Role of pulmonary circulation assessment in CT imaging in evaluating the severity and tendency of severe and critical COVID-19 pneumonia

Objective: To investigate the value of changes of pulmonary circulation in CT imaging in evaluating the severity and tendency of 2019 novel coronavirus disease (COVID-19) pneumonia. Methods: This retrospective study analyzed 99 severe and critical COVID-19 pneumonia patients including the 47 improved cases and 52 dead cases. Demographic data, laboratory ndings, comorbidities, and CT imaging features including the diameters of pulmonary vein (PV), pulmonary artery (PA) and ascending aorta were collected and assessed. Results: The PV diameters of the deceased patients were larger than recovered patients in the severe phase. Compared with the severe phase in the improvement group, the diameters of the pulmonary veins during the improved phase were smaller, and the total CT scores were signicantly decreased (p < 0.001). Instead, there was no signicant difference in the ratio of main PA to aorta diameter between the recovered group and the deceased group, nor did the self control of the recovered group and the deceased group (p > 0.05). Construction of a ROC curve yielded an optimal cut-off value of the PV diameters for prediction of survival (p < 0.05). Conclusion: The changes of the PV diameters might indirectly reect the activity of pulmonary inammation and cardiac insuciency. Pulmonary manifestations of severe and critical COVID-19 pneumonia might be related to myocardial injury and cardiac insuciency, expecially accompanied by dilated PVs. Evaluation of changes in pulmonary circulation by chest CT images may be considered as a useful tool for determining the severity, fatal outcome and tendency of COVID-19.


Introduction
Key Points: The changes of the pulmonary vein diameters might indirectly re ect the activity of pulmonary in ammation and cardiac insu ciency.
The dilation of pulmonary vein was a predictor of poor clinical outcomes and tendency of COVID-19.
Pulmonary manifestations of severe and critical COVID-19 pneumonia might be related to myocardial injury and cardiac insu ciency.
Since December 2019, the ongoing coronavirus disease 2019 (COVID-19) has spread globally, and evolved into a pandemic [1]. In almost four months, coronavirus disease 2019  has resulted in considerable morbidity and mortality all over the world. As of 7:10pm CEST, 14 May 2020, it has spread CT Protocol All patients underwent non-enhanced chest CT scanning using standard-dose chest CT protocols (GE Healthcare, Philips, or Toshiba Medical Systems) in the supine position during end-inspiration. Highresolution computed tomography (HRCT) could overcome heart beat artifacts, and cases with excessive artifacts were excluded. The main scanning parameters were as follow: tube voltage =80-120 kVp, automatic tube current modulation(60-300 mA), pitch = 0.984:1, matrix = 512×512, slice thickness = 1.25mm, which were all selected according to the machine model.

CT Review
For the recovered group, two CT scans for each patient were collected, including the CT in the severe phase and in the recovered phase. In addition, the rst available CT after symptoms onset and the last CTs before death were collected for the deceased group. The radiologic assessments included vascular measurements, chest CT imaging features and CT score, which were independently assessed and counted by two chest radiologists (HXG and QJH, with 20 and 10 years of work experience, respectively). Final decisions were reached by consensus. Diameters of pulmonary arteries and pulmonary veins were measured by CT images. The main pulmonary artery (PA) and ascending aorta were measured in the axial plane at the level of the pulmonary artery bifurcation. The ratio of the main pulmonary artery to the ascending aorta was calculated. Dividing the main PA by the aortic diameter measured at the same chest level and during the same phase of the cardiac cycle can adjust for some potential differences due to the association between the mean PA diameter and body surface area [8,9]. Using multiple-plane reconstruction, we could delineate clearly the junction of each pulmonary vein (PV) to left atrial (LA). The diameter of the PV was measured at the position 5mm distance to the junction of the LA and each PV in axial images [10,11]. Pulmonary vein pulmonary artery and ascending aorta diameters were measured by two independent observers with no knowledge of the patients' clinical data.
All CT scans were evaluated for the following characteristics: (a) ground-glass opacities, (b) consolidation, (c) number of lobes affected by ground-glass or consolidative opacities, (d) degree of lobe involvement in addition to overall lung "total severity score", (e) nodules, (f) pleural effusion and pleural thickening, (g) pericardial effusion thoracic, (h) lymphadenopathy(de ned as lymph node size of 10 mm in short-axis dimension) and (i) hilar involvement. The degree of lung involvement was visually assessed from 0 to 5 as: none (0%), minimal (1%-25%), mild (26%-50%), moderate (51%-75%), or severe (76%-100%). No involvement corresponded to a lobe score of 0, minimal involvement to a lobe score of 1, mild involvement to a lobe score of 2, moderate involvement to a lobe score of 3, and severe involvement to a lobe score of 4. The total CT scores were reached by summing the ve lobe scores (range, 0 none, 25 maximum) [12]. Ground-glass opaci cation was de ned as increased lung attenuation with blurred bronchi and vascular margins, and consolidation was de ned as blurred vascular margins and airway walls [13]. The criterion of hilar involvement is that the CT image shows enlarged hilar, unclear structure, and blurred blood vessel edges.

Statistical Analysis
All categorical variables were compared for the study outcome by using the Fisher exact test or χ2 test, and continuous variables were compared using the t test or the Mann-Whitney U test, as appropriate.
Wilcoxon signed-rank tests were used to analyze quantitative data for every patient's two CT scans. Kappa consistency test was used for self-comparison of qualitative data. To analyze the PV diameters as a predictor of survival, receiver operator characteristic (ROC) curves and the corresponding area under the curve (AUC) was used to determine the cut-off value of the PVs yielding the highest sensitivity and specificity. All statistical analyses were performed with SPSS, version 21.0 (IBM Corp) for Windows and p < 0.05 was considered statistically signi cant.

CT Review
The measurement results of PVs and PAs are listed in Table 3 and Table 4. The PV diameters of the deceased patients were larger than recovered patients in the severe phase, especially the right inferior pulmonary vein and the left superior pulmonary vein, the differences were signi cant (all p < 0.05). For patients in the recovered group, the PV diameters in the improved phase were signi cantly smaller than those in the severe phase (all p < 0.001). Only 12 deceased patients underwent two effective chest CT scans: the initial CT and the last CT before death. For those cases, compared with the initial CT, PVs were more dilated on the last CT (all p < 0.05). Construction of a ROC curve yielded an optimal cut-off value of the PV diameters for prediction of survival (all p < 0.05) ( Table 5).
Instead, there was no signi cant difference in the ratio of main PA to aorta diameter between the recovered group and the deceased group, nor did the self control of the recovered group and the deceased group (all p > 0.05). 36 cases (69%) of 52 deceased patients and 20 cases (42.55%) of 47 recovered patients shown hilar involvement on chest CT images in the severe phase (p =0.009).
The total CT scores in the recovered group were lower than that in the deceased group in the severe phase (medium 15 vs 19.5, p=0.001). The CT scores of the right lobe and LU (the left upper lobe) in deceased patients were signi cantly higher than those of the recovered patients (all p < 0.05). Besides, in the recovered group, the CT scores of each lung lobe in the improved phase were signi cantly smaller compared with those in the severe phase (all p < 0.001) and the difference in total CT scores was also signi cant (medium 10 vs 15, p < 0.001). Compared with the initial CT, the CT scores of the last CT exams in deceased group were signi cantly increased (medium 21.5 vs 8.5, p = 0.001).
Other major CT ndings including pleural effusion, pericardial effusion, and pleural thickening were also evaluated. Compared with deceased patients, recovered patients were more likely to have pericardial effusion (20[43%] vs 7[13%], p = 0.001). Whereas the occurrence rates of pleural thickening in deceased patients were signi cantly higher than those of the recovered patients (40[77%] vs 14[30%]). There was a signi cant difference in pericardial effusion and pleural thickening between the severe phase and improved phase for the recovered group (both p < 0.001).

Discussion
This retrospective study aimed to determine the value of pulmonary circulation in the assessment of the severity and tendency of severe and critical patients with COVID-19 pneumonia through chest CT images. Our study comprehensively described the major differences in pulmonary circulation on chest CT images between deceased patients and recovered patients with COVID-19. The severe phases and the improved phases of COVID-19 patients in the recovered group were also evaluated and compared. We found that evaluation in the diameters of pulmonary veins on chest CT images might be a useful tool for determining the severity and tendency of COVID-19. Furthermore, the dilation of PV was a predictor of poor clinical outcomes.
COVID-19 has caused far more infections and deaths than severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) due to its high infectivity, which have mortality rates of 9.6% [14] and 34.4% [15], respectively. Because no specialized drugs have been found to treat severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, the main treatment is timely symptomatic and supportive treatment [ 16 ]. Among patients with COVID-19, especially those who are severely and critically ill, SARS-CoV-2 primarily affects the lungs, it also affects multiple organs, such as the cardiovascular system [17]. However, to our knowledge, no study has been performed to evaluate pulmonary circulation on chest CT in COVID-19 pneumonia patients.
In our study, the diameters of pulmonary vein of the deceased patients were larger than recovered patients in the severe phase (p < 0.001), and compared with the initial CT, PVs were more dilated on the last CT before death in the deceased patients (all p < 0.05), which may indicate that the dilation of pulmonary vein in severe and critical patients with COVID-19 are at higher risk of death. Furthermore, construction of a ROC curve yielded an optimal cut-off value of the PV diameters for prediction of survival. We proposed that the dilation of PV was a predictor of poor clinical outcomes. The dilation of PV is seen in pulmonary congestion caused by increased pulmonary venous return and left heart insu ciency [18,19]. Pulmonary in ammation leads to congestion of the lung tissue, which has been reported in the early studies. Severe and critically patients have severe pulmonary infections, manifested by abnormal laboratory and radiographic ndings, such as higher levels of WBC and h-CRP; more groundglass opacities, consolidation and higher total CT stores. Therefore, obvious congestion might increase PV regurgitation, leading to the enlargement of PV. In recovered patients, the reduction of PV diameter accompanied by lower total CT score in the improved phase compared with in the severe phase, indicates the reduced of pulmonary circulation re ux and the absorption and improvement of pulmonary in ammation. In addition, the main target organ of the COVID-19 infection is the lung, and some recent studies have reported it could also cause myocardial damage [17,20]. Mechanistically, SARS-CoV-2, binds to the transmembrane angiotensin-converting enzyme 2 (ACE2), and then enters type 2 pneumocytes, macrophages, pericytes and cardiomyocytes [17]. High expression of ACE2 in pericytes could lead to development of microvascular dysfunction. ACE2 expression is up-regulated in failing human hearts, suggesting a plausible explanation for a higher infectivity of virus and a higher mortality in patients with heart failure. The signi cant increase of myocardial enzyme spectrum indicates that the myocardium is involved [5]. In our study, compared with the recovered patients, the deceased patients showed the signi cant increase of high-sensitivity troponin, lactic dehydrogenase and NT-proBNP, which was consistent with relevant reports and suggested myocardial injury and cardiac dysfunction [21]. Mechanisms underlying myocardial injury remain unclear. It is worthy of attention whether they re ect systemic or local process and ischaemic or in ammatory process. Also, since ACE2 (a homologue of ACE) is expressed in cardiomyocytes, direct cardiomyocyte infection by SARS-CoV-2 may be a possibility [17,22 ]. Pulmonary congestion indirectly indicated left heart insu ciency, which is usually manifested as blurred hilar shadows and dilated PVs on CT images. CT examination clearly showed the hilar structure, so it has a unique advantage in determining whether the lesion involves the hilum. In our study, 36 cases (69%) of 52 deceased patients and 20 cases (42.55%) of 47 recovered patients shown hilar involvement on chest CT images(p=0.009). Some attempts to treat COVID-19 cardiac injury have done, the detection of pulmonary circulation, expecially pulmonary vein, might help identify a subset of patients at greater risk of COVID-19 complications and successful therapies. Therefore, the changes of PVs might indirectly re ect the activity of pulmonary in ammation and cardiac insu ciency. This new radiological evidence provided an innovative new method compared with previous investigations in the assessment of the severity of severe and critical patients with COVID-19.
The representative chest CT ndings of patients with COVID-19 were peripheral and/or subpleural groundglass opacities or consolidation for common patients [1]. Furthermore, in severe and critical COVID-19 pneumonia patients, more and more ground-glass attenuation and airspace consolidation as well as involvement of multiple lung lobes and high CT scores were common chest CT nding [23], which was similar in our study. Traditionally, more and more larger lesions on CT images indicates that the lung in ammation was more serious. However, we posed a hypothetical question which pulmonary manifestations of severe and critical COVID-19 pneumonia might be related to myocardial injury and cardiac insu ciency, expecially accompanied by dilated PVs.
This study has the following limitations: Firstly, it was a retrospective study. Secondly, there was a small sample size of COVID-19, thus some conclusions are preliminary. More powerful studies with pooled data from multiple centers is needed in our subsequent studies. Thirdly, no lung biopsy or autopsy was performed to re ect the histopathological changes.
In conclusion, the retrospective study suggests that chest CT should be considered as a quick and effective method to assess the changes in pulmonary circulation and the progression of COVID-19 pneumonia. Evaluation of changes in pulmonary circulation by chest CT images might be considered as a useful tool for determining the severity and tendency of COVID-19. The changes of PVs might indirectly re ect the activity of pulmonary in ammation and cardiac insu ciency. Pulmonary manifestations of severe and critical COVID-19 pneumonia might be related to myocardial injury and cardiac insu ciency, expecially accompanied by dilated PVs. In this way, the evolution of the patient's condition could be detected in time, and targeted treatment can be adopted, so as to improve the prognosis of patients and reduce the risk of death of severe patients, especially critical patients.

Declarations
Informed consent SARS-Cov-2 is a continuing public health outbreak investigation globally. Therefore, written informed consent was waived by the Institutional Review Board.       left inferior pulmonary vein (LPV) were 14.23mm and 11.18mm, respectively, which were more dilated compared with the initial CT. Four days later, the patient died. pulmonary vein (LPV) were 9.11mm and 7.78 mm, respectively, and the diameters of pulmonary vein was signi cantly smaller than those in the severe phase. The patient was discharged two day later. Figure 3. HRCT ndings in a 40-year-old man with severe COVID-19 pneumonia who survived and recovered. (a) and (b) In the severe phase (day 9 after the onset of initial symptoms), the images exhibited showed diffuse mixed ground-glass opacity and consolidation lesions. Meanwhile, the diameter of the right inferior pulmonary vein (RPV) and left inferior pulmonary vein (LPV) were 14.24mm and 10.33 mm, respectively . (c) and (d) Day 42 after the onset, the lesions were absorbed. Meanwhile, the diameter of the right inferior pulmonary vein (RPV) and left inferior pulmonary vein (LPV) were 9.11mm and 7.78 mm, respectively, and the diameters of pulmonary vein was signi cantly smaller than those in the severe phase. The patient was discharged two day later.