key: cord-0701026-5zu0c32g
authors: Riyahi, Sadjad; Dev, Hreedi; Behzadi, Ashkan; Kim, Jinhye; Attari, Hanieh; Raza, Syed I.; Margolis, Daniel J.; Jonisch, Ari; Megahed, Ayah; Bamashmos, Anas; Elfatairy, Kareem; Prince, Martin R.
title: Pulmonary Embolism in Hospitalized Patients with COVID-19: A Multicenter Study
date: 2021-07-13
journal: Radiology
DOI: 10.1148/radiol.2021210777
sha: ba38c81b0269e4782dd09d63dafd3b865c97930d
doc_id: 701026
cord_uid: 5zu0c32g

BACKGROUND: Pulmonary embolism (PE) commonly complicates SARS-CoV-2 infection but there is heterogeneity in incidence and mortality in the single center reports and risk factors. PURPOSE: To determine the incidence of PE in COVID-19 and its associations with clinical and laboratory parameters. METHODS: Electronic medical records were searched retrospectively for demographic, clinical and laboratory data and outcomes in patients admitted with COVID-19 at 4 hospitals March-June 2020. PE on CT pulmonary angiography (CTPA) and perfusion scintigraphy was correlated with clinical and laboratory parameters. D-dimer was used to predict PE and the obtained threshold underwent an external validation on 85 hospitalized patients with COVID-19 at a 5(th) hospital. We also assessed the association between right heart strain and embolic burden in patients with PE undergoing echocardiography. RESULTS: Four-hundred-thirteen patients with COVID-19 (230 men, aged 20-98 years, mean + SD = 60+16 years) were evaluated. PE was diagnosed in 25% (102/413, 95%CI: 21%-29%) of hospitalized patients with COVID-19, undergoing CTPA or perfusion scintigraphy. PE was observed in 29% (21/73, 95%CI:19% -41%) of ICU vs. 24% (81/340, 95%CI:20% -29%) of non-ICU patients (p=0.37). PE was associated with male sex(Odds Ratio=OR[95% CI]: 1.7[1.1–2.8], p=0.02), smoking (OR[95% CI]:1.8[1.01–3.4], p=0.04) and increased d-dimer (p<0.001), lactate dehydrogenase (p<0.001), ferritin (p=0.001) and IL-6 (p=0.02). Mortality in hospitalized patients was similar between those with PE and without PE (14%[13/102, 95%CI: 8% - 22%]vs 13%[40/311, 95%CI: 9% - 17%], p=0.98), suggesting that diagnosis and treatment of PE was not associated with excess mortality. D-dimer>1600 ng/mL predicts PE with 100% sensitivity and 62% specificity in an external validation cohort. Embolic burden was higher in patients with right heart strain among the patients with PE undergoing echocardiogram (p=0.03). CONCLUSION: Pulmonary embolism incidence was 25% in patients hospitalized with COVID-19 suspected of PE. D-dimer>1600ng/mL was sensitive for identifying which patients need CTPA. See also the editorial by Ketai.

I n p r e s s Abstract Background: Pulmonary embolism (PE) commonly complicates SARS-CoV-2 infection but there is heterogeneity in incidence and mortality in the single center reports and risk factors.

Purpose: To determine the incidence of PE in COVID-19 and its associations with clinical and laboratory parameters.

Methods: Electronic medical records were searched retrospectively for demographic, clinical and laboratory data and outcomes in patients admitted with COVID-19 at 4 hospitals March-June 2020. PE on CT pulmonary angiography (CTPA) and perfusion scintigraphy was correlated with clinical and laboratory parameters. D-dimer was used to predict PE and the obtained threshold underwent an external validation on 85 hospitalized patients with COVID-19 at a 5 th hospital. We also assessed the association between right heart strain and embolic burden in patients with PE undergoing echocardiography.

Results: Four-hundred-thirteen patients with COVID-19 (230 men, aged 20-98 years, mean + SD = 60+16 years) were evaluated. PE was diagnosed in 25% (102/413, 95%CI: 21%-29%) of hospitalized patients with COVID-19, undergoing CTPA or perfusion scintigraphy. PE was observed in 29% ( that diagnosis and treatment of PE was not associated with excess mortality. D-dimer>1600 ng/mL predicts PE with 100% sensitivity and 62% specificity in an external validation cohort. Embolic burden was higher in patients with right heart strain among the patients with PE undergoing echocardiogram (p=0.03).

Introduction SARS COV-2 binds the ACE-2 receptors on endothelial cells, especially within kidney heart, lung, and liver.(1) Endothelial cell damage leads to thrombosis, which can be a defense mechanism that compartmentalizes infection and prevents further dissemination. (2) The purpose of this research is to determine the multi-center incidence of pulmonary embolism in COVID-19 and its associations with clinical and laboratory parameters. Secondary objectives are to assess the predictive value of d-dimer and the relationship between right heart strain and clot burden. We evaluate patients with COVID-19 admitted at 4 hospitals and develop a predictive model for identifying patients with COVID-19 at high risk for PE, validating with data from another institution. There is some overlap in the patient population with a letter utilizing 41 patients with PE from the same COVID-19 database to assess the risk and d-dimer cut-off of venous thromboembolism (9) .

Data extracted included RT-PCR diagnosis date as well as demographic, clinical and laboratory data, collected within 72 hours of imaging. These included age, gender, race/ethnicity, body mass index (BMI), smoking, d-dimer, fibrinogen, ferritin, C-reactive protein, Creatinine Kinase, platelets, prothrombin time (PT) and activated partial thromboplastin time (PTT), heart rate, systolic and diastolic blood pressure, arterial %O2 saturation, duration and route of supplemental O 2 administration, past medical history of chronic lung disease, cardiovascular disease and chronic kidney disease, hypercoagulability and DVT, diabetes mellitus, human immunodeficiency virus (HIV) and anticoagulation use at least 48 hours before imaging. For hypercoagulability, diagnosis was further confirmed by checking the ICD10 codes including D68 and I82 and their derivatives for thrombophilia and chronic thromboembolism in the electronic medical record.

We also extracted, intubation date, # of days on ventilator, discharge status and/or death. Prospectively generated chest CT reports were reviewed to extract the following imaging features: pulmonary embolism, ground glass opacities and its locations, crazy-paving, septal thickening without mention of crazy paving, consolidation, air bronchograms, pleural effusion, right heart strain and any artefacts. Right heart strain was defined based upon reports from echocardiography performed within 1 day of CT or perfusion scintigraphy for those patients in whom it was available. The prospectively reported location of pulmonary emboli was used to determine a score of embolic burden using the Qanadli method (10), reviewing the images as necessary to resolve any emboli location reporting ambiguities.

CTPA was performed on 64 or 256 slice scanners (Optima, Discovery or Revolution GE Healthcare) during breath holding with injection of 70ml to 100ml iohexol (omnipaque) 300 or ioversol (optiray) 350.

Bolus timing to peak pulmonary artery contrast enhancement was performed using smartprep sampling on the mid superior vena cava and right pulmonary artery. Images were reconstructed in the axial plane at 1.25mm and 2.5mm slice thicknesses as well as 8mm MIPs and in coronal and sagittal planes at 2mm slice thickness and with 5mm MIPs. Reporting of the exams was performed primarily by fellowship trained chest radiologists working together with residents at 2 of the hospitals and by general radiologists at the other hospitals. All nuclear medicine scans were interpreted by fellowship trained nuclear medicine radiologists.

Perfusion scintigraphy was performed using 4mC Technetium 99m macro-aggregated albumin. Imaging was obtained using spect-CT or multiple planar images.

Continuous variables were tested for normality and reported as either mean ± standard deviation or median (interquartile range). Student's t-test (or Mann-Whitney U test) was used to compare the I n p r e s s difference between the groups for these variables. Categorical variables were reported as frequency (percentage) and the significance of between-group comparisons was assessed using chi-squared test.

We used a subsample of 408 patients who underwent a chest CTPA to build a random forest model predicting acute PE. We did not include the five patients with perfusion scintigraphy in this analysis to maintain the consistency in the method of outcome detection. Missing values were imputed in variables with less than 10% missing data points using the missForest(11) method of imputation. The total study sample was randomly divided into a training set maintaining the outcome proportion, including 2/3 of the patients, and a test set (i.e. the remaining 1/3) for model building and internal validation, respectively. As acute PE was a rare event, synthetic minority oversampling technique (SMOTE) was applied in the training set to balance the data, by up-sampling the minority group (12) . A random forest model was fitted on the balanced training set for prediction of acute PE. The independent variables were age, gender, BMI, smoking, past medical history of chronic lung disease, cardiovascular disease and VTE/hypercoagulability; COVID-19 symptoms to CTPA interval (days), Hospital stay (days), thromboprophylaxis at least 48 hours before CTPA, days on ventilator, intubation, need for supplemental oxygen at arrival, noninvasive ventilation, heart rate, systolic blood pressure, %O2 saturation, d-dimer, CRP, fibrinogen, ferritin, platelets, PT and PTT. Variable importance was determined using permutation importance (i.e. mean decrease in accuracy). A 10 by 10-fold cross-validation was used to train the model on the training set. Predictions were then made using the random forest model on the test set. The area under (AUC) the receiver operating characteristic (ROC) curve was calculated to assess the model performance on the validation set. Youden's index was then used to threshold the predictions for diagnosis of acute PE.

Univariate logistic regression was used to assess the predictive performance of d-dimer as a continuous variable. Subsequently, sensitivity analysis was performed using Youden's J index to determine the optimal cut point in d-dimer levels for PE prediction using d-dimer alone. This cutoff was then externally validated using the external validation cohort described above.

The significance level was set to 0.05 and the statistical analysis was performed in R version 4.0.2 (R Core Team, Vienna, Austria, 2020).

From March 3 to June 5, 2020, SARS-CoV-2 RT-PCR tests (n=25,335) at 4 hospitals identified 8,460 positive cases of which 4131 were admitted (Figure 1 ). The research data repository had detailed information including laboratory data and vital signs of all patients with positive RT-PCR tests. For a subset of 413 patients who underwent CTPA (n=408) or perfusion scintigraphy (n=5), laboratory data and vital signs data within 72 hours of ordering the imaging were collected. CT reports described reduced diagnostic confidence in 40 studies due to motion artifact (n=30), sub-optimal bolus timing (n=11), parenchymal disease (n=6) and streak artifact (n=5; note some patients had more than one issue). In these scans diagnostic confidence was reduced primarily for segmental and subsegmental arteries. Nuclear medicine perfusion scintigraphy was used to evaluate for PE in 5 patients hospitalized with COVID-19 infection. Ventilation scanning was not performed due to COVID-19 contamination risk so the assessment for matched defects had to be made based on comparing to chest X-ray findings and any other imaging available.

The incidence of PE diagnosis was 25% (102/413, 95%CI: 21% -29%) in patients who underwent CTPA or perfusion scintigraphy. For patients in the intensive care unit (ICU) at the time of imaging the incidence was 29% (21/73, 95% CI:19% -40%) and for non-ICU patients, the incidence of PE was 24% (81/340, 95% CI:20% -29%). This difference was not significant (p=0.37). The demographic, laboratory, vital signs, pre-existing conditions, and outcome data for 102 patients with PE and for 311 patients with negative CTPA or perfusion scintigraphy are shown in Table 1a Table   2 ). There was a trend towards significantly higher levels of PT and creatinine kinase among those with PE. CTPA findings are summarized in Table E1 (Appendix E1). (Figures 2 and 3) . Table E2 (Appendix E1) .

Subsequently, we analyzed the utility of d-dimer alone for predicting PE in 383 patients without missing d-dimer measurement (PE+=98, PE-=285, d-dimer was missing in 30 patients, 7%). The univariable logistic regression model showed AUC=0.75 (95% CI: 69% -80%), sensitivity= 83% (95% CI: 73% -I n p r e s s accuracy=63% (95%CI: 58% -68%), using d-dimer as a continuous variable. The optimal cutoff point was 1600 ng/mL using the Youden's index in the entire internal sample.

Applying the d-dimer cutoff above to the external validation cohort, showed a sensitivity= 100% (True positive rate=32/32, 95%CI: 89% -100%) , a specificity= 62% (True negative rate=20/53, 95%CI: 48% -73%) and accuracy=76% ( 95%CI : %66% -85%) for diagnosing PE.

In this multi-center study of 413 patients hospitalized with COVID-19 and suspected of PE, we found pulmonary embolism in 25% (95%CI: 21% -29%). Although use of CT pulmonary angiography and perfusion scintigraphy in COVID-19 can be cumbersome due to the complex logistics of transporting sick infected patients to the radiology department, there may be a benefit since we observed no excess mortality when pulmonary embolism was diagnosed and treated. Our random forest model demonstrated d-dimer to be the dominant predictor with d-dimer >1600 ng/mL having100% sensitivity and 62% specificity for diagnosing pulmonary embolism on external validation, when used alone. PE positive patients with echocardiography confirmed right heart strain had higher embolic burden compared to those without right heart strain.

Our 25% incidence of PE in all hospitalized patients with COVID-19 is higher than the 17.6% (95% CI: 12.3% -23.5%) overall PE incidence, and closer to 21.7% (95%CI:14.8% -29.3%) incidence in the severe group, reported by Liu et al in a systematic review and meta-analysis. (7) We also did not find significantly higher PE incidence in ICU patients. This might reflect that hospitalized patients in our multicentric cohort were all on the severe end of the spectrum. The large range in PE incidence reported by Liu et al reflects the heterogeneity and small sample size of the studies included, with most studies having less than 100 subjects from single institutions.

We found that male gender was significantly associated with PE, in line with previous reports. (7, I n p r e s s analysis including over three million reported global COVID-19 cases, men were almost three times more likely to require intensive treatment unit and had 40% higher odds of death than women. (15) This female advantage in COVID-19 is suggested to be accounted for by the sex differences in the immune system. (15) While the effects of androgen on the endothelial functioning was also suspected as a contributing factor, making the male gender a potential risk factor for VTE, more so in the context of COVID-19. (7) D-dimer was higher in the patients with COVID-19 and PE compared to negative studies (16) and this has been confirmed in nearly every publication on this topic (9, 17) . D-dimer measures fibrin break down products and has been utilized for prediction and prognosis of venous thromboembolism among patients with COVID-19. (7, 9, (17) (18) (19) (20) (21) One study showed d-dimer peaks as the measurements approached the day of ultrasound examination for DVT, following an inverse-U shape. (19) It is thought that the sensitivity of d-dimer for diagnosis of VTE diminishes with time. (17, 22) The 1600 ng/mL d-dimer threshold we found was lower than most of the thresholds previously reported for PE or VTE prediction, ranging from 2600 to 7500 ng/mL (9, 16, 23) , and closer to the 1500 ng/mL cutoff proposed by Cui et al. Prior studies report about half of the patients with PE have some degree of right ventricular compromise(25) and increased right heart strain is associated with PE and syncope as well as higher mortality.(26) Among 31 PE positive patients with COVID-19 with available echocardiography, right heart strain was associated with higher embolic burden. Since an estimated 25% of the pulmonary vasculature needs to be occluded to result in pulmonary hypertension, and acute right heart failure requires more than 50% occlusion (27), it is not surprising to see higher rates of RV strain in these patients whose pulmonary vasculature is already compromised by infection. Right ventricle systolic dysfunction was previously reported in patients with COVID-19 without PE,(28, 29) due to the release of I n p r e s s vasoactive mediators such as serotonin, thromboxane and histamine in response to the acute hypoxic injury and platelet rich clots (30) which may also contribute to right heart strain in these patients.

Retrospective data collection was the main limitation of this study, making it difficult to control for factors influencing the outcomes, including severity of the disease, treatment protocols and regular laboratory and clinical data collection. Further, only including patients undergoing imaging for suspected PE might not reflect the true incidence of PE in the hospitalized patients with COVID-19. Several variables were excluded from the random forest analysis (including PaO2, IL-6, CRP, BNP, Fibrinogen, ESR, CK and Troponin I) for having more than 10% missing values and we used imputation for those with missing values less than this threshold. Prophylactic anticoagulation may have influenced the level of d-dimer in patients suspected for PE. Further, due to the limited data on prophylactic anticoagulation dosage and a lack of randomization in treatment assignment, our analysis of the effect of prophylactic anticoagulation dose is limited. Finally, lack of echocardiography within 24 hours in all PE positive patients prevented us from calculating incidence of right heart strain among all patients.

In conclusion, our results indicate a high incidence of pulmonary embolism in hospitalized patients with COVID-19 undergoing CTPA or perfusion scintigraphy which is even more prevalent in males and smokers. D-dimer >1600 ng/mL is useful for identifying patients likely to have PE. Right ventricular strain was associated with higher embolic burden.

All authors declared no conflict of interest.

Guarantors Ordieres-Ortega L, Demelo-Rodríguez P, 

ACE2 receptor polymorphism: Susceptibility to SARS-CoV-2, hypertension, multi-organ failure, and COVID-19 disease outcome

Endothelial cell infection and endotheliitis in COVID-19

Venous thromboembolism among hospitalized patients with COVID-19 undergoing thromboprophylaxis: a systematic review and meta-analysis

Clinical characteristics and outcomes of venous thromboembolism in patients hospitalized for COVID-19: systematic review and meta-analysis. Thrombosis Update

Acute pulmonary embolism associated with COVID-19 pneumonia detected with pulmonary CT angiography

The Incidence, Prognosis and Laboratory Indicators of Venous Thromboembolism in Hospitalized Patients with COVID-19: A Systematic Review and Meta-analysis

COVID-19 complicated by acute pulmonary embolism and right-sided heart failure

D-dimer cut-off points and risk of venous thromboembolism in adult hospitalized patients with COVID-19

New CT index to quantify arterial obstruction in pulmonary embolism: comparison with angiographic index and echocardiography

MissForest-non-parametric missing value imputation for mixedtype data

SMOTE: synthetic minority over-sampling technique

Deep vein thrombosis in hospitalized patients with coronavirus disease 2019

Pulmonary embolism in COVID-19 patients: a French multicentre cohort study

Male sex identified by global COVID-19 metaanalysis as a risk factor for death and ITU admission

Incidence of pulmonary embolism in noncritically ill COVID-19 patients. Predicting factors for a challenging diagnosis

D-dimer in patients infected with COVID-19 and suspected pulmonary embolism

D-dimer is associated with severity of coronavirus disease 2019: a pooled analysis

Clinical characteristics of acute lower extremity deep venous thrombosis diagnosed by duplex in patients hospitalized for coronavirus disease 2019

Deep vein thrombosis in hospitalized patients with coronavirus disease 2019 (COVID-19) in Wuhan, China: prevalence, risk factors, and outcome. Circulation

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. The lancet

U test was used due to non-normal distribution

I n p r e s s 

The ROC curve analysis for the prediction of acute PE among the 136 patients in the test sample revealed an AUC of 0.78 (95%CI: 0.69 -0.87). Using the Youden's index, the model classified the patients into PE and no PE outcomes with an accuracy of 71% (95%CI: 62% -78%) and sensitivity and specificity = 82% (95%CI: 63% -94%) and 68% (95%CI: 58% -76%), respectively (Supplemental table   2 ). The specificity was 32% at the sensitivity of 95%, and the sensitivity was 18% at the specificity of 95%.I n p r e s s