Comparison of the mitral valve morphologies of Cavalier King Charles Spaniels and dogs of other breeds using 3D transthoracic echocardiography

2.1 Animals

In this prospective cross-sectional study, we enrolled dogs from two institutions: VA-MD College of Veterinary Medicine (VMCVM), and Swedish University of Agricultural Sciences (SLU). After interview of the pet-owner, each dog had a physical examination including comprehensive cardiac auscultation, conventional 2D/color Doppler echocardiography, and acquisition of a RT–3DTTE dataset. To be enrolled in the study, dogs had to be older than 1 year of age, and without a history of cardiac disease. Additional inclusion criteria were tolerance of a complete echocardiographic examination without need of sedation, absence of evidence of MMVD or other cardiac disease at 2DTTE and color Doppler examination, and acquisition of RT–3DTTE datasets free from significant stitching artifacts. Dogs with murmurs graded 1/6 or 2/6 were not excluded provided all other inclusion criteria were met.

2.2 Methods

The RT–3DTTE datasets were acquired using two different ultrasound units (Artida, Toshiba Medical Systems, Tokyo, Japan; IE33, Philips Medical Systems, Andover, MA) equipped with matrix probes (PST-25SX matrix-array transducer, Toshiba Medical Systems, Tokyo, Japan; X2-7, Philips Medical Systems, Andover, MA; X1-3, Philips Medical Systems, Andover, MA) from a left parasternal apical window, as previously described.7, 18 MMVD was identified when color Doppler echocardiography demonstrated a mitral regurgitant jet that was consistently evident in a right parasternal four-chamber view, and/or a left apical four chamber view, and associated with MV thickening and/or prolapse. All RT–3DTTE datasets were imported into a computer workstation equipped with a dedicated software for MV analysis (4D MV-Analysis 2.3, TomTec Imaging Systems, Unterschleissheim, Germany), and analyzed offline by a single operator (GM). Mitral models were created using a semi-automated procedure, and manually reviewed to identify and correct tracing errors, as previously described.7 Morphologic variables were automatically measured and calculated on the MV models, and variables known to be significantly related to body size were normalized using allometric equations reported elsewhere.7 Variables obtained were: normalized antero-posterior annulus diameter (nAPD), normalized anterolateral-posteromedial annulus diameter (nALPMD), normalized commissural diameter (nCmD), annulus sphericity index (SI) calculated as APD/ALPMD, annulus height (AnH), normalized annulus circumference (nAnCirc), normalized annulus area (nAnA), normalized anterior leaflet length (nALL), normalized anterior leaflet area (nALA), posterior leaflet length (PLL), normalized posterior leaflet area (nPLA), nonplanar angle (NPA), tenting height (TnH), tenting area (TnA), normalized tenting volume (nTnV), angle between aortic annulus and anterior leaflet (AAo-AP Angle). The ratio of AnH and CmD (AHCWR) was calculated as an index of nonplanarity of the mitral annulus.8 Detailed figures illustrating each variable are reported as supplemental material elsewhere.7

2.3 Data analysis

Data describing sex, age, body weight (BW), breed, institution of image acquisition, and the values of the MV morphologic variables, were entered into a spreadsheet and analyzed using commercially available software (JMP Pro 11, SAS Institute Inc., Cary, NC, 1989-2013). Dogs were divided in two groups: CKCSs and dogs of other breeds. Cavalier King Charles Spaniels were subdivided based on the institution at which images were acquired—VMCVM (VMCVM-CKCSs), or SLU (SLU-CKCSs). Proportions of males, age, BW, and the MV morphologic variables were compared between CKCSs and dogs of other breeds, and between VMCVM–CKCSs and SLU–CKCSs. Normal distribution of data was assessed by visual inspection of normal quantile plots. Homoscedasticity of data was assessed using a Levene's test. Normally distributed data are presented as mean ± standard deviation, while data that were not normally distributed are presented as median (25th-75th percentile).

Differences in the proportion of males between two groups were assessed using Fisher's exact test. Differences in continuous variables between two groups were assessed using an unpaired Student's t-test when data had a normal distribution, Mann-Whitney U test when data were not normally distributed, and Welch's test when the homoscedasticity assumption was not met. Given the finding of an age difference between CKCSs and dogs of other breeds, a possible relationship between the MV morphologic variables and age was investigated using simple linear regression. A P-value Bonferroni-corrected for multiple comparisons ≤ .05 was considered significant for all tests.

3.1 Seventy-six dogs were enrolled in the study

Thirty-five CKCSs and 41 dogs of other breeds, of which the most represented (count ≥ 3) breeds were mixed breed (n = 17), Beagle (n = 4), and Border Collie (n = 3). Fourteen CKCSs and 33 other breed dogs were enrolled at the VMCVM, and twenty-one CKCSs and 8 other breed dogs were enrolled at the SLU. Of the 35 CKCS 28 were female and 7 were male. For the 41 other breed dogs, 28 were female and 16 were males.

3.2 CKCSs versus dogs of other breeds

Differences between CKCSs and dogs of other breeds are summarized in Table 1. The median age for all dogs was 3.4 years (2.6-5.8 years), and CKCSs were significantly younger than dogs of other breeds. The median BW for all dogs was 10.1 kg (8.5-19.5 kg), and CKCSs weighed significantly less than dogs of other breeds. The AnH, nPLA, TnH, TnA, and nTnV were all smaller in CKCSs, compared to dogs of other breeds (Figure 1). The AAo-AP Angle was greater (less acute) in CKCS, compared to dogs of other breeds. No difference was found between CKCSs and dogs of other breeds regarding all the other study variables.

3.3 VMCVM–CKCSs versus SLU–CKCSs

Differences between VMCVM–CKCSs and SLU–CKCSs are summarized in Table 2. The PLL was significantly greater in VMCVM–CKCSs than in SLU–CKCSs. Normalized tenting volume was significantly smaller in SLU–CKCSs compared to VMCVM–CKCSs. No difference was found between VMCVM–CKCSs and SLU–CKCSs regarding all the other study variables.

There was no relationship between age and any MV morphologic variable in the entire study population (all P values > .16), nor among CKCSs (all P values > .13) and dogs of other breeds (all P values > .11).

4.1 Limitations

Datasets analyzed in this study were acquired in two different centers, equipped with two different ultrasound units, and by multiple operators. This could have introduced some variability in the data. Therefore, the differences in MV morphologies found between CKCSs examined in the two centers, could be attributed to a center bias. The influence of these factors however, should have been minimized by the nature of the three-dimensional technique, in which the operator has very limited influence on the final dataset. Moreover, one operator using the same software, analyzed all the datasets, which should have further minimized the variability. However, an influence of a parameter intrinsic to the two different acquisition platforms, such as the volume-rate of the acquisition, cannot be excluded. An assessment of the intercenter variation, would require the same dogs to be examined in both institutions, which was not feasible. Also, although the breed of the dog was not clearly identifiable by the operator at the time of dataset analysis, no attempt was made to anonymize the datasets; therefore, the dataset analysis was not conducted in a blind fashion. Furthermore, dog's pedigrees information was not acquired, therefore whether the CKCSs examined belong to different lineages is unknown. Lastly, BW and age differed significantly between CKCSs and dogs of other breeds. Some measurements of MV morphology have been reported to be significantly related to BW⁷; however, in the current study, the variables known to be significantly related to body size had been standardized using allometric indexes derived from a population of healthy dogs,7 minimizing the effect of body size on the morphologic variables. Furthermore, as mentioned in the previous section, regression analyses failed to identify any effect of age on MV morphologic variables. Therefore, we do not expect the age difference between groups to have affected the morphologic comparisons.