Persistent infection and chronic neurological disease: reservoirs, CNS seeding, entrenchment, and neural erosion

A unified causal grammar for how infections — past, latent, or distant — can quietly erode the brain over years or decades.

The core intuition is simple but easy to miss: a pathogen doesn't need to be actively replicating in your brain to damage it. Something pathogen-derived can persist elsewhere in the body — in sensory neurons, immune cells, gut communities, or even as inherited genomic remnants — and periodically deliver organisms, infected cells, antigens, or pathogen-specific immune pressure into the nervous system. The brain mounts a defense. Usually that defense resolves. The paper's central question is: when does it fail to resolve, and why?

The authors organize this into a six-step sequence: reservoir → seeding → engagement → effector response → entrenchment → erosion. Each step is a concrete, falsifiable claim. A "reservoir" is whatever preserves pathogen-derived causal capacity between neurologically relevant events — it could be latent herpesvirus in a ganglion, HIV hiding in brain macrophages, or the oral microbiome chronically releasing bacterial products into the bloodstream. "Seeding" is when reservoir output reaches the nervous system. "Entrenchment" is the critical failure mode: the local response doesn't shut off, either because the pathogen keeps driving it, or because the host's own feedback loops (glial activation, protein aggregation, vascular damage) become self-sustaining even after the microbial trigger has waned. The paper is careful to distinguish these regimes — they have very different implications for when and whether antimicrobial treatment could help.

The framework is anchored in well-established cases — SSPE, neuroHIV, EBV-associated multiple sclerosis, VZV vasculopathy, post-HSV autoimmune encephalitis — which serve as "reference architectures" proving that each link in the chain is biologically real in humans. SARS-CoV-2 is treated as a useful bridge case: persistent viral products and post-acute neurovascular abnormalities are documented in subsets of patients, but the paper explicitly refuses to conclude that chronic brain infection causes long COVID. That restraint is characteristic throughout. For more speculative branches — HSV-1 in Alzheimer's disease, periodontal bacteria, HCMV, HHV-6/7, endogenous retroviruses — the authors grade each causal link independently (established / supportive / mechanistic only / missing / contradicted) rather than giving an overall verdict. The VALAD trial, which found worse cognitive outcomes in HSV-seropositive Alzheimer's patients given high-dose valacyclovir, is reported honestly and not explained away.

A key insight is that the same reservoir output can resolve in one host and become entrenched in another. Aging is not a single variable here — it's a composite of blood-brain barrier dysfunction (worsened by hypertension and diabetes), declining proteostatic capacity (the machinery that clears misfolded proteins), changing microglial identity (recent data show yolk-sac-derived microglia are progressively replaced by marrow-derived cells in aging human hippocampus), and accumulated injury history including repetitive head impacts. The paper also makes a non-obvious point about proteins like amyloid-beta, tau, and alpha-synuclein: they appear to have genuine antimicrobial and antiviral functions, meaning their initial accumulation may be a defense response. Entrenchment happens when clearance fails and these proteins begin self-amplifying — a second step that requires its own explanation, and one that can occur without any infection at all (as familial Alzheimer's disease demonstrates).

The framework's value is disciplinary as much as scientific. It forces each proposed disease-pathogen chain to name what persists, how it reaches the nervous system, what local response follows, why that response fails to resolve, and — crucially — what observation would falsify the chain. The paper explicitly lists claims the evidence does not support, including the idea that microbial DNA in postmortem brain proves viable infection, or that a late negative PCR rules out an earlier focal trigger. The decisive experiments it calls for are prospective and stage-specific: measure active reservoir output before neurological conversion, not serostatus; use spatial, cell-resolved tissue analysis rather than bulk sequencing; and run biomarker-stratified trials that verify target engagement before testing clinical endpoints.