency distributions and the strong age- and sex-dependence of CAC levels. To search for regions that may contain genes influencing CAC quantity, a genome-wide scan was conducted on a sample enriched for hypertension from the Epidemiology of Coronary Artery Calcification Study. The genome-wide linkage scan was based on 370 anonymous markers spanning the 22 autosomal chromosomes in 29 families. The analysis sample consisted of 20 affected sib-pairs, 8 affected sib-trios and 1 family with 6 affected siblings. Here, affection status was defined as being above the 70th sex- and age-specific percentile based on a reference population. Mode-of-inheritance-free qualitative linkage analysis was used to analyze sib-pair data. Maximum multipoint lod scores were calculated using the computer software Genehunter-2. Weighted analyses were conducted using Holmans “possible triangle” constraint. The overall maximum lod scores were observed on chromosomes 6 (lod=2.22) and 10 (lod=3.24). Additional lod scores greater than 1.0 were observed on chromosomes 2, 14, 15 and 17. These results identify regions that may harbor genes involved in high sex- and age-specific levels of CAC (supported by HL46292).

ABSTRACTNext‐generation sequencing technologies have afforded unprecedented characterization of low‐frequency and rare genetic variation. Due to low power for single‐variant testing, aggregative methods are commonly used to combine observed rare variation within a single gene. Causal variation may also aggregate across multiple genes within relevant biomolecular pathways. Kernel‐machine regression and adaptive testing methods for aggregative rare‐variant association testing have been demonstrated to be powerful approaches for pathway‐level analysis, although these methods tend to be computationally intensive at high‐variant dimensionality and require access to complete data. An additional analytical issue in scans of large pathway definition sets is multiple testing correction. Gene set definitions may exhibit substantial genic overlap, and the impact of the resultant correlation in test statistics on Type I error rate control for large agnostic gene set scans has not been fully explored. Herein, we first outline a statistical strategy for aggregative rare‐variant analysis using component gene‐level linear kernel score test summary statistics as well as derive simple estimators of the effective number of tests for family‐wise error rate control. We then conduct extensive simulation studies to characterize the behavior of our approach relative to direct application of kernel and adaptive methods under a variety of conditions. We also apply our method to two case‐control studies, respectively, evaluating rare variation in hereditary prostate cancer and schizophrenia. Finally, we provide open‐source R code for public use to facilitate easy application of our methods to existing rare‐variant analysis results.

ABSTRACTRare variants (RVs) have been shown to be significant contributors to complex disease risk. By definition, these variants have very low minor allele frequencies and traditional single‐marker methods for statistical analysis are underpowered for typical sequencing study sample sizes. Multimarker burden‐type approaches attempt to identify aggregation of RVs across case‐control status by analyzing relatively small partitions of the genome, such as genes. However, it is generally the case that the aggregative measure would be a mixture of causal and neutral variants, and these omnibus tests do not directly provide any indication of which RVs may be driving a given association. Recently, Bayesian variable selection approaches have been proposed to identify RV associations from a large set of RVs under consideration. Although these approaches have been shown to be powerful at detecting associations at the RV level, there are often computational limitations on the total quantity of RVs under consideration and compromises are necessary for large‐scale application. Here, we propose a computationally efficient alternative formulation of this method using a probit regression approach specifically capable of simultaneously analyzing hundreds to thousands of RVs. We evaluate our approach to detect causal variation on simulated data and examine sensitivity and specificity in instances of high RV dimensionality as well as apply it to pathway‐level RV analysis results from a prostate cancer (PC) risk case‐control sequencing study. Finally, we discuss potential extensions and future directions of this work.

Abstract— Coronary artery disease (CAD) is the leading cause of mortality in the developed world. Although several CAD risk factors, including measures of lipid metabolism, obesity, and blood pressure, have a genetic basis, many genes for CAD susceptibility have yet to be identified. Coronary atherosclerosis is the major cause of CAD, but many with coronary atherosclerosis lack symptoms. Thus, a major limitation of using symptomatic CAD endpoints (eg, sudden coronary death, myocardial infarction) as a study outcome is substantial disease misclassification. Coronary artery calcification (CAC) is part of the atherosclerotic process and is an independent predictor of CAD endpoints. In the present study, CAC was noninvasively quantified by electron beam computed tomography. We performed genome-wide multipoint mode-of-inheritance-free linkage analysis on affected sib pairs, defined as being ≥ the 70th sex- and age-specific percentile for CAC quantity, in a sample of 29 families enriched for hypertension. Almost 95% of participants were asymptomatic for CAD. Our LOD score (log 10 odds in favor of linkage) results provide evidence that chromosomal regions 6p21.3 (maximum LOD score=2.22, P =0.00070) and 10q21.3 (maximum LOD score=3.24, P =0.000057) may harbor genes associated with subclinical coronary atherosclerosis.