For most of the twentieth century, an Alzheimer’s diagnosis was a kind of verdict. The neurologist delivered it, the family absorbed it, and everyone — patient included, for as long as the patient could understand anything — waited for the decline that medicine had decided was inevitable. There was nothing to do except manage the symptoms of a brain that was, by consensus, already lost.
Gayatri Devi has spent thirty years watching that consensus dissolve.
Devi is a neurologist and psychiatrist who runs a memory disorders practice on the Upper East Side of Manhattan, where she has been seeing patients since 1996. She is board-certified in neurology, psychiatry, brain injury medicine, pain medicine, and behavioral neurology — a range that reflects less a collector’s instinct than a clinician’s recognition that the brain does not respect subspecialty boundaries. Her patients are, by her own description, mostly high-functioning: lawyers, physicians, executives, people who are still working and intend to stay that way, people whose brains are so practiced at compensation that standard cognitive screening tools miss what is happening inside them entirely.
What has changed in her field over those three decades is not incremental. It is, she says, the difference between practicing medicine before and after penicillin. There are now drugs that can clear the pathological proteins that define Alzheimer’s disease from the brain. There are blood tests that can detect the disease’s biological signature years before a person forgets a single word. There are scanning technologies that can map amyloid and tau deposition with a precision that would have been unimaginable when Devi finished her training. And there is, slowly, a reconceptualization of Alzheimer’s itself — not as a single disease with a single trajectory, but as a spectrum condition as heterogeneous as the brains it inhabits, one that in some patients can be slowed, stabilized, and in rare cases partially reversed.
None of this means Alzheimer’s has been solved. It has not. But the therapeutic nihilism that defined the field for generations is no longer defensible, and Devi is among the clinicians most responsible for demonstrating why.
Not a Single Disease
The word “dementia” describes a process, not a diagnosis. It refers to a progressive loss of cognitive function severe enough to interfere with daily life — a deterioration in the brain’s connectivity that can arise from several distinct pathological processes, often occurring simultaneously. Alzheimer’s disease is the most common of these, accounting for roughly 60 to 80 percent of dementia cases, but it almost never presents in isolation. Autopsy studies have found that between 98 and 99 percent of patients with Alzheimer’s pathology have at least one concurrent brain condition — most commonly cerebrovascular disease (Schneider JA et al., Annals of Neurology, 2007).
Alzheimer’s disease is defined by three pathological features: the accumulation of extracellular amyloid plaques between neurons, the formation of intracellular neurofibrillary tangles made of hyperphosphorylated tau protein, and — increasingly recognized as a third pillar — neuroinflammation driven by activated microglial and other glial cells. These changes disrupt synaptic connectivity, the brain’s fundamental mechanism of communication, and that disruption is what produces cognitive symptoms.
But the relationship between pathology and symptoms is not straightforward, and this is one of the field’s most important and least appreciated facts. A person can carry a substantial burden of amyloid plaques and tau tangles and remain cognitively intact. Another person with a lighter pathological load may be significantly impaired. The difference lies largely in what researchers call cognitive reserve — the brain’s capacity to maintain functional connectivity despite structural damage. Highly educated individuals, people who have spent decades in cognitively demanding work, people with large and active social networks, tend to have more of it. Their brains are better at routing around damage.
This creates a diagnostic paradox that Devi encounters constantly. Her most cognitively sophisticated patients are often her most diagnostically challenging ones. A standard cognitive screening tool — the Mini-Mental State Examination, the Montreal Cognitive Assessment — may return a near-perfect score in someone whose brain is already meaningfully compromised, because the test is not sensitive enough to detect the subtle deficits that high reserve masks. Devi’s evaluation protocol is consequently far more extensive: neuropsychological testing that takes several hours, electroencephalography to detect early slowing in brain activity, transcranial Doppler ultrasound to assess cerebral blood flow, specialized MRI sequences that can measure hippocampal and parietal lobe volumes, amyloid and tau PET scans where indicated, dopamine transporter scans to rule out Lewy body pathology, and a full laboratory workup including APOE genotyping and inflammatory markers. For patients traveling from out of town, this process unfolds over two to three days.
The goal is not to arrive at a stage — Devi is explicitly resistant to staging, which she considers reductionist — but to build a multidimensional picture of where each cognitive domain stands relative to where it should be for that particular person. A patient whose overall intellectual ability sits at the 99th percentile but whose language scores have fallen to the 10th percentile has lost something significant, even if a standard screening tool would never flag it. The decline is visible only against the baseline of what that brain was capable of.
The Order of Events
For decades, the dominant model of Alzheimer’s pathogenesis placed amyloid deposition at the beginning of the causal chain. Amyloid accumulated first, tau pathology followed, synaptic dysfunction came after that, and cognitive symptoms arrived last — sometimes twenty or thirty years after the first amyloid plaques appeared. This “amyloid cascade hypothesis,” first formally articulated by Hardy and Higgins in 1992, shaped virtually every major drug development program in the field for the next three decades.
The model is not wrong, exactly, but it is increasingly understood to be incomplete. There is growing evidence that changes in neuroinflammatory pathways may precede amyloid deposition by a substantial margin — that the brain’s immune system, specifically its microglial cells, begins responding to some upstream insult before the first plaques form (Heneka MT et al., The Lancet Neurology, 2015). What that upstream insult is remains an active area of investigation, but one hypothesis with accumulating support involves viral pathogens with a predilection for neural tissue.
The evidence linking herpes viruses to Alzheimer’s risk has been building for years, but it reached a new level of rigor in 2023, when a study published in Nature exploited a natural experiment in the United Kingdom’s shingles vaccination program. Because the UK’s varicella zoster vaccine was initially offered only to people aged 70 to 79 — with those just above the age cutoff ineligible — Eyting and colleagues were able to compare dementia rates in people who received the vaccine against a near-identical group who did not, controlling for the confounding that plagues most observational studies in this area. The result: vaccination against varicella zoster virus was associated with a 20 percent reduction in dementia diagnoses over seven years (Eyting M et al., Nature, 2023). A separate large Taiwanese cohort study found that antiviral treatment for herpes simplex virus infections was associated with roughly a 90 percent reduction in subsequent dementia risk (Tzeng NS et al., Neurotherapeutics, 2018).
The mechanism is presumed to be inflammatory. Both herpes simplex virus type 1 and varicella zoster virus establish latent infections in neural tissue and reactivate periodically, triggering immune responses in the central nervous system. Whether this chronic low-grade neuroinflammation is sufficient to initiate the amyloid cascade, or whether it accelerates a process already underway through other means, is not yet established. But the epidemiological signal is strong enough that some clinicians — Devi among them — have incorporated antiviral management into their preventive approach for high-risk patients.
The same inflammatory logic extends to oral health. Porphyromonas gingivalis, the primary bacterial driver of periodontal disease, has been detected in postmortem Alzheimer’s brains, and its toxic proteases — gingipains — have been found to correlate with tau burden and neuronal damage markers in brain tissue (Dominy SS et al., Science Advances, 2019). The gums are not far from the brain, and when they are chronically inflamed, the systemic consequences are not confined to the cardiovascular system.
Devi has two pairs of siblings in her practice, each pair in which one sibling has significant Alzheimer’s pathology and the other does not. In each case, the unaffected sibling has a history of aggressive immunological treatment for an unrelated condition — treatment that has, as a side effect, modulated their immune response for decades. One of these patients carries two copies of the APOE4 allele, the strongest common genetic risk factor for Alzheimer’s, and has no detectable amyloid in their brain in their seventies. The observation is anecdotal and the mechanism speculative. But it is the kind of clinical anomaly that points toward where the field may be heading.
The Genetics of Risk
The APOE gene encodes apolipoprotein E, a protein involved in lipid transport and, critically, in the clearance of amyloid from the brain. It comes in three common variants — ε2, ε3, and ε4 — and the ε4 allele is the most significant common genetic risk factor for late-onset Alzheimer’s disease. Carrying one copy of ε4 increases lifetime risk substantially; carrying two copies has long been understood to increase it further.
How much further became considerably clearer in 2024, when Juan Fortea and colleagues published an analysis of more than 10,000 individuals in Nature Medicine that reframed APOE4 homozygosity not as a risk factor but as a distinct genetic form of the disease. By age 55, virtually all ε4/ε4 carriers in the study showed abnormal CSF biomarkers. By age 65, nearly all had amyloid pathology detectable by PET scan. Lifetime risk of Alzheimer’s in ε4/ε4 carriers approached 60 percent by age 85, compared to roughly 9 percent in ε3/ε3 carriers. The authors argued that APOE4 homozygosity should be reclassified accordingly — not a predisposition, but a near-deterministic genetic trajectory (Fortea J et al., Nature Medicine, 2024).
This matters clinically in several ways. It changes the calculus of preclinical screening. It changes the risk-benefit analysis of anti-amyloid therapy. And it changes the conversation a clinician has with a patient who is, as Devi describes one of hers, a physician in her fifties with two copies of ε4 and a father whose Alzheimer’s diagnosis was confirmed by biomarker testing — not just clinical impression.
That patient came to Devi without symptoms. She came because she knew her genetics and she did not want to wait. A lumbar puncture showed low CSF amyloid — the signature of amyloid accumulating in the brain rather than clearing into the spinal fluid — along with borderline tau. Her cognitive scores were high overall, but her language performance gave Devi pause. They started her on a GLP-1 receptor agonist to address weight, and on a monoclonal antibody to begin clearing the amyloid that was already building in her brain. The goal was not to treat a disease that had declared itself. It was to intervene before it did.
The Biomarker Problem
The ability to detect Alzheimer’s pathology before symptoms appear is one of the genuine advances of the past two decades, and it has created a diagnostic problem that the field has not yet resolved.
Blood-based biomarkers — primarily the ratio of amyloid-beta 42 to amyloid-beta 40, and phosphorylated tau 217 — are now commercially available from multiple laboratories. A 2024 study in Nature Aging reported that plasma p-tau217 assays can achieve greater than 90 percent accuracy against amyloid PET in independent cohorts (Ashton NJ et al., Nature Aging, 2024). These tests are increasingly requested by patients and ordered by clinicians who may not specialize in memory disorders.
The problem is what a positive result means — or more precisely, what it does not mean. All blood-based biomarkers are validated primarily against amyloid PET imaging. And amyloid PET imaging detects amyloid. Amyloid, as it turns out, is present in the brains of a substantial proportion of cognitively normal older adults: roughly 25 percent of people in their seventies, more than 30 percent of people in their eighties, and approaching 44 percent of community-dwelling individuals by age 90 (Villemagne VL et al., The Lancet Neurology, 2013). Most of these people will never develop dementia.
This has produced a genuine schism in the diagnostic community. The Alzheimer’s Association’s revised 2024 criteria define Alzheimer’s disease primarily on biological grounds — the presence of amyloid pathology, whether detected in blood, CSF, or brain imaging, is sufficient for diagnosis even in the absence of symptoms (Jack CR et al., Alzheimer’s & Dementia, 2024). The International Working Group, a largely European consortium with some American participation, takes a different position: amyloid alone is not enough. Diagnosis requires amyloid plus tau plus clinical symptoms — a functional deficit that the biology is actually producing (Dubois B et al., JAMA Neurology, 2024). Under the IWG framework, a cognitively normal person with amyloid in their brain is not a patient. They are, as the group puts it, a “patient in waiting.”
Devi’s position is closer to the IWG’s, though she is careful about how she applies it. For most asymptomatic patients, she does not test. For patients with strong family histories and known genetic risk — particularly ε4/ε4 carriers — she does, because the pretest probability is high enough and the potential benefit of early intervention is real enough to justify it. But she is explicit that a positive blood test alone is not a diagnosis, and she has seen the harm that premature diagnosis can cause: a high-functioning man in his seventies who spent three weeks believing he had Alzheimer’s based on an abnormal blood test, until an amyloid PET scan came back negative.
Her preference, for patients in whom she has genuine concern, is to go beyond the blood test to either a tau and amyloid PET scan or a lumbar puncture. The spinal tap, she notes, has the advantage of detecting early tau changes that PET imaging misses, and it is less expensive than scanning. It is also, she acknowledges, something most patients would rather avoid — though she has found it useful to point out that any woman who has had an epidural has already experienced the reverse of the procedure.
Clearing the Plaques
The approval of aducanumab by the FDA in June 2021 was one of the most contested regulatory decisions in recent pharmaceutical history. The drug — a monoclonal antibody that clears amyloid plaques from the brain — had failed to demonstrate significant clinical benefit in two phase III trials, leading the FDA’s own advisory committee to vote overwhelmingly against approval. The FDA approved it anyway, under its accelerated approval pathway, on the grounds that amyloid clearance was a reasonable surrogate endpoint for clinical benefit. Several advisory committee members resigned in protest.
Devi was among the first clinicians to use it.
Her reasoning was not that the drug had proven clinical benefit — it hadn’t, not convincingly. It was that the drug did what it was supposed to do biologically, that the trials may have enrolled patients too late in the disease process for amyloid clearance to translate into meaningful cognitive improvement, and that if she herself had Alzheimer’s and understood the biology, she would want the drug. She also recognized immediately that the drug’s most significant liability — a greater than 40 percent incidence of amyloid-related imaging abnormalities, or ARIA, in clinical trials — was a problem that could be addressed through the speed of dose escalation.
ARIA is the collective term for two related complications: ARIA-E, edema caused by fluid leaking from blood vessels as the monoclonal antibody strips amyloid from arterial walls, and ARIA-H, microhemorrhages caused by actual rupture of those vessel walls. The mechanism is direct: the antibody enters the brain and begins clearing amyloid not only from the brain parenchyma but from the tunica media of small cerebral arteries, where amyloid deposits in a condition called cerebral amyloid angiopathy. This disrupts vascular integrity. In most cases the disruption is subclinical — detectable only on MRI — but in severe cases it can produce symptomatic neurological deficits, and in rare cases a massive cerebral hemorrhage.
APOE4 carriers are at substantially higher risk for ARIA, because they tend to have more severe cerebral amyloid angiopathy at baseline. ε4/ε4 homozygotes are at the highest risk of all — the same patients who most need the drug are the ones most likely to be harmed by it at standard dosing.
Devi’s response was to slow everything down. Amyloid has been accumulating in these patients’ brains for decades, she reasoned. There is no urgency that justifies rapid dose escalation. She developed an extended titration protocol, starting patients at doses far below the standard initiation dose and advancing them slowly over months rather than weeks. The result, published in the Journal of Alzheimer’s Disease in 2024, was an ARIA incidence of approximately 4 percent in ε4/ε4 patients — compared to the 35 to 40 percent rates reported in clinical trials (Devi G et al., Journal of Alzheimer’s Disease, 2024). Symptomatic ARIA, across more than five years of treating patients with monoclonal antibodies, has occurred in a single case.
The FDA has since approved a slower titration schedule for donanemab, the third drug in the class. Lecanemab, the second, begins at 10 mg/kg with no lower starting option — a protocol that insurance companies will reimburse and that Devi considers too aggressive for her highest-risk patients. She has had ε4/ε4 patients develop significant ARIA on just 3 mg/kg of lecanemab. The standard starting dose is more than three times that.
Aducanumab was withdrawn from the market in 2024, not because of safety but because the manufacturer could not sustain the economics of the required post-approval monitoring program. Lecanemab and donanemab remain available. Both have demonstrated statistically significant slowing of clinical decline in phase III trials — 27 percent for lecanemab on the CDR-SB scale at 18 months (van Dyck CH et al., NEJM, 2023), and 35 percent for donanemab in the low-to-medium tau subgroup (Sims JR et al., JAMA, 2023). The absolute differences are modest. Whether they are clinically meaningful for individual patients remains a genuine debate.
What Devi believes, based on her clinical experience, is that the trials have been measuring the wrong patients at the wrong time. The drugs were tested primarily in people with early symptomatic Alzheimer’s — mild cognitive impairment or mild dementia. By that point, tau pathology is already established, synaptic loss is already underway, and clearing amyloid may be too late to reverse the downstream damage. The more interesting question — whether clearing amyloid in a presymptomatic ε4/ε4 carrier in their fifties can prevent the disease from ever declaring itself — has not been answered, because no adequately powered trial has asked it.
When the Brain Improves
One of the things Devi did not believe, for most of her career, was that patients with confirmed Alzheimer’s disease could get better. The teaching was unambiguous: Alzheimer’s was a one-way decline. When she saw patients improve, she assumed she had made a diagnostic error.
Then, after 2007, when lumbar puncture-based biomarker testing made it possible to confirm Alzheimer’s pathology in living patients rather than at autopsy, she began to see something she could no longer explain away. Patients with confirmed amyloid and tau pathology were improving — sometimes with interventions as modest as structured cognitive exercises. The diagnosis was not wrong. The trajectory was.
She has a patient now, a woman in her seventies, who came to her after seeing a television program about Alzheimer’s. Cognitive testing placed her overall ability at the 70th percentile — not exceptional, but functional. A lumbar puncture confirmed amyloid and tau pathology. She was started on lecanemab. She cleared her amyloid. A subsequent tau PET scan showed no detectable tau. Devi is waiting on a follow-up lumbar puncture before drawing conclusions, but the clinical picture is remarkable: a woman who entered her practice with confirmed Alzheimer’s pathology and who now, by every available measure, appears to have none.
Devi is careful about what she claims. Individual cases are not evidence of a general principle. Immune system variation, the spectrum nature of the disease, the possibility that some patients have a biological resilience that is not yet understood — all of these complicate interpretation. But she no longer believes that improvement is impossible, and she believes it because she has seen it, repeatedly, in patients whose diagnoses were not in doubt.
The field’s 30 percent misdiagnosis rate in pre-biomarker clinical trials — the proportion of patients enrolled in Alzheimer’s drug studies who turned out, on autopsy or later biomarker verification, not to have had Alzheimer’s at all (Beach TG et al., Journal of Neuropathology & Experimental Neurology, 2012) — means that some of the historical “improvement” cases were almost certainly misdiagnoses. But not all of them. And the ones that weren’t are worth understanding.
The Other Dementias
Alzheimer’s disease dominates the public conversation about dementia, but it is not the only pathological process Devi manages, and in clinical practice the boundaries between conditions are rarely clean.
Vascular dementia — cognitive impairment driven primarily by cerebrovascular disease, strokes, and the accumulation of white matter damage — is theoretically a distinct entity, but Devi finds pure vascular dementia rarer than its prevalence statistics suggest. What she sees more often is vascular pathology as a comorbid contributor to Alzheimer’s: small vessel disease that interrupts the brain’s connective architecture and reduces its resilience to the amyloid and tau burden it is simultaneously carrying. She treats it aggressively — statins, blood pressure control, anticoagulation where indicated — and in patients with atrial fibrillation who are at high fall risk, she has advocated for left atrial appendage closure devices (the WATCHMAN procedure) as a way to eliminate the need for anticoagulation and thereby open the door to monoclonal antibody therapy.
Lewy body dementia is, in her view, the most frequently misdiagnosed condition in her practice. It shares its pathological substrate — abnormal aggregation of alpha-synuclein protein — with Parkinson’s disease, and the clinical distinction between the two rests on a somewhat arbitrary temporal criterion: if cognitive symptoms appear within a year of motor symptoms, the diagnosis is Lewy body dementia; if motor symptoms precede cognitive symptoms by more than a year, it is Parkinson’s disease. The distinction matters enormously for treatment, because the dopaminergic medications that are standard in Parkinson’s disease can worsen psychosis in Lewy body patients.
Lewy body dementia presents with a characteristic cluster: Parkinsonian motor features, fluctuating consciousness, cognitive impairment, and visual hallucinations that are often vivid and detailed. Crucially, patients with Lewy body disease typically retain insight into their hallucinations — they know, on some level, that what they are seeing is not real, even as they cannot stop seeing it. Alzheimer’s patients rarely have this metacognitive awareness. Devi has never seen a Lewy body patient present with the classic pill-rolling rest tremor of Parkinson’s disease, and she uses its absence as a clinical discriminator, though she acknowledges this is an empirical observation rather than an established diagnostic criterion.
Diagnosis involves dopamine transporter (DaT) scanning to assess basal ganglia dopamine uptake, and increasingly, skin punch biopsies to detect alpha-synuclein in cutaneous nerves — a technique that exploits the fact that the same protein pathology that accumulates in the brain travels along peripheral neural pathways into the skin. The more alpha-synuclein detected at sites progressively distant from the central nervous system, the more widespread the disease.
The prognosis for Lewy body dementia is better than its reputation. The common clinical teaching — that it is rapidly progressive and fatal within five to six years — is, Devi says flatly, not true. Her patients with Lewy body disease have life expectancies comparable to her Alzheimer’s patients, and some respond to treatment more dramatically. The key is getting the diagnosis right before someone has spent years on dopaminergic medications that have made them worse.
Menopause and the Brain
Women develop Alzheimer’s disease at higher rates than men, and the disparity persists even after adjusting for women’s longer average lifespan (Livingston G et al., The Lancet, 2024). The reasons are not fully understood, but one of the most compelling hypotheses centers on estrogen.
Estrogen receptors are co-localized throughout the brain, with particularly high density in the hippocampus — the structure most critical to memory formation. Estrogen drives synaptic sprouting in hippocampal tissue; its withdrawal produces measurable cognitive effects. Dominique Toran-Allerand at Columbia established the neurobiological basis for this in foundational work beginning in the 1980s (Toran-Allerand CD et al., Brain Research, 1980). Barbara Sherwin at McGill demonstrated the clinical consequences: women who underwent surgical menopause — abrupt estrogen withdrawal following oophorectomy — showed deficits in short-term memory, executive function, and verbal fluency that were reversible with estrogen replacement (Sherwin BB, Psychoneuroendocrinology, 1988).
Devi has coined a term for the syndrome she sees in her practice: menopause-related cognitive impairment, or MeRCI. It presents identically to early Alzheimer’s disease — word-finding difficulty, short-term memory lapses, executive dysfunction — and it can produce objective findings on cognitive testing and neuroimaging that are indistinguishable from early Alzheimer’s pathology. She has patients who were diagnosed with Alzheimer’s elsewhere and referred to her who turned out, on full evaluation, to have MeRCI. Treated with hormone replacement, they recovered.
The immunological dimension of women’s higher Alzheimer’s risk is less well characterized but worth noting. Women are substantially more susceptible to autoimmune conditions — multiple sclerosis, lupus, rheumatoid arthritis — suggesting immunological differences that may also influence neuroinflammatory risk. The cardiovascular dimension is also relevant: Alzheimer’s disease and cerebrovascular disease share virtually every risk factor, and women who survive cardiovascular events that kill men at younger ages carry that vascular burden into the decades when Alzheimer’s pathology is accumulating.
The Women’s Health Initiative, published in 2002, produced a generation of women who were denied hormone replacement therapy based on a misreading of its findings — a misreading that has been extensively documented and partially corrected in the years since, but not before those women spent the critical perimenopausal window without estrogen. Many of them are now in their seventies. The cognitive consequences of that lost decade of estrogen exposure are, Devi believes, showing up in her practice.
For women who cannot take estrogen — those with active hormone-sensitive cancers, or those for whom the risk-benefit calculation comes out differently — there are other approaches. Targeted cognitive exercises, structured to force engagement with the specific domains that are declining, can produce meaningful recovery through mechanisms analogous to constraint-induced movement therapy in stroke rehabilitation. Cholinesterase inhibitors, which increase acetylcholine availability in the brain, address the underlying neurotransmitter deficit that menopause-related cognitive impairment shares with Alzheimer’s disease; Devi’s own double-blind trial of donepezil versus placebo in postmenopausal women with cognitive impairment found a trend toward improvement in the treated group (Devi G, Obstetrics & Gynecology, 2018). Transcranial magnetic stimulation, targeted to specific cortical regions using MRI-guided neuronavigation, is a further option she has used since 2008.
The Multimodal Logic
Before the monoclonal antibodies arrived, Devi was already treating Alzheimer’s as a condition that required a cocktail of interventions rather than a single drug. She trained during the AIDS crisis, and the parallel has stayed with her: HIV was not cured by any single agent, but patients who would have died within months were kept alive and functional by combinations of drugs that addressed different aspects of the same disease. The logic applies.
Her current toolkit includes cholinesterase inhibitors and memantine for symptomatic management; antiviral therapy for patients with herpes virus exposure; immune-modulating treatments where indicated; aggressive management of vascular risk factors; GLP-1 receptor agonists for patients with obesity or metabolic dysfunction; and, for appropriate candidates, monoclonal antibody therapy. She uses MRI-guided TMS to maintain neuronal connectivity in circuits that are threatened by pathology, stimulating the dorsolateral prefrontal cortex, Broca’s area, the precuneus, and Wernicke’s area depending on the patient’s specific pattern of deficit.
The GLP-1 question is one she finds genuinely interesting. The observational data linking GLP-1 receptor agonist use to reduced dementia risk is favorable, but observational data in this area is confounded by the healthy user bias and by the fact that any drug producing weight loss and metabolic improvement will reduce Alzheimer’s risk through those mechanisms alone. The more interesting question is whether GLP-1 agonists offer a benefit through neuroinflammatory pathways independent of their metabolic effects — a possibility that animal model data supports but that has not been established in humans.
The IV immunoglobulin story is instructive about where the field may be heading. Relkin and colleagues at Cornell ran a phase III trial of pooled immunoglobulin in Alzheimer’s patients that failed to show benefit on primary endpoints (Relkin NR et al., Neurology, 2017). But the post-hoc subgroup analysis told a different story: patients carrying the APOE4 allele showed significantly better responses than those who did not. Two of Devi’s patients who received IV Ig — both with confirmed Alzheimer’s pathology by spinal tap — became amyloid-negative. One has remained stable for seventeen years.
The trial failed. The drug was not approved. But the signal in the APOE4 subgroup was real, and it points toward the same principle that Devi and Attia discuss at length: the heterogeneity of Alzheimer’s disease means that treatments which fail in unselected populations may succeed in precisely defined subgroups. The field has been treating Alzheimer’s the way oncology once treated cancer — as a single disease requiring a single approach — when the evidence increasingly suggests it is more like breast cancer, where ER-positive, HER2-positive, and triple-negative subtypes are, for practical purposes, different diseases requiring different treatments.
What the Next Decade Looks Like
Twin studies have established that the heritability of Alzheimer’s disease is high — roughly 79 percent in the Swedish Twin Registry data (Gatz M et al., Archives of General Psychiatry, 2006) — but that identical twins can diverge in onset by a decade or more. Two people with the same genome, raised in the same environment, carrying the same genetic risk, can have dramatically different disease trajectories. The difference lies in the accumulated effects of inflammation, epigenetics, metabolic health, viral exposures, and factors not yet identified.
This is simultaneously humbling and hopeful. It means the disease is not simply written in the genome. It means there is room for intervention. And it means that the precision medicine framework that has transformed oncology — where treatment is matched to the molecular subtype of the disease in the individual patient, not to the average patient in a clinical trial — is not just applicable to Alzheimer’s but may be the only framework adequate to its complexity.
Devi’s vision for the next decade centers on two developments. The first is AI-assisted early detection: tools that can identify subtle changes in speech patterns, cognitive performance, and behavioral markers years before clinical symptoms appear, enabling intervention at the point where it is most likely to matter. The second is a new generation of anti-amyloid and anti-inflammatory therapies — drugs that can be taken orally, administered at home, and titrated in ways that minimize vascular risk — that make the current intravenous monoclonal antibody protocols look like a first draft.
What she believes now, that she did not believe thirty years ago, is that the trajectory of Alzheimer’s disease is not fixed. She has seen patients with confirmed pathology stabilize. She has seen patients improve. She has seen a woman in her seventies whose tau scan, after treatment, showed nothing. She is not claiming a cure. She is claiming something more modest and, in the context of this disease’s history, more radical: that what happens next is not already decided.
Sources
- Fortea J, Pegueroles J, Alcolea D, et al. “APOE4 homozygosity represents a distinct genetic form of Alzheimer’s disease.” Nature Medicine. 2024. https://doi.org/10.1038/s41591-024-02931-w
- van Dyck CH, Swanson CJ, Aisen P, et al. “Lecanemab in Early Alzheimer’s Disease.” New England Journal of Medicine. 2023;388(1):9–21. https://doi.org/10.1056/NEJMoa2212948
- Sims JR, Zimmer JA, Evans CD, et al. “Donanemab in Early Symptomatic Alzheimer’s Disease.” JAMA. 2023;330(6):512–527. https://doi.org/10.1001/jama.2023.13239
- Eyting M, Xie M, Heier M, et al. “Causal evidence that herpes zoster vaccination prevents a proportion of dementia cases.” Nature. 2023. https://doi.org/10.1038/s41586-023-06321-3
- Tzeng NS, Chung CH, Lin FH, et al. “Anti-herpetic Medications and Reduced Risk of Dementia in Patients with Herpes Simplex Virus Infections.” Neurotherapeutics. 2018;15(2):417–429. https://doi.org/10.1007/s13311-018-0611-x
- Dominy SS, Lynch C, Ermini F, et al. “Porphyromonas gingivalis in Alzheimer’s disease brains: Evidence for disease causation and treatment with small-molecule inhibitors.” Science Advances. 2019;5(1):eaau3333. https://doi.org/10.1126/sciadv.aau3333
- Jack CR Jr, Andrews JS, Beach TG, et al. “Revised criteria for diagnosis and staging of Alzheimer’s disease: Alzheimer’s Association Workgroup.” Alzheimer’s & Dementia. 2024;20(8):5143–5169. https://doi.org/10.1002/alz.13859
- Dubois B, Villain N, Frisoni GB, et al. “Alzheimer Disease as a Clinical-Biological Construct — An International Working Group Recommendation.” JAMA Neurology. 2024. https://doi.org/10.1001/jamaneurol.2024.1826
- Livingston G, Huntley J, Liu KY, et al. “Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission.” The Lancet. 2024;404(10452):572–628. https://doi.org/10.1016/S0140-6736(24)01296-0
- Ashton NJ, Brum WS, Di Molfetta G, et al. “Diagnostic accuracy of a plasma phosphorylated tau 217 immunoassay for Alzheimer disease pathology.” Nature Aging. 2024. https://doi.org/10.1038/s43587-024-00659-1
- Heneka MT, Carson MJ, El Khoury J, et al. “Neuroinflammation in Alzheimer’s disease.” The Lancet Neurology. 2015;14(4):388–405. https://doi.org/10.1016/S1474-4422(15)70016-5
- Villemagne VL, Burnham S, Bourgeat P, et al. “Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: a prospective cohort study.” The Lancet Neurology. 2013;12(4):357–367. https://doi.org/10.1016/S1474-4422(13)70044-9
- Gatz M, Reynolds CA, Fratiglioni L, et al. “Role of genes and environments for explaining Alzheimer disease.” Archives of General Psychiatry. 2006;63(2):168–177. https://doi.org/10.1001/archpsyc.63.2.168
- Relkin NR, Thomas RG, Rissman RA, et al. “A phase 3 trial of IV immunoglobulin for Alzheimer disease.” Neurology. 2017;88(18):1768–1775. https://doi.org/10.1212/WNL.0000000000003904
- Devi G. “A how-to guide for a precision medicine approach to the diagnosis and treatment of Alzheimer’s disease.” Frontiers in Aging Neuroscience. 2023. https://doi.org/10.3389/fnagi.2023.1101453
- Devi G. “Menopause-Related Cognitive Impairment.” Obstetrics & Gynecology. 2018. https://doi.org/10.1097/AOG.0000000000002848
- Devi G. “The Certain Uncertainty of an Alzheimer Disease Diagnosis.” Annals of Internal Medicine. 2026.
- Devi G, et al. “Slow Titration of Anti-Amyloid Monoclonal Antibodies Reduces the Risk of Amyloid-Related Imaging Abnormalities.” Journal of Alzheimer’s Disease. 2024.
- Sherwin BB. “Estrogen and/or androgen replacement therapy and cognitive functioning in surgically menopausal women.” Psychoneuroendocrinology. 1988;13(4):345–357. https://doi.org/10.1016/0306-4530(88)90060-1
- Toran-Allerand CD, Gerlach JL, McEwen BS. “Autoradiographic localization of alpha-estradiol binding in the developing rat brain and pituitary.” Brain Research. 1980;184(2):517–522. https://doi.org/10.1016/0006-8993(80)90819-3
- Beach TG, Monsell SE, Phillips LE, Kukull W. “Accuracy of the clinical diagnosis of Alzheimer disease at two state psychiatric hospitals.” Journal of Neuropathology & Experimental Neurology. 2012;71(4):266–273. https://doi.org/10.1097/NEN.0b013e31824b211b
- Schneider JA, Arvanitakis Z, Bang W, Bennett DA. “Mixed brain pathologies account for most dementia cases in community-dwelling older persons.” Annals of Neurology. 2007;62(6):616–623. https://doi.org/10.1002/ana.21sprache
- de la Torre JC. “Alzheimer disease as a vascular disorder: nosological evidence.” Stroke. 2002;33(4):1152–1162. https://doi.org/10.1161/01.STR.0000014421.15948.67
- O’Shea D, Dukart J, Bhatt P, et al. “Practical use of DAT SPECT imaging in diagnosing dementia with Lewy bodies: a US perspective of current guidelines and future directions.” Frontiers in Neurology. 2024. https://doi.org/10.3389/fneur.2024.1362547
