ABSTRACT
Background. Intensive Care Resources are heavily utilized during the COVID-19 pandemic. However, risk stratification and prediction of SARS-CoV-2 patient clinical outcomes upon ICU admission remain inadequate. This study aimed to develop a machine learning model, based on retrospective & prospective clinical data, to stratify patient risk and predict ICU survival and outcomes.Methods. A Germany-wide electronic registry was established to pseudonymously collect admission, therapeutic and discharge information of SARS-CoV-2 ICU patients retrospectively and prospectively. Machine learning approaches were evaluated for the accuracy and interpretability of predictions. The Explainable Boosting Machine approach was selected as the most suitable method. Individual, non-linear shape functions for predictive parameters and parameter interactions are reported. Results. 1,039 patients were included in the Explainable Boosting Machine model, 596 patients retrospectively collected, and 443 patients prospectively collected. The model for prediction of general ICU outcome was shown to be more reliable to predict “survival”. Age, inflammatory and thrombotic activity, and severity of ARDS at ICU admission were shown to be predictive of ICU survival. Patients’ age, pulmonary dysfunction and transfer from an external institution were predictors for ECMO therapy. The interaction of patient age with D-dimer levels on admission and creatinine levels with SOFA score without GCS were predictors for renal replacement therapy. Conclusions. Using Explainable Boosting Machine analysis, we confirmed and weighed previously reported and identified novel predictors for outcome in critically ill COVID-19 patients. Using this strategy, predictive modeling of COVID-19 ICU patient outcomes can be performed overcoming the limitations of linear regression models.Trial registration. “ClinicalTrials” (clinicaltrials.gov) under NCT04455451
Subject(s)
Lung Diseases , Severe Acute Respiratory Syndrome , Thrombosis , Learning Disabilities , COVID-19ABSTRACT
Background: Some recipients of ChAdOx1 nCoV-19 COVID-19 Vaccine AstraZeneca develop antibody-mediated vaccine-induced thrombotic thrombocytopenia (VITT), associated with cerebral venous and other unusual thrombosis resembling autoimmune heparin-induced thrombocytopenia. A prothrombotic predisposition is also observed in Covid-19. We explored whether antibodies against the SARS-CoV-2 spike protein induced by Covid-19 cross-react with platelet factor 4 (PF4/CXLC4), the protein targeted in both VITT and autoimmune heparin-induced thrombocytopenia.Methods: Immunogenic epitopes of PF4 and SARS-CoV-2 spike protein were compared via prediction tools and 3D modelling software (IMED, SIM, MacMYPOL). Sera from 222 PCR-confirmed Covid-19 patients from five European centers were tested by PF4/heparin ELISA, heparin-dependent and PF4-dependent platelet activation assays. Immunogenic reactivity of purified anti-PF4 and anti-PF4/heparin antibodies from patients with VITT were tested against recombinant SARS-CoV-2 spike protein. Results: Three motifs within the spike protein sequence share a potential immunogenic epitope with PF4. Nineteen of 222 (8.6%) Covid-19 patient sera tested positive in the IgG-specific PF4/heparin ELISA, none of which showed platelet activation in the heparin-dependent activation assay, including 10 (4.5%) of the 222 Covid-19 patients who developed thromboembolic complications. Purified anti-PF4 and anti-PF4/heparin antibodies from two VITT patients did not show cross-reactivity to recombinant SARS-CoV-2 spike protein. Conclusions: The antibody responses to PF4 in SARS-CoV-2 infection and after vaccination with COVID-19 Vaccine AstraZeneca differ. Antibodies against SARS-CoV-2 spike protein do not cross-react with PF4 or PF4/heparin complexes through molecular mimicry. These findings make it very unlikely that the intended vaccine-induced immune response against SARS-CoV-2 spike protein would itself induce VITT.