Echocardiographic mitral valve association with morphometric measurements in Cavalier King Charles Spaniels via Inverse Probability Weighting analysis

Development and progression of myxomatous mitral valve disease (MMVD) in Cavalier King Charles Spaniels (CKCS) are difficult to predict. Identification at a young age of dogs with a morphotype associated with more severe mitral disease is desirable. The aims of this study were to: 1) describe the physical, morphometric, and echocardiographic features of MMVD affected Cavalier King Charles Spaniels (CKCS) in American College of Veterinary Internal Medicine (ACVIM) class B1; 2) evaluate the influence of morphometric physical measurements on murmur intensity, mitral valve prolapse (MVP), regurgitant jet size and indexed mitral valve and annulus measurements. Fifty-two MMVD affected CKCS in ACVIM class B1 were included. This is a prospective clinical cross-sectional study. Morphometric measurements, which included body, thorax, and the head sizing of each dog, have been investigated to establish the association with heart murmur intensity, valvular and annular echocardiographic measurements, MVP and regurgitant jet size using inverse probability weighting (IPW) analyses to adjust for confounding. The IPW analyses showed that when head length and nose length decreased, dogs had more severe regurgitant jet size. Furthermore, subjects with more pronounced head stop angle had thicker anterior mitral valve leaflets. A brachycephalic morphotype, with dogs more similar to King Charles Spaniel breed in cephalic morphology, is associated with a more severe regurgitant jet size and with valvular characteristics related to worse forms of MMVD.

showed that there is 55 a negative correlation between body weight and mitral valve prolapse (MVP) in this breed [9]. 56 Furthermore, in the CKCS, MMVD is associated with earlier onset and thus with potentially greater 57 cardiac morbidity and mortality compared to other breeds [8,10,11]. Yet, the preclinical period 58 often varies markedly among subjects, making it challenging for clinicians to identify those that 59 will eventually develop clinical signs [12,13]. For these reasons, the early identification of a 60 morphotype associated with a more severe MMVD can have several advantages. This could allow 61 clinicians to monitor the dogs in a very targeted way and to educate breeders in the selection of 62 subjects without some phenotypical characteristics related to a more severe MMVD and/or more 63 rapid progression of the disease. This could be possible in the context of a breeding selection 64 program, which should also consider all the heritable disorders of the CKCS. 65 To the best of our knowledge, only one study evaluated the prevalence and severity of MVP related 66 to the size of the thorax in dogs, and in particular in Dachshunds,3 but no study has ever analyzed 67 the association between MVP or MMVD severity and morphometric measurements in CKCS. 68 The aim of this study was to investigate morphometric measurements in relation with (ACVIM) class B1 CKCS [14]. This class, in fact, includes the majority of the breeding population 71 and is very heterogeneous from clinical, morphological and echocardiographic points of view [15]. 72 The identification of phenotypic characteristics associated with more severe forms of MMVD 73 could be useful for setting breeding selection programs aimed at reducing the prevalence of the 74 disease in this breed.

76
Study population, research question and statistical framework 77 In this prospective clinical cross-sectional study, we carefully described the morphometry 78 of a small Italian study population of CKCS and then we evaluated the influence of body, thorax, 79 and head dimension on different clinical (heart murmur intensity) and echocardiographic 80 measures/indexes of the severity of MMVD (MVP, regurgitant jet size, and indexed mitral valve 81 and annulus measurements). Furthermore, we investigated the severity of MMVD including a 82 score, assigned according to the degree of MVP, mitral regurgitation jet size and age [8]. 83 To investigate the association between morphometric measures and severity of MMVD we used a 84 method adopted from the causal inference framework [16]. The framework proposes methods to 85 address causal questions accounting for the confounding which affects the association between the 86 exposure and the outcome of interest. In this study we used inverse probability weighting (IPW) 87 analyses that, via weighted regression modelling, adjust for confounders [17][18]. The confounders 88 are used to estimate the probability of being exposed conditional on the values of the confounders 89 and a function of this probability is used to construct weights. The weights are assigned to the 90 subjects in the study population to balance them with respect to the confounders used in the 91 analysis. Balancing subjects in the study population allows to estimate an association that is 92 unbiased from the confounders considered in the analysis while constructing the weights.  All the dogs underwent physical examination, echocardiography, and morphometric evaluation.

101
The data regarding the dates of birth and the genealogy were verified by checking each animal 102 microchip number and family tree in the Italian regional registry and ENCI's (Ente Nazionale della 103 Cinofilia Italiana) pedigree database. Cardiac auscultation was performed by two well-trained 104 operators (MB and PGB) and the dogs were restrained in standing position in a quiet room by the 105 owners. The detection of a left apical systolic murmur was not considered a mandatory inclusion 106 criterion. The evaluated auscultatory findings were presence/absence, timing, and intensity 107 (0=absent; 1=I-II/VI left apical systolic or soft murmur; 2=III-IV/VI bilateral systolic or moderate 108 and loud murmur respectively; 3=V-VI/VI bilateral systolic or palpable murmur) of murmur [19].

109
Unless otherwise stated, hereinafter the term murmur refers to a left apical systolic murmur.   Echocardiographic measurements were performed by one operator (MB) to reduce potential biases.

161
All measurements of interest were repeated on 3 consecutive cardiac cycles, and the mean value  Table 1 [36-38]. Body size, cephalic, thoracic and volume indexes were also calculated 182 (Table 1) [36]. During the morphometric evaluation, the dogs were kept calm and in standing 183 position by owners, with the four limbs perpendicular, hand stacking as if they were in exposition.

184
Morphometric evaluation was always performed by the same operator (MB), to reduce systematic 9 185 errors, on the left side of each dog, to reduce potential biases. Intra-observer variability was < 10%. 186 The circumference of the thorax was measured using a measuring tape. Body and thorax 187 evaluations were performed using a custom-made sliding gauge (Fig 1).

217
The body condition score (BCS) was recorded for each dog using a 1 to 9 score, and scores 4 and  In 26 dogs (50%), no murmurs were found. severity score of all included subjects. Moreover, MVAd was larger in subjects older than 6 years 285 than in dogs younger than 3 years (P = .03), whereas MVAs was larger in subjects older than 6 286 years than in those between 3 and 6 years (P < .001). Sphericity index was lower in subjects older 287 than 6 years compared to subjects with age between 3 and 6 years (P = .01).      Settings for IPW analysis 334 The IPW analysis was performed including 49 of the 52 subjects, due to lack of some 335 values, both for ordinal and continuous variables. Furthermore, the covariate "coat color type" has 336 been simplified in "Blenheim" and "other colors" due to the small number of subjects with coat 337 different from Blenheim. For the same reason, the information "neutered/sterilized", originally 338 incorporated in the covariate "sex", has not been used.

339
The PS model with interactions between covariates/confounders provided a better balance in terms 340 of observed confounders between exposure groups (increasing comparability) and accordingly this 341 model was used to build SIPW.

IPW analyses for ordinal variables 343
Obtained results were significant for two of the three considered ordinal variables, 344 particularly for heart murmur intensity and jet size. The IPW analysis, in fact, showed that body 345 length (P = .03) and nose length (P < .01) had negative influence on heart murmur intensity (shorter 346 body length and shorter nose were associated to higher murmur intensity). Furthermore, head 347 length (P < .001) had negative influence on jet size (shorter head was associated to larger jet size).

348
However, morphometric measurements had no effects on MVP severity.

349
The results of the regression analysis for ordinal variables, applied to the included population, are 350 summarized in Table 3a.

IPW analyses for continuous variables 352
Head length (P < .001) had positive influence on anterior mitral valve length (longer head 353 was associated to longer anterior mitral valve leaflet).

363
Anterior or axillary thoracic circumference (P = .01) and head length (P = .000) had a positive 364 influence on sphericity index.

365
The derivation of results of IPW analysis respect to the morphometric indexes showed that only 366 thoracic index was negatively associated with mitral valve annulus in systole and diastole.

367
The results of the regression analysis for continuous variables, applied to the included population, 368 are summarized in Table 3b.

369
The performance of the outcome models regarding normality, linearity and variance homogeneity 370 were acceptable according to the diagnostic plots and there were no clues of violation of 371 proportional odds assumption for ordinal models, whereas some multicollinearity problems were 372 detected in the models including interaction terms (data not shown).  Level of statistical significance: *** = P < .001; ** = P < .01; * = P < .05.  positively associated with measures of the degree of MMVD (e.g., jet size, leaflet length, and 482 murmur intensity). In the present study, thorax height had a negative association with AMVL.

483
Furthermore, thorax width and TC 1 had a positive association on MVAd and SI, respectively.

484
Surely, the most interesting result obtained is that subjects with shorter head were associated with 485 a higher jet size and subjects with shorter body and nose length had higher heart murmur intensity.

486
Regarding mitral valve and mitral annulus measurements, subjects with more barrel-shaped thorax