One-year real-world experience with mavacamten and its physiologic effects on obstructive hypertrophic cardiomyopathy.

Mavacamten is a first-in-class cardiac myosin ATPase inhibitor, approved by the United States Food and Drug Administration for the treatment of hypertrophic cardiomyopathy with obstructive physiology (oHCM). Here, we present the real-world use of mavacamten in 50 patients with oHCM at a tertiary care referral center. In both our highlighted case and in our aggregate data, we report significant improvement in wall thickness, mitral regurgitation, left ventricular outflow tract obstruction and New York Heart Association symptom class. Moreover, in our center's experience, neither arrhythmia burden, nor contractility have worsened in the vast majority of patients: we note a clinically insignificant mean decrease in left ventricular ejection fraction (LVEF), with only two patients requiring temporary mavacamten discontinuance for LVEF < 50%. Adverse events were rare, unrelated to mavacamten itself, and seen solely in patients with disease too advanced to have been represented in clinical trials. Moreover, our multidisciplinary pathway enabled us to provide a large number of patients with a novel closely-monitored therapeutic within just a few months of commercial availability. These data lead us to conclude that mavacamten, as a first-in-class cardiac myosin inhibitor, is safe and efficacious in real-world settings.

Disease prediction with multi-omics and biomarkers empowers case-control genetic discoveries in the UK Biobank.

The emergence of biobank-level datasets offers new opportunities to discover novel biomarkers and develop predictive algorithms for human disease. Here, we present an ensemble machine-learning framework (machine learning with phenotype associations, MILTON) utilizing a range of biomarkers to predict 3,213 diseases in the UK Biobank. Leveraging the UK Biobank's longitudinal health record data, MILTON predicts incident disease cases undiagnosed at time of recruitment, largely outperforming available polygenic risk scores. We further demonstrate the utility of MILTON in augmenting genetic association analyses in a phenome-wide association study of 484,230 genome-sequenced samples, along with 46,327 samples with matched plasma proteomics data. This resulted in improved signals for 88 known (P < 1 × 10-8) gene-disease relationships alongside 182 gene-disease relationships that did not achieve genome-wide significance in the nonaugmented baseline cohorts. We validated these discoveries in the FinnGen biobank alongside two orthogonal machine-learning methods built for gene-disease prioritization. All extracted gene-disease associations and incident disease predictive biomarkers are publicly available ( http://milton.public.cgr.astrazeneca.com ).

Charting the Molecular Terrain of Exercise: The Power of Multi-Omic Mapping.

Physical activity plays a fundamental role in human health and disease. Exercise has been shown to improve a wide variety of disease states, and the scientific community is committed to understanding the precise molecular mechanisms that underlie the exquisite benefits. This review provides an overview of molecular responses to acute exercise and chronic training, particularly energy mobilization and generation, structural adaptation, inflammation, and immune regulation. Further it offers a detailed discussion on known molecular signals and systemic regulators activated during various forms of exercise and their role in orchestrating health benefits. Critically, the increasing use of multi-omic technologies is explored with an emphasis on how multi-omic and multi-tissue studies contribute to a more profound understanding of exercise biology. These data inform anticipated future advancement in the field and highlight the prospect of integrating exercise with pharmacology for personalized disease prevention and treatment.

Deciphering glial contributions to CSF1R-related disorder via single-nuclear transcriptomic profiling: a case study.

CSF1R-related disorder (CSF1R-RD) is a neurodegenerative condition that predominantly affects white matter due to genetic alterations in the CSF1R gene, which is expressed by microglia. We studied an elderly man with a hereditary, progressive dementing disorder of unclear etiology. Standard genetic testing for leukodystrophy and other neurodegenerative conditions was negative. Brain autopsy revealed classic features of adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP), including confluent white matter degeneration with axonal spheroids and pigmented glial cells in the affected white matter, consistent with CSF1R-RD. Subsequent long-read sequencing identified a novel deletion in CSF1R that was not detectable with short-read exome sequencing. To gain insight into potential mechanisms underlying white matter degeneration in CSF1R-RD, we studied multiple brain regions exhibiting varying degrees of white matter pathology. We found decreased CSF1R transcript and protein across brain regions, including intact white matter. Single nuclear RNA sequencing (snRNAseq) identified two disease-associated microglial cell states: lipid-laden microglia (expressing GPNMB, ATG7, LGALS1, LGALS3) and inflammatory microglia (expressing IL2RA, ATP2C1, FCGBP, VSIR, SESN3), along with a small population of CD44+ peripheral monocyte-derived macrophages exhibiting migratory and phagocytic signatures. GPNMB+ lipid-laden microglia with ameboid morphology represented the end-stage disease microglia state. Disease-associated oligodendrocytes exhibited cell stress signatures and dysregulated apoptosis-related genes. Disease-associated oligodendrocyte precursor cells (OPCs) displayed a failure in their differentiation into mature myelin-forming oligodendrocytes, as evidenced by upregulated LRP1, PDGFRA, SOX5, NFIA, and downregulated NKX2-2, NKX6.2, SOX4, SOX8, TCF7L2, YY1, ZNF488. Overall, our findings highlight microglia-oligodendroglia crosstalk in demyelination, with CSF1R dysfunction promoting phagocytic and inflammatory microglia states, an arrest in OPC differentiation, and oligodendrocyte depletion.

Local read haplotagging enables accurate long-read small variant calling.

Long-read sequencing technology has enabled variant detection in difficult-to-map regions of the genome and enabled rapid genetic diagnosis in clinical settings. Rapidly evolving third-generation sequencing platforms like Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) are introducing newer platforms and data types. It has been demonstrated that variant calling methods based on deep neural networks can use local haplotyping information with long-reads to improve the genotyping accuracy. However, using local haplotype information creates an overhead as variant calling needs to be performed multiple times which ultimately makes it difficult to extend to new data types and platforms as they get introduced. In this work, we have developed a local haplotype approximate method that enables state-of-the-art variant calling performance with multiple sequencing platforms including PacBio Revio system, ONT R10.4 simplex and duplex data. This addition of local haplotype approximation simplifies long-read variant calling with DeepVariant.

Simple models vs. deep learning in detecting low ejection fraction from the electrocardiogram.

Aims: Deep learning methods have recently gained success in detecting left ventricular systolic dysfunction (LVSD) from electrocardiogram (ECG) waveforms. Despite their high level of accuracy, they are difficult to interpret and deploy broadly in the clinical setting. In this study, we set out to determine whether simpler models based on standard ECG measurements could detect LVSD with similar accuracy to that of deep learning models. Methods and results: Using an observational data set of 40 994 matched 12-lead ECGs and transthoracic echocardiograms, we trained a range of models with increasing complexity to detect LVSD based on ECG waveforms and derived measurements. The training data were acquired from the Stanford University Medical Center. External validation data were acquired from the Columbia Medical Center and the UK Biobank. The Stanford data set consisted of 40 994 matched ECGs and echocardiograms, of which 9.72% had LVSD. A random forest model using 555 discrete, automated measurements achieved an area under the receiver operator characteristic curve (AUC) of 0.92 (0.91-0.93), similar to a deep learning waveform model with an AUC of 0.94 (0.93-0.94). A logistic regression model based on five measurements achieved high performance [AUC of 0.86 (0.85-0.87)], close to a deep learning model and better than N-terminal prohormone brain natriuretic peptide (NT-proBNP). Finally, we found that simpler models were more portable across sites, with experiments at two independent, external sites. Conclusion: Our study demonstrates the value of simple electrocardiographic models that perform nearly as well as deep learning models, while being much easier to implement and interpret.

Loss of function of FAM177A1, a Golgi complex localized protein, causes a novel neurodevelopmental disorder.

PURPOSE: The function of FAM177A1 and its relationship to human disease is largely unknown. Recent studies have demonstrated FAM177A1 to be a critical immune-associated gene. One previous case study has linked FAM177A1 to a neurodevelopmental disorder in 4 siblings. METHODS: We identified 5 individuals from 3 unrelated families with biallelic variants in FAM177A1. The physiological function of FAM177A1 was studied in a zebrafish model organism and human cell lines with loss-of-function variants similar to the affected cohort. RESULTS: These individuals share a characteristic phenotype defined by macrocephaly, global developmental delay, intellectual disability, seizures, behavioral abnormalities, hypotonia, and gait disturbance. We show that FAM177A1 localizes to the Golgi complex in mammalian and zebrafish cells. Intersection of the RNA sequencing and metabolomic data sets from FAM177A1-deficient human fibroblasts and whole zebrafish larvae demonstrated dysregulation of pathways associated with apoptosis, inflammation, and negative regulation of cell proliferation. CONCLUSION: Our data shed light on the emerging function of FAM177A1 and defines FAM177A1-related neurodevelopmental disorder as a new clinical entity.

The mitochondrial multi-omic response to exercise training across rat tissues.

Mitochondria have diverse functions critical to whole-body metabolic homeostasis. Endurance training alters mitochondrial activity, but systematic characterization of these adaptations is lacking. Here, the Molecular Transducers of Physical Activity Consortium mapped the temporal, multi-omic changes in mitochondrial analytes across 19 tissues in male and female rats trained for 1, 2, 4, or 8 weeks. Training elicited substantial changes in the adrenal gland, brown adipose, colon, heart, and skeletal muscle. The colon showed non-linear response dynamics, whereas mitochondrial pathways were downregulated in brown adipose and adrenal tissues. Protein acetylation increased in the liver, with a shift in lipid metabolism, whereas oxidative proteins increased in striated muscles. Exercise-upregulated networks were downregulated in human diabetes and cirrhosis. Knockdown of the central network protein 17-beta-hydroxysteroid dehydrogenase 10 (HSD17B10) elevated oxygen consumption, indicative of metabolic stress. We provide a multi-omic, multi-tissue, temporal atlas of the mitochondrial response to exercise training and identify candidates linked to mitochondrial dysfunction.

Prediction of diagnosis and diastolic filling pressure by AI-enhanced cardiac MRI: a modelling study of hospital data.

BACKGROUND: With increasing numbers of patients and novel drugs for distinct causes of systolic and diastolic heart failure, automated assessment of cardiac function is important. We aimed to provide a non-invasive method to predict diagnosis of patients undergoing cardiac MRI (cMRI) and to obtain left ventricular end-diastolic pressure (LVEDP). METHODS: For this modelling study, patients who had undergone cardiac catheterisation at University Hospital Heidelberg (Heidelberg, Germany) between July 15, 2004 and March 16, 2023, were identified, as were individual left ventricular pressure measurements. We used existing patient data from routine cardiac diagnostics. From this initial group, we extracted patients who had been diagnosed with ischaemic cardiomyopathy, dilated cardiomyopathy, hypertrophic cardiomyopathy, or amyloidosis, as well as control individuals with no structural phenotype. Data were pseudonymised and only processed within the university hospital's AI infrastructure. We used the data to build different models to predict either demographic (ie, AI-age and AI-sex), diagnostic (ie, AI-coronary artery disease and AI-cardiomyopathy [AI-CMP]), or functional parameters (ie, AI-LVEDP). We randomly divided our datasets via computer into training, validation, and test datasets. AI-CMP was not compared with other models, but was validated in a prospective setting. Benchmarking was also done. FINDINGS: 66 936 patients who had undergone cardiac catheterisation at University Hospital Heidelberg were identified, with more than 183 772 individual left ventricular pressure measurements. We extracted 4390 patients from this initial group, of whom 1131 (25·8%) had been diagnosed with ischaemic cardiomyopathy, 1064 (24·2%) had been diagnosed with dilated cardiomyopathy, 816 (18·6%) had been diagnosed with hypertrophic cardiomyopathy, 202 (4·6%) had been diagnosed with amyloidosis, and 1177 (26·7%) were control individuals with no structural phenotype. The core cohort only included patients with cardiac catherisation and cMRI within 30 days, and emergency cases were excluded. AI-sex was able to predict patient sex with areas under the receiver operating characteristic curves (AUCs) of 0·78 (95% CI 0·77-0·78) and AI-age was able to predict patient age with a mean absolute error of 7·86 years (7·77-7·95), with a Pearson correlation of 0·57 (95% CI 0·56-0·57). The AUCs for the classification tasks ranged between 0·82 (95% CI 0·79-0·84) for ischaemic cardiomyopathy and 0·92 (0·91-0·94) for hypertrophic cardiomyopathy. INTERPRETATION: Our AI models could be easily integrated into clinical practice and provide added value to the information content of cMRI, allowing for disease classification and prediction of diastolic function. FUNDING: Informatics for Life initiative of the Klaus-Tschira Foundation, German Center for Cardiovascular Research, eCardiology section of the German Cardiac Society, and AI Health Innovation Cluster Heidelberg.

Molecular Transducers of Physical Activity Consortium (MoTrPAC): human studies design and protocol.

Physical activity, including structured exercise, is associated with favorable health-related chronic disease outcomes. Although there is evidence of various molecular pathways that affect these responses, a comprehensive molecular map of these molecular responses to exercise has not been developed. The Molecular Transducers of Physical Activity Consortium (MoTrPAC) is a multicenter study designed to isolate the effects of structured exercise training on the molecular mechanisms underlying the health benefits of exercise and physical activity. MoTrPAC contains both a preclinical and human component. The details of the human studies component of MoTrPAC that include the design and methods are presented here. The human studies contain both an adult and pediatric component. In the adult component, sedentary participants are randomized to 12 wk of Control, Endurance Exercise Training, or Resistance Exercise Training with outcomes measures completed before and following the 12 wk. The adult component also includes recruitment of highly active endurance-trained or resistance-trained participants who only complete measures once. A similar design is used for the pediatric component; however, only endurance exercise is examined. Phenotyping measures include weight, body composition, vital signs, cardiorespiratory fitness, muscular strength, physical activity and diet, and other questionnaires. Participants also complete an acute rest period (adults only) or exercise session (adults, pediatrics) with collection of biospecimens (blood only for pediatrics) to allow for examination of the molecular responses. The design and methods of MoTrPAC may inform other studies. Moreover, MoTrPAC will provide a repository of data that can be used broadly across the scientific community.NEW & NOTEWORTHY The Molecular Transducers of Physical Activity Consortium (MoTrPAC) will be the first large trial to isolate the effects of structured exercise training on the molecular mechanisms underlying the health benefits of exercise and physical activity. By generating a compendium of the molecular responses to exercise, MoTrPAC will lay the foundation for a new era of biomedical research on Precision Exercise Medicine. Presented here is the design, protocols, and procedures for the MoTrPAC human studies.

Learning epistatic polygenic phenotypes with Boolean interactions.

Detecting epistatic drivers of human phenotypes is a considerable challenge. Traditional approaches use regression to sequentially test multiplicative interaction terms involving pairs of genetic variants. For higher-order interactions and genome-wide large-scale data, this strategy is computationally intractable. Moreover, multiplicative terms used in regression modeling may not capture the form of biological interactions. Building on the Predictability, Computability, Stability (PCS) framework, we introduce the epiTree pipeline to extract higher-order interactions from genomic data using tree-based models. The epiTree pipeline first selects a set of variants derived from tissue-specific estimates of gene expression. Next, it uses iterative random forests (iRF) to search training data for candidate Boolean interactions (pairwise and higher-order). We derive significance tests for interactions, based on a stabilized likelihood ratio test, by simulating Boolean tree-structured null (no epistasis) and alternative (epistasis) distributions on hold-out test data. Finally, our pipeline computes PCS epistasis p-values that probabilisticly quantify improvement in prediction accuracy via bootstrap sampling on the test set. We validate the epiTree pipeline in two case studies using data from the UK Biobank: predicting red hair and multiple sclerosis (MS). In the case of predicting red hair, epiTree recovers known epistatic interactions surrounding MC1R and novel interactions, representing non-linearities not captured by logistic regression models. In the case of predicting MS, a more complex phenotype than red hair, epiTree rankings prioritize novel interactions surrounding HLA-DRB1, a variant previously associated with MS in several populations. Taken together, these results highlight the potential for epiTree rankings to help reduce the design space for follow up experiments.

Integration of transcriptomics and long-read genomics prioritizes structural variants in rare disease.

Rare structural variants (SVs) - insertions, deletions, and complex rearrangements - can cause Mendelian disease, yet they remain difficult to accurately detect and interpret. We sequenced and analyzed Oxford Nanopore long-read genomes of 68 individuals from the Undiagnosed Disease Network (UDN) with no previously identified diagnostic mutations from short-read sequencing. Using our optimized SV detection pipelines and 571 control long-read genomes, we detected 716 long-read rare (MAF < 0.01) SV alleles per genome on average, achieving a 2.4x increase from short-reads. To characterize the functional effects of rare SVs, we assessed their relationship with gene expression from blood or fibroblasts from the same individuals, and found that rare SVs overlapping enhancers were enriched (LOR = 0.46) near expression outliers. We also evaluated tandem repeat expansions (TREs) and found 14 rare TREs per genome; notably these TREs were also enriched near overexpression outliers. To prioritize candidate functional SVs, we developed Watershed-SV, a probabilistic model that integrates expression data with SV-specific genomic annotations, which significantly outperforms baseline models that don't incorporate expression data. Watershed-SV identified a median of eight high-confidence functional SVs per UDN genome. Notably, this included compound heterozygous deletions in FAM177A1 shared by two siblings, which were likely causal for a rare neurodevelopmental disorder. Our observations demonstrate the promise of integrating long-read sequencing with gene expression towards improving the prioritization of functional SVs and TREs in rare disease patients.

Quantifying assumptions underlying peak oxygen consumption equations across the body mass spectrum.

The goal of this study is to quantify the assumptions associated with the Wasserman-Hansen (WH) and Fitness Registry and the Importance of Exercise: A National Database (FRIEND) predictive peak oxygen consumption (pVO2) equations across body mass index (BMI). Assumptions in pVO2 for both equations were first determined using a simulation and then evaluated using exercise data from the Stanford Exercise Testing registry. We calculated percent-predicted VO2 (ppVO2) values for both equations and compared them using the Bland-Altman method. Assumptions associated with pVO2 across BMI categories were quantified by comparing the slopes of age-adjusted VO2 ratios (pVO2/pre-exercise VO2) and ppVO2 values for different BMI categories. The simulation revealed lower predicted fitness among adults with obesity using the FRIEND equation compared to the WH equations. In the clinical cohort, we evaluated 2471 patients (56.9% male, 22% with BMI >30 kg/m2, pVO2 26.8 mlO2/kg/min). The Bland-Altman plot revealed an average relative difference of -1.7% (95% CI: -2.1 to -1.2%) between WH and FRIEND ppVO2 values with greater differences among those with obesity. Analysis of the VO2 ratio to ppVO2 slopes across the BMI spectrum confirmed the assumption of lower fitness in those with obesity, and this trend was more pronounced using the FRIEND equation. Peak VO2 estimations between the WH and FRIEND equations differed significantly among individuals with obesity. The FRIEND equation resulted in a greater attributable reduction in pVO2 associated with obesity relative to the WH equations. The outlined relationships between BMI and predicted VO2 may better inform the clinical interpretation of ppVO2 values during cardiopulmonary exercise test evaluations.

Regional Variation in Cardiovascular Genes Enables a Tractable Genome Editing Strategy.

BACKGROUND: To realize the potential of genome engineering therapeutics, tractable strategies must be identified that balance personalized therapy with the need for off-the-shelf availability. We hypothesized that regional clustering of pathogenic variants can inform the design of rational prime editing therapeutics to treat the majority of genetic cardiovascular diseases with a limited number of reagents. METHODS: We collated 2435 high-confidence pathogenic/likely pathogenic (P/LP) variants in 82 cardiovascular disease genes from ClinVar. We assessed the regional density of these variants by defining a regional clustering index. We then combined a highly active base editor with prime editing to demonstrate the feasibility of a P/LP hotspot-directed genome engineering therapeutic strategy in vitro. RESULTS: P/LP variants in cardiovascular disease genes display higher regional density than rare variants found in the general population. P/LP missense variants displayed higher average regional density than P/LP truncating variants. Following hypermutagenesis at a pathogenic hotspot, mean prime editing efficiency across introduced variants was 57±27%. CONCLUSIONS: Designing therapeutics that target pathogenic hotspots will not only address known missense P/LP variants but also novel P/LP variants identified in these hotspots as well. Moreover, the clustering of P/LP missense rather than truncating variants in these hotspots suggests that prime editing technology is particularly valuable for dominant negative disease. Although prime editing technology in relation to cardiac health continues to improve, this study presents an approach to targeting the most impactful regions of the genome for inherited cardiovascular disease.

Deep learning to detect left ventricular structural abnormalities in chest X-rays.

BACKGROUND AND AIMS: Early identification of cardiac structural abnormalities indicative of heart failure is crucial to improving patient outcomes. Chest X-rays (CXRs) are routinely conducted on a broad population of patients, presenting an opportunity to build scalable screening tools for structural abnormalities indicative of Stage B or worse heart failure with deep learning methods. In this study, a model was developed to identify severe left ventricular hypertrophy (SLVH) and dilated left ventricle (DLV) using CXRs. METHODS: A total of 71 589 unique CXRs from 24 689 different patients completed within 1 year of echocardiograms were identified. Labels for SLVH, DLV, and a composite label indicating the presence of either were extracted from echocardiograms. A deep learning model was developed and evaluated using area under the receiver operating characteristic curve (AUROC). Performance was additionally validated on 8003 CXRs from an external site and compared against visual assessment by 15 board-certified radiologists. RESULTS: The model yielded an AUROC of 0.79 (0.76-0.81) for SLVH, 0.80 (0.77-0.84) for DLV, and 0.80 (0.78-0.83) for the composite label, with similar performance on an external data set. The model outperformed all 15 individual radiologists for predicting the composite label and achieved a sensitivity of 71% vs. 66% against the consensus vote across all radiologists at a fixed specificity of 73%. CONCLUSIONS: Deep learning analysis of CXRs can accurately detect the presence of certain structural abnormalities and may be useful in early identification of patients with LV hypertrophy and dilation. As a resource to promote further innovation, 71 589 CXRs with adjoining echocardiographic labels have been made publicly available.

APOE loss-of-function variants: Compatible with longevity and associated with resistance to Alzheimer's disease pathology.

The ε4 allele of apolipoprotein E (APOE) is the strongest genetic risk factor for sporadic Alzheimer's disease (AD). Knockdown of ε4 may provide a therapeutic strategy for AD, but the effect of APOE loss of function (LoF) on AD pathogenesis is unknown. We searched for APOE LoF variants in a large cohort of controls and patients with AD and identified seven heterozygote carriers of APOE LoF variants. Five carriers were controls (aged 71-90 years), one carrier was affected by progressive supranuclear palsy, and one carrier was affected by AD with an unremarkable age at onset of 75 years. Two APOE ε3/ε4 controls carried a stop-gain affecting ε4: one was cognitively normal at 90 years and had no neuritic plaques at autopsy; the other was cognitively healthy at 79 years, and lumbar puncture at 76 years showed normal levels of amyloid. These results suggest that ε4 drives AD risk through the gain of abnormal function and support ε4 knockdown as a viable therapeutic option.

Improving Reporting of Exercise Capacity Across Age Ranges Using Novel Workload Reference Equations.

Exercise capacity (EC) is an important predictor of survival in the general population and in subjects with cardiopulmonary disease. Despite its relevance, considering the percent-predicted workload (%pWL) given by current equations may overestimate EC in older adults. Therefore, to improve the reporting of EC in clinical practice, our main objective was to develop workload reference equations (pWL) that better reflect the relation between workload and age. Using the Fitness Registry and the Importance of Exercise National Database (FRIEND), we analyzed a reference group of 6,966 apparently healthy participants and 1,060 participants with heart failure who underwent graded treadmill cardiopulmonary exercise testing. For the first group, the mean age was 44 years (18 to 79); 56.5% of participants were males and 15.4% had obesity. Peak oxygen consumption was 11.6 ± 3.0 METs in males and 8.5 ± 2.4 METs in females. After partition analysis, we first developed sex-specific pWL equations to allow comparisons to a healthy weight reference. For males, pWL (METs) = 14.1-0.9×10-3×age2 and 11.5-0.87×10-3×age2 for females. We used those equations as denominators of %pWL, and based on their distribution, we determined thresholds for EC classification, with average EC defined by the range corresponding to 85% to 115%pWL. Compared with %pWL using current equations, the new equations yielded better-calibrated %pWL across different age ranges. We also derived body mass index-adjusted pWL equations that better assessed EC in subjects with heart failure. In conclusion, the novel pWL equations have the potential to impact the report of EC in practice.

Knowledge and attitudes on implementing cardiovascular pharmacogenomic testing.

Pharmacogenomics has the potential to inform drug dosing and selection, reduce adverse events, and improve medication efficacy; however, provider knowledge of pharmacogenomic testing varies across provider types and specialties. Given that many actionable pharmacogenomic genes are implicated in cardiovascular medication response variability, this study aimed to evaluate cardiology providers' knowledge and attitudes on implementing clinical pharmacogenomic testing. Sixty-one providers responded to an online survey, including pharmacists (46%), physicians (31%), genetic counselors (15%), and nurses (8%). Most respondents (94%) reported previous genetics education; however, only 52% felt their genetics education prepared them to order a clinical pharmacogenomic test. In addition, most respondents (66%) were familiar with pharmacogenomics, with genetic counselors being most likely to be familiar (p < 0.001). Only 15% of respondents had previously ordered a clinical pharmacogenomic test and a total of 36% indicated they are likely to order a pharmacogenomic test in the future; however, the vast majority of respondents (89%) were interested in pharmacogenomic testing being incorporated into diagnostic cardiovascular genetic tests. Moreover, 84% of providers preferred pharmacogenomic panel testing compared to 16% who preferred single gene testing. Half of the providers reported being comfortable discussing pharmacogenomic results with their patients, but the majority (60%) expressed discomfort with the logistics of test ordering. Reported barriers to implementation included uncertainty about the clinical utility and difficulty choosing an appropriate test. Taken together, cardiology providers have moderate familiarity with pharmacogenomics and limited experience with test ordering; however, they are interested in incorporating pharmacogenomics into diagnostic genetic tests and ordering pharmacogenomic panels.

Almanac - Retrieval-Augmented Language Models for Clinical Medicine.

BACKGROUND: Large language models (LLMs) have recently shown impressive zero-shot capabilities, whereby they can use auxiliary data, without the availability of task-specific training examples, to complete a variety of natural language tasks, such as summarization, dialogue generation, and question answering. However, despite many promising applications of LLMs in clinical medicine, adoption of these models has been limited by their tendency to generate incorrect and sometimes even harmful statements. METHODS: We tasked a panel of eight board-certified clinicians and two health care practitioners with evaluating Almanac, an LLM framework augmented with retrieval capabilities from curated medical resources for medical guideline and treatment recommendations. The panel compared responses from Almanac and standard LLMs (ChatGPT-4, Bing, and Bard) versus a novel data set of 314 clinical questions spanning nine medical specialties. RESULTS: Almanac showed a significant improvement in performance compared with the standard LLMs across axes of factuality, completeness, user preference, and adversarial safety. CONCLUSIONS: Our results show the potential for LLMs with access to domain-specific corpora to be effective in clinical decision-making. The findings also underscore the importance of carefully testing LLMs before deployment to mitigate their shortcomings. (Funded by the National Institutes of Health, National Heart, Lung, and Blood Institute.).