What Are the Potential Side Effects of Aducanumab?

Aducanumab’s most significant side effect is a set of brain changes visible on MRI scans, collectively called amyloid-related imaging abnormalities, or ARIA. In the high-dose group from its two large clinical trials, roughly one in three participants developed ARIA involving brain swelling or fluid buildup, and the risk climbed even higher for people carrying a specific genetic variant tied to Alzheimer’s disease. Most of these episodes produced no noticeable symptoms, but when they did, the effects ranged from headaches and confusion to, in rare cases, seizures or death. The drug’s safety profile became one of the central controversies in its brief time on the market, and understanding what ARIA is, who gets it, and what it feels like remains relevant as newer drugs in the same class continue to be prescribed.

What ARIA Actually Feels Like

ARIA comes in two flavors. ARIA-E refers to edema or fluid leakage in the brain, and ARIA-H refers to tiny bleeds (microhemorrhages) or iron deposits on the brain’s surface. Both are detected primarily through routine MRI scans rather than through a patient’s own complaints. Most ARIA episodes are asymptomatic, meaning they show up on imaging but the person feels nothing unusual. When symptoms do occur, they tend to include headache, confusion, dizziness, and nausea.¹ Seizures have been reported in a small number of cases, and deaths linked to ARIA, while rare, have occurred.

The disconnect between what the MRI shows and what the patient feels is one of the more unsettling aspects of aducanumab treatment. A person could have visible brain swelling on a scan and report feeling perfectly fine. This is why treatment guidelines call for regular MRI monitoring rather than relying on patient-reported symptoms alone. A side effect you cannot feel is still a side effect your brain is experiencing.

How Common ARIA Is

The incidence numbers come primarily from aducanumab’s two phase 3 trials, known as EMERGE and ENGAGE, which together enrolled over 3,000 people with early Alzheimer’s disease. At the highest dose tested (10 mg/kg), about 35% of participants developed ARIA-E, compared with roughly 21% in a middle-dose group and under 3% in placebo-treated participants.² A separate meta-analysis pooling data across anti-amyloid drug trials put aducanumab’s overall ARIA-E rate at about 31%.³

The relationship between dose and ARIA is clear and consistent: the more drug a person receives, the more likely they are to develop brain swelling or microbleeds. This dose-dependent pattern was one reason the drug’s clinical usefulness came under scrutiny. The dose thought to have the best chance of slowing cognitive decline was also the dose most likely to cause ARIA.⁴

Who Faces the Highest Risk

Three factors stand out as major predictors of developing ARIA: the dose of aducanumab, whether a person carries the ApoE ε4 gene variant, and whether pre-existing microbleeds are already visible on a baseline brain MRI.⁵

The genetic piece is especially striking. ApoE ε4 is already the strongest known genetic risk factor for developing late-onset Alzheimer’s disease, so many of the people prescribed aducanumab carry it. In the high-dose trial group, about 43% of ApoE ε4 carriers developed ARIA-E, compared with roughly 20% of non-carriers.⁶ People who carry two copies of the ε4 variant (homozygotes) face an even steeper risk. A genome-wide study of the same trial participants found that ε4 homozygotes had roughly four times the odds of developing ARIA-E and nearly eight times the odds of a specific type of ARIA-H called superficial siderosis, compared with non-carriers. The genetic effect was especially pronounced for radiographically severe ARIA, where odds ratios among homozygotes ranged from about 7 to nearly 25 times higher than in non-carriers.⁷

Because of this gene-dose relationship, clinical use recommendations call for ApoE genotyping before starting treatment. The idea is not necessarily to exclude all ε4 carriers, but to give clinicians and patients a clearer picture of the risk they are taking on and to set an appropriate monitoring schedule.⁸

Pre-existing microbleeds on a baseline MRI also raise the stakes. If the brain’s small blood vessels are already showing signs of damage, particularly from a condition called cerebral amyloid angiopathy (CAA), the risk of developing ARIA-H during treatment goes up. More extensive amyloid deposits in and around blood vessels correlate with a higher chance of treatment-related bleeding.⁹

When Side Effects Tend to Appear

ARIA is not evenly distributed across the course of treatment. It tends to cluster during the early months, typically between weeks 12 and 32, which corresponds to the dose-escalation (titration) phase when the drug is gradually being ramped up to its target dose.¹⁰ Real-world safety data from the FDA’s adverse event reporting system found a median time to any adverse event onset of about 146 days after starting treatment, roughly five months.¹¹

Appropriate use recommendations suggested performing MRI scans before the 5th, 7th, 9th, and 12th infusions to catch ARIA as early as possible during this vulnerable window.¹² If ARIA was detected, the guidance called for pausing or stopping treatment depending on severity. Symptomatic ARIA and moderate-to-severe ARIA on imaging generally warranted dose interruption or discontinuation.

Most ARIA-E episodes do resolve over time. Long-term extension data from the trials indicated that the large majority of edema events cleared within about three to four months, and nearly all had resolved by the end of the longer follow-up period. That said, “resolved on imaging” does not necessarily mean “no lasting consequence.” Whether repeated or prolonged ARIA episodes leave behind subtle brain damage that compounds over time remains an open question in the field.

What Happens Inside the Brain

The mechanism behind ARIA helps explain why it is so closely tied to drugs that clear amyloid from the brain. Aducanumab works by binding to clumps of amyloid-beta protein and promoting their removal. The problem is that amyloid does not just accumulate in brain tissue; it also builds up in the walls of small blood vessels, a condition common in Alzheimer’s patients. When the drug mobilizes amyloid from plaques in brain tissue and blood vessel walls, it can temporarily destabilize those vessels.¹³

The body’s immune system recognizes the disrupted amyloid and launches an inflammatory response against the affected vessel walls. This inflammation increases vascular permeability, meaning the walls of blood vessels become leaky. Proteinaceous fluid and blood products seep out into surrounding brain tissue.¹⁴ When what leaks out is mostly clear fluid, it shows up as ARIA-E (edema). When red blood cells escape, the result is ARIA-H (microhemorrhages or iron staining).

People with more extensive pre-existing amyloid in their blood vessel walls, the condition referred to as cerebral amyloid angiopathy, have more vulnerable vessels to begin with. The drug essentially stresses a system that is already compromised, which is why baseline vascular amyloid burden is such a strong predictor of who develops ARIA.¹⁵

Real-World Safety Signals Beyond the Trials

Clinical trials are controlled environments with carefully selected participants. Once aducanumab reached real-world patients, additional safety data emerged through the FDA’s adverse event reporting system (FAERS). An analysis of 510 aducanumab-related reports confirmed that nervous system problems were the most frequently reported category, making up about 53% of all reports. ARIA-E and ARIA-H remained the dominant signals by a wide margin.¹⁶

The safety profile looked broadly similar across men and women. Age mattered, though. ARIA signals were especially strong in patients 75 and older, a group that was underrepresented in the original trials. Since Alzheimer’s disease disproportionately affects older adults, the fact that the oldest patients carry the most ARIA risk is a practical concern for treatment decisions.¹⁷

A separate pharmacovigilance analysis using a broader FAERS dataset found extremely strong statistical associations between aducanumab and ARIA-E, ARIA-H, and cerebral microhemorrhage. The analysis also flagged a small number of reports of subdural hematoma, though with only three cases, the researchers cautioned against drawing firm conclusions from such limited data.¹⁸ These post-market reports did not fundamentally change the risk picture painted by the clinical trials, but they reinforced that ARIA is not a controlled-setting artifact. It occurs in routine clinical practice too.

The Financial Side of Safety Monitoring

ARIA monitoring is not optional. The drug’s prescribing information requires brain MRI scans before treatment begins and at regular intervals throughout. This is not a one-time screening; for a patient who develops ARIA, additional scans are needed to track whether the swelling or bleeding resolves, stabilizes, or worsens. The cost of this monitoring adds up. One analysis estimated that the additional cost of three brain MRIs for patients who develop ARIA comes to about $766 per person, on top of the drug’s own cost and the fees for intravenous infusion sessions.¹⁹

For a disease that already imposes enormous caregiving and healthcare costs on families, the added financial burden of monitoring for drug side effects was a genuine concern. This was compounded by aducanumab’s initial list price of $56,000 per year (later reduced by the manufacturer), which drew widespread criticism from insurers, patient advocates, and health economists alike. Medicare ultimately agreed to cover the drug only for patients enrolled in clinical trials, severely limiting real-world access and making the monitoring cost question somewhat academic for most families.

How Aducanumab Compares to Similar Drugs

Aducanumab is not the only anti-amyloid antibody to cause ARIA. Lecanemab (Leqembi) and donanemab (Kisunla) are newer drugs in the same class, and both carry ARIA risk as well. The drugs differ in their side effect profiles, though. A systematic review and meta-regression analysis found that aducanumab had the highest ARIA-E rate among the anti-amyloid therapies studied, at roughly 31%. However, it had the lowest rate of infusion-related reactions, at about 1%.²⁰

A network meta-analysis comparing the drugs across randomized trials ranked the highest ARIA-E risk with donanemab and aducanumab, followed by gantenerumab and lecanemab.²¹ Lecanemab’s ARIA-E rate is lower than aducanumab’s, though it comes with higher rates of infusion reactions, which are flu-like symptoms occurring during or shortly after the IV drip.

These comparisons matter because aducanumab is no longer commercially available. Biogen withdrew the drug from the U.S. market in early 2024, citing commercial viability rather than safety concerns as the primary reason. But its safety data remains relevant because the side effects it causes, particularly ARIA, are a class-wide phenomenon. Patients considering lecanemab or donanemab face the same fundamental tradeoff: the potential for modest cognitive slowing in exchange for a real risk of brain swelling and microbleeds. The lessons learned from monitoring and managing aducanumab’s ARIA have directly shaped the monitoring protocols now used for its successors.

Infusion Reactions and Other Non-ARIA Side Effects

While ARIA dominates the conversation around aducanumab’s safety, it is not the only side effect. Infusion-related reactions, where the body reacts during or shortly after the intravenous drip, do occur, though at a notably low rate compared with the other anti-amyloid drugs. The systematic review mentioned above pegged aducanumab’s infusion reaction rate at about 1.2%, making it the mildest in the class on that front.²² Infusion reactions can include chills, flushing, nausea, and changes in blood pressure, and they are generally manageable by slowing the infusion rate or pre-medicating with antihistamines.

Headache shows up in both the ARIA-related symptom list and as a standalone complaint from patients receiving infusions. Distinguishing a headache caused by brain edema from one caused by the infusion process itself, or from something entirely unrelated, is part of the clinical challenge. This ambiguity reinforces why imaging rather than symptoms remains the primary tool for detecting ARIA.

Ongoing Questions About Cumulative Risk

One area where the evidence remains thin is the long-term consequence of repeated ARIA episodes. Some patients in the clinical trials experienced more than one episode of ARIA-E or ARIA-H. The appropriate use recommendations acknowledged that recurrent or serious ARIA can occur and suggested additional criteria for stopping treatment in those cases.²³ But the trials were not designed to track whether microbleeds that “resolve” on MRI leave behind lasting damage to brain tissue, and the extension studies that ran longer were not large enough to provide definitive answers.

This gap matters because Alzheimer’s patients are already losing brain volume and function over time. If treatment-related microbleeds contribute even modestly to cumulative vascular injury, the net benefit calculation could shift. Researchers studying ARIA across the anti-amyloid drug class continue to investigate whether radiographic resolution truly equals biological resolution, or whether each episode leaves a mark the MRI cannot see. For now, the working assumption in clinical practice is that resolved ARIA is benign, but some neurologists remain skeptical, and it is a legitimate area of uncertainty that patients and families deserve to know about.