# Brain of Woman Who Died on Leqembi Shows Worst-Case Scenario
Date: 26 Jan 2024
What happens in the brain of a person with fatal ARIA? Researchers now have one of their first detailed looks at the pathology of this rare and calamitous condition. In the December 12 Nature Communications, researchers led by Matthew Schrag at Vanderbilt University Medical Center, Nashville, Tennessee, published postmortem data from a woman who received three doses of the anti-amyloid antibody lecanemab in the open-label extension of the Phase 3 Clarity study. The autopsy showed extensive vascular inflammation and numerous ruptured blood vessels that led to bleeding throughout her brain. She likely died of complications from lecanemab treatment, the authors concluded.
The second neuropathological study of active ARIA, like the first, finds vascular inflammation throughout the brain.
The condition resembles inflammation related to cerebral amyloid angiopathy.
Aβ-immune complexes trigger perivascular macrophages to damage vessels.
Costantino Iadecola at Weill Cornell Medical College, New York, said the inflammatory findings resemble those seen in trials of earlier immunotherapies, such as bapineuzumab and the active vaccine AN-1792. He suggested that the term ARIA, i.e., amyloid-related imaging abnormalities, is a misnomer for what is actually Aβ-related vascular inflammation. “The emerging picture is of an innate immune attack on blood vessels, triggered by Aβ,” he told Alzforum.
Why did treatment go so badly for this woman? Schrag and colleagues found she had previously unrecognized risk factors that made her a poor candidate for immunotherapy, being an APOE4 homozygote with severe cerebral amyloid angiopathy.
Several experts told Alzforum the presence of CAA is the critical factor sparking ARIA. “The data … support the hypothesis that the mechanism underlying ARIA closely relates to CAA and likely shares many mechanistic links with CAA-related inflammation,” Mariel Kozberg at Massachusetts General Hospital, Boston, wrote to Alzforum.
Researchers Alzforum spoke with emphasized that to avoid future deaths and other serious outcomes, clinicians will need to screen potential immunotherapy patients carefully, looking in particular for signs of CAA. With thousands of people nationwide now being treated with lecanemab for early Alzheimer’s disease, clinicians must quickly learn to do this well.
Bursting Pipes. In a woman who died from ARIA after lecanemab treatment, blood vessels (green) in the meninges were packed with amyloid (red) and bulging with multiple aneurysms (arrows, inset). [Courtesy of Solopova et al., Nature Communications.]
### Catastrophic Vascular Breakdown
Previous autopsy data have been sparse, but a few cases from the AN-1792 and bapineuzumab trials had already shown that amyloid immunotherapy can worsen CAA. This is believed to happen because Aβ freed from plaques drains into the basement membrane around blood vessels and gets stuck there. These clogs are associated with inflammation and vascular damage. Nonetheless, this inflammation had not been directly linked to ARIA, which was first defined in conjunction with the bapineuzumab trials (Jul 2011 conference news). Autopsies of people taking bapineuzumab were done long after their ARIA had resolved.
More recently, three deaths in the Clarity OLE were blamed on ARIA, raising fresh concerns. An autopsy report from one of those deaths, a 65-year-old woman who suffered an ischemic stroke while on lecanemab and received the clot-busting drug tPA in the emergency room, showed massive cerebral bleeding as the cause of death. She was also an APOE4 homozygote with CAA. Her large number of hemorrhages would be unlikely to be caused by tPA and CAA alone, the pathologists concluded.
The new paper describes the pathology of another one of the OLE deaths, a 79-year-old woman in Florida who had been on placebo in the Clarity trial. Transitioned to lecanemab in the OLE, after each dose she experienced severe headaches that required a day or two in bed to recover. After the third dose, she became confused, and said she had “brain fog.” She started seizing, was taken to the hospital, and died five days later.
Black Dots. Pathologists counted more than 30 microhemorrhages (arrows) scattered throughout the brain of a woman who died with ARIA (left). Not all of them showed up on MRI (right). [Courtesy of Solopova et al., Nature Communications.]
Joint first authors Elena Solopova and Wilber Romero-Fernandez, both at Vanderbilt, compared her MRI at the time of death to her baseline MRI before the start of OLE. At baseline, four microhemorrhages were visible, just below the cutoff for treatment with lecanemab. Significantly, however, her baseline MRI met the criteria for probable CAA, the authors noted. CAA was not an exclusion criteria in the Clarity trial. Emerging evidence suggests severe CAA should contraindicate amyloid immunotherapy.
After lecanemab treatment, edema was evident throughout her temporal, parietal, and occipital lobes, and more than 30 microhemorrhages were scattered across these areas. As expected, the autopsy identified even more ruptured vessels that the MRI had missed. Routine clinical MRIs are lower-resolution than postmortem examination.
“The amount of white-matter lesions was surprising to me,” Iadecola noted. “They blossomed all over her brain. Cases like this highlight how badly things can go.”
Rupture and Spill. Blood leaks from a split in a vessel wall (left). On right, same vessel is shown with amyloid deposits (red), the gap where it ruptured, and diffuse leakage (yellow). Vessels stained with isolectin (green). [Courtesy of Solopova et al., Nature Communications.]
Similar to the previous case reports, the Florida woman’s cerebral blood vessels were inflamed. Where there was edema, numerous activated macrophages and T cells crowded around arterioles in the parenchyma. Some of the macrophages were multinucleated giant cells, which usually form when several immune cells fuse in response to infection. Vessel walls were degenerating, their dying cells interspersed with antibody-Aβ complexes and deposits of fibrin. This is called fibroid necrosis. In the leptomeninges as well, the authors found leaking vessels packed with CAA and surrounded by activated macrophages and microglia.
The overall findings resembled CAA-related inflammation, the authors said. They noted that the neuropathology matched that of the previous APOE4 homozygote case described. “These two cases are the only detailed neuropathological descriptions of active ARIA in the existing literature,” Solopova and colleagues wrote.
### Time to Retire the Term “ARIA”?
Iadecola considers these neuropathological data indispensable for piecing together the puzzle of ARIA mechanisms, especially because there are no good animal models for the condition. He thinks the findings suggest that either Aβ or Aβ-antibody complexes around blood vessels trigger perivascular macrophages to spew cytokines and free radicals, which damage the vessels, causing them to spring leaks.
Other researchers believe excess Aβ itself is the trigger. “The findings offer support for the hypothesis that following immunotherapy against Aβ, the solubilized Aβ from plaques becomes entrapped in the pathways for intramural periarterial drainage,” said Roxana Carare at the University of Southampton, U.K. When those drainage pathways are compromised, for example in someone with cerebrovascular damage, the person is unable to clear Aβ, and ARIA develops, she wrote.
Meanwhile, recent work from Eli Lilly places more blame on antibody complexes than Aβ alone. In mouse models, they found that antibodies interacted with vascular amyloid to activate perivascular macrophages. “These findings again suggest the importance of perivascular macrophages in the vascular inflammation associated with ARIA,” Kozberg noted.
Iadecola also points an accusing finger at these cells. “Leptomeningeal and perivascular macrophages are really what’s causing the problem,” he told Alzforum.
As it is dawning on researchers what ARIA really is, some researchers think the term may have outlived its usefulness. “It’s no longer an ‘imaging abnormality,'” Iadecola said. Fabrizio Piazza of the University of Milano-Bicocca, Italy, believes it should be called iatrogenic CAA-related inflammation. “ARIA-E is the imaging manifestation of the complex (auto)immune biological mechanisms of CAA-ri,” he wrote.
### How To Avoid ARIA?
It’s stating the obvious that prescribers want screening protocols for amyloid immunotherapy that are keenly sensitive to preventing such outcomes in the future. One approach could be to set a lower cutoff for microhemorrhages, which often indicate the presence of CAA. Solopova and colleagues wrote in their paper’s discussion that the U.S. Food and Drug Administration had initially suggested excluding anyone with two or more microhemorrhages from participating in amyloid immunotherapy trials. Early on, a working group convened by the Alzheimer’s Association recommended the current standard of four.
Recent data hint that a lower number could improve safety. Analyzing data from donanemab trials, Steven Greenberg at Massachusetts General Hospital, Boston, found that the presence of two or more microhemorrhages at baseline doubled a person’s odds of developing ARIA-E during the trial.
For their part, Solopova and colleagues recommended standardizing the MRI protocol to enhance detection of microhemorrhages, as well as diagnosis of probable CAA. They suggested a minimum field strength of 3 Tesla, using susceptibility-weighted imaging at a slice thickness of no more than 5 mm. This would offer higher resolution for spotting tiny bleeds. Many hospitals and outpatient centers have been upgrading from 1.5T to 3T scanners in the past decade, though not all have them.
Some programs are already doing this. “We’ve worked with our neuroradiologist to develop an MRI sequence with a susceptibility-weighted imaging sequence that is very sensitive for microbleeds,” Erik Musiek at Washington University in St. Louis told Alzforum.
Kozberg noted that where in the brain microbleeds are matters, as well. Leakage in deep brain regions is not linked to CAA, whereas lobar bleeds often indicate the presence of extensive vascular amyloid.
In hindsight, researchers now believe that some of the deaths in immunotherapy trials to date could have been prevented with better baseline screening or by stopping infusions sooner.
The FDA has reserved judgment, saying the evidence so far is inconclusive for amyloid immunotherapy causing death. In briefing documents for lecanemab approval, the agency wrote that because the two deaths occurred in the OLE, there was no placebo control for comparison. In addition, the FDA noted the existence of many confounding factors, such as other medications, stroke, and falls, that surrounded these deaths. “Any role for an interaction between lecanemab and underlying severe CAA or CAA-related inflammation/vasculitis cannot be determined,” the agency wrote.
Clinician-researchers believe caution is warranted. Stephen Salloway at Butler Hospital in Providence, Rhode Island, would like to see more training available for physicians, for example webinars on the topic, or an ARIA support line to call with questions. He also believes automated reads of MRIs could offer value to community physicians inexperienced in recognizing ARIA. For example, the company Icometrix, based in Leuven, Belgium, and in Boston, has developed machine-learning algorithms for MRI that help distinguish several neurological conditions and are FDA-approved for clinical use. “It would help clinicians to have that as a backup,” Salloway said.
### How to Make Treatment Better?
Besides more rigorous baseline evaluation, researchers emphasized the importance of following patients closely during treatment, and performing an MRI if ARIA symptoms show up. “I want to encourage clinicians to be attentive to these changes,” Salloway said. He noted that ARIA symptoms can mimic stroke, and stressed the importance of educating emergency-room physicians to be alert for this so they do not administer clot-busting drugs to someone on amyloid immunotherapy.
To help raise awareness, Greenberg organized an ARIA session at the International Stroke Conference to be held February 9 in Phoenix. Talks will cover mechanisms and neuropathology of ARIA, as well as how to recognize and treat it in hospitals.
What is the treatment for severe ARIA? Physicians typically administer steroids, but Iadecola noted that those do not always work. For example, the woman described in Solopova et al. received steroids, but continued to worsen. In CAA-ri, steroids do help, perhaps because they squelch production of the endogenous antibodies that are causing the problem. In ARIA caused by immunotherapy, however, antibodies are exogenous and thus not affected, Iadecola said.
“The best treatment for ARIA and the effectiveness of steroids is not currently known,” Greenberg agreed.
Kozberg noted that the vascular inflammation seen in recent case studies affected the vessel wall. This is known as Aβ-related angiitis (ABRA), and is distinct from CAA-ri, where inflammation is mostly perivascular. ABRA is often treated more aggressively than is CAA-ri, by adding cyclophosphamide to the usual intravenous steroids. Perhaps this could be tried for ARIA, Kozberg suggested.
Iadecola recommended scientists investigate ways to calm down perivascular macrophages, perhaps by suppressing receptors that bind Aβ. “That would be a more precision approach than steroids,” he noted. That work is yet to be done, however. “Until we figure that out, screening is key,” Iadecola added. Prevention is the name of the game, Salloway agreed.
Salloway noted that more neuropathology studies from ARIA cases are in press, including one about a death in the aducanumab trial. Together, the data may provide more clues as to how to make amyloid immunotherapy safer.—Madolyn Bowman Rogers
