The virus–Alzheimer’s tug of war continues. New data across several studies weaken the proposed, and much-debated, association; its proponents are holding fast.

  • Short-term anti-herpes drug use slightly lowered dementia risk.
  • Infection without antiviral treatment slightly upped risk.
  • But results not consistent across European countries.

A new epidemiology study reports a weak link between herpes and dementia. Researchers led by Richard Lathe and Jürgen Haas at the University of Edinburgh combed through four European population-wide healthcare databases and, in a medRxiv preprint uploaded December 4, 2020, describe equivocal data. In Denmark and Wales, short-term antiviral drug use came with slightly fewer future dementia cases. Alas, in Germany and Scotland, this association did not hold. The opposite was also true; infected people who were not prescribed an antiviral had a slightly higher risk of dementia—but only in the German cohort. “The results are not very encouraging,” Lathe wrote to Alzforum.

Some of these associations held no matter what type of dementia or virus was considered. “Because neither dementia subtype nor herpes subtype modified the association, the small but significant decrease in dementia incidence with antiherpetic administration may reflect confounding and misclassification,” the authors concluded.

Herpes Simplex. [Image courtesy Swetha Vijayakrishnan et al., Scientific Reports 2020]

Twenty-five years ago, Ruth Itzhaki linked herpes simplex virus type 1 (HSV-1) infection with an increased risk of AD (Itzhaki et al., 1997). She later spotted the virus hiding in amyloid plaques in brain tissue, and postulated that it may trigger the deposits (Wozniak et al., 2009). Since then, other connections have emerged. Scientists linked viral DNA in the brain to expression changes of genes involved in amyloid metabolism; others proposed that amyloid acts as an antimicrobial (Jun 2018 newsMay 2020 news).

Still, compelling evidence that viruses, particularly herpesviruses, cause AD remains elusive, although some researchers believe that the teeny irritants could speed disease along (Jul 2019 conference news). The debate has taken on a new sense of urgency since reports that COVID-19 causes long-term neurological symptoms in a fraction of people who contract the disease (Pilotto et al., 2021). Scientists are just beginning to study this aspect of the infection (see Jan 2021 news).

Previous population studies have been conflicting, and this latest one is no different. Co-first authors Christian Schnier and Janet Janbeck, University of Copenhagen, combed through health data from the Welsh Secure Anonymised Information Linkage (SAIL) Databank, the German IMS Disease Analyzer, the Danish National Registry (DNR), and the Scottish Electronic Data Research and Innovation Service (eDRIS).

They calculated, at best, a faint connection between herpes infection and subsequent dementia. In the Danish and Welsh cohorts, people who were prescribed at least one dose of herpes antiviral medication had up to an 11 percent decrease in dementia risk, with the two other cohorts showing no altered risk. It did not matter which antiherpetic drug they took, although acyclovir and valacyclovir were most prescribed. Herpes subtype also had no bearing on dementia outcomes.

“The fact that they saw any protection at all from extremely short durations of antiherpetic medications is quite interesting,” Joel Dudley, Icahn School of Medicine at Mount Sinai, New York, wrote to Alzforum.

The opposite was true too, but only in the German cohort. Namely, people with a herpes infection who were not prescribed an antiherpetic drug had an 18 percent higher risk of dementia compared to people who had no medical history of a herpes infection. People untreated for shingles, which is caused by the varicella zoster herpes virus, were 20 percent more likely to develop dementia, whereas those with a history of HSV-1 were just as likely to be diagnosed with dementia as controls.

Controlling for socioeconomic status (SES), including level of education, quality of health insurance, or poverty level, did not change the dementia risk in any country. Ben Readhead, Arizona State University, Tempe, thought this was a strength of analyzing multiple diverse cohorts. “The persistent protective effect after accounting for these heterogeneous SES measures suggests the primary effect is not obviously confounded by SES,” he wrote.

Schnier and colleagues were more cautious in interpreting the signal. To their mind, inconsistencies across cohorts, and similar trends for both AD and vascular dementia risk despite their having different neuropathologies, indicate that confounding factors are to blame.

Readhead warned against overemphasizing associations with dementia subtypes because these were likely based on clinical diagnoses, which can be unreliable. Subtypes often co-occur in the same person and can be misclassified. “I found the authors’ interpretation a little at odds with the results, which I thought were quite encouraging for at least a modest protective effect of antiherpetic medications on dementia diagnosis,” he wrote.

“Much like in previous meta-analyses, their conclusions were mixed,” wrote Rudolph Tanzi and William Eimer, both at Massachusetts General Hospital, Charlestown. “Heterogeneous results are not terribly surprising given the complex nature of AD etiology and pathogenesis, which, so far, does not exclude an infectious component.”

A Tenuous Signal Emerges
Lathe’s new findings somewhat jibe with three prior studies of the Taiwanese national healthcare database. They found a 2.5 to three times higher risk of dementia associated with severe herpes simplex virus (HSV) infection or severe shingles involving the eye (Tsai et al., 2017; Chen et al., 2018; Tzeng et al., 2018; for commentary, see Itzhaki and Lathe, 2018). Acyclovir and valacyclovir dramatically diminished the odds of getting dementia. Treatment for fewer than 30 days reportedly reduced risk by 40 percent and longer treatment by a whopping 90 percent.

Was this too good to be true? “The Taiwanese results were extremely puzzling because antivirals are quite short-lived in the body, so explaining a long-term effect was difficult,” Itzhaki wrote to Alzforum.

One possible explanation why the associations may not have held in the European cohorts could be treatment duration. In Taiwan, people took antivirals for 90 days on average, whereas most Europeans took them for fewer than 30 days.

More recent studies in Asia and Europe reported weak associations between microbes and dementia. In South Korea, older adults diagnosed with shingles were 12 percent more likely to develop dementia than those who weren’t, and antivirals reduced risk of dementia by 24 percent (Bae et al., 2020). In three Finnish population cohorts and in the U.K. Biobank, hospitalization for any viral or bacterial infection increased dementia risk 1.6-fold. Intriguingly, herpes and Gram-negative bacterial infections popped up as the main culprits, increasing dementia risk two- and 1.6-fold, respectively. Severe infections were more closely tied to subsequent vascular dementia than to AD (Sipilä et al., 2020).

However, not all population studies of herpes and dementia have found a link. Last year, a meta-analysis found inconsistencies (Warren-Gash et al., 2019). Some studies weakly associated HSV, cytomegalovirus (CMV), and human herpesvirus 6 (HHV-6) with future dementia but other studies did not. Higher-quality evidence is needed, claimed the authors.

Biases and confounds are major sticking points for epidemiological studies. A recent genetic analysis study using Mendelian randomization, a methodology designed to eliminate such confounds, searched several genome-wide association study (GWAS) datasets for links between people’s genetic susceptibility to viral infection since birth and dementia or cognitive decline late in life. As a proxy for the former, the scientists searched for single-nucleotide polymorphisms associated with infection. The authors found 129 of these SNPs among the more than 700,000 participants, of whom 300,486 came from the CHARGE, COGENT, and UK Biobank GWAS of cognition, and 455,258 came from the PGC-ALZ, IGAP, and ADSP GWAS of late-onset AD. Importantly, the researchers found no association between a person’s genetically predicted odds of herpes infection and subsequent cognitive decline or AD (Kwok and Schooling, 2020).

Neither this nor any of the other studies took ApoE genotype into account, even though APOE4, the strongest genetic risk factor for late-onset AD, has been linked to more robust immune responses against invading herpes viruses (Jul 2019 conference news). “HSV-1 infection tends to cause worse damage in those with the APOE4 allele,” Itzhaki said.

AD Antiviral Trials
Itzhaki noted that the epidemiology studies did not investigate the effect of antivirals on people with dementia. “People were treated years before dementia developed; it is a different matter than treating them when they are symptomatic,” she said. Indeed, multiple clinical studies are now evaluating if long-term antiviral treatment can slow cognitive decline in people who already have it.

Two Phase 2 trials, one in Sweden and one in the U.S., tested daily valacyclovir for one and 18 months, respectively (Devanand et al., 2020).

The Swedish trial wrapped up in March 2020. “We have completed CSF and plasma analyses and plan to submit a manuscript of the main results within the next few months,” Hugo Lövheim, Umeå University, Sweden, wrote to Alzforum. The American trial was held up by COVID-19 but is now back on track. “Recruitment will hopefully finish this year, then we will follow the last participant for 18 months,” Davangere Devanand, Columbia University, New York, wrote.

A Phase 2 trial in Poland tested the picornavirus antiviral pleconaril as an add-on to acetylcholinesterase inhibitors or memantine for 12 months (Nov 2018 conference news). That trial, sponsored by Apodemus AB in Solna, Sweden, ended prematurely in March 2020. Issues with data quality led to legal charges against the contract research organization managing the trial. “As of now, we have no plans to restart,” Nina Lindblom of Apodemus told Alzforum.

More funding might help settle the question of what role, if any, viruses play in AD. Last year the Infectious Diseases Society of America Foundation awarded $100,000 to each of 10 researchers studying how microbes are linked to AD (press release). Earlier this month, the National Institute on Aging opened an R01 grant funding opportunity  for the Infectious Etiology of Alzheimer’s Disease. It comes on top of the more than $8 million in research grants for ongoing studies of herpesviruses and dementia listed on the NIH RePORTER. Finally, anyone “who provides persuasive evidence that an infectious agent is the root cause of Alzheimer's disease,” can claim the $1 million prize offered by Leslie Norins,, Naples, Florida.—Chelsea Weidman Burke


  1. In this study Schnier et al. examined four different European observational cohort studies for association between herpesvirus infection and dementia. Expanding upon the conclusions of Warren-Gash et al. that current published evidence is insufficient to draw conclusive associations between human herpesvirus infections and dementia, Schnier et al. refined their target to HSV1/2 or zoster ophalmicus (VZV) and short-term treatment with antiherpetic medications. Much like previous meta-analyses, their conclusions were mixed. Some study populations displayed reduced incidences of dementia with any treatment, and other study populations had no increased risk of dementia with any herpes infection independent of treatment.

    The authors carried out a significant refinement of target populations from the study pools and accounted for covariates with what data is available. One concerning statistic is the lower percentage of their study population being identified as herpes simplex positive. The two studies listing this information listed 14 percent and 39 percent of the total population, but seroprevalence studies in Europe of individuals 40+ years of age place those percentages closer to 60-80 percent depending on the country. Consequently, this forces this study into a more restricted examination of association, focusing on impacts of active viral infection and short-term application of anti-herpetics.

    The heterogeneity of the results is not terribly surprising given the nature of Alzheimer’s disease. Continued focus is being given to detection of amyloid or tau well before symptoms appear in patients. This application is likely no different for history of both anti-viral treatment and viral infections, i.e. decreased risk with long term anti-viral use (as observed in the Taiwanese epidemiological studies) and increased risk with long term reactivating herpes infections (as observed in some but not all IgG seropositivity studies).

    These results are further complicated by our own studies examining the amyloid beta protein as an antimicrobial peptide (AMP) of the brain’s innate immune system. Amyloid’s AMP activities against bacterial, fungal, and viral pathogens (including HSV1) open the possibility that no single pathogen is causative for AD, and that any sufficient insult to the brain’s immune system (mono or polymicrobial) may be sufficient to drive AD etiology. And, of course, genetic susceptibility factors may preclude the need for microbes and infection in many AD cases.

    In summary, Schnier et al. present important new insights into the growing number of population studies of the association of microbial pathogens with risk for Alzheimer’s disease. While no firm conclusions concerning new pathogenic covariates in AD can yet be drawn, this study further highlights the need for focused, large-scale studies of the complex nature of AD etiology and pathogenesis, which, so far, does not exclude an infectious component.

  2. Schnier et al. have performed a careful and expansive study examining the protective effects of antiherpetic medication upon subsequent dementia diagnosis. Following recent studies that reported the protective effects of antiherpetic medications on subsequent development of dementia (Chen et al., 2018Tzeng et al., 2018Bae et al., 2020), evaluating whether these associations persist in other cohorts and using other formulations of related questions is of great interest to many in the field.

    The authors report a quite consistent, modest protective effect of antiherpetic medications upon subsequent diagnosis of dementia. A critique of these associations historically has been that access to antiherpetic medication is a proxy for better access to healthcare and higher socioeconomic status (SES), both of which could drive lower rates of incident dementia. The authors used several proxies for SES depending on data availability across cohorts, including multiple deprivation index, insurance status, civil status, and education. Though each of these measures is an imperfect proxy for SES, the persistence of a protective effect when accounting for these heterogeneous measures collectively suggests that the primary effect is not obviously confounded by SES, and highlights the strength of the authors approach looking across several cohorts with diverse data structures.

    Overall, I found the interpretation a little at odds with the results, which I thought were quite encouraging for at least a modest protective effect of antiherpetic medications on dementia diagnosis. The authors note that these associations are not modified by consideration of dementia subtype (Alzheimer's disease (AD) and vascular dementia (VaD)) or duration/frequency of antiherpetic medication dosage, and suggest that this might indicate unobserved confounding factors driving the appearance of a false protective effect.

    There are limitations to whether these aspects of the data can be used in this manner however. Within these data, diagnosis of AD or VaD is likely to be made in the course of clinical management, not as the result of post-mortem neuropathological evaluation. Although "pure" AD and VaD neuropathologies have very distinct pathomechanisms, they are frequently comorbid, and misclassification on clinical grounds alone is very common. For example, Nolan et al. reported that among dementia-clinic patients enrolled in a prospective study of VaD who later came to autopsy, 87 percent were subsequently diagnosed either with AD alone, or with comorbid AD and VaD (Nolan et al., 1998).

    Other autopsy series also support the notion that VaD in the absence of concomitant AD neuropathology is uncommon (Hulette et al., 1997). This suggests that accurate classification of VaD from clinical records alone may not currently be possible, which would weaken inferences that might follow from the absence of a dementia-subtype specific interaction reported in this study. An alternative interpretation is that the authors demonstrated the same dementia-protective effects of antiherpetic medications in two patient subpopulations that are likely to represent an admixture of dementia subtypes, with AD neuropathology likely to predominate.

    Inferences about the absence of a modifying effect of anti-herpetic dose duration may have similar limitations. As the authors note, approximately 1 percent of subjects received three or more prescriptions of antiherpetic medication. Given that these treatment regimens are typically 1-2 weeks, and require primary care review, it is likely that subjects requiring multiple treatments have experienced a commensurately more severe or persistent viral perturbation, which may obviate the otherwise beneficial effects of longer treatment exposure.

    It's interesting to note the distribution of antiherpetic medication treatment durations between these cohorts and those reported by Tzeng et al. (Tzeng et al., 2018), where prolonged treatment protocols seem to have been more frequent. To the extent that they exist, it may be fruitful to examine dosage/duration dependent protective effects among populations with more permissive antiherpetic protocols, particularly if it allows the capture of potential dementia protective effects in subjects with prolonged drug exposure, but only moderate viral pathology.


    . Herpes Zoster and Dementia: A Nationwide Population-Based Cohort Study. J Clin Psychiatry. 2018 Jan/Feb;79(1) PubMed.

    . Anti-herpetic Medications and Reduced Risk of Dementia in Patients with Herpes Simplex Virus Infections-a Nationwide, Population-Based Cohort Study in Taiwan. Neurotherapeutics. 2018 Apr;15(2):417-429. PubMed.

    . Absence of vascular dementia in an autopsy series from a dementia clinic. J Am Geriatr Soc. 1998 May;46(5):597-604. PubMed.

    . Clinical-neuropathologic findings in multi-infarct dementia: a report of six autopsied cases. Neurology. 1997 Mar;48(3):668-72. PubMed.

    . Association of herpes zoster with dementia and effect of antiviral therapy on dementia: a population-based cohort study. Eur Arch Psychiatry Clin Neurosci. 2020 Jul 1; PubMed.

  3. The viral vector hypothesis of Alzheimer’s disease has been one of the most provocative and promising research avenues in this field (Doty 2008; Haas and Lathe 2018). With the sequential failures in finding effective treatments for AD in the past 20 years, the prospect of repurposing drugs in individuals with past-recurrent herpetic infection has become particularly attractive, but it has failed to return strong and conclusive evidence in the clinical settings.

    Early preclinical studies demonstrated the ability of herpesviruses to colonize the brain, causing amyloid-beta aggregation (Itzhaki et al. 1997; Wozniak, Mee, and Itzhaki 2009). Most recently, HHV6 traces were also found in AD brain specimens (Readhead et al. 2018). However, this was recently disputed by a large post-mortem brain screening finding no traces of either HHV6 or HHV7 in AD and by a re-analysis of the original data fueling further the dispute (Allnutt et al. 2020; Chorlton 2020).

    Furthermore, ApoEe4 carriers with higher susceptibility for herpes infections are also at higher risk of dementia (Itzhaki et al. 1997), establishing a positive genetic predisposition link. Most recently a mechanistic association has been demonstrated showing that amyloid-β functions as an agglutinating anti-microbial trap, indirectly explaining the positive association between herpes-virus infection and AD pathogenesis and inspiring further investigations of causative link (Eimer et al. 2018).

    Turning away from preclinical investigations, in the last 3-4 years, observational studies in large population cohorts have interrogated the risk of dementia conversion in aged individuals with herpetic outbreaks and estimated the protection conferred by the treatment with the antiherpetic medications. Three Taiwanese studies were the first independent reports analyzing retrospectively the effect of herpetic outbreaks on dementia through the Taiwan National Health Insurance Database. The first study assessed the increase in dementia risk for 846 subjects with Herpes Ophtalmicus (HR=2.83) as compared to more than 2500 controls without, considering medication (Tsai et al. 2017). The second study analyzed 39,000 reported cases of Herpes Zoster infections from 50 to 90 years old and follow-up visits of about 6 years (Chen et al. 2018). The group observed a very small but significant risk of converting to dementia as a general condition (HR=1.1) in subjects with reported Herpes Zoster, which was substantially curbed by the use of antiviral therapy (HR=0.55).

    A comparable result was obtained very recently by a South Korean group who analyzedg more than 200,000 individuals aged more than 50, using the National Health Insurance Service Cohort of South Korea, and who were followed up for about 10 years (Bae et al. 2020). Another report from Taiwan in 2018, analyzing 8,362 subjects with reported Herpes Simplex over 10 years showed an even stronger risk factor for the conversion to dementia (HR=2.6) with comparable risk for AD and Vascular dementia, which was significantly reduced by the use of AVT (HR=0.09) (Tzeng et al. 2018). None of these studies reported the type or duration of ATV medication, limiting the clinical applicability of the data.

    The present retrospective observational cohorts’ study from Schnier is the first work using multicentric health records from four European regions (SAIL, Wales; Disease Analyzer, Germany; DNR, Denmark and eDRIS, Scotland) to study the association between herpes infections and the risk of dementia considering the largest population ever examined ( more than 2.5 million individuals). The Health records were chosen to be comparable in terms of demographics, years of follow up visits (1937-1953), type of datasets warranting the highest reproducibility, and increased the ability to detect confounders. Furthermore, it’s the first report analyzing at the same time the effect of the most common types of herpes infections (Herpes Simplex and Herpes Zoster) on dementia as an outcome, and further stratifying the risk in respect to the two most prevalent form of dementia subtypes (AD and VaD) (SAIL and Disease Analyzer). It also considers the type of medication and the duration of treatments, which was never accomplished before in such a systematic way.

    Overall only the Disease Analyzer across the 4 datasets analyzed indicated an increased risk of dementia in subjects with herpes virus infection, with Herpes Zoster representing a higher risk (HR=1.2) than Herpes Simplex, which could be explained also by the extent and clinical consequences of Herpes Zoster outbreaks especially in the elderly (Bowsher 1999). Nevertheless, no casual association could be confirmed as treatment showed no reversal of dementia. Overall, with no substantial effect of herpetic infection, antiherpetic medication (Acyclovir versus alternative treatment), or its duration, lead to significant differences.

    This study confirms the evidence-based skepticism of a recent meta-analysis of 57 papers, reporting no significant causative association between herpesviruses and dementia, which claimed the effect too small to be relevant after adjusting for confounders, and questioned whether the effects observed in the Taiwanese or Korean studies have been appropriately controlled or interpreted (Warren-Gash et al. 2019). Overall, observational retrospective analyses are never ideal, even if a large number of patients can be included. Informative biological data, such as APOEe genotype or inflammatory grade, is often missing, which would be essential to evaluate the effect of low-grade herpetic infection over a lifetime (Lövheim et al. 2015). At the same time, the dementia diagnosis in none of these studies provides any information about the staging of dementia, which may influence the outcome of the results.

    Finally, the medication should be also considered as patients may have comorbidities altering the effect of antiherpetics and the outcome. That said, this is the best observational evidence to date investigating and negating casual associations between herpesviridae and the risk for dementia. With increased biological depth, self-reports through the spread of mobile technologies, and matched health records, herpetic subjects have the opportunity to actively contribute to the interoperable databases that should increase the depth and insight of such retrospective analysis in the future.


    . Human Herpesvirus 6 Detection in Alzheimer's Disease Cases and Controls across Multiple Cohorts. Neuron. 2020 Mar 18;105(6):1027-1035.e2. Epub 2020 Jan 23 PubMed.

    . Association of herpes zoster with dementia and effect of antiviral therapy on dementia: a population-based cohort study. Eur Arch Psychiatry Clin Neurosci. 2020 Jul 1; PubMed.

    . The lifetime occurrence of Herpes zoster and prevalence of post-herpetic neuralgia: A retrospective survey in an elderly population. Eur J Pain. 1999 Dec;3(4):335-342. PubMed.

    . Herpes Zoster and Dementia: A Nationwide Population-Based Cohort Study. J Clin Psychiatry. 2018 Jan/Feb;79(1) PubMed.

    . Reanalysis of Alzheimer's brain sequencing data reveals absence of purported HHV6A and HHV7. J Bioinform Comput Biol. 2020 Feb;18(1):2050012. PubMed.

    . The olfactory vector hypothesis of neurodegenerative disease: is it viable?. Ann Neurol. 2008 Jan;63(1):7-15. PubMed.

    . Alzheimer's Disease-Associated β-Amyloid Is Rapidly Seeded by Herpesviridae to Protect against Brain Infection. Neuron. 2018 Jul 11;99(1):56-63.e3. PubMed.

    . Microbes and Alzheimer's Disease: New Findings Call for a Paradigm Change. Trends Neurosci. 2018 Sep;41(9):570-573. Epub 2018 Jul 19 PubMed.

    . Herpes simplex virus type 1 in brain and risk of Alzheimer's disease. Lancet. 1997 Jan 25;349(9047):241-4. PubMed.

    . Reactivated herpes simplex infection increases the risk of Alzheimer's disease. Alzheimers Dement. 2015 Jun;11(6):593-9. Epub 2014 Jul 17 PubMed.

    . Multiscale Analysis of Independent Alzheimer's Cohorts Finds Disruption of Molecular, Genetic, and Clinical Networks by Human Herpesvirus. Neuron. 2018 Jul 11;99(1):64-82.e7. Epub 2018 Jun 21 PubMed.

    . Increased risk of dementia following herpes zoster ophthalmicus. PLoS One. 2017;12(11):e0188490. Epub 2017 Nov 22 PubMed.

    . Anti-herpetic Medications and Reduced Risk of Dementia in Patients with Herpes Simplex Virus Infections-a Nationwide, Population-Based Cohort Study in Taiwan. Neurotherapeutics. 2018 Apr;15(2):417-429. PubMed.

    . Human herpesvirus infections and dementia or mild cognitive impairment: a systematic review and meta-analysis. Sci Rep. 2019 Mar 18;9(1):4743. PubMed.

    . Herpes simplex virus type 1 DNA is located within Alzheimer's disease amyloid plaques. J Pathol. 2009 Jan;217(1):131-8. PubMed.

  4. Retrospective observational data analysis such as the kind performed in this study is known to be quite difficult due to the challenging set of assumptions and confounders that need to be accounted for. I commend these authors for taking on this challenging work and for asking important questions from these large and interesting data sets.

    My interpretation of the findings is quite different from the measured tone of the authors. Mainly, I am surprised there was any observed association between antiherpetic medications and dementia risk. This tells me that the association must have a substantial effect to rise above the many issues with this type of analysis. e.g.,

    • This type of study is very sensitive to the phenotypic definition of dementia being used to define cases. It is well known that creating "electronic phenotypes" for dementia from diagnosis codes is an unsolved challenge. The authors of this study did a commendable job of addressing that problem, but much remains to be improved in this area. So that is one significant source of confound for the signal.
    • The propensity score matching used in this analysis appears to be quite simplistic; it does not appear that the authors took a sophisticated approach to control for non-dementia comorbidities and additional medications such as anti-hypertensives, diabetes medications, statins, etc.
    • My baseline hypothesis would be that a long duration of antiherpetic medication exposure would be required to see any effect, such as was found in the Taiwanese prophylaxis study. The fact that we saw any protective signal at all of anti-herpetic medications in this study, which evaluated extremely short durations of antiherpetic medications, is quite interesting.

    It is unclear to me how the inclusion criteria affected the recording of anti-herpetic medication exposure. One would expect that any duration of anti-herpetic exposure prior to age 65 would be expected to have a protective effect. Indeed, one hypothesis would be that anti-herpetic medication exposure much earlier in disease pathophysiology would be more effective. So I would like to see risk differences propensity-score matched cohorts of people receiving anti-herpetics in age ranges of 40s and 50s.

    As noted by the authors, the inability to control for APOE genotype status is a substantial limitation of this study. I would argue that even a weak signal for antiherpetic medication, despite not being able to control for APOE, is quite interesting. Previous studies suggest that the relationship between neurological viral load and APOE status is quite significant. Another limitation of this study not mentioned by the authors is that they were quite restricted in their ability to designate herpes infection status. Previous studies implied that HHV6 and HHV7 are among the herpes species most strongly associated with dementia pathology. As these herpes variants are not regularly diagnosed or recorded in clinical records, it's likely that this has a significant confounding effect on the results and interpretation.

  5. This paper is of interest and underscores the need to increase our knowledge on herpesvirus implications in dementia. A point that might be considered is also the classification of the patients on the basis of their genetic background.

    We have previously demonstrated that a gene, KIR2DL2, which is expressed as a receptor on the surface of natural killer cells, modifies the susceptibility to herpesvirus infection and might have a role also in the response to anti-viral therapies (Rizzo et al., 2020). The subdivision of the population on the basis of this gene, which can be present or absent in the individual genome, might increase the accuracy of the results obtained.


    . Controversial role of herpesviruses in Alzheimer's disease. PLoS Pathog. 2020 Jun;16(6):e1008575. Epub 2020 Jun 18 PubMed.

  6. I must challenge the statement “New data across several studies weaken the proposed, and much-debated, association” of virus–Alzheimer’s, as the score is now about several hundred for strengthening the association and about 5 for weakening it. The Schnier et al. paper is not one of the latter as, in at least two respects, when comparing data with the relevant Taiwan papers, it does not compare like with like: firstly, in examining the risk of AD/dementia in those infected with HSV1, and secondly in comparing the risk after antiviral treatment for HSV1.

    A major feature (which could not be considered as there were no APOE genotypes in the databases) is the effect of APOE-e4, as the Alzforum report mentions. There is good evidence that this allele, in combination with HSV1 infection, is a risk for AD. Thus, those HSV1-infected subjects who are APOE-e4 carriers would be at greater risk of AD/dementia than the HSV1-infected group as a whole, who, in contrast, would show little or no risk because the proportion of APOE-e4 carriers is low anyway and decreases with old age.

    As APOE-e4 is a major risk for herpes labialis (cold sores) (accounting for about 65 percent of cases- Itzhaki et al., 1997), cold sore sufferers are presumably at high risk of AD. This would explain the large hazard ratio (HR) of 2.56 found by Tzeng et al. for their HSV1-infected subjects who, having been selected on the basis of at least three outpatient visits within the index year, must have had severe cold sores, and so, presumably, would have been mainly APOE-e4 carriers. ApoE would explain also the difference in magnitude of Tzeng et al.’s result from that of Schnier et al., who found only a slight increase in risk of AD/dementia (HR 1.18) for their HSV1-infected subjects (those not selected for cold sores, and not treated with an antiviral), compared with subjects not examined for HSV1 but most of whom very probably were infected.

    As for antiviral treatment, the difference between the results of Tzeng et al. and Schnier et al. on the effect of antiviral treatment of HSV1 on risk of subsequent AD/dementia is very probably a result of the large difference in duration and frequency of treatment: in Taiwan, many were treated for at least 30 days, whereas those investigated by Schnier at al. had merely a single dose - which the authors mention but only in passing.

    In the case of herpes zoster (shingles) sufferers, Bae and colleagues found a clear risk of AD/dementia, and a significantly lower risk in those treated with antivirals (Bae et al., 2020). As to whether an allele of APOE confers a risk of shingles is uncertain: Pirtilla and colleagues found that no APOE allele increased the risk (Pirtilla et al., 2000). In contrast, my group (in one of several studies on diseases in which the pathogen involved shared cell surface receptors with a specific ApoE isoform and thus might compete for cell entry), found that women homozygous for APOE-e4 had a higher-than-normal risk of shingles (Wozniak et al., 2007). However, both studies had only a small number of participants so they need to be repeated. As to the risk of AD/dementia conferred by shingles, although it suggests a viral role it might not necessarily indicate that varicella zoster virus (VZV) is present in the brain and causing damage on reactivation, since an early PCR study did not detect VZV DNA in brain, instead, it might reflect shingles-induced inflammation causing reactivation of HSV1 in brain (Lin et al., 1997).

    It is worth mentioning that epidemiological studies on VZV are relatively straight-forward, as shingles is an easily diagnosed illness, whereas with HSV1, cold sores if mild are often not reported to a physician, let alone diagnosed; also, they can vary in frequency, severity and duration and quite often are treated only by the sufferer. Such details would not be recorded in a databank, so cold sores are scarcely suitable as markers of HSV1 infection for epidemiological studies (unlike shingles – good makers for VZV), especially as they affect only some 25-40 percent of HSV1-infected people; the remaining 60-75 percent are asymptomatic even though infected.

    Importantly, studies on cold sores have the further disadvantage that it is not known whether they indicate reactivations of latent HSV1 in the brain as well as of latent HSV1 in the periphery; a similar uncertainty exists also about whether serum antibodies, which indicate the presence of HSV1 antigens in the periphery, indicate also activity of HSV1 in the brain.


    . Herpes simplex virus type 1 in brain and risk of Alzheimer's disease. Lancet. 1997 Jan 25;349(9047):241-4. PubMed.

    . Apolipoprotein E (APOE) phenotype and APOE concentrations in multiple sclerosis and acute herpes zoster. Acta Neurol Scand. 2000 Aug;102(2):94-8. PubMed.

    . Does apolipoprotein E determine outcome of infection by varicella zoster virus and by Epstein Barr virus?. Eur J Hum Genet. 2007 Jun;15(6):672-8. Epub 2007 Mar 14 PubMed.

    . Neurotropic viruses and Alzheimer's disease: a search for varicella zoster virus DNA by the polymerase chain reaction. J Neurol Neurosurg Psychiatry. 1997 Jun;62(6):586-9. PubMed.

  7. Many thanks to Chelsea Weidman Burke for a great summary of our research and to you, colleagues, for excellent comments.

    One of the first things Chelsea asked us was to comment on the heterogeneity of results between the four cohorts, and our response was very similar to Eimer and Tanzi's comment: It's not terribly surprising. There are substantial differences in the way different national health systems work; for example, when and how people are diagnosed (with dementia and herpes), when and how people are medicated, and how the information ends up in electronic health records. Indeed, if anything, we were surprised in how little variation there was—whichever way we looked at the data, there was no strong, convincing argument for a “substantial” association between exposure to antiherpetic medication and subsequent dementia. We did find some “significant P-values,” but the interpretation of P-values in large cohort studies such as ours is problematic. Thus, this was our main conclusion, and it was in contrast to the results from the Taiwanese and South Korean studies.

    It is tempting to speculate about the reasons behind these differences (we probably didn't do a good enough job in presenting these speculations in our manuscript), and we fully agree with comments from Dudley, Eimer and Tanzi, Readhead, and Itzhaki that it might have been caused by different (substantially longer) treatment regimes (although Tzeng et al. found substantial effects even in treatments <30 days). Additionally, differences might have been caused by different medications. In the Taiwanese study, almost 40 percent of patients were treated with ganciclovir and valganciclovir, potential teratogens and carcinogens that, in the U.K., are only indicated for severe cytomegalovirus retinitis and for immunosuppressed transplant patients with cytomegalovirus, but  not for herpes (HSV/zoster) infections.

    We agree with Itzhaki that because of the inclusion criteria used in the study by Tzeng et al., but also because of differences in the national health systems, the study population in the Taiwanese study might have comprised patients with different herpes types (higher proportion of treated herpes labialis); if the association of antiherpetic treatment with dementia was substantially different in patients with herpes labialis than in patients with any other herpes infection, this might explain some of the differences; however, this would not invalidate our conclusions that, in our study populations, antiherpetic treatment was not substantially associated with incident dementia. Some of the differences might be explained by the analytic design—one of our main concerns was to properly correct for the effect of age: given an almost exponential increase in dementia incidence with increasing age, even small differences in age by exposure have a detrimental effect; studies categorizing people in age groups (e.g., ≤65 and 65+) seem doomed. To a lesser extent there were differences in the definitions of treatment (frequency versus duration) that are not directly comparable; however, we do not believe such differences majorly explain the big variation in our results. Finally, we note some unexplainable statistical irregularities in the presentation of hazard ratios in Tzeng et al.

    It is tempting to try to infer causality from our research (for example, Readhead stated that we report a quite consistent, modest protective effect of antiherpetic medications), but this is something we would certainly warn against: Observational cohort studies such as ours are not designed to infer causality. There might be subpopulations (for example, people with specific alleles, as suggested by Rizzo and Itzhaki, or people with low-grade herpetic infection, as suggested by Alberi) for whom treatment has a protective (or detrimental) effect—however, in our study populations (that are representative of the underlying general population, a great advantage of studies using electronic health records), we did not see any such associations.


    . Anti-herpetic Medications and Reduced Risk of Dementia in Patients with Herpes Simplex Virus Infections-a Nationwide, Population-Based Cohort Study in Taiwan. Neurotherapeutics. 2018 Apr;15(2):417-429. PubMed.

  8. Schnier et al. present a new, impressive epidemiological analysis in four large European cohorts to investigate a potential association of herpesvirus infection with the later development of dementia. Results from this study are, as previous ones, equivocal.

    We find it problematic that basic assumptions of many epidemiological studies, such as this one is, are 1. that anti-herpetic medication is taken as a proxy for an underlying biological herpesvirus infection but that, alongside, its variable severity is not quantified, and 2. that it is taken for granted that an acute anti-herpetic medication was effective against a chronic infection. For example, if an anti-herpetic medication was ineffective or too briefly administered, then the treatment effect would be marginal and infection may have persisted, thereby confounding any outcomes. The inclusion of biological parameters such as IgM levels against herpesviruses that indicate herpesvirus reactivation strengthens such studies (Lövheim et al., 2015). 

    We believe that the merit of epidemiological studies correlating phenomena in an interesting way, such as linking herpesvirus infection with later Alzheimer’s disease (AD), should trigger experimentally controlled studies aimed at establishing causality. So far, unfortunately, such studies are underrepresented relative to the importance of the issue.

    Experimental inoculations in common AD mouse models have supported claims of a causal relationship between herpesvirus infection and AD-like neuropathology (Eimer et al., 2018). One may rightfully object that the current AD animal models are based on genetic AD cases that insufficiently reflect the majority of sporadic AD cases, even if at this point there are no good alternatives. Experimental herpesvirus infections of a subclinical and chronic kind have been conducted in wild-type mice that likewise demonstrated induction of Aβ aggregation and increased tau phosphorylation (De Chiara et al., 2019), even though mouse Aβ peptide and tau protein do not behave similarly to their human counterparts.

    The next obvious step for reconstructing the molecular pathways from herpesvirus infection to AD-like cellular neuropathology, apart from refining modeling of herpesvirus latency and identifying more valid animal models, will be to identify overlapping cellular networks disrupted both during herpesvirus infection and AD (Readhead et al., 2018). 

    We recently demonstrated that influenza infection causes α-synuclein aggregation in vitro by inhibiting autophagy and dramatically augmenting its expression in vivo (Marreiros et al., 2020). This suggests that, in addition to the virus-induced neuroimmune activation, direct interactions of the virus with cellular host proteins could play a role. In fact, host cell proteins involved in both replication and assembly of neurotropic viruses and in the cellular pathways of neurodegenerative diseases could provide the molecular link that would propel the interesting connection between herpesviruses and AD to the next level and reveal unexpected novel drug targets (Müller-Schiffmann et al., 2020). 


    . Recurrent herpes simplex virus-1 infection induces hallmarks of neurodegeneration and cognitive deficits in mice. PLoS Pathog. 2019 Mar;15(3):e1007617. Epub 2019 Mar 14 PubMed.

    . Alzheimer's Disease-Associated β-Amyloid Is Rapidly Seeded by Herpesviridae to Protect against Brain Infection. Neuron. 2018 Jul 11;99(1):56-63.e3. PubMed.

    . Reactivated herpes simplex infection increases the risk of Alzheimer's disease. Alzheimers Dement. 2015 Jun;11(6):593-9. Epub 2014 Jul 17 PubMed.

    . Disruption of cellular proteostasis by H1N1 influenza A virus causes α-synuclein aggregation. Proc Natl Acad Sci U S A. 2020 Mar 24;117(12):6741-6751. Epub 2020 Mar 9 PubMed.

    . Viruses as 'Truffle Hounds': Molecular Tools for Untangling Brain Cellular Pathology. Trends Neurosci. 2020 Dec 11; PubMed.

    . Multiscale Analysis of Independent Alzheimer's Cohorts Finds Disruption of Molecular, Genetic, and Clinical Networks by Human Herpesvirus. Neuron. 2018 Jul 11;99(1):64-82.e7. Epub 2018 Jun 21 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Herpes Triggers Amyloid—Could This Virus Fuel Alzheimer’s?
  2. Herpes Simplex Virus Triggers Amyloidosis in 3D Neural Cultures
  3. Going Viral: Alzheimer’s Research at Herpes Conference
  4. How Does COVID-19 Affect the Brain?
  5. Fits and Starts: Trial Results from the CTAD Conference

Therapeutics Citations

  1. Valacyclovir

Paper Citations

  1. . Herpes simplex virus type 1 in brain and risk of Alzheimer's disease. Lancet. 1997 Jan 25;349(9047):241-4. PubMed.
  2. . Herpes simplex virus type 1 DNA is located within Alzheimer's disease amyloid plaques. J Pathol. 2009 Jan;217(1):131-8. PubMed.
  3. . COVID-19 severity impacts on long-term neurological manifestation after hospitalisation. MedRxiv. Posted January 02, 2021.
  4. . Increased risk of dementia following herpes zoster ophthalmicus. PLoS One. 2017;12(11):e0188490. Epub 2017 Nov 22 PubMed.
  5. . Herpes Zoster and Dementia: A Nationwide Population-Based Cohort Study. J Clin Psychiatry. 2018 Jan/Feb;79(1) PubMed.
  6. . Anti-herpetic Medications and Reduced Risk of Dementia in Patients with Herpes Simplex Virus Infections-a Nationwide, Population-Based Cohort Study in Taiwan. Neurotherapeutics. 2018 Apr;15(2):417-429. PubMed.
  7. . Herpes Viruses and Senile Dementia: First Population Evidence for a Causal Link. J Alzheimers Dis. 2018;64(2):363-366. PubMed.
  8. . Association of herpes zoster with dementia and effect of antiviral therapy on dementia: a population-based cohort study. Eur Arch Psychiatry Clin Neurosci. 2020 Jul 1; PubMed.
  9. . Hospital-treated infectious diseases and the risk of dementia: multicohort study with replication in the UK Biobank. MedRxiv. Posted April 24, 2020.
  10. . Human herpesvirus infections and dementia or mild cognitive impairment: a systematic review and meta-analysis. Sci Rep. 2019 Mar 18;9(1):4743. PubMed.
  11. . Herpes simplex virus and Alzheimer's disease: a Mendelian randomization study. Neurobiol Aging. 2021 Mar;99:101.e11-101.e13. Epub 2020 Oct 9 PubMed.
  12. . Antiviral therapy: Valacyclovir Treatment of Alzheimer's Disease (VALAD) Trial: protocol for a randomised, double-blind,placebo-controlled, treatment trial. BMJ Open. 2020 Feb 6;10(2):e032112. PubMed.

External Citations

  1. one in Sweden
  2. one in the U.S.
  3. Phase 2 trial in Poland
  4. press release
  5. R01 grant
  6. $1 million prize

Further Reading

Primary Papers

  1. . Antiherpetic medication and incident dementia: observational cohort studies in four countries. Eur J Neurol. 2021 Mar 3; PubMed.