Prediction of clinically important acquired cardiac disease without an echocardiogram in large breed dogs using a combination of clinical, radiographic and electrocardiographic variables
Wesselowski, S. et al (2021) Prediction of clinically important acquired cardiac disease without an echocardiogram in large breed dogs using a combination of clinical, radiographic and electrocardiographic variables. Journal of Veterinary Cardiology.
The objectives of this retrospective study were to (1) assess the clinical utility of thoracic radiographs and (2) to explore the potential utility of prediction models incorporating radiographic measurements, physical examination (PE) and electrocardiogram (ECG) to predict clinically important dilated cardiomyopathy (DCM) or myxomatous mitral valve disease (MMVD) in the absence of an echocardiogram in large-breed (LB) dogs.
Records for LB dogs weighing ≥ 20 kg that had echocardiograms and thoracic radiographs taken within 30 days of each other that had attended a US veterinary teaching hospital between 2010-2019 were retrieved.
The radiographs of included dogs were measured for vertebral heart size (VHS) and vertebral left atrial size (VLAS) by cardiologists blinded to the echocardiographic diagnosis. A single cardiologist, blinded to the radiographs, remeasured the echocardiographic cine loops and assigned the dog to one of six study groups: normal heart size (NHS); preclinical DCM (PC-DCM); clinical DCM (C-DCM); preclinical MMVD (PC-MMVD); clinical MMVD (C-MMVD); and equivocal (EQ).
455 dogs were included in the study. The overall prevalence of DCM and MMVD (clinical and preclinical) was 24.8% and 39.3%, respectively. However, 67.0% of dogs with MMVD in this population had echocardiographic evidence of MMVD, but no secondary heart enlargement. When considering dogs with clinical disease, DCM was more likely to be diagnosed than MMVD.
The ability of VHS to discriminate between NHS and clinical DCM/MMVD or preclinical DCM/MMVD as determined by echocardiography was 0.861 and 0.712, respectively, while for VLAS, it was 0.891 and 0.722, respectively. Dogs with scores of VLAS ≥2.8 or VHS ≥12.3 were shown to have an 89.7% and 89.2% chance of having underlying cardiac disease.
Predictive models incorporating physical examination and electrocardiography findings in addition to VHS/VLAS increased area under the curve to 0.978 (NHS vs. clinical DCM/MMVD) and 0.829 (NHS vs. preclinical DCM/MMVD). The three most important predictors from the physical examination were breed, murmur, and auscultable arrhythmia.
Limitations of the study are the retrospective nature of the study, incomplete data and issues relating to disease classification in LB dogs.
This study provides some evidence that thoracic radiographs can be used for the prediction of clinically important DCM or MMVD in large-breed dogs, in the absence of echocardiograms. The discriminatory ability will be improved if the findings from a physical examination are also considered.
Image copyright attribute: melnyk58