Eosinophilic pulmonary vasculitis as a manifestation of the hyperinflammatory phase of COVID-19.

To the Editor:

The role of eosinophils in coronavirus disease 2019 (COVID-19) is still a matter of debate and eosinopenia is regarded as a negative predictive factor. ,

A 60-year-old male patient with severe COVID-19 was mechanically ventilated for 6 days before transfer to our hospital. Comorbidities were hypertension and diabetes, but no allergic disorders or asthma. Body mass index was 26.3 kg/m2. He was a former smoker. Initially, C-reactive protein was 202 mg/L, procalcitonin 0.693 ng/mL, and white blood cell count 9.45 gpt/L, with normal percentages of eosinophils and lymphocytes.

He received antibiotics and fluid management. Computed tomography on day 5 postintubationem (PI) showed bipulmonary ground glass opacities and basal consolidations, with marked progression on day 16 PI, necessitating extracorporeal membrane oxygenation starting on day 22 PI. Severe acute respiratory syndrome coronavirus 2 PCR was still positive on day 33 PI and negative from day 37 PI. He showed only a very slight increase in anti–severe acute respiratory syndrome coronavirus 2 antibodies in plasma, about one-third in comparison to other patients with a severe course of COVID-19. The arbitrary result of the assay amounted to 13 on day 16 PI with no significant increase in the following month (Elecsys immunoassay; Roche, Grenzach-Wyhlen, Germany; cutoff index 1.0).

Bronchoalveolar lavage on day 32 PI yielded 30% lymphocytes (97% CD3+ T cells, 53% cytotoxic CD8+ CD3+ T cells), 25% neutrophils, and 36% eosinophils as well as 130 pg/mL IL-6 (Elecsys IL-6; Roche), 2.4 pg/mL IL-5 (ELISA MAX Human IL-5; BioLegend, San Diego, Calif), and 4.5 pg/mL TNF-α (Human TNF-alpha Quantikine ELISA Kit; R&D, Minneapolis, Minn).

Therefore, prednisolone treatment was initiated on day 32 PI. Notably, there had been a constant increase in peripheral blood eosinophils (0% day 6 PI; 2.3% day 41 PI; peak 4.8% day 23 PI), which however did not exceed the upper limit of normal.

Because of intrathoracic bleeding, thoracotomy was performed 8 days after initiation of prednisolone therapy (day 40 PI). Lung parenchyma showed marked eosinophilic vasculitis, organizing diffuse alveolar damage, patchy interstitial inflammation, and acute intraalveolar hemorrhage, without evidence for ongoing infection. This histopathologic pattern has not been described so far (Fig 1 ).

Fig 1

Histology of lung parenchyma resected on day 40 after intubation (8 days after initiation of prednisolone therapy) histology showed diffuse alveolar damage of proliferative organizing phase: polypoid plugs of loose organizing connective tissue protruding into the alveolar ducts (arrows) (A, hematoxylin-eosin, B, Masson-Trichrome), prominent type 2 pneumocyte hyperplasia with cytologic atypia (C, pancytokeratine, red), and alveolar macrophage accumulation (C, CD68, brown). Focally, peribronchial interstitial chronic inflammation with dominating CD3+ CD8+ T cells (D, CD3, E, CD8) and single CD20+ B cells (F) was observed. Patchy subpleural accentuated acute hemorrhage (asterisks) and vasculitis with endothelial swelling, fibrinoid necrosis, and transmural eosinophilic infiltrate (arrows) (G-I) were observed.

Serum analysis confirmed hyperinflammation after viral clearance with markedly elevated IL-6 (71 pg/mL day 8 PI; 111 pg/mL day 46 PI; peak 344 pg/mL day 27 PI; Elecsys IL-6; Roche). On day 41 PI, HLA-DR–activated CD8+ T cells in the blood were increased (282/μL), B lymphocytes reduced (87/μL), and the plasma concentration of IL-5 was 0.5 pg/mL (Cytometric Bead Array Human Th1/Th2/Th17 Kit; BD, Franklin Lakes, NJ). The later result might be influenced by prednisolone therapy. Autoantibodies (antinuclear antibodies, extractable nuclear antigens, perinuclear anti-neutrophil cytoplasmic antibody, and cytoplasmic anti-neutrophil cytoplasmic antibody [indirect immunofluorescence test]) as well as anti-proteinase 3 and anti-myeloperoxidase ELISAs were negative on day 41 PI.

After 2 weeks of prednisolone therapy (day 45 PI), a second BAL yielded 8% lymphocytes (52% CD3+ CD8+ cytotoxic T cells, 60.2% HLA-DR+ indicating an activated phenotype), 3% eosinophils, and 62% neutrophils as well as 175 pg/mL IL-6 (Elecsys IL-6; Roche), 2.3 pg/mL IL-5 (ELISA MAX Human IL-5; BioLegend), and 3.6 pg/mL TNF-α (Human TNF-alpha Quantikine ELISA Kit; R&D).

On day 46 PI, flow cytometric analysis confirmed massive HLA-DR activation in peripheral CD8+ T cells and strong alterations in neutrophils and monocytes as well as dendritic-cell subsets.

In summary, this case shows that the hyperinflammatory phase of COVID-19 may present as eosinophilic inflammation of the pulmonary vasculature without distinct peripheral eosinophilia. Pulmonary eosinophilia might be a potential therapeutic target (eg, corticosteroids and anti–IL-5 antibodies), and this case underlines the diagnostic value of BAL cytology. Furthermore, we assume that even a slight increase in peripheral eosinophils not exceeding upper limit of normal may be meaningful.

Notably, the patient showed a weak humoral immune response in combination with an exceptionally strong activation of CD8+ T cells detectable in blood and BAL. Our observations indicate that besides the spectrum of hyperinflammation as described very recently, pulmonary eosinophilic vasculitis without preexisting allergic disorders may occur. The cause of the eosinophilia at this point remains elusive. Although it may indicate a type 2 response, we did not detect a significant elevation of type 2 cytokines. Altogether this case demonstrates the vast breadth of heterogeneous immunological mechanisms at play in severe COVID-19.