Will Unpredictable Side Effects Dim the Promise of New Alzheimer’s Drugs?
Despite landmark antibody approval, research into potentially serious brain swelling and bleeding still lags
Jennifer Couzin-Frankel
Aug. 2, 2023 (Science.org) – A sea change is underway in the treatment of Alzheimer’s disease, where for the first time a drug that targets the disease’s pathology and clearly slows cognitive decline has hit the U.S. market. A related therapy will likely be approved in the coming months. As many neurologists, patients, and brain scientists celebrate, they’re also nervously eyeing complications from treatment: brain swelling and bleeding, which in clinical trials affected up to about one-third of patients and ranged from asymptomatic to fatal.
The side effect—amyloid-related imaging abnormalities, or ARIA—remains mysterious. “We don’t really understand what it is, what causes it, and what we can do about it,” says neurologist R. Scott Turner, director of the Memory Disorders Program at Georgetown University.
The drugs triggering the problem are monoclonal antibodies that mop up beta amyloid, a protein that accumulates in the brains of people with Alzheimer’s. One such antibody, lecanemab, marketed as Leqembi by Eisai and Biogen, won full approval in the United States last month after trials showed it modestly slowed cognitive decline in early disease. Another, Eli Lilly & Co.’s donanemab, had similar results and is under regulatory review. These therapies aim to clear clumps of amyloid, called plaques, from between neurons. But such plaques can also build up in brain blood vessels, and breaking them down there may set off ARIA.
The condition comes in two forms: brain swelling (ARIA-E for edema) and brain bleeding (ARIA-H for hemorrhage). Both show up in MRI and some patients also experience symptoms, which can include headaches, confusion, vomiting, and seizures. In the largest lecanemab trial, 21% of those on the drug had ARIA, and 3% had ARIA with symptoms. In donanemab’s biggest trial, 37% on therapy got ARIA and 6% had symptoms.
These trials and others of similar antibodies have helped identify risk factors for ARIA, in particular having a gene variant, APOE4, that also heightens risk of Alzheimer’s. Anticoagulants that treat or prevent blood clots appear to increase ARIA risk as well.
As patients and families express enthusiasm for the treatments, neurologists are worrying not only about serious ARIA, but the challenges that may accompany milder cases, which often call for pausing treatment and monitoring brain health with MRIs. “We’ve got to get our hands around this,” says Walter Koroshetz, director of the National Institute of Neurological Disorders and Stroke (NINDS). Yet few researchers are studying ARIA. NINDS is now funding one ARIA-focused grant, but has new applications under review. Along with the National Institute on Aging (NIA), NINDS is also planning a meeting on ARIA next month to sketch out scientific priorities.
Interest in ARIA blossomed 20 years ago, when antiamyloid antibodies were drawing increasing attention. In 2002, a brief report in Science described 10 mice engineered to develop Alzheimer’s-like symptoms and treated with antiamyloid therapy; plaques faded, but the mice had more frequent and severe brain hemorrhages than controls. “I call that the first-ever report of ARIA,” says Donna Wilcock, a neuroscientist who directs the Center for Neurodegenerative Disorders at the Indiana University School of Medicine. She joined other scientists who began to study the complication. But as excitement grew about the therapies’ potential and they entered clinical trials, the basic science “just kind of dried up,” she says.
Wilcock has returned to ARIA research and holds the lone NINDS ARIA grant. She is examining whether side effects in her antibody-exposed mice align with what most doctors believe occurs in people on treatment: ARIAE, followed in some cases by ARIA-H.
Wilcock and others think a condition called cerebral amyloid angiopathy (CAA), a buildup of amyloid in brain blood vessels that’s common in older people with Alzheimer’s, helps set the stage for ARIA. CAA can lead to small brain hemorrhages; in the drug trials, the rate of ARIA-H alone was roughly the same in the treated and placebo groups. CAA also raises risk of ARIA for those on antiamyloid antibodies, and the U.S. Food and Drug Administration recommends that people with signs of CAA exercise caution before taking lecanemab. But CAA isn’t always easy to detect on MRIs, especially when it’s mild.
What happens, Wilcock suspects, is that as the antibodies break down amyloid deposits in blood vessels, they activate microglia. These cells support brain repair and likely help antibodies clear plaques, but they can also cause inflammation. Such inflammation could prompt blood vessels to leak, producing swelling. In the trials, brain swelling, especially with symptoms, was far more common in the treatment than the placebo groups. Blood vessel ruptures can also cause bleeding. But, “There’s a lot of what-ifs” in this cascade of events, Wilcock acknowledges, and there are other theories, too.
She and other scientists suspect an uncommon complication of CAA, called CAA-related inflammation (CAA-ri), may offer further insight. Here, people not taking the antibodies spontaneously develop what looks like ARIA-E with symptoms. Last year, Fabrizio Piazza, a translational researcher at the University of Milano-Bicocca, reported in Neurology that in patients with CAA-ri, microglia are activated in areas with brain swelling. That finding, he says, supports the idea that these cells may contribute to treatment- related ARIA.
Piazza also found spontaneously occurring antiamyloid antibodies in the cerebrospinal fluid (CSF) of people with CAA-ri. He’s setting up a global registry of CAA-ri patients to see whether these CSF markers can help predict and track ARIA-E, given that abnormalities on MRIs don’t always reflect symptoms.
For now, the risk of complications raises many questions for doctors and patients, some of them life or death. At Northwestern University, chief of neurocritical care Sherry Chou is struggling with what she knows is an inevitable scenario: A patient on antibody treatment for early Alzheimer’s is having a stroke and urgently needs a clot-busting drug to stop brain cells dying en masse. Withholding the drug could mean devastating brain damage, but giving it may raise the risk of catastrophic hemorrhage. In February, Chou and her colleagues published the case of a woman she helped care for, who was taking Leqembi through a trial, received this stroke treatment, and died days later after suffering significant hemorrhages.
Given that about 800,000 people in the United States have a stroke every year, Chou anticipates “natural experiments” where doctors make emergency medical decisions without data. “The research is just getting started, but the drug is already in clinical use,” Chou says. “What happens now?”
Another quandary is when or whether to return a patient to antibody treatment after stopping it because of ARIA. “There’s a lot of guesswork” and a need for more data, Turner says. MRIs may be reassuring, but the risk of a repeat scenario could remain. In the donanemab trial, one participant died from ARIA months after an episode apparently resolved and therapy restarted.
Such guesswork is complicated by another mystery: “I want to know what, if anything, ARIA did to a patient’s cognition at the end of the clinical trial,” especially in serious cases, says Madhav Thambisetty, chief of the clinical and translational neuroscience section at NIA and a neurologist at Johns Hopkins University. One concern is whether some patients with marked ARIA might have worsening memory or other cognitive functions by the trial’s end. Data on that question exist but the companies haven’t shared them, Thambisetty says.
Doctors also want to know how best to treat ARIA and whether it can be prevented. Again, CAA-ri may provide clues. Steven Greenberg, a vascular neurologist and director of the Hemorrhagic Stroke Research Program at Massachusetts General Hospital, recently found that CAA-ri patients with brain swelling reminiscent of ARIA-E generally had quick improvement and a reduced chance of relapse when treated with high-dose steroids, a treatment also used for severe ARIA. Depending on how steroids work in CAA-ri, Greenberg even wonders whether they might be given alongside antiamyloid therapy or at the earliest sign of ARIA to head off severe effects; such strategies would need testing, he stresses.
Researchers are especially eager to help people with two copies of APOE4. They are at higher risk for ARIA but also develop Alzheimer’s at a younger age. Many doctors are unsure about prescribing the antibodies to these patients (lecanemab data suggest they may also benefit less than patients without APOE4), and the Department of Veterans Affairs has said it won’t typically cover antiamyloid treatment for them, but will consider requests for exceptions. But, “Just saying, ‘Oh, sorry,’” isn’t good enough, Wilcock says. “We need to figure this out so we can treat those people without the risk.”
Ultimately, the risk calculus will be personal. Nancy Meserve, a retired special education administrator who carries two copies of APOE4, is enrolled in a trial to see whether lecanemab can prevent Alzheimer’s. She doesn’t know whether she’s on the drug or the placebo, but based on side effects suspects the drug; she also developed ARIA-H with mild symptoms, which resolved. “People take risks with things that are important to them,” Meserve says. “To me, this was important.”