Immune boosting to combat Alzheimer's disease: Position statement

Abstract

The societal cost of Alzheimer's disease (AD) and related dementias is immense, but there are no effective treatments. Emerging evidence suggests that AD brain pathology may be triggered by microbial infection, and that the diagnostic protein deposits in AD brain represent an antimicrobial defense mechanism. This raises the prospect of protection against AD through stimulation of the immune system. Specifically, non-specific activation through ‘trained immunity’ or ‘immunopotentiation’ is now known to offer protection against unrelated pathogens as well as against diverse diseases. Earlier findings that vaccine recipients are at reduced risk of developing AD have now been replicated around the world. Protection has been reported for diverse types of vaccines, and the effect sizes are clinically meaningful. Although these studies are predominantly observational, and are thus at risk of confounders such as ‘healthy vaccinee’ bias, several studies based on ‘natural experiments’ avoid this bias and provide stronger evidence that vaccine administration can reduce dementia risk. Most such vaccines can be obtained for a few dollars, although others are a little more expensive. But the cost of vaccination is entirely trivial compared to the financial burden of AD. Both clinical trials of immunopotentiation in AD and further research into the underlying mechanisms will be essential. This Position Statement briefly summarizes current knowledge and recommends that immune boosting should be placed at the top of the AD public health R&D agenda.

The annual societal cost of Alzheimer's disease (AD) and related dementias, including healthcare and lost productivity, is estimated to exceed USD 1.3 trillion worldwide. Even a 10% reduction in AD rates would save USD 38 billion each year in a single country, the USA, and £4 billion in the UK. Immune boosting potentially offers an immediate means to reduce the societal burden of AD but has not so far been taken up by public health authorities or funding agencies. In this concise Position Statement, we first briefly review the background and recent evidence and then offer recommendations.

Background to trained immunity

A century ago it was found that individuals immunized with smallpox vaccine showed regression of lesions caused by an entirely unrelated pathogen, herpes simplex virus. ¹ At the same time it was discovered that laboratory animals infected with the tuberculosis-causing microbe Mycobacterium tuberculosis showed a 20-fold increase in antibodies against an unrelated antigen. ² The inventor of the Bacille Calmette–Guérin (BCG) vaccine, Calmette, noted that children died far less after BCG vaccination than their unvaccinated peers and speculated: ‘does the harbouring of BCG, followed by its digestion and elimination, confer on the organism a special aptitude to resist those other infections which are so frequent in young children?’ ( ³ , p. 92).

These and other findings have spurred the idea that immunological challenge enhances the ability of the immune system to ward off unrelated infections, a concept now known as ‘trained immunity’ or ‘immunopotentiation’. ⁴ In addition to invigorating the immune system to eliminate pathogens, there is evidence that vaccines can modulate systemic inflammation and induce beneficial long-term changes in regulatory immune cells.^5–7

The concept of trained immunity attracted the attention of the medical community in part because studies from low-income settings showed that inoculation of the tuberculosis vaccine BCG could markedly reduce neonatal mortality from infectious disease and sepsis,^8–10 as Calmette observed, and this finding that has been reiterated in other settings (e.g., ¹⁰ ).

The idea gained further momentum during the COVID-19 pandemic when inoculation of BCG was found to be associated with improved outcomes following SARS-CoV-2 infection. Two randomized double-blinded clinical trials for both COVID-19 and infectious disease prevention in the USA relied on at-risk subjects who received four BCG vaccines with the potent Tokyo 174 strain of BCG, ¹¹ Vaccine dosing was 3 years prior to the pandemic, and treated subjects were matched with a BCG never-vaccinated group. The efficacy of protection was similar to, or better than, that provided by commercial SARS-CoV-2 mRNA vaccines. ¹² However, the extent of protection in these studies has been somewhat variable, perhaps reflecting the different vaccine subtypes in use^11,13,14 and/or differential underlying susceptibilities of the study participants including BCG vaccination status. ¹⁵ Nevertheless, in a combined analysis of all the BCG trials performed during the COVID pandemic, receipt of BCG versus placebo was associated with a 51% (95% CI: 15–72%) reduction in all-cause mortality. ¹⁶

For completeness, however, we stress that some studies on BCG specifically in COVID-19 found no benefits. Some negative studies reported lack of protection against adult COVID-19 following historical childhood BCG vaccination. However, protection is likely to wane with age: childhood vaccination is insufficient to protect against bladder cancer (where BCG is the standard of care), and re-instillation of BCG is essential. Nevertheless, the placebo-controlled trials BCG-Corona, BRACE, and BCG-PRIME found no benefits of recent vaccination.^17–20 We return later to why, in contrast to many conditions including dementia (below), there is no agreed consensus on whether immunopotentiation is protective in COVID-19.

In addition to reducing neonatal mortality, recent trials have indicated that BCG inoculation can also extend healthspan in the elderly, and moreover has protective effects against diverse conditions such as eczema, diabetes, and multiple sclerosis^21–24 that have also been linked to immunological/inflammatory factors.

For AD, the unusual protein deposits found in AD brain (amyloid-β, Aβ) are now considered to play a central role in antimicrobial defense. ²⁵ Although this is not yet universally accepted, and Aβ may well have other roles that remain to be uncovered, recent findings argue that the second abnormal protein found in AD brain—phosphorylated tau (p-Tau)—also has prominent antimicrobial activity, ²⁶ These findings begin to provide a framework for deciphering the pathoetiology of AD, a condition that is associated with neuroinflammation, where brain infection with diverse microbes including bacteria, fungi, and viruses may play a role in advancing the disease. ²⁷ Indeed, multiple pathogens are positively associated with AD development ( ²⁸ for brief review), and infections can increase AD odds by up to 30-fold. ²⁹ The novel idea that infections contribute to AD has met with skepticism, sometimes vituperative, but is becoming more widely accepted. If correct, then immunopotentiation through the non-specific effects of vaccines would be a viable option to combat the disorder.

Although historically the tuberculosis vaccine BCG has been a paradigm for trained immunity, in the sections following we highlight that many other vaccines, including herpes zoster, influenza, pneumococcus, and Tdap (tetanus, diphtheria, and pertussis), as well as associated adjuvants, may also have beneficial immunopotentiating effects.

Vaccines and AD

Growing evidence argues in support of the anti-dementia effects of vaccines. It was first reported in Canada 25 years ago that recipients of different vaccines were at reduced risk of developing AD. ³⁰ Interest resurfaced when it was found that AD rates in individuals treated with BCG as the standard of care for bladder cancer were reduced by half. ³¹ although a more recent study reported a lower level of protection, with 12% reduction across the BCG group and 27% reduction for individuals over 75 years of age. ³² Protective findings have now been broadly replicated around the world, although replication of observational associations does not establish causality. Nevertheless, significant protective effects have been reported for diverse types of vaccines, from diphtheria to influenza to pneumococcus, ²⁸ and, when given by injection, recipients appeared to be relatively protected from dementia even when followed for 15 years. ³³

Although there have been a small number of contradictory studies, earlier reports that influenza vaccine appeared to increase (rather than decrease) dementia progression have been overturned by a more recent study. ³⁴ Moreover, in a study of over 700 routinely administered medications, only four were protective against dementia, and all were vaccines. ³⁵ Vaccines against shingles, a painful skin condition caused by reactivation of varicella zoster virus (VZV) that often afflicts the elderly, also appear to be beneficial, although the modern Shingrix vaccine appeared to give better results than the earlier alternative, Zostavax, presumably due to differences in formulation. ³⁶ Two doses of Shingrix plus Zostavax were associated with the maximum protection against dementia reported to date (>80%), ³⁷ although this remains to be confirmed. Of note, there is no direct evidence so far that VZV is causally implicated in AD because antiherpetics that prevent/treat shingles have no effect on AD, suggesting that the vaccine acts independently of inhibiting VZV propagation, such as through trained immunity. Consistent with a possible dose–response relationship, recipients of high-dose influenza vaccination demonstrated lower AD incidence than recipients of standard-dose vaccination ( ³⁴ ; discussed later), and greater protection was observed in individuals in receipt of two vaccines (herpes zoster and Tdap) than either one alone. ³⁸ A recent meta-analysis concluded that adult vaccinations, particularly with herpes zoster, influenza, pneumococcus, and Tdap vaccines, are associated with a lower risk of dementia. ³⁹

We emphasize that the majority of the vaccine–dementia literature relates to standard observational epidemiology and thus cannot adequately account for unmeasured confounding, and therefore should be interpreted as correlational only. In addition, many studies have evaluated dementia or AD-related dementia outcomes rather than biomarker-confirmed AD specifically. Nevertheless, the consistency of findings across vaccine classes, populations, and study designs supports investigation of immunopotentiation as a potential strategy for AD prevention.

A further study in press in Communications Medicine ⁴⁰ has shown that BCG induces trained immunity-like changes in cerebrospinal fluid immune cells, including enhanced innate immune-responsiveness after ex vivo stimulation and sustained transcriptional reprogramming of central nervous system immune populations, together with longitudinal changes in AD-related biomarkers. Moreover, it has recently been reported that vaccines are also protective against early-onset dementia in individuals with Down syndrome (DS; trisomy for chromosome 21), a condition linked to immunodeficiency, where one half of individuals develop AD-like dementia by the age of 60 years. DS individuals in receipt of Tdap vaccine showed a 26% reduction in dementia onset during the follow-up period (median of 5.6 years). ⁴¹

Presentations at a recent Clinical Trials in AD Meeting (CTAD2025) in San Diego support the protective effects of multi-dose BCG vaccines in at-risk diabetic adults with long-standing disease. Although these 10-year randomized clinical trials are still underway, the 5-year data were recently unblinded from a large placebo-controlled clinical trial in the USA (NCT02081326). Six BCG inoculations over a 5-year period significantly improved AD biomarkers (p-Tau217 and Aβ42/Aβ40 ratio) in blood starting at about year 2 and persisting for at least 5 years. Because these are the same FDA-approved biomarkers that clinicians use to diagnose AD, it is predicted that AD incidence will also be reduced. However, although suggestive, this study has not yet been formally published and the full analysis is keenly awaited.

Other immunological strategies being explored in AD include immune checkpoint inhibitors (e.g., ⁴² ; NCT05551741), additives (‘adjuvants’) that are coadministered with antigen to enhance the immune response (reviewed in ⁴³ ), and the immune-boosting cytokine G-CSF. ⁴⁴ Potential immunological treatments distinct from vaccines were recently reviewed. ⁴⁵ By contrast, trials of monoclonal antibodies (mAbs) designed to remove Aβ have generally been unproductive. Recent studies on mAbs such as donanemab and lecanemab have reported some small benefits, but there is no evidence that Aβ removal is associated with slower disease progression. The most authoritative study recently concluded that the drugs yield trivial or clinically non-meaningful effects on cognitive decline and increase the risk of brain swelling and bleeding. ⁴⁶

Natural experiments

Many of the vaccine studies discussed above were not double-blinded placebo-controlled clinical trials. Although this does not necessarily negate their conclusions, they are potentially open to residual confounding such as ‘healthy vaccinee’ bias (caused by the general healthier profile of people choosing to be vaccinated) and bias caused by unmeasured confounders. Indeed, many important variables including socioeconomic status, medication adherence, frailty and cognitive ability, and lifestyle factors such as physical activity and diet, among others, are often absent from the public health datasets used in these studies. A legitimate concern is that a potential confounder might determine who receives the vaccine and who does not, which could distort the findings. Nevertheless, important corroborating evidence comes from natural experiments (where a variable is manipulated by nature, policy, or circumstance rather than by the investigators themselves) that eliminate or strongly reduce the risk of ‘healthy vaccinee’ bias and also substantially mitigate the risk of other unmeasured confounders.

Taquet et al. took advantage of the fact that there was an abrupt transition in the USA from the traditional live attenuated VZV vaccine (Zostavax) to the adjuvanted recombinant vaccine (Shingrix) in October 2017. They then compared dementia rates in those vaccinated before the transition date to vaccine recipients after the transition. Recipients of the recombinant vaccine were found to have a significantly lower rate of AD-related dementia over the follow-up period (6 years), with a 17% increase in time lived without a dementia diagnosis. ³⁶ Because both cohorts were shingles vaccine recipients, this study eliminated the healthy vaccinee bias. Moreover, because there was no abrupt change in the characteristics of people choosing to be vaccinated against shingles at the transition from the live to the recombinant vaccine, it strongly mitigates the risk of unmeasured confounding.

In a second such experiment, Eyting et al. exploited the finding that individuals in Wales (UK) who were born before 2 September 1933 have been ineligible for any zoster vaccine and remained ineligible for life. By contrast, individuals born on or after that date became eligible to receive the vaccine from age 70 onwards, and it was estimated that 47% of the latter group actually received the vaccine. The researchers compared dementia rates in the two groups (now in their 80 s) flanking the eligibility date and found a 20% reduction after normalization in the vaccine-eligible group. ⁴⁷ These findings have been corroborated in Canada and Australia using similar designs.^48,49

In a third study, Bukhbinder et al. compared AD rates in adults receiving high-dose (HD) versus low-dose (LD) inactivated influenza vaccine in the period 2014–2019. At the end of follow-up, the AD rate was ∼16% reduced in HD versus LD recipients. ³⁴ Because the patients (and probably even their physicians) did not know whether they were receiving the HD or the LD formulation, the authors concluded that the effect could not be attributed to healthy vaccinee or similar bias.

Although these studies rule out healthy vaccinee effects and mitigate related biases, residual confounding cannot be excluded. For example, in the study by Taquet et al. Shingrix and Zostavax were administered during different calendar periods (7 months apart), raising the possibility of systematic differences between recipients. However, baseline differences were small and likely reflect secular trends in shingles vaccination uptake rather than preferential receipt of one vaccine over the other. Moreover, dementia rates among Tdap recipients were unchanged across the transition period (M. Taquet, unpublished data), which further militates against this explanation. Likewise, in the study by Eyting et al. there is an assumption that dementia rate increases linearly with age, and this might not be fully met in practice. Nevertheless, the results were robust when different age-windows were used to capture the discontinuity, which argues against this source of bias. The Bukhbinder et al. study also argues against potential bias in the recipient groups. The strongest evidence in our view comes from the two natural experiments with VZV vaccines. Although it could be held that the two independent studies differ in evidentiary strength, we believe that their combined weight is more persuasive than either study alone. Despite using two very different approaches with distinct caveats, they reached the same conclusion—that shingles vaccination is likely to be protective against dementia. Nevertheless, in future studies it is essential that researchers remain vigilant of potential confounding when interpreting their data.

Conclusions

These findings prompt the following general conclusions. First, despite caveats that some studies are purely observational, many vaccines, irrespective of their specific type, do appear to offer a degree of protection against dementia. Nevertheless, the different vaccines have different efficacies and may operate by distinct mechanisms. Trained immunity has broad effects across myeloid cells, natural killer cells, and innate lymphoid cells, as well as on pathogen-recognition and inflammatory mechanisms that affect both innate and adaptive immunity. ⁴ Indeed, BCG inoculation is known to boost both antibody responses and innate immunity, and some vaccines (e.g., Tdap and pneumococcus) may plausibly operate through partly distinct mechanisms involving adaptive as well as innate immunity. Moreover, the immune system has been implicated in brain processes such as axon regeneration following injury and vascular remodeling, particularly during sleep ( ⁵⁰ for review), but conversely may sometimes act to aggravate inflammation and prevent regeneration, ⁵¹ all of which could be modulated by vaccination.

The mechanism of action is an important consideration. Although multiple conditions including neonatal mortality, eczema, diabetes, multiple sclerosis, and dementia appear to be broadly amenable to vaccine-elicited protection, there is no consensus regarding protection against COVID-1. However, this disease may be an exception because much of the pathology is caused by overexuberant inflammatory responses, and host-induced over-reaction is ultimately the cause of morbidity and mortality. ⁵² This highlights the fact that non-specific immunopotentiation may potentially be counterproductive in cases of excess (rather than insufficient) immune activation.

Second, irrespective of the exact mechanism of protection, more vaccine doses appear to offer more protection, and the highest level of protection against dementia reported to date was for three doses of shingles vaccine (see earlier), and greater protection was observed in individuals in receipt of two vaccines (herpes zoster and Tdap) than either one alone. ³⁸ In addition, six doses of BCG vaccine in a randomized clinical trial showed for the first time an improvement in AD biomarkers indicative of AD prevention. Nevertheless, because health-engaged individuals are more likely to receive multiple vaccines and may have a lower baseline dementia risk, any potential dose–response relationship will need to be validated through placebo-controlled trials.

Third, the effect sizes are clinically highly meaningful, and relative AD risk reductions of up to 50% were observed within 5 years of administration (reviewed in ²⁸ ), although such studies are prone to bias and the extent of protection remains to be confirmed in controlled trials. However, in terms of cost, most of these vaccines can be obtained for a few dollars, although others are a little more expensive. But the cost of vaccination is entirely trivial compared to the financial burden of AD.

Several countries have already initiated programs of vaccination of the elderly against diseases such as influenza, but there is a need to assess which specific vaccines, subtypes, and associated formulations are the most protective. In addition, the beneficial effects of immune boosting may be modulated by factors such as race, age, sex, and genotype; further research in this direction will be invaluable. We acknowledge that controlled clinical trials will be essential to exclude potential confounding factors and demonstrate the efficacy of immunopotentiation against AD.

There are of course many uncertainties. From a strategic perspective, the identification of the underlying microbes, inflammation pathways, and/or defense mechanisms, and how are they are dependent on specific types of vaccines, would be an important step to optimize vaccines as an anti-AD strategy. There is also a limitation in that many of the cited studies do not clearly distinguish between different types of dementia. Dementia is not a unitary condition and includes AD, frontotemporal dementia, vascular dementia, Parkinson's disease dementia, Lewy body dementia, and mixed dementia. AD itself has been classified into different subtypes. It remains to be established whether vaccination offers comparable benefits in different types of dementia. Nevertheless, the finding that dementia onset in Down syndrome is reduced by Tdap vaccine suggests that the benefits may extend beyond strictly defined AD. Future studies should aim to clearly distinguish between dementia subtypes and report subtype-specific outcomes. It will also be necessary to elucidate the roles of the vaccine itself versus that associated with adjuvants. In addition, studies to date have only addressed prevention, and it is not yet known whether immunopotentiation can combat already established disease, although there are hints that it may. ⁵³

We recognize that there is unlikely to be significant interest from the pharma sector because most preparations are cheap and off-patent, and the prospects of big financial rewards to manufacturers are slim, irrespective of the potential benefits to society.

Nevertheless, the accumulated data are in our view so compelling that we urge the relevant authorities worldwide to give their fullest consideration to placing immune boosting at the top of the AD public health R&D agenda.