• Open Access

Left Ventricular Twist and Circumferential Strain in Dogs with Myxomatous Mitral Valve Disease

Authors


  • The work was performed at the Department of Veterinary Clinical and Animal Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark; Din Veterinär Animal Hospital, Helsingborg, Sweden, and Blå Stjärnans Animal Hospital, Gothenburg, Sweden. Preliminary results from this study were presented at the 22nd European College of Veterinary Internal Medicine Congress in Maastricht, Holland, 2012

Corresponding author: Nora E. Zois, DVM, Department of Veterinary Clinical and Animal Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, 7 Grønnegårdsvej, Frederiksberg C 1870, Denmark; e-mail: nora@zois.dk.

Abstract

Background

During the cardiac cycle, the ventricle undergoes a twisting motion because of the oblique orientation of the left ventricular (LV) myofibers. This can be quantified by speckle-tracking echocardiography (STE). In mitral regurgitation (MR) in humans, the short axis deformation has been suggested as being pivotal to LV function. Decreased and delayed LV twist has been described in experimental MR, but has not been studied in myxomatous mitral valve disease (MMVD).

Hypotheses

(1) Magnitude (CSt) and rate (CSRs) of systolic circumferential deformation decrease before the onset of congestive heart failure (CHF); (2) magnitude and rate of LV twist decrease, and onset of untwist is delayed, with increasing MMVD severity.

Animals

A total of 97 privately owned small- to medium-sized dogs.

Methods

Severity of MMVD was assessed by echocardiography and presence of clinical signs of CHF. Magnitude and rate of LV twist and circumferential deformation were evaluated by STE.

Results

Dogs with CHF receiving treatment had increased CSt, CSRs, early diastolic untwisting rate, and delayed onset of untwist compared to dogs with minimal MMVD and increased systolic twist compared to dogs with mild MMVD (all P < .01). CSt and time to onset of untwist increased with echocardiographic variables of MR severity (all P < .002). CSRs and several LV twist variables decreased with increasing systolic LV internal diameter (all P < .01).

Conclusions and Clinical Importance

No STE-derived variable was decreased before onset of CHF. In dogs with CHF receiving treatment, the delayed onset of relaxation might indicate LV dysfunction and the hyperdynamic CSt and LV twist reflect compensatory mechanisms.

Abbreviations
BW

body weight

CHF

congestive heart failure

CKCS

Cavalier King Charles Spaniel

CSRs

global circumferential strain rate in systole

CSt

global circumferential strain

CV

coefficient of variation

h/R

left ventricular caudal wall thickness in diastole divided by left ventricular radius

HR

heart rate

LA/Ao

left atrial-to-aortic root ratio

LV

left ventricular

LVIDdinc

percentage increase from expected median LV internal diameter in diastole

LVIDsinc

percentage increase from expected median LV internal diameter in systole

LVPWd

LV caudal wall thickness in diastole indexed to body weight

MMVD

myxomatous mitral valve disease

MR

mitral regurgitation

STE

speckle-tracking echocardiography

During the cardiac cycle, the myocardium deforms in three dimensions and because of the oppositely oriented subepicardial and subendocardial layers, it furthermore undergoes a twisting motion. In systole, the apex rotates counterclockwise whereas the base rotates clockwise resulting in a systolic twist, followed by a recoiling and untwisting motion primarily during early diastole.[1, 2] The magnitude and rate of left ventricular (LV) deformation, defined as strain and strain rate, as well as the twisting motion (LV twist) can be quantified noninvasively by the use of speckle-tracking echocardiography (STE). Quantification of systolic twist by the use of STE has been validated against sonomicrometry,[3] and acceptable repeatability and reproducibility were found in healthy awake dogs.[4]

Myxomatous mitral valve disease (MMVD), resulting in mitral regurgitation (MR)[5] is the most prevalent naturally occurring cardiac disease in dogs, and certain breeds, such as the Cavalier King Charles Spaniel (CKCS), are more likely to develop the disease compared with other breeds such as the Beagle.[6] Even though MMVD is considered a valvular disease with relatively well-preserved global LV function, especially in small breed dogs,[7, 8] intramyocardial fibrosis and arteriosclerosis have been reported in dogs with congestive heart failure (CHF) caused by MMVD.[9] Furthermore, intrinsic contractile dysfunction has been reported before the onset of CHF in experimental animal models.[10] However, the presence of subclinical LV dysfunction has not been reported by conventional echocardiography, likely because incipient LV dysfunction is masked by favorable loading conditions and increased sympathetic tone.

Quantification of LV short axis function, represented by the global circumferential systolic strain rate (CSRs), ie, the shortening rate along the perimeter of the ventricle, has been suggested as having potential for detecting latent LV dysfunction in human patients with primary MR,[11] but has not been reported in dogs with MMVD. The LV twist, important for the maintenance of overall LV function and equalization of fiber stress and oxygen demand,[12] similarly has not been reported in canine MMVD. Therefore, the aims of this study were to examine LV twist, CSRs, and the magnitude of global systolic circumferential strain (CSt) in small- to medium-sized dogs with different severities of naturally occurring MMVD. Dogs with no or minimal MR from a breed highly likely to develop MMVD as well as from a breed less predisposed were included. It was hypothesized that CSt and CSRs decrease before the onset of clinical signs of CHF, and the magnitude and rate of systolic twist and early diastolic untwist are decreased, and time to onset of untwist delayed, with increasing MMVD severity.

Materials and Methods

Recruitment and Conventional Echocardiographic Examination of Dogs

The prospective observational study included 97 privately owned dogs ≥ 4 years of age. All dogs were recruited as part of a previous study on radial and longitudinal LV deformation.[13] The postprocessing of images was carried out on separate occasions and not by use of identical image loops. The resultant inclusion groups were overlapping but not identical, as fewer dogs with asymptomatic MR were included in this study.

Pregnancy, lactation, signs of clinical systemic disease, apart from CHF caused by MMVD, and systemic hypertension (defined as systolic arterial blood pressure > 175 mmHg) were exclusion criteria. Dogs with clinical signs of CHF were allowed cardiac treatment, whereas the remaining dogs received no medication.

Dogs were recruited based on degree of MR (defined as the maximal regurgitant jet area measured in percentage of the left atrial area),[14] breed, and presence of clinical signs: (1) Beagles with no or minimal (≤ 15%) MR; (2) CKCS with no or minimal MR; (3) CKCS with mild (> 15% and ≤ 50%) MR; (4) CKCS with moderate or severe (> 50%) MR and no clinical signs; (5) CKCS and (6) dogs of different breeds with clinical signs of CHF because of MR. Groups 5 and 6 were defined by moderate or severe MR, left atrial-to-aortic root ratio (LA/Ao) > 1.7, and having or having had clinical signs, including cough, dyspnea, and exercise intolerance, responsive to furosemide treatment. Written consent was obtained from all dog owners, and the study was approved by the Danish Animal Experiments Inspectorate and the Local Ethical Committee in Gothenburg, Sweden.

The examination comprised structured owner interview, blood sampling for CBC and serum biochemistry, noninvasive blood pressure measurements, clinical examination, and echocardiography, performed in lateral recumbency without sedation. Doppler and 2D echocardiography was performed by a Vivid i ultrasound system1 with continuous ECG recording. Standard apical and parasternal images were recorded and stored digitally.[15] Severity of MR was evaluated from the left apical 4-chamber view by color flow mapping[14] by use of a sector array transducer by second harmonic settings (1.7/3.4 MHz). The same color gain settings and Nyquist limit (± 82 cm/s) were used in all dogs. All remaining echocardiographic images (including those intended for STE) were recorded by use of a higher frequency sector array transducer also by second harmonic settings (2.5/5.0 MHz). From the 2D short axis view, LA/Ao[16] and LV dimensions, presented here as percentage increase from the median expected dimensions,[17] were measured. The LV dimensions also were used for the calculation of fractional shortening (FS). The h/R ratio, representing LV caudal wall thickness in diastole relative to LV radius, was calculated as described previously.[18] All echocardiographic examinations and off-line measurements of conventional echocardiographic variables were performed by the same experienced operator (LHO) blinded to dog identity.

Speckle-Tracking Echocardiography

Digital 2D cine loops acquired by means of 84–100 frames per second in the parasternal short axis view at the basal, apical, and midpapillary level[15] were used for the evaluation of the magnitude and rate of LV twist and CSt, respectively. An effort was made to acquire basal and apical images with a similar frame rate and sector width. Postprocessing was performed by EchoPAC software2 by 1 observer (NEZ), blinded to dog identity and conventional echocardiographic results. In accordance with previous studies,[4, 19] the endocardial border was manually defined and an automatic frame-by-frame tracking of speckle patterns was performed. All measurements were averaged over 3 cardiac cycles, and only images displaying adequate tracking quality, as assessed by automated and manual accept of all 6 myocardial segments, were included. The CSt and CSRs were calculated automatically from the midpapillary level as an average over the entire myocardium delineated by the observer. From the basal and apical levels, respectively, the basal and apical rotations were automatically computed as angular displacements of speckles along the ventricular centroid averaged over the entire region of interest to achieve global basal and apical rotation angles. As previously described,[20] counter-clockwise rotation was expressed as a positive value, whereas clockwise rotation was expressed as a negative value. Rotation rates were calculated automatically by differentiation with respect to time. Data plots of the ECG, rotation, and rotation rate profiles as well at the LV twist and twist rate profiles were exported to a spreadsheet program.3 The magnitude of LV twist was calculated as the net difference between apical and basal rotation angles. By use of the concomitant ECG, the magnitude of systolic twist was defined as the peak of the upward deflection of the LV twist curve during systole. The magnitude of untwist during early diastole (early untwist) was calculated as the net angular difference between the systolic twist and the peak downward deflection of the LV twist curve during early diastole. Similarly, the magnitude of late diastolic untwist (late untwist) was calculated as the net angular difference between the peak downward deflections of the twist curve during early and late diastole (Fig 1). Peak systolic, early and late diastolic twisting rates were calculated as the maximal net difference between peak basal and apical rotation rates in systole, early and late diastole, respectively. Finally, time to onset of untwist was measured as time from the start of QRS-complex until the peak of systolic twist as previously described (Fig 1).[21] To standardize for differences in heart rate (HR) among dogs, the time to onset of untwist was normalized to RR-interval as follows: time to onset of untwist (ratio) = time to onset of untwist (ms)/(201.2 + 0.15 × RR-interval [ms]). This normalization was based on the linear association found between the time to onset of untwist and duration of the RR interval in the 26 dogs with no or minimal MR.

Figure 1.

Example of postprocessing of left ventricular twist in a Cavalier King Charles Spaniel with clinical signs of congestive heart failure. After tracking of basal (upper left corner) and apical (lower left corner) short axis views, the basal (purple) and apical (turquoise) rotations, and the resultant LV twist curve (white) were displayed. Systolic twist was defined as the peak upward deflection during systole, whereas magnitude of early untwist was calculated as the angular difference between systolic twist and the peak downward deflection during early diastole. Magnitude of late untwist was calculated as the angular difference between downward deflections during early and late diastole. By differentiation with respect to time, the rates of rotation and twist were obtained. The yellow arrow depicts the time to onset of untwist, which was indexed to RR interval for further analysis. Timing was defined by use of the ECG as illustrated.

To assess the repeatability of the STE analyses, interexamination variability was evaluated by repeated echocardiographic examination of 5 dogs (3 CKCS with no or minimal MR and 2 CKCS with clinical signs of CHF) at 6 different time points on the same day by same examiner (LHO). Postprocessing of resultant images was performed by one observer (NEZ). Furthermore, one examination of each dog was postprocessed repeatedly on 6 different days to evaluate intraobserver between-day variability.

Statistical Analysis

Statistical analyses were performed by use of statistical software4 and by numerical values of all STE-derived variables. Homogeneity and normal distribution of residuals were evaluated by inspection of residual plots and Shapiro–Wilk W-tests. When appropriate, logarithmic and power transformations were used to achieve normality and homogeneity of the residuals. However, normal distribution and homogeneity of residuals were not achieved with biologically interpretable transformations of observations within dog groups; therefore, nonparametrical statistical tests were applied for group-wise comparisons and values are reported as medians and interquartile range. Initially, the 2 recruitment groups with no or minimal MR, represented by Beagles and CKCS, and the 2 recruitment groups with clinical signs of CHF, represented by mixed breeds and CKCS, were compared by the Mann–Whitney U-test. If no statistically significant differences were found, dogs were pooled into 2 groups, regardless of breed, for further overall analysis. Overall group-wise differences in descriptive data, conventional echocardiographic variables, and STE-derived variables were investigated by the nonparametric Kruskal–Wallis test, followed by pair-wise comparisons by the Mann–Whitney U-test with appropriate Bonferroni adjustment of the significance level.

Normal distribution and homogeneity of residuals were found for all STE-derived variables in multiple linear regression analysis models and therefore, associations between STE-derived variables and conventional echocardiographic variables of MMVD severity, including percentage MR, LA/Ao, percentage increase in LV internal diameter in systole (LVIDsinc) and diastole (LVIDdinc), were investigated by use of least squares multiple linear regression analyses. Age, body weight (BW), sex, HR (calculated from the ECG during the cardiac cycles used for postprocessing), and CKCS/other breed were included as covariates in all models, except models predicting onset of untwist, which did not include HR. Reductions were performed in a backward, stepwise manner until only statistically significant effects remained. Because each of the 4 explanatory variables was entered separately, the significance level for multiple regression analyses was corrected to P = .0125.

For calculation of interexamination coefficient of variation (CV), the mean and standard deviation (SD) values from the 6 examinations of each dog were used for the calculation of a CV for each STE variable in every dog, which were averaged to an overall CV for each STE variable. Similarly, an average CV of each STE variable was computed from the repeated postprocessing of the same examination for the between-day CV. The variability in all STE-derived variables moreover was tested by use of the following ANOVA model with random effects: Yi = μ + a (exami) + b (dogi) + εi, where μ was the general mean; a was the fixed effect of examination number (1–6); b(1),.., b(5) were random intercepts for the 5 dogs; and εi was the model error.

Results

Adequate tracking of the midpapillary short axis view was obtained in 85 dogs (88%), whereas LV twist could be measured in 77 dogs (79%). Inadequate tracking of all 3 parasternal image levels was encountered in 12 dogs, representing all recruitment groups. Thus, a total of 45 females and 40 males were included in the study, represented by 62 CKCS, 10 Beagles, 5 Dachshunds, 2 Jack Russell Terriers, and 6 other small- to medium-sized breeds, each represented by 1 individual. Beagles with no or minimal MR had significantly higher BW and dogs with clinical signs of CHF were significantly older than the other dog groups (Table 1). Dogs with clinical signs of CHF had increased left atrial size, eccentric remodeling, and were treated with combinations of furosemide (31), pimobendan (18), benazepril (23), spironolactone (7), digoxin (4), and propranolol (3). All dogs in CHF had clinical benefit of diuretic treatment either at the time of inclusion or before inclusion. Repeatability of the STE-derived variables is summarized in Table 2. No effect of examination number was found for any STE-variable.

Table 1. Descriptive data of included dogs
 Beagle, No/Minimal MRCKCS, No/Minimal MRCKCS, Mild MRCKCS, Moderate/Severe MRCKCS, CHFDifferent Breeds, CHF
  1. CHF, clinical signs of congestive heart failure; FS, fractional shortening; h/R, caudal wall thickness divided by left ventricular (LV) radius in diastole; LA/Ao, left atrial-to-aortic root ratio; LVIDdinc and LVIDsinc, percentage increase from expected median LV internal diameter in diastole and systole, respectively; LVPWd, LV caudal wall thickness in diastole indexed to body weight.

  2. Within each row, numeral superscripts indicate that the group in question is statistically significantly different from Beagles with no or minimal mitral regurgitation (MR)1, from Cavalier King Charles Spaniels (CKCS) with no or minimal MR2, from CKCS mild MR3, and from CKCS with moderate or severe asymptomatic MR4, respectively.

n101512171813
Sex (m/f)5/55/105/711/610/89/4
Age (years)5.6 (4.8–7.2)4.6 (4.1–5.5)5.4 (4.2–6.3)6.82 (6.0–7.6)11.01,2,3,4 (8.6–11.7)11.61,2,3,4 (10.9–12.6)
Body weight (kg)13.7 (13.0–15.0)9.51 (8.9–11.8)9.61 (8.3–10.8)10.01 (8.8–11.3)10.61 (9.7–11.6)9.91 (7.5–12.3)
LA/Ao1.3 (1.2–1.5)1.2 (1.2–1.4)1.3 (1.3–1.4)1.61,2,3 (1.5–1.7)1.91,2,3 (1.8–2.6)2.21,2,3 (1.9–2.4)
LVIDdinc (%)5.6 (−9.3 to 11.3)−0.9 (−4.9 to 4.7)0.0 (−6.7 to 10.4)3.8 (1.1–15.6)39.71,2,3,4 (27.8–52.7)41.21,2,3,4 (35.6–44.2)
LVIDsinc (%)14.4 (8.9–21.8)6.4 (2.1–12.8)12.9 (10.3–17.1)14.2 (2.0–20.7)20.82 (11.8–32.7)24.02,3 (15.8–48.0)
LVPWd5.1 (4.9–5.2)4.4 (4.2–5.2)4.5 (4.2–4.7)4.6 (4.3–5.3)4.7 (4.2–5.1)4.7 (4.6–4.9)
h/R0.53 (0.49–0.59)0.52 (0.46–0.58)0.51 (0.44–0.56)0.49 (0.43–0.52)0.381,2,3 (0.34–0.43)0.401,2 (0.36–0.40)
FS (%)25.6 (19.5–32.6)28.5 (23.4–37.0)28.0 (21.4–32.2)33.0 (26.0–36.9)42.31,2,3,4 (40.5–45.5)40.41,2,3 (29.4–44.5)
Table 2. Repeatability of speckle tracking-derived variables based on 6 examinations of each of 5 dogs (interexamination variability) and repeated postprocessing of 1 examination from each dog (between-day variability)
VariableBetween-DayInterexamination
SDMean CV and (Range) (%)SDMean CV and (Range) (%)
  1. Please see text for the description of the calculation of speckle tracking-derived variables. CKCS, Cavalier King Charles Spaniel; CV, coefficient of variation; CSRs, global circumferential systolic strain rate; CSt, global circumferential strain; SD, standard deviation.

  2. a

    Normalized to RR-interval duration as: time to onset of untwist/(201.2 + 0.15 × RR-interval).

Systolic twist (°)0.54.8 (4.0–5.4)1.917.2 (11.4–21.4)
Early untwist (°)0.56.8 (3.4–12.0)1.417.9 (11.0–19.2)
Late untwist (°)0.728.9 (10.0–53.1)1.831.3 (23.9–39.1)
Systolic twisting rate (°/s)6.75.5 (3.4–7.9)26.617.7 (8.5–23.5)
Early diastolic untwisting rate (°/s)6.05.0 (2.1–10.9)27.320.2 (11.0–36.5)
Late diastolic untwisting rate (°/s)9.38.2 (6.2–11.3)26.526.9 (19.0–45.7)
Time to onset of untwist (ratio)a0.012.0 (0.7–3.4)0.044.7 (2.5–5.8)
CSt (%)0.84.0 (2.8–5.2)1.46.8 (3.9–12.9)
CSRs (s−1)0.14.8 (3.2–5.6)0.27.6 (5.7–9.9)

LV Twist and CSt among MMVD Severity Groups

No STE-derived variable differed between CKCS and Beagles with no or minimal MR or between CKCS and dogs of mixed breeds with clinical signs of CHF; the groups thus were pooled, and group-wise comparisons were performed using the resultant 4 MMVD severity groups (Table 3).

Table 3. Magnitude and rate of left ventricular twist (n = 77) and circumferential strain (n = 85) in included dogs divided into MMVD severity groups
 Minimal MMVDMild MMVDModerate or Severe MMVDCHF Caused by MMVDOverall P-Value
  1. If a significant overall difference among groups was found on Kruskal–Wallis analysis, posthoc pairwise comparisons were performed.

  2. CHF, clinical signs of congestive heart failure, CKCS, Cavalier King Charles Spaniel, HR, heart rate measured during the cine loop used for postprocessing of global circumferential strain (CSt) and strain rate (CSRs).

  3. a

    Significantly different from minimal myxomatous mitral valve disease (MMVD).

  4. b

    significantly different from mild MMVD (all P <0.008).

  5. c

    Normalized to RR-interval duration as: time to onset of untwist/(201.2 + 0.15 × RR-interval).

CKCS/other breed (n)15/1012/017/018/13
Systolic twist (°)8.3 (6.9–11.4)8.0 (6.7–9.9)8.4 (5.2–10.7)11.9b (8.5–13.4)P = .03
Early untwist (°)7.5 (5.8–9.1)6.6 (4.9–8.3)7.4 (6.2–8.5)9.2 (5.8–10.7)P = .07
Late untwist (°)2.5 (1.4–4.2)2.0 (0.7–2.7)2.7 (1.5–4.6)2.7 (2.1–4.0)P = .51
Systolic twisting rate (°/s)106.7 (88.5–134.2)125.3 (95.7–144.4)125.6 (81.3–148.8)133.5 (111.3–164.5)P = .12
Early diastolic untwisting rate (°/s)−104.6 (−83.0 to −128.3)−111.9 (−93.2 to −135.8)−111.2 (−83.9 to −137.3)−142.4a (−121.7 to −167.5)P = .003
Late diastolic untwisting rate (°/s)−83.7 (−60.3 to −103.4)−81.6 (−63.7 to −102.9)−84.4 (−45.0 to −100.2)−88.8 (−81.8 to −109.1)P = .40
Time to onset of untwistc (ratio)0.49 (0.43–0.51)0.47 (0.43–0.47)0.51 (0.45–0.56)0.52a,b (0.49–0.59)P = .003
CSt (%)−19.5 (−17.2 to −21.2)−19.7 (−18.1 to −21.3)−21.8 (−23.7 to −18.7)−24.8a,b (−21.8 to −27.0)P < .0001
CSRs (s−1)−2.1 (−1.8 to −2.4)−2.1 (−2.0 to −2.3)−2.2 (−2.0 to −2.4)−2.4a (−2.1 to −2.8)P = .04
HR (min−1)112 (87–120)117 (103–130)131 (110–158)131a (116–155)P = .004

Time to onset of untwist was increased in dogs with clinical signs of CHF compared to dogs with minimal and mild MMVD (P = .006 and P = .002, respectively). Dogs with clinical signs of CHF had increased magnitude of systolic twist compared to dogs with mild MMVD (P = .002) and increased early untwisting rate compared to dogs with minimal MR (P = .001). Furthermore, dogs with clinical signs of CHF had increased CSt and CSRs compared to dogs with minimal MMVD (P < .0001 and P = .008, respectively); CSt was moreover increased as compared to dogs with mild MMVD (P = .0002) (Table 3).

Associations between LV Twist, CSt, and Conventional Echocardiographic Variables

Time to onset of untwist increased with increasing LA/Ao and LVIDdinc (both P < .002; Fig 2). Magnitude and rate of systolic twist and early untwist decreased with increasing LVIDsinc (all P < .01). All final models with LVIDsinc as predictor also included age as a significant covariate. Age had positive model estimates, thus oppositely directed from LVIDsinc, and had the highest semipartial correlation coefficient in these models. The CSt increased with degree of MR (Fig 3), LA/AO, and LVIDdinc (all P < .0001), whereas CSRs decreased with LVIDsinc. The latter model included BW and HR; both had positive parameters estimates and the highest semipartial correlation coefficient was found for HR (Table 4).

Table 4. Multiple linear regression analyses of the magnitude and rate of twist (n = 77) and CSt (n = 85). Conventional echocardiographic variables of MMVD severity and left ventricular dimensions were entered separately as predictors. Age, body weight, sex, heart rate and breed (CKCS/other) were entered as covariates
 MRLVIDdincLVIDsincLA/Ao
  1. In parentheses, the adj. R2 values for the final models are reported. A positive association between the main predictor and the outcome is marked by ↑, while a negative association is marked by ↓. The superscripts indicate significant covariates: heart rate (HR) (P < .005), body weight (BW) (P < .0008), and age (P < .003) were left in the respective final models and all had positive model estimates.

  2. CKCS, Cavalier King Charles Spaniel, CSRs, global circumferential strain rate; CSt, global circumferential strain; LA/Ao, left atrial-to-aortic root ratio; LVIDdinc and LVIDsinc, percentage increase from expected median left ventricular internal diameter in diastole and systole, respectively; MMVD, myxomatous mitral valve disease; MR, mitral regurgitation measured in percentage of the left atrial area.

  3. a

    Normalized to RR-interval duration as time to onset of untwist/(201.2 + 0.15 × RR-interval).

Systolic twist (°)NSNSP = .007 (0.22)AgeNS
Early untwist (°)NSNSP = .01 (0.17)AgeNS
Late untwist (°)NSNSNSNS
Systolic twisting rate (°/s)NSNSP = .004 (0.29)Age, HRNS
Early diastolic untwisting rate (°/s)NSNSP = .003 (0.27)Age, HRNS
Late diastolic untwisting rate (°/s)NSNSNSNS
Time to onset of untwist (ratio)aNSP = .002 (0.16)↑NSP < .0001 (0.18)↑
CSt (%)P < .0001 (0.30)↑P < .0001 (0.32)BWNSP < .0001 (0.24)↑
CSRs (s−1)NSNSP = .005 (0.33)HR,BWNS
Figure 2.

Line of fit with 95% confidence interval from the multiple linear regression of the time to onset of untwist on left atrial-to-aortic root ratio (LA/Ao) in 77 dogs. Time to untwist (ratio) = 0.07 × LA/Ao + 0.32.

Figure 3.

Line of fit with 95% confidence interval from the multiple linear regression of the global circumferential strain (CSt) on mitral regurgitation (MR) measured as the area of the regurgitant jet in percentage of the left atrial area in 85 dogs. Please note that numerical values were used for statistical analysis. CSt = 0.05 × MR + 18.6.

All multiple linear regression analyses were repeated in the subset of dogs without CHF (n = 54). Consistency with the results presented in Table 4 was found for all variables regarding significance and parameter estimates except for the association between time to onset of untwist and LVIDdinc which was no longer significant (data not shown).

Discussion

Evaluation of LV short axis function, assessed by CSt and CSRs, was feasible in the majority of dogs with different severities of MR secondary to MMVD, whereas LV twist was less readily obtainable, mostly because of inadequate tracking quality of the apical views or excessively large differences in heart rate between apical and basal image loops. Onset of untwist was delayed, whereas CSt, CSRs, systolic twist, and early diastolic untwisting rate were increased in dogs with clinical signs of CHF receiving treatment. CSt and time to onset of untwist increased with variables reflecting MMVD severity, whereas CSRs, magnitude and rate of systolic twist and early untwist decreased with LVIDsinc.

The twisting motion evaluated by use of STE has not been reported previously in dogs with naturally occurring MR. However, the finding of a delayed onset of untwist with increasing MMVD severity is in agreement with findings in human patients with primary MR[21] and an experimental canine model of chronic MR evaluating LV twist by use of cineflouroscopy.[22] Because the initiation of diastole is an energy-dependent and sensitive phase of the cardiac cycle,[23] the timing of contraction–relaxation is a very vulnerable period of myofiber mechanics.[24, 25] Based on clinical studies of inducible ischemia and cardiomyopathies in humans, it has been suggested that a progressive loss of ability to modulate the timing of onset of relaxation is an early sign of LV dysfunction.[25, 26]

Previous studies evaluating magnitude and rate of LV twist have shown conflicting results. Preserved magnitude and rate of systolic twist and early untwisting rate have been reported in human patients with primary MR,[21] whereas a decrease in magnitude of systolic twist and untwist was demonstrated 12 weeks after surgical disruption of the mitral valve in an experimental canine model.[22] Regarding the preservation of systolic twisting rate, our findings are concordant with the findings in human patients. We did, however, find significantly greater magnitude of systolic twist and rate of early untwist in dogs with clinical signs of CHF receiving treatment compared to dogs with mild and minimal MMVD, respectively. However, on multiple linear regression analyses, both magnitude and rate of systolic twist and early untwist decreased with LVIDsinc, likely indicating an association with deteriorating LV systolic function, which is in agreement with the findings in human patients with primary MR.[21]

In this study, increased CSt was found in dogs with clinical signs of CHF receiving treatment compared to dogs with minimal and mild MMVD. Furthermore, CSt increased linearly with indices of LV and left atrial remodeling. A recent study reported increased CSt in dogs with compensated MR compared to healthy control dogs.[19] Thus, our study supports these findings and extends them to symptomatic stages of MR caused by MMVD. To the best of our knowledge, CSRs has not previously been investigated in dogs with MMVD. In this study, CSRs was found to be higher in dogs with clinical signs of CHF receiving treatment as compared to dogs with minimal MMVD. By invasively determined peak rate of LV pressure rise (+dP/dt), a preserved CSRs was found in human patients with primary MR and maintained contractile function, whereas patients with impaired latent contractile function exhibited decreased CSRs. This led to the hypothesis of LV short axis function playing a pivotal, and maybe even compensatory role, for the maintenance of global LV contractile function.[11] Thus, in contrast to findings in human patients with MR, this study did not detect subclinical short axis dysfunction, because CSRs was increased in dogs with clinical signs of CHF receiving treatment. The significant inverse relationship with LVIDsinc does, however, support the hypothesis of CSRs being important for the preservation of overall LV systolic function in dogs with MMVD.

Because the short axis function and LV twist[27-29] have been proposed as having a compensatory function in latent LV dysfunction, the finding of a delayed untwist and a preserved or even increased CSt, CSRs, and LV twist are not necessarily conflicting. Instead, the findings might indicate subtle LV dysfunction in dogs with increasing severity of MMVD. The net LV twist is determined by the subepicardial myofiber direction whereas the subendocardial myofibers in part serve to counteract this motion. Preservation of LV twist thus is not unexpected with selective subendocardial damage.[30] The intramyocardial fibrosis and arteriolar narrowing reported in dogs with CHF because of MMVD were most pronounced in the subendocardium,[9] and therefore selective subendocardial myofiber damage might explain the preservation of LV twist and circumferential function along with a prolonged time to onset of untwist.

In accordance with studies in people,[31, 32] increasing age was associated with increasing magnitude and rate of LV twist in this study. The reason for this age-related increase in LV twist can only be speculated upon, but subendocardial impairment previously has been described in the elderly,[33] and might therefore potentially explain this finding too.

Differences in methodology could account for the divergent findings between the study of LV twist in canine experimental MR[22] and this study. Moreover, duration and magnitude of the volume overload imposed on the myocardium in acutely induced experimental MR in contrast to the slowly progressing spontaneous MR could explain the differences. However, differential progression of LV dysfunction has been reported within the canine species, and large breed dogs with MMVD apparently develop an earlier and more severe LV dysfunction than small breed dogs.[18] Among healthy dogs, a lower magnitude of systolic twist and circumferential strain was found in dogs weighing > 20 kg compared to small- and medium-sized dogs.[34] Interestingly, the experimental study examining LV twist included considerably larger dogs (25–30 kg)5 than this study, and differences in body size might therefore also explain the discordant findings. It is possible that the divergence regarding CSRs between this study and findings in human patients with primary MR also could be related to differences in body size. However, comparative studies are needed to elucidate this subject.

The high CV of the LV twist variables must be emphasized. In the light of such large variation among repeated examinations, the interpretation of an absence of significant changes as being reflective of a preservation of LV twist variables should be made with great caution. On the other hand, the CV of the time to onset of untwist, CSt, and CSRs were within acceptable ranges. Furthermore, the findings in this study might, at least partially, reflect the hemodynamic consequences of MR, because preload and afterload affect LV twist and deformation.[35-37] Similarly, the delay in untwist might reflect the ongoing regurgitation of blood into the left atrium after aortic valve closure, because the twisting motion might continue along with the regurgitation to the atrium during the earliest phase of ventricular diastole.

Secondly, the administration of drugs affecting cardiovascular load and cardiac contractility to dogs with CHF is an important limitation. Remaining dogs did not receive any medication and the preservation or increase in LV twist and CSt variables in dogs with CHF might therefore merely be the result of the administration of cardiovascular medications. However, the multiple linear regression analyses were repeated in the subset of dogs without CHF (and thus not receiving treatment), and consistency was found for all variables except one, regarding significance and parameter estimates.

Another limitation of this study is the timing based on the concomitant ECG as it holds no information on the isovolumetric contraction and relaxation. Furthermore, because combination of 2 different heart cycles is inherent in the calculation of LV twist by use of STE, a difference in RR interval was inevitable. Moreover, the size of the regurgitant jet was evaluated by color Doppler mapping of the mitral valve. This semiquantitative method is not accurate and is furthermore affected by both physiological and methodological factors. The upper limit and gain settings of the Doppler shift were, however, standardized in all dogs, the size of the jet area was optimized by inclusion of oblique image views when necessary, and only dogs with normal systemic blood pressure were included in the study.

Finally, the associations found by regression analyses were few and weak, and covariates such as age had higher potential of predicting change in STE variables than LVIDsinc.

In conclusion, this study could not detect LV dysfunction, as indicated by magnitude and rate of circumferential strain and LV twist, before the onset of CHF. However, a delayed onset of untwist, potentially being an indication of early LV dysfunction, was found with gradual increase in MMVD severity in small- to medium-sized dogs. The concomitant increase in magnitude and rate of circumferential strain and LV twist might therefore be a compensatory mechanism and of importance for the preservation of overall LV function and thus survival time.

Acknowledgments

The authors thank Christina Kjempff, Vibeke Christensen, Dennis Jensen, and Hanne Carlsson at the Department of Veterinary Clinical and Animal Sciences, University of Copenhagen, Frederiksberg, Denmark for excellent technical assistance. In addition, Björn Åblad and Heidi Young Nilsen at Blå Stjärnans Animal Hospital, Gothenburg, Sweden, are acknowledged for their indispensable help on the recruitment of dogs. The study was supported by a PhD study grant from the Faculty of Health and Medical Sciences, University of Copenhagen and by the Danish Council of Independent Research, Medical Sciences (project no. 271-08-0998).

Conflict of Interest: Authors disclose no conflict of interest.

Footnotes

  1. 1

    Vivid®i echocardiograph, GE Healthcare, Milwaukee, WI

  2. 2

    EchoPAC PC. Version 108 1.5, GE Healthcare Vingmed Ultrasound AS, Horten, Norway

  3. 3

    Microsoft Office Excel 2007, Microsoft Corporation, Redmond, WA

  4. 4

    SAS statistical software, version 9.2, SAS Institute, Cary, NC

  5. 5

    F. Tibayan, personal communication, March 1, 2012

Ancillary