As of **April 16, 2026**, the practical answer is: **plaques and tangles are unlikely to be just a consequence of cholinergic dysfunction**. Human data support a model where **cholinergic failure is mostly a downstream and amplifying vulnerability state**, not the primary upstream lesion. The strongest reason is that **anti-amyloid drugs do slow clinical decline**, albeit modestly, while decades of cholinergic-targeted programs have mostly delivered **symptomatic** benefit, not disease modification.
**Practical readout on the 7 hypotheses**
| Hypothesis | Druggable? | Real chemical matter / programs | Practical verdict |
|---|---|---|---|
| `α7-nAChR / APP` | Yes, receptor is druggable | `encenicline/EVP-6124` (Forum; Ph3 AD `NCT01969123`, `NCT01969136`, both terminated after clinical hold), `ABT-126` (AbbVie; Ph2 negative) | **Biology interesting, asset class de-risked negatively for AD.** Also mechanistically conflicted because the field mostly pursued **agonism/PAMs**, not blockade. |
| `EPHB2` | Poorly, for this use | No credible CNS-ready EphB2 agonist program in AD | **Low tractability.** Hard target because the therapeutic idea needs tuned **restoration/agonism**, not simple inhibition. |
| `P2X7` | Yes | `JNJ-54175446`, `JNJ-55308942` (Janssen CNS P2X7 agents; brain penetration/occupancy shown, but not advanced in AD) | **Reasonable neuroinflammation target, but likely glia-first not cholinergic-first.** Best fit as adjunct, not core thesis. |
| `PDK1` | Marginally | `dichloroacetate` as a blunt PDK inhibitor; no meaningful AD clinical development | **Repurposable for a biomarker study, but weak IP, weak selectivity, neuropathy/hepatotoxicity baggage.** |
| `NLRP3` | Yes | `dapansutrile` (systemic oral NLRP3 inhibitor; PD trial `NCT07157735`), `ACI-19764` (AC Immune; Ph1 healthy volunteers `NCT07463196`) | **Best small-molecule class here.** Still, this is an inflammation program, not a clean cholinergic-causality program. |
| `GAT3/SLC6A13` | Poorly | No validated **enhancer** chemistry; transporter field mostly has inhibitors, which would go the wrong way | **Not practical today.** Biology can be studied, but I would not build a company on this now. |
| `CDC37/HSP90` | HSP90 yes, CDC37 no | `icapamespib/PU-AD` (Samus; Ph1 `NCT03935568`, sponsor ceased operations) | **Mechanistically sophisticated but high execution risk.** Better framed as epichaperome/proteostasis than cholinergic selectivity. |
**What is actually investable**
1. **`NLRP3`**: best balance of druggability, current chemistry, and strategic relevance. Main risk is infection/immunology liabilities and the fact that benefit may be broad anti-inflammatory rather than cholinergic-specific.
2. **`P2X7`**: second-best. Real CNS chemistry exists, but the biology points more to **microglia/inflammasome** than cholinergic terminals. Competitive landscape is thinner than NLRP3, but so is human efficacy evidence.
3. **`PDK1 / metabolic rescue`**: only attractive as a fast, cheap translational experiment using repurposed matter, not as a durable differentiated program.
4. **`α7-nAChR`**: target is druggable, but the AD field already spent a lot of capital here and got little. Hard to justify unless you have a very different modality, such as a cholinergic-neuron-selective degrader/blocker or a structural biology-based interface disruptor.
5. **`EPHB2`, `GAT3`, `CDC37`**: good paper targets, weak program targets.
**Safety reality**
- `α7 nAChR`: GI and cholinergic adverse effects, desensitization/tachyphylaxis risk, seizure concern in fragile brains.
- `P2X7` / `NLRP3`: infection risk, immunomodulation, liver signal risk, uncertain chronic CNS tolerability in elderly AD patients.
- `DCA/PDK`: peripheral neuropathy, hepatotoxicity, mitochondrial off-target concerns.
- `HSP90/CDC37`: proteostasis toxicity, DDIs, narrow therapeutic index; oncology-style HSP90 liabilities are a warning.
- `EPHB2`: synaptic plasticity/off-target kinase biology makes safety and dose window unclear.
- `GAT3`: if you ever find an enhancer, circuit-level sedation/seizure liability would need close monitoring.
**Cost / timeline**
- Repurposed `DCA` biomarker study: **12-18 months, $3M-$8M**.
- Existing-asset `P2X7` or `NLRP3` proof-of-concept in AD biomarkers: **18-30 months, $10M-$25M**.
- New discovery program for `EPHB2`, `GAT3`, or `CDC37`: **3-5 years to IND, $25M-$60M+**.
- Real disease-modification readout in AD: usually **Phase 2b/3 time horizon**, so **5-8 years** from fresh start.
**Recommendation**
If the goal is to ground the gap in drug-development reality, I would reframe the causal conclusion as:
`Aβ/tau are partly upstream; cholinergic dysfunction is a vulnerable amplifier and clinically important mediator, not the sole initiating event.`
Then prioritize:
1. `NLRP3`
2. `P2X7`
3. `metabolic rescue` as a cheap translational side-bet
I would **not** prioritize `EPHB2`, `GAT3`, or direct `CDC37` for a serious AD program today.
Sources: [Lecanemab approval](https://www.eisai.com/news/2023/news202301.html), [Clarity AD NCT03887455](https://www.clinicaltrials.gov/study/NCT03887455), [Donanemab approval](https://investor.lilly.com/news-releases/news-release-details/lillys-kisunlatm-donanemab-azbt-approved-fda-treatment-early), [TRAILBLAZER-ALZ 2 NCT04437511](https://www.clinicaltrials.gov/study/NCT04437511), [Encenicline Ph3 AD NCT01969123](https://clinicaltrials.gov/study/NCT01969123), [Encenicline Ph3 AD NCT01969136](https://clinicaltrials.gov/study/NCT01969136), [ABT-126 trial](https://pubmed.ncbi.nlm.nih.gov/29854928/), [JNJ-54175446 Phase 1](https://pubmed.ncbi.nlm.nih.gov/30260294/), [JNJ-55308942 occupancy study NCT03437590](https://clinicaltrials.gov/study/NCT03437590), [ACI-19764 Phase 1 NCT07463196](https://clinicaltrials.gov/study/NCT07463196), [Dapansutrile in PD NCT07157735](https://clinicaltrials.gov/study/NCT07157735), [Icapamespib/PU-AD Ph1 NCT03935568](https://clinicaltrials.gov/study/NCT03935568), [AAV2-NGF trial NCT00876863](https://pmc.ncbi.nlm.nih.gov/articles/PMC5885277/).
If you want, I can turn this into a **ranked investment memo** with `mechanism / asset / trial design / biomarkers / kill criteria` for the top 3.