Background/Objective: The high predisposition to pulmonary embolism (PE) in CoViD-19-patients increases with its seriousness, even under anticoagulation. Because neither clinical nor laboratory parameters seem to be specific in this case, the CTPA stays the gold standard for further diagnostic, with the risk however of overuse. Our study investigates the incidence of PE in CoViD-19-patients, who underwent aCTPA, and to assess the link of several clinical and laboratory parameters with its occurrence.
Methods: We led a retrospective monocentric study in Johanniter Hospital in Duisburg-Rheinhausen between April 1st, 2020, and October 26th, 2021, in patients CoViD-19 positive tested with a RT-PCR. Data about examined CoViD-19 patients with CTPA by PE suspicion or accidentally revealed PE were collected, as well as clinical and laboratory parameters too.
Results: among 330 CoViD-19-patients 52 (15.8%) underwent a CTPA. 9 Patients (17.3% of 52/2.7% of 330) showed a PE. By 4 patients (7.7% of 52/1.2% of 330) PE could not be excluded. Only the Caucasian race (OR – 6.25; 95% CI 1.140-34.290; p = 0.035) was found to be independently associated with PE.
Conclusion: Clinical and laboratory parameters are insufficient to clarify the occurrence of PE in patients with CoViD-19. Because of its seriousness and its relative higher incidence in CoViD-19-patients a lower barrier based on clinical appearance of the disease degree is required in the future to evocate the possibility of a PE and to include earliest possible the performance of a CTPA for evaluation.
CoVid-19, Pulmonary embolism, Anticoagulation, Computer tomography pulmonary angiogram, Venous thromboembolism
VTE: Venous Thromboembolism; PE: Pulmonary Embolism; CTPA: Computed Tomography Pulmonary Angiogram; RT-PCR: Reverse Transcriptase Polymerase Chain Reaction; DVT: Deep Vein Thrombosis; CrCl: Creatinine Clearance; RHS: Right Heart Strain; ICU: Intensive Care Unit; CRP: C-Reactive Protein; LMWH: Low-Molecular-Weight Heparin; UFH: Unfractionated Heparin
The elevated morbidity and mortality of CoViD-19  seems to be associated i.a. with a high predisposition to VTE [2,3], principally by severe and critical course [3-6], and especially to PE instead of other patients .
Since both the differentiation between CoViD-19-related symptoms and aggravation due to PE by clinical appearance is not easy and laboratory parameters fail to allow a clear differentiation, the use of CTPA remains ineluctable. CTPA is indeed the most common and practical imaging modality, regarding its availability, its high specificity and sensibility. It allows furthermore to state the ill severity and to confirm or to eliminate differential diagnosis.
The challenge is to avoid the overuse of the CTPA, because of the requirement of contrast media (induction/aggravation of renal disease), of a high level of radiation and of additional costs.
The aim of this study was to evaluate the detection rate in the CTPA of PE in CoViD-19-patients who were examined with a CTPA, either presenting a clinical and/or laboratory PE-suspicion, or for other indications, and to determine the risk of PE considering the clinical and laboratory criteria, in order to refine the diagnostic procedure.
In a single centre setting we assessed the results of CTPA in CoViD-19-patients and the association of several parameters with the risk of PE.
We conducted a retrospective study in Johanniter-Krankenhaus-Rheinhausen - Duisburg (Germany), a medium-sized hospital (282 beds), evaluating all CTPA performed between April 1st 2020 and October 26th 2021 in admitted/examined patients, positive tested with RT-PCR for SARS-CoV-2 on nasopharyngeal swab, sputum, tracheobronchial secretion, oropharyngeal and on bronchoalveolar lavage. Excluded were patients who only were CoViD-19 positive tested, but neither admitted nor examined.
In order to compare our results and in the interests of reproductivity, we focused on the same criteria as prior studies and meta-analysis [7-10], an especially on the study of Filippi, et al. , that we considered as model. Patient records were analysed for information reported in Table 1, Table 2, Table 3, Table 4 and Table 5.
Table 1: Clinical and demographic features of the study population before CT-scan. View Table 1
Table 2: Multivariable regression analyses for the occurrence of pulmonary embolism. View Table 2
Table 3: Pulmonary embolism localisation. View Table 3
Table 4: In-hospital adverse events. View Table 4
Table 5: Findings in PE-Patients. View Table 5
CoViD-19-Patients in our study are classified in three ward categories as they were tested (Table 6):
Table 6: Care ward definition. View Table 6
Care setting definition:
The patients were then immediately isolated and assigned to an adapted care intensity before the CT-scan. This assignment may have been revised after the CT-scan depending on the infect burden and/or on the location of the pulmonary embolism. These care assignment criteria were defined in Table 7.
Table 7: Care setting definition. View Table 7
The classical assignment in the different care categories was defined by the DIVI (German Interdisciplinary Association for Intensive Care and Emergency Medicine) in 2017 . In an unprecedented situation such as the CoViD-19-pandemia, the RKI (Robert-Koch-Institute) published adapted WHO-Guidelines [12,13] to the CoViD-19-management, in which the assignment criteria to sort CoViD-19-patients were detailed. Although, we had to modify permanently our intern care assignment criteria, according to the pandemic situation, and took decision on a case-by-case basis, in order to avoid patients' overflow and to save resources for the serious to critical patients, especially at the beginning of the pandemic and during the different waves. In the same optic were the patients discharged as soon as possible.
The doses of heparin were defined as following (Table 8):
Table 8: Heparin doses. View Table 8
a) CT-indication: With incidental findings as exception, the indication to CTA was decided in correlation with the symptomatic primarily, the clinical information, including the Wells-score, and the laboratory results. Some patients were immediately examined with a CT-scan after admission. We defined the admission day as d0. Other patients underwent a CTA during the hospitalisation, mainly by worsening or delayed recovery.
b) CT-Protocol: A 16-row scanner was used for the acquisition. The bolus was tracked in the truncus of the pulmonary artery after injection of 50 to 100 mL of non-ionic monomeric iodine contrast media, with a threshold of 160 to 250 HU.
Mediastinal and parenchymal windows were multiplanar reconstructed with slice-thickness of minimum 0.75 mm.
c) PE-Classification: PE was classified as main pulmonary, lobar, segmental or subsegmental based on the location of the most proximal luminal defect.
d) CT-Infect burden evaluation: Many scores were described in the literature to assess the severity of the infect burden in the CT-Scan, with different cut-off values [16-19].
We used the CT-Severity index score to visually assess the infect burden in the lung parenchyma on the CT-scan. We choose to distribute the patients in four categories depending on the estimated percentage of the affected lung volume (Table 9). Our cut-off value to define severe infect was 50%, as advocated by Ran, et al. and Fonseca, et al. [16,17]. However, without the help of AI remains our estimation unprecise.
Table 9: CT-Severity index score. View Table 9
Ultrasound: An echocolour Doppler of the lower extremities was performed by patients with concomitant symptoms of DVT.
Patients with hemodynamic instability underwent an echocardiography before the CT-scan.
Normally distributed variables were communicated as mean ± standard deviation. Non-normal distributed parameters were reported as median and interquartile range.
Categorical data were expressed as quantity (total number) and percentage.
The Student t-test was used to compare continuous normally distributed variables, whereas chi-squared test or Fisher exact test was used to compare categorical variables.
To assess the individual and independent association of PE with clinical and laboratory parameters, univariable and multivariable logistic regression analyses were set and presented as odds ratio (OR) with 95% confidence interval (CI).
No relevant collinearity was found among the covariates by the variance inflation factors.
As a candidate for the multivariate analysis a parsimonious model was employed including variables with p < 0.05 by the univariate test.
In order to evaluate the in-hospital mortality risk in patients with vs. those without PE, multivariable logistic regression analyses were applied and the results were communicated as odds ratio (OR) with 95% confidence intervals (CI).
330 patients admitted/examined in our institution were tested positive for SARS-CoV-2 with RT-PCR. The mean age was 68.4 and the male sex rate was 54%.
In 52 patients (15.8% of 330) a CTPA was performed. The PE was verified in 9 CoViD-19-patients (17.3%).
In 4 patients (7.7% of 52) the CTPA was non-conclusive as a widespread thromboembolic burden couldn't be excluded at the subsegmental level of PE. Because of respiratory artefacts, beam hardening, insufficient contrasting and/or by pronounced consolidations, their CTPA showed multiple defects in the vascular lumen, without any proven evidence of a complete abort of the distal contrasting.
We summarized our results in the Table 1, Table 2, Table 3, Table 4 and Table 5.
A low pCO2 (< 35 mmHg) before the CT scan was found in none of our PE-patients, but in 17.9% of the non-PE-patients (p < 0.003).
No further significant differences concerning gender, age, nutritional status, in-hospital setting, or several clinical and laboratory parameters could be determined.
The thrombus was in 55.6% of cases central, i.e. in the main and/or the lobar pulmonary arteries, and 44.4% in the subsegmental arteries. There was no correlation between the location/burden of the thrombus and the symptomatic. None of the PE was restricted to the segmental arteries (Table 2).
In our study we observed an incidence of PE in 17.3%, which is higher compared to other publications [7-9], but lower to a recent meta-analysis . It remains higher than the one observed in general ward in non-CoViD-19-patient [20-23]. The proportion of severely to critically affected CoViD-19-patients (38.5% of our cohort) may contribute to the higher incidence. Studies found a proportional increase of PE-rate to the severity of CoViD-19 [2,9,10].
Most of the CTPA was performed in CoVid-19-patients who presented unspecific clinical symptomatic. They may occur due to the infection itself or to a PE, principally by worsening respiratory pattern such as hypocapnia, dyspnoea or tachypnoea. A further hint was a positive preclinical score result. Surprisingly, PE-patients were less likely to have a high (> 4) Wells-score of PE (33.3% versus 38.5%), and the PE was found fortuitously in three asymptomatic patients who underwent CTPA for other indications. No statistically significant correlation was found between the development of the PE and clinical presentation, or between PE and clinical complications, such as intensification of care or mortality, which is consistent with the finding of Filippi, et al. .
Primarily in elderly, the SARS-CoV-2 infection involves the whole organism, which alters significantly the haemostatics  and may explain the advanced age our cohort. Compared to other studies [3,25], age and sex have no influence on the risk of PE in our patients. From all of the investigated parameters, only the Caucasian race turned out as a statistically significant positive correlation (p = 0.04). This may be a reflection of the well-known correlation to different mutations (such as Factor V Leiden and prothrombin gene G20210A) but out of the available clinical data, a further differentiation of our patients was not possible.
This risk may further increase through some medications like steroids .
Most reports found the frequent involvement of the peripheral arteries without signs of DVT, which favours the local origin of the thrombosis [2,3,6,10,27,28]. 44.4% of the PE in our study were subsegmental.
A study showed that DVT does not necessarily lead to thromboembolism, and remains “sleeping” without clinical manifestations . The COVID-19-induced DVT are able to cause complications and are resistant to anticoagulant therapy , as shown in one of our patients.
As found in the literature, the majority of our CoViD-19-patients presented elevated D-dimer and CRP . In most of PE-Patients the D-dimer level varied in-between 1500 and 4000 ng/mL (44.4%). Thus, the sensibility of the D-dimer as a standalone parameter is argued. Whereas some authors rely on D-Dimer analysis , sometimes with different cut-off levels [3,10], others doubt whether D-dimer is rather a marker of thromboembolic complications in a clinical scenario dominated by (severe) infectious/inflammatory disorder . In fact, D-dimer increases proportionally with the severity of CoViD-19, independent of the presence of PE .
Furthermore, in spite of prophylactic/therapeutic anticoagulation, many reports underlined the high incidence of thrombotic COVID-19-complications such as microvascular thrombosis, stroke and venous thromboembolic disease [2,9,27,30], especially PE [3,31-36], most notably in severely ill CoViD-19-patients with pneumonia admitted to the ICU , in comparison to other diseases . This was the reason why many studies support the use of higher doses of anticoagulants, while some advocate an immediate onset of therapeutic anticoagulation in CoViD-19-patients, at least in those admitted to the ICU [25,26,38,39]. The results of a randomised clinical trial reported no clear advantage of intermediate over preventive doses of enoxaparin in severe ill patients . Our findings may be concordant with these observations, since our coagulated PE-patients received exclusively therapeutical heparin doses. It remains to determine the heparin dosage with the best benefit-to-risk ratio . Only in two patients a central PE was resistant to a long-term intensive anticoagulation and to a thrombolytic treatment, which is consistent with the finding of Kerbikov, et al.  and of Nahum, et al. .
The potential effect of antiplatelet drugs on the prevention of the inflammatory activation of the thrombocytes is still controversial [41-44]. Because of the small number of concerned patients, we could not debate on this subject.
Some authors suspect a synergic effect of CoViD-19 and active cancer [45-47] since the cancer and some oncological therapy itself are thrombogenic. In our cohort no relevant difference was found between the PE- and non-PE-population concerning the prevalence of an active cancer (22.2% vs. 15.4%, p = 0.337).
The benefit of the cardiac echography was underlined in a recent study . We could not verify this observation, as the echocardiography was performed in only one PE-patient and only in 3 non-PE-patients.
Since the literature is controversial and our results do not indicate a certain pattern, the impact of PE on patient prognosis remains unclear [9,27]. Table 4 indicates the lack of influence of PE on death from any cause, on increase of oxygenation modality or on increase of care intensity.
The limits of our study are related once to the retrospective design of the study but even more important to the limited number of patients in a single centre setting, which implies numerous possible biases that are not accounted for. Nevertheless, the findings about the lack of a flow-chart like decision-making are important information.
Finally, because of the high risk of exposure at the time, thromboembolic events may have been underreported due to the inability to pursue the appropriate imaging tests.
Aim of our study was to find a couple or certain parameters indicating a high incidence for PE in CoViD-19-patients, who presented clinical and/or laboratory criteria, which could indicate a VTE. We failed to show any relevant parameter able to predict the prevalence of PE. Most of them are unspecific, principally by a constellation dominated by infection and respiratory compromise.
Many studies support local inflammation being the onset of PE in CoViD-19-patients since the majority of lesions affects the peripheral level. This level is associated with a limited detection by CTPA. This may contribute to an underestimation of the proportion of PE in CoViD-19-patients. Nevertheless, if the PE extends to the subsegmental level, CTPA represents a valuable imaging technique, which enables a tailored therapy in critical ill CoViD-19-patients.
Since we aspire constantly to optimise patient care and to adapt to the still changing SARS-CoV-19, we could try to reconsider our PE-diagnostic strategy by refining our clinical evaluation and by using other diagnostic means, such as echocardiography in haemodynamic unstable patients, as recommended in the guidelines of the ESC, to avoid the overuse of the CTPA.
Special thanks to Dr. Welskop and his team preparation and providing the test results.
No funds, grants, or other support was received.
All authors have no relevant financial or non-financial interests to disclose.
According to § 15 of the professional regulations in North Rhine-Westphalia (Germany), an ethics consultation is not required in anonymised retrospective data collections studies.
All authors have contributed equally.