Dementia and Alzheimer's Disease: A Comprehensive Evidence-Based Guide

Executive Summary

The conceptual frame

Dementia is not a disease. It is an acquired, progressive syndrome that involves more than one cognitive domain and reaches a severity that impairs independent living — and behind it lie at least a dozen distinct molecular pathologies. The most important conceptual shift of the past decade is the separation of syndrome from biology: Alzheimer's disease is now defined biologically by amyloid-β and tau biomarkers rather than by the presence of dementia (Jack et al., 2024). This is a reframing that changes everything from diagnosis to trial design to who is a candidate for treatment.

The burden

In 2019 roughly 57.4 million people worldwide were living with dementia; the projection for 2050 is 152.8 million — a 166% increase, almost all of it driven by population ageing and growth (GBD 2019 Dementia Forecasting Collaborators, 2022). The global annual cost exceeded 1.3 trillion dollars in 2019 and is projected to exceed 2.8 trillion by 2030 (WHO, 2021). Roughly half of that cost is unpaid informal care — and the overwhelming majority of that care is provided by women.

Five developments that define the picture in 2026

1. Disease modification is real, but modest. Anti-amyloid monoclonal antibodies — lecanemab (CLARITY-AD; van Dyck et al., 2023) and donanemab (TRAILBLAZER-ALZ 2; Sims et al., 2023) — slow clinical worsening in early symptomatic Alzheimer's by roughly 27–36% over 18 months. Verified, replicated, biologically coherent — and clinically small. Differences of 0.45–0.7 points on the CDR-SB sit below or at the edge of most estimates of the minimum clinically important difference for this population. These drugs slow the disease; they do not stop or reverse it.
2. Diagnosis has been transformed — this is where the real revolution lies. Plasma p-tau217 now approaches the accuracy of CSF and amyloid PET in detecting Alzheimer's pathology. The FDA approved the first blood-based in vitro diagnostic test (Lumipulse G pTau217/Aβ1-42 plasma ratio) on 16 May 2025. In a Swedish real-world study presented at AAIC 2026 (London, 12–15 July), family physicians who saw the blood test result identified Alzheimer's with 93% accuracy — almost identical to the 94% specialists achieved in the same patients. Before the test, family physicians were 65% accurate.
3. Prevention now has positive randomised evidence in large and diverse populations. The US POINTER trial (n=2,111; 2025) showed that a structured multidomain lifestyle programme improved global cognition compared with a self-guided version. The LatAm-FINGERS trial, presented at AAIC 2026 and published simultaneously in the Lancet, replicated this in more than 1,000 participants across 11 Latin American countries: the group receiving structured coaching showed 55% greater improvement on the cognitive composite over two years. Combined with the Lancet Commission's estimate that roughly 45% of dementia risk is attributable to 14 modifiable risk factors (Livingston et al., 2024), prevention is today the highest-return public health target in the field.
4. Non-amyloid pharmacology is largely unproven — and the most expensive test of it failed. The evoke/evoke+ trials of oral semaglutide (n=3,808, 104 weeks) failed decisively in November 2025: no difference on CDR-SB (−0.08 in evoke; +0.10 in evoke+), no difference on secondary cognitive and functional endpoints — and yet CSF tau markers and inflammatory markers improved significantly. This is the cleanest refutation we have of the assumption that "the biomarker improved, therefore the patient improved".
5. The first genuine signal has arrived on the tau side. Diranersen (BIIB080) is an antisense oligonucleotide targeting tau mRNA. The phase 2 CELIA trial (topline in May 2026, detailed at AAIC in July 2026) failed its primary endpoint of demonstrating a dose–response relationship; it did, however, show strong reductions in tau PET and CSF tau at every dose and a slowing of clinical decline — with the clearest response at the lowest dose. Biogen is proceeding to phase 3. The tau field today stands roughly where amyloid stood in 2016: clear target engagement, an encouraging direction, and no positive pivotal trial yet.

Tiers of evidence — what we know today

The honest answer to the question patients ask

Dementia due to neurodegeneration cannot today be cured, halted or reversed. Roughly 1–5% of dementia presentations have a potentially reversible cause (B12 deficiency, hypothyroidism, normal pressure hydrocephalus, autoimmune encephalitis, chronic subdural haematoma, drug effects) and even in those, complete recovery is uncommon. Neurogenesis does occur in the adult human hippocampus, but it is quantitatively trivial next to the neuronal loss of established dementia. No stem cell or gene therapy has restored cognitive function in a controlled human study.

The realistic near-term goal is not repair; it is prevention and early biological intervention.

What Is Dementia? Definitions and Classification

Definition

DSM-5-TR replaced the term "dementia" with major neurocognitive disorder: a significant decline from a previous level of performance in one or more cognitive domains (complex attention, executive function, learning and memory, language, perceptual-motor, social cognition); based on both the concern of the patient, an informant or a clinician and on objective impairment on standardised testing; severe enough to interfere with independence in daily life; not occurring in the context of delirium and not better explained by another mental disorder (APA, 2022). Mild neurocognitive disorder corresponds roughly to MCI: measurable decline, but independence preserved.

ICD-11 (6D80–6D8Z) retains the term "dementia": an acquired brain syndrome with decline in two or more cognitive domains, not attributable to normal ageing, that impairs independence (WHO, 2019).

Distinctions that must be kept sharp:

These distinctions are not bureaucratic. "Alzheimer's disease" is now the name of a biology; "dementia" is a description of a level of function. A person can carry Alzheimer's pathology and die without ever developing dementia — and a substantial share of the older population does exactly that.

The organising principle of classification: proteinopathy, not syndrome

Modern classification is molecular. Four protein systems account for the overwhelming majority of neurodegenerative dementias; vascular damage is the principal non-protein contributor:

• Amyloidopathy + tauopathy → Alzheimer's disease
• Primary tauopathy → PSP, CBD, PART, FTLD-tau/MAPT
• Synucleinopathy → DLB, PDD, MSA
• TDP-43 proteinopathy → FTLD-TDP, LATE, C9orf72, GRN
• Prion → sCJD, vCJD, GSS, FFI
• Vascular → large vessel/multi-infarct, small vessel (CADASIL, Binswanger), strategic infarct, haemorrhagic/cerebral amyloid angiopathy
• Secondary/other → metabolic, infectious, autoimmune, structural (NPH, tumour, subdural), toxic, traumatic (CTE)

The main subtypes

Alzheimer's disease (AD). 60–70% of dementias; present as a contributing pathology in a far higher proportion. The defining pathology: extracellular Aβ plaques (Thal phases 1–5) plus intracellular neurofibrillary tangles of hyperphosphorylated tau (Braak stages I–VI). Typical phenotype: hippocampal-type episodic memory impairment — a delayed-recall deficit that does not improve with cueing. Atypical variants (~15%, more common in early onset): posterior cortical atrophy (visuospatial), logopenic variant PPA (word-finding, sentence repetition), behavioural/dysexecutive AD, corticobasal syndrome.

Vascular cognitive impairment (VaD/VCID). 15–20% as a primary cause; contributes in 30–50% of all dementias. The most common subtype is subcortical ischaemic small vessel disease. The clinical signature: dominance of executive function and processing speed deficits, early gait disturbance, urinary urgency, apathy, pseudobulbar affect. Imaging reporting is standardised by STRIVE-2 (Duering et al., 2023): lacunar infarcts, white matter hyperintensities, perivascular spaces, microbleeds, cortical superficial siderosis.

Dementia with Lewy bodies (DLB). 4–8% of clinical diagnoses, but 15–25% at autopsy — the most commonly missed of the common dementias. Core features (McKeith et al., 2017): marked fluctuation in attention and alertness, recurrent and well-formed visual hallucinations, REM sleep behaviour disorder (which can begin decades before cognitive decline), spontaneous parkinsonism. Supportive biomarkers: reduced dopamine transporter uptake on DaTscan, abnormal ¹²³I-MIBG myocardial scintigraphy, REM sleep without atonia on polysomnography. In recent years α-synuclein seed amplification assays (SAA) have been able to detect misfolded α-synuclein in CSF and skin biopsy with sensitivity and specificity of around 90% or better.

Frontotemporal dementia (FTD). 5–10% of all dementias, but the most common dementia under the age of 60. Mean onset is 58 years. 30–40% are familial. Subsyndromes: behavioural variant (disinhibition, apathy, loss of empathy, hyperorality), semantic variant PPA, non-fluent/agrammatic PPA, FTD-MND, PSP-Richardson, corticobasal syndrome.

Mixed dementia — the most important fact about classification. In community-based autopsy series (the Religious Orders Study, Rush MAP), fewer than one third of cases diagnosed clinically as AD carry pure AD pathology; the majority carry additional Lewy body, TDP-43/LATE, hippocampal sclerosis or vascular pathology (Schneider et al., 2007). Each additional pathology independently accelerates cognitive decline.

LATE (limbic-predominant age-related TDP-43 encephalopathy). The TDP-43 proteinopathy of the oldest old; its amnestic phenotype is clinically indistinguishable from AD; it is found in 20–50% of brains over the age of 80 (Nelson et al., 2019). It is the principal source of "amyloid-negative amnestic dementia" and cannot be identified in life — the largest biomarker gap in the field.

Potentially reversible and secondary dementias. Roughly 1–5% of presentations. Every dementia work-up should include at least B12, TSH, calcium and structural imaging. Autoimmune panels are indicated in rapidly progressive, fluctuating or seizure-associated presentations.

A comparison matrix of the subtypes

Epidemiology: The Global Picture and Turkey

Global prevalence and projections

GRADE: High certainty for the direction and approximate magnitude of the projection; Moderate for regional point estimates — primary data from low- and middle-income countries are sparse and the models rely heavily on imputation.

The age gradient and the "incidence paradox"

Age-specific incidence roughly doubles every 5–6 years from the age of 65: about 2–4 cases per 1,000 person-years at 65–69, and 70–90 cases above 90. This exponential relationship is the dominant epidemiological fact in the field; every other risk factor rides on top of it.

Two things are true at once, and they are constantly conflated in public discussion:

1. Age-specific incidence is falling in high-income countries. The Framingham Heart Study reported a relative decline of about 20% per decade (Satizabal et al., 2016); the Rotterdam Study, the UK CFAS I/II and the Health and Retirement Study report consistent declines of 13–24% per decade. The attributed causes: rising educational attainment (cognitive reserve), improved cardiovascular risk management, falling smoking rates, stroke prevention.
2. The absolute number of cases is rising rapidly, because population ageing far outweighs the per-person decline.

A critical caveat: this decline appears confined to high-income countries. In many low- and middle-income countries with worsening cardiometabolic risk profiles, age-specific rates are flat or rising.

Sex differences

• Prevalence: roughly two thirds of people with AD dementia are women. The dominant explanation is that more women reach the highest-risk age bands.
• Incidence: after adjustment for age, the sex difference in Western cohorts is small and inconsistent.
• Beyond demography: faster tau accumulation and faster cognitive decline after an MCI diagnosis have been reported in women; APOE ε4 carries stronger risk in women within the 65–75 window (Neu et al., 2017).
• Care burden: women provide roughly 70% of global informal dementia care hours.
• Male-predominant patterns: vascular dementia, alcohol-related dementia, CTE and FTD-MND.

The picture in Turkey

According to TurkStat's 2022 Turkey Health Survey, the prevalence of Alzheimer's disease among people over 65 in Turkey is approximately 5.5% (Turkish Ministry of Health, General Directorate of Public Health, 2025). The Turkish Alzheimer's Association estimates that roughly 600,000 families are living with the disease. Studies conducted in Turkey suggest that prevalence, particularly in large cities, is at levels similar to Western countries.

Two points matter especially for Turkey:

1. The share of the older population is rising rapidly, and because of the exponential shape of the incidence curve the rise in case numbers will be far steeper than population growth.
2. The modifiable risk profile — hypertension, diabetes, obesity, smoking and a large share of cohorts with low educational attainment — makes Turkey one of the countries that could benefit most from the Lancet Commission framework. Prevention has a higher return here than in most high-income countries.

The economic burden

The cost curve is dominated by care, not by drugs. The health-economic consequence is this: even an expensive treatment that delays entry into institutional care by 12 months can be cost-neutral or cost-saving. That is the central economic argument for disease modification — and it is also why NICE did not recommend lecanemab and donanemab for the NHS: the price does not make that calculation work.

The diagnostic gap

Globally, between 50% and 90% of dementia is undiagnosed; the higher figure belongs to low- and middle-income countries (WHO, 2021). Blood biomarkers are the first technology with the potential to close that gap at scale. That is the most practical message of this article.

Neurobiology: How the Disease Progresses in the Brain

An integrated model

No single-pathway model fits all the data. The defensible contemporary synthesis is a multi-hit cascade in which proteinopathy is the initiator, innate immunity the amplifier, and synaptic and network-level failure the direct cause of symptoms:

The steps of the cascade run as follows:

1. Genetic risk (APOE, TREM2, polygenic burden), environmental and vascular exposures, and the age-related decline in proteostasis and clearance converge; protein misfolding and aggregation begin.
2. Misfolded protein shows prion-like seeding and spread along interconnected networks.
3. Microglial and astrocytic activation follows; chronic neuroinflammation and complement-mediated synapse pruning begin.
4. Blood–brain barrier breakdown and hypoperfusion on one side, oxidative stress and mitochondrial failure on the other, load onto the same process.
5. Synapse loss emerges — this is the strongest structural correlate of cognitive decline.
6. Neuronal death and regional atrophy turn into network disconnection.
7. The clinical dementia syndrome appears; cognitive and brain reserve delay this final step.

Amyloid-β

Production. APP is cleaved either along the non-amyloidogenic route (α-secretase/ADAM10) or the amyloidogenic route (β-secretase/BACE1, then the γ-secretase/presenilin complex), yielding Aβ peptides of 37–43 residues. Aβ42 aggregates most readily; the diagnostically informative quantity, however, is the Aβ42/Aβ40 ratio.

Aggregation. Monomer → oligomer → protofibril → fibril → plaque. The most important modern correction to the amyloid cascade hypothesis is this: the truly synaptotoxic species are not fibrillar plaques but soluble oligomers and protofibrils. That is precisely the rationale for lecanemab's protofibril selectivity, and it explains why plaque burden correlates poorly with cognition while oligomer burden correlates well.

Clearance failure. Late-onset AD is predominantly a clearance disorder, not an overproduction disorder. Stable isotope labelling kinetics show that in sporadic AD, Aβ42 production is not increased but its clearance is reduced (Mawuenyega et al., 2010). The clearance routes: enzymatic degradation (neprilysin, IDE), LRP1-mediated transcytosis, perivascular/glymphatic drainage and microglial phagocytosis. APOE ε4 impairs several of these at once.

Where does the amyloid hypothesis stand in 2026? The evidence that amyloid is a necessary initiating factor is strong: every autosomal dominant mutation increases Aβ42 or the Aβ42/40 ratio; the APP A673T variant protects against AD (Jonsson et al., 2012); Down syndrome (trisomy of APP) produces almost universal AD pathology; and amyloid removal produces measurable, dose-dependent clinical slowing. The evidence that amyloid is not a sufficient explanation is equally strong: 20–40% of cognitively normal older adults are amyloid-positive; plaque burden plateaus early while cognition continues to decline; and complete plaque clearance yields only about 30% slowing.

Tau

Tau (MAPT, chromosome 17q21) stabilises axonal microtubules. Alternative splicing of exon 10 produces the 3R and 4R isoforms; the 3R:4R ratio defines the classes of tauopathy.

The pathogenic sequence: hyperphosphorylation (GSK-3β, CDK5, MARK) → detachment from microtubules → conformational change → oligomerisation → filament formation → tangle.

Braak staging (I–VI): transentorhinal → limbic → neocortical. Unlike amyloid, the distribution and burden of tau map closely onto clinical severity and regional atrophy. That is why tau PET is the superior staging biomarker and why tau has become the primary target of the second wave of treatments.

Prion-like spread. Tau seeds released from neurons are taken up by connected neurons and template the misfolding of local tau — a connectome-constrained, trans-synaptic spread. This is the rationale for extracellular anti-tau antibodies (bepranemab, E2814); the need to reach intracellular tau as well is the rationale for the ASO approach (diranersen).

α-Synuclein and TDP-43

α-Synuclein (SNCA) is a 140-amino-acid presynaptic protein that regulates synaptic vesicle traffic. It misfolds into β-sheet-rich fibrils to form Lewy bodies. Braak PD staging proposes caudo-rostral spread (dorsal motor nucleus of the vagus → locus coeruleus → substantia nigra → limbic → neocortex); the "gut-first" subtype, beginning in the enteric nervous system and ascending via the vagus, is supported by the sequence of constipation, hyposmia and RBD that precedes motor onset by years to decades. SAA assays are now the first true molecular biomarker of synucleinopathy.

TDP-43 (TARDBP) is a nuclear DNA/RNA-binding protein that regulates splicing and mRNA stability. In pathology it is cleared from the nucleus and accumulates in the cytoplasm in phosphorylated and ubiquitinated form — that is, both a loss of nuclear function and a gain of cytoplasmic toxicity. The central discovery of the past decade is that TDP-43 suppresses cryptic exon inclusion; its loss permits cryptic exon inclusion in STMN2 and UNC13A. TDP-43 pathology underlies most of FTLD, about 97% of ALS and LATE; it is a frequent co-pathology in AD and independently worsens memory decline.

The vascular contribution and the origin of ARIA

The mechanisms: large vessel infarct; small vessel arteriolosclerosis and lipohyalinosis; cerebral amyloid angiopathy (CAA); chronic hypoperfusion; blood–brain barrier leakage; impaired neurovascular coupling.

The amyloid–vascular interaction is bidirectional and synergistic: hypoperfusion increases APP processing and impairs perivascular Aβ clearance; vascular Aβ deposition further impairs clearance and vessel reactivity. This is the mechanistic basis of mixed dementia — and the reason vascular risk control confers benefit independently of amyloid status.

Neuroinflammation: where genetics redirected the field

The majority of late-onset AD GWAS loci point not to neuronal biology but to microglial and myeloid function: TREM2, CD33, MS4A, ABI3, PLCG2, INPP5D, SPI1, BIN1, CR1.

• Microglia shift from a homeostatic state to a disease-associated microglia (DAM) state; this is a TREM2-dependent, two-stage transcriptional programme (Keren-Shaul et al., 2017). The functional consequence is dual: early activation is protective (plaque compaction, phagocytic clearance), sustained activation is destructive (complement C1q/C3-mediated synapse pruning, pro-inflammatory cytokines, NLRP3 inflammasome activation). Timing determines the sign of the effect — and this is the single most important reason anti-inflammatory drug development in AD has failed repeatedly.
• Astrocytes enter reactive states; they impair glutamate uptake (reduced EAAT2 → excitotoxicity), impair potassium buffering and weaken the blood–brain barrier. The biomarker read-out is plasma GFAP, and it is one of the earliest detectable changes in the AD cascade.
• Peripheral immunity: the evoke/evoke+ failure is instructive here — semaglutide markedly lowered plasma hs-CRP, and the clinical benefit was zero. It shows that a purely peripheral anti-inflammatory effect is insufficient, and that approaches must enter the CNS and be cell-type-specific and temporally targeted.

The blood–brain barrier and the glymphatic system

Blood–brain barrier breakdown is detectable in the hippocampus early in cognitive decline, independently of amyloid and tau (Nation et al., 2019), and is exacerbated by APOE ε4 via the cyclophilin A–MMP9 pathway in pericytes. Pericyte loss is an early and functionally important event.

The glymphatic system — perivascular exchange of CSF and interstitial fluid, driven by AQP4-dependent astrocytic water transport and markedly increased during slow-wave sleep — provides a mechanistic bridge between sleep disruption and amyloid accumulation. A single night of sleep deprivation produces a PET-measurable rise in Aβ in the human hippocampus and thalamus (Shokri-Kojori et al., 2018). Glymphatic anatomy and flow dynamics in humans remain contested; read the model as strongly suggestive rather than settled.

Synapse loss — the heart of the matter

Synapse loss is the strongest structural correlate of cognitive impairment in AD — stronger than plaque count, tangle count or gross atrophy (Terry et al., 1991). The mechanisms: binding of Aβ oligomers to PrP^C, NMDAR and mGluR5; suppression of LTP; complement-mediated microglial engulfment of synapses; mislocalisation of tau to dendritic spines; and neurotransmitter system failure (cholinergic basal forebrain degeneration, noradrenergic loss in the locus coeruleus — which may be the earliest site of tau pathology in the human brain).

The network degeneration hypothesis

Neurodegenerative diseases target intrinsic connectivity networks rather than random regions (Seeley et al., 2009):

Hub regions — highly connected, metabolically expensive and the last to myelinate — are selectively vulnerable.

Genetics and Risk Factors

Autosomal dominant Alzheimer's disease (ADAD)

Fewer than 1% of all AD cases. Near-complete penetrance; age at onset is largely determined by the specific mutation and shows high concordance within families.

Down syndrome (trisomy 21, triplication of APP) is the largest genetically determined AD population: more than 90% develop AD pathology by the age of 40 and most develop dementia in their sixties. This population is now central to the design of prevention trials.

Protective variants — the field's most informative "natural experiments": APP A673T (the "Icelandic variant") reduces BACE1 cleavage by about 40% and protects against AD (Jonsson et al., 2012) — the strongest genetic evidence that lowering amyloid prevents AD. The Christchurch APOE3 variant (R136S) was found homozygous in a Colombian PSEN1 E280A carrier and delayed dementia onset by roughly 30 years despite an extreme amyloid burden (Arboleda-Velasquez et al., 2019); in a second individual the RELN-COLBOS variant conferred similar protection (Lopera et al., 2023). These two cases directly triggered APOE-directed and reelin-pathway treatment programmes.

APOE — the dominant common risk locus

The mechanisms: isoform-dependent impairment of Aβ clearance and aggregation kinetics; lipid transport and cholesterol homeostasis; amplification of tau-mediated neurodegeneration independently of amyloid; blood–brain barrier breakdown via cyclophilin A–MMP9; altered microglial lipid handling.

Clinical and ethical consequences in 2026:

• It has been proposed that APOE ε4/ε4 homozygosity represents a distinct genetic form of AD, with near-complete biomarker penetrance by the age of 65 (Fortea et al., 2024) — a biologically defensible but clinically contested claim, because penetrance for dementia itself is not complete.
• Genotyping is now mandatory practice before anti-amyloid antibody therapy, because ARIA risk stratifies by genotype.
• The effect size is ancestry-dependent: markedly weaker in populations of African and Hispanic ancestry, intermediate in East Asian ancestry. This is the principal limitation of risk models derived from European-ancestry data.
• Predictive testing in asymptomatic individuals is not recommended outside research or a structured counselling context.

FTD and related genes

TREM2 and microglial genetics

Rare TREM2 variants (R47H, R62H, D87N) carry odds ratios comparable to a single APOE ε4 allele (2–4×). TREM2 is the master regulator of the DAM programme; loss of function impairs microglial plaque compaction, lipid sensing and survival signalling. High levels of soluble TREM2 in CSF are associated with slower decline — supporting the "protective microglial response" model. Yet the direct test of that model, the phase 2 INVOKE-2 trial of the TREM2 agonist AL002, failed to show benefit and the programme was terminated.

The polygenic architecture

The largest AD GWAS meta-analyses have identified more than 75 genome-wide significant loci (Bellenguez et al., 2022); these cluster in coherent biological pathways: amyloid processing (APP, ADAM10, SORL1, BIN1), tau biology (MAPT, FERMT2), lipid metabolism (APOE, CLU, ABCA7), innate immunity and microglia (TREM2, CD33, MS4A, CR1, PLCG2), endocytosis and vesicle traffic (BIN1, PICALM, CD2AP).

Polygenic risk scores (PRS): AUC ~0.70–0.75 with APOE included; ~0.60–0.65 with it excluded. Clinical utility is limited: PRSs transfer poorly across ancestries (they are overwhelmingly derived from European-ancestry cohorts), add only modest discrimination over age and APOE, and yield no actionable result. Their most defensible near-term use is trial enrichment, not individual risk communication. GRADE: Low for clinical utility.

Genetic testing: a practical framework

Modifiable risk factors: the 2024 Lancet Commission framework

The 2024 Lancet Commission expanded its model to 14 modifiable risk factors and reported that these account for roughly 45% of the population attributable fraction (PAF) of dementia worldwide (Livingston et al., 2024). Two factors were added to the twelve on the 2020 list: high LDL cholesterol and untreated vision loss.

Three caveats the PAF framework requires — omit them and the table becomes misleading:

1. PAFs are calculated on the assumptions of causality and of complete elimination of the risk factor. Neither is realistic, and causality has not been established for several of the factors.
2. PAFs are not in fact additive; the factors are heavily correlated (obesity–diabetes–hypertension–inactivity form a single cluster). Reading the 45% as "tick 14 boxes and your risk falls by 45%" is wrong.
3. Reverse causality is a live threat for depression, hearing loss, social isolation and physical inactivity — all of them may be prodromal features of the disease they appear to cause.

A closer look at the leading factors

Hypertension. Midlife hypertension (systolic >130–140 mmHg between the ages of 45 and 65) is consistently associated with later dementia; late-life blood pressure shows a U-shaped or even inverse relationship. SPRINT MIND (Williamson et al., 2019) is the strongest interventional evidence in the whole prevention literature: intensive systolic targeting (<120 vs <140 mmHg) significantly reduced new MCI (HR 0.81) and the composite of MCI-or-probable-dementia (HR 0.85); the dementia-alone endpoint did not reach significance — most likely because the trial was stopped early for cardiovascular benefit, with the consequent loss of power. GRADE: Moderate–High.

Diabetes. Type 2 diabetes carries a risk of roughly 1.5–2×. Yet glycaemic control trials have not demonstrated a reduction in dementia risk, and the definitive test of GLP-1 receptor agonism in established early AD (evoke/evoke+) failed. Observational signals for metformin, GLP-1RAs and SGLT2 inhibitors are compromised by confounding by indication and immortal time bias.

Obesity. Midlife obesity increases risk (~1.6×); in later life a low BMI is associated with higher dementia risk — a textbook reverse-causality trap that reflects prodromal weight loss.

Alcohol. Heavy consumption increases risk and causes a distinct alcohol-related brain injury. The historical protective effect attributed to light-to-moderate drinking has been largely refuted by Mendelian randomisation studies and by recognition of the "sick quitter" bias. The defensible current position: no level of alcohol has been proven neuroprotective.

Physical inactivity. Robust observationally (RR ~1.4). But the definitive RCT in people who already have dementia — DAPA (Lamb et al., 2018) — found that a structured exercise programme improved fitness but not cognition, with a small worsening on ADAS-Cog. Exercise is supported as a primary prevention intervention (and at every stage for mood, falls and function); not as a treatment for established dementia.

Nutrition. The Mediterranean and MIND diets are observationally associated with slower cognitive decline. But the 3-year RCT of the MIND diet against mild calorie restriction (Barnes et al., 2023, NEJM) found no significant between-group difference — both groups improved. Nutritional supplements (vitamin E, B vitamins, omega-3, ginkgo) have failed consistently in adequately powered RCTs. GRADE: Low–Moderate.

Sleep. Both short (<6 hours) and long (>9 hours) sleep duration are associated with dementia; obstructive sleep apnoea is an independent and treatable risk factor. RBD is not a risk factor but a marker of prodromal synucleinopathy — a critical distinction.

Hearing loss. One of the single largest PAFs (7%). ACHIEVE (Lin et al., 2023) is the pivotal trial: hearing intervention did not slow cognitive decline in the randomised population as a whole, but produced a 48% reduction in 3-year cognitive decline in the higher-risk ARIC subcohort. The interpretation is contested — a real effect in high-risk individuals, or a subgroup artefact? Because hearing aids are safe, inexpensive and independently beneficial, the practical recommendation is unchanged. GRADE: Low for dementia prevention; High for quality of life.

Air pollution. PM2.5 exposure shows a consistent dose–response relationship across cohorts; natural experiments in air quality improvement support causality. GRADE: Moderate. In Turkey this is a policy lever rather than an individual one.

Education and cognitive reserve. Higher education delays clinical expression without altering the underlying pathology — the "reserve" phenomenon. More educated individuals tolerate a greater pathological burden before crossing the clinical threshold, and then decline faster once they cross it.

Traumatic brain injury and sport. The relationship is dose-dependent. In the first large study of its kind, presented at AAIC 2026, 142 retired professional footballers aged 30–60 were compared with 56 healthy controls: depression and anxiety were significantly more common among former players (clinically significant depression 31% versus 9%), and brain scans showed lower volume in regions important for memory, attention and emotional regulation. The same meeting heard that even in amateur play, heading was associated with short-lived rises in markers of brain cell injury.

Emerging and contested factors. Herpes simplex type 1 and varicella zoster — including observational data that the shingles vaccine reduces dementia incidence, among them the Welsh natural experiment exploiting an age-based eligibility threshold (Eyting et al., 2025, Nature); periodontal disease; the gut microbiome; proton pump inhibitors and anticholinergic burden (this association is robust and clinically actionable); benzodiazepines; general anaesthesia; and the timing of hormone replacement therapy.

The Clinical Picture: What Is Impaired, and How

Cognitive domains

Memory — it is the pattern, not the presence, that matters:

• Amnestic/hippocampal type (AD, LATE): rapid forgetting, impaired delayed recall, no benefit from cueing or recognition — a storage failure.
• Retrieval/frontosubcortical type (VaD, PDD, bvFTD, depression): impaired free recall but marked improvement with cueing and recognition — a retrieval failure.
• The most specific instrument designed to make this distinction is the Free and Cued Selective Reminding Test (FCSRT).

Executive function. Planning, set-shifting, inhibition, abstraction, working memory. Earliest and most prominent in VaD, bvFTD, PDD/DLB and dysexecutive AD. Instruments: Trail Making B, Stroop, verbal fluency, digit span backwards.

Language. Anomia is almost universal and non-specific. The discriminating features are: fluency (reduced in nfvPPA), repetition (impaired in lvPPA), single-word comprehension (impaired in svPPA), grammar (impaired in nfvPPA) and motor speech (apraxia of speech in nfvPPA).

Visuospatial function. Early and prominent in DLB and PCA. In PCA, look specifically for simultanagnosia, optic ataxia and oculomotor apraxia (Balint's syndrome) together with the Gerstmann tetrad.

Social cognition. Impaired early in bvFTD: loss of empathy, failure to detect social faux pas, deficits in emotion recognition. It is formally recognised as a cognitive domain in DSM-5 and is routinely not tested.

Behavioural and psychological symptoms of dementia (BPSD)

These affect more than 90% of patients over the course of the illness. They are the dominant determinant of caregiver burden, institutionalisation and healthcare cost — more so than cognitive severity itself.

Staging and course

• Global Deterioration Scale (GDS) — 7 stages.
• FAST — a parallel 7-stage functional scale; FAST 7a is a commonly used hospice eligibility threshold.
• Clinical Dementia Rating (CDR) — global 0/0.5/1/2/3; the CDR Sum of Boxes (CDR-SB, 0–18) is today the standard primary endpoint in trials. Estimates of the minimum clinically important difference in early AD range from about 0.5 to about 1.6 points depending on the derivation method — and that range directly determines whether current anti-amyloid effect sizes count as clinically meaningful.

The typical AD course: a decline of about 2–4 MMSE points per year in the moderate stage; a preclinical phase of 15–20 years; an MCI phase of 3–7 years; a dementia phase of 4–10 years. A decline faster than 6 points per year requires the diagnosis to be revisited: prion disease, autoimmune encephalitis, mixed pathology or unrecognised delirium.

Diagnosis: From Syndrome to Biology

The diagnostic algorithm

The work-up proceeds as follows:

1. Begin with a cognitive concern raised by the patient, an informant or a clinician.
2. Take a history: including an informant, a functional assessment, a medication review and a three-generation family history.
3. If the picture is acute, fluctuating and attention-dominant, assess and treat delirium first; re-measure cognition after it resolves.
4. Perform cognitive screening — MoCA, MMSE, Mini-Cog or RUDAS.
5. Request basic laboratory tests: full blood count, urea and electrolytes, liver function tests, TSH, B12, folate, calcium, glucose/HbA1c, lipids; syphilis and HIV serology where indicated.
6. Perform structural imaging; MRI is preferred, CT if MRI is contraindicated.
7. If red flags are present (onset under 65, rapid progression, focal signs, seizures, atypical features), proceed to specialist investigation: CSF examination (including RT-QuIC), autoimmune panel, EEG, genetic testing, FDG-PET.
8. Determine the syndrome from the neuropsychological profile. In amnestic, visuospatial or language-predominant presentations, proceed to Alzheimer's biomarkers — first-line plasma p-tau217 where available. In a dysexecutive or behavioural presentation, DaTscan, MIBG scintigraphy, polysomnography and an FTD gene panel come to the fore.
9. If the biomarker is positive, Alzheimer's disease is accepted; stage the disease and assess eligibility for anti-amyloid treatment. If the result is indeterminate, confirm with CSF or amyloid PET. If negative, consider non-Alzheimer pathology — LATE, vascular disease, Lewy body disease, FTLD or a reversible cause.

Clinical assessment

The informant history is the highest-yield diagnostic instrument in the entire work-up — higher than any biomarker. It should clarify: onset (insidious or abrupt), tempo, the first symptom (which usually indicates the network of origin), the functional trajectory (instrumental activities of daily living decline before basic ones), behavioural change, hallucinations, RBD (ask the bed partner directly: "does he act out his dreams, shout or thrash in his sleep?"), falls, autonomic symptoms, and a complete medication review with particular attention to anticholinergic burden.

Cognitive assessment instruments

Structural imaging

MRI (preferred). Required sequences: 3D T1 (volumetry), T2/FLAIR (white matter, infarcts), DWI (prion disease, acute infarct), SWI or GRE (microbleeds, superficial siderosis — mandatory before anti-amyloid treatment) and coronal oblique T1 (medial temporal lobe).

Visual rating scales: Scheltens MTA (0–4, with age-adjusted thresholds), Koedam posterior atrophy (0–3), Fazekas (white matter hyperintensity, 0–3), global cortical atrophy (0–3). Quantitative volumetry (FreeSurfer and CE/FDA-cleared commercial tools) adds normative percentile reporting.

CT is acceptable where MRI is unavailable or contraindicated; it excludes tumour, subdural haematoma and hydrocephalus but is markedly inadequate for medial temporal atrophy, small vessel disease and microbleeds.

Molecular and functional imaging

The appropriate use criteria for amyloid PET restrict it to situations where the result will change management: unexplained persistent MCI, atypical or mixed presentations, young-onset dementia, and determination of eligibility for anti-amyloid treatment. It is not appropriate in asymptomatic individuals, for APOE-based risk assessment, or for grading severity.

CSF biomarkers

Blood biomarkers — the central advance of the decade

Regulatory and implementation milestones:

• 16 May 2025 — the FDA approved the Lumipulse G pTau217/β-Amyloid 1-42 Plasma Ratio test (Fujirebio) for adults over 55 with signs and symptoms of cognitive decline, for use in specialised care settings. Validation: 499 samples; 91.7% of positives were confirmed by amyloid PET or CSF; 97.3% of negatives were confirmed negative; the indeterminate zone was under 20%. It is not a screening test and not a stand-alone diagnostic.
• December 2025 — a second FDA clearance was granted to a plasma p-tau181 test positioned as a rule-out test for primary care (specificity ~98%, but a positive predictive value of only ~22% in primary care — a design that can only be interpreted if its asymmetry is understood).
• July 2026 (AAIC, London) — a Swedish real-world study of 1,310 patients and 165 physicians: family physicians were 65% accurate without the blood test result and 93% accurate with it; specialists were 76% and 94% respectively. The test changed the diagnosis or the care plan in more than half of patients, and was most useful in primary care for ruling Alzheimer's out.
• July 2026 — a prognostic analysis published simultaneously in JAMA: roughly 2,700 cognitively healthy older adults followed for up to 10 years. Among those with very high p-tau217 levels, the estimated probability of developing cognitive impairment was 78% within 10 years and 33% within 5 years (Buckley et al., 2026).

Implementation caveats — omit these and the test does harm:

• Performance degrades in chronic kidney disease, high BMI and certain comorbidities.
• The assays are not interchangeable; each requires its own validated thresholds.
• The most defensible clinical algorithm is a two-threshold one: definitely positive / definitely negative / intermediate zone — with CSF or PET confirmation for the intermediate zone.
• Screening asymptomatic populations is not supported. At low prevalence the positive predictive value collapses, and there is no treatment that can be offered outside a trial to an asymptomatic individual who tests positive. The Alzheimer's Association's own guidance states that blood tests should be ordered only for patients who are experiencing memory or thinking problems.

GRADE: High certainty for diagnostic accuracy in symptomatic specialist populations; Moderate for primary care; Low for asymptomatic screening (and there the evidence does not support use).

Differential diagnosis: discriminating patterns

Current Treatments

Cholinesterase inhibitors

Efficacy (Cochrane meta-analyses): approximately 2.4–2.7 points of improvement on ADAS-Cog against placebo at 6 months (on a 70-point scale), ~1.4 points on MMSE; small but consistent benefit in global impression, function and behaviour. The number needed to treat for global improvement is roughly 10–12. GRADE: High certainty for a small effect. Oxford Level 1a.

Practical points: titrate slowly; assess response at month 3; the absence of dramatic improvement is not treatment failure — stabilisation counts too; do not stop abruptly (the DOMINO-AD trial showed significant cognitive worsening on withdrawal); check the pulse at baseline and thereafter, and consider an ECG if there is cardiac history; and watch for this prescribing cascade: treating ChEI-induced urinary urgency with an anticholinergic directly erases the effect of the ChEI.

Memantine

A non-competitive, moderate-affinity, fast-off NMDA receptor antagonist; it reduces pathological tonic glutamatergic activation while preserving physiological phasic signalling.

• Indication: moderate-to-severe AD (MMSE <20). The evidence in mild AD is negative and it should not be used there.
• Dosing: 5 mg/day, increased weekly to 10 mg twice daily; renal dose adjustment is required.
• Efficacy: small benefit in cognition, function and behaviour; particularly useful for agitation and aggression in moderate-to-severe disease. GRADE: Moderate.
• Non-AD dementias: not recommended in MCI, vascular dementia or FTD (negative trials).

Anti-amyloid monoclonal antibodies

ARIA — the class-defining toxicity. ARIA-E (vasogenic oedema/effusion) and ARIA-H (microbleeds/superficial siderosis) arise from mobilisation of vascular amyloid on a background of cerebral amyloid angiopathy. Most cases are asymptomatic and detected radiologically; some cause headache, confusion, visual change or seizures; rare cases — particularly with concurrent anticoagulation or thrombolysis — are fatal.

Risk stratification:

• APOE ε4/ε4 homozygotes: highest risk (in TRAILBLAZER-ALZ 2, donanemab ARIA-E occurred in ~40% of homozygotes versus ~16% of non-carriers). The EU restricted lecanemab's indication accordingly.
• More than 4 microbleeds at baseline, cortical superficial siderosis or suspected CAA: exclude, or proceed with extreme caution.
• Anticoagulation: a relative contraindication.
• Monitoring: MRI before doses 5, 7 and 14 for lecanemab; before doses 2, 3, 4 and 7 for donanemab; and always whenever symptoms occur.
• Patients should carry a wallet card: thrombolysis for stroke in a patient on anti-amyloid therapy is potentially catastrophic.

Critical appraisal. These agents cleared a real bar: prespecified primary endpoints were met, target engagement was confirmed by biomarker, a dose–response was demonstrated, and the result replicated across two independent molecules. This is the strongest evidence the amyloid hypothesis has ever received. The case against is equally serious:

1. effects of 0.45–0.7 CDR-SB points sit below or at the edge of most MCID estimates;
2. functional unblinding via ARIA and infusion reactions may have inflated the observed effect;
3. trial populations were younger, healthier, less comorbid and markedly less ethnically diverse than the real outpatient population;
4. 18-month trials cannot determine whether treatment and placebo curves subsequently diverge further, run parallel or converge;
5. the accelerated brain volume loss seen on treatment remains incompletely explained (amyloid removal, or a form of injury?);
6. the infrastructure demands — biomarker confirmation, serial MRI, infusion/injection capacity, ARIA expertise — restrict access sharply and inequitably.

GRADE: Moderate certainty for a small clinical benefit; High for amyloid removal; Moderate for ARIA harm.

There is also real-world data: the LEADER study presented at AAIC 2026 reported that more than 75% of early-AD patients receiving lecanemab for a mean of 17 months remained stable, and about 7% improved. Registry data of this kind lack a control group and are open to selection bias; they do not substitute for randomised evidence, but they are informative about feasibility.

Pharmacological management of BPSD

First principle: always non-pharmacological first. The DICE approach (Describe–Investigate–Create–Evaluate) is the best-validated structured framework. Delirium, pain, infection, constipation, retention, drug effects and sensory deprivation must be excluded before the behaviour is attributed to the dementia.

Non-pharmacological interventions

The care model matters more than the drugs

Effective dementia care is less a drug problem than a care-model problem. The core components: a named care coordinator; timely diagnosis and structured information; advance care planning while capacity is intact; driving assessment; financial and legal protection (power of attorney); nutrition and swallowing assessment as the disease advances; falls prevention; continence management; education, respite and mental health support for the caregiver; and a documented plan for crisis, hospital admission and the end of life.

Palliative care principles should be introduced early: advanced dementia is a terminal illness with a course comparable to metastatic cancer. In advanced dementia, tube feeding does not improve survival, aspiration risk, pressure sores or comfort — the evidence-based alternative is careful hand feeding.

Subtype-Specific Treatment Approaches

Alzheimer's disease — a stage-based framework

Vascular cognitive impairment

There is no licensed pharmacotherapy for vascular dementia. Management is aggressive secondary prevention:

• Blood pressure control — the only intervention with RCT support for reducing new cognitive impairment (SPRINT MIND). A systolic target below 130 mmHg in those who tolerate it; individualise in the frail and the very old.
• Antithrombotics according to standard guidelines for ischaemic stroke and TIA; in the absence of a vascular indication there is no evidence of cognitive benefit from antiplatelet agents.
• Statins for cardiovascular indications. Statin trials for the prevention or treatment of dementia are negative; the Lancet Commission's inclusion of LDL as a risk factor reflects epidemiology and Mendelian randomisation, not positive intervention trials.
• ChEIs and memantine: small, inconsistent cognitive effects and no clear functional benefit in VaD trials. Reasonable in mixed AD-VaD; not routinely indicated in pure VaD.

Dementia with Lewy bodies

Frontotemporal dementia

There is no approved disease-modifying or symptomatic treatment.

• ChEIs are not indicated and may worsen behaviour in bvFTD. Memantine trials are negative.
• SSRIs (sertraline, citalopram, fluvoxamine, paroxetine) and trazodone have modest evidence for disinhibition, compulsive behaviour and hyperorality.
• Non-pharmacological management is essential: structured routine, environmental modification, behavioural analysis, speech and language therapy for PPA, safety planning (the risk of financial exploitation is very high in bvFTD) and cessation of driving.
• Genetic counselling is central, given the high familial rate.

A negative result worth knowing: Alector's phase 3 INFRONT-3 trial (October 2025) failed its clinical co-primary endpoint of slowing progression on the CDR plus NACC-FTLD in FTD-GRN, despite latozinemab successfully raising progranulin. This is the first phase 3 read-out of a mechanistically clean target-engagement strategy in genetic FTD, and a serious blow to the "replace the missing protein" model.

Parkinson's disease dementia

• Rivastigmine is licensed (the EXPRESS trial; NNT ~10). Oral or patch.
• Simplify dopaminergic therapy: withdraw anticholinergics, then amantadine, then dopamine agonists, then MAO-B and COMT inhibitors, in that order; reduce levodopa last.
• Psychosis: pimavanserin, quetiapine or clozapine. Avoid other antipsychotics.

Rapidly progressive and rare dementias

• Prion disease: no disease-modifying treatment. Supportive and palliative care; clonazepam or levetiracetam for myoclonus. ASO approaches targeting PRNP (for example ION717) are in early clinical development — and with the possibility of presymptomatic treatment in known mutation carriers, they represent the strongest future option in the field.
• Autoimmune encephalitis: urgent immunotherapy (high-dose corticosteroids, IVIg, plasma exchange; rituximab or cyclophosphamide second line) and tumour screening. Early treatment is the dominant determinant of outcome — this is the most outcome-changing diagnosis in the entire differential.
• NPH: ventriculoperitoneal shunt with a programmable valve in selected patients. GRADE: Moderate for gait, Low for cognition.
• Wilson's disease, Wernicke's encephalopathy, neurosyphilis, HIV: specific treatment; genuinely reversible if caught early.

Emerging Treatments and the Clinical Trial Landscape

Next-generation anti-amyloid agents

Trontinemab (Roche) is the most important near-term development. It is a 2+1 bispecific antibody ("Brainshuttle") that combines gantenerumab's binding site with a Fab that binds the transferrin receptor; by receptor-mediated transcytosis it enters the brain at doses an order of magnitude lower than conventional antibodies. What was observed in the phase Ib/IIa Brainshuttle AD trial:

• At the 3.6 mg/kg dose, 91% of participants were amyloid-PET negative at 28 weeks — faster and deeper clearance than any approved agent; 72% fell below 11 centiloids.
• Plaque clearance in deep brain regions that conventional antibodies reach poorly.
• ARIA-E in fewer than 5% of participants — compare 13% for lecanemab and 24% for donanemab. The mechanistic explanation: capillary-mediated entry bypasses the amyloid-laden penetrating arteries from which ARIA arises.
• One safety signal to watch: 10% anaemia in the 3.6 mg/kg cohort.

The development programme: TRONTIER 1 and 2 (phase 3, ~1,600 participants, early symptomatic AD, begun in September 2025) and PrevenTRON, announced at AAIC 2026 (phase 3, ~1,600 cognitively normal individuals aged 55–80, CDR 0, at high risk as screened by plasma p-tau217; the primary endpoint is time to clinical progression). If the ARIA advantage holds at scale, this class removes the principal obstacle both to treating APOE ε4 homozygotes and to preventive dosing.

Remternetug (Eli Lilly), donanemab's N3pG-Aβ successor, is in phase 3 TRAILRUNNER-ALZ 1 with subcutaneous self-administration.

An abandoned approach: BACE1 inhibition. Verubecestat, atabecestat, lanabecestat and umibecestat all failed with cognitive worsening — a definitive negative result for the "reduce production" strategy in symptomatic and at-risk populations.

Tau — the second wave

The tau field today stands roughly where amyloid stood in 2016: clear target engagement, an encouraging direction, and no positive pivotal trial yet. The critical open question is this: does tau lowering have to be done before appreciable tangle accumulation in order to work?

α-Synuclein

Prasinezumab missed the primary endpoint in phase 2 PASADENA but showed signals in motor progression in prespecified subgroups; phase 2b PADOVA also fell short of formal significance. Sinpanemab failed in SPARK. No α-synuclein therapy has been tested in a way that would reach a positive endpoint in DLB. Yet the accuracy of α-synuclein SAA now makes it possible for the first time to enrich DLB trials biologically — potentially one of the most important structural changes of the next five years.

Gene therapy and genome editing

CRISPR and genome editing are entirely preclinical for dementia. The persuasive avenues: APOE4→APOE3/2 base editing, selective disruption of the APP Swedish mutation, allele-specific PSEN1 editing, C9orf72 repeat excision, and the insertion of the protective APP A673T variant by base editing. The obstacles are severe and unsolved: delivery reaching a sufficient fraction of neurons across the whole brain, off-target editing in non-dividing post-mitotic cells, immune responses to Cas proteins and AAV capsids, irreversibility, and the ethics of intervening decades in advance in a person who may never fall ill.

Cell-based therapies — an honest appraisal

No stem cell therapy has restored cognitive function in a controlled human study in any dementia. The mechanistic plausibility of MSCs is immunomodulatory, not regenerative — and that point is routinely distorted by unregulated clinics selling "stem cell treatment for Alzheimer's". This is currently a form of exploitation and patients should be actively warned. GRADE: Very Low for efficacy; Moderate for short-term safety.

RNA therapeutics and targeted protein degradation

Antisense oligonucleotides (ASOs) are the most clinically advanced non-antibody modality in neurodegeneration; they have been validated by nusinersen in SMA and tofersen in SOD1-ALS. In dementia: diranersen (MAPT), NIO752 (MAPT), tominersen (HTT), ION717 (PRNP). BIIB078 and WVE-004, both targeting C9orf72, were discontinued for lack of efficacy.

PROTACs (proteolysis-targeting chimeras) and molecular glues offer catalytic, sub-stoichiometric degradation of aggregation-prone proteins. Tau-directed PROTACs can degrade pathological tau in patient-derived neurons. Their advantages over antibodies: intracellular access, catalytic efficiency, and the potential for oral bioavailability. Status: preclinical.

Blood–brain barrier delivery platforms

This is the highest-leverage technical problem in the field, because it multiplies the value of every other treatment modality.

• Receptor-mediated transcytosis: the transferrin receptor (Roche Brainshuttle — clinically validated with trontinemab; Denali's TransportVehicle platform), CD98hc, the insulin receptor.
• Microbubble-assisted focused ultrasound (FUS): transient, targeted, reversible opening of the blood–brain barrier. Clinical trials have demonstrated safe opening of the hippocampus and entorhinal cortex in AD patients; amyloid reduction with FUS alone and increased antibody delivery in combination with aducanumab have been reported (Rezai et al., 2024, NEJM). Currently the most convincing physical delivery method; the open questions are durability, scalability and the long-term safety of repeated opening.
• Nanomedicine: an extensive preclinical literature and almost no clinical translation in dementia. The gap between publication volume and clinical impact in this subfield is unusually wide and should be read as a warning sign.

Neuromodulation and device-based treatments

Inflammation, microglia and metabolic targets

• TREM2 agonism: the phase 2 INVOKE-2 trial of AL002 failed to show benefit and the programme was halted.
• NLRP3 inflammasome inhibitors: multiple molecules in phase 1/2; a strong preclinical rationale links ASC specks to Aβ seeding.
• NSAIDs: definitively negative (the ADAPT trial suggested possible harm in some analyses). Do not use them.
• GLP-1 receptor agonists: the evoke/evoke+ programme is the defining result — 3,808 participants, 104 weeks, oral semaglutide 14 mg. No difference on CDR-SB (evoke: −0.08 [95% CI −0.35 to 0.20], p=0.57; evoke+: +0.10 [−0.17 to 0.38], p=0.46); no difference on functional endpoints — and yet there were reductions of up to 10% in CSF p-tau181, p-tau217, total tau, neurogranin and YKL-40, and a significant fall in plasma hs-CRP. Novo Nordisk cancelled the planned one-year extension. The trials were published in the Lancet in 2026.

• Antivirals: valacyclovir in HSV-positive mild AD, together with the epidemiological signal around the shingles vaccine, keeps the infection hypothesis alive. An adequately powered vaccination trial would be among the highest-value studies the field could run.
• Sodium oligomannate (GV-971): approved in China on the basis of a 36-week trial; its gut-microbiome mechanism and trial design were widely criticised internationally, and the confirmatory global trial did not deliver a result.

Combination and precision approaches

The strongest strategic consensus of 2025–2026 is that monotherapy will not suffice. Anti-amyloid agents deliver about 30% slowing; the remaining 70% requires complementary mechanisms. The precedent is drawn explicitly from oncology and HIV. Platform trials designed for this purpose: DIAN-TU-002 (adaptive, multi-arm), the ACTC infrastructure, EPAD. Likely first combinations: anti-amyloid plus anti-tau; anti-amyloid plus anti-inflammatory; and biomarker-triggered sequential therapy (clear amyloid, then suppress tau while maintaining on a low-burden agent).

The key trials to watch

Why do trials fail? A structural analysis

Prevention: The Most Powerful Tool We Have Today

The evidence hierarchy of multidomain prevention trials

The World-Wide FINGERS network is now adapting the FINGER protocol in more than 40 countries. LatAm-FINGERS's positive result in 2026 is the first large randomised evidence that the approach works outside Western Europe and North America — and it is directly relevant to rapidly ageing upper-middle-income countries such as Turkey.

A practical prevention prescription

The Questions Patients Actually Ask

Can dementia be cured?

No — not for any neurodegenerative cause. A "cure" means eliminating the pathology and restoring function. Neither has been achieved in any human being for AD, DLB, FTD, PDD or prion disease.

The qualified exception: dementia syndromes due to treatable conditions — B12 deficiency, hypothyroidism, neurosyphilis, autoimmune encephalitis, chronic subdural haematoma, NPH, drug toxicity, severe sleep apnoea — can improve markedly, particularly when treated early, and rarely resolve completely. These account for roughly 1–5% of presentations; and even here, complete reversal is uncommon if dementia-level impairment has been present for months. This is precisely why investigating for reversible causes is mandatory rather than optional.

Can progression be halted?

Not at present. It can be slowed by roughly a quarter to a third over a limited period of observation. Open-label extension data suggest that treatment and placebo trajectories continue to diverge; that is encouraging, but open-label extensions are vulnerable to survivorship bias and lack a concurrent control group.

The intellectually honest formulation is this: halting the disease is a coherent goal, not a fantasy — but it will require either much earlier intervention (before appreciable tau spread and neuronal loss) or combination therapy addressing several pathways at once. The preclinical-stage trials (AHEAD 3-45, TRAILBLAZER-ALZ 3, PrevenTRON) are a direct test of the "is earlier enough?" hypothesis, and their read-outs between 2027 and the early 2030s will set the direction of the field.

Can dementia be reversed?

Not for established neurodegeneration. Two nuances matter.

First, cognitive fluctuation is not reversal. Patients with DLB, or patients with superimposed delirium, depression or drug toxicity, can improve markedly once the reversible layer is removed; this is frequently misread as reversal of the disease.

Second, some biomarker abnormalities genuinely are reversible: amyloid plaque can be brought below the positivity threshold, and CSF and plasma tau can be markedly lowered. Reversing the biology is now demonstrably possible; reversing the clinical syndrome is not. The gap between those two facts is the central scientific problem of the field.

Can neurons be regenerated?

Only marginally in the adult human brain — and not at a scale that would be relevant to dementia.

• Adult hippocampal neurogenesis has been reported in the dentate gyrus into the ninth decade (Boldrini et al., 2018; Moreno-Jiménez et al., 2019) and has been contested in parallel by studies finding negligible neurogenesis after childhood (Sorrells et al., 2018). The best current synthesis: adult human hippocampal neurogenesis is real but low-rate, declines with age, and declines further in AD.
• Even the most generous estimates speak of a few hundred to a few thousand new granule cells per day in a single hippocampal subregion. Established AD involves the loss of billions of neurons and far more synapses.
• There is no neurogenesis in the neocortex, entorhinal cortex or basal forebrain — that is, in none of the regions that generate the dementia syndrome.

The conclusion: endogenous regeneration is not a plausible therapeutic route for repairing a demented brain. More realistic regenerative targets are synaptic (restoring function in surviving neurons) and protective (preventing further loss).

Can stem cells restore cognitive function?

There is no human evidence supporting that claim. Phase 1/2 trials of MSCs in AD have shown acceptable short-term safety and occasional biomarker or exploratory cognitive signals; none has progressed to convincing controlled efficacy. The mechanistic rationale for MSCs is paracrine immunomodulation and trophic support, not cell replacement — MSCs do not become neurons.

The Parkinson's comparison is illuminating: iPSC- and ESC-derived dopaminergic neuron grafts have entered clinical trials because PD involves the loss of a relatively homogeneous single cell population projecting to a single target structure. AD involves the loss of a heterogeneous, long-range, precisely wired cortical circuit that carries decades of learned information. Replacing that is not the same class of problem, and no existing technology comes close to it.

Can gene editing prevent neurodegeneration?

Probably, eventually, in monogenic disease — but in nobody yet. The strongest rationale is presymptomatic intervention in highly penetrant monogenic disorders where the causal variant is known, the natural history is predictable and the at-risk individual can be identified decades in advance: PSEN1/APP autosomal dominant AD, GRN and C9orf72 FTD, PRNP prion disease and Huntington's disease.

For sporadic late-onset AD, gene editing is not a coherent strategy: there is no single causal variant to correct.

What actually distinguishes Alzheimer's from the other dementias?

Four separate answers:

1. Molecularly: AD uniquely requires both Aβ plaques and neurofibrillary tau tangles. Tau without amyloid is PART; amyloid without tau is not AD.
2. Clinically: AD typically begins with hippocampal-type amnesia — a delayed-recall deficit that does not improve with cueing; VaD, PDD and bvFTD begin with retrieval-type memory failure that does improve with cueing, and with executive or behavioural primacy.
3. Diagnostically: AD is the only dementia with accessible, validated in vivo molecular biomarkers in blood, CSF and on PET. This asymmetry distorts the whole field: AD is disproportionately studied and treated because it is disproportionately measurable.
4. Therapeutically: AD is the only dementia with an approved disease-modifying treatment.

A caveat that disrupts all four: over the age of 80, mixed pathology is the norm. In practice, "Alzheimer's disease" in an 85-year-old usually means "Alzheimer's pathology plus at least one more".

What is the real opportunity in early diagnosis?

The blood biomarker revolution creates four concrete opportunities:

1. Closing the diagnostic gap. Globally, between half and 90% of dementia is undiagnosed. A 50–200 dollar blood test usable in primary care is the first technology that could change that.
2. Treatment eligibility at scale. Anti-amyloid treatment requires biomarker confirmation; PET and lumbar puncture capacity cannot serve the eligible population. The blood test is the only feasible triage.
3. Prognosis before symptoms. Plasma p-tau217 predicts progression to impairment 5–10 years in advance in cognitively normal adults — making prevention trials possible in precisely the population that stands to benefit most.
4. Trial efficiency. Screen failure rates in amyloid-confirmed trials were running at 55–68%; blood pre-screening cuts the cost sharply.

The counter-risks are serious and under-discussed: false positives in low-prevalence populations; the psychological and insurance consequences of a positive result in an asymptomatic person; direct-to-consumer commercial testing outrunning clinical infrastructure; and the creation of a large "biomarker-positive, symptom-free" population for whom no approved treatment exists. Screening asymptomatic people outside research cannot currently be justified.

What will personalised treatment realistically look like?

A defensible projection for around 2030:

• Risk stratification in midlife — using APOE, a polygenic score and the modifiable risk profile → prevention matched in intensity to risk.
• Blood biomarker surveillance from age 55–60 in high-risk individuals.
• Biomarker-triggered intervention at the preclinical stage — conditional on the ongoing prevention trials reading out positive.
• Pathology-matched treatment: amyloid-directed in the amyloid-positive; tau-directed according to tau PET stage; α-synuclein-directed in the SAA-positive patient; gene-specific therapy in monogenic disease.
• Genotype-matched safety selection: brain-shuttle antibodies in APOE ε4/ε4 homozygotes because of the ARIA profile.
• Biomarker-defined stopping and restarting rules — rather than indefinite dosing.

Ranking the Evidence (a GRADE-Style Synthesis)

The field's two recurring failure patterns are identical to those in the longevity literature: (1) the species gap — impressive in the transgenic mouse, absent in humans; (2) the surrogate–outcome gap — the biomarker improves, function does not (evoke/evoke+ is the textbook example).

Controversies, Limitations and Bias

Live controversies

Limitations

• The effect sizes are small. A CDR-SB difference of 0.45–0.7 is below or at the edge of most MCID estimates. Whether patients and families experience this as meaningful is genuinely unresolved, and it is not answered by statistical significance.
• Short trial durations cannot characterise a 20-year disease.
• Unrepresentative recruitment. Most pivotal AD trials enrolled fewer than 5% Black and Hispanic participants; multimorbid, anticoagulated and very old patients — that is, the real outpatient population — were systematically excluded.
• The MCID problem is unsolved. Different derivation methods yield estimates that differ by up to threefold, and sponsors reliably cite the most favourable one.
• Publication and reporting bias. Negative trials are published later, less visibly and sometimes not at all; the emphasis on subgroups in press releases routinely outruns the prespecified analysis.
• Autopsy confirmation is rare, so the true diagnostic composition of trial populations is inferred rather than known.
• Mixed pathology, although it is the biological norm, is inadequately modelled in trial design.

Reproducibility

The field has notable replication failures: the effects of 40 Hz gamma stimulation on amyloid have not been robustly reproduced outside the originating laboratory; adult human hippocampal neurogenesis is contested; and a great many single-laboratory preclinical results in transgenic mice have failed to translate to humans.

The structural driver is this: transgenic mouse models overexpressing mutant human APP model familial, plaque-dominant amyloidosis — not sporadic human AD — and they have an almost perfect record of predicting drugs that will fail in humans. Models with better construct validity (knock-in models, human iPSC-derived neurons and organoids, non-human primates) are necessary and under-used.

Bias and conflicts of interest

Industry sponsorship dominates therapeutic trials in this field. Its documented consequences across medicine: selective outcome reporting, favourable framing of subgroup findings, ghostwriting and delayed publication of null results. In dementia specifically, patient advocacy organisations receive substantial industry funding, and several played visible roles in the aducanumab approval controversy.

A reader's checklist for any dementia trial report: was the primary endpoint prespecified and met? Was multiplicity controlled? Can the effect size be interpreted against a declared MCID? Who funded it and who wrote it? Were negative secondary endpoints reported with equal prominence? Does the population represent the intended treatment population? Was blinding maintained, and was it tested?

Research gaps — in order of priority

Clinical Recommendations

Graded by strength of recommendation (Strong/Conditional) and certainty of evidence (GRADE).

Diagnosis

1. Strong / High — Take a structured informant history and functional assessment in every case of suspected cognitive impairment. It outperforms any biomarker.
2. Strong / High — Exclude delirium and reversible contributors before making a diagnosis of dementia: a full medication review with assessment of anticholinergic burden, B12, TSH, calcium and structural imaging.
3. Strong / High — Perform structural imaging in all new diagnoses (MRI preferred).
4. Strong / Moderate — Use a validated cognitive instrument appropriate to the patient's language, literacy and culture. Do not rely on MMSE alone to exclude MCI.
5. Strong / High — Use biomarker confirmation where AD is suspected and the result will change management. In specialist settings, plasma p-tau217-based tests are appropriate first line; confirm indeterminate results with CSF or amyloid PET.
6. Strong / Moderate — Do not screen asymptomatic individuals with AD blood biomarkers outside research.
7. Strong / High — In rapidly progressive dementia, investigate urgently for prion disease (RT-QuIC), autoimmune, neoplastic and infectious causes.
8. Strong / Moderate — Ask every patient's bed partner about dream enactment. RBD radically changes the differential diagnosis.
9. Strong / High — In dementia under 65, take a three-generation family history and consider a genetic panel.
10. Strong / Moderate — Disclose the diagnosis clearly, face to face, with a support person present, with written follow-up and a defined next point of contact.

Treatment

1. Strong / High — Offer a cholinesterase inhibitor in AD, DLB and PDD; frame the expected modest benefit clearly and honestly.
2. Strong / Moderate — Add memantine in moderate-to-severe AD. Do not use it in mild AD or MCI.
3. Strong / Moderate — Do not use cholinesterase inhibitors in bvFTD.
4. Conditional / Moderate — Offer anti-amyloid antibody therapy only to patients with biomarker-confirmed early symptomatic AD who meet trial-equivalent eligibility criteria; after APOE genotyping and a baseline MRI; at a centre with ARIA monitoring capacity; and following a documented shared decision-making conversation that covers the magnitude of the expected benefit.
5. Strong / Moderate — Treat anticoagulation, more than 4 microbleeds or cortical superficial siderosis as a strong caution against anti-amyloid therapy. Provide a wallet card warning against thrombolysis.
6. Strong / High — Use a structured non-pharmacological framework (DICE) as first line in BPSD; exclude pain, infection, constipation, retention and drug effects before escalating.
7. Strong / High — Reserve antipsychotics for severe distress or risk of harm only; use the lowest effective dose, document the indication and attempt withdrawal at least every 3 months.
8. Strong / High — In DLB and PDD, avoid antipsychotics other than quetiapine, clozapine or pimavanserin.
9. Strong / High — Avoid benzodiazepines.
10. Strong / Moderate — Offer cognitive stimulation therapy in mild-to-moderate dementia.
11. Strong / High — Offer a structured caregiver intervention (START-type). This is among the highest-value interventions available and is systematically under-delivered.

Prevention and risk reduction

1. Strong / Moderate-High — Treat midlife hypertension to a systolic target below 130 mmHg where tolerated.
2. Strong / Moderate — Recommend regular aerobic and resistance exercise for prevention, and for function at every stage.
3. Conditional / Low-Moderate — Recommend a Mediterranean or MIND dietary pattern. Do not recommend supplements.
4. Strong / Moderate — Screen for and treat hearing and vision impairment.
5. Strong / High — Support smoking cessation and minimise alcohol.
6. Strong / Moderate — Recognise and treat obstructive sleep apnoea.
7. Strong / Moderate — Actively reduce anticholinergic and sedative medications.
8. Conditional / Low — Encourage vaccination including shingles; state that the dementia signal is observational.
9. Strong / Moderate — Where accessible, offer at-risk older adults a structured multidomain programme (the FINGER/POINTER model).

Care and systems

1. Strong / High — Begin advance care planning at the point of diagnosis, while capacity is intact.
2. Strong / High — Assess driving safety and revisit it at every review.
3. Strong / High — Do not use tube feeding in advanced dementia; provide careful hand feeding.
4. Strong / Moderate — Recognise advanced dementia as a terminal illness and involve palliative care.
5. Strong / High — Provide a named coordinator and a defined crisis plan.

The Outlook

Closing: an honest accounting

The field has crossed a real threshold, and that should be said without overstatement. For the first time a treatment alters the biology of Alzheimer's disease and produces a measurable, replicated clinical effect; for the first time that biology can be detected in a tube of blood drawn in primary care; and for the first time, large randomised trials in diverse populations have shown that a structured lifestyle programme improves cognition in people at risk. These are genuine achievements, and a review that failed to acknowledge them would be as misleading as one that exaggerated them.

But the honest accounting is sobering. The anti-amyloid effect is a slowing of roughly a quarter to a third over eighteen months; it is bought with serious cost, an infrastructure burden and a class toxicity that has killed patients. The second wave — tau — has produced encouraging biology and no positive pivotal trial yet. The most expensive non-amyloid trial ever conducted in this disease moved every biomarker it was supposed to move and changed nothing that patients or families would notice. We have nothing that cures, halts or reverses neurodegeneration, and nothing credible in development that promises to rebuild a brain that has already lost its circuitry.

Three conclusions follow, in order of practical importance.

First, prevention is today the highest expected-value strategy in the field. Forty-five per cent of dementia risk is attributable to modifiable factors. Blood pressure control, hearing correction, physical activity, education, air quality and deprescribing are available now, inexpensive, and radically under-delivered.

Second, in the near term the diagnostic transformation matters more than the therapeutic one. Between half and 90% of dementia worldwide is undiagnosed. A blood test that lets a family physician identify Alzheimer's pathology with near-specialist accuracy changes what is possible at population scale — for access to treatment, for trial recruitment, and above all for the basic dignity of knowing what is happening. That capability arrived between 2025 and 2026 and its consequences have not yet been absorbed.

Third, the therapeutic future is combination and early. Amyloid is the trigger, tau the executioner, and inflammation and vascular damage set the tempo. A single agent given at a single node of a decades-long cascade, after symptoms have appeared, was never a plausible route to a large effect.

Meanwhile, the interventions that most reliably improve the lives of people living with dementia today are not molecular. They are timely diagnosis, honest information, coordinated care, caregiver support, stopping unnecessary drugs, sensory correction and competent palliative care at the end of life. Unglamorous, well evidenced, inexpensive and neglected. A serious response to dementia has to deliver both — the science and the care; and the credibility of the field depends on not confusing progress in one with progress in the other.

References

1. Alzheimer's Association. (2026). Research advances from the 2026 Alzheimer's Association International Conference. https://aaic.alz.org/releases-2026/overview.asp
2. Alzheimer's Association. (2026). Alzheimer's blood test could bring highly accurate diagnosis into everyday clinical care. AAIC 2026. https://aaic.alz.org/releases-2026/alzheimers-blood-test-accurate-diagnosis-pcp.asp
3. Alzheimer's Association. (2026). Blood test may help predict Alzheimer's risk a decade before symptoms appear. AAIC 2026. https://aaic.alz.org/releases-2026/blood-test-p-tau217-predicts-alzheimers-risk.asp
4. Alzheimer's Association. (2026). LiBBY trial: THC/CBD for agitation in late-stage dementia. AAIC 2026. https://aaic.alz.org/releases-2026/libby-trial-thc-cbd-agitation-late-stage-dementia.asp
5. Alzheimer's Association. (2026). LatAm-FINGERS improves brain health in Latin America. AAIC 2026. https://aaic.alz.org/releases-2026/latam-fingers-improves-brain-health-latin-america.asp
6. Alzheimer's Association. (2026). Study finds signs of brain health changes in retired professional soccer players. AAIC 2026. https://aaic.alz.org/releases-2026/brain-health-changes-soccer-football-players.asp
7. American Psychiatric Association. (2022). Diagnostic and Statistical Manual of Mental Disorders (5th ed., text rev.). APA Publishing.
8. Arboleda-Velasquez, J. F., Lopera, F., O'Hare, M., et al. (2019). Resistance to autosomal dominant Alzheimer's disease in an APOE3 Christchurch homozygote. Nature Medicine, 25(11), 1680–1683. https://doi.org/10.1038/s41591-019-0611-3
9. Barnes, L. L., Dhana, K., Liu, X., et al. (2023). Trial of the MIND diet for prevention of cognitive decline in older persons. New England Journal of Medicine, 389(7), 602–611. https://doi.org/10.1056/NEJMoa2302368
10. Bellenguez, C., Küçükali, F., Jansen, I. E., et al. (2022). New insights into the genetic etiology of Alzheimer's disease and related dementias. Nature Genetics, 54(4), 412–436. https://doi.org/10.1038/s41588-022-01024-z
11. Biogen. (2026, 14 July). Biogen presents Phase 2 CELIA data at AAIC demonstrating meaningful clinical outcomes and robust tau reduction with diranersen in early Alzheimer's disease. https://investors.biogen.com/news-releases/news-release-details/biogen-presents-phase-2-celia-data-aaic-demonstrating-meaningful
12. BioArctic. (2026, 13 July). FDA approves Leqembi Iqlik (lecanemab-irmb) subcutaneous injection as a starting dose for early Alzheimer's disease. https://www.bioarctic.com/en/fda-approves-leqembi-iqlik-lecanemab-irmb-subcutaneous-injection-as-a-starting-dose-for-early-alzheimers-disease/
13. Boldrini, M., Fulmore, C. A., Tartt, A. N., et al. (2018). Human hippocampal neurogenesis persists throughout aging. Cell Stem Cell, 22(4), 589–599.e5. https://doi.org/10.1016/j.stem.2018.03.015
14. Buckley, R. F., Townsend, D. L., Birkenbihl, C. J., et al. (2026). The prognostic value of p-tau217 levels on progression to clinical impairment over 2, 5, and 10 years. JAMA. https://doi.org/10.1001/jama.2026.12556
15. Duering, M., Biessels, G. J., Brodtmann, A., et al. (2023). Neuroimaging standards for research into small vessel disease (STRIVE-2). The Lancet Neurology, 22(7), 602–618. https://doi.org/10.1016/S1474-4422(23)00131-X
16. World Health Organization. (2021). Global status report on the public health response to dementia. WHO.
17. World Health Organization. (2025). Dementia fact sheet. https://www.who.int/news-room/fact-sheets/detail/dementia
18. Eyting, M., Xie, M., Michalik, F., et al. (2025). A natural experiment on the effect of herpes zoster vaccination on dementia. Nature, 641, 438–446.
19. Fortea, J., Pegueroles, J., Alcolea, D., et al. (2024). APOE4 homozygosity represents a distinct genetic form of Alzheimer's disease. Nature Medicine, 30(5), 1284–1291. https://doi.org/10.1038/s41591-024-02931-w
20. Foutz, A., Appleby, B. S., Hamlin, C., et al. (2017). Diagnostic and prognostic value of human prion detection in cerebrospinal fluid. Annals of Neurology, 81(1), 79–92. https://doi.org/10.1002/ana.24833
21. GBD 2019 Dementia Forecasting Collaborators. (2022). Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050. The Lancet Public Health, 7(2), e105–e125. https://doi.org/10.1016/S2468-2667(21)00249-8
22. Hely, M. A., Reid, W. G. J., Adena, M. A., et al. (2008). The Sydney multicenter study of Parkinson's disease: The inevitability of dementia at 20 years. Movement Disorders, 23(6), 837–844. https://doi.org/10.1002/mds.21956
23. Jack, C. R., Andrews, J. S., Beach, T. G., et al. (2024). Revised criteria for diagnosis and staging of Alzheimer's disease: Alzheimer's Association Workgroup. Alzheimer's & Dementia, 20(8), 5143–5169. https://doi.org/10.1002/alz.13859
24. Jonsson, T., Atwal, J. K., Steinberg, S., et al. (2012). A mutation in APP protects against Alzheimer's disease and age-related cognitive decline. Nature, 488(7409), 96–99. https://doi.org/10.1038/nature11283
25. Keren-Shaul, H., Spinrad, A., Weiner, A., et al. (2017). A unique microglia type associated with restricting development of Alzheimer's disease. Cell, 169(7), 1276–1290.e17. https://doi.org/10.1016/j.cell.2017.05.018
26. Lamb, S. E., Sheehan, B., Atherton, N., et al. (2018). Dementia and Physical Activity (DAPA) trial. BMJ, 361, k1675. https://doi.org/10.1136/bmj.k1675
27. Lin, F. R., Pike, J. R., Albert, M. S., et al. (2023). Hearing intervention versus health education control to reduce cognitive decline (ACHIEVE). The Lancet, 402(10404), 786–797. https://doi.org/10.1016/S0140-6736(23)01406-X
28. Livingston, G., Huntley, J., Liu, K. Y., et al. (2024). Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet, 404(10452), 572–628. https://doi.org/10.1016/S0140-6736(24)01296-0
29. Lopera, F., Marino, C., Chandrahas, A. S., et al. (2023). Resilience to autosomal dominant Alzheimer's disease in a Reelin-COLBOS heterozygous man. Nature Medicine, 29(5), 1243–1252. https://doi.org/10.1038/s41591-023-02318-3
30. Mawuenyega, K. G., Sigurdson, W., Ovod, V., et al. (2010). Decreased clearance of CNS β-amyloid in Alzheimer's disease. Science, 330(6012), 1774. https://doi.org/10.1126/science.1197623
31. McKeith, I. G., Boeve, B. F., Dickson, D. W., et al. (2017). Diagnosis and management of dementia with Lewy bodies: Fourth consensus report of the DLB Consortium. Neurology, 89(1), 88–100. https://doi.org/10.1212/WNL.0000000000004058
32. Montine, T. J., Phelps, C. H., Beach, T. G., et al. (2012). National Institute on Aging–Alzheimer's Association guidelines for the neuropathologic assessment of Alzheimer's disease. Acta Neuropathologica, 123(1), 1–11. https://doi.org/10.1007/s00401-011-0910-3
33. Moreno-Jiménez, E. P., Flor-García, M., Terreros-Roncal, J., et al. (2019). Adult hippocampal neurogenesis is abundant in neurologically healthy subjects and drops sharply in patients with Alzheimer's disease. Nature Medicine, 25(4), 554–560. https://doi.org/10.1038/s41591-019-0375-9
34. Nation, D. A., Sweeney, M. D., Montagne, A., et al. (2019). Blood–brain barrier breakdown is an early biomarker of human cognitive dysfunction. Nature Medicine, 25(2), 270–276. https://doi.org/10.1038/s41591-018-0297-y
35. Nelson, P. T., Dickson, D. W., Trojanowski, J. Q., et al. (2019). Limbic-predominant age-related TDP-43 encephalopathy (LATE): Consensus working group report. Brain, 142(6), 1503–1527. https://doi.org/10.1093/brain/awz099
36. Neu, S. C., Pa, J., Kukull, W., et al. (2017). Apolipoprotein E genotype and sex risk factors for Alzheimer disease: A meta-analysis. JAMA Neurology, 74(10), 1178–1189. https://doi.org/10.1001/jamaneurol.2017.2188
37. Ngandu, T., Lehtisalo, J., Solomon, A., et al. (2015). A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER). The Lancet, 385(9984), 2255–2263. https://doi.org/10.1016/S0140-6736(15)60461-5
38. Novo Nordisk / evoke & evoke+ investigators. (2026). Efficacy and safety of oral semaglutide 14 mg in early-stage symptomatic Alzheimer's disease (evoke and evoke+): Two phase 3, randomised, placebo-controlled trials. The Lancet. https://doi.org/10.1016/S0140-6736(26)00459-9
39. Rascovsky, K., Hodges, J. R., Knopman, D., et al. (2011). Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain, 134(9), 2456–2477. https://doi.org/10.1093/brain/awr179
40. Rezai, A. R., D'Haese, P.-F., Finomore, V., et al. (2024). Ultrasound blood–brain barrier opening and aducanumab in Alzheimer's disease. New England Journal of Medicine, 390(1), 55–62. https://doi.org/10.1056/NEJMoa2308719
41. Roche / NeurologyLive. (2026). Roche unveils PrevenTRON, a phase 3 prevention trial of trontinemab in cognitively unimpaired individuals at high risk of Alzheimer decline. https://www.neurologylive.com/view/roche-unveils-preventron-phase-3-prevention-trial-trontinemab-cognitively-unimpaired-individuals-high-risk-alzheimers
42. Satizabal, C. L., Beiser, A. S., Chouraki, V., et al. (2016). Incidence of dementia over three decades in the Framingham Heart Study. New England Journal of Medicine, 374(6), 523–532. https://doi.org/10.1056/NEJMoa1504327
43. Schneider, J. A., Arvanitakis, Z., Bang, W., ve Bennett, D. A. (2007). Mixed brain pathologies account for most dementia cases in community-dwelling older persons. Neurology, 69(24), 2197–2204. https://doi.org/10.1212/01.wnl.0000271090.28148.24
44. Seeley, W. W., Crawford, R. K., Zhou, J., Miller, B. L., ve Greicius, M. D. (2009). Neurodegenerative diseases target large-scale human brain networks. Neuron, 62(1), 42–52. https://doi.org/10.1016/j.neuron.2009.03.024
45. Shokri-Kojori, E., Wang, G.-J., Wiers, C. E., et al. (2018). β-Amyloid accumulation in the human brain after one night of sleep deprivation. PNAS, 115(17), 4483–4488. https://doi.org/10.1073/pnas.1721694115
46. Sims, J. R., Zimmer, J. A., Evans, C. D., et al. (2023). Donanemab in early symptomatic Alzheimer disease: The TRAILBLAZER-ALZ 2 randomized clinical trial. JAMA, 330(6), 512–527. https://doi.org/10.1001/jama.2023.13239
47. Sorrells, S. F., Paredes, M. F., Cebrian-Silla, A., et al. (2018). Human hippocampal neurogenesis drops sharply in children to undetectable levels in adults. Nature, 555(7696), 377–381. https://doi.org/10.1038/nature25975
48. Republic of Türkiye Ministry of Health, General Directorate of Public Health. (2025). World Alzheimer's Day, 21 September. https://hsgm.saglik.gov.tr/tr/haberler-16/dunya-alzheimer-gunu-21-eylul-2025.html
49. Terry, R. D., Masliah, E., Salmon, D. P., et al. (1991). Physical basis of cognitive alterations in Alzheimer's disease: Synapse loss is the major correlate of cognitive impairment. Annals of Neurology, 30(4), 572–580. https://doi.org/10.1002/ana.410300410
50. van Dyck, C. H., Swanson, C. J., Aisen, P., et al. (2023). Lecanemab in early Alzheimer's disease. New England Journal of Medicine, 388(1), 9–21. https://doi.org/10.1056/NEJMoa2212948
51. Williamson, J. D., Pajewski, N. M., Auchus, A. P., et al. (2019). Effect of intensive vs standard blood pressure control on probable dementia (SPRINT MIND). JAMA, 321(6), 553–561. https://doi.org/10.1001/jama.2018.21442
52. U.S. Food and Drug Administration. (2025, 16 May). FDA clears first blood test used in diagnosing Alzheimer's disease [Press release]. fda.gov