What it is:A progressive, irreversible neurodegenerative disease of unknown cause that erodes memory, cognition, and eventually the ability to do anything for yourself. It's the most common cause of dementia by a wide margin, and it affects roughly 7.5 million Americans.
The core problem:Two abnormal proteins, β-amyloid and tau, build up faster than the brain can clear them. Amyloid clumps outside neurons into plaques, tau tangles up inside them, and the net result is dying synapses, dying neurons, and a brain that's visibly shrunk (cortical atrophy) by the time symptoms are obvious.
What you do about it:Nothing available today reverses or stops AD. Cholinesterase inhibitors and memantine buy modest, temporary symptomatic ground. The new anti-amyloid antibodies target the actual pathology but their real-world clinical payoff is still small relative to their cost, monitoring burden, and risk.
Every drug in this chapter is a patch, not a repair. Cholinesterase inhibitors boost what's left of a dying cholinergic system. Memantine dampens excitotoxic glutamate signaling. Anti-amyloid antibodies clear plaque but only modestly slow the clinical slide. Keep that framing in mind and the whole treatment section stops feeling like a drug list and starts feeling like a logical response to the pathophysiology.
AD isn't a switch that flips on, it's a continuum. Patients move through an asymptomatic preclinical phase, then a symptomatic preclinical phase(this is where mild cognitive impairment, MCI, lives), and finally the dementia phasewhere functional independence is actually lost. Only about 1 in 5 patients with MCI progress to AD, so MCI itself is not automatically treated pharmacologically.
Once a patient is in the dementia phase, severity is staged clinically and tracked with MMSE score:
| Stage | MMSE | What you'll see |
|---|---|---|
| Mild | 26–21 | Trouble with recent events, finances, meal prep, other household tasks. May get lost driving. Withdraws from hobbies and difficult tasks. Often denies there's a problem. |
| Moderate | 20–10 | Needs help with ADLs. Disoriented to time. Recent memory badly impaired, may forget names of family. Day-to-day function fluctuates. Loses ability to drive safely. Agitation, paranoia, and delusions become common. |
| Severe | 9–0 | Loses speech, ambulation, and self-feeding. Incontinent of urine and stool. Needs 24/7 care. |
These aren't just for the exam, they gate therapy. Donepezilis the only cholinesterase inhibitor labeled all the way down through severe disease. Memantineonly enters the picture at moderate-to-severe disease, never for mild AD alone. An untreated patient loses roughly 2–4 MMSE points per year; a successfully treated patient loses less than 2 points per year, that's literally how you define "successful treatment" on a cholinesterase inhibitor.
Genetic susceptibility for late-onset (typical) AD tracks mainly with the APOE genotype, layered on top of environmental factors. True early-onset AD (before age 65) accounts for less than 1% of cases and comes from dominantly inherited chromosomal changes that disrupt how amyloid precursor protein (APP) gets processed. Risk factors for the more common late-onset form: age (the single biggest driver, prevalence roughly doubles every 5 years after 65), reduced brain reserve, head injury, Down syndrome, depression, MCI, and the whole vascular risk cluster (hypertension, elevated homocysteine, high LDL, low HDL, obesity, metabolic syndrome, diabetes).
APP sits in the neuronal membrane and gets clipped by secretase enzymes. Which enzyme does the clipping determines whether you get something harmless or something toxic.
| Pathway | Enzymes | Product | Consequence |
|---|---|---|---|
| Non-amyloidogenic | α-secretase | β-amyloid40 | More soluble, less plaque-forming, relatively harmless |
| Amyloidogenic | β-secretase → γ-secretase | β-amyloid42 | Highly insoluble, aggregates into extracellular plaques, associated with neuronal injury |
The theory is a simple imbalance: β-amyloid42production outpaces clearance, it aggregates into plaques in the cortex and medial temporal lobe, and that aggregation kicks off the rest of the pathology. It's still unclear whether amyloid is the actual primary driver in most AD or just an early bystander, which is exactly why the anti-amyloid antibodies below have been such a mixed bag clinically. Notably, AD patients accumulate a heavier plaque and tangle burden even in their youngeryears compared to age-matched people who never develop AD.
Tau is a structural protein that normally stabilizes the microtubule "scaffolding" inside neurons. In AD, tau becomes hyperphosphorylated, stops doing its job, and instead aggregates into paired helical filaments that form the intracellular neurofibrillary tangles (NFTs)you'll see on pathology slides. Both plaques and tangles correlate with neurodegeneration, synaptic loss, and the macroscopic cortical atrophy seen on imaging.
Cholinergic loss is the single most prominent neurotransmitter deficitand tracks closely with disease severity. It's important to get the causality right: cholinergic cell death is a consequenceof the amyloid/tau pathology, not the original cause. That's exactly why cholinesterase inhibitors only manage symptoms rather than modifying the disease.
Map the drugs onto the pathology and the whole chapter clicks: cholinesterase inhibitorsslow the breakdown of what little acetylcholine is left. Memantineblocks NMDA receptors to tone down excitotoxic glutamate signaling without shutting down the physiologic signaling needed for learning. Anti-amyloid antibodiesattack the plaque itself. None of them touch tau, and none of them reliably slow neuronal death, they just buy time against the symptoms of a process that keeps running underneath.
Memory impairment is the cardinal, most common presenting symptom.Early on, executive dysfunction (planning, organizing) and visuospatial impairment are frequently present too, sometimes before memory loss is obvious to family. Language deficits and behavioral symptoms tend to show up later as the disease advances.
| Stage | What you'll see |
|---|---|
| Early | Trouble learning and retaining new information, planning and organizing, and mood changes |
| Progressive | Worsening of the above, plus personality changes, impaired judgment, speech deficits, spatial disorientation |
| Late | Functional collapse: gait disturbance, dysphagia, incontinence, marked behavioral change. Total dependence for self-care |
This is one of the most tested distinctions in the chapter. Dementia(including AD) is an insidious, chronic, generally irreversible decline in cognition. Deliriumis an acute, fluctuating disturbance in attention and awareness that develops over hours to days, is a medical emergency, and is often reversible once you find and fix the cause (infection, metabolic derangement, drug effect, withdrawal). A demented patient can absolutely develop superimposed delirium on top of their baseline, which is exactly why acute, sudden decline always warrants a fresh workup rather than being written off as "the dementia getting worse."
AD is a clinical diagnosis. There's no single confirmatory test, it's built from history, a physical exam, targeted labs to exclude mimics, and structured cognitive testing that documents deficits in two or more cognitive domains, corroborated by both the patient and a caregiver.
Screen the medication list before you blame the disease. Contributors to dementia-like symptoms:anticholinergics, sedatives, hypnotics, opioids, antipsychotics, anticonvulsants. Contributors to delirium specifically:digoxin, NSAIDs, H2 receptor antagonists, amiodarone, antihypertensives, corticosteroids. A medication list cleanup is sometimes the highest-yield intervention you can make before assuming irreversible decline.
These are short, standardized tests, not full neuropsychological batteries, and they're what establishes a baseline you can track decline against over time.
| Tool | Scale | Abnormal cutoff | Domains tested |
|---|---|---|---|
| MMSE | 0–30, higher is better | <24 (single cutoff); education-adjusted cutoffs also used | Orientation (10 pts), memory (6), attention/concentration (5), language (8), visuospatial (1) |
| MoCA | 0–30, higher is better | <26/30 | Delayed recall (5), visuospatial/executive incl. clock draw (7), language (6), attention (6), orientation (6) |
They test similar domains but MoCA is more sensitive to subtle, mild impairment, which is why it's often preferred for catching early or borderline cases the MMSE misses. The MMSE also relies heavily on verbal/language skill, so it under-performs in patients with limited education, which is exactly why education-adjusted cutoffs exist for it. One more practical wrinkle: the MMSE is copyrighted, use requires registration and a per-use fee. MoCA is free.
Goals of treatment:preserve cognitive function and activities of daily living for as long as possible. The secondary goal is managing the psychiatric and behavioral symptoms that tend to show up as the disease advances. Nothing here reverses damage already done, so "success" is realistically defined as slowed declineand delayed long-term care placement, not improvement.
Donepezil, rivastigmine, galantamine. First-line for mild-to-moderate AD.
NMDA antagonist. Added for moderate-to-severe disease, alone or on top of a CEI.
Aducanumab and newer agents. For MCI due to AD and mild AD.
For MCI, consider an anti-amyloid mAb; there is no cholinesterase inhibitor indication here. For mild-to-moderate AD, start a cholinesterase inhibitor oran anti-amyloid mAb. For moderate-to-severe AD, add memantine to the cholinesterase inhibitor (or use memantine or a CEI alone). A reasonable, honest goal to set with the family is a slowed rate of declineand delayed nursing home placement, not reversal of symptoms already present.
Practical points that apply across the whole cognitive-symptom category: simplify the regimenand factor in caregiver burden to protect adherence and persistence, since these are long-term, unglamorous daily medications. Whether stopping therapy causes a real loss of benefit is controversial, that ambiguity is exactly why deprescribing decisions are hard and individualized. And remember, behavioral symptoms often need a separatepharmacologic approach on top of the cognitive drugs (see BPSD below).
| Drug | Start | Usual range / target | Key adjustment |
|---|---|---|---|
| Cholinesterase inhibitors (mild-moderate AD; donepezil also severe) | |||
| Donepezil | 5 mg daily, evening | 5–10 mg/day (mild-mod) · 10–23 mg/day (mod-severe) | None recommended. Interruption → retitrate. |
| Rivastigmine | 1.5 mg BID (cap/soln) or 4.6 mg/day patch | 3–6 mg BID (cap/soln) · 9.5–13.3 mg/day (patch) | Renal/hepatic impairment or <50 kg: lower dose. Patch + hepatic/low weight: cap at 4.6 mg/24h. |
| Galantamine | 4 mg BID (tab/soln) or 8 mg daily AM (ER) | 8–12 mg BID (tab/soln) · 16–24 mg/day (ER) | Moderate renal/hepatic: max 16 mg/day. Severe: not recommended. |
| NMDA receptor antagonist (moderate-severe AD, never mild alone) | |||
| Memantine | 5 mg daily (tab/soln) or 7 mg daily (ER) | 10 mg BID (tab/soln) · 28 mg daily (ER) | Severe renal impairment: 5 mg BID (tab/soln) or 14 mg daily (ER). Severe hepatic: caution. |
| Combination (moderate-severe AD) | |||
| Memantine ER / donepezil | 7 mg/10 mg (stabilized on donepezil alone) | 28 mg/10 mg daily (stabilized on both) | Severe renal: 14 mg/10 mg daily. |
| Anti-amyloid monoclonal antibody (MCI due to AD, mild AD) | |||
| Aducanumab | 1 mg/kg IV | Titrated to 10 mg/kg IV every 4 weeks | No dosage adjustments recommended. |
Rivastigmine and galantamine have short half-lives: interrupt therapy for several days or more and you must retitrate from the starting dose. Same rule applies to donepezil and memantine, and to the combination product. When switching between cholinesterase inhibitors, a washout period is recommended. Abrupt discontinuation of a cholinesterase inhibitor can worsen cognition and behavior.
All three (donepezil, rivastigmine, galantamine) are considered roughly equally effective, there's no trial evidence favoring one over another head-to-head. Onset of benefit is delayed, typically 3–6 months, and the effect is purely symptomatic: they slow decline, they don't reverse it and don't modify the underlying disease. A "successful" response is an MMSE decline of less than 2 points per year, with benefit generally lasting somewhere in the 3–24 month range.
Applying more than one rivastigmine patch at the same time is associated with hospitalization and death from cholinergic toxicity. This shows up as a distractor answer choice constantly, patients and caregivers need to be told explicitly: one patch at a time, remove the old one before applying the new one.
Class-wide adverse effects to monitor:dizziness, syncope, bradycardia, atrial arrhythmias, SA/AV block, and even MI have been reported. Check pulse and blood pressure (including orthostatics) at baseline, monthly during titration, and every 6 months thereafter. Most dizziness is mild, transient, and not cardiovascular in nature, but you still have to rule that out with vitals. GI effects (nausea, vomiting, diarrhea, anorexia, weight loss) are usually transient around initiation, up-titration, or a drug switch, but frail or low-body-weight patients can lose significant weight, especially on rivastigmine or when titrating donepezil to 23 mg. Watch for peptic ulcer disease or GI bleeding, especially in patients with an ulcer history or on concurrent NSAIDs. Insomnia and vivid or abnormal dreams can occur; adjusting donepezil's dose timing can help.
Galantamine carries risk of Stevens-Johnson syndrome and acute generalized exanthematous pustulosis. Discontinue at the first sign of a skin rash unless it's clearly unrelated, and don't rechallenge if the reaction looks serious.
Memantine is an NMDA receptor antagonist, a mechanistically distinct class from the cholinesterase inhibitors. Glutamate is the principal excitatory neurotransmitter in the hippocampus, and chronic overactivation of NMDA receptors is thought to be neurotoxic (excitotoxicity). Memantine dampens that pathologic excess signal while, importantly, sparing the normal physiologic glutamate signaling needed for learning and memory, which is why it doesn't just blunt cognition across the board.
It's used both as monotherapy and layered on top of a cholinesterase inhibitor, individually or as the fixed-dose combination (memantine ER/donepezil, brand Namzaric). It's indicated for moderate-to-severe AD only, never for mild AD. Combination therapy slows cognitive and functional decline more than cholinesterase inhibitor monotherapy or no treatment at all, and as a bonus, memantine may partially offset the GI side effects that come with the cholinesterase inhibitors.
Memantine is not hepatically metabolized, it's cleared renally, so it needs dose adjustment in renal impairment (in contrast to donepezil, which needs none). It's generally well tolerated: headache, confusion, dizziness, and hallucinations can occur, and confusion during dose titration is usually transient. Constipation is the most notable GI complaint, the opposite problem from what cholinesterase inhibitors cause, which is part of why the combination can feel more tolerable than a CEI alone.
Practical tip for patients with swallowing difficulty: the extended-release capsule (memantine alone or the combination product) can be opened and the contents sprinkled on applesauce.
Four humanized IgG1 monoclonal antibodies against β-amyloid have earned FDA breakthrough therapy designation: aducanumab, lecanemab, donanemab, and gantenerumab.These are the first drugs designed to hit the actual pathology (amyloid plaque) rather than just patch over the downstream neurotransmitter deficit.
| Agent | Approval context | What it does |
|---|---|---|
| Aducanumab | Controversial accelerated approval for MCI due to AD and mild AD | Substantially reduces brain amyloid on PET (a surrogate endpoint); clinical benefit was inconsistent across its two pivotal trials, one showed a smaller decline with high dose, the other showed no difference from placebo |
| Lecanemab, Donanemab | Approved 2023 | Highly effective at reducing amyloid burden on PET, considered disease-modifying, but clinical efficacy is described as "modest." Neither cures AD, which continues to progress in every patient |
ARIA (amyloid-related imaging abnormalities), meaning brain edema, microhemorrhage, or superficial siderosis, is the signature risk of this class. MRI is required at baseline, prior to the 7th and 12th infusions, and any time symptoms suggest it (headache, confusion, dizziness, visual disturbances, nausea). Vigilance for ARIA and focal neurologic changes should be highest during dose titration and the first eight doses. Hypersensitivity reactions (angioedema, urticaria) are rare but possible.
Dosing (aducanumab):requires a recent (within one year) brain MRI before starting, then administered by IV infusion over roughly one hour, diluted in 100 mL of 0.9% NaCl, once every 4 weeks with dose titration (a common titration pattern is lower doses at weeks 0 and 4, stepping up at weeks 8 and 16, continuing to a maintenance dose up to 10 mg/kg). Additional MRIs are required during the escalation period.
The honest clinical framing your lecture material pushed: these agents are expensive, require intensive imaging surveillance, and carry real safety risk, including reported deaths, for a benefit that's real on a brain scan but only modest and sometimes clinically imperceptible to the patient. Professional guidance has generally leaned toward notroutinely recommending aducanumab specifically, while acknowledging that individual patients and physicians may reasonably choose it given the (uncertain) potential upside. The degree and durability of benefit across this whole drug class remains genuinely unsettled.
Estrogen, NSAIDs, prednisone, statins, and Ginkgo biloba are not recommendedto prevent or treat dementia. Don't use Ginkgo biloba in patients on anticoagulants or antiplatelets, and use it cautiously with NSAIDs (bleeding risk stacks). Vitamin E is still being studied for prevention but isn't recommended for treatment. There's currently insufficient evidence for omega-3 fatty acids or medical foods marketed for AD (Axona, Souvenaid, Cerefolin NAC). Curcumin has also not been shown effective.
No medication is FDA-approved specifically for BPSD, which spans four rough categories: psychotic symptoms, hyperactive/disruptive behavior, affective symptoms (like depression), and apathy. General principles apply across all of these:reserve pharmacologic treatment for when nonpharmacologic strategies have failed, start at reduced doses and titrate slowly, monitor closely, periodically attempt to taper and discontinue, and document everything carefully.
Cholinesterase inhibitors and memantine can help BPSD generally, but they do notreduce acute agitation, don't reach for them expecting a fast behavioral fix. Avoid anticholinergic medications in this population altogether, they'll worsen cognition on top of whatever they're being given for.
| Symptom | Notes | Reasonable options |
|---|---|---|
| Anxiety | Partly driven by the patient's own awareness they can't remember things | Buspirone, SSRI/SNRI, gabapentin. Limit benzodiazepines. |
| Apathy | Often one of the earliest symptoms | Cholinesterase inhibitors, methylphenidate, modafinil |
| Depression | Present in up to 80% of AD patients | SSRIs (sertraline, citalopram have the best evidence), mirtazapine. Avoid TCAs. |
| Insomnia | Circadian rhythm disruption is common | Melatonin. Limit benzodiazepines and zolpidem. |
| Paranoia / hallucinations | Frequently accusatory (e.g., accusing a partner of stealing) | Atypical antipsychotics, at very low doses, and only if refractory to other measures |
| Aggression | The hardest symptom cluster to manage | Prazosin has been used; nonpharmacologic de-escalation remains first-line |
Antidepressants aren't only for frank depression here, they're also used to manage anxiety, apathy, agitation, and aggression. An SSRIis the go-to, sertraline and citalopram have the strongest evidence base. Tricyclics are generally avoided given their anticholinergic burden in an already vulnerable brain.
Second-generation antipsychotics (aripiprazole, risperidone, olanzapine, quetiapine) beat placebo for psychotic and hyperactive symptoms, but the higher risk of adverse effects, including increased mortality(a black-box warning in this population), largely offsets that benefit. Reserve them for severe symptoms unresponsive to everything else, taper as early as possible, and rarely continue them beyond 12 weeks. Watch for somnolence, extrapyramidal symptoms, abnormal gait, worsening cognition, and cerebrovascular events.
Evidence for benzodiazepinesin BPSD is lacking and they're not advised given the real risk of falls, sedation, and paradoxical agitation in this population. Antiseizure medications used as mood stabilizers (carbamazepine, lamotrigine, pregabalin, gabapentin) are sometimes tried for agitation but the evidence is conflicting. Valproic acid is no longer recommendedhere given its adverse effect profile.
AD is the most common dementia by far, but the others have distinguishing features and, critically, different drug sensitivities that show up as distractors:
At baseline, interview both the patient and caregiver together to identify target symptoms and set concrete therapeutic goals. Document cognitive status, physical status, functional performance, mood, thought processes, and behavior so you actually have something to compare against later. Validated scales help quantify change over time: the Bristol Activities of Daily Living Scalefor function, and the Neuropsychiatric Inventory Questionnairefor behavioral disturbance.
Reassess medication changes and dose adjustments at 2–4 weeks and again at 8–12 weeksafter starting a new agent, then repeat full assessments every 3–6 monthsthereafter. Give it time: it can take several months to a full year of treatment before you can honestly say whether a cognitive medication is providing benefit.
Deprescribing guidelines exist to help decide when to stop these medications, but whether and when to stop for lack of efficacy remains genuinely controversial, and stopping abruptly risks a rebound worsening of cognition and behavior. This isn't a settled area, expect nuance rather than a hard rule here.
| Parameter | When | Watching for |
|---|---|---|
| MMSE / MoCA score | Baseline, then periodically (roughly every 6 months) | Rate of decline. Success = decline <2 points/year on a CEI |
| Pulse and BP (incl. orthostatics) | Baseline, monthly during CEI titration, then every 6 months | Bradycardia, syncope, arrhythmia, falls |
| Weight and GI symptoms | Every visit, especially during CEI initiation/titration/switch | Nausea, diarrhea, anorexia, weight loss, especially in frail/low-weight patients |
| Skin | Every visit on galantamine or rivastigmine patch | Rash (galantamine → SJS/AGEP risk) or spreading local reaction beyond patch size (rivastigmine) |
| Renal function | Baseline and periodically on memantine or galantamine | Need for dose adjustment |
| Mental status changes | Every visit, especially early in memantine titration | Confusion, hallucinations, headache, dizziness (usually transient on memantine) |
| MRI | Baseline, before the 7th and 12th aducanumab infusions, or if symptomatic | ARIA: edema, microhemorrhage, superficial siderosis |
| Functional status / ADLs | Every visit, using a validated scale | Trajectory of independence, need for caregiver support or LTC placement |
| Behavioral symptoms | Every visit, via caregiver interview | New psychosis, agitation, depression, changes warranting a taper attempt of any BPSD drug |