Performance of Echocardiographic Algorithms for Assessment of High Aortic Bioprosthetic Valve Gradients

BACKGROUND: Bioprosthetic aortic valve dysfunction (BAVD) is a challenging diagnosis. Commonly used algorithms to classify high-gradient BAVD are the 2009 American Society of Echocardiography (ASE), 2014 Blauwet-Miller, and 2016 European Association of Cardiovascular Imaging (EACVI). We sought 1) to evaluate the accuracy of existing algorithms against objectively proven BAVD and 2) to propose an improved algorithm.

METHODS: Retrospective study of 266 patients with objectively proven BAVD (pathology of explanted valves, 4D-CT prior to transcatheter valve-in-valve, or therapeutically confirmed bioprosthetic thrombosis) who were treated. Of those, 191 had obstruction, 48 had regurgitation, 15 had mixed stenosis and regurgitation, and 12 had patient-prosthesis mismatch (PPM). Normal controls were matched 1:1 (age, prosthesis size and type), of which 43 had high gradients (PPM in 30, high flow in 9 and normal prosthesis in 9). Algorithm assignment was based on the echocardiogram leading to BAVD diagnosis and the pre-discharge "fingerprint" echocardiogram after surgical or transcatheter aortic valve replacement. A novel algorithm (Mayo Clinic algorithm) incorporating valve appearance in addition to Doppler parameters was developed to improve observed deficiencies.

RESULTS: The accuracy of existing algorithms was suboptimal (2009 ASE: 62%; 2014 Blauwet-Miller: 62%; 2016 EACVI: 57%). The most common overdiagnosis was PPM (22-29% of patients and controls with high gradients). The novel Mayo Clinic algorithm correctly identified the mechanism in 256 of 307 patients and controls (83%). Recognition of regurgitation was substantially improved (42 of 47 patients, 89%) and the number of PPM misdiagnoses significantly reduced (5 patients).

CONCLUSION: Currently recommended algorithms misclassify a significant number of BAVD patients. The accuracy was improved by a newly proposed algorithm.