AI-enhanced echocardiography improves early detection of cardiac amyloidosis
![Ultromics EchoGo Amyloidosis artificial intelligence. A first of its kind AI model developed by Mayo Clinic and Ultromics Ltd., was found to be highly accurate in screening for cardiac amyloidosis in a large 2,612 patient clinical trial. The findings were published in the European Heart Journal.[1] Mayo Clinic, investigators at the University of Chicago Medicine and collaborators around the world at other sites, validated and tested the model on a large and multi-ethnic population.](/sites/default/files/styles/top_stories/public/2024-11/screenshot_2024-11-21_at_8.03.58_am.png.webp?itok=t6ancPrf)
The first artificial intelligence (AI) model of its kind, developed by Mayo Clinic and Ultromics, was found to be highly accurate in screening for cardiac amyloidosis in a new clinical trial of more than 2,600 patients. The findings were published in full European Heart Journal.[1]
Mayo Clinic, investigators at the University of Chicago Medicine and collaborators around the world at other sites validated and tested the model on a large and multi-ethnic patient population and compared its abilities to other diagnostic methods for cardiac amyloidosis. The AI model was found to be highly accurate, with 85% sensitivity for correctly identifying those with the disease and 93% specificity in correctly identifying those without the disease. Using a single echocardiography videoclip, the AI model was effective across all major types of cardiac amyloidosis and distinguished it from other conditions with similar characteristics.
"This AI model is a breakthrough tool that can help us identify patients earlier so they can receive the treatment they need. We have not previously had an accurate way of screening for amyloid cardiomyopathy and have relied on previously validated clinical scores for screening. These scores, the transthyretin cardiac amyloidosis score and increased wall thickness score, use echo measurements and clinical variables to predict risk of cardiac amyloid. However, accuracy of these scores is limited. Our model solidly outperformed both models in international testing. Our model interfaces with the echo PACS system and could be employed in echo labs anywhere," Patricia A. Pellikka, MD, a key investigator in the trial, told Cardiovascular Business. She is the Betty Knight Scripps Professor of Cardiovascular Disease Clinical Research, president of Mayo Clinic Officers and Councilors, editor-in-chief of the Journal of the American Society of Echocardiography and a consultant for the department of cardiovascular medicine at Mayo Clinic.
Cardiac amyloidosis is a life-threatening condition where an abnormal protein, called amyloid, builds up in the heart, causing it to stiffen. This then leads to heart failure. Cardiac amyloidosis is often missed because the symptoms and imaging features can be similar to other heart conditions. In the past, not much could be done for patients with the condition, but there has been an explosion of interest the past few years with new drug therapies now available. However, early diagnosis is crucial because the new drug therapies can slow or stop the disease's progression, but cannot reverse the impact of the disease progression that has already occurred.
Another ongoing challenge with diagnosing this condition is correctly identifying the type of amyloid, because the drug treatments are specific to the phenotype of the disease. These include light chain amyloidosis (AL) and transthyretin amyloidosis (ATTR). ATTR is further divided into wild-type (wtATTR) and hereditary (hATTR) forms.
"We used a very detailed schema to design our study. Patients with cardiac amyloid, both AL and ATTR types, were matched with controls. The matching plan was detailed and complex and included controls with echo features that mimic cardiac amyloid. This contributed to the excellent performance of the model in external testing. Our model performed well in detecting all types of amyloid cardiomyopathy with sensitivity of 84%, 85%, and 86%, for AL, ATTR wild type, and ATTR hereditary, respectively," Pellikka explained.
This work builds on the previous experience of Mayo Clinic and Ultromics in developing an AI echocardiography model to detect heart failure with preserved ejection fraction (HFpEF), which received Food and Drug Administration (FDA) clearance in 2022. HFpEF is a common type of heart failure associated with high morbidity and mortality; it can be challenging to diagnose. An estimated 15% of patients with HFpEF have cardiac amyloidosis.
The study's authors noted that consensus guidelines currently recommend patients with suspected cardiac amyloid undergo initial clinical evaluation using ECG and echocardiography. The presence of red flag features should then prompt further testing with cardiac magnetic resonance, technetium pyrophosphate scintigraphy (Tc-PYP), and/or invasive tissue biopsy. But given its low cost and widespread availability of echo, they said it would make sense to be able to pull more precise information out of the initial gatekeeper echo evaluation.
The amyloid AI model is FDA-cleared and is currently being used at multiple centers in the U.S. Pellikka said she looks forward to applying this technology in the clinical practice at Mayo Clinic.
Watch a video interview on using AI to detect cardiac amyloidosis with the lead author.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: [email protected]