Citations

AbstractIntroductionWe sought to assess longitudinal electronic health records (EHRs) using machine learning (ML) methods to computationally derive probable Alzheimer's Disease (AD) and related dementia subphenotypes.MethodsA retrospective analysis of EHR data from a cohort of 7587 patients seen at a large, multi‐specialty urban academic medical center in New York was conducted. Subphenotypes were derived using hierarchical clustering from 792 probable AD patients (cases) who had received at least one diagnosis of AD using their clinical data. The other 6795 patients, labeled as controls, were matched on age and gender with the cases and randomly selected in the ratio of 9:1. Prediction models with multiple ML algorithms were trained on this cohort using 5‐fold cross‐validation. XGBoost was used to rank the variable importance.ResultsFour subphenotypes were computationally derived. Subphenotype A (n = 273; 28.2%) had more patients with cardiovascular diseases; subphenotype B (n = 221; 27.9%) had more patients with mental health illnesses, such as depression and anxiety; patients in subphenotype C (n = 183; 23.1%) were overall older (mean (SD) age, 79.5 (5.4) years) and had the most comorbidities including diabetes, cardiovascular diseases, and mental health disorders; and subphenotype D (n = 115; 14.5%) included patients who took anti‐dementia drugs and had sensory problems, such as deafness and hearing impairment.The 0‐year prediction model for AD risk achieved an area under the receiver operating curve (AUC) of 0.764 (SD: 0.02); the 6‐month model, 0.751 (SD: 0.02); the 1‐year model, 0.752 (SD: 0.02); the 2‐year model, 0.749 (SD: 0.03); and the 3‐year model, 0.735 (SD: 0.03), respectively. Based on variable importance, the top‐ranked comorbidities included depression, stroke/transient ischemic attack, hypertension, anxiety, mobility impairments, and atrial fibrillation. The top‐ranked medications included anti‐dementia drugs, antipsychotics, antiepileptics, and antidepressants.ConclusionsFour subphenotypes were computationally derived that correlated with cardiovascular diseases and mental health illnesses. ML algorithms based on patient demographics, diagnosis, and treatment demonstrated promising results in predicting the risk of developing AD at different time points across an individual's lifespan.

Abstract Background: Whether circulating polyunsaturated fatty acid (PUFA) levels are associated with pancreatic cancer risk is uncertain. Mendelian randomization (MR) represents a study design using genetic instruments to better characterize the relationship between exposure and outcome. Methods: We utilized data from genome-wide association studies within the Pancreatic Cancer Cohort Consortium and Pancreatic Cancer Case–Control Consortium, involving approximately 9,269 cases and 12,530 controls of European descent, to evaluate associations between pancreatic cancer risk and genetically predicted plasma n-6 PUFA levels. Conventional MR analyses were performed using individual-level and summary-level data. Results: Using genetic instruments, we did not find evidence of associations between genetically predicted plasma n-6 PUFA levels and pancreatic cancer risk [estimates per one SD increase in each PUFA-specific weighted genetic score using summary statistics: linoleic acid odds ratio (OR) = 1.00, 95% confidence interval (CI) = 0.98–1.02; arachidonic acid OR = 1.00, 95% CI = 0.99–1.01; and dihomo-gamma-linolenic acid OR = 0.95, 95% CI = 0.87–1.02]. The OR estimates remained virtually unchanged after adjustment for covariates, using individual-level data or summary statistics, or stratification by age and sex. Conclusions: Our results suggest that variations of genetically determined plasma n-6 PUFA levels are not associated with pancreatic cancer risk. Impact: These results suggest that modifying n-6 PUFA levels through food sources or supplementation may not influence risk of pancreatic cancer.

AbstractStaphylococcal enterotoxin B (SEB), which is produced by the major human pathogen, Staphylococcus aureus, represents a powerful superantigenic toxin and is considered a bioweapon. However, the contribution of SEB to S. aureus pathogenesis has never been directly demonstrated with genetically defined mutants in clinically relevant strains. Many isolates of the predominant Asian community-associated methicillin-resistant S. aureus lineage sequence type (ST) 59 harbor seb, implying a significant role of SEB in the observed hypervirulence of this lineage. We created an isogenic seb mutant in a representative ST59 isolate and assessed its virulence potential in mouse infection models. We detected a significant contribution of seb to systemic ST59 infection that was associated with a cytokine storm. Our results directly demonstrate that seb contributes to S. aureus pathogenesis, suggesting the value of including SEB as a target in multipronged antistaphylococcal drug development strategies. Furthermore, they indicate that seb contributes to fatal exacerbation of community-associated methicillin-resistant S. aureus infection.

ABSTRACT Introduction: We hypothesize that a patient (pt) with accelerated thrombin generation, time to peak height (ttPeak), will have a greater odds of meeting critical administration threshold (CAT) criteria (> 3 packed red blood cell [pRBC] transfusions [Tx] per 60 min interval), within the first 24 h after injury, independent of international normalized ratio (INR). Methods: In a prospective cohort study, trauma patients were enrolled over a 4.5-year period and serial blood samples collected at various time points. We retrospectively stratified pts into three categories: CAT+, CAT− but receiving some pRBC Tx, receiving no Tx within the first 24 h. Blood collected prior to Tx was analyzed for thrombin generation parameters and prothrombin time (PT)/INR. Results: A total of 484 trauma pts were analyzed: injury severity score = 13 [7,22], age = 48 [28, 64] years, and 73% male. Fifty pts met criteria for CAT+, 64 pts CAT−, and 370 received no Tx. Risk factors for meeting CAT+: decreased arrival systolic blood pressure (OR 2.82 [2.17, 3.67]), increased INR (OR 2.09, [1.66, 2.62]) and decreased time to peak OR 2.27 [1.74, 2.95]). These variables remained independently associated with increased risk of requiring Tx in a multivariable logistic model, after adjusting for sex and trauma type. Conclusions: Pts in hemorrhagic shock, who meet CAT+ criteria, are characterized by accelerated thrombin generation. In our multivariable analysis, both ttPeak and PT/INR have a complementary role in predicting those injured patients who will require a high rate of Tx.

High-grade neuroendocrine carcinomas (NEC) of the endometrium are rare and account for <1% of all endometrial carcinomas. Both small cell neuroendocrine carcinoma (SCNEC) and large cell neuroendocrine carcinoma (LCNEC) morphologies have been reported. Little is known regarding the molecular features of endometrial NEC including how they compare to pulmonary NEC (the most common site for these neoplasms) and the more common endometrial carcinoma histotypes. In this study, we investigated the molecular alterations in a series of endometrial NEC using a targeted next generation sequencing panel (Oncopanel). Fourteen NEC were sequenced; pure NEC (n=4) and mixed (n=10) with endometrioid adenocarcinoma (n=9) or carcinosarcoma (n=1). The NEC components of mixed tumors comprised LCNEC (n=6) and SCNEC (n=4). The 4 pure NEC comprised LCNEC (n=2) and SCNEC (n=2). Molecular analysis classified tumors into the 4 The Cancer Genome Atlas groups: (1) POLE-mutated/ultramutated (1/14; 7%), (2) microsatellite instability/hypermutated (6/14; 43%), (3) TP53 mutated/copy number high (2/14; 14%), or (4) no specific molecular profile (5/14; 36%). Overall, 50% of cases were ultramutated or hypermutated. In 8 cases of mixed carcinomas, the different histologic components were macrodissected and separately sequenced; molecular alterations were nearly identical among the 2 components, with the non-NEC component harboring slightly increased tumor mutational burden. Only 2 carcinomas (both with pure SCNEC morphology) had a molecular profile that would be expected in typical pulmonary SCNEC (RB1 deletion and TP53 mutations). Our findings, similar to data from NECs of other anatomic sites, suggest that the molecular context may be important when selecting therapies for women with endometrial NEC. Immune checkpoint inhibition may be a reasonable approach to treatment of microsatellite instability-NEC and we thus recommend that all endometrial NEC be tested for mismatch repair abnormalities, either molecularly or by mismatch repair protein immunohistochemistry.

AbstractObjectiveTo identify depression subphenotypes from Electronic Health Records (EHRs) using machine learning methods, and analyze their characteristics with respect to patient demographics, comorbidities, and medications.Materials and MethodsUsing EHRs from the INSIGHT Clinical Research Network (CRN) database, multiple machine learning (ML) algorithms were applied to analyze 11 275 patients with depression to discern depression subphenotypes with distinct characteristics.ResultsUsing the computational approaches, we derived three depression subphenotypes: Phenotype_A (n = 2791; 31.35%) included patients who were the oldest (mean (SD) age, 72.55 (14.93) years), had the most comorbidities, and took the most medications. The most common comorbidities in this cluster of patients were hyperlipidemia, hypertension, and diabetes. Phenotype_B (mean (SD) age, 68.44 (19.09) years) was the largest cluster (n = 4687; 52.65%), and included patients suffering from moderate loss of body function. Asthma, fibromyalgia, and Chronic Pain and Fatigue (CPF) were common comorbidities in this subphenotype. Phenotype_C (n = 1452; 16.31%) included patients who were younger (mean (SD) age, 63.47 (18.81) years), had the fewest comorbidities, and took fewer medications. Anxiety and tobacco use were common comorbidities in this subphenotype.ConclusionComputationally deriving depression subtypes can provide meaningful insights and improve understanding of depression as a heterogeneous disorder. Further investigation is needed to assess the utility of these derived phenotypes to inform clinical trial design and interpretation in routine patient care.

Abstract BackgroundDeep venous thrombosis (DVT) is a severe complication of the coronavirus disease 2019 (COVID-19). It may interfere with COVID-19 treatment and delay the recovery, but there is less data about the anticoagulant therapy and sex difference of VTE in patients with COVID-19. The purpose of this study is to study the prevalence, risk factors, anticoagulant therapy and sex difference of deep venous thrombosis (DVT) in patients with COVID-19.MethodsThe enrolled 121 patients were confirmed positive for COVID-19. All suspected patients with a high Caprini index (≥4) or PADUA index (≥4) received color Doppler Ultrasound (US) to screen DVT in both lower extremities. Clinical characteristics of DVT-COVID-19 patients were analyzed. Multivariate logistic regression was performed to identify risk factors related to DVT in COVID-19 patients. The distribution of DVT locations, anticoagulation therapy with sex difference, and the outcomes were also analyzed.ResultsDVT was found in 48% asymptomatic COVID-19 patients with increased PAUDA index or Caprini index by US scanning. Multivariate logistic regression determined that age, CRP and baseline D-dimer were risk factors among COVDI-19 patients. Although the most common DVT location was infrapopliteal (Class I and Class II), higher mortality in DVT-COVID-19 patients was confirmed. DVT-COVID-19 patients presented significant increases in the CRP, neutrophil count and D-dimer throughout the whole inpatient period compared to non-DVT-COVID-19 patients. Although anticoagulation therapy accelerated the recovery of lymphocytopenia condition in DVT patients, men DVT-COVID-19 patients showed higher CRP and neutrophil count vs. lymphocyte count (N/L) ratio but lower lymphocyte count compared to women DVT-COVID-19 patients. ConclusionsDVT is common in COVID-19 patients with high risk factors, especially for older age, higher CRP and baseline D-dimer populations. It is important to consider sex differences in the anticoagulant therapy among DVT-COVID-19 patients.

Background: Racial disparity in pathological consequences in skull growth may impact the treatment plan for different patient populations. This study attempts to explore the differences between Asian and Caucasian cranial morphology in Crouzon syndrome. Method: Ninety-one computed tomographic scans were included (Asian Crouzon syndrome, n = 12; Asian controls, n = 22; Caucasian Crouzon syndrome, n = 16; Caucasian controls, n = 41), and measured using Mimics and 3-matics software. Results: The entire cranial base length was reduced 11.92 mm (P = 0.004) in Asian Crouzon patients, and 14.58 mm (P < 0.001) in Caucasian Crouzon patients, compared to respective controls. The cranial base angle on the facial side of basicranium was more narrowed in Crouzon syndrome in both races, with similar changes of degrees (9.61°, P = 0.002, in Asian Crouzon; 9.20°, P = 0.019, in Caucasian Crouzon). However, the intracranial side was statistically more narrowed only in the Asian group (9.86°, P = 0.003). Both Asian and Caucasian Crouzon patients developed reduced posterior fossa volume, by 15% (P = 0.034) and 17% (P = 0.004), respectively. However, Caucasian Crouzon patients developed a more shortened anterior and middle cranial base, than that of Asian patients. The separation of lateral pterygoids was only significantly increased in Caucasian patients (5.49°, P < 0.001). Conclusion: Crouzon syndrome causes a shortened, widened, and kyphotic cranial base across both races. It also restricts the development of the posterior cranial fossa. However, the skull malformation is not the same between populations: Asian Crouzon patients developed more narrowed cranial base angulation on the intracranial side of basicranium, while Caucasian Crouzon patients developed more widened lateral pterygoids.

AbstractIntroductionElectronic health record (EHR)‐driven phenotyping is a critical first step in generating biomedical knowledge from EHR data. Despite recent progress, current phenotyping approaches are manual, time‐consuming, error‐prone, and platform‐specific. This results in duplication of effort and highly variable results across systems and institutions, and is not scalable or portable. In this work, we investigate how the nascent Clinical Quality Language (CQL) can address these issues and enable high‐throughput, cross‐platform phenotyping.MethodsWe selected a clinically validated heart failure (HF) phenotype definition and translated it into CQL, then developed a CQL execution engine to integrate with the Observational Health Data Sciences and Informatics (OHDSI) platform. We executed the phenotype definition at two large academic medical centers, Northwestern Medicine and Weill Cornell Medicine, and conducted results verification (n = 100) to determine precision and recall. We additionally executed the same phenotype definition against two different data platforms, OHDSI and Fast Healthcare Interoperability Resources (FHIR), using the same underlying dataset and compared the results.ResultsCQL is expressive enough to represent the HF phenotype definition, including Boolean and aggregate operators, and temporal relationships between data elements. The language design also enabled the implementation of a custom execution engine with relative ease, and results verification at both sites revealed that precision and recall were both 100%. Cross‐platform execution resulted in identical patient cohorts generated by both data platforms.ConclusionsCQL supports the representation of arbitrarily complex phenotype definitions, and our execution engine implementation demonstrated cross‐platform execution against two widely used clinical data platforms. The language thus has the potential to help address current limitations with portability in EHR‐driven phenotyping and scale in learning health systems.