Cytomegalovirus antibody levels and mortality among hospitalised elderly patients

Abstract

Background

The impact of cytomegalovirus infection in elderly subjects remains unclear. This study examined the relationship between humoral immune response to cytomegalovirus (CMV) and all-cause mortality in a cohort of elderly hospitalised patients.

Methods

Data were obtained from a random sample of 715 patients (≥65 years old) admitted for any cause in a third level hospital. Serum IgG antibody against CMV was determined by enzyme-linked immunosorbent (ELISA) assay.

Results

A total of 480 deaths occurred in seropositive patients (n = 671) during a follow-up of 7.6 years (mean, 4.6); of which 112 patients died in-hospital or within 30 days after discharge (short-term mortality). For patients with CMV IgG antibody levels in the highest quartile compared with lower quartile, fully adjusted models showed that mortality was 1.40 times (95% CI 1.05–1.86) and 2.20 times (95% CI 1.15–4.21) higher, respectively. The exclusion of patients with cardiovascular disease (angina, myocardial infarction, heart failure, peripheral artery disease, or stroke) increases the risk of long-term (HR 2.22, 95% CI 1.36–3.62) and short-term mortality (OR 3.18, 95% CI 1.40–7.24).

Conclusions

Increased IgG antibody levels against CMV are associated with increased short and long-term mortality in elderly hospitalised patients, especially in patients without cardiovascular disease.

Key Messages

1. The outcome of elderly hospitalised patients in relation to CMV is unknown.

2. We demonstrate an association between increased anti-CMV IgG levels and mortality.

3. This association is greater in elderly patients without cardiovascular disease.

Keywords:

• cytomegalovirus
• CMV
• IgG antibodies
• mortality
• hospitalised elderly

Introduction

Human cytomegalovirus (CMV) is a ubiquitous DNA virus of the Herpesviridae family with the ability of remaining latent for a long time, reactivating and causing severe complications [1]. Primary CMV infection in most immunocompetent adults is asymptomatic or non-specifically symptomatic, leaving no clinically observable trace of the infection except seroconversion [1]. Its prevalence ranges from 40 to 100% in the general population, and varies with geographical regions, socioeconomic status, and age [2]. Evidence of past CMV infection is the presence of immunoglobulin G (IgG) antibodies for CMV in the peripheral blood (seropositivity) [3]. If CMV escapes immunological control and reactivates from latency, it can cause severe disease and increased mortality, as has been observed among severely ill [4], and immunocompromised individuals [5,6].

CMV has also been linked to the development of several chronic diseases and geriatric syndromes associated with adverse outcomes, such as atherosclerotic cardiovascular disease [7,8], functional impairment [9], cognitive decline [10–12], prolonged hospitaliation [4], frailty [13], and cancer [14]. CMV has also been implicated in immune impairment associated with age, also called immunosenescence [15,16]. As a persistent infection, CMV is continually captured, processed, and presented to T cells, leading to clonal expansion and contraction of the adaptive immune system. Over time, this process leads to clonal exhaustion whereby CMV specific T cells are present but anergic, leaving fewer naïve T cells to combat novel pathogens [16]. On the other hand, immunosenescence may contribute to the reactivation of CMV and increase susceptibility to others infections [15,16]. Previous studies have attempted to examine the association between CMV IgG antibody levels and mortality in elderly people. These studies have been limited to highly selected groups including older adults with stable cardiovascular disease [17], community-dwelling older women [13], and Latinos over the age of 60 with low comorbidity [18]. Community-dwelling subjects and long-term effect of CMV infection have been a consistent focus of these studies. The short-term effect of CMV infection in acutely ill or hospitalised elderly has never been investigated.

Given the high prevalence of CMV infection worldwide and its potential implications on clinical outcomes, we conducted a cross-sectional and prospective cohort study to determine the effect of persistent CMV infection on a cohort of hospitalised elderly patients. Since CMV antibody, rather than viral DNA, is detectable in the serum of immunocompetent elderly persons and may potentially correlate with cumulative viral burden [19,20], we hypothesised that increased CMV antibody levels would be associated with higher rates of short and long-term mortality, even after accounting for conventional risk factors and comorbidity in many of these patients.

Material and methods

Study design and population

This study involved 715 randomly selected hospitalised elderly patients (aged 65 or older) who participated in previous complementary prospective observational studies started in 2011. Details on the study methods and sampling design of these studies have been published elsewhere [21,22]. The research protocols were approved by the Hospital Universitario de Burgos (Spain) institutional review board. Informed consent was obtained from all participants.

This study consisted of a cross-sectional component, in which we examined the relation between serum CMV immunoglobulin G (IgG) antibody levels and the risk of short-term mortality (in-hospital or within 30 days after discharge), and a longitudinal component, in which we examined the relation between CMV IgG levels and the risk of 7.6-year all-cause mortality. Of the 715 recruited patients none were excluded because of lack of serum specimens, or because baseline covariates were missing.

Covariates at baseline

Covariates were screened as confounders based on hypothesised relations and were included in models if they were associated with the exposure and associated with the outcome among the unexposed defined as the lowest quartile of anti-CMV IgG [23]. Age and gender, were assessed as covariates because they are associated with death and CMV serostatus or age at infection [2]. Smoking status, hypertension and hypercholesteronemia were assessed as covariates because they are associated with death [24] and CMV infection. It is thought that CMV may be related to these risk factors through socioeconomic status. Low socioeconomic status may influence CMV IgG antibody levels by increasing the frequency of exposure and transmission via poor living conditions, poorer nutrition, higher levels of stress, diabetes, and smoking [2,25]. CMV has been implicated in the progression of many chronic diseases and may also reactivate as a consequence of chronic disease, including many of the conditions compiled in our comorbidity index [9–14]. For this reason, these diseases can mediate or confound the relation between CMV and mortality.

Smoking was dichotomised as having never smoked or having ever smoked, because only 155 (10.6%) of the sample currently smoked. Hypertension was defined as antihypertensives use or a sitting systolic blood pressure greater than or equals to 140 mm Hg and/or a diastolic blood pressure greater than or equal to 90 mm Hg. Comorbid conditions were prospectively assessed for each elderly patient according to the Charlson Comorbidity Index. This index is a good predictor of long-term mortality in medical patients [26]. According to its criteria, participants received one point for each of the following: coronary heart disease, congestive heart failure, peripheral artery disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, peptic ulcer disease, mild liver disease, and diabetes mellitus, and two points for each of the following: hemiplegia, moderate or severe renal disease, diabetes with end-organ damage, any tumour, leukaemia, and lymphoma. Three points were given for moderate or severe liver disease and six points for metastatic solid tumour or acquired immunodeficiency syndrome. The points were summed, and participants were assigned a score between 0 and 11, reflecting the number of conditions reported at baseline. Immunosuppressive therapy was defined as any dose of steroids, immunosuppressive drugs, or biological therapy administered regularly for at least the last three months. To investigate whether a systemic inflammatory reaction mediated the CMV–mortality association, serum albumin levels, C-reactive protein, white-cell count, and lymphocytes count were analysed [27]. In addition, inflammatory biomarkers were used as covariates because they correlated significantly with mortality [28], CMV infection via reactivation [19,20], and immune dysregulation and associated diseases, such as infection [16], cancer [14] and chronic diseases [7–13].

CMV Antibody

A serum sample was obtained from each patient by venipuncture within the first 48 h after hospital admission, and stored at −20 °C until analysis. A commercial enzyme-linked immunosorbent assay (ELISA) kit (Vircell® microbiologist) was used for the assessment of IgG CMV antibodies, and measured using optical density units. The coefficient of variation for the assay is <9%, specificity 100%, and sensitivity 100%/96% (compared to another (s) ELISA kits, respectively). The assay was performed and interpreted according to manufacturer recommendations. Seropositivity for CMV was defined as a serum IgG concentration of ≥7 UA/ml. The blood specimen was also used to determine C-reactive protein, serum albumin, and white-cell and lymphocytes counts.

Outcomes

Vital status was obtained through follow-up interviews and matching with the Public Health System during a median follow-up of 54 months. Death certificates were obtained for all patients.

Statistical analysis

The distributions of mortality, causes of admission and covariates were compared across levels (lowest, medium and highest) of CMV IgG antibody by using chi-squared tests for general association. All differences were significant at the .05 alpha level using 2-tailed significance tests.

CMV IgG antibody levels were parameterised as a dummy variable comparing the highest levels (high quartile) and the medium levels (2° and 3° quartile) with the lowest levels (low quartile). This categorisation was used en base on previous literature [9,13,17], and because examination of results showed no differences between the last two groups of patients (those with medium and high CMV IgG levels) regarding short and long-term mortality. For the cross-sectional analysis, a multinomial logistic regression model was constructed to control for potential confounders: age, gender, comorbidity (Charlson score), immunosuppressive drugs, C-reactive protein (CRP) levels, serum albumin levels, and white-cell and lymphocytes count. The definition of and rationale for choosing the covariates are provided above.

For the longitudinal analysis, Cox proportional hazards models were fit to investigate the association between CMV antibody levels and time until death, controlling for baseline covariates as in the cross-sectional analysis. Tests of the proportional hazards assumption based on Schoenfeld residuals indicated no violation of the assumption for all covariates [29]. Plotting martingale residuals from the Cox model without covariates against each covariate produced approximately linear smooth curves and validated the functional form for the covariates entered into the models. All analyses/graphic plots were conducted/created using SPSS 15.0 software (SPSS, Chicago, IL).

Results

Table 1 presents the demographic and clinical characteristics of our sample stratified by anti-CMV IgG levels. The mean age of the cohort was 81.2 years, with 51.3% men. Of the 715 participants, 44 had a baseline serum CMV IgG concentration of <7 UA/ml, reflecting no prior CMV infection, and 671 had a concentration of >7 UA/ml, reflecting persistent infection. Participants with evidence of persistent CMV infection were further categorised into 3 groups according to quartiles of CMV antibody concentration (UA/ml): ≤ 41 (lowest quartile = low levels), 42–80 (2 and 3 quartiles = middle levels), and ≥81 (highest quartile = high levels). Those with higher CMV IgG antibody levels were more likely to die from all causes in both the short and long term; were female; had more stroke and heart failure; and had lower levels of serum albumin. There were no differences in CMV IgG antibody levels based on age, hypertension, lipid levels, and number of other basal health conditions (Table 1).

Table 1. Characteristics of the study population (n = 715 patients) in function of CMV IgG antibody levels.

		Positive for CMV IgG Antibodies
	Negative (n = 44)	Low antibody (n = 157)	Middle antibody (n = 354)	High antibody (n = 160)	p value
Age, years (mean ± SD)	77.9 (±7.0)	80.0 (±8.0)	81.6 (±7.6)	81.7 (±7.6)	.067
Male sex, n (%)	34 (77.3)	102 (65.0)	177 (49.2)	65 (42.5)	.000
Hypertension	30 (62.8)	113 (72.0)	248 (72.1)	120 (75.0)	.52
Hypercholesteronemia	19 (43.2)	75 (47.8)	143 (40.4)	59 (36.9)	.13
Smoking status	25 (56.8)	79 (50.3)	133 (37.6)	52 (32.5)	.003
Immunosuppressive drugs	1 (2.3)	12 (7.6)	23 (6.5)	12 (7.5)	.84
Comorbidity^a
Coronary heart disease	11 (25.0)	37 (23.6)	63 (17.5)	138 (17.5)	.24
Congestive heart failure	13 (29.5)	39 (24.8)	122 (33.9)	54 (35.0)	.081
Peripheral artery disease	6 (13.6)	16 (102)	46 (12.7)	15 (10.0)	.59
Stroke	6 (13.6)	36 (22.9)	46 (13.0)	44 (27.5)	.00
Dementia	8 (18.2)	19 (12.1)	68 (19.2)	27 (16.9)	.14
Chronic pulmonary disease	16 (36.4)	39 (24.8)	100 (27.7)	46 (30.0)	.58
Connective tissue disease	1 (2.3)	7 (4.5)	13 (3.7)	5 (3.1)	.79
Peptic ulcer disease	1 (2.3)	21 (13.4)	38 (10.7)	17 (10.6)	.64
Mild liver disease	1 (2.3)	1 (0.6)	2 (0.6)	3 (1.9)	.31
Diabetes Mellitus	9 (20.5)	43 (27.4)	101 (28.5)	38 (23.8)	.53
Hemiplegia	2 (4.5)	6 (3.8)	11 (3.1)	10 (6.3)	.25
Moderate or severe renal disease	2 (4.5)	5 (3.2)	13 (3.7)	6 (3.8)	1.00
Diabetes with end organ damage	3 (6.8)	8 (5.1)	20 (5.6)	7 (4.4)	.87
Any tumour	11 (25.0)	31 (19.7)	54 (15.0)	22 (14.4)	.33
Leukaemia	0 (0.0)	2 (1.3)	6 (1.7)	3 (1.9)	1.00
Lymphoma	1 (2.3)	2 (1.3)	5 (1.4)	1 (0.6)	.80
Moderate or severe liver disease	0 (0.0)	2 (1.3)	4 (1.1)	7 (4.4)	.058
Metastatic solid tumour	1 (2.3)	11 (7.0)	22 (6.2)	12 (7.5)	.82
AIDS	0 (0.0)	0 (0.0)	4 (1.1)	2 (1.3)	.55
Charlson score (n)	2 (0-7)	2 (1-4)	2 (1-4)	2 (1-4)	.62
Causes of admission
Infectious diseases	12 (27.3)	59 (37.6)	133 (37.6)	49 (30.6)	.46
Cardiovascular diseases	10 (22.7)	23 (14.6)	63 (17.8)	33 (20.6)	.43
Tumors	2 (4.5)	14 (8.9)	30 (8.5)	56 (7.8)	.71
Inflammatory biomarkers
C-reactive protein (mg/L)^b	26 (6-77)	25 (8-59)	31 (11-70)	26 (7-72)	.27
Serum albumin (mg/dL)	3092 ± 694	3020 ± 573	2879 ± 650	2817 ± 624	.010
White-cell count (per µl)	8.4 (±4.0)	8.1 (±3.6)	9.1 (±9.5)	8.3 (±4.7)	.57
Lymphocytes count (per µl)	1.01 (±0.5)	1.3 (±0.7)	1.4 (±1.2)	1.4 (± 0.9)	.63
All-cause mortality
In-hospital/within 30 days after discharge	8 (18.2)	18 (11.5)	59 (16.7)	35 (21.9)	.045
92 month follow-up	31 (70.5)	98 (62.4)	261 (73.7)	121 (75.6)	.016

CMV: cytomegalovirus; low antibody: levels within the first quartile or <41 AU/ml; middle: 2 and 3 quartiles or antibody levels between 41 and 81 AU/ml; high: top quartile or antibody levels > 81 AU/ml; AIDS: acquired immunodeficiency syndrome.

The distributions of mortality, causes of admission and covariates were compared across levels (low, middle and high) of CMV IgG antibody by using the Chi-squared/ Kruskal–Wallis tests for general association and according to the type of variable.

^aThe comorbidities were standardised according to the Charlson criteria [26].

^bMedian (interquartile range).

A total of 112 patients (16.7%) died in-hospital or within 30 days after discharge: 18 (11.5%) patients with low anti-CMV IgG levels, 59 (16.7%) patients with middle levels, and 35(21.9%) patients with high levels. Table 2 presents odds ratios and 95% confidence intervals from regression models of all-cause mortality. Except for immunosuppressive drugs, all the other covariates were associated with mortality in the bivariate analysis. In model 1, CMV IgG antibody levels (high compared with low levels) were associated with an increased death (odds ratio = 2.16, 95% confidence interval: 1.16–4.01). When we adjusted for age and gender, the top quartile of CMV IgG antibody levels was associated with a 27% (95% confidence interval: 19–33) increased odds of all-cause mortality (model 2). In our final model, only heart failure was added, since it was the only covariate that was associated with both CMV serology and short-term mortality. Model 3 (heart failure added) showed an odds ratio for CMV slightly lower than the previous (HR = 2.20, 95% confidence interval: 1.15–4.21).

Table 2. Odds Ratios for all-cause of short-term mortality in CMV seropositive (n = 671 patients) hospitalised elderly patients.

	Adjusted OR	p value
Model 1^a
High vs. lowest antibody levels	2.16 (1.16-4.01)	.014
Middle vs. lowest antibody levels	1.54 (0.87-2.71)	.13
High/middle vs. lowest antibody levels	1.72 (1.00-2.96)	.047
Model 2^b
High vs. lowest antibody levels	2.27 (1.19-4.33)	.012
Middle vs. lowest antibody levels	1.59 (0.89-2.83)	.011
High/middle vs. lowest antibody levels	1.81 (1.05-3.15)	.033
Model 3^c
High vs. lowest antibody levels	2.20 (1.15-4.21)	.016
Middle vs. lowest antibody levels	1.49 (0.83-2.66)	.017
High/middle vs. lowest antibody levels	1.73 (1.00-3.02)	.050

Short-term mortality, in-hospital mortality or within 30 days after discharge. OR: odds ratio (95% confidence interval); CMV: Cytomegalovirus.

^aModel 1: bivariate relation with mortality.

^bModel 2: include age (>81 years) and gender.

^cModel 3: covariates associated with anti-CMV IgG levels and mortality included: heart failure.

After a median time of follow-up of 76 months (range: 56–92), 480 (71.5%) deaths occurred, consisting of 98 (62.4%) patients with low anti-CMV IgG levels, 261(73.7%) patients with middle levels, and 121(75.6%) patients with high levels. Table 3 presents hazard ratios and 95% confidence intervals from Cox proportional hazards models of all-cause mortality. In bivariate models, all variables were significantly associated with all-cause mortality. In model 1, CMV IgG antibody levels (high compared with low levels) were associated with an increased hazard of death (hazard ratio = 1.46, 95% confidence interval: 1.11–1.90). When we adjusted for age and gender, the top quartile of CMV IgG antibody levels was associated with a 51% (95% confidence interval: 14–20) increased hazard of all-cause mortality (model 2). Our final model excluded hypertension, hypercholesteronemia, immunosupresive drugs, C-reactive protein, white-cell count, and lymphocytes count because they were unassociated with CMV in our population. Smoking status was also excluded due to heavy dependency on sex. In fact, only 4.3% of women were smokers compared to 72.7% of men. In the absence of association between the comorbidity index (Charlson score) and CMV serology, comorbidities that were associated with CMV IgG levels were included in the model (stroke and heart failure; this latter is tangentially associated). Model 3 added the stroke, heart failure and serum albumin levels; the hazard ratio for CMV decreases slightly (hazard ratio = 1.40, 95% confidence interval: 1.05–1.86). No interaction terms were found between the anti-CMV-IgG levels and short and long-term mortality with respect to the covariates included in the study.

Table 3. Hazard ratios for all-cause mortality at 7,6 years follow-up in 671 CMV seropositive hospitalised elderly patients.

	Adjusted HR	p value
Model 1^a
High vs. lowest antibody levels	1.46 (1.11–1.90)	.005
Middle vs. lowest antibody levels	1.33 (1.06–1.68)	.014
High/middle vs. lowest antibody levels	1.37 (1.10–1.71)	.005
Model 2^b
High vs. lowest antibody levels	1.51 (1.14–2.00)	.003
Middle vs. lowest antibody levels	1.46 (1.15–1.81)	.002
High/middle vs. lowest antibody levels	1.49 (1.18–1.86)	.001
Model 3^c
High vs. lowest antibody levels	1.40 (1.05–1.86)	.019
Middle vs. lowest antibody levels	1.42 (1.11–1.81)	.004
High/middle vs. lowest antibody levels	1.41 (1.12–1.77)	.003

^aModel 1: bivariate relation with mortality.

^bModel 2: include age (>81 years) and gender.

^cModel 3: covariates associated with anti-CMV IgG levels and mortality included: serum albumin levels (<2900 mg/dL), stroke and heart failure.

Given the known association between CMV infection and cardiovascular disease [7,8], we performed additional sensitivity analyses to assess the contribution of cardiovascular disease to the higher incidence mortality in participants with high serum CMV antibody concentration. After excluding participants with acute (cause of admission) or baseline cardiovascular disease (angina, myocardial infarction, congestive heart failure, peripheral artery disease, or stroke), we repeated the mortality analyses, in which the sample sizes were consequently reduced by 22.8% and 55.5%, respectively (Table 4). Cardiovascular-disease-free older patients had higher short and long-term mortality rates. Thus, models 1, 2, and 3 increased the relation between high CMV IgG levels (compared with low levels) and short term mortality by 103%, 141%, and 98%, respectively; and long-term mortality by 46%, 70%, and 82%, respectively.

Table 4. Odds and hazard ratios for all-cause mortality in CMV seropositive hospitalised elderly patients without cardiovascular disease.

	Adjusted OR/HR	p value
Model 1^a
High vs. lowest antibody levels	3.19 (1.47–6.93)^d	.003
	1.92 (1.20–3.06)^e	.006
Middle vs. lowest antibody levels	2.26 (1.10–4.65)^d	.010
	1.74 (1.16–2.62)^e	.007
High/middle vs. lowest antibody levels	2.53 (1.26–5.07)^d	.008
	1.79 (1.21–2.65)^e	.001
Model 2^b
High vs. lowest antibody levels	3.68 (1.64–8.27)^d	.002
	2.21 (1.36–3.60)^e	.001
Middle vs. lowest antibody levels	2.24 (1.17–5.10)^d	.017
	1.92 (1.27–2.11)^e	.002
High/middle vs. lowest antibody levels	2.77 (1.36–5.62)^d	.005
	2.01 (1.35–2.99)^e	.001
Model 3^c
High vs. lowest antibody levels	3.18 (1.40–7.24)^d	.006
	2.22 (1.36–3.62)^e	.001
Middle vs. lowest antibody levels	2.29 (1.09–4.81)^d	.028
	1.87 (1.24–2.84)^e	.003
High/middle vs. lowest antibody levels	2.53 (1.24–5.18)^d	.11
	1.97 (1.32–2.94)^e	.001

^aModel 1: bivariate relation with mortality.

^bModel 2: include age (>81 years), and gender.

^cModel 3: Covariates for short-term mortality included: heart failure and serum albumin levels <2900 mg/dL; and covariates for long-term mortality included: serum albumin levels <2900 mg/dL, stroke and heart failure.

^dOdds ratio between short-term mortality and CMV IgG levels.

^eHazards ratio between long-term mortality and CMV IgG levels.

Smaller, although significant, results are obtained when comparing 2–3 (middle levels) and 2–4 (high/middle levels) quartiles with lower quartile in the total sample and in patients without cardiovascular disease (Tables 3 and 4).

Discussion

In this prospective study of elderly hospitalised patients, those in the top quartile of CMV IgG antibody concentration had higher risks of short-term mortality (in-hospital/within 30 days after discharge) than those with the lowest quartile. After adjustment for potential confounders, middle and high values of CMV antibody independently predicted a higher risk of 7.6-year mortality. The exclusion of patients with cardiovascular disease significantly increased death associated with CMV infection. Survival curves suggest significantly worse all-cause for individuals in the highest quartile of CMV IgG antibody levels over follow-up time.

Given the high prevalence of CMV infection worldwide, a better understanding of its short and long-term clinical effect in immunocompetent elderly has an important public health impact. To our knowledge, this study is the first to report on the relation between CMV IgG antibody levels and mortality in a hospital cohort. These results are in agreement with the few available studies examining CMV seropositivity and all-cause mortality in older subjects. In the Swedish longitudinal OCTO Immune Study, mortality was related to CMV-driven characteristic immune system changes [30]. Similar findings were replicated in the Swedish NONA Immune Study [31]. These two studies had modest sample sizes (102 and 138 participants, respectively), utilised non-institutionalized/community-dwelling subjects, and examined seropositivity rather than antibody levels. CMV serostatus was associated with a near doubling of cardiovascular deaths, whereas there was no increase in mortality from other causes, in 511 individuals aged at least 65 years who were followed up for 18 years [32]. However, CMV seropositivity was not associated with an increased risk for all-cause mortality in 549 community-dwelling persons aged 80 and older recruited from Belgium [33]. CMV seropositive subjects with high anti-CMV levels were at higher risk for all-cause mortality in this study [33]. Previously, in a population-based cohort study that included 13.090 participants (aged 40–79 years), CMV seropositivity was associated with all-cause mortality (HR 1.16, 95% CI 1.07–1.26), and mortality rates increased as CMV IgG antibody levels rose [34]. However, both this and previous studies recruit subjects living in the community and focus their attention on investigating long-term mortality.

Our study extends these earlier findings by testing the association with antibody levels in elderly patients hospitalised for any cause. Given that approximately between 80% and 97% of individuals aged 65 years or older are seropositive for CMV [2,18], it is unlikely that infection serostatus alone will explain mortality outcomes. Measuring antibody response, on the other hand, may capture subclinical reactivation of CMV and therefore may be more relevant for examining immune system dysfunction, and its consequences, than seropositivity alone [19,20]. Thus, our novel results concerning mortality support earlier works showing that immune response to CMV is related not only with to myriad of comorbidities and chronic diseases [7–13], but also with acute illnesses and hospitalisation.

Several mechanisms may explain why a higher CMV antibody level is associated with an increased risk of hospital mortality. First, reactivation of CMV likely contributes to cell damage through inflammatory pathways and/or direct pathogenicity [4,35]. However, this mechanism seems to be uncommon as demonstrated in critically ill patients [4]. Indirect effects not attributed to end-organ disease (lung, digestive tract, etc.) have also been hypothesised. In this way and as previously mentioned, CMV has also been associated with cardiovascular disease and other chronic diseases of aging [7–13]. The specific mechanisms responsible for these associations have not been fully elucidated, but are likely to have an immune and inflammatory component. Indeed, CMV seropositivity belongs to a cluster of immune factors constituting an “immune risk profile” (IRP) associated with all-cause mortality at 2, 4 and 6-year follow-up in elderly Swedes in the OCTO/NONA longitudinal studies [30,31]. IRP and other risk factors, such as cognitive decline, can accumulate and amplify the effect on non-survival; as demonstrated in 138 elderly Swedish followed for 4 years [31]. Inflammatory changes included in the IRP (cytokine IL-6 increased) may be involved in this effect [31,36]. Inflammatory data of chronic, sterile, and low-grade inflammation that develops with age – called inflammaging – surpasses those enrolled in the IRP [36,37]; and its impact on survival is less, at least in the short term, than those that develop in response to acute processes and/or multiple comorbidities [37]. Persistent infection with CMV has been associated with deficits in cellular immune responses and increased health risks with aging [16,38]. Repeated response to CMV infection leads to inflation of the memory compartment with CMV-specific clones and may place considerable limitations on the responsiveness of the CD8 + repertoire towards other antigens, as this oligoclonal expansion minimises the space and resources necessary to maintain T cells with other specificities [38]. As a result, there is a reduction of peripheral naïve CD8 + T cells available for fighting new infections [36,39]. However, our patients with infection (as the cause of admission) do not have higher levels of CMV IgG antibodies.

An extended hypothesis is that the higher mortality rate among subjects with the highest CMV antibody levels could be mediated partially or completely by cardiovascular disease, known to be associated with high CMV antibody levels [7,8,17]. In the present study, patients with stroke and, to a lesser extent, patients with heart failure, showed increased levels of CMV IgG antibodies. However, the exclusion of patients with these and other cardiovascular diseases substantially increases of risk of mortality associated with CMV IgG levels. Death from competitive causes linked to CMV infection could be involved in these results. In other words, chronic diseases [9–13] and immune dysregulation [15,16], both repeatedly linked to CMV infection, could gain protagonism as a cause of death in the absence of cardiovascular disease.

Although unlikely, measurement error might have affected our results. The assay used is not affected by other common viral antibodies and has good performance characteristics. Error cannot have been introduced by inaccuracies in mortality information, since the exact date is immediately registered in the Social Security database. Residual confounding might have also affected our results. In this study, we measured CMV IgG antibody levels at a single time point and did not directly measure CMV reactivation. Although it is postulated that higher CMV IgG antibody levels represent more frequent or intense subclinical CMV reactivation from latency, this has not been conclusively proven. CMV IgG antibody levels correlate well with CMV-specific B-cell numbers [40], but significant intraindividual variation exists [40], and antibody levels do not always reflect the presence of CMV DNA [41]. This variation may explain the wide range of CMV IgG antibody levels (middle and high) associated with mortality in our study. Other events that might affect CMV IgG levels acutely or chronically have not been studied. It is possible, therefore, that CMV IgG antibody levels are elevated through an unknown mechanism and that the underlying cause is also associated with the risk of death. In this context, CMV infection might even be an innocent bystander or a measure of a failing immune system [42].

In conclusion, this study contributes to growing evidence that CMV immune response is related to the deleterious processes involved in aging, contributing to premature mortality. The mechanisms underlying these associations remain to be elucidated, but cardiovascular disease could have a lower role than expected in ill elderly with high comorbidity. Further studies elucidating the pathophysiological and biologic mechanisms underlying the association between CMV infection, disease and death are needed. These findings provide a strong rationale for well designed longitudinal hospital population studies investigating the clinical effects and pathogenic mechanisms of persistent CMV infection, the results of which can guide public health efforts on potential preventive and therapeutic strategies for this widely prevalent infection. Given the improvements in clinical efficacy that licenced vaccines for older adults have already been demonstrated [43], we support the need – already expressed by other authors – to develop effective vaccines against CMV infection applicable to seronegative subjects of any age [44].

Disclosure statement

No potential conflict of interest was reported by the author(s).