Last week, the pharmaceutical company Merck pulled the plug on a closely watched Alzheimer’s drug trial. The drug verubecestat, an outside committee concluded, had “virtually no chance” of benefit for patients with the disease.

The failure of one drug is of course disappointing, but verubecestat is only the latest in a string of failed trials all attempting the same strategy to battle Alzheimer’s. That pattern of failure has provoked some rather public soul-searching about the basic hypothesis that has guided Alzheimer’s research for the past quarter century.

The “amyloid hypothesis” began with a simple observation: Alzheimer’s patients have an unusual buildup of the protein amyloid in their brains. Thus, drugs that prevent or remove the amyloid should slow the onset of dementia. Yet all drugs targeting amyloid—including solanezumab from Eli Lilly and bapineuzumab from Pfizer and Johnson & Johnson, to add a few more high-profile flameouts to the fail pile—have not worked so far.

After Merck’s announcement last week, one neurologist told Bloomberg that “there is mounting evidence—of which this is another piece—that removing amyloid once people have established dementia is closing the barn door after the cows have left.” An advisor to a life-sciences venture-capital firm tweeted, “I've been a long-term adherent of the amyloid hypothesis, but starting to feel like this”: “This” was a gif of the Black Knight from Monty Python, arms missing but still adamant he had suffered nothing worse than a flesh wound.

And well, the amyloid hypothesis is not dead yet. Large clinical trials targeting amyloid are still underway—either using new, potentially more powerful anti-amyloid drugs or trying out the previously failed drugs in patients with less advanced Alzheimer’s. These trials will likely affirm the amyloid hypothesis or kill it for good.

With the benefit of hindsight, the story of the amyloid hypothesis will be written either as one where scientists soldiered on despite setbacks, or one where a wrong idea derailed a field for 25 years. And the field of Alzheimer’s research is no stranger to ideas inflated, abandoned, and sometimes resurrected.

* * *

In the 1980s, the importance of amyloid was not the dominant idea in a field that might need shaking up. It was the upstart idea doing the shaking up. At the time Alzheimer’s researchers were considering the cholinergic hypothesis, which posits that a decline in the neurotransmitter acetylcholine is a cause of the disease. The handful of available Alzheimer’s drugs come out of this line of research. But it never produced an outright cure, and lack of acetylcholine has since been abandoned as a root cause of Alzheimer’s.

Dennis Selkoe—who is now a leading proponent of the amyloid hypothesis and a neurologist at Harvard—was not that interested in acetylcholine at the time either. He actually started out studying the buildup of tau protein, which is behind yet another hypothesis for the cause of Alzheimer’s. But he moved on to the nascent amyloid field when he met George Glenner, who was gathering data on how amyloid can build up into waxy plaques throughout the body. Amyloid deposits in the kidneys can cause kidney failure; in hearts, it can cause heart failure. Wouldn’t it follow then, that amyloid deposits in the brain can cause brain failure? The apparent a-ha moment came in 1984, when Glenner indeed found the protein in the brain tissue of Alzheimer’s patients and purified it. (Glenner himself had systemic senile amyloidosis, where amyloid builds up in the body. He died in 1995.)

Selkoe’s lab eventually worked out some of the molecular mechanisms behind the amyloid build-up. Then, geneticists found that people with a family history of Alzheimer’s also had mutations in the very genes that encode for making amyloid. “The genetic evidence is so strongly supported,” says Selkoe, “People gravitated toward it.” A 1992 paper in Science by John Hardy and Gerald Higgins laid out the case for the amyloid hypothesis.

The discovery of these genes created a sense of optimism in Alzheimer’s research. Scientists had a roadmap to a cure. Drug companies just needed to follow it. But of course, a quarter century later, there are still no drugs for Alzheimer’s that target amyloid.

“When you have a setback, there’s understandably questioning, are you sure about the science,” Selkoe says about the latest Merck trial. But he remains convinced that targeting amyloid could still succeed with a few changes—like when patients start treatment. By the time an Alzheimer’s patient starts suffering memory loss, that may be too late. So the A4 trial, coordinated out of the University of Southern California, is testing an previously unsuccessful drug in patients with elevated amyloid levels but no outward cognitive symptoms yet. Even if anti-amyloid drugs can’t reverse symptoms, perhaps they can prevent full-blown Alzheimer’s. The prevailing view in the field, Selkoe says, is still that dealing with amyloid can treat the disease.

One of the loudest contrarian voices is George Perry, a neuroscientist at the University of Texas at San Antonio, who has gone so far as to declare the amyloid hypothesis “dead.” Perry favors an explanation where neurons are damaged by an imbalance of overreactive molecules containing oxygen—also known as oxidative stress. Of course, the evidence that amyloid has something to do with Alzheimer’s is strong, and Perry is not discounting that. Perhaps, he says, amyloid buildup is a protective response to oxidative stress, though the root cause of the stress is unknown.

To be clear, this is a minority view in the research community. Perry says  his trainees’ research grant applications have negative feedback from reviewers for working with him because of his outspokenness against the amyloid hypothesis. The problem, he contends, is that people invested in the hypothesis are entrenched. “The government and the pharmaceutical industry have invested almost all their resources, and some of the brightest people have been in the amyloid area,” he says. “Think about all that talent that has been invested and that so far as yielded zero therapeutic value.”

Other skeptics of the amyloid hypothesis are coming back to tau, the protein Selkoe left decades ago to focus on amyloid. In the brains of Alzheimer’s patients, tau gets twisted into tangles that block the internal transport system of neurons. A recent failed trial aside, several drugs targeting tau are in early phases of clinical trials.

The waxing and waning of animating ideas is just how science works. For scientists, says Bart de Strooper, an Alzheimer’s researcher at the University of Leuven, these failed trials still have something to teach. “The failed trial doesn’t have to be a failure. We learn from it,” he says. “Only because a trial is done do we know now we have to go further.” The question is how much further. Refine the amyloid hypothesis, or abandon it altogether? Without the benefit of hindsight, it’s hard to know which is the side of history.