[Interview] When Alzheimer’s patients can’t stay on anti-amyloid drugs, Illimis sees an opening

Leqembi (lecanemab) and Kisunla (donanemab) changed Alzheimer’s drug development. They showed that amyloid beta -- the protein that builds up in the brains of people with Alzheimer’s -- could be cleared, and that clearing it could slow decline enough to win approval.

They also left doctors with a narrower problem: what to do when a patient develops ARIA, the brain swelling or microbleeds seen on MRI after anti-amyloid treatment, and cannot stay on the drug.

“There is no second-line drug,” Park Sang-hoon, chief executive of Illimis Therapeutics, said in a recent interview with Korea Biomedical Review. In amyloid-beta therapy, he said, “if the first drug cannot be used anymore, there is no second option to give.”

The Korean biotech is developing an Alzheimer’s candidate called GAIA-Aβ. It still targets amyloid beta. The difference, Illimis says, is how the protein gets cleared.

Current antibodies clear amyloid through Fc receptors, proteins on immune cells that help drive plaque removal. Illimis says its drug is designed to use TAM receptors instead, a different pathway the company says can remove plaque without causing the same inflammatory damage.

Illimis presented animal data on the drug at the AD/PD 2026 International Conference on Alzheimer’s and Parkinson’s Diseases on March 19.

In a conference abstract, the company said the candidate reached the brain, cut amyloid plaque in mouse models and lowered pro-inflammatory cytokines, proteins released when the immune system is activated.

Park Sang-hoon, chief executive of Illimis Therapeutics, presenting the company’s Alzheimer’s data at the AD/PD 2026 conference on March 19 in Copenhagen. (Courtesy of Illimis Therapeutics)

Park argues that approved drugs already do the first part of the job: they lower amyloid. The unresolved question is how much damage comes with doing it.

Brain swelling, microbleeds, repeated MRI monitoring and a narrower eligible patient pool still hang over the class. “In my view, the currently approved drugs do not cover even half of all Alzheimer’s patients,” he said, pointing both to patients screened out before the first infusion and to those who start treatment but do not stay on it.

Some patients already show microbleeds on baseline scans. Others have heavier amyloid in the blood vessels or carry APOE4, a genetic variant linked to higher Alzheimer’s risk and, in some patients, greater concern about treatment-related bleeding. Others qualify, then face repeated MRI scans, more hospital visits and the chance that one troubling result ends treatment.

He said Korean doctors, especially at larger hospitals, have been cautious. In many cases, patients and families are the ones bringing up Leqembi first after reading about it and deciding they are willing to pay. That is a market that is still being pulled by demand, not pushed by routine prescribing.

Park said the current class is boxed in by dosing: raise the dose and ARIA gets worse; lower it and efficacy slips. That, he said, is also why so much work is going into shuttle approaches designed to get more drug into the brain at a lower dose.

At this year’s conference, he said, researchers kept presenting versions of the same result: change part of the antibody, pair it with an anti-inflammatory approach, adjust the dose, and ARIA falls. Illimis is arguing for something broader. “If you use our platform,” Park said, “all of that can in principle be solved at once.”

Illimis, he said, is not trying to just reduce ARIA. “We do not trigger the inflammatory response that is the root cause of ARIA in the first place.” In the company’s preclinical work, he added, the molecule showed “zero ARIA.”

A shuttle may reduce the degree of ARIA, he said. Illimis’s drug, by contrast, did not cause it at all in animal testing, even at high concentrations. That, he argued, could make it more applicable in high-risk patients -- people with APOE4, prior hemorrhage or prior ARIA -- the same patients current drugs often struggle to reach.

The last several years, Park said, went to two basic problems: whether using this immune pathway would create a different safety risk, and whether a GAS6-based molecule could be made, stored and handled like a practical drug.

He said the company has now set up manufacturing through a contract manufacturer and has not seen major preliminary toxicity signals in mice or monkeys. The next step is a phase 1 trial, which the company is aiming to start next year.

Park said the questions he kept hearing after the AD/PD 2026 presentation were narrower than the usual platform-company discussion. Could the drug be used after ARIA, in patients who stopped treatment, recovered and still could not safely restart the same antibody?

“That is a very good question, and in fact we are trying to approach it that way,” he said.