Genome-wide study of resistant hypertension identified from electronic health records

AbstractIn this study, we perform a full genome-wide association study (GWAS) to identify statistically significantly associated single nucleotide polymorphisms (SNPs) with three red blood cell (RBC) components and follow it with two independent PheWASs to examine associations between phenotypic data (case-control status of diagnoses or disease), significant SNPs, and RBC component levels. We first identified associations between the three RBC components: mean platelet volume (MPV), mean corpuscular volume (MCV), and platelet counts (PC), and the genotypes of approximately 500,000 SNPs on the Illumina Infimum® DNA Human OmniExpress-24 BeadChip using a single cohort of 4,700 Northern Nevadans. Twenty-one SNPs in five major genomic regions were found to be statistically significantly associated with MPV, two regions with MCV, and one region with PC, with p<5x10-8. Twenty-nine SNPs and nine chromosomal regions were identified in 30 previous GWASs, with effect sizes of similar magnitude and direction as found in our cohort. The two strongest associations were SNP rs1354034 with MPV (p=2.4x10-13) and rs855791 with MCV (p=5.2x10-12). We then examined possible associations between these significant SNPs and incidence of 1,488 phenotype groups mapped from International Classification of Disease version 9 and 10 (ICD9 and ICD10) codes collected in the extensive electronic health record (EHR) database associated with Healthy Nevada Project consented participants. Further leveraging data collected in the EHR, we performed an additional PheWAS to identify associations between continuous red blood cell (RBC) component measures and incidence of specific diagnoses. The first PheWAS illuminated whether SNPs associated with RBC components in our cohort were linked with other hematologic phenotypic diagnoses or diagnoses of other nature. Although no SNPs from our GWAS were identified as strongly associated to other phenotypic components, a number of associations were identified with p-values ranging between 1x10-3 and 1x10-4 with traits such as respiratory failure, sleep disorders, hypoglycemia, hyperglyceridemia, GERD and IBS. The second PheWAS examined possible phenotypic predictors of abnormal RBC component measures: a number of hematologic phenotypes such as thrombocytopenia, anemias, hemoglobinopathies and pancytopenia were found to be strongly associated to RBC component measures; additional phenotypes such as (morbid) obesity, malaise and fatigue, alcoholism, and cirrhosis were also identified to be possible predictors of RBC component measures.Author SummaryThe combination of electronic health records and genomic data have the capability to revolutionize personalized medicine. Each separately contains invaluable data; however, combined, the two are able to identify new discoveries that may have long-term health benefits. The Healthy Nevada Project is a non-profit initiative between Renown Medical Center and the Desert Research Institute in Reno, NV. The project has so far collected a cohort of 6,500 Northern Nevadans, with extensive medical electronic health records in the Renown Health database. Combining the genotypes of these participants with the clinical data, this study’s aim is to find associations between genotypes (genes) and phenotypes (diagnoses and lab records). Here, we identify and examine clinical associations with red blood cell components such as platelet counts and mean platelet volume. These are components that have clinical relevance for several diseases, such as anemia, atherothrombosis and cancer. Our results from genome wide association studies mirror previous studies, and identify new associations. The extensive electronic health records enabled us to perform phenome wide associations to discover strong associations with hematologic components, as well as other important traits and diagnoses.

Download Full-text

Abstract P185: Using Electronic Health Records To Identify Patients With Apparent Treatment Resistant Hypertension

Hypertension ◽

10.1161/hyp.76.suppl_1.p185 ◽

2020 ◽

Vol 76 (Suppl_1) ◽

Author(s):

Priyanka Solanki ◽

Imran Ajmal ◽

Xiruo Ding ◽

Jordana Cohen ◽

Debbie Cohen ◽

...

Keyword(s):

Random Forest ◽

Electronic Health Records ◽

Resistant Hypertension ◽

Cross Validation ◽

Treatment Resistant ◽

Health Records ◽

Forest Model ◽

Diagnosis Codes ◽

Electronic Health ◽

Secondary Causes

Introduction: Apparent treatment resistant hypertension (aTRH) affects 10-20% of hypertensive adults and increases risk of cardiovascular events and mortality. Fewer than half of these patients have true resistant hypertension. The majority experience pseudo-resistant hypertension due to inadequate medication adherence, white coat hypertension, and secondary causes of hypertension. We hypothesize that electronic health records can be leveraged to identify aTRH patients who would benefit from targeted counseling, medication reconciliation, and screening for secondary causes of hypertension. Methods: We studied electronic health record (EHR) data from 395 hypertensive adults in our primary care population who received longitudinal care between 2007 and 2017. Patients who met the 2008 AHA definition of resistant hypertension by chart review were considered to have aTRH. We also included 100 patients identified by heuristics targeting secondary hypertension. We extracted from the EHR demographics, vitals, laboratory results, diagnosis codes, and medications. Results outside of physiologic range were excluded and median imputation was used to handle missing data. Random forest model performance was assessed by 5-fold cross validation. Model discrimination was evaluated at an estimated positive predictive value of 75%. Results: The prevalence of aTRH in our randomly selected and full cohorts was 20.3% (n=295) and 25.8% (n=395), respectively. In cross-validation, the random forest model demonstrated a median sensitivity of 65% (IQR: 60% - 65%) and a median AUROC of 0.92 (IQR: 0.90 - 0.92). The most influential variables were related to the prescription of three or more hypertension medications; number of days on diuretics, angiotensin-converting enzyme inhibitors, or angiotensin II receptor blockers; systolic blood pressure measurements; and hypertension or diabetes diagnosis codes. Conclusion: EHR data can be used to accurately identify patients with aTRH. We expect the implementation of a clinical decision support system leveraging such models could lead to the improved care for aTRH patients.

Download Full-text

Genome-wide survey of parent-of-origin specific associations across clinical traits derived from electronic health records

10.1101/2020.12.08.20246199 ◽

2020 ◽

Author(s):

Hye In Kim ◽

Bin Ye ◽

Jeffrey Staples ◽

Anthony Marcketta ◽

Chuan Gao ◽

...

Keyword(s):

Electronic Health Records ◽

Genetic Variants ◽

Additive Model ◽

European Ancestry ◽

Health Records ◽

Significance Threshold ◽

Specific Effects ◽

Genome Wide ◽

Parent Of Origin ◽

Electronic Health

AbstractParent-of-origin (PoO) effects refer to the differential phenotypic impact of genetic variants dependent on their parental inheritance. Genetic variants in imprinted genes can have PoO specific effects on complex traits, but these effects may be poorly captured by models that do not differentiate the parental origin of the variant. The aim of this study was to screen genome-wide imputed sequence for PoO effects on electronic health records (EHR) derived clinical traits in 134,049 individuals of European ancestry from the DiscovEHR study.Using pairwise kinship estimates from genetic data and demographic data, we identified 22,051 offspring with at least one parent present in the DiscovEHR study. We then assigned the PoO of ∼9 million variants in the heterozygous offspring using two methods. First, when one of the parental genotypes was homozygous, we determined PoO based on apparent Mendelian segregation. Second, we estimated PoO by comparing parental and offspring haplotypes around the variant allele. Using these PoO assignments, we performed genome-wide PoO association analyses across 154 quantitative traits including lab test results and biometric measures and 612 binary traits of ICD10 3-digit codes extracted from EHR in the DiscovEHR study. Out of 732 PoO associations meeting a significance threshold of P <5×10−8, we attempted to replicate 274 PoO associations in the UK Biobank study, consisting of 462,453 individuals and including 5,015 offspring with at least one parent, and replicated 9 PoO associations with nominal significance threshold P <0.05.In summary, the current study characterizes PoO effects of genetic variants genome-wide on a broad range of clinical traits derived from EHR in a large population study enriched for familial relationships. Our results suggest that 1) PoO specific effects are frequently captured by a standard additive model and that 2) statistical power to detect PoO specific effects remains modest even in large studies. Nonetheless, accurately modeling PoO effects of genetic variants has the potential to improve our understanding of the mechanism of the association and finding new associations that are not captured by the additive model.

Download Full-text