Features of studied group
All the 59 donors recruited were more than 60 years old, with a mean age of 72.15 years (SD: 12 years), of whom 39 were female (66%) and 20 were male (34%). Details of underlying diseases (64.4% of the patients) and symptoms of patients that had mild symptomatic SARS CoV-2 infection (61%) are summarised in Table 1. Laboratory characteristics, such as the level of the leukocyte subsets, D-Dimer and NT-proBNP, were also measured in most patients. Regarding influenza immunology status, 33 patients had received the influenza vaccine some months before the samples were collected, while 26 had not been vaccinated in the most recent vaccination campaign. Forty-nine patients were CMV-seropositive and 10 were CMV-seronegative (Table 1).
Table 1. Participant characteristics
Variables
|
Patients (n=59)
|
AGE ± SD (years)
|
72.2 ± 12
|
FEMALES
|
39 (66)
|
Underlying disease
|
38/56 (64.4)
|
Hypertension
|
25 (65.8)
|
Diabetes mellitus
|
13 (34.2)
|
Asthma
|
7 (18.4)
|
Heart disease
|
14 (36.8)
|
Renal disease
|
4 (10.5)
|
Smoker
|
3 (7.9)
|
Drinker
|
4 (10.5)
|
ALTERED D-DIMER (>500 ng/mL)
|
18/49 (36.7)
|
ALTERED NT-proBNP (>300 pg/mL)
|
9/58 (15.5)
|
INFLUENZA-VACCINATED
|
33 (55.9)
|
CMV-SEROPOSITIVE
|
49 (83)
|
Symptomatic of COVID-19 disease
|
36/56 (61)
|
Fever
|
24 (66.7)
|
Cough
|
24 (66.7)
|
Dyspnoea
|
9 (25)
|
Chest pain
|
5 (13.9)
|
Ageusia
|
9 (25)
|
Anosmia
|
11 (30.6)
|
Mean days with symptoms ± SD
|
3.3 ± 2.7
|
Numbers in brackets represent percentages.
Haemogram and biochemical characteristic recovery after SARS-CoV-2 infection
Since recruited patients had recently been infected with SARS-CoV-2, we wanted to determine whether some of the characteristics that are known to be altered during this infection had already returned to normal. As is well known, SARS-CoV-2 infection produces lymphopenia, as was seen in five of the patients for whom these data were available during the infection process (Supplementary Figure 1). We wanted to establish whether patients showed normal levels of leukocytes when the study began. It was possible to obtain the basal haemogram information from 2 months to 4 years before the SARS-CoV-2 infection in 52 of the 59 patients. When the levels of the different leukocyte populations were compared before and after the infection, no statistically significant differences, or statistically significant but not biologically significant differences, were seen in the total leukocytes (Wilcoxon test, p<0.01), lymphocytes (paired-samples t test, p<0.01), monocytes (Wilcoxon test, p<0.01) or neutrophils (Wilcoxon test, p>0.05) (Figure 1A). SARS-CoV-2 infection appeared not to affect leukocyte populations substantively once patients had completely recovered.
We now consider the biochemical characteristics that are known to be altered in SARS-CoV-2-infected patients and are considered indicators of a more severe COVID-19 disease. We measured D-Dimer and NT-proBNP levels in serum samples of 49 and 58 of the 59 patients, respectively. Eighteen and nine individuals had an elevated level of D-Dimer (>500ng/mL) and NT-proBNP (>300 pg/mL), respectively, when the study began (Table 1). This finding was not associated with the patients’ gender or COVID-19 symptomatology. Elevated levels of both factors were more likely to occur in older patients, since individuals with altered characteristics were significantly older than those whose parameters were within the normal range (Student’s t test, p<0.05, for both) (Figure 1B). With respect to individuals’ existing underlying pathologies a higher prevalence of altered NT-proBNP was observed in patients who were suffering from an underlying cardiopathy (Pearson’s chi-squared test, p<0.05) (Figure 1C).
Cellular and humoral responses to the three virus models
Cellular and humoral memory responses to the three viral antigens were determined. Cellular responses were measured by IFN-g ELISpot and granzyme B ELISpot. Humoral responses were measured as the level of specific antibodies against the different viruses in sera. It was not possible to obtain all results for all patients, especially in the case of granzyme B ELISpot since not enough cells could be isolated from some patients. Complete results of the three measurements were achieved in 74.6% of patients for CMV, 84.7% for influenza and 83% for SARS-CoV-2 of the 59 patients enrolled in the study.
No responses were obtained for some measurements, especially cellular memory measured by granzyme B ELIspot (Figure 2A). No cellular or humoral response was detected in 17% of CMV (corresponding to seronegative CMV patients) and 3% of SARS-CoV-2 cases. However, most patients showed both cellular (with at least one of the two measurements obtained) and humoral responses to CMV, influenza and SARS-CoV-2 viruses (83%, 83% and 85%, respectively) (Figure 2B).
Isolated cellular responses with no accompanying humoral response were seen for influenza in 17% of the patients (Figure 2B). Conversely, an isolated humoral response with no detectable cellular response was observed for SARS-CoV-2 in 12% of the patients. Of these, 14% showed anti-N antibodies and the remaining 86% had both anti-N and anti-S antibodies. Likewise, 96% of the patients with humoral and cellular responses had both anti-N and anti-S antibodies, while the other 4% had isolated anti-N (2%) or anti-S (2%) antibodies (Figure 2B).
In all cases, positive cellular responses occurred mainly at the expense of IFN-g-ELISpot measurement or both IFN-g and granzyme B ELIspot. Just in the case of cellular responses to SARS-CoV-2, 5% of the patients showed granzyme B-ELISpot-positive results with negative IFN-g-ELISpot. (Figure 2B).
a. Anti-CMV responses
Eighty three percent of the patients were CMV-seropositive. Distributions of the different response measurements against CMV are represented in Figure 3A. As mentioned above, humoral and cellular responses to CMV matched perfectly, whereby all seronegative patients were also negative for cellular responses, while the seropositive group of patients showed cellular responses. There was a positive correlation in the seropositive CMV patients between the two cellular responses measured as IFN-g and granzyme B-producer-specific T lymphocytes (Spearman test; r=0.6, p<0.01). However, no significant correlation was observed between the anti-CMV cellular and humoral responses (Supplementary Figure 2A).
b. Anti-influenza responses
The influenza-vaccinated status of all patients was known: 55.9% of them had been vaccinated in the most recent vaccination campaign. Nevertheless, it can be assumed that all the patients, vaccinated and unvaccinated, had been exposed to this seasonal virus during their lives. In fact, the cellular memory response to influenza (IFN-g and/or granzyme-B-producing T lymphocytes) was detected in all the patients. Regarding the humoral response, 94% of vaccinated and 69% of unvaccinated patients had antibodies against influenza at a level above our detection threshold. The great majority (80%) of the 17% patients who showed cellular responses without antibodies to the influenza vaccine were unvaccinated (Figure 2B). As expected, the group of vaccinated individuals had a significantly higher specific cellular memory response (Mann–Whitney test, p<0.05 for IFN-g and p<0.01 for granzyme B producer T lymphocytes per 106 T lymphocytes) and anti-influenza antibody titre (Mann–Whitney test, p<0.01) compared with the unvaccinated group (Figure 3B).
The two different measurements of cellular response (IFN-g and granzyme B production) and the values for IFN-g-producer T lymphocytes and humoral response against influenza virus were both positively correlated (Spearman test; r=0.6, p<0.01 and r= 0.3, p<0.05, respectively) (Supplementary Figure 2B).
c. Anti-SARS-CoV-2 responses
All the patients had been infected with SARS-CoV-2 for the first time and an adequate response to the virus was expected as they were asymptomatic or suffered only very mild symptoms. As described above, cellular or humoral responses to SARS-CoV-2 virus were observed in 97% of the cases (85% showed both types of response, while 12% presented only a humoral response). No response was detected in two patients studied by any of the methods used (Figure 2B). Distributions of the different response measurements against SARS-CoV-2 are represented in figure 3C. As described above, some biochemical characteristics related to the severity of the SARS-CoV-2 infection, D-Dimer and NT-proBNP levels were measured. No significant differences were found in the responses to SARS-CoV-2 between patients with and without altered parameters (Supplementary Figure 3).
Positive correlations were noted between cellular responses, measured as specific IFN-g-producer T lymphocytes, and humoral responses to SARS-CoV-2. The correlation was higher for anti-N antibodies (Spearman test, r=0.5, p<0.01) than for anti-S antibodies (Spearman test; r=0.3, p<0.05). As expected, the two humoral response measurements were significantly positively correlated (Spearman test, r=0.5, p<0.01). However, no significant correlation was observed between the two cellular response measurements (Supplementary Figure 2C).
Relation between viral antigen model responses
Comparing the cellular responses to the three viruses showed that the specific anti-CMV cellular response was stronger, in the CMV-seropositive group of patients, than the cellular response to the influenza virus, and both were stronger than the response to SARS-CoV-2, particularly when the cellular response was measured with IFN-g ELIspot, since the differences between all the three viral antigens responses were statistically significant (Bonferroni test, p<0.01). When measured by Granzyme B ELIspot, there were significant differences between CMV and SARS-CoV-2 (Bonferroni test, p<0.05) and between influenza and SARS-CoV-2 (Bonferroni test, p<0.01) but not between CMV and influenza in either of the unvaccinated and vaccinated patient groups (Figure 4). Humoral responses could not be compared because they were measured in different ways.
We found a significant positive correlation between cellular responses to influenza and SARS-CoV-2 for IFN-g-producer T lymphocytes (Spearman test; r=0.4, p<0.01) and granzyme B-producer T lymphocytes (Spearman test; r=0.4, p<0.05) although these were not apparent at the humoral level (Figure 5A). On the other hand, there were no significant correlations of any of the measurements of the responses between CMV and SARS-CoV-2 or between CMV and influenza. However, comparing influenza and SARS-CoV-2 specific responses in seropositive-CMV and seronegative-CMV patients showed them to be consistently lower in CMV-seropositive than in CMV seronegative patients, in both cases (Figure 5B and 5C). This comparison was not significant probably because there were too few CMV-seronegative individuals in the sample for a difference of that magnitude to be significant.
Characterization of the responses to the three antigen models in relation to the T lymphocyte phenotype
Immunosenescence may be related to the intensity of responses to the different virus infections. The immunophenotype of T lymphocytes by their degree of maturation (naïve/memory) and functional differentiation of CD4+ T lymphocytes (Th1/Th2/Th17) and their correlations with the viral responses were analysed.
a. CMV
Agreeing with what is already well known and stablished, differences in the distribution of the T lymphocyte subpopulations were observed between CMV-seropositive and CMV-seronegative individuals (Supplementary Figure 4A). CMV-seronegative patients had a significantly lower proportion of CD4+EM3 than did CMV-seropositive individuals (median: 0.03% vs 6.8%; Mann–Whitney test, p<0.01). However, the median proportions of CD4+CM and CD8+N were significantly higher in CMV-seronegative than in CMV-seropositive patients (30.4% vs 22.6%; Mann–Whitney test, p<0.05, and 16.2% vs 7.1%; Mann Whitney test, p<0.01, respectively) (Supplementary Figure 4A). With respect to the distribution of the functional subpopulations of CD4+ memory T lymphocytes, there was a significantly lower mean percentage of Th1 in CMV-seronegative than in CMV-seropositive patients (29.2% vs 39.2%; Student’s t test, p<0.01) (Supplementary Figure 4B).
Considering the CMV-seropositive group of patients, the cellular response (IFN-g and/or granzyme B-producing T lymphocytes) was negatively correlated with CD4+N, CD4+EM1 and CD8+N while there was a significant positive correlation with CD4+EM4. Regarding the humoral responses to CMV, there were negative correlations with CD4+CM, CD4+EM1, CD8+N CD8+CM and CD8+EM1. However, a significant positive correlation was seen with the CD4+EM4, CD4+EM3, CD4+EMRA, CD8+EM3 and CD8+EMRA subsets (Figure 6A). In the case of the functional differentiation of memory CD4+ T lymphocytes, the humoral response to CMV was positively correlated with Th1 type and negatively correlated with Th2 and Th17 (Figure 6B).
b. Influenza
It is expected that recently vaccinated patients are a more homogeneous group with respect to the immune responses generated against influenza seasonal pathogen. In vaccinated patients, the cellular memory responses to the influenza virus (IFN-g and/or granzyme B-producing T lymphocytes) were negatively correlated with CD4+EM3, CD8+EM3 and CD8+EMRA cells, and positively correlated with CD4+EM4 and CD4+EM1 cells. No significant correlations were noted between the level of antibodies and the distribution of any of the subpopulations (Figure 6A). Regarding the functional differentiation of memory CD4+ T lymphocytes, the cellular response measured as IFN-g-producer-specific T lymphocytes was negatively correlated with the Th2 subpopulation but positively correlated with the Th1.17 subpopulation (Figure 6B).
c. SARS-CoV-2
T lymphocyte cellular responses to SARS-CoV-2 (IFN-g and/or granzyme B-producing T lymphocytes) were correlated negatively with CD4+EM4 and positively with CD4+EMRA. Regarding CD8+ T lymphocyte responses, there was a negative correlation between cellular responses and the CD8+EM3 and CD8+EMRA subsets. Analysis of the humoral responses revealed that the level of anti-S antibodies was negatively correlated with the CD4+N, CD4+CM, CD4+EM1 and CD8+N subsets, whereas anti-N antibodies were not significantly correlated with the distribution of any of the subpopulations (Figure 6A). In terms of the functional differentiation of memory CD4+ T lymphocytes, the humoral response, measured as anti-S antibodies, was negatively correlated with the Th2 and Th17 subpopulations. The negative correlation with Th17 lymphocytes was also seen at the cellular level (Figure 6B).