NeuroVoices: Cynthia Lemere, PhD, on Exploring the Role of Complement in ARIA and CAA

Cynthia A. Lemere, PhD

As Alzheimer disease (AD) research advances, there is growing interest in the relationship between complement, cerebral amyloid angiopathy (CAA), and amyloid-related imaging abnormalities (ARIA). Anti-amyloid immunotherapy has shown promise in targeting beta-amyloid plaques, but these treatments can also cause ARIA, a vascular side effect ranging from asymptomatic edema to microhemorrhages. New studies are beginning to uncover how the complement system may play a key role in these adverse reactions.

CAA, where amyloid plaques accumulate in brain blood vessels, is closely linked to ARIA, especially with anti-amyloid treatments. The complement system, a critical part of the immune response, is increasingly being studied for its involvement in ARIA development. Understanding this process not only helps us better grasp the side effects of these therapies but also informs strategies to reduce ARIA while maintaining the benefits of anti-amyloid treatments for Alzheimer's patients.

As part of a new NeuroVoices, NeurologyLive® sat down with Cynthia A. Lemere, PhD, a dementia expert, who plans to present a talk on complement, CAA, and anti-amyloid induced ARIA at the 2025 AD/PD International Conference on Alzheimer’s and Parkinson’s Diseases, held April 1-5 in Vienna, Austria. Lemere, a professor of neurology at Brigham and Women’s Hospital, provided context on her talk, as well as her longstanding research on this topic, and how early involvement of complement C1q might be a precursor to the development of ARIA. She also discussed the potential for mitigating ARIA through targeted therapies, including modulating complement activation and identifying new biomarkers.

NeurologyLive: Why was this a topic of interest for you?

Cynthia Lemere, PhD: For many years, in fact, decades now, I've been working on anti-amyloid immunotherapy in the non-clinical world—mostly testing it in mice and also in some tissue culture assays. My goal has always been to help it move to the clinic, which is important. I started working on this in 1997, and then, as we became aware of the vascular side effect known as amyloid-related imaging abnormalities (ARIA), I became very interested in that. While I was pleased that anti-amyloid immunotherapy was moving forward, I also realized that ARIA could be a serious side effect in about one to one-and-a-half percent of patients. In the rest of the patients, it tends to be asymptomatic. There are two forms of ARIA: ARIA-E, which is edema or effusion, and ARIA-H, which is hemorrhage or microhemorrhages. ARIA-E typically occurs within the first two to four immunizations or doses, while ARIA-H tends to appear later in treatment. My lab became interested in understanding the cause of this vascular side effect.

At the same time, my lab has been studying complement since, I think, our first publication in 2000. We’ve been very interested in complement, particularly its relation to synapse loss in the aging brain and in Alzheimer's disease. In learning about complement, we knew that the classical complement cascade has a strong role in removing immune complexes. Immune complexes are when an antibody binds to its antigen or protein. For example, an anti-amyloid antibody binding to beta-amyloid in the brain activates an immune response, and complement C1q should come in and bind to its binding site on the antibody. All IgG antibodies have a portion that binds to C1q, which activates the classical complement cascade and helps clear or phagocytose the immune complex. This process also happens with viruses or pathogens. The antibody binds to the pathogen, and then C1q tags it, initiating the classical complement cascade.

We were very interested to see if the same process might be happening in the brains of mice and people receiving anti-amyloid antibodies. In our studies, we've found that very early on—keep in mind, ARIA-E happens very early in humans—we can see this process in mice. When you immunize a mouse for 13 to 16 weeks in a row, you can actually detect microhemorrhages with the murine version of bapinezumab. Bapinezumab was the first antibody to go into human clinical trials targeting beta-amyloid, and it induced ARIA. This is when the definition of ARIA first emerged. The murine version of bapinezumab is very reliable and always induces microhemorrhages in old mice that have CAA (cerebral amyloid angiopathy), which is vascular amyloid.

We started looking at what happens in the very early stages after immunization, because in humans, ARIA happens early. We immunized mice on day one, then again on day four, and euthanized them on day seven. What we found was that the antibodies were only binding to vascular amyloid; they didn’t get into the parenchyma to bind plaques. Where they did bind to vascular amyloid, we saw a big increase in complement C1q binding. We also saw co-localization of the antibody with CAA and C1q in the brain, which provides strong evidence that complement is activated very early on. Complement has downstream effects, including phagocytosis of the immune complex, but other components of the cascade can also recruit immune cells to the area. We’re now starting to investigate which immune cells are being recruited. We’re doing RNA sequencing to look at this.

In one part of my presentation, after seven immunizations, we looked at bulk RNA sequencing and saw very strong upregulation of innate immune genes, including complement, macrophage, and astrocytic genes. So, there’s definitely a local immune response, at least pathologically. When we looked at blood samples from these mice, we saw no difference in C1q levels between mice immunized three to six times and control antibody mice. However, when we looked at complement C3, which is further downstream in the cascade, we saw a huge increase in C3 in the brain, and a very significant reduction in C3 in the blood. When we looked at C3 levels in the brain and compared them to microhemorrhages, we found a striking correlation. This relationship is something we’re very interested in, and we think there may be a temporal pattern to ARIA. Early on, it may be activated by C1q binding to the antibody and CAA, and as that progresses, there is more leakage around the blood-brain barrier, leading to more antibodies binding to plaques in the tissue. Over time, this results in upregulation of C3, which seems to be more involved in the microhemorrhage effects.

What had been researched previously about complement and its role in ARIA? How much did you add to the puzzle, or did you start the puzzle?

I don't think anybody else had proposed the complement hypothesis for ARIA before I did. I first proposed it almost two years ago at a different meeting when I gave a plenary talk on complement. I suggested that complement might be involved in ARIA, and I gave one example from a recent study. Now, we’re finishing up a very large paper that we hope to submit next week. It contains a lot more data, and I believe it strongly supports the idea that complement is involved in ARIA. There's also now evidence from others. For example, David Gates is presenting at this meeting, and he published a paper in Nature showing that complement is one of the genes that’s strongly upregulated in human brains with ARIA. However, in his study, the patients had been treated with TPA in the emergency room, and unfortunately, they succumbed to ARIA. This type of situation is different from what happens early on in patients who, for example, don’t know they’re having ARIA. Most patients with ARIA-E don’t have symptoms; it’s only detected by MRI. These are the people I want to better understand—what’s happening early on that doesn’t cause major symptoms? I believe ARIA-E precedes ARIA-H, and that it sets the stage for ARIA-H later on.

There are microhemorrhages in CAA and in Alzheimer’s brains without any anti-amyloid antibodies, so one risk factor for ARIA is the presence of microhemorrhages prior to anti-amyloid immunotherapy. This suggests that the process may already be underway, with complement already involved, and the vascular remodeling may already be in motion. Adding an antibody could tip the balance.

How do we take what we know and continue to advance it? Obviously, we’re not suggesting we need to bring this into the clinic right now, but how do we continue to advance this research?

I’m collaborating with a group in our hospital system, and for a long time, I’ve wanted to get blood samples to look at complement levels before, during, and after ARIA in patients. If I were to guess, I’d say that C3 might be the key biomarker. We don't know yet in humans, but the studies are underway, and there’s a huge interest in this now. If we can find better biomarkers for CAA, it could help determine who is at risk. Right now, it’s hard to detect CAA in the living brain. If someone has microhemorrhages, they likely have CAA, but you can’t distinguish between vascular amyloid and plaque amyloid with a PET scan. We need better biomarkers for CAA.

I also think going earlier in the disease is important. There’s a strong movement to treat earlier, and I believe it’s the right approach. Treating patients with amyloid or CAA pathology before clinical symptoms appear may improve results. In the future, I envision a 40-year-old getting an anti-amyloid vaccine during a routine checkup, protecting them against Alzheimer's later on.

What additional research should we prioritize to understand the relationship between complement, CAA, and ARIA?

We need more information about how ARIA, CAA, and complement interact. But we also need to think about mitigation strategies. For example, if C1q is involved early, is there a way to block it and still clear beta-amyloid? I’ve shown a small pilot study in my talk where we mutated the antibody at the C1q binding site and saw reduced microhemorrhages, though it wasn’t statistically significant. We also found that this mutation almost completely prevented red blood cell extravasation, which suggests that blocking C1q might reduce ARIA. But there are other ways to block complement, and other groups are investigating them.

We’re also looking at treatments that target the vasculature, such as semaglutide (Ozempic), which has pro-vascular and anti-inflammatory properties. We have an NIH grant to investigate whether combining semaglutide with anti-amyloid immunotherapy can help suppress ARIA.

Do you think the clinical community would be more open to therapeutics that might not seem as fundamentally safe, if we’re able to mitigate ARIA?

I think clinicians are focused on the benefit-risk ratio. While some see the benefits as modest, others may view the positive effects on daily living activities—such as slowing the decline in ADLs—as significant. People need to look at the overall cognitive and behavioral benefits, not just the primary endpoint of clinical trials.

There are efforts underway to reduce ARIA with the antibodies themselves. For example, Roche has developed a bispecific antibody that uses a transferrin receptor to bring the antibody into the brain via the choroid plexus. This approach could avoid targeting vascular amyloid in the larger blood vessels, reducing the risk of ARIA. I believe there will be successes in reducing ARIA in the next few years, either by modifying the antibodies themselves or through mitigation strategies. Combining anti-amyloid immunotherapy with other treatments, like antihypertensives, could also help lower the risk of ARIA. Additionally, going earlier in treatment will likely improve outcomes. I’m hopeful that combining anti-amyloid immunotherapy with other strategies will boost both safety and cognitive outcomes in the long term.