Development and validation of prognostic machine learning models for short- and long-term mortality among acutely admitted patients based on blood tests.

Several scores predicting mortality at the emergency department have been developed. However, all with shortcomings either simple and applicable in a clinical setting, with poor performance, or advanced, with high performance, but clinically difficult to implement. This study aimed to explore if machine learning algorithms could predict all-cause short- and long-term mortality based on the routine blood test collected at admission. METHODS: We analyzed data from a retrospective cohort study, including patients > 18 years admitted to the Emergency Department (ED) of Copenhagen University Hospital Hvidovre, Denmark between November 2013 and March 2017. The primary outcomes were 3-, 10-, 30-, and 365-day mortality after admission. PyCaret, an automated machine learning library, was used to evaluate the predictive performance of fifteen machine learning algorithms using the area under the receiver operating characteristic curve (AUC). RESULTS: Data from 48,841 admissions were analyzed, of these 34,190 (70%) were randomly divided into training data, and 14,651 (30%) were in test data. Eight machine learning algorithms achieved very good to excellent results of AUC on test data in a of range 0.85-0.93. In prediction of short-term mortality, lactate dehydrogenase (LDH), leukocyte counts and differentials, Blood urea nitrogen (BUN) and mean corpuscular hemoglobin concentration (MCHC) were the best predictors, whereas prediction of long-term mortality was favored by age, LDH, soluble urokinase plasminogen activator receptor (suPAR), albumin, and blood urea nitrogen (BUN). CONCLUSION: The findings suggest that measures of biomarkers taken from one blood sample during admission to the ED can identify patients at high risk of short-and long-term mortality following emergency admissions.

Biomarkers and the post-thrombotic syndrome: A systematic review of biomarkers associated with the occurrence of the post-thrombotic syndrome after lower extremity deep venous thrombosis.

INTRODUCTION: Post-thrombotic syndrome (PTS) is a frequent chronic complication of deep venous thrombosis (DVT). Biomarkers are potentially valuable clinical tools for handling PTS. The purpose of this review was to examine which biomarkers are associated with the development of PTS in adults with lower extremity DVT. METHODS: We performed a systematic review of all English language prospective studies of biomarkers and PTS published in PubMed and EMBASE. Studies were included if diagnosing DVT by diagnostic imaging and assessing PTS by clinical scales, for example, the Villalta scale. Biomarkers of thrombophilia and pathological clot properties were not assessed. Data was reported qualitatively. RESULTS: 15 prospective studies were included. Studies varied widely in study design and methods of data analysis. Forty-six different biomarkers were examined, with seven being measured in two or more studies. The most frequently studied biomarkers were D-dimer, CRP, and IL-6. Associations between PTS and D-dimer were predominantly significant, while results on CRP and IL-6 were inconsistent. ICAM-1 was consistently associated with PTS in all studies and at all timepoints. IL-10 was significantly related to PTS development in the largest study and at all time points. Adiponectin, tPA, HRG and TAFI, MMP-1 and -8, and TIMP-1 and -2 were significantly associated with PTS in single studies. CONCLUSION: (1) Further research on biomarkers and PTS is clearly warranted. (2) Significant differences in study designs made it difficult to draw reliable conclusions regarding individual biomarkers. We suggest the implementation of a standardized framework for the study of biomarkers and PTS, to make comparison of future studies more feasible. (3) D-dimer, ICAM-1, IL-10, MMP-1 and 8, TIMP-1, TIMP-2, and adiponectin are clinical biomarkers of particular interest to include in future studies of PTS. Large scale systemic quantitative proteomic analyses of DVT patients could help identify novel biomarkers of interest in PTS-patients.

Soluble Urokinase Plasminogen Activator Receptor and Venous Thromboembolism in COVID-19.

Background Venous thromboembolism (VTE) contributes significantly to COVID-19 morbidity and mortality. The urokinase receptor system is involved in the regulation of coagulation. Levels of soluble urokinase plasminogen activator receptor (suPAR) reflect hyperinflammation and are strongly predictive of outcomes in COVID-19. Whether suPAR levels identify patients with COVID-19 at risk for VTE is unclear. Methods and Results We leveraged a multinational observational study of patients hospitalized for COVID-19 with suPAR and D-dimer levels measured on admission. In 1960 patients (mean age, 58 years; 57% men; 20% Black race), we assessed the association between suPAR and incident VTE (defined as pulmonary embolism or deep vein thrombosis) using logistic regression and Fine-Gray modeling, accounting for the competing risk of death. VTE occurred in 163 (8%) patients and was associated with higher suPAR and D-dimer levels. There was a positive association between suPAR and D-dimer (ß=7.34; P=0.002). Adjusted for clinical covariables, including D-dimer, the odds of VTE were 168% higher comparing the third with first suPAR tertiles (adjusted odds ratio, 2.68 [95% CI, 1.51-4.75]; P<0.001). Findings were consistent when stratified by D-dimer levels and in survival analysis accounting for death as a competing risk. On the basis of predicted probabilities from random forest, a decision tree found the combined D-dimer <1 mg/L and suPAR <11 ng/mL cutoffs, identifying 41% of patients with only 3.6% VTE probability. Conclusions Higher suPAR was associated with incident VTE independently of D-dimer in patients hospitalized for COVID-19. Combining suPAR and D-dimer identified patients at low VTE risk. Registration URL: https://www.clinicaltrials.gov; Unique identifier: NCT04818866.

suPAR Cut-Offs for Risk Stratification in Patients With Symptoms of COVID-19.

OBJECTIVES: Elevated soluble urokinase Plasminogen Activator Receptor (suPAR) is a biomarker associated with adverse outcomes. We aimed to investigate the associations between plasma suPAR levels (testing the cut-offs ⩽4, 4-6, and ⩾6 ng/mL) with risk of 14-day mortality, and with the risk of mechanical ventilation in patients that tested positive for SARS-CoV-2. METHODS: Observational cohort study of patients presenting with symptoms of COVID-19 at Department of Emergency Medicine, Amager and Hvidovre Hospital, Denmark from March 19th, 2020 to April 3rd, 2020. Plasma suPAR was measured using suPARnostic technologies. Patients were followed for development of mechanical ventilation and mortality for 14 days. Validation of our findings were carried out in a similar sized COVID-19 patient cohort from Mikkeli Central Hospital, Finland. RESULTS: Among 386 patients with symptoms of COVID-19, the median (interquartile range) age was 64 years (46-77), 57% were women, median suPAR was 4.0 ng/mL (2.7-5.9). In total, 35 patients (9.1%) died during the 14 days follow-up. Patients with suPAR ⩽4 ng/mL (N = 196; 50.8%) had a low risk of mortality (N = 2; 1.0%; negative predictive value of 99.0%, specificity 55.3%, sensitivity 95.2%, positive predictive value 17.4%). Among patients with suPAR ⩾6 ng/mL (N = 92; 23.8%), 16 died (17.4%). About 99 patients (25.6%) tested positive for SARS CoV-2 and of those 12 (12.1%) developed need for mechanical ventilation. None of the SARS-CoV-2 positive patients with suPAR ⩽4 ng/mL (N = 28; 38.8%) needed mechanical ventilation or died. The Mikkeli Central Hospital validation cohort confirmed our findings concerning suPAR cut-offs for risk of development of mechanical ventilation and mortality. CONCLUSIONS: Patients with symptoms of COVID-19 and suPAR ⩽4 or ⩾6 ng/mL had low or high risk, respectively, concerning the need for mechanical ventilation or mortality. We suggest cut-offs for identification of risk groups in patients presenting to the ED with symptoms of or confirmed COVID-19.

Soluble Urokinase Plasminogen Activator Receptor as a Decision Marker for Early Discharge of Patients with COVID-19 Symptoms in the Emergency Department.

BACKGROUND: Severe acute respiratory syndrome coronavirus 2 (The Covid-19 pandemic) strains health care capacity. Better risk stratification, with discharge of patients with a predicted mild disease trajectory, can ease this burden. Elevated blood-soluble urokinase plasminogen activator receptor (suPAR) has previously been shown to be associated with risk of intubation in confirmed COVID-19 patients. OBJECTIVE: To evaluate whether point-of-care measures of suPAR in patients presenting to the emergency department (ED) with symptoms of COVID-19 can identify patients that can be safely discharged. METHODS: Observational cohort study including all patients in the ED with symptoms of COVID-19 from March 19 to April 3, 2020. SuPAR was measured at first presentation. Review of electronic patient records 14 days after admission was used to assess disease trajectory. Primary endpoints were mild, moderate, severe, or very severe trajectory. The predictive value of suPAR, National Early Warning Score (NEWS), C-reactive protein (CRP), and duration of symptoms was calculated using receiver operating characteristics (ROC). RESULTS: Of 386 patients, 171 (44%) had a mild disease trajectory, 79 (20%) a moderate, 63 (16%) a severe, and 73 (19%) a very severe disease trajectory. Low suPAR was a strong marker of mild disease trajectory. Results suggest a cut-off for discharge for suPAR < 2.0 ng/mL if suPAR is used as a single parameter, and <3.0 ng/mL when combined with NEWS ≤ 4 and CRP < 10 mg/L. CONCLUSION: suPAR is a potential biomarker for triage and safe early discharge of patients with COVID-19 symptoms in the ED. suPAR can be used even before SARS-CoV-2 status is known.

Soluble Urokinase Receptor (SuPAR) in COVID-19-Related AKI.

BACKGROUND: AKI commonly occurs in patients with coronavirus disease 2019 (COVID-19). Its pathogenesis is poorly understood. The urokinase receptor system is a key regulator of the intersection between inflammation, immunity, and coagulation, and soluble urokinase plasminogen activator receptor (suPAR) has been identified as an immunologic risk factor for AKI. Whether suPAR is associated with COVID-19-related AKI is unknown. METHODS: In a multinational observational study of adult patients hospitalized for COVID-19, we measured suPAR levels in plasma samples from 352 adult patients that had been collected within 48 hours of admission. We examined the association between suPAR levels and incident in-hospital AKI. RESULTS: Of the 352 patients (57.4% were male, 13.9% were black, and mean age was 61 years), 91 (25.9%) developed AKI during their hospitalization, of whom 25 (27.4%) required dialysis. The median suPAR level was 5.61 ng/ml. AKI incidence rose with increasing suPAR tertiles, from a 6.0% incidence in patients with suPAR <4.60 ng/ml (first tertile) to a 45.8% incidence of AKI in patients with suPAR levels >6.86 ng/ml (third tertile). None of the patients with suPAR <4.60 ng/ml required dialysis during their hospitalization. In multivariable analysis, the highest suPAR tertile was associated with a 9.15-fold increase in the odds of AKI (95% confidence interval [95% CI], 3.64 to 22.93) and a 22.86-fold increase in the odds of requiring dialysis (95% CI, 2.77 to 188.75). The association was independent of inflammatory markers and persisted across subgroups. CONCLUSIONS: Admission suPAR levels in patients hospitalized for COVID-19 are predictive of in-hospital AKI and the need for dialysis. SuPAR may be a key component of the pathophysiology of AKI in COVID-19.