ABSTRACT
Clinical deterioration of COVID-19 patients is still a challenging event to predict in the emergency department (ED). The present study developed an artificial neural network using textual and tabular data from ED electronic medical reports. Predicted outcomes were 30-day mortality and ICU admission. Consecutive patients between February 20 and May 5, 2020, from Humanitas Research Hospital and San Raffaele Hospital, in the Milan area, were included. COVID-19 patients were 1296. Textual predictors were patient history, physical exam, and radiological reports. Tabular predictors were age, creatinine, C-reactive protein, hemoglobin, and platelet count. Tabular-textual model performance indices were compared to a model implementing only tabular data. For 30-day mortality, the combined model yielded slightly better performances than the tabular model, with AUC 0.84 ± 0.02, F-1 score 0.56 ± 0.04 and an MCC 0.44 ± 0.04. Tabular model performance was: AUC 0.84 ± 0.02, F-1 score 0.55 ± 0.03 and MCC 0.43 ± 0.04. As for ICU admission, the combined model was not superior to the tabular one. The present data points to the effectiveness of a textual and tabular model for COVID-19 prognosis prediction. Also, it may support the ED physician in their decision-making process.
Subject(s)
COVID-19ABSTRACT
BackgroundQuantitative CT (QCT) analysis is an invaluable diagnostic tool to assess lung injury and predict prognosis of patients affected by COVID-19 pneumonia. PTX3 was recently described as one of the most reliable serological predictors of clinical deterioration and short-term mortality. The present study was designed to evaluate a correlation between serological biomarkers of inflammation and lung injury measured by QCT. MethodsThis retrospective monocentric study analysed a cohort of patients diagnosed with COVID-19 and admitted because of respiratory failure, or significant radiological involvement on chest CT scan. All patients, males and non-pregnant females older than 18 years, underwent chest CT scan and laboratory testing at admission. Exclusion criteria were defined by concurrent acute pathological processes and ongoing specific treatments which could interfere with immune activity. The cohort was stratified based on severity in mild and severe forms. Compromised lung at QCT was then correlated to serological biomarkers representative of SARS-CoV-2. We further developed a multivariable logistic model to predict CT data and clinical deterioration based on a specific molecular signature. Internal cross-validation led to evaluate discrimination, calibration, and clinical utility of the tool that was provided by a score to simplify its application. Findings592 patients were recruited between March 19th and December 1st, 2020. Applying exclusion criteria which consider confounders, the cohort resulted in 366 individuals characterized by 177 mild and 189 severe forms. In our predictive model, blood levels of PTX3, CRP and LDH were found to correlate with QCT values in mild COVID-19 disease. A signature of these three biomarkers had a high predictive accuracy in detecting compromised lungs as assessed by QCT. The score was elaborated and resulted representative of lung CT damage leading to clinical deterioration and oxygen need in mild disease. InterpretationThe LDH, PTX3, CRP blood signature can serve as a strong correlate of compromised lung in COVID-19, possibly integrating cellular damage, systemic inflammation, myeloid and endothelial cell activation. FundingThis work was supported by a philanthropic donation by Dolce & Gabbana fashion house (to A.M., C.G.) and by a grant from Italian Ministry of Health for COVID-19 (to A.M. and C.G.). Research in contextO_ST_ABSEvidence before this studyC_ST_ABSBesides nasopharyngeal swab and serological test, chest CT scan represents one of the most useful tools to confirm COVID-19 diagnosis; moreover, QCT has been demonstrated to foresee oxygen need as well as deterioration of health status. Several clinical and serological parameters have been studied alone or combined in scores to be applied as prognostic tools of SARS-CoV-2 pneumonia; however, no one has yet reached the everyday practice. Recently, our group has investigated the expression and clinical significance of PTX3 in COVID-19 demonstrating the correlation with short-term mortality independently of confounders. The result was confirmed by other studies in different settings increasing evidence of PTX3 as a strong biomarker of severity; noteworthy, a recent report analysed proteomic data with a machine learning approach identifying age with PTX3 or SARS-CoV-2 RNAemia as the best binary signatures associated to 28-days mortality. Added value of this studyThe present study was designed to investigate associations between markers of damage and the CT extension of SARS-CoV-2 pneumonia in order to provide a biological footprint of radiological results in paucisymptomatic patients. QCT data were considered in a binary form identifying a threshold relevant for clinical deterioration, as already proved by literature. Our findings demonstrate a significant correlation with three peripheral blood proteins (PTX3, LDH and CRP) which result representative of COVID-19 severity. The study presents a predictive model of radiological lung involvement which performs with a high level of accuracy (cvAUC of 0{middle dot}794{+/-}0{middle dot}107; CI 95%: 0{middle dot}74-0{middle dot}87) and a simple score was provided to simplify the interpretation of the three biomarkers. Besides additional finding on PTX3 role in SARS-CoV2 pathology, its prognostic value was confirmed by data on clinical deterioration; indeed, paucisymptomatic subjects showed a 11{middle dot}9% deaths. The model offers the possibility to quickly assess patients resulted positive for SARS-CoV-2 and estimate people at risk of deterioration despite normal clinical and blood gases analysis, with potential to identify those who need better clinical monitoring and interventions. Implications of all the available evidencePredicting the extension, severity, and clinical deterioration in COVID-19 patients its pivotal to allocate enough resources in emergency and to avoid health system burden. Despite the urgent clinical need of biomarkers, SARS-CoV-2 pneumonia still lacks something able to provide an easy measure of its severity. Some multiparametric scores have been proposed for severe COVID-19 and rely on deep assessment of patients status (clinical, serological, and radiological data). Our model represents an unprecedented effort to provide a tool which could predict CT pneumonia extension, oxygen requirement and clinical deterioration in mild COVID-19. Based on the measurement of three proteins on peripheral blood, this score could improve early assessment of asymptomatic patients tested positive by SARS-CoV2 specifically in first level hospitals as well in developing countries.
Subject(s)
Lung Diseases , Pneumonia , Severe Acute Respiratory Syndrome , COVID-19 , Inflammation , Respiratory InsufficiencyABSTRACT
Due to the highly variable clinical phenotype of Coronavirus disease 2019 (COVID-19), deepening the host genetic contribution to severe COVID-19 may further improve our understanding about underlying disease mechanisms. Here, we describe an extended GWAS meta-analysis of 3,260 COVID-19 patients with respiratory failure and 12,483 population controls from Italy, Spain, Norway and Germany, as well as hypothesis-driven targeted analysis of the human leukocyte antigen (HLA) region and chromosome Y haplotypes. We include detailed stratified analyses based on age, sex and disease severity. In addition to already established risk loci, our data identify and replicate two genome-wide significant loci at 17q21.31 and 19q13.33 associated with severe COVID-19 with respiratory failure. These associations implicate a highly pleiotropic ~0.9-Mb 17q21.31 inversion polymorphism, which affects lung function and immune and blood cell counts, and the NAPSA gene, involved in lung surfactant protein production, in COVID-19 pathogenesis.
Subject(s)
COVID-19 , Respiratory InsufficiencyABSTRACT
PurposeTo evaluate the post- coronavirus disease-19 (COVID-19) outcome of thyroid function in patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-related thyrotoxicosis. MethodsThis was a single-center prospective study involving 29 patients (11 females, 18 males; median age 64 years, range: 43-85) with thyrotoxicosis diagnosed after hospitalization for COVID-19 and then followed-up for a median period of 90 days (range: 30-120) after hospital discharge. At the follow-up, patients were evaluated for serum thyrotropic (TSH), free-thyroxine (FT4), free-triiodiothyronine (FT3), TSH receptor antibodies (TRAb), thyroglobulin antibodies (TgAb), thyroperoxidase antibodies (TPOAb) and ultrasonographic thyroid structure.ResultsAfter recovery of COVID-19, serum TSH values significantly increased (P<0.001) and FT4 values significantly decreased (P=0.001), without significant change in serum FT3 (P=0.572). At the follow-up, 28 subjects (96.6%) became euthyroid whereas overt hypothyroidism developed in one case. At the ultrasound evaluation of thyroid gland, hypoecogenicity was found in 10 patients (34.5%) with a prevalence that was significantly higher in cases with serum TSH > 3.0 mU/l as compared to those with TSH values below 1.0 mU/L (P=0.039). All subjects resulted to be negative for TgAb, TPOAb and TRAb. ConclusionIn a short-term follow-up, thyroid function spontaneously normalized in most subjects with SARS-CoV-2-related thyrotoxicosis. However, thyroid hypoecogenicity was found in a remarkable number of them and future longer-term studies are needed to clarify whether this ultrasonographic alteration may predispose to develop late-onset thyroid dysfunction.
Subject(s)
Thyrotoxicosis , Coronavirus Infections , COVID-19 , Thyroid Diseases , HypothyroidismABSTRACT
Background. Respiratory failure is a key feature of severe Covid-19 and a critical driver of mortality, but for reasons poorly defined affects less than 10% of SARS-CoV-2 infected patients. Methods. We included 1,980 patients with Covid-19 respiratory failure at seven centers in the Italian and Spanish epicenters of the SARS-CoV-2 pandemic in Europe (Milan, Monza, Madrid, San Sebastian and Barcelona) for a genome-wide association analysis. After quality control and exclusion of population outliers, 835 patients and 1,255 population-derived controls from Italy, and 775 patients and 950 controls from Spain were included in the final analysis. In total we analyzed 8,582,968 single-nucleotide polymorphisms (SNPs) and conducted a meta-analysis of both case-control panels. Results. We detected cross-replicating associations with rs11385942 at chromosome 3p21.31 and rs657152 at 9q34, which were genome-wide significant (P<5x10-8) in the meta-analysis of both study panels, odds ratio [OR], 1.77; 95% confidence interval [CI], 1.48 to 2.11; P=1.14x10-10 and OR 1.32 (95% CI, 1.20 to 1.47; P=4.95x10-8), respectively. Among six genes at 3p21.31, SLC6A20 encodes a known interaction partner with angiotensin converting enzyme 2 (ACE2). The association signal at 9q34 was located at the ABO blood group locus and a blood-group-specific analysis showed higher risk for A-positive individuals (OR=1.45, 95% CI, 1.20 to 1.75, P=1.48x10-4) and a protective effect for blood group O (OR=0.65, 95% CI, 0.53 to 0.79, P=1.06x10-5). Conclusions. We herein report the first robust genetic susceptibility loci for the development of respiratory failure in Covid-19. Identified variants may help guide targeted exploration of severe Covid-19 pathophysiology.