A new drug for the treatment of Alzheimer’s disease should be given to a much narrower group of patients than the federal approval permits, Alzheimer’s experts — including those who strongly supported approval of the medication — said on Monday.

Since the Food and Drug Administration approved the controversial and expensive drug, Aduhelm, made by Biogen, this month, much discussion has focused on the fact that many scientists, and the F.D.A.’s own independent advisory committee, say the evidence does not convincingly show that the drug works.

But another major issue has received less attention: which patients should receive the drug and what doctors should do to prescribe it responsibly and safely.

The F.D.A. has so far imposed strikingly few limitations on Aduhelm, a monthly intravenous infusion that requires patients to have regular M.R.I. scans because the drug can cause swelling or hemorrhaging in the brain.

While the only patients who received the drug during clinical trials were those with very mild Alzheimer’s or an even milder pre-Alzheimer’s impairment, the F.D.A.’s label for Aduhelm says simply that the drug is “for the treatment of Alzheimer’s disease.” Under “contraindications,” the term for health conditions or other characteristics that should prevent patients from taking a drug, the label says “None.”

The broadness of the label has surprised and concerned even the biggest champions of the drug.

“Oy,” said one enthusiastic supporter of Aduhelm’s approval, Dr. Stephen Salloway, describing his reaction “when I saw from the label that there are no contraindications.”

Dr. Salloway, director of neurology and the Memory and Aging Program at Butler Hospital in Providence, R.I., spoke on Monday in a forum sponsored by the Alzheimer’s Association, a large patient advocacy group that pushed for approval of the drug. He and the five other experts answering questions about the use of the drug emphasized that the use of Aduhelm should be limited to certain patients: those in early stages of the disease whose brains contain high levels of amyloid, a protein that clumps into plaques in people with Alzheimer’s disease.

The panelists, who had varying opinions about whether Aduhelm should have been approved, agreed that the drug’s potential brain side effects must be monitored carefully and that doctors should disclose to interested patients that there are many unknowns about Aduhelm, including whether it can provide any benefit.

Aduhelm was designed to slow the progression of memory and thinking problems in people with mild cognitive symptoms, but its approval has been contentious. A number of scientists objected because only one of two clinical trials showed any hint of benefit, and in that trial the high dose of the medication slowed cognitive decline only slightly — by about four months in an 18-month period.

Dr. Salloway, a site principal investigator for trials of the drug, wasn’t paid for that work but has received research and consulting fees from Biogen. He said doctors should use the drug only for patients whose statuses match those in the clinical trials.

“There’s no evidence that it could be beneficial for any other stage of Alzheimer’s,” he said.

Mary Sano, director of the Mount Sinai Alzheimer’s Disease Research Center in New York City, said the criteria that she and other panelists outlined were “very important” and meant that “it’s going to be very restrictive and the ability to share this drug with a wide range of people will be significantly limited, at least at this time.”

Treating people only with mild symptoms would mean that for dementia clinicians, “most of your people in your current practice are probably not eligible,” Dr. Sano said.

In its decision, the F.D.A. acknowledged that there was not the level of evidence of benefit that the agency usually requires. As a result, it is making Aduhelm available under a program called accelerated approval, citing the drug’s ability to reduce levels of amyloid in the brain. But reducing amyloid is not the same thing as slowing symptoms of dementia. Many amyloid-reducing drugs have failed to slow decline in clinical trials, a history that makes some experts especially wary of placing confidence in Aduhelm based on the evidence produced so far.

Given the agency’s emphasis on amyloid in its approval decision, and the fact that all of the clinical trial participants had to have high amyloid levels, experts have also been surprised that the F.D.A. label does not require patients to be screened for the protein. Doctors at the Alzheimer’s Association forum all said that high levels of amyloid, typically measured by PET scan or spinal tap, should be a condition of treatment.

Several of the panelists said that, at least at the outset, relatively few doctors and clinics would have the ability to adequately diagnose, screen and treat patients.

“This is not a simple medication to use,” said Dr. Paul Aisen, director of the Alzheimer’s Therapeutic Research Institute at the University of Southern California and a co-author of an article that urged the F.D.A. to approve the drug. “I think that establishing the appropriate individuals for treatment, and monitoring treatment, requires knowledge and benefits from experience, and there are very few clinicians who have this experience.”

The panelists devoted considerable discussion to the possibility of brain swelling and hemorrhages, which occurred in about 40 percent of participants who received the high dose in the two large clinical trials. Many cases were mild or asymptomatic, but Dr. Alireza Atri, director of the Banner Sun Health Research Institute in Phoenix and another co-author of the article supporting approval of the drug, said that it was possible that “one out of 200 or 300 individuals can have a serious side effect and need to be in a hospital.”

Dr. Salloway said that it would be “more challenging” for doctors to safely monitor for brain side effects than it was within the strict standards of the clinical trials. He said that people should not be given the drug if they have had a macro-hemorrhage in the brain; more than five micro-hemorrhages; a significant stroke; or “unstable medical conditions that could interfere with treatment.”

Dr. David S. Knopman, a clinical neurologist at the Mayo Clinic and a site principal investigator for one of the trials, who did not support approval, said people who were taking blood thinners should also be excluded.

“We know that this treatment carries considerable risks,” said Suzanne Craft, co-director of the Roena B. Kulynych Center for Memory and Cognition Research at the Wake Forest University School of Medicine. And assessing how and whether it helps patients could be tricky, Dr. Craft and others said.

It will be important to have comprehensive discussions with patients and families about “how to weigh the inconvenience and cost and risk against the possible benefit,” Dr. Aisen said.

“Managing expectations is a huge challenge here,” he said, adding that “our expectation is a modest slowing of the rate of decline. It is impossible to determine on an individual patient level whether someone is benefiting or not.”