Citations

AbstractStatistical methods to integrate multiple layers of data, from exposures to intermediate traits to outcome variables, are needed to guide interpretation of complex data sets for which variables are likely contributing in a causal pathway from exposure to outcome. Statistical mediation analysis based on structural equation models provide a general modeling framework, yet they can be difficult to apply to high‐dimensional data and they are not automated to select the best fitting model. To overcome these limitations, we developed novel algorithms and software to simultaneously evaluate multiple exposure variables, multiple intermediate traits, and multiple outcome variables. Our penalized mediation models are computationally efficient and simulations demonstrate that they produce reliable results for large data sets. Application of our methods to a study of vascular disease demonstrates their utility to identify novel direct effects of single‐nucleotide polymorphisms (SNPs) on coronary heart disease and peripheral artery disease, while disentangling the effects of SNPs on the intermediate risk factors including lipids, cigarette smoking, systolic blood pressure, and type 2 diabetes.

Abstract The public health impact of genomic screening can be enhanced by cascade testing. However, cascade testing depends on communication of results to family members. While the barriers and facilitators of family communication have been researched following clinical genetic testing, the factors impacting the dissemination of genomic screening results are unknown. Using the pragmatic Electronic Medical Records and Genomics Network‐3 (eMERGE‐3) study, we explored the reported sharing practices of participants who underwent genomic screening across the United States. Six eMERGE‐3 sites returned genomic screening results for mostly dominant medically actionable disorders and surveyed adult participants regarding communication of results with first‐degree relatives. Across the sites, 279 participants completed a 1‐month and/or 6‐month post‐results survey. By 6 months, only 34% of the 156 respondents shared their results with all first‐degree relatives and 4% did not share with any. Over a third (39%) first‐degree relatives were not notified of the results. Half (53%) of participants who received their results from a genetics provider shared them with all first‐degree relatives compared with 11% of participants who received their results from a non‐genetics provider. The most frequent reasons for sharing were a feeling of obligation (72%) and that the information could help family members make medical decisions (72%). The most common reasons indicated for not sharing were that the family members were too young (38%), or they were not in contact (25%) or not close to them (25%). These data indicate that the professional returning the results may impact sharing patterns, suggesting that there is a need to continue to educate healthcare providers regarding approaches to facilitate sharing of genetic results within families. Finally, these data suggest that interventions to increase sharing may be universally effective regardless of the origin of the genetic result.

Background: Peripheral artery disease (PAD) affects >200 million people worldwide and is associated with high mortality and morbidity. We sought to identify genomic variants associated with PAD overall and in the contexts of diabetes and smoking status. Methods: We identified genetic variants associated with PAD and then meta-analyzed with published summary statistics from the Million Veterans Program and UK Biobank to replicate their findings. Next, we ran stratified genome-wide association analysis in ever smokers, never smokers, individuals with diabetes, and individuals with no history of diabetes and corresponding interaction analyses, to identify variants that modify the risk of PAD by diabetic or smoking status. Results: We identified 5 genome-wide significant ( P association ≤5×10 −8 ) associations with PAD in 449 548 (N cases =12 086) individuals of European ancestry near LPA (lipoprotein [a]), CDKN2BAS1 (CDKN2B antisense RNA 1), SH2B3 (SH2B adaptor protein 3) - PTPN11 (protein tyrosine phosphatase non-receptor type 11), HDAC9 (histone deacetylase 9), and CHRNA3 (cholinergic receptor nicotinic alpha 3 subunit ) loci (which overlapped previously reported associations). Meta-analysis with variants previously associated with PAD showed that 18 of 19 published variants remained genome-wide significant. In individuals with diabetes, rs116405693 at the CCSER1 (coiled-coil serine rich protein 1 ) locus was associated with PAD (odds ratio [95% CI], 1.51 [1.32–1.74], P diabetes =2.5×10 −9 , P interactionwithdiabetes =5.3×10 −7 ). Furthermore, in smokers, rs12910984 at the CHRNA3 locus was associated with PAD (odds ratio [95% CI], 1.15 [1.11–1.19], P smokers =9.3×10 −10 , P interactionwithsmoking =3.9×10 −5 ). Conclusions: Our analyses confirm the published genetic associations with PAD and identify novel variants that may influence susceptibility to PAD in the context of diabetes or smoking status.

Abstract IntroductionCurrently, one of the commonly used methods for disseminating electronic health record (EHR)-based phenotype algorithms is providing a narrative description of the algorithm logic, often accompanied by flowcharts. A challenge with this mode of dissemination is the potential for under-specification in the algorithm definition, which leads to ambiguity and vagueness. MethodsThis study examines incidents of under-specification that occurred during the implementation of 34 narrative phenotyping algorithms in the electronic Medical Record and Genomics (eMERGE) network. We reviewed the online communication history between algorithm developers and implementers within the Phenotype Knowledge Base (PheKB) platform, where questions could be raised and answered regarding the intended implementation of a phenotype algorithm. ResultsWe developed a taxonomy of under-specification categories via an iterative review process between two groups of annotators. Under-specifications that lead to ambiguity and vagueness were consistently found across narrative phenotype algorithms developed by all involved eMERGE sites. Discussion & ConclusionOur findings highlight that under-specification is an impediment to the accuracy and efficiency of the implementation of current narrative phenotyping algorithms, and we propose approaches for mitigating these issues and improved methods for disseminating EHR phenotyping algorithms.

Background: Lp(a) (lipoprotein [a]) levels are higher in individuals of African ancestry (AA) than in individuals of European ancestry (EA). We examined associations of genetically predicted Lp(a) levels with (1) atherosclerotic cardiovascular disease subtypes: coronary heart disease, cerebrovascular disease, peripheral artery disease, and abdominal aortic aneurysm and (2) nonatherosclerotic cardiovascular disease phenotypes, stratified by ancestry. Methods: We performed (1) Mendelian randomization analyses for previously reported cardiovascular associations and (2) Mendelian randomization–phenome-wide association analyses for novel associations. Analyses were stratified by ancestry in electronic Medical Records and Genomics, United Kingdom Biobank, and Million Veteran Program cohorts separately and in a combined cohort of 804 507 EA and 103 580 AA participants. Results: In Mendelian randomization analyses using the combined cohort, a 1-SD genetic increase in Lp(a) level was associated with atherosclerotic cardiovascular disease subtypes in EA—odds ratio and 95% CI for coronary heart disease 1.28 (1.16–1.41); cerebrovascular disease 1.14 (1.07–1.21); peripheral artery disease 1.22 (1.11–1.34); abdominal aortic aneurysm 1.28 (1.17–1.40); in AA, the effect estimate was lower than in EA and nonsignificant for coronary heart disease 1.11 (0.99–1.24) and cerebrovascular disease 1.06 (0.99–1.14) but similar for peripheral artery disease 1.16 (1.01–1.33) and abdominal aortic aneurysm 1.34 (1.11–1.62). In EA, a 1-SD genetic increase in Lp(a) level was associated with aortic valve disorders 1.34 (1.10–1.62), mitral valve disorders 1.18 (1.09–1.27), congestive heart failure 1.12 (1.05–1.19), and chronic kidney disease 1.07 (1.01–1.14). In AA, no significant associations were noted for aortic valve disorders 1.08 (0.94–1.25), mitral valve disorders 1.02 (0.89–1.16), congestive heart failure 1.02 (0.95–1.10), or chronic kidney disease 1.05 (0.99–1.12). Mendelian randomization–phenome-wide association analyses identified novel associations in EA with arterial thromboembolic disease, nonaortic aneurysmal disease, atrial fibrillation, cardiac conduction disorders, and hypertension. Conclusions: Many cardiovascular associations of genetically increased Lp(a) that were significant in EA were not significant in AA. Lp(a) was associated with atherosclerotic cardiovascular disease in four major arterial beds in EA but only with peripheral artery disease and abdominal aortic aneurysm in AA. Additionally, novel cardiovascular associations were detected in EA.

Background: Perceived personal control (PPC) represents the belief that a person can alter his/her own situation or state. Higher PPC has been associated with increased health-related quality of life and helps facilitate healthy behavioral changes. We investigated whether disclosure of coronary heart disease (CHD) genetic risk influences PPC and genetic counseling satisfaction (GCS) in the myocardial infarction genes (MI-GENES) study, a randomized controlled trial of disclosing genetic risk of CHD. Methods: Participants (40-65 year-old, at intermediate 10-year CHD risk, and not on statins) were randomized to receive estimated 10-year risk of CHD based on their Framingham Risk Score (FRS) or FRS plus a 28 SNP genetic risk score (FRS*GRS). CHD risk was disclosed in each arm by a genetic counselor during a 30-minute scripted session that included a discussion of the impact of family history on CHD risk. For FRS*GRS participants, the genetic counselor also reviewed how their genetic risk score altered their FRS. Each participant then met with a physician to engage in shared-decision making regarding the need for statin therapy. Afterwards, study participants were asked to complete validated surveys to assess PPC and GCS. The 9-item PPC questionnaire was scored 0-9, with higher scores indicating greater control beliefs. The 5-item GCS questionnaire was scored 0-10, with greater scores indicating greater satisfaction with the disclosure session. Results: 207 patients (mean age 58.8±5 years, 46.7% males) were randomized to receive either FRS or FRS*GRS. Patients randomized to receive FRS*GRS had higher PPC compared to FRS although the absolute difference was small (8.85±0.77 vs. 8.54±1.31, P=0.016). Patients randomized to receive FRS*GRS also had a higher GCS score than FRS patients (9.08±2.67 vs. 8.3±3.67, P=0.050]. Conclusions: Disclosure of genetic risk for CHD did not adversely affect PPC or GCS. In fact, patients who received CHD genetic risk information had higher PPC and GCS. Our findings suggest that disclosure of CHD genetic risk is appreciated by patients and may empower them to improve health-related behaviors.

Background: Arterial stiffness is associated with increased risk for adverse cardiovascular outcome. Ankle-brachial index (ABI) is a measure reflective of systemic atherosclerosis. Whether it can predict central aortic stiffness and its progression is unknown. As such, we investigated the impact of ABI on carotid-femoral pulse wave velocity (cfPWV) and aortic characteristic impedance (Z c ) in apparently healthy adults from the community. Methods: Participants were from the GENOA study without cardiovascular disease at baseline (December 2000 to September 2006) and completed follow-up (September 2009 to December 2012) assessment (n=358, 63%women, 69% hypertensive). ABI was measured by standard protocol at baseline. cfPWV was measured by applanation tonometry. Difference in log-cfPWV between two visits ([[Unable to Display Character: &#8710;]]log-cfPWV) was used to estimate progression of aortic stiffness. Z c was ascertained at 2 nd visit using carotid tonometry followed by echocardiography (Cardiovascular Engineering Inc., Norwood). Results: After a mean follow-up of 8.5±0.9 years, cfPWV increased from 9.3±2.4m/s to 11.1±5.0m/s. After adjusted for age, sex and time-interval between two visits, lower ABI was associated with greater [[Unable to Display Character: &#8710;]]log-cfPWV (β±SE:-0.41±0.15), and increased risk for higher log-cfPWV (β±SE:-0.38±0.12) and higher log-Z c at 2 nd visit (β±SE:-0.38±0.18) (all P≤0.04). Diabetes and renal function at baseline were predictors for higher cfPWV and Z c ; hypertension and metabolic syndrome were only predictors of cfPWV; while age and male gender were major predictors for [[Unable to Display Character: &#8710;]]log-cfPWV (all P≤0.04). The association of ABI with cfPWV or Z c remained significant after adjustment for these covariates (both P<0.05). Greater [[Unable to Display Character: &#8710;]]log-cfPWV was associated with increased risk for higher log-Z c ; longer time-interval amplified this effect (both p for [[Unable to Display Character: &#8710;]]log-cfPWV and [[Unable to Display Character: &#8710;]]log-cfPWV x time-interval ≤0.03). Conclusions: Lower ABI predicts progression of aortic stiffness, leading to both increased aortic wall stiffness (PWV) and greater aortic area-flow mismatch (Z c ), suggesting the impact of subclinical atherosclerosis on proximal aortic stiffening.

Background: Familial hypercholesterolemia (FH) is an important public health burden as it is a relatively common Mendelian genetic disorder which is associated with dramatically increased lifetime risk for premature atherosclerotic cardiovascular disease (ASCVD) due to elevated plasma low-density lipoprotein cholesterol (LDL-C) levels. Little is known about prevalence and control of FH in the US. Objective: To develop an electronic phenotyping algorithm for rapid identification of FH in electronic health records (EHRs) and thereby address knowledge gaps in prevalence and control of FH. Methods: We identified patients in the Mayo Employee and Community Health (ECH) system who met the following inclusion criteria: 1) gave research authorization, 2) any LDL-C in EHR ≥190 mg/dL; 2) triglycerides <400 mg/dL; 3) absence of secondary causes of hyperlipidemia (hypothyroidism, cholestatic liver diseases, nephrotic syndrome, renal failure, and pregnancy). Next we applied the Dutch Lipid Clinic Network (DLCN) criteria to ascertain FH cases. ASCVD-related diagnosis codes, laboratory data, and medications were obtained from structured EHR datasets. We used the Mayo natural language processing (NLP) system to ascertain presence of family and personal history of premature ASCVD, xanthomas and corneal arcus from clinical text notes. Results: Of a total of 131,000 patients seen in ECH clinics between July 1993 and December 2014, 6018 met the inclusion criteria. Implementing the electronic phenotyping algorithm for FH in these patients identified 178 definite and 369 probable cases (DLCN score >8 and 6-8 points, respectively) with an overall FH prevalence of 0.4% (1:240). Blinded expert review of 160 randomly chosen patients showed positive and negative predictive values for electronic phenotyping algorithm at 85% and 90%, respectively. Only 40% of these patients achieved LDL-C ≤100 mg/dL on treatment. Conclusions: 1) An EHR-based phenotyping algorithm that included NLP had reasonable accuracy in ascertaining FH cases. 2) Implementing this algorithm revealed the prevalence of FH in the study cohort to be nearly twice the current estimate of 1:500 in the general population. 3) Less than half of the FH patients had optimal LDL-C levels on treatment.

Background: Whether disclosing genetic risk for coronary heart disease (CHD) to individuals influences information seeking and information sharing is not known. Methods: The myocardial infarction genes (MI-GENES) trial randomized participants aged 45-65 years who were at intermediate risk for CHD based on conventional risk factors and not on statins, to receive their conventional risk score (CRS) or their CRS plus a genetic risk score (GRS) based on 28 susceptibility variants. CHD risk was disclosed by a genetic counselor and then discussed with a physician. Surveys to assess information seeking (including internet use and accessing electronic health records (EHR)) were completed before and three and six months after risk disclosure. Information sharing parameters were assessed after risk disclosure. We assessed whether these behaviors differed by GRS disclosure, or by high (≥1.1) or low (<1.1) GRS. Adjustments were made for age, sex, family history of CHD, baseline CRS and GRS, and education. Results were reported as the mean difference (and standard error) in the score for each survey response between the GRS and CRS participants, with significance determined by regression analysis. Results: GRS participants accessed their EHR to obtain information related to their CHD risk more than CRS participants (0.14 ± 0.06, p=0.03). Overall internet use (0.61 ± 0.23, p=0.01), as well as internet use to seek information about heart disease (0.14 ± 0.06, p=0.02) and how genetic factors affect risk of having a heart attack (0.23 ± 0.07, p=0.002), was significantly higher in the GRS participants. GRS participants shared information about heart attack risk with others (0.35 ± 0.13, p=0.007), particularly family members (0.1 ± 0.04, p=0.02), (V4: 0.10 ± 0.05, p=0.05), and their primary care provider (V4: 0.15 ± 0.07, p=0.03) more than CRS participants. Internet use, EHR access, and information sharing did not differ significantly between the high and low GRS groups. Conclusions: Disclosure of GRS for CHD resulted in greater information seeking (including internet use and EHR access) and information sharing by study participants. Disclosure of genetic risk for CHD may help advance patient engagement in precision medicine.

Fibromuscular dysplasia (FMD) is a nonatherosclerotic vascular disease leading to arterial stenosis, aneurysm and dissection, mainly in renal and carotid arteries. FMD has higher prevalence in females (80-90%) and is associated with hypertension and stroke. The pathophysiology of FMD is unclear and a genetic origin is suspected. We performed a genetic association study in European ancestry individuals. The discovery included 249 cases and 689 controls, in which we analyzed ∼26K common variants (MAF>0.05) using an exome-chip array. We followed up 13 loci (P<10-4) in 393 cases and 2537 controls and replicated a signal on Chr6. Three additional studies (combined Ncases=512, Ncontrols=669) confirmed this association, with an overall OR of 1.39, (P=7.4х10e-10 , All Ncases=1154, All Ncontrols=3895). The FMD risk variant is intronic to the phosphatase and actin regulator 1 gene (PHACTR1), involved in angiogenesis and cell migration. PHACTR1 is a risk locus for coronary artery disease, migraine, and cervical artery dissection, which may occur in FMD. We found a significant association between the risk allele and higher central pulse pressure (P=0.0009), increased intima media thickness (P=0.001) and wall cross-sectional area (P=0.003) of carotids assessed by echotracking in 3800 population-based individuals. The correlation of genotypes with the expression of PHACTR1 in primary cultured human fibroblasts showed higher expression in FMD risk allele carriers, compared to non carriers (N=57, P=0.02). Finally, Phactr1 knockdown of zebrafish showed significantly dilated vessels (P=0.003) indicating impaired vascular development. In conclusion, we report the first risk locus for FMD with the largest genetic association study conducted so far. Our data reveal a common genetic variant at PHACTR1 providing indices of shared pathophysiology between FMD and other cardiovascular and neurovascular diseases.