Utility of radiographic measurements to predict echocardiographic left heart enlargement in dogs with preclinical myxomatous mitral valve disease

Abstract Background Evaluation of left heart size helps determine disease severity in dogs with myxomatous mitral valve disease (MMVD). Hypothesis/Objectives Determine the ability of radiographic vertebral heart size (VHS) and vertebral left atrial size (VLAS) to predict LHEECHO in dogs with preclinical MMVD. Animals Seventy client‐owned dogs with MMVD and no historical or present clinical or radiographic evidence of congestive heart failure (CHF). Methods Retrospective cross‐sectional study of dogs with same‐day echocardiography and thoracic radiography. Receiver‐operating characteristic (ROC) curves were used to assess the ability of VHS, VLAS, and VHS + VLAS to discern dogs with and without LHEECHO, and clinically relevant cutpoints for these radiographic measurements were selected. Results The ability of VHS and VHS + VLAS to predict LHEECHO was moderate (area under the curve [AUC]VHS = 0.851; 95% CI, 0.762‐0.941; AUCVHS + VLAS = 0.865; 0.783‐0.947), and performance of VLAS and VHS + VLAS was not different from that of VHS alone. A VHS cutpoint of >10.8 had sensitivity = 91.1% (76.3%‐98.1%) and specificity = 69.4% (51.9%‐83.7%) for predicting LHEECHO. A cutpoint of >11.7 had sensitivity = 32.4% (17.4%‐50.5%) and specificity = 97.2% (85.5%‐99.9%) for predicting LHEECHO. Thirty (43%) of the 70 dogs had a VHS value of 10.9 to 11.7. Conclusions and Clinical Importance Vertebral heart size >11.7 identified dogs with LHEECHO and VHS ≤ 10.8 excluded dogs with LHEECHO. A large percentage of dogs had VHS values intermediate to these cutpoints.


| INTRODUCTION
Myxomatous mitral valve disease (MMVD) is the most common cardiovascular disease in dogs, 1,2 with higher prevalence in small breed dogs (≤20 kg). 3 Resultant mitral regurgitation (MR) can lead to cardiac remodeling and dilatation of the left ventricular and left atrial chambers. Evaluation of left heart size in asymptomatic dogs is performed to determine severity of disease and identify dogs that could benefit from medical treatment. Administration of pimobendan to dogs with preclinical MMVD and specific radiographic and echocardiographic evidence of left sided cardiomegaly, including normalized LV internal diameter in diastole (LVIDDN) ≥ 1.7, as measured on M-mode echocardiogram obtained from the right parasternal short-axis view, and left atrial to aortic (LA:Ao) ratio ≥ 1.6), as measured from the right parasternal short-axis 2D view, and 1 radiographic measurement, vertebral heart size (VHS) >10.5, resulted in prolongation of the preclinical period. 4 Echocardiography is considered the gold standard for evaluating left atrial and left ventricular chamber size, however the skill and expertise required to perform and interpret echocardiographic examinations is often not available. In contrast, thoracic radiography is widely available and vertebral heart size (VHS) is a technique used to quantify cardiac size in dogs, including dogs with MMVD. 5 There are parallel increases in VHS and echocardiographic left atrial and left ventricular dimensions before onset of congestive heart failure dogs with MMVD. [6][7][8] Vertebral left atrial size (VLAS), a radiographic measurement specifically targeted to reflect left atrial size, is a useful and repeatable predictor of echocardiographically confirmed left atrial enlargement in dogs with MMVD. 9 The primary objective of this study was to evaluate the ability of VHS and VLAS to predict echocardiographic criteria for left heart enlargement (LHE ECHO ) in dogs with preclinical MMVD. We hypothesized that radiographic measurements would predict LHE ECHO and would help determine severity of disease in dogs with preclinical MMVD. We aimed to identify clinically relevant values of VHS, VLAS, and sum of VHS + VLAS that predicted presence and absence of LHE ECHO .

| MATERIALS AND METHODS
Records of client-owned dogs with MMVD and no current or previous clinical or radiographic signs of congestive heart failure presenting to Matthew J. Ryan Veterinary Hospital of the University of Pennsylvania from July 2016 to December 2018 were retrospectively reviewed.
Dogs were included based on the following criteria: ≥5 years of age, body weight ≤ 15 kg, systolic heart murmur (≥grade 2/6) with maximal intensity over the mitral valve, echocardiographic evidence of MMVD defined as characteristic thickening and prolapse of the mitral valve leaflets and MR on color flow Doppler, and fractional shortening >30%. Dogs with current or previous clinical or radiographic evidence of cardiogenic pulmonary edema, cardiac disease other than MMVD (including congenital abnormalities, severe pulmonary hypertension or right heart disease, pericardial effusion) were excluded. Severe pulmonary hypertension was defined by a Doppler-derived systolic pressure gradient between the right ventricle and right atrium ≥70 mmHg, and any dog with more than mild right atrial or right ventricular chamber dilatation or concentric hypertrophy based on subjective assessment was excluded. Dogs were ineligible if receiving cardiac medications or parenteral fluid therapy at the time of enrollment. While sedation is not routinely used for echocardiographic and radiographic testing at this institution, if deemed necessary to achieve light sedation, dogs who received butorphanol IV or IM were included in the study. The measurements were made from inner edge (blood-tissue interface) to inner edge. LVIDD was normalized to body size (LVIDDN) as previously described. 10 Left atrial and aortic root dimensions were measured by the 2-D right parasternal short axis view as previously described 11 and normalized to body weight by the following formula: The change in probability is approximated as ln(LR) x .19 15 . Based on this formula, one can approximate likelihood ratios of 2, 4, and 6 as increasing the pretest probability of disease by an absolute value of 15%, 25%, and 35%, respectively, whereas likelihood ratios of .5, .3, and .1 decrease pretest probability of disease by an absolute value of 15%, 25%, and 45%, respectively. 15 For instance, if the pretest probability of disease is 50%, a likelihood ratio of 6 increases the posttest probability of disease to 85%, whereas a likelihood ratio of .1 decreases the posttest probability of disease to 5%. Univariate and multivariate logistic regression was used to calculate the odds ratios associated with radiographic measurements and presence or absence of LHE ECHO or LA:Ao ≥1.6. The level of agreement between 2 observers measuring VLAS was determined by calculation of the interclass correlation coefficient (ICC) by a model of absolute agreement. The level of agreement was described as none, slight, fair, moderate, and substantial for ICC values of 0 to .10, .11 to .40, .41 to .60, .61 to .81, and .81 to 1.0, respectively. 16 The strength of correlation between LA:Ao and VLAS was determined by Spearman correlation coefficient (r S ) and described as very high, high, moderate, low, and negligible, for values of .9 to 1.0, .7 to .9, .5 to .7, .3 to .5, and 0 to .3, respectively. 17 Figure 1A) and .865 (95% CI, .783-.947, Figure 1B  Level of agreement between the 2 observers for measurement of VLAS was substantial with ICC of 0.932 (95% CI, .901-.961). Correlation between LA:Ao and VLAS was low (R s = .480; P < .0001) (Figure 3).

CONFLICT OF INTEREST DECLARATION
Authors declare no conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION
Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION
Authors declare human ethics approval was not needed for this study.