News
Article
Author(s):
The software, designed using thousands of brain MRI scans, is used for automated detection and severity grading of ARIA-E and ARIA-H for safety monitoring of new Alzheimer disease therapies.
According to an announcement, the FDA has cleared icobrain aria (icometrix), the first artificial-intelligence (AI) software approved to detect, measure, and grade amyloid-related imaging abnormalities (ARIA), a potential serious adverse effect of new anti-amyloid treatments for Alzheimer disease (AD).1
This software, trained using thousands of brain MRI scans, generates a quantitative summary report with includes a severity rating score that is integrated into the hospital workflow. For ARIA-edema (ARIA-E), the number of ARIA-E sites, their locations, and the longest axis of the largest site are measured and combined into an area severity score, graded according to clinical guidelines. For ARIA-hemorrhages (ARIA-H), the number of new microhemorrhages and areas of superficial siderosis are individually counted and localized, with an ARIA-H severity score reported, graded per clinical guidelines.2
"Amyloid-lowering treatments represent an important advance in the treatment of Alzheimer's disease but they are associated with a risk of brain swelling and hemorrhage called ARIA," Stephen Salloway, MD, professor of neurology and psychiatry at the Warren Alpert Medical School of Brown University, said in a statement.1 "New standardized tools are needed, such as icobrain aria, to assist radiologists and treating clinicians in detecting and managing ARIA to optimize patient safety."
He added, "I am excited that icobrain aria has received FDA approval, clearing the way for wider use in clinical practice."
Icobrain aria was validated in a large-scale study that was presented earlier this year at the American Academy of Neurology Annual Meeting. In the study, 16 radiologists evaluated ARIA for 199 cases, assisted and unassisted by icobrain aria. All told, sensitivity and detection for ARIA-E and ARIA-E increased significantly when reading assisted by icobrain aria compared with unassisted reads.3
This was true across mild ARIA (from 47.2% to 70.2%), mild-or-severe ARIA-E (83.7% to 95.3%), mild ARIA-H (60.0% to 71.0%), and moderate-or-severe ARIA-H (85.5% to 94.4%). In ARIA-free cases, assisted reads had specificity of 83.0% for mild ARIA-E, 80.3% for mild ARIA-H, 94.0% for moderate-or-severe ARIA-E, and 99.4% for moderate-or-severe ARIA-H. Overall, the investigators concluded that the software may help improve safety monitoring for patients on antiamyloid treatments, as well as help bridge the gap between expert and non-expert readers.
"We are thrilled to announce FDA authorization of the first-ever computer-aided detection and diagnosis solution in neuroradiology,” Dirk Smeets, chief officer at icometrix, said in a statement.1 "The FDA holds devices in this category to exceptionally high standards of evidence and performance, and this approval reflects the strong evidence behind icobrain aria's ability to help radiologists assess important safety events that can impact treatment and patient care."
READ MORE: Phase 2 cAPPricorn Study to Assess RNA Therapeutic Mivelsiran in Cerebral Amyloid Angiopathy
The effects of ARIA have been an issue with antiamyloid treatments throughout their early development even until the latest approvals. They were further amplified with the 2021 approval of aducanumab (Aduhelm; Biogen), which showed AREA-E rates of 25.7% and 25.5% in the notable phase 3 EMERGE and ENGAGE low-dose groups, respectively, and rates of 34% and 35.5% in the high-dose groups, respective. The drug was approved but rarely used, had high costs associated, and was subsequently taken off the market earlier this year by Biogen in efforts to divert funds and resources to other pipeline agents.4,5
The second of the antiamyloid treatments, lecanemab (Leeqembi; Eisai), had less ARIA-E observed in its confirmatory phase 3 study (Clarity AD; NCT03887455), at a rate of 12.5%.6 ARIA-H occurred at a rate of 17.3% in lecanemab-treated individuals and 9.0% for those on placebo. The drug, which is approved for the earliest stage of AD, has prescribing information that states that testing for apolipoprotein e4 status should be performed before starting treatment to inform the risk of ARIA.7
Donanemab (Kisulna; Eli Lilly), the latest of the antiamyloid treatments, was approved based on data from the TRAILBLAZER-ALZ-2 trial (NCT04437511), large-scale, phase 3, double-blind, placebo-controlled trial of 1736 patients with early-stage AD. In total, 24% of patients in the study experienced ARIA-E, with most ARIA-E events found to be largely mild to moderate radiographically. In terms of timing, ARIA-E first occurred after receiving up to 3 donanemab infusions in most cases (58%). First ARIA-E events radiographically resolved in 98% of participants, with a mean resolution time of around 10 weeks. Of note, 6% of patients on active treatment experienced recurrent ARIA-E.8
"For families facing Alzheimer's disease, the new disease-modifying therapies have brought renewed hope. But to help as many people as possible, we do need to standardize care pathways and make sure we have access to cognitive tests, blood biomarkers, and imaging exams that can ensure a timely diagnosis and a safe drug prescription,” George Vradenburg, chairman of UsAgainstAlzheimer’s, said in a statement.1 "Thanks to the FDA clearance of icobrain aria, Alzheimer therapies are more accessible and safer for patients, as it provides clinicians with the tools they need to make informed, effective treatment decisions."