Novel Role for Cardiolipin as a Target of Therapy to Mitigate Myocardial Injury Caused by Venoarterial Extracorporeal Membrane Oxygenation.

BACKGROUND: Cardiolipin is a mitochondrial-specific phospholipid that maintains integrity of the electron transport chain (ETC) and plays a central role in myocardial ischemia/reperfusion injury. Tafazzin is an enzyme that is required for cardiolipin maturation. Venoarterial extracorporeal membrane oxygenation (VA-ECMO) use to provide hemodynamic support for acute myocardial infarction has grown exponentially, is associated with poor outcomes, and is under active clinical investigation, yet the mechanistic effect of VA-ECMO on myocardial damage in acute myocardial infarction remains poorly understood. We hypothesized that VA-ECMO acutely depletes myocardial cardiolipin and exacerbates myocardial injury in acute myocardial infarction. METHODS: We examined cardiolipin and tafazzin levels in human subjects with heart failure and healthy swine exposed to VA-ECMO and used a swine model of closed-chest myocardial ischemia/reperfusion injury to evaluate the effect of VA-ECMO on cardiolipin expression, myocardial injury, and mitochondrial function. RESULTS: Cardiolipin and tafazzin levels are significantly reduced in the left ventricles of individuals requiring VA-ECMO compared with individuals without VA-ECMO before heart transplantation. Six hours of exposure to VA-ECMO also decreased left ventricular levels of cardiolipin and tafazzin in healthy swine compared with sham controls. To explore whether cardiolipin depletion by VA-ECMO increases infarct size, we performed left anterior descending artery occlusion for a total of 120 minutes followed by 180 minutes of reperfusion in adult swine in the presence and absence of MTP-131, an amphipathic molecule that interacts with cardiolipin to stabilize the inner mitochondrial membrane. Compared with reperfusion alone, VA-ECMO activation beginning after 90 minutes of left anterior descending artery occlusion increased infarct size (36±8% versus 48±7%; P<0.001). VA-ECMO also decreased cardiolipin and tafazzin levels, disrupted mitochondrial integrity, reduced electron transport chain function, and promoted oxidative stress. Compared with reperfusion alone or VA-ECMO before reperfusion, delivery of MTP-131 before VA-ECMO activation reduced infarct size (22±8%; P=0.03 versus reperfusion alone and P<0.001 versus VA-ECMO alone). MTP-131 restored cardiolipin and tafazzin levels, stabilized mitochondrial function, and reduced oxidative stress in the left ventricle. CONCLUSIONS: We identified a novel mechanism by which VA-ECMO promotes myocardial injury and further identify cardiolipin as an important target of therapy to reduce infarct size and to preserve mitochondrial function in the setting of VA-ECMO for acute myocardial infarction.

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.

Circulating Proteome Analysis Identifies Reduced Inflammation After Initiation of Hemodynamic Support with Either Veno-Arterial Extracorporeal Membrane Oxygenation or Impella in Patients with Cardiogenic Shock.

In-hospital mortality associated with cardiogenic shock (CS) remains high despite the use of percutaneous assist devices. We sought to determine whether support with VA-ECMO or Impella in patients with CS alters specific components of the plasma proteome. Plasma samples were collected before device implantation and 72 h after initiation of support in 11 CS patients receiving ECMO or Impella. SOMAscan was used to detect 1305 circulating proteins. Sixty-seven proteins were changed after ECMO (18 upregulated and 49 downregulated, p < 0.05), 38 after Impella (10 upregulated and 28 downregulated, p < 0.05), and only eight proteins were commonly affected. Despite minimal protein overlap, both devices were associated with markers of reduced inflammation and increased apoptosis of inflammatory cells. In summary, ECMO and Impella are associated with reduced expression of inflammatory markers and increased markers of inflammatory cell death. These circulating proteins may serve as novel targets of therapy or biomarkers to tailor AMCS use.

Maximum Likelihood Estimation of Flexible Survival Densities with Importance Sampling.

Survival analysis is a widely-used technique for analyzing time-to-event data in the presence of censoring. In recent years, numerous survival analysis methods have emerged which scale to large datasets and relax traditional assumptions such as proportional hazards. These models, while being performant, are very sensitive to model hyperparameters including: (1) number of bins and bin size for discrete models and (2) number of cluster assignments for mixture-based models. Each of these choices requires extensive tuning by practitioners to achieve optimal performance. In addition, we demonstrate in empirical studies that: (1) optimal bin size may drastically differ based on the metric of interest (e.g., concordance vs brier score), and (2) mixture models may suffer from mode collapse and numerical instability. We propose a survival analysis approach which eliminates the need to tune hyperparameters such as mixture assignments and bin sizes, reducing the burden on practitioners. We show that the proposed approach matches or outperforms baselines on several real-world datasets.

ALK1 Deficiency Impairs the Wound-Healing Process and Increases Mortality in Murine Model of Myocardial Infarction.

The functional role of TGFß type I receptor, activin-like kinase (ALK)-1 in post-myocardial infarction (MI) cardiac remodeling is unknown. We hypothesize that reduced ALK1 activity reduces survival and promotes cardiac fibrosis after MI. MI was induced in wild-type (WT), and ALK+/- mice by left coronary ligation. After 14 days ALK1+/- mice had reduced survival with a higher rate of cardiac rupture compared to WT mice. ALK1+/- left ventricles (LVs) had increased volumes at the end of systole and at the end of diastole. After MI ALK1+/- LVs had increased profibrotic SMAD3 signaling, type 1 collagen, and fibrosis as well as increased levels of TGFß1 co-receptor, endoglin, VEGF, and ALK1 ligands BMP9 and BMP10. ALK1+/- LVs had decreased levels of stromal-derived factor 1α. These data identify the critical role of ALK1 in post-MI survival and cardiac remodeling and implicate ALK1 as a potential therapeutic target to improve survival after MI.

Loss of Bone Morphogenetic Protein-9 Reduces Survival and Increases MMP Activity After Myocardial Infarction.

No studies have explored a functional role for bone morphogenetic protein (BMP)-9, a transforming growth factor-ß superfamily ligand, in cardiac remodeling after myocardial infarction (MI). Using BMP-9 null mice, we observed that loss of BMP-9 decreases survival and increases cardiac rupture after MI. We further observed that loss of BMP-9 not only increases collagen abundance, but also promotes matrix metalloproteinase-9 activity and collagen degradation after MI. These findings identify BMP-9 as a necessary component of cardiac remodeling after MI and a potentially important target of therapy to improve outcomes after MI.

Transvalvular Unloading Mitigates Ventricular Injury Due to Venoarterial Extracorporeal Membrane Oxygenation in Acute Myocardial Infarction.

Whether extracorporeal membrane oxygenation (ECMO) with Impella, known as EC-Pella, limits cardiac damage in acute myocardial infarction remains unknown. The authors now report that the combination of transvalvular unloading and ECMO (EC-Pella) initiated before reperfusion reduced infarct size compared with ECMO alone before reperfusion in a preclinical model of acute myocardial infarction. EC-Pella also reduced left ventricular pressure-volume area when transvalvular unloading was applied before, not after, activation of ECMO. The authors further observed that EC-Pella increased cardioprotective signaling but failed to rescue mitochondrial dysfunction compared with ECMO alone. These findings suggest that ECMO can increase infarct size in acute myocardial infarction and that EC-Pella can mitigate this effect but also suggest that left ventricular unloading and myocardial salvage may be uncoupled in the presence of ECMO in acute myocardial infarction. These observations implicate mechanisms beyond hemodynamic load as part of the injury cascade associated with ECMO in acute myocardial infarction.

Assessing Phenotype Definitions for Algorithmic Fairness.

Phenotyping is a core, routine activity in observational health research. Cohorts impact downstream analyses, such as how a condition is characterized, how patient risk is defined, and what treatments are studied. It is thus critical to ensure that cohorts are representative of all patients, independently of their demographics or social determinants of health. In this paper, we propose a set of best practices to assess the fairness of phenotype definitions. We leverage established fairness metrics commonly used in predictive models and relate them to commonly used epidemiological metrics. We describe an empirical study for Crohn's disease and diabetes type 2, each with multiple phenotype definitions taken from the literature across gender and race. We show that the different phenotype definitions exhibit widely varying and disparate performance according to the different fairness metrics and subgroups. We hope that the proposed best practices can help in constructing fair and inclusive phenotype definitions.

Myocardial Injury Promotes Matrix Metalloproteinase-9 Activity in the Renal Cortex in Preclinical Models of Acute Myocardial Infarction.

New mechanistic insight into how the kidney responds to cardiac injury during acute myocardial infarction (AMI) is required. We hypothesized that AMI promotes inflammation and matrix metalloproteinase-9 (MMP9) activity in the kidney and studied the effect of initiating an Impella CP or veno-arterial extracorporeal membrane oxygenation (VA-ECMO) before coronary reperfusion during AMI. Adult male swine were subjected to coronary occlusion and either reperfusion (ischemia-reperfusion; IR) or support with either Impella or VA-ECMO before reperfusion. IR and ECMO increased while Impella reduced levels of MMP-9 in the myocardial infarct zone, circulation, and renal cortex. Compared to IR, Impella reduced myocardial infarct size and urinary KIM-1 levels, but VA-ECMO did not. IR and VA-ECMO increased pro-fibrogenic signaling via transforming growth factor-beta and endoglin in the renal cortex, but Impella did not. These findings identify that AMI increases inflammatory activity in the kidney, which may be attenuated by Impella support.

Deep Learning Electrocardiographic Analysis for Detection of Left-Sided Valvular Heart Disease.

BACKGROUND: Valvular heart disease is an important contributor to cardiovascular morbidity and mortality and remains underdiagnosed. Deep learning analysis of electrocardiography (ECG) may be useful in detecting aortic stenosis (AS), aortic regurgitation (AR), and mitral regurgitation (MR). OBJECTIVES: This study aimed to develop ECG deep learning algorithms to identify moderate or severe AS, AR, and MR alone and in combination. METHODS: A total of 77,163 patients undergoing ECG within 1 year before echocardiography from 2005-2021 were identified and split into train (n = 43,165), validation (n = 12,950), and test sets (n = 21,048; 7.8% with any of AS, AR, or MR). Model performance was assessed using area under the receiver-operating characteristic (AU-ROC) and precision-recall curves. Outside validation was conducted on an independent data set. Test accuracy was modeled using different disease prevalence levels to simulate screening efficacy using the deep learning model. RESULTS: The deep learning algorithm model accuracy was as follows: AS (AU-ROC: 0.88), AR (AU-ROC: 0.77), MR (AU-ROC: 0.83), and any of AS, AR, or MR (AU-ROC: 0.84; sensitivity 78%, specificity 73%) with similar accuracy in external validation. In screening program modeling, test characteristics were dependent on underlying prevalence and selected sensitivity levels. At a prevalence of 7.8%, the positive and negative predictive values were 20% and 97.6%, respectively. CONCLUSIONS: Deep learning analysis of the ECG can accurately detect AS, AR, and MR in this multicenter cohort and may serve as the basis for the development of a valvular heart disease screening program.

Human SMAD4 Genomic Variants Identified in Individuals with Heritable and Early-Onset Thoracic Aortic Disease.

Thoracic aortic aneurysms (TAAs) that progress to acute thoracic aortic dissections (TADs) are life threatening vascular events that have been associated with altered transforming growth factor (TGF) ß signaling. In addition to TAA, multiple genetic vascular disorders, including hereditary hemorrhagic telangiectasia (HHT), involve altered TGFß signaling and vascular malformations. Due to the importance of TGFß, genomic variant databases have been curated for activin receptor-like kinase 1 (ALK1) and endoglin (ENG). This case report details seven variants in SMAD4 that are associated with either heritable or early onset aortic dissections and compares them to pathogenic exon variants in gnomAD v2.1.1. The TAA and TAD variants were identified through whole exome sequencing of 346 unrelated heritable thoracic aortic disease (HTAD) and 355 individuals of early onset (age ≤ 56 years old) of thoracic aortic dissection (ESTAD). An allele frequency filter of less than 0.05% was applied in the Genome Aggregation Database (gnomAD exome v2.1.1) with a combined annotation dependent depletion score (CADD) greater than 20. These seven variants also have a higher REVEL score (>0.2), indicating pathogenic potential. Further in vivo and in vitro analysis is needed to evaluate how these variants affect mRNA stability and SMAD4 protein activity in association with thoracic aortic disease.

Development and validation of prediction models for mechanical ventilation, renal replacement therapy, and readmission in COVID-19 patients.

OBJECTIVE: Coronavirus disease 2019 (COVID-19) patients are at risk for resource-intensive outcomes including mechanical ventilation (MV), renal replacement therapy (RRT), and readmission. Accurate outcome prognostication could facilitate hospital resource allocation. We develop and validate predictive models for each outcome using retrospective electronic health record data for COVID-19 patients treated between March 2 and May 6, 2020. MATERIALS AND METHODS: For each outcome, we trained 3 classes of prediction models using clinical data for a cohort of SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2)-positive patients (n = 2256). Cross-validation was used to select the best-performing models per the areas under the receiver-operating characteristic and precision-recall curves. Models were validated using a held-out cohort (n = 855). We measured each model's calibration and evaluated feature importances to interpret model output. RESULTS: The predictive performance for our selected models on the held-out cohort was as follows: area under the receiver-operating characteristic curve-MV 0.743 (95% CI, 0.682-0.812), RRT 0.847 (95% CI, 0.772-0.936), readmission 0.871 (95% CI, 0.830-0.917); area under the precision-recall curve-MV 0.137 (95% CI, 0.047-0.175), RRT 0.325 (95% CI, 0.117-0.497), readmission 0.504 (95% CI, 0.388-0.604). Predictions were well calibrated, and the most important features within each model were consistent with clinical intuition. DISCUSSION: Our models produce performant, well-calibrated, and interpretable predictions for COVID-19 patients at risk for the target outcomes. They demonstrate the potential to accurately estimate outcome prognosis in resource-constrained care sites managing COVID-19 patients. CONCLUSIONS: We develop and validate prognostic models targeting MV, RRT, and readmission for hospitalized COVID-19 patients which produce accurate, interpretable predictions. Additional external validation studies are needed to further verify the generalizability of our results.

Zero-Shot Clinical Acronym Expansion via Latent Meaning Cells.

We introduce Latent Meaning Cells, a deep latent variable model which learns contextualized representations of words by combining local lexical context and metadata. Metadata can refer to granular context, such as section type, or to more global context, such as unique document ids. Reliance on metadata for contextualized representation learning is apropos in the clinical domain where text is semi-structured and expresses high variation in topics. We evaluate the LMC model on the task of zero-shot clinical acronym expansion across three datasets. The LMC significantly outperforms a diverse set of baselines at a fraction of the pre-training cost and learns clinically coherent representations. We demonstrate that not only is metadata itself very helpful for the task, but that the LMC inference algorithm provides an additional large benefit.

Transvalvular Ventricular Unloading Before Reperfusion in Acute Myocardial Infarction.

BACKGROUND: Myocardial damage due to acute ST-segment elevation myocardial infarction (STEMI) remains a significant global health problem. New approaches to limit myocardial infarct size and reduce progression to heart failure after STEMI are needed. Mechanically reducing left ventricular (LV) workload (LV unloading) before coronary reperfusion is emerging as a potential approach to reduce infarct size. OBJECTIVES: Given the central importance of mitochondria in reperfusion injury, we hypothesized that compared with immediate reperfusion (IR), LV unloading before reperfusion improves myocardial energy substrate use and preserves mitochondrial structure and function. METHODS: To explore the effect of LV unloading duration on infarct size, we analyzed data from the STEMI-Door to Unload (STEMI-DTU) trial and then tested the effect of LV unloading on ischemia and reperfusion injury, cardiac metabolism, and mitochondrial function in swine models of acute myocardial infarction. RESULTS: The duration of LV unloading before reperfusion was inversely associated with infarct size in patients with large anterior STEMI. In preclinical models, LV unloading reduced the expression of hypoxia-sensitive proteins and myocardial damage due to ischemia alone. LV unloading with a transvalvular pump (TV-P) but not with venoarterial extracorporeal membrane oxygenation (ECMO) reduced infarct size. Using unbiased and blinded metabolic profiling, TV-P improved myocardial energy substrate use and preserved mitochondrial structure including cardiolipin content after reperfusion compared with IR or ECMO. Functional testing in mitochondria isolated from the infarct zone showed an intact mitochondrial structure including cardiolipin content, preserved activity of the electron transport chain including mitochondrial complex I, and reduced oxidative stress with TV-P-supported reperfusion but not with IR or ECMO. CONCLUSIONS: These novel findings identify that transvalvular unloading limits ischemic injury before reperfusion, improves myocardial energy substrate use, and preserves mitochondrial structure and function after reperfusion.

CpG methylation and the methyl CpG binding protein 2 (MeCP2) are required for restraining corticotropin releasing hormone (CRH) gene expression.

The hypothalamic-pituitary-adrenal (HPA) axis plays a critical role in mounting a stress response and maintaining homeostasis. A dysregulated HPA axis and elevated levels of CRH are associated with a number of disorders. Although extensive research has been devoted to understanding molecular events associated with stimulated CRH gene, less is known about the mechanisms that restrain CRH expression. Using a cell culture system, we report here two molecular aspects of CRH gene regulation that are required for maintenance of basal level of CRH gene expression. These are a specific CpG methylation at a single CpG, and adequate levels of the methyl CpG binding protein 2 (MeCP2). The single site methylation allows the recruitment of MeCP2 to the CRH gene promoter region, and MeCP2 knockdown leads to increased expression of CRH gene. Taken together, the results indicate that site-specific methylation and MeCP2 are required for maintenance of basal levels of CRH gene expression.

Mechanisms by Which 17ß-Estradiol (E2) Suppress Neuronal cox-2 Gene Expression.

E2 attenuates inflammatory responses by suppressing expression of pro-inflammatory genes. Given that inflammation is increasingly being associated with neurodegenerative and psychiatric processes, we sought to elucidate mechanisms by which E2 down-regulates a component of an inflammatory response, cyclooxygenase- 2 (COX-2) expression. Although inflammatory processes in the brain are usually associated with microglia and astrocytes, we found that the COX-2 gene (cox-2) was expressed in a neuronal context, specifically in an amygdalar cell line (AR-5). Given that COX-2 has been reported to be in neurons in the brain, and that the amygdala is a site involved in neurodegenerative and neuropsychiatric processes, we investigated mechanisms by which E2 could down-regulate cox-2 expression in the AR-5 line. These cells express estrogen receptors alpha (ERα) and beta (ERß), and as shown here cox-2. At the level of RNA, E2 and the ERß selective ligand diarylpropionitrile (DPN) both attenuated gene expression, whereas the ERα selective ligand propyl pyrazole triol (PPT) had no effect. Neither ligand increased ERß at the cox-2 promoter. Rather, DPN decreased promoter occupancy of NF-κB p65 and histone 4 (H4) acetylation. Treatment with the non-specific HDAC inhibitor Trichostatin A (TSA) counteracted DPN's repressive effects on cox-2 expression. In keeping with the TSA effect, E2 and DPN increased histone deacetylase one (HDAC1) and switch-independent 3A (Sin3A) promoter occupancy. Lastly, even though E2 increased CpG methylation, DPN did not. Taken together, the pharmacological data indicate that ERß contributes to neuronal cox-2 expression, as measured by RNA levels. Furthermore, ER ligands lead to increased recruitment of HDAC1, Sin3A and a concomitant reduction of p65 occupancy and Ac-H4 levels. None of the events, however, are associated with a significant recruitment of ERß at the promoter. Thus, ERß directs recruitment to the cox-2 promoter, but does so in the absence of being recruited itself.

Dexamethasone induces a putative repressor complex and chromatin modifications in the CRH promoter.

Glucocorticoids down-regulate expression of hypothalamic CRH; however, mechanisms by which they do so are not fully understood. The proximal promoter cAMP response element, negative glucocorticoid response element (nGRE), and methylated CpG islands all play a role in crh down-regulation. Dexamethasone (Dex)-repressed crh expression is associated with glucocorticoid receptor (GR) and histone deacetylase 1 (HDAC1) recruitment to the region of the crh promoter. Given that HDAC1 may be present in methylated CpG binding protein 2 (MeCP2) complexes, and that MeCP2 is known to play a role in regulating crh expression, we sought to determine whether or not HDAC1 and/or MeCP2 could interact with the GR. Dex enhanced GR interactions with both proteins. Glucocorticoid regulation of crh has also been associated with CpG methylation; thus we assessed whether GR could interact with a DNA methyltransferase (DnMT). Indeed, the GR interacted with DnMT3b, but not DnMT3a. In addition, Dex-induced occupancy of the crh promoter by HDAC1, MeCP2, and DnMT3b was associated with an increased level of promoter methylation, which appeared to be CpG site specific. Lastly, to extend previous assessment of chromatin modifications in this promoter region, the degree of histone methylation was measured. Dex increased trimethylation of histone 3-lysine 9, a marker of gene suppression; however, levels of di- and trimethylated histone 3-lysine 4, markers of gene activation, were not significantly changed. Taken together, the data suggest that Dex-mediated crh suppression involves formation of a repressor complex consisting of GR, MeCP2, and HDAC1, recruitment of DnMT3b, and associated changes in proximal promoter CpG methylation.