An artificial intelligence (AI) tool can help identify heart transplant rejection and estimate its severity, results of a pilot study suggest.
The Cardiac Rejection Assessment Neural Estimator (CRANE) simultaneously addresses detection, subtyping, and grading of allograft rejection in H&E-stained whole-slide images of endomyocardial biopsy samples and is intended to be used in conjunction with the heart transplant team to more quickly and accurately diagnose rejection.
"Our retrospective pilot study demonstrated that combining artificial intelligence and human intelligence can improve expert agreement and reduce the time needed to evaluate biopsies," senior author Faisal Mahmood, PhD, Department of Pathology, Brigham and Women’s Hospital, Boston, said in a news release.
The study was published in the March issue of Nature Medicine.
Improving on the Standard of Care
Endomyocardial biopsy screening is the standard of care for detecting cardiac allograft rejection, but manual interpretation of surveillance endomyocardial biopsies remains a challenge, the authors note. Experts often disagree on whether or not the patient is rejecting the allograft and on the degree of severity of rejection when present.
Overestimation of rejection can lead to patient anxiety, overtreatment and unnecessary follow-up biopsies, whereas underestimation can lead to delays in treatment and worse outcomes.
"CRANE is an AI model which can act as an assistive tool to decrease such observer variability," Mahmood told theheart.org | Medscape Cardiology.
"While the final assessment will still be subjective, the CRANE model provides experts with an AI-driven prediction, as well as how confident it is in making these predictions," he said.
"To further instil confidence, the model also highlights the regions in the image it makes the prediction from," he added.
CRANE was trained on more than 5000 gigapixel whole-slide images from nearly 1700 patient endomyocardial biopsy samples collected at Brigham and Women's Hospital. Model performance was determined using a separate large dataset from the United States, as well as independent test cohorts from Turkey and Switzerland.
"These independent international test sets were deliberately constructed to reflect the high data variability present across populations and medical centers, as they used different biopsy protocols, slide preparation and staining mechanisms, and scanner vendors, which are all known contributors to image variability," the researchers note in their report.
The results show that CRANE detects allograft rejection, with an area under the receiver operating characteristic curve (AUC) of 0.962; assesses cellular and antibody-mediated rejection type with AUCs of 0.958 and 0.874, respectively; detects Quilty B lesions (benign mimics of rejection) with an AUC of 0.939; and differentiates between low- and high-grade rejection with an AUC of 0.833.
Moving From Glass Slides to Digital Scans
CRANE shows "non-inferior performance to conventional assessment and reduced inter-observer variability and assessment time," the researchers report.
Mahmood said further studies and clinical trials are needed to establish the overall efficacy of AI-assisted endomyocardial biopsy assessment for cardiac allograft rejection and its potential for improving heart transplant outcomes.
He noted that the biggest barrier to the broad applicability of such AI models is that most pathology departments in hospitals around the country still use glass slides under a microscope to make diagnostic assessments.
"First, there needs to be a broad transition from using physical glass slides to digitally scanned images and then we will see AI for pathology much more broadly applicable," he said.
Reached for comment, Preethi Pirlamarla, MD, a heart failure and transplantation cardiologist at Mount Sinai Queens in New York, said CRANE is a "very promising" tool that uses deep learning to enable better assessment of endomyocardial biopsies.
"One of the drawbacks with endomyocardial biopsies is there is very high inter-reader and inter-observer variability, as well as issues with sampling that may confound the results. This tool could serve as another layer to assess for cardiac allograft rejection," Pirlamarla told theheart.org | Medscape Cardiology.
A key strength of the study is the use of three different cohorts from the United States, Turkey, and Switzerland, with different protocols. "This really speaks to the stability of the program, even when you're using it across different countries and different protocols," Pirlamarla said.
This study "should springboard to further larger-scale trials," she added.
This work was supported in part by the Brigham and Women's (BWH) President's Fund, internal funds from BWH and Massachusetts General Hospital, National Institutes of Health, and the National Science Foundation. Mahmood and Pirlamarla report no relevant conflicts of interest.
Nature Med. 2022;28:575-582. Full text
Lead Image: Dreamstime
Medscape Medical News © 2022
Cite this: AI System Helps Spot Signs of Heart Transplant Rejection - Medscape - Mar 29, 2022.