What amyloid threshold level is required for optimal clinical benefit in early AD?

Mechanistically-Specific Hypotheses: Amyloid Threshold Definition for Donanemab

Hypothesis 1: Synaptic Resilience Recovery Threshold

Title: PSD-95 Normalization Defines Amyloid Cessation Point

Mechanism: Donanemab-mediated amyloid clearance must reach a threshold (estimated 20-25 Centiloids) that permits synaptic protein synthesis recovery. PSD-95 (encoded by DLG4) requires sustained amyloid reduction to stabilize at postsynaptic densities. Below this threshold, excitatory synapse loss reverses; above it, synaptic degeneration continues despite treatment. This explains the non-linear dose-response curve observed in TRAILBLAZER-ALZ 2.

Key Evidence:

• Reduced CSF neurogranin predicts faster cognitive decline (PMID: 29739473)
• Amyloid burden inversely correlates with PSD-95 density in postmortem AD cortex (PMID: 29030421)

Testable Prediction: In TRAILBLAZER-ALZ 2 participants, CSF neurogranin (synaptic marker) will normalize (return to age-matched control levels) only when amyloid PET reaches <20 Centiloids. Patients stopping donanemab at 40 Centiloids will show persistent synaptic biomarker elevation, while those reaching <20 Centiloids will demonstrate biomarker normalization—measurable within 6 months of achieving threshold.

Primary Target: DLG4 (PSD-95) / Synaptic proteostasis machinery

Hypothesis 2: TREM2-Dependent Microglial State Transition Threshold

Title: TREM2 Activation State Determines Benefit Floor

Mechanism: Microglial TREM2 signaling requires amyloid reduction below a critical threshold to complete the transition from disease-associated microglia (DAM)/"dark microglia" back toward homeostatic states. Above this threshold (~30 Centiloids), chronic TREM2 engagement drives maladaptive inflammation via NLRP3 inflammasome activation (requires NLRP3, ASC specks) and sustained IL-1β secretion. Below the threshold, TREM2 signaling shifts toward phagocytic clearance without inflammatory exacerbation.

Key Evidence:

• TREM2 loss-of-function variants abolish amyloid plaque compaction and increase dystrophic neurites (PMID: 29263245)
• DAM signature requires TREM2 for amyloid-dependent induction (PMID: 28134504)

Testable Prediction: CSF GFAP (astrocyte) and IL-1β levels will demonstrate a biphasic pattern: elevated during active amyloid clearance (inflammatory response), but declining significantly only after amyloid reaches <30 Centiloids. Patients who discontinue treatment at higher thresholds (40-50 Centiloids) will show persistent neuroinflammation. TSPO-PET in a validation cohort will confirm microglial activation resolution at this threshold.

Primary Target: TREM2 / NLRP3 inflammasome axis

Hypothesis 3: Amyloid-Tau Seeding Quiescence Threshold

Title: Amyloid Below 15 Centiloids Halts Tau Propagation

Mechanism: Amyloid-β oligomers maintain a critical concentration threshold above which they sustain GSK3β and CDK5-mediated tau hyperphosphorylation at pathogenic sites (Thr231, Ser396). The "amyloid threshold for tau spreading" (~15 Centiloids, corresponding to ~50% plaque reduction from baseline) represents the point where monomeric Aβ concentrations fall below the critical micellar concentration required for toxic oligomer formation. Below this, soluble Aβ can no longer potentiate tau seeding activity in entorhinal cortex.

Key Evidence:

• Aβ oligomers directly activate GSK3β in neurons (PMID: 23728510)
• Amyloid reduction >50% correlates with slowed tau accumulation on PET (PMID: 34437572)

Testable Prediction: Longitudinal tau-PET (flortaucipir) will show annualized accumulation rate approaching zero exclusively in patients achieving amyloid <15 Centiloids. Those stopping at 25-40 Centiloids will continue tau spreading despite

Skeptic's Evaluation of Amyloid Threshold Hypotheses

Hypothesis 1: PSD-95 Normalization Threshold

Strongest Weakness: Presumed Causal Mechanism Without Direct Evidence

The hypothesis invokes PSD-95 (DLG4) stabilization as the mechanistic link between amyloid reduction and cognitive recovery, but the causal chain is underspecified. PSD-95 is a downstream synaptic protein whose expression reflects neuronal health—not a driver of recovery. The critical mechanistic question remains unanswered: What molecular pathway connects extracellular amyloid clearance to increased postsynaptic PSD-95 protein synthesis or stability?

Without this linkage, the hypothesis conflates correlation (synaptic loss correlates with amyloid) with mechanism (amyloid reduction causes PSD-95 recovery).

Counter-Evidence and Complications

• Alternative解释了 non-linear dose-response: The TRAILBLAZER-ALZ 2 dose-response may reflect pharmacodynamic saturation or patient selection effects rather than a specific biological threshold. Bimodal therapeutic effects are common in complex biological systems (PMID: 29739473 noted associations, not causation between neurogranin and outcomes).
• Synaptic biomarkers as surrogate markers: CSF neurogranin and PSD-95 changes may be epiphenomena of general neuronal health rather than mechanistic drivers. Patients with amyloid reduction but continued neurodegeneration would not fit this model.
• Temporal assumption unvalidated: The "6-month" prediction window is arbitrary. Synaptic protein turnover rates vary substantially between patients based on baseline neuronal integrity, age, and APOE genotype (PMID: 29030421 examined cross-sectional correlations, not temporal dynamics).

Pointed Question

If PSD-95 requires amyloid <20-25 Centiloids for "stabilization," what is the molecular mechanism by which extracellular amyloid plaques (removed from direct synaptic contact) signal intracellularly to regulate postsynaptic protein synthesis? Does this invoke a circulating soluble species (Aβ oligomers, Aβ*56), selective transport dysfunction, or network-level activity changes? The mechanism must connect peripheral amyloid plaques to dendritic protein translation machinery.

Confidence Rating: Weak

Rationale: The hypothesis identifies a legitimate observation (synaptic biomarkers correlate with amyloid burden) but inserts an unspecific protein (PSD-95) as the causal mediator without establishing the signaling pathway. The proposed threshold (20-25 Centiloids) is suspiciously precise given no direct measurements from the source trial. The field lacks direct human evidence linking amyloid PET thresholds to postsynaptic protein stabilization—this would require serial postmortem validation or live synaptic imaging, neither of which exists in TRAILBLAZER-ALZ 2.

Hypothesis 2: TREM2-Dependent Microglial State Transition

Strongest Weakness: TREM2 Biology Does Not Support a Binary State Transition Model

The hypothesis implies that TREM2-driven microglia exist in two discrete states (disease-associated vs. homeostatic) separable by a ~30 Centiloid amyloid threshold. This misrepresents current understanding of microglial biology. TREM2-dependent microglia operate on a spectrum, and their function is context-dependent based on ligand (TDP-43, myelin debris, apoptotic neurons, amyloid) rather than amyloid burden alone (PMID: 33355664).

Counter-Evidence and Complications

• DAM may be protective, not maladaptive: Disease-associated microglia in mouse models exhibit amyloid plaque localization and

Domain Expert Assessment: Amyloid Threshold Hypotheses for Donanemab

Executive Summary

The research gap you've identified is among the most operationally critical outstanding questions in anti-amyloid therapeutics. Current FDA approval language permits stopping when amyloid is "clinically appropriate," but this ambiguity creates real-world prescribing chaos. The hypotheses presented are mechanistically intriguing but carry significant translational gaps. I'll prioritize based on what can actually move clinical practice within the next 5 years.

Part 1: Hypotheses with Highest Translational Potential

Hypothesis A: CSF p-tau217 as Surrogate Threshold Marker

Why This Has Highest Translational Potential:

While the Theorist focused on synaptic markers, the current biomarker landscape strongly favors phosphorylated tau as the primary therapeutic monitoring tool. The mechanism is straightforward: amyloid reduction precedes tau reduction in the causal cascade, making p-tau217 a downstream integrator of both amyloid clearance and downstream neurodegeneration.

Current Clinical Evidence:

• P-tau217 demonstrates 85-90% sensitivity/specificity for amyloid positivity (Jansen et al., 2022; Palmqvist et al., 2020)
• In TRAILBLAZER-ALZ 2, plasma p-tau217 declined significantly with donanemab treatment and correlated with amyloid change
• The phase 2 data suggest p-tau217 normalization may precede or parallel cognitive stabilization—making it a candidate for threshold-based stopping rules
• Multiple CLIA-certified plasma assays now available (ALZpath, Fujirebio, Roche)

Safety Considerations:

• Low-risk biomarker monitoring—blood draw poses no direct patient harm
• Potential indirect risk: if stopping rules are too aggressive, patients might discontinue treatment before full benefit achieved
• Conversely, stopping at wrong threshold could leave patients with ongoing amyloid-mediated toxicity

Patient Population Fit:

• Excellent fit for early symptomatic AD (MCI-mild dementia) as studied in TRAILBLAZER-ALZ 2
• Challenging for preclinical populations where p-tau217 may not yet be elevated
• Particularly relevant for APOE4 carriers, who show different tau trajectories

Translational Pathway: The most actionable hypothesis because:

Assays are already validated and commercially available

Correlation with clinical endpoints exists in Phase 2/3 data

Could be incorporated into treatment guidelines within 2-3 years with prospective validation

Hypothesis B: Minimal Chronic Amyloid Burden (MCAB) Model

**Why This Des