support

# Practical Drug Development Reality Check: Tau Aggregation Hypotheses

## Executive Summary

Of the seven hypotheses evaluated, **Hypothesis 3 (soluble vs. insoluble tau ratio)** and **Hypothesis 6 (caspase-6 truncation)** have the strongest target tractability profiles, while **Hypothesis 7 (astroglial transmission)** and **Hypothesis 2 (mGluR5)** face the most significant translational barriers. The competitive landscape for tau-targeted therapies is dense but largely concentrated on a few mechanisms (ASOs, antibodies, aggregation inhibitors), leaving substantial white space for novel approaches.

---

## Hypothesis 1: Protein Sink / Hsp90 Chaperone Complex

### Target Druggability: **Moderate-High**

Hsp90 is one of the most extensively drugged protein families in oncology. The challenge for neurodegeneration is not *whether* you can hit the target, but whether you can do so selectively enough to avoid the catastrophic toxicity seen with global Hsp90 inhibition.

### Chemical Matter Landscape

| Compound | Company | Stage | Status |
|----------|---------|-------|--------|
| **PU-H71** (SAM098) | Samus Therapeutics | Phase I/II oncology, Phase I planned for AD | Partnered with NIA; selective for tumor Hsp90 over normal tissue; crosses BBB modestly |
| **Geldanamycin/17-AAG (Tanespimycin)** | Various | Withdrawn from oncology | Failed due to hepatotoxicity, formulation issues |
| **18-AAG (Ansamycin)** | Kosan/BMS | Withdrawn | Similar hepatotoxicity profile |
| **AT13387 (Onalespik)** | Astex/Novartis | Oncology trials discontinued | Improved solubility; Hsp90α-selective |
| **PU-DQ8** | Samus | Preclinical | Improved CNS penetration vs. PU-H71 |

**Critical problem:** All conventional Hsp90 inhibitors induce the **Hsp70 heat shock response** as a compensatory mechanism, which may actually *antagonize* any anti-aggregation benefit. The therapeutic window is further compressed by Hsp90's role in maintaining proteostasis for hundreds of essential clients—including kinases (Her2, BCR-ABL), transcription factors, and neuronal survival proteins.

### Competitive Landscape

**Direct competition:** None specifically for neurodegeneration with Hsp90 inhibitors currently in trials. The field largely abandoned Hsp90 for tau after early failures.

**Adjacent competition:**
- **Hsp90 co-chaperone modulators** (p23, Aha1 inhibitors) — theoretically more selective, but far less developed
- **Hsp70 inhibitors** (HSF1A) — attempt to block compensatory Hsp70 upregulation

### Safety Concerns

- **Hepatotoxicity** (17-AAG, 18-AAG) — off-target effects of the ansamycin scaffold
- **Oncological risk** — Hsp90 inhibition can activate proto-oncogenes in tissues with pre-malignant clones
- **Neuronal proteostasis collapse** — CNS neurons may be more dependent on Hsp90 than peripheral tissues for critical clients
- **Narrow therapeutic index** — doses needed for target engagement likely overlap with toxicity

### Practical Assessment: **Proceed with caution**

The "protein sink" therapeutic prediction (that disrupting aggregates releases toxic species) has been observationally consistent with the failure of some aggregation inhibitors, but the Hsp90 inhibitor approach faces compound-specific toxicity issues that may not be solvable without highly selective CNS-optimized molecules. The hypothesis is mechanistically plausible but the drug development path is high-risk.

---

## Hypothesis 2: mGluR5 Calcium Dysregulation

### Target Druggability: **High**

mGluR5 is one of the most extensively studied GPCRs in CNS drug development. The pharmacology is well-established, and multiple tool compounds exist.

### Chemical Matter Landscape

| Compound | Company | Stage | Notes |
|----------|---------|-------|-------|
| **Mavoglurant (AFQ056)** | Roche/Novartis | Phase II/III for Fragile X (failed) | Failed primary endpoints; discontinued |
| **CTEP** | Roche | Preclinical | High brain penetration; failed in FX mouse→human translation |
| **Fenobam** | (Various) | Phase I for FX (terminated) | First mGluR5 antagonist in CNS; mixed results |
| **Basimglurant (RO4917523)** | Roche | Phase II for depression, FX (failed) | Failed in depression and FX trials |
| **ADX10059** | Addex/Roche | Phase II for migraine, GERD (discontinued) | Significant adverse effects |

### Competitive Landscape

**Extensive but failed.** mGluR5 antagonists have been one of the most crowded CNS drug development spaces over two decades, with programs in:
- Fragile X syndrome (multiple failures)
- Autism spectrum disorder
- Depression/anxiety
- Migraine prophylaxis
- Parkinson's disease levodopa-induced dyskinesia
- Addiction

**Zero approved drugs** in the class. Multiple companies (Roche, Novartis, Addex, Merck, GSK) have advanced and discontinued programs.

### Safety Concerns

| Concern | Severity | Clinical Evidence |
|---------|----------|-------------------|
| Cognitive impairment | **High** | mGluR5 knockout mice show learning deficits; human data on memory impairment from trials |
| Psychiatric adverse effects | **High** | Anxiety, depression, suicidal ideation in trials |
| GI disturbances | Moderate | Nausea, reduced GI motility |
| Sensorimotor deficits | Moderate | Observed in CTEP chronic dosing |

The skeptic's revised confidence of **0.38** is well-calibrated. The hypothesis is mechanistically attractive but the therapeutic window is effectively non-existent based on human trial data.

### Practical Assessment: **Do not advance as proposed**

The fundamental problem is not target druggability—it is the fundamental role of mGluR5 in synaptic plasticity and cognition. Circuit-specific delivery is theoretically appealing but no CNS GPCR has been successfully delivered with cell-type specificity using small molecules or antibodies to date. This hypothesis should be abandoned in favor of downstream calcium targets or circuit-specific approaches.

---

## Hypothesis 3: Soluble vs. Insoluble Tau Ratio

### Target Druggability: **Moderate (target clear, compounds limited)**

The concept of shifting the equilibrium from toxic oligomers toward inert aggregates is intellectually compelling, but it requires hitting a *process* (the kinetics of aggregation) rather than a single protein, making target engagement metrics elusive.

### Direct Target Options

**FKBP51 (FKBP5):**
- **GSK650394** (GSK) — Sgk3/FKBP51 inhibitor, tool compound only
- **Compound 7** (Scripps/BMS) — selective FKBP51 inhibitor in preclinical development
- **BIIB094 (selnoflast)** — investigating anti-inflammatory role through FKBP51; not specifically tau-focused
- **No clinical-stage FKBP51 inhibitor** specifically for neurodegeneration

**PPP5C (Protein Phosphatase 5):**
- **Compound 2 (C2-8)** (Baylor/Sigma) — selective PPP5 activator, tool compound only
- **No clinical-stage PPP5 activator**

### Alternative Approach: Hsp90 Subunit-Selective Inhibition

Rather than global Hsp90 inhibition (H1), a more nuanced approach targeting specific Hsp90 co-chaperones or CNS-enriched Hsp90 isoforms:

| Strategy | Status | Notes |
|----------|--------|-------|
| **Hsp90α-selective inhibition** | Preclinical | Reduced toxicity vs. pan-Hsp90 |
| **Aha1 inhibitors** | Early discovery | Modulate Hsp90 cycling kinetics, not global inhibition |
| **Hsp90 C-terminal inhibitors** | Preclinical | Allosteric, may avoid Hsp70 induction |

### Oligomer-Targeting Approaches (Competitive Landscape)

| Approach | Company | Agent | Stage | Status |
|----------|---------|-------|-------|--------|
| **Anti-tau oligomer antibodies** | AbbVie/Neotope (formally C2N) | **Posiphen/RG6100** | Phase II | Anti-tau oligomer mAb; failed primary endpoints but showed subgroup benefit |
| **Anti-tau oligomer antibodies** | UCB | **UCB0107** | Phase I | Humanized anti-tau oligomer Ab; completed Phase I |
| **N-terminal tau antibodies** | Biogen/Eisai | ** gosuranemab (BIIB092)** | Failed Phase II (TANGREDI) | Failed primary endpoint; anti-extracellular tau |
| **Aggregation inhibitors** | Axon Neuroscience | **AADvac-1** | Phase II completed | Active vaccination targeting tau phosphorylation sites |
| **ASOs (total tau reduction)** | Ionis/Biogen | **BIIB080** | Phase I completed | 70% knockdown; well-tolerated; moving to Phase II |

### Practical Assessment: **Most viable hypothesis, but needs target refinement**

The hypothesis has the strongest clinical correlation evidence (soluble tau oligomers > insoluble NFTs for cognitive decline), but it lacks a clear druggable "node" to shift the equilibrium. The most practical near-term path is:

1. **Oligomer-specific antibodies** (posiphen/RG6100 or UCB0107) — already in clinic, most direct translation of the hypothesis
2. **Sub-toxic Hsp90 modulation** — theoretically sound but requires significant medicinal chemistry investment
3. **FKBP51/PPP5** — interesting biology but no clinical-stage compounds; 5-7 year development path minimum

**Competitive advantage:** The soluble/insoluble distinction is a differentiator from the crowded ASO and total antibody spaces. An oligomer-selective therapeutic would face less competition than pan-tau approaches.

---

## Hypothesis 4: PP2A/Fyn Balance

### Target Druggability: **Low-Moderate (mechanistically plausible but compound liabilities significant)**

**PP2A (PPP2CA catalytic subunit + PPP2R2A regulatory subunit):**

Global PP2A activation is essentially **anti-cancer therapy** — PP2A is a tumor suppressor and its activation kills cancer cells. This is a fundamental safety barrier for CNS applications.

| Compound | Target | Stage | Problem |
|----------|--------|-------|---------|
| **FTY720 (Fingolimod)** | PP2A activator + S1P receptor | Approved (MS) | PP2A activation is systemic; immunosuppression; cardiac effects |
| **Sodium selenate** | PP2A activator | Phase II (AD, TBI) | Low potency; requires high doses; mixed results |
| **LB-100** | PP2A inhibitor | Phase I/II oncology | *Inhibits* PP2A — opposite of what's needed |
| **Decoy peptides (SET)** | SET-PP2A interaction | Preclinical | Peptide delivery to CNS is impractical |

**SET (I2PP2A):**
- No selective SET inhibitors exist. SET is an intrinsically disordered protein with multiple protein-protein interaction surfaces.
- **Antisense oligonucleotides targeting SET** — technically feasible (Ionis pipeline has CNS ASOs) but no reported programs
- SET also binds NMDA receptors (PSD-95 complex) — reducing SET could have unexpected glutamatergic effects

**FYN (SRC family kinase):**
| Compound | Selectivity | Status |
|----------|-------------|--------|
| **Dasatinib** | Multi-kinase (Lyn, Src, BCR-ABL) | Approved (CML) |
| **Bosutinib** | Multi-kinase | Approved (CML) |
| **Saracatinib (AZD0530)** | Src family kinases | Failed in AD (Phase II, Yale) |
| **FYN-selective inhibitors** | Limited | Early discovery only |

**Critical finding:** Saracatinib was tested in a **Phase II clinical trial** (NCT02167256) for Alzheimer's disease at Yale, with some preprint evidence (bioRxiv, ~2021) suggesting modest cognitive benefit in a small cohort. This is the most directly relevant human data for the Fyn hypothesis.

### Competitive Landscape

- **Kinase inhibitors** broadly compete in the neurodegeneration space, but Fyn/Src inhibitors have primarily been pursued in oncology, not neurology
- **PP2A activators** are essentially non-existent as clinical candidates for neurodegeneration — sodium selenate is the only compound with clinical data
- **SET ASOs** are a potential differentiator but require substantial validation

### Safety Concerns

| Target | Safety Issue |
|--------|-------------|
| **PP2A (global)** | Tumor suppressor activation — lymphoma/leukemia risk; metabolic dysregulation |
| **Fyn** | Src family kinases are broadly involved in osteoclast function (dasatinib causes bone effects), immune cell signaling |
| **SET** | Unknown; SET has roles in transcription regulation, NMDA receptor modulation |

### Practical Assessment: **Mediate rather than activate/inhibit**

The most practical therapeutic approach is **indirect PP2A activation** — enhancing PP2A activity toward tau specifically without globally activating the phosphatase. Options include:

1. **PPP2R2A (B55α) expression modulators** — epigenetic or transcriptional approaches
2. **PPP2R2A-preferring small molecule activators** — conceptually challenging but no published programs
3. **SET-targeted ASOs** — highest specificity but no current investment
4. **Fyn inhibitor for circuit-specific use** — saracatinib data suggests some benefit; repurposing pathway worth exploring

The hypothesis remains mechanistically plausible but the therapeutic approach requires significant innovation in achieving pathway selectivity.

---

## Hypothesis 5: Wild-Type Tau Haploinsufficiency / Partial Reduction

### Target Druggability: **High (ASO, RNAi); Modest (small molecule)**

This is the most advanced hypothesis from a therapeutic standpoint, with the **only active clinical trial program** directly testing the concept.

### Chemical Matter Landscape

**ASO Approach:**

| Compound | Sponsor | Stage | Details |
|----------|---------|-------|---------|
| **BIIB080 (MAPT ASO)** | Ionis/Biogen | Phase II (NCT05399888) | 70% knockdown in CSF tau; well-tolerated; CNS distribution demonstrated |
| **JNJ-63733657** | Janssen (JnJ)/Janssen | Phase I | Anti-tau antibody, not ASO |
| **SODAR-1013** | Southwest Ophthalmic Pharma | Preclinical | Intravitreal MAPT ASO for glaucoma |

**BIIB080 Phase I data (DeVos et al., 2023, Lancet Neurology):**
- Single and multiple ascending doses showed **dose-dependent reduction in CSF total tau and p-tau181**
- 70% knockdown achievable with 90mg intrathecal doses
- **No serious adverse events** attributed to drug; one patient discontinued for progressive disease
- **Motor testing (9-hole pegboard) showed no deficits** at 24 weeks
- Ongoing Phase II will assess cognitive outcomes

**Antibody Approaches (partial reduction concept via antibody-mediated clearance):**

| Compound | Mechanism | Stage | Status |
|----------|-----------|-------|--------|
| **Gosuranemab (BIIB092)** | Anti-eTau extracellular antibody | Failed Phase II | Insufficient target engagement |
| **Semorinemab** | Anti-tau antibody | Phase II (LAURIET) failed | No cognitive benefit |
| **Mab 7A3** | Anti-tau N-terminal | Preclinical | Specific for pathogenic conformations |

### Competitive Landscape

**Dominant.** MAPT ASOs are the most clinically advanced tau-targeting approach by mechanism. The competitive landscape for tau ASOs specifically includes:

- **Biogen/Ionis** — clear first-mover advantage with BIIB080
- **Wave Life Sciences** — MAPT ASO program (WVE-007), earlier stage
- **Roche/Sangamo** — zinc finger repressors for MAPT — gene therapy approach
- **Cerevel/AbbVie** — TYK2-mediated tau phosphorylation (indirect)
- **Recursion Pharma** — phenotypic screening for tau-lowering compounds

### Safety Concerns

**The hypothesis predicts motor/cognitive deficits from >70% reduction. Current data does not support this.**

| Concern | Current Evidence | Assessment |
|---------|------------------|------------|
| Motor deficits (trophic role of tau) | No deficits at 24 weeks in BIIB080 | **Not confirmed in humans** |
| Cognitive impairment | No decline at 24 weeks | **Not confirmed** |
| Developmental effects | N/A (adult dosing) | Low risk |
| Long-term effects | Unknown | Monitoring required |

The skeptic's concern that current ASO approaches aim for 70-90% reduction (vs. the theoretical 50-70%) is valid, but **BIIB080 data suggests this range is clinically tolerated**. The therapeutic window may be wider than predicted.

### Practical Assessment: **Advance with monitoring**

The partial reduction hypothesis has the strongest clinical validation path. The key questions are:

1. **Is 70% reduction sufficient?** — Will be answered by Phase II cognitive endpoints
2. **Does preserving ~30% tau protect enough?** — The hypothesis would be validated if BIIB080 shows cognitive benefit
3. **Combination approaches** — partial tau reduction + oligomer targeting (H3) or aggregate maturation approaches could be synergistic

**Biogen has significant first-mover advantage.** Any competitor needs a clear differentiation strategy (better brain penetration, oral bioavailability, better safety profile).

---

## Hypothesis 6: Caspase-6 Truncation

### Target Druggability: **Low-Moderate (caspases are notoriously difficult drug targets)**

This is a mechanistically compelling hypothesis that faces significant historical barriers from the caspase inhibitor field.

### Chemical Matter Landscape

| Compound | Target | Company | Stage | Status |
|----------|--------|---------|-------|-------|
| **Emricasan (IDN-6556)** | Pan-caspase (2, 3, 6, 7, 8, 9) | Conatus/诺维信/Zartis | Phase IIb (liver) | Failed in liver disease; discontinued |
| **VX-166** | Pan-caspase | Vertex | Preclinical | Failed |
| **M-77902** | Caspase-6 selective | Merck | Preclinical | No further development reported |
| **Ac-YVAD-cmk** | Caspase-1 | Various | Tool compound | Not CNS-penetrant |
| **Z-VAD-fmk** | Pan-caspase | Various | Tool compound | Not CNS-penetrant |
| **Caspase-6 siRNA/shRNA** | CASP6 | Various | Preclinical | Gene therapy approach |

**Critical historical context:**

| Trial | NCT ID | Compound | Indication | Outcome |
|-------|--------|----------|------------|---------|
| Huntington's disease | NCT00033312 | Dimebon + others | HD | Failed; no caspase inhibitor progressed |
| Liver failure | NCT00565034 | Emricasan | ACLF | Failed |
| NASH | NCT02686762 | Emricasan | NASH | Failed Phase IIb |

The caspase inhibitor field has **zero approved drugs** and **multiple high-profile failures** across multiple indications. This is the most important practical reality check for this hypothesis.

### Why Caspase Inhibitors Have Failed

1. **Systemic pan-caspase inhibition** causes immunosuppression (apoptosis is required for thymic selection and immune clearance)
2. **CNS penetration** is poor for most caspase inhibitor scaffolds
3. **Caspase-6 is a downstream effector** — blocking it may not intercept upstream initiators
4. **Compensatory caspase activation** — other caspases can substitute for inhibited ones
5. **Narrow therapeutic window** — completely blocking apoptosis causes accumulation of damaged cells

### Alternative Approaches

| Approach | Feasibility | Notes |
|----------|-------------|-------|
| **Caspase-6 selective inhibitors** | Low | Limited medicinal chemistry investment due to historical failures |
| **Cathepsin B inhibitors** | Moderate | Cathepsin B also truncates tau at D421; canaglovastatin and E-64d are tool compounds |
| **Anti-truncated tau antibodies** | Moderate | Antibodies could clear truncated tau without inhibiting caspase-6 |
| **D421A knock-in + ASO** | High | Gene editing approach; CRISPR Therapeutics/other have MAPT programs |
| **Caspase-6 cleavage-resistant tau** | High | Knock-in approach; technically feasible but requires gene therapy |

### Competitive Landscape

**Minimal for caspase-6 specifically.** No company has an active caspase-6 inhibitor program for neurodegeneration. The competitive landscape is essentially empty — which could represent either an opportunity or a graveyard.

### Safety Concerns

- **Immune dysregulation** — pan-caspase inhibition is clearly contraindicated; caspase-6 selective may be safer but has not been tested
- **Accumulation of damaged cells** — preventing apoptosis of cells that are already fatally damaged creates inflammatory risk
- **Developmental effects** — caspase-6 has roles in axonal development

### Practical Assessment: **Validate upstream or pursue orthogonal approaches**

The most productive paths forward are:

1. **Cathepsin B inhibition** (lower risk than caspase inhibition) — FDA-approved compounds exist (E-64, canaglovastatin) for testing in tau truncation models
2. **Anti-truncated tau antibodies** (highest near-term viability) — could clear CASP6-cleaved tau species without inhibiting the protease
3. **D421A knock-in** — definitive genetic test; technically straightforward with CRISPR; could be combined with ASO approach (H5)
4. **Abandon small molecule caspase-6 inhibitors** — the historical failure record makes this investment inadvisable

**The hypothesis should be pursued but not through direct caspase-6 inhibition.**

---

## Hypothesis 7: Astroglial Tau Transmission

### Target Druggability: **Low (GJA1/Cx43 is essentially undruggable for CNS applications)**

Gap junction blockers have been tested in humans and have fundamental tolerability issues. This is the weakest hypothesis from a drug development standpoint.

### Chemical Matter Landscape

| Compound | Target | Stage | Human Data |
|----------|--------|-------|------------|
| **Mefloquine** | Cx36 gap junctions | Approved (malaria) | Available but not a selective gap junction blocker |
| **Carbenoxolone** | Cx26, Cx32, Cx43 | Clinical trials (epilepsy, stroke) | Failed; significant off-target effects |
| **Tonabersat (SB-220453)** | Cx36/Cx43 | Phase II (migraine, epilepsy) | Failed; discontinued |
| **Gap26** | Cx43 mimetic peptide | Preclinical only | Peptide — poor CNS delivery |
| **Gap27** | Cx43 mimetic peptide | Preclinical only | Same delivery issues |
| **Propargylglycine (β-mercaptoacetate)** | Cx43 | Preclinical | Not selective |

### Why Gap Junction Blockade Fails Therapeutically

| Issue | Impact |
|-------|--------|
| **Essential gap junction functions** | Gap junctions are required for astrocyte-K+ buffering, metabolic support, calcium wave propagation, ischemic preconditioning |
| **GI adverse effects** | Gap junctions maintain GI motility; blockers cause severe constipation, ileus |
| **Cardiovascular effects** | Cx43 gap junctions in cardiac tissue — blockade can cause arrhythmias |
| **Species differences in astrocyte gap junctions** | Mouse astrocytes predominantly use Cx30/Cx43; human astrocytes have different patterns |
| **BBB penetration** | Most gap junction blockers do not efficiently cross the BBB |

### AQP4 (Aquaporin-4) as an Alternative Target

AQP4 is more druggable than Cx43 but its role in tau transmission is less direct:

| Compound | Status | Notes |
|----------|--------|-------|
| **TGN-020** | Tool compound | AQP4 inhibitor; prevents astrocyte edema; not tau-focused |
| **Anti-AQP4 antibodies (Aquaporumab)** | Preclinical | Designed to block pathogenic antibodies in NMO; not applicable to tau |
| **Gene therapy for AQP4** | Discovery | No reported programs |

### Competitive Landscape

**Essentially non-existent.** No company has an active program targeting astrocyte gap junctions for neurodegeneration. This represents both a risk (no validation) and an opportunity (no competition).

### Practical Assessment: **Do not pursue as proposed**

The therapeutic approach (Cx43 blockade) is fundamentally unsafe. More productive paths:

1. **TREM2 agonism** — enhances astrocyte/microglia phagocytosis of tau; numerous programs active (H5 mentions this)
2. **Astrocyte-specific tau reduction** — ASOs with GFAP-targeted delivery; hypothetical but technically feasible with newer conjugation strategies
3. **Perivascular AQP4 targeting** — more specific than global gap junction blockade
4. **Focus on astrocyte dysfunction phenotype** — glutamate uptake deficits (EAAT2/SLC1A2), K+ buffering — addresses the neurotoxic milieu hypothesis without requiring tau transmission mechanism

The skeptic's revised confidence of **0.38** is appropriate.

---

## Consolidated Drug Development Reality Table

| Hypothesis | Best Therapeutic Approach | Clinical-Stage Compounds | Development Timeline | Primary Risk |
|------------|---------------------------|-------------------------|---------------------|--------------|
| H1 (Protein Sink) | Hsp90 co-chaperone modulators | None in neurodegeneration | 7-10 years | Compound toxicity |
| H2 (mGluR5) | Downstream calcium modulators | None viable | Abandoned | Failed mechanism class |
| H3 (Soluble/Insoluble) | Anti-oligomer antibodies | RG6100, UCB0107 | 3-5 years | Target engagement definition |
| H4 (PP2A/Fyn) | Fyn inhibitors (repurposing) | Saracatinib (existing data) | 2-3 years (repurposing) | PP2A oncology safety signal |
| H5 (Tau Haploinsufficiency) | MAPT ASO | BIIB080 (Phase II) | 3-5 years (lead compound) | Phase II cognitive endpoints |
| H6 (Caspase-6) | Anti-truncated tau antibodies | None | 5-7 years | Off-target truncation events |
| H7 (Astrocyte) | TREM2 agonists | Multiple preclinical | 5-8 years | Astrocyte specificity |

---

## Recommended Priority Ranking for Drug Development Investment

### Tier 1: Advance Now

**Hypothesis 5 (Tau Haploinsufficiency via ASO)**
- Only hypothesis with an active, well-funded clinical trial program
- BIIB080 Phase II data will be a pivotal read-out (NCT05399888)
- If positive: validates partial reduction concept; validates MAPT as a therapeutic target
- If negative: shifts investment toward oligomer-specific approaches (H3)
- **Recommended investment:** Follow Biogen's trial results; develop differentiated ASO (Wave Life Sciences); explore combination with H3 approach

**Hypothesis 3 (Soluble vs. Insoluble Tau Ratio)**
- Most scientifically validated correlative evidence
- Anti-oligomer antibodies (RG6100, UCB0107) are in clinical development and represent a near-term translational path
- Differentiated from crowded ASO space
- **Recommended investment:** License or develop anti-truncated tau antibodies in parallel with ASO program; consider combination therapy

### Tier 2: Validate Mechanistically

**Hypothesis 6 (Caspase-6 Truncation)**
- Focus on **antibodies against truncated tau** (D421), not caspase inhibitors
- Develop **Cathepsin B inhibitor** programs as an alternative upstream approach
- **D421A knock-in mouse model** to definitively test causation
- **Recommended investment:** Mechanistic validation studies; develop anti-D421-tau antibody; de-risk before committing to clinical development

**Hypothesis 4 (PP2A/Fyn Balance)**
- **Immediate opportunity:** Analyze saracatinib repurposing data from completed AD trial (NCT02167256)
- Develop **SET-targeted ASO** as a differentiated PP2A activation strategy
- Investigate **PPP2R2A-preferring small molecule modulators**
- **Recommended investment:** Moderate; focus on the SET ASO approach given Ionis's ASO platform capability

### Tier 3: De-prioritize

**Hypotheses 1, 2, 7** — face compound-specific toxicity issues, failed clinical precedent, or fundamental target tractability problems that require either breakthrough innovation or are effectively non-viable as proposed.

---

## Integrated Therapeutic Strategy

Based on the competitive landscape and drug development reality, the **most viable integrated strategy** combines:

```
Foundational: MAPT ASO (H5) → partial tau reduction, 50-70%
    │
    ├── Add: Anti-oligomer antibody (H3) → clear released toxic species
    │
    └── Consider: Fyn inhibitor (H4) → protect vulnerable circuits
         (only if saracatinib data is compelling)

Exploratory: Anti-truncated tau antibody (H6) → intercept initiation
```

This combination addresses:
- The protective function of wild-type tau (H5 — preserve 30-50%)
- The toxic oligomer species (H3 — direct clearance)
- Circuit-specific vulnerability (H4 — Fyn)
- Initiation of pathology (H6 — truncated tau seeds)

**Single most critical near-term data point:** BIIB080 Phase II cognitive outcomes — this read-out will define the therapeutic viability of the entire tau reduction hypothesis space.