The FDA approval of Leqembi represented a pivotal moment for the Alzheimer’s disease (AD) community by providing stronger evidence than its predecessor, Aduhelm, that anti-amyloid antibodies can meaningfully slow cognitive decline. However, excitement around this news was tempered by two important observations. First, similar to Aduhelm, Leqembi does not reverse cognitive decline but slows it by about 30% over 18 months, which translates to approximately four to five additional months of preserved function in areas such as memory and activities of daily living. Second, like Aduhelm, its successor Kisunla, and prior experimental anti-amyloid immunotherapies, Leqembi can cause potentially serious side effects involving brain swelling and bleeding termed ARIA (Amyloid Related Imaging Abnormalities).
ARIA comes in two types: ARIA-E (brain swelling) and ARIA-H (brain bleeding). In clinical trials and consistent with real-world evidence since Leqembi’s approval, imaging scans found ARIA in up to 25% of all people who took Leqembi. For most of these people, ARIA had no symptoms and resolved without intervention. Less than 1% of treated patients had severe ARIA-E, and less than 2% had ARIA-E serious enough to merit stopping Leqembi. Kisunla presents very similar risks of asymptomatic and clinically meaningful ARIA. However, while severe ARIA is rare, it can be fatal, and carriers of the AD risk gene variant APOE4 face significantly increased risk of severe ARIA of both types. Although carrying APOE4 does not guarantee the development of AD, most Americans diagnosed with Alzheimer’s carry at least one copy. This creates a troubling paradox: those at the highest risk for the disease are also at the greatest risk for serious treatment complications. Understanding how amyloid-lowering antibodies cause ARIA is crucial for developing safer immunotherapies.
Dr. Lemere is interested in how anti-amyloid immunotherapies trigger ARIA. Her team discovered that an earlier experimental anti-amyloid immunotherapy antibody called 3D6 consistently caused brain microbleeds when given to aged amyloid mouse models that carried APOE4, reminiscent of the ARIAs seen in humans. 3D6 also led to an increase in immune cells in the brain vasculature and activation of the complement cascade, a pro-inflammatory immune pathway. Interestingly, these reactions only occurred in mouse models with amyloid aggregation inside the walls of the blood vessels (cerebral amyloid angiopathy or CAA). These observations suggested that 3D6 was binding to vascular amyloid deposits and activating complement proteins there, a mechanism that could explain ARIA in human patients with similar vascular amyloid pathology. Based on these findings, Dr. Lemere hypothesizes that ARIA is triggered when an antibody binds to amyloid aggregates in blood vessel walls (CAA) and recruits a complement protein called C1q, which causes inflammation and bleeding. She also suggests that intravenous (IV) delivery, as used in humans, may cause more severe vascular inflammation than other routes because it exposes the vasculature immediately to the full administered dose, whereas other routes provide slower, more gradual access to the vasculature.
In this proposal, Dr. Lemere and her team are investigating whether complement activation drives the cascade from anti-amyloid antibody administration to ARIA development, and whether different delivery methods can reduce ARIA side effects. The lab is using 5XE4 mice (5XFAD amyloid mice with human APOE4 gene), which develop CAA. Dr. Lemere’s team proposed two aims. In the first aim, they are determining if C1q binding to the antibody/amyloid complex is required for antibody-induced ARIA. For these experiments, they are engineering a modified version of 3D6 (called 3D6-k) that cannot bind C1q but retains its ability to bind and clear amyloid plaques from the brain and blood vessels. They are treating mice with 3D6, 3D6-k, or another control antibody, and measuring key outcomes related to inflammation, the complement pathway, immune and other blood cell markers, and amyloid levels after acute (10-day) and chronic (8-week) dosing schedules. In the second aim, they are testing whether the route of administration of 3D6—subcutaneous (under the skin) injection, intraperitoneal (into the abdomen) injection, or via IV infusion (the current standard for Leqembi)—yields different ARIA incidence or beta amyloid clearance. Overall, this work could provide a better mechanistic understanding of antibody-induced ARIA and identify practical strategies to improve the safety profile of current and future Alzheimer’s treatments.
In the first year of funding, Dr. Lemere’s team made several important findings related to the mechanisms underlying ARIA and the current amyloid mouse models used in studies of AD and ARIA throughout the field. They found an increase in complement proteins near amyloid plaques in the brain following the binding of anti-amyloid antibodies. This finding supports Dr. Lemere’s central hypothesis that the amyloid/antibody complex recruits complement proteins, which generate inflammation and vascular damage, ultimately leading to ARIA. Her team also discovered that certain breeding considerations had a previously unrecognized and significant impact on the level and timing of vascular amyloid pathology in this mouse model. This surprising finding highlights how subtle differences in mouse breeding can affect pathology phenotype—an important insight that will help other researchers save resources and improve experimental interpretation by accounting for these differences. In addition to these findings, Dr. Lemere’s team optimized several experimental pipelines, including plasma proteomics to examine changes in the blood of the mice throughout treatment. Further breeding of mice is currently underway, and Dr. Lemere anticipates that the remaining experiments will be completed during the second year of funding.
