Dual-energy CT angiography reveals high prevalence of perfusion defects unrelated to pulmonary embolism in COVID-19 lesions

The severe acute respiratory distress syndrome-associated coronavirus-2 (SARS-Cov-2) was first identified in Wuhan, Hubei, China in December 2019 [1]. Soon afterward, progressive respiratory failure and a systemic coagulopathy were found to be critical aspects of the morbidity and mortality in the coronavirus disease 2019 (COVID-19) pandemic [2‐4]. The elevation of the D-dimer level is now a widely published biological feature of the disease [5], linked to both inflammatory syndrome and hypercoagulable state [4]. Several publications have also suggested a high prevalence of pulmonary embolism (PE) in COVID-19 patients (2.6%—30%) [6‐11]. However, increased D-dimer levels are seen even in the absence of PE.

Interestingly, in the acute respiratory distress syndrome linked to COVID-19, lung compliance seems to be unusually preserved, suggesting that perfusion abnormalities might participate in the occurrence of severe hypoxemia along with the alveolar damage [12]. Corroborating this hypothesis, several autopsy series reported distinctive vascular characteristics associated with COVID-19, including endothelialitis and capillary microthrombi, which were not seen in other conditions such as influenza, along with more anticipated features such as exudative diffuse alveolar damage and surimposed bronchopneumonia [13‐15]. Lung perfusion defects (PDs), which could be a radiological correlate of these histological findings, have previously been described in 24 out of 25 patients with COVID-19 using dual-energy computed tomography pulmonary angiography (DE-CTPA) [16, 17]. In light of these recent biological, histological and radiological data, we hypothesized that PDs could be a distinctive radiological finding of COVID-19 related pneumonia. However, these pulmonary vascular manifestations in COVID-19 have not yet been compared to non-COVID-19 lung infections. The relationship between the occurrence of PE-unrelated perfusion abnormalities, perhaps due to distal microthrombosis, and the elevation of the D-dimer level is also unsolved.

Here, we aimed to assess the prevalence and characteristics of perfusion defects unrelated to PE in COVID-19, by comparing patients with COVID-19 and non-COVID-19 community acquired pneumonia with suspected PE and a CTPA negative for PE. Secondary objectives were to investigate the clinical and biological parameters associated with the occurrence PDs unrelated to PE in COVID-19.

Our retrospective study comparing DE-CTPA scans negative for PE of COVID-19 and non-COVID-19 pneumonias found a higher prevalence of perfusion defects within lung opacities in COVID-19 patients. This finding is consistent with the first published series of DE-CTPA in COVID-19 [17] reporting a very high prevalence of oligemia; however, our study included more COVID-19 patients with DE-CTPA scans (67 versus 25) and a control group. These differences in terms of lung perfusion do not seem solely explained by the different types of parenchymal lesions observed between COVID-19 and non-COVID-19 pneumonia. Indeed, consolidations exhibited perfusion defects in the COVID-19 group more frequently than in the non-COVID-19 group (90% versus 28%), suggesting an underlying pathological process specific to COVID-19.

To avoid confusion with perfusion defects linked to macro-thrombosis and embolism, we excluded PE patients from the study. As might have been expected considering the high prevalence of PE in COVID-19 [6‐9], a greater number of PE patients were excluded from the COVID-19 group than from the non-COVID-19 group. The male predominance, known to be a feature of the disease is also reflected in our study sample (64%) [21]. The PD pattern was more frequently heterogenous in the COVID-19 group than in the non-COVID-19 group. This could perhaps be due to the wider spectrum of lesion types within each patient individually in COVID-19, whereas lesion types were more uniform in community acquired pneumonia. Unlike Lang et al. [17], we did not observe any “hyperemic halo” surrounding oligemic pulmonary opacities (found in 9 out of 25 patients). The use of different scanners between our two studies might have played a role in these differences as this finding was illustrated on pulmonary blood volumes (PBV) images obtained using a scanner from a different vendor.

By acquiring near-simultaneous CT images at two x-ray energy levels [22], DE-CTPA allows the decomposition of images into two base materials (e.g., iodine and water) and thereby separate regions of iodine uptake from unenhanced soft tissues. Hence, iodine maps can be obtained to demonstrate the distribution of pulmonary perfusion [23]. The main clinical application of DE-CTPA is the evaluation of pulmonary embolism (PE) [24]. The spectrum of pulmonary infections has been far less explored. Some authors described increased or decreased heterogenous iodine distribution in pneumonia [19, 25], matching the abnormalities seen on conventional lung attenuation images, as opposed to pulmonary embolism in which the mismatch between perfusion-ventilation images is a key diagnostic feature. However, to our knowledge, the physiopathological significance and clinical implication of these findings have not been further investigated.

We observed a homogeneous hypoperfusion pattern in a majority of PD + non-COVID patients. Little is known about the perfusion profile of community acquired pneumonias on DE-CTPA and it was out of the scope of this study. However; we hypothesize that hypoperfusion, less frequent than in COVID-19, could be associated with suppurative or necrotizing forms (i.e. in pneumococcal or tuberculous pneumonia), but this topic would require further specific research. Another important finding of the study is that PDs were not associated with a worse outcome in terms of oxygen requirement, need for invasive ventilation, length of hospitalization nor death. The D-dimer level and anticoagulant therapy prior DE-CTPA were also similar between PD + and PD- patients. The extent of PDs was not correlated to the D-dimer level nor to the patients’ oxygen requirements. According to these findings, PDs could therefore be seen as a feature suggestive of COVID-19 pneumonia, along with other typical CT findings [26], rather than a marker of severity.

Several recent autopsy studies have described widespread vascular thrombosis with microangiopathy and occlusion of alveolar capillaries in post-mortem COVID-19 lungs [13, 14]. Menter et al. reported alveolar capillary microthrombi 9 times as prevalent in patients with Covid-19 as in patients with influenza. Furthermore, clinically, many patients have elevated D-dimer levels [5], as well as cutaneous changes in their extremities, suggesting thrombotic microangiopathy [27] and generalized thrombotic microvascular injury. These distinctive pathology vascular findings substantiate the perfusion patterns we observed on DE-CTPA and support the global hypothesis of the impact of diffuse vascular abnormalities in COVID-19 pneumonia. However, the exact nature of these PDs (e.g. micro-emboli, in-situ thrombosis, or vasocontriction) remains unelucidated.

This study has several limitations. First, the inclusion of patients who underwent DE-CTPA for suspect PE automatically induces a selection bias. In the COVID-19 group, DE-CTPAs were performed when the clinical state of patients worsened, the study sample therefore included more severe patients and did not reflect the entire spectrum of COVID-19 pneumonia. Also, during the first epidemic wave, the selection criteria for CTPA may have been stricter in order to minimize staff exposure. The level of D-dimer was also susceptible to be higher in these PE-suspected patients. This bias may have flattened the differences between the PD + and PD- COVID-19 patients. In the non-COVID-19 group, the selection of patients referred for suspected PE resulted in the inclusion of a large majority of patients with bacterial rather than viral pneumonia (only one patient), more susceptible to present with chest pain or D-dimer elevation linked to the inflammatory syndrome. Comparing COVID-19 to other viral infections, exhibiting rather diffuse alveolar damage than suppurative bronchopneumonia, would have been of great interest, but such cases are less prevalent, and patients rarely undergo DE-CTPA, so the data were not available to us. Although it is a major limitation of this study, it wouldn’t have been ethical to collect prospectively DE-CTPA in patients with viral pneumonia and no PE suspicion. Secondly, although the readers were blinded to the diagnosis, COVID-19 CT findings are quite typical and the evaluation of perfusion abnormalities subjective, potentially leading to a misclassification bias. We tried to reduce this bias by establishing a consensus between readers in difficult cases. The use of an iodine concentration threshold would have allowed for a quantitative analysis, but we preferred a subjective evaluation using experts in the absence of a standard criterion and vendor recommendation. Another limitation is the lack of temporal perspective on these DE-CTPA findings: are perfusion defects an early or a tardive feature of the COVID-19 pneumonia? We might think that perfusion defects appear at a later stage since more patients with PD were diagnosed on a follow-up rather than on a baseline DE-CTPA, and occurred in consolidations and organizing pneumonia’s suggestive lesions rather than in ground glass opacities seen at an earlier stage [28], but this would need further investigation. Finally, for clarity purposes, we did not assess the occurrence of hyperemia or hyper-perfusion in this study. This could be a matter of interest in both COVID-19 and non-COVID-19 pneumonias.

Beyond offering novel insights relevant to our understanding of the disease, the clinical application of these findings remains to be determined. DE-CTPA could potentially help diagnosing litigious COVID-19-suspected cases. However, in our opinion the use of DE-CTPA has to be motivated by a clinical PE suspicion.

In conclusion, unlike in bacterial pneumonia, heterogeneous perfusion defects within lung opacities could be a specific feature of COVID-19 pneumonia, in line with previously published histological data.