Epigenetic and proteomic signatures associate with clonal hematopoiesis expansion rate.

Clonal hematopoiesis of indeterminate potential (CHIP), whereby somatic mutations in hematopoietic stem cells confer a selective advantage and drive clonal expansion, not only correlates with age but also confers increased risk of morbidity and mortality. Here, we leverage genetically predicted traits to identify factors that determine CHIP clonal expansion rate. We used the passenger-approximated clonal expansion rate method to quantify the clonal expansion rate for 4,370 individuals in the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) cohort and calculated polygenic risk scores for DNA methylation aging, inflammation-related measures and circulating protein levels. Clonal expansion rate was significantly associated with both genetically predicted and measured epigenetic clocks. No associations were identified with inflammation-related lab values or diseases and CHIP expansion rate overall. A proteome-wide search identified predicted circulating levels of myeloid zinc finger 1 and anti-Müllerian hormone as associated with an increased CHIP clonal expansion rate and tissue inhibitor of metalloproteinase 1 and glycine N-methyltransferase as associated with decreased CHIP clonal expansion rate. Together, our findings identify epigenetic and proteomic patterns associated with the rate of hematopoietic clonal expansion.

High-throughput functional mapping of variants in an arrhythmia gene, KCNE1, reveals novel biology.

BACKGROUND: KCNE1 encodes a 129-residue cardiac potassium channel (IKs) subunit. KCNE1 variants are associated with long QT syndrome and atrial fibrillation. However, most variants have insufficient evidence of clinical consequences and thus limited clinical utility. METHODS: In this study, we leveraged the power of variant effect mapping, which couples saturation mutagenesis with high-throughput sequencing, to ascertain the function of thousands of protein-coding KCNE1 variants. RESULTS: We comprehensively assayed KCNE1 variant cell surface expression (2554/2709 possible single-amino-acid variants) and function (2534 variants). Our study identified 470 loss- or partial loss-of-surface expression and 574 loss- or partial loss-of-function variants. Of the 574 loss- or partial loss-of-function variants, 152 (26.5%) had reduced cell surface expression, indicating that most functionally deleterious variants affect channel gating. Nonsense variants at residues 56-104 generally had WT-like trafficking scores but decreased functional scores, indicating that the latter half of the protein is dispensable for protein trafficking but essential for channel function. 22 of the 30 KCNE1 residues (73%) highly intolerant of variation (with > 70% loss-of-function variants) were in predicted close contact with binding partners KCNQ1 or calmodulin. Our functional assay data were consistent with gold standard electrophysiological data (ρ = - 0.64), population and patient cohorts (32/38 presumed benign or pathogenic variants with consistent scores), and computational predictors (ρ = - 0.62). Our data provide moderate-strength evidence for the American College of Medical Genetics/Association of Molecular Pathology functional criteria for benign and pathogenic variants. CONCLUSIONS: Comprehensive variant effect maps of KCNE1 can both provide insight into I Ks channel biology and help reclassify variants of uncertain significance.

Overcome the Limitation of Phenome-Wide Association Studies (PheWAS): Extension of PheWAS to Efficient and Robust Large-Scale ICD Codes Analysis.

The Phenome-wide association studies (PheWAS) have become widely used for efficient, high-throughput evaluation of relationship between a genetic factor and a large number of disease phenotypes, typically extracted from a DNA biobank linked with electronic medical records (EMR). Phecodes, billing code-derived disease case-control status, are usually used as outcome variables in PheWAS and logistic regression has been the standard choice of analysis method. Since the clinical diagnoses in EMR are often inaccurate with errors which can lead to biases in the odds ratio estimates, much effort has been put to accurately define the cases and controls to ensure an accurate analysis. Specifically in order to correctly classify controls in the population, an exclusion criteria list for each Phecode was manually compiled to obtain unbiased odds ratios. However, the accuracy of the list cannot be guaranteed without extensive data curation process. The costly curation process limits the efficiency of large-scale analyses that take full advantage of all structured phenotypic information available in EMR. Here, we proposed to estimate relative risks (RR) instead. We first demonstrated the desired nature of RR that overcomes the inaccuracy in the controls via theoretical formula. With simulation and real data application, we further confirmed that RR is unbiased without compiling exclusion criteria lists. With RR as estimates, we are able to efficiently extend PheWAS to a larger-scale, phenome construction agnostic analysis of phenotypes, using ICD 9/10 codes, which preserve much more disease-related clinical information than Phecodes.

Determinants of mosaic chromosomal alteration fitness.

Clonal hematopoiesis (CH) is characterized by the acquisition of a somatic mutation in a hematopoietic stem cell that results in a clonal expansion. These driver mutations can be single nucleotide variants in cancer driver genes or larger structural rearrangements called mosaic chromosomal alterations (mCAs). The factors that influence the variations in mCA fitness and ultimately result in different clonal expansion rates are not well understood. We used the Passenger-Approximated Clonal Expansion Rate (PACER) method to estimate clonal expansion rate as PACER scores for 6,381 individuals in the NHLBI TOPMed cohort with gain, loss, and copy-neutral loss of heterozygosity mCAs. Our mCA fitness estimates, derived by aggregating per-individual PACER scores, were correlated (R2 = 0.49) with an alternative approach that estimated fitness of mCAs in the UK Biobank using population-level distributions of clonal fraction. Among individuals with JAK2 V617F clonal hematopoiesis of indeterminate potential or mCAs affecting the JAK2 gene on chromosome 9, PACER score was strongly correlated with erythrocyte count. In a cross-sectional analysis, genome-wide association study of estimates of mCA expansion rate identified a TCL1A locus variant associated with mCA clonal expansion rate, with suggestive variants in NRIP1 and TERT.

Meta-Analysis of Genome-Wide Association Studies Reveals Genetic Mechanisms of Supraventricular Arrhythmias.

BACKGROUND: Substantial data support a heritable basis for supraventricular tachycardias, but the genetic determinants and molecular mechanisms of these arrhythmias are poorly understood. We sought to identify genetic loci associated with atrioventricular nodal reentrant tachycardia (AVNRT) and atrioventricular accessory pathways or atrioventricular reciprocating tachycardia (AVAPs/AVRT). METHODS: We performed multiancestry meta-analyses of genome-wide association studies to identify genetic loci for AVNRT (4 studies) and AVAP/AVRT (7 studies). We assessed evidence supporting the potential causal effects of candidate genes by analyzing relations between associated variants and cardiac gene expression, performing transcriptome-wide analyses, and examining prior genome-wide association studies. RESULTS: Analyses comprised 2384 AVNRT cases and 106 489 referents, and 2811 AVAP/AVRT cases and 1,483 093 referents. We identified 2 significant loci for AVNRT, which implicates NKX2-5 and TTN as disease susceptibility genes. A transcriptome-wide association analysis supported an association between reduced predicted cardiac expression of NKX2-5 and AVNRT. We identified 3 significant loci for AVAP/AVRT, which implicates SCN5A, SCN10A, and TTN/CCDC141. Variant associations at several loci have been previously reported for cardiac phenotypes, including atrial fibrillation, stroke, Brugada syndrome, and electrocardiographic intervals. CONCLUSIONS: Our findings highlight gene regions associated with ion channel function (AVAP/AVRT), as well as cardiac development and the sarcomere (AVAP/AVRT and AVNRT) as important potential effectors of supraventricular tachycardia susceptibility.

Large language models facilitate the generation of electronic health record phenotyping algorithms.

OBJECTIVES: Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts. MATERIALS AND METHODS: We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (ie, type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network. RESULTS: GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values). CONCLUSION: GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

Impact of Insurance Status and Region on Angiotensin Receptor-Neprilysin Inhibitor Prescription During Heart Failure Hospitalizations.

BACKGROUND: An angiotensin receptor-neprilysin inhibitor (ARNI) is the preferred renin-angiotensin system (RAS) inhibitor for heart failure with reduced ejection fraction (HFrEF). Among eligible patients, insurance status and prescriber concern regarding out-of-pocket costs may constrain early initiation of ARNI and other new therapies. OBJECTIVES: In this study, the authors sought to evaluate the association of insurance and other social determinants of health with ARNI initiation at discharge from HFrEF hospitalization. METHODS: The authors analyzed ARNI initiation from January 2017 to June 2020 among patients with HFrEF eligible to receive RAS inhibitor at discharge from hospitals in the Get With The Guidelines-Heart Failure registry. The primary outcome was the proportion of ARNI prescription at discharge among those prescribed RAS inhibitor who were not on ARNI on admission. A logistic regression model was used to determine the association of insurance status, U.S. region, and their interaction, as well as self-reported race, with ARNI initiation at discharge. RESULTS: From 42,766 admissions, 24,904 were excluded for absolute or relative contraindications to RAS inhibitors. RAS inhibitors were prescribed for 16,817 (94.2%) of remaining discharges, for which ARNI was prescribed in 1,640 (9.8%). Self-reported Black patients were less likely to be initiated on ARNI compared to self-reported White patients (OR: 0.64; 95% CI: 0.50-0.81). Compared to Medicare beneficiaries, patients with third-party insurance, Medicaid, or no insurance were less likely to be initiated on ARNI (OR: 0.47 [95% CI: 0.31-0.72], OR: 0.41 [95% CI: 0.25-0.67], and OR: 0.20 [95% CI: 0.08-0.47], respectively). ARNI therapy varied by hospital region, with lowest utilization in the Mountain region. An interaction was demonstrated between the impact of insurance disparities and hospital region. CONCLUSIONS: Among patients hospitalized between 2017 and 2020 for HFrEF who were prescribed RAS inhibitor therapy at discharge, insurance status, geographic region, and self-reported race were associated with ARNI initiation.

Interoperability of phenome-wide multimorbidity patterns: a comparative study of two large-scale EHR systems.

Background: Electronic health records (EHR) are increasingly used for studying multimorbidities. However, concerns about accuracy, completeness, and EHRs being primarily designed for billing and administrative purposes raise questions about the consistency and reproducibility of EHR-based multimorbidity research. Methods: Utilizing phecodes to represent the disease phenome, we analyzed pairwise comorbidity strengths using a dual logistic regression approach and constructed multimorbidity as an undirected weighted graph. We assessed the consistency of the multimorbidity networks within and between two major EHR systems at local (nodes and edges), meso (neighboring patterns), and global (network statistics) scales. We present case studies to identify disease clusters and uncover clinically interpretable disease relationships. We provide an interactive web tool and a knowledge base combining data from multiple sources for online multimorbidity analysis. Findings: Analyzing data from 500,000 patients across Vanderbilt University Medical Center and Mass General Brigham health systems, we observed a strong correlation in disease frequencies ( Kendall's τ = 0.643) and comorbidity strengths (Pearson ρ = 0.79). Consistent network statistics across EHRs suggest similar structures of multimorbidity networks at various scales. Comorbidity strengths and similarities of multimorbidity connection patterns align with the disease genetic correlations. Graph-theoretic analyses revealed a consistent core-periphery structure, implying efficient network clustering through threshold graph construction. Using hydronephrosis as a case study, we demonstrated the network's ability to uncover clinically relevant disease relationships and provide novel insights. Interpretation: Our findings demonstrate the robustness of large-scale EHR data for studying phenome-wide multimorbidities. The alignment of multimorbidity patterns with genetic data suggests the potential utility for uncovering shared biology of diseases. The consistent core-periphery structure offers analytical insights to discover complex disease interactions. This work also sets the stage for advanced disease modeling, with implications for precision medicine. Funding: VUMC Biostatistics Development Award, the National Institutes of Health, and the VA CSRD.

Clinical associations with a polygenic predisposition to benign lower white blood cell counts.

Polygenic variation unrelated to disease contributes to interindividual variation in baseline white blood cell (WBC) counts, but its clinical significance is uncharacterized. We investigated the clinical consequences of a genetic predisposition toward lower WBC counts among 89,559 biobank participants from tertiary care centers using a polygenic score for WBC count (PGSWBC) comprising single nucleotide polymorphisms not associated with disease. A predisposition to lower WBC counts was associated with a decreased risk of identifying pathology on a bone marrow biopsy performed for a low WBC count (odds-ratio = 0.55 per standard deviation increase in PGSWBC [95%CI, 0.30-0.94], p = 0.04), an increased risk of leukopenia (a low WBC count) when treated with a chemotherapeutic (n = 1724, hazard ratio [HR] = 0.78 [0.69-0.88], p = 4.0 × 10-5) or immunosuppressant (n = 354, HR = 0.61 [0.38-0.99], p = 0.04). A predisposition to benign lower WBC counts was associated with an increased risk of discontinuing azathioprine treatment (n = 1,466, HR = 0.62 [0.44-0.87], p = 0.006). Collectively, these findings suggest that there are genetically predisposed individuals who are susceptible to escalations or alterations in clinical care that may be harmful or of little benefit.

Physical Activity and Incident Obesity Across the Spectrum of Genetic Risk for Obesity.

Importance: Despite consistent public health recommendations, obesity rates in the US continue to increase. Physical activity recommendations do not account for individual genetic variability, increasing risk of obesity. Objective: To use activity, clinical, and genetic data from the All of Us Research Program (AoURP) to explore the association of genetic risk of higher body mass index (BMI) with the level of physical activity needed to reduce incident obesity. Design, Setting, and Participants: In this US population-based retrospective cohort study, participants were enrolled in the AoURP between May 1, 2018, and July 1, 2022. Enrollees in the AoURP who were of European ancestry, owned a personal activity tracking device, and did not have obesity up to 6 months into activity tracking were included in the analysis. Exposure: Physical activity expressed as daily step counts and a polygenic risk score (PRS) for BMI, calculated as weight in kilograms divided by height in meters squared. Main Outcome and Measures: Incident obesity (BMI ≥30). Results: A total of 3124 participants met inclusion criteria. Among 3051 participants with available data, 2216 (73%) were women, and the median age was 52.7 (IQR, 36.4-62.8) years. The total cohort of 3124 participants walked a median of 8326 (IQR, 6499-10 389) steps/d over a median of 5.4 (IQR, 3.4-7.0) years of personal activity tracking. The incidence of obesity over the study period increased from 13% (101 of 781) to 43% (335 of 781) in the lowest and highest PRS quartiles, respectively (P = 1.0 × 10-20). The BMI PRS demonstrated an 81% increase in obesity risk (P = 3.57 × 10-20) while mean step count demonstrated a 43% reduction (P = 5.30 × 10-12) when comparing the 75th and 25th percentiles, respectively. Individuals with a PRS in the 75th percentile would need to walk a mean of 2280 (95% CI, 1680-3310) more steps per day (11 020 total) than those at the 50th percentile to have a comparable risk of obesity. To have a comparable risk of obesity to individuals at the 25th percentile of PRS, those at the 75th percentile with a baseline BMI of 22 would need to walk an additional 3460 steps/d; with a baseline BMI of 24, an additional 4430 steps/d; with a baseline BMI of 26, an additional 5380 steps/d; and with a baseline BMI of 28, an additional 6350 steps/d. Conclusions and Relevance: In this cohort study, the association between daily step count and obesity risk across genetic background and baseline BMI were quantified. Population-based recommendations may underestimate physical activity needed to prevent obesity among those at high genetic risk.

Machine Learning Prediction of Treatment Response to Inhaled Corticosteroids in Asthma.

BACKGROUND: Although inhaled corticosteroids (ICS) are the first-line therapy for patients with persistent asthma, many patients continue to have exacerbations. We developed machine learning models to predict the ICS response in patients with asthma. METHODS: The subjects included asthma patients of European ancestry (n = 1371; 448 children; 916 adults). A genome-wide association study was performed to identify the SNPs associated with ICS response. Using the SNPs identified, two machine learning models were developed to predict ICS response: (1) least absolute shrinkage and selection operator (LASSO) regression and (2) random forest. RESULTS: The LASSO regression model achieved an AUC of 0.71 (95% CI 0.67-0.76; sensitivity: 0.57; specificity: 0.75) in an independent test cohort, and the random forest model achieved an AUC of 0.74 (95% CI 0.70-0.78; sensitivity: 0.70; specificity: 0.68). The genes contributing to the prediction of ICS response included those associated with ICS responses in asthma (TPSAB1, FBXL16), asthma symptoms and severity (ABCA7, CNN2, PTRN3, and BSG/CD147), airway remodeling (ELANE, FSTL3), mucin production (GAL3ST), leukotriene synthesis (GPX4), allergic asthma (ZFPM1, SBNO2), and others. CONCLUSIONS: An accurate risk prediction of ICS response can be obtained using machine learning methods, with the potential to inform personalized treatment decisions. Further studies are needed to examine if the integration of richer phenotype data could improve risk prediction.

Recurrence After Atrial Fibrillation Ablation and Investigational Biomarkers of Cardiac Remodeling.

BACKGROUND: Recurrence after atrial fibrillation (AF) ablation remains common. We evaluated the association between recurrence and levels of biomarkers of cardiac remodeling, and their ability to improve recurrence prediction when added to a clinical prediction model. METHODS AND RESULTS: Blood samples collected before de novo catheter ablation were analyzed. Levels of bone morphogenetic protein-10, angiopoietin-2, fibroblast growth factor-23, insulin-like growth factor-binding protein-7, myosin-binding protein C3, growth differentiation factor-15, interleukin-6, N-terminal pro-brain natriuretic peptide, and high-sensitivity troponin T were measured. Recurrence was defined as ≥30 seconds of an atrial arrhythmia 3 to 12 months postablation. Multivariable logistic regression was performed using biomarker levels along with clinical covariates: APPLE score (Age >65 years, Persistent AF, imPaired eGFR [<60 ml/min/1.73m2], LA diameter ≥43 mm, EF <50%; which includes age, left atrial diameter, left ventricular ejection fraction, persistent atrial fibrillation, and estimated glomerular filtration rate), preablation rhythm, sex, height, body mass index, presence of an implanted continuous monitor, year of ablation, and additional linear ablation. A total of 1873 participants were included. A multivariable logistic regression showed an association between recurrence and levels of angiopoietin-2 (odds ratio, 1.08 [95% CI, 1.02-1.15], P=0.007) and interleukin-6 (odds ratio, 1.02 [95% CI, 1.003-1.03]; P=0.02). The area under the receiver operating characteristic curve of a model that only contained clinical predictors was 0.711. The addition of any of the 9 studied biomarkers to the predictive model did not result in a statistically significant improvement in the area under the receiver operating characteristic curve. CONCLUSIONS: Higher angiopoietin-2 and interleukin-6 levels were associated with recurrence after atrial fibrillation ablation in multivariable modeling. However, the addition of biomarkers to a clinical prediction model did not significantly improve recurrence prediction.

The electrophysiologic effects of KCNQ1 extend beyond expression of IKs: evidence from genetic and pharmacologic block.

AIMS: While variants in KCNQ1 are the commonest cause of the congenital long QT syndrome, we and others find only a small IKs in cardiomyocytes from human-induced pluripotent stem cells (iPSC-CMs) or human ventricular myocytes. METHODS AND RESULTS: We studied population control iPSC-CMs and iPSC-CMs from a patient with Jervell and Lange-Nielsen (JLN) syndrome due to compound heterozygous loss-of-function (LOF) KCNQ1 variants. We compared the effects of pharmacologic IKs block to those of genetic KCNQ1 ablation, using JLN cells, cells homozygous for the KCNQ1 LOF allele G643S, or siRNAs reducing KCNQ1 expression. We also studied the effects of two blockers of IKr, the other major cardiac repolarizing current, in the setting of pharmacologic or genetic ablation of KCNQ1: moxifloxacin, associated with a very low risk of drug-induced long QT, and dofetilide, a high-risk drug. In control cells, a small IKs was readily recorded but the pharmacologic IKs block produced no change in action potential duration at 90% repolarization (APD90). In contrast, in cells with genetic ablation of KCNQ1 (JLN), baseline APD90 was markedly prolonged compared with control cells (469 ± 20 vs. 310 ± 16 ms). JLN cells displayed increased sensitivity to acute IKr block: the concentration (µM) of moxifloxacin required to prolong APD90 100 msec was 237.4 [median, interquartile range (IQR) 100.6-391.6, n = 7] in population cells vs. 23.7 (17.3-28.7, n = 11) in JLN cells. In control cells, chronic moxifloxacin exposure (300 µM) mildly prolonged APD90 (10%) and increased IKs, while chronic exposure to dofetilide (5 nM) produced greater prolongation (67%) and no increase in IKs. However, in the siRNA-treated cells, moxifloxacin did not increase IKs and markedly prolonged APD90. CONCLUSION: Our data strongly suggest that KCNQ1 expression modulates baseline cardiac repolarization, and the response to IKr block, through mechanisms beyond simply generating IKs.

Leveraging generative AI to prioritize drug repurposing candidates for Alzheimer's disease with real-world clinical validation.

Drug repurposing represents an attractive alternative to the costly and time-consuming process of new drug development, particularly for serious, widespread conditions with limited effective treatments, such as Alzheimer's disease (AD). Emerging generative artificial intelligence (GAI) technologies like ChatGPT offer the promise of expediting the review and summary of scientific knowledge. To examine the feasibility of using GAI for identifying drug repurposing candidates, we iteratively tasked ChatGPT with proposing the twenty most promising drugs for repurposing in AD, and tested the top ten for risk of incident AD in exposed and unexposed individuals over age 65 in two large clinical datasets: (1) Vanderbilt University Medical Center and (2) the All of Us Research Program. Among the candidates suggested by ChatGPT, metformin, simvastatin, and losartan were associated with lower AD risk in meta-analysis. These findings suggest GAI technologies can assimilate scientific insights from an extensive Internet-based search space, helping to prioritize drug repurposing candidates and facilitate the treatment of diseases.

Multifocal Ectopic Purkinje Premature Contractions due to neutralization of an SCN5A negative charge: structural insights into the gating pore hypothesis.

Background: We identified a novel SCN5A variant, E171Q, in a neonate with very frequent ectopy and reduced ejection fraction which normalized after arrhythmia suppression by flecainide. This clinical picture is consistent with multifocal ectopic Purkinje-related premature contractions (MEPPC). Most previous reports of MEPPC have implicated SCN5A variants such as R222Q that neutralize positive charges in the S4 voltage sensor helix of the channel protein NaV1.5 and generate a gating pore current. Methods and Results: E171 is a highly conserved negatively-charged residue located in the S2 transmembrane helix of NaV1.5 domain I. E171 is a key component of the Gating Charge Transfer Center, a region thought to be critical for normal movement of the S4 voltage sensor helix. We used heterologous expression, CRISPR-edited induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs), and molecular dynamics simulations to demonstrate that E171Q generates a gating pore current, which was suppressed by a low concentration of flecainide (IC50 = 0.71±0.07 µM). R222Q shifts voltage dependence of activation and inactivation in a negative direction but we observed positive shifts with E171Q. E171Q iPSC-CMs demonstrated abnormal spontaneous activity and prolonged action potentials. Molecular dynamics simulations revealed that both R222Q and E171Q proteins generate a water-filled permeation pathway that underlies generation of the gating pore current. Conclusion: Previously identified MEPPC-associated variants that create gating pore currents are located in positively-charged residues in the S4 voltage sensor and generate negative shifts in the voltage dependence of activation and inactivation. We demonstrate that neutralizing a negatively charged S2 helix residue in the Gating Charge Transfer Center generates positive shifts but also create a gating pore pathway. These findings implicate the gating pore pathway as the primary functional and structural determinant of MEPPC and widen the spectrum of variants that are associated with gating pore-related disease in voltage-gated ion channels.

Prognostic Value of Multiplexed Assays of Variant Effect and Automated Patch-clamping for KCNH2-LQTS Risk Stratification.

Background: Long QT syndrome (LQTS) is a lethal arrhythmia condition, frequently caused by rare loss-of-function variants in the cardiac potassium channel encoded by KCNH2. Variant-based risk stratification is complicated by heterogenous clinical data, incomplete penetrance, and low-throughput functional data. Objective: To test the utility of variant-specific features, including high-throughput functional data, to predict cardiac events among KCNH2 variant heterozygotes. Methods: We quantified cell-surface trafficking of 18,323 variants in KCNH2 and recorded potassium current densities for 506 KCNH2 variants. Next, we deeply phenotyped 1150 KCNH2 missense variant patients, including ECG features, cardiac event history (528 total cardiac events), and mortality. We then assessed variant functional, in silico, structural, and LQTS penetrance data to stratify event-free survival for cardiac events in the study cohort. Results: Variant-specific current density (HR 0.28 [0.13-0.60]) and estimates of LQTS penetrance incorporating MAVE data (HR 3.16 [1.59-6.27]) were independently predictive of severe cardiac events when controlling for patient-specific features. Risk prediction models incorporating these data significantly improved prediction of 20 year cardiac events (AUC 0.79 [0.75-0.82]) over patient-only covariates (QTc and sex) (AUC 0.73 [0.70-0.77]). Conclusion: We show that high-throughput functional data, and other variant-specific features, meaningfully contribute to both diagnosis and prognosis of a clinically actionable monogenic disease.

Unbiased characterization of atrial fibrillation phenotypic architecture provides insight to genetic liability and clinically relevant outcomes.

Background: Atrial Fibrillation (AF) is a common and clinically heterogeneous arrythmia. Machine learning (ML) algorithms can define data-driven disease subtypes in an unbiased fashion, but whether the AF subgroups defined in this way align with underlying mechanisms, such as high polygenic liability to AF or inflammation, and associate with clinical outcomes is unclear. Methods: We identified individuals with AF in a large biobank linked to electronic health records (EHR) and genome-wide genotyping. The phenotypic architecture in the AF cohort was defined using principal component analysis of 35 expertly curated and uncorrelated clinical features. We applied an unsupervised co-clustering machine learning algorithm to the 35 features to identify distinct phenotypic AF clusters. The clinical inflammatory status of the clusters was defined using measured biomarkers (CRP, ESR, WBC, Neutrophil %, Platelet count, RDW) within 6 months of first AF mention in the EHR. Polygenic risk scores (PRS) for AF and cytokine levels were used to assess genetic liability of clusters to AF and inflammation, respectively. Clinical outcomes were collected from EHR up to the last medical contact. Results: The analysis included 23,271 subjects with AF, of which 6,023 had available genome-wide genotyping. The machine learning algorithm identified 3 phenotypic clusters that were distinguished by increasing prevalence of comorbidities, particularly renal dysfunction, and coronary artery disease. Polygenic liability to AF across clusters was highest in the low comorbidity cluster. Clinically measured inflammatory biomarkers were highest in the high comorbid cluster, while there was no difference between groups in genetically predicted levels of inflammatory biomarkers. Subgroup assignment was associated with multiple clinical outcomes including mortality, stroke, bleeding, and use of cardiac implantable electronic devices after AF diagnosis. Conclusion: Patient subgroups identified by unsupervised clustering were distinguished by comorbidity burden and associated with risk of clinically important outcomes. Polygenic liability to AF across clusters was greatest in the low comorbidity subgroup. Clinical inflammation, as reflected by measured biomarkers, was lowest in the subgroup with lowest comorbidities. However, there were no differences in genetically predicted levels of inflammatory biomarkers, suggesting associations between AF and inflammation is driven by acquired comorbidities rather than genetic predisposition.

Large Language Models Facilitate the Generation of Electronic Health Record Phenotyping Algorithms.

Objectives: Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts. Materials and Methods: We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (i.e., type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network. Results: GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values). Conclusion: GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

The impact of damaging epilepsy and cardiac genetic variant burden in sudden death in the young.

BACKGROUND: Sudden unexpected death in children is a tragic event. Understanding the genetics of sudden death in the young (SDY) enables family counseling and cascade screening. The objective of this study was to characterize genetic variation in an SDY cohort using whole genome sequencing. METHODS: The SDY Case Registry is a National Institutes of Health/Centers for Disease Control and Prevention surveillance effort to discern the prevalence, causes, and risk factors for SDY. The SDY Case Registry prospectively collected clinical data and DNA biospecimens from SDY cases < 20 years of age. SDY cases were collected from medical examiner and coroner offices spanning 13 US jurisdictions from 2015 to 2019. The cohort included 211 children (median age 0.33 year; range 0-20 years), determined to have died suddenly and unexpectedly and from whom DNA biospecimens for DNA extractions and next-of-kin consent were ascertained. A control cohort consisted of 211 randomly sampled, sex- and ancestry-matched individuals from the 1000 Genomes Project. Genetic variation was evaluated in epilepsy, cardiomyopathy, and arrhythmia genes in the SDY and control cohorts. American College of Medical Genetics/Genomics guidelines were used to classify variants as pathogenic or likely pathogenic. Additionally, pathogenic and likely pathogenic genetic variation was identified using a Bayesian-based artificial intelligence (AI) tool. RESULTS: The SDY cohort was 43% European, 29% African, 3% Asian, 16% Hispanic, and 9% with mixed ancestries and 39% female. Six percent of the cohort was found to harbor a pathogenic or likely pathogenic genetic variant in an epilepsy, cardiomyopathy, or arrhythmia gene. The genomes of SDY cases, but not controls, were enriched for rare, potentially damaging variants in epilepsy, cardiomyopathy, and arrhythmia-related genes. A greater number of rare epilepsy genetic variants correlated with younger age at death. CONCLUSIONS: While damaging cardiomyopathy and arrhythmia genes are recognized contributors to SDY, we also observed an enrichment in epilepsy-related genes in the SDY cohort and a correlation between rare epilepsy variation and younger age at death. These findings emphasize the importance of considering epilepsy genes when evaluating SDY.