IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 pp65-Specific T-Cell Response Is Independently Associated With Time to Death in Elderly Humans

Abstract

Persistent cytomegalovirus (CMV) infection has been suggested to be a major driving force in the immune deterioration and an underlying source of age-related diseases in the elderly. CMV antibody titers are associated with lower responses to vaccination, cardiovascular diseases, frailty, and mortality. CMV infection is also associated with shorter T-cell telomeres and replicative senescence. Although an age-related deregulation of CMV-specific T-cell responses could be an underlying cause of the relationship between CMV and immune defects, strong and polyfunctional responses are observed in elderly individuals, casting uncertainty on their direct role in age-related immune frailty. In this study, we longitudinally followed a cohort of healthy donors aged over 50 years, assessing their mortality rates and time to death during a 2-year period. Specific T-cell responses to the immunodominant antigen pp65 (IFNγ, TNFα, IL2, MIP1α, CD107a, and perforin production) were analyzed at the beginning of the 2-year observation period. A cytotoxic CD8 pp65-specific T-cell response, without cytokine or chemokine coexpression, was independently associated with all-cause mortality in these elderly individuals. This pp65-specific CD8 T-cell response could be a useful tool to identify individuals with depressed immune function and a higher risk of death.

Cytomegalovirus (CMV) is a herpesvirus with a worldwide distribution that establishes persistent infection in more than 80% of the population. CMV can be transmitted both vertically and horizontally, but immunocompetent hosts carry a life-long asymptomatic infection (1). Despite the lack of apparent disease, elderly individuals commit large percentages of the CD8 T-cell pool to the CMV infection (2). The immune risk profile (3), a term coined to encompass the results from the OCTO and NONA studies (4,5), linked CMV infection with immune deterioration in elderly individuals. The increase in CD8 T-cell number—and the consequent decrease in the CD4:CD8 T-cell ratio—is not the only misbalance in CD8 T-cell numbers impacting the elderly frailty. The BELFRAIL study showed that a CD4:CD8 T-cell ratio over 5 was also associated with impaired physical functioning in CMV-infected elderly individuals (6). As a result, it has been argued that persistent viral infections (mainly CMV) are a major cause of the age-related immune deterioration (7). Clinical outcomes of this immune decline, including increased incidence and prevalence of infectious diseases and cancer (8), have raised interest in determining the role of CMV infection in triggering age-related immune system defects.

Several studies have associated the presence and plasma levels of CMV antibodies with impaired responses to the influenza vaccine (9), cardiovascular diseases (10), geriatric frailty (11), and increased mortality (10,12). However, despite the association between CMV seropositivity and immune system frailty, studies focusing on CMV-specific T-cell responses are still controversial (13). At the cellular level, CMV infection has been related to T-cell telomere shortening (14) and T-cell replicative senescence (15). In addition, the large number of CD8 T cells specifically committed to CMV (16) displaces other T-cell subsets (17), favoring CD8 repertoire shrinkage (18). In murine models, CMV infection impairs T-cell responses (19) to nonrelated viral infections (20,21). Despite all these associations, neither higher antibody level nor CD8 CMV-specific T-cell expansion are associated with the onset of symptomatic CMV disease. In addition, in both human and nonhuman primate models, elderly subjects have well-maintained cytokine-based (IFNγ, TNFα, and/or IL2) CMV-specific T-cell responses (17) with a preserved degranulation capacity (22). Nonetheless, CMV-specific CD8 T cells show a uniquely mature phenotype (23), and CMV-specific CD4 T cells achieve effector capacities usually observed in CD8 cytolytic responses (24).

Despite the clinical importance of immunosenescence, the potential implication of CMV-specific T-cell responses in immune system exhaustion is poorly understood. Studies focusing on morbidity or mortality endpoints in elderly cohorts, rather than age, could be helpful in defining the role of CMV-specific T-cell alterations in age-related immune dysfunction. In this study, we examined the association between the CMV-specific T-cell response quality and all-cause 2-year mortality in healthy individuals aged over 50 years.

Methods

Ethics Statement

This study was reviewed and approved by the Ethics Committee of the Virgen del Rocio University Hospital/Institute of Biomedicine of Seville, Seville, Spain. All participants, or their caregivers when appropriate, provided written informed consent prior to enrollment in this study.

Subjects

Peripheral blood mononuclear cells (PBMCs) were obtained during the first quarter of 2008 from healthy volunteers who were residing in the community or in nursing homes in the Aljarafe health area (Seville, Spain). All participants were active, middle-class individuals. The following inclusion/exclusion criteria applied for enrollment of the subjects: We selected functional individuals aged 50 years or older, excluding any subject with diagnosis of dementia or with any of the following conditions in the 6 months prior to sampling: (i) clinical evidence of active infections, (ii) any hospital admission, (iii) antitumor therapy or (iv) treatment judged to affect their immune status, such as the administration of corticosteroids. This cohort, originally named the CARRERITAS cohort of healthy elderly, has been partially reported in a previous study (25). PBMCs and demographic data were obtained at baseline. PBMCs were cryopreserved in liquid nitrogen in fetal calf serum with 10% dimethyl sulfoxide until further use. After a median follow-up of 87 weeks (interquartile range = 82–103 weeks), all participants were contacted again to assess survival rates. Sixty-seven consecutive individuals who fulfilled the inclusion criteria (50.7–96.9 years old) and had been previously exposed to CMV (as measured by positive plasmatic IgG antibodies) were selected for this study according to sample availability.

Cell Stimulations and Flow Cytometry

Frozen PBMCs were thawed, stimulated, and stained for flow cytometry as previously reported (26). Briefly, cells were thawed, washed, and rested for 2h in DNase I (Roche Diagnostics, Indianapolis, IN)-containing R-10 complete medium. Two million PBMCs per mL, including 1 μg/mL of αCD28 and 1 μg/mL of αCD49d (BD Biosciences, San Jose CA), 10 μg/mL of brefeldin A (Sigma Chemical Co, St. Louis MO), and 0.7 μg/mL of monensin (BD Biosciences), were incubated for 6 hours in the absence or presence of pp65 peptide antigen. Directly conjugated monoclonal αCD107a was added at the beginning of incubation.

Monoclonal antibodies α-IL2-Cy55PerCP, α-CD3-Cy7APC, α-IFNγ–PE, α-TNFα-Cy7PE, α-CD14-PB, α-CD19-PB, and MIP1α-APC were from BD Biosciences; α-CD45RO TRPE and α-CD27-Cy5PE were form Beckman Coulter; α-CD4-Cy55PE was from Caltag; and α-PRF1-FITC clone B-D48 was from Cell Sciences. The monoclonal antibodies α-CD107a-Alexa680, α-CD8-QD655, and α-CD57-QD565 were conjugated in our laboratory according to standard protocols (http://drmr.com/abcon/index.html).

Stimulated PBMCs were washed and stained with a pre-titered quantity of surface antibodies. Cells were then washed and permeabilized using a Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturers’ instructions, intracellularly stained, washed, and fixed in phosphate-buffered saline containing 1% paraformaldehyde. Thirteen-color/15-parameter flow cytometry was performed on an LSRII (BD Immunocytometry Systems) equipped for the detection of 18 fluorescence parameters. A minimum of 3×10⁵ events was collected for each sample. Electronic compensation was conducted with antibody capture beads (BD Biosciences) stained separately with individual monoclonal antibodies. Analysis was performed using FlowJo version 9.2 (Tree Star) as previously described (26). Briefly, singlet viable CD3⁺CD8⁻CD4⁺ or CD3⁺CD8⁺CD4⁻ T cells were sequentially selected. pp65-specific T cells were identified by IFNγ, TNFα, IL2, MIP1α, CD107a, and PRF1 production. A representative example of the complete gating strategy is shown in Supplementary Figure 1. All analyses included background subtraction using a nonstimulated sample as negative control. Graphs were made using Pestle Version 1.6.2 and Spice Version 5.2 (both provided by M. Roederer, NIH, Bethesda, MD) (27).

Peptide Pools

Peptide pools consisting of 15-mer peptides that overlapped by 11 residues and covered the entire pp65 proteins were constructed from individual proteins with greater than 70% purity (JPT labs, Berlin, Germany). Each pool contained 400 µg of each peptide, and when added to incubations, each peptide was present at a concentration of 2 μg/mL.

Absolute Cell Counts

CD4 and CD8 absolute numbers were determined in fresh whole blood using the Epic XL-MCL flow cytometer (Beckman Coulter Inc., Brea, CA) according to the manufacturer’s instructions. CD4:CD8 T-cell ratios lower than 1 (28) were selected as a simplified surrogate marker of the immune risk profile.

C-Reactive Protein Quantification

Serum C-reactive protein (CRP) was quantified with the CRPLX C-Reactive Protein kit (Latex) for the Cobas C 711 automated analyzer (Roche Diagnostics, Mannheim, Germany) according to the manufacturers’ instructions. Normal CRP levels for adults with this kit are < 5mg/L.

sj/β-TREC Ratio Quantification

Thymic function was indirectly calculated in peripheral PBMCs DNA using the sj/β-T-cell receptor rearrangement excision circles (TREC) ratio previously described (29) with minor modifications. Briefly, the six DβJβ-TRECs from cluster one were amplified together in the same PCR tube, whereas the sj-TREC was amplified in a different PCR tube. Twenty-one amplification rounds were performed to guarantee an accurate quantification at the real-time PCR step. All amplicons (DβJβ- and sj-TRECs) were then amplified together in a second PCR using a LightCycler 480 system (Roche). Six microliters of a 1/10 dilution of the first-round PCR were amplified in a 20 μL final volume. Förster resonance energy transfer specific probes for the sj-TRECs (30) and the DβJβ-TRECs (31) were previously reported.

Statistical Analysis

Continuous variables are expressed as median (interquartile range) and categorical variables as percentages. Nonparametric linear regression analysis was done using the Spearman rank test. The Mann–Whitney U-test was used to analyze differences between unpaired groups. Differences between paired samples were determined by the Wilcoxon signed-rank test. p Values <.05 were considered significant. Bivariate and multivariate Cox proportional hazards regressions were performed to analyze the association between the immune frailty-related parameters and time to death. Multivariate regression was performed with all variables reaching p < .1 in the bivariate analysis, with the exception of age, which as the critical confounder was included in all multivariate models regardless of previous p values. A p value <.05 was again considered to indicate statistical significance. Hazard ratios and 95% confidence intervals were estimated. Relative operating characteristic curves were performed to determine values with increased risk of death. Values were selected to maximize specificity while maintaining sensitivity. Kaplan–Meier curves and log-rank test bivariate analysis were performed for all time-to-event analyses. Statistical analysis was executed using the Statistical Package for the Social Sciences software (SPSS 17.0; SPSS, Chicago, IL). Prism Version 5.0 (GraphPad Software, Inc.) was used to generate graphs.

Results

Advanced Age Is Associated With Higher Magnitude But Similar Polyfunctionality of pp65-Specific T-Cell Responses

To study age-related changes in CMV-specific T-cell responses, we analyzed 67 healthy individuals aged over 50 years (min = 50.7; max = 96.9). All subjects had been previously exposed to CMV, confirmed by the presence of anti-CMV IgG antibodies. The characteristics of the cohort are described in Supplementary Table 1. Individuals were divided in two age groups: one group under the median age of 81 years (median age = 69.3 years, interquartile range = 64.2–74.6, n = 34) and one group over 81 years old (median age = 88.2 years, interquartile range = 84.4–90.2, n = 33). A standard virus-specific response was defined by intracellular cytokine production of IFNγ and/or TNFα and/or IL2 in response to the pp65 peptide pool. As shown in Figure 1A, a higher frequency of the pp65-specific CD4 and CD8 T-cell responses was observed in the more elderly group.

Figure 1.

Magnitude of the pp65-specific T-cell response. (A) Pooled data showing the percentage of CD45RO⁺CD27⁺ memory CD4 and CD8 T cells expressing IFNγ and/or TNFα and/or IL2 in response to pp65 peptide pool stimulation from younger (age < median, n = 33) and older individuals (age ≥ median, n = 34). Median age = 81 y. (B) Polyfunctionality patterns of pp65-specific CD8 T cells producing one (1+), two (2+), three (3+), four (4+), five (5+), or six (6+) simultaneous functions (combination of IFNγ-, TNFα-, IL2-, MIP1α-, CD107a-, and PRF1-expressing cells) among the different age groups.

We then analyzed cytokine (IFNγ, TNFα, IL2) and chemokine (MIP1α) production together with cytolytic marker (perforin 1 [PRF1]) production and CD107a surface mobilization—from now on called “functions”—in response to stimulation with the pp65 peptide pool. Strikingly, our results showed that T-cell responses were as polyfunctional in elderly individuals as they were in their younger counterparts (Figure 1B).

IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-Cell Responses Are Associated With Higher Differentiation Patterns and Higher All-Cause 2-Year Mortality

We assessed all-cause mortality in our cohort after a median follow-up period of 87 (76–102) weeks. Twenty-two of the 67 donors (32.8%) died during this period (Supplementary Table 1). pp65-specific responses at baseline were compared without consideration of age for survivors and nonsurvivors, to assess the potential relationship between pp65-specific responses and risk of death. As shown in Figure 2A, pp65-specific CD8 T-cell responses were less polyfunctional in individuals who died during this study. In the nonsurvivor group, responses consisting of five (gray), four (blue), and three functions (red) were significantly diminished, whereas the two-function responses (orange) were increased (p < .001). Although a trend toward less polyfunctional pp65-specific CD4 T-cell responses was also found in the nonsurvivor group, this difference did not reach statistical significance (p = .184, data not shown). The magnitude of the pp65-specific T-cell response for each of the 64 possible combinations of our six functions was then analyzed (Supplementary Figure 2). Among these responses, the cytokine- and chemokine-independent IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 pp65-specific T-cell response was significantly overrepresented in the nonsurvivor group in terms of frequency as well as absolute numbers (Figure 2B and C and Supplementary Figure 2). The level of the IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ pp65-specific CD8 T-cell response was associated with age when the nonsurvivor group was independently analyzed, but we found no association for the complete cohort (Supplementary Figure 3A). In addition, the percentage of the cytokine- and chemokine-independent IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell response was not associated with the cytokine-including standard response that is usually analyzed (Supplementary Figure 3B). In addition, this subset was increased in the CD4 T-cell responses of nonsurvivors, but this difference did not reach statistical significance (p = .075, data not shown).

Figure 2.

Nonsurvivors show deregulated CD8⁺ pp65-specific T-cell responses. (A) Polyfunctionality patterns of pp65-specific CD8 T cells producing one (1+), two (2+), three (3+), four (4+), five (5+), or six (6+) simultaneous functions (combination of IFNγ-, TNFα-, IL2-, MIP1α-, CD107a-, and PRF1-expressing cells) of the survivor and nonsurvivor groups. (B) Pooled data showing the frequencies and absolute numbers of the IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell response in the survivor and nonsurvivor groups. (C) Pie charts show the relative frequencies of this IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ response (purple) among the different groups.

The age-related accumulation of CMV-specific CD8 T-cell clones is associated with a higher differentiation status of the immune system, usually defined by the loss of costimulatory molecules (CD27 and/or CD28) and the concomitant increase of terminally differentiated CD57⁺ effector T cells. Increased levels of CD57-expressing CD8 T cells increased with age (Supplementary Figure 3C). In our cohort, when the nonsurvivor group was analyzed, the expression of an IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ functional phenotype, without cytokine or chemokine coexpression, by pp65-specific CD8 T cells was strongly associated with expression of CD57 (Figure 3A). Thus, we analyzed the polyfunctionality of CD57⁺/CD57⁻pp65-specific CD8 T cells. In the survivor group, pp65-specific CD8 T-cell polyfunctionality was similar in CD57⁺ and CD57⁻ cells (Figure 3B). In nonsurvivors, pp65-specific CD8 T cells showed markedly diminished polyfunctionality compared with CD57⁻ cells. Furthermore, an IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ response, without cytokine or chemokine coexpression, was observed in the CD57-expressing CD8 T-cell subset (Figure 3C).

Figure 3.

The IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ response increase is associated with a more differentiated CD8 T-cell phenotype. (A) Relationship, in the nonsurvivor group, between percentage of the pp65-specific IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell response (negative for IFNγ, TNFα, IL2, and MIP1α) and CD57-expressing CD8 T cells. (B) Polyfunctionality patterns (possible combinations of IFNγ, TNFα, IL2, MIP1α, CD107a, and PRF1 production in response to cognate antigen) according to survival and CD57 expression. (C) Relative frequency of the IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ response (purple) according to CD57 expression.

IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 pp65-Specific T-Cell Accumulation Is Independently Associated With Time to Death in Elderly Individuals

To better assess the association between IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ accumulation and risk of death, different parameters previously correlated with all-cause mortality in elderly individuals were selected for further study. Gender, thymic function (25), CRP (10,25), and a CD4:CD8 T-cell ratio of <1 (28) were analyzed for all participants at baseline. The relationship between these parameters (including the accumulation of the IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺CD8 pp65-specific T-cell response) and time to death was analyzed using Cox regression analysis. Bivariate analysis showed that all analyzed parameters, except for gender, were associated with time to death (Table 1, p < .1). However, when multivariate analysis was performed, only the frequency of the CD107a⁺PRF1⁺ CD8 pp65-specific T-cell response was independently associated with time to death (adjusted p < .001). The relationship between the IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell level and time to death is shown in Supplementary Figure 3D. Since 65 years old is the cut-off age usually selected for elderly studies and none of the individuals between 50 and 65 years old (9/67, 13.4%) died in the follow-up period, we repeated the Cox regression analysis in the subgroup of individuals over 65 years old. As Table 2 shows, overrepresentation of the CD107a⁺PRF1⁺ CD8 pp65-specific T-cell response was the only factor independently associated with time to death (adjusted p < .001) in this cohort. Next, we performed the same analysis including only individuals aged over 85 years, the cut-off age for which the immune risk profile was first described. Twenty-four individuals were included in this model. Ten of these individuals died during the follow-up period (10/24, 41.7%). When the analysis was restricted to the very old (Table 3), the accumulation of the IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell response remained a strong independent risk factor for death. Consistent with our previous reports (25), increased CRP was also an independent risk factor in the very old.

Table 1.

Cox Regression Analysis Between Survivorship-Related Parameters and Time to Death; All Individuals (age > 50 y, n = 67, 22 deaths) Were Included in This Model

Variable	Univariate			Bivariate
	p	HR	95% CI	p	HR	95% CI
Age (years)	.061	1.041	0.998–1.085	.250
Sex (women)	.806	1.112	0.475–2.604	.850
% CD107a+PRF1+ CD8 T cells*	<.001	1.202	1.123–1.286	<.001	1.231	1.114–1.360
sj/β-TREC ratio	.029	0.949	0.906–0.995	.314
CD4:CD8 > 1	.002	0.256	0.108–0.603	.962
CRP (mg/L)	.006	1.024	1.007–1.042	.070

Notes: Boldface values indicate variable independently associated with time to death. CRP = C-reactive protein; HR = hazard ratio; 95% CI = 95% confidence interval.

*CD107a⁺PRF1⁺ T cells are negative for IFNγ, TNFα, IL2, and MIP1α.

Table 2.

Cox Regression Analysis Between Survivorship-Related Parameters and Time to Death; Individuals Aged ≥ 65 y (n = 58, 22 deaths) Were Included in This Model

Variable	Univariate			Bivariate
	p	HR	95% CI	p	HR	95% CI
Age (years)	.535	1.015	0.967–1.066	.501
Sex (women)	.594	0.794	0.309–1.858	.841
%CD107a+PRF1+ CD8 T cells*	<.001	1.184	1.106–1.268	<.001	1.218	1.099–1.347
sj/β-TREC ratio	.047	.953	0.909–0.999	.353
CD4:CD8 > 1	.003	0.278	0.118–0.606	.970
CRP (mg/L)	.011	1.023	1.005–1.040	.068

Notes: Boldface values indicate variable independently associated with time to death. CRP = C-reactive protein; HR = hazard ratio; 95% CI = 95% confidence interval.

*CD107a⁺PRF1⁺ T cells are negative for IFNγ, TNFα, IL2, and MIP1α.

Table 3.

Cox Regression Analysis Between Survivorship-Related Parameters and Time to Death; Individual aged ≥ 85 (n = 24, 10 deaths) Were Included in This Model

Variable	Univariate			Bivariate
	p	HR	95% CI	p	HR	95% CI
Age (years)	.754	0.966	0.778–1.200	0.392
Sex (women)	.879	0.900	0.232–3.497
% CD107a+PRF1+ CD8 T cells*	.013	1.214	1.042–1.415	.006	1.381	1.096–1.740
sj/β-TREC ratio	.382	0.960	0.875–1.052
CD4:CD8 > 1	.084	0.309	0.081–1.170	.431
CRP (mg/L)	.011	1.019	1.004–1.034	.007	1.024	1.007–1.042

Notes: Boldface values indicate variable independently associated with time to death. CRP = C-reactive protein; HR = hazard ratio; 95% CI = 95% confidence interval.

*CD107a⁺PRF1⁺ T cells are negative for IFNγ, TNFα, IL2, and MIP1α.

To further confirm this result, we performed a Kaplan–Meier analysis for all times to events. First, individuals were categorized by the median level of the pp65-specific IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell response (0.296%). Survival rates for the group under the median were 87.9% because 4 out of 33 individuals under the median died during the follow-up period (4/33, 12.1% mortality). On the other hand, survival significantly decreased to 47.1% in individuals over the median (18/34, 52.9% mortality; chi-square test p = .001). In addition, individuals with IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ levels over the median had a significantly lower survival time (log-rank p = .006, Figure 4A). A subsequent relative operating characteristic curve analysis showed that the percentage of IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ was able to discriminate between survivors and nonsurvivors (area = 0.789, p = .001). IFNγ⁻ TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell responses over 3% showed 66.7% sensitivity and 100% specificity in identifying nonsurvivors at 2-years’ follow-up (Figure 4B). Individuals with IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺MIP1b⁺ CD8 T-cell responses lower than 3% showed a survival rate of 84.9% (8/53, 15.1% mortality), whereas survival dropped to 0% for individuals with IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ T-cell responses over 3% (chi-square test p < .001). These data suggest that the frequency of pp65-specific IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T cells may be a clinically significant marker of increased short-term mortality.

Figure 4.

The IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺response increase is associated with time to death in elderly humans. Kaplan–Meier curves for all time-to-event analyses of the frequency of the pp65-specific CD107a⁺PRF1⁺ (negative for IFNγ, TNFα, IL2, and MIP1α) CD8 T-cell response. (A) The response magnitude was dichotomized using the median (median = 0.296%; IQR [0.149–1.147]) as the cut-off value. Specificity = 62.9%. Sensitivity = 20%. (B) Cut-off value for CD107a⁺PRF1⁺ = 3% of the total response. Specificity = 100%. Sensitivity = 66.7%. Bivariate analysis using a log-rank test was performed to assess significant differences among the curves. IQR = interquartile range.

Discussion

Here, we show that individuals who died within 2 years of entry into our study had a higher frequency of pp65-specific CD8 T cells with a cytokine-deficient IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ functional phenotype than did individuals who survived. This response was significantly associated with expression of CD57, a marker of CD8 T cells senescence. Survivors showed a polyfunctional and cytokine-based complete response. These data identify a surrogate marker of the immune system deregulation in late stages of life, providing a new tool to recognize individuals with a higher risk of death.

Consistent with previous reports in mice (17), rhesus macaques (22), and humans (32), we did not find differences in the quality of the pp65-specific T-cell response in younger (69.3 [64.2–74.5] years old) versus older individuals (88.2 [84.4–90.2] years old). Significantly, both the duration of CMV infection and the rate of aging are different for each individual. To overcome this limitation, we analyzed a longitudinal prospective cohort where we could identify individuals at higher risk of death and evaluate the difference in CMV-specific T-cell response between these two groups. It should be noted that causes of death are not known for this cohort, and potentially unrelated deaths, such as accidental deaths, were not excluded. However, in small cohorts of elderly individuals, where it is difficult to distinguish real accidental deaths from accidents caused by an underlying health problem (eg, strokes), the analysis of all-cause mortality rates gives both a stricter and broader perspective. Thus, when the function response of pp65-specific CD8 T cells from individuals who died during this study were compared with those who survived, we observed that those who died during the study showed a less polyfunctional functional phenotype than those who survived. In fact, the cytokine-deficient IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 T-cell response was markedly increased in the nonsurvivor group. The kinetics of the CMV-specific T-cell response show that while IFNγ-containing T-cell responses need several hours to be produced, a faster response that includes degranulation markers (CD107a) but does not involve IFNγ occurs within an hour of T-cell stimulation (33). This prompt response is produced by more differentiated CD8 T cells and contains higher levels of granzyme and perforin, suggesting higher cytolytic properties (33). In fact, the accumulation of these highly differentiated cytolytic CD8 T cells in elderly individuals has been proposed as a mediator of the immunopathogenesis of influenza infection (34), whereas CMV-specific responses including proinflammatory molecules (such as IFNγ or TNFα) are associated with longer survival in the very old (35). Notably, some pp65-specific CD8 T-cell responses that included only one of the studied functions increased in the survivor group (Supplementary Figure 2). These results suggest that specific combinations of cytokines or cytolytic markers could be more informative than the polyfunctionality index itself. Thus, although the analysis of T-cell polyfunctionality could be informative, the magnitude of the individual response should be always considered.

Our data suggest that CD57-expressing CD8 T cells are the main producers of the cytokine-deficient IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ response. This functional phenotype appears to become increasingly significant as individuals approach the end of life. Accumulation of pp65-specific T cells expressing a cytokine-deficient IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺phenotype is independently associated with time to death in this cohort of healthy elderly. These results were significant even when the model was adjusted for sex, age, thymic function, CD4:CD8 T-cell ratio, and CRP level, markers that have been previously reported as predictors of mortality in elderly individuals (10,25,28).

Collectively, our results suggest that CMV infection has an active role in age-related T-cell dysfunction observed immunosenescence, only responses to the pp65 CMV protein were analyzed in this study. Whether these alterations are observed for all CMV-specific epitopes should be clarified in the future. This response may not be limited to CMV antigens, but further studies are needed to determine whether this is a generalized phenomenon that occurs with other chronic pathogens in individuals at higher risk of death. The mechanisms determining why some individuals showed higher misbalances in the CMV-specific T-cell response should also be clarified. Some reports suggest that offspring of centenarians are genetically less susceptible to accumulate immunosenescence markers late in life (36), but other stressors such as household income or daily activities cannot be ruled out (37).

In conclusion, our results suggest that the accumulation of IFNγ⁻TNFα⁻IL2⁻MIP1α⁻CD107a⁺PRF1⁺ CD8 pp65-specific T-cell responses could be a useful tool to identify individuals with age-related immune dysfunction and a higher risk of death. Further studies are needed to differentiate whether this aberrant response is a consequence of life-long CMV immunologic pressure or an indirect marker of generalized immune system weakness. This immune dysfunction, or exhaustion, could detrimentally influence the prognosis of other prevalent infectious diseases in elderly individuals.

Supplementary Material

Supplementary material can be found at: http://biomedgerontology.oxfordjournals.org/

Funding

S.F.-M. and E.R.-M. received grants from the Fondo de Investigaciones Sanitarias (CD10/00382 and CP08/00172, respectively). R.S. de P.-B. is supported by a Proyecto de Excelencia (CTS-6313) and B.D.-M. by the RETIC program. This study has been supported by Redes Temáticas de Investigación en SIDA (ISCIII RETIC RD12/0017/0029 and RD12/0017/0037), Proyecto de Excelencia (CTS-6313), and Consejería de Salud (PI-0278). This work was funded in part by the Intramural Research Program of the Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health.