When the U.S. Food and Drug Administration approved Biogen Inc.’s controversial Alzheimer’s drug aducanumab last month without firm evidence that the drug helps patients, it created concern that other unproven treatments might follow. That scenario is already here.

Eli Lilly & Co. announced last week that it plans to file this year for accelerated approval for its experimental Alzheimer’s treatment, donanemab, based on its ability to clear amyloid brain plaques, believed by some to cause the disease. That’s the same unprecedented justification and pathway the FDA used for aducanumab. Lilly is clearly reacting to the agency’s shifted goalposts; in April, the company told investors it wouldn’t seek quick approval because the regulator said there wasn’t enough data.

But that was the old FDA. It’s now possible that the agency might approve another medicine based on an unproven idea. Lilly shareholders seem to think so – the company’s shares rose 8 percent the day that Lilly announced it was seeking accelerated approval of donanemab. And yet, although approval would boost Lilly’s revenue prospects, it would compound the mistakes of the aducanumab decision and further damage future Alzheimer’s research and America’s health budget.

Amyloid plaque removal has failed to help patients in many previous trials, and using it as the sole basis for approval of Biogen’s treatment was a surprise. The FDA released guidance on Alzheimer’s drugs in 2018 rejecting such secondary measures because of scientific uncertainty. A panel of external experts asked to evaluate Biogen’s drug in November were told that the agency wasn’t considering that option. But when the group voted that the contradictory evidence from Biogen’s trials didn’t support approval, the FDA found a new way forward. Three members of the expert panel resigned in protest.

The FDA released a memo last Tuesday describing its reasoning for going all in on amyloid. It cites analyses supporting a link between plaque reduction and slower patient decline and explaining away Biogen’s messy data. But those analyses weren’t reviewed by external experts and are disturbingly biased.

Lilly’s approval pursuit won’t clear things up. Donanemab may be even better than Biogen’s drug at clearing amyloid. But the evidence of patient benefit is once again mixed and comes from only 257 patients. The trial appeared to slow cognitive decline, but the effect was mild and visible only on a scale that hasn’t previously been used extensively in Alzheimer’s trials. The drug showed no benefit at all on a variety of other measures.

The consequences of another accelerated approval would be dire. It would attract more R&D dollars toward amyloid drugs, which appear to have a very mild effect at best. Billions of dollars that would be better spent better working on an understanding of Alzheimer’s and truly transformational drugs will be spent on possible placebos. And the whole biopharma industry will get even more incentive to keep studying weak drugs.

The FDA can right the ship. Lilly is already running a larger trial that could provide better evidence for its medicine and the amyloid hypothesis as soon as 2023. Results from other amyloid-targeting drugs from Biogen and Roche Holding AG could be available by then as well. By waiting a bit and conducting a public evaluation of the evidence for amyloid with more data, the agency could reset expectations and its standards instead of opening the barn door wider.