Javier Palarea-Albaladejo ¹ Elizabeth F. Bode ^2,3 Catheryn Partington ^2,4 Mattia Basili ^2,3 Elzbieta Mederska ^2,5 Hannah Hodgkiss-Geere ² Paul Capewell ^6,7 Caroline Chauché ^7,8 Robert M. Coultous ⁷ Eve Hanks ⁷ Joanna Dukes-McEwan ^9*

• ¹ Dept Computer Sciences, Applied Mathematics and Statistics, University of Girona, Girona, Catalonia, Spain
• ² Department of Small Animal Clinical Sciences, School of Veterinary Science, Leahurst Campus, University of Liverpool, Liverpool, United Kingdom
• ³ ChesterGates Veterinary Specialists, Chester, United Kingdom
• ⁴ Department of Veterinary Medicine, Queen's Veterinary School Hospital, University of Cambridge, Cambridge, England, United Kingdom
• ⁵ Blaise Referrals Veterinary Hospital, Birmingham, England, United Kingdom
• ⁶ School of Molecular Biosciences, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, Scotland, United Kingdom
• ⁷ Other, Edinburgh, United Kingdom
• ⁸ Centre for Inflammation Research, The Queen’s Medical Research Institute, University of Edinburgh, Edinburgh, Scotland, United Kingdom
• ⁹ Dept Small Animal Clinical Sciences, School of Veterinary Sciences, Leahurst Campus, University of Liverpool, Liverpool, United Kingdom

Background: Myxomatous Mitral Valve Disease (MMVD) is a common, acquired, and progressive canine heart disease. Presence of heart murmur and current cardiac biomarkers are useful in MMVD cases but are not sufficiently discriminatory for staging an individual patient.Objectives: Preliminary aAssessment of the use of canine serum and plasma expression profiles of 15 selected miRNA markers as a method to accurately discriminate MMVD patients from healthy controls. Additionally, an assessment of the same method to discriminate pre-clinical (stage B1/B2) from clinical (stage C/D) MMVD patients.Animals: Client-owned dogs (n = 123) were recruited. Following sample exclusions (n = 26), healthy controls (n = 50) and MMVD cases (n = 47) were analyzed.Methods: Multicenter, cross-sectional, prospective investigation. MicroRNA expression profiles were compared between dogs, and the performance ofwere used as input in predictive modeling to distinguish healthy controls from MMVD patients, as well as to finer-grained discrimination between , and pre-clinical from and clinical MMVD patients, was evaluated.Results: Ordinary performance metrics Analysis of miRNA expression patterns by predictive classification algorithms could revealed a compelling ability of the method to differentiate healthy controls from dogs with MMVD (sensitivity .85; specificity .82; accuracy .83). Discrimination of pre-clinical (n = 29) from clinical (n = 18) MMVD cases resulted in promising results (sensitivity.61; specificity.79; accuracy.73). The method also compared advantageously to current biomarkers in a limited population.The analysis use of miRNA expression profiles in combination withby customized probabilistic predictive classification algorithmsmodelling provides shows good scope to devise a reliable useful diagnostic tool to distinguish healthy controls from MMVD cases (stage B1 to D). Investigation into the ability to dDiscriminatetion between pre-clinical and clinical MMVD cases by the same method yielded promising early results, which could be further enhanced with data from an increased study population.