Researchers have identified nearly three dozen blood biomarkers associated with Alzheimer’s disease (AD) that undergo significant changes during treatment with lecanemab, offering clues to how the brain responds to the drug and the removal of amyloid plaques. 

The real-world cohort study included analysis of 130 proteins in nearly 200 patients who received lecanemab infusions every 2 weeks for up to 2 years. Significant changes were noted in 34 proteins as patients received more lecanemab infusions, with some moving toward levels seen in cognitively healthy adults without amyloid buildup and others moving further away.

Levels of phosphorylated tau (p-tau) declined toward those seen in a control group of cognitively healthy, amyloid-negative participants, although after 2 years of treatment, p-tau remained 15%-44% higher than levels seen in healthy adults. The decline in one form, p-tau217, is consistent with clearance of amyloid plaques, researchers suggested. 

Several inflammatory proteins, including TREM2, a marker of activation of microglia in the brain, moved further from healthy concentrations with more infusions. None of the inflammatory changes were associated with cognitive decline during treatment, which suggests they may reflect the brain’s immune response to plaque removal rather than effects on neuronal function, the researchers noted.

The findings are an early step toward filling a gap in clinical management of AD — the lack of a blood test to monitor whether an anti-amyloid drug is working — although researchers caution that further validation is needed.

“P-tau217 is an outstanding biomarker of whether you have amyloid pathology, but once you start treating with these drugs, things get much more complicated, and there may be other biomarkers that are more informative,” co-senior investigator Suzanne Schindler, MD, PhD, professor of neurology at the Washington University School of Medicine in St. Louis, told Medscape Medical News. “Our ability to monitor the effects of treatment is very limited right now.”

The study was published online on September 24 in The Lancet Neurology.

Broader Biomarker Panel

Monitoring whether drugs like lecanemab are helping to clear amyloid plaques in an individual patient is difficult. Effects can vary from person to person, and amyloid PET scans, the only way to confirm plaques have cleared, are costly and hard to access in many places. A blood test that tracks specific proteins could address that problem.

Earlier research on blood biomarkers in patients on anti-amyloid therapy examined only a handful of proteins in studies with a short follow-up.

For this study, researchers analyzed a broader panel of blood proteins that reflect amyloid and tau pathology, inflammation, and nerve cell damage, looking for changes in patients treated in routine care that might signal whether the treatment was working. 

Researchers included 197 patients who received lecanemab 10 mg/kg intravenously every 2 weeks between July 2023 and October 2025. The median age was 73 years, 53% were women, and 98% of those with available data were White. Blood was drawn approximately every 6 months.

The healthy control group included 458 cognitively healthy adults whose amyloid-negative status was confirmed by PET or cerebrospinal fluid testing. A second control group included 1312 untreated adults with cognitive impairment.

Shifting Levels

Thirteen proteins normalized, moving toward concentrations in the healthy control group without reaching them, including p-tau, glial fibrillary acidic protein (GFAP), and the tau protein MAPT.

Before treatment, p-tau, which tracks amyloid and tau pathology, was 130%-192% higher than in the healthy control group. P-tau231 changed the most (P = .0020 vs p-tau217). In treated patients, GFAP was 131% above concentrations in the healthy control group before the first infusion and 39% above after 24 months of treatment.

Ten proteins overcorrected, moving past concentrations in the healthy control group, led by amyloid-beta 42, 40, and 38, which started more than 100% below levels in the healthy control group and rose to more than 100% above them after 18 months of treatment before declining. The investigators proposed that lecanemab binds and stabilizes amyloid-beta in the blood, raising its concentration above healthy levels, a pattern the authors associated with treatment-related amyloid clearance. 

Schindler said the team expected most biomarkers to simply normalize. “People have had this idea: ‘Oh, you’re just removing amyloid.’ No, actually, there’s lots of things going on when you remove amyloid,” she said.

Eleven proteins moved in the opposite direction, further from concentrations in the healthy control group with more infusions, including TREM2, which rose over the first year of treatment and then declined.

Such proteomic changes aren’t necessarily harmful, said co-senior investigator Carlos Cruchaga, PhD, founding director of the NeuroGenomics and Informatics Center at Washington University School of Medicine.

“If someone now is being treated and we are seeing these changes, we don’t need to be worried about them,” Cruchaga told Medscape Medical News. Cruchaga added the rise reflects microglia mobilizing to clear plaques and “trying to reset themselves.” 

Among patients scanned at 18 months, the decline was faster in those who remained amyloid positive, and untreated patients showed no similar rise. 

Clearance and Cognition

Of 57 patients who underwent amyloid PET after 18 months of treatment, 29 had converted to amyloid negative.

If validated, those biomarkers could eventually help clinicians predict amyloid clearance without repeating a PET scan, Schindler said, but “you need hundreds of people from multiple cohorts to assess that.”

A separate analysis examined whether baseline protein concentrations predicted the rate of cognitive decline during treatment, which could eventually help clinicians counsel patients and families about what to expect.

Among 126 patients with longitudinal cognitive data, higher baseline growth differentiation factor 15 (GDF15), a marker of oxidative stress and cellular aging, predicted faster decline. Patients in the highest GDF15 quartile worsened by 2.00 points on the Clinical Dementia Rating-Sum of Boxes over 50 infusions compared with 0.66 points in the lowest quartile (P = .0058). For NPTXR, which is involved in synapse organization, the highest quartile worsened by 1.03 points and the lowest by 2.05 points (P = .023).

Amyloid PET is currently the only way to confirm that plaques have cleared in an individual patient, but it costs about $3000 and exposes patients to radiation. Most patients in the study initially qualified for amyloid PET, but plasma p-tau217 testing, at $200-$300, has since become the more common route for confirming amyloid-positive status before starting treatment, Cruchaga said.

Blood markers that signal when a patient is likely to become amyloid negative could let clinicians time a single follow-up scan, and developing a laboratory test from a candidate protein in the current study would be straightforward, Schindler said.

The harder step is clinical validation, which requires multiple diverse cohorts, large sample sizes, and a clear reference standard.

It could take a few years to develop and validate a monitoring test based on these new candidate proteins, Cruchaga said. “The question is, how good will it be, and will it be good enough to be clinically useful?”

Existing p-tau217 blood tests are not suited to tracking amyloid clearance, she noted, “but it looks like there are better proteins for doing that, so I’m hopeful. We just need a lot more data.”

Limitations included enrollment from a single clinic and a largely White, non-Hispanic cohort. Neither control group was followed over time, so protein changes with treatment could not be directly compared with changes during untreated disease progression. Of the 197 treated patients, 45 (23%) developed amyloid-related imaging abnormalities, but blood was not drawn at those events.

The study included only patients on lecanemab, but the team is running similar analyses with donanemab, Cruchaga said.

Schindler cautioned against clinical use of any of these markers now. “Right now, I wouldn’t recommend any of these for treatment monitoring. I really wouldn’t,” she said. 

‘A Lot of Work to Do’

The study is a first step whose value for now is mainly conceptual, David Knopman, MD, professor of neurology at Mayo Clinic in Rochester, Minnesota, told Medscape Medical News.

Measuring 130 proteins without an independent data set to replicate the results raises the odds that some associations appeared by chance, Knopman noted.

“When you’re dealing with 130 proteins, the capitalization on chance in this situation is very high,” said Knopman, who was not involved in the study.

In addition, the analyses were based on group averages, and the spread of individual values was too wide to predict how a single patient would respond, he said. 

Whether the markers indicate that lecanemab is working or only how fast a patient’s disease will progress can’t be determined from this analysis, Knopman said.

Amyloid PET and lecanemab’s phase 3 trial results already show that the drug engages its target and produces a clinical benefit at the group level, but “the proteomic analysis doesn’t really address the issue of whether lecanemab worked. “It merely shows that there were changes that went along with the PET biomarker changes and the clinical changes, at least on a groupwise basis,” Knopman said.

“Maybe one of these proteins upon replication will really shine through as being predictive of clinical outcome. But you can’t tell that here,” he added.

Knopman added that a clinical test being available within a few years is unrealistic. The more useful direction, he said, is identifying what amyloid removal fails to address.

“The clinical trial evidence is pretty strong that there’s something there, but it’s weak,” he said. “What isn’t the amyloid removal doing that needs to be done?”

The study was funded by the National Institute on Aging, Cure Alzheimer’s Fund, and Washington University. Cruchaga reported serving on the scientific advisory boards of Admit Therapeutics, Alamar Biosciences, and Circular Genomics, in which he owns stock; consulting for Sanofi, Novo Nordisk, and Owkin; and receiving research support from GSK, Bristol Myers Squibb, Danaher, and Eisai. Other authors’ disclosures are listed in the original article. Knopman reported having no relevant financial disclosures.