SARS-CoV-2 pathogenesis in an angiotensin II-induced heart-on-a-chip disease model and extracellular vesicle screening.

Adverse cardiac outcomes in COVID-19 patients, particularly those with preexisting cardiac disease, motivate the development of human cell-based organ-on-a-chip models to recapitulate cardiac injury and dysfunction and for screening of cardioprotective therapeutics. Here, we developed a heart-on-a-chip model to study the pathogenesis of SARS-CoV-2 in healthy myocardium established from human induced pluripotent stem cell (iPSC)-derived cardiomyocytes and a cardiac dysfunction model, mimicking aspects of preexisting hypertensive disease induced by angiotensin II (Ang II). We recapitulated cytopathic features of SARS-CoV-2-induced cardiac damage, including progressively impaired contractile function and calcium handling, apoptosis, and sarcomere disarray. SARS-CoV-2 presence in Ang II-treated hearts-on-a-chip decreased contractile force with earlier onset of contractile dysfunction and profoundly enhanced inflammatory cytokines compared to SARS-CoV-2 alone. Toward the development of potential therapeutics, we evaluated the cardioprotective effects of extracellular vesicles (EVs) from human iPSC which alleviated the impairment of contractile force, decreased apoptosis, reduced the disruption of sarcomeric proteins, and enhanced beta-oxidation gene expression. Viral load was not affected by either Ang II or EV treatment. We identified MicroRNAs miR-20a-5p and miR-19a-3p as potential mediators of cardioprotective effects of these EVs.

Paracrine cross-talk between human adipose tissue-derived endothelial cells and perivascular cells accelerates the endothelialization of an electrospun ionomeric polyurethane scaffold.

The ex vivo endothelialization of small diameter vascular prostheses can prolong their patency. Here, we demonstrate that heterotypic interactions between human adipose tissue-derived endothelial cells and perivascular cells can be exploited to accelerate the endothelialization of an electrospun ionomeric polyurethane scaffold. The scaffold was used to physically separate endothelial cells from perivascular cells to prevent their diffuse neo-intimal hyperplasia and spontaneous tubulogenesis, yet enable their paracrine cross-talk to accelerate the integration of the endothelial cells into a temporally stable endothelial lining of a continuous, elongated, and aligned morphology. Perivascular cells stimulated endothelial basement membrane protein production and suppressed their angiogenic and inflammatory activation to accelerate this biomimetic morphogenesis of the endothelium. These findings demonstrate the feasibility and underscore the value of exploiting heterotypic interactions between endothelial cells and perivascular cells for the fabrication of an endothelial lining intended for small diameter arterial reconstruction. STATEMENT OF SIGNIFICANCE: Adipose tissue is an abundant, accessible, and uniquely dispensable source of endothelial cells and perivascular cells for vascular tissue engineering. While their spontaneous self-assembly into microvascular networks is routinely exploited for the vascularization of engineered tissues, it threatens the temporal stability of an endothelial lining intended for small diameter arterial reconstruction. Here, we demonstrate that an electrospun polyurethane scaffold can be used to physically separate endothelial cells from perivascular cells to prevent their spontaneous capillary morphogenesis, yet enable their cross-talk to promote the formation of a stable endothelium. Our findings demonstrate the feasibility of engineering an endothelial lining from human adipose tissue, poising it for the rapid ex vivo endothelialization of small diameter vascular prostheses in an autologous, patient-specific manner.

Noninvasive Quantification of Contractile Dynamics in Cardiac Cells, Spheroids, and Organs-on-a-Chip Using High-Frequency Ultrasound.

Cell-based models that mimic in vivo heart physiology are poised to make significant advances in cardiac disease modeling and drug discovery. In these systems, cardiomyocyte (CM) contractility is an important functional metric, but current measurement methods are inaccurate and low-throughput or require complex setups. To address this need, we developed a standalone noninvasive, label-free ultrasound technique operating at 40-200 MHz to measure the contractile kinetics of cardiac models, ranging from single adult CMs to 3D microtissue constructs in standard cell culture formats. The high temporal resolution of 1000 fps resolved the beat profile of single mouse CMs paced at up to 9 Hz, revealing limitations of lower speed optical based measurements to resolve beat kinetics or characterize aberrant beats. Coupling of ultrasound with traction force microscopy enabled the measurement of the CM longitudinal modulus and facile estimation of adult mouse CM contractile forces of 2.34 ± 1.40 µN, comparable to more complex measurement techniques. Similarly, the beat rate, rhythm, and drug responses of CM spheroid and microtissue models were measured, including in configurations without optical access. In conclusion, ultrasound can be used for the rapid characterization of CM contractile function in a wide range of commonly studied configurations ranging from single cells to 3D tissue constructs using standard well plates and custom microdevices, with applications in cardiac drug discovery and cardiotoxicity evaluation.

Multi-omic analyses and network biology in cardiovascular disease.

Heart disease remains a leading cause of death in North America and worldwide. Despite advances in therapies, the chronic nature of cardiovascular diseases ultimately results in frequent hospitalizations and steady rates of mortality. Systems biology approaches have provided a new frontier toward unraveling the underlying mechanisms of cell, tissue, and organ dysfunction in disease. Mapping the complex networks of molecular functions across the genome, transcriptome, proteome, and metabolome has enormous potential to advance our understanding of cardiovascular disease, discover new disease biomarkers, and develop novel therapies. Computational workflows to interpret these data-intensive analyses as well as integration between different levels of interrogation remain important challenges in the advancement and application of systems biology-based analyses in cardiovascular research. This review will focus on summarizing the recent developments in network biology-level profiling in the heart, with particular emphasis on modeling of human heart failure. We will provide new perspectives on integration between different levels of large "omics" datasets, including integration of gene regulatory networks, protein-protein interactions, signaling networks, and metabolic networks in the heart.

Dilated cardiomyopathy variant R14del increases phospholamban pentamer stability, blunting dynamic regulation of cardiac calcium handling.

The sarco(endo)plasmic reticulum Ca 2+ ATPase (SERCA) is a membrane transporter that creates and maintains intracellular Ca 2+ stores. In the heart, SERCA is regulated by an inhibitory interaction with the monomeric form of the transmembrane micropeptide phospholamban (PLB). PLB also forms avid homo-pentamers, and dynamic exchange of PLB between pentamers and the regulatory complex with SERCA is an important determinant of cardiac responsiveness to exercise. Here, we investigated two naturally occurring pathogenic mutations of PLB, a cysteine substitution of arginine 9 (R9C) and an in-frame deletion of arginine 14 (R14del). Both mutations are associated with dilated cardiomyopathy. We previously showed that the R9C mutation causes disulfide crosslinking and hyperstabilization of pentamers. While the pathogenic mechanism of R14del is unclear, we hypothesized that this mutation may also alter PLB homo-oligomerization and disrupt the PLB-SERCA regulatory interaction. SDS-PAGE revealed a significantly increased pentamer:monomer ratio for R14del-PLB when compared to WT-PLB. In addition, we quantified homo-oligomerization and SERCA-binding in live cells using fluorescence resonance energy transfer (FRET) microscopy. R14del-PLB showed an increased affinity for homo-oligomerization and decreased binding affinity for SERCA compared to WT, suggesting that, like R9C, the R14del mutation stabilizes PLB in its pentameric form, decreasing its ability to regulate SERCA. Moreover, the R14del mutation reduces the rate of PLB unbinding from the pentamer after a transient Ca 2+ elevation, limiting the rate of re-binding to SERCA. A computational model predicted that hyperstabilization of PLB pentamers by R14del impairs the ability of cardiac Ca 2+ handling to respond to changing heart rates between rest and exercise. We postulate that impaired responsiveness to physiological stress contributes to arrhythmogenesis in human carriers of the R14del mutation.

Proteomics and phosphoproteomics of failing human left ventricle identifies dilated cardiomyopathy-associated phosphorylation of CTNNA3.

The prognosis and treatment outcomes of heart failure (HF) patients rely heavily on disease etiology, yet the majority of underlying signaling mechanisms are complex and not fully elucidated. Phosphorylation is a major point of protein regulation with rapid and profound effects on the function and activity of protein networks. Currently, there is a lack of comprehensive proteomic and phosphoproteomic studies examining cardiac tissue from HF patients with either dilated dilated cardiomyopathy (DCM) or ischemic cardiomyopathy (ICM). Here, we used a combined proteomic and phosphoproteomic approach to identify and quantify more than 5,000 total proteins with greater than 13,000 corresponding phosphorylation sites across explanted left ventricle (LV) tissue samples, including HF patients with DCM vs. nonfailing controls (NFC), and left ventricular infarct vs. noninfarct, and periinfarct vs. noninfarct regions of HF patients with ICM. Each pair-wise comparison revealed unique global proteomic and phosphoproteomic profiles with both shared and etiology-specific perturbations. With this approach, we identified a DCM-associated hyperphosphorylation cluster in the cardiomyocyte intercalated disc (ICD) protein, αT-catenin (CTNNA3). We demonstrate using both ex vivo isolated cardiomyocytes and in vivo using an AAV9-mediated overexpression mouse model, that CTNNA3 phosphorylation at these residues plays a key role in maintaining protein localization at the cardiomyocyte ICD to regulate conductance and cell-cell adhesion. Collectively, this integrative proteomic/phosphoproteomic approach identifies region- and etiology-associated signaling pathways in human HF and describes a role for CTNNA3 phosphorylation in the pathophysiology of DCM.

Pressure overload induces ISG15 to facilitate adverse ventricular remodeling and promote heart failure.

Inflammation promotes adverse ventricular remodeling, a common antecedent of heart failure. Here, we set out to determine how inflammatory cells affect cardiomyocytes in the remodeling heart. Pathogenic cardiac macrophages induced an IFN response in cardiomyocytes, characterized by upregulation of the ubiquitin-like protein IFN-stimulated gene 15 (ISG15), which posttranslationally modifies its targets through a process termed ISGylation. Cardiac ISG15 is controlled by type I IFN signaling, and ISG15 or ISGylation is upregulated in mice with transverse aortic constriction or infused with angiotensin II; rats with uninephrectomy and DOCA-salt, or pulmonary artery banding; cardiomyocytes exposed to IFNs or CD4+ T cell-conditioned medium; and ventricular tissue of humans with nonischemic cardiomyopathy. By nanoscale liquid chromatography-tandem mass spectrometry, we identified the myofibrillar protein filamin-C as an ISGylation target. ISG15 deficiency preserved cardiac function in mice with transverse aortic constriction and led to improved recovery of mouse hearts ex vivo. Metabolomics revealed that ISG15 regulates cardiac amino acid metabolism, whereas ISG15 deficiency prevented misfolded filamin-C accumulation and induced cardiomyocyte autophagy. In sum, ISG15 upregulation is a feature of pathological ventricular remodeling, and protein ISGylation is an inflammation-induced posttranslational modification that may contribute to heart failure development by altering cardiomyocyte protein turnover.

An efficient and cost-effective purification protocol for Staphylococcus aureus Cas9 nuclease.

Here, we describe a protocol for purifying functional clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (Cas9) from Staphylococcus aureus within 24 h and over 90% purity. SaCas9 purification begins with immobilized metal affinity chromatography, followed by cation exchange chromatography, and ended with centrifugal concentrators. The simplicity, cost-effectiveness, and reproducibility of such protocols will enable general labs to produce a sizable amount of Cas9 proteins, further accelerating CRISPR research.

Tmem65 is critical for the structure and function of the intercalated discs in mouse hearts.

The intercalated disc (ICD) is a unique membrane structure that is indispensable to normal heart function, yet its structural organization is not completely understood. Previously, we showed that the ICD-bound transmembrane protein 65 (Tmem65) was required for connexin43 (Cx43) localization and function in cultured mouse neonatal cardiomyocytes. Here, we investigate the functional and cellular effects of Tmem65 reductions on the myocardium in a mouse model by injecting CD1 mouse pups (3-7 days after birth) with recombinant adeno-associated virus 9 (rAAV9) harboring Tmem65 shRNA, which reduces Tmem65 expression by 90% in mouse ventricles compared to scrambled shRNA injection. Tmem65 knockdown (KD) results in increased mortality which is accompanied by eccentric hypertrophic cardiomyopathy within 3 weeks of injection and progression to dilated cardiomyopathy with severe cardiac fibrosis by 7 weeks post-injection. Tmem65 KD hearts display depressed hemodynamics as measured echocardiographically as well as slowed conduction in optical recording accompanied by prolonged PR intervals and QRS duration in electrocardiograms. Immunoprecipitation and super-resolution microscopy demonstrate a physical interaction between Tmem65 and sodium channel ß subunit (ß1) in mouse hearts and this interaction appears to be required for both the establishment of perinexal nanodomain structure and the localization of both voltage-gated sodium channel 1.5 (NaV1.5) and Cx43 to ICDs. Despite the loss of NaV1.5 at ICDs, whole-cell patch clamp electrophysiology did not reveal reductions in Na+ currents but did show reduced Ca2+ and K+ currents in Tmem65 KD cardiomyocytes in comparison to control cells. We conclude that disrupting Tmem65 function results in impaired ICD structure, abnormal cardiac electrophysiology, and ultimately cardiomyopathy.

Immunomagnetic Isolation and Enrichment of Microvascular Endothelial Cells from Human Adipose Tissue.

Human adipose tissue-resident microvascular endothelial cells are not only garnering attention for their emergent role in the pathogenesis of obesity-related metabolic disorders, but are also of considerable interest for vascular tissue engineering due, in part, to the abundant, accessible, and uniquely dispensable nature of the tissue. Here, we delineate a protocol for the acquisition of microvascular endothelial cells from human fat. A cheaper, smaller, and simpler alternative to fluorescence-assisted cell sorting for the immunoselection of cells, our protocol adapts magnet-assisted cell sorting for the isolation of endothelial cells from enzymatically digested adipose tissue and the subsequent enrichment of their primary cultures. Strategies are employed to mitigate the non-specific uptake of immunomagnetic microparticles, enabling the reproducible acquisition of human adipose tissue-resident microvascular endothelial cells with purities ≥98%. They exhibit morphological, molecular, and functional hallmarks of endothelium, yet retain a unique proteomic signature when compared with endothelial cells derived from different vascular beds. Their cultures can be expanded for >10 population doublings and can be maintained at confluence for at least 28 days without being overgrown by residual stromal cells from the cell sorting procedure. The isolation of human adipose tissue-resident microvascular endothelial cells can be completed within 6 hours and their enrichment within 2 hours, following approximately 7 days in culture. Graphical abstract.

Vascular tissue engineering from human adipose tissue: fundamental phenotype of its resident microvascular endothelial cells and stromal/stem cells.

Adipose tissue is an abundant, accessible, and uniquely dispensable source of cells for vascular tissue engineering. Despite its intrinsic endothelial cells, considerable effort is directed at deriving endothelium from its resident stem and progenitor cells. Here, we investigate the composition of human adipose tissue and characterize the phenotypes of its constituent cells in order to help ascertain their potential utility for vascular tissue engineering. Unsupervised clustering based on cell-surface protein signatures failed to detect CD45-CD31-VEGFR2+ endothelial progenitor cells within adipose tissue, but supported further investigation of its resident CD45-CD31+ microvascular endothelial cells (HAMVECs) and CD45-CD31- stromal/stem cells (ASCs). The endothelial differentiation of ASCs altered their proteome, but it remained distinct from that of primary endothelial cell controls - as well as HAMVECs - regardless of their arterial-venous specification or macrovascular-microvascular origin. Rather, ASCs retained a proteome indicative of a perivascular phenotype, which was supported by their ability to facilitate the capillary morphogenesis of HAMVECs. This study supports the use of HAMVECs for the generation of endothelium. It suggests that the utility of ASCs for vascular tissue engineering lies in their capacity to remodel the extracellular matrix and to function as mural cells.

Self-Assembled Oligo-Urethane Nanoparticles: Their Characterization and Use for the Delivery of Active Biomolecules into Mammalian Cells.

Developing safe and effective strategies to deliver biomolecules such as oligonucleotides and proteins into cells has grown in importance over recent years, with an increasing demand for non-viral methods that enable clinical translation. Here, we investigate uniquely configured oligo-urethane nanoparticles based on synthetic chemistries that minimize the release of pro-inflammatory biomarkers from immune cells, show low cytotoxicity in a broad range of cells, and efficiently deliver oligonucleotides and proteins into mammalian cells. The mechanism of cell uptake for the self-assembled oligo-urethane nanoparticles was shown to be directed by caveolae-dependent endocytosis in murine myoblasts (C2C12) cells. Inhibiting caveolae functions with genistein and methyl-ß-cyclodextrin limited nanoparticle internalization. The nanoparticles showed a very high delivery efficiency for the genetic material (a 47-base oligonucleotide) (∼80% incorporation into cells) as well as the purified protein (full length firefly luciferase, 67 kDa) into human embryonic kidney (HEK293T) cells. Luciferase enzyme activity in HEK293T cells demonstrated that intact and functional proteins could be delivered and showed a significant extension of activity retention up to 24 h, well beyond the 2 h half-life of the free enzyme. This study introduces a novel self-assembled oligo-urethane nanoparticle delivery platform with very low associated production costs, enabled by their scalable chemistry (the benchwork cost is $ 0.152/mg vs $ 974.6/mg for typical lipid carriers) that has potential to deliver both oligonucleotides and proteins for biomedical purposes.

Deletion of type VIII collagen reduces blood pressure, increases carotid artery functional distensibility and promotes elastin deposition.

Arterial stiffening is a significant predictor of cardiovascular disease development and mortality. In elastic arteries, stiffening refers to the loss and fragmentation of elastic fibers, with a progressive increase in collagen fibers. Type VIII collagen (Col-8) is highly expressed developmentally, and then once again dramatically upregulated in aged and diseased vessels characterized by arterial stiffening. Yet its biophysical impact on the vessel wall remains unknown. The purpose of this study was to test the hypothesis that Col-8 functions as a matrix scaffold to maintain vessel integrity during extracellular matrix (ECM) development. These changes are predicted to persist into the adult vasculature, and we have tested this in our investigation. Through our in vivo and in vitro studies, we have determined a novel interaction between Col-8 and elastin. Mice deficient in Col-8 (Col8-/-) had reduced baseline blood pressure and increased arterial compliance, indicating an enhanced Windkessel effect in conducting arteries. Differences in both the ECM composition and VSMC activity resulted in Col8-/- carotid arteries that displayed increased crosslinked elastin and functional distensibility, but enhanced catecholamine-induced VSMC contractility. In vitro studies revealed that the absence of Col-8 dramatically increased tropoelastin mRNA and elastic fiber deposition in the ECM, which was decreased with exogenous Col-8 treatment. These findings suggest a causative role for Col-8 in reducing mRNA levels of tropoelastin and the presence of elastic fibers in the matrix. Moreover, we also found that Col-8 and elastin have opposing effects on VSMC phenotype, the former promoting a synthetic phenotype, whereas the latter confers quiescence. These studies further our understanding of Col-8 function and open a promising new area of investigation related to elastin biology.

Neuronatin promotes SERCA uncoupling and its expression is altered in skeletal muscles of high-fat diet-fed mice.

Neuronatin (NNAT) is a transmembrane protein in the endoplasmic reticulum involved in metabolic regulation. It shares sequence homology with sarcolipin (SLN), which negatively regulates the sarco(endo)plasmic reticulum Ca2+ -ATPase (SERCA) that maintains energy homeostasis in muscles. Here, we examined whether NNAT could uncouple the Ca2+ transport activity of SERCA from ATP hydrolysis, similarly to SLN. NNAT significantly reduced Ca2+ uptake without altering SERCA activity, ultimately lowering the apparent coupling ratio of SERCA. This effect of NNAT was reversed by the adenylyl cyclase activator forskolin. Furthermore, soleus muscles from high fat diet (HFD)-fed mice showed a significant downregulation in NNAT content compared with chow-fed mice, whereas an upregulation in NNAT content was observed in fast-twitch muscles from HFD- versus chow- fed mice. Therefore, NNAT is a SERCA uncoupler in cells and may function in adaptive thermogenesis.

Mitigating the non-specific uptake of immunomagnetic microparticles enables the extraction of endothelium from human fat.

Endothelial cells are among the fundamental building blocks for vascular tissue engineering. However, a clinically viable source of endothelium has continued to elude the field. Here, we demonstrate the feasibility of sourcing autologous endothelium from human fat - an abundant and uniquely dispensable tissue that can be readily harvested with minimally invasive procedures. We investigate the challenges underlying the overgrowth of human adipose tissue-derived microvascular endothelial cells by stromal cells to facilitate the development of a reliable method for their acquisition. Magnet-assisted cell sorting strategies are established to mitigate the non-specific uptake of immunomagnetic microparticles, enabling the enrichment of endothelial cells to purities that prevent their overgrowth by stromal cells. This work delineates a reliable method for acquiring human adipose tissue-derived microvascular endothelial cells in large quantities with high purities that can be readily applied in future vascular tissue engineering applications.

An organ-on-a-chip model for pre-clinical drug evaluation in progressive non-genetic cardiomyopathy.

Angiotensin II (Ang II) presents a critical mediator in various pathological conditions such as non-genetic cardiomyopathy. Osmotic pump infusion in rodents is a commonly used approach to model cardiomyopathy associated with Ang II. However, profound differences in electrophysiology and pharmacokinetics between rodent and human cardiomyocytes may limit predictability of animal-based experiments. This study investigates the application of an Organ-on-a-chip (OOC) system in modeling Ang II-induced progressive cardiomyopathy. The disease model is constructed to recapitulate myocardial response to Ang II in a temporal manner. The long-term tissue cultivation and non-invasive functional readouts enable monitoring of both acute and chronic cardiac responses to Ang II stimulation. Along with mapping of cytokine secretion and proteomic profiles, this model presents an opportunity to quantitatively measure the dynamic pathological changes that could not be otherwise identified in animals. Further, we present this model as a testbed to evaluate compounds that target Ang II-induced cardiac remodeling. Through assessing the effects of losartan, relaxin, and saracatinib, the drug screening data implicated multifaceted cardioprotective effects of relaxin in restoring contractile function and reducing fibrotic remodeling. Overall, this study provides a controllable platform where cardiac activities can be explicitly observed and tested over the pathological process. The facile and high-content screening can facilitate the evaluation of potential drug candidates in the pre-clinical stage.

Myocardial Infarction Induces Cardiac Fibroblast Transformation within Injured and Noninjured Regions of the Mouse Heart.

Heart failure (HF) is associated with pathological remodeling of the myocardium, including the initiation of fibrosis and scar formation by activated cardiac fibroblasts (CFs). Although early CF-dependent scar formation helps prevent cardiac rupture by maintaining the heart's structural integrity, ongoing deposition of the extracellular matrix in the remote and infarct regions can reduce tissue compliance, impair cardiac function, and accelerate progression to HF. In our study, we conducted mass spectrometry (MS) analysis to identify differentially altered proteins and signaling pathways between CFs isolated from 7 day sham and infarcted murine hearts. Surprisingly, CFs from both the remote and infarct regions of injured hearts had a wide number of similarly altered proteins and signaling pathways that were consistent with fibrosis and activation into pathological myofibroblasts. Specifically, proteins enriched in CFs isolated from MI hearts were involved in pathways pertaining to cell-cell and cell-matrix adhesion, chaperone-mediated protein folding, and collagen fibril organization. These results, together with principal component analyses, provided evidence of global CF activation postinjury. Interestingly, however, direct comparisons between CFs from the remote and infarct regions of injured hearts identified 15 differentially expressed proteins between MI remote and MI infarct CFs. Eleven of these proteins (Gpc1, Cthrc1, Vmac, Nexn, Znf185, Sprr1a, Specc1, Emb, Limd2, Pawr, and Mcam) were higher in MI infarct CFs, whereas four proteins (Gstt1, Gstm1, Tceal3, and Inmt) were higher in MI remote CFs. Collectively, our study shows that MI injury induced global changes to the CF proteome, with the magnitude of change reflecting their relative proximity to the site of injury.

Machine Learning Compared With Conventional Statistical Models for Predicting Myocardial Infarction Readmission and Mortality: A Systematic Review.

BACKGROUND: Machine learning (ML) methods are increasingly used in addition to conventional statistical modelling (CSM) for predicting readmission and mortality in patients with myocardial infarction (MI). However, the two approaches have not been systematically compared across studies of prognosis in patients with MI. METHODS: Following PRISMA guidelines, we systematically reviewed the literature via Medline, EPub, Cochrane Central, Embase, Inspec, ACM Digital Library, and Web of Science. Eligible studies included primary research articles published from January 2000 to March 2020, comparing ML and CSM for prognostication after MI. RESULTS: Of 7,348 articles, 112 underwent full-text review, with the final set composed of 24 articles representing 374,365 patients. ML methods included artificial neural networks (n = 12 studies), random forests (n = 11), decision trees (n = 8), support vector machines (n = 8), and Bayesian techniques (n = 7). CSM included logistic regression (n = 19 studies), existing CSM-derived risk scores (n = 12), and Cox regression (n = 2). Thirteen of 19 studies examining mortality reported higher C-indexes with the use of ML compared with CSM. One study examined readmissions at 2 different time points, with C-indexes that were higher for ML than CSM. Across all studies, a total of 29 comparisons were performed, but the majority (n = 26, 90%) found small (< 0.05) absolute differences in the C-index between ML and CSM. With the use of a modified CHARMS checklist, sources of bias were identifiable in the majority of studies, and only 2 were externally validated. CONCLUSION: Although ML algorithms tended to have higher C-indexes than CSM for predicting death or readmission after MI, these studies exhibited threats to internal validity and were often unvalidated. Further comparisons are needed, with adherence to clinical quality standards for prognosis research. (Trial registration: PROSPERO CRD42019134896).

Membrane proteomic profiling of the heart: past, present, and future.

Cardiovascular diseases remain the most rapidly rising contributing factor of all-cause mortality and the leading cause of inpatient hospitalization worldwide, with costs exceeding $30 billion annually in North America. Cell surface and membrane-associated proteins play an important role in cardiomyocyte biology and are involved in the pathogenesis of many human heart diseases. In cardiomyocytes, membrane proteins serve as critical signaling receptors, Ca2+ cycling regulators, and electrical propagation regulators, all functioning in concert to maintain spontaneous and synchronous contractions of cardiomyocytes. Membrane proteins are excellent pharmaceutical targets due to their uniquely exposed position within the cell. Perturbations in cardiac membrane protein localization and function have been implicated in the progression and pathogenesis of many heart diseases. However, previous attempts at profiling the cardiac membrane proteome have yielded limited results due to poor technological developments for isolating hydrophobic, low-abundance membrane proteins. Comprehensive mapping and characterization of the cardiac membrane proteome thereby remains incomplete. This review will focus on recent advances in mapping the cardiac membrane proteome and the role of novel cardiac membrane proteins in the healthy and the diseased heart.

Machine learning vs. conventional statistical models for predicting heart failure readmission and mortality.

AIMS: This study aimed to review the performance of machine learning (ML) methods compared with conventional statistical models (CSMs) for predicting readmission and mortality in patients with heart failure (HF) and to present an approach to formally evaluate the quality of studies using ML algorithms for prediction modelling. METHODS AND RESULTS: Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we performed a systematic literature search using MEDLINE, EPUB, Cochrane CENTRAL, EMBASE, INSPEC, ACM Library, and Web of Science. Eligible studies included primary research articles published between January 2000 and July 2020 comparing ML and CSMs in mortality and readmission prognosis of initially hospitalized HF patients. Data were extracted and analysed by two independent reviewers. A modified CHARMS checklist was developed in consultation with ML and biostatistics experts for quality assessment and was utilized to evaluate studies for risk of bias. Of 4322 articles identified and screened by two independent reviewers, 172 were deemed eligible for a full-text review. The final set comprised 20 articles and 686 842 patients. ML methods included random forests (n = 11), decision trees (n = 5), regression trees (n = 3), support vector machines (n = 9), neural networks (n = 12), and Bayesian techniques (n = 3). CSMs included logistic regression (n = 16), Cox regression (n = 3), or Poisson regression (n = 3). In 15 studies, readmission was examined at multiple time points ranging from 30 to 180 day readmission, with the majority of studies (n = 12) presenting prediction models for 30 day readmission outcomes. Of a total of 21 time-point comparisons, ML-derived c-indices were higher than CSM-derived c-indices in 16 of the 21 comparisons. In seven studies, mortality was examined at 9 time points ranging from in-hospital mortality to 1 year survival; of these nine, seven reported higher c-indices using ML. Two of these seven studies reported survival analyses utilizing random survival forests in their ML prediction models. Both reported higher c-indices when using ML compared with CSMs. A limitation of studies using ML techniques was that the majority were not externally validated, and calibration was rarely assessed. In the only study that was externally validated in a separate dataset, ML was superior to CSMs (c-indices 0.913 vs. 0.835). CONCLUSIONS: ML algorithms had better discrimination than CSMs in most studies aiming to predict risk of readmission and mortality in HF patients. Based on our review, there is a need for external validation of ML-based studies of prediction modelling. We suggest that ML-based studies should also be evaluated using clinical quality standards for prognosis research. Registration: PROSPERO CRD42020134867.