Age-related changes in plasma biomarkers and their association with mortality in COVID-19

Abstract

Background: Coronavirus disease 2019 (COVID-19)-induced mortality occurs predominantly in older patients. Several immunomodulating therapies seem less beneficial in these patients. The biological substrate behind these observations is unknown. The aim of this study was to obtain insight into the association between ageing, the host response and mortality in patients with COVID-19.

Methods: We determined 43 biomarkers reflective of alterations in four pathophysiological domains: endothelial cell and coagulation activation, inflammation and organ damage, and cytokine and chemokine release. We used mediation analysis to associate ageing-driven alterations in the host response with 30-day mortality. Biomarkers associated with both ageing and mortality were validated in an intensive care unit and external cohort.

Results: 464 general ward patients with COVID-19 were stratified according to age decades. Increasing age was an independent risk factor for 30-day mortality. Ageing was associated with alterations in each of the host response domains, characterised by greater activation of the endothelium and coagulation system and stronger elevation of inflammation and organ damage markers, which was independent of an increase in age-related comorbidities. Soluble tumour necrosis factor receptor 1, soluble triggering receptor expressed on myeloid cells 1 and soluble thrombomodulin showed the strongest correlation with ageing and explained part of the ageing-driven increase in 30-day mortality (proportion mediated: 13.0%, 12.9% and 12.6%, respectively).

Conclusions: Ageing is associated with a strong and broad modification of the host response to COVID-19, and specific immune changes likely contribute to increased mortality in older patients. These results may provide insight into potential age-specific immunomodulatory targets in COVID-19.

Overview of the study findings. TNF-R1: tumour necrosis factor receptor 1; TREM-1: triggering receptor expressed on myeloid cells 1. Figure partially created with BioRender.com.

FIGURE 1

Overview of the three COVID-19 cohorts in which plasma biomarkers were measured. The external validation cohort was derived from a publicly available dataset of nonintubated COVID-19 patients in whom plasma proteins were measured by the Olink proximity extension assay [21]. ICU: intensive care unit. Figure partially created with BioRender.com.

FIGURE 2

Mortality analysis of COVID-19 patients admitted to the general ward stratified by age decade. a) Kaplan–Meier plot of patients stratified by age decade. b) Risk of 30-day mortality with age modelled as a continuous variable. Given the nonlinear relationship between age and mortality, a restricted cubic spline function with three inner knots at default quantile locations was used. To calculate the odds ratio, the reference was set to 60 years of age. c) The same method as b), but now the 30-day mortality odds ratio is adjusted for demographics (inclusion hospital, sex and inclusion wave), age-related comorbidities (hypertension, diabetes, malignancies, immunosuppression, and chronic cardiac, neurological, respiratory and kidney disease), age-related chronic medication (antiplatelet and anticoagulant drugs), and COVID-19-related treatments both before and after sampling (corticosteroids including dexamethasone, anti-interleukin-6, imatinib, remdesivir and antibiotics).

FIGURE 3

Principal component analysis (PCA) of host response domain differences between age groups: a) endothelial and coagulation response, b) systemic inflammation and organ damage, c) cytokines and d) chemokines. Principal components (PCs) 1 and 2 are plotted per domain. For each domain, the x- and y-axes are labelled with the percentage of the total variance within that domain that is explained by PC1 and PC2, respectively. The complete contribution of each biomarker to a PC score is depicted in supplementary table S6. The ellipse indicates the central 10% of each age group, colour coded as indicated in the key at the bottom of the figure. The arrows indicate the direction (arrow orientation) and strength (arrow length) of the correlation between each biomarker and the PCs. Next to each PCA plot are box plots with 1.5 interquartile range whiskers of PC1 and PC2. Herein, upper p-values were obtained by ANOVA between age groups: ρ-values with accompanying p-values were generated using a Spearman's correlation with ageing on a continuous scale. Note that a negative association of a PC with ageing may still reflect a positive association with biomarker concentration, as reflected by the direction of the arrows. Post-hoc testing was done with a Tukey test. *: p<0.05; **: p<0.01; ***: p<0.001. ANG: angiopoietin; sTie-2: soluble Tie-2; sE-selectin: soluble E-selectin; sThrombomodulin: soluble thrombomodulin; sVCAM-1: soluble vascular cellular adhesion molecule 1; PAI-1: plasminogen activator inhibitor 1; sCD31: soluble cluster of differentiation 31; sRAGE: soluble receptor for advanced glycation end-products; sTNF-R1: soluble tumour necrosis factor receptor 1; sTREM-1: soluble triggering receptor expressed on myeloid cells 1; SP-D: surfactant protein D; CD40L: CD40 ligand; PD-L1: programmed death ligand 1; CCL:C-C motif chemokine ligand; CXCL: C-X-C motif chemokine ligand; IL: interleukin; TNF: tumour necrosis factor; GM-CSF: granulocyte–macrophage colony-stimulating factor; IFN: interferon.

FIGURE 4

Association of host response biomarkers with ageing. a) Heatmap depicting the magnitude of biomarker differences (Hedges’ g) between patients ≥70 years compared with other age groups. p-values were obtained from a linear (if linear) or cubic spline regression analysis (if nonlinear) in which age was modelled as a continuous variable. The adjusted model included demographics, age-related comorbidities, age and biomarker-related chronic medication, and COVID-19-related immunomodulating treatments before sampling (see Methods for details). Red indicates higher levels in patients ≥70 years and blue indicates lower levels in patients ≥70 years. ^#: biomarkers with a nonlinear relationship with ageing on a continuous scale. b) Volcano plot depicting the strength of the correlation between a biomarker and ageing. Red dots represent a significant positive correlation, blue dots represent a significant negative correlation and grey dots represent a nonsignificant correlation. Both the adjusted and unadjusted p-values are multiple testing corrected using the Benjamini–Hochberg (BH) procedure for testing 43 biomarkers. *: p<0.05; **: p<0.01; ***: p<0.001. ANG: angiopoietin; sTie-2: soluble Tie-2; sE-selectin: soluble E-selectin; sThrombomodulin: soluble thrombomodulin; sVCAM-1: soluble vascular cellular adhesion molecule 1; PAI-1: plasminogen activator inhibitor 1; sCD31: soluble cluster of differentiation 31; sRAGE: soluble receptor for advanced glycation end-products; sTNF-R1: soluble tumour necrosis factor receptor 1; sTREM-1: soluble triggering receptor expressed on myeloid cells 1; SP-D: surfactant protein D; CD40L: CD40 ligand; PD-L1: programmed death ligand 1; CCL: C-C motif chemokine ligand; CXCL: C-X-C motif chemokine ligand; IL: interleukin; TNF: tumour necrosis factor; GM-CSF: granulocyte–macrophage colony-stimulating factor; IFN: interferon.

FIGURE 5

Mediation analysis for ageing-associated mortality and host response biomarkers upon admission to the general ward. a) Quadrant plot. The x-axis depicts the increase in the 30-day mortality odds ratio per 25% increase of the biomarker derived from an unadjusted logistic regression with the log-transformed biomarker as the explanatory variable and 30-day mortality as the response variable. The y-axis shows the percentage change in the biomarker concentration per 1-year increase in age which was derived from an unadjusted linear regression analysis with the log-transformed biomarker as the response variable. Biomarkers in both the top-right and bottom-left corners are most likely associated with an age-dependent increase in 30-day mortality. The significance of the coefficient was multiple testing corrected using the Benjamini–Hochberg procedure for testing 43 biomarkers. b) Quadrant plot. The same method as in a), but both coefficients are now adjusted for demographics (inclusion hospital, sex and inclusion wave), age-related comorbidities (hypertension, diabetes, malignancies, immunosuppression, and chronic cardiac, neurological, respiratory and kidney disease), age- and biomarker-related chronic medication (antiplatelet and anticoagulant drugs), and COVID-19-related treatments both before and after sampling (corticosteroids including dexamethasone, anti-interleukin (IL)-6, imatinib, remdesivir and antibiotics). c) Diagram of mediation analysis. The adjusted model contained the same covariates as in b). d) Unadjusted and e) adjusted mediation analysis results. Only biomarkers and principal components (PCs) significantly associated with ageing and 30-day mortality were analysed. Confidence intervals were obtained from 1000 times bootstrapping. The higher the proportion of mediation, the stronger the association of the age-dependent differences in that biomarker and the age-dependent increase in 30-day mortality. The PCs and their contributing biomarkers are depicted in figure 3. The complete contribution of each biomarker to a PC score is depicted in supplementary table S6. Endocoag: endothelial and coagulation; ANG: angiopoietin; sTie-2: soluble Tie-2; sE-selectin: soluble E-selectin; sThrombomodulin: soluble thrombomodulin; sVCAM-1: soluble vascular cellular adhesion molecule 1; PAI-1: plasminogen activator inhibitor 1; sCD31: soluble cluster of differentiation 31; sRAGE: soluble receptor for advanced glycation end-products; sTNF-R1: soluble tumour necrosis factor receptor 1; sTREM-1: soluble triggering receptor expressed on myeloid cells 1; SP-D: surfactant protein D; CD40L: CD40 ligand; PD-L1: programmed death ligand 1; CCL: C-C motif chemokine ligand; CXCL: C-X-C motif chemokine ligand; IL: interleukin; TNF: tumour necrosis factor; GM-CSF: granulocyte–macrophage colony-stimulating factor; IFN: interferon.