Temporal dynamics of the multi-omic response to endurance exercise training

Regular exercise promotes whole-body health and prevents disease, but the underlying molecular mechanisms are incompletely understood1-3. Here, the Molecular Transducers of Physical Activity Consortium4 profiled the temporal transcriptome, proteome, metabolome, lipidome, phosphoproteome, acetylproteome, ubiquitylproteome, epigenome and immunome in whole blood, plasma and 18 solid tissues in male and female Rattus norvegicus over eight weeks of endurance exercise training. The resulting data compendium encompasses 9,466 assays across 19 tissues, 25 molecular platforms and 4 training time points. Thousands of shared and tissue-specific molecular alterations were identified, with sex differences found in multiple tissues. Temporal multi-omic and multi-tissue analyses revealed expansive biological insights into the adaptive responses to endurance training, including widespread regulation of immune, metabolic, stress response and mitochondrial pathways. Many changes were relevant to human health, including non-alcoholic fatty liver disease, inflammatory bowel disease, cardiovascular health and tissue injury and recovery. The data and analyses presented in this study will serve as valuable resources for understanding and exploring the multi-tissue molecular effects of endurance training and are provided in a public repository ( https://motrpac-data.org/ ).

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.

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.

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.

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.

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.).

A 3'UTR Insertion Is a Candidate Causal Variant at the TMEM106B Locus Associated with Increased Risk for FTLD-TDP

Background and Objectives

Single nucleotide variants near TMEM106B associate with risk of frontotemporal lobar dementia with TDP-43 inclusions (FTLD-TDP) and Alzheimer's disease (AD) in genome-wide association studies (GWAS), but the causal variant at this locus remains unclear. Here we asked whether a novel structural variant on TMEM106B is the causal variant.

Methods

An exploratory analysis identified structural variants on neurodegeneration-related genes. Subsequent analyses focused on an Alu element insertion on the 3'UTR of TMEM106B. This study included data from longitudinal aging and neurogenerative disease cohorts at Stanford University, case-control cohorts in the Alzheimer's Disease Sequencing Project (ADSP), and expression and proteomics data from Washington University in St. Louis (WUSTL). 432 individuals from two Stanford aging cohorts were whole-genome long-read and short-read sequenced. 16,906 samples from ADSP were short-read sequenced. Genotypes, transcriptomics, and proteomics data were available in 1,979 participants from an aging and dementia cohort at WUSTL. Selection criteria were specific to each cohort. In primary analyses, the linkage disequilibrium between the TMEM106B locus variants in the FTLD-TDP GWAS and the 3'UTR insertion was estimated. We then estimated linkage by ancestry in the ADSP and evaluated the effect of the TMEM106B lead variant on mRNA and protein levels.

Results

The primary analysis included 432 participants (52.5% females, age range 45-92 years old). We identified a 316 bp Alu insertion overlapping the TMEM106B 3'UTR tightly linked with top GWAS variants rs3173615(C) and rs1990622(A). In ADSP European-ancestry participants, this insertion is in equivalent linkage with rs1990622(A) (R2=0.962, D'=0.998) and rs3173615(C) (R2=0.960, D'=0.996). In African-ancestry participants, the insertion is in stronger linkage with rs1990622(A) (R2=0.992, D'=0.998) than with rs3173615(C) (R2=0.811, D'=0.994). In public datasets, rs1990622 was consistently associated with TMEM106B protein levels but not with mRNA expression. In the WUSTL dataset, rs1990622 is associated with TMEM106B protein levels in plasma and cerebrospinal fluid, but not with TMEM106B mRNA expression.

Discussion

We identified a novel Alu element insertion in the 3'UTR of TMEM106B in tight linkage with the lead FTLD-TDP risk variant. The lead variant is associated with TMEM106B protein levels, but not expression. The 3'UTR insertion is a lead candidate for the causal variant at this complex locus, pending confirmation with functional studies.

The Stanford Medicine data science ecosystem for clinical and translational research

Objective

To describe the infrastructure, tools, and services developed at Stanford Medicine to maintain its data science ecosystem and research patient data repository for clinical and translational research.

Materials and Methods

The data science ecosystem, dubbed the Stanford Data Science Resources (SDSR), includes infrastructure and tools to create, search, retrieve, and analyze patient data, as well as services for data deidentification, linkage, and processing to extract high-value information from healthcare IT systems. Data are made available via self-service and concierge access, on HIPAA compliant secure computing infrastructure supported by in-depth user training.

Results

The Stanford Medicine Research Data Repository (STARR) functions as the SDSR data integration point, and includes electronic medical records, clinical images, text, bedside monitoring data and HL7 messages. SDSR tools include tools for electronic phenotyping, cohort building, and a search engine for patient timelines. The SDSR supports patient data collection, reproducible research, and teaching using healthcare data, and facilitates industry collaborations and large-scale observational studies.

Discussion

Research patient data repositories and their underlying data science infrastructure are essential to realizing a learning health system and advancing the mission of academic medical centers. Challenges to maintaining the SDSR include ensuring sufficient financial support while providing researchers and clinicians with maximal access to data and digital infrastructure, balancing tool development with user training, and supporting the diverse needs of users.

Conclusion

Our experience maintaining the SDSR offers a case study for academic medical centers developing data science and research informatics infrastructure.

A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease

The electrocardiogram (ECG) is the most frequently performed cardiovascular diagnostic test, but it is unclear how much information resting ECGs contain about long term cardiovascular risk. Here we report that a deep convolutional neural network can accurately predict the long-term risk of cardiovascular mortality and disease based on a resting ECG alone. Using a large dataset of resting 12-lead ECGs collected at Stanford University Medical Center, we developed SEER, the Stanford Estimator of Electrocardiogram Risk. SEER predicts 5-year cardiovascular mortality with an area under the receiver operator characteristic curve (AUC) of 0.83 in a held-out test set at Stanford, and with AUCs of 0.78 and 0.83 respectively when independently evaluated at Cedars-Sinai Medical Center and Columbia University Irving Medical Center. SEER predicts 5-year atherosclerotic disease (ASCVD) with an AUC of 0.67, similar to the Pooled Cohort Equations for ASCVD Risk, while being only modestly correlated. When used in conjunction with the Pooled Cohort Equations, SEER accurately reclassified 16% of patients from low to moderate risk, uncovering a group with an actual average 9.9% 10-year ASCVD risk who would not have otherwise been indicated for statin therapy. SEER can also predict several other cardiovascular conditions such as heart failure and atrial fibrillation. Using only lead I of the ECG it predicts 5-year cardiovascular mortality with an AUC of 0.80. SEER, used alongside the Pooled Cohort Equations and other risk tools, can substantially improve cardiovascular risk stratification and aid in medical decision making.

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 this allele 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 older controls and patients with AD and identified six heterozygote carriers of APOE LoF variants. Five carriers were controls (ages 71-90) and one was an AD case with an unremarkable age-at-onset between 75-79. Two APOE ε3/ε4 controls (Subjects 1 and 2) carried a stop-gain affecting the ε4 allele. Subject 1 was cognitively normal at 90+ and had no neuritic plaques at autopsy. Subject 2 was cognitively healthy within the age range 75-79 and underwent lumbar puncture at between ages 75-79 with normal levels of amyloid. The results provide the strongest human genetics evidence yet available suggesting that ε4 drives AD risk through a gain of abnormal function and support knockdown of APOE ε4 or its protein product as a viable therapeutic option.

Respiratory gas kinetics in patients with congestive heart failure during recovery from peak exercise

BACKGROUND

Cardiopulmonary Exercise Testing (CPX) is essential for the assessment of exercise capacity for patients with Chronic Heart Failure (CHF). Respiratory gas and hemodynamic parameters such as Ventilatory Efficiency (VE/VCO₂ slope), peak oxygen uptake (peak VO₂), and heart rate recovery are established diagnostic and prognostic markers for clinical populations. Previous studies have suggested the clinical value of metrics related to respiratory gas collected during recovery from peak exercise, particularly recovery time to 50% (T1/2) of peak VO₂. The current study explores these metrics in detail during recovery from peak exercise in CHF.

METHODS

Patients with CHF who were referred for CPX and healthy individuals without formal diagnoses were assessed for inclusion. All subjects performed CPX on cycle ergometers to volitional exhaustion and were monitored for at least five minutes of recovery. CPX data were analyzed for overshoot of respiratory exchange ratio (RER=VCO₂/VO₂), ventilatory equivalent for oxygen (VE/VO₂), end-tidal partial pressure of oxygen (PETO₂), and T1/2 of peak VO₂ and VCO₂.

RESULTS

Thirty-two patients with CHF and 30 controls were included. Peak VO₂ differed significantly between patients and controls (13.5 ± 3.8 vs. 32.5 ± 9.8 mL/Kg*min^-1, p < 0.001). Mean Left Ventricular Ejection Fraction (LVEF) was 35.9 ± 9.8% for patients with CHF compared to 61.1 ± 8.2% in the control group. The T1/2 of VO₂, VCO₂ and VE was significantly higher in patients (111.3 ± 51.0, 132.0 ± 38.8 and 155.6 ± 45.5s) than in controls (58.08 ± 13.2, 74.3 ± 21.1, 96.7 ± 36.8s; p < 0.001) while the overshoot of PETO₂, VE/VO₂ and RER was significantly lower in patients (7.2 ± 3.3, 41.9 ± 29.1 and 25.0 ± 13.6%) than in controls (10.1 ± 4.6, 62.1 ± 17.7 and 38.7 ± 15.1%; all p < 0.01). Most of the recovery metrics were significantly correlated with peak VO₂ in CHF patients, but not with LVEF.

CONCLUSIONS

Patients with CHF have a significantly blunted recovery from peak exercise. This is reflected in delays of VO₂, VCO₂, VE, PETO₂, RER and VE/VO₂, reflecting a greater energy required to return to baseline. Abnormal respiratory gas kinetics in CHF was negatively correlated with peak VO₂ but not baseline LVEF.

Almanac: Retrieval-Augmented Language Models for Clinical Medicine

Large-language models have recently demonstrated impressive zero-shot capabilities in a variety of natural language tasks such as summarization, dialogue generation, and question-answering. Despite many promising applications in clinical medicine, adoption of these models in real-world settings has been largely limited by their tendency to generate incorrect and sometimes even toxic statements. In this study, we develop Almanac, a large language model framework augmented with retrieval capabilities for medical guideline and treatment recommendations. Performance on a novel dataset of clinical scenarios (n= 130) evaluated by a panel of 5 board-certified and resident physicians demonstrates significant increases in factuality (mean of 18% at p-value < 0.05) across all specialties, with improvements in completeness and safety. Our results demonstrate the potential for large language models to be effective tools in the clinical decision-making process, while also emphasizing the importance of careful testing and deployment to mitigate their shortcomings.

Cardiac splicing as a diagnostic and therapeutic target

Despite advances in therapeutics for heart failure and arrhythmias, a substantial proportion of patients with cardiomyopathy do not respond to interventions, indicating a need to identify novel modifiable myocardial pathobiology. Human genetic variation associated with severe forms of cardiomyopathy and arrhythmias has highlighted the crucial role of alternative splicing in myocardial health and disease, given that it determines which mature RNA transcripts drive the mechanical, structural, signalling and metabolic properties of the heart. In this Review, we discuss how the analysis of cardiac isoform expression has been facilitated by technical advances in multiomics and long-read and single-cell sequencing technologies. The resulting insights into the regulation of alternative splicing - including the identification of cardiac splice regulators as therapeutic targets and the development of a translational pipeline to evaluate splice modulators in human engineered heart tissue, animal models and clinical trials - provide a basis for improved diagnosis and therapy. Finally, we consider how the medical and scientific communities can benefit from facilitated acquisition and interpretation of splicing data towards improved clinical decision-making and patient care.

Defining left ventricular remodeling using lean body mass allometry: a UK Biobank study

PURPOSE

The geometric patterns of ventricular remodeling are determined using indexed left ventricular mass (LVM), end-diastolic volume (LVEDV) and concentricity, most often measured using the mass-to-volume ratio (MVR). The aims of this study were to validate lean body mass (LBM)-based allometric coefficients for scaling and to determine an index of concentricity that is independent of both volume and LBM.

METHODS

Participants from the UK Biobank who underwent both CMR and dual-energy X-ray absorptiometry (DXA) during 2014-2015 were considered (n = 5064). We excluded participants aged ≥ 70 years or those with cardiometabolic risk factors. We determined allometric coefficients for scaling using linear regression of the logarithmically transformed ventricular remodeling parameters. We further defined a multiplicative allometric relationship for LV concentricity (LVC) adjusting for both LVEDV and LBM.

RESULTS

A total of 1638 individuals (1057 female) were included. In subjects with lower body fat percentage (< 25% in males, < 35% in females, n = 644), the LBM allometric coefficients for scaling LVM and LVEDV were 0.85 ± 0.06 and 0.85 ± 0.03 respectively (R² = 0.61 and 0.57, P < 0.001), with no evidence of sex-allometry interaction. While the MVR was independent of LBM, it demonstrated a negative association with LVEDV in (females: r = - 0.44, P < 0.001; males: - 0.38, P < 0.001). In contrast, LVC was independent of both LVEDV and LBM [LVC = LVM/(LVEDV^0.40 × LBM^0.50)] leading to increased overlap between LV hypertrophy and higher concentricity.

CONCLUSIONS

We validated allometric coefficients for LBM-based scaling for CMR indexed parameters relevant for classifying geometric patterns of ventricular remodeling.

Association of Titin Variations With Late-Onset Dilated Cardiomyopathy

IMPORTANCE

Dilated cardiomyopathy (DCM) is frequently caused by genetic factors. Studies identifying deleterious rare variants have predominantly focused on early-onset cases, and little is known about the genetic underpinnings of the growing numbers of patients with DCM who are diagnosed when they are older than 60 years (ie, late-onset DCM).

OBJECTIVE

To investigate the prevalence, type, and prognostic impact of disease-associated rare variants in patients with late-onset DCM.

DESIGN, SETTING, AND PARTICIPANTS

A population of patients with late-onset DCM who had undergone genetic testing in 7 international tertiary referral centers worldwide were enrolled from March 1990 to August 2020. A positive genotype was defined as the presence of pathogenic or likely pathogenic (P/LP) variants.

MAIN OUTCOMES AND MEASURES

The study outcome was all-cause mortality.

RESULTS

A total of 184 patients older than 60 years (103 female [56%]; mean [SD] age, 67 [6] years; mean [SD] left ventricular ejection fraction, 32% [10%]) were studied. Sixty-six patients (36%) were carriers of a P/LP variant. Titin-truncating variants were the most prevalent (present in 46 [25%] of the total population and accounting for 46 [69%] of all genotype-positive patients). During a median (interquartile range) follow-up of 42 (10-115) months, 23 patients (13%) died; 17 (25%) of these were carriers of P/LP variants, while 6 patients (5.1%) were genotype-negative.

CONCLUSIONS AND RELEVANCE

Late-onset DCM might represent a distinct subgroup characterized by and a high genetic variation burden, largely due to titin-truncating variants. Patients with a positive genetic test had higher mortality than genotype-negative patients. These findings support the extended use of genetic testing also in older patients.

SARS-CoV-2 RNAemia Predicts Clinical Deterioration and Extrapulmonary Complications from COVID-19

BACKGROUND

The determinants of coronavirus disease 2019 (COVID-19) disease severity and extrapulmonary complications (EPCs) are poorly understood. We characterized relationships between severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNAemia and disease severity, clinical deterioration, and specific EPCs.

METHODS

We used quantitative and digital polymerase chain reaction (qPCR and dPCR) to quantify SARS-CoV-2 RNA from plasma in 191 patients presenting to the emergency department with COVID-19. We recorded patient symptoms, laboratory markers, and clinical outcomes, with a focus on oxygen requirements over time. We collected longitudinal plasma samples from a subset of patients. We characterized the role of RNAemia in predicting clinical severity and EPCs using elastic net regression.

RESULTS

Of SARS-CoV-2-positive patients, 23.0% (44 of 191) had viral RNA detected in plasma by dPCR, compared with 1.4% (2 of 147) by qPCR. Most patients with serial measurements had undetectable RNAemia within 10 days of symptom onset, reached maximum clinical severity within 16 days, and symptom resolution within 33 days. Initially RNAemic patients were more likely to manifest severe disease (odds ratio, 6.72 [95% confidence interval, 2.45-19.79]), worsening of disease severity (2.43 [1.07-5.38]), and EPCs (2.81 [1.26-6.36]). RNA loads were correlated with maximum severity (r = 0.47 [95% confidence interval, .20-.67]).

CONCLUSIONS

dPCR is more sensitive than qPCR for the detection of SARS-CoV-2 RNAemia, which is a robust predictor of eventual COVID-19 severity and oxygen requirements, as well as EPCs. Because many COVID-19 therapies are initiated on the basis of oxygen requirements, RNAemia on presentation might serve to direct early initiation of appropriate therapies for the patients most likely to deteriorate.

418 Titin mutations and female sex characterize dilated cardiomyopathy in the elderly

Aims

Dilated cardiomyopathy (DCM) is frequently caused by genetic factors. Studies identifying deleterious rare variants have predominantly focused on early-onset cases, and little is known about the genetic underpinnings of the growing numbers of patients with DCM who are diagnosed after 60 years of age (i.e. late-onset DCM). The aim is to investigate the prevalence, type, and prognostic impact of disease-associated rare variants in late-onset DCM patients.

Methods and results

We analysed a population of late-onset DCM patients who had undergone genetic testing in seven international tertiary referral centres worldwide. A positive genotype was defined as the presence of ‘pathogenic’ or ‘likely pathogenic’ (P/LP) variants. The study outcome was all-cause mortality. 184 patients over age 60 years (56% females, mean age 67 ± 6 years, mean left ventricular ejection fraction 32 ± 10%) were studied. Sixty-six patients (36%) were carriers of a P/LP variant. Titin truncating variants (TTNtv) were the most prevalent (present in 25% of the total population and accounting for 69% of all genotype-positive patients). During a median follow-up of 42 months (interquartile range: 10–115), 23 patients (13%) died; 17 of these (25%) were carriers of P/LP variants while six patients (5.1%) were genotype-negative (P &lt; 0.001).

Conclusions

In the largest series worldwide, to date, of patients with late-onset DCM, we found a high prevalence of female sex and a high genetic mutation burden, largely due to TTNtv. Patients with a positive genetic test had higher mortality than genotype-negative patients. These findings support the extended use of genetic testing also in the elderly.

Comparison of the FRIEND and Wasserman-Hansen Equations in Predicting Outcomes in Heart Failure

Background

Percentage of age‐predicted peak oxygen uptake (VO₂) achieved (ppVO₂) has been widely used to stratify risk in patients with heart failure. However, there are limitations to traditional normal standards. We compared the recently derived FRIEND (Fitness Registry and the Importance of Exercise: A National Data Base) equation to the widely used Wasserman‐Hansen (WH) ppVO₂ equation to predict outcomes in patients with heart failure.

Methods and Results

A subgroup of 4055 heart failure patients from the FRIEND registry (mean age 53±15 years) was followed for a mean of 28±16 months. The FRIEND and WH equations along with measured peak VO₂ expressed in mL/kg⁻¹ per min⁻¹ were compared for mortality and composite cardiovascular events. ppVO₂ was higher for the FRIEND versus the WH equation (66±30% versus 58±25%; P<0.001). The areas under the receiver operating characteristic curves were slightly but significantly higher for the FRIEND equation for mortality (0.74 versus 0.72; P=0.03) and cardiac events (0.70 versus 0.68; P=0.008). Area under the receiver operating characteristic curve for measured peak VO₂ was 0.70 (P<0.001) for mortality and 0.73 (P<0.001) for cardiovascular events. For each 1‐SD higher ppVO₂ for the FRIEND equation, mortality was reduced by 18% (hazard ratio, 0.82; 95% CI, 0.69–0.97; P<0.02); for each 1‐SD higher ppVO₂ for the WH equation, the mortality was reduced by 17% (hazard ratio, 0.83; 95% CI, 0.71–0.97; P=0.02). The corresponding reductions in risk per 1 SD for cardiovascular events for the FRIEND and WH equations were 23 and 21%, respectively (both P<0.001).

Conclusions

Peak VO₂ expressed as percentage of an age‐predicted standard strongly predicts mortality and major cardiovascular events in patients with heart failure. The FRIEND registry equation exhibited test characteristics slightly superior to the commonly used WH equation.

Smartphone-Based VO2max Measurement With Heart Snapshot in Clinical and Real-world Settings With a Diverse Population: Validation Study

BACKGROUND

Maximal oxygen consumption (VO₂max) is one of the most predictive biometrics for cardiovascular health and overall mortality. However, VO₂max is rarely measured in large-scale research studies or routine clinical care because of the high cost, participant burden, and requirement for specialized equipment and staff.

OBJECTIVE

To overcome the limitations of clinical VO₂max measurement, we aim to develop a digital VO₂max estimation protocol that can be self-administered remotely using only the sensors within a smartphone. We also aim to validate this measure within a broadly representative population across a spectrum of smartphone devices.

METHODS

Two smartphone-based VO₂max estimation protocols were developed: a 12-minute run test (12-MRT) based on distance measured by GPS and a 3-minute step test (3-MST) based on heart rate recovery measured by a camera. In a 101-person cohort, balanced across age deciles and sex, participants completed a gold standard treadmill-based VO₂max measurement, two silver standard clinical protocols, and the smartphone-based 12-MRT and 3-MST protocols in the clinic and at home. In a separate 120-participant cohort, the video-based heart rate measurement underlying the 3-MST was measured for accuracy in individuals across the spectrum skin tones while using 8 different smartphones ranging in cost from US $99 to US $999.

RESULTS

When compared with gold standard VO₂max testing, Lin concordance was p_c=0.66 for 12-MRT and p_c=0.61 for 3-MST. However, in remote settings, the 12-MRT was significantly less concordant with the gold standard (p_c=0.25) compared with the 3-MST (p_c=0.61), although both had high test-retest reliability (12-MRT intraclass correlation coefficient=0.88; 3-MST intraclass correlation coefficient=0.86). On the basis of the finding that 3-MST concordance was generalizable to remote settings whereas 12-MRT was not, the video-based heart rate measure within the 3-MST was selected for further investigation. Heart rate measurements in any of the combinations of the six Fitzpatrick skin tones and 8 smartphones resulted in a concordance of p_c≥0.81. Performance did not correlate with device cost, with all phones selling under US $200 performing better than p_c>0.92.

CONCLUSIONS

These findings demonstrate the importance of validating mobile health measures in the real world across a diverse cohort and spectrum of hardware. The 3-MST protocol, termed as heart snapshot, measured VO₂max with similar accuracy to supervised in-clinic tests such as the Tecumseh (p_c=0.94) protocol, while also generalizing to remote and unsupervised measurements. Heart snapshot measurements demonstrated fidelity across demographic variation in age and sex, across diverse skin pigmentation, and between various iOS and Android phone configurations. This software is freely available for all validation data and analysis code.

Assessing post-exercise respiratory gas kinetics in clinical sample - a pilot study

Funding Acknowledgements

Type of funding sources: None.

Background

Cardiopulmonary exercise testing (CPX) is established in the evaluation of patients with cardiac and pulmonary diseases, and its clinical utility seems to be expanding.  Currently the most important diagnostic and prognostic ventilatory metrics of CPX rely on the exercise phase. Nevertheless, a consistent body of evidence suggests that important information can be derived from the recovery phase, especially in the first few minutes after exercise. In this context, patients with heart failure (HF) demonstrate a slower recovery of the oxygen consumption (VO2) compared with healthy individuals. Purpose: To comprehensively investigate the behavior of respiratory gases during recovery from CPX in a diverse cohort of HF patients. Methods: All individuals who performed CPX at the department of cardiology of Stanford University Hospital were eligible for the study. Patients were included in the experimental group if they (i) were recorded for five minutes after the exercise phase of CPX and (ii) had documented heart failure. They were excluded if they had other clinical diagnoses which may be responsible for exercise intolerance or symptoms or were unable to give informed consent. Healthy controls were recruited from the local community and were included if they did not have documented or suspected disease. Respiratory gases were collected on a breath-by-breath basis and analysed after applying a 30 second rolling average filter. Metrics were analyzed as absolute values, percentage change from peak and the half-time of recovery (T ½; i.e. the duration until a metric had returned to ½ of its value at peak). Data was analyzed over time within patients and averages between groups using parametric statistical methods. In accordance with previous studies, the amount of change in a metric after exercise is presented as the "magnitude" of overshoot. Results: 32 patients with HF (11 Female, 47 ± 13 yrs) and 30 healthy subjects (14 Female, 43 ± 12 yrs) were included. A comparison of ventilatory metrics during recovery between HF and controls is depicted in Figure 1. Peak VO2 was 1135 ± 419 mL/min (13.5 ± 3.8 mL/Kg/min) vs 2408 ± 787 mL/min (32.5 ± 9.0 mL/Kg/min); P &lt;0.01. A significant difference between patients with HF and healthy subjects was found in T ½ of VO2 (111.3 ± 51.0s vs 58.0 ± 13.2s, p &lt; 0.01) and VCO2 (132.0 ± 38.8s vs 74.3 ± 21.1s, p &lt; 0.01). The magnitude of the overshoot was also found to be significantly reduced in patients with HF for VE/VO2 (41.9 ± 29.1% vs 62.1 ± 17.7%, P &lt; 0.01), RQ (25.0 ± 13.6% vs 38.7 ± 15.1%, p &lt; 0.01) and PETO2 (7.2 ± 3.3% vs 10.1 ± 4.6%, p &lt; 0.01). Finally, the magnitude of the RQ overshoot showed a moderate correlation with peak VO2 (ϱ=0.58, p &lt; 0.01). Conclusions: We observed that ventilatory kinetics measured in early recovery after CPX differ significantly between healthy subjects and patients with HF. The assessment of post exercise respiratory gases in a clinical setting may add to the prognostic and diagnostic value of CPX in heart failure.

Abstract Figure.

Disease-specific variant pathogenicity prediction significantly improves variant interpretation in inherited cardiac conditions

PURPOSE

Accurate discrimination of benign and pathogenic rare variation remains a priority for clinical genome interpretation. State-of-the-art machine learning variant prioritization tools are imprecise and ignore important parameters defining gene-disease relationships, e.g., distinct consequences of gain-of-function versus loss-of-function variants. We hypothesized that incorporating disease-specific information would improve tool performance.

METHODS

We developed a disease-specific variant classifier, CardioBoost, that estimates the probability of pathogenicity for rare missense variants in inherited cardiomyopathies and arrhythmias. We assessed CardioBoost's ability to discriminate known pathogenic from benign variants, prioritize disease-associated variants, and stratify patient outcomes.

RESULTS

CardioBoost has high global discrimination accuracy (precision recall area under the curve [AUC] 0.91 for cardiomyopathies; 0.96 for arrhythmias), outperforming existing tools (4-24% improvement). CardioBoost obtains excellent accuracy (cardiomyopathies 90.2%; arrhythmias 91.9%) for variants classified with >90% confidence, and increases the proportion of variants classified with high confidence more than twofold compared with existing tools. Variants classified as disease-causing are associated with both disease status and clinical severity, including a 21% increased risk (95% confidence interval [CI] 11-29%) of severe adverse outcomes by age 60 in patients with hypertrophic cardiomyopathy.

CONCLUSIONS

A disease-specific variant classifier outperforms state-of-the-art genome-wide tools for rare missense variants in inherited cardiac conditions ( https://www.cardiodb.org/cardioboost/ ), highlighting broad opportunities for improved pathogenicity prediction through disease specificity.

SARS-CoV-2 RNAaemia predicts clinical deterioration and extrapulmonary complications from COVID-19

Background

The determinants of COVID-19 disease severity and extrapulmonary complications (EPCs) are poorly understood. We characterise the relationships between SARS-CoV-2 RNAaemia and disease severity, clinical deterioration, and specific EPCs.

Methods

We used quantitative (qPCR) and digital (dPCR) PCR to quantify SARS-CoV-2 RNA from nasopharyngeal swabs and plasma in 191 patients presenting to the Emergency Department (ED) with COVID-19. We recorded patient symptoms, laboratory markers, and clinical outcomes, with a focus on oxygen requirements over time. We collected longitudinal plasma samples from a subset of patients. We characterised the role of RNAaemia in predicting clinical severity and EPCs using elastic net regression.

Findings

23·0% (44/191) of SARS-CoV-2 positive patients had viral RNA detected in plasma by dPCR, compared to 1·4% (2/147) by qPCR. Most patients with serial measurements had undetectable RNAaemia 10 days after onset of symptoms, but took 16 days to reach maximum severity, and 33 days for symptoms to resolve. Initially RNAaemic patients were more likely to manifest severe disease (OR 6·72 [95% CI, 2·45 - 19·79]), worsening of disease severity (OR 2·43 [95% CI, 1·07 - 5·38]), and EPCs (OR 2·81 [95% CI, 1·26 - 6·36]). RNA load correlated with maximum severity (r = 0·47 [95% CI, 0·20 - 0·67]).

Interpretation

dPCR is more sensitive than qPCR for the detection of SARS-CoV-2 RNAaemia, which is a robust predictor of eventual COVID-19 severity and oxygen requirements, as well as EPCs. Since many COVID-19 therapies are initiated on the basis of oxygen requirements, RNAaemia on presentation might serve to direct early initiation of appropriate therapies for the patients most likely to deteriorate.

Clinical utility of genomic sequencing: a measurement toolkit

Whole-genome sequencing (WGS) is positioned to become one of the most robust strategies for achieving timely diagnosis of rare genomic diseases. Despite its favorable diagnostic performance compared to conventional testing strategies, routine use and reimbursement of WGS are hampered by inconsistencies in the definition and measurement of clinical utility. For example, what constitutes clinical utility for WGS varies by stakeholder's perspective (physicians, patients, families, insurance companies, health-care organizations, and society), clinical context (prenatal, pediatric, critical care, adult medicine), and test purpose (diagnosis, screening, treatment selection). A rapidly evolving technology landscape and challenges associated with robust comparative study design in the context of rare disease further impede progress in this area of empiric research. To address this challenge, an expert working group of the Medical Genome Initiative was formed. Following a consensus-based process, we align with a broad definition of clinical utility and propose a conceptually-grounded and empirically-guided measurement toolkit focused on four domains of utility: diagnostic thinking efficacy, therapeutic efficacy, patient outcome efficacy, and societal efficacy. For each domain of utility, we offer specific indicators and measurement strategies. While we focus on diagnostic applications of WGS for rare germline diseases, this toolkit offers a flexible framework for best practices around measuring clinical utility for a range of WGS applications. While we expect this toolkit to evolve over time, it provides a resource for laboratories, clinicians, and researchers looking to characterize the value of WGS beyond the laboratory.

Cardiac Imaging of Aortic Valve Area From 34 287 UK Biobank Participants Reveals Novel Genetic Associations and Shared Genetic Comorbidity With Multiple Disease Phenotypes

Background:

The aortic valve is an important determinant of cardiovascular physiology and anatomic location of common human diseases.

Methods:

From a sample of 34 287 white British ancestry participants, we estimated functional aortic valve area by planimetry from prospectively obtained cardiac magnetic resonance imaging sequences of the aortic valve. Aortic valve area measurements were submitted to genome-wide association testing, followed by polygenic risk scoring and phenome-wide screening, to identify genetic comorbidities.

Results:

A genome-wide association study of aortic valve area in these UK Biobank participants showed 3 significant associations, indexed by rs71190365 (chr13:50764607, DLEU1 , P =1.8×10 −9 ), rs35991305 (chr12:94191968, CRADD , P =3.4×10 −8 ), and chr17:45013271:C:T ( GOSR2 , P =5.6×10 −8 ). Replication on an independent set of 8145 unrelated European ancestry participants showed consistent effect sizes in all 3 loci, although rs35991305 did not meet nominal significance. We constructed a polygenic risk score for aortic valve area, which in a separate cohort of 311 728 individuals without imaging demonstrated that smaller aortic valve area is predictive of increased risk for aortic valve disease (odds ratio, 1.14; P =2.3×10 −6 ). After excluding subjects with a medical diagnosis of aortic valve stenosis (remaining n=308 683 individuals), phenome-wide association of >10 000 traits showed multiple links between the polygenic score for aortic valve disease and key health-related comorbidities involving the cardiovascular system and autoimmune disease. Genetic correlation analysis supports a shared genetic etiology with between aortic valve area and birth weight along with other cardiovascular conditions.

Conclusions:

These results illustrate the use of automated phenotyping of cardiac imaging data from the general population to investigate the genetic etiology of aortic valve disease, perform clinical prediction, and uncover new clinical and genetic correlates of cardiac anatomy.

Is stopping secondary prophylaxis safe in HIV-positive talaromycosis patients? Experience from Myanmar

Objectives

The aim of the study was to determine whether it is safe to stop secondary prophylaxis in patients with talaromycosis after immune reconstitution with a sustained increase in CD4 count to ≥ 100 cells/µL after antiretroviral therapy (ART).

Methods

A retrospective cohort analysis was performed in HIV‐infected patients treated for talaromycosis between June 2009 and June 2017 in Medical Action Myanmar (MAM) clinics.

Results

Among a cohort of 5466 HIV‐infected patients, 41 patients were diagnosed with and treated for clinical talaromycosis. All the patients were on ART and had a CD4 count < 100 cells/µL. Of these 41 patients, 24 patients (71%) were skin smear positive for talaromycosis, while results were negative in 17 patients. Median CD4 count and haemoglobin concentration were 24 cells/µL and 7.7 g/dL, respectively. Seventy‐three per cent (30) were male. Among the 41 patients, 11 (27%) died and six (15%) were transferred to other centres. Twenty‐four patients (58% of the total diagnosed) stopped itraconazole secondary prophylaxis after starting active ART with CD4 counts > 100 cells/µL for at least 1 year. Throughout the duration of follow‐up post itraconazole cessation, the observed incidence of relapse was zero with a total follow‐up of 93.8 person‐years (95% confidence interval 0–4 per 100 person‐years). The median (25th, 75th percentile) duration of follow‐up post‐prophylaxis discontinuation was 2.8 (2.1, 6.3) years.

Conclusions

Secondary prophylaxis can be safely stopped in patients with talaromycosis after immune reconstitution with a sustained increase in CD4 count to ≥ 100 cells/µL after highly active antiretroviral therapy.

Silencing of MYH7 ameliorates disease phenotypes in human iPSC-cardiomyocytes

Allele-specific RNA silencing has been shown to be an effective therapeutic treatment in a number of diseases, including neurodegenerative disorders. Studies of allele-specific silencing in hypertrophic cardiomyopathy (HCM) to date have focused on mouse models of disease. We here examine allele-specific silencing in a human-cell model of HCM. We investigate two methods of silencing, short hairpin RNA (shRNA) and antisense oligonucleotide (ASO) silencing, using a human induced pluripotent stem cell-derived cardiomyocyte (hiPSC-CM) model. We used cellular micropatterning devices with traction force microscopy and automated video analysis to examine each strategy's effects on contractile defects underlying disease. We find that shRNA silencing ameliorates contractile phenotypes of disease, reducing disease-associated increases in cardiomyocyte velocity, force, and power. We find that ASO silencing, while better able to target and knockdown a specific disease-associated allele, showed more modest improvements in contractile phenotypes. These findings are the first exploration of allele-specific silencing in a human HCM model and provide a foundation for further exploration of silencing as a therapeutic treatment for MYH7-mutation-associated cardiomyopathy.

The Medical Genome Initiative: moving whole-genome sequencing for rare disease diagnosis to the clinic

Clinical whole-genome sequencing (WGS) offers clear diagnostic benefits for patients with rare disease. However, there are barriers to its widespread adoption, including a lack of standards for clinical practice. The Medical Genome Initiative consortium was formed to provide practical guidance and support the development of standards for the use of clinical WGS.

Impact of the distance from the chest wall to the heart on surface ECG voltage in athletes

Objective

Available ECG criteria for detection of left ventricular (LV) hypertrophy have been reported to have limited diagnostic capability. Our goal was to describe how the distance between the chest wall and the left ventricle determined by echocardiography affected the relationship between ECG voltage and LV mass (LVM) in athletes.

Methods

We retrospectively evaluated digitised ECG data from college athletes undergoing routine echocardiography as part of their preparticipation evaluation. Along with LV mass and volume, we determined the chest wall–LV distance in the parasternal short-axis and long-axis views from two-dimensional transthoracic echocardiographic images and explored the relation with ECG QRS voltages in all leads, as well as summed voltages as included in six major ECG-LVH criteria.

Results

239 athletes (43 women) were included (age 19±1 years). In men, greater LV–chest wall distance was associated with higher R-wave amplitudes in leads aVL and I (R=0.20 and R=0.25, both p<0.01), while in women greater distance was associated with higher R-amplitudes in V5 and V6 (R=0.42 and R=0.34, both p<0.01). In women, the chest wall–LV distance was the only variable independently (and positively) associated with R V5 voltage, while LVM, height and weight contributed to the relationship in men.

Conclusions

The chest wall–LV distance was weakly associated with ECG voltage in athletes. Inconsistent associations in men and women imply different intrathoracic factors affecting impedance and conductance between sexes. This may help explain the poor relationship between QRS voltage and LVM in athletes.

Classifying Drugs by their Arrhythmogenic Risk Using Machine Learning

All medications have adverse effects. Among the most serious of these are cardiac arrhythmias. Current paradigms for drug safety evaluation are costly, lengthy, conservative, and impede efficient drug development. Here, we combine multiscale experiment and simulation, high-performance computing, and machine learning to create a risk estimator to stratify new and existing drugs according to their proarrhythmic potential. We capitalize on recent developments in machine learning and integrate information across 10 orders of magnitude in space and time to provide a holistic picture of the effects of drugs, either individually or in combination with other drugs. We show, both experimentally and computationally, that drug-induced arrhythmias are dominated by the interplay between two currents with opposing effects: the rapid delayed rectifier potassium current and the L-type calcium current. Using Gaussian process classification, we create a classifier that stratifies drugs into safe and arrhythmic domains for any combinations of these two currents. We demonstrate that our classifier correctly identifies the risk categories of 22 common drugs exclusively on the basis of their concentrations at 50% current block. Our new risk assessment tool explains under which conditions blocking the L-type calcium current can delay or even entirely suppress arrhythmogenic events. Using machine learning in drug safety evaluation can provide a more accurate and comprehensive mechanistic assessment of the proarrhythmic potential of new drugs. Our study paves the way toward establishing science-based criteria to accelerate drug development, design safer drugs, and reduce heart rhythm disorders.

Myosin Sequestration Regulates Sarcomere Function, Cardiomyocyte Energetics, and Metabolism, Informing the Pathogenesis of Hypertrophic Cardiomyopathy

BACKGROUND

Hypertrophic cardiomyopathy (HCM) is caused by pathogenic variants in sarcomere protein genes that evoke hypercontractility, poor relaxation, and increased energy consumption by the heart and increased patient risks for arrhythmias and heart failure. Recent studies show that pathogenic missense variants in myosin, the molecular motor of the sarcomere, are clustered in residues that participate in dynamic conformational states of sarcomere proteins. We hypothesized that these conformations are essential to adapt contractile output for energy conservation and that pathophysiology of HCM results from destabilization of these conformations.

METHODS

We assayed myosin ATP binding to define the proportion of myosins in the super relaxed state (SRX) conformation or the disordered relaxed state (DRX) conformation in healthy rodent and human hearts, at baseline and in response to reduced hemodynamic demands of hibernation or pathogenic HCM variants. To determine the relationships between myosin conformations, sarcomere function, and cell biology, we assessed contractility, relaxation, and cardiomyocyte morphology and metabolism, with and without an allosteric modulator of myosin ATPase activity. We then tested whether the positions of myosin variants of unknown clinical significance that were identified in patients with HCM, predicted functional consequences and associations with heart failure and arrhythmias.

RESULTS

Myosins undergo physiological shifts between the SRX conformation that maximizes energy conservation and the DRX conformation that enables cross-bridge formation with greater ATP consumption. Systemic hemodynamic requirements, pharmacological modulators of myosin, and pathogenic myosin missense mutations influenced the proportions of these conformations. Hibernation increased the proportion of myosins in the SRX conformation, whereas pathogenic variants destabilized these and increased the proportion of myosins in the DRX conformation, which enhanced cardiomyocyte contractility, but impaired relaxation and evoked hypertrophic remodeling with increased energetic stress. Using structural locations to stratify variants of unknown clinical significance, we showed that the variants that destabilized myosin conformations were associated with higher rates of heart failure and arrhythmias in patients with HCM.

CONCLUSIONS

Myosin conformations establish work-energy equipoise that is essential for life-long cellular homeostasis and heart function. Destabilization of myosin energy-conserving states promotes contractile abnormalities, morphological and metabolic remodeling, and adverse clinical outcomes in patients with HCM. Therapeutic restabilization corrects cellular contractile and metabolic phenotypes and may limit these adverse clinical outcomes in patients with HCM.

Classifying Drugs by their Arrhythmogenic Risk Using Machine Learning

All medications have adverse effects. Among the most serious of these are cardiac arrhythmias. Current paradigms for drug safety evaluation are costly, lengthy, conservative, and impede efficient drug development. Here, we combine multiscale experiment and simulation, high-performance computing, and machine learning to create a risk estimator to stratify new and existing drugs according to their proarrhythmic potential. We capitalize on recent developments in machine learning and integrate information across 10 orders of magnitude in space and time to provide a holistic picture of the effects of drugs, either individually or in combination with other drugs. We show, both experimentally and computationally, that drug-induced arrhythmias are dominated by the interplay between two currents with opposing effects: the rapid delayed rectifier potassium current and the L-type calcium current. Using Gaussian process classification, we create a classifier that stratifies drugs into safe and arrhythmic domains for any combinations of these two currents. We demonstrate that our classifier correctly identifies the risk categories of 22 common drugs exclusively on the basis of their concentrations at 50% current block. Our new risk assessment tool explains under which conditions blocking the L-type calcium current can delay or even entirely suppress arrhythmogenic events. Using machine learning in drug safety evaluation can provide a more accurate and comprehensive mechanistic assessment of the proarrhythmic potential of new drugs. Our study paves the way toward establishing science-based criteria to accelerate drug development, design safer drugs, and reduce heart rhythm disorders.

Limitations of Electrocardiography for Detecting Left Ventricular Hypertrophy or Concentric Remodeling in Athletes

BACKGROUND

Electrocardiography (ECG) is used to screen for left ventricular hypertrophy (LVH), but common ECG-LVH criteria have been found less effective in athletes. The purpose of this study was to comprehensively evaluate the value of ECG for identifying athletes with LVH or a concentric cardiac phenotype.

METHODS

A retrospective analysis of 196 male Division I college athletes routinely screened with ECG and echocardiography within the Stanford Athletic Cardiovascular Screening Program was performed. Left-ventricular mass and volume were determined using echocardiography. LVH was defined as left ventricular mass (LVM) >102 g/m²; a concentric cardiac phenotype as LVM-to-volume (M/V) ≥1.05 g/mL. Twelve-lead electrocardiograms including high-resolution time intervals and QRS voltages were obtained. Thirty-seven previously published ECG-LVH criteria were applied, of which the majority have never been evaluated in athletes. C-statistics, including area under the receiver operating curve (AUC) and likelihood ratios were calculated.

RESULTS

ECG lead voltages were poorly associated with LVM (r = 0.18-0.30) and M/V (r = 0.15-0.25). The proportion of athletes with ECG-LVH was 0%-74% across criteria, with sensitivity and specificity ranging between 0% and 91% and 27% and 99.5%, respectively. The average AUC of the criteria in identifying the 11 athletes with LVH was 0.57 (95% confidence interval [CI] 0.56-0.59), and the average AUC for identifying the 8 athletes with a concentric phenotype was 0.59 (95% CI 0.56-0.62).

CONCLUSION

The diagnostic capacity of all ECG-LVH criteria were inadequate and, therefore, not clinically useful in screening for LVH or a concentric phenotype in athletes. This is probably due to the weak association between LVM and ECG voltage.

Echocardiographic Assessment of Left Ventricular Remodeling in American Style Footballers

Several athletic programs incorporate echocardiography during pre-participation screening of American Style Football (ASF) players with great variability in reported echocardiographic values. Pre-participation screening was performed in National Collegiate Athletic Association Division I ASF players from 2008 to 2016 at the Division of Sports Cardiology. The echocardiographic protocol focused on left ventricular (LV) mass, mass-to-volume ratio, sphericity, ejection fraction, and longitudinal Lagrangian strain. LV mass was calculated using the area-length method in end-diastole and end-systole. A total of two hundred and thirty players were included (18±1 years, 57% were Caucasian, body mass index 29±4 kg/m²) after four players (2%) were excluded for pathological findings. Although there was no difference in indexed LV mass by race (Caucasian 78±11 vs. African American 81±10 g/m², p=0.089) or sphericity (Caucasian 1.81±0.13 vs. African American 1.78±0.14, p=0.130), the mass-to-volume ratio was higher in African Americans (0.91±0.09 vs. 0.83±0.08, p<0.001). No race-specific differences were noted in LV longitudinal Lagrangian strain. Player position appeared to have a limited role in defining LV remodeling. In conclusion, significant echocardiographic differences were observed in mass-to-volume ratio between African American and Caucasian players. These demographics should be considered as part of pre-participation screening.

Prevalence and Progression of Late Gadolinium Enhancement in Children and Adolescents With Hypertrophic Cardiomyopathy

BACKGROUND

Late gadolinium enhancement (LGE) on cardiac magnetic resonance imaging (CMR) is believed to represent dense replacement fibrosis. It is seen in ≈60% of adult patients with hypertrophic cardiomyopathy (HCM). However, the prevalence of LGE in children and adolescents with HCM is not well established. In addition, longitudinal studies describing the development and evolution of LGE in pediatric HCM are lacking. This study assesses the prevalence, progression, and clinical correlations of LGE in children and adolescents with, or genetically predisposed to, HCM.

METHODS

CMR scans from 195 patients ≤21 years of age were analyzed in an observational, retrospective study, including 155 patients with overt HCM and 40 sarcomere mutation carriers without left ventricular (LV) hypertrophy. The extent of LGE was quantified by measuring regions with signal intensity >6 SD above nulled remote myocardium.

RESULTS

Patients were 14.3±4.5 years of age at baseline and 68% were male. LGE was present in 70 (46%) patients with overt HCM (median extent, 3.3%; interquartile range, 0.8-7.1%), but absent in mutation carriers without LV hypertrophy. Thirty-one patients had >1 CMR (median interval between studies, 2.4 years; interquartile range, 1.5-3.2 years). LGE was detected in 13 patients (42%) at baseline and in 16 patients (52%) at follow-up CMR. The median extent of LGE increased by 2.4 g/y (range, 0-13.2 g/y) from 2.9% (interquartile range, 0.8-3.2%) of LV mass to 4.3% (interquartile range, 2.9-6.8%) ( P=0.02). In addition to LGE, LV mass and left atrial volume, indexed to body surface area, and z score for LV mass, as well, increased significantly from first to most recent CMR.

CONCLUSIONS

LGE was present in 46% of children and adolescents with overt HCM, in contrast to ≈60% typically reported in adult HCM. In the subset of patients with serial imaging, statistically significant increases in LGE, LV mass, and left atrial size were detected over 2.5 years, indicating disease progression over time. Further prospective studies are required to confirm these findings and to better understand the clinical implications of LGE in pediatric HCM.

A novel noninvasive method for remote heart failure monitoring: the EuleriAn video Magnification apPLications In heart Failure studY (AMPLIFY)

Current remote monitoring devices for heart failure have been shown to reduce hospitalizations but are invasive and costly; accurate non-invasive options remain limited. The EuleriAn Video Magnification ApPLications In Heart Failure StudY (AMPLIFY) pilot aimed to evaluate the accuracy of a novel noninvasive method that uses Eulerian video magnification. Video recordings were performed on the neck veins of 50 patients who were scheduled for right heart catheterization at the Palo Alto VA Medical Center. The recorded jugular venous pulsations were then enhanced by applying Eulerian phase-based motion magnification. Assessment of jugular venous pressure was compared across three categories: (1) physicians who performed bedside exams, (2) physicians who reviewed both the amplified and unamplified videos, and (3) direct invasive measurement of right atrial pressure from right heart catheterization. Motion magnification reduced inaccuracy of the clinician assessment of central venous pressure compared to the gold standard of right heart catheterization (mean discrepancy of −0.80 cm H₂O; 95% CI −2.189 to 0.612, p = 0.27) when compared to both unamplified video (−1.84 cm H₂O; 95% CI −3.22 to −0.46, p = 0.0096) and the bedside exam (−2.90 cm H₂O; 95% CI −4.33 to 1.40, p = 0.0002). Major categorical disagreements with right heart catheterization were significantly reduced with motion magnification (12%) when compared to unamplified video (25%) or the bedside exam (27%). This novel method of assessing jugular venous pressure improves the accuracy of the clinical exam and may enable accurate remote monitoring of heart failure patients with minimal patient risk.

Artificial Intelligence in Cardiology

Artificial intelligence and machine learning are poised to influence nearly every aspect of the human condition, and cardiology is not an exception to this trend. This paper provides a guide for clinicians on relevant aspects of artificial intelligence and machine learning, reviews selected applications of these methods in cardiology to date, and identifies how cardiovascular medicine could incorporate artificial intelligence in the future. In particular, the paper first reviews predictive modeling concepts relevant to cardiology such as feature selection and frequent pitfalls such as improper dichotomization. Second, it discusses common algorithms used in supervised learning and reviews selected applications in cardiology and related disciplines. Third, it describes the advent of deep learning and related methods collectively called unsupervised learning, provides contextual examples both in general medicine and in cardiovascular medicine, and then explains how these methods could be applied to enable precision cardiology and improve patient outcomes.

Time based versus strain based myocardial performance indices in hypertrophic cardiomyopathy, the merging role of left atrial strain

AIMS

The myocardial performance index (MPI) is a time-based index of global myocardial performance. In this study, we sought to compare the prognostic value of the MPI with other strain and remodelling indices in hypertrophic cardiomyopathy (HCM).

METHODS AND RESULTS

We enrolled 126 patients with HCM and 50 age- and sex-matched controls. Along with traditional echocardiographic assessment, MPI, left ventricular global longitudinal strain (LVGLS), E/e' ratio, and total left atrial (LA) global strain (LAS) were also measured. Time-based MPI was calculated from flow or tissue-based pulse wave Doppler (PWD and TDI) as the (isovolumic-relaxation and contraction time)/systolic-time. We used hierarchical clustering and network analysis to better visualize the relationship between parameters. The primary endpoint was the composite of all-cause death, heart transplantation, left ventricular assist device implantation, and clinical worsening. Left ventricular outflow tract (LVOT) obstruction was present in 56% of patients. Compared with controls, patients with HCM had worse LVGLS (-14.0 ± 3.4% vs. -19.6 ± 1.5%), higher E/e' (12.9 ± 7.2 vs. 6.1 ± 1.5), LA volume index (LAVI) (36.4 ± 13.8 ml/m2 vs. 25.6 ± 6.7 ml/m2), and MPI (0.55 ± 0.17 vs. 0.40 ± 0.11 for PWD and 0.59 ± 0.22 vs. 0.46 ± 0.09 for TDI) (all P < 0.001). During a median follow-up of 55 months, 47 endpoints occurred. PWD or TDI-based MPI was not associated with outcome, while LAVI, LAS, LVGLS, and E/e' were (all P < 0.01). On multivariable analysis, LVOT obstruction (P < 0.001), LAS (P < 0.001), and E/e' (P = 0.02) were retained as independent associates. They were in different clusters suggesting complemental relationship between them.

CONCLUSION

Time-based index is less predictive of outcome than strain or tissue Doppler indices. LAS may be a promising prognostic marker in HCM.

Approaching Higher Dimension Imaging Data Using Cluster-Based Hierarchical Modeling in Patients with Heart Failure Preserved Ejection Fraction

Heart failure with preserved ejection fraction (HFpEF) is a major cause of morbidity and mortality, accounting for the majority of heart failure (HF) hospitalization. To identify the most complementary predictors of mortality among clinical, laboratory and echocardiographic data, we used cluster based hierarchical modeling. Using Stanford Translational Research Database, we identified patients hospitalized with HFpEF between 2005 and 2016 in whom echocardiogram and NT-proBNP were both available at the time of admission. Comprehensive echocardiographic assessment including left ventricular longitudinal strain (LVLS), right ventricular function and right ventricular systolic pressure (RVSP) was performed. The outcome was defined as all-cause mortality. Among patients identified, 186 patients with complete echocardiographic assessment were included in the analysis. The cohort included 58% female, with a mean age of 78.7 ± 13.5 years, LVLS of −13.3 ± 2.5%, an estimated RVSP of 38 ± 13 mmHg. Unsupervised cluster analyses identified six clusters including ventricular systolic-function cluster, diastolic-hemodynamic cluster, end-organ function cluster, vital-sign cluster, complete blood count and sodium clusters. Using a stepwise hierarchical selection from each cluster, we identified NT-proBNP (standard hazard ratio [95%CI] = 1.56 [1.17–2.08]) and RVSP (1.37 [1.09–1.78]) as independent correlates of outcome. When adding these parameters to the well validated Get with the Guideline Heart Failure risk score, the Chi-square was significantly improved (p = 0.01). In conclusion, NT-proBNP and RVSP were independently predictive in HFpEF among clinical, imaging, and biomarker parameters. Cluster-based hierarchical modeling may help identify the complementally predictive parameters in small cohorts with higher dimensional clinical data.

Weakly supervised classification of aortic valve malformations using unlabeled cardiac MRI sequences

Biomedical repositories such as the UK Biobank provide increasing access to prospectively collected cardiac imaging, however these data are unlabeled, which creates barriers to their use in supervised machine learning. We develop a weakly supervised deep learning model for classification of aortic valve malformations using up to 4,000 unlabeled cardiac MRI sequences. Instead of requiring highly curated training data, weak supervision relies on noisy heuristics defined by domain experts to programmatically generate large-scale, imperfect training labels. For aortic valve classification, models trained with imperfect labels substantially outperform a supervised model trained on hand-labeled MRIs. In an orthogonal validation experiment using health outcomes data, our model identifies individuals with a 1.8-fold increase in risk of a major adverse cardiac event. This work formalizes a deep learning baseline for aortic valve classification and outlines a general strategy for using weak supervision to train machine learning models using unlabeled medical images at scale.

Comparison of QT Interval Measurement Methods and Correction Formulas in Atrial Fibrillation

Antiarrhythmic drugs used in atrial fibrillation (AF) cause QT prolongation and are associated with torsades de pointes, a deadly ventricular arrhythmia. No consensus exists on the optimal method of QT measurement or correction in AF. Therefore, we compared common methods to measure and correct QT in AF to identify the most accurate approach. We identified patients who had electrocardiograms done at Stanford Hospital (Stanford, California) between January 2014 and October 2016 with conversion from AF to sinus rhythm (SR) within a 24-hour period. QT intervals were determined using different measurement methods and corrected using the Bazett's, Framingham, Fridericia, or Hodges formulas for heart rate (HR). Comparisons were made between QT in a patient's last instance of AF to SR. Computerized measurements were taken from 715 patients. Manual measurements were taken from a 50-patient subset. Bazett's formula produced the longest corrected QT in AF compared with other formulas (p <0.005). Measuring QT as an average over multiple beats resulted in a smaller difference between AF and SR than choosing a single beat. Determining QT from a 5-beat average resulted in a QTc that was 19.0 ms higher (interquartile range 0.30 to 43.7) in AF than SR. After correcting for residual effect of HR on QTc, there was not a significant difference between QTc in AF to SR. In conclusion, measuring QT over multiple beats produces a more accurate measurement of QT in AF. Differences between QTc in AF and SR exist because of imperfect HR correction formula and not due to an independent effect of AF.

Blood pressure in athletic preparticipation evaluation and the implication for cardiac remodelling

OBJECTIVES

To explore blood pressure (BP) in athletes at preparticipation evaluation (PPE) in the context of recently updated US and European hypertension guidelines, and to determine the relationship between BP and left ventricular (LV) remodelling.

METHODS

In this retrospective study, athletes aged 13-35 years who underwent PPE facilitated by the Stanford Sports Cardiology programme were considered. Resting BP was measured in both arms; repeated once if ≥140/90 mm Hg. Athletes with abnormal ECGs or known hypertension were excluded. BP was categorised per US/European hypertension guidelines. In a separate cohort of athletes undergoing routine PPE echocardiography, we explored the relationship between BP and LV remodelling (LV mass, mass/volume ratio, sphericity index) and LV function.

RESULTS

In cohort 1 (n=2733, 65.5% male), 34.3% of athletes exceeded US hypertension thresholds. Male sex (B=3.17, p<0.001), body mass index (BMI) (B=0.80, p<0.001) and height (B=0.25, p<0.001) were the strongest independent correlates of systolic BP. In the second cohort (n=304, ages 17-26), systolic BP was an independent correlate of LV mass/volume ratio (B=0.002, p=0.001). LV longitudinal strain was similar across BP categories, while higher BP was associated with slower early diastolic relaxation.

CONCLUSION

In a large contemporary cohort of athletes, one-third presented with BP levels above the current US guidelines' thresholds for hypertension, highlighting that lowering the BP thresholds at PPE warrants careful consideration as well as efforts to standardise measurements. Higher systolic BP was associated with male sex, BMI and height and with LV remodelling and diastolic function, suggesting elevated BP in athletes during PPE may signify a clinically relevant condition.