Background
The recent COVID-19 pandemic highlighted the role of ageing as a risk factor for severe disease after infection. Indeed, age has been described as an independent risk factor for both severity and mortality during COVID-19 [
1,
2]. Ageing leads to modifications in adaptive immunity with thymic involution and reduction of naive T cells while memory and differentiated T cells (i.e. CD27
−CD28
− and EMRA) accumulate [
3].
CD8
+ T cells are the main cytotoxic effectors and are major players of antiviral defence through their killing activity of infected cells after antigen recognition [
4]. Ageing is accompanied by a CD4 and CD8 lymphopenia, and a decrease of both CD8
+ T cell priming and generation after a newly encountered antigen [
5]. This decreased specific CD8
+ T cell response may be responsible for poor vaccine efficacy and may lead to severe manifestations from seasonal infection or latent virus reactivation in some individuals [
6]. For instance, both Th1 response and CD8
+ T cell cytotoxicity are impaired following an in vitro influenza virus exposure in older population [
7]. After SARS-CoV-2 infection, T cells confer protective immunity [
8] and CD8
+ T cell memory persist up to 6 to 8 months in half of the affected individuals [
9]. A coordinated adaptive immune response is protective against severe forms of COVID-19 [
1]. In older COVID-19
+ patients, there is (i) an impaired capacity of CD8
+ T cell priming after in vitro stimulation [
10] and (ii) less secretion of granzyme A and B by CD8
+ T cells after SARS-CoV2 infection than in younger patients [
11].
Biomarkers of T cell senescence have emerged recently. CD8
+ T cells in particular, can overexpress (i) receptors usually found on natural killer (NK) cells such as CD57, KLRG1, CD56 and NKG2a and (ii) a Sestrin-2/MAP-Kinases complex [
3,
12]. SARS-CoV-2 infection has been associated with higher and persistent CD57 expression by CD8
+ T cells [
13], but a characterization of the CD8 senescence profile with all the above-mentioned markers has not been performed.
To our knowledge, the association between immune ageing and COVID-19 in older population has not been reported yet.
The dysregulated immune response observed in severe COVID-19 cases could also be partly due to a non-specific inflammation [
1,
14]. In COVID-19, a cytokine storm is involved in the severe cases and is in part responsible for tissue lesions particularly in lungs [
15]. Ageing can be accompanied by inflammaging which is a low-grade chronic inflammation that is observed in healthy individuals even without any infection. This inflammation is characterized by an elevation of C reactive protein (CRP) and cytokines such as IL-6, IL-1β, CCL2 (MCP-1), CXCL8 (IL-8) and CXCL1 (GRO-A) even in the absence of any detectable stimulus [
3,
16]. Additionally, some cytokines such as IL-1Ra and IL-10 seem to display anti-inflammaging properties and may favour longevity [
17]. These cytokines are also produced by senescent non-lymphoid cells comprising the senescence-associated secretory phenotype (SASP). The SASP is a secretome of cytokines including the ones mentioned above as well as TNFα, IL-1Ra, CCL20 (MIP-3A), CXCL-2 (GRO-B), GM-CSF, G-CSF and VEGF [
16]. It is interesting to note that all these cytokines associated with inflammaging and SASP favour the myeloid response (i.e. myelopoiesis and cell recruitment and proliferation).
Besides the lymphoid response which is protective, the myeloid response (neutrophils and monocytes) is also implicated in disease severity and tissue lesions during COVID-19 [
18‐
20]. Neutrophil and macrophage circulating biomarkers are associated with mortality in older patients [
21].
Thus, in addition to the CD8 senescence profile, we also characterized the inflammaging and myeloid lineage with (i) other cytokines involved in myeloid cell migration, activation or differentiation such as G-CSF, IL-33, fractalkine or others secreted by myeloid cells such as CXCL10 (IP-10) and (ii) neutrophil and monocyte cell counts.
In this study, we performed comprehensive immune phenotyping in older patients during COVID-19, by exploring the cytotoxic T cell senescent phenotype and soluble inflammaging markers. We compared hospitalized patients over 70 years old affected (COVID-19+ cohort stratified on severity) or not (control cohort). A younger cohort was used to compare the age-differential positive association of cytokines with severity. We aimed to better characterize CD8+ T cells and cytokines factors that are associated with COVID-19 severity in older patients.
Material and methods
Study design
Hospitalized patients were prospectively enrolled from 24/04/2020 to 19/03/2021 in the Georges Pompidou European Hospital (HEGP), Paris, France during the first and second COVID-19 waves. The local Ethical Committee approved this study (CERAPHP “Centre Comité d'éthique de la recherche AP-HP Centre”, IRB registration: #00011928). One hundred and four patients were included in the study: 81 patients in the COVID-19+ group and 23 in the control group (COVID-19 negative).
Inclusion criteria in the COVID-19+ group were as follows: > 70 years old, confirmed SARS-CoV-2 infection (positive RT-PCR test on a respiratory sample: nasopharyngeal swab or invasive respiratory sample), initial onset of COVID-19 signs/symptoms within 8 days prior to the day of inclusion.
A comparative control group matched on age and sex was constituted within the same period including 23 patients negative for COVID-19, hospitalized in the geriatric department for non-infectious diseases. Exclusion criteria for controls were confirmed or active infection and a positive SARS-CoV-2 RT-PCR test.
In the whole cohort, exclusion criteria were the presence of hemopathy or a treatment by long-term immunosuppressive drug.
A flow chart illustrates the study design (Figure S
1).
Data and sample collection
Sociodemographic characteristics (age, gender, housing: home or nursing home), previous medical history, daily treatment, clinical symptoms in the COVID-19+ cohort, corticoid treatment introduced during the COVID-19 care, in-hospital mortality and laboratory data were collected and extracted from a secured and standardized electronic case report form (eCRF Redcap).
Routine blood tests, including complete blood counts (CBC), plasmatic biochemical tests (including markers of renal and liver function, lactate dehydrogenase (LDH) and electrolytes), C-reactive protein (CRP), ultrasensitive I-troponin were collected.
Blood samples were drawn as soon as the COVID-19 diagnosis was confirmed and sent the same day to our laboratory for immunophenotyping and cytokines measurement. Thus, biological samples were taken before any COVID-19 treatment (including dexamethasone) was started.
Comorbidity severity was assessed with the Charlson Comorbidity Index (CCI) [
22] and frailty was assessed with the adjusted Rockwood Frailty Scale (RFS) [
23].
The time of COVID-19 onset was defined as the date when the first signs or symptoms were noticed. Signs or symptoms included fever, shortness of breath, increased respiratory rate, dry cough, chest tightness, fatigue, myalgia, hypotension, pharyngalgia, diarrhea, nausea, vomiting, abdominal pain, dizziness, delirium, headache, anosmia and agueusia. A quick Sequential Organ Failure Assessment (qSOFA) score was calculated for COVID-19
+ patients [
24]. The qSOFA score ranges from 0 (best) to 3 (worst) points, with one point allocated for: (1) systolic blood pressure ≤ 100 mm Hg, (2) respiratory rate ≥ 22 breaths/min (3) altered mental status (Glasgow Coma Score < 14).
COVID-19+groups stratification based on severity
Patients included in the COVID-19
+ groups were classified, in the first 48 h following diagnosis, according to the WHO Clinical Progression Scale (0–10) [
25] adapted to the elderly population, then stratified into 3 groups:
-
- Group 1: Non-severe COVID-19 defined as asymptomatic or symptomatic without oxygen requirement (WHO score range 2–3).
-
- Group 2: Non-critical COVID-19 defined as median oxygen requirement (≥ 3L/min) with nasal or mask prongs (WHO score range 4–5).
-
- Group 3: Critical COVID-19 defined as high flow oxygen therapy (HFOT) or Non-Invasive Ventilation (NIV) or Invasive Mechanical Ventilation (IMV) (WHO score range 6–9).
Being a real-life geriatric cohort with a lot of patients with multiple comorbidities and loss of autonomy, some patients in group 3, who developed a severe form of COVID-19 and had a theorical indication to be admitted to an intensive care unit, were not, based on multidisciplinary decision. Indeed, the benefit of such admission was collegially considered too minimal by the physicians. Such decision was taken in accordance with the family and the wishes of the patient when possible. They were nevertheless classified in the COVID-19 critical group (group 3).
T cell immunophenotype by flow cytometry
Fresh whole blood (100 µL) was labelled for surface and intracellular markers distributed in 3 flow cytometry panels: panel 1: CD3, CD8, CD4, CD45RA, CD27, CD28 and CD95, panel 2: CD3, CD8, CD4, CD45RA, CD27, CD28, Ki-67 and Sestrin-2 and panel 3: CD3, CD8, CD4, CD56, CD57, NKG2A, KLRG1 and DAP12 (see Table S
4 for the specific clones and brand). All were direct stainings except for Sestrin-2. All stainings were performed according to the manufacturer’s recommendations, with Sestrin-2 expression being assessed after fixation and permeabilization (Biolegend FoxP3 Perm and Fix). After labelling, red blood cells were lysed using Versalyse lysing solution (Beckman Coulter, Inc., USA) according to the manufacturer’s recommendations. Surface and intracellular markers were analysed by flow cytometry using a Navios® flow cytometer (Beckman Coulter, Inc., USA) according to the manufacturer’s recommendations. Flow set and Flow-check fluorosphere (Beckman Coulter, Inc., USA) were used to calibrate our cytometer on days of experiment. Fluorescence minus one (FMO) controls were used to verify the absence of spillover after applying the compensation matrix and as gating controls. Additionally, isotype controls were used for Sestrin-2, Dap-12, Ki-67, NKG2A and CD95.
Characterization of CD8+T cell differentiation and senescence by flow cytometry
As highly differentiated memory T cells accumulate with ageing, we studied T cell differentiation. We used CD45RA, CD27 and CD28 cell surface markers to define: naive N (CD45RA
+CD27
+), central memory CM (CD27
+CD45RA
−), effector memory EM (CD45RA
−CD27
−) and EMRA (CD45RA
+CD27
−) T cells [
12]. We used CD95 to discriminate stem-like memory T cell TSCM from naive T cells [
26]. Thus, TSCM were defined as CD45RA
− CD27
+ CD28
+ CD95
+ [
13,
26] (Figure S
2). As initially described by Lugli et al., TSCM CD95 expression levels were lower compared with the CM CD95 expression levels on CD8 (
p < 0.0001 Figure S
2) [
26]. As NK receptors and Sestrins accumulate in differentiated T cells [
12], we studied CD56, CD57, NKG2A, KLRG1 and Sestrin-2 expression. We characterized the CD8 senescence profile of each patient according to the percentage of EMRA and CD27
−CD28
− subsets, percentage of CD56, CD57, KLRG1 and NKG2A expression and Sestrin-2 expression level in mean fluorescence intensity (MFI) normalized on control isotype (nMFI). The gating strategy is shown in Figure S
2.
Cytokine assays
Plasma samples were put on ice and collected as soon as possible after arrival in the lab. Before protein analysis, plasma and nasal samples were treated in a P3 laboratory for viral decontamination using a protocol previously described for SARS-CoV, which we validated for SARS-CoV-2 [
14]. Briefly, samples were treated with 1% TRITON X100 (vol/vol) and 0.3% tri-N-butyl phosphate (vol/vol) for 2 h at room temperature. Tri-N-butyl phosphate was removed before cytokine analysis by passing the treated samples though C18 columns. IL-6, TNF and IL-10 were measured with a commercial triplex assay (Quanterix) on a Simoa HD-1 analyzer (Quanterix). An additional 38 cytokines and chemokines were measured in plasma supernatants with a commercial Luminex multi-analyte assay (Biotechne, R&D systems). Data were acquired on a Bio-Plex 200 System (Bio-Rad) and analyzed with Bio-Plex Manager v5 (Bio-Rad).
Statistical methods
The significance of associations between the severity groups and biological or clinical variables was assessed using either t-test or ANOVA for continuous variables, and likelihood-ratio test of logistic regression coefficients for categorical ones. For each association, statistical tests were computed both in univariate and bivariate settings, with the latter incorporating age into a multivariable model. This inclusion of age was essential due to the notable age difference between groups and its potential association with various biological and clinical markers under study. Correlations between the different biological markers was asserted using Pearson correlation coefficient (presented as correlation matrices). The threshold for statistical significance was set at 5%.
Furthermore, Principal Component Analysis (PCA) followed by k-means clustering were employed to investigate (i) the association between severity groups and CD8+ T cell senescence, and (ii) the association between severity groups and cytokines. A PCA was carried-out in a first step, and the individual patients were then grouped together using k-means based on the first two-dimensions of the PCA.
A Manhattan plot summarizes the statistical associations for the variables pertaining to either CD8+ T cell immunosenesence, or clinical group. The 5% statistical significance threshold was corrected on this figure to account for the multiple testing using the False Discovery Rate method.
External dataset of independent younger COVID-19+cohort
For the cytokine dosage, we compared our data set (median age of the current cohort: 86 years old) with an independent younger cohort (median age of 55 years old) already described and published using the same method of dosage by the same lab [
14]. The dosage were performed by the same operator. We chose the closest timepoint from the symptom onset and the closest degree of COVID-19 severity for the comparisons (group 2 and 3).
Discussion
Older age is associated with a significantly higher risk of severe COVID-19 leading to increased mortality. However, immune parameters associated with COVID-19 in older patients are poorly described.
To address this, we performed an observational study in which we characterized CD8+ T cell phenotypes and cytokines associated with senescence, during COVID-19, in a cohort of older patients. We compared a COVID-19 positive cohort with an uninfected control cohort. We then investigated the factors associated with COVID-19 severity between groups of increasing severity from group 1 to 3. We found that a high level of the senescence-associated cytokines GM-CSF, CXCL10, VEGF and CCL2, NLR, and a high CD27−CD28− percentage among CD8+ T cells, were positively associated with increased COVID-19 severity. CD8+ stem memory T cells (TSCM) and CD27+CD28+ CD8+ T cell were negatively associated with severity.
We found that gender, BMI and serotonin reuptake inhibitor treatment were positively associated with more severe COVID-19 disease as previously described [
31]. In our cohort, comorbidities such as cardiovascular disease, chronic lung infection and diabetes were not associated with COVID-19 severity. Accordingly, the Charlson Comorbidity Index was not a risk factor for severe forms of COVID-19. Surprisingly, the COVID-19
+ critical severity group (group 3) was younger than the other groups (groups 1 and 2). Thus, in very old people, the association between age and severity may no longer be linear. These observations are in line with the few existing reports on older patients cohort or sub-group analysis [
2,
32].
We found that our COVID-19
+ cohort had higher neutrophil counts, NLR, monocyte counts, CRP levels and a lower CD8 count compared to the control cohort, which is consistent with existing data [
13,
19]. In our study, these parameters positively correlated with COVID-19 severity. The peripheral CD8
+ T cell lymphopenia may reflect CD8
+ T cell migration to the lungs. The lymphopenia in CM CD8
+ cells appears to be particularly pronounced, as the difference was found in both count and percentage, suggesting that this was not solely a reflection of the global lymphopenia, in contrast to what was observed for EM and naïve cells.
The CD8
+ T cell senescence parameters we examined (CD27
−CD28
−, EMRA, CD57, KLRG1, CD56% among CD8 and Sestrin-2 expression by CD8) were positively correlated together and defined a CD8 senescence profile. The NLR, known to be increased with age and associated with bad prognosis during infection [
27], was positively correlated with this CD8 senescence profile whereas the CD4/CD8 ratio was inversely correlated. This is in concordance with previous findings showing that, in older people, CD4/CD8 ratio inferior to 1 is an immune risk phenotype for lung nosocomial infection [
33]. Interestingly, we found that the CD8 senescence profile was inversely associated with early differentiated CD27
+CD28
+ and TSCM-like CD8
+ T cells.
CD27
−CD28
−, CD57 and EMRA frequencies among CD8
+ T cells were increased in the higher severity groups. This difference persisted for CD27
−CD28
− after correction for multiple testing. To note, the comparison of absolute values did not reveal any difference between the groups, suggesting an over-occupation of the CD8 pool by senescent T cells in the group 3 that is particularly lymphopenic. Even if the statistical comparison could not be performed, the EMRA frequency among CD8
+ T cell was increased in our older COVID 19
+ cohort compared to a younger COVID 19
+ cohort described in the literature (30.3% ± 23.4 in the geriatrics
versus 20.0% ± 18.4 in younger adults, from Divij Mathew dataset) [
34]. We observed increased expression of Sestrin-2 in the COVID-19
+ cohort compared to the control cohort. This elevation could be associated with acute infection and cellular stress, which is known to upregulate Sestrin-2 expression [
12]. Additionally, Sestrin-2 expression was found to be higher in CD8
+ T cells from patients in severity group-2. However, given that the group 2 patients were older than those in the other groups, the lack of difference after age adjustment suggests that this may be attributed to increased expression with age, even within a cohort of patients over 70 years old.
Interestingly, the percentage and number of TSCM-like CD8
+ T cells, which are early differentiated cells with a great self-renewal and differentiation potential after antigen exposure [
35], was highly associated with a lower severity, even after FDR correction for multiple testing. The fact that the difference was strongly observed in percentage and in absolute value, despite comparable CD8 lymphocyte number between the groups 1 and 2, suggests that it is not a consequence of global lymphopenia. Furthermore, the numbers of EM, EMRA and CM were comparable between the severity groups, suggesting that they occupy a consistent position within the CD8 pool. Also, TSCM-like cells were more abundant in COVID-19
+ group 1 (but not in groups 2 and 3) compared to the control group (data not shown). This suggests that CD8
+ TSCM-like cells, even in low numbers, could proliferate early during in the course of the infection to provide protection against severe manifestations.
CXCL10, GM-CSF, VEGF, IL-1β and CCL2 were clearly positively associated with a higher COVID-19 severity in elderly patients even after adjustment for age and correction for multiple testing. In COVID-19
+ patients, concentrations were all higher in our cohort compared to a younger cohort [
14]. GM-CSF, VEGF, IL-1β and CCL2 were previously associated with ageing [
16,
17,
29,
30] and our study confirms their association with senescence. In a clustering model (Principal Component Analysis), the severity group 3 appeared mostly grouped within one single cluster based on cytokine concentrations, confirming their importance in the development of severe infection. Accordingly, some reports suggest that the inflammaging could contribute to the cytokine storm [
36], which is involved in the disease severity.
CXCL10 had previously been associated with frailty in older population [
37,
38] We found in our study that CXCL10 could be an interesting marker of COVID-19 severity and also seems to be associated with ageing. Indeed, CXCL10 levels were higher in our older cohort compared to a younger one in which it was not associated with severity [
14]. This suggests that it could be a marker of inflammaging with prognostic value in older patients. Interestingly, it negatively correlated with CD8
+ T cell specific response in acute cases [
1].
GM-CSF could contribute to the tissue infiltration by inflammatory myeloid cells described in moderate-to-severe forms of COVID19 [
39]. We found in our study that GM-CSF concentration was higher compared to the younger cohort and positively associated with COVID-19 severity. Smith et al. study [
14] reported that GM-CSF was not associated with severity in a younger cohort. Therefore, our results suggest a specific prognostic value in older patients. Accordingly, in an adult cohort, Thwaistes et al. found that GM-CSF could be a differential mediator of COVID-19 severity compared to flu severity particularly in patients over 70 years old [
40]. Interestingly, a recent study examined the role of GM-CSF antagonism for COVID-19 care. Although no role was found in the whole population in a phase II clinical trial, the antagonism led to a clinical improvement (mortality and time to recovery from respiratory failure) in the sub-group of patients over 70 years old [
41].
Single-cell analysis of bronchoalveolar lavages from critical COVID-19 patients revealed the abundance of inflammatory IL-1β-secreting myeloid cells, which could be involved in lung damages [
42]. Indeed, NLRP3 inflammasome activation, which induces IL-1β, has been described in neutrophils from severe COVID-19 patients and this pathway seems to be particularly involved in older patients [
27]. In a younger adults cohort (aged 53 to 72 years old), Del Valle et al. found that IL-1β had a low predictive value of COVID-19 mortality [
43]. This suggests that IL-1β association with severity is age-specific.
During severe COVID-19, CCL2 secretion in the lung likely participates to unconventional monocyte lung infiltration linked with pathogenesis [
18]. This mechanism was previously observed in aged skin where CCL2 over-secretion by senescent fibroblast decreased memory T cell activation and proliferation [
44].
Our study has some limitations. First, we studied a real-life cohort and this study was exploratory in nature, thus the development of a statistical model with adjustment for variables of interest such as age, BMI, sex ratio, comorbidities or where patients live (in patient versus out patient) was not possible due to the low number of subjects in the cohort. Second, some socio-demographic criteria, such as ethnic origin, would have been a relevant criteria to include in the statistical analysis, but due to a lack of official data collection this could not be included in the model. However, we assume that the likely impact of this demographic parameter in our study would be minor. Finally, the size of the control cohort is not comparable to the size of the cohort of interest but, we assume that comparison with a larger cohort would have confirmed our key findings. However, our study has strengths to highlight. In the literature, cohorts of older people with significant size are rare. To our knowledge, no study of immunological ageing in COVID-19+ older people cohort is available. Our study has the advantage of being a prospective single-centre study, with a homogeneous inclusion and sampling time point, which limits certain biases. In our statistical approach, all results were adjusted for age, and strongest results were confirmed by correcting for multiple testing.(FDR). We believe that the findings of our study, highlighting the association with the severity of COVID-19 and the high levels of GMCSF and CXCL10 and low levels of CD8+ TSCM, require further investigation and confirmation on a larger scale.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.