CT as a Tool to Depict Pulmonary Fibrosis in Patients With COVID-19: a Radiopathological Correlation

CT ndings of COVID-19 infected patients has been well described, but it it’s roll in depicting signs of brosis in critically ill patients remains unclear. To our knowledge, there are no radiopathological correlations of the pulmonary pathology. Exudative and proliferative diffuse alveolar damage (DAD) are the most commonly reported injury. Few studies describe brosis, the last phase of DAD. Our study correlates post-mortem chest US and CT ndings of COVID-19 infected patients with the histopathology from biopsies taken of the lung. It focuses on the role of CT to depict brosis. METHODS This is a prospective observational study of deceased patients infected with COVID-19. Post-mortem chest CTs and were were biopsies of different radiological Pre-mortem CT were retrospectively


Introduction
Despite an increasing number of studies describing the radiological ndings of COVID-19, there is a lack of post-mortem studies crucial in understanding the pathological changes that occur during the disease.
Autopsies of COVID-19 infected patients have not been performed systematically worldwide 1 .
Most patients infected with COVID-19 present with respiratory symptoms. Up to 20% of the patients develop severe disease. Some of these will develop an acute respiratory distress syndrome (ARDS) with high mortality rates 2 . CT ndings have been described in recent publications [3][4][5][6][7] . Ground-glass opacities (GGO) with or without consolidation with a peripheral, posterior and lower distribution are commonly seen 3 . These ndings are speci c in grading the severity of the disease. GGO appear between 0 and 4 days after symptom onset. Areas of GGO become coalescent, and other patterns, including consolidations appear and peak at 6-13 days 8 . Signs of brosis has been described at the end-stage of the disease or as a sequalae 9,5 .
The use of US of the lung of COVID-19 infected patients has been reported, but its role in the assessment of the disease needs to be fully established. US features include pleural thickening, consolidation, and Bpattern, in a variety of patterns including focal, multifocal, and con uent. 10 .
Small series of autopsies report diffuse alveolar damage (DAD) as the fundamental injury in the lung [11][12][13][14][15][16][17] . Lung brosis is an uncommon feature in these studies, but it has been described as a progression of the disease that can worsen the prognosis 9 . Lung damage in DAD occurs in three phases. The exudative phase, in which hyaline membranes predominate lasts 7 days. The next phase is the proliferative phase, characterized by a thickening of the alveolar septa with proliferation of myo broblasts and a duration of two weeks. Week three is characterized by the brotic phase. Fibroblast activation and collagen formation lead to thickening of the alveolar septa 20,21 To our knowledge, nothing has been reported on the radiological appearance of post-mortem US or CT.
Moreover, there are no studies correlating these radiological ndings with tissue samples of the infected lungs. 2 This study reports the CT and US ndings of deceased patients infected with COVID-19. It correlates these ndings with pathology samples obtained with US and CT-guided biopsies of different areas of the lung. We hypothesize that CT correlates with pathology ndings of pulmonary brosis, and therefore could be a tool to depict brosis in critically ill patients.

Methods
This is a prospective observational study of six consecutive deceased patients infected with COVID-19 (range 64-100 years, mean age 73.1), who had died as a result of an ARDS. All cases tested positive for COVID-19 by nasopharyngeal swab at time of admission. Patients were recruited from April to June of 2020. The study was approved by the Ethical Committee of the hospital. Oral informed consent was obtained from the relatives in all cases.

Radiological examinations
Post-mortem examinations and biopsies were performed within 24 hours of the death of the patient.
POST-MORTEM US: US examinations (Esaote TM Mylab70 Vision, Milano It) were performed in the CT room by a single senior radiologist. Patients were examined only in the supine position, so mostly anterior segments of the lung were imaged. We evaluated the following signs and patterns in both lungs: Normal ventilated lung, B-lines, consolidations, pleural effusion, and nodular pleural thickening.
POST-MORTEM CT: CTs (OptimaCT660 GE Healthcare, Inc. Boston MA USA) were also performed before the biopsies. We evaluated the presence of: GGO, consolidation, pleural effusion, enlarged lymph nodes, halo and inverted halo sign, crazy paving pattern), consolidation with an organizing cryptogenetic pneumonia pattern, septal thickening, bronchial wall thickening, and pericardial effusion. We de ned honey-comb pattern, microcystic pattern, and traction bronchiectasis as signs of brosis. All axial and reconstructed CT images were reviewed by two experienced radiologists independently.
PRE-MORTEM CT: The existing premortem CT examinations performed during admission to rule out pulmonary embolism were compared to the post-mortem examinations.

Post-mortem biopsies
Both CT and US were used to identify areas to be biopsied. 16G needles (Bard ® Magnum Reusable Core System, Arizona USA) were used for all CT guided biopsies. Overall, 15 tissue samples were taken. These included areas of consolidation, of GGO and of parenchymal distortion (honeycomb pattern, microcystic pattern, traction bronchiectasis). We also performed US guided biopsies from aerated lung with B-lines and areas of consolidation. Tissues were xed in 10% buffered formalin for at least 24 h. They were embedded in para n to obtain sections of 3-μm in thickness stained with hematoxylin and eosin (H&E). Immunohistochemical staining was done in selected cases for detection of TTF1 (8G7G2/1), Muscle Speci cActin (HHF35) and CD-68 (KP-1). All antibodies were ready-to-use monoclonal antibodies (Ventana, Roche Diagnostics, Basel, Switzerland). Samples were stained with the BenchMark Ultra IHC/ISH System (Roche, Basel, Switzerland) in accordance with the standard protocols supplied by the manufacturer. Masson trichome staining was applied to characterize collagen deposition and brosis. All samples were reviewed by the senior pathologist of the hospital.

Results
The patients were admitted to the hospital between 3 and 10 days after the onset of symptoms. The hospital stays ranged from 9 to 52 days (mean 33,8 days). Five of the patients required mechanical ventilation requiring in the intensive care unit (ICU). Stays ranged from 23 to 45 days (mean of 35.2 days). One patient did not meat criteria for ventilation and admission to the intensive care unit based on age and comorbidities. On average patients died 40.3 days after onset of symptoms, with a range of 19-59 days. All patients died as a result of an ARDS. Table 1  Days from admission to death 9-52 Patients in intensive care unit (mechanical ventilation) 5 83% Table 1 shows the characteristics of the patients included in the study.

Radiological ndings
Post-mortem US: Areas of nodular pleural thickening and extensive areas of subpleural consolidation were the most common nding. B-pattern was present in the areas that remained aerated, mostly in the antero-superior segments of the lobes. Two patients showed small areas of normally aerated lung with no pathologic ndings. Pleural effusion was depicted in only one patient.
Post-mortem CT: At least 80% of the lung was affected by pathologic changes, including: extensive and bilateral areas of consolidation, GGO with or without crazy paving and bilateral pleural effusion. In all cases, only the anterior and mostly apical segment of the lung remained partially aerated. All patients showed partial collapse of the inferior lower lobes. In 4 patients additional partial collapse of the middle and upper lobes was observed. Signs of brosis were seen in 4 cases including: Traction bronchiectasis, honey-comb pattern, and a microcystic pattern, initially with a subpleural distribution (seen in examinations during admission) that became multifocal or patchy in the postmortem CT ( gure 1). Enlarged mediastinal lymph nodes were depicted in 2 patients. EXTENT of the lesion >80% 6 100% Table 2 summarizes the US and CT findings present on post-mortem US and CT.
Comparing the post-mortem study and the CT during admission, we observed a progression of the areas of consolidation, collapse, and pleural effusion, as shown in gure 2.

Histopathological ndings
We studied 15 lung biopsies from different lung regions of 6 patients. All patients showed proliferative DAD in at least one of the lung samples, in combination with brotic DAD in four patients, exudative DAD in one patient, and AFOP in three patients. Only one patient presented a proliferative DAD as the only form of DAD. The rest presented combinations of different patterns of DAD or with AFOP. Only two patients showed microthrombosis. Only one of the patients had a superimposed bacterial pneumonia.

Discussion
The severity of the disease caused by COVID-19 correlates well with the radiological presentations 3 . CT ndings have been meticulously described 4 . Initial radiological ndings appear to be very speci c for COVID-19 in the context of the current pandemic. Very little has been published on the end stage of the disease, as critically ill patients usually do not undergo US or CT examinations unless they show signs of complications (pulmonary embolism or a superimposed bacterial pneumonia). In post-mortem examinations the damage is so extensive that some of the speci c ndings reported in the literature, such as the halo sign, or the distribution of the opacities are not identi able. Pleural effusions were present in all 6 patients. A pleural effusion is an uncommon or nonspeci c feature at early stages of COVID, but it has been identi ed as a sign of bad prognosis if shown 10 . Our ndings are consistent with both alveolar and interstitial damage on the subsequent histological studies. Signs of pulmonary brosis are rare in patients that fully recover not requiring long hospital stays and/or mechanical ventilation 6 . In our study, there is a perfect correlation between the signs of brosis on CT and the subsequent histopathologic study, as discussed below.
The role of US in the management of COVID-19 infected patients has not been fully established. Our study shows a correlation of both US and CT: Areas of pleural thickening were shown as thickened pleura on CT, a B-pattern on ultrasound presented as GGO or crazy paving on CT, and subpleural consolidations as areas of consolidation in contact with the pleura. In our study, US was not as sensitive as CT to detect pleural effusions because examinations only included the anterior segments of the lung.
Our study con rms DAD as the predominant pattern in the lungs of patients infected with COVID-19 virus, most commonly in the proliferative phase. Other publications report exudative DAD as the predominant injury, sometimes associated with the proliferative phase. The presence of the brotic phase has been reported by very few groups 14 . In contrast, we demonstrated brotic DAD in most patients, always in association of proliferative DAD. Exudative DAD alone was only seen in a severely ill patient that died after nine days of admission. We believe that the differences found with other studies are related to longer courses of the disease of our cohort.. Our patients presented with symptoms for an average of 40,3 days before deceasing, whilst other groups reported symptoms from 16 days until death 17 . What is more, our patients were in the intensive care unit for an average of 35,2 days. Other groups report stays in the ICU for 7 to 30 days with an average of 12,5 days. 17,16,15,23,14 Only Schaller et al. reports a case with areas of brotic DAD in a patient with a 26-day disease-to-death course and mechanical ventilation for 21 days 14 . AFOP alone, with no DAD, was the only lesion described by one author 24 . In our study, AFOP was present in two of our cases, but always associated with DAD.
Two patients in our study had been diagnosed with pulmonary embolism during admission, one which had an aortic mural thrombus. Vascular involvement cannot be evaluated on post-mortem CT's. Therefore, radiopathological correlation was not possible. On histology, two patients showed microvascular thrombosis, a common nding in DAD, but not a necessarily speci c feature in patients infected with COVID-19.
Our study ndings are limited by different factors. First, the number of patients included in our study is very small. Second, whilst the cause of death of these patients was a result of ARDS, other conditions could have contributed to their death, such a bacterial infection, drug toxicity, or barotrauma. Finally, we have no prior history or examinations of the patients. The fact that brosis was present in almost all the mechanically ventilated for over 30 days may indicate that it is factor that could worsen the prognosis. However, brosis is not necessarily present in patients with a bad prognosis if they die at an early stage of the alveolar damage.
In conclusion, signs of brosis on CT correlated with the histopathological ndings. Fibrosis in the context of COVID-19 infection may be an indicator of poor prognosis. CT may be a tool to identify the group of patients that develop brosis in the late stage of the disease. Further bigger studies are required to examine our observations. Figure 1 CT and US ndings of patient 8 days before and 4 hours after death. a. Axial contrast-enhanced CT in the arterial phase during admission shows postero-basal areas of consolidation, microcystic lesions with a subpleural distribution, areas of GGO, and pleural effusion. b. Post-mortem non-enhanced axial CT shows areas of consolidation affecting most of the lung, with very few remaining aerated areas depictingcrazy paving pattern. Traction bronchiectasis and microcystic lesions had also progressed. A large pleural effusion was present. c and d. Post-mortem US shows a "white lung" as a sign of diffuse consolidation (c) and areas of subpleural nodular thickening (d).  CT examinations of a patient 6 days before (a and b) and 24 hours after death (c and d). a. Axial contrast-enhanced CT in arterial phase before death to rule out pulmonary embolism shows peripheral GGO, crazy paving opacities and posterobasal areas of condensation. Subpleural microcystic lesions as an initial sign of brosis were also present. b. Coronal multiplanar reconstruction of the CT in arterial phase shows embolism of the lower right lobe artery (arrow) and a mural thrombus in the thoracic aorta (arrowhead). c. Post-mortem axial non-enhanced CT shows progression of the areas of consolidation, traction bronchiectasis and of the microcystic pattern that becomes diffuse. A small pleural effusion is also present. d. Axial post-mortem on-enhanced CT showing enlarged mediastinal lymph nodes (arrows).