PheWAS: demonstrating the feasibility of a phenome-wide scan to discover gene-disease associations

Inter-labeler and intra-labeler variability of condition severity classification models using active and passive learning methods

BACKGROUND AND OBJECTIVES

Labeling instances by domain experts for classification is often time consuming and expensive. To reduce such labeling efforts, we had proposed the application of active learning (AL) methods, introduced our CAESAR-ALE framework for classifying the severity of clinical conditions, and shown its significant reduction of labeling efforts. The use of any of three AL methods (one well known [SVM-Margin], and two that we introduced [Exploitation and Combination_XA]) significantly reduced (by 48% to 64%) condition labeling efforts, compared to standard passive (random instance-selection) SVM learning. Furthermore, our new AL methods achieved maximal accuracy using 12% fewer labeled cases than the SVM-Margin AL method. However, because labelers have varying levels of expertise, a major issue associated with learning methods, and AL methods in particular, is how to best to use the labeling provided by a committee of labelers. First, we wanted to know, based on the labelers' learning curves, whether using AL methods (versus standard passive learning methods) has an effect on the Intra-labeler variability (within the learning curve of each labeler) and inter-labeler variability (among the learning curves of different labelers). Then, we wanted to examine the effect of learning (either passively or actively) from the labels created by the majority consensus of a group of labelers.

METHODS

We used our CAESAR-ALE framework for classifying the severity of clinical conditions, the three AL methods and the passive learning method, as mentioned above, to induce the classifications models. We used a dataset of 516 clinical conditions and their severity labeling, represented by features aggregated from the medical records of 1.9 million patients treated at Columbia University Medical Center. We analyzed the variance of the classification performance within (intra-labeler), and especially among (inter-labeler) the classification models that were induced by using the labels provided by seven labelers. We also compared the performance of the passive and active learning models when using the consensus label.

RESULTS

The AL methods: produced, for the models induced from each labeler, smoother Intra-labeler learning curves during the training phase, compared to the models produced when using the passive learning method. The mean standard deviation of the learning curves of the three AL methods over all labelers (mean: 0.0379; range: [0.0182 to 0.0496]), was significantly lower (p=0.049) than the Intra-labeler standard deviation when using the passive learning method (mean: 0.0484; range: [0.0275-0.0724). Using the AL methods resulted in a lower mean Inter-labeler AUC standard deviation among the AUC values of the labelers' different models during the training phase, compared to the variance of the induced models' AUC values when using passive learning. The Inter-labeler AUC standard deviation, using the passive learning method (0.039), was almost twice as high as the Inter-labeler standard deviation using our two new AL methods (0.02 and 0.019, respectively). The SVM-Margin AL method resulted in an Inter-labeler standard deviation (0.029) that was higher by almost 50% than that of our two AL methods The difference in the inter-labeler standard deviation between the passive learning method and the SVM-Margin learning method was significant (p=0.042). The difference between the SVM-Margin and Exploitation method was insignificant (p=0.29), as was the difference between the Combination_XA and Exploitation methods (p=0.67). Finally, using the consensus label led to a learning curve that had a higher mean intra-labeler variance, but resulted eventually in an AUC that was at least as high as the AUC achieved using the gold standard label and that was always higher than the expected mean AUC of a randomly selected labeler, regardless of the choice of learning method (including a passive learning method). Using a paired t-test, the difference between the intra-labeler AUC standard deviation when using the consensus label, versus that value when using the other two labeling strategies, was significant only when using the passive learning method (p=0.014), but not when using any of the three AL methods.

CONCLUSIONS

The use of AL methods, (a) reduces intra-labeler variability in the performance of the induced models during the training phase, and thus reduces the risk of halting the process at a local minimum that is significantly different in performance from the rest of the learned models; and (b) reduces Inter-labeler performance variance, and thus reduces the dependence on the use of a particular labeler. In addition, the use of a consensus label, agreed upon by a rather uneven group of labelers, might be at least as good as using the gold standard labeler, who might not be available, and certainly better than randomly selecting one of the group's individual labelers. Finally, using the AL methods: when provided by the consensus label reduced the intra-labeler AUC variance during the learning phase, compared to using passive learning.

Polygenic loading for major depression is associated with specific medical comorbidity

Major depressive disorder frequently co-occurs with medical disorders, raising the possibility of shared genetic liability. Recent identification of 15 novel genetic loci associated with depression allows direct investigation of this question. In cohorts of individuals participating in biobanks at two academic medical centers, we calculated polygenic loading for risk loci reported to be associated with depression. We then examined the association between such loading and 50 groups of clinical diagnoses, or topics, drawn from these patients' electronic health records, determined using a novel application of latent Dirichilet allocation. Three topics showed experiment-wide association with the depression liability score; these included diagnostic groups representing greater prevalence of mood and anxiety disorders, greater prevalence of cardiac ischemia, and a decreased prevalence of heart failure. The latter two associations persisted even among individuals with no mood disorder diagnosis. This application of a novel method for grouping related diagnoses in biobanks indicate shared genetic risk for depression and cardiac disease, with a pattern suggesting greater ischemic risk and diminished heart failure risk.

Efficient genome-wide association in biobanks using topic modeling identifies multiple novel disease loci

Biobanks and national registries represent a powerful tool for genomic discovery, but rely on diagnostic codes that may be unreliable and fail to capture the relationship between related diagnoses. We developed an efficient means of conducting genome-wide association studies using combinations of diagnostic codes from electronic health records (EHR) for 10845 participants in a biobanking program at two large academic medical centers. Specifically, we applied latent Dirichilet allocation to fit 50 disease topics based on diagnostic codes, then conducted genome-wide common-variant association for each topic. In sensitivity analysis, these results were contrasted with those obtained from traditional single-diagnosis phenome-wide association analysis, as well as those in which only a subset of diagnostic codes are included per topic. In meta-analysis across three biobank cohorts, we identified 23 disease-associated loci with p<1e-15, including previously associated autoimmune disease loci. In all cases, observed significant associations were of greater magnitude than for single phenome-wide diagnostic codes, and incorporation of less strongly-loading diagnostic codes enhanced association. This strategy provides a more efficient means of phenome-wide association in biobanks with coded clinical data.

Using Personalized Medicine in the Management of Diabetes Mellitus

Diabetes mellitus is a worldwide problem with an immense pharmacoeconomic burden. The multifactorial and complex nature of the disease lends itself to personalized pharmacotherapeutic approaches to treatment. Variability in individual risk and subsequent development of diabetes has been reported in addition to differences in response to the many oral glucose lowering therapies currently available for diabetes pharmacotherapy. Pharmacogenomic studies have attempted to uncover the heritable components of individual variability in risk susceptibility and response to pharmacotherapy. We review the current pharmacogenomics evidence as it relates to common oral glucose lowering therapies and how they can be utilized in the management of polygenic and monogenic forms of diabetes. Evidence supports the use of genetic testing and personalized approaches to the treatment of monogenic diabetes of the young. The data are not as robust for the current application of pharmacogenetic approaches to the treatment of polygenic type 2 diabetes mellitus, but there are suggestions as to future applications in this regard. We reviewed pertinent primary literature sources as well as current evidence-based guidelines on diabetes management.

Clinical Data Reuse or Secondary Use: Current Status and Potential Future Progress

Objective: To perform a review of recent research in clinical data reuse or secondary use, and envision future advances in this field. Methods: The review is based on a large literature search in MEDLINE (through PubMed), conference proceedings, and the ACM Digital Library, focusing only on research published between 2005 and early 2016. Each selected publication was reviewed by the authors, and a structured analysis and summarization of its content was developed. Results: The initial search produced 359 publications, reduced after a manual examination of abstracts and full publications. The following aspects of clinical data reuse are discussed: motivations and challenges, privacy and ethical concerns, data integration and interoperability, data models and terminologies, unstructured data reuse, structured data mining, clinical practice and research integration, and examples of clinical data reuse (quality measurement and learning healthcare systems). Conclusion: Reuse of clinical data is a fast-growing field recognized as essential to realize the potentials for high quality healthcare, improved healthcare management, reduced healthcare costs, population health management, and effective clinical research.

TEPAPA: a novel in silico feature learning pipeline for mining prognostic and associative factors from text-based electronic medical records

Vast amounts of clinically relevant text-based variables lie undiscovered and unexploited in electronic medical records (EMR). To exploit this untapped resource, and thus facilitate the discovery of informative covariates from unstructured clinical narratives, we have built a novel computational pipeline termed Text-based Exploratory Pattern Analyser for Prognosticator and Associator discovery (TEPAPA). This pipeline combines semantic-free natural language processing (NLP), regular expression induction, and statistical association testing to identify conserved text patterns associated with outcome variables of clinical interest. When we applied TEPAPA to a cohort of head and neck squamous cell carcinoma patients, plausible concepts known to be correlated with human papilloma virus (HPV) status were identified from the EMR text, including site of primary disease, tumour stage, pathologic characteristics, and treatment modalities. Similarly, correlates of other variables (including gender, nodal status, recurrent disease, smoking and alcohol status) were also reliably recovered. Using highly-associated patterns as covariates, a patient's HPV status was classifiable using a bootstrap analysis with a mean area under the ROC curve of 0.861, suggesting its predictive utility in supporting EMR-based phenotyping tasks. These data support using this integrative approach to efficiently identify disease-associated factors from unstructured EMR narratives, and thus to efficiently generate testable hypotheses.

Comparative analyses of population-scale phenomic data in electronic medical records reveal race-specific disease networks

Motivation: Underrepresentation of racial groups represents an important challenge and major gap in phenomics research. Most of the current human phenomics research is based primarily on European populations; hence it is an important challenge to expand it to consider other population groups. One approach is to utilize data from EMR databases that contain patient data from diverse demographics and ancestries. The implications of this racial underrepresentation of data can be profound regarding effects on the healthcare delivery and actionability. To the best of our knowledge, our work is the first attempt to perform comparative, population-scale analyses of disease networks across three different populations, namely Caucasian (EA), African American (AA) and Hispanic/Latino (HL).

Results: We compared susceptibility profiles and temporal connectivity patterns for 1988 diseases and 37 282 disease pairs represented in a clinical population of 1 025 573 patients. Accordingly, we revealed appreciable differences in disease susceptibility, temporal patterns, network structure and underlying disease connections between EA, AA and HL populations. We found 2158 significantly comorbid diseases for the EA cohort, 3265 for AA and 672 for HL. We further outlined key disease pair associations unique to each population as well as categorical enrichments of these pairs. Finally, we identified 51 key ‘hub’ diseases that are the focal points in the race-centric networks and of particular clinical importance. Incorporating race-specific disease comorbidity patterns will produce a more accurate and complete picture of the disease landscape overall and could support more precise understanding of disease relationships and patient management towards improved clinical outcomes.

Contacts: rong.chen@mssm.edu or joel.dudley@mssm.edu

Supplementary information:Supplementary data are available at Bioinformatics online.

Towards a more molecular taxonomy of disease

Background

Disease taxonomies have been designed for many applications, but they tend not to fully incorporate the growing amount of molecular-level knowledge of disease processes, inhibiting research efforts. Understanding the degree to which we can infer disease relationships from molecular data alone may yield insights into how to ultimately construct more modern taxonomies that integrate both physiological and molecular information.

Results

We introduce a new technique we call Parent Promotion to infer hierarchical relationships between disease terms using disease-gene data. We compare this technique with both an established ontology inference method (CliXO) and a minimum weight spanning tree approach. Because there is no gold standard molecular disease taxonomy available, we compare our inferred hierarchies to both the Medical Subject Headings (MeSH) category C forest of diseases and to subnetworks of the Disease Ontology (DO). This comparison provides insights about the inference algorithms, choices of evaluation metrics, and the existing molecular content of various subnetworks of MeSH and the DO. Our results suggest that the Parent Promotion method performs well in most cases. Performance across MeSH trees is also correlated between inference methods. Specifically, inferred relationships are more consistent with those in smaller MeSH disease trees than larger ones, but there are some notable exceptions that may correlate with higher molecular content in MeSH.

Conclusions

Our experiments provide insights about learning relationships between diseases from disease genes alone. Future work should explore the prospect of disease term discovery from molecular data and how best to integrate molecular data with anatomical and clinical knowledge. This study nonetheless suggests that disease gene information has the potential to form an important part of the foundation for future representations of the disease landscape.

Electronic supplementary material

The online version of this article (doi:10.1186/s13326-017-0134-0) contains supplementary material, which is available to authorized users.

IL-6 variant is associated with metastasis in breast cancer patients

INTRODUCTION

Although tumor metastases remain significant drivers of mortality, the genetic factors that increase the risks of metastases are not fully identified. Interleukin 6 (IL-6) has emerged as an important factor in breast cancer progression with IL-6 single nucleotide polymorphism (SNP) variants shown to affect survival. We hypothesized that SNPs of the IL-6 promoter at rs1800795 in breast cancer patients are associated with distant metastases.

METHODS

We performed an initial case-control study using Vanderbilt University Medical Center's BioVU, a genomic biobank linked to de-identified electronic medical records in the Synthetic Derivative database, to identify germline SNPs that may predict the development of metastatic disease to any site from any solid tumor including breast cancer. We identified a SNP in IL-6: rs1800795 to be of significance and evaluated this finding using a separate, matched-pair cohort of breast cancer patients with and without metastases from The Ohio State University Wexner Medical Center.

RESULTS

The initial study suggested that GG relative to CG at rs1800795 (OR 1.52; 95% CI 1.14-2.02; p = 0.004) was significantly associated with the development of metastases. This association was also observed in the Ohio State University cohort (OR 2.23; 95% CI 1.06-4.71; p = 0.001). There were no significant relationships between rs1800795 status and any patient or tumor characteristics, including estrogen receptor status.

CONCLUSIONS

These findings suggest that GG SNP at IL-6: rs1800795 may indicate an increased risk of metastasis of primary breast cancer. Further studies in larger population sets are warranted as advanced screening and prophylactic intervention might be employed in GG carriers.

Distribution and clinical impact of functional variants in 50,726 whole-exome sequences from the DiscovEHR study

The DiscovEHR collaboration between the Regeneron Genetics Center and Geisinger Health System couples high-throughput sequencing to an integrated health care system using longitudinal electronic health records (EHRs). We sequenced the exomes of 50,726 adult participants in the DiscovEHR study to identify ~4.2 million rare single-nucleotide variants and insertion/deletion events, of which ~176,000 are predicted to result in a loss of gene function. Linking these data to EHR-derived clinical phenotypes, we find clinical associations supporting therapeutic targets, including genes encoding drug targets for lipid lowering, and identify previously unidentified rare alleles associated with lipid levels and other blood level traits. About 3.5% of individuals harbor deleterious variants in 76 clinically actionable genes. The DiscovEHR data set provides a blueprint for large-scale precision medicine initiatives and genomics-guided therapeutic discovery.

Phenome-Wide Association Study of Rheumatoid Arthritis Subgroups Identifies Association Between Seronegative Disease and Fibromyalgia

OBJECTIVE

The differences between seronegative and seropositive rheumatoid arthritis (RA) have not been widely reported. We performed electronic health record (EHR)-based phenome-wide association studies (PheWAS) to identify disease associations in seropositive and seronegative RA.

METHODS

A validated algorithm identified RA subjects from the de-identified version of the Vanderbilt University Medical Center EHR. Serotypes were determined by rheumatoid factor (RF) and anti-cyclic citrullinated peptide antibody (ACPA) values. We tested EHR-derived phenotypes using PheWAS comparing seropositive RA and seronegative RA, yielding disease associations. PheWAS was also performed in RF-positive versus RF-negative subjects and ACPA-positive versus ACPA-negative subjects. Following PheWAS, select phenotypes were then manually reviewed, and fibromyalgia was specifically evaluated using a validated algorithm.

RESULTS

A total of 2,199 RA individuals with either RF or ACPA testing were identified. Of these, 1,382 patients (63%) were classified as seropositive. Seronegative RA was associated with myalgia and myositis (odds ratio [OR] 2.1, P = 3.7 × 10^-10 ) and back pain. A manual review of the health record showed that among subjects coded for Myalgia and Myositis, ∼80% had fibromyalgia. Follow-up with a specific EHR algorithm for fibromyalgia confirmed that seronegative RA was associated with fibromyalgia (OR 1.8, P = 4.0 × 10^-6 ). Seropositive RA was associated with chronic airway obstruction (OR 2.2, P = 1.4 × 10^-4 ) and tobacco use (OR 2.2, P = 7.0 × 10^-4 ).

CONCLUSION

This PheWAS of RA patients identifies a strong association between seronegativity and fibromyalgia. It also affirms relationships between seropositivity and chronic airway obstruction and between seropositivity and tobacco use. These findings demonstrate the utility of the PheWAS approach to discover novel phenotype associations within different subgroups of a disease.

Phenome-Wide Association Study of Autoantibodies to Citrullinated and Noncitrullinated Epitopes in Rheumatoid Arthritis

OBJECTIVE

Patients with rheumatoid arthritis (RA) develop autoantibodies against a spectrum of antigens, but the clinical significance of these autoantibodies is unclear. Using a phenome-wide association study (PheWAS) approach, we examined the association between autoantibodies and clinical subphenotypes of RA.

METHODS

This study was conducted in a cohort of RA patients identified from the electronic medical records (EMRs) of 2 tertiary care centers. Using a published multiplex bead assay, we measured 36 autoantibodies targeting epitopes implicated in RA. We extracted all International Classification of Diseases, Ninth Revision (ICD-9) codes for each subject and grouped them into disease categories (PheWAS codes), using a published method. We tested for the association of each autoantibody (grouped by the targeted protein) with PheWAS codes. To determine significant associations (at a false discovery rate [FDR] of ≤0.1), we reviewed the medical records of 50 patients with each PheWAS code to determine positive predictive values (PPVs).

RESULTS

We studied 1,006 RA patients; the mean ± SD age of the patients was 61.0 ± 12.9 years, and 79.0% were female. A total of 3,568 unique ICD-9 codes were grouped into 625 PheWAS codes; the 206 PheWAS codes with a prevalence of ≥3% were studied. Using the PheWAS method, we identified 24 significant associations of autoantibodies to epitopes at an FDR of ≤0.1. The associations that were strongest and had the highest PPV for the PheWAS code were autoantibodies against fibronectin and obesity (P = 6.1 × 10^-4 , PPV 100%), and that between fibrinogen and pneumonopathy (P = 2.7 × 10^-4 , PPV 96%). Pneumonopathy codes included diagnoses for cryptogenic organizing pneumonia and obliterative bronchiolitis.

CONCLUSION

We demonstrated application of a bioinformatics method, the PheWAS, to screen for the clinical significance of RA-related autoantibodies. Using the PheWAS approach, we identified potentially significant links between variations in the levels of autoantibodies and comorbidities of interest in RA.

Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record

OBJECTIVE

To compare three groupings of Electronic Health Record (EHR) billing codes for their ability to represent clinically meaningful phenotypes and to replicate known genetic associations. The three tested coding systems were the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, the Agency for Healthcare Research and Quality Clinical Classification Software for ICD-9-CM (CCS), and manually curated "phecodes" designed to facilitate phenome-wide association studies (PheWAS) in EHRs.

METHODS AND MATERIALS

We selected 100 disease phenotypes and compared the ability of each coding system to accurately represent them without performing additional groupings. The 100 phenotypes included 25 randomly-chosen clinical phenotypes pursued in prior genome-wide association studies (GWAS) and another 75 common disease phenotypes mentioned across free-text problem lists from 189,289 individuals. We then evaluated the performance of each coding system to replicate known associations for 440 SNP-phenotype pairs.

RESULTS

Out of the 100 tested clinical phenotypes, phecodes exactly matched 83, compared to 53 for ICD-9-CM and 32 for CCS. ICD-9-CM codes were typically too detailed (requiring custom groupings) while CCS codes were often not granular enough. Among 440 tested known SNP-phenotype associations, use of phecodes replicated 153 SNP-phenotype pairs compared to 143 for ICD-9-CM and 139 for CCS. Phecodes also generally produced stronger odds ratios and lower p-values for known associations than ICD-9-CM and CCS. Finally, evaluation of several SNPs via PheWAS identified novel potential signals, some seen in only using the phecode approach. Among them, rs7318369 in PEPD was associated with gastrointestinal hemorrhage.

CONCLUSION

Our results suggest that the phecode groupings better align with clinical diseases mentioned in clinical practice or for genomic studies. ICD-9-CM, CCS, and phecode groupings all worked for PheWAS-type studies, though the phecode groupings produced superior results.

A Fast and Accurate Algorithm to Test for Binary Phenotypes and Its Application to PheWAS

The availability of electronic health record (EHR)-based phenotypes allows for genome-wide association analyses in thousands of traits and has great potential to enable identification of genetic variants associated with clinical phenotypes. We can interpret the phenome-wide association study (PheWAS) result for a single genetic variant by observing its association across a landscape of phenotypes. Because a PheWAS can test thousands of binary phenotypes, and most of them have unbalanced or often extremely unbalanced case-control ratios (1:10 or 1:600, respectively), existing methods cannot provide an accurate and scalable way to test for associations. Here, we propose a computationally fast score-test-based method that estimates the distribution of the test statistic by using the saddlepoint approximation. Our method is much (∼100 times) faster than the state-of-the-art Firth's test. It can also adjust for covariates and control type I error rates even when the case-control ratio is extremely unbalanced. Through application to PheWAS data from the Michigan Genomics Initiative, we show that the proposed method can control type I error rates while replicating previously known association signals even for traits with a very small number of cases and a large number of controls.

Phenome-wide association studies: a new method for functional genomics in humans

In experimental physiological research, a common study design for examining the functional role of a gene or a genetic variant is to introduce that genetic variant into a model organism (such as yeast or mouse) and then to search for phenotypic consequences. The development of DNA biobanks linked to dense phenotypic information enables such an experiment to be applied to human subjects in the form of a phenome-wide association study (PheWAS). The PheWAS paradigm takes advantage of a curated medical phenome, often derived from electronic health records, to search for associations between 'input functions' and phenotypes in an unbiased fashion. The most commonly studied input function to date has been single nucleotide polymorphisms (SNPs), but other inputs, such as sets of SNPs or a disease or drug exposure, are now being explored to probe the genetic and phenotypic architecture of human traits. Potential outcomes of these approaches include defining subsets of complex diseases (that can then be targeted by specific therapies) and drug repurposing.

Text mining applied to electronic cardiovascular procedure reports to identify patients with trileaflet aortic stenosis and coronary artery disease

BACKGROUND

Interrogation of the electronic health record (EHR) using billing codes as a surrogate for diagnoses of interest has been widely used for clinical research. However, the accuracy of this methodology is variable, as it reflects billing codes rather than severity of disease, and depends on the disease and the accuracy of the coding practitioner. Systematic application of text mining to the EHR has had variable success for the detection of cardiovascular phenotypes. We hypothesize that the application of text mining algorithms to cardiovascular procedure reports may be a superior method to identify patients with cardiovascular conditions of interest.

METHODS

We adapted the Oracle product Endeca, which utilizes text mining to identify terms of interest from a NoSQL-like database, for purposes of searching cardiovascular procedure reports and termed the tool "PennSeek". We imported 282,569 echocardiography reports representing 81,164 individuals and 27,205 cardiac catheterization reports representing 14,567 individuals from non-searchable databases into PennSeek. We then applied clinical criteria to these reports in PennSeek to identify patients with trileaflet aortic stenosis (TAS) and coronary artery disease (CAD). Accuracy of patient identification by text mining through PennSeek was compared with ICD-9 billing codes.

RESULTS

Text mining identified 7115 patients with TAS and 9247 patients with CAD. ICD-9 codes identified 8272 patients with TAS and 6913 patients with CAD. 4346 patients with AS and 6024 patients with CAD were identified by both approaches. A randomly selected sample of 200-250 patients uniquely identified by text mining was compared with 200-250 patients uniquely identified by billing codes for both diseases. We demonstrate that text mining was superior, with a positive predictive value (PPV) of 0.95 compared to 0.53 by ICD-9 for TAS, and a PPV of 0.97 compared to 0.86 for CAD.

CONCLUSION

These results highlight the superiority of text mining algorithms applied to electronic cardiovascular procedure reports in the identification of phenotypes of interest for cardiovascular research.

The Diagnosis-Wide Landscape of Hospital-Acquired AKI

BACKGROUND AND OBJECTIVES

The exploration of electronic hospital records offers a unique opportunity to describe in-depth the prevalence of conditions associated with diagnoses at an unprecedented level of comprehensiveness. We used a diagnosis-wide approach, adapted from phenome-wide association studies (PheWAS), to perform an exhaustive analysis of all diagnoses associated with hospital-acquired AKI (HA-AKI) in a French urban tertiary academic hospital over a period of 10 years.

DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS

We retrospectively extracted all diagnoses from an i2b2 (Informatics for Integrating Biology and the Bedside) clinical data warehouse for patients who stayed in this hospital between 2006 and 2015 and had at least two plasma creatinine measurements performed during the first week of their stay. We then analyzed the association between HA-AKI and each International Classification of Diseases (ICD)-10 diagnostic category to draw a comprehensive picture of diagnoses associated with AKI. Hospital stays for 126,736 unique individuals were extracted.

RESULTS

Hemodynamic impairment and surgical procedures are the main factors associated with HA-AKI and five clusters of diagnoses were identified: sepsis, heart diseases, polytrauma, liver disease, and cardiovascular surgery. The ICD-10 code corresponding to AKI (N17) was recorded in 30% of the cases with HA-AKI identified, and in this situation, 20% of the diagnoses associated with HA-AKI corresponded to kidney diseases such as tubulointerstitial nephritis, necrotizing vasculitis, or myeloma cast nephropathy. Codes associated with HA-AKI that demonstrated the greatest increase in prevalence with time were related to influenza, polytrauma, and surgery of neoplasms of the genitourinary system.

CONCLUSIONS

Our approach, derived from PheWAS, is a valuable way to comprehensively identify and classify all of the diagnoses and clusters of diagnoses associated with HA-AKI. Our analysis delivers insights into how diagnoses associated with HA-AKI evolved over time. On the basis of ICD-10 codes, HA-AKI appears largely underestimated in this academic hospital.

Phenome-wide scanning identifies multiple diseases and disease severity phenotypes associated with HLA variants

Although many phenotypes have been associated with variants in human leukocyte antigen (HLA) genes, the full phenotypic impact of HLA variants across all diseases is unknown. We imputed HLA genomic variation from two populations of 28,839 and 8431 European ancestry individuals and tested association of HLA variation with 1368 phenotypes. A total of 104 four-digit and 92 two-digit HLA allele phenotype associations were significant in both discovery and replication cohorts, the strongest being HLA-DQB1*03:02 and type 1 diabetes. Four previously unidentified associations were identified across the spectrum of disease with two- and four-digit HLA alleles and 10 with nonsynonymous variants. Some conditions associated with multiple HLA variants and stronger associations with more severe disease manifestations were identified. A comprehensive, publicly available catalog of clinical phenotypes associated with HLA variation is provided. Examining HLA variant disease associations in this large data set allows comprehensive definition of disease associations to drive further mechanistic insights.

The joint effect of air pollution exposure and copy number variation on risk for autism

Autism spectrum disorder is a complex trait with a high degree of heritability as well as documented susceptibility from environmental factors. In this study the contributions of copy number variation, exposure to air pollutants, and the interaction between the two on autism risk, were evaluated in the population-based case-control Childhood Autism Risks from Genetics and Environment (CHARGE) Study. For the current investigation, we included only those CHARGE children (a) who met criteria for autism or typical development and (b) for whom our team had conducted both genetic evaluation of copy number burden and determination of environmental air pollution exposures based on mapping addresses from the pregnancy and early childhood. This sample consisted of 158 cases of children with autism and 147 controls with typical development. Multiple logistic regression models were fit with and without environmental variable-copy number burden interactions. We found no correlation between average air pollution exposure from conception to age 2 years and the child's CNV burden. We found a significant interaction in which a 1SD increase in duplication burden combined with a 1SD increase in ozone exposure was associated with an elevated autism risk (OR 3.4, P < 0.005) much greater than the increased risks associated with either genomic duplication (OR 1.85, 95% CI 1.25-2.73) or ozone (OR 1.20, 95% CI 0.93-1.54) alone. Similar results were obtained when CNV and ozone were dichotomized to compare those in the top quartile relative to those having a smaller CNV burden and lower exposure to ozone, and when exposures were assessed separately for pregnancy, the first year of life, and the second year of life. No interactions were observed for other air pollutants, even those that demonstrated main effects; ozone tends to be negatively correlated with the other pollutants examined. While earlier work has demonstrated interactions between the presence of a pathogenic CNV and an environmental exposure [Webb et al., 2016], these findings appear to be the first indication that global copy number variation may increase susceptibility to certain environmental factors, and underscore the need to consider both genomics and environmental exposures as well as the mechanisms by which each may amplify the risks for autism associated with the other. Autism Res 2017, 10: 1470-1480. © 2017 International Society for Autism Research, Wiley Periodicals, Inc.

Investigating the Genetic Architecture of the PR Interval Using Clinical Phenotypes

BACKGROUND

One potential use for the PR interval is as a biomarker of disease risk. We hypothesized that quantifying the shared genetic architectures of the PR interval and a set of clinical phenotypes would identify genetic mechanisms contributing to PR variability and identify diseases associated with a genetic predictor of PR variability.

METHODS AND RESULTS

We used ECG measurements from the ARIC study (Atherosclerosis Risk in Communities; n=6731 subjects) and 63 genetically modulated diseases from the eMERGE network (Electronic Medical Records and Genomics; n=12 978). We measured pairwise genetic correlations (rG) between PR phenotypes (PR interval, PR segment, P-wave duration) and each of the 63 phenotypes. The PR segment was genetically correlated with atrial fibrillation (rG=-0.88; P=0.0009). An analysis of metabolic phenotypes in ARIC also showed that the P wave was genetically correlated with waist circumference (rG=0.47; P=0.02). A genetically predicted PR interval phenotype based on 645 714 single-nucleotide polymorphisms was associated with atrial fibrillation (odds ratio=0.89 per SD change; 95% confidence interval, 0.83-0.95; P=0.0006). The differing pattern of associations among the PR phenotypes is consistent with analyses that show that the genetic correlation between the P wave and PR segment was not significantly different from 0 (rG=-0.03 [0.16]).

CONCLUSIONS

The genetic architecture of the PR interval comprises modulators of atrial fibrillation risk and obesity.

Defining the complex phenotype of severe systemic loxoscelism using a large electronic health record cohort

Objective

Systemic loxoscelism is a rare illness resulting from the bite of the recluse spider and, in its most severe form, can lead to widespread hemolysis, coagulopathy, and death. We aim to describe the clinical features and outcomes of the largest known cohort of individuals with moderate to severe loxoscelism.

Methods

We performed a retrospective, cross sectional study from January 1, 1995, to December 31, 2015, at a tertiary-care academic medical center, to determine individuals with clinical records consistent with moderate to severe loxoscelism. Age-, sex-, and race-matched controls were compared. Demographics, clinical characteristics, laboratory measures, and outcomes of individuals with loxoscelism are described. Case and control groups were compared with descriptive statistics and phenome-wide association study (PheWAS).

Results

During the time period, 57 individuals were identified as having moderate to severe loxoscelism. Of these, only 33% had an antecedent spider bite documented. Median age of individuals diagnosed with moderate to severe loxoscelism was 14 years old (IQR 9.0–24.0 years). PheWAS confirmed associations of systemic loxoscelism with 29 other phenotypes, e.g., rash, hemolytic anemia, and sepsis. Hemoglobin level dropped an average of 3.1 g/dL over an average of 2.0 days (IQR 2.0–6.0). Lactate dehydrogenase and total bilirubin levels were on average over two times their upper limit of normal values. Eighteen individuals of 32 tested had a positive direct antiglobulin (Coombs’) test. Mortality was 3.5% (2/57 individuals).

Conclusion

Systemic loxoscelism is a rare but devastating process with only a minority of patients recalling the toxic exposure; hemolysis reaches a peak at 2 days after admission, with some cases taking more than a week before recovery. In endemic areas, suspicion for systemic loxoscelism should be high in individuals, especially children and younger adults, presenting with a cutaneous ulcer and hemolysis or coagulopathy, even in the absence of a bite exposure history.

Combining clinical and genomics queries using i2b2 - Three methods

We are fortunate to be living in an era of twin biomedical data surges: a burgeoning representation of human phenotypes in the medical records of our healthcare systems, and high-throughput sequencing making rapid technological advances. The difficulty representing genomic data and its annotations has almost by itself led to the recognition of a biomedical "Big Data" challenge, and the complexity of healthcare data only compounds the problem to the point that coherent representation of both systems on the same platform seems insuperably difficult. We investigated the capability for complex, integrative genomic and clinical queries to be supported in the Informatics for Integrating Biology and the Bedside (i2b2) translational software package. Three different data integration approaches were developed: The first is based on Sequence Ontology, the second is based on the tranSMART engine, and the third on CouchDB. These novel methods for representing and querying complex genomic and clinical data on the i2b2 platform are available today for advancing precision medicine.

Accelerating Precision Drug Development and Drug Repurposing by Leveraging Human Genetics

The potential impact of using human genetic data linked to longitudinal electronic medical records on drug development is extraordinary; however, the practical application of these data necessitates some organizational innovations. Vanderbilt has created resources such as an easily queried database of >2.6 million de-identified electronic health records linked to BioVU, which is a DNA biobank with more than 230,000 unique samples. To ensure these data are used to maximally benefit and accelerate both de novo drug discovery and drug repurposing efforts, we created the Accelerating Drug Development and Repurposing Incubator, a multidisciplinary think tank of experts in various therapeutic areas within both basic and clinical science as well as experts in legal, business, and other operational domains. The Incubator supports a diverse pipeline of drug indication finding projects, leveraging the natural experiment of human genetics.

Mining the topography and dynamics of the 4D Nucleome to identify novel CNS drug pathways

The pharmacoepigenome can be defined as the active, noncoding province of the genome including canonical spatial and temporal regulatory mechanisms of gene regulation that respond to xenobiotic stimuli. Many psychotropic drugs that have been in clinical use for decades have ill-defined mechanisms of action that are beginning to be resolved as we understand the transcriptional hierarchy and dynamics of the nucleus. In this review, we describe spatial, temporal and biomechanical mechanisms mediated by psychotropic medications. Focus is placed on a bioinformatics pipeline that can be used both for detection of pharmacoepigenomic variants that discretize drug response and adverse events to improve pharmacogenomic testing, and for the discovery of novel CNS therapeutics. This approach integrates the functional topology and dynamics of the transcriptional hierarchy of the pharmacoepigenome, gene variant-driven identification of pharmacogenomic regulatory domains, and mesoscale mapping for the discovery of novel CNS pharmacodynamic pathways in human brain. Examples of the application of this pipeline are provided, including the discovery of valproic acid (VPA) mediated transcriptional reprogramming of neuronal cell fate following injury, and mapping of a CNS pathway glutamatergic pathway for the mood stabilizer lithium. These examples in regulatory pharmacoepigenomics illustrate how ongoing research using the 4D nucleome provides a foundation to further insight into previously unrecognized psychotropic drug pharmacodynamic pathways in the human CNS.

Identifying collaborative care teams through electronic medical record utilization patterns

OBJECTIVE

The goal of this investigation was to determine whether automated approaches can learn patient-oriented care teams via utilization of an electronic medical record (EMR) system.

MATERIALS AND METHODS

To perform this investigation, we designed a data-mining framework that relies on a combination of latent topic modeling and network analysis to infer patterns of collaborative teams. We applied the framework to the EMR utilization records of over 10 000 employees and 17 000 inpatients at a large academic medical center during a 4-month window in 2010. Next, we conducted an extrinsic evaluation of the patterns to determine the plausibility of the inferred care teams via surveys with knowledgeable experts. Finally, we conducted an intrinsic evaluation to contextualize each team in terms of collaboration strength (via a cluster coefficient) and clinical credibility (via associations between teams and patient comorbidities).

RESULTS

The framework discovered 34 collaborative care teams, 27 (79.4%) of which were confirmed as administratively plausible. Of those, 26 teams depicted strong collaborations, with a cluster coefficient > 0.5. There were 119 diagnostic conditions associated with 34 care teams. Additionally, to provide clarity on how the survey respondents arrived at their determinations, we worked with several oncologists to develop an illustrative example of how a certain team functions in cancer care.

DISCUSSION

Inferred collaborative teams are plausible; translating such patterns into optimized collaborative care will require administrative review and integration with management practices.

CONCLUSIONS

EMR utilization records can be mined for collaborative care patterns in large complex medical centers.

Informatics and Data Analytics to Support Exposome-Based Discovery for Public Health

The complexity of the human exposome-the totality of environmental exposures encountered from birth to death-motivates systematic, high-throughput approaches to discover new environmental determinants of disease. In this review, we describe the state of science in analyzing the human exposome and provide recommendations for the public health community to consider in dealing with analytic challenges of exposome-based biomedical research. We describe extant and novel analytic methods needed to associate the exposome with critical health outcomes and contextualize the data-centered challenges by drawing parallels to other research endeavors such as human genomics research. We discuss efforts for training scientists who can bridge public health, genomics, and biomedicine in informatics and statistics. If an exposome data ecosystem is brought to fruition, it will likely play a role as central as genomic science has had in molding the current and new generations of biomedical researchers, computational scientists, and public health research programs.

Hypothesis-Free Search for Connections between Birth Month and Disease Prevalence in Large, Geographically Varied Cohorts

We have sought to replicate and extend the Season-wide Association Study (SeaWAS) of Boland, et al.1 in identifying birth month-disease associations from electronic health records (EHRs). We used methodology similar to that implemented by Boland on three geographically distinct cohorts, for a total of 11.8 million individuals derived from multiple data sources. We were able to identify eleven out of sixteen literature-supported birth month associations as compared to seven of sixteen for SeaWAS. Of the nine novel cardiovascular birth month associations discovered by SeaWAS, we were able to replicate four. None of the novel non-cardiovascular associations discovered by SeaWAS emerged as significant relations in our study. We identified thirty birth month disease associations not previously reported; of those, only six associations were validated in more than one cohort. These results suggest that differences in cohort composition and location can cause consequential variation in results of hypothesis-free searches.

Phenotypic Analysis of Clinical Narratives Using Human Phenotype Ontology

Phenotypes are defined as observable characteristics and clinical traits of diseases and organisms. As connectors between medical experimental findings and clinical practices, phenotypes play vital roles in translational medicine. To facilitate the translation between genotype and phenotype, Human Phenotype Ontology (HPO) was developed as a semantically computable vocabulary to capture phenotypic abnormalities found in human diseases discovered through biomedical research. The use of HPO in annotating phenotypic information in clinical practice remains unexplored. In this study, we investigated the use of HPO to annotate phenotypic information in clinical domain by leveraging a corpus of 12.8 million clinical notes created from 2010 to 2015 for 729 thousand patients at Mayo Clinic Rochester campus and assessed the distribution information of HPO terms in the corpus. We also analyzed the distributional difference of HPO terms among demographic groups. We further demonstrated the potential application of the annotated corpus to support knowledge discovery in precision medicine through Wilson's Disease.

How to Become a Smart Patient in the Era of Precision Medicine?

The objective of this paper is to define the definition of smart patients, summarize the existing foundation, and explore the approaches and system participation model of how to become a smart patient. Here a thorough review of the literature was conducted to make theory derivation processes of the smart patient; "data, information, knowledge, and wisdom (DIKW) framework" was performed to construct the model of how smart patients participate in the medical process. The smart patient can take an active role and fully participate in their own health management; DIKW system model provides a theoretical framework and practical model of smart patients; patient education is the key to the realization of smart patients. The conclusion is that the smart patient is attainable and he or she is not merely a patient but more importantly a captain and global manager of one's own health management, a partner of medical practitioner, and also a supervisor of medical behavior. Smart patients can actively participate in their healthcare and assume higher levels of responsibility for their own health and wellness which can facilitate the development of precision medicine and its widespread practice.

Using nationwide ‘big data’ from linked electronic health records to help improve outcomes in cardiovascular diseases: 33 studies using methods from epidemiology, informatics, economics and social science in the ClinicAl disease research using LInked Bespoke studies and Electronic health Records (CALIBER) programme

Background

Electronic health records (EHRs), when linked across primary and secondary care and curated for research use, have the potential to improve our understanding of care quality and outcomes.

Objective

To evaluate new opportunities arising from linked EHRs for improving quality of care and outcomes for patients at risk of or with coronary disease across the patient journey.

Design

Epidemiological cohort, health informatics, health economics and ethnographic approaches were used.

Setting

230 NHS hospitals and 226 general practices in England and Wales.

Participants

Up to 2 million initially healthy adults, 100,000 people with stable coronary artery disease (SCAD) and up to 300,000 patients with acute coronary syndrome.

Main outcome measures

Quality of care, fatal and non-fatal cardiovascular disease (CVD) events.

Data platform and methods

We created a novel research platform [ClinicAl disease research using LInked Bespoke studies and Electronic health Records (CALIBER)] based on linkage of four major sources of EHR data in primary care and national registries. We carried out 33 complementary studies within the CALIBER framework. We developed a web-based clinical decision support system (CDSS) in hospital chest pain clinics. We established a novel consented prognostic clinical cohort of SCAD patients.

Results

CALIBER was successfully established as a valid research platform based on linked EHR data in nearly 2 million adults with > 600 EHR phenotypes implemented on the web portal (seehttps://caliberresearch.org/portal). Despite national guidance, key opportunities for investigation and treatment were missed across the patient journey, resulting in a worse prognosis for patients in the UK compared with patients in health systems in other countries. Our novel, contemporary, high-resolution studies showed heterogeneous associations for CVD risk factors across CVDs. The CDSS did not alter the decision-making behaviour of clinicians in chest pain clinics. Prognostic models using real-world data validly discriminated risk of death and events, and were used in cost-effectiveness decision models.

Conclusions

Emerging ‘big data’ opportunities arising from the linkage of records at different stages of a patient’s journey are vital to the generation of actionable insights into the diagnosis, risk stratification and cost-effective treatment of people at risk of, or with, CVD.

Future work

The vast majority of NHS data remain inaccessible to research and this hampers efforts to improve efficiency and quality of care and to drive innovation. We propose three priority directions for further research. First, there is an urgent need to ‘unlock’ more detailed data within hospitals for the scale of the UK’s 65 million population. Second, there is a need for scaled approaches to using EHRs to design and carry out trials, and interpret the implementation of trial results. Third, large-scale, disease agnostic genetic and biological collections linked to such EHRs are required in order to deliver precision medicine and to innovate discovery.

Study registration

CALIBER studies are registered as follows: study 2 – NCT01569139, study 4 – NCT02176174 and NCT01164371, study 5 – NCT01163513, studies 6 and 7 – NCT01804439, study 8 – NCT02285322, and studies 26–29 – NCT01162187. Optimising the Management of Angina is registered as Current Controlled Trials ISRCTN54381840.

Funding

The National Institute for Health Research (NIHR) Programme Grants for Applied Research programme (RP-PG-0407-10314) (all 33 studies) and additional funding from the Wellcome Trust (study 1), Medical Research Council Partnership grant (study 3), Servier (study 16), NIHR Research Methods Fellowship funding (study 19) and NIHR Research for Patient Benefit (study 33).

Natural Selection in the Great Apes

Natural selection is crucial for the adaptation of populations to their environments. Here, we present the first global study of natural selection in the Hominidae (humans and great apes) based on genome-wide information from population samples representing all extant species (including most subspecies). Combining several neutrality tests we create a multi-species map of signatures of natural selection covering all major types of natural selection. We find that the estimated efficiency of both purifying and positive selection varies between species and is significantly correlated with their long-term effective population size. Thus, even the modest differences in population size among the closely related Hominidae lineages have resulted in differences in their ability to remove deleterious alleles and to adapt to changing environments. Most signatures of balancing and positive selection are species-specific, with signatures of balancing selection more often being shared among species. We also identify loci with evidence of positive selection across several lineages. Notably, we detect signatures of positive selection in several genes related to brain function, anatomy, diet and immune processes. Our results contribute to a better understanding of human evolution by putting the evidence of natural selection in humans within its larger evolutionary context. The global map of natural selection in our closest living relatives is available as an interactive browser at http://tinyurl.com/nf8qmzh.

Meta-Analysis of Genome-Wide Association Studies for Abdominal Aortic Aneurysm Identifies Four New Disease-Specific Risk Loci

Supplemental Digital Content is available in the text.

Rationale:

Abdominal aortic aneurysm (AAA) is a complex disease with both genetic and environmental risk factors. Together, 6 previously identified risk loci only explain a small proportion of the heritability of AAA.

Objective:

To identify additional AAA risk loci using data from all available genome-wide association studies.

Methods and Results:

Through a meta-analysis of 6 genome-wide association study data sets and a validation study totaling 10 204 cases and 107 766 controls, we identified 4 new AAA risk loci: 1q32.3 (SMYD2), 13q12.11 (LINC00540), 20q13.12 (near PCIF1/MMP9/ZNF335), and 21q22.2 (ERG). In various database searches, we observed no new associations between the lead AAA single nucleotide polymorphisms and coronary artery disease, blood pressure, lipids, or diabetes mellitus. Network analyses identified ERG, IL6R, and LDLR as modifiers of MMP9, with a direct interaction between ERG and MMP9.

Conclusions:

The 4 new risk loci for AAA seem to be specific for AAA compared with other cardiovascular diseases and related traits suggesting that traditional cardiovascular risk factor management may only have limited value in preventing the progression of aneurysmal disease.

A principal component meta-analysis on multiple anthropometric traits identifies novel loci for body shape

Large consortia have revealed hundreds of genetic loci associated with anthropometric traits, one trait at a time. We examined whether genetic variants affect body shape as a composite phenotype that is represented by a combination of anthropometric traits. We developed an approach that calculates averaged PCs (AvPCs) representing body shape derived from six anthropometric traits (body mass index, height, weight, waist and hip circumference, waist-to-hip ratio). The first four AvPCs explain >99% of the variability, are heritable, and associate with cardiometabolic outcomes. We performed genome-wide association analyses for each body shape composite phenotype across 65 studies and meta-analysed summary statistics. We identify six novel loci: LEMD2 and CD47 for AvPC1, RPS6KA5/C14orf159 and GANAB for AvPC3, and ARL15 and ANP32 for AvPC4. Our findings highlight the value of using multiple traits to define complex phenotypes for discovery, which are not captured by single-trait analyses, and may shed light onto new pathways.

Biomechanisms of Comorbidity: Reviewing Integrative Analyses of Multi-omics Datasets and Electronic Health Records

OBJECTIVES

Disease comorbidity is a pervasive phenomenon impacting patients' health outcomes, disease management, and clinical decisions. This review presents past, current and future research directions leveraging both phenotypic and molecular information to uncover disease similarity underpinning the biology and etiology of disease comorbidity.

METHODS

We retrieved ~130 publications and retained 59, ranging from 2006 to 2015, that comprise a minimum number of five diseases and at least one type of biomolecule. We surveyed their methods, disease similarity metrics, and calculation of comorbidities in the electronic health records, if present.

RESULTS

Among the surveyed studies, 44% generated or validated disease similarity metrics in context of comorbidity, with 60% being published in the last two years. As inputs, 87% of studies utilized intragenic loci and proteins while 13% employed RNA (mRNA, LncRNA or miRNA). Network modeling was predominantly used (35%) followed by statistics (28%) to impute similarity between these biomolecules and diseases. Studies with large numbers of biomolecules and diseases used network models or naïve overlap of disease-molecule associations, while machine learning, statistics, and information retrieval were utilized in smaller and moderate sized studies. Multiscale computations comprising shared function, network topology, and phenotypes were performed exclusively on proteins.

CONCLUSION

This review highlighted the growing methods for identifying the molecular mechanisms underpinning comorbidities that leverage multiscale molecular information and patterns from electronic health records. The survey unveiled that intergenic polymorphisms have been overlooked for similarity imputation compared to their intragenic counterparts, offering new opportunities to bridge the mechanistic and similarity gaps of comorbidity.

PheKB: a catalog and workflow for creating electronic phenotype algorithms for transportability

OBJECTIVE

Health care generated data have become an important source for clinical and genomic research. Often, investigators create and iteratively refine phenotype algorithms to achieve high positive predictive values (PPVs) or sensitivity, thereby identifying valid cases and controls. These algorithms achieve the greatest utility when validated and shared by multiple health care systems.Materials and Methods We report the current status and impact of the Phenotype KnowledgeBase (PheKB, http://phekb.org), an online environment supporting the workflow of building, sharing, and validating electronic phenotype algorithms. We analyze the most frequent components used in algorithms and their performance at authoring institutions and secondary implementation sites.

RESULTS

As of June 2015, PheKB contained 30 finalized phenotype algorithms and 62 algorithms in development spanning a range of traits and diseases. Phenotypes have had over 3500 unique views in a 6-month period and have been reused by other institutions. International Classification of Disease codes were the most frequently used component, followed by medications and natural language processing. Among algorithms with published performance data, the median PPV was nearly identical when evaluated at the authoring institutions (n = 44; case 96.0%, control 100%) compared to implementation sites (n = 40; case 97.5%, control 100%).

DISCUSSION

These results demonstrate that a broad range of algorithms to mine electronic health record data from different health systems can be developed with high PPV, and algorithms developed at one site are generally transportable to others.

CONCLUSION

By providing a central repository, PheKB enables improved development, transportability, and validity of algorithms for research-grade phenotypes using health care generated data.

Minimax Rate-optimal Estimation of High-dimensional Covariance Matrices with Incomplete Data

Missing data occur frequently in a wide range of applications. In this paper, we consider estimation of high-dimensional covariance matrices in the presence of missing observations under a general missing completely at random model in the sense that the missingness is not dependent on the values of the data. Based on incomplete data, estimators for bandable and sparse covariance matrices are proposed and their theoretical and numerical properties are investigated. Minimax rates of convergence are established under the spectral norm loss and the proposed estimators are shown to be rate-optimal under mild regularity conditions. Simulation studies demonstrate that the estimators perform well numerically. The methods are also illustrated through an application to data from four ovarian cancer studies. The key technical tools developed in this paper are of independent interest and potentially useful for a range of related problems in high-dimensional statistical inference with missing data.

Defining a Contemporary Ischemic Heart Disease Genetic Risk Profile Using Historical Data

BACKGROUND

Continued reductions in morbidity and mortality attributable to ischemic heart disease (IHD) require an understanding of the changing epidemiology of this disease. We hypothesized that we could use genetic correlations, which quantify the shared genetic architectures of phenotype pairs and extant risk factors from a historical prospective study to define the risk profile of a contemporary IHD phenotype.

METHODS AND RESULTS

We used 37 phenotypes measured in the ARIC study (Atherosclerosis Risk in Communities; n=7716, European ancestry subjects) and clinical diagnoses from an electronic health record (EHR) data set (n=19 093). All subjects had genome-wide single-nucleotide polymorphism genotyping. We measured pairwise genetic correlations (rG) between the ARIC and EHR phenotypes using linear mixed models. The genetic correlation estimates between the ARIC risk factors and the EHR IHD were modestly linearly correlated with hazards ratio estimates for incident IHD in ARIC (Pearson correlation [r]=0.62), indicating that the 2 IHD phenotypes had differing risk profiles. For comparison, this correlation was 0.80 when comparing EHR and ARIC type 2 diabetes mellitus phenotypes. The EHR IHD phenotype was most strongly correlated with ARIC metabolic phenotypes, including total:high-density lipoprotein cholesterol ratio (rG=-0.44, P=0.005), high-density lipoprotein (rG=-0.48, P=0.005), systolic blood pressure (rG=0.44, P=0.02), and triglycerides (rG=0.38, P=0.02). EHR phenotypes related to type 2 diabetes mellitus, atherosclerotic, and hypertensive diseases were also genetically correlated with these ARIC risk factors.

CONCLUSIONS

The EHR IHD risk profile differed from ARIC and indicates that treatment and prevention efforts in this population should target hypertensive and metabolic disease.

Transcription factor ETV1 is essential for rapid conduction in the heart

Rapid impulse propagation in the heart is a defining property of pectinated atrial myocardium (PAM) and the ventricular conduction system (VCS) and is essential for maintaining normal cardiac rhythm and optimal cardiac output. Conduction defects in these tissues produce a disproportionate burden of arrhythmic disease and are major predictors of mortality in heart failure patients. Despite the clinical importance, little is known about the gene regulatory network that dictates the fast conduction phenotype. Here, we have used signal transduction and transcriptional profiling screens to identify a genetic pathway that converges on the NRG1-responsive transcription factor ETV1 as a critical regulator of fast conduction physiology for PAM and VCS cardiomyocytes. Etv1 was highly expressed in murine PAM and VCS cardiomyocytes, where it regulates expression of Nkx2-5, Gja5, and Scn5a, key cardiac genes required for rapid conduction. Mice deficient in Etv1 exhibited marked cardiac conduction defects coupled with developmental abnormalities of the VCS. Loss of Etv1 resulted in a complete disruption of the normal sodium current heterogeneity that exists between atrial, VCS, and ventricular myocytes. Lastly, a phenome-wide association study identified a link between ETV1 and bundle branch block and heart block in humans. Together, these results identify ETV1 as a critical factor in determining fast conduction physiology in the heart.

SZGR 2.0: a one-stop shop of schizophrenia candidate genes

SZGR 2.0 is a comprehensive resource of candidate variants and genes for schizophrenia, covering genetic, epigenetic, transcriptomic, translational and many other types of evidence. By systematic review and curation of multiple lines of evidence, we included almost all variants and genes that have ever been reported to be associated with schizophrenia. In particular, we collected ∼4200 common variants reported in genome-wide association studies, ∼1000 de novo mutations discovered by large-scale sequencing of family samples, 215 genes spanning rare and replication copy number variations, 99 genes overlapping with linkage regions, 240 differentially expressed genes, 4651 differentially methylated genes and 49 genes as antipsychotic drug targets. To facilitate interpretation, we included various functional annotation data, especially brain eQTL, methylation QTL, brain expression featured in deep categorization of brain areas and developmental stages and brain-specific promoter and enhancer annotations. Furthermore, we conducted cross-study, cross-data type and integrative analyses of the multidimensional data deposited in SZGR 2.0, and made the data and results available through a user-friendly interface. In summary, SZGR 2.0 provides a one-stop shop of schizophrenia variants and genes and their function and regulation, providing an important resource in the schizophrenia and other mental disease community. SZGR 2.0 is available at https://bioinfo.uth.edu/SZGR/.

Integration of complex data sources to provide biologic insight into pulmonary vascular disease (2015 Grover Conference Series)

The application of complex data sources to pulmonary vascular diseases is an emerging and promising area of investigation. The use of -omics platforms, in silico modeling of gene networks, and linkage of large human cohorts with DNA biobanks are beginning to bear biologic insight into pulmonary hypertension. These approaches to high-throughput molecular phenotyping offer the possibility of discovering new therapeutic targets and identifying variability in response to therapy that can be leveraged to improve clinical care. Optimizing the methods for analyzing complex data sources and accruing large, well-phenotyped human cohorts linked to biologic data remain significant challenges. Here, we discuss two specific types of complex data sources-gene regulatory networks and DNA-linked electronic medical record cohorts-that illustrate the promise, challenges, and current limitations of these approaches to understanding and managing pulmonary vascular disease.

Identification of type 2 diabetes subgroups through topological analysis of patient similarity

Type 2 diabetes (T2D) is a heterogeneous complex disease affecting more than 29 million Americans alone with a rising prevalence trending toward steady increases in the coming decades. Thus, there is a pressing clinical need to improve early prevention and clinical management of T2D and its complications. Clinicians have understood that patients who carry the T2D diagnosis have a variety of phenotypes and susceptibilities to diabetes-related complications. We used a precision medicine approach to characterize the complexity of T2D patient populations based on high-dimensional electronic medical records (EMRs) and genotype data from 11,210 individuals. We successfully identified three distinct subgroups of T2D from topology-based patient-patient networks. Subtype 1 was characterized by T2D complications diabetic nephropathy and diabetic retinopathy; subtype 2 was enriched for cancer malignancy and cardiovascular diseases; and subtype 3 was associated most strongly with cardiovascular diseases, neurological diseases, allergies, and HIV infections. We performed a genetic association analysis of the emergent T2D subtypes to identify subtype-specific genetic markers and identified 1279, 1227, and 1338 single-nucleotide polymorphisms (SNPs) that mapped to 425, 322, and 437 unique genes specific to subtypes 1, 2, and 3, respectively. By assessing the human disease-SNP association for each subtype, the enriched phenotypes and biological functions at the gene level for each subtype matched with the disease comorbidities and clinical differences that we identified through EMRs. Our approach demonstrates the utility of applying the precision medicine paradigm in T2D and the promise of extending the approach to the study of other complex, multifactorial diseases.

An information model for computable cancer phenotypes

BACKGROUND

Standards, methods, and tools supporting the integration of clinical data and genomic information are an area of significant need and rapid growth in biomedical informatics. Integration of cancer clinical data and cancer genomic information poses unique challenges, because of the high volume and complexity of clinical data, as well as the heterogeneity and instability of cancer genome data when compared with germline data. Current information models of clinical and genomic data are not sufficiently expressive to represent individual observations and to aggregate those observations into longitudinal summaries over the course of cancer care. These models are acutely needed to support the development of systems and tools for generating the so called clinical "deep phenotype" of individual cancer patients, a process which remains almost entirely manual in cancer research and precision medicine.

METHODS

Reviews of existing ontologies and interviews with cancer researchers were used to inform iterative development of a cancer phenotype information model. We translated a subset of the Fast Healthcare Interoperability Resources (FHIR) models into the OWL 2 Description Logic (DL) representation, and added extensions as needed for modeling cancer phenotypes with terms derived from the NCI Thesaurus. Models were validated with domain experts and evaluated against competency questions.

RESULTS

The DeepPhe Information model represents cancer phenotype data at increasing levels of abstraction from mention level in clinical documents to summaries of key events and findings. We describe the model using breast cancer as an example, depicting methods to represent phenotypic features of cancers, tumors, treatment regimens, and specific biologic behaviors that span the entire course of a patient's disease.

CONCLUSIONS

We present a multi-scale information model for representing individual document mentions, document level classifications, episodes along a disease course, and phenotype summarization, linking individual observations to high-level summaries in support of subsequent integration and analysis.

Replicating Cardiovascular Condition-Birth Month Associations

Independent replication is vital for study findings drawn from Electronic Health Records (EHR). This replication study evaluates the relationship between seasonal effects at birth and lifetime cardiovascular condition risk. We performed a Season-wide Association Study on 1,169,599 patients from Mount Sinai Hospital (MSH) to compute phenome-wide associations between birth month and CVD. We then evaluated if seasonal patterns found at MSH matched those reported at Columbia University Medical Center. Coronary arteriosclerosis, essential hypertension, angina, and pre-infarction syndrome passed phenome-wide significance and their seasonal patterns matched those previously reported. Atrial fibrillation, cardiomyopathy, and chronic myocardial ischemia had consistent patterns but were not phenome-wide significant. We confirm that CVD risk peaks for those born in the late winter/early spring among the evaluated patient populations. The replication findings bolster evidence for a seasonal birth month effect in CVD. Further study is required to identify the environmental and developmental mechanisms.

Functional Analysis of Mouse G6pc1 Mutations Using a Novel In Situ Assay for Glucose-6-Phosphatase Activity and the Effect of Mutations in Conserved Human G6PC1/G6PC2 Amino Acids on G6PC2 Protein Expression

Elevated fasting blood glucose (FBG) has been associated with increased risk for development of type 2 diabetes. Single nucleotide polymorphisms (SNPs) in G6PC2 are the most important common determinants of variations in FBG in humans. Studies using G6pc2 knockout mice suggest that G6pc2 regulates the glucose sensitivity of insulin secretion. G6PC2 and the related G6PC1 and G6PC3 genes encode glucose-6-phosphatase catalytic subunits. This study describes a functional analysis of 22 non-synonymous G6PC2 SNPs, that alter amino acids that are conserved in human G6PC1, mouse G6pc1 and mouse G6pc2, with the goal of identifying variants that potentially affect G6PC2 activity/expression. Published data suggest strong conservation of catalytically important amino acids between all four proteins and the related G6PC3 isoform. Because human G6PC2 has very low glucose-6-phosphatase activity we used an indirect approach, examining the effect of these SNPs on mouse G6pc1 activity. Using a novel in situ functional assay for glucose-6-phosphatase activity we demonstrate that the amino acid changes associated with the human G6PC2 rs144254880 (Arg79Gln), rs149663725 (Gly114Arg) and rs2232326 (Ser324Pro) SNPs reduce mouse G6pc1 enzyme activity without affecting protein expression. The Arg79Gln variant alters an amino acid mutation of which, in G6PC1, has previously been shown to cause glycogen storage disease type 1a. We also demonstrate that the rs368382511 (Gly8Glu), rs138726309 (His177Tyr), rs2232323 (Tyr207Ser) rs374055555 (Arg293Trp), rs2232326 (Ser324Pro), rs137857125 (Pro313Leu) and rs2232327 (Pro340Leu) SNPs confer decreased G6PC2 protein expression. In summary, these studies identify multiple G6PC2 variants that have the potential to be associated with altered FBG in humans.