From a testable Alzheimer’s model to an Ion-Terrain map of dementia

From (DISSAD → DISSAD+)

In Part 1 (2025-08-14) I described how a “lithium” sidewalk moment in April 2024 turned into a testable Alzheimer’s model: DISSAD — Default-Mode Network Ion Seed-and-Sink in Alzheimer’s Disease The short version:

• Amyloid plaques in DMN hubs act as lithium sinks.
• Local Li⁺ depletion releases the brake on GSK3β, pushing tau into a hyper-phosphorylated state.
• That drives microglial stress and DMN network failure.
• The model comes with a concrete imaging plan: ^7Li-MRI, QSM for iron, amyloid/tau PET, p-tau217 slope, region-by-region.

At the end of that post, I wrote: “And now we wait.” …We didn’t wait very long.

What happened next: DISSAD+

Between August and late September, two things became obvious:

1. The DMN “terrain” piece wasn’t Alzheimer’s-specific. The same hot, leaky, iron-tilted hubs show up across multiple dementias.
2. DISSAD was just one configuration on that terrain. Alzheimer’s looked like Config-1 on a shared trunk, not its own isolated planet.

So on 2025-09-21 I created a new OSF project and promoted DISSAD from “just an AD theory” to a full trunk-and-fork framework: DISSAD+ — Default-Mode Network Ion-Terrain Trunk & Fork Theory
OSF project: https://osf.io/u26d9/ That project currently contains two PDFs:

• Default-Mode Network Ion-Terrain Trunk & Fork Theory (DISSAD+) — Detailed Outline
• Chemistry Appendix + Assay Plan

Those two PDFs are deliberately frozen at their September 2025 state. Since then, a wave of late-2025 papers (on lithium, DMN gradients, microglial states, BBB mechanics, etc.) has landed that I read as independent tests of that v0.3 trunk and chemistry. Rather than quietly editing the OSF files every time a new result appears, I’m treating them as a time-stamped snapshot. This blog post is where I track how later work lines up — or fails to line up — with that snapshot. Below is the plain-language version of what changed.

1. From single disease to shared trunk

DISSAD+ starts from a simple trunk hypothesis:

• The default-mode network (DMN) and a few coupled hubs run “hot” and wear down early.
• Over decades, a combination of micro-BBB leak, sleep/glymphatic failures, perfusion quirks, and iron-tilted microenvironments produces a shared vulnerable terrain.

In the more technical language of the outline, that’s a DMN susceptibility index – a common “ion-terrain” trunk. On top of that trunk, different substrates and networks give you different diseases.

2. Alzheimer’s becomes Config-1

In DISSAD+ language, Alzheimer’s is: Config-1: Aβ plaques + Li-sink ON in DMN hubs Same core idea as in Part 1, now embedded in a bigger structure:

• Seed: terrain crosses a vulnerability threshold in DMN hubs.
• Sink: amyloid plaques on that terrain behave like charged matrices that partition cations and locally lower free Li⁺.
• Switch: reduced Li⁺ disinhibits GSK3β → tau hyper-phosphorylation → inflammatory/myelin stress.
• Spread: DMN connectivity and structure fail.
• Clinical: MCI → Alzheimer’s dementia.

DISSAD+ turns this into a stage map (DISSAD-0 → DISSAD-6) and explicit predictions:

• Where the Li-sink should appear.
• How it should line up with plaques, iron and p-tau.
• Which regions must not show the effect (primary visual cortex as a registered negative control).

3. Other dementias become forks on the same terrain

Once the trunk and Config-1 were written down, the rest followed. DISSAD+ treats other dementias as forks on the same DMN-anchored terrain, with different seeds and network biases:

• DLB / Parkinson’s dementia: same trunk, different substrate (α-syn), more posterior/visual bias. Li-sink should stay OFF.
• LATE: TDP-43 in limbic territory attached to the DMN axis. Li-sink OFF.
• bvFTD: salience / fronto-insular network dominates; still coupled into the DMN, but with a different protein mix. Li-sink OFF.
• PCA: an Alzheimer’s variant where posterior visual networks are more involved, but with AD-style Aβ and Li-sink in the background.
• VCI / mixed: vascular lesions attacking the same hub system; Aβ-negative arms should not show a Li-sink.

DISSAD+ makes this precise enough that cohorts can be labelled as: Stage (0–6) × Config (AD / DLB / LATE / bvFTD / VCI / Mixed), with “fingerprint” rules for each config and explicit falsifiers if the fingerprints don’t hold.

4. The chemistry is now pinned down instead of hand-waved

Part 1 sketched the Li-sink idea conceptually. The Chemistry Appendix + Assay Plan turns it into math and experiments:

• Reaction–diffusion and Donnan-style partitioning around plaques.
• Bounds on how much local Li⁺ can realistically drop in DMN tissue.
• A bridge from Δ[Li⁺] → GSK3β inhibition → expected change in p-tau217 slope.
• A pass/fail sheet: if no plausible parameter set produces enough Δ[Li⁺] to move GSK3β, the sink idea fails.

That theory side is tied directly to an assay plan:

• Ex vivo Li mapping in plaques vs peri-plaque vs plaque-free tissue (DMN vs controls).
• In-vitro binding and competition experiments across pH, ionic strength, and Li formulations.
• ^7Li-MRI + QSM in Aβ-positive cohorts to define when a parcel is “Li-sink ON”.
• p-tau217 slopes as the main downstream signal.
• Falsifiers that explicitly say: “If X doesn’t happen, the model is wrong.”

5. Where we are now

So, what’s changed since the first blog post?

• The original DISSAD Alzheimer’s model is still there, but it has moved from “nice idea” to a fully specified Config-1 on a shared trunk.
• The theory now covers the broader dementia landscape as a set of forks on a DMN-anchored ion-terrain.
• We’ve written down enough chemistry and imaging detail that other groups can actually try to break it.

Since September 2025, several independent groups have published work on lithium, microglia, DMN gradients, BBB mechanics, vascular pulses and network-level biomarker cascades that appear to land directly on this trunk-and-fork map. Rather than revising the OSF PDFs every time a new paper arrives, I’m keeping v0.3 as the “this is what we committed to on 2025-09-21” snapshot. Any convergence, contradictions or updates after that point will be tracked in blog posts, talks and (if needed) a future v0.4 — but only if the structure itself needs to change, not just because another paper agrees with it. If you’re curious (or you run cohorts, scanners, or wet labs), everything so far is wrapped here: DISSAD+ OSF project (trunk + forks + chemistry + falsifiers): https://osf.io/u26d9/ Cheers,
Marcus