Figure
Figure S1: Peripheral blood biomarkers and clinical data measured at hospital admission in severe COVID-19 fatal patients identifies distinct disease trajectories associated with the days of symptoms until death. (A,B) PCA biplot of longitudinal clinical variables showing patients' samples color-coded by unsupervised k-means clustering (left panel) and by disease progression (right panel). Principal component analysis (PCA) shows the relationship between the
Source publication
Linking clinical biomarkers and lung pathology still is necessary to understand COVID-19 pathogenesis and the basis of progression to lethal outcomes. Resolving these knowledge gaps enables optimal treatment approaches of severe COVID-19. We present an integrated analysis of longitudinal clinical parameters, blood biomarkers and lung pathology in C...
Contexts in source publication
Context 1
... lymphocyte ratio (NLCR), a recognized feature of severe COVID-19 ( Laing et al., 2020;Lucas et al., 2020;Mann et al., 2020). Fatal cases also had increased levels of tissue injury markers, (e.g., urea, creatinine and LDH) in peripheral blood, and were more anaemic during hospitalization (reduced hematocrit and hemoglobin levels, Figures 1E left panel, S1A). The high variability across clinical parameters in fatal cases suggests higher heterogeneity in the trajectories of disease progression leading to a fatal outcome. ...Context 2
... characterize the drivers of these distinct trajectories in disease progression in our cohort, we first applied Exploratory Factor Analysis (EFA) to the clinical data recorded at and up to 28 days of hospitalization. We identified three clinical signatures based on factor loadings and mapped how they varied during disease progression (Figures 2A, S1D). Clinical signature 1 (CS1) is positively associated with (i) markers of tissue injury, creatinine, urea, LDH and CRP and (ii) neutrophil counts and NLCR. ...Context 3
... clinical signatures indicate that patients following the fatal trajectory (both early and late death) show higher values of CS 1 and a rapid decrease of CS 2 compared to recovered patients during the whole period of hospitalization. However, within the fatal group, the early death progression showed higher levels of clinical signature 1 in the first days of hospital admission and a significantly faster drop of the clinical signature 2 during the first week of hospitalization, when compared to the late death cases (Figures 2A, S1D). Next, we applied EFA to multiplex plasma profiling using bead-based assays (Luminex Panel 2 - Table S1) on a series of plasma from PB samples collected every 2-3 days during hospitalization (at and up to 14 days of hospitalization) from 11 COVID-19 fatal patients (early death=4; late death=7) and 11 recovered patients ( Figures 1A, 2B, S1E, S1F). ...Context 4
... we applied EFA to multiplex plasma profiling using bead-based assays (Luminex Panel 2 - Table S1) on a series of plasma from PB samples collected every 2-3 days during hospitalization (at and up to 14 days of hospitalization) from 11 COVID-19 fatal patients (early death=4; late death=7) and 11 recovered patients ( Figures 1A, 2B, S1E, S1F). The EFA output showed a rapid and sustained increase of PB signatures 1 and 2 in the early death progression during hospitalization (Figures 2B, S1F). The parameters with high positive factor loadings for these signatures (i.e., It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint ...Context 5
... /2023 characterizing the early death progression) were related to myeloid cell response, myeloid chemoattraction, EC activation, vascular damage, coagulopathy, inflammasome activation and inhibition of T-cell responses (Figures 2B, S1F). In comparison, the late death progression was characterized by a slower (when compared to the early death group) and sustained increase (when compared to the recovered group) of PB signature 1. ...Context 6
... diffuse alveolar damage, fibrin deposition, microthrombi and low levels of cell infiltration and MS3 is characterized by high neutrophil infiltration ( Figure 3A). These findings fit with the strong procoagulation phenotype (PB signature 1) observed in the plasma of early death patients during life (Figures 2B, S1F). Morphological signatures 4 and 5 were only found in the lung of fatal patients following the late death progression (9% and 27.6% of late death cases, respectively) ( Figure 3B). ...Context 7
... the single-cell spatially resolved data increases the granularity of tissue features characterizing lung pathology and the biological processes underlying and discriminating the distinct disease progression groups. Figures 5A, 5B, S10A, S10B). The MFA output showed a distinct separation of the early and late death patients in the dimensionality . ...Context 8
... https://doi.org/10. 1101/2023 reduction plot (Figures 5A, S10A). The tissue signatures enriched in the early death progression comprise those highly positively correlated to Dim1 and contributing to Dim 1, such as (i) abundance of and cellular interactions involving SARS-CoV-2 + alveolar macrophages, (ii) mean expression of SARS-CoV-2 antigens in these cells; ...Context 9
... abundance of and cellular interactions involving activated ECs and mean expression of ICAM1, vWF, MHCI in these cells (Figures 5B, S10A, S10B). PB biomarkers capturing most of the variances defining the early death progression 5C-F, S10C-E). ...Context 10
... abundance of and cellular interactions involving activated ECs and mean expression of ICAM1, vWF, MHCI in these cells (Figures 5B, S10A, S10B). PB biomarkers capturing most of the variances defining the early death progression 5C-F, S10C-E). The MOFA factors 1-3, henceforth "integrated signatures" (IS) 1-3, capture variation between 20% and 70% of the total variance across all data modalities (Figures 5C, S10C). ...Context 11
... biomarkers capturing most of the variances defining the early death progression 5C-F, S10C-E). The MOFA factors 1-3, henceforth "integrated signatures" (IS) 1-3, capture variation between 20% and 70% of the total variance across all data modalities (Figures 5C, S10C). Most of the IMC variability driving differences between early and late death progression are well represented in IS 1 and 2, the PB biomarkers variability is mostly captured by IS 2, and finally, clinical data variability is mainly captured by IS 3 ( Figure 5C). ...Context 12
... inferred integrated signatures values for each disease progression were plotted against the days of symptoms ( Figure 5D) and days of sampling ( Figure S10D) during hospitalization. Values of IS 2 and 3 are significantly higher in COVID-. ...Context 13
... 19 fatal patients following the early death progression, while IS 1 is enriched in the late death progression. Once fitted, the model allows the identification of the S11A-C, S12). Figure S11: Integration of signatures in the Amazonian and north American COVID-19 fatal cohorts with MOFA and validation of the predictive potential of the integration signatures in the Amazonian fatal COVID-19. (A) MOFA quality control metrics. ...Context 14
... fitted, the model allows the identification of the S11A-C, S12). Figure S11: Integration of signatures in the Amazonian and north American COVID-19 fatal cohorts with MOFA and validation of the predictive potential of the integration signatures in the Amazonian fatal COVID-19. (A) MOFA quality control metrics. ...Context 15
... survival analysis confirms our observations that hospitalized COVID-19 patients showing elevated levels of the clinical and PB parameters in the integrated signature 2 have a faster disease progression, thus, shorter survival times (early death) ( Figure 5H, right panel). To validate and to verify the practical value of the top 10 clinical and PB parameters in the IS 2 (here referred as "big panel") ( Figures 5E, 5F), in predicting disease progression, when measured at or early after hospital admission (up to 3 days), we trained a RF model and then evaluated its prediction performance in a test dataset (Figures S11E, S11F). Aiming at translating discoveries into a clinically actionable assay that is broadly accessible, we sought to determine the minimum threshold of markers that could be used to predict progression without compromising accuracy, with the goal of reducing our panel to approximately three clinical parameters and three PB parameters, which is more likely to be amenable to clinical routine. ...Context 16
... or five markers, that can be measured by laboratory tests in a blood sample, and tested whether combinations could predict progression in our validation dataset with high accuracy. Using this approach, we achieved similar high accuracy (big panel AUROC train set 0.78; test set 0.83; small panel AUROC train set 0.84; test set 0.84) when we used three clinical parameters (creatinine, CRP and urea) and three PB protein markers (IL11, E-selectin and TPO) in the integrated signature 2 (here referred as "small panel"), in comparison to big panel (Figures S11E, S11F). ...Context 17
... of decision trees and cut-off values for the top three clinical and PB biomarker parameters measured up to three days of hospitalization exemplify how these parameters, such as creatinine, urea, E-selectin and TPO, could assist in prediction of disease progression and patient stratification in the clinical setting with the aim to evaluate the best therapeutic strategies ( Figure S11E). ...Context 18
... lymphocyte ratio (NLCR), a recognized feature of severe COVID-19 ( Laing et al., 2020;Lucas et al., 2020;Mann et al., 2020). Fatal cases also had increased levels of tissue injury markers, (e.g., urea, creatinine and LDH) in peripheral blood, and were more anaemic during hospitalization (reduced hematocrit and hemoglobin levels, Figures 1E left panel, S1A). The high variability across clinical parameters in fatal cases suggests higher heterogeneity in the trajectories of disease progression leading to a fatal outcome. ...Context 19
... characterize the drivers of these distinct trajectories in disease progression in our cohort, we first applied Exploratory Factor Analysis (EFA) to the clinical data recorded at and up to 28 days of hospitalization. We identified three clinical signatures based on factor loadings and mapped how they varied during disease progression (Figures 2A, S1D). Clinical signature 1 (CS1) is positively associated with (i) markers of tissue injury, creatinine, urea, LDH and CRP and (ii) neutrophil counts and NLCR. ...Context 20
... clinical signatures indicate that patients following the fatal trajectory (both early and late death) show higher values of CS 1 and a rapid decrease of CS 2 compared to recovered patients during the whole period of hospitalization. However, within the fatal group, the early death progression showed higher levels of clinical signature 1 in the first days of hospital admission and a significantly faster drop of the clinical signature 2 during the first week of hospitalization, when compared to the late death cases (Figures 2A, S1D). Next, we applied EFA to multiplex plasma profiling using bead-based assays (Luminex Panel 2 - Table S1) on a series of plasma from PB samples collected every 2-3 days during hospitalization (at and up to 14 days of hospitalization) from 11 COVID-19 fatal patients (early death=4; late death=7) and 11 recovered patients ( Figures 1A, 2B, S1E, S1F). ...Context 21
... we applied EFA to multiplex plasma profiling using bead-based assays (Luminex Panel 2 - Table S1) on a series of plasma from PB samples collected every 2-3 days during hospitalization (at and up to 14 days of hospitalization) from 11 COVID-19 fatal patients (early death=4; late death=7) and 11 recovered patients ( Figures 1A, 2B, S1E, S1F). The EFA output showed a rapid and sustained increase of PB signatures 1 and 2 in the early death progression during hospitalization (Figures 2B, S1F). The parameters with high positive factor loadings for these signatures (i.e., It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint ...Context 22
... /2023 characterizing the early death progression) were related to myeloid cell response, myeloid chemoattraction, EC activation, vascular damage, coagulopathy, inflammasome activation and inhibition of T-cell responses (Figures 2B, S1F). In comparison, the late death progression was characterized by a slower (when compared to the early death group) and sustained increase (when compared to the recovered group) of PB signature 1. ...Context 23
... diffuse alveolar damage, fibrin deposition, microthrombi and low levels of cell infiltration and MS3 is characterized by high neutrophil infiltration ( Figure 3A). These findings fit with the strong procoagulation phenotype (PB signature 1) observed in the plasma of early death patients during life (Figures 2B, S1F). Morphological signatures 4 and 5 were only found in the lung of fatal patients following the late death progression (9% and 27.6% of late death cases, respectively) ( Figure 3B). ...Context 24
... the single-cell spatially resolved data increases the granularity of tissue features characterizing lung pathology and the biological processes underlying and discriminating the distinct disease progression groups. Figures 5A, 5B, S10A, S10B). The MFA output showed a distinct separation of the early and late death patients in the dimensionality . ...Context 25
... https://doi.org/10. 1101/2023 reduction plot (Figures 5A, S10A). The tissue signatures enriched in the early death progression comprise those highly positively correlated to Dim1 and contributing to Dim 1, such as (i) abundance of and cellular interactions involving SARS-CoV-2 + alveolar macrophages, (ii) mean expression of SARS-CoV-2 antigens in these cells; ...Context 26
... abundance of and cellular interactions involving activated ECs and mean expression of ICAM1, vWF, MHCI in these cells (Figures 5B, S10A, S10B). PB biomarkers capturing most of the variances defining the early death progression 5C-F, S10C-E). ...Context 27
... abundance of and cellular interactions involving activated ECs and mean expression of ICAM1, vWF, MHCI in these cells (Figures 5B, S10A, S10B). PB biomarkers capturing most of the variances defining the early death progression 5C-F, S10C-E). The MOFA factors 1-3, henceforth "integrated signatures" (IS) 1-3, capture variation between 20% and 70% of the total variance across all data modalities (Figures 5C, S10C). ...Context 28
... biomarkers capturing most of the variances defining the early death progression 5C-F, S10C-E). The MOFA factors 1-3, henceforth "integrated signatures" (IS) 1-3, capture variation between 20% and 70% of the total variance across all data modalities (Figures 5C, S10C). Most of the IMC variability driving differences between early and late death progression are well represented in IS 1 and 2, the PB biomarkers variability is mostly captured by IS 2, and finally, clinical data variability is mainly captured by IS 3 ( Figure 5C). ...Context 29
... inferred integrated signatures values for each disease progression were plotted against the days of symptoms ( Figure 5D) and days of sampling ( Figure S10D) during hospitalization. Values of IS 2 and 3 are significantly higher in COVID-. ...Context 30
... 19 fatal patients following the early death progression, while IS 1 is enriched in the late death progression. Once fitted, the model allows the identification of the S11A-C, S12). Figure S11: Integration of signatures in the Amazonian and north American COVID-19 fatal cohorts with MOFA and validation of the predictive potential of the integration signatures in the Amazonian fatal COVID-19. (A) MOFA quality control metrics. ...Context 31
... fitted, the model allows the identification of the S11A-C, S12). Figure S11: Integration of signatures in the Amazonian and north American COVID-19 fatal cohorts with MOFA and validation of the predictive potential of the integration signatures in the Amazonian fatal COVID-19. (A) MOFA quality control metrics. ...Context 32
... survival analysis confirms our observations that hospitalized COVID-19 patients showing elevated levels of the clinical and PB parameters in the integrated signature 2 have a faster disease progression, thus, shorter survival times (early death) ( Figure 5H, right panel). To validate and to verify the practical value of the top 10 clinical and PB parameters in the IS 2 (here referred as "big panel") ( Figures 5E, 5F), in predicting disease progression, when measured at or early after hospital admission (up to 3 days), we trained a RF model and then evaluated its prediction performance in a test dataset (Figures S11E, S11F). Aiming at translating discoveries into a clinically actionable assay that is broadly accessible, we sought to determine the minimum threshold of markers that could be used to predict progression without compromising accuracy, with the goal of reducing our panel to approximately three clinical parameters and three PB parameters, which is more likely to be amenable to clinical routine. ...Context 33
... or five markers, that can be measured by laboratory tests in a blood sample, and tested whether combinations could predict progression in our validation dataset with high accuracy. Using this approach, we achieved similar high accuracy (big panel AUROC train set 0.78; test set 0.83; small panel AUROC train set 0.84; test set 0.84) when we used three clinical parameters (creatinine, CRP and urea) and three PB protein markers (IL11, E-selectin and TPO) in the integrated signature 2 (here referred as "small panel"), in comparison to big panel (Figures S11E, S11F). ...Context 34
... of decision trees and cut-off values for the top three clinical and PB biomarker parameters measured up to three days of hospitalization exemplify how these parameters, such as creatinine, urea, E-selectin and TPO, could assist in prediction of disease progression and patient stratification in the clinical setting with the aim to evaluate the best therapeutic strategies ( Figure S11E). ...Citations
... Overall, we found that infiltrating macrophages and proliferating epithelial cells constitute most of the cells in the affected areas of the lung parenchyma of hamsters infected with virulent SARS-CoV-2 variants. We consistently observed epithelial hyperplasia and macrophage infiltrates also in a separate study based on histopathology of the lungs of patients who died as result of the first wave of COVID-19 [60]. These two features represent therefore exceedingly good markers that by themselves provide an unbiased quantification of the virus-induced pulmonary lesions, and by extension virus virulence. ...
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has continued to evolve throughout the coronavirus disease-19 (COVID-19) pandemic, giving rise to multiple variants of concern (VOCs) with different biological properties. As the pandemic progresses, it will be essential to test in near real time the potential of any new emerging variant to cause severe disease. BA.1 (Omicron) was shown to be attenuated compared to the previous VOCs like Delta, but it is possible that newly emerging variants may regain a virulent phenotype. Hamsters have been proven to be an exceedingly good model for SARS-CoV-2 pathogenesis. Here, we aimed to develop robust quantitative pipelines to assess the virulence of SARS-CoV-2 variants in hamsters. We used various approaches including RNAseq, RNA in situ hybridization, immunohistochemistry, and digital pathology, including software assisted whole section imaging and downstream automatic analyses enhanced by machine learning, to develop methods to assess and quantify virus-induced pulmonary lesions in an unbiased manner. Initially, we used Delta and Omicron to develop our experimental pipelines. We then assessed the virulence of recent Omicron sub-lineages including BA.5, XBB, BQ.1.18, BA.2, BA.2.75 and EG.5.1. We show that in experimentally infected hamsters, accurate quantification of alveolar epithelial hyperplasia and macrophage infiltrates represent robust markers for assessing the extent of virus-induced pulmonary pathology, and hence virus virulence. In addition, using these pipelines, we could reveal how some Omicron sub-lineages (e.g., BA.2.75 and EG.5.1) have regained virulence compared to the original BA.1. Finally, to maximise the utility of the digital pathology pipelines reported in our study, we developed an online repository containing representative whole organ histopathology sections that can be visualised at variable magnifications ( https://covid-atlas.cvr.gla.ac.uk ). Overall, this pipeline can provide unbiased and invaluable data for rapidly assessing newly emerging variants and their potential to cause severe disease.
