1. Introduction

COVID-19 pneumonia, caused by the virus SARS-CoV-2 (severe acute respiratory syndrome coronavirus-2), was first reported in China in December 2019 and spread rapidly in many countries until it was declared a pandemic by the WHO on 11th March 2020 [1]. SARS-CoV-2 infection is highly contagious and is transmitted from human-to-human through respiratory droplets and contact [2]. The main symptoms are fever, dry cough, fatigue, and malaise. In the advanced stage of the disease, patients may develop dyspnoea and respiratory distress syndrome (ARDS). However, it is the heterogeneity of the symptoms which poses a serious challenge to the healthcare providers as regards the most appropriate clinical management [3].

Computed Tomography (CT) plays an important role in the diagnosis and follow-up of COVID-19 pneumonia, thus allowing the best clinical management. Current guidelines advocate the use of the high resolution unenhanced chest CT (HRCT) in COVID-19 patients thanks to its elevated accuracy in detecting pulmonary changes in viral pneumonia [3, 4], also permitting differential diagnoses [5–7].

According to the literature, repeated CT follow-ups in COVID-19 patients demonstrate the transition from pure ground glass opacities (GGO) to GGO with consolidation as the disease’s most frequent course [8, 9]. Reports of acute pulmonary artery embolism (PAE) associated with COVID-19 have emerged in the literature since the onset of the disease [10–13]. The hypothetical pathogenesis for SARS-CoV-2-induced thrombosis includes a disease-specific hypercoagulable state, diffused cytokine-mediated microvascular damage, and, in a few cases, reactive thrombocytosis [14, 15].

Indeed, due to its elevated accuracy in detecting the embolic filling defects in the pulmonary arteries, CT Pulmonary Angiography (CTPA) has become fundamental in COVID-19 patients with suspected PAE [2, 3, 16, 17].

The purpose of our study was to assess the role of CTPA in the detection of pulmonary artery filling defect (PAFD) consistent with PAE and to investigate possible radiological or clinical predictors of PAFD in a cohort of patients with COVID-19 infection in a “COVID hospital” in southern Italy during the pandemic.

2. Materials and Methods

3. Results

The clinical, biological, and imaging features are summarized in Tables 1, 2, and 3.

PAFDs were diagnosed in 15/43 (35%) patients. PAFD was located in the main pulmonary artery in 1/15 (7%) patient (Figure 1), in 2/15 (13%) patients in lobar arteries (Figure 2), in 6/15 (40%) patients in segmental arteries, and in 6/15 (40%) in subsegmental branches (Figure 3). The number of cases receiving prophylactic enoxaparin therapy at the moment of CTPA as well as the dosage of enoxaparin was not significantly different between patients with and without PAFD.

Parenchymal pathological findings related to pulmonary COVID-19 disease were evident in 39/43 (91%) patients. In 32/43 patients (74%), more than 3 pulmonary lobes were involved. The number of patients with pure GGO or pure consolidation patterns was the same (15/43, 35%), while GGO+consolidation was detected in 7/43 (15%) patients (Figures 1 and 2). Crazy paving pattern and linear opacity were found, respectively, in 5/43 (12%) and 22/43 (51%) patients (Figure 3). Pleural effusion was diagnosed in 11/43 (26%) patients.

D-dimer values in both measurements recorded (at admission and within 24 h of CTPA) and the mean value of the same were significantly higher in patients with PAE. Similarly, the revised Geneva score was higher in the patients with acute pulmonary embolus than in those without embolus (mean 2 ± 2 versus 4 ± 2, p = 0.011). No other statistically significant differences were found between patients with and without PAFD for all the remaining variables; namely, no differences were found in D-dimer values for patients with PAFD located in arterial branches of different grades according to an ANOVA test. ROC analysis demonstrated that the variable with a higher predictive value was the mean D-dimer value (AUC 0.831) and that a cut-off value > 1800 μg/l of this parameter predicted the probability of PAFD with a sensitivity and specificity of 70% and 78%, respectively (Table 4 and Figure 4).

4. Discussion

Our study demonstrated PAFD, consistent with PAE, in 15/43 (35%) consecutive CTPAs performed in COVID-19 patients; our results are in agreement with previous reports by Grillet et al. (a 23% prevalence of PAE on 100 CTPAs) and by Leonard-Lorant et al. (a 30% prevalence of PAE on 106 CTPAs) [2, 16]. This rate of PAE is higher than that of 3%-10% encountered in normal circumstances, when the ever-increasing availability of CTPA favours its inappropriate use, resulting in a low prevalence of PAE [21, 22]. The high prevalence of PAFD in our study is in agreement with other authors’ hypothesis of a coagulopathy associated with COVID-19 infection and with the hypothesis that pulmonary vessel obstruction, detected by CTPA, may represent pulmonary thrombi rather than emboli, as they are not fully occlusive and are related to a low number of deep venous thromboses [23]. It is still unknown if this phenomenon is related to a direct endothelial cell involvement by the virus or to an inflammatory reaction following alveolar damage [24, 25]. This pathogenetic mechanism might also explain the overall lower predictivity of clinical scores, such as the Wells score, compared with recent papers [26]. Similar conclusions were reached by Poissy et al. who demonstrated a 22% prevalence of PAE in patients admitted to ICU during the pandemic period [27]. In our experience, no CT scans were performed in patients requiring intubation, who were mostly managed clinically, using pulmonary US for imaging. Consequently, our results confirm a higher prevalence of PAE even in less severe patients, strengthening the possibility of a disease-specific hypercoagulable condition. Most of the arterial filling defects found in our study were in segmental and subsegmental vessels. However, the severity of the pneumonia and the parenchymal CT findings was not significantly different in patients with or without PAE. Indeed, most of them showed a bilateral and multilobar involvement of the lungs. Subpleural consolidation, which was advocated as a parenchymal predictor of PAE, was not significantly associated with this complication [28]. In our study, BMI does not correlate with PAFD, in contrast with a recent paper by Poyiadji et al. [29]. The main reason for this discrepancy, in our opinion, is the high number of obese patients (BMI > 30) in Poyiadji et al.’s study, in comparison to the highest BMI of our patients which was 27.8. In agreement with previous literature, the D-dimer values at admission and within 24 h of CTPA, in addition to the mean value of these two measurements, were significantly higher in patients with PAE [2, 16, 29]. The high values of the D-dimer could be related to the activation of blood coagulation in COVID-19 patients following a systemic inflammatory response syndrome [23]. The revised Geneva scores were also significantly higher in COVID-19 patients showing PAE. ROC analysis showed that the mean D-dimer value was the variable with the highest predictivity in our study (AUC 0.831). These findings, in our opinion, are meaningful as they suggest that, though most of the patients have high D-dimer levels (often >1000 μg/l), those with constantly higher levels may also show a higher risk for pulmonary artery obstruction. Our study has some limitations. First of all, it is a retrospective study from a single centre with a lower number of patients if compared with similar reports. Although Italy has been one of the most involved countries in the SARS-CoV-2 pandemia, southern regions have experienced a by far lower number of cases. Furthermore, in our centre, most patients were studied using an unenhanced chest CT while CTPA was performed on patients whose respiratory conditions were worsening. However, the selection criteria for CTPA were not established beforehand, and CTPA was obtained during the hospitalization (13 median days after admission) when the majority of the patients were already receiving prophylactic enoxaparin therapy. This data poses issues concerning the possible underestimation of PAE present at admission and the role of enoxaparin therapy in the management of this complication. Further investigation with prospective studies is required. Finally, in our study, no information is available regarding the ICU patients, as none of these underwent a CTPA. However, this drawback makes our experience different from other similar papers, as it underlines an increased prevalence of PAE unrelated to an admission to ICU, thus supporting the hypothesis of SARS-CoV-2-related endovascular thrombosis.

5. Conclusion

In conclusion, our results are in agreement with previous literature as it shows a higher risk of pulmonary artery obstruction in COVID-19 patients; our data suggest that patients with persistently high values of D-dimer should undergo CTPA, rather than HRCT, due to the higher risk of pulmonary artery obstruction. Nevertheless, further studies are needed to confirm our data and to explain the high prevalence of PAFD during SARS-CoV-2 infection.

Conflicts of Interest

The authors declare no conflict of interest.