Human Herpes viruses and potential links to neurological disease

 

Virus

Common name / illness

Neurological links (strength of evidence)

HSV-1

Cold sores

Strong link to encephalitis (a rare, severe brain infection). Increasing evidence connects recurrent infection or reactivation to Alzheimer’s disease.

HSV-2

Genital herpes

Can cause meningitis or neonatal encephalitis; less studied for dementia.

VZV

Chickenpox → shingles

Causes postherpetic neuralgia and, rarely, stroke or meningitis after reactivation. Currently being studied for a possible role in dementia.

EBV (Epstein–Barr)

Mononucleosis

Strongly linked to multiple sclerosis (MS); most convincing viral–neurological link so far.

CMV (Cytomegalovirus)

Often silent infection

In newborns or immunocompromised people can cause hearing loss and brain damage; mixed evidence for cognitive aging.

HHV-6 & HHV-7

Roseola in kids

Occasionally cause encephalitis; suspected minor role in chronic fatigue or epilepsy, not proven.

HHV-8

Kaposi’s sarcoma virus

No major neurological connection.



Caveats

Caveat 1: Residual confounding (time-shift) analysis
In Extended Data Figure 1, the authors do a clever “time-shift” test to check for residual confounding. They pretend everyone received their exposure (the shingles vaccine or shingles itself) one year earlier than they actually did, and then they compare dementia rates between groups for up to 1 year. 

If the shingles vaccine truly protects against dementia—or if having fewer shingles cases truly matters—we shouldn’t see any effect before the exposure happens. So any association in this pre-exposure window can only come from residual confounding.

But some of the relative risks (RRs) in the time-shift test look a lot like the RRs in the main analysis. For example:

  • Shingrix (2 doses) vs. pneumococcal vaccine: RR ≈ 0.80 at the “fake” one-year mark

  • Main analysis RRs: 0.73 (3 years) and 0.83 (5 years)

That similarity raises flags: it suggests there may be systematic differences between vaccinated and comparison groups that matching and weighting didn’t fully remove.

Here’s a full table with the results of the residual confounding analysis. The authors emphasize that residual confounding appears smaller when Shingrix is compared to another active vaccine rather than to no vaccine. That makes sense — active-comparator designs are generally stronger than comparing to people who get no vaccine at all.

But the key issue isn't whether the active comparator is better than nothing — it’s whether confounding is gone. And in these results, it’s not. Some of the pre-exposure relative risks for the active-comparator groups still cluster near the effects seen in the main analyses. 

Note that the confidence intervals are wide and most results are not statistically significant due to the shorter follow up time and fewer dementia cases. So it's difficult to draw any firm conclusions from these results.

Residual Confounding — Relative Risks (estimates are roughly estimated from the figure)

Comparison

1 mo

6 mo

12 mo

Zostavax vs PPSV23

0.60

1.00

1.20

Shingrix (2+ doses) vs PPSV23

0.80

1.00

0.90

Shingrix (1 dose) vs PPSV23

0.80

0.95

0.80*

Shingrix (2+ doses) vs Zostavax

1.30

1.05

1.00

shingles multiple vs shingles single

0.90

1.00

0.85*

Shingrix (2+ doses) vs Not exposed

0.60*

0.55*

0.55*

Zostavax vs Not exposed

0.55*

0.55*

0.55*

 

*significant reduction in risk in the pre-exposure period

Caveat 2: 

Preregistered protocol outcome mismatch

The study was preregistered in the EU PAS Register (now the HMA–EMA real-world data registry) as EUPAS107206

 In the published protocol,the authors list two outcomes:

  • Primary: dementia

  • Secondary: mild cognitive impairment (MCI)

But in the Nature Medicine paper, MCI isn’t reported at all. It may be slated for a future paper, but preregistering an outcome and then having it disappear without comment isn't ideal. Transparent reporting of all prespecified outcomes matters—especially when studying subtle, controversial effects like dementia prevention.