Diffuse alveolar damage patterns reflect the immunological and molecular heterogeneity in fatal COVID-19

Abstract

Background: Severe COVID-19 lung disease exhibits a high degree of spatial and temporal heterogeneity, with different histological features coexisting within a single individual. It is important to capture the disease complexity to support patient management and treatment strategies. We provide spatially decoded analyses on the immunopathology of diffuse alveolar damage (DAD) patterns and factors that modulate immune and structural changes in fatal COVID-19.

Methods: We spatially quantified the immune and structural cells in exudative, intermediate, and advanced DAD through multiplex immunohistochemistry in autopsy lung tissue of 18 COVID-19 patients. Cytokine profiling, viral, bacteria, and fungi detection, and transcriptome analyses were performed.

Findings: Spatial DAD progression was associated with expansion of immune cells, macrophages, CD8+ T cells, fibroblasts, and (lymph)angiogenesis. Viral load correlated positively with exudative DAD and negatively with disease/hospital length. In all cases, enteric bacteria were isolated, and Candida parapsilosis in eight cases. Cytokines correlated mainly with macrophages and CD8+T cells. Pro-coagulation and acute repair were enriched pathways in exudative DAD whereas intermediate/advanced DAD had a molecular profile of elevated humoral and innate immune responses and extracellular matrix production.

Interpretation: Unraveling the spatial and molecular immunopathology of COVID-19 cases exposes the responses to SARS-CoV-2-induced exudative DAD and subsequent immune-modulatory and remodeling changes in proliferative/advanced DAD that occur side-by-side together with secondary infections in the lungs. These complex features have important implications for disease management and the development of novel treatments.

Funding: CNPq, Bill and Melinda Gates Foundation, HC-Convida, FAPESP, Regeneron Pharmaceuticals, and the Swedish Heart & Lung Foundation.

Keywords: Autopsy; COVID-19; Diffuse alveolar damage; Immunopathology; SARS-Cov-2.

Figure 1

H&E-stained lung biopsies from ultrasound-guided minimally invasive autopsies confirm typical pulmonary histological findings of fatal cases of COVID-19. (A) Exudative diffuse alveolar damage with hyaline membranes (arrows) in the alveolar space, alveolar oedema, and congestion. (B) Mixed pattern of diffuse alveolar damage showing the combination of areas of haemorrhage and thickened alveolar septa with loose collagen deposition. Asterisks show the septal thickening. (C) Proliferative phase of diffuse alveolar damage showing thickened alveolar septa with deposition of collagen, lymphocytic infiltrate, and reactive and proliferative type II pneumocytes during alveolar reepithelization. Circled area shows the lymphocytic infiltrate; arrow points to collagen deposition and arrowheads indicate the proliferation of type II pneumocytes. (D) Secondary suppurative bacterial pneumonia characterized by intra alveolar exudation with macrophages and polymorphonuclear infiltrate, indicated by the asterisks. (E) Pulmonary artery with thrombus, indicated by the asterisk. Scale bar = 100 µm.

Figure 2

Correlation plot showing the significant correlations between immune and structural cell quantifications, cytokines, and clinical data of COVID-19. The specific correlation coefficients and p-values are shown in Table S5. DAD, Diffuse Alveolar Damage; DC, Dendritic cells; ICU, Intensive Care Unit; MV, Mechanical Ventilation.

Figure 3

Multiplex imaging identifies complex patchworks of parallel histopathological microenvironments as a common feature in fatal COVID-19: Exemplification of the histopathological heterogeneity by combined H&E and multiplex IHC imaging in a single COVID-19 lung biopsy. (A) Low power image from a COVID-19 lung biopsy where the main structural cell markers, neutrophils, and macrophages are highlighted. The other immune cell markers are combined into a blue label to reduce the visual complexity. The marked patchy and compartmentalized histopathological heterogeneity is seen as distinct microenvironments with concomitant exudative DAD areas, localized neutrophilia (PMN), hemorrhage and necrosis (HN), macrophage and fibroblast-rich clusters (MQC), alveolar epithelial hyperplasia (EPH) and fibrotic lung tissue in advanced DAD areas. (B-H) Zoomed-in pairwise images with routine H&E staining (upper panel) and corresponding multiplex IHC image (lower panel). (B) Area with exudative DAD and marked ongoing shedding of the alveolar epithelium (green, arrowheads). (C) Exudative DAD with almost complete denudation of epithelial cells (asterisk) and mild early influx of neutrophils (yellow). (D) Pulmonary blood vessel with focal loss of endothelium. (E) Blood vessel occluded with neutrophils (yellow) and monocytes (red) in an area with alveolar epithelial and vascular disarray. (F) Intermediate DAD with epithelial hyperplasia accompanied by macrophages (red) and fibroblasts (brown; asterisks denote intraluminal fibrosis; arrowheads denote microthrombosis). (G) Neutrophil and hemorrhage space (asterisk) flanked with dense macrophage and fibroblast sheets (bracket). (H) Advanced DAD with epithelial hyperplasia (Hep) and emergence of organized fibroblasts, solitary smooth muscle cells, and myofibroblasts (asterisks). (I-K) High power images exemplifying fibroblasts (I), solitary smooth muscle cells (J), and aSMA⁺, vimentin⁺ double-positive myofibroblast (K) associated with COVID-19 associated advanced DAD. BV= blood vessel, Alv = alveolar space, Hep = Hyperplastic epithelium. Scale bars: A = 0.6 mm; B-H = 70 µm; I-K = 20 µm.

Figure 4

Formation of heterogeneous and compartmentalized immune cell niches in COVID lungs. (A) Adjusted coefficient of spatial variation across the IHC multiplex markers. Each dot represents the mean coefficient of variation per marker for the hundreds of analyzed pre-defined 1000 × 1000-pixel lung tissue areas explored per sample. Statistical differences between markers (see p-values in the text) were determined by a non-parametric Dunn´s test for all pairs with Bonferroni adjustment. (B) Different spatial distribution patterns among immune cell populations. Panels exemplify color-coded weighted spatial density maps for B-lymphocytes, neutrophils, monocytes/macrophages, myeloid DCs, eosinophils, and CD8 T-lymphocytes. The density plots are based on cell x,y coordinates within the same tissue section and display density gradients from low (blue) to high (red). (C-D) Example of AI-based non-supervised identification of spatial immune cell patterns in a single lung section. The analysis was performed by the CytoMAP platform using marker identity and x,y coordinates for individual immune cells. The spatial distribution of regions with identified and color-coded immune cell neighborhoods is shown in C, whereas the corresponding relative immune cell compositions are shown in panel D. (E-H) Corresponding multiplex micrograph images from AI-identified tissue regions are outlined in panel C. (E) Intermediate DAD; (F) marked epithelial hyperplasia and neutrophil infiltration; (G) advanced DAD, (H) region with pigmented macrophages and lymphoid tissue. AP = anthracotic pigment, ECP = the eosinophil marker eosinophil cationic protein, Trypt/Chym = Mast cell markers, Lang = langerin/CD207 DCs, MPO = neutrophil marker, vim = the fibroblast marker vimentin, Glycoph = red blood cell marker, Cytoker = epithelial marker, SMA smooth muscle actin, vim = the fibroblast marker vimentin, Collag = Collagen 1, AL=alveolar lumen, FT= fibrotic tissue, LT = lymphoid tissue. Scale bars: E and F = 65 µm; G and H = 50 µm.

Figure 5

Decoding of cell marker compositions during DAD progression. (A-C) Zoomed-in micrographs of diffuse alveolar damage (DAD) patterns, as viewed by traditional H&E staining and corresponding multiplex immunohistochemistry. Panels A-C show paired H&E and multiplex images and typical patterns during exudative, intermediate, and advanced DAD, respectively. (D-E) Quantitative data on the density of immune cells (D) and structural cell markers (E) across the DAD patterns. The data are from marker density analysis in 95 multiplex-stained tissue regions of interest (ROIs) that were selected from H&E-stained sections with the criteria of having a uniform and distinct DAD histopathology. Statistical comparisons were determined by a non-parametric Kruskal–Wallis test, followed by Bonferroni post-hoc test. (F) Multivariate analysis of individual DAD region marker content and identification of 3 clusters of marker constellations by principal component analysis (PCA) and unsupervised K-mean clustering (Clusters 1-3). Individual ROIs within the PCA-defined clusters are color-coded according to previously H&E-confirmed DAD patterns. (G) Individual ROIs sorted for increasing abundance of the 4 markers that had the most statistical influence on initial cluster identification, again individual ROIs are color-coded according to DAD category. (H) Cell plots with relative marker densities across the identified clusters. Each horizontal line represents one DAD region; with its DAD category color-coding to the right. *p<0·05 and **p<0·01.