Beth Israel Deaconess Medical Center Research Articles

Characterization of Post-COVID-19 Definitions and Clinical Coding Practices: Longitudinal Study.

Post-COVID-19 condition (colloquially known as "long COVID-19") characterized as postacute sequelae of SARS-CoV-2 has no universal clinical case definition. Recent efforts have focused on understanding long COVID-19 symptoms, and electronic health record (EHR) data provide a unique resource for understanding this condition. The introduction of the International Classification of Diseases, Tenth Revision (ICD-10) code U09.9 for "Post COVID-19 condition, unspecified" to identify patients with long COVID-19 has provided a method of evaluating this condition in EHRs; however, the accuracy of this code is unclear. This study aimed to characterize the utility and accuracy of the U09.9 code across 3 health care systems-the Veterans Health Administration, the Beth Israel Deaconess Medical Center, and the University of Pittsburgh Medical Center-against patients identified with long COVID-19 via a chart review by operationalizing the World Health Organization (WHO) and Centers for Disease Control and Prevention (CDC) definitions. Patients who were COVID-19 positive with either a U07.1 ICD-10 code or positive polymerase chain reaction test within these health care systems were identified for chart review. Among this cohort, we sampled patients based on two approaches: (1) with a U09.9 code and (2) without a U09.9 code but with a new onset long COVID-19-related ICD-10 code, which allows us to assess the sensitivity of the U09.9 code. To operationalize the long COVID-19 definition based on health agency guidelines, symptoms were grouped into a "core" cluster of 11 commonly reported symptoms among patients with long COVID-19 and an extended cluster that captured all other symptoms by disease domain. Patients having ≥2 symptoms persisting for ≥60 days that were new onset after their COVID-19 infection, with ≥1 symptom in the core cluster, were labeled as having long COVID-19 per chart review. The code's performance was compared across 3 health care systems and across different time periods of the pandemic. Overall, 900 patient charts were reviewed across 3 health care systems. The prevalence of long COVID-19 among the cohort with the U09.9 ICD-10 code based on the operationalized WHO definition was between 23.2% and 62.4% across these health care systems. We also evaluated a less stringent version of the WHO definition and the CDC definition and observed an increase in the prevalence of long COVID-19 at all 3 health care systems. This is one of the first studies to evaluate the U09.9 code against a clinical case definition for long COVID-19, as well as the first to apply this definition to EHR data using a chart review approach on a nationwide cohort across multiple health care systems. This chart review approach can be implemented at other EHR systems to further evaluate the utility and performance of the U09.9 code.

The association between ondansetron use and mortality risk of traumatic brain injury patients: a population-based study.

Traumatic brain injury (TBI) patients suffer high risks of mortality. Ondansetron has been verified to be effective in improving the prognosis of some kinds of critically ill patients. We design this study to explore whether ondansetron use is associated with lower risks of mortality among TBI patients. TBI patients from the Medical Information Mart for Intensive Care-III were collected. The usage of ondansetron, including intravenous injection and oral tablet, since admission to the Beth Israel Deaconess Medical Center between 2001 and 2012 was identified. Univariate and multivariate logistic regression were performed to analyze the relationship between the ondansetron use and mortality of TBI patients. Propensity score matching (PSM) was utilized to generate balanced cohorts of the non-ondansetron use group and ondansetron use group. Sub-group analysis was performed to verify the association between the ondansetron use and mortality of TBI patients in different TBI severity levels after PSM. In TBI cohorts before PSM, the usage incidence of ondansetron was 37.2%. The 30-day mortality was significantly lower in the ondansetron group (p < 0.001). The multivariate logistic regression showed that ondansetron was associated with the lower mortality of TBI patients (p = 0.008). In TBI cohorts after PSM, the 30-day mortality of the ondansetron group was lower than that of the non-ondansetron group, although without statistical significance (p = 0.079). Logistic regression indicated ondansetron use was significantly associated with the lower mortality of moderate-to-severe TBI (p < 0.001) but not mild TBI (p = 0.051). In addition, Cox regression also presented that ondansetron use was significantly associated with the lower mortality of moderate-to-severe TBI (p < 0.001) but not mild TBI (p = 0.052). Ondansetron usage is associated with a lower mortality risk of moderate-to-severe TBI but not mild TBI patients. Ondansetron may be a novel adjunctive therapeutic strategy to improve the prognosis of moderate-to-severe TBI patients.

The relationship between red cell distribution width, serum calcium ratio, and in-hospital mortality among patients with acute respiratory failure: A retrospective cohort study of the MIMIC-IV database.

To investigate the impact of RDW/CA (the ratio of red cell distribution width to calcium) on in-hospital mortality in patients with acute respiratory failure (ARF). This retrospective cohort study analyzed the data of 6981 ARF patients from the Medical Information Mart for Intensive Care (MIMIC-IV) database 2.0. Critically ill participants between 2008 and 2019 at the Beth Israel Deaconess Medical Center in Boston. The primary outcome of interest was in-hospital mortality. A Cox proportional hazards regression model was used to determine whether the RDW/CA ratio independently correlated with in-hospital mortality. The Kaplan-Meier method was used to plot the survival curves of the RDW/CA. Subgroup analyses were performed to measure the mortality across various subgroups. After adjusting for potential covariates, we found that a higher RDW/CA was associated with an increased risk of in-hospital mortality (HR = 1.17, 95% CI: 1.01-1.35, P = .0365) in ARF patients. A nonlinear relationship was observed between RDW/CA and in-hospital mortality, with an inflection point of 1.97. When RDW/CA ≥ 1.97 was positively correlated with in-hospital mortality in patients with ARF (HR = 1.554, 95% CI: 1.183-2.042, P = .0015). The Kaplan-Meier curve indicated the higher survival rates for RDW/CA < 1.97 and the lower for RDW/CA ≥ 1.97 after adjustment for age, gender, body mass index, and ethnicity. RDW/CA is an independent predictor of in-hospital mortality in patients with ARF. Furthermore, a nonlinear relationship was observed between RDW/CA and in-hospital mortality in patients with ARF.

How Would You Manage HIV Pre-exposure Prophylaxis in This Patient With Medical Comorbidities? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Despite advances in treatment, HIV infection remains an important cause of morbidity and mortality, with more than 30 000 new cases diagnosed in the United States each year. There are several interventions traditionally used to prevent HIV transmission, but these vary in effectiveness and there are challenges to their implementation. In 2014, the Centers for Disease Control and Prevention published initial guidance on the use of antiretroviral pre-exposure prophylaxis (PrEP) to prevent transmission of HIV infection in persons at risk based on multiple studies that showed it to be highly efficacious in various populations. It was updated in 2021 to reflect new drug options. The U.S. Preventive Services Task Force also recently updated its recommendations for PrEP, which strongly support its use in persons at risk. Despite its well-established effectiveness, the implementation of PrEP in clinical practice has been variable, especially among populations underserved by the medical system and marginalized by society. Fewer than one third of persons in the United States who are eligible for PrEP currently receive it. Here, 2 physicians experienced in HIV PrEP debate how best to identify patients who might benefit from PrEP, how to decide what regimen to use, and how to monitor therapy.

Vitamin D supplementation may be beneficial in improving the prognosis of patients with chronic obstructive pulmonary disease in the intensive care unit: a retrospective study.

Vitamin D is a crucial fat-soluble vitamin that has garnered significant attention due to its potential impact on respiratory health. It is noteworthy that many patients with chronic obstructive pulmonary disease (COPD) often experience deficiencies or insufficiencies of vitamin D. To address this issue, our retrospective study aimed to explore the potential association between serum 25-hydroxyvitamin D concentration and the prognoses of COPD patients in the Intensive Care Unit (ICU). This study utilised data from the Medical Information Marketplace in Intensive Care IV (MIMIC-IV), a database of patients admitted to the Intensive Care Unit at Beth Israel Deaconess Medical Center (BIDMC) in the United States of America, with a focus on patients with a diagnosis of COPD. These patients were categorized into two groups: those who received vitamin D supplementation during their ICU stay and those who did not. We assessed in-hospital mortality and ICU mortality outcomes. Our analysis involved various analytical tools, including Kaplan-Meier survival curves, Cox proportional risk regression models, and subgroup analyses, to investigate the relationship between vitamin D supplementation and these outcomes. Additionally, we employed propensity-score matching (PSM) to enhance the reliability of our findings. The study included a total of 3,203 COPD patients, with 587 in the vitamin D group and 2,616 in the no-vitamin D group. The Kaplan-Meier survival curve demonstrated a significant difference in survival probability between the two groups. After adjusting for potential confounders using Cox regression models, the vitamin D group exhibited a substantially lower risk of in-hospital and ICU mortalities compared to the no-vitamin D group. The hazard ratios for in-hospital and ICU mortalities in the vitamin D group were 1.7 (95% CI: 1.3, 2.3) and 1.8 (95% CI: 1.2, 2.6), respectively. Propensity-score matching (PSM) estimation yielded consistent results. Furthermore, in the subgroup analysis, female patients who received vitamin D supplementation showed a reduced risk of in-hospital mortality. The study suggests that vitamin D supplementation may be linked to a reduction in in-hospital and ICU mortalities among COPD patients in the ICU. Of particular note is the potential benefit observed in terms of in-hospital mortality, especially for female patients.

Development and Validation of an Interpretable Conformal Predictor to Predict Sepsis Mortality Risk: Retrospective Cohort Study.

Early and reliable identification of patients with sepsis who are at high risk of mortality is important to improve clinical outcomes. However, 3 major barriers to artificial intelligence (AI) models, including the lack of interpretability, the difficulty in generalizability, and the risk of automation bias, hinder the widespread adoption of AI models for use in clinical practice. This study aimed to develop and validate (internally and externally) a conformal predictor of sepsis mortality risk in patients who are critically ill, leveraging AI-assisted prediction modeling. The proposed approach enables explaining the model output and assessing its confidence level. We retrospectively extracted data on adult patients with sepsis from a database collected in a teaching hospital at Beth Israel Deaconess Medical Center for model training and internal validation. A large multicenter critical care database from the Philips eICU Research Institute was used for external validation. A total of 103 clinical features were extracted from the first day after admission. We developed an AI model using gradient-boosting machines to predict the mortality risk of sepsis and used Mondrian conformal prediction to estimate the prediction uncertainty. The Shapley additive explanation method was used to explain the model. A total of 16,746 (80%) patients from Beth Israel Deaconess Medical Center were used to train the model. When tested on the internal validation population of 4187 (20%) patients, the model achieved an area under the receiver operating characteristic curve of 0.858 (95% CI 0.845-0.871), which was reduced to 0.800 (95% CI 0.789-0.811) when externally validated on 10,362 patients from the Philips eICU database. At a specified confidence level of 90% for the internal validation cohort the percentage of error predictions (n=438) out of all predictions (n=4187) was 10.5%, with 1229 (29.4%) predictions requiring clinician review. In contrast, the AI model without conformal prediction made 1449 (34.6%) errors. When externally validated, more predictions (n=4004, 38.6%) were flagged for clinician review due to interdatabase heterogeneity. Nevertheless, the model still produced significantly lower error rates compared to the point predictions by AI (n=1221, 11.8% vs n=4540, 43.8%). The most important predictors identified in this predictive model were Acute Physiology Score III, age, urine output, vasopressors, and pulmonary infection. Clinically relevant risk factors contributing to a single patient were also examined to show how the risk arose. By combining model explanation and conformal prediction, AI-based systems can be better translated into medical practice for clinical decision-making.

Association between body mass index and delirium incidence in critically ill patients: a retrospective cohort study based on the MIMIC-IV Database

ObjectivesDelirium is a form of brain dysfunction with high incidence and is associated with many negative outcomes in the intensive care unit. However, few studies have been large enough to...

Corticosteroids to Reduce Inflammation in Severe Pancreatitis (CRISP) protocol and statistical analysis plan: a prospective, multicentre, double-blind, randomized, placebo controlled clinical trial

IntroductionAcute pancreatitis is a common disease which, in its severe form, is associated with significant morbidity and mortality. Currently, there is no specific therapy known to attenuate organ failure in severe pancreatitis and treatment consists primarily of supportive care. Corticosteroids have been shown to be beneficial in disease processes associated with systemic inflammation and could potentially improve outcomes in severe acute pancreatitis. MethodsThe Corticosteroids to Reduce Inflammation in Severe Pancreatitis (CRISP) trial is a multi-centre, double-blind, randomized, placebo-controlled clinical trial that aims to determine the impact of corticosteroids versus placebo on organ injury in patients with severe acute pancreatitis. Patients are randomized to receive 100 mg of hydrocortisone parenterally versus matching placebo every 8 h for 3 days. Clinical and laboratory data are collected at the time of study enrollment, at 24, 48 and 72 h. The primary end-point for the trial is the difference in 72-h change in the Sequential Organ Failure Assessment (SOFA) score between hydrocortisone and placebo groups. Additional key secondary outcomes include ventilator free days and 28-day mortality. DiscussionThis study will add to the evidence base in the treatment of severe acute pancreatitis. The results will inform clinical practice and future studies in the field.Trial registration numberThe trial is registered on clinicaltrials.gov (NCT05160506). It was posted on December 16th, 2021. The study protocol was approved by the Beth Israel Deaconess Medical Center Committee on Clinical Investigation (CCI) (protocol 2021 P-000803).

Generalized functional linear model with a point process predictor.

Point process data have become increasingly popular these days. For example, many of the data captured in electronic health records (EHR) are in the format of point process data. It is of great interest to study the association between a point process predictor and a scalar response using generalized functional linear regression models. Various generalized functional linear regression models have been developed under different settings in the past decades. However, existing methods can only deal with functional or longitudinal predictors, not point process predictors. In this article, we propose a novel generalized functional linear regression model for a point process predictor. Our proposed model is based on the joint modeling framework, where we adopt a log-Gaussian Cox process model for the point process predictor and a generalized linear regression model for the outcome. We also develop a new algorithm for fast model estimation based on the Gaussian variational approximation method. We conduct extensive simulation studies to evaluate the performance of our proposed method and compare it to competing methods. The performance of our proposed method is further demonstrated on an EHR dataset of patients admitted into the intensive care units of the Beth Israel Deaconess Medical Center between 2001 and 2008.

A systematic evaluation of the performance of GPT‐4 and PaLM2 to diagnose comorbidities in MIMIC‐IV patients

AbstractBackgroundGiven the strikingly high diagnostic error rate in hospitals, and the recent development of Large Language Models (LLMs), we set out to measure the diagnostic sensitivity of two popular LLMs: GPT‐4 and PaLM2. Small‐scale studies to evaluate the diagnostic ability of LLMs have shown promising results, with GPT‐4 demonstrating high accuracy in diagnosing test cases. However, larger evaluations on real electronic patient data are needed to provide more reliable estimates.MethodsTo fill this gap in the literature, we used a deidentified Electronic Health Record (EHR) data set of about 300,000 patients admitted to the Beth Israel Deaconess Medical Center in Boston. This data set contained blood, imaging, microbiology and vital sign information as well as the patients' medical diagnostic codes. Based on the available EHR data, doctors curated a set of diagnoses for each patient, which we will refer to as ground truth diagnoses. We then designed carefully‐written prompts to get patient diagnostic predictions from the LLMs and compared this to the ground truth diagnoses in a random sample of 1000 patients.ResultsBased on the proportion of correctly predicted ground truth diagnoses, we estimated the diagnostic hit rate of GPT‐4 to be 93.9%. PaLM2 achieved 84.7% on the same data set. On these 1000 randomly selected EHRs, GPT‐4 correctly identified 1116 unique diagnoses.ConclusionThe results suggest that artificial intelligence (AI) has the potential when working alongside clinicians to reduce cognitive errors which lead to hundreds of thousands of misdiagnoses every year. However, human oversight of AI remains essential: LLMs cannot replace clinicians, especially when it comes to human understanding and empathy. Furthermore, a significant number of challenges in incorporating AI into health care exist, including ethical, liability and regulatory barriers.

Thiamine as adjunctive therapy for diabetic ketoacidosis (DKAT) trial protocol and statistical analysis plan: a prospective, single-centre, double-blind, randomised, placebo-controlled clinical trial in the USA

IntroductionDiabetic ketoacidosis (DKA) is a potentially life-threatening diabetic complication. Despite the high prevalence of DKA and the substantial associated healthcare burden, limited research on strategies to improve outcomes currently exists.Thiamine...

Intraoperative Opioid Waste and Association of Intraoperative Opioid Dose with Postoperative Adverse Outcomes: A Hospital Registry Study.

Perioperative opioid use has been associated with adverse clinical outcomes. Additionally, opioid disposal carries significant costs, due to the waste of pharmaceutical products and the time needed by skilled labor to report the waste. In this study, we aimed to estimate costs and predict factors of opioid-associated intraoperative product waste, as well as to evaluate whether higher intraoperative opioid doses are associated with increased risk of adverse postoperative outcomes. We included 170,607 patients undergoing general anesthesia and receiving intraoperative fentanyl, hydromorphone, or morphine at Beth Israel Deaconess Medical Center, Boston, MA, USA, between January 2010 and June 2020. We estimated product waste-associated costs based on various opioid syringe sizes and determined predictors of opioid waste. Further, we evaluated whether higher opioid doses were associated with postoperative adverse events according to the severity-indexed, incident report-based medication error-reporting program classification. The primary outcome included post-extubation desaturation, postoperative nausea or vomiting, or postoperative somnolence or sedation. The use of the smallest syringe sizes (50 mcg for fentanyl, 0.2mg for hydromorphone, and 2mg for morphine) resulted in the lowest product waste-associated costs. The main predictor of opioid waste was the administration of more than one intraoperative opioid (adjusted odds ratio [aOR] = 7.64, 95% CI 7.40-7.89, P < 0.001). Intraoperative doses of fentanyl > 50-100 mcg (aOR = 1.17 [1.10-1.25], P < 0.001, adjusted risk difference [ARD] 2%) and > 100 mcg (aOR = 1.24 [1.16-1.33], P < 0.001, ARD 3%), hydromorphone > 1mg (aOR = 1.13 [1.06-1.20], P < 0.001, ARD 2%), and morphine > 2-4mg (aOR = 1.26 [1.02-1.56], P = 0.04, ARD 3%) and > 4mg (aOR = 1.45 [1.18-1.77], P < 0.001, ARD 5%) were associated with higher risk of the primary outcome. Smaller syringe sizes of intraoperative opioids may help to reduce product waste and associated costs, as well postoperative adverse events through utilization of lower intraoperative opioid doses.

Clinical course of multiple sclerosis with comorbid endometriosis: A matched cohort study

BackgroundEndometriosis (EMS) is pain syndrome in which endometrial tissue grows outside the uterus. EMS is associated with an increased risk of multiple sclerosis (MS), a demyelinating disease of the central nervous system. ObjectiveTo characterize clinical phenotypes of a cohort of patients with both EMS and MS compared to a cohort of matched controls with only MS. MethodsWe retrospectively identified patients with EMS and MS at Beth Israel Deaconess Medical Center (BIDMC). We collected data on EMS treatments and analyzed differences in histories of gynecological cancer, smoking, fatigue, anxiety, depression, headache, and neuropathic pain compared to matched controls. We used Wilcoxon signed rank tests for paired samples to compare Expanded Disability Status Scores (EDSS) and timed 25-foot walk values (T25FW). ResultsUsing a case-control methodology, we found significantly increased EDSS (p < 0.001) and T25FW (p = 0.01) in the EMS-MS group compared to the MS group. More patients in the EMS-MS group had histories of smoking, anxiety, depression, and headaches, while more patients in the MS group had histories of fatigue and neuropathic pain. ConclusionWhen controlling for age, race, and MS therapy, those with EMS-MS experience more MS disability than controls, suggesting this population requires more monitoring and efficacious treatment.

Calcium administration appears not to benefit acute pancreatitis patients with hypocalcemia.

Hypocalcemia occurs commonly among patients with acute pancreatitis (AP) in the intensive care unit (ICU). Calcium therapy could be used to correct hypocalcemia and maintain calcium levels, but its impact on the prognosis has not been demonstrated. Our study aimed to determine whether calcium therapy could benefit the multiple outcomes of AP patients with hypocalcemia. We extracted 807 AP patients with hypocalcemia from the Beth Israel Deaconess Medical Center (MIMIC-IV) database and performed retrospective analyses. The outcomes were in-hospital, 28 days, ICU mortality, and the length of stay (LOS) in the hospital and ICU. We performed propensity matching (PSM) and inverse probability weighting (IPTW) to balance the baseline differences and conducted multivariate regression to investigate the impact of calcium therapy. A total of 620 patients (76.8%) received calcium treatment (calcium group) during hospitalization, while 187 patients (non-calcium group) did not. Patients in the calcium group did not present significant survival differences between groups before and after matching. After including covariates, calcium administration had no association with patients' in-hospital (HR: 1.03, 95% Cl: 0.47-2.27, p = .942), 28 days and ICU mortality and was significantly associated with prolonged length of stay in the hospital (effect estimate: 6.18, 95% Cl: 3.27-9.09, p < .001) and ICU (effect estimate: 1.72, 95% Cl: 0.24-3.20, p < .001). Calcium therapy could not benefit patients in subgroups with exclusive parenteral infusion, early calcium therapy (<48 h), or various degrees of hypocalcemia. AP patients with hypocalcemia could not benefit from calcium administration, which has no association with multiple mortality and is significantly associated with prolonged LOS in the hospital and ICU.

Low HDL-C can be a biomarker to predict persistent severe AKI in septic patients? A retrospective cohort study

PurposesLow HDL-C is associated with an increased risk of sepsis-associated AKI and subsequent decline in eGFR. HDL-C possesses anti-inflammatory, antioxidant, and endothelial repair-promoting properties. The use of Apo A-I mimetic peptides, which are the main structural components of HDL-C, has been shown to improve renal function in animal models of sepsis. However, the diagnostic value of low HDL-C in persistent sepsis-associated AKI remains unclear.MethodsThis is a retrospective cohort study based on MIMIC IV (V 2.2). The study population consisted of all adult septic patients admitted to the Beth Israel Deaconess Medical Center Intensive Care Unit from 2008 to 2019, with plasma HDL-C measured within 24 h of ICU admission. The primary endpoint was persistent severe sepsis-associated acute kidney injury (SA-AKI) and the secondary endpoint is kidney replacement therapy (KRT). Logistic regression was used to assess the correlation between HDL-C and persistent severe SA-AKI and KRT, and receiver operating characteristic (ROC) curve analysis was performed to evaluate predictive ability.ResultsA total of 604 cases of SA-AKI patients were included in the analysis, among which 88 cases (14.5%) experienced persistent severe SA-AKI. The median (IQR) HDL-C level in the group with persistent severe SA-AKI was lower (33.0 [24.0–45.5]) compared to the non-persistent severe SA-AKI group (42.0 [31.0–53.0]). However, HDL-C showed poor discriminatory ability with an AUROC [95%CI] of 0.62 [0.56–0.69]. Clinical prediction models based on serum creatinine concentration, 24-h creatinine change, APSIIIscore, lactate levels, APTT, and heart rate performed well in predicting persistent severe SA-AKI with an AUROC [95%CI] of 0.876 [0.84–0.91]. However, adding HDL-C to this model did not improve predictive performance.ConclusionsThe plasma HDL-C measured within 24 h after admission to the ICU does not provide a good prediction for persistent severe SA-AKI, and it does not improve the clinical predictive ability compared to conventional variables.

Thrombosis and Intracranial Hemorrhage in Patients with Metastatic Brain Cancer Treated with Immune Checkpoint Inhibitors Versus Chemotherapy

Background Metastatic brain cancer increases the risk of thrombosis and intracranial hemorrhage (ICH). Immune Checkpoint Inhibitors (ICI) have emerged as breakthrough therapies for the treatment of metastatic disease. Thrombosis has been increasingly reported as a potential adverse event associated to ICI treatment. There is conflicting evidence on the comparative risk of thrombosis between ICI and chemotherapy, and limited data on bleeding outcomes during ICI treatment. We aimed to compare the rate of thrombosis and ICH in patients with metastatic brain tumors treated with ICI versus chemotherapy. Methods A retrospective, single-center cohort study was performed at Beth Israel Deaconess Medical Center. Patients with metastatic brain tumors, treated from 2010 to 2020 with ICI or chemotherapy were included. Primary endpoints included development of thrombotic and ICH events within 12 months of starting treatment. Thrombotic events included venous thromboembolism (VTE), or arterial thrombosis confirmed by imaging. ICH events were classified by a neuro-oncologist blinded to patient information. Intracranial bleeding due to surgical interventions or radiation therapy were excluded. Primary exposure was treatment group, defined as ICI with or without chemotherapy compared to chemotherapy alone. Proportions were compared using Chi square or Fisher's exact tests. Log-binomial regression was used to estimate relative risk (RR) and 95% confidence interval (CI) associated with each additional point of the Khorana score (KS). Cox hazard models were used to calculate the hazard of thrombosis and ICH in each group, accounting for death as a competing risk. Models were weighted by the inverse probability of treatment, estimated using age, gender, cancer, KS, elevated creatinine, and prior exposure to anticoagulants and/or antiplatelets (AC/AP) as covariates; weighting (IPTW) to calculate the cumulative incidence (CumI) of thrombosis and ICH among treatment groups, and stratified by use of AC/AP at the start of treatment. Results We included 264 patients with a median age of 65.4 years (Table 1). Overall, 59.8% were female. Major primary malignancies included non-small cell lung cancer (NSCLC) 189 (71.6%), SCLC 31 (11.7%), melanoma 13 (4.9%), and renal cell carcinoma 13 (2.9%). Regarding treatment, 78 (29.5%) patients received ICI, and 186 (70.5) chemotherapy; within the ICI group, 65.4% of patients also received chemotherapy. Overall, 28 patients developed thrombotic events, 14 pulmonary embolisms (50%), followed by lower extremity VTE (35.7%). Ten patients developed ICH during the study period, including three patients with major bleeds (&amp;gt;10cm3 in size and/or symptomatic). The majority of bleeding events were noted to be intraparenchymal (80%). The 12-month weighted CumI of thrombosis was similar in the ICI (11.1%, 95%CI 4.5-17.3) and chemotherapy (9.2%, 95%CI 5.6-12.6) groups. The 12-month CumI of ICH was also similar between ICI (7.2%, 95%CI 1.7-12.4) and chemotherapy (8.9%, 95%CI 5.4-12.3) groups. (Figure 1) Median time (MT) to thrombosis was comparable for patients on ICI (58.5 days, IQR26.0-239) and chemotherapy (64.5 days, IQR 26.0-154) (P=1.00). Conversely, ICH occurred earlier in the chemotherapy group (MT 49.0 days, IQR 12-112) compared to the ICI group (MT 280 days, IQR 266-292) (P=0.01). In weighted competing risk models, the hazard of thrombosis was similar among both treatment groups (aHR: 1.05, 95% CI: 0.5-2.1). The risk of ICH was not significantly lower with ICI compared to chemotherapy (aHR: 0.66, 95% CI 0.2-1.5). Within the ICI group, AC/AP use was not associated with significant difference in thrombotic risk (aHR: 0.63, 95%CI 0.1-2.3) but there was a significantly increased risk of ICH (aHR: 9.0, 95%CI 1.0-79.3). Within the chemotherapy group, use of AC/AP was not associated with increased risk of thrombosis (aHR: 0.96, 95%CI 0.3-2.5) or ICH (aHR: 0.16, 95% CI 0.0-1.0). Conclusions Rates of thrombosis and ICH were high in patients with metastatic brain tumors treated with ICI or cytotoxic chemotherapy but did not differ significantly between the two treatment groups. In our cohort, the time to intracranial bleeding was significantly shorter among those on chemotherapy compared to those on ICI. Contrasting risks of ICH and thrombosis in patients with metastatic brain cancer should be considered during treatment.

Stratification of Multiple Myeloma Patients in the Commpass Study By Single Cell Transcriptomics of Plasma Cells Associates with Clinical Outcome

Introduction: There is substantial variation in the prognosis and response to therapy among multiple myeloma (MM) patients. This diversity is influenced by both intrinsic tumor features and the tumor microenvironment. We hypothesize that the clinical outcomes post IMiD and/or PI are linked to molecular subtypes dictated by tumor sample transcriptomics. We previously showed that molecular subtyping based on the CoMMpass bulk RNAseq data has an association with time to second-line therapy (log rank p = 0.0018; Hamidi et al., Blood 140: 7208-7210). Here, we applied the same molecular subtyping schema to single cell RNAseq isolated plasma cell transcriptomic profiles. Methods: Bone marrow biopsies from newly diagnosed MM patients (N = 200) in the MMRF CoMMpass study (NCT01454297) were collected, enriched by CD138 negative bead selection, and scRNAseq was performed by MMRF Immune Profiling Consortium, a collaboration between MMRF and laboratories at Beth Israel Deaconess Medical Center, Icahn School of Medicine at Mt. Sinai, Mayo Clinic, Washington University in St. Louis, and Emory University. After quality control and unsupervised clustering, plasma cells were subset out and pseudo-bulk expression profiles were generated for baseline samples with more than 100 plasma cells (N = 95). Unsupervised machine learning approach based on consensus non-negative matrix factorization (cNMF) was applied to identify the transcriptomics-based molecular subtypes. A random forest model, built from scRNAseq-based molecular subtypes (scNMF), was applied to bulk RNAseq data with a larger sample size (N = 754) to train a predictive molecular subtyping (pNMF). Potential cell-cell communication events based on scRNAseq were also predicted by LIgand-receptor ANalysis frAmework (LIANA). Results: 705,959 single cells passed the quality control and were broadly annotated as T/NK cells, B cells, plasma cells, myeloid cells, erythroid, progenitor, and stromal cells. Consistent with previous reports, CD138 negative bead selection did not completely remove plasma cells, resulting in an average of 18.5% plasma cells (n = 130,213). Subsequent unsupervised clustering of plasma cells exhibited a patient specific clustering pattern, suggestive of the heterogeneity of these tumor-derived cells. Unsupervised NMF on the pseudo-bulk expressions of baseline plasma cells identified five patient subsets. In the same cohort of patients (N = 95), subtypes assigned by the single cell had a better association with time to second-line therapy ( p-value = 0.0027 in scNMF as compared to p = 0.13 in bulk NMF). The median time to 2nd line of 2,443 days in scNMF5.1 versus 585 days in scNMF5.4; and the median time to 2nd line of 1,597 days in NMF5.2 versus 850 days in NMF5.3. A predicted subtyping based on scRNAseq data (pNMF) had a significantly better association with time to second-line therapy (Figure 1A, log rank p = 0.00088), compared to the published molecular subtyping based on bulk RNAseq (log rank p = 0.0018). The median time to 2nd line of 1,342 days in pNMF5.1 versus 597 days in pNMF5.5; compared to the median time to 2nd line of 1,347 days in NMF5.2 versus 850 days in NMF5.3. The best clinical outcome subset pNMF5.1 was characterized with higher levels of IFN pathways (Figure 1B, Hallmark interferon gamma and alpha pathways Wilcoxon p = 0.003 and p = 0.010, respectively) and higher levels of tumor-promoting inflammatory microenvironment (Hallmarks inflammatory response and TNFα signaling via NF-κB p = 0.003 and p = 0.010, respectively). The worst outcome subset pNMF5.5, on the other hand, displayed higher levels of glycolysis, a promoting factor of tumor development ( p &amp;lt; 0.001). Moreover, the total number of cell-cell interactions between the major cell types differed based on scRNAseq subtypes, informing the underlying biology driving the molecular heterogeneity. We identified a positive feedback loop of CXCL12/CXCR4 signaling between stromal and plasma cells, enhancing tumor cell proliferation. Conclusions: The identification of molecular subtypes based on single-cell data through an innovative unsupervised machine learning approach can have significant implications for determining which NDMM patients are likely to respond to established therapies; and the cellular level resolution allows deconvolution of cell-cell signaling, enhancing our understanding of the underlying molecular biology and mechanisms of resistance.

Pre-Existing Cardiovascular Disease Increases the Risk of Cardiovascular Adverse Events during Bruton Tyrosine Kinase Inhibitor Therapy

Background Bruton tyrosine kinase inhibitors (BTKis) have revolutionized the treatment of B-cell neoplasms. BTKi treatment has been associated with a higher incidence of cardiovascular adverse events (CVAE), which contribute significantly to morbidity and mortality in this population. Ibrutinib has been more strongly associated with CVAE compared to more selective 2 nd-generation BTKis. Although predisposing factors are not clear, pre-existing cardiovascular disease (CVD) appears to increase the risk of atrial arrhythmias and mortality during BTKi treatment. Our primary objective was to evaluate and compare the rate of CVAE during BTKi therapy, in individuals with and without pre-existing CVD. In addition, we aimed to compare the outcomes of patients treated with 1 st vs. 2 nd generation BTKis and characterize management decisions. Methods A retrospective review was conducted on patients treated with BTKis for hematologic malignancies from 2013 to 2022 at Beth Israel Deaconess Medical Center. Exclusion criteria included prior treatment with anthracyclines and BTKis, prescription of BTKi therapy without administration, prior BTKi treatment for graft-vs-host disease, and loss to follow-up. CVAE were defined as new-onset or worsening atrial fibrillation (AF), new-onset or worsening hypertension (HTN), supraventricular or ventricular arrhythmias, and sudden cardiac death. Pre-existing CVD included coronary artery disease, congestive heart failure, cerebrovascular disease, and pulmonary HTN. Student's t-test and Fisher's exact test were performed to compare the differences in continuous and categorical variables between patients with and without pre-existing CVD, respectively. Logistic regression analyses were performed to evaluate the association between pre-existing CVD and CVAE. All statistics were performed using R version 4.3. Results We identified 274 patients and 180 met inclusion criteria. Primary indications were Chronic Lymphocytic Leukemia (CLL) in 106 patients (60%), Mantle Cell Lymphoma (MCL) in 31 (17.2%), Diffuse Large B-Cell Lymphoma (DLBCL) in 17 (9.4%), and Waldenström Macroglobulinemia (WM) in 16 (8.9%). In terms of BTKi selection, 141 patients (78.3%) received ibrutinib, 35 (19.4%) acalabrutinib, and 4 (2.2%) zanubrutinib. Prevalent comorbidities were HTN in 103 patients (57%), dyslipidemia in 80 (44%), and diabetes in 26 (14%). Overall, 49 patients (27.2%) had preexisting CVD, and 21 (11.6%) experienced CVAE during BTKi therapy. In our cohort, (11/49) patients with pre-existing CVD developed CVAE, compared to (10/131) patients without pre-existing CVD, (22% vs 8% respectively), OR 3.38 (95%CI = [1.3, 8.5], p-value= 0.01). The OR remained robust ranging from 3.4 to 4.1 (p &amp;lt; 0.05) in multiple adjusted logistic regression models, accounting for imbalances in gender, comorbidities, type of BTKi, and baseline medications. Patients with pre-existing CVD had significantly higher odds of developing new or worsening AF, (16% vs. 5% respectively), OR = 3.94, (95%CI = [1.3, 12.0], p-value= 0.02) (Table 1). In subgroup analysis, patients treated with 1 st generation BTKis had a numerically higher rate of CVAE than those treated with 2 nd generation BTKis, OR = 6.28 (95%CI = [0.8, 48.3], p-value= 0.08) (Table 2). Regarding management, BTKi treatment was discontinued in 13 patients (62%) and continued in 8 (38%). Medical interventions were required in 17 patients (80%). In contrast, the event resolved in 4 patients (19%) with only BTKi discontinuation. Following CVAE, 3 patients (14%) were transitioned to a 2 nd generation BTKi with complete resolution. Conclusions CVAE remain significant contributors to treatment discontinuation. Patients treated with 1st-generation BTKis and pre-existing CVD had significantly higher odds of developing CVAE. Our study is the first to evaluate the impact of pre-existing CVD on CVAE during treatment with 2nd-generation BTKis. Patients on Ibrutinib developed more CVAE compared to patients on 2nd-generation BTKis, though differences were not statistically significant. This study was limited due to its retrospective and single-centered nature. Results highlight the need for structured approaches to prevent and promptly detect CVAE, particularly in patients with pre-existing CVD.

Estimating the Burden of Venous and Arterial Thrombotic Events in Hospitalized Adults with COVID-19; A National Multicenter Cohort Study

Background: Patients infected with Coronavirus Disease 2019 (COVID-19) have clinical and laboratory features consistent with a hypercoaguable state, particularly in more severely ill patients that require hospitalization. However, estimates of thrombotic complications in this high risk population vary widely. A geographic representative cohort to explore the rates of thrombotic events in patients hospitalized with COVID-19 in the United States would be useful to better understand the burden of thrombosis in this population. We aimed to assess the rates of venous and arterial thrombosis events during hospitalization in patients with COVID-19 in a national multicenter cohort study. Methods: We conducted a retrospective cohort study of hospitalized patients diagnosed with COVID-19 from January 1 2020 to January 2023. IRB approval was obtained at each site. Inclusion criteria included adults (&amp;gt; 21 years), hospitalized for &amp;gt;1 day and laboratory confirmation of polymerase chain reaction testing of severe acute respiratory syndrome coronavirus 2. Data were extracted by participating centers in a pre-specified instrument from the electronic medical record at each institution and pooled prior to analysis. Data sources included manual extraction by study investigators in 7 sites and electronic extraction based on billing codes at one site (Cleveland Clinic). Primary outcome was venous thrombotic events, including deep vein thrombosis (DVT), pulmonary embolism (PE). Unusual sites including splanchnic vein thrombosis and cerebral venous thrombosis were included if objectively demonstrated on imaging. Arterial thrombotic events including ischemic stroke, acute coronary syndrome were included as secondary outcomes. Data are presented as proportions, with binomial 95% confidence intervals (CI). Results: We included 33,769 patients from eight medical centers across the United States. Participating hospitals included all four census regions from the North East (Beth Israel Deaconess Medical Center, MA (n=1021) Rochester Regional Hospital, NY (n=99)), South (John Hopkins Medical Institute, ML (n=334); Baycare Health System, FL (n=3734), MidWest (Cleveland Clinic Foundation, OH (n=25467); Henry Ford Health System, MI (973); University of Illinois, IL (1192)) and West (University of California, CA (n=613)). (Figure) Of the total 33,433 hospitalized patients with COVID-19 included, 1684 patients developed a venous thrombotic event during the index hospitalization (5.0%, 95% CI 4.8-5.2) and 261 developed an arterial thrombotic event (0.8%, 95% CI, 95% CI 0.7-0.9). Rates of venous events by individual sites ranged from 2.6-8% and arterial events from 0.3-6.6%. Conclusion: In this national multicenter cohort study that included academic and community hospitals from all four US census regions, we estimated the rates of VTE were over five times more frequent than arterial events in hospitalized patients with COVID-19. Accurate estimates of thrombotic rates and trends can help plan targeted interventions related to strategies around thromboprophylaxis interventions and resource allocation for future surges of COVID-19 cases.

2665. Health Care Provider Characteristics Associated with Anal Cancer Screening in an Academic Medical System

Abstract Background Anal cancer disproportionally affects persons living with HIV (PWH), sexual and gender minorities, and those with gynecologic cancers, condyloma acuminata, and solid organ transplantation. While expert guidance on anal cancer screening exists for PWH, there are no national guidelines for screening any of these populations. We analyzed clinician characteristics associated with anal cancer screening among providers who care for high-risk populations at an academic medical center including infectious disease (ID) specialists, primary care providers (PCPs), and obstetrician/gynecologists (OBGYNs). Methods We retrospectively reviewed medical charts for patients with HIV, gynecologic cancers, condyloma acuminata, solid organ transplantation, and sexual or gender minority status at Beth Israel Deaconess Medical Center and two affiliated clinics from 01/01/2015 - 08/01/2022. We extracted provider specialties, clinic locations, patient panel demographics, and dates of anal cytology and human papillomavirus (HPV) screening tests. We used chi-squared tests to identify provider factors associated with screening. Results Of 1,093 providers, at least one anal cytology test was performed by 93.8% (75/80) of ID providers, 55.2% (376/681) of PCPs, 28.3% (41/145) of OBGYNs, and 24.1% (45/187) of other providers (p &amp;lt; 0.001); among screeners, anal HPV co-testing was performed equally across specialties (54.2%, p = 0.23). Providers were more likely to screen if they cared for PWH (91.8% with 10-49 PWH vs 3.7% for no PWH, p &amp;lt; 0.001), majority cis-men (55.2%, p=0.003), or majority publicly insured patients (51.8%, p &amp;lt; 0.001). Providers were less likely to screen if they cared for majority cis-women (40.9%, p&amp;lt; 0.001) or majority privately insured patients (24.0%, p&amp;lt; 0.001). Screening practices did not differ by the racial composition of providers’ patient panels (p=.21). Conclusion While many ID specialists performed anal cancer screenings, other providers were less likely to screen. Providers caring for many PWH, cis-men, or publicly insured patients were more likely to screen. Additional research is needed to clarify screening guidance and optimize screening practices for all higher-risk populations, especially non-PWH groups. Disclosures Douglas Krakower, MD, Gilead: Grant/Research Support|Merck: Grant/Research Support|U. North Texas Health Sciences Center: Funding for mentoring|UAB: Advisor/Consultant|UpToDate, Inc.: Royalties|Virology Education: Honoraria