Did the World Just Gain Hundreds of Millions of Alzheimer’s Patients?
This July, some ten thousand people descended on London for the Alzheimer’s Association International Conference. Another three thousand of us joined virtually. I’m an anthropologist, accustomed to conferences where croissants at coffee breaks count as a luxury. This one opened with six trumpeters dressed as Buckingham Palace guards, followed by dancers who discoed, vogued, and high-kicked through the history of British pop. After the first day’s sessions, participants dispersed to parties; at one, singers in sequins belted out covers while someone played an illuminated, transparent violin. One night, the London Eye was lit purple to honor the people in attendance.
The mood matched the spectacle. Joanne Pike, the Alzheimer’s Association’s president and C.E.O., announced that recent advances had brought the campaign against the disease to a “pivotal moment.” Maria Carrillo, its chief science officer, hailed a groundswell of progress.
That optimism followed decades of disappointment. Between 1974 and 2020, the U.S. government spent roughly twenty billion dollars on Alzheimer’s research without seeing the approval of a single treatment designed to act on the disease’s underlying biology, let alone stop or reverse its course. Today, Alzheimer’s afflicts roughly one in nine Americans over sixty-five and, along with related dementias, has been estimated to cost the economy about three-quarters of a trillion dollars a year. In surveys, it ranks with cancer as the disease people fear most.
That’s understandable. Alzheimer’s dismantles the self by degrees. It typically starts by robbing patients of recent memories—conversations they just had, where they put important objects—before corroding the basic functions of adult independence, from paying bills to managing medications. Eventually, it leaves patients unable to speak, move, swallow, or recognize close relatives. “We estimate a man by how much he remembers,” Ralph Waldo Emerson wrote, about a decade before an Alzheimer’s-like dementia left him scarcely able to write his name.
But at the conference many seemed to think that the disease could soon join scurvy and smallpox in the annals of yesterday’s afflictions. That excitement is widely shared. In June, Bill Gates had posted about “a golden age of Alzheimer’s research.”
These high hopes rest largely on two developments. First: drugs. Since 2021, the Food and Drug Administration has approved three antibody treatments meant to slow the disease’s progression. All target amyloid, a protein fragment that can accumulate into the plaques that have long been associated with Alzheimer’s and that many researchers believe plays an important role in its development. Two drugs—lecanemab, sold as Leqembi, and donanemab, sold as Kisunla—are on the market and are covered by Medicare.
The second development concerns diagnosis. Until recently, measuring amyloid levels in living patients required expensive PET scans or invasive spinal taps. Blood tests developed over the past five years approach the accuracy of those methods. The F.D.A. cleared the first such test in May, 2025; three more have followed. “For the first time,” Gates wrote, “we have simple blood tests that tell you whether you have Alzheimer’s.”
A disease that could once be confirmed only after death is now, in Gates’s telling, as easy to detect as high cholesterol. Combined with drugs that promise to slow deterioration, such tests are raising hopes of achieving a long-sought goal: treating prospective patients early enough that symptoms never appear. The result has been to give new urgency to one of the most fiercely debated questions in contemporary medicine: What, exactly, is Alzheimer’s disease?
At Donald Kewman’s home, outside Sacramento, there’s a sticker that reads “ALL ARE WELCOME” beside the front door. Inside, the living room shows signs of having been colonized by grandchildren. Bookshelves and a Persian-style rug share space with toy trucks, a purple plastic kitchen, and an easel with a basketball hoop attached to its back. Kewman, a seventy-six-year-old retired clinical psychologist, is “supremely good with little kids,” his wife, Sandy, told me—“willing to voice all the stuffies in ways that they prefer.” Kewman told me, “I like babies.” They were attending a grandchild’s eighth-birthday party later that day.
Kewman was adopted as an infant, and his concerns about Alzheimer’s began in 2016, when he signed up for genetic testing through 23andMe. When his results came back, he was given the option of learning his genetic risk for late-onset Alzheimer’s. “Of course, being a curious guy, I said yes,” he told me, his hazel eyes alert behind brow-line rectangular glasses. It turned out that he carried two copies of a gene called APOE4, making him much more likely than the average person to develop Alzheimer’s dementia. “It took a while to sort of absorb that information,” he said. Several years later, he was informed that his brain contained a moderate accumulation of amyloid plaques.
Alzheimer’s, he told me, is “at the bottom of my desirable ways to die.” A Quaker from the age of seventeen, he worries less about what it might do to him than about what it might mean for his loved ones. He and Sandy do not have long-term-care insurance, and he knows that the disease can impose “a terrible burden on family,” one that “can deplete your assets pretty dang fast.”
Sandy spent more than twenty-five years as an intensive-care nurse and understands the difficulties that could come with the disease. Yet Kewman remains cognitively normal. “We’ve taken care of our bucket lists and keep making bucket lists,” she said. When I asked him whether he thought of himself as having Alzheimer’s, he paused. “It’s a good question,” he said. “I have some of the changes associated with Alzheimer’s disease.” He meant the amyloid deposits. “So, in some sense, I do.” He quickly qualified his answer: many people with those changes never develop the “full-blown disease,” with its deterioration of “memory, behavior, functioning, independence.”
I asked Sandy the same question. “No,” she said. Not long ago, most physicians would have agreed.
Since antiquity, people have known that old age can bring lapses of memory and judgment. Alzheimer’s disease, however, was named barely a century ago. Coca-Cola and the automobile predate it, and its boundaries have repeatedly been contested and rewritten. The story conventionally begins on November 26, 1901, in an asylum in Frankfurt, where the physician Alois Alzheimer examined Auguste Deter, a housewife in her early fifties who had undergone a rapid mental decline. She had become suspicious of her neighbors and accused her husband of adultery. A photograph of her from around this time suggests an overwhelming sense of weariness. Deep lines mark her forehead as her heavy-lidded eyes gaze down. Her thin hands clutch each other over a white tunic. What most struck Alzheimer was Deter’s “utter perplexity.” His notes preserve their exchanges:
At times, Deter seemed to recognize what was happening to her. “So anxious, so anxious,” she said when Alzheimer asked whether his questions were difficult. Told to write the number 8, she repeated, “I have lost myself, so to say.”
Deter never left the asylum. Her speech deteriorated into shouts and hums and eventually almost disappeared. She died in 1906, roughly five years after Alzheimer first saw her. By then, he had moved to Munich, but he had her brain sent to him. Examining it, he detected three abnormalities: deposits of a substance, later identified as amyloid beta; tangles of fibrils inside neurons, now known to be composed largely of tau proteins; and a reduced number of neurons. In Deter, features that would become closely associated with Alzheimer’s appeared together: amyloid plaques, tau tangles, and profound cognitive decline. Both amyloid beta and tau are normally present in healthy brains; the plaques and tangles are abnormal aggregations of these molecules.
At the time, however, Alzheimer’s disease did not mean what it does today. When Emil Kraepelin, Alzheimer’s supervisor, named the condition in the 1910 edition of his influential psychiatry textbook, he conjectured that it differed from dementia, then the catchall term for cognitive decline in old age. Although Kraepelin recognized that plaques, tangles, and neuronal loss could also occur in the ordinary dementia of old age, he noted that Alzheimer’s patients were unusually young, sometimes in their forties or fifties, and that their illness could take an especially severe course. For much of the twentieth century, “Alzheimer’s disease” was therefore virtually synonymous with “presenile dementia.”
It was not until the nineteen-seventies, more than sixty years after Kraepelin named Alzheimer’s disease, that it came to encompass late-onset dementia. The figure most responsible for this was Robert Katzman, a reserved, soft-spoken neurologist who chaired the department at the Albert Einstein College of Medicine, in the Bronx. In the nineteen-sixties, his mother-in-law developed confusion and memory problems. She was sixty-four—just below the conventional cutoff for senile dementia—and so she was diagnosed with “presenile dementia,” or Alzheimer’s disease. The line struck Katzman as suspiciously arbitrary. “It became obvious that Alzheimer’s disease was a single entity regardless of age of onset,” he later recalled.
Medicine had a powerful model for deciding that apparently different illnesses were really one disease: identify a common cause. In the nineteenth century, the physician Robert Koch showed that pulmonary consumption, intestinal disease, and swollen lymph nodes could all result from the same bacterium, Mycobacterium tuberculosis. In a two-page editorial published in 1976, Katzman argued that some instances of early- and late-onset dementia should be grouped together as Alzheimer’s disease. His evidence was chiefly descriptive. Both could involve plaques and tangles. He could not show that a single causal process produced both conditions, as Koch had done with tuberculosis. But the shared features, he maintained, were enough.
The consequences were enormous. A condition once considered rare now encompassed most forms of old-age dementia. Katzman himself came to regard the editorial as the most important paper he had written. The shift was institutional as well as scientific. In 1974, Congress had created the National Institute on Aging, as part of the National Institutes of Health, but it had struggled for attention. Aging was diffuse and abstract, a difficult cause around which to mobilize resources, in the manner of Richard Nixon’s War on Cancer. But Katzman’s enlarged conception of Alzheimer’s provided a similarly compelling foe. In the opening paragraph of his 1976 editorial, he described the disease as a societal “malignancy” that ranked among the country’s leading causes of death. “We had to make it a household word,” Robert Butler, the first director of the National Institute on Aging, recalled, because funding didn’t flow toward vague biological processes. “People don’t die from basic research, they don’t suffer from basic research. They suffer from specific diseases.”
In 1977, Katzman, Butler, and others convened a conference at the N.I.H., in Bethesda, in part to settle the question of classification. Katzman remarked that he had expected “more argument,” but the participants seemed to agree “that Alzheimer’s disease and senile dementia are one.” When the group discussed what to recommend to the American Psychiatric Association for the next edition of the Diagnostic and Statistical Manual of Mental Disorders, one attendee asked, “Does anybody here seriously believe they are two different diseases?” Three years later, the Alzheimer’s Association was founded; four years after that, the new definition of the disease was formally codified, when six experts, including Katzman, published the first widely accepted diagnostic criteria for the disease.
These criteria remained centered on symptoms. A diagnosis of “probable Alzheimer’s disease” depended primarily on progressive cognitive decline, particularly memory loss. Plaques and tangles could confirm it, but only after death. Although the diagnosis covered a much larger population, a living patient still had to show cognitive impairment to qualify. Back then, researchers would have had no way of knowing about Kewman’s amyloid deposits. All they would have seen was a cognitively normal person—by their criteria, someone without Alzheimer’s.
Then, in 1984, researchers at the University of California, San Diego, isolated the main component of the plaques that Alzheimer had first observed: aggregations of the peptide now known as amyloid beta. Months later, the same group identified a similar protein fragment in the brain of a person with Down syndrome, a condition linked to an unusually high risk of Alzheimer’s-like pathology and dementia.
The discoveries helped elevate amyloid from a hallmark of the disease into a possible culprit. In 1992, the molecular biologists John Hardy and Gerald Higgins proposed what became known as the “amyloid-cascade hypothesis.” Amyloid, they argued, was the “causative agent of Alzheimer’s disease.” Its accumulation, in their account, triggered tau abnormalities, killed neurons, and produced dementia.
The hypothesis soon became the leading model of Alzheimer’s. Its champions, branded Baptists for their emphasis on beta-amyloid deposits, marshalled plenty of supporting evidence. Rare inherited mutations that virtually guaranteed early-onset Alzheimer’s affected the production of amyloid. People with Down syndrome carried an extra copy of the gene for the amyloid precursor protein. These observations suggested that when amyloid accumulated more readily, Alzheimer’s would follow. The therapeutic strategy seemed obvious: remove amyloid. If that worked, a treatment would be of enormous commercial value.
Promising evidence emerged from studies in mice. Scientists inserted genes associated with Alzheimer’s into mouse genomes, breeding animals that had high levels of amyloid in their brains; the mice then began to struggle with spatial navigation. In 1999, researchers at Elan Pharmaceuticals reported that they had vaccinated mice against amyloid beta, provoking their immune systems to clear it. Plaque burdens plummeted. Howard Feldman, a neurologist at the University of California, San Diego, remembers thinking, “If this works out, it’s a Nobel Prize.”
The excitement spread widely. “A cure for Alzheimer’s might be within the grasp of researchers,” New Scientist announced. Elan quickly began testing its vaccine in people. In 2002, however, the trial was halted after some participants developed meningoencephalitis, a dangerous inflammatory condition.
Meanwhile, drugmakers explored a related approach: supplying the antibodies directly. They infused patients with laboratory-made monoclonal antibodies, or mAbs, designed to bind to amyloid and help clear it. The approach would eventually produce lecanemab and donanemab.
For years, though, it failed. The first anti-amyloid antibody tested in large clinical trials appeared to affect amyloid but did not slow cognitive decline. MRIs also revealed, in some patients, brain swelling and microhemorrhages. (These were given the benign-sounding name “amyloid-related imaging abnormalities,” or ARIA—which feels as abstruse as calling a knife wound a “blade-related imaging abnormality.”) Other antibodies, such as gantenerumab and crenezumab, were also abandoned, after failing to produce clinical benefits. Researchers had learned how to attack one of Alzheimer’s defining pathologies without meaningfully altering the illness.
The next problem was measurement: How could investigators be sure that trial participants actually had amyloid, or determine how much an antibody affected it? In 2004, the N.I.H., pharmaceutical companies, and nonprofit groups launched the Alzheimer’s Disease Neuroimaging Initiative, in part to find biomarkers that could track the disease. With funding from the Alzheimer’s Association, the project adopted a new tracer that, in combination with a PET scan, could measure amyloid burdens. Between 2007 and 2009, more than a hundred participants underwent amyloid scans.
The results were surprising. Some people with classic Alzheimer’s dementia had little or no amyloid, while substantial amyloid deposits appeared in certain people who remained cognitively normal. “That was the revolutionary moment for me,” Carrillo, the chief scientific officer of the Alzheimer’s Association, told me.
One way of interpreting the findings was that amyloid had been assigned too large a role in Alzheimer’s. Carrillo eventually drew a very different conclusion. The negative trial results, she thought, showed not that targeting amyloid was futile but that researchers had been treating the wrong people at the wrong time. She began to think, “We’re too late.” Amyloid seemed to accumulate years before dementia appeared, so perhaps drugs were needed in this presymptomatic period. And the fact that many participants in amyloid-targeting-drug trials turned out not to have amyloid at all? To her, it demonstrated that patients could look as though they had Alzheimer’s and yet, biologically speaking, have something else.
Reisa Sperling, a Harvard neurologist, encountered the same puzzle from another angle. She had been studying the brain networks involved in memory and found that roughly a third of the cognitively normal older people she’d examined had amyloid plaques. Some of them showed patterns of brain activity like those of people with symptoms of Alzheimer’s. To Sperling, the scans suggested that Alzheimer’s could already be altering the brain before memory problems appeared. She remembers sitting at her computer at three in the morning, comparing images and thinking, “Oh, my gosh, oh, my gosh.”
Cases like Kewman’s forced the issue. In 2011, the N.I.A. and the Alzheimer’s Association published new recommendations. One paper, led by Sperling and co-authored by Carrillo, proposed a category that would once have sounded contradictory: “preclinical Alzheimer’s disease.” A cognitively normal person with the relevant biological abnormalities could now be classified as being in an early stage of the disease. “It seemed heretical,” Sperling told me. The authors were cautious. They described it as a research category, not a clinical diagnosis, and acknowledged that no “firm link” had yet been established between the biomarkers and later decline. But the conceptual break had occurred.
In 2018, another N.I.A.-Alzheimer’s Association working group went further, defining the disease as a “biological construct,” identifiable by evidence of amyloid plaques and tau tangles, independent of a person’s cognitive state. In 2024, criteria published by the Alzheimer’s Association went further still. Now amyloid alone was sufficient to establish the disease, and the criteria were meant to help inform clinical diagnosis. If a blood test could reliably detect amyloid, then a person with no memory problems could, in principle, have a needle stick and emerge with an Alzheimer’s diagnosis.
Not everyone accepted this. The National Institute on Aging, which had issued the 2011 and 2018 recommendations jointly with the Alzheimer’s Association, dropped its name from the 2024 criteria, citing an N.I.H. policy against endorsing private entities or services. The American Geriatrics Society, in response to a draft of the guidelines, called their potential use in clinical practice “premature.” (The final criteria do state that they are not intended to provide “step-by-step clinical practice guidelines.”) The society granted that biomarkers can define diseases, as with certain cancers, before symptoms appear, but only when their meaning is sufficiently well understood. Amyloid, it argued, had not reached that point. “As of now,” the group stated, “there is no evidence that removing amyloid helps a cognitively normal person who is biomarker positive.”
A related dispute, of intense interest to drugmakers, concerned what should count as evidence that a treatment worked. Was it enough for a drug to reduce amyloid? Or did it have to affect a patient’s ability to remember and function?
Their most promising candidate was aducanumab, an antibody licensed by Biogen which aggressively stripped amyloid from the brain. In 2016, Nature put before-and-after PET scans from an early study on its cover, showing that, as the dose increased, bright patches marking amyloid seemed to melt away. The study was small and had not been designed to establish whether aducanumab preserved cognition; in addition, more than a quarter of participants experienced brain swelling or microbleeds. But there was also a tantalizing suggestion that patients receiving higher doses had a slower decline in mental functioning. Around this time, Biogen’s market capitalization rose by twenty-nine billion dollars over the course of three months.
In March, 2019, Biogen and a Japanese partner, Eisai, halted two large trials of aducanumab after an independent analysis concluded that the drug wasn’t working. Biogen’s share price plunged. Seven months later, however, the company reversed itself. Newer data had been analyzed, producing an almost bewildering result: one of the two nearly identical trials remained negative, but, in the other, patients receiving a high dose appeared to decline more slowly. Biogen decided to seek F.D.A. approval.
The F.D.A. convened an advisory committee that included eleven outside experts. Aaron Kesselheim, a physician and a professor at Harvard Medical School, spent two weeks studying the conflicting results before the committee met, remotely, on November 6, 2020. He expected the meeting to revolve around the uncertainty. Instead, the F.D.A.’s biostatisticians, who seemed to have reservations about Biogen’s interpretation, appeared only in a prerecorded presentation and a Q. & A. (A spokesperson for the F.D.A. said that the team’s participation was typical of similar meetings.) Otherwise, Kesselheim said, “the F.D.A.’s presentation was about how the efficacy of the drug is obvious, and the company’s presentation was about how the efficacy of the drug was obvious.” He recalled thinking, “Well, it’s not obvious to me.”
The other outside experts largely agreed with him. The committee was asked whether the positive trial, considered alongside the other analyses presented, provided evidence that aducanumab worked. Ten members voted no. One was uncertain. None voted yes.
Seven months later, the F.D.A. granted the drug “accelerated approval” anyway. This status allows a drug for a serious illness to reach the market on the basis of a surrogate measure considered reasonably likely to predict clinical benefit. Here, the surrogate was the reduction of amyloid plaques.
The decision was explosive. Kesselheim resigned from the advisory committee. In a Times opinion piece, he and a colleague accused the agency of having “shifted the goal posts.” Subsequent reporting uncovered an unusually close relationship between Biogen and Billy Dunn, the head of the F.D.A.’s neuroscience office, who had met with the company ahead of the drug’s approval. A congressional investigation documented “atypical collaboration” between the agency and the company. The potential financial consequences were enormous. According to the investigation, Biogen estimated that the company would be able to bill Medicare twelve billion dollars a year for the drug.
In fact, aducanumab, sold as Aduhelm, was tainted by the controversy and never achieved broad use. Medicare restricted coverage to clinical trials, and Biogen discontinued the drug in 2024. But the principle behind its approval survived. In January, 2023, the F.D.A. granted accelerated approval to Biogen and Eisai’s lecanemab, again “on the basis of a reduction in amyloid beta plaque.”
What has been hard to miss is the way pharmaceutical companies have been orbiting around the redefinition of Alzheimer’s. Of the twenty authors of the 2018 framework, twelve disclosed professional relationships with Biogen, Eisai, or Lilly, the companies whose anti-amyloid antibodies would later reach the F.D.A. One author, Samantha Budd Haeberlein, was a Biogen executive involved in developing aducanumab. Another, Eric Siemers, had been a medical director at Lilly. A third was Billy Dunn.
None of those relationships establishes why the authors favored a biological definition of Alzheimer’s; there were scientific arguments for the position. But the question of where risk ends and disease begins leaves room for interpretation. Lilly and Biogen have both sent comments to the F.D.A. on draft guidelines for the pharmaceutical industry, encouraging the agency to categorize symptomatic people with positive biomarkers as having “preclinical” Alzheimer’s. Both have regularly been “platinum sponsors” of the conferences organized by the Alzheimer’s Association.
One morning near the end of summer, Kewman handed me a pair of binoculars and led me from his house into nearby wetlands and open spaces. He carried a camera and wore a sun hat adorned with a peacock feather. At seventy-six, he set a brisk pace.
Not long into the walk, he stopped. “Belted kingfisher,” he said, directing my gaze toward a tree across the water. “Around eleven o’clock. About three-quarters of the way up.” I raised the binoculars and eventually found it.
Kewman walks about three miles most mornings, usually along the same network of paths. Familiarity had given him a sense of what to expect, and he paid attention to departures from the routine. White pelicans, he told me, had lingered unusually long that year. He knew where the beavers built their lodges and where the herons fished. As we walked, he picked individual birdcalls out of the mixture of natural and suburban noise. I depended on him to tell me what I was hearing.
Other walkers passed us on the paths; he knew most by name. There was something faintly absurd about accompanying him through this place while trying to understand whether he had a disease associated with losing one’s bearings.
Howard Feldman, the U.C. San Diego neurologist who once thought an amyloid vaccine might garner a Nobel Prize, has spent nearly two decades helping develop diagnostic criteria through the International Working Group on Alzheimer’s Disease, or I.W.G., which helped pioneer the use of biomarkers to identify the disease. Feldman told me that biomarkers generally improve diagnostic accuracy for patients already experiencing cognitive problems. The I.W.G. would describe Kewman’s amyloid as a risk factor for Alzheimer’s, rather than evidence of the disease itself.
Whether an amyloid-positive person eventually develops dementia depends on many factors, including the amount and distribution of tau pathology, vascular health, and age. Feldman favors individualized risk assessments rather than making amyloid deposits alone the basis of a diagnosis. Otherwise, he said, people may “experience the full brunt” of an Alzheimer’s diagnosis without ever developing Alzheimer’s dementia.
During our walk, Kewman and I came upon a flock of California quail beside the trail. He pointed out a male strutting among them—“a sentinel” keeping watch while his mate stayed close to cover. Birding, he said, was “a very soothing, almost spiritual thing.” On a morning like this, the wish to preserve such moments was easy to understand, as was the hope that animated researchers who favored the biological definition. The difficulty was knowing whether Kewman was likely to develop dementia and whether treating the amyloid in his brain would make it any less likely.
The I.W.G. and the Alzheimer’s Association have divided the field into two camps, but the camps agree about many of the underlying facts. Sperling, the Harvard neurologist, told me that separating the world into amyloid-positive and amyloid-negative people is “simplistic.” She, too, thinks risk depends on tau, vascular disease, and other factors, including the amount of amyloid in brain tissues. She acknowledges that researchers still cannot explain why some people accumulate substantial amyloid without developing dementia. Yet, for Sperling, the disease process has already begun in anyone with amyloid deposits. From a treatment perspective, she said, “to pretend that they don’t have Alzheimer’s disease in the brain, to me, doesn’t make sense.”
Carrillo, the chief scientific officer of the Alzheimer’s Association, was even more direct when I asked her for the strongest argument for defining the disease through biomarkers. “Treatment,” she said. She echoed Feldman’s observation that biomarkers help physicians distinguish Alzheimer’s from other causes of cognitive decline and deliver more appropriate care. But, she pointed out, by the time a person has dementia, extensive neurodegeneration has occurred. “We can’t bring those cells back to life,” Carrillo said. “You can’t get those memories back.” If drugs are ever going to prevent Alzheimer’s dementia, rather than merely slow it, they may have to be administered while people look much as Kewman did on our walk—training binoculars on a distant bird, with no outward sign of the process the drug is meant to stop.
“We don’t understand—I don’t, and the authors of the criteria don’t understand—how you can treat a person ‘at risk,’ ” Carrillo told me. “You’re treating a disease. You’re not treating a person at risk.” Doctors, she noted, do not wait for a heart attack before treating high cholesterol. Why wait for dementia before treating the biological processes thought to cause it?
But lowering cholesterol has been shown to prevent heart attacks. No treatment has yet been shown to prevent decline in cognitively normal amyloid-positive people like Kewman. There are trials under way of the two new F.D.A.-approved antibodies that are meant to test that possibility, but no positive results have yet been reported.
The biomarker definition works both ways. Just as asymptomatic people were drawn into the disease, some patients with Alzheimer’s symptoms were kept out. “Why are we diagnosing people with Alzheimer’s dementia,” Carrillo asked, “when they don’t have Alzheimer’s because they don’t have amyloid?” There was a good experimental reason to exclude amyloid-negative patients from an anti-amyloid trial. But using that logic to define the disease risks begging the question: if dementia without amyloid never counted as Alzheimer’s, evidence that might complicate the amyloid theory would disappear by fiat.
Carrillo also offered a practical argument for the disease label. Calling an asymptomatic person merely “at risk,” she said, “doesn’t work for the U.S. in terms of approvals or coverage.” The F.D.A. and the Centers for Medicare and Medicaid Services, she told me, “do not provide coverage for at-risk drugs. . . . F.D.A. approves drugs for a disease.” The Alzheimer’s Association, she added, worked closely with both agencies and therefore had “a better understanding of some of those regulatory processes” than some other researchers did.
I was surprised to hear Carrillo invoke the rules governing approval and coverage in justifying the new definition of Alzheimer’s. As it happens, her account of those rules was mistaken. (A spokesperson for the association clarified that the F.D.A. can and does approve drugs to reduce disease risk, and that regulatory considerations are a consequence of defining the disease by biomarkers, not a reason for doing so.) Still, the exchange revealed how entwined the debate about diagnosis had become with the development and marketing of treatments.
For all that, Carrillo herself stopped short of recommending that cognitively normal people be given an Alzheimer’s diagnosis today. “There’s no need,” she told me, “because there’s nothing actionable.” The biological criteria, she said, were meant, in part, to encourage trials in asymptomatic people, which would help determine “if we were on the right path or not.”
Well in advance of any such results, however, the definition has made an enormous population eligible for the label. One in four cognitively unimpaired people aged sixty-five and older is thought to have amyloid accumulation in the brain. Feldman estimated that there are hundreds of millions of such individuals worldwide. But only five to ten per cent of them will develop Alzheimer’s dementia in the next five years. The proportion who will develop it over the rest of their lives is harder to pin down, but published studies generally put it at a quarter to a third. In other words, most people who might now qualify as having Alzheimer’s disease will never develop the symptom that once defined it.
Kewman, with his two copies of APOE4, faces higher odds than those population figures suggest. For him, the more urgent question is what his biomarkers actually say about what will happen to him. Blood tests don’t answer the question that Don and Sandy Kewman care about most: Will he develop dementia?
Soon after receiving his genetic results, Kewman began looking for clinical trials. It was a way, he told me, “to turn something that’s maybe not so positive into something positive, and contribute.”
The first study he joined, at Stanford, was testing an experimental drug intended to prevent amyloid buildup in cognitively normal people. The trial was halted early, though, when results indicated that the drug was worsening people’s cognition. Kewman later learned that he had been receiving a placebo.
His study coördinator told him about another trial. It aimed to determine whether lecanemab could prevent or delay cognitive decline in people with brain amyloid but no evident impairment. Joining took months. Kewman underwent blood work, MRIs, a physical exam, psychological testing, and two PET scans. Eventually, a neurologist and his study coördinator sat down with him and told him that his brain contained a moderate amount of amyloid buildup. The coördinator had tears in her eyes. Gayatri Devi, a neurologist who sees patients seeking second opinions after amyloid-based diagnoses, called the label “a psychologically large burden for someone to carry.” Kewman was prepared for the news, though, having had time to process his genetic results. “It was sobering,” he said. “But it wasn’t surprising.”
For the next four years, roughly once a month, a Lyft driver took Kewman to Stanford, where a nurse infused either lecanemab or saline into his bloodstream. Neither he nor the researchers knew whether the liquid entering his vein contained the drug. He completed cognitive tests and underwent, by his count, more than twenty MRIs. His cognition remained, as he put it, “pretty steady.”
In 2025, the blinded phase of the trial ended, and Kewman entered an open-label extension, in which everyone receives lecanemab. He suspects that, until then, he had been on the placebo. During his career, he had studied placebo effects, and so it seemed to be “karma” that he landed in that group. But there was also physical evidence of a switch during the new phase. After each of his first two open-label infusions, he started shaking with chills, a reaction that can occur when patients begin receiving lecanemab.
Kewman showed me a photograph from one of his infusion days. He sits upright in a padded chair, an I.V. port in his hand, smiling broadly. An infusion pump hangs from a pole beside him; sunlight pours through tall windows in the background.
A day after one of his visits to Stanford, I asked whether he thought of himself as a patient.
“I did yesterday,” he said. He reconsidered. “So the answer is: Yes, I do, because I’m receiving a treatment.”
Half a year after lecanemab secured accelerated approval, the drug received traditional approval on the strength of a larger trial suggesting that it modestly slowed decline in people with early Alzheimer’s. At eighteen months, participants on the placebo had worsened by an average of 1.66 points on an eighteen-point dementia-rating scale. Those on lecanemab worsened by 1.21 points. The difference, 0.45 points, was modest, but qualified as statistically significant.
It was harder to say whether patients or their families would notice it. In 2019, Lilly researchers had proposed a yardstick for clinically meaningful decline: about one point on the same scale for someone with mild cognitive impairment, and one and a half points for someone with mild dementia. Lecanemab’s advantage over the placebo was less than half a point; donanemab’s was about 0.70 points. Neither cleared those thresholds. But Lilly disputed the comparison. The thresholds described how much an individual patient’s condition had worsened, a company employee told me; they couldn’t be used to judge the average difference between patients receiving a drug and those receiving a placebo.
Not everyone accepts this argument. Edo Richard, a neurologist in the Netherlands who serves on an independent advisory group to the European Medicines Agency, balks at that distinction. “You can do a clinical trial at the group level, but you cannot apply a minimally clinically important difference at the group level?” he asked me when we met in Amsterdam over the summer. “That’s so inconsistent.” Pria Anand, a neurologist at Boston University, said, “We’ve shown that we can change some of the underlying biology. The harder question is whether we can meaningfully change patients’ lives.”
The disagreement within the field widened this year. In April, Richard and other researchers published what’s known as a Cochrane review, analyzing seventeen randomized trials of seven amyloid-targeting antibodies, involving more than twenty thousand participants. Cochrane reviews are prized in medicine partly because their methods are specified in advance, limiting researchers’ freedom to search for an analysis that produces the desired result. “You cannot torture the data until they confess,” Richard told me. “You just do what you promised you would do.” The reviewers concluded that the antibodies “probably result in little to no difference” in the decline of memory and of cognitive ability, even as they increase the risk of brain swelling and microhemorrhages.
The response was fierce. In July, a group led by the Alzheimer’s Association published a long rebuttal accusing the review of “serious flaws” and a “fundamental misunderstanding” of amyloid biology and calling for its retraction. Their strongest objection was that pooling antibodies that effectively cleared amyloid with ones that didn’t could obscure the benefits of the newer treatments.
Richard agreed that the drugs were different. “Absolutely,” he said. But the review had also analyzed each antibody separately. His conclusion did not change: “None of the effect sizes comes even close to what’s clinically meaningful.”
The Cochrane authors and their Alzheimer’s Association critics also differed in their ties to the companies whose drugs were being assessed. None of the eleven Cochrane authors reported a relevant financial conflict. Of the thirty-seven authors of the rebuttal, six were employees of the Alzheimer’s Association, and nearly two-thirds disclosed recent relationships with Biogen, Eisai, or Lilly.
Richard pointed out that another amyloid-targeting antibody, solanezumab, had already been tested in cognitively normal, amyloid-positive people. In a large trial led by Sperling, the drug slowed amyloid accumulation by about forty per cent and yet showed no evidence of slowing cognitive decline. Solanezumab was, as Richard acknowledged, not the most effective at removing amyloid; in theory, a more potent amyloid-clearer might have fared differently. But the result, published in 2023, was a disappointment for enthusiasts of early intervention. Richard objected to the habit of describing such trials as “failed.” He observed, “These trials have not failed. These are very rigorously conducted, successful clinical trials. But they show a null effect, and it is equally important in science as a positive effect.”
He saw a pattern in how disappointing results were minimized. He listed the explanations: “We’re too late. The outcome measure is not sensitive enough. The intervention was not intense enough. It was not the right target population.” Each possibility might be true. But, he argued, another possibility received much less consideration: “It simply doesn’t work.”
Carrillo and others at the Alzheimer’s Association regard the major role of amyloid in Alzheimer’s as essentially confirmed, but a number of researchers wonder whether it confuses an effect with a cause. Some disease models depict amyloid accumulation as downstream of failures in the cellular machinery that processes and recycles proteins. Other scientists argue that vascular dysfunction hinders the brain’s ability to clear amyloid and injures neurons through other routes. In such accounts, removing the deposits may be like battling a fire by clearing the smoke.
Alzheimer’s might be better understood as a syndrome, several researchers I spoke with suggested; they saw a pattern of impairment arising from more than one biological process. The rare inherited forms differ importantly from the usual late-onset disease. The distribution of damage, the pace of decline, and even the molecular structure of amyloid filaments can vary. A therapy effective in one case may do little in another. Kewman, who has spent years reading about the disease that may determine his future, was sympathetic to this view. “There are many pathways to Alzheimer’s,” he told me. “I don’t think it’s a unitary disease.”
The prospect of Alzheimer’s has, in any case, already led Kewman to reorganize his life. He and Sandy have visited memory-care facilities, trying to imagine what they might someday need. He drinks less, goes to the gym three times a week, and makes a point of staying socially engaged. He’s a volunteer for the Alzheimer’s Association, and he advocates for more research funding and broader access to testing and treatment. He is, as he puts it, “the unusual person” in those advocacy spaces, being neither a caregiver nor someone in cognitive decline but an asymptomatic man whose genes and biomarkers have placed him inside the expanding boundaries of the disease.
Earlier this year, he went so far as to write his own obituary, or “memorial minute,” as Quakers call it. It recounts his birth in San Francisco and adoption as an infant; his early love of nature; his career as a psychologist. It describes Sandy and their two sons; the discovery of his biological family and, later, of his genetic risk for Alzheimer’s; his years in clinical trials. It ends with his becoming “Opa” to five grandchildren, “the frosting on his cake of life.”
There’s another way in which the trials have folded his past into his present. Kewman grew up in Palo Alto and attended Stanford. In the early nineteen-seventies, not long after graduating, he worked with a patient—a retired physics professor who had been diagnosed with what was then called “presenile dementia”—at a nearby nursing home. Once a week, they spent an hour or so together, sometimes walking through the oak and eucalyptus groves between Palo Alto and Stanford. “It allowed him to just wander,” Kewman recalled.
The oak and eucalyptus groves are still there. More than fifty years later, on the days that Kewman goes to Stanford for lecanemab infusions, he tries to arrive early and spend an hour walking those same paths. “It’s a full-circle sort of experience,” he told me. ♦