The New Era of Alzheimer Disease Therapy — A 2026 Neurologist's Guide

Bottom Line: Two anti-amyloid antibodies — lecanemab and donanemab — are FDA-approved for early Alzheimer disease and deliver a 27–35% slowing of cognitive decline over 18 months. In absolute terms, that's roughly a 0.45-point CDR-SB difference or 4–7 months of preserved function. ARIA-E occurs in 12–14% of treated patients (up to 32–40% in APOE ε4 homozygotes). Patient selection, MRI monitoring, and APOE-stratified consent are now non-negotiable. Brexpiprazole is the first labeled option for AD-related agitation. Plasma p-tau217 is quietly rewriting the diagnostic workflow. What the trials show and what the field often claims about them are not the same thing — so let's look at what the trials actually showed.

Why This Matters

For two decades, AD pharmacotherapy meant donepezil, rivastigmine, galantamine, and memantine — symptomatic agents with marginal effect sizes. Aducanumab's 2021 accelerated approval was a regulatory event, not a clinical one; EMERGE was nominally positive while ENGAGE was negative, leaving the field unconvinced. Biogen withdrew the drug in January 2024.

That story is now over. Lecanemab (Leqembi) and donanemab (Kisunla) have full FDA approval based on statistically significant and clinically interpretable slowing of decline. Brexpiprazole (Rexulti) — approved May 2023 — filled a 30-year gap for AD-related agitation.

The shift is real. It is also small. Selecting the right patient is now where the neurologist earns the consult.

What the Trials Showed — Anti-Amyloid Antibodies

Lecanemab — Clarity AD (van Dyck, NEJM 2023)

• n = 1,795 early-AD patients (MCI-due-to-AD or mild dementia-due-to-AD)
• Primary endpoint: CDR-SB at 18 months — lecanemab 1.21 vs placebo 1.66
• Absolute difference: −0.45 points (27% relative slowing; P<0.001)
• All key secondaries (ADAS-cog14, ADCOMS, ADCS-MCI-ADL) favored lecanemab
• Amyloid PET: −55.5 centiloids vs +3.6 placebo
• ARIA-E 12.6% vs 1.7%; ARIA-H 17.3% vs 9.0%
• Symptomatic ARIA-E 2.8%; infusion-related reactions 26.4%

The effect size matters more than the P-value. A 0.45-point CDR-SB difference over 18 months is at the low end of what most cognitive specialists call clinically meaningful — not nothing, but not transformative.

Donanemab — TRAILBLAZER-ALZ 2 (Sims, JAMA 2023)

• n = 1,736 early symptomatic AD, stratified by tau PET
• Primary endpoint: iADRS at 76 weeks — 35.1% slowing in the low/medium tau population (Δ 3.25; P<0.001)
• CDR-SB in the low/medium tau population: 36% slowing (Δ −0.67)
• Combined population (including high tau): 22.3% slowing on iADRS
• Effect attenuated in high-tau participants — biology matters
• 80.1% of low/medium tau participants achieved amyloid clearance (<24.1 centiloids on PET) at 76 weeks — the basis for the FDA-endorsed stopping rule
• ARIA-E 24% with original titration; 14% with modified titration (TRAILBLAZER-ALZ 6)
• ARIA-H 31.4% (vs 13.6% placebo); symptomatic ARIA-E 6.1%; three ARIA-related deaths in the pivotal trial
• Unique feature: FDA-endorsed stopping rule when amyloid PET clears below threshold — typically 1–2 years of therapy

Head-to-Head — The Practical Comparison

There is no head-to-head trial. Cross-trial relative risk reduction comparisons should be read with skepticism — different endpoints (CDR-SB vs iADRS), different populations, different durations.

ARIA — The Defining Safety Topic

Amyloid-related imaging abnormalities are the new clinical skill. Every neurologist treating AD now needs a protocol.

• Timing: Most ARIA occurs within the first 6 months, especially around infusions 3–5
• Risk factors: APOE ε4 dose (homozygote > heterozygote > non-carrier), anticoagulation, t-PA exposure, baseline microbleed burden
• Hard rule: Avoid systemic thrombolytics in patients on anti-amyloid therapy — post-marketing fatal hemorrhages have been reported

The Pre-Treatment Workup — Nine Things to Confirm

1. Stage confirmed as MCI-due-to-AD or mild dementia (typically MMSE ≥22)
2. Amyloid pathology confirmed (PET or CSF Aβ42/40 + p-tau)
3. Baseline brain MRI within ~12 months — >4 microhemorrhages or any macrohemorrhage typically excludes
4. APOE ε4 genotyping — required for informed consent under the boxed warning
5. Anticoagulation review — warfarin and DOACs generally avoided; aspirin acceptable
6. Stroke/CAA history — recent stroke or active CAA is a contraindication
7. Caregiver, transport, and infusion-center feasibility
8. Insurance authorization and CMS registry enrollment
9. Documented shared decision-making — modest benefit, real risk, calibrated family expectations

The Symptomatic Side — Don't Forget Brexpiprazole

Approved May 2023 for AD-related agitation — the first label-indicated option for one of the most disabling problems in late disease.

• Phase 3 evidence (Lee, JAMA Neurol 2023): n=345 randomized 2:1 to brexpiprazole (2 or 3 mg) or placebo over 12 weeks
• Mean CMAI reduction: −22.6 with brexpiprazole vs −17.6 with placebo
• Placebo-adjusted difference: −5.3 points (effect size 0.35; P=0.003)
• Dosing: 0.5 mg/day × 7 days → 1 mg/day × 7 days → 2 mg/day target; max 3 mg/day after 14 days at target
• Boxed warning: increased mortality in elderly patients with dementia-related psychosis on antipsychotics (class effect)
• Common adverse events: headache, dizziness, urinary tract infection, nasopharyngitis, somnolence/insomnia

Non-pharmacologic interventions remain first-line. But after 30 years of off-label haloperidol, risperidone, and quetiapine roulette, having an evidence-based, labeled option matters.

AChEIs and memantine remain standard background therapy through the disease course.

The Decision Tree

Cognitive complaint → confirm MCI or mild dementia stage → confirm AD pathology (amyloid PET, CSF, or plasma p-tau217 for triage) → screen for agitation (brexpiprazole if present) → assess anti-amyloid eligibility against the nine-point workup → APOE-stratified consent → choose lecanemab or donanemab based on patient priorities (treatment duration, home administration, tau status).

Not eligible? Continue AChEI + memantine, manage symptoms, address vascular risk, plan for caregiver support.

Strengths and Limitations

Strengths

• Both pivotal trials were large, well-powered, placebo-controlled, with biomarker-confirmed populations
• Primary endpoints were prespecified and clinically interpretable (CDR-SB, iADRS)
• Biomarker reduction (amyloid PET centiloids) was unambiguous and dose-dependent
• TRAILBLAZER-ALZ 6 directly addressed a safety signal with a label-changing modification

Limitations — Read These Before You Quote the Press Release

• Effect size is modest. A 27–35% relative slowing translates to ~4–7 months of preserved function over 18 months. The absolute CDR-SB difference (0.45 points on an 18-point scale) sits at the threshold of what neuropsychologists consider clinically meaningful
• Trial populations were highly selected — real-world memory-clinic eligibility is <30%
• Industry-sponsored, though event adjudication was independent
• Long-term safety beyond 18–24 months remains thin, especially for indefinite lecanemab maintenance
• No head-to-head data. Cross-trial comparisons of "27% vs 35%" should not drive prescribing decisions
• APOE ε4 homozygotes — ~15% of treatment candidates — carry ARIA-E rates of 32–40%. Whether they should be treated at all is unsettled
• Tau-high patients showed attenuated benefit in TRAILBLAZER-ALZ 2 — biology is not uniform
• Stopping criteria for lecanemab are undefined; progression to moderate dementia is the only consensus off-ramp

Real-World Implementation

• Screening yield: <30% of memory-clinic referrals end up amyloid-positive AND eligible. The Lumipulse plasma p-tau217/Aβ42 ratio (FDA-cleared May 16, 2025) has dramatically improved triage efficiency and will likely displace LP and PET in screening within 3–5 years
• Cost: ~$26,500/year in the US for lecanemab; donanemab similar. CMS reimbursement requires registry participation
• Infrastructure burden: infusion capacity, MRI scheduling, radiologist familiarity with ARIA — these are the rate-limiting steps, not the prescription
• SC lecanemab (Leqembi IQLIK) approved August 29, 2025 for weekly home maintenance; SC starting-dose sBLA accepted January 2026 (PDUFA August 24, 2026). If approved, home-only therapy becomes a real option

What's Coming in the Next 12–24 Months

• Trontinemab (Roche) — transferrin-receptor brain-shuttle anti-amyloid; lower ARIA in early data; phase 3 TRONTIER underway
• Remternetug (Lilly) — next-generation N3pG anti-amyloid in subcutaneous formulation
• Anti-tau agents — bepranemab, E2814, HMTM (hydromethylthionine) with phase 2/3 readouts
• Secondary prevention — DIAN-TU, API ADAD Colombia, and AHEAD 3–45 will define the pre-symptomatic intervention paradigm
• LATTICE — low-dose lithium for prevention (results pending 2026–2027)
• VALAD — valacyclovir / HSV-1 hypothesis
• Plasma p-tau217 — triage today, screening tomorrow, plausibly a treatment-selection biomarker within five years

What This Means for Practice

The neurologist's job has changed. The question is no longer "which symptomatic agent?" but "is this patient a candidate to modify the trajectory?"

• Benefit is real but modest — most defensible in MCI or mild dementia with biomarker-confirmed amyloid, APOE-stratified consent, and reliable MRI access
• ARIA management is the new core skill — every prescriber needs an MRI-monitoring protocol and a hold/discontinue rule
• Brexpiprazole filled the agitation gap — labeled, evidence-based, but still under the antipsychotic boxed warning
• Plasma p-tau217 is changing the workflow now — adopt it for triage
• Donanemab's treat-to-clearance design is likely a template; lecanemab's indefinite maintenance is the open question
• Calibrate family expectations explicitly — 4–7 months of preserved function over 18 months is the conversation, not "a cure"

Practice should follow the trial evidence, not the marketing of it.