Vaccination Against Pneumonia May Provide Genotype-Specific Protection Against Alzheimer’s Disease

INTRODUCTION

Accumulating evidence suggests that infections could play a major role in Alzheimer’s disease (AD); however, mechanism is poorly understood. Diverse studies linked different microorganisms (viruses, bacteria, fungi) to AD and related traits [1–4]. This indicates a possibility that the culprit may not (or not only) be a specific microbe, but compromised immunity that may increase vulnerability of the brain to various infections promoting neurodegeneration. Examples of factors that may compromise immunity include aging of the immune system, certain exposures (e.g., to air pollution), and genetic variation, among others [5–9]. For instance, aging is characterized by immunosenescence, which is accompanied by depletion of reserves of naive immune cells, slower and impaired responses to antigens, declining immune surveillance, increased permeability of the blood-brain-barrier (BBB), and chronic inflammation [5, 10, 11], all of which may increase vulnerability of the brain to a variety of microbes and their damaging byproducts. If compromised immunity (due to aging or other factors) does underly the connection between infections and AD, then approaches to strengthening immunity might confer protection against AD.

Vaccination is one such approach that may improve immunity broadly, in addition to making individuals more resistant/resilient to specific infections. Recent studies suggested that some vaccines may have beneficial heterologous (a.k.a. off-target, non-specific) effects on immunity that span beyond protection against the target disease [12, 13]. Mechanisms of such effects may involve heterologous lymphocyte responses, trained innate immunity, stabilization of the BBB, and other yet unknown factors [9, 12]. Several studies reported the beneficial off-target effects of pneumococcal and influenza vaccines on various health traits, including AD-related [14–17].

Here we conducted a retrospective cohort study among a subset of individuals who survived without AD as of age 75 in the Cardiovascular Health Study (CHS), to estimate associations of vaccinations against pneumonia and influenza (flu) received between ages 65 and 75 with AD onset at age ≥75 years, taking into account individual genotype, to explore the potential of these vaccines as candidates for personalized, genotype-tailored, prevention of AD. We focused on the rs6859 polymorphism of NECTIN2 (nectin cell adhesion molecule 2, a.k.a. herpes virus entry mediator B) gene on chromosome 19 involved in adherens junctions that are important for controlling BBB permeability and protecting the brain from infections. We selected NECTIN2 because it plays role in both AD and vulnerability to infections, and its rs6859 polymorphism is an established genetic risk factor for AD, which may also act independently of APOE4, as shown by our and others’ previous genome-wide association studies and linkage disequilibrium analysis [7, 18]. We propose that this polymorphism may modulate the effects of vaccines on AD risk.

MATERIALS AND METHODS

RESULTS

Characteristics of CHS participants are summarized in Table 1. Results of associations of vaccinations against pneumonia and flu with AD onset, unstratified, as well as stratified by carrier status of the rs6859 A allele, are shown in Table 2. Pneumonia vaccine received between ages 65–75 was associated with 33% lower odds of AD onset at age 75 and older in carriers of A allele (OR = 0.67; p = 0.0496), but not in non-carriers, or in the total sample. The influenza vaccine alone did not show a statistically significant association with AD in any group, and the number of flu shots was only marginally statistically significantly (i.e., p-value slightly exceeding 0.05) associated with a lower AD risk in unstratified sample, and in carriers of A allele. The total count of vaccinations against pneumonia and flu was associated with lower odds of AD in the overall sample (OR = 0.94; p = 0.049), and in carriers of the A allele (OR = 0.93; p = 0.048), but not in non-carriers. Notably, all associations in the non-carrier group were not statistically significant, and most ORs were close to 1 (Table 2C). Kaplan-Meier survival trajectories of the probability of staying AD-free at ages 75 and older, following the vaccination against pneumonia between ages 65–75, additionally illustrated the protective effect of pneumonia vaccine on AD in carriers of the rs6859 A allele, but not in non-carriers (Supplementary Figure 1).

DISCUSSION

This retrospective cohort study among a subset of CHS participants found that pneumococcal polysaccharide vaccine, as well as total count of vaccinations against pneumonia and flu, received between ages 65–75, have genotype-specific associations with odds of AD onset at age ≥75 years, such that vaccinations were AD-protective only in carriers of the rs6859 A allele, but not in non-carriers. Results of this study are broadly consistent with the idea that strengthening immunity by the off-target effects of vaccines may provide protection against AD, and also suggest that variation in genes involved in AD and vulnerability to infections may influence such effects of vaccines and should be taken into account in personalized prevention of AD.

Other genes and polymorphisms beyond the rs6859 in NECTIN2 might also influence the effects of vaccines on AD, e.g., our unpublished data (a conference abstract) suggested that rs2075650 in TOMM40 is another candidate polymorphism that may modulate the effect of pneumococcal vaccine on AD. The rs2075650, however, is in linkage disequilibrium with multiple SNPs in APOE and NECTIN2, and is also the eQTL SNP that influences expression of these genes, so its biological effects, including possible role in vulnerability to infections, may potentially be related to functions of APOE and/or NECTIN2, which deserves separate investigation.

In addition to the possibility of a beneficial heterologous effect of pneumococcal vaccine on immunity, there is also a possibility that this vaccine may contribute to a reduction in AD risk by simply preventing cases of the originally targeted disease. For example, it was shown that bacterial pneumonia is associated with increased risk of AD [19]. If so, then preventing the pneumonia cases might reduce AD risk. Such mechanism seems plausible. Some studies, however, suggested that the off-target effects of vaccines cannot be fully explained by pathogen-specific immune protection, and that non-specific immune system modulation by vaccines may be a common phenomenon [20]. It is also possible that both mechanisms (i.e., boosting immunity broadly, along with preventing cases of target disease that may otherwise promote dementia) could be responsible for lowering the risk of AD by the pneumonia vaccine.

One potential reason for the lack of a significant effect of the flu vaccine on AD could be that this vaccine has a relatively short-term (seasonal) effect, so multiple flu shots might be needed to produce a noticeable long-term beneficial effect on immunity [21]. Another reason could be differences between off-target effects of pneumonia and flu vaccines, among other factors [9, 12].

We acknowledge that this study has limitations. Some are inherent to the CHS design, which includes potential response and participation biases [22], and unavailability of clinically verified AD diagnoses. A common limitation of studies that use hospital discharge records is that identifying disease cases based on such records may potentially lead to disease misclassification. For example, exclusion of participants with AD onset before age 75 might be incomplete because some participants with no prior hospitalizations would not be identified as having AD based on hospital records, and so could be misclassified as free of AD at the start of the follow-up (i.e., age 75). In such cases, there may be a possibility of reverse causality. Unavailability of sampling weights that would allow generalizability of findings to the general US population of respective ages is another limitation that is inherent to the CHS design. In addition, our analysis did not consider competing risks (e.g., due to mortality). One should also note that even though we adjusted all analyses for relevant covariates, such as sex, race, birth cohort, education, and smoking, we cannot fully exclude the possibility that people who obtain vaccinations, as compared to those who do not, may have a higher socioeconomic status, or seek vaccinations due to other, yet unknown, factors. We also did not adjust for comorbidities, which may, potentially, impact likelihood of vaccination and/or risk of AD. Indeed, there may be many unknown factors that could impact propensity to vaccination and chances of AD.

We conclude that pneumococcal polysaccharide vaccine is a promising candidate for repurposing for genotype-tailored personalized prevention of AD. Multiple seasonal flu shots may potentially strengthen its effect, due to, e.g., additional stimulation of the immune system. This possibility, however, needs further investigation and confirmation. It is important to note that our study suggested the protective effect of the pneumococcal vaccine on AD onset among CHS participants specifically, so its results should not be generalized to other groups. Prospective, randomized, or pseudo-randomized (using propensity score matching) studies with adjudicated AD cases could help further support these findings.