Diagnostic accuracy and inter-observer variability of cardiac and left atrial radiographic measurements in cats with and without cardiomyopathies
Inroduction Radiographic assessment of cardiac size and left atrial enlargement (LAE) can be challenging in cats with suspected heart disease. Several radiographic indices, including vertebral heart size (VHS) and left atrial-specific indices, such as left atrial vertebral heart size (LA-VHS), vertebral left atrial size (VLAS), and radiographic left atrial dimension (RLAD), have been proposed to assess LAE. However, their reproducibility, repeatability and diagnostic accuracy in cats have not been fully established. The aims of this study were to evaluate the inter-observer reliability, agreement and intra-observer repeatability of VHS, LA-VHS, VLAS, and RLAD in cats and to assess their diagnostic accuracy for identifying moderate-to-severe LAE. Methods In this retrospective, multicenter study, right lateral thoracic radiographs of cats of any age, sex, or body weight were retrieved from the medical databases of two institutions. Only cats that underwent radiographic and echocardiographic examinations during the same clinical evaluation were included. Echocardiography was used to diagnose feline cardiomyopathies (FCMPs) and grade the severity of LAE. Results A total of 86 cats were included, comprising 33 cardiologically healthy controls and 53 cats with FCMPs. Of these, 33 cats (38.4%) had no LAE, 12 (13.9 %) had mild LAE, 6 (7%) had moderate LAE, and 35 (40.7%) had severe LAE. Each radiograph was independently evaluated by three observers with different levels of clinical expertise in veterinary radiology and cardiology. Inter-observer reliability, intra-observer repeatability and agreement of the radiographic measurements, correlations between radiographic indices and echocardiographic assessment of LAE, and their diagnostic accuracy for detecting moderate-to-severe LAE were evaluated. All radiographic indices showed good-to-excellent inter-observer reliability, with intraclass correlation coefficients (ICCs) ranging from 0.85 to 0.97 and median coefficients of variation (CVs) ranging from 3% to 11%. Intra-observer repeatability was generally good to excellent, with ICCs ranging from 0.70 to 0.99 and median CVs ranging from 0.7% to 9.3%. However, significant differences in VHS and RLAD measurements were observed among observers (both p < 0.001). Diagnostic accuracy for detecting moderate-to-severe LAE was consistent across observers, with areas under the curve ranging from 0.73 to 0.83. The VHS, LA-VHS, VLAS and RLAD showed moderate-to-good diagnostic accuracy in cats. Conclusion These indices may therefore serve as useful supportive tools for screening and initial clinical assessment when LAE or underlying cardiac disease is suspected, although their diagnostic accuracy remains insufficient for use as stand-alone diagnostic tests in clinical practice.
Authors
- Carlotta Valente (ORCID: https://orcid.org/0000-0002-0189-7812)
- Helen Poser (ORCID: https://orcid.org/0000-0003-2323-7708)
- Tommaso Banzato (ORCID: https://orcid.org/0000-0002-1261-9033)
- Alessandro Zotti (ORCID: https://orcid.org/0000-0002-9279-5140)
- Barbara Contiero (ORCID: https://orcid.org/0000-0002-6586-5035)
- David Chiavegato
- Roberto Venturini (ORCID: https://orcid.org/0009-0002-6907-9928)
- Carlo Guglielmini (ORCID: https://orcid.org/0000-0001-7575-3646)
- Nicolò Mastromattei
Institutions
- University of Padua (IT)
Publication Details
- Journal
- Frontiers in Veterinary Science
- Published
- 2026-11-18
- DOI
- https://doi.org/10.3389/fvets.2026.1995426
- Primary Topic
- Cardiomyopathy and Myosin Studies
- Type
- article
- Field-Weighted Citation Impact
- 0.00