STREAK: A supervised cell surface receptor abundance estimation strategy for single cell RNA-sequencing data using feature selection and thresholded gene set scoring.

The accurate estimation of cell surface receptor abundance for single cell transcriptomics data is important for the tasks of cell type and phenotype categorization and cell-cell interaction quantification. We previously developed an unsupervised receptor abundance estimation technique named SPECK (Surface Protein abundance Estimation using CKmeans-based clustered thresholding) to address the challenges associated with accurate abundance estimation. In that paper, we concluded that SPECK results in improved concordance with Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) data relative to comparative unsupervised abundance estimation techniques using only single-cell RNA-sequencing (scRNA-seq) data. In this paper, we outline a new supervised receptor abundance estimation method called STREAK (gene Set Testing-based Receptor abundance Estimation using Adjusted distances and cKmeans thresholding) that leverages associations learned from joint scRNA-seq/CITE-seq training data and a thresholded gene set scoring mechanism to estimate receptor abundance for scRNA-seq target data. We evaluate STREAK relative to both unsupervised and supervised receptor abundance estimation techniques using two evaluation approaches on six joint scRNA-seq/CITE-seq datasets that represent four human and mouse tissue types. We conclude that STREAK outperforms other abundance estimation strategies and provides a more biologically interpretable and transparent statistical model.

SPECK: an unsupervised learning approach for cell surface receptor abundance estimation for single-cell RNA-sequencing data.

Summary: The rapid development of single-cell transcriptomics has revolutionized the study of complex tissues. Single-cell RNA-sequencing (scRNA-seq) can profile tens-of-thousands of dissociated cells from a tissue sample, enabling researchers to identify cell types, phenotypes and interactions that control tissue structure and function. A key requirement of these applications is the accurate estimation of cell surface protein abundance. Although technologies to directly quantify surface proteins are available, these data are uncommon and limited to proteins with available antibodies. While supervised methods that are trained on Cellular Indexing of Transcriptomes and Epitopes by Sequencing data can provide the best performance, these training data are limited by available antibodies and may not exist for the tissue under investigation. In the absence of protein measurements, researchers must estimate receptor abundance from scRNA-seq data. Therefore, we developed a new unsupervised method for receptor abundance estimation using scRNA-seq data called SPECK (Surface Protein abundance Estimation using CKmeans-based clustered thresholding) and primarily evaluated its performance against unsupervised approaches for at least 25 human receptors and multiple tissue types. This analysis reveals that techniques based on a thresholded reduced rank reconstruction of scRNA-seq data are effective for receptor abundance estimation, with SPECK providing the best overall performance. Availability and implementation: SPECK is freely available at https://CRAN.R-project.org/package=SPECK. Supplementary information: Supplementary data are available at Bioinformatics Advances online.

Postdischarge Functional Capacity, Health-Related Quality of Life, Depression, Anxiety, and Post-traumatic Stress Disorder in Patients Receiving a Long-term Left Ventricular Assist Device.

BACKGROUND: There is a paucity of data on depression, anxiety and post-traumatic stress disorder after left ventricular assist device (LVAD) implantation. We designed an observational study to integrate these with functional capacity and health-related quality of life (HR-QOL) in surviving LVAD patients. METHODS AND RESULTS: Consenting patients between 1 month and 9 years after LVAD implantation (n = 121) were screened for functional capacity (World Health Organization Disability Assessment Schedule 2.0 [WHODAS 2.0)]); HR-QOL (European Quality of Life [EQ-5D] and Visual Assessment Scales [EQ-VAS]), depression (Patient Health Questionnaire [PHQ-9], anxiety (Generalized Anxiety Disorder Scale [GAD-7]) and post-traumatic stress disorder (Impact of Event Scale Revised [IES-R]). Of the 94% of patients who consented, 34.7% reported impaired functional capacity (WHODAS 2.0 score of ≥25%), 23.1%-34.7% HR-QOL problems (domain EQ-5D of ≥3), 10.7% "poor health" (EQ-VAS of ≤40), 14.9% depression (PHQ-9 of >14), 11.7% suicidal ideation and 17.5% anxiety (GAD-7 of >10). Among these patients, 23.5% had a positive screen for post-traumatic stress disorder (IES-R of ≥24). An EQ-VAS of 80 or greater predicted good functional capacity (P < .001). CONCLUSIONS: One-third of discharged LVAD patients reported impaired function, HR-QOL, and psychological issues. A standardized evaluation before and after LVAD implantation could facilitate psychologic prehabilitation, inform decision-making, and identify indications for mental health intervention.

Levels of Trimethylamine N-Oxide Remain Elevated Long Term After Left Ventricular Assist Device and Heart Transplantation and Are Independent From Measures of Inflammation and Gut Dysbiosis.

BACKGROUND: Trimethylamine N-oxide (TMAO)-a gut-derived metabolite-is elevated in heart failure (HF) and linked to poor prognosis. We investigated variations in TMAO in HF, left ventricular assist device (LVAD), and heart transplant (HT) and assessed its relation with inflammation, endotoxemia, oxidative stress, and gut dysbiosis. METHODS: We enrolled 341 patients. TMAO, CRP (C-reactive protein), IL (interleukin)-6, TNF-α (tumor necrosis factor alpha), ET-1 (endothelin-1), adiponectin, lipopolysaccharide, soluble CD14, and isoprostane were measured in 611 blood samples in HF (New York Heart Association class I-IV) and at multiple time points post-LVAD and post-HT. Gut microbiota were assessed via 16S rRNA sequencing among 327 stool samples. Multivariable regression models were used to assess the relationship between TMAO and (1) New York Heart Association class; (2) pre- versus post-LVAD or post-HT; (3) biomarkers of inflammation, endotoxemia, oxidative stress, and microbial diversity. RESULTS: ln-TMAO was lower among HF New York Heart Association class I (1.23 [95% CI, 0.52-1.94] µM) versus either class II, III, or IV (1.99 [95% CI, 1.68-2.30], 1.97 [95% CI, 1.71-2.24], and 2.09 [95% CI, 1.83-2.34] µM, respectively; all P<0.05). In comparison to class II-IV, ln-TMAO was lower 1 month post-LVAD (1.58 [95% CI, 1.32-1.83] µM) and 1 week and 1 month post-HT (0.97 [95% CI, 0.60-1.35] and 1.36 [95% CI, 1.01-1.70] µM). ln-TMAO levels in long-term LVAD (>6 months: 1.99 [95% CI, 1.76-2.22] µM) and HT (>6 months: 1.86 [95% CI, 1.66-2.05] µM) were not different from symptomatic HF. After multivariable adjustments, TMAO was not associated with biomarkers of inflammation, endotoxemia, oxidative stress, or microbial diversity. CONCLUSIONS: TMAO levels are increased in symptomatic HF patients and remain elevated long term after LVAD and HT. TMAO levels were independent from measures of inflammation, endotoxemia, oxidative stress, and gut dysbiosis.

Presence of Intracardiac Thrombus at the Time of Left Ventricular Assist Device Implantation Is Associated With an Increased Risk of Stroke and Death.

BACKGROUND: Heart failure predisposes to intracardiac thrombus (ICT) formation. There are limited data on the prevalence and impact of preexisting ICT on postoperative outcomes in left ventricular assist device patients. We examined the risk for stroke and death in this patient population. METHODS AND RESULTS: We retrospectively studied patients who were implanted with HeartMate (HM) II or HM3 between February 2009 and March 2019. Preoperative transthoracic echocardiograms, intraoperative transesophageal echocardiograms and operative reports were reviewed to identify ICT. There were 525 patients with a left ventricular assist device (median age 60.6 years, 81.8% male, 372 HMII and 151 HM3) included in this analysis. An ICT was identified in 44 patients (8.4%). During the follow-up, 43 patients experienced a stroke and 55 died. After multivariable adjustment, presence of ICT increased the risk for the composite of stroke or death at 6-month (hazard ratio [HR] 1.82, 95% confidence interval [CI] 1.00-3.33, P = .049). Patients with ICT were also at higher risk for stroke (HR 2.45, 95% CI 1.14-5.28, P = .021) and death (HR 2.36, 95% CI 1.17-4.79 P = .016) at 6 months of follow-up. CONCLUSIONS: The presence of ICT is an independent predictor of stroke and death at 6 months after left ventricular assist device implantation. Additional studies are needed to help risk stratify and optimize the perioperative management of this patient population.

Association of preoperative infections, nasal Staphylococcus aureus colonization and gut microbiota with left ventricular assist device outcomes.

AIMS: Infections are common following left ventricular assist device (LVAD) implantation and predict adverse events. Infections are frequent prior to LVAD implantation although their impact on postoperative outcomes remains unknown. Gut and nasal microbial imbalance may predispose to mucosal colonization with pathogens. Herein, we investigated the predictive role of pre-LVAD infections, and explored the association of nasal Staphylococcus aureus (SA) colonization and gut microbiota, on postoperative outcomes. METHODS AND RESULTS: Overall, 254 LVAD patients were retrospectively categorized based on pre-LVAD infection status: Group 1, bacterial/fungal bloodstream infection (BSI); Group 2, other bacterial/fungal; Group 3, viral; and Group 4, no infection. In a subset of patients, nasal SA colonization (n = 140) and pre-LVAD stool (n = 25) were analysed using 16S rRNA sequencing. A total of 75 (29%) patients had a pre-LVAD infection [Group 1: 22 (29%); Group 2: 41 (55%); Group 3: 12 (16%)]. Pre-LVAD BSIs were independent predictors of 1-year postoperative mortality and infections [Group 1 vs. 4: hazard ratio (HR) 2.70, P = 0.036 vs. HR 1.8, P = 0.046]. In an unadjusted analysis, pre-LVAD infections other than BSIs, INTERMACS profile ≤2, higher serum creatinine, lower serum albumin and nasal SA colonization were also significantly associated with postoperative infections. Patients with early post-LVAD infections exhibited decreased microbial diversity (P < 0.05). CONCLUSIONS: Pre-LVAD infections are common. BSIs independently predict postoperative mortality and infections. Additional studies are needed to confirm our findings that pre-LVAD SA nasal colonization and gut microbial composition can help stratify patients' risk for infectious complications after LVAD implantation.

Increased Aortic Stiffness Is Associated With Higher Rates of Stroke, Gastrointestinal Bleeding and Pump Thrombosis in Patients With a Continuous Flow Left Ventricular Assist Device.

BACKGROUND: In the general population, increased aortic stiffness is associated with an increased risk of cardiovascular events. Previous studies have demonstrated an increase in aortic stiffness in patients with a continuous flow left ventricular assist device (CF-LVAD). However, the association between aortic stiffness and common adverse events is unknown. METHODS AND RESULTS: Forty patients with a HeartMate II (HMII) (51 $ 11 years; 20% female; 25% ischemic) implanted between January 2011 and September 2017 were included. Two-dimensional transthoracic echocardiograms of the ascending aorta, obtained before HMII placement and early after heart transplant, were analyzed to calculate the aortic stiffness index (AO-SI). The study cohort was divided into patients who had an increased vs decreased AO-SI after LVAD support. A composite outcome of gastrointestinal bleeding, stroke, and pump thrombosis was defined as the primary end point and compared between the groups. While median AO-SI increased significantly after HMII support (AO-SI 4.4-6.5, P = .012), 16 patients had a lower AO-SI. Patients with increased (n = 24) AO-SI had a significantly higher rate of the composite end point (58% vs 12%, odds ratio 9.8, P < .01). Similarly, those with increased AO-SI tended to be on LVAD support for a longer duration, had higher LVAD speed and reduced use of angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers. CONCLUSIONS: Increased aortic stiffness in patients with a HMII is associated with a significantly higher rates of adverse events. Further studies are warranted to determine the causality between aortic stiffness and adverse events, as well as the effect of neurohormonal modulation on the conduit vasculature in patients with a CF-LVAD.

Carotid artery structure and hemodynamics and their association with adverse vascular events in left ventricular assist device patients.

Left ventricular assist devices (LVADs) are associated with major vascular complications including stroke and gastrointestinal bleeding (GIB). These adverse vascular events may be the result of widespread vascular dysfunction resulting from pre-LVAD abnormalities or continuous flow during LVAD therapy. We hypothesized that pre-existing large artery atherosclerosis and/or abnormal blood flow as measured in carotid arteries using ultrasonography are associated with a post-implantation composite adverse outcome including stroke, GIB, or death. We retrospectively studied 141 adult HeartMate II patients who had carotid ultrasound duplex exams performed before and/or after LVAD surgery. Structural parameters examined included plaque burden and stenosis. Hemodynamic parameters included peak-systolic, end-diastolic, and mean velocity as well as pulsatility index. We examined the association of these measures with the composite outcome as well as individual subcomponents such as stroke. After adjusting for established risk factors, the composite adverse outcome was associated with pre-operative moderate-to-severe carotid plaque (OR 5.08, 95% CI 1.67-15.52) as well as pre-operative internal carotid artery stenosis (OR 9.02, 95% CI 1.06-76.56). In contrast, altered hemodynamics during LVAD support were not associated with the composite outcome. Our findings suggest that pre-existing atherosclerosis possibly in combination with LVAD hemodynamics may be an important contributor to adverse vascular events during mechanical support. This encourages greater awareness of carotid morphology pre-operatively and further study of the interaction between hemodynamics, pulsatility, and structural arterial disease during LVAD support.

Frailty subtypes and recovery in older survivors of acute respiratory failure: a pilot study.

BACKGROUND: Identifying subtypes of acute respiratory failure survivors may facilitate patient selection for post-intensive care unit (ICU) follow-up clinics and trials. METHODS: We conducted a single-centre prospective cohort study of 185 acute respiratory failure survivors, aged ≥ 65 years. We applied latent class modelling to identify frailty subtypes using frailty phenotype and cognitive impairment measurements made during the week before hospital discharge. We used Fine-Gray competing risks survival regression to test associations between frailty subtypes and recovery, defined as returning to a basic Activities of Daily Living disability count less than or equal to the pre-hospitalisation count within 6 months. We characterised subtypes by pre-ICU frailty (Clinical Frailty Scale score ≥ 5), the post-ICU frailty phenotype, and serum inflammatory cytokines, hormones and exosome proteomics during the week before hospital discharge. RESULTS: We identified five frailty subtypes. The recovery rate decreased 49% across each subtype independent of age, sex, pre-existing disability, comorbidity and Acute Physiology and Chronic Health Evaluation II score (recovery rate ratio: 0.51, 95% CI 0.41 to 0.63). Post-ICU frailty phenotype prevalence increased across subtypes, but pre-ICU frailty prevalence did not. In the subtype with the slowest recovery, all had cognitive impairment. The three subtypes with the slowest recovery had higher interleukin-6 levels (p=0.03) and a higher prevalence of ≥ 2 deficiencies in insulin growth factor-1, dehydroepiandrostersone-sulfate, or free-testosterone (p=0.02). Exosome proteomics revealed impaired innate immunity in subtypes with slower recovery. CONCLUSIONS: Frailty subtypes varied by prehospitalisation frailty and cognitive impairment at hospital discharge. Subtypes with the slowest recovery were similarly characterised by greater systemic inflammation and more anabolic hormone deficiencies at hospital discharge.

Outcomes and Mortality Prediction Model of Critically Ill Adults With Acute Respiratory Failure and Interstitial Lung Disease.

BACKGROUND: We aimed to examine short- and long-term mortality in a mixed population of patients with interstitial lung disease (ILD) with acute respiratory failure, and to identify those at lower vs higher risk of in-hospital death. METHODS: We conducted a single-center retrospective cohort study of 126 consecutive adults with ILD admitted to an ICU for respiratory failure at a tertiary care hospital between 2010 and 2014 and who did not undergo lung transplantation during their hospitalization. We examined associations of ICU-day 1 characteristics with in-hospital and 1-year mortality, using Poisson regression, and examined survival using Kaplan-Meier curves. We created a risk score for in-hospital mortality, using a model developed with penalized regression. RESULTS: In-hospital mortality was 66%, and 1-year mortality was 80%. Those with connective tissue disease-related ILD had better short-term and long-term mortality compared with unclassifiable ILD (adjusted relative risk, 0.6; 95% CI, 0.3-0.9; and relative risk, 0.6; 95% CI, 0.4-0.9, respectively). Our prediction model includes male sex, interstitial pulmonary fibrosis diagnosis, use of invasive mechanical ventilation and/or extracorporeal life support, no ambulation within 24 h of ICU admission, BMI, and Simplified Acute Physiology Score-II. The optimism-corrected C-statistic was 0.73, and model calibration was excellent (P = .99). In-hospital mortality rates for the low-, moderate-, and high-risk groups were 33%, 65%, and 96%, respectively. CONCLUSIONS: We created a risk score that classifies patients with ILD with acute respiratory failure from low to high risk for in-hospital mortality. The score could aid providers in counseling these patients and their families.

Race, Ethnicity, Health Insurance, and Mortality in Older Survivors of Critical Illness.

OBJECTIVES: To determine whether minority race or ethnicity is associated with mortality and mediated by health insurance coverage among older (≥ 65 yr old) survivors of critical illness. DESIGN: A retrospective cohort study. SETTING: Two New York City academic medical centers. PATIENTS: A total of 1,947 consecutive white (1,107), black (361), and Hispanic (479) older adults who had their first medical-ICU admission from 2006 through 2009 and survived to hospital discharge. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We obtained demographic, insurance, and clinical data from electronic health records, determined each patient's neighborhood-level socioeconomic data from 2010 U.S. Census tract data, and determined death dates using the Social Security Death Index. Subjects had a mean (SD) age of 79 years (8.6 yr) and median (interquartile range) follow-up time of 1.6 years (0.4-3.0 yr). Blacks and Hispanics had similar mortality rates compared with whites (adjusted hazard ratio, 0.92; 95% CI, 0.76-1.11 and adjusted hazard ratio, 0.92; 95% CI, 0.76-1.12, respectively). Compared to those with commercial insurance and Medicare, higher mortality rates were observed for those with Medicare only (adjusted hazard ratio, 1.43; 95% CI, 1.03-1.98) and Medicaid (adjusted hazard ratio, 1.30; 95% CI, 1.10-1.52). Medicaid recipients who were the oldest ICU survivors (> 82 yr), survivors of mechanical ventilation, and discharged to skilled-care facilities had the highest mortality rates (p-for-interaction: 0.08, 0.03, and 0.17, respectively). CONCLUSIONS: Mortality after critical illness among older adults varies by insurance coverage but not by race or ethnicity. Those with federal or state insurance coverage only had higher mortality rates than those with additional commercial insurance.