Claudia Ravaglia, Claudio Doglioni, Marco Chilosi, Sara Piciucchi, Alessandra Dubini, Giulio Rossi, Federica Pedica, Silvia Puglisi, Luca Donati, Sara Tomassetti, Venerino Poletti
European Respiratory Journal 2022; DOI: 10.1183/13993003.02411-2021
Several patients experience pulmonary sequelae after Sars-Cov-2 infection, ranging from self-limited abnormalities to major lung diseases. Morphological analysis of lung tissue may help in understanding pathogenic mechanisms and provide consistent personalised management. Aim of the study was to ascertain morphologic and immuno-molecular features of lung tissue. Transbronchial lung cryobiopsy was carried out in patients with persistent symptoms and computed tomography suggestive of residual lung disease after recovery from Sars-CoV-2 infection. 164 patients were referred for suspected pulmonary sequelae after COVID-19; 10 patients with parenchymal lung disease extent >5% underwent lung biopsy. Histological pattern was not homogeneous, as three different case clusters could be evidenced, which were mirrored in clinical and radiological features: cluster one (“chronic fibrosing”) characterised by post-infection progression of pre-existing interstitial pneumonias; cluster two (“acute/subacute injury”) characterised by different types and grades of lung injury, ranging from organising pneumonia and fibrosing NSIP to diffuse alveolar damage; cluster three (“vascular changes”) characterised by diffuse vascular increase, dilatation and distortion (capillaries and venules) within otherwise normal parenchyma. Clusters two and three were characterised by immunophenotypical changes similar to those observed in early/mild covid-19 pneumonias (abnormal expression of STAT3 in hyperplastic pneumocytes and PD-L1, IDO and STAT3 in endothelial cells). This is the first study correlating histological/immunohistochemical patterns with clinical and radiological pictures of patients with post-COVID lung disease. Different phenotypes with potential different underlying pathogenic mechanisms have been identified.
As COVID-19 pandemic has progressed, some patients are experiencing prolonged multi-organ symptoms and complications (long COVID or post-COVID syndrome). The implications and consequences of such ongoing clinical manifestations are a growing health concern. Evidence of residual organ damage following COVID-19 infection applies especially to pulmonary sequelae, with a spectrum ranging from self-limited abnormalities to a clinical profile of major lung diseases, with variable extent of significant inflammatory and/or fibrotic abnormalities. The histological evaluation of lung tissue in this late phase may help reveal peculiar morpho-phenotypical changes which, together with future studies examining patterns evolution, may help in better understanding pathogenic mechanisms and provide consistent personalised therapy. Here, we examine the morphologic and immuno-molecular features of transbronchial lung cryobiopsies (TBLC) performed in patients with persistent lung disease after recovery from Sars-CoV-2 infection and the potential explanation of the observed lung abnormalities.
We conducted a comparative, prospective, multicentric, investigator-initiated and observational study, from January 1st 2021 through March 31st 2021, with the aim to evaluate histological profile and immunohistochemical/molecular features of lung tissues in patients with persistent lung involvement on HRCT (high resolution computed tomography) scan and persistent symptoms (respiratory and/or systemic) after recovery from Sars-CoV-2 infection. Eligible subjects were aged between 18 and 80 years, had a history of previous molecular confirmed diagnosis of COVID-19 related pneumonia, with recovery at least 30 days before, persistent lung involvement on HRCT scan analysis >5%, persistent symptoms (respiratory and/or systemic) and no contraindications for lung biopsy. Recovery was defined according to national directives updated in December 2020 by the presence of two consecutive negative tests for Sars-CoV-2 by rRT-PCR in 24 h associated with possible improvement/stabilization of symptoms. Key exclusion criteria were severely compromised lung function or resting hypoxaemia, high bleeding risk, pulmonary hypertension assessed by echocardiography and severe comorbidities (details are reported in the online data supplement); finally, patients who could not provide consent or refused biopsy were excluded. This study was in accordance with regulations issued by the Helsinki Declaration; the protocol was approved by the Institutional Review Board (IRB) of the Area Vasta Romagna Ethical Committee (Prot. 5285/2021) and patients have provided their informed consent. High Resolution Computed Tomography and Transbronchial Lung Cryobiopsy were performed as described previously and reported in supplemented data. Morphologic examination was based on conventional hematoxylin-eosin stains and immunostaining with cytokeratin 7. Five dedicated pulmonary pathologists (A.D., G.R., V.P., M.C., and C.D.) blindly and independently reviewed the slides from all biopsies and quoted several morphological features involving different compartments constituting the lung parenchyma: epithelial and vascular components, alveolar and interstitial spaces, inflammatory cells. Divergent opinions were then overtly discussed and a final consensus was reached in all cases. All immunohistochemical tests were performed in the ULTRA Benchmark automated immunostainer (Ventana Medical Systems/Roche, Tucson, AZ) using standard procedures and reagents are described in Table S1. We have analyzed 15 control cases with diffuse parenchymal lung diseases as uninfected control specimens: three Cryptogenic Organising Pneumonia (OP), three Non-Specific Interstitial Pneumonia (NSIP), two OP/NSIP, two Usual Interstitial Pneumonia (UIP), two fibrotic Hypersensitivity Pneumonitis (fHP), one idiopathic Acute Fibrinous and Organising pneumonia (AFOP), one Sarcoidosis, one Smoking-Related Interstitial Fibrosis (SR-IF).
164 patients were referred to our post-acute outpatient service for suspected pulmonary sequelae after COVID-19. In 141 patients, parenchymal lung disease was absent or <5%, 13 patients had contraindications for lung biopsy or did not provide consent to the procedure; 10 patients underwent transbronchial lung cryobiopsy. Median age was 61 years (range, 38–76 years) and 78% were men. 7/10 patients had been hospitalised but no one had been intubated. Clinical characteristics are summarised in Table S2. Patients underwent bronchoscopy on average 3.5 months after recovery from Sars-CoV-2 infection. At the time of biopsy all cases resulted negative at Sars-Cov-2 molecular swab testing. Most patients reported fatigue and low grade fever, a minority reported joint and muscle pain and depression (Table S2). Median forced vital capacity (FVC) was 77% of predicted value (range 46–116); median carbon-monoxide diffusion coefficient (DLCO) was 51% of predicted value (range 43–80). The most representative laboratory tests are shown in Table S3.
Radiologic findings are summarised in table 1. In a sub-group (cluster 1), CT scan showed fibrotic appearances with some aspects of interstitial involvement, architectural distortion and traction bronchiectasis; in one of these patients, a previous CT scan had shown a pre-existing background of fibrotic interstitial lung abnormalities (ILA), which were stable and persistent after COVID-19 pneumonia (fig. 1), while in the second patient, acute consolidations masked and covered the possible presence of underlying pre-existing traction bronchiectasis.
Radiological and pathological findings in patients with persistent lung disease
CT scan in a 60 y/o male, smoker. CT scan during acute infection (a–c) a mild peripheral ground glass is present in both upper lobes, mainly in the left hemithorax (red circle). Scattered areas of cystic changes suggestive of airspace enlargement with associated fibrosis (AEF) with mild architectural distortion is present bilaterally. CT scan performed 4 months later (d–f) shows reduction of ground glass attenuation, with residual ground glass attenuation and areas of AEF (f, green arrow).
In five cases (cluster 2), CT scan showed persistence of ground glass with perilobular pattern and consolidations as observed in organising pneumonia (OP) or fibrotic NSIP (figs. 2 and 3); a common feature was residual vessel enlargement (gravity dependent perilobular pattern as assessed after prone positioning) as supposed venular dilatation in the peri-lobule. Finally, third sub-group (cluster 3) was characterised by only mild residual radiological changes, with mild peripheral ground glass associated with perilobular pattern in 2/3. In a single case, features suggesting pleuro-parenchymal fibroelastosis in the upper lobes were observed.
Male, 62 y/o, with acute Covid infection characterised by bilateral, peripheral consolidations (a, b, blue arrow) and perilobular pattern (c, red arrow). Two months later (d–f) CT scan shows mild peripheral reticulation and minimal perilobular pattern, mainly in the right lower lobe (f, yellow circle).
Male 50 y/o, with Covid-related acute pneumonia (a–c) with extensive, peribronchovascular consolidations, in both lungs (a–c) and lobular sparing in both lower lobes. Moderate ground glass attenuation is present in both upper lobes associated with vessel enlargement (c, blue circle). Two months later, a mild, diffuse ground glass attenuation is present in both lungs associated with central bronchiectasis in middle lobe and in both lower lobes.
Histological evaluation on H&E stained slides revealed significant modifications of the pulmonary structure in all ten investigated cases, matching the three-cluster separation. All cases were tested and resulted negative for Sars-Cov-2 [in situ hybridisation in tissue (smFISH method, HuluFISH, Pixelbiotech Germany) and real-time reverse transcriptase polymerase chain reaction (RT-PCR) assay in BAL (bronchoalveolar lavage)].
This cluster included two cases with features of either UIP (usual interstitial pneumonia), as defined by architectural distortion, spatial and temporal heterogeneity of scarring modifications, fibroblastic foci and microscopic honeycombing (fig. 4), or fibrosing smoking-related interstitial pneumonia, as defined by the accumulation of macrophages containing smoking-related light brown pigment in alveolar spaces and interstitial collagenous fibrosis (Table S4). The immunohistochemical analysis in the first case was consistent with UIP, including TUBB3 in myofibroblasts and focal p16 expression in pneumocytes. The second case (showing pre-existing background of smoking related ILD at HRCT which persisted after COVID-19 pneumonia), showed histological features consistent with a smoking-related ILD, with also lymphoid nodules.
Cluster 1. A case with UIP pattern: architectural distortion, spatial and temporal heterogeneity of scarring modifications, and microscopic honeycombing. H&E (a); Cluster 2. A case with morphological evidence of ongoing interstitial fibrosis and extended AECII hyperplasia as observed in diffuse alveolar damage (DAD), proliferative phase. H&E (b). Cytokeratin-7 immunostaining of epithelial cells, at low magnification, showing the severe effacement of parenchymal structure. (c); Cluster 2. A case with organising pneumonia pattern. H&E (d). Most epithelial cells and endothelial cells express nuclear pSTAT3 (e).
Cluster two was characterised by different types and grades of lung injury, ranging from organising pneumonia or overlapping OP/NSIP to diffuse alveolar damage (DAD – proliferative phase, 2 cases) (fig. 4) . These cases shared morphological and immunohistochemical similarities with cases of severe COVID-19 pneumonia previously described . In the two cases consistent with DAD (proliferative/organising without hjaline membranes), the interstitial spaces were diffusely thickened by dense fibrosis and aggregates of myofibroblasts (fig. 4). Residual alveolar spaces were lined by patchy collections of hyperplastic AECII, expressing pSTAT3 and Ki67. Four cases were characterised by the occurrence of endoalveolar Masson's bodies and various amounts of interstitial thickening (OP/NSIP) (fig. 4) (Table S4). Inflammatory cells: abundant immune infiltrates, composed mainly of lymphocytes were present in all cluster 2 cases, either diffusely distributed in interstitial spaces, or forming perivascular aggregates; lymphoid infiltrates were mainly composed of T and B cells; in T cells, CD4 cells and CD8 subsets were almost equally represented. Macrophages were less represented, neutrophils were rare and no eosinophils were observed. Alveolar cells: hyperplastic AECII were characterised by a high Ki67 index, and nuclear pSTAT3. TUBB3 staining was also expressed in alveolar cells, mainly in the cases with the most severe damage. Vascular bed: diffuse IDO, PD-L1 and pSTAT3 immunostaining was detected in the endothelial cells of both interstitial capillary vessels and venules, this representing the most relevant finding in this subgroup .
Cluster 3 was characterised by diffuse vascular increase, dilatation and distortion (both capillaries and venules) within an otherwise normal parenchyma (minimal patchy AECII hyperplasia in one case) (Table S4) (fig. 5). This pattern was comparable to that described in early-phase COVID-19 pneumonia with conserved parenchymal structure. Inflammatory cells: a small number of macrophages and interstitial T lymphocytes were present, without obvious predominance of either CD4 or CD8. PD-L1 was expressed in alveolar macrophages (as may commonly be observed in alveolar macrophages in normal conditions), but no PD-L1 immunoreactivity was observed in the epithelial alveolar cells. Vascular bed: diffuse PD-L1 staining decorated the capillary lining along with IDO reactivity in endothelial cells (fig. 5). Stromal/interstitial compartment: no TUBB3 reactivity was observed. These cases shared similarities with some cases of early/mild COVID-19 pneumonia previously described, in particular the persistence of diffuse and strong expression of PD-L1 and indoleamine 2,3-dioxygenase (IDO) in endothelial cells of both interstitial capillaries and venules in post-acute phase cases. This peculiar expression profile may be appears as strongly more related with SARS-CoV-2 pneumonias, since these markers are not typically observed absent in lung tissue obtained in healthy subjects and/or in control cases with diffuse parenchymal lung diseases as uninfected control specimens. These morphologic abnormalities were mirrored in BAL fluid (Table S3); in fact, all patients with lymphocytosis in BAL (>20%) were characterised by organising pneumonia in lung tissue, with abundant immune infiltrates, composed mainly of lymphocytes, located in perivascular and interstitial spaces, sometimes forming nodular aggregates.
Cluster 3. A case with diffuse vascular increase, dilatation and distortion (both capillaries and venules) within an otherwise normal parenchyma. H&E (a); Cytokeratin-7 immunostaining (b); diffuse and strong endothelial expression of pSTAT3 (c), IDO (d) and PD-L1 (e).
To our knowledge, this is the first prospective analysis of morpho-phenotypical changes in patients with persistent symptoms and residual parenchymal lung disease after recovery from COVID-19. Most published studies have been small and mainly focused on clinical sequelae in patients admitted to hospital for COVID-19. Little is known of the extent and features of lung pathology in COVID-19 survivors, mostly those who had milder infections not requiring hospitalisation. Three cases with post-COVID histologically confirmed OP have been published so far (two trans-bronchial lung biopsy and one surgical lung biopsy) and eleven asymptomatic patients who had previously contracted COVID-19 underwent elective lung resections for other unrelated indications (recovered asymptomatic survivors were confirmed to have no discernible histopathological changes suggesting parenchymal damage). As is the case with other interstitial lung diseases (ILDs), lung biopsy may have a significant role in post-COVID-ILDs, in order to better define histological patterns for the purposes of management. The ILDs community has long been aware that fibrotic nonspecific interstitial pneumonia and organising pneumonia progressing to fibrosis are not infrequently seen in survivors of DAD/acute lung injury. Current algorithms mostly view post-COVID lung disease as a single entity; however, the existence of different COVID-induced pathways may underlie separate post-COVID lung entities, which may require separate management. In this multidisciplinary study, we have identified three different subtypes or sub-groups (table 2).
Identification of possible sub-groups (clusters) of patients with post-acute Covid-19
One group (cluster 1) was characterised by features suggesting the occurrence of interstitial lung disease before SARS-CoV2 infection. Both cases included were male and former smokers. Clinical behavior was characterised by persistent respiratory symptoms after COVID-19 pneumonia, shortness of breath on exertion and cough, with less frequent systemic symptoms (Table S2). One case, whose pattern was consistent with UIP, was characterised by fibrotic appearances on CT scan and lung tissue, architectural distortion, traction bronchiectasis, spatial and temporal heterogeneity of scarring modifications, and extended honeycombing (fig. 4 and table 1). It is possible to argue that the acute consolidations in this case masked and covered the presence of underlying pre-existing fibrosing ILD. The second patient had a documented pre-existing background of fibrotic interstitial lung abnormalities (ILA), which were stable and persistent after COVID-19 pneumonia. Previous pulmonary function tests were normal (FVC 103% predicted, FEV1 104% predicted, DLCO 76% predicted). Lung cryobiopsy showed a pattern consistent with smoking-related interstitial fibrosis, SR-IF). We hypothesise that a number of post-covid-19 cases, characterised by radiological and/or pathological finding of fibrotic appearances might be actually more an expression of a pre-existing pulmonary interstitial disease (which may or may not be accelerated by the virus infection itself) than strictly related to the viral infection. Acceleration of pulmonary fibrosis by respiratory virus has been described in the bleomycin model of pulmonary fibrosis; interestingly, in these cases, foci of cells expressing the cell-senescence associated marker p16 were observed as in UIP/IPF.
The second cluster included patients with a more “auto-inflammatory” phenotype, with different clinical presentation, radiological features and immuno-morphological pattern. The inflammatory pathway may be triggered by the virus itself causing type 2 pneumocytes proliferation/activation, influx of inflammatory cells and vascular dysfunction. CT scan showed residual ground glass with perilobular pattern, sometimes consolidations, halo signs and reverse halo sign, with no fibrotic distortion; these changes can be included in a OP-like pattern (table 1). All patients had different morphological types and grades of lung injury, ranging from organising pneumonia to diffuse alveolar damage (DAD - proliferative phase) and had in common abundant immune infiltrates, composed mainly of lymphocytes, in the perivascular and interstitial spaces, sometimes forming nodular aggregates. In this cluster, patients had more frequently systemic symptoms, such as fever and fatigue, together with respiratory symptoms (Table S2). Cluster two patients were treated with steroids following lung biopsy results (details are reported in Supplement). Patients with organising pneumonia on lung biopsy experienced significant improvement in both symptoms and lung function impairment; 3-months follow-up CT scan showed a significant reduction in ground glass areas and perilobular pattern in all cases. Patients with diffuse alveolar damage / proliferative phase features on lung biopsy showed moderate improvement, both clinical and functional, with complete resolution of fever and reduction of shortness of breath, although persistent on mild to moderate exertion; the 3-months follow-up CT scan showed a slight reduction of the extension of parenchymal involvement.
Third subgroup included patients with persistent respiratory and systemic symptoms but minimal changes on CT scan. The histological picture was characterised by dilatation of the lumen of alveolar capillaries and venules and hyperplasia of capillaries within an otherwise normal or minimally abnormal parenchyma (minimal patchy AECII hyperplasia and heterogeneous interstitial thickening in one case). We considered this “pattern” as similar in some aspects to pulmonary capillary hemangiomatosis (hyperplastic interstitial capillaries along and in the alveolar walls, frequently dilated lumens of the capillaries and capillary congestion); however, we think they are clearly two distinct conditions. In fact, the capillary proliferation observed in cluster 3 cases was not typically centered around bronchovascular bundles and did not create any nodular appearance; endothelial cells were negative for proliferation cues (i.e. Ki-67) and vascular changes involved also post-capillary venules, which showed dilated and tortuous lumen, with thickened edematous walls and a patchy perivascular lymphocyte infiltrate, all changes that are not observed in capillary hemangiomatosis. In addition, hemosiderin-laden macrophages and intimal thickening of small muscular arteries were not observed. All these patients underwent echocardiography and pulmonary artery hypertension was excluded. Interestingly, the expression of molecules involved in immune derangement and vascular tone control (STAT3, IDO, PD-L1) were altered as in early phase COVID-19 pneumonia . In order to assess whether these findings were related to Sars-CoV-2 infection, we have analyzed the control cases with diffuse parenchymal lung diseases as uninfected control specimens and we found that the endothelial expression of these markers was faint and/or restricted to rare cells.
One of the most relevant findings of our study was the persistent elevated expression of nuclear pSTAT3 in both alveolar epithelial cells and endothelial cells of interstitial capillary and venules months after the clearing of SARS-CoV-2 infection. Nuclear pSTAT3 was particularly overexpressed in cluster 2 patients, characterised by more prominent systemic symptoms (mostly fever and fatigue) associated with persistence of ground glass opacities on CT scan, perilobular pattern and consolidations. These cases shared similarities with cases of severe COVID-19 pneumonia previously described, in which alveolar epithelial type 2 cells and endothelial cells expressed nuclear pSTAT3. A significant reduction in pSTAT3 after administration of Baricitinib in patients with severe Covid-19 pneumonia, preventing progression, with a safer and more favorable clinical outcome, has been previously demonstrated. These results might suggest a potential rationale for the use of inhibitors of the Janus kinases JAK1 and JAK2 in a subgroup of patients with post-Covid persistent symptoms. All patients included in the study were found negative for Sars-CoV-2 in BAL, confirming the fact that pulmonary manifestations are not associated with the persistence of virus in the lung, but are related to pathogenic mechanisms irrespective of viral clearance and continuing after infections. We have observed that vascular abnormalities (disordered angiogenesis with luminal enlargement and dilated vessel wall, perivascular T-cell infiltration) are distinctly present from the beginning of COVID-19 pneumonia, can persist through progressive disease stages and may linger also several weeks/months after infection. The co-regulatory ligand PD-L1 and the tolerogenic enzyme IDO are part of negative feed-back circuits that restrain immune responses and maintain peripheral tolerance and are both implicated in defense mechanisms from lung injury. Tryptophan is a substrate for IDO enzymatic activity, and abnormal levels of tryptophan metabolites have been demonstrated in Covid-19, with normalisation at recovery. As previously demonstrated, IDO is involved in the regulation of vascular remodelling and relaxation and according to this study its potential role in reducing pulmonary vascular resistance may also be involved in long COVID or post-COVID syndrome. These observations might explain the persistence on CT scan, in some post-COVID cases, of areas of ground-glass attenuation, with or without a “crazy paving” pattern, associated with enlarged veins contained within ground-glass areas, partially disappearing with a change from the supine to the prone position (“venoplegic/hyperemic pattern”). In some post-COVID cases of this series, we observed the persistence of these features on CT scan (table 1), through progressive disease stages several weeks/months after infection.
This study has some limitations. Firstly, the baseline data of pulmonary function tests and chest CT scan were not available, the number of enrolled patients was limited and patients group was heterogeneous. Secondly, the short timeframe between the post-acute sampling and the acute COVID-19 infection maybe a further issue of concern as it may not be representative of the long term abnormalities. In future, a larger sample would be needed and longer longitudinal observational studies will be critical to elucidate the health consequences attributable to COVID-19.
In conclusion, this is the first study showing correlation between histological patterns and immunohistochemistry with clinical and radiological pictures of patients with post-COVID syndrome, which identified different subtypes or phenotypes with potential different underlying pathogenic mechanisms (table 2). Our results confirm that a multidisciplinary evaluation of these patients is needed (instead of a limited imagine evaluation) to identify key patient subgroups and their current optimal management. In particular, immune-histochemical evaluation of lung tissue may help revealing peculiar morphological and morpho-phenotypical changes of this new “Post-COVID/Long-COVID syndrome”, but also help in better understanding of pathogenic mechanisms and lay the groundwork to efficiently conducting therapeutic intervention studies and designing future follow-up plans as previously described. Although the pathogenic role of different cell types and the mechanisms leading to different disease endotypes are not fully understood, this heterogeneity is likely to reflect different COVID-induced phenotypes of interstitial lung disease, probably in predisposed subjects to abnormal wound healing. Post-COVID lung disease is not a single entity, but includes different subtypes or sub-groups], each of them potentially requiring separate and different management.
We thank Fondazione Cariverona (ENACT Project) for the financial support and AMMP (Associazione Morgagni Malattie Polmonari) for patients support.
Declaration of interests: We report no competing interests
Author Contributors: CR, CD, MC and VP conceived of and designed the study; CR, CD and MC wrote the paper; CR, CD, MC, SP, AD, GR and VP contributed to data interpretation; all other authors commented on drafts of the paper and contributed to writing of the final version of the manuscript
Support statement: Not applicable