Aduhelm Side Effects: ARIA Risks, Symptoms, and Timeline

The dominant side effect of Aduhelm (aducanumab) is a set of brain changes visible on MRI scans, collectively called amyloid-related imaging abnormalities, or ARIA. In the two pivotal phase 3 trials, roughly 41% of patients receiving the drug developed some form of ARIA, most commonly brain swelling or fluid accumulation around the brain’s surface. While the majority of these episodes resolved on their own and many patients never felt symptoms, a meaningful fraction experienced headaches, confusion, or dizziness, and the required MRI monitoring adds both cost and complexity to treatment.

Two Types of Brain Changes on MRI

ARIA comes in two distinct flavors, and understanding the difference matters because they carry different implications. ARIA-E refers to edema and effusion, meaning fluid builds up in or around the brain tissue. ARIA-H refers to hemorrhage, which shows up as tiny bleeds (microhemorrhages) or iron deposits on the brain’s surface called superficial siderosis. Both types are thought to result from an inflammatory response triggered when the drug clears amyloid protein from blood vessel walls, making those vessels temporarily leaky.

ARIA-E tends to be the more common and clinically significant of the two. It usually appears as patchy bright areas on a type of MRI sequence called FLAIR imaging, often in regions where amyloid tends to accumulate along blood vessels. ARIA-H, by contrast, shows up as small dark spots on a different MRI sequence. A patient can develop one type, both types, or neither. In the phase 3 trials, ARIA-E occurred in about 35% of treated patients, while microhemorrhages appeared in about 19% and superficial siderosis in roughly 15%.

What Patients Actually Feel

Here is the reassuring part and the unsettling part wrapped into one: most ARIA episodes are “silent,” meaning the patient has no symptoms and the changes are only caught on routine MRI. In the combined phase 3 data, only about 26% of patients who developed ARIA-E actually noticed anything wrong. But when symptoms did appear, they were not trivial.

Among the roughly 100 patients who experienced symptomatic ARIA in the trials, the most frequently reported complaints were:

Headache: reported by about 47% of symptomatic cases
Confusion: reported by about 15%
Dizziness: reported by about 11%
Nausea: reported by about 8%

These symptoms can overlap with the cognitive decline of Alzheimer’s disease itself, which makes them tricky to distinguish without imaging. A person on Aduhelm who suddenly seems more confused than usual could be experiencing disease progression, a medication side effect, or both. That ambiguity is one reason regular MRI monitoring is built into the treatment protocol.

Timing and Resolution

ARIA tends to appear early in treatment. Nearly three-quarters of initial ARIA-E events in the phase 3 trials occurred within the first eight doses of Aduhelm, which corresponds roughly to the first several months of the dose-escalation period when the infusion amount is being gradually increased. This clustering early in the treatment course is important clinically because it means the highest-risk window is somewhat predictable, and MRI surveillance can be concentrated during that period.

⁴

The good news is that ARIA-E typically resolves. In the trials, about 98% of ARIA-E episodes resolved on imaging, and roughly 83% cleared within 16 weeks. Serious cases requiring hospitalization were uncommon, affecting about 1.4% of treated patients overall. ARIA-H behaves a bit differently: microhemorrhages and superficial siderosis may persist on imaging even after the acute episode passes. Whether residual ARIA-H on MRI has long-term consequences is still an open question, and real-world data collection continues to try to answer it.

⁵

The APOE ε4 Connection

Not everyone faces the same risk. The strongest known predictor of developing ARIA on Aduhelm is whether you carry one or two copies of the APOE ε4 gene variant. This is the same genetic variant that increases Alzheimer’s risk in the first place, creating an uncomfortable situation: the patients most likely to develop the disease are also the most likely to experience the drug’s main side effect.

In the phase 3 trials, ARIA-E rates were highest among APOE ε4 carriers receiving Aduhelm. The risk follows a dose-response pattern tied to gene copies, meaning people with two copies of the variant face a higher rate than those with one, who in turn face a higher rate than non-carriers. Because of this strong genetic link, expert recommendations advise that clinicians perform APOE genotyping before starting treatment, not to exclude patients, but to calibrate monitoring intensity and have more informed conversations about risk.

⁶

It is worth noting that even in the placebo group of the trials, a small percentage of participants (about 2.7%) developed ARIA-E, a reminder that some degree of these brain changes can occur spontaneously in people with Alzheimer’s pathology, independent of any drug.

⁷

How Doctors Handle ARIA When It Appears

The standard clinical approach when ARIA is detected on a scheduled MRI depends on severity. For mild, asymptomatic ARIA-E, doctors may continue treatment under closer surveillance, with more frequent MRI scans until the finding resolves. For moderate to severe cases, or when symptoms are present, the typical response is to temporarily suspend infusions and wait for the brain changes to clear before considering restarting at a lower dose. In the most serious cases, treatment is permanently discontinued.

⁸

The required MRI monitoring schedule adds a logistical layer to treatment. Patients typically need brain MRIs before starting Aduhelm, then at regular intervals during the first year, particularly during and just after the dose-escalation phase. For many patients, especially those in rural areas or those who have difficulty lying still in an MRI machine, this is a real practical barrier. The monitoring is not optional; without it, asymptomatic ARIA could go undetected and potentially progress.

Long-term extension data from the earlier Phase 1b trial of aducanumab confirmed that dose titration, gradually increasing the dose rather than starting at the full amount, was associated with lower rates of ARIA-E. This finding shaped the dosing protocols used in clinical practice.

⁹

Genotype-Guided Dosing Shows Promise

Some clinicians have gone further than standard dose titration by tailoring the escalation schedule to a patient’s APOE genotype, slowing the dose ramp-up more aggressively for ε4 carriers. A retrospective analysis from a private practice setting examined 25 patients treated with aducanumab under this kind of individualized protocol and found ARIA-E and ARIA-H rates of just 4% each, dramatically lower than the 35% and 19% rates seen in the phase 3 clinical trials.

¹⁰

The tradeoff is meaningful, though. In that same analysis, plaque clearance rates were less robust with the slower titration, suggesting that the gentler dosing approach that reduces ARIA risk may also reduce the drug’s ability to remove amyloid from the brain. This is the central tension in managing Aduhelm’s side effects: the drug works by provoking an immune-mediated attack on amyloid plaques, and ARIA appears to be a byproduct of that same immune response. Dialing down the intensity of the attack may protect against side effects at the cost of some efficacy. The sample size in that real-world analysis was small, so the exact magnitude of this tradeoff remains uncertain.

The Financial Side of Safety Monitoring

Managing ARIA has a price tag that goes beyond the drug itself. A cost-effectiveness analysis modeled the expenses associated with ARIA monitoring and treatment. All patients who developed ARIA incurred the cost of at least one additional physician visit and monthly MRI scans until the abnormality resolved, typically averaging about three months. For the subset of patients whose ARIA was severe enough to require hospitalization, the modeled cost was substantial, with hospital stays averaging nearly 12 days at roughly $3,000 per day.

¹¹

For most patients, the financial impact lands somewhere between those extremes: the baseline MRI monitoring schedule is a fixed cost of treatment, and an ARIA episode adds a few extra scans and office visits. But for the small percentage who experience symptomatic, severe ARIA, the costs escalate quickly. That same analysis used delirium as an analogy for how ARIA symptoms affect quality of life, applying a meaningful quality-of-life reduction lasting about three months for symptomatic cases. For patients and families already managing the emotional and financial weight of an Alzheimer’s diagnosis, these additional costs and disruptions are not trivial considerations.

How Aduhelm’s Side Effect Profile Compares to Newer Drugs

Aduhelm was the first anti-amyloid antibody approved for Alzheimer’s, but it was quickly followed by lecanemab (Leqembi) and donanemab (Kisunla). All three drugs share the same basic side effect profile because they all work by targeting amyloid, but the rates and severity differ somewhat. A network meta-analysis comparing these drugs found that aducanumab was less well tolerated than placebo, consistent with the ARIA burden seen in its trials. Donanemab and lecanemab were found to be significantly less acceptable and less tolerable than placebo as well.

¹²

A real-world analysis using the FDA’s adverse event reporting system compared the safety signals of lecanemab and aducanumab and found that for both drugs, the only organ system class that triggered a strong safety signal across all detection methods was nervous system disorders. Both drugs showed strong individual signals for cerebral microbleeds, ARIA, and superficial siderosis. One difference worth noting: the median time from starting treatment to the onset of an adverse event was considerably shorter with lecanemab (about 33 days) than with aducanumab (about 146 days), likely reflecting differences in dosing schedules and how quickly therapeutic levels are reached.

¹³

These comparisons matter because Biogen withdrew Aduhelm from the market in early 2024, citing business rather than safety reasons, so most patients today are being treated with one of the newer agents. But the ARIA risk is a class-wide phenomenon for anti-amyloid antibodies, and the lessons learned from Aduhelm’s side effect profile directly informed the monitoring guidelines and dosing strategies now used for its successors.

Gaps in What We Know

Despite the extensive clinical trial data, there are still meaningful unknowns about Aduhelm’s side effects. The clinical trials enrolled carefully selected patients: relatively young for an Alzheimer’s population, with few other medical conditions and limited use of blood thinners. Real-world patients are messier. They tend to be older, take anticoagulants, and have conditions like uncontrolled hypertension or prior strokes that were exclusion criteria in the trials. How ARIA behaves in these higher-risk patients is not well characterized by the existing data.

Pathological and clinical data on what predicts who will develop severe or symptomatic ARIA, as opposed to the mild, silent variety, remain limited. An update from the Alzheimer’s Association ARIA workgroup noted that this gap underscores the need for continued real-world data collection.

¹⁴

There is also uncertainty about whether repeated episodes of ARIA, even ones that resolve on imaging, leave behind subtle neurological damage that accumulates over time. The trials were not long enough and did not have the right tools to answer that question definitively. For a drug class that may need to be given for years to maintain amyloid clearance, this is not a small question. Until longer-term follow-up studies report results, clinicians and patients are making decisions with incomplete information about the long-range consequences of these brain changes.