Pathogenic load and frailty in older adults: Singapore longitudinal ageing study

Introduction

Frailty is an age-associated state of health characterized by increased vulnerability to the effects of stress resulting eventually in increased risks of adverse health outcomes including multi-morbidity, cognitive impairment, disability, hospitalization, institutionalization and mortality [1, 2]. Frailty may be viewed as a pathophysiological state of diminished ability to maintain homeostasis due to increased allostatic load from chronic stress responses [3–6]. Physiological dysregulation in multiple organ systems underlying the concept of frailty [7–9] is the basis for two main approaches to measuring frailty in clinical research. One approach is the frailty index, an accumulated deficits construct using cumulated counts of age-related health deficits across multiple physiological and functional systems, and encompasses multiple dimensions of physical, psychological, biological and social functioning [10]. The other approach is the physical frailty phenotype, [11] clinically defined by the presence of unintentional weight loss, self-reported exhaustion, weakness, slowness, and low levels of physical activity. Frailty may therefore be examined globally or in terms of multiple phenotypes or domains.

The causes of age-related frailty are multifactorial, and chronic systemic inflammation is believed to play a central pathogenic role. The systemic pro-inflammatory milieu seen in old age is partly the result of the dysregulation of immune response associated with ageing (“immunosenescence”) involving both innate and (more extensively studied) adaptive immunity, suggesting the erosion of immune surveillance over time [12]. This is phenotypically characterized by the accumulated number and proportion of late-stage differentiated T cells, especially CD8+T cells. T cells may be driven to late differentiation and/or exhaustion by the continuous antigen stimulation from persistent latent infections from microbial agents. Persistent and latent infections which are highly prevalent worldwide include those caused by Herpesvirus such as herpes simplex viruses 1 and 2 (HSV-1 and HSV-2) and varicella-zoster virus (VZV); cytomegalovirus (CMV), Epstein-Barr virus (EBV), and other viruses such as hepatitis B virus (HBV), human papilloma virus (HPV), human immunodeficiency virus (HIV), as well as other microbes such as helicobacter pylori and mycoplasma pneumoniae.

Evidence for the role of continuous antigenic stimulation from persistent and latent infections in human frailty is limited. HIV patients show characteristic surface markers and functionality profiles on CD4+ or CD8+ T cells known to drive premature immunosenescence [13, 14]. High levels of CD8+ and low levels of CD4+ T-cells contributing to pro-inflammatory cytokine increase is reported to predict frailty in older adults [15] CMV infection is the most well studied viral factor driving the differentiation of the CD8+ T cells towards immunosenescence (Fulop et al, 2013), but human studies that examined the association between persistent CMV infection and clinical outcomes have produced contradictory results. Some studies have shown that CMV was associated with prevalent frailty, [16–18] functional and cognitive impairment, [19–21] and all-cause and cardiovascular disease mortality [22–24] but other studies have not supported these findings [15–28]. A recent systematic review and meta-analysis of six studies [16–18, 26, 27, 29] analysing the association between CMV, EBV, VZV, and HSV seropositivity and frailty in older people found an association between CMV infection and frailty only in older persons aged 60–79 years, but not in the oldest-old subjects. No association was found between EBV, VZV, and HSV infections and frailty [30].

So far, studies have selectively investigated less than a handful of individual viruses singly for their individual associations with frailty and related outcomes, and no studies have considered latent infections by bacterial or other pathogens. In this study, we assessed latent infection by ten viral, bacterial and mycoplasma agents (Cytomegalovirus, Herpes simplex 1, Herpes simplex 2, Varicella-zoster virus, Epstein-Barr virus, Respiratory syncytial virus, Chikungunya, Dengue, H. Pylori, and Mycoplasma pneumonia). As equivocal results in previous studies of single infections may be due to confounding by other pathogenic infections and the modifying effect of age, we investigated the association of a pathogenic load of multiple latent infections on measures of prevalent frailty and mortality in a study sample of young-old and old adults.

Results

The mean age of the study participants was 72.7 years (range: 57-92). The mean PL count was 6.1 (SD=1.2), individually ranging from 3 to 9. The most common pathogens (CMV, H. Pylori, EBV, M. Pneumoniae, and HSV-1) were prevalent in 80 to 99% of the participants, while other pathogens were prevalent in less than 65% of the participants, with the least prevalent pathogen being Chikungunya virus (4.2%). Figure 1.

Figure 1. Latent infections by pathogens.

The mean CD-FI scores was 0.26 (SD=0.086, range: 0.09 to 0.49); 54.2% were pre-frail (1-2 components), and 14.1% were frail (3-5 components). Participants with high PL (7-9) were significantly older, and tended to show more smoking histories, to live alone, have greater number of chronic diseases, and higher CRP levels, although the differences were not statistically significant at p<0.05. Table 1. However, they showed significantly greater mean scores on the CD-FI (p=0.001 and p=0.002), compared to those with lower PL. A quarter (24.6%) of those with high PL were physically frail (PP-FI score of 3-5), significantly more than their low PL counterparts (6.2%, p=0.001). They were significantly more likely to be weak (57.4%) compared to the low PL counterparts (29.6%, p=0.001), and to be slow (23.0%) compared to their counterparts (8.6%), p=0.017. Otherwise, there were no significant differences for exhaustion, inactivity or shrinking.

The estimated effect sizes of association of high PL with these frailty measures are shown in Table 2. Controlling for the confounding influences of sex, age, education, smoking and alcohol history, these associations were attenuated but remained statistically significant. A high PL was associated with an estimated increase of 0.035 points in the CD-FI (Cohen’s D=0.035 / 0.086, or 0.41). A high PL was associated with 8.5 times odds of being physically frail (p=0.001), 2.8 times odds of being weak (p=0.010); and 3.4 times odds of being slow (p=0.024). The mortality hazard ratio associated with a high PL, controlling for age, sex, education, smoking and alcohol histories, was 1.53 (95% CI, 1.01-2.1). See Figure 2.

Figure 2. Cumulative survival by pathogenic load category.

In additional analyses, we explored the association of latent infection by specific pathogens with frailty. We found no significant associations for all except marginal associations for EBV and Chikungunya. In preliminary analyses of variance, EBV seropositivity was significantly associated with PP-FI (mean ± SD of 1.27 ± SD 1.16 versus 0.62 ± 0.77, p=0.012), but the association was not significant (p=0.095) when controlled for confounding variables. Chikungunya seropositivity was associated with CD-FI (mean ± SD, 0.329 ± 0.108 versus 0.257 ± 0.084, p=0.014, but the association was not significant (p=0.053) when controlled for confounding variables. We also compared participants with high anti-CMV IgG titers in the top tertile range versus those with low titers, and observed no difference in CD-FI (0.256±SD 0.083 vs 0.266 ±SD 0.092, p=0.546).

Discussion

To our knowledge, this is the first study that links a high pathogenic load of latent infections to an increased risk of frailty and mortality. Thus far, only a handful of studies have investigated a link with frailty for latent infection by a few individual pathogens of interest, namely CMV, EBV, VZV and HSV, and in single order [16–18, 26, 27, 29]. We found notably in this study population sample that virtually all but 2 participants were tested positive on CMV serology, and neither CMV seropositivity nor high antibody titers was associated with frailty. Hence for latent infection by pathogens like CMV, H. Pylori, EBV, M. Pneumoniae, and HSV-1 that are ubiquitous in most populations including ours, it will be exceptional for studies to detect a significant link with frailty. The same may apply for rare infections by pathogens such as Chikungunya in this population, unless the study has a very large sample size. In this study we took a different line of inquiry away from the traditional approach of establishing unique associations with frailty for specific pathogenic infection on its own. Instead we hypothesized that multiple latent infections by a variety of pathogens each with varying small effects may contribute collectively to a large effect on frailty and mortality when present in the same individual.

By the nature of latent infection, the potential reactivation of pathogens is kept in check by constant immune surveillance, and wherewith, the induced responses of virus-specific CD8+T cells can be detected long after the virus is controlled [37]. In HIV disease where a basal viral load is constantly present, or CMV infection which intermittently reactivates during the lifetime, constant or repeated viral replication and antigenic stimulation have been shown to cause an antigen-dependent clonal expansion of the memory T cells resembling immunosenescence. Most CMV-specific CD8+ T cells are of TEMRA phenotype with decreased CD28, CD27, CCR7, CD62L and increased CD57 and re-expressed CD45RA surface markers, and are largely pro-inflammatory producing a large amount of IL-6, TNFa, and IL-1b [38]. We have previously reported an association between physical frailty and markers of T cell senescence [32] and herewith we provide evidence for immunological history linking this previous observation. The clinically latent or asymptomatic viral disease under viral load control is thus characterized by a persistent systemic inflammation, which may thus contribute to age-associated diseases and the frailty state. Also, the maintenance of viral infection in the latent state carries a high energy cost, as the virus reactivation potency [39] and the expansion of CD8+ T cells at the expense of other T cell subpopulations both rely on the host’s mTOR pathway, thus depleting energy reserves, which also characterizes the frailty state.

The immune system response to chronic stress originating from latent infection represent an allostatic process to maintain homeostasis [3]. The allostatic overload from chronic antigenic stimulation from pathogenic sources is akin to the multisystem physiological dysregulation that researchers believe underlies ageing and frailty. Studies show that higher allostatic load scores based the count of multiple biomarkers of cardiovascular, endocrine, metabolic and immune function are associated with higher levels of the physical frailty syndrome [5]. Our finding of the link between pathogenic load and frailty and mortality in this study may thus be understood in the same context.

Our study does not exclude the possibility that certain specific viral agents such as CMV may be linked to frailty at least in some populations. Rather, it supports the allostatic load hypothesis that the increasing numbers of latent infection from different pathogens including viruses, bacterial and mycoplasma was associated with increased frailty and mortality risks. General and perhaps varied processes of immune surveillance responses to multiple pathogen-specific antigenic stimulations may cause gradual erosion of the immune system and other systems over the lifetime, and explain the development of age-related frailty. The caveat to this generalization is that while viruses such as CMV, HIV and HBV are likely to drive the accumulation of memory T cells towards senescence as described above, other viruses such as EBV and VZV as well as non-viral pathogens with their different life cycle and pathophysiology [40, 41] may have different mechanisms, which have not been as well studied.

There are some limitations in this study to note. Serological testing was performed according to manufacturer’s instructions and seropositivity was determined according to the antibodies titre cut-offs provided by the manufacturer. A lack of test standardization between laboratories and variations in cut-off levels may affect the sensitivity and specificity for a specific serological test. Variations in sensitivity and specificity across different serological tests may result in varying seropositivity leading to under-counting or over-counting of the pathogenic load. While serologic testing indicate previous infection was present at some time prior to 2003, it does not provide information on the duration of infection. Hence, it does not perfectly measure the degree of antigenic stimulation that the donor has been experiencing over a long period of time.

Additionally, there may have been additional infections during the period of follow up from 2003 to 2017, that were not captured by the serological data. Our study has the limitation of small sample size and case-control design, with the possibility of inherent prevalence bias and the odds ratio over-estimation of risks. We thus suggest that our results be viewed as preliminary findings and larger prospective cohort studies should be conducted.

Materials and Methods

Supplementary Materials

Supplementary Tables