In plain terms

The claim being assessed: Primary dysfunction of neuronal insulin/IGF-1 signalling initiates and drives sporadic Alzheimer's disease, and correcting this dysfunction modifies the disease trajectory.

Where the evidence stands: Γ — Plausible mechanism. There is a believable biological story and partial support, but the causal step has not been demonstrated.

What this mechanism does: This is a result of the disease process, not a cause of it.

Strongest evidence for it: Γ follows the preregistered weakest-link bundle (Γ,Ε,Γ,Γ); primary-SC3 and scope-matched sensitivities also return Γ. Model-specific causal contribution earns Ε, while primary sporadic neuronal initiation and clinically proven correction remain unestablished. Functional tissue resistance, acute memory effects, peripheral prevention and FDG changes are kept distinct. Positive models and valid human causal equivalents are assessed together, without automatically re-capping their weaker observational members.

What decisive evidence would look like

In plain terms, researchers would change the proposed cause, confirm it changed as intended, and measure a meaningful health outcome while checking other explanations. The result should be able to support or count against the claim.

Relative explanatory fit: Alternative ranked ahead

What this review means for Metabolic / 'Type 3 Diabetes' Hypothesis

Claim being assessed: Primary dysfunction of neuronal insulin/IGF-1 signalling initiates and drives sporadic Alzheimer's disease, and correcting this dysfunction modifies the disease trajectory.. This review asks whether the proposed factor causes the health outcome—not just whether the two appear together. For the wording assessed here, the evidence is rated Plausible mechanism (HSEF Γ; Moderate confidence). There is a believable biological story and partial support, but the causal step has not been demonstrated. That rating applies to this claim and scope; it does not mean the biological process is irrelevant in every setting.

Implications for health and treatment: This is an assessment of a causal explanation, not proof that a particular supplement, diet, test, or treatment improves health. A treatment claim needs studies that test the intervention in people and measure meaningful benefits and harms. These findings can help frame research and informed clinical questions, but they do not identify what is right for one person.

Limits for individual care: Evidence from cells, animals, associations, or a plausible mechanism may not predict what happens in people. The grade is limited to the exact claim, outcomes, and sources reviewed, and it is not a diagnosis or an individual forecast. Do not start, stop, or change care based on this page; discuss personal decisions with a qualified clinician.

Medical information notice: This review provides general health information, not medical advice. It cannot diagnose a condition or replace an individual discussion with a qualified clinician.

This assessment rests on 68 sources and 73 individually checked claims. 73 of those claims were verified against the original paper.

A low grade is not an accusation. It means the causal work has not been done yet, which is the ordinary state of most actively studied mechanisms in biology.

At a glance

Sub-claims assessed separately

ID	Sub-claim	Type	Level	Pivotal card
SC1	Brain insulin resistance is present independently of peripheral type2 diabetes.	existence	Γ	—
SC2	Failure of neuronal insulin signaling contributes causally to AD pathology.	mechanism	Ε	Card ↗
SC3	Neuronal insulin resistance is upstream of amyloid and tau.	primacy	Γ	Card ↗
SC4	Insulin-targeting therapy modifies clinical disease trajectory.	application	Γ	Card ↗

Visual Evidence Cards · Pivotal Subclaims

Pivotal subclaim illustrations

Dark-theme simulated-micrograph renders and concise causal evidence summaries for the pivotal subclaims that govern this review's grade. Click any card to inspect the master 2400×2400 render.

Master 2400×2400 render ↗

Master 2400×2400 render ↗

Master 2400×2400 render ↗

Figures

Explanatory figures summarizing this review. They carry no evidential weight of their own; where a figure and the review differ, the review governs.

Full assessment

From here the review is written for a technical reader: caps and thresholds, competing models, the independence basis, and the citation ledger.

Sources: 68 primary reports; claims: 73; verified: 73; unverified: 0. Verification is bounded attribution. Access: 6 partial-body and 62 complete abstract/status examinations. Literature cutoff: 2026-10-05; source examination UTC date: 2026-10-06. No complete-paper audit or blinded calibration claimed. Adversarial review: Generated by Sol 6.1; adversarially reviewed by Claude Opus 5.5 (Anthropic) on 2026-10-06. 2 corrections applied (see ADVERSARIAL REVIEW AND APPROVAL). Status: Approved as revised.

AT A GLANCE

Brain insulin signalling is altered in some AD assays and can modify injury in models, but a primary neuronal insulin/IGF defect initiating sporadic AD and proven clinical correction remain unestablished.

Deciding experiment: Predefine an amyloid/tau-negative sporadic-risk cohort with validated live neuron-specific insulin-response measurements; triangulate a selective neuronal exposure instrument and an early CNS-delivered intervention that corrects that response, with matched adult conditional model loss/rescue and repeated pathology/function follow-up.

Belief versus evidence: Γ follows the preregistered weakest-link bundle (Γ,Ε,Γ,Γ); primary-SC3 and scope-matched sensitivities also return Γ. Model-specific causal contribution earns Ε, while primary sporadic neuronal initiation and clinically proven correction remain unestablished. Functional tissue resistance, acute memory effects, peripheral prevention and FDG changes are kept distinct. Positive models and valid human causal equivalents are assessed together, without automatically re-capping their weaker observational members.

Best-supported rival: Amyloid/tau-induced resistance with feedback. A selective adult neuronal IR perturbation/rescue with verified live target engagement, repeated prelesion amyloid/tau measures and long-term function separates primary failure from feedback or permissive support. Branch, cell type and systemic effects must be identified independently.

COMPETING MECHANISTIC MODELS

Rank 1: Amyloid/tau-induced resistance with feedback (Ε). AD-related injury can disrupt insulin-related signalling, which can amplify subsequent dysfunction. Oligomer induction, TNFR1/PKR intervention, tau deletion and insulin/GLP1 rescue establish direction and contribution in defined systems (PMID22476196,24315369,28652303). Sole temporal ordering and neuronal mediation in sporadic humans remain unproved.

Rank 2: Permissive neurotrophic support (Ε). Insulin-related signalling supports synapses and memory without constituting a primary AD trigger. IR dosage changes and normal-human memory modulation show functional support; NIRKO markers without degeneration separate vulnerability from initiation (PMID22661254,15288712,14981233). Adult neuron-specific necessity in AD is not established, and IGF-reduction protection requires branch-specific explanation.

Rank 3: Neurovascular/peripheral metabolic contribution (Γ). Metabolic risk can act through systemic and vascular routes without primary neuronal receptor failure. Diabetes incidence cohorts, brain-infarct associations, peripheral sensitivity MR and ancillary randomized dementia contrasts fit this route (PMID10599761,33858515,39479764,35229024). The particular mediator remains unidentified and diabetes-liability MR is null in several tests.

Rank 4: Reduced synaptic demand and secondary glucose change (Γ). Loss of synapses/cells or altered transport can produce FDG and tissue-glucose abnormalities. BLSA glucose/GLUT3 and Talbot glucose-uptake controls distinguish metabolic readouts from peripheral-style insulin-dependent uptake (PMID29055815,22476197). These observations do not themselves isolate a demand-only cause or exclude signalling feedback.

Rank 5: Primary neuronal insulin/IGF failure (Γ). A native neuronal signalling defect precedes and drives sporadic AD, and correcting it modifies clinical trajectory. Functional postmortem resistance, induced metabolic injury and small therapeutic signals provide the strongest case. Native prelesion ordering is absent; IR deletion need not produce degeneration, IGF attenuation can protect, and large clinical correction predictions fail.

The tournament does not alter the nominal level. Compatible pathways and flexible alternatives are not automatically proved by explaining observations.

FULL REVIEW

Registered claim and scope

Primary dysfunction of neuronal insulin/IGF-1 signalling initiates and drives sporadic Alzheimer's disease, and correcting this dysfunction modifies the disease trajectory.

Retain the exact index primary metabolic rubric with co-governing neurodegeneration rules. Presence, causal contribution, upstream primacy and clinical application are scored separately; the conjunctive wording takes the weakest link, with direction and treatment sensitivities displayed. Neither FDG deficit nor peripheral diabetes alone establishes neuronal insulin resistance.

Selection: public-interest rank 20 from the editorial ranking compiled 2026-09-23; categorical prior verdicts were visible. Private fresh review; earlier documents are preserved.

Evidence and causal interpretation

Human brain insulin resistance is a scoped finding rather than a diagnosis of type3 diabetes. Talbot used functional ex-vivo challenge and diabetes-free cases; original glucose-uptake experiments found no basal response to insulin but a glutamate positive control. Peripheral resistance was not measured in those cohorts. ROS midfrontal assays differ and do not reproduce the same IRS1/AD association; higher AKT tracks pathology and prior decline. Region, site and cell composition matter, and glucose/FDG measures do not uniquely measure neuronal insulin action (PMID22476197,32557841,38029396,29055815).

Prediagnostic putative neural EV IRS1 changes and peripheral insulin/IGF cohorts supply temporal clinical clues, but not ordering before amyloid/tau. Original single-EV positive validation includes detergent/isotype and neuronal-marker controls; later negative validation uses the common 5G3 clone with calnexin/free-protein controls and finds mostly soluble L1CAM. The disagreement concerns assay recovery/enrichment and does not establish absence of all neural EVs. Neither approach directly measures living neuronal insulin responsiveness (PMID25342129,33160263,38868956,40210837).

Causal contributions are real but branch- and cell-dependent. NIRKO has tau changes without neuronal loss or memory deficit; global IRS2 loss has developmental/peripheral effects. Adult neuronal IGF1R deletion protects APP/PS1 function, while astrocyte IGFR or microglial IR loss has adverse effects. Oligomers and tau manipulation can produce resistance; insulin/GLP1 and PTP1B interventions can protect downstream function. Bomfim original weight controls reduce a simple caloric explanation but do not isolate neuronal insulin mediation (PMID14981233,12904469,26290229,29398615,40388612,22476196,31915254).

Human peripheral genetic and prevention evidence deserves causal consideration without relabelling its exposure. Insulin-sensitivity MR suggests protection, overall T2D liability can be null, and a mechanism-specific signal does not replicate significantly in HRS cognition. Randomized cardiovascular-trial ancillary dementia diagnoses are lower with GLP1RA, while matched semaglutide EHR associations remain vulnerable to indication/ascertainment. None identifies a neuron-selective initiating mediator; prevention in diabetes and progression after amyloid-confirmed AD are different estimands (PMID39479764,39350373,26650880,35229024,39445596).

Small insulin/metformin trials show cognitive signals, often without daily-function or robust group biomarker effects. SNIFF original Methods define device2 as primary before lock and apply gatekeeping; its primary result is null. The 49-person device1 cohort has nominal six-month but nonsignificant twelve-month cognition, and supplies later WMH/inflammatory reports. MRI volumes are not uniformly protective. CSF insulin did not change, so delivery/engagement uncertainty bounds a universal mechanistic interpretation. Human lispro exposure failure and a newer radiotracer delivery study show why devices must be validated prospectively (PMID21911655,28372335,32568367,34101779,35079029,27553192,40703476).

ELAD original body reports 204 randomized, 154 SUV paired scans and 101 arterial spectral sensitivity scans, a null primary FDG effect, an unadjusted ADAS-Exec secondary signal, and null CDR/ADL. CSF was not collected; planned tau/TSPO PET was not completed. EVOKE/EVOKE+ are published 2026 results, not pending readouts: 3808 amyloid-confirmed early AD participants show no 104-week CDR-SB benefit in either trial. Rosiglitazone phase3 and low-dose preventive TOMMORROW also fail. These results challenge treatment predictions while leaving early risk modification and properly engaged different neuronal interventions open; a four-week safety-primary factorial signal is a lead for confirmation (PMID41326666,41865758,20733306,21592048,34146512,41057918).

HYPOTHESIS STATE EVALUATION FRAMEWORK (HSEF v5.3)

Subclaim matrix and aggregation

Preregistered aggregation: weakest-link. Γ follows the preregistered weakest-link bundle (Γ,Ε,Γ,Γ); primary-SC3 and scope-matched sensitivities also return Γ. Model-specific causal contribution earns Ε, while primary sporadic neuronal initiation and clinically proven correction remain unestablished. Functional tissue resistance, acute memory effects, peripheral prevention and FDG changes are kept distinct. Positive models and valid human causal equivalents are assessed together, without automatically re-capping their weaker observational members. Sensitivities: weakest-link Γ; primary-subclaim Γ; scope-matched Γ.

Decisive-perturbation inventory and causal tally

[2] AD hippocampal tissue without diabetes responds less to insulin and IGF challenge; IRS1 serine measures associate with cognition and pathology. Original Methods validate low-PMI tissue and insulin dose responses, but peripheral IR data were unavailable. Original Results find no basal glucose-uptake response to insulin in either group, with glutamate as a positive control. Functional postmortem resistance does not order it before amyloid/tau or establish trajectory benefit.

[3] In 150 matched ROS brains, IRS1 measures and ex-vivo response were not associated with global AD pathology; higher AKT phosphorylation was associated with pathology and lower cognition. Different region, phosphorylation site and assay prevent a simple exact replication claim; AKT integrates other growth-factor signals and tissue was obtained after disease.

[10] Orthogonal biofluid fractionation and single-molecule assays find L1CAM largely outside extracellular vesicles, challenging common neuronal immunocapture assumptions. This validates an assay limitation rather than absence of all neural EVs or falsity of every IRS1 signal; protocol and antibody-specific follow-up is necessary.

[11] Single-EV methods find L1CAM epitopes co-expressed with neuronal proteins and enrichment relative to L1CAM-negative EVs. Original Results include detergent, isotype, background, tetraspanin and hepatocyte-marker controls; 5G3 immunocapture enriches a subset co-carrying neuronal markers. Most soluble L1CAM remains outside EVs. This supports specific procedures rather than universal neuronal purity or live-cell insulin response.

[12] Across plasma, CSF and induced-neuron media, generic EVs and neuronal cargo are recovered but L1CAM predominantly remains free and is not enriched by immunocapture. Original Methods/Results use the same common 5G3 clone, calnexin-negative capture, EV markers, neuronal versus hematopoietic media and free-protein fractions. CSF proteomics is one pooled experiment. Negative enrichment differs from single-EV positive procedures; neither result measures live neuronal insulin causality.

[13] NIRKO loses insulin-mediated signalling and increases tau phosphorylation while showing no neuronal survival, memory or basal brain glucose deficit. Developmental deletion and compensation may mask adult functions, but marker changes alone are insufficient to generate AD-like functional degeneration in this tested context.

[14] Global IRS2 loss reduces developmental neuronal proliferation and older mice accumulate phosphorylated tau pathology. Major developmental growth and peripheral diabetes effects prevent isolation of an adult neuronal initiating mechanism or specificity to sporadic AD.

[15] Neural insulin-receptor deletion alters feeding, obesity and reproductive control without altered brain development or neuronal survival. Systemic consequences and developmental timing matter; metabolic regulation by brain IR is not sufficient proof of AD initiation.

[16] Adult neuron-specific IGF1R deletion in APP/PS1 mice improves spatial memory and reduces amyloid/inflammation, with neuronal structural changes and higher plasma amyloid. The direction opposes simple restoration of deficient IGF signalling. An adult genetic intervention is credited, but an APP/PS1 disease substrate and cell-size/proteostasis effects do not establish sporadic human primacy.

[17] IRS2 or neuronal IGF1R loss reduces amyloid and prevents premature Tg2576 death; neuronal IR loss does not alter lethality. Insulin and IGF branches differ. Developmental manipulation, sex differences and transgenic premature mortality constrain interpretation as natural aging or sporadic AD.

[18] Reduced IGF signalling protects AD-model mice against behavioural deficits, inflammation and neuronal loss while promoting dense amyloid aggregates. Protection from reduced signalling challenges a uniformly deficient-signalling model; developmental, aggregation and transgenic-context effects remain.

[19] Tau deletion impairs hippocampal insulin response through IRS1/PTEN and changes hypothalamic insulin effects; human tau haplotypes associate with glycemic traits. Provides a tau-to-insulin direction in models, not proof of the sole human ordering; lifelong deletion and peripheral glycemic traits do not reproduce pathological tau loss selectively.

[20] Amyloid oligomers activate JNK/TNF, inhibit IRS1 signalling in neurons and primate challenge, while exendin4 restores neuronal transport and improves transgenic-mouse cognition. Original Methods use male APP/PS1 mice and imposed primate oligomers; behavioural treatment starts at nine months. Comparable slight weight loss limits a simple caloric explanation, but GLP1 pleiotropy and no neuronal-IR-specific epistasis remain. Amyloid can induce resistance; rescue does not establish primary sporadic initiation.

[21] TNFR1/PKR disruption blocks oligomer-induced translation-signalling changes and memory impairment; insulin support rescues pathway abnormalities. Challenge genetics establish an amyloid-to-inflammatory/IRS1 injury route, not endogenous sporadic neuronal insulin failure as the first lesion.

[22] Insulin prevents oligomer-induced surface IR loss, oxidative stress and spine deterioration; kinase inhibition abolishes protection and rosiglitazone potentiates it. Culture rescue does not establish organismal trajectory change or natural initiation. A linked unidentified erratum remains unresolved.

[23] Active AKT, GSK3 inhibition and exendin4 prevent or reverse oligomer-induced BDNF transport deficits in wild-type and tau-null neurons. A transport rescue after imposed amyloid is narrower than survival or clinical benefit and is compatible with a reinforcing feedback pathway.

[24] Oligomers transiently inhibit AMPK/ATP and remove dendritic glucose transporters; NMDA-related intervention and insulin protect these measures. Acute metabolic responses after imposed amyloid are not proof of primary neuronal insulin resistance or sustained organismal disease modification.

[25] Reducing fly insulin signalling suppresses expressed-Aβ toxicity; Aβ elevates brain insulin signalling while peripheral signalling falls. A different species adds direction discrimination but is an imposed peptide system; IDE competition and branch/tissue effects oppose a simple low-signalling master mechanism.

[26] IR beta-subunit haploinsufficiency impairs late LTP and recognition memory. A lifelong global dosage change demonstrates a memory contribution in this background, not adult neuron-specific AD initiation or neurodegenerative trajectory.

[27] Neuronal PTP1B ablation and systemic trodusquemine prevent neuron loss and spatial memory deficits in hAPP-J20 mice; neuronal ablation preserves benefit without reducing inflammation. An endogenous signalling regulator is tested within an amyloid transgenic substrate, but PTP1B affects multiple receptors; restoration of insulin signalling specifically is not isolated by the abstract.

[28] Adult neuronal IGF1R suppression protects against oligomer-induced memory and inflammatory injury; related APP/PS1 transcriptional signatures converge. Shares the adult IGF-deletion programme with PMID26290229. Transcriptomic similarity is a lead, not independent causal proof; loss of the IGF branch again protects rather than harms.

[29] Central, but not systemic, oligomer infusion induces peripheral glucose intolerance; TNFR1 loss and central inflammatory/ER-stress inhibition prevent it. Demonstrates brain injury can drive peripheral metabolic change. Imposed exposure and transgenics do not establish native human direction or exclude reciprocal metabolic risk.

[30] Astrocyte IR deletion alters ATP exocytosis and dopamine-related behaviour; central ATP analog rescues depressive-like behaviour. A cell-type-specific glial route is functional but is neither neuronal insulin resistance nor AD initiation, and the behavioural endpoint is not AD cognitive trajectory.

[31] Inducible astrocyte IGFR loss impairs working memory, mitochondrial function and glucose/amyloid uptake. Astrocytic support differs from adult neuronal IGFR loss; single-context functional benefit and cultured mitochondrial changes do not establish a uniform neuronal mechanism.

[32] Inducible microglial IR loss impairs amyloid uptake and increases plaque/inflammation in 5xFAD, with mood/social changes. Microglial rather than neuronal perturbation, an imposed familial substrate and no demonstrated AD cognitive rescue constrain the registered neuronal claim. Updated preprint is not counted again.

[33] Prolonged high-fat diet worsens cognition and soluble amyloid in APP/PS1 mice while central and peripheral transporter responses differ. Diet begins at one month and changes lipids/inflammation/intake; broad developmental and systemic exposure cannot isolate adult neuronal insulin resistance.

[34] Intracerebroventricular STZ worsens Tg2576 spatial cognition, mortality and amyloid/tau measures. STZ is a toxic imposed trigger with effects beyond insulin signalling. Combining it with APP overexpression is not selective sufficiency for sporadic AD.

[35] Induced diabetes increases subicular amyloid plaques in middle-aged 3xTg mice but does not worsen learning/memory or AT8 burden. A functional negative comparator prevents biomarker-only changes earning trajectory credit; dose, timing and model baseline bound the null result.

[40] Genetically higher peripheral insulin sensitivity associates with lower AD risk in two outcome datasets, with attenuation after adjusting for T2D. MR is credited as an upper-ladder equivalent on its exposure, but peripheral sensitivity is not a selective neuronal signalling instrument and multivariable attenuation does not by itself prove mediation.

[41] Large T2D genetic instruments do not show an AD effect in the primary or FinnGen analysis. Null peripheral-liability estimates do not refute neuronal insulin resistance; largely overlapping FinnGen exposure/outcome samples and disease ascertainment matter.

[42] Overall T2D polygenic liability is null for AD, while a mechanism-specific insulin-sensitivity score predicts AD risk in IGAP. The corresponding HRS cognition association is nonsignificant. Instrument subclass, pleiotropy and shared GWAS require caution, and peripheral sensitivity is not neuron-specific primacy.

[43] Semaglutide improves short-term cognition and glucose/pathology measures in 3xTg mice; SIRT1 inhibition blocks cultured metabolic effects. Single-background, short-horizon functional benefit and cell-level epistasis do not establish human disease modification or that insulin failure was upstream of transgenic amyloid/tau.

[44] Seven-day intranasal insulin in 3xTg mice restores signalling/synaptic proteins and lowers Aβ40 and microglial activation. No functional benefit is measured here, and tau phosphorylation, glucose transporters and O-GlcNAcylation are unchanged; biomarker effects do not cross the trajectory boundary.

[45] In 38 healthy participants, eight-week intranasal insulin improves delayed recall without plasma insulin/glucose change. Normal memory modulation is a credible central-function effect but does not demonstrate AD disease modification or receptor-specific neuronal engagement.

[46] With the tested Aero Pump lispro regimen, human CSF drug is below quantification despite detection in beagle CSF. Direct delivery limitation is regimen/species-specific; CSF absence does not universally exclude local brain delivery, but animal exposure cannot validate human delivery.

[47] Intranasal radiolabelled insulin with an Aptar device reaches multiple brain regions in 16 older CN/MCI participants. Radiotracer distribution validates this device/exposure route, not therapeutic receptor phosphorylation, sustained bioactive concentration or clinical efficacy in other devices.

[48] Twenty-one-day randomized insulin pilot improves delayed recall, attention and caregiver function in 25 early AD/MCI participants without changed fasting insulin/glucose. Small short-term clinical changes can be symptomatic; multiple outcomes, diagnostic heterogeneity and an unresolved linked erratum limit trajectory inference.

[49] Four-month 104-person pilot finds improved delayed memory at 20 IU and preserved caregiver function at both doses, with exploratory FDG and CSF correlations. Small endpoint subsets, subgroup effects and no group CSF biomarker shift do not establish durable disease modification or selective neuronal insulin correction.

[50] In a 60-person 21-day detemir trial, high-dose memory effects differ by APOE, with improvement in carriers and worsening in noncarriers; daily function is null. Short-term, genotype-dependent effects require replication; baseline peripheral resistance is not brain target engagement. Linked erratum PMID25869922 is not a primary study.

[51] In a 36-person four-month trial, regular insulin improves memory while detemir and daily function are null; MRI and CSF-ratio signals accompany regular insulin. Small groups and outcomes constrain durable inference; a formulation-specific symptomatic response does not identify primary aetiology.

[52] Thirty-five-person glulisine trial shows no cognitive, functional or mood benefit at three/six months. Enrollment/duration and age imbalance limit precision; no selective neuronal engagement result is provided in the abstract, so this regimen's failure is bounded.

[53] SNIFF randomizes 289; the prespecified device2 cohort of 240 has no 12-month ADAS-cog benefit or secondary clinical/CSF benefit, with difference 0.0258 and P=.98. Original Methods define device2 as primary before lock with serial clinical gatekeeping. Device1 has nominal six-month but nonsignificant twelve-month cognition and greater entorhinal volume loss; device2 has small greater hippocampal loss. Primary CSF insulin is unchanged, leaving delivery/engagement uncertain rather than proving no transient neuronal effect.

[56] Thirty-eight-person six-month liraglutide trial reports preserved FDG metabolism but no between-group amyloid effect or cognitive change. Underpowered cognition and predominantly biomarker effects do not establish clinical trajectory modification; between-arm evidence must be distinguished from within-arm decline.

[58] ELAD randomizes 204 and finds no primary FDG difference or CDR/ADL benefit; ADAS-Exec has an unadjusted secondary signal. Original Methods use 154 complete SUV scans and 101 spectral sensitivity scans without outcome imputation; no CSF was collected and planned tau/TSPO PET was not done. Exploratory MRI, weight loss and limited brain-penetration arguments do not directly verify neuronal insulin engagement.

[59] Early-terminated exenatide pilot has 18 completers and no clinical, cognitive, MRI or main fluid response; EV Aβ42 alone falls. Severe power limitation and disputed neural EV provenance constrain interpretation; isolated EV biomarker improvement is not clinical disease modification.

[60] Thirty-two MCI participants randomized to weekly exenatide or no treatment show no ADAS-Cog benefit; weight/glucose fall and a female-worsening interaction is reported. Small open comparator and subgroup multiplicity limit the interaction; systemic engagement does not demonstrate correction of brain neuronal insulin resistance.

[61] EVOKE and EVOKE+ randomize 3808 amyloid-confirmed early AD participants; semaglutide does not improve 104-week CDR-SB in either trial, with differences -.08 and +.10. Primary abstract/status was examined; original numerical body was unavailable. Strong regimen-level clinical negatives do not demonstrate selective restoration of deficient neuronal signalling or refute every earlier metabolic-risk contribution.

[63] Pooled diabetes cardiovascular trials and Danish prescriptions both show lower recorded dementia with GLP1RA exposure; randomized estimate HR=.47. Random allocation deserves causal credit for this ancillary dementia outcome, not dismissal as observational. Endpoint capture, mixed source classes and systemic/vascular effects do not identify neuronal insulin mediation or established AD progression.

[64] 693-person rosiglitazone monotherapy phase3 trial fails prespecified cognition/global endpoints overall and in prospectively stratified APOE4-negative participants. Failure of this dose/PPAR regimen is not proof of an engaged neuronal IR defect's irrelevance; the earlier genotype rescue prediction did not replicate.

[65] Two rosiglitazone adjunct phase3 trials of 1496 and 1485 participants fail 48-week cognition/global endpoints with no APOE interaction. Two trials in one report strengthen clinical nonreplication, while PPAR pleiotropy and uncertain brain exposure prevent a universal insulin-mechanism exclusion.

[66] TOMMORROW enrolls 3494 cognitively healthy participants; low-dose pioglitazone does not delay MCI and the trial stops for futility. 0.8 mg regimen, incomplete event accrual and genetic-risk selection constrain prevention inference; the risk algorithm also misses its prespecified significance threshold and neuronal engagement is not established.

[67] In 80 overweight/obese aMCI adults without treated diabetes, metformin improves one adjusted recall outcome but not ADAS-cog or other endpoints. Baseline imbalance, low maximal-dose tolerability, multiple endpoints and no neuron-specific target engagement limit disease-modification inference.

[68] Four-week factorial insulin/empagliflozin trial meets safety aims and reports promising secondary cognition, perfusion and fluid-marker changes. Primary outcome is adverse events; short duration and secondary outcomes require larger longer confirmation. Empagliflozin vascular/systemic effects and fluid changes do not identify neuronal initiating resistance.

Tally: 53 discriminating contrasts. Null and negative contrasts are discriminators, not positive threshold crossings. Reports and replications are not independent causal lines. Assay controls and clinical nulls are discriminators, not positive neuronal causal threshold crossings. Two trials in one report and follow-up outcomes are reported explicitly rather than inflated counts.

Cap audit, precedence and Β/Γ/Δ boundary

The exact index primary rubric06 is retained with neurodegeneration01 co-governing. AD functional trajectory is the disease-specific healthspan analogue; FDG, phosphorylation, stress resistance and brief cognition alone do not establish modification. Evidence labels use rubric01 Strong for adequately randomized hard clinical regimen tests, including negatives; they are not positive Strong causal proof of neuronal mediation. Adult conditional and pathway-rescue experiments are credited within their systems, with single-background/pleiotropy limits. Cross-species direction discrimination, human functional assays, genetic instruments and several drug classes are evaluated together, not re-capped by one weak leg. Insulin and IGF branches have nonconcordant directions; the human instruments and drug effects do not selectively measure the claimed neuronal initiating exposure.

Gamma hard-rule audit: Biomarker shifts and acute cellular physiology do not establish disease/healthspan trajectory change. Developmental deletion, toxic STZ, familial transgenes and single-background outcomes constrain broader inference. Clinical negative trials are credited for their regimen/population, with CNS delivery and mediation limits explicit.

SC1 field-valid definition: “ - Genetic or pharmacologic intervention producing reproducible lifespan/healthspan extension - Must separate developmental effects from true aging effects - Biomarker change alone does NOT satisfy this threshold” Positive full-scope result: no. Functional postmortem challenge supports insulin/IGF resistance in selected AD hippocampal samples without diabetes, beyond static marker abundance. Other region/site assays are null or show higher downstream AKT activity. Γ records support for this finding within the causal review; it does not deny the observed functional resistance. No disease-trajectory crossing is inferred from tissue responsiveness alone.

SC2 field-valid definition: “ - Genetic or pharmacologic intervention producing reproducible lifespan/healthspan extension - Must separate developmental effects from true aging effects - Biomarker change alone does NOT satisfy this threshold” Positive full-scope result: yes. Loss, rescue and pathway support affect synaptic or cognitive function in defined models, reaching constrained causal contribution. Adult glial deletions demonstrate cell-type differences and neuronal PTP1B/GLP1 manipulations protect disease-model function. The IR and IGF branches cannot be merged: neuronal IGF attenuation can instead be protective, and selective adult neuronal IR mediation remains incompletely isolated.

SC3 field-valid definition: “ - Genetic or pharmacologic intervention producing reproducible lifespan/healthspan extension - Must separate developmental effects from true aging effects - Biomarker change alone does NOT satisfy this threshold” Positive full-scope result: no. Primary upstream failure remains a mechanistic possibility supported by imposed metabolic injury and antecedent peripheral/putative EV clues, but no selective neuronal instrument or serial preamyloid/tau functional assay establishes sporadic human order. Models demonstrate the opposite direction as well. Γ is an unmet positive primacy burden, not a blanket observational cap imposed on the entire convergent set.

SC4 field-valid definition: “ - Genetic or pharmacologic intervention producing reproducible lifespan/healthspan extension - Must separate developmental effects from true aging effects - Biomarker change alone does NOT satisfy this threshold” Positive full-scope result: no. Small insulin/metformin and recent short factorial trials show scoped clinical signals, but larger insulin, rosiglitazone, pioglitazone and GLP1 trials do not establish reproducible disease modification. SNIFF delivery uncertainty and pleiotropic/systemic engagement prevent interpreting failed regimens as a universal negative neuronal mechanism test. ELAD secondary cognition is unadjusted and EVOKE clinical outcomes are null.

Upward threshold ladder

SC1 Δ: “Requires at least one intervention changing lifespan or healthspan in a way consistent with the hypothesis.” Satisfied: no. Independently replicate cell-resolved ante-mortem and ex-vivo insulin-response assays with region, phosphorylation-site, vascular and peripheral-metabolic controls, then connect resistance to selective functional rescue.

SC2 Δ: “Requires at least one intervention changing lifespan or healthspan in a way consistent with the hypothesis.” Satisfied: yes.

SC2 Ε: “Requires causal effect in constrained contexts, backgrounds, or species.” Satisfied: yes.

SC2 Ζ: “Requires: - Reproducible intervention effects - Mechanistic narrowing - Evidence not reducible to general sickness, caloric reduction, or developmental artifacts” Satisfied: no. Adult neuronal IR-specific loss and restoration with receptor epistasis change disease-model function in independent backgrounds while controlling intake, vascular/metabolic effects and IGF branch direction.

SC3 Δ: “Requires at least one intervention changing lifespan or healthspan in a way consistent with the hypothesis.” Satisfied: no. Prospective functional neuronal signalling before amyloid/tau, or a neuron-selective genetic/natural instrument, predicts and mediates AD initiation independently of systemic and vascular routes.

SC4 Δ: “Requires at least one intervention changing lifespan or healthspan in a way consistent with the hypothesis.” Satisfied: no. A sufficiently powered biomarker-confirmed trial with validated brain delivery and neuronal pathway engagement demonstrates replicated durable cognition and daily-function benefit, with multiplicity and disease-versus-symptomatic outcomes specified.

Upper ladder: Peripheral insulin-sensitivity/T2D MR, multiple insulin/PPAR/GLP1 clinical drug classes and ancillary randomized dementia prevention were examined as legitimate upper-ladder candidates. MR exposures are not neuronal resistance and drugs have divergent/null clinical effects without linked neuronal mediator correction. Dose/genotype/device effects are not concordant independent replications; no unethical neuronal knock-out trial is required.

Independence: 68 reports are not 68 independent causal lines. ROS150, SNIFF device1 follow-ups, one liraglutide trial reanalysis, NIRKO and adult IGF programmes and shared GWAS are identified. EVOKE and REFLECT each report two trials but count once as a primary report. Orthogonal directions expose distinct errors without proving a common neuronal mechanism.

Functional tissue versus biomarker validation [2, 3, 11, 12]: Ex-vivo responses and orthogonal EV methods exclude different assay errors; postmortem timing and sample overlap remain.

Adult/cell and branch genetic contrasts [16, 31, 32, 13]: Distinguish developmental, neuronal, glial, IR and IGF explanations, though disease transgenes and mouse backgrounds constrain generality.

Genetic/population exposure instruments [40, 41, 42, 63]: Reduce conventional exposure confounding but do not isolate a neuronal initiating mediator.

Randomized clinical regimen contrasts [53, 58, 61, 65, 66]: Separate treatment assignment from indication bias while CNS engagement, timing and drug specificity remain.

Falsification and pitfalls

Signalling loss does not produce AD-like functional degeneration. NIRKO preserves survival/memory and basal glucose despite tau changes; neuronal IR loss does not alter Tg2576 mortality. Additional downgrade: False.

Reduced IGF signalling protects rather than harms. Adult neuronal deletion, Tg2576 and proteotoxicity mice, plus fly insulin reduction, oppose a uniformly deficient-signalling model. Additional downgrade: False.

Large therapeutic predictions fail. SNIFF primary cohort, rosiglitazone phase3, TOMMORROW, ELAD FDG and EVOKE cognition are null; small secondary signals do not replace primary outcomes. Additional downgrade: False.

Peripheral HOMA/MR relabelled neuronal insulin resistance. Risk cohorts and glycemic instruments.

Prediagnosis timing relabelled preamyloid/tau initiation. EV and peripheral biomarker cohorts.

Insulin and IGF branches or neurons and glia collapsed. Genetic interventions with opposing directions.

FDG or plasma marker change treated as neuronal receptor engagement. GLP1 trials and metabolic scans.

Shared selected device cohort counted as new efficacy replication. SNIFF WMH and CSF follow-ups.

Exploratory or nominal secondary outcome replaces a null primary. Device1, ELAD MRI/cognition and small pilots.

Developmental or STZ injury treated as endogenous aging/AD. Global IRS2 and toxic challenge models.

Adversarial steelman: above, then below

Δ for the complete claim would require at least one selective endogenous/equivalent initiating neuronal intervention with meaningful trajectory change and credible clinical correction. Functional models and ancillary dementia prevention are real, but do not identify that same mediator or temporal role.

Β would undercredit the assembled pathway perturbations, adult/cell genetic contrasts, ex-vivo functional challenge and human randomized signals. The model is more than a glucose-colocalization narrative, even though its primary human interpretation remains unestablished.

Γ retains mechanistic plausibility while separating an Ε constrained contribution from the unestablished primacy/application bundle. It neither equates AD with peripheral diabetes nor declares every insulin-related contribution absent.

Final score, causal direction and M/S/D

Γ follows the preregistered weakest-link bundle (Γ,Ε,Γ,Γ); primary-SC3 and scope-matched sensitivities also return Γ. Model-specific causal contribution earns Ε, while primary sporadic neuronal initiation and clinically proven correction remain unestablished. Functional tissue resistance, acute memory effects, peripheral prevention and FDG changes are kept distinct. Positive models and valid human causal equivalents are assessed together, without automatically re-capping their weaker observational members.

For the nominal initiating defect, first-match step1 is supported: imposed amyloid/tau disturbance can produce signalling resistance, and disease-model pathology can persist or improve when IGF signalling is blocked. This direction does not exclude reinforcing feedback or permissive insulin support in particular systems; it does not establish a sole native human sequence or a rate-limiting primary driver.

Momentum: Mechanistic only. The named programmes in this corpus — EVOKE and EVOKE+, the ELAD liraglutide trial and the intranasal insulin/empagliflozin factorial study, together with adult neuronal IGF1R, PTP1B and GLP1 perturbation work — are completed and published rather than running. No presently recruiting study addressing the SC3 boundary check, serial prelesion measurement of human neuronal insulin responsiveness with a selective neuronal instrument, is identified.

Stability: Stable. No named pending result is asserted to be likely to move the registered score imminently.

Divergence: Contested; no single implied-level arithmetic. These dimensions do not change level or confidence.

PROPOSED DECISIVE EXPERIMENTS

Predefine an amyloid/tau-negative sporadic-risk cohort with validated live neuron-specific insulin-response measurements; triangulate a selective neuronal exposure instrument and an early CNS-delivered intervention that corrects that response, with matched adult conditional model loss/rescue and repeated pathology/function follow-up.

Readout: Incident biomarker-confirmed AD and durable cognitive/daily-function trajectory with temporal neuronal mediator correction. Controls: Peripheral glycemia, vascular perfusion, intake/weight, IGF branch, glial engagement, validated delivery device, inactive exposure and prespecified multiplicity. Supporting: Resistance precedes lesions and selective correction prevents conversion through measured neuronal mediation, independently replicated. Falsifying: Resistance follows lesions or selective correction changes signalling without pathology/functional benefit, while systemic/vascular effects explain treatment associations.

RESEARCH TIMELINE

KNOWLEDGE GAPS

Validated live neuron-specific insulin-response assay with orthogonal EV provenance controls.

Prospective ordering before amyloid/tau rather than before clinical diagnosis.

Adult neuronal IR-specific loss/rescue separated from IGF, glial and developmental effects.

Human CNS delivery and receptor/pathway engagement linked to durable benefit in the same participants.

Prevention in diabetes separated from disease modification in amyloid-confirmed AD and from symptomatic memory effects.

Independent replication with diet/weight/vascular controls, shared cohorts and outcome multiplicity explicitly accounted for.

METHODS AND CONSENSUS CONTRIBUTION

Consensus was used through the signed-in browser for discovery, counterevidence, research-gap and timeline suggestions, and attempted checks of specific indexed full-text passages. Returned Methods/Results passages were reconciled with original-paper passages where available; failed retrievals, source substitutions and unverified aspects are recorded in the case reconciliation. Successful passage checks are distinguished from provider summaries. Provider screening counts, citation counts and meters were not used as primary counts or HSEF scores. This source log records access depth; it does not claim complete-paper or supplement review. The monthly API allowance was conserved: zero Consensus API calls in this batch.

REVIEWER NOTES

68 primary reports, six selected original partial-body examinations; zero complete-paper or supplement audits. Original 6 includes three web-section reviewer paraphrase receipts and three XML passage receipts. Two Pro browser searches used zero API calls; provider full-text narration and future-trial error are reconciled explicitly. No new independent blinded calibration. Exact v5.3 validation is structural rather than scientific peer review.

Search leads, reviews, protocols and correction notices do not pad the primary count. Downloaded XML does not earn examination credit. Linked notices remain attached; unexamined correction details limit inference. This is a coordinator source/claim consistency assessment, not independent blinded calibration.

ADVERSARIAL REVIEW AND APPROVAL

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

Scope of checks. All 68 reference PMIDs re-resolved against PubMed (68/68 correct for first author, title, journal and DOI; no retraction or expression of concern; three linked errata already recorded in the ledger's status_flags). Eleven load-bearing claims spot-checked against primary abstracts: EVOKE and EVOKE+ (3808 randomized of 9981 screened; week-104 CDR-SB differences −0.08, 95 per cent CI −0.35 to 0.20, and 0.10, −0.17 to 0.38; both trials discontinued for negative clinical outcome), ELAD (204 randomized, null primary cerebral glucose metabolism, unadjusted ADAS-Exec secondary signal, null ADCS-ADL), the two REFLECT rosiglitazone phase 3 trials (N = 1496 and N = 1485, no relevant differences at 48 weeks, no APOE interaction), the rosiglitazone monotherapy phase 3 study, the four-week intranasal insulin/empagliflozin factorial trial with a safety primary endpoint, and the Talbot ex-vivo functional-resistance, NIRKO, IRS2 and neural-EV sources. Scoring logic re-derived independently before the existing letter was read.

Findings.

Rejected findings (checked, no change).

1. Weakest-link aggregation is harsh for a conjunctive claim. Rejected: the registered claim joins primary neuronal initiation to clinical correction with an "and", SKILL.md Step 1 states that "A conjunctive hypothesis is scored on the conjunction, not on its easier half", and the primary-subclaim and scope-matched sensitivities return Γ as well, so the letter does not depend on the rule.
2. Role of mechanism as "downstream consequence". Checked against the framework's ordered derivation; step 1 is satisfied on the review's own evidence (imposed amyloid and tau disturbance produces signalling resistance; blocking neuronal IGF signalling can protect), and the review states the qualification explicitly.
3. The −0.08 and 0.10 CDR-SB differences are printed as "-.08" and "+.10". House numeric style; the values match the source and were left unchanged.
4. Reference 25 title differs from the PubMed title field. PubMed drops the italicised species name; the review's title is the published one.
5. Citation-year differences of ±1 between printed issue year and PubMed e-pub year (references 6, 8, 9, 23, 35, 40, 48, 49, 58). Publication-convention difference, not an error.

Residual limitations. Verification depth is bounded by access: 6 partial-body examinations and 62 abstract/status examinations. The EVOKE numerical body was not available to the original review and was verified here against the published abstract only.

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

REFERENCES

1. Steen E, Terry BM, Rivera EJ et al. Impaired insulin and insulin-like growth factor expression and signaling mechanisms in Alzheimer's disease--is this type 3 diabetes? Journal of Alzheimer's disease : JAD. 2005. PMID 15750215. DOI: 10.3233/jad-2005-7107. Access: abstract_metadata. Study: PMID15750215.

2. Talbot K, Wang HY, Kazi H et al. Demonstrated brain insulin resistance in Alzheimer's disease patients is associated with IGF-1 resistance, IRS-1 dysregulation, and cognitive decline. The Journal of clinical investigation. 2012. PMID 22476197. DOI: 10.1172/JCI59903. Access: full_text_partial. Study: PMID22476197.

3. Arvanitakis Z, Wang HY, Capuano AW et al. Brain Insulin Signaling, Alzheimer Disease Pathology, and Cognitive Function. Annals of neurology. 2020. PMID 32557841. DOI: 10.1002/ana.25826. Access: abstract_metadata. Study: ROS-150-matched-brains.

4. Tong H, Capuano AW, Carmichael OT et al. Brain Insulin Signaling is Associated with Late-Life Cognitive Decline. Aging and disease. 2024. PMID 38029396. DOI: 10.14336/AD.2023.1117. Access: abstract_metadata. Study: ROS-150-matched-brains.

5. Yarchoan M, Toledo JB, Lee EB et al. Abnormal serine phosphorylation of insulin receptor substrate 1 is associated with tau pathology in Alzheimer's disease and tauopathies. Acta neuropathologica. 2014. PMID 25107476. DOI: 10.1007/s00401-014-1328-5. Access: abstract_metadata. Study: PMID25107476.

6. An Y, Varma VR, Varma S et al. Evidence for brain glucose dysregulation in Alzheimer's disease. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2018. PMID 29055815. DOI: 10.1016/j.jalz.2017.09.011. Access: abstract_metadata. Study: PMID29055815.

7. Femminella GD, Livingston NR, Raza S et al. Does insulin resistance influence neurodegeneration in non-diabetic Alzheimer's subjects? Alzheimer's research & therapy. 2021. PMID 33597002. DOI: 10.1186/s13195-021-00784-w. Access: abstract_metadata. Study: PMID33597002.

8. Kapogiannis D, Boxer A, Schwartz JB et al. Dysfunctionally phosphorylated type 1 insulin receptor substrate in neural-derived blood exosomes of preclinical Alzheimer's disease. FASEB journal : official publication of the Federation of American Societies for Experimental Biology. 2015. PMID 25342129. DOI: 10.1096/fj.14-262048. Access: abstract_metadata. Study: PMID25342129.

9. Walker KA, Chawla S, Nogueras-Ortiz C et al. Neuronal insulin signaling and brain structure in nondemented older adults: the Atherosclerosis Risk in Communities Study. Neurobiology of aging. 2021. PMID 33160263. DOI: 10.1016/j.neurobiolaging.2020.09.022. Access: abstract_metadata. Study: PMID33160263.

10. Norman M, Ter-Ovanesyan D, Trieu W et al. L1CAM is not associated with extracellular vesicles in human cerebrospinal fluid or plasma. Nature methods. 2021. PMID 34092791. DOI: 10.1038/s41592-021-01174-8. Access: abstract_metadata. Study: PMID34092791.

11. Nogueras-Ortiz CJ, Eren E, Yao P et al. Single-extracellular vesicle (EV) analyses validate the use of L1 Cell Adhesion Molecule (L1CAM) as a reliable biomarker of neuron-derived EVs. Journal of extracellular vesicles. 2024. PMID 38868956. DOI: 10.1002/jev2.12459. Access: full_text_partial. Study: PMID38868956.

12. Kadam V, Wacker M, Oeckl P et al. Most L1CAM Is not Associated with Extracellular Vesicles in Human Biofluids and iPSC-Derived Neurons. Molecular neurobiology. 2025. PMID 40210837. DOI: 10.1007/s12035-025-04909-2. Access: full_text_partial. Study: PMID40210837.

13. Schubert M, Gautam D, Surjo D et al. Role for neuronal insulin resistance in neurodegenerative diseases. Proceedings of the National Academy of Sciences of the United States of America. 2004. PMID 14981233. DOI: 10.1073/pnas.0308724101. Access: abstract_metadata. Study: PMID14981233.

14. Schubert M, Brazil DP, Burks DJ et al. Insulin receptor substrate-2 deficiency impairs brain growth and promotes tau phosphorylation. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2003. PMID 12904469. DOI: 10.1523/JNEUROSCI.23-18-07084.2003. Access: abstract_metadata. Study: PMID12904469.

15. Brüning JC, Gautam D, Burks DJ et al. Role of brain insulin receptor in control of body weight and reproduction. Science (New York, N.Y.). 2000. PMID 11000114. DOI: 10.1126/science.289.5487.2122. Access: abstract_metadata. Study: PMID11000114.

16. Gontier G, George C, Chaker Z et al. Blocking IGF Signaling in Adult Neurons Alleviates Alzheimer's Disease Pathology through Amyloid-β Clearance. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2015. PMID 26290229. DOI: 10.1523/JNEUROSCI.0343-15.2015. Access: abstract_metadata. Study: Adult-neuronal-IGF1R-programme.

17. Freude S, Hettich MM, Schumann C et al. Neuronal IGF-1 resistance reduces Abeta accumulation and protects against premature death in a model of Alzheimer's disease. FASEB journal : official publication of the Federation of American Societies for Experimental Biology. 2009. PMID 19487308. DOI: 10.1096/fj.09-132043. Access: abstract_metadata. Study: PMID19487308.

18. Cohen E, Paulsson JF, Blinder P et al. Reduced IGF-1 signaling delays age-associated proteotoxicity in mice. Cell. 2009. PMID 20005808. DOI: 10.1016/j.cell.2009.11.014. Access: abstract_metadata. Study: PMID20005808.

19. Marciniak E, Leboucher A, Caron E et al. Tau deletion promotes brain insulin resistance. The Journal of experimental medicine. 2017. PMID 28652303. DOI: 10.1084/jem.20161731. Access: abstract_metadata. Study: PMID28652303.

20. Bomfim TR, Forny-Germano L, Sathler LB et al. An anti-diabetes agent protects the mouse brain from defective insulin signaling caused by Alzheimer's disease- associated Aβ oligomers. The Journal of clinical investigation. 2012. PMID 22476196. DOI: 10.1172/JCI57256. Access: full_text_partial. Study: PMID22476196.

21. Lourenco MV, Clarke JR, Frozza RL et al. TNF-α mediates PKR-dependent memory impairment and brain IRS-1 inhibition induced by Alzheimer's β-amyloid oligomers in mice and monkeys. Cell metabolism. 2013. PMID 24315369. DOI: 10.1016/j.cmet.2013.11.002. Access: abstract_metadata. Study: PMID24315369.

22. De Felice FG, Vieira MN, Bomfim TR et al. Protection of synapses against Alzheimer's-linked toxins: insulin signaling prevents the pathogenic binding of Abeta oligomers. Proceedings of the National Academy of Sciences of the United States of America. 2009. PMID 19188609. DOI: 10.1073/pnas.0809158106. Access: abstract_metadata. Study: PMID19188609.

23. Takach O, Gill TB, Silverman MA et al. Modulation of insulin signaling rescues BDNF transport defects independent of tau in amyloid-β oligomer-treated hippocampal neurons. Neurobiology of aging. 2015. PMID 25543463. DOI: 10.1016/j.neurobiolaging.2014.11.018. Access: abstract_metadata. Study: PMID25543463.

24. Seixas da Silva GS, Melo HM, Lourenco MV et al. Amyloid-β oligomers transiently inhibit AMP-activated kinase and cause metabolic defects in hippocampal neurons. The Journal of biological chemistry. 2017. PMID 28302722. DOI: 10.1074/jbc.M116.753525. Access: abstract_metadata. Study: PMID28302722.

25. Huang Y, Wan Z, Wang Z et al. Insulin signaling in Drosophila melanogaster mediates Aβ toxicity. Communications biology. 2019. PMID 30652125. DOI: 10.1038/s42003-018-0253-x. Access: abstract_metadata. Study: PMID30652125.

26. Nisticò R, Cavallucci V, Piccinin S et al. Insulin receptor β-subunit haploinsufficiency impairs hippocampal late-phase LTP and recognition memory. Neuromolecular medicine. 2012. PMID 22661254. DOI: 10.1007/s12017-012-8184-z. Access: abstract_metadata. Study: PMID22661254.

27. Ricke KM, Cruz SA, Qin Z et al. Neuronal Protein Tyrosine Phosphatase 1B Hastens Amyloid β-Associated Alzheimer's Disease in Mice. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2020. PMID 31915254. DOI: 10.1523/JNEUROSCI.2120-19.2019. Access: abstract_metadata. Study: PMID31915254.

28. George C, Gontier G, Lacube P et al. The Alzheimer's disease transcriptome mimics the neuroprotective signature of IGF-1 receptor-deficient neurons. Brain : a journal of neurology. 2017. PMID 28595357. DOI: 10.1093/brain/awx132. Access: abstract_metadata. Study: Adult-neuronal-IGF1R-programme.

29. Clarke JR, Lyra E Silva NM, Figueiredo CP et al. Alzheimer-associated Aβ oligomers impact the central nervous system to induce peripheral metabolic deregulation. EMBO molecular medicine. 2015. PMID 25617315. DOI: 10.15252/emmm.201404183. Access: abstract_metadata. Study: PMID25617315.

30. Cai W, Xue C, Sakaguchi M et al. Insulin regulates astrocyte gliotransmission and modulates behavior. The Journal of clinical investigation. 2018. PMID 29664737. DOI: 10.1172/JCI99366. Access: abstract_metadata. Study: PMID29664737.

31. Logan S, Pharaoh GA, Marlin MC et al. Insulin-like growth factor receptor signaling regulates working memory, mitochondrial metabolism, and amyloid-β uptake in astrocytes. Molecular metabolism. 2018. PMID 29398615. DOI: 10.1016/j.molmet.2018.01.013. Access: abstract_metadata. Study: PMID29398615.

32. Chen W, Liu X, Muñoz VR et al. Loss of insulin signaling in microglia impairs cellular uptake of Aβ and neuroinflammatory response exacerbating AD-like neuropathology. Proceedings of the National Academy of Sciences of the United States of America. 2025. PMID 40388612. DOI: 10.1073/pnas.2501527122. Access: abstract_metadata. Study: PMID40388612.

33. Guo Y, Ma X, Li P et al. High-fat diet induced discrepant peripheral and central nervous systems insulin resistance in APPswe/PS1dE9 and wild-type C57BL/6J mice. Aging. 2020. PMID 33291072. DOI: 10.18632/aging.202262. Access: abstract_metadata. Study: PMID33291072.

34. Plaschke K, Kopitz J, Siegelin M et al. Insulin-resistant brain state after intracerebroventricular streptozotocin injection exacerbates Alzheimer-like changes in Tg2576 AbetaPP-overexpressing mice. Journal of Alzheimer's disease : JAD. 2010. PMID 20110613. DOI: 10.3233/JAD-2010-1270. Access: abstract_metadata. Study: PMID20110613.

35. Hayashi-Park E, Ozment BN, Griffith CM et al. Experimentally induced diabetes worsens neuropathology, but not learning and memory, in middle aged 3xTg mice. Behavioural brain research. 2017. PMID 27173425. DOI: 10.1016/j.bbr.2016.05.020. Access: abstract_metadata. Study: PMID27173425.

36. Barthold D, Gibbons LE, Marcum ZA et al. Alzheimer's Disease-Related Neuropathology Among Patients with Medication Treated Type 2 Diabetes in a Community-Based Autopsy Cohort. Journal of Alzheimer's disease : JAD. 2021. PMID 34420950. DOI: 10.3233/JAD-210059. Access: abstract_metadata. Study: PMID34420950.

37. Arvanitakis Z, Capuano AW, Wang HY et al. Brain insulin signaling and cerebrovascular disease in human postmortem brain. Acta neuropathologica communications. 2021. PMID 33858515. DOI: 10.1186/s40478-021-01176-9. Access: abstract_metadata. Study: ROS-150-matched-brains.

38. Ott A, Stolk RP, van Harskamp F et al. Diabetes mellitus and the risk of dementia: The Rotterdam Study. Neurology. 1999. PMID 10599761. DOI: 10.1212/wnl.53.9.1937. Access: abstract_metadata. Study: PMID10599761.

39. Westwood AJ, Beiser A, Decarli C et al. Insulin-like growth factor-1 and risk of Alzheimer dementia and brain atrophy. Neurology. 2014. PMID 24706014. DOI: 10.1212/WNL.0000000000000382. Access: abstract_metadata. Study: PMID24706014.

40. Xu F, Wu S, Gao S et al. Causal association between insulin sensitivity index and Alzheimer's disease. Journal of neurochemistry. 2025. PMID 39479764. DOI: 10.1111/jnc.16254. Access: abstract_metadata. Study: PMID39479764.

41. Liu D, Baranova A, Zhang F et al. Evaluating the Causal Effect of Type 2 Diabetes on Alzheimer's Disease Using Large-Scale Genetic Data. The journal of prevention of Alzheimer's disease. 2024. PMID 39350373. DOI: 10.14283/jpad.2024.148. Access: abstract_metadata. Study: PMID39350373.

42. Walter S, Marden JR, Kubzansky LD et al. Diabetic Phenotypes and Late-Life Dementia Risk: A Mechanism-specific Mendelian Randomization Study. Alzheimer disease and associated disorders. 2016. PMID 26650880. DOI: 10.1097/WAD.0000000000000128. Access: abstract_metadata. Study: PMID26650880.

43. Wang ZJ, Li XR, Chai SF et al. Semaglutide ameliorates cognition and glucose metabolism dysfunction in the 3xTg mouse model of Alzheimer's disease via the GLP-1R/SIRT1/GLUT4 pathway. Neuropharmacology. 2023. PMID 37730113. DOI: 10.1016/j.neuropharm.2023.109716. Access: abstract_metadata. Study: PMID37730113.

44. Chen Y, Zhao Y, Dai CL et al. Intranasal insulin restores insulin signaling, increases synaptic proteins, and reduces Aβ level and microglia activation in the brains of 3xTg-AD mice. Experimental neurology. 2014. PMID 24918340. DOI: 10.1016/j.expneurol.2014.06.004. Access: abstract_metadata. Study: PMID24918340.

45. Benedict C, Hallschmid M, Hatke A et al. Intranasal insulin improves memory in humans. Psychoneuroendocrinology. 2004. PMID 15288712. DOI: 10.1016/j.psyneuen.2004.04.003. Access: abstract_metadata. Study: PMID15288712.

46. Lowe S, Sher E, Wishart G et al. An assessment of the central disposition of intranasally administered insulin lispro in the cerebrospinal fluid of healthy volunteers and beagle dogs. Drug delivery and translational research. 2017. PMID 27553192. DOI: 10.1007/s13346-016-0325-8. Access: abstract_metadata. Study: PMID27553192.

47. Sai KKS, Erichsen JM, Gollapelli KK et al. First-in-human positron emission tomography study of intranasal insulin in aging and MCI. Alzheimer's & dementia (New York, N. Y.). 2025. PMID 40703476. DOI: 10.1002/trc2.70123. Access: abstract_metadata. Study: PMID40703476.

48. Reger MA, Watson GS, Green PS et al. Intranasal insulin improves cognition and modulates beta-amyloid in early AD. Neurology. 2008. PMID 17942819. DOI: 10.1212/01.WNL.0000265401.62434.36. Access: abstract_metadata. Study: PMID17942819.

49. Craft S, Baker LD, Montine TJ et al. Intranasal insulin therapy for Alzheimer disease and amnestic mild cognitive impairment: a pilot clinical trial. Archives of neurology. 2012. PMID 21911655. DOI: 10.1001/archneurol.2011.233. Access: abstract_metadata. Study: PMID21911655.

50. Claxton A, Baker LD, Hanson A et al. Long-acting intranasal insulin detemir improves cognition for adults with mild cognitive impairment or early-stage Alzheimer's disease dementia. Journal of Alzheimer's disease : JAD. 2015. PMID 25374101. DOI: 10.3233/JAD-141791. Access: abstract_metadata. Study: PMID25374101.

51. Craft S, Claxton A, Baker LD et al. Effects of Regular and Long-Acting Insulin on Cognition and Alzheimer's Disease Biomarkers: A Pilot Clinical Trial. Journal of Alzheimer's disease : JAD. 2017. PMID 28372335. DOI: 10.3233/JAD-161256. Access: abstract_metadata. Study: PMID28372335.

52. Rosenbloom M, Barclay TR, Kashyap B et al. A Phase II, Single-Center, Randomized, Double-Blind, Placebo-Controlled Study of the Safety and Therapeutic Efficacy of Intranasal Glulisine in Amnestic Mild Cognitive Impairment and Probable Mild Alzheimer's Disease. Drugs & aging. 2021. PMID 33719017. DOI: 10.1007/s40266-021-00845-7. Access: abstract_metadata. Study: PMID33719017.

53. Craft S, Raman R, Chow TW et al. Safety, Efficacy, and Feasibility of Intranasal Insulin for the Treatment of Mild Cognitive Impairment and Alzheimer Disease Dementia: A Randomized Clinical Trial. JAMA neurology. 2020. PMID 32568367. DOI: 10.1001/jamaneurol.2020.1840. Access: full_text_partial. Study: SNIFF-NCT01767909.

54. Kellar D, Lockhart SN, Aisen P et al. Intranasal Insulin Reduces White Matter Hyperintensity Progression in Association with Improvements in Cognition and CSF Biomarker Profiles in Mild Cognitive Impairment and Alzheimer's Disease. The journal of prevention of Alzheimer's disease. 2021. PMID 34101779. DOI: 10.14283/jpad.2021.14. Access: abstract_metadata. Study: SNIFF-NCT01767909.

55. Kellar D, Register T, Lockhart SN et al. Intranasal insulin modulates cerebrospinal fluid markers of neuroinflammation in mild cognitive impairment and Alzheimer's disease: a randomized trial. Scientific reports. 2022. PMID 35079029. DOI: 10.1038/s41598-022-05165-3. Access: abstract_metadata. Study: SNIFF-NCT01767909.

56. Gejl M, Gjedde A, Egefjord L et al. In Alzheimer's Disease, 6-Month Treatment with GLP-1 Analog Prevents Decline of Brain Glucose Metabolism: Randomized, Placebo-Controlled, Double-Blind Clinical Trial. Frontiers in aging neuroscience. 2016. PMID 27252647. DOI: 10.3389/fnagi.2016.00108. Access: abstract_metadata. Study: Liraglutide-NCT01469351.

57. Gejl M, Brock B, Egefjord L et al. Blood-Brain Glucose Transfer in Alzheimer's disease: Effect of GLP-1 Analog Treatment. Scientific reports. 2017. PMID 29235507. DOI: 10.1038/s41598-017-17718-y. Access: abstract_metadata. Study: Liraglutide-NCT01469351.

58. Edison P, Femminella GD, Ritchie C et al. Liraglutide in mild to moderate Alzheimer's disease: a phase 2b clinical trial. Nature medicine. 2026. PMID 41326666. DOI: 10.1038/s41591-025-04106-7. Access: full_text_partial. Study: PMID41326666.

59. Mullins RJ, Mustapic M, Chia CW et al. A Pilot Study of Exenatide Actions in Alzheimer's Disease. Current Alzheimer research. 2019. PMID 31518224. DOI: 10.2174/1567205016666190913155950. Access: abstract_metadata. Study: PMID31518224.

60. Dei Cas A, Micheli MM, Aldigeri R et al. Long-acting exenatide does not prevent cognitive decline in mild cognitive impairment: a proof-of-concept clinical trial. Journal of endocrinological investigation. 2024. PMID 38565814. DOI: 10.1007/s40618-024-02320-7. Access: abstract_metadata. Study: PMID38565814.

61. Cummings JL, Atri A, Sano M et al. Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials. Lancet (London, England). 2026. PMID 41865758. DOI: 10.1016/S0140-6736(26)00459-9. Access: abstract_metadata. Study: PMID41865758.

62. Wang W, Wang Q, Qi X et al. Associations of semaglutide with first-time diagnosis of Alzheimer's disease in patients with type 2 diabetes: Target trial emulation using nationwide real-world data in the US. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2024. PMID 39445596. DOI: 10.1002/alz.14313. Access: abstract_metadata. Study: PMID39445596.

63. Nørgaard CH, Friedrich S, Hansen CT et al. Treatment with glucagon-like peptide-1 receptor agonists and incidence of dementia: Data from pooled double-blind randomized controlled trials and nationwide disease and prescription registers. Alzheimer's & dementia (New York, N. Y.). 2022. PMID 35229024. DOI: 10.1002/trc2.12268. Access: abstract_metadata. Study: PMID35229024.

64. Gold M, Alderton C, Zvartau-Hind M et al. Rosiglitazone monotherapy in mild-to-moderate Alzheimer's disease: results from a randomized, double-blind, placebo-controlled phase III study. Dementia and geriatric cognitive disorders. 2010. PMID 20733306. DOI: 10.1159/000318845. Access: abstract_metadata. Study: PMID20733306.

65. Harrington C, Sawchak S, Chiang C et al. Rosiglitazone does not improve cognition or global function when used as adjunctive therapy to AChE inhibitors in mild-to-moderate Alzheimer's disease: two phase 3 studies. Current Alzheimer research. 2011. PMID 21592048. DOI: 10.2174/156720511796391935. Access: abstract_metadata. Study: PMID21592048.

66. Burns DK, Alexander RC, Welsh-Bohmer KA et al. Safety and efficacy of pioglitazone for the delay of cognitive impairment in people at risk of Alzheimer's disease (TOMMORROW): a prognostic biomarker study and a phase 3, randomised, double-blind, placebo-controlled trial. The Lancet. Neurology. 2021. PMID 34146512. DOI: 10.1016/S1474-4422(21)00043-0. Access: abstract_metadata. Study: PMID34146512.

67. Luchsinger JA, Perez T, Chang H et al. Metformin in Amnestic Mild Cognitive Impairment: Results of a Pilot Randomized Placebo Controlled Clinical Trial. Journal of Alzheimer's disease : JAD. 2016. PMID 26890736. DOI: 10.3233/JAD-150493. Access: abstract_metadata. Study: PMID26890736.

68. Erichsen JM, Register TC, Sutphen C et al. A phase 2A/B randomized trial of metabolic modulators intranasal insulin and empagliflozin for MCI and early AD. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2025. PMID 41057918. DOI: 10.1002/alz.70704. Access: abstract_metadata. Study: PMID41057918.

Source-specific evidence cards

[1] Impaired insulin and insulin-like growth factor expression and signaling mechanisms in Alzheimer's disease--is this type 3 diabetes?

Shows: AD postmortem brains show lower insulin/IGF pathway expression and altered survival-signalling markers, motivating the type3-diabetes framing.

Does not show: Terminal expression and cell-composition changes do not establish an initiating neuronal endocrine defect or temporal direction.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[2] Demonstrated brain insulin resistance in Alzheimer's disease patients is associated with IGF-1 resistance, IRS-1 dysregulation, and cognitive decline.

Shows: AD hippocampal tissue without diabetes responds less to insulin and IGF challenge; IRS1 serine measures associate with cognition and pathology.

Does not show: Original Methods validate low-PMI tissue and insulin dose responses, but peripheral IR data were unavailable. Original Results find no basal glucose-uptake response to insulin in either group, with glutamate as a positive control. Functional postmortem resistance does not order it before amyloid/tau or establish trajectory benefit.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: Results:ex-vivo validation and glucose uptake; Methods:Subjects, tissue preparation, insulin stimulation. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[3] Brain Insulin Signaling, Alzheimer Disease Pathology, and Cognitive Function.

Shows: In 150 matched ROS brains, IRS1 measures and ex-vivo response were not associated with global AD pathology; higher AKT phosphorylation was associated with pathology and lower cognition.

Does not show: Different region, phosphorylation site and assay prevent a simple exact replication claim; AKT integrates other growth-factor signals and tissue was obtained after disease.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[4] Brain Insulin Signaling is Associated with Late-Life Cognitive Decline.

Shows: In the same 150 ROS individuals, higher terminal AKT phosphorylation tracked faster prior cognitive decline; IRS1 signals were largely null.

Does not show: Longitudinal cognition with a terminal pathway assay is not longitudinal neuronal insulin measurement; this is shared-cohort follow-up rather than independent replication.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[5] Abnormal serine phosphorylation of insulin receptor substrate 1 is associated with tau pathology in Alzheimer's disease and tauopathies.

Shows: IRS1 serine phosphorylation occurs in AD and other tauopathies; tau transgenics and high-fat diet link it with abnormal tau.

Does not show: Disease specificity and direction are unresolved; abnormal tau can precede the marker and diet affects several metabolic routes.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[6] Evidence for brain glucose dysregulation in Alzheimer's disease.

Shows: BLSA autopsies link higher tissue glucose, lower glycolytic flux and GLUT3 with AD severity; prior plasma glucose relates to tissue glucose.

Does not show: Glucose accumulation and predominantly insulin-independent transport do not diagnose neuronal insulin resistance; preclinical symptoms are not prelesion order.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[7] Does insulin resistance influence neurodegeneration in non-diabetic Alzheimer's subjects?

Shows: Peripheral HOMA2 associates with lower hippocampal FDG metabolism and total grey matter in 130 nondiabetic AD participants.

Does not show: HOMA2 measures peripheral resistance, not neuronal response; cross-sectional FDG and volume do not identify initiation or causal mediation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[8] Dysfunctionally phosphorylated type 1 insulin receptor substrate in neural-derived blood exosomes of preclinical Alzheimer's disease.

Shows: Immunocaptured putative neural exosome IRS1 measures differ in AD, diabetes and FTD; 22 future AD cases show differences 1-10 years before diagnosis.

Does not show: Preclinical clinical timing does not precede amyloid/tau necessarily. Small selected case-control series, classifier validation and disputed L1CAM enrichment constrain neuronal provenance and prediction generalization.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[9] Neuronal insulin signaling and brain structure in nondemented older adults: the Atherosclerosis Risk in Communities Study.

Shows: L1CAM-enriched vesicle insulin-signalling composite associates with less WMH and greater temporal volume in 88 nondemented ARIC-PET participants.

Does not show: Cross-sectional association, amyloid-dependent attenuation and uncertain neuronal vesicle isolation do not establish causal neuronal signalling or prelesion initiation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[10] L1CAM is not associated with extracellular vesicles in human cerebrospinal fluid or plasma.

Shows: Orthogonal biofluid fractionation and single-molecule assays find L1CAM largely outside extracellular vesicles, challenging common neuronal immunocapture assumptions.

Does not show: This validates an assay limitation rather than absence of all neural EVs or falsity of every IRS1 signal; protocol and antibody-specific follow-up is necessary.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[11] Single-extracellular vesicle (EV) analyses validate the use of L1 Cell Adhesion Molecule (L1CAM) as a reliable biomarker of neuron-derived EVs.

Shows: Single-EV methods find L1CAM epitopes co-expressed with neuronal proteins and enrichment relative to L1CAM-negative EVs.

Does not show: Original Results include detergent, isotype, background, tetraspanin and hepatocyte-marker controls; 5G3 immunocapture enriches a subset co-carrying neuronal markers. Most soluble L1CAM remains outside EVs. This supports specific procedures rather than universal neuronal purity or live-cell insulin response.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph6, 1:RESULTS:paragraph9, 1:RESULTS:paragraph12, 1:RESULTS:paragraph15, 1:RESULTS:paragraph18. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[12] Most L1CAM Is not Associated with Extracellular Vesicles in Human Biofluids and iPSC-Derived Neurons.

Shows: Across plasma, CSF and induced-neuron media, generic EVs and neuronal cargo are recovered but L1CAM predominantly remains free and is not enriched by immunocapture.

Does not show: Original Methods/Results use the same common 5G3 clone, calnexin-negative capture, EV markers, neuronal versus hematopoietic media and free-protein fractions. CSF proteomics is one pooled experiment. Negative enrichment differs from single-EV positive procedures; neither result measures live neuronal insulin causality.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Materials and Methods:paragraph3, 1:Materials and Methods:paragraph9, 1:Materials and Methods:paragraph13, 2:Results:paragraph2, 2:Results:paragraph3, 2:Results:paragraph7, 2:Results:paragraph9, 2:Results:paragraph10, 2:Results:paragraph12. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[13] Role for neuronal insulin resistance in neurodegenerative diseases.

Shows: NIRKO loses insulin-mediated signalling and increases tau phosphorylation while showing no neuronal survival, memory or basal brain glucose deficit.

Does not show: Developmental deletion and compensation may mask adult functions, but marker changes alone are insufficient to generate AD-like functional degeneration in this tested context.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[14] Insulin receptor substrate-2 deficiency impairs brain growth and promotes tau phosphorylation.

Shows: Global IRS2 loss reduces developmental neuronal proliferation and older mice accumulate phosphorylated tau pathology.

Does not show: Major developmental growth and peripheral diabetes effects prevent isolation of an adult neuronal initiating mechanism or specificity to sporadic AD.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[15] Role of brain insulin receptor in control of body weight and reproduction.

Shows: Neural insulin-receptor deletion alters feeding, obesity and reproductive control without altered brain development or neuronal survival.

Does not show: Systemic consequences and developmental timing matter; metabolic regulation by brain IR is not sufficient proof of AD initiation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[16] Blocking IGF Signaling in Adult Neurons Alleviates Alzheimer's Disease Pathology through Amyloid-β Clearance.

Shows: Adult neuron-specific IGF1R deletion in APP/PS1 mice improves spatial memory and reduces amyloid/inflammation, with neuronal structural changes and higher plasma amyloid.

Does not show: The direction opposes simple restoration of deficient IGF signalling. An adult genetic intervention is credited, but an APP/PS1 disease substrate and cell-size/proteostasis effects do not establish sporadic human primacy.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[17] Neuronal IGF-1 resistance reduces Abeta accumulation and protects against premature death in a model of Alzheimer's disease.

Shows: IRS2 or neuronal IGF1R loss reduces amyloid and prevents premature Tg2576 death; neuronal IR loss does not alter lethality.

Does not show: Insulin and IGF branches differ. Developmental manipulation, sex differences and transgenic premature mortality constrain interpretation as natural aging or sporadic AD.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[18] Reduced IGF-1 signaling delays age-associated proteotoxicity in mice.

Shows: Reduced IGF signalling protects AD-model mice against behavioural deficits, inflammation and neuronal loss while promoting dense amyloid aggregates.

Does not show: Protection from reduced signalling challenges a uniformly deficient-signalling model; developmental, aggregation and transgenic-context effects remain.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[19] Tau deletion promotes brain insulin resistance.

Shows: Tau deletion impairs hippocampal insulin response through IRS1/PTEN and changes hypothalamic insulin effects; human tau haplotypes associate with glycemic traits.

Does not show: Provides a tau-to-insulin direction in models, not proof of the sole human ordering; lifelong deletion and peripheral glycemic traits do not reproduce pathological tau loss selectively.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[20] An anti-diabetes agent protects the mouse brain from defective insulin signaling caused by Alzheimer's disease- associated Aβ oligomers.

Shows: Amyloid oligomers activate JNK/TNF, inhibit IRS1 signalling in neurons and primate challenge, while exendin4 restores neuronal transport and improves transgenic-mouse cognition.

Does not show: Original Methods use male APP/PS1 mice and imposed primate oligomers; behavioural treatment starts at nine months. Comparable slight weight loss limits a simple caloric explanation, but GLP1 pleiotropy and no neuronal-IR-specific epistasis remain. Amyloid can induce resistance; rescue does not establish primary sporadic initiation.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: Results:Figure8 and exendin4 rescue; Methods:primate infusion, APP/PS1 animals and behavioural tests. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[21] TNF-α mediates PKR-dependent memory impairment and brain IRS-1 inhibition induced by Alzheimer's β-amyloid oligomers in mice and monkeys.

Shows: TNFR1/PKR disruption blocks oligomer-induced translation-signalling changes and memory impairment; insulin support rescues pathway abnormalities.

Does not show: Challenge genetics establish an amyloid-to-inflammatory/IRS1 injury route, not endogenous sporadic neuronal insulin failure as the first lesion.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[22] Protection of synapses against Alzheimer's-linked toxins: insulin signaling prevents the pathogenic binding of Abeta oligomers.

Shows: Insulin prevents oligomer-induced surface IR loss, oxidative stress and spine deterioration; kinase inhibition abolishes protection and rosiglitazone potentiates it.

Does not show: Culture rescue does not establish organismal trajectory change or natural initiation. A linked unidentified erratum remains unresolved.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMIDnot supplied. Subclaims: SC2, SC3.

[23] Modulation of insulin signaling rescues BDNF transport defects independent of tau in amyloid-β oligomer-treated hippocampal neurons.

Shows: Active AKT, GSK3 inhibition and exendin4 prevent or reverse oligomer-induced BDNF transport deficits in wild-type and tau-null neurons.

Does not show: A transport rescue after imposed amyloid is narrower than survival or clinical benefit and is compatible with a reinforcing feedback pathway.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[24] Amyloid-β oligomers transiently inhibit AMP-activated kinase and cause metabolic defects in hippocampal neurons.

Shows: Oligomers transiently inhibit AMPK/ATP and remove dendritic glucose transporters; NMDA-related intervention and insulin protect these measures.

Does not show: Acute metabolic responses after imposed amyloid are not proof of primary neuronal insulin resistance or sustained organismal disease modification.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[25] Insulin signaling in Drosophila melanogaster mediates Aβ toxicity.

Shows: Reducing fly insulin signalling suppresses expressed-Aβ toxicity; Aβ elevates brain insulin signalling while peripheral signalling falls.

Does not show: A different species adds direction discrimination but is an imposed peptide system; IDE competition and branch/tissue effects oppose a simple low-signalling master mechanism.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[26] Insulin receptor β-subunit haploinsufficiency impairs hippocampal late-phase LTP and recognition memory.

Shows: IR beta-subunit haploinsufficiency impairs late LTP and recognition memory.

Does not show: A lifelong global dosage change demonstrates a memory contribution in this background, not adult neuron-specific AD initiation or neurodegenerative trajectory.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2.

[27] Neuronal Protein Tyrosine Phosphatase 1B Hastens Amyloid β-Associated Alzheimer's Disease in Mice.

Shows: Neuronal PTP1B ablation and systemic trodusquemine prevent neuron loss and spatial memory deficits in hAPP-J20 mice; neuronal ablation preserves benefit without reducing inflammation.

Does not show: An endogenous signalling regulator is tested within an amyloid transgenic substrate, but PTP1B affects multiple receptors; restoration of insulin signalling specifically is not isolated by the abstract.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[28] The Alzheimer's disease transcriptome mimics the neuroprotective signature of IGF-1 receptor-deficient neurons.

Shows: Adult neuronal IGF1R suppression protects against oligomer-induced memory and inflammatory injury; related APP/PS1 transcriptional signatures converge.

Does not show: Shares the adult IGF-deletion programme with PMID26290229. Transcriptomic similarity is a lead, not independent causal proof; loss of the IGF branch again protects rather than harms.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[29] Alzheimer-associated Aβ oligomers impact the central nervous system to induce peripheral metabolic deregulation.

Shows: Central, but not systemic, oligomer infusion induces peripheral glucose intolerance; TNFR1 loss and central inflammatory/ER-stress inhibition prevent it.

Does not show: Demonstrates brain injury can drive peripheral metabolic change. Imposed exposure and transgenics do not establish native human direction or exclude reciprocal metabolic risk.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[30] Insulin regulates astrocyte gliotransmission and modulates behavior.

Shows: Astrocyte IR deletion alters ATP exocytosis and dopamine-related behaviour; central ATP analog rescues depressive-like behaviour.

Does not show: A cell-type-specific glial route is functional but is neither neuronal insulin resistance nor AD initiation, and the behavioural endpoint is not AD cognitive trajectory.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2.

[31] Insulin-like growth factor receptor signaling regulates working memory, mitochondrial metabolism, and amyloid-β uptake in astrocytes.

Shows: Inducible astrocyte IGFR loss impairs working memory, mitochondrial function and glucose/amyloid uptake.

Does not show: Astrocytic support differs from adult neuronal IGFR loss; single-context functional benefit and cultured mitochondrial changes do not establish a uniform neuronal mechanism.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2.

[32] Loss of insulin signaling in microglia impairs cellular uptake of Aβ and neuroinflammatory response exacerbating AD-like neuropathology.

Shows: Inducible microglial IR loss impairs amyloid uptake and increases plaque/inflammation in 5xFAD, with mood/social changes.

Does not show: Microglial rather than neuronal perturbation, an imposed familial substrate and no demonstrated AD cognitive rescue constrain the registered neuronal claim. Updated preprint is not counted again.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. UpdateOf: PMID39229221. Subclaims: SC2, SC3.

[33] High-fat diet induced discrepant peripheral and central nervous systems insulin resistance in APPswe/PS1dE9 and wild-type C57BL/6J mice.

Shows: Prolonged high-fat diet worsens cognition and soluble amyloid in APP/PS1 mice while central and peripheral transporter responses differ.

Does not show: Diet begins at one month and changes lipids/inflammation/intake; broad developmental and systemic exposure cannot isolate adult neuronal insulin resistance.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[34] Insulin-resistant brain state after intracerebroventricular streptozotocin injection exacerbates Alzheimer-like changes in Tg2576 AbetaPP-overexpressing mice.

Shows: Intracerebroventricular STZ worsens Tg2576 spatial cognition, mortality and amyloid/tau measures.

Does not show: STZ is a toxic imposed trigger with effects beyond insulin signalling. Combining it with APP overexpression is not selective sufficiency for sporadic AD.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[35] Experimentally induced diabetes worsens neuropathology, but not learning and memory, in middle aged 3xTg mice.

Shows: Induced diabetes increases subicular amyloid plaques in middle-aged 3xTg mice but does not worsen learning/memory or AT8 burden.

Does not show: A functional negative comparator prevents biomarker-only changes earning trajectory credit; dose, timing and model baseline bound the null result.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

[36] Alzheimer's Disease-Related Neuropathology Among Patients with Medication Treated Type 2 Diabetes in a Community-Based Autopsy Cohort.

Shows: In 118 medication-treated ACT autopsies, conventional AD neuropathology is not associated with drug class; some adjusted quantitative amyloid measures are lower.

Does not show: Small descriptive analysis and indication/severity confounding cannot establish therapeutic neuronal insulin mediation or clinical disease modification.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC4.

[37] Brain insulin signaling and cerebrovascular disease in human postmortem brain.

Shows: In the same 150 ROS brains, stimulated AKT and cytoplasmic IRS1 measures associate with infarcts, with a separate amyloid-angiopathy association.

Does not show: Shared cohort adds vascular specificity rather than independent replication; infarction can cause cognition/AD-like symptoms without proving neuronal insulin primacy.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

[38] Diabetes mellitus and the risk of dementia: The Rotterdam Study.

Shows: Prospective Rotterdam data link diabetes to incident clinically diagnosed dementia and AD over mean 2.1 years.

Does not show: Peripheral diabetes, clinical diagnosis and short prediagnosis follow-up do not distinguish vascular contribution, preclinical reverse causation or neuronal initiating resistance; insulin treatment marks severity.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

[39] Insulin-like growth factor-1 and risk of Alzheimer dementia and brain atrophy.

Shows: Lower serum IGF1 predicts higher AD dementia incidence in Framingham and higher IGF1 associates with greater brain volume.

Does not show: Serum hormone is not neuronal receptor responsiveness; cohort confounding and clinically dementia-free enrollment do not establish preamyloid initiation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

[40] Causal association between insulin sensitivity index and Alzheimer's disease.

Shows: Genetically higher peripheral insulin sensitivity associates with lower AD risk in two outcome datasets, with attenuation after adjusting for T2D.

Does not show: MR is credited as an upper-ladder equivalent on its exposure, but peripheral sensitivity is not a selective neuronal signalling instrument and multivariable attenuation does not by itself prove mediation.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

[41] Evaluating the Causal Effect of Type 2 Diabetes on Alzheimer's Disease Using Large-Scale Genetic Data.

Shows: Large T2D genetic instruments do not show an AD effect in the primary or FinnGen analysis.

Does not show: Null peripheral-liability estimates do not refute neuronal insulin resistance; largely overlapping FinnGen exposure/outcome samples and disease ascertainment matter.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

[42] Diabetic Phenotypes and Late-Life Dementia Risk: A Mechanism-specific Mendelian Randomization Study.

Shows: Overall T2D polygenic liability is null for AD, while a mechanism-specific insulin-sensitivity score predicts AD risk in IGAP.

Does not show: The corresponding HRS cognition association is nonsignificant. Instrument subclass, pleiotropy and shared GWAS require caution, and peripheral sensitivity is not neuron-specific primacy.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3.

[43] Semaglutide ameliorates cognition and glucose metabolism dysfunction in the 3xTg mouse model of Alzheimer's disease via the GLP-1R/SIRT1/GLUT4 pathway.

Shows: Semaglutide improves short-term cognition and glucose/pathology measures in 3xTg mice; SIRT1 inhibition blocks cultured metabolic effects.

Does not show: Single-background, short-horizon functional benefit and cell-level epistasis do not establish human disease modification or that insulin failure was upstream of transgenic amyloid/tau.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC4.

[44] Intranasal insulin restores insulin signaling, increases synaptic proteins, and reduces Aβ level and microglia activation in the brains of 3xTg-AD mice.

Shows: Seven-day intranasal insulin in 3xTg mice restores signalling/synaptic proteins and lowers Aβ40 and microglial activation.

Does not show: No functional benefit is measured here, and tau phosphorylation, glucose transporters and O-GlcNAcylation are unchanged; biomarker effects do not cross the trajectory boundary.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC4.

[45] Intranasal insulin improves memory in humans.

Shows: In 38 healthy participants, eight-week intranasal insulin improves delayed recall without plasma insulin/glucose change.

Does not show: Normal memory modulation is a credible central-function effect but does not demonstrate AD disease modification or receptor-specific neuronal engagement.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC4.

[46] An assessment of the central disposition of intranasally administered insulin lispro in the cerebrospinal fluid of healthy volunteers and beagle dogs.

Shows: With the tested Aero Pump lispro regimen, human CSF drug is below quantification despite detection in beagle CSF.

Does not show: Direct delivery limitation is regimen/species-specific; CSF absence does not universally exclude local brain delivery, but animal exposure cannot validate human delivery.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[47] First-in-human positron emission tomography study of intranasal insulin in aging and MCI.

Shows: Intranasal radiolabelled insulin with an Aptar device reaches multiple brain regions in 16 older CN/MCI participants.

Does not show: Radiotracer distribution validates this device/exposure route, not therapeutic receptor phosphorylation, sustained bioactive concentration or clinical efficacy in other devices.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[48] Intranasal insulin improves cognition and modulates beta-amyloid in early AD.

Shows: Twenty-one-day randomized insulin pilot improves delayed recall, attention and caregiver function in 25 early AD/MCI participants without changed fasting insulin/glucose.

Does not show: Small short-term clinical changes can be symptomatic; multiple outcomes, diagnostic heterogeneity and an unresolved linked erratum limit trajectory inference.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMIDnot supplied. Subclaims: SC4.

[49] Intranasal insulin therapy for Alzheimer disease and amnestic mild cognitive impairment: a pilot clinical trial.

Shows: Four-month 104-person pilot finds improved delayed memory at 20 IU and preserved caregiver function at both doses, with exploratory FDG and CSF correlations.

Does not show: Small endpoint subsets, subgroup effects and no group CSF biomarker shift do not establish durable disease modification or selective neuronal insulin correction.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[50] Long-acting intranasal insulin detemir improves cognition for adults with mild cognitive impairment or early-stage Alzheimer's disease dementia.

Shows: In a 60-person 21-day detemir trial, high-dose memory effects differ by APOE, with improvement in carriers and worsening in noncarriers; daily function is null.

Does not show: Short-term, genotype-dependent effects require replication; baseline peripheral resistance is not brain target engagement. Linked erratum PMID25869922 is not a primary study.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID25869922. Subclaims: SC4.

[51] Effects of Regular and Long-Acting Insulin on Cognition and Alzheimer's Disease Biomarkers: A Pilot Clinical Trial.

Shows: In a 36-person four-month trial, regular insulin improves memory while detemir and daily function are null; MRI and CSF-ratio signals accompany regular insulin.

Does not show: Small groups and outcomes constrain durable inference; a formulation-specific symptomatic response does not identify primary aetiology.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[52] A Phase II, Single-Center, Randomized, Double-Blind, Placebo-Controlled Study of the Safety and Therapeutic Efficacy of Intranasal Glulisine in Amnestic Mild Cognitive Impairment and Probable Mild Alzheimer's Disease.

Shows: Thirty-five-person glulisine trial shows no cognitive, functional or mood benefit at three/six months.

Does not show: Enrollment/duration and age imbalance limit precision; no selective neuronal engagement result is provided in the abstract, so this regimen's failure is bounded.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[53] Safety, Efficacy, and Feasibility of Intranasal Insulin for the Treatment of Mild Cognitive Impairment and Alzheimer Disease Dementia: A Randomized Clinical Trial.

Shows: SNIFF randomizes 289; the prespecified device2 cohort of 240 has no 12-month ADAS-cog benefit or secondary clinical/CSF benefit, with difference 0.0258 and P=.98.

Does not show: Original Methods define device2 as primary before lock with serial clinical gatekeeping. Device1 has nominal six-month but nonsignificant twelve-month cognition and greater entorhinal volume loss; device2 has small greater hippocampal loss. Primary CSF insulin is unchanged, leaving delivery/engagement uncertain rather than proving no transient neuronal effect.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: Methods:Participants, Statistical Analysis; Results:CSF, Secondary Analyses. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[54] Intranasal Insulin Reduces White Matter Hyperintensity Progression in Association with Improvements in Cognition and CSF Biomarker Profiles in Mild Cognitive Impairment and Alzheimer's Disease.

Shows: In SNIFF's 49-person device1 subgroup, insulin reduces deep/frontal WMH progression associated with clinical and CSF measures.

Does not show: This is a secondary analysis of the same selected device cohort, not independent trial replication or direct neuronal signalling engagement.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[55] Intranasal insulin modulates cerebrospinal fluid markers of neuroinflammation in mild cognitive impairment and Alzheimer's disease: a randomized trial.

Shows: SNIFF's cohort showing benefit has treatment-related CSF IL6, interferon-gamma and eotaxin changes with exploratory clinical/imaging associations.

Does not show: Selected shared cohort, multiple marker tests and inflammatory rather than neuron-specific engagement limit mechanistic inference; the large primary cohort remains null.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[56] In Alzheimer's Disease, 6-Month Treatment with GLP-1 Analog Prevents Decline of Brain Glucose Metabolism: Randomized, Placebo-Controlled, Double-Blind Clinical Trial.

Shows: Thirty-eight-person six-month liraglutide trial reports preserved FDG metabolism but no between-group amyloid effect or cognitive change.

Does not show: Underpowered cognition and predominantly biomarker effects do not establish clinical trajectory modification; between-arm evidence must be distinguished from within-arm decline.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[57] Blood-Brain Glucose Transfer in Alzheimer's disease: Effect of GLP-1 Analog Treatment.

Shows: Reanalysis of the same 38-person liraglutide trial estimates increased blood-brain glucose transport capacity relative to placebo.

Does not show: Model-derived BBB transfer is a distinct downstream endpoint, not direct neuronal IR correction or a second independent clinical trial.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[58] Liraglutide in mild to moderate Alzheimer's disease: a phase 2b clinical trial.

Shows: ELAD randomizes 204 and finds no primary FDG difference or CDR/ADL benefit; ADAS-Exec has an unadjusted secondary signal.

Does not show: Original Methods use 154 complete SUV scans and 101 spectral sensitivity scans without outcome imputation; no CSF was collected and planned tau/TSPO PET was not done. Exploratory MRI, weight loss and limited brain-penetration arguments do not directly verify neuronal insulin engagement.

Access/status: Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph0, 1:Results:paragraph15, 1:Results:paragraph17, 1:Results:paragraph19, 1:Results:paragraph20, 1:Results:paragraph22, 1:Results:paragraph24, 3:Methods:paragraph1, 3:Methods:paragraph3, 3:Methods:paragraph5, 3:Methods:paragraph6, 3:Methods:paragraph10, 3:Methods:paragraph12, 3:Methods:paragraph63, 3:Methods:paragraph65, 3:Methods:paragraph66. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[59] A Pilot Study of Exenatide Actions in Alzheimer's Disease.

Shows: Early-terminated exenatide pilot has 18 completers and no clinical, cognitive, MRI or main fluid response; EV Aβ42 alone falls.

Does not show: Severe power limitation and disputed neural EV provenance constrain interpretation; isolated EV biomarker improvement is not clinical disease modification.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[60] Long-acting exenatide does not prevent cognitive decline in mild cognitive impairment: a proof-of-concept clinical trial.

Shows: Thirty-two MCI participants randomized to weekly exenatide or no treatment show no ADAS-Cog benefit; weight/glucose fall and a female-worsening interaction is reported.

Does not show: Small open comparator and subgroup multiplicity limit the interaction; systemic engagement does not demonstrate correction of brain neuronal insulin resistance.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[61] Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials.

Shows: EVOKE and EVOKE+ randomize 3808 amyloid-confirmed early AD participants; semaglutide does not improve 104-week CDR-SB in either trial, with differences -.08 and +.10.

Does not show: Primary abstract/status was examined; original numerical body was unavailable. Strong regimen-level clinical negatives do not demonstrate selective restoration of deficient neuronal signalling or refute every earlier metabolic-risk contribution.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[62] Associations of semaglutide with first-time diagnosis of Alzheimer's disease in patients with type 2 diabetes: Target trial emulation using nationwide real-world data in the US.

Shows: Matched US diabetes EHR emulations associate semaglutide with lower first AD diagnosis than seven active medication comparators.

Does not show: Target-trial emulation remains nonrandomized; indication, residual severity/adherence/ascertainment and clinically unrecognized preexisting pathology persist. Prevention in diabetes is a different estimand from EVOKE symptomatic AD.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

[63] Treatment with glucagon-like peptide-1 receptor agonists and incidence of dementia: Data from pooled double-blind randomized controlled trials and nationwide disease and prescription registers.

Shows: Pooled diabetes cardiovascular trials and Danish prescriptions both show lower recorded dementia with GLP1RA exposure; randomized estimate HR=.47.

Does not show: Random allocation deserves causal credit for this ancillary dementia outcome, not dismissal as observational. Endpoint capture, mixed source classes and systemic/vascular effects do not identify neuronal insulin mediation or established AD progression.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

[64] Rosiglitazone monotherapy in mild-to-moderate Alzheimer's disease: results from a randomized, double-blind, placebo-controlled phase III study.

Shows: 693-person rosiglitazone monotherapy phase3 trial fails prespecified cognition/global endpoints overall and in prospectively stratified APOE4-negative participants.

Does not show: Failure of this dose/PPAR regimen is not proof of an engaged neuronal IR defect's irrelevance; the earlier genotype rescue prediction did not replicate.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[65] Rosiglitazone does not improve cognition or global function when used as adjunctive therapy to AChE inhibitors in mild-to-moderate Alzheimer's disease: two phase 3 studies.

Shows: Two rosiglitazone adjunct phase3 trials of 1496 and 1485 participants fail 48-week cognition/global endpoints with no APOE interaction.

Does not show: Two trials in one report strengthen clinical nonreplication, while PPAR pleiotropy and uncertain brain exposure prevent a universal insulin-mechanism exclusion.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[66] Safety and efficacy of pioglitazone for the delay of cognitive impairment in people at risk of Alzheimer's disease (TOMMORROW): a prognostic biomarker study and a phase 3, randomised, double-blind, placebo-controlled trial.

Shows: TOMMORROW enrolls 3494 cognitively healthy participants; low-dose pioglitazone does not delay MCI and the trial stops for futility.

Does not show: 0.8 mg regimen, incomplete event accrual and genetic-risk selection constrain prevention inference; the risk algorithm also misses its prespecified significance threshold and neuronal engagement is not established.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC3, SC4.

[67] Metformin in Amnestic Mild Cognitive Impairment: Results of a Pilot Randomized Placebo Controlled Clinical Trial.

Shows: In 80 overweight/obese aMCI adults without treated diabetes, metformin improves one adjusted recall outcome but not ADAS-cog or other endpoints.

Does not show: Baseline imbalance, low maximal-dose tolerability, multiple endpoints and no neuron-specific target engagement limit disease-modification inference.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

[68] A phase 2A/B randomized trial of metabolic modulators intranasal insulin and empagliflozin for MCI and early AD.

Shows: Four-week factorial insulin/empagliflozin trial meets safety aims and reports promising secondary cognition, perfusion and fluid-marker changes.

Does not show: Primary outcome is adverse events; short duration and secondary outcomes require larger longer confirmation. Empagliflozin vascular/systemic effects and fluid changes do not identify neuronal initiating resistance.

Access/status: Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC4.

Documentary record

Every grade on this site is reproducible from its record. The files below are the posted edition exactly as assessed.

• SHA256.txt
• codex-correction-provenance.json
• consensus-addendum.md
• consensus-validation.json
• edition-approval-audit.json
• type3-diabetes-alz.review.md
• type3-diabetes-alz.scorecard.json
• validation-report.json
• validation-round2-2026-10-14.json
• validation.txt

Sources: 68 · load-bearing claims: 73 · verified: 73 · unverified: 0 · contradicted: 0 · schema 5.3