# HSEF v5.3 review: Amyloid Cascade Hypothesis

> Sources: 56 primary reports; claims: 63; verified: 63; unverified: 0. Verification means bounded source attribution. Access: 4 selected partial-body examinations and 52 complete abstract/status examinations. Publication cutoff: 2026-10-04. No completed blinded calibration is claimed.
>
> **Adversarial review:** Generated by Sol 6.1; adversarially reviewed by Claude Opus 5.5 (Anthropic) on 2026-10-06. 1 corrections applied (see ADVERSARIAL REVIEW AND APPROVAL). Status: **Approved as revised.**

## AT A GLANCE

**Amyloid is a causal contributor in defined early AD, but the universal primary-initiator cascade remains only partly identified.**

| Level | Confidence | Causal role | Momentum | Stability | Divergence |
|---|---|---|---|---|---|
| Γ — Mechanistically plausible, partial system support | Moderate | reinforcing / stabilizing factor | Decisive work underway | Uncertain | Contested |

**Deciding experiment:** Randomize preclinical amyloid-positive participants to verified aggregate reduction with prospectively defined clinical follow-up, baseline tau/vascular stratification and blinded outcomes. Include a species-selective mediator contrast where feasible. Prevention of tau-linked injury and clinical onset would strengthen initiation; adequate sustained engagement with unchanged clinical onset would constrain it. Preserve non-Aβ APP/BACE functions and monitor harms.

**Belief versus evidence:** contribution, exact species, initiation and universal primacy require different causal tests.

**Best-supported rival:** Heterogeneous multi-hit AD; Γ. APOE3 Christchurch heterozygosity (PMID38899694) separates the heterogeneous model from a simple plaque-primary trajectory: delayed impairment and limited tau occur despite appreciable plaque in selected autopsies. This is a bounded natural contrast, not definitive universal refutation.

## COMPETING MECHANISTIC MODELS

**Rank 1: Heterogeneous multi-hit AD (Γ).** Multiple interacting initiating and amplifying pathways produce AD trajectories, with context-dependent amyloid contribution. Mixed pathology, familial/sporadic differences, resilience and clinical stage dependence fit heterogeneity. A flexible multi-hit model can absorb observations without a uniquely discriminating intervention; it is not itself proved.

**Rank 2: Amyloid-primary cascade (Γ).** Amyloid accumulation is the primary initiating driver upstream of tau and clinical injury across AD. APP genetics, knock-in/conditional perturbations and positive clinical antibodies support a causal contribution. Prevention/agent nulls, nonuniform tau order and unisolated human species/primacy block the complete initiating claim.

**Rank 3: Tau-primary/selective vulnerability (Γ).** Tau-associated injury and neuronal vulnerability determine degeneration; amyloid facilitates rather than universally initiates it. Tau reduction with amyloid preserved and tau-linked atrophy/resilience support a proximate injury role. Model rescue does not demonstrate tau is the earliest human initiating event; amyloid-directed clinical benefits constrain an amyloid-inert account.

**Rank 4: Vascular/clearance failure (Γ).** Vascular and clearance disturbances drive some injury and amyloid retention. APOE4 BBB prognosis and endothelial/model perturbations support a vascular contribution. Human observational prediction and 5xFAD endothelial biology do not establish vascular primacy for all AD.

**Rank 5: Inflammatory initiation (Γ).** Glial innate pathways initiate injury and drive amyloid/tau downstream. STING-related functional intervention and glial marker coupling support inflammatory involvement. Pathway blockade can interrupt amplification without proving first initiation; the human marker results remain observational.

Tournament ranking does not alter the nominal level. Models can be partially compatible; a flexible rival is not validated merely by explaining heterogeneous observations.

## FULL REVIEW

### Registered claim and scope

Aβ accumulation is the primary upstream driver of Alzheimer's disease, initiating tau pathology, synaptic failure and progressive neurodegeneration in familial and sporadic disease.

The index prescribes scope-matched assessment; SC4 carries the full sporadic primacy/generalization scope. SC5 is an added narrower therapeutic subclaim and cannot carry initiation.

Selection is rank 1 in the editorial public-interest list compiled 2026-09-23. Existing categorical verdicts were visible; grading was not blinded. Public pages and prior prose remain unchanged.

### Evidence and causal interpretation

The central question is primacy. Lecanemab and donanemab changed clinical cognition/function, not only PET. Lecanemab's primary CDR-SB difference was −0.45 points at 18 months, with both arms worsening [1]. Donanemab favored the prespecified clinical populations, also with decline in both arms [2]. These results are positive causal evidence for bounded contribution; modest effect, harms and missing mediation do not turn them into nulls.

The negatives remain equally visible. Gantenerumab lowered plaque without significant primary clinical slowing [3]. Aducanumab trials disagreed [4]. Preclinical and symptomatic solanezumab were null [5,6]. Verubecestat, lanabecestat and atabecestat were ineffective or adverse [7–10]. Target form, engagement, stage, duration and BACE's other substrates constrain how broadly these results can falsify the nominal. The 2026 Colombian crenezumab report is a genuine completed prevention trial; it did not significantly move either primary clinical endpoint and did not clearly remove plaques [18].

Natural genetic instruments provide upstream evidence beyond postmortem association. APP missense, duplication and protective processing variation [15–17] are strong within their scopes. Early-onset apparently sporadic cases can carry de novo pathogenic mutations [20], which does not make them representative late-onset sporadic disease. APP and presenilins also affect precursor fragments, autophagy and endolysosomal physiology [21–24]. A genetic instrument is not automatically a single-species selective perturbation.

Defined model causal results are not discarded. Tau reduction prevented APP-model behavioral deficits despite preserved Aβ [43]. The 2024 inducible oligodendrocyte BACE1 study restored early neuronal firing in APP knock-in mice, with tamoxifen treatment after most developmental myelination and matched treated controls; Aβ-depleted conditioned medium did not trigger the same acute hyperactivity [46]. This supports bounded physiological injury beyond overexpression. It does not isolate oligomer species, eliminate BACE pleiotropy, or measure human clinical neurodegeneration.

Tau chronology is not one observation. Amyloid-related longitudinal p-tau, connectivity and GFAP findings [34–36] support an interacting cascade but do not turn statistical mediation into causal manipulation. Early tau-positive preclinical atrophy [37], age-related medial-temporal tau [51], APP-stage mitochondrial alterations [48], and BBB-related prognostic injury [54] complicate a single sequence. A biomarker threshold crossed late is not molecular absence beforehand [40].

Longer follow-up is encouraging and less decisive than new placebo-controlled efficacy. Clarity and TRAILBLAZER extensions reuse participants and add open-label or external-control limitations [26,27,29]. The 2025 DIAN extension's final primary endpoint was plaque PET; its main-control clinical confidence intervals included no effect, while one extended-control CDR-SB contrast favored long exposure [33]. The prior randomized DIAN test was clinically null and its primary model did not converge, although sensitivity analyses agreed [52]. Its small selected autopsy series showed plaque lowering without detectable tau/glial response [55]. These distinctions prevent a possible delay signal from being relabeled proven prevention.

The Christchurch kindred is a strong constraint on sufficiency of plaque burden as a trajectory explanation [19]. Matched impairment ages differed, while selected autopsy plaque/tau comparisons were qualitative rather than quantitative. The resilient PSEN2 case [56] is hypothesis-generating, not a frequency estimate. The best-supported broad interpretation is heterogeneous interacting disease with a real amyloid contribution; its flexible breadth does not earn a high causal level by itself.

## HYPOTHESIS STATE EVALUATION FRAMEWORK (HSEF v5.3)

<!-- consensus-consistency-2026-10-09 -->
### Consensus provenance reconciliation — 9 October 2026

Case-specific Consensus discovery records were located and linked to this final edition. This supplements the review's provenance; it leaves the original assessed claim, grades and source counts intact. Provider summaries and full-text badges do not establish examination of original primary passages by this reviewer. Source access and incorporation retain the limits stated in the original methods and any dated research update.

Recorded evidence artifacts:

- `deliverables/hsef_v53_popularity_batch_2026-10-04/amyloid-cascade/consensus/evidence-cards.json`
- `deliverables/hsef_v53_popularity_batch_2026-10-04/amyloid-cascade/consensus/evidence-cards.md`
- `deliverables/hsef_v53_popularity_batch_2026-10-04/amyloid-cascade/discovery/consensus-targeted.json`
- `deliverables/hsef_v53_popularity_batch_2026-10-04/amyloid-cascade/discovery/consensus.json`
- `deliverables/prepublication-enrichment-2026-10-05/website-series/amyloid-cascade/consensus-browser-links.json`

The located artifacts do not state a reliable source-comparison count. No count is inferred.

**Existing source reconciliation:** Yes; 11 timeline primary identifiers cross-checked per dated ledger; 1 selected body check plus 10 abstract excerpts; timeline, gaps, access limits documented. Selected checks are not complete-paper examinations (ledger says 0 complete papers). Raw Consensus is discovery/provenance, not the independently verified passage evidence.

Matched primary identifiers in the preserved dated ledger: PMID 1671712, PMID 16369530, PMID 17478722, PMID 22784036, PMID 22801501, PMID 31686034, PMID 36449413, PMID 37459141, PMID 37966285, PMID 38381674, PMID 40120616. These are historical checks, not newly credited complete-paper examinations.

Shared template wording about an amyloid-specific Deep follow-up is not credited as a case-specific run for this topic. The actual raw case records above establish provider use.

A dated adversarial approval record exists for the separately identified amyloid-cascade.review.md (2026-10-06; see edition-approval-audit.json). Its scope is that recorded edition. This supplement has no separate approval and no coverage is inferred from topic identity. No score or confidence change is made by this reconciliation.
<!-- end-consensus-consistency -->


### Subclaim matrix and aggregation

| Subclaim | Scope | Level | Decision |
|---|---|---|---|
| SC1 Aβ plaque accumulation is the primary driver of neurodegeneration in AD. | familial and sporadic AD | Γ | Plaque-targeting treatment changes early clinical trajectories, but a primary plaque-specific initiation mechanism is not isolated across AD. Plaques, soluble aggregates and immune clearance are not interchangeable mediators. |
| SC2 Soluble Aβ oligomers are the causally active species. | AD-related injury, with human attribution scored separately | Γ | Human extracts and selective model manipulations support soluble Aβ toxicity. BACE1 suppression is not oligomer-selective, and human CSF oligomer engagement has not isolated oligomer-specific clinical mediation. Defined mouse hyperactivity causality is retained separately at Ε. |
| SC3 Aβ is upstream of tau in the pathogenic cascade. | familial and sporadic AD | Γ | Amyloid can facilitate model injury and modulate tau markers; wider tau spread associates with amyloid in people. Early medial-temporal tau, human resilience, regional/null downstream responses and absent selective human mediation limit universal temporal primacy. |
| SC4 Familial Aβ-cascade causality generalizes to sporadic AD as its primary initiating explanation. | sporadic AD across baseline disease stages | Γ | Lifelong APP processing instruments and early-AD randomized clinical effects establish more than association, but neither proves that the full familial primary-initiation cascade explains sporadic disease across stages. This is the scope-matched gate. |
| SC5 Amyloid-directed intervention changes clinical trajectory in biomarker-confirmed early symptomatic AD. | baseline amyloid-positive MCI or mild dementia due to AD | Ε | Positive lecanemab and donanemab clinical comparisons cross Δ, with a defined early amyloid-positive symptomatic population and contrasting stages/agents. Ε records mapped context and pharmacology boundaries. Antibody replications alone are one mechanistic modality; familial instruments do not independently establish this specific treatment subclaim. |

**Preregistered aggregation:** scope-matched. Scope-matched SC4 returns Γ for primary initiation/generalization; SC5 separately returns Ε for early clinical contribution. Human genetics and clinical perturbations are considered together and are not re-capped as mere association. Their convergence supports bounded amyloid contribution, while sole/primary initiation and exact pathogenic-species mediation remain unresolved. Weakest-link: Γ; primary-subclaim SC4: Γ; scope-matched: Γ.

### Decisive-perturbation inventory and causal tally

[1] Lecanemab slowed cognitive and functional decline in amyloid-confirmed early AD, with adverse events. Clinical slowing does not establish complete mediation, universal initiation, halt or reversal of disease.

[2] Donanemab favored clinical outcomes in early symptomatic amyloid/tau-positive AD. Both groups worsened; this is stage-specific contribution rather than proof of a sole initiator.

[3] Gantenerumab reduced plaque and fluid tau markers without significant primary clinical slowing in either GRADUATE trial. Incomplete plaque removal and target/stage differences limit a universal null inference; tau PET was similar.

[5] Solanezumab did not slow preclinical cognitive decline; amyloid PET increased in both arms. Monomer targeting and incomplete prevention of deposition do not test complete plaque eradication.

[8] Prodromal verubecestat was ineffective and the higher dose worsened the primary clinical outcome. Early stopping, BACE substrate pleiotropy and drug harms prevent equating this with selective Aβ harm removal.

[15] APP A673T was associated with protection and reduced amyloidogenic processing in functional assays. A rare lifelong processing variant is stronger than pathology co-occurrence but does not isolate every APP product or all sporadic cases.

[18] The completed Colombian crenezumab prevention trial found no significant benefit on either primary clinical endpoint and no clear plaque-removal effect. It cannot be described as a successful plaque-clearance prevention test; limited engagement constrains interpretation.

[19] APOE3 Christchurch heterozygotes in the PSEN1 kindred had delayed impairment; selected autopsies showed substantial plaque with limited tau. Related-kindred sampling, sparse imaging/autopsy and qualitative unquantified pathology do not prove amyloid is causally irrelevant.

[43] Endogenous tau reduction prevented APP-transgenic behavioral deficits without lowering Aβ. Tau dependency in this model does not prove tau is the earliest human initiator; transgenic Aβ exposure still limits scope.

[46] Inducible oligodendrocyte BACE1 deletion reduced plaque burden and restored early neuronal firing in APP knock-in mice; Aβ-depleted conditioned medium lost acute hyperactivity effects. BACE1 is pleiotropic, exposure began in young mice, and neuronal firing is not human clinical progression; injected medium is a separate forced sufficiency test.

[52] DIAN randomized antibodies engaged targets without clinical benefit; gantenerumab moved tau and NfL markers. The primary disease model did not converge, but sensitivity analyses agreed; late dose escalation and no asymptomatic decline limit prevention inference.

Tally: 11 discriminating perturbation/natural contrasts; nulls and negatives remain separate from positive causal support. None selectively identifies the entire universal sporadic initiating cascade. Replication and reports do not equal independent lines.

### Cap audit, precedence and Β/Γ/Δ boundary

Universal checks: “In vitro evidence alone cannot exceed Γ.” “Pure association evidence — however large, however replicated — cannot exceed Β.” “Biomarker change without functional/phenotypic change caps at Γ.” These govern isolated inferences, not the assembled genetic/clinical set. Overexpression and forced extract exposure remain bounded; conditional endogenous suppression and human clinical outcomes are retained. Familial-to-sporadic scope is independently adjudicated.

**SC1.** Field-valid definition: 
- Endogenous genetic deletion, suppression, or degron depletion of the proposed mediator in vivo
- Selective pathway inhibition that changes onset/progression while preserving broader viability
- Vagotomy-type natural experiment, only if confounding is rigorously controlled AND effect is disease-initiation relevant rather than merely associated Qualifying full-scope evidence: unclear; positive full-scope result: no; phenotype relevance: yes. Narrow positive results are retained; no positive perturbation identifies the complete nominal primary-initiation inference for this subclaim. Ceiling: Γ.

**SC2.** Field-valid definition: 
- Endogenous genetic deletion, suppression, or degron depletion of the proposed mediator in vivo
- Selective pathway inhibition that changes onset/progression while preserving broader viability
- Vagotomy-type natural experiment, only if confounding is rigorously controlled AND effect is disease-initiation relevant rather than merely associated Qualifying full-scope evidence: unclear; positive full-scope result: no; phenotype relevance: yes. Narrow positive results are retained; no positive perturbation identifies the complete nominal primary-initiation inference for this subclaim. Ceiling: Γ.

**SC3.** Field-valid definition: 
- Endogenous genetic deletion, suppression, or degron depletion of the proposed mediator in vivo
- Selective pathway inhibition that changes onset/progression while preserving broader viability
- Vagotomy-type natural experiment, only if confounding is rigorously controlled AND effect is disease-initiation relevant rather than merely associated Qualifying full-scope evidence: unclear; positive full-scope result: no; phenotype relevance: yes. Narrow positive results are retained; no positive perturbation identifies the complete nominal primary-initiation inference for this subclaim. Ceiling: Γ.

**SC4.** Field-valid definition: 
- Endogenous genetic deletion, suppression, or degron depletion of the proposed mediator in vivo
- Selective pathway inhibition that changes onset/progression while preserving broader viability
- Vagotomy-type natural experiment, only if confounding is rigorously controlled AND effect is disease-initiation relevant rather than merely associated Qualifying full-scope evidence: unclear; positive full-scope result: no; phenotype relevance: yes. Narrow positive results are retained; no positive perturbation identifies the complete nominal primary-initiation inference for this subclaim. Ceiling: Γ.

**SC5.** Field-valid definition: 
- Endogenous genetic deletion, suppression, or degron depletion of the proposed mediator in vivo
- Selective pathway inhibition that changes onset/progression while preserving broader viability
- Vagotomy-type natural experiment, only if confounding is rigorously controlled AND effect is disease-initiation relevant rather than merely associated Qualifying full-scope evidence: yes; positive full-scope result: yes; phenotype relevance: yes. Lecanemab and donanemab favor clinical cognition/function, not only biomarkers. Ceiling: Δ or higher.

SC1 cap: “Maximum if mechanism is biologically plausible, supported by in vitro or simplified animal paradigms, but no disease-relevant necessity is shown in vivo.” Plaque-targeting treatment changes early clinical trajectories, but a primary plaque-specific initiation mechanism is not isolated across AD. Plaques, soluble aggregates and immune clearance are not interchangeable mediators. Narrow causal support does not isolate full primacy/species/generalization.

SC2 cap: “Maximum if mechanism is biologically plausible, supported by in vitro or simplified animal paradigms, but no disease-relevant necessity is shown in vivo.” Human extracts and selective model manipulations support soluble Aβ toxicity. BACE1 suppression is not oligomer-selective, and human CSF oligomer engagement has not isolated oligomer-specific clinical mediation. Defined mouse hyperactivity causality is retained separately at Ε. Narrow causal support does not isolate full primacy/species/generalization.

SC3 cap: “Maximum if mechanism is biologically plausible, supported by in vitro or simplified animal paradigms, but no disease-relevant necessity is shown in vivo.” Amyloid can facilitate model injury and modulate tau markers; wider tau spread associates with amyloid in people. Early medial-temporal tau, human resilience, regional/null downstream responses and absent selective human mediation limit universal temporal primacy. Narrow causal support does not isolate full primacy/species/generalization.

SC4 cap: “Maximum if mechanism is biologically plausible, supported by in vitro or simplified animal paradigms, but no disease-relevant necessity is shown in vivo.” Lifelong APP processing instruments and early-AD randomized clinical effects establish more than association, but neither proves that the full familial primary-initiation cascade explains sporadic disease across stages. This is the scope-matched gate. Narrow causal support does not isolate full primacy/species/generalization.

Γ hard-rule checks: “In vitro aggregation alone”; “In vitro seeding alone”; “Droplet aging in vitro”; “Organoid toxicity without organismal follow-through”; “Exogenous seeding models alone”. None describes the complete genetic/clinical set. Simplified and forced assays remain insufficient for the broad human inference. A perturbation crosses within its actual system; scope generalization requires separate support.

### Upward threshold ladder

SC1 Δ: “Requires at least one genuine causal perturbation in a disease-relevant in vivo system, or equivalent human natural experiment that materially supports the mechanism.” Satisfied: no. Clinical trajectory following selective plaque perturbation with functional endpoints, specificity and prospectively defined scope.

SC2 Δ: “Requires at least one genuine causal perturbation in a disease-relevant in vivo system, or equivalent human natural experiment that materially supports the mechanism.” Satisfied: no. Oligomer-selective depletion with plaque burden preserved with functional endpoints, specificity and prospectively defined scope.

SC3 Δ: “Requires at least one genuine causal perturbation in a disease-relevant in vivo system, or equivalent human natural experiment that materially supports the mechanism.” Satisfied: no. Aβ perturbation preventing or delaying tau and clinical injury with functional endpoints, specificity and prospectively defined scope.

SC4 Δ: “Requires at least one genuine causal perturbation in a disease-relevant in vivo system, or equivalent human natural experiment that materially supports the mechanism.” Satisfied: no. Human genetics and clinical intervention separating contribution from initiation/primacy with functional endpoints, specificity and prospectively defined scope.

SC5 Δ: “Requires at least one genuine causal perturbation in a disease-relevant in vivo system, or equivalent human natural experiment that materially supports the mechanism.” Satisfied: yes. 

SC5 Ε: “Requires causal support in a defined context, but with clear restriction by subtype, disease stage, model, or tissue.

Use Ε when there is a real causal contribution, but generalization to the entire disease class fails.” Satisfied: yes. 

SC5 Ζ: “Requires:
- Multiple independent causal lines in vivo
- Replication across at least two relevant systems
- Narrowing of major alternative models” Satisfied: no. Randomized clinical outcomes with target engagement with functional endpoints, specificity and prospectively defined scope.

The upper ladder remains reachable. Human genetics and therapeutic clinical perturbations are eligible equivalents. Their shared/distinct failure modes are considered; they do not isolate the complete nominal primary initiation or exact species mediation. No deliberate human initiation RCT is required.

### Falsification and pitfalls

Subtype heterogeneity shows the mechanism is non-general but the hypothesis is framed as general. The source cards retain counterexamples and intervention nulls; these constrain the same generalization gap already used in the Γ assessment. No additional downgrade duplicates an already governing scope limitation.

Blocking the proposed mechanism does not alter onset/progression in vivo. Negative agents and stages do not erase the separately positive lecanemab/donanemab contexts. Engagement/specificity limits forbid treating every null as total-mechanism blockade. No additional downgrade duplicates an already governing scope limitation.

6. Downstream effect treated as upstream driver. Clinical slowing is contribution, not automatic earliest initiation.

10. Mechanism conflation. Plaques, oligomers, BACE and precursor/presenilin functions are distinguished.

12. Temporal order inferred rather than demonstrated. The 2012 DIAN baseline analysis is cross-sectional, despite prospective study enrollment.

13. Therapeutic effect misread as mechanistic proof. Target engagement and clinical benefit do not isolate every mediator.

17. Negative field-valid perturbation read as a crossed threshold. Negative drug results do not count as positive Δ support.

21. A universal cap re-applied to a convergent evidence set. Human instruments and clinical evidence are evaluated together.

22. Scope fitted to the result. Baseline stage and genotype define contexts; responding patients do not define scope.

### Adversarial steelman: above, then below

For Δ, lifelong protective APP processing variation, familial gene-dose instruments, endogenous mouse depletion and positive human clinical intervention form a compelling causal set. A charitable reading would treat this as proof of the primary upstream cascade in sporadic disease.

For Β, broad human initiation/primacy is still largely inferred from pathology chronology; prevention and BACE trials are null or harmful, and severe amyloid can coexist with delayed cognitive impairment.

Reject Δ for the exact primary-initiation gate: clinical contribution and APP processing risk are not selective identification of the full initiating cascade in sporadic AD. Reject Β for the mechanistic whole: verified intervention, functional depletion and genetic instruments are not mere co-occurrence. Γ does not downgrade the independently positive SC5.

### Final score, causal direction and M/S/D

Scope-matched SC4 returns Γ for primary initiation/generalization; SC5 separately returns Ε for early clinical contribution. Human genetics and clinical perturbations are considered together and are not re-capped as mere association. Their convergence supports bounded amyloid contribution, while sole/primary initiation and exact pathogenic-species mediation remain unresolved.

At sporadic disease scope, production solely by established disease is not demonstrated (step 1), and intervention-sensitive clinical contribution rejects co-occurrence alone (step 2). Removal, initiation and rate limitation are not jointly established (step 3), nor universal necessity (step 4). Amyloid amplifies or maintains injury in the observed early-disease contexts, so step 5 is the first defensible match. Familial APP pathogenicity is stronger and separately scoped.

Momentum: Decisive work underway. TRAILBLAZER-ALZ 3 (NCT05026866): preclinical donanemab prevention; screening/baseline report PMID40955720 is a discovery lead and is not counted as a completed outcome trial.

Stability: Uncertain. The named preclinical plaque-removal prevention trial could strengthen or weaken amyloid-initiation inference depending on clinical outcomes and verified target engagement. Neither direction is presumed.

Divergence: Contested; no single field-implied level or signed arithmetic gap is asserted. M/S/D were assigned last and change neither level nor confidence.

## PROPOSED DECISIVE EXPERIMENTS

Randomize preclinical amyloid-positive participants to verified aggregate reduction with prospectively defined clinical follow-up, baseline tau/vascular stratification and blinded outcomes. Include a species-selective mediator contrast where feasible. Prevention of tau-linked injury and clinical onset would strengthen initiation; adequate sustained engagement with unchanged clinical onset would constrain it. Preserve non-Aβ APP/BACE functions and monitor harms.

## KNOWLEDGE GAPS

Selective human oligomer-versus-plaque clinical mediation remains unisolated.

Preventive target engagement and clinical outcomes require a coherent prospective bridge.

Familial, late-onset sporadic, tau-defined and vascular/mixed scopes need prospective stratification.

Most of this bounded corpus was examined at complete-abstract/status depth, not full report/supplement depth; linked correction bodies remain incompletely examined.

## REVIEWER NOTES

Targeted critical review of 56 primary reports, not an exhaustive systematic review or a blinded calibration. Search leads, reviews, protocols and notices do not pad the primary count. Selected body access is limited to the displayed sections recorded in receipts; downloaded XML without examination earns no body credit. Corrections attach to parent papers. Exact correction contents for the Study201, titration, DIAN chronology and DIAN-OLE notices were not fully examined; their fine numerical details are not used to establish a level. Lanabecestat’s PET-sign correction was verified in the preceding batch and is not used as a new outcome source. Shared study families are explicit in the ledger.

## ADVERSARIAL REVIEW AND APPROVAL

**Authoring model:** Sol 6.1. **Adversarial reviewer:** Claude Opus 5.5 (Anthropic). **Date:** 2026-10-06.

**Scope of checks:** All 56 reference PMIDs re-fetched with the PubMed MCP tool get_article_metadata and cross-checked against NCBI E-utilities (title, first author, journal, year, DOI): 56/56 resolve correctly; no retraction or expression of concern; 5 linked errata, all already disclosed on the evidence cards. 20 load-bearing claims spot-checked against PubMed abstracts and, for PMIDs 38899694 (sections S14) and 40120616 (S17, S19), the local PMC full-text passages: refs 1, 2, 3, 4, 5, 8, 12, 15, 18, 19, 20, 33, 37, 43, 46, 51, 52, 54, 55, 56. Scoring logic checked against SKILL.md Step 3 and hsef-framework.md (universal caps, cap precedence, Β/Γ/Δ boundary, scope-matched aggregation and reported alternatives).

**Findings and corrections**

| ID | Severity | Location | Original → corrected | Evidence |
|---|---|---|---|---|
| AC-1 | Editorial | Evidence card [53] status (also scorecard ledger, evidence.json) | 'ErratumIn: PMIDnot supplied' → 'ErratumIn: N Engl J Med 2012;367(8):780 (erratum notice has no PMID)' | PubMed CommentsCorrections for PMID 22784036 lists the erratum by citation only, with no PMID |

**Rejected findings (checked, no change)**

- SC1/SC4 Γ rather than Δ (the framework's worked example scores amyloid Δ): the review grades the primacy and initiation subclaims, and the lecanemab/donanemab Δ-qualifying evidence is scored separately in SC5 at Ε, as hypothesis-index.md asks ('Δ-qualifying for the trajectory-modification subclaim'). This is a defensible judgement, and no framework text forces Δ for a primacy subclaim. Scope-matched SC1/SC3/SC4 = Γ; weakest-link Γ; alternatives are reported correctly.
- APOE3 Christchurch claim ('substantial plaque with limited tau' in selected autopsies): confirmed in the PMC body, section S14 ('greater amyloid-β plaque burden and a relatively limited tau burden … qualitative visual inspection … not quantified').
- DIAN-TU OLE claim that one extended-control CDR-SB contrast favoured long exposure: confirmed in section S19 (HR 0.53, 95% CI 0.29–0.97), and the main-control CIs include 1.
- Crenezumab API ADAD (PMID 41579901, 2026): the null dual primary endpoints and the absence of significant plaque removal match the abstract.
- Year differences [1] (2023 print vs 2022 e-pub) and [17] (2006 vs 2005): acceptable under the print/e-pub convention.
- Lecanemab −0.45 CDR-SB, gantenerumab GRADUATE nulls with p-tau181 lowering and similar tau PET, A4 solanezumab null with PET rising in both arms, prodromal verubecestat worsening at 40 mg, A673T ~40% reduction in vitro: all match the abstracts.

**Residual limitations**

- Most sources were verified at abstract depth only. Partial-body verification covered the 4 receipted papers, plus the 2 passages re-read in this review.
- Scorecard provenance_header.generated_by reads 'Codex; HSEF v5.3'. Authorship is recorded here as Sol 6.1 per the review mandate; the field was not altered.
- Batch ZIP archives and pre-review hashes are pre-review snapshots and were not rebuilt.

**Approval:** Following adversarial review and the corrections listed above, the revised document is approved. — Claude Opus 5.5 (Anthropic), 2026-10-06.

## REFERENCES

1. van Dyck CH, Swanson CJ, Aisen P et al. Lecanemab in Early Alzheimer's Disease. The New England journal of medicine. 2023. [PMID 36449413](https://pubmed.ncbi.nlm.nih.gov/36449413/). DOI: 10.1056/NEJMoa2212948. Access: abstract_metadata. Study: NCT03887455.

2. Sims JR, Zimmer JA, Evans CD et al. Donanemab in Early Symptomatic Alzheimer Disease: The TRAILBLAZER-ALZ 2 Randomized Clinical Trial. JAMA. 2023. [PMID 37459141](https://pubmed.ncbi.nlm.nih.gov/37459141/). DOI: 10.1001/jama.2023.13239. Access: abstract_metadata. Study: NCT04437511.

3. Bateman RJ, Smith J, Donohue MC et al. Two Phase 3 Trials of Gantenerumab in Early Alzheimer's Disease. The New England journal of medicine. 2023. [PMID 37966285](https://pubmed.ncbi.nlm.nih.gov/37966285/). DOI: 10.1056/NEJMoa2304430. Access: abstract_metadata. Study: PMID37966285.

4. Budd Haeberlein S, Aisen PS, Barkhof F et al. Two Randomized Phase 3 Studies of Aducanumab in Early Alzheimer's Disease. The journal of prevention of Alzheimer's disease. 2022. [PMID 35542991](https://pubmed.ncbi.nlm.nih.gov/35542991/). DOI: 10.14283/jpad.2022.30. Access: abstract_metadata. Study: PMID35542991.

5. Sperling RA, Donohue MC, Raman R et al. Trial of Solanezumab in Preclinical Alzheimer's Disease. The New England journal of medicine. 2023. [PMID 37458272](https://pubmed.ncbi.nlm.nih.gov/37458272/). DOI: 10.1056/NEJMoa2305032. Access: abstract_metadata. Study: PMID37458272.

6. Honig LS, Vellas B, Woodward M et al. Trial of Solanezumab for Mild Dementia Due to Alzheimer's Disease. The New England journal of medicine. 2018. [PMID 29365294](https://pubmed.ncbi.nlm.nih.gov/29365294/). DOI: 10.1056/NEJMoa1705971. Access: abstract_metadata. Study: PMID29365294.

7. Egan MF, Kost J, Tariot PN et al. Randomized Trial of Verubecestat for Mild-to-Moderate Alzheimer's Disease. The New England journal of medicine. 2018. [PMID 29719179](https://pubmed.ncbi.nlm.nih.gov/29719179/). DOI: 10.1056/NEJMoa1706441. Access: abstract_metadata. Study: NCT01739348.

8. Egan MF, Kost J, Voss T et al. Randomized Trial of Verubecestat for Prodromal Alzheimer's Disease. The New England journal of medicine. 2019. [PMID 30970186](https://pubmed.ncbi.nlm.nih.gov/30970186/). DOI: 10.1056/NEJMoa1812840. Access: abstract_metadata. Study: NCT01953601.

9. Sperling R, Henley D, Aisen PS et al. Findings of Efficacy, Safety, and Biomarker Outcomes of Atabecestat in Preclinical Alzheimer Disease: A Truncated Randomized Phase 2b/3 Clinical Trial. JAMA neurology. 2021. [PMID 33464300](https://pubmed.ncbi.nlm.nih.gov/33464300/). DOI: 10.1001/jamaneurol.2020.4857. Access: abstract_metadata. Study: PMID33464300.

10. Wessels AM, Tariot PN, Zimmer JA et al. Efficacy and Safety of Lanabecestat for Treatment of Early and Mild Alzheimer Disease: The AMARANTH and DAYBREAK-ALZ Randomized Clinical Trials. JAMA neurology. 2020. [PMID 31764959](https://pubmed.ncbi.nlm.nih.gov/31764959/). DOI: 10.1001/jamaneurol.2019.3988. Access: abstract_metadata. Study: AMARANTH+DAYBREAK.

11. Mintun MA, Lo AC, Duggan Evans C et al. Donanemab in Early Alzheimer's Disease. The New England journal of medicine. 2021. [PMID 33720637](https://pubmed.ncbi.nlm.nih.gov/33720637/). DOI: 10.1056/NEJMoa2100708. Access: abstract_metadata. Study: NCT03367403.

12. Shcherbinin S, Evans CD, Lu M et al. Association of Amyloid Reduction After Donanemab Treatment With Tau Pathology and Clinical Outcomes: The TRAILBLAZER-ALZ Randomized Clinical Trial. JAMA neurology. 2022. [PMID 36094645](https://pubmed.ncbi.nlm.nih.gov/36094645/). DOI: 10.1001/jamaneurol.2022.2793. Access: abstract_metadata. Study: NCT03367403.

13. Yang T, Dang Y, Ostaszewski B et al. Target engagement in an alzheimer trial: Crenezumab lowers amyloid β oligomers in cerebrospinal fluid. Annals of neurology. 2019. [PMID 31168802](https://pubmed.ncbi.nlm.nih.gov/31168802/). DOI: 10.1002/ana.25513. Access: abstract_metadata. Study: ABBY+BLAZE.

14. Vandenberghe R, Rinne JO, Boada M et al. Bapineuzumab for mild to moderate Alzheimer's disease in two global, randomized, phase 3 trials. Alzheimer's research & therapy. 2016. [PMID 27176461](https://pubmed.ncbi.nlm.nih.gov/27176461/). DOI: 10.1186/s13195-016-0189-7. Access: abstract_metadata. Study: PMID27176461.

15. Jonsson T, Atwal JK, Steinberg S et al. A mutation in APP protects against Alzheimer's disease and age-related cognitive decline. Nature. 2012. [PMID 22801501](https://pubmed.ncbi.nlm.nih.gov/22801501/). DOI: 10.1038/nature11283. Access: abstract_metadata. Study: PMID22801501.

16. Goate A, Chartier-Harlin MC, Mullan M et al. Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer's disease. Nature. 1991. [PMID 1671712](https://pubmed.ncbi.nlm.nih.gov/1671712/). DOI: 10.1038/349704a0. Access: abstract_metadata. Study: PMID1671712.

17. Rovelet-Lecrux A, Hannequin D, Raux G et al. APP locus duplication causes autosomal dominant early-onset Alzheimer disease with cerebral amyloid angiopathy. Nature genetics. 2006. [PMID 16369530](https://pubmed.ncbi.nlm.nih.gov/16369530/). DOI: 10.1038/ng1718. Access: abstract_metadata. Study: PMID16369530.

18. Tariot PN, Lopera FS, Ríos-Romenets S et al. Safety and efficacy of crenezumab in cognitively unimpaired carriers of the PSEN1Glu280Ala mutation at risk for autosomal-dominant Alzheimer's disease in Colombia (API ADAD Colombia Trial): a phase 2, randomised, double-blind, placebo-controlled trial. The Lancet. Neurology. 2026. [PMID 41579901](https://pubmed.ncbi.nlm.nih.gov/41579901/). DOI: 10.1016/S1474-4422(25)00426-0. Access: abstract_metadata. Study: PMID41579901.

19. Quiroz YT, Aguillon D, Aguirre-Acevedo DC et al. APOE3 Christchurch Heterozygosity and Autosomal Dominant Alzheimer's Disease. The New England journal of medicine. 2024. [PMID 38899694](https://pubmed.ncbi.nlm.nih.gov/38899694/). DOI: 10.1056/NEJMoa2308583. Access: full_text_partial. Study: PMID38899694.

20. Lanoiselée HM, Nicolas G, Wallon D et al. APP, PSEN1, and PSEN2 mutations in early-onset Alzheimer disease: A genetic screening study of familial and sporadic cases. PLoS medicine. 2017. [PMID 28350801](https://pubmed.ncbi.nlm.nih.gov/28350801/). DOI: 10.1371/journal.pmed.1002270. Access: abstract_metadata. Study: PMID28350801.

21. Arber C, Toombs J, Lovejoy C et al. Familial Alzheimer's disease patient-derived neurons reveal distinct mutation-specific effects on amyloid beta. Molecular psychiatry. 2020. [PMID 30980041](https://pubmed.ncbi.nlm.nih.gov/30980041/). DOI: 10.1038/s41380-019-0410-8. Access: abstract_metadata. Study: PMID30980041.

22. Ringman JM, Monsell S, Ng DW et al. Neuropathology of Autosomal Dominant Alzheimer Disease in the National Alzheimer Coordinating Center Database. Journal of neuropathology and experimental neurology. 2016. [PMID 26888304](https://pubmed.ncbi.nlm.nih.gov/26888304/). DOI: 10.1093/jnen/nlv028. Access: abstract_metadata. Study: PMID26888304.

23. Bustos V, Pulina MV, Kelahmetoglu Y et al. Bidirectional regulation of Aβ levels by Presenilin 1. Proceedings of the National Academy of Sciences of the United States of America. 2017. [PMID 28533411](https://pubmed.ncbi.nlm.nih.gov/28533411/). DOI: 10.1073/pnas.1705235114. Access: abstract_metadata. Study: PMID28533411.

24. Perdok A, Van Acker ZP, Vrancx C et al. Altered expression of Presenilin2 impacts endolysosomal homeostasis and synapse function in Alzheimer's disease-relevant brain circuits. Nature communications. 2024. [PMID 39613768](https://pubmed.ncbi.nlm.nih.gov/39613768/). DOI: 10.1038/s41467-024-54777-y. Access: abstract_metadata. Study: PMID39613768.

25. Swanson CJ, Zhang Y, Dhadda S et al. A randomized, double-blind, phase 2b proof-of-concept clinical trial in early Alzheimer's disease with lecanemab, an anti-Aβ protofibril antibody. Alzheimer's research & therapy. 2021. [PMID 33865446](https://pubmed.ncbi.nlm.nih.gov/33865446/). DOI: 10.1186/s13195-021-00813-8. Access: abstract_metadata. Study: NCT01767311.

26. Zimmer JA, Sims JR, Evans CD et al. Donanemab in early symptomatic Alzheimer's disease: results from the TRAILBLAZER-ALZ 2 long-term extension. The journal of prevention of Alzheimer's disease. 2026. [PMID 41330788](https://pubmed.ncbi.nlm.nih.gov/41330788/). DOI: 10.1016/j.tjpad.2025.100446. Access: abstract_metadata. Study: NCT04437511.

27. van Dyck CH, Sperling R, Johnson K et al. Long-term safety and efficacy of lecanemab in early Alzheimer's disease: Results from the clarity AD open-label extension study. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2025. [PMID 41355080](https://pubmed.ncbi.nlm.nih.gov/41355080/). DOI: 10.1002/alz.70905. Access: abstract_metadata. Study: NCT03887455.

28. Wang H, Serap Monkul Nery E, Ardayfio P et al. Modified titration of donanemab reduces ARIA risk and maintains amyloid reduction. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2025. [PMID 40172303](https://pubmed.ncbi.nlm.nih.gov/40172303/). DOI: 10.1002/alz.70062. Access: abstract_metadata. Study: PMID40172303.

29. McDade E, Cummings JL, Dhadda S et al. Lecanemab in patients with early Alzheimer's disease: detailed results on biomarker, cognitive, and clinical effects from the randomized and open-label extension of the phase 2 proof-of-concept study. Alzheimer's research & therapy. 2022. [PMID 36544184](https://pubmed.ncbi.nlm.nih.gov/36544184/). DOI: 10.1186/s13195-022-01124-2. Access: abstract_metadata. Study: NCT01767311.

30. Zimmer JA, Ardayfio P, Wang H et al. Amyloid-Related Imaging Abnormalities With Donanemab in Early Symptomatic Alzheimer Disease: Secondary Analysis of the TRAILBLAZER-ALZ and ALZ 2 Randomized Clinical Trials. JAMA neurology. 2025. [PMID 40063015](https://pubmed.ncbi.nlm.nih.gov/40063015/). DOI: 10.1001/jamaneurol.2025.0065. Access: abstract_metadata. Study: TRAILBLAZER-ALZ+ALZ2+addendum.

31. Sur C, Kost J, Scott D et al. BACE inhibition causes rapid, regional, and non-progressive volume reduction in Alzheimer's disease brain. Brain : a journal of neurology. 2020. [PMID 33253354](https://pubmed.ncbi.nlm.nih.gov/33253354/). DOI: 10.1093/brain/awaa332. Access: abstract_metadata. Study: NCT01739348.

32. Wessels AM, Lines C, Stern RA et al. Cognitive outcomes in trials of two BACE inhibitors in Alzheimer's disease. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2020. [PMID 33049114](https://pubmed.ncbi.nlm.nih.gov/33049114/). DOI: 10.1002/alz.12164. Access: abstract_metadata. Study: APECS+AMARANTH.

33. Bateman RJ, Li Y, McDade EM et al. Safety and efficacy of long-term gantenerumab treatment in dominantly inherited Alzheimer's disease: an open-label extension of the phase 2/3 multicentre, randomised, double-blind, placebo-controlled platform DIAN-TU trial. The Lancet. Neurology. 2025. [PMID 40120616](https://pubmed.ncbi.nlm.nih.gov/40120616/). DOI: 10.1016/S1474-4422(25)00024-9. Access: full_text_partial. Study: DIAN-TU-001+OLE.

34. Mattsson-Carlgren N, Janelidze S, Palmqvist S et al. Longitudinal plasma p-tau217 is increased in early stages of Alzheimer's disease. Brain : a journal of neurology. 2020. [PMID 33068398](https://pubmed.ncbi.nlm.nih.gov/33068398/). DOI: 10.1093/brain/awaa286. Access: abstract_metadata. Study: PMID33068398.

35. Roemer-Cassiano SN, Wagner F, Evangelista L et al. Amyloid-associated hyperconnectivity drives tau spread across connected brain regions in Alzheimer's disease. Science translational medicine. 2025. [PMID 39841807](https://pubmed.ncbi.nlm.nih.gov/39841807/). DOI: 10.1126/scitranslmed.adp2564. Access: abstract_metadata. Study: PMID39841807.

36. Pereira JB, Janelidze S, Smith R et al. Plasma GFAP is an early marker of amyloid-β but not tau pathology in Alzheimer's disease. Brain : a journal of neurology. 2021. [PMID 34259835](https://pubmed.ncbi.nlm.nih.gov/34259835/). DOI: 10.1093/brain/awab223. Access: abstract_metadata. Study: PMID34259835.

37. Xie L, Wisse LEM, Das SR et al. Longitudinal atrophy in early Braak regions in preclinical Alzheimer's disease. Human brain mapping. 2020. [PMID 32845545](https://pubmed.ncbi.nlm.nih.gov/32845545/). DOI: 10.1002/hbm.25151. Access: abstract_metadata. Study: PMID32845545.

38. Mattsson-Carlgren N, Salvadó G, Ashton NJ et al. Prediction of Longitudinal Cognitive Decline in Preclinical Alzheimer Disease Using Plasma Biomarkers. JAMA neurology. 2023. [PMID 36745413](https://pubmed.ncbi.nlm.nih.gov/36745413/). DOI: 10.1001/jamaneurol.2022.5272. Access: abstract_metadata. Study: PMID36745413.

39. Hall S, Janelidze S, Londos E et al. Plasma Phospho-Tau Identifies Alzheimer's Co-Pathology in Patients with Lewy Body Disease. Movement disorders : official journal of the Movement Disorder Society. 2021. [PMID 33285015](https://pubmed.ncbi.nlm.nih.gov/33285015/). DOI: 10.1002/mds.28370. Access: abstract_metadata. Study: PMID33285015.

40. Insel PS, Mattsson N, Donohue MC et al. The transitional association between β-amyloid pathology and regional brain atrophy. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2015. [PMID 25499535](https://pubmed.ncbi.nlm.nih.gov/25499535/). DOI: 10.1016/j.jalz.2014.11.002. Access: abstract_metadata. Study: PMID25499535.

41. Satir TM, Agholme L, Karlsson A et al. Partial reduction of amyloid β production by β-secretase inhibitors does not decrease synaptic transmission. Alzheimer's research & therapy. 2020. [PMID 32456694](https://pubmed.ncbi.nlm.nih.gov/32456694/). DOI: 10.1186/s13195-020-00635-0. Access: abstract_metadata. Study: PMID32456694.

42. Saito T, Matsuba Y, Mihira N et al. Single App knock-in mouse models of Alzheimer's disease. Nature neuroscience. 2014. [PMID 24728269](https://pubmed.ncbi.nlm.nih.gov/24728269/). DOI: 10.1038/nn.3697. Access: abstract_metadata. Study: PMID24728269.

43. Roberson ED, Scearce-Levie K, Palop JJ et al. Reducing endogenous tau ameliorates amyloid beta-induced deficits in an Alzheimer's disease mouse model. Science (New York, N.Y.). 2007. [PMID 17478722](https://pubmed.ncbi.nlm.nih.gov/17478722/). DOI: 10.1126/science.1141736. Access: abstract_metadata. Study: PMID17478722.

44. Shankar GM, Li S, Mehta TH et al. Amyloid-beta protein dimers isolated directly from Alzheimer's brains impair synaptic plasticity and memory. Nature medicine. 2008. [PMID 18568035](https://pubmed.ncbi.nlm.nih.gov/18568035/). DOI: 10.1038/nm1782. Access: abstract_metadata. Study: PMID18568035.

45. Chung S, Jeong JH, Park JC et al. Blockade of STING activation alleviates microglial dysfunction and a broad spectrum of Alzheimer's disease pathologies. Experimental & molecular medicine. 2024. [PMID 39218977](https://pubmed.ncbi.nlm.nih.gov/39218977/). DOI: 10.1038/s12276-024-01295-y. Access: abstract_metadata. Study: PMID39218977.

46. Rajani RM, Ellingford R, Hellmuth M et al. Selective suppression of oligodendrocyte-derived amyloid beta rescues neuronal dysfunction in Alzheimer's disease. PLoS biology. 2024. [PMID 39042667](https://pubmed.ncbi.nlm.nih.gov/39042667/). DOI: 10.1371/journal.pbio.3002727. Access: full_text_partial. Study: PMID39042667.

47. Pickett EK, Koffie RM, Wegmann S et al. Non-Fibrillar Oligomeric Amyloid-β within Synapses. Journal of Alzheimer's disease : JAD. 2016. [PMID 27258414](https://pubmed.ncbi.nlm.nih.gov/27258414/). DOI: 10.3233/JAD-160007. Access: abstract_metadata. Study: PMID27258414.

48. Naia L, Shimozawa M, Bereczki E et al. Mitochondrial hypermetabolism precedes impaired autophagy and synaptic disorganization in App knock-in Alzheimer mouse models. Molecular psychiatry. 2023. [PMID 37907591](https://pubmed.ncbi.nlm.nih.gov/37907591/). DOI: 10.1038/s41380-023-02289-4. Access: abstract_metadata. Study: PMID37907591.

49. Lee E, Kim S, Zhu CL et al. Angiopoietin-2 aggravates Alzheimer's disease by promoting blood-brain barrier dysfunction and neuroinflammation. Cell reports. 2026. [PMID 41529684](https://pubmed.ncbi.nlm.nih.gov/41529684/). DOI: 10.1016/j.celrep.2025.116621. Access: abstract_metadata. Study: PMID41529684.

50. de Weerd L, Hummel S, Müller SA et al. Early intervention anti-Aβ immunotherapy attenuates microglial activation without inducing exhaustion at residual plaques. Molecular neurodegeneration. 2025. [PMID 40830489](https://pubmed.ncbi.nlm.nih.gov/40830489/). DOI: 10.1186/s13024-025-00878-1. Access: abstract_metadata. Study: PMID40830489.

51. Schöll M, Lockhart SN, Schonhaut DR et al. PET Imaging of Tau Deposition in the Aging Human Brain. Neuron. 2016. [PMID 26938442](https://pubmed.ncbi.nlm.nih.gov/26938442/). DOI: 10.1016/j.neuron.2016.01.028. Access: abstract_metadata. Study: PMID26938442.

52. Salloway S, Farlow M, McDade E et al. A trial of gantenerumab or solanezumab in dominantly inherited Alzheimer's disease. Nature medicine. 2021. [PMID 34155411](https://pubmed.ncbi.nlm.nih.gov/34155411/). DOI: 10.1038/s41591-021-01369-8. Access: full_text_partial. Study: DIAN-TU-001.

53. Bateman RJ, Xiong C, Benzinger TL et al. Clinical and biomarker changes in dominantly inherited Alzheimer's disease. The New England journal of medicine. 2012. [PMID 22784036](https://pubmed.ncbi.nlm.nih.gov/22784036/). DOI: 10.1056/NEJMoa1202753. Access: abstract_metadata. Study: PMID22784036.

54. Montagne A, Nation DA, Sagare AP et al. APOE4 leads to blood-brain barrier dysfunction predicting cognitive decline. Nature. 2020. [PMID 32376954](https://pubmed.ncbi.nlm.nih.gov/32376954/). DOI: 10.1038/s41586-020-2247-3. Access: abstract_metadata. Study: PMID32376954.

55. Chen CD, Franklin EE, Li Y et al. Immunohistochemical evaluation of a trial of gantenerumab or solanezumab in dominantly inherited Alzheimer disease. Acta neuropathologica. 2025. [PMID 40459787](https://pubmed.ncbi.nlm.nih.gov/40459787/). DOI: 10.1007/s00401-025-02890-7. Access: abstract_metadata. Study: DIAN-TU-001-autopsy-subset.

56. Llibre-Guerra JJ, Fernandez MV, Joseph-Mathurin N et al. Longitudinal analysis of a dominantly inherited Alzheimer disease mutation carrier protected from dementia. Nature medicine. 2025. [PMID 39930140](https://pubmed.ncbi.nlm.nih.gov/39930140/). DOI: 10.1038/s41591-025-03494-0. Access: abstract_metadata. Study: PMID39930140.

## Source-specific evidence cards

### [1] Lecanemab in Early Alzheimer's Disease.

**Shows:** Lecanemab slowed cognitive and functional decline in amyloid-confirmed early AD, with adverse events.

**Does not show:** Clinical slowing does not establish complete mediation, universal initiation, halt or reversal of disease.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [2] Donanemab in Early Symptomatic Alzheimer Disease: The TRAILBLAZER-ALZ 2 Randomized Clinical Trial.

**Shows:** Donanemab favored clinical outcomes in early symptomatic amyloid/tau-positive AD.

**Does not show:** Both groups worsened; this is stage-specific contribution rather than proof of a sole initiator.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4, SC5.

### [3] Two Phase 3 Trials of Gantenerumab in Early Alzheimer's Disease.

**Shows:** Gantenerumab reduced plaque and fluid tau markers without significant primary clinical slowing in either GRADUATE trial.

**Does not show:** Incomplete plaque removal and target/stage differences limit a universal null inference; tau PET was similar.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3, SC5.

### [4] Two Randomized Phase 3 Studies of Aducanumab in Early Alzheimer's Disease.

**Shows:** Aducanumab's EMERGE clinical comparison was positive and ENGAGE was null despite marker engagement in both.

**Does not show:** Discordance, early futility stopping and exposure differences prevent treating the pair as consistent clinical replication.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [5] Trial of Solanezumab in Preclinical Alzheimer's Disease.

**Shows:** Solanezumab did not slow preclinical cognitive decline; amyloid PET increased in both arms.

**Does not show:** Monomer targeting and incomplete prevention of deposition do not test complete plaque eradication.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4, SC5.

### [6] Trial of Solanezumab for Mild Dementia Due to Alzheimer's Disease.

**Shows:** Solanezumab did not meet its mild-AD primary clinical endpoint.

**Does not show:** Hierarchical secondary outcomes were descriptive; this agent's null does not erase antibody-specific positives.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC5.

### [7] Randomized Trial of Verubecestat for Mild-to-Moderate Alzheimer's Disease.

**Shows:** Verubecestat did not improve cognitive or functional decline in mild-to-moderate AD and increased adverse events.

**Does not show:** BACE inhibition affects non-Aβ substrates and is not a selective experiment on pathological aggregates.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4, SC5.

### [8] Randomized Trial of Verubecestat for Prodromal Alzheimer's Disease.

**Shows:** Prodromal verubecestat was ineffective and the higher dose worsened the primary clinical outcome.

**Does not show:** Early stopping, BACE substrate pleiotropy and drug harms prevent equating this with selective Aβ harm removal.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4, SC5.

### [9] Findings of Efficacy, Safety, and Biomarker Outcomes of Atabecestat in Preclinical Alzheimer Disease: A Truncated Randomized Phase 2b/3 Clinical Trial.

**Shows:** Atabecestat caused dose-related cognitive worsening with evidence of reversibility after withdrawal.

**Does not show:** A truncated hepatic-safety-stopped trial does not isolate the specific Aβ versus other BACE-dependent functions.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4, SC5.

### [10] Efficacy and Safety of Lanabecestat for Treatment of Early and Mild Alzheimer Disease: The AMARANTH and DAYBREAK-ALZ Randomized Clinical Trials.

**Shows:** Lanabecestat failed to slow clinical decline in AMARANTH and DAYBREAK.

**Does not show:** Premature termination and substrate specificity constrain the mechanistic inference; a PET-sign erratum is attached.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. ErratumIn: PMID32658253. Subclaims: SC4, SC5.

### [11] Donanemab in Early Alzheimer's Disease.

**Shows:** Phase-2 donanemab favored its primary integrated clinical score; most secondary comparisons were mixed.

**Does not show:** This primary contrast is not the same as all participants in later post-hoc analyses, and tau change was small globally.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3, SC5.

### [12] Association of Amyloid Reduction After Donanemab Treatment With Tau Pathology and Clinical Outcomes: The TRAILBLAZER-ALZ Randomized Clinical Trial.

**Shows:** Post-hoc donanemab analyses linked clearance to regional tau slowing and clinical model effects in APOE4 carriers.

**Does not show:** Postrandomization associations and model forecasts are not randomized mediator manipulation or a new trial.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3, SC5.

### [13] Target engagement in an alzheimer trial: Crenezumab lowers amyloid β oligomers in cerebrospinal fluid.

**Shows:** Crenezumab lowered measured CSF oligomer levels in phase-2 trial specimens.

**Does not show:** Fluid target engagement does not establish brain-species selective depletion or clinical efficacy.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2.

### [14] Bapineuzumab for mild to moderate Alzheimer's disease in two global, randomized, phase 3 trials.

**Shows:** Two terminated global bapineuzumab trials were null for clinical outcomes and their examined biomarker contrasts.

**Does not show:** Dose, stage and absent biomarker movement limit claims that successful selective removal was clinically ineffective.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [15] A mutation in APP protects against Alzheimer's disease and age-related cognitive decline.

**Shows:** APP A673T was associated with protection and reduced amyloidogenic processing in functional assays.

**Does not show:** A rare lifelong processing variant is stronger than pathology co-occurrence but does not isolate every APP product or all sporadic cases.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [16] Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer's disease.

**Shows:** An APP missense variant segregated with early-onset familial AD in two unrelated families.

**Does not show:** The instrument concerns selected familial disease, not all late-onset sporadic initiation.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [17] APP locus duplication causes autosomal dominant early-onset Alzheimer disease with cerebral amyloid angiopathy.

**Shows:** APP-locus duplications occurred in families with early-onset AD and prominent cerebral amyloid angiopathy.

**Does not show:** Gene dosage affects the precursor and vascular/parenchymal biology, not exclusively plaque toxicity.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [18] Safety and efficacy of crenezumab in cognitively unimpaired carriers of the PSEN1Glu280Ala mutation at risk for autosomal-dominant Alzheimer's disease in Colombia (API ADAD Colombia Trial): a phase 2, randomised, double-blind, placebo-controlled trial.

**Shows:** The completed Colombian crenezumab prevention trial found no significant benefit on either primary clinical endpoint and no clear plaque-removal effect.

**Does not show:** It cannot be described as a successful plaque-clearance prevention test; limited engagement constrains interpretation.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4, SC5.

### [19] APOE3 Christchurch Heterozygosity and Autosomal Dominant Alzheimer's Disease.

**Shows:** APOE3 Christchurch heterozygotes in the PSEN1 kindred had delayed impairment; selected autopsies showed substantial plaque with limited tau.

**Does not show:** Related-kindred sampling, sparse imaging/autopsy and qualitative unquantified pathology do not prove amyloid is causally irrelevant.

**Access/status:** Complete primary abstract and linked-status record; selected primary-body sections: S7, S13, S14. No full-paper or supplement audit. none_found. Subclaims: SC1, SC3, SC4.

### [20] APP, PSEN1, and PSEN2 mutations in early-onset Alzheimer disease: A genetic screening study of familial and sporadic cases.

**Shows:** Early-onset screening identified familial and de novo APP/presenilin variants, including apparently sporadic presentations.

**Does not show:** Apparently sporadic early-onset mutation carriers are not equivalent to general late-onset sporadic AD; some variant pathogenicity was unresolved.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4.

### [21] Familial Alzheimer's disease patient-derived neurons reveal distinct mutation-specific effects on amyloid beta.

**Shows:** Patient-derived neurons showed distinct mutation-specific Aβ processing signatures with raised Aβ42:40.

**Does not show:** Cellular secretion signatures do not demonstrate organismal degeneration or a single uniform familial mechanism.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [22] Neuropathology of Autosomal Dominant Alzheimer Disease in the National Alzheimer Coordinating Center Database.

**Shows:** Autosomal-dominant and sporadic autopsies differed in plaque, tau and vascular burden, with mutation-location heterogeneity.

**Does not show:** Terminal pathology differences do not identify causal initiation or validate identical mechanisms across scopes.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4.

### [23] Bidirectional regulation of Aβ levels by Presenilin 1.

**Shows:** Presenilin phosphorylation altered Aβ through βCTF degradation and autophagy as well as catalytic processing.

**Does not show:** Amyloid-marker effects alone do not establish clinical primacy; precursor-fragment mechanisms remain alternatives.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [24] Altered expression of Presenilin2 impacts endolysosomal homeostasis and synapse function in Alzheimer's disease-relevant brain circuits.

**Shows:** PSEN2 loss or mutation altered endolysosomal and synaptic physiology in female APP knock-in mice.

**Does not show:** Sex/model restriction and non-Aβ presenilin functions prevent attributing all dysfunction to extracellular amyloid.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2, SC3, SC4.

### [25] A randomized, double-blind, phase 2b proof-of-concept clinical trial in early Alzheimer's disease with lecanemab, an anti-Aβ protofibril antibody.

**Shows:** Phase-2 lecanemab missed its primary probability threshold but favored prespecified later clinical/biomarker comparisons.

**Does not show:** The positive later analyses must not replace the negative primary; an erratum is linked and not separately counted.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. ErratumIn: PMID35598024. Subclaims: SC5.

### [26] Donanemab in early symptomatic Alzheimer's disease: results from the TRAILBLAZER-ALZ 2 long-term extension.

**Shows:** Donanemab long-term extension reported sustained early-start advantage and slowing versus a weighted external cohort.

**Does not show:** Extension completers and external ADNI comparisons are not a new long-duration concurrent placebo experiment.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC5.

### [27] Long-term safety and efficacy of lecanemab in early Alzheimer's disease: Results from the clarity AD open-label extension study.

**Shows:** Clarity AD's open-label extension reported maintained early/delayed-start separation and favorable longer-term outcomes.

**Does not show:** Open-label, selected continuation and subgroup analyses do not establish universal prevention or independent causal replication.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC5.

### [28] Modified titration of donanemab reduces ARIA risk and maintains amyloid reduction.

**Shows:** Modified donanemab titration lowered ARIA risk while retaining comparable amyloid reduction.

**Does not show:** The safety/engagement endpoint does not establish superior clinical efficacy; linked erratum impact on fine details remains unexamined.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. ErratumIn: PMID40775664. Subclaims: SC5.

### [29] Lecanemab in patients with early Alzheimer's disease: detailed results on biomarker, cognitive, and clinical effects from the randomized and open-label extension of the phase 2 proof-of-concept study.

**Shows:** Study-201 follow-up found retained clinical separation during a treatment gap while plasma markers returned sooner than plaque PET.

**Does not show:** The same cohort, nonrandomized gap/extension and differing randomized/treated denominators limit causal withdrawal or mediation claims.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3, SC5.

### [30] Amyloid-Related Imaging Abnormalities With Donanemab in Early Symptomatic Alzheimer Disease: Secondary Analysis of the TRAILBLAZER-ALZ and ALZ 2 Randomized Clinical Trials.

**Shows:** Donanemab secondary safety analyses documented ARIA and baseline risk associations.

**Does not show:** Pooled trial/addendum specimens are reused; observational risk associations do not identify ARIA as the efficacy mediator.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC5.

### [31] BACE inhibition causes rapid, regional, and non-progressive volume reduction in Alzheimer's disease brain.

**Shows:** EPOCH exploratory imaging found rapid nonprogressive verubecestat-associated regional volume loss without a matching NfL rise.

**Does not show:** MRI volume change is not automatically progressive neuronal death or selective loss of protective amyloid function.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC5.

### [32] Cognitive outcomes in trials of two BACE inhibitors in Alzheimer's disease.

**Shows:** APECS and AMARANTH secondary tests showed several worsened domains and improved verbal fluency with BACE inhibitors.

**Does not show:** Mixed cognitive tasks and reused trial data do not form a new mechanistic line or a general beneficial aging effect.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC5.

### [33] Safety and efficacy of long-term gantenerumab treatment in dominantly inherited Alzheimer's disease: an open-label extension of the phase 2/3 multicentre, randomised, double-blind, placebo-controlled platform DIAN-TU trial.

**Shows:** DIAN gantenerumab extension achieved substantial plaque removal and suggested possible clinical delay after long exposure.

**Does not show:** The final primary was PET; main-control clinical confidence intervals included no effect, with an extended-control contrast favorable. OLE/external controls and linked correction constrain proof.

**Access/status:** Complete primary abstract and linked-status record; selected primary-body sections: S10, S12, S17, S19. No full-paper or supplement audit. UpdateOf: PMID39974075; ErratumIn: PMID40651485. Subclaims: SC4, SC5.

### [34] Longitudinal plasma p-tau217 is increased in early stages of Alzheimer's disease.

**Shows:** Longitudinal p-tau217 rose in amyloid-positive preclinical/prodromal groups and correlated with decline.

**Does not show:** Temporal marker coupling does not selectively manipulate amyloid or establish tau's first origin.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3.

### [35] Amyloid-associated hyperconnectivity drives tau spread across connected brain regions in Alzheimer's disease.

**Shows:** Longitudinal cohorts associated amyloid-related hyperconnectivity with later tau accumulation.

**Does not show:** Statistical mediation and replicated association are not a causal intervention on connectivity or amyloid.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3.

### [36] Plasma GFAP is an early marker of amyloid-β but not tau pathology in Alzheimer's disease.

**Shows:** Plasma GFAP associated with amyloid burden and longitudinal change after accounting for tau.

**Does not show:** A glial biomarker and observational mediation do not establish inflammation's causal direction.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3.

### [37] Longitudinal atrophy in early Braak regions in preclinical Alzheimer's disease.

**Shows:** Preclinical medial-temporal atrophy was elevated in participants with concomitant tau, not the tau-negative group.

**Does not show:** Baseline strata and imaging outcomes are prognostic evidence rather than tau necessity or amyloid inertness.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [38] Prediction of Longitudinal Cognitive Decline in Preclinical Alzheimer Disease Using Plasma Biomarkers.

**Shows:** Plasma p-tau217 predicted later cognitive decline in amyloid-positive participants across cohorts.

**Does not show:** Prediction conditional on amyloid positivity does not determine the universal initiating mechanism.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3, SC4.

### [39] Plasma Phospho-Tau Identifies Alzheimer's Co-Pathology in Patients with Lewy Body Disease.

**Shows:** Phospho-tau markers detected Alzheimer co-pathology in a Lewy-body dementia sample.

**Does not show:** Cross-sectional biomarker overlap cannot establish which pathology caused each person's dementia.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC4.

### [40] The transitional association between β-amyloid pathology and regional brain atrophy.

**Shows:** Spline analyses found accelerated regional atrophy before conventional CSF Aβ-positivity thresholds.

**Does not show:** Subthreshold Aβ and measurement sensitivity remain possible; a diagnostic threshold is not true molecular absence.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [41] Partial reduction of amyloid β production by β-secretase inhibitors does not decrease synaptic transmission.

**Shows:** Partial BACE inhibition reduced Aβ without impairing transmission in cortical cultures, whereas stronger suppression impaired transmission.

**Does not show:** This dose boundary is not a human therapeutic window or an isolated test of Aβ-specific synaptic requirements.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2, SC5.

### [42] Single App knock-in mouse models of Alzheimer's disease.

**Shows:** Physiological-promoter APP knock-in mice developed age-dependent Aβ pathology, inflammation and memory impairment.

**Does not show:** Humanized familial mutations remain a genotype model; eliminating overexpression does not make all sporadic disease equivalent.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [43] Reducing endogenous tau ameliorates amyloid beta-induced deficits in an Alzheimer's disease mouse model.

**Shows:** Endogenous tau reduction prevented APP-transgenic behavioral deficits without lowering Aβ.

**Does not show:** Tau dependency in this model does not prove tau is the earliest human initiator; transgenic Aβ exposure still limits scope.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2, SC3.

### [44] Amyloid-beta protein dimers isolated directly from Alzheimer's brains impair synaptic plasticity and memory.

**Shows:** Human brain-derived Aβ dimers impaired rodent plasticity and memory; antibodies prevented selected effects and intact plaque cores were relatively inactive.

**Does not show:** Forced extract administration demonstrates capacity, not endogenous human species necessity or propagation.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2.

### [45] Blockade of STING activation alleviates microglial dysfunction and a broad spectrum of Alzheimer's disease pathologies.

**Shows:** STING inhibition improved memory and multiple inflammatory/amyloid/tau features in a double-knock-in model.

**Does not show:** Broad pathway effects do not identify a sole amyloid-primary or inflammation-primary initiator in sporadic humans.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [46] Selective suppression of oligodendrocyte-derived amyloid beta rescues neuronal dysfunction in Alzheimer's disease.

**Shows:** Inducible oligodendrocyte BACE1 deletion reduced plaque burden and restored early neuronal firing in APP knock-in mice; Aβ-depleted conditioned medium lost acute hyperactivity effects.

**Does not show:** BACE1 is pleiotropic, exposure began in young mice, and neuronal firing is not human clinical progression; injected medium is a separate forced sufficiency test.

**Access/status:** Complete primary abstract and linked-status record; selected primary-body sections: sec006, sec007, sec010, sec033. No full-paper or supplement audit. none_found. Subclaims: SC1, SC2, SC3.

### [47] Non-Fibrillar Oligomeric Amyloid-β within Synapses.

**Shows:** Several high-resolution imaging assays localized oligomeric Aβ inside synapses of a plaque-bearing model.

**Does not show:** Localization and agreement between imaging modalities do not establish synaptic species necessity.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC2.

### [48] Mitochondrial hypermetabolism precedes impaired autophagy and synaptic disorganization in App knock-in Alzheimer mouse models.

**Shows:** APP knock-in time courses showed early mitochondrial hypermetabolism and later synaptic/autophagy abnormalities.

**Does not show:** Model temporal order does not isolate the causal mediator or establish a universal human sequence.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3.

### [49] Angiopoietin-2 aggravates Alzheimer's disease by promoting blood-brain barrier dysfunction and neuroinflammation.

**Shows:** Endothelial Angpt2 deletion lowered amyloid in 5xFAD; vascular and inflammatory pathways accompanied behavioral abnormalities.

**Does not show:** Overexpression and knockout contrasts have different inferential strength; 5xFAD and vascular pleiotropy limit sporadic-human initiation claims.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [50] Early intervention anti-Aβ immunotherapy attenuates microglial activation without inducing exhaustion at residual plaques.

**Shows:** Early anti-Aβ treatment in APP-SAA knock-in mice reduced plaque and broad disease-associated glial signals, with active glia at residual plaques.

**Does not show:** The reported markers do not by themselves demonstrate preserved cognition, mediator necessity or complete disease prevention.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [51] PET Imaging of Tau Deposition in the Aging Human Brain.

**Shows:** Medial-temporal tau occurred with aging, while wider cortical tau was associated with amyloid and global decline.

**Does not show:** Cross-sectional PET supports stage distinctions but cannot prove amyloid-free tau initiates AD or amyloid causes its expansion.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC3.

### [52] A trial of gantenerumab or solanezumab in dominantly inherited Alzheimer's disease.

**Shows:** DIAN randomized antibodies engaged targets without clinical benefit; gantenerumab moved tau and NfL markers.

**Does not show:** The primary disease model did not converge, but sensitivity analyses agreed; late dose escalation and no asymptomatic decline limit prevention inference.

**Access/status:** Complete primary abstract and linked-status record; selected primary-body sections: S3, S4, S8, S9. No full-paper or supplement audit. none_found. Subclaims: SC3, SC4, SC5.

### [53] Clinical and biomarker changes in dominantly inherited Alzheimer's disease.

**Shows:** DIAN baseline cross-sectional data inferred biomarker changes decades before expected symptoms.

**Does not show:** A prospective cohort label does not turn this baseline analysis into observed within-person chronology; familial scope and an erratum are retained.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. ErratumIn: N Engl J Med 2012;367(8):780 (erratum notice has no PMID). Subclaims: SC3, SC4.

### [54] APOE4 leads to blood-brain barrier dysfunction predicting cognitive decline.

**Shows:** APOE4-associated BBB injury predicted decline after accounting for measured amyloid and tau.

**Does not show:** This human prediction is not a vascular intervention or proof that every case is vascular-primary; pathology measurements are imperfect.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC4.

### [55] Immunohistochemical evaluation of a trial of gantenerumab or solanezumab in dominantly inherited Alzheimer disease.

**Shows:** Small selected DIAN autopsies showed dose-related gantenerumab plaque reduction without detectable tau/glial differences.

**Does not show:** Nonrandom death/tissue selection and small groups constrain null downstream inference; this is reused DIAN evidence.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3, SC5.

### [56] Longitudinal analysis of a dominantly inherited Alzheimer disease mutation carrier protected from dementia.

**Shows:** A longitudinal PSEN2 mutation-carrier case remained clinically resilient with restricted tau pathology.

**Does not show:** A single case and candidate genetic/environmental explanations do not establish a protective mechanism or population frequency.

**Access/status:** Complete primary abstract and linked-status metadata examined; no Methods/body/supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.