• Open Access

Comparison of 3 Ultrasound Methods for Quantifying Left Ventricular Systolic Function: Correlation with Disease Severity and Prognostic Value in Dogs with Mitral Valve Disease


  • F. Serres,

    1. Unité de Cardiologie d'Alfort, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    Search for more papers by this author
  • V. Chetboul,

    1. Unité de Cardiologie d'Alfort, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    2. Unité Mixte de Recherche INSERM U841, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    Search for more papers by this author
  • R. Tissier,

    1. Unité Mixte de Recherche INSERM U841, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    2. Unité de Pharmacie-Toxicologie, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France
    Search for more papers by this author
  • L. Poujol,

    1. Unité de Médecine, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France, and
    Search for more papers by this author
  • V. Gouni,

    1. Unité de Cardiologie d'Alfort, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    2. Unité Mixte de Recherche INSERM U841, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    Search for more papers by this author
  • C. Carlos Sampedrano,

    1. Unité de Cardiologie d'Alfort, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    2. Unité Mixte de Recherche INSERM U841, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    Search for more papers by this author
  • J.-L. Pouchelon

    1. Unité de Cardiologie d'Alfort, Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort cedex, France,
    Search for more papers by this author

Corresponding author: Valérie Chetboul, DVM, PhD, Dipl-ECVIM-CA (cardiology), Unité de Cardiologie d'Alfort, Ecole Nationale Vétérinaire d'Alfort, 7 avenue du Général de Gaulle, 94704 Maisons-Alfort cedex, France; e-mail: vchetboul@vet-alfort.fr.


Background: End-systolic volume index (ESVI) is a marker of systolic function, which can be assessed by the geometric (GM, based on Teichholz formula) or 2 planimetric methods (PM, Simpson's derived and length area methods).

Hypothesis: Systolic dysfunction (SyD) may be observed in dogs with mitral valve disease (MVD) and is better assessed by PM than GM, which does not take into account the longitudinal left ventricular systolic shortening.

Animals: Six healthy dogs were used to determine the variability of the tested variables (Study 1). These variables were then prospectively assessed (Study 2) in 101 small breed dogs: 77 dogs with MVD and 24 healthy controls (CD).

Methods: ESVI was measured by GM and PM in awake dogs.

Results: All within- and between-day coefficients of variation were <11% (Study 1). For Study 2, a nonlinear overestimation of ESVI was observed by GM compared with PM. PM-derived ESVI was significantly increased in ISACHC class 3 dogs compared with ISACHC class 1 dogs and exerted a significant influence on cardiac events at 5 months in dogs with MVD from ISACHC classes 2 and 3 (P < .05).

Conclusions and Clinical Importance: ESVI can be calculated by GM and PM with good repeatability and reproducibility. However, GM overestimates ESVI in a nonlinear way. Therefore, PM-derived ESVI should be preferred for the detection of SyD that is present at the late stages of the disease.

Mitral valve disease (MVD) is the most common acquired heart disease in the dog.1–4 The natural history of this pathologic condition is usually slowly progressive with an asymptomatic period of variable and unpredictable duration. Various complications such as pulmonary arterial hypertension (PH),5 chordae tendinae rupture,6 and cardiac arrhythmiasa may influence prognosis and clinical progression of the disease. Another potential MVD complication also described in humans with mitral regurgitation (MR) is left ventricular (LV) systolic dysfunction (SyD).7–9 However, the detection of MR-associated LV SyD remains challenging in both humans and dogs. Assessment of SyD in humans with MR may rely on comparison between pre- and postoperative values of echographic variables when mitral valve replacement or repair is performed.7–9 However, such studies are currently difficult to undertake in the dog, because mitral valve surgery is a very rare treatment option for canine MVD.10,11

Several M-mode echocardiographic indices such as fractional shortening (FS) or ejection fraction (EF) are commonly used for the noninvasive evaluation of systolic function in the dog.12 However, these parameters are dependent on both wall stress and loading conditions, which may be altered in the case of MR.13–15

End-systolic volume index (ESVI) is another echocardiographic variable, which has been shown to be a load-independent predictor of postoperative SyD in humans.7,8 This index has also been demonstrated to be relatively independent of experimental increased preload in the dog.16 Three ultrasound methods may be used to evaluate ESVI: the Teichholz method, also called the geometric method (GM), using M-mode linear measurement of the LV systolic diameter,12,17 and 2 planimetric methods (PM), including the Simpson's derived method of disks (SDM) and the length-area method (LAM).18,19

Unlike SDM and LAM, the Teichholz method does not involve direct measurement of longitudinal LV dimension, and this may lead to either overestimation or underestimation of LV volume in dogs with globoid or elongated hearts, respectively. Using SDM, heart volume is measured as the summation of parallel cylinders, whose diameters are derived from endocardial border tracing performed on 1 or 2 orthogonal LV apical views. The LAM relies on the following simple formula:


where V is the LV volume, A is the LV area, and L is the LV length measured on a single plane apical view.

Unlike the Teichholz method, both SDM and LAM have been validated against invasive methods in the anesthetized dog with good correlations between stroke volumes and EF assessed by PM and those calculated either by thermodilution or cineangiography in both normal and ischemic states.20,21 Additionally, in 1 study, GM and PM were used and compared with the conductance catheter technique in healthy anesthetized dogs with various drug-induced changes in loading conditions.b In this report, PM were more closely correlated to the invasive method than GM. However, the within- and between-day variability of SDM- and LAM-derived variables (ESVI, end-diastolic volume index [EDVI], and EF) has never been assessed in the awake dog, nor that of GM variables. Moreover, these 3 echocardiographic techniques have never been compared with each other in dogs with naturally occurring heart diseases. In other words, whether or not LV volumes could be indistinctly assessed by PM or GM in cardiac dogs remains unknown. Lastly, the comparative correlation between MVD severity and these LV shape and volume indices as well as the prognostic value of the latter have never been assessed in the context of spontaneous canine MVD.

The aims of this study were therefore to (1) determine the comparative within- and between-day variability of ESVI, EDVI, and EF assessed by both GM and PM in healthy dogs (Study 1), (2) quantify and compare these LV variables in a large population of healthy and diseased dogs affected by MVD, and (3) analyze their correlation with several clinical and echo-Doppler markers of MVD severity as well as their comparative prognostic value (Study 2).

Materials and Methods

Experimental Design

Two separate studies were performed.

Study 1 (Validation Protocol). The within- and between-day variability of GM and PM variables (ESVI, EDVI, and EF) were determined by performing 36 echocardiographic examinations on 6 healthy, awake dogs (age: 3.1 ± 1.8 years [0.8 – 5.0 years]; weight: 22.7 ± 11.2 kg [11 – 40 kg]) on 4 different days over a 2-week period: 1 Cane Corso (neutered female), 1 Beagle (neutered female), 1 Cocker Spaniel (neutered female), 1 Labrador (neutered female), 1 Brittany Spaniel (neutered male), and 1 Boxer (neutered male). The within- and between-day variability of the LV index of sphericity22 (LVSI) was also assessed. On a given day, 3 dogs were examined at 3 nonconsecutive times. Each variable was measured 3 times on 3 consecutive cardiac cycles using the same frame for GM, and using the same loop for the PM. The mean values were used to determine the within- and between-day variability. This protocol involved 1 single trained observer (VC).

Study 2 (Prospective Study). The study population consisted of client-owned dogs that prospectively underwent a complete echocardiographic and Doppler examination at the Cardiology Unit (National Veterinary School of Alfort, France) during 2006 and 2007. These ultrasound examinations were performed by the observer involved in Study 1 (VC) and by 3 other observers (FS, VG, and CCS). The latter 3 observers had been trained by the 1st observer in the use of GM and PM for at least 1 year before performing the echocardiographic examinations. For each dog, the owner's consent was obtained before enrollment in the study. In order to limit influence of the breed effect on systolic function, only small breed dogs (≤15 kg) were included in the study. Dogs receiving pimobendan were not included in the study because of the positive inotropic effect of this drug. Dogs from Study 2 were assigned to either Group 1 or 2. Group 1 consisted of healthy control dogs characterized by both normal physical and echo-Doppler examinations, whereas Group 2 included dogs with MVD. Diagnosis of MVD was based on the following criteria: (1) left systolic apical heart murmur of late appearance (age >1 year old), (2) no history of infectious disease, and (3) echocardiographic and Doppler signs of MVD, including irregularly thickened mitral valve leaflets (observed on the right parasternal 4-chamber view) and a color-flow jet of systolic mitral insufficiency in the left atrium (observed on the left parasternal 4-chamber view). Dogs with MVD (Group 2) were included in the study only if the color-flow jet of systolic mitral insufficiency was adequate for color and continuous Doppler mode examination, allowing quantification of MR using the proximal isovelocity surface area (PISA) method as previously described and validated by our group.23,24 For each dog from Group 2, the degree of heart failure was also classified according to ISACHC recommendations,25 leading to 3 subgroups (subgroups 2-1, 2-2, and 2-3 corresponding to ISACHC classes 1, 2, and 3, respectively).

Echocardiography and Standard Doppler Examination

Conventional echocardiographic and Doppler examinations were performed in awake standing dogs using continuous ECG monitoring with an ultrasound unitc equipped with 7.5–10, 5–7.5, and 2–5-MHz phased-array transducers as previously described and validated.26

Left Atrial and Ventricular Dimensions. Measurements of the aortic and left atrial diameters were obtained by a two-dimensional (2D) method using the right parasternal transaortic short axis view as previously described.27 The left atrium/aorta ratio (LA/Ao) was then calculated. LV measurements were first obtained from the right parasternal location using 2D-guided M-mode echocardiography according to recommendations of the American Society of Echocardiography (Fig 1A).28 LV systolic and diastolic diameters were then used for the respective calculation of ESV and EDV using the Teichholz formula as previously described.12 Ventricular volumes measured by this GM (GM-ESV and GM-EDV) were then indexed to body surface area (GM-ESVI and GM-EDVI), which was derived from body weight using the described equation.29 Additionally, GM-derived ventricular volumes were used to calculate the LV ejection fraction (GM-EF) according to the following formula:12

Figure 1.

 Echographic measurement of left ventricular (LV) volumes. (A) Geometric method performed using 2D-guided M-mode echocardiography from the right parasternal short axis transventricular view according to recommendations of the American Society of Echocardiography.28 (B) Simpson's derived method (SDM) performed on left parasternal apical 4-chamber view optimizing LV length and area. Using frame-by-frame analysis, end-diastolic frame (corresponding to onset of qRS, ie, at the time of mitral valve closure, left image) and end-systolic frame (corresponding to end of T wave, ie, last frame before mitral valve opening, right image) were selected. Left ventricular area was measured by tracing the endocardial border on each selected image and maximal LV length was measured from the mitral annulus to the endocardial border of the LV apex. (C) Length area method performed on the same view and using the same end-diastolic (left image) and end-systolic frame (right image) as SDM. LA, left atrium; LV, left ventricle.

LV volumes and indexed volumes were also measured by the 2 PM using the left apical 4-chamber view (Fig 1B and 1C). Briefly, long-axis images optimizing LV length and area were recorded and digitally stored. Frame-by-frame analysis was performed off-line, with selection of end-diastolic frames (corresponding to onset of qRs, ie, at the time of mitral valve closure) and end-systolic frames (corresponding to end of T wave, ie, last frame before mitral valve opening).19 The endocardial border on each selected image was traced and closed around the mitral annulus, thus delimiting the end-diastolic and end-systolic LV areas. Maximal diastolic and systolic LV lengths were also measured from the mitral annulus to the endocardial border of the LV apex. The LV volumes were then automatically calculated by a specific softwared and the subsequent EF (SDM-EF) by Simpson's rules. As for GM, the 2 SDM-derived end-diastolic and end-systolic LV volumes were also indexed to body surface area (SDM-ESVI and SDM-EDVI, respectively).29

The same end-diastolic and end-systolic LV lengths and LV areas obtained from the left apical 4-chamber view were used to calculate LV volumes and EF using the LAM formula as described above, and LV volumes were then indexed to the body surface area (LAM-ESVI, LAM-EDVI, and LAM-EF, respectively).29 Lastly, LVSI defined as the ratio of LV end-diastolic length to the M-mode LV end-diastolic diameter was calculated according to the ESVC taskforce recommendations using the left apical 4-chamber view.22

Assessment of MR Severity. MR was quantitatively assessed on each dog from Group 2 by the PISA method as previously described and validated.23 The studied PISA variable was the regurgitation fraction (RF). Color-flow Doppler examination of the tricuspid valve was performed in all cases by use of the left apical 4-chamber view. When tricuspid regurgitation (TR) was identified, peak systolic TR velocity was quantitatively assessed by continuous-wave Doppler mode. Bernouilli's equation was then applied to calculate the systolic right ventricle-to-right atrium pressure gradient across the tricuspid valve. Lastly, as previously described,5,6 the systolic pulmonary arterial pressure (SPAP) was determined by adding the estimated right atrial pressure (5 mmHg in dogs with a nonenlarged right atrium, 10 mmHg in dogs with an enlarged right atrium but no right-sided heart failure, and 15 mmHg in dogs with right-sided heart failure) to the systolic right ventricle-to-right atrium pressure gradient.


Follow-up of all symptomatic dogs (subgroups 2-2 and 2-3) was assessed 5 months after initial presentation. Dogs were classified as survivors (S) if they were still alive or nonsurvivors (NS) if they died or were euthanized for unresponsive heart failure. Additionally, dogs were classified as having “stable disease” (StD) if they either remained in the same ISACHC class or improved their clinical status. Dogs that either died or for which progression of heart failure class was observed during the 5-month period (from ISACHC class 2 to 3 or from ISACHC class 3a to 3b) were classified as having “progressing disease” (PD). Dogs lost at follow-up or that died from noncardiac diseases were censored for the statistical analysis.

Statistical Analysis

Data are expressed as mean ± standard deviation (SD). Statistical analyses were performed by computer software.e

The following linear model was used to determine the within-day and between-day variability of each GM and PM variable, and also that of LVSI (Study 1):


where Yijk is the kth value measured for dog j on day i, μ is the general mean, dayi is the differential effect of day i, dogj is the differential effect of dog j, (day × dog)ij is the interaction term between day and dog, and ɛijk is the model error. The SD of repeatability was estimated as the residual SD of the model and the SD of reproducibility as the SD of the differential effect of day. The corresponding coefficients of variation (CV) were determined by dividing each SD by the mean.

For Study 2, the descriptive statistical analysis was used for age, heart rate, body weight, sex, and clinical signs. The 3 different ultrasound techniques used to measure ESVI (GM, SDM, and LAM) were compared using Bland-Altman analysis.30 Correlations between markers of MR severity (LA/Ao, RF, and SPAP) and GM or PM variables were examined by applying the Pearson correlation analysis. Differences in continuous variables among groups (groups 1, 2-1, 2-2, and 2-3) were evaluated by a one way analysis of variance (ANOVA), followed if necessary by a Student's t-test with Bonferroni's correction. For all comparisons, P< .05 was considered significant.


Study 1: Within- and Between-Day Intraobserver Variability of GM and PM Variables

The within- and between-day CV values (n = 10) and the corresponding SD values are presented in Table 1. No significant interaction was observed between day and dog except for 1 GM variable (GM-EF).

Table 1.   Within- and between-day variability, expressed as standard deviations (SD) and coefficients of variation (CV), of left ventricular index of sphericity, left ventricular volumes, and ejection fraction assessed by 3 different echocardiographic techniques: the Teichholz formula method (geometric method, GM), the Simpson's derived, and the length area methods (planimetric methods, PM).
Independent VariableMethod of MeasurementWithin-DayBetween-Day
SDCV (%)SDCV (%)
  • *

    Significant day-dog interaction (P= .035).

  • ESVI, end-systolic indexed left ventricular volume; EDVI, end-diastolic indexed left ventricular volume.

ESVI (mL/m2)Teichholz formula method (GM)1.925.23.579.7
Simpson's derived method (PM)0.883.21.214.4
Length area method (PM)2.448.13.1810.6
EDVI (mL/m2)Teichholz formula method (GM)
Simpson's derived method (PM)1.632.31.802.5
Length area method (PM)4.315.63.74.8
Ejection fraction (%)Teichholz formula method* (GM)2.363.94.046.7
Simpson's derived method (PM)
Length area method (PM)
Left ventricular index of sphericity0.0453.30.0846.2

Study 2: Prospective Study of GM and PM Variables on a Large Canine Population

Ninety-five dogs affected by MVD with MR quantified by the PISA method were prospectively recruited from October 2006 to July 2007. Among those dogs, 13 (14%) were not included in the study because of a body weight >15 kg and 5 (5%) because of a concurrent treatment that might have exerted an influence on systolic function (pimobendan). Therefore, 77 dogs with MVD (Group 2) were included in the study. Additionally, 24 healthy small breed dogs were recruited during the same period (Group 1).

Epidemiologic and Clinical Characteristics of the Healthy Control Group (Group 1, n = 24) and Dogs with MVD (Group 2, n = 77). The epidemiologic characteristics of dogs from Groups 1 and 2 are presented in Table 2. As expected, Group 2 was mostly composed of males (n = 60, 78%), aged adult dogs (10.9 ± 3.8 years, range: 3.0–17.0 years). English Toy Spaniels, Poodles, Yorkshire Terriers, and cross-breed dogs were overrepresented in both groups.

Table 2.   Demographic characteristics of healthy control dogs (Group 1, n = 24) and dogs with mitral valve disease (Group 2, n = 77).
(n = 101)
Group 1
Healthy Control
(n = 24)
Group 2 Dogs
with MVD
(n = 77)
  • *

    , P < .05 versus Group 1.

Age (years) (mean ± SD, [range])9.4 ± 4.54.7 ± 2.810.9 ± 3.8*
Body weight (kg) (mean ± SD, [range])8.2 ± 3.08.1 ± 2.88.2 ± 3.1
 Cavalier and KingCharles Spaniel28%29%27%
 Cross breed15%13%16%
 Yorkshire Terrier11%8%12%
 Other breeds30%42%27%

As shown in Table 3, 49% (38/77) of the dogs with MVD were asymptomatic and therefore belonged to ISACHC class 1 (subgroup 2-1). The remaining 39 dogs were symptomatic and assigned to either ISACHC classes 2 (n = 24/77, 31%) or 3 (n = 15/77, 20%), ie, subgroups 2-2 and 2-3, respectively. Exercise intolerance and chronic cough were by far the most common clinical signs reported (respectively 31/39 and 24/39, ie, 79 and 62%, respectively) in those animals. Clinical signs indicative of right-sided congestive heart failure were also reported (ascites in 15% of the symptomatic dogs, n = 6).

Table 3.   Standard echo-Doppler variables assessed in healthy control dogs (Group 1, n = 24) and dogs with mitral valve disease (Group 2, divided in subgroups 2-1, 2-2, and 2-3 for corresponding ISACHC classes 1, 2, and 3, respectively; n = 77).
Echo-Doppler Quantitative
Healthy Control
(Group 1) (n = 24)
Whole MVD Population
(Group 2) (n = 77)
(Subgroup 2-1)
(n = 38)
(Subgroup 2-2)
(n = 24)
(Subgroup 2-3)
(n = 15)
  • *

    : P < .05 versus Group 1.

  • : P < .05 versus Subgroup 2-1.

  • #

    : P < .05 versus Subgroup 2-2.

  • SPAP, systolic pulmonary arterial pressure; n, number of dogs for which variables were available; EDVI, end-diastolic volume indexed to body surface area; ESVI, end-systolic volume indexed to body surface area; EF, ejection fraction of the left ventricle; LVSI, left ventricular index of sphericity; SPAP, systolic pulmonary arterial pressure; NM, not measured.

Left atrium to aorta ratio
Mean ± SD0.86 ± 0.171.58 ± 0.70*1.04 ± 0.27*1.89 ± 0.54*,‡2.45 ± 0.49*,‡,#
Regurgitant fraction (%)
Mean ± SDNM38.5 ± 17.827.8 ± 14.746.1 ± 13.953.2 ± 13.8
[range] [6.6–79.6][6.6–70.3][28.3–79.6][31.6–73.2]
SPAP (mmHg)
Mean ± SD14 ± 6.552.1 ± 27.8*39.1 ± 17.0*56.0 ± 31.1*,‡74.5 ± 27.4*,‡
(Assessed in n dogs)(n = 13)(n = 72)(n = 33)(n = 24)(n = 15)
Fractional shortening
Mean ± SD38.8 ± 4.845.9 ± 8.7*42.0 ± 7.047.8 ± 6.9*,‡52.7 ± 10.1*,‡
Geometric method (GM)
 GM-EDVI (mL/m2)
 Mean ± SD80.4 ± 18.6130.8 ± 52.5*100.8 ± 31.1*146.7 ± 42.0*,‡181.1 ± 62.8*,‡,#
 GM-ESVI (mL/m2)
 Mean ± SD23.4 ± 7.328.8 ± 14.726.6 ± 11.129.8 ± 13.032.8 ± 23.3
 GM-EF (%)
 Mean ± SD70.9 ± 6.077.4 ± 8.7*73.6 ± 7.879.5 ± 7.3*,‡83.7 ± 8.7*,‡
Planimetric methods
 Simpson's derived method (SDM)
  SDM-EDVI (mL/m2)
  Mean ± SD47.6 ± 8.475.2 ± 27.5*63.6 ± 17.9*77.4 ± 25.0*,‡101.1 ± 33.8*,‡,#
  SDM-ESVI (mL/m2)
  Mean ± SD15.9 ± 3.919.2 ± 7.2*17.9 ± 5.218.9 ± 7.223.0 ± 10.1*,‡
  SDM-EF (%)
  Mean ± SD66.5 ± 6.473.6 ± 6.4*71.2 ± 6.1*75.3 ± 6.3*,‡77.2 ± 5.2*,‡
 Length area method (LAM)
  LAM-EDVI (mL/m2)
  Mean ± SD49.5 ± 10.377.4 ± 27.9*65.2 ± 19.0*81.5 ± 25.0*,‡101.5 ± 34.5*,‡,#
  LAM-ESVI (mL/m2)
  Mean ± SD16.9 ± 4.320.3 ± 7.6*19.2 ± 5.619.4 ± 6.824.5 ± 11.6*,‡
  LAM-EF (%)
  Mean ± SD65.5 ± 6.772.6 ± 7.4*69.6 ± 7.475.3 ± 6.5*,‡75.9 ± 6.3*,‡
LVSI mean ± SD
1.43 ± 0.121.31 ± 0.17*1.38 ± 0.171.24 ± 0.14*,‡1.25 ± 0.17*,‡

None of the dogs from Group 1 received any treatment at the time of diagnosis, whereas 18/38 dogs from subgroup 2-1 (47%), 17/24 dogs from subgroup 2-2 (71%), and 14/15 dogs from subgroup 2-3 (93%) received at least 1 treatment for heart disease at the time of diagnosis. Treatment included angiotensin-converting enzyme inhibitors such as benazepril, enalapril, imidapril, and ramipril (47/77 dogs [61%]), furosemide (22/77 dogs [29%]), spironolactone (19/77 dogs [25%]), theophylline (6/77 dogs [8%]), amiodarone (2/77 dogs [3%]), and diltiazem (1/77 dog [1%]).

Echocardiographic and Doppler Assessment of MR Severity in Dogs with MVD (Group 2, n = 77). As an inclusion criterion, MR severity was evaluated by PISA method with calculation of RF (38.5 ± 17.8%, reference range: 6.6–79.6%) on all the dogs from Group 2 (Table 3). TR adequate for indirect assessment of SPAP was possible in 13/24 dogs from Group 1 (54%) and 72/77 dogs from Group 2 (94%). Within Group 2, a significant correlation (P < .05) was found between RF and the LA/Ao ratio (r= 0.67), between RF and SPAP (r= 0.57), and between SPAP and the LA/Ao ratio (r= 0.52) (Fig 2A–2C). A significant increase in LA/Ao ratio, RF, and SPAP was found from subgroup 2-1 to subgroup 2-3 (P < .05, Table 3). Additionally, LA/Ao and SPAP were significantly higher in all MVD groups (ie, Group 2, and subgroups 2-1, 2-2, and 2-3) than in Group 1 (Table 3).

Figure 2.

 Correlations between 3 echographic markers of mitral valve disease severity: left atrium to aorta ratio (LA/Ao), regurgitant fraction calculated by the PISA method (RF), and Doppler derived estimation of systolic pulmonary arterial pressure (SPAP). Figures represent correlations between LA/Ao and RF (A), SPAP and RF (B), and LA/Ao and SPAP (C).

Assessment of LV Shape, Volumes, and Systolic Function in the Healthy Control Group (Group 1, n = 24) and Dogs with MVD (Group 2, n = 77). A significant decrease in LVSI was found in symptomatic dogs with MVD (subgroups 2-2 and 2-3) compared with asymptomatic dogs (both Group 1 and subgroup 2-1), indicating that progression of MVD was associated with increased sphericity of the LV (P < .05, Table 3). Concomitantly to this increased LV sphericity and whatever the method used (GM or PM), a significant increase in EDVI was observed in subgroup 2-3 and in subgroup 2-2 compared with both Groups 1 and subgroup 2-1 (P < .05, Table 3). Additionally, SDM-ESVI and LAM-ESVI were significantly higher in subgroup 2-3 compared with both Group 1 and subgroup 2-1 (P < .05) (Fig 3). Finally, a significant increase in FS and EF calculated by the 3 echocardiographic methods was found from Group 1 to subgroup 2-3 (P < .05). These variables were also significantly positively correlated (P < .05) with RF in dogs from Group 2 (r= 0.37, 0.35, 0.39, and 0.49 for FS, GM-EF, SDM-EF, and LAM-EF, respectively).

Figure 3.

 Distribution of end-systolic volume index (ESVI) values assessed using the geometric method, the Simpson's derived method, and the length area method (A, B, C, respectively) within the 4 clinical groups considered (ie, Group 1 and subgroups 2-1, 2-2 and 2-3 for healthy control and ISACHC class 1, 2, and 3 MVD dogs, respectively). MVD, mitral valve disease.

Comparison between GM and PM for Assessment of LV Volumes in the Whole Study Population (n = 101) Including 24 Healthy Dogs and 77 Dogs with MVD. When GM and PM were compared, a significant correlation (P < .05) was found between all methods for measurement of ESVI, with the correlation between LAM-ESVI and SDM-ESVI (r= 0.92) being higher than between LAM-ESVI and GM-ESVI (r= 0.71), and SDM-ESVI and GM-ESVI (r= 0.67). For EDVI and EF, correlations were also significant among all groups (P < 0.05) with the correlation between LAM and SDM (r= 0.96 and 0.71) being higher than between LAM and GM (r= 0.84 and 0.53), and also between SDM and GM (r= 0.86 and 0.56). For measurement of ESVI, GM gave significantly higher values than LAM and SDM in all groups (P < .05), whereas both PM gave nonsignificantly different results. In addition, Bland-Altman analysis revealed that the difference between PM and GM increased as the mean ESVI value increased (Fig 4A–4C). This finding was associated with high limits of agreements when PM and GM were compared because the 95% confidence interval (CI) of the difference between the values assessed by these techniques was 9.6 ± 11.2 (for GM and SDM) and 8.5 ± 10.6 (for GM and LAM) (Fig 4A and 4B). Conversely, the comparison between LAM and SDM showed tight limits of agreements (95% CI=− 1.1 ± 3.0, Fig 4C).

Figure 4.

 Bland-Altman analysis illustrating the difference between ESVI values assessed by 2 methods against the average of these 2 values. The mean differences are indicated as horizontal dashed lines, and limits of agreement (mean ± 1.96 SD or 95% CI) are indicated as horizontal continuous gray lines. ESVI, end-systolic volume index. (A) Comparison of the geometric method and Simpson's derived method. (B) Comparison of the geometric method and length area method. (C) Comparison of the Simpson's derived method and length area method.

Correlation of LV Volumes to MR Severity (Group 2, n = 77). Correlation analysis failed to identify a significant correlation between ESVI (assessed by GM, SDM, and LAM) and RF (r= 0.11, 0.2, and 0.15, respectively). Conversely, EDVI was significantly correlated to RF (r= 0.46, 0.43, and 0.44, for GM, SDM, and LAM respectively, P < .05). Similarly, SPAP was not correlated to GM-, SDM-, and LAM-ESVI (r= 0.19, 0.15, and 0.11, respectively). However, a significant positive correlation (P < .05) was found between SPAP and EDVI (r= 0.34, 0.34, and 0.35, for GM, SDM, and LAM, respectively). Finally, LA/Ao was significantly correlated (P < .05) to both ESVI (r= 0.37, 0.35, and 0.34, for GM, SDM, and LAM, respectively) and EDVI (r= 0.74, 0.64, and 0.62, for GM, SDM, and LAM, respectively).

Survival in Dogs with Symptomatic MVD (SubGroups 2-2 and 2-3, n = 39). At the time of writing, follow-up at 5 months was available for 28/39 symptomatic dogs (72%). Five more of those dogs were excluded from the statistical analysis owing to noncardiac-related death, ie, euthanasia because of neoplasia (n = 2), or neurologic disease, trauma, and hepatic disease (1 dog each). Among the 23 dogs available for statistical analysis, 15 were still alive (S group) and 8 (NS group: NS) either died from heart failure (n = 6) or were euthanized owing to unresponsive heart failure (n = 2). In addition, in 5 dogs from the S group, cardiac conditions were assessed as worsening, with progression from ISACHC class 2 to class 3 in 4 dogs and progression from ISACHC class 3a to ISACHC 3b in 1 dog. Those dogs together with NS dogs formed the “progressing disease” (PD) group (n = 13), and were compared with the remaining 10 dogs classified as presenting a stable heart disease (StD). Clinical and echocardiographic characteristics of S, NS, PD, and StD dogs are presented in Table 4. A marked effect of the clinical status on survival was found. Dogs from the PD group had statistically higher RF, SPAP, FS, LAM-EDVI, SDM-EDVI, and LAM-EF than StD dogs at the time of inclusion (P < .05). Dogs from the NS group had statistically higher LA/Ao, EDVI (assessed by all methods), and ESVI (assessed by LAM and SDM) than S dogs (P < .05).

Table 4.   Clinical and standard echo-Doppler variables of dogs with symptomatic mitral valve disease (ISACHC class 2 or 3 at the time of diagnosis) for which follow-up at 5 months was available (n = 23).
Clinical and Echocardiographic
Quantitative Variables
Whole Population
(n = 23)
Survivor Group
(S) (n = 15)
Nonsurvivor Group
(NS) (n = 8)
Stable Disease
(StD) (n = 10)
Progressing Disease
(PD) (n = 13)
  • LA/Ao, left atrial to aortic ratio; SPAP, systolic pulmonary arterial pressure; n, number of dogs for which variables were available; EDVI, end-diastolic volume indexed to body surface area; ESVI, end-systolic volume indexed to body surface area; EF, ejection fraction of the left ventricle; GM, geometric method.

  • *

    : P < .05 versus corresponding Group S.

  • : P < .05 versus corresponding Group StD.

ISACHC class 217 (74%)14/15 (%)3/8 (37%)10/10 (100%)7/13 (54%)
ISACHC class 36 (26%)1/15 (%)5/8 (63%)0/10 (0%)6/13 (46%)
Presence of ascites5 (22%)1/15 (7%)4/8 (50%)0/10 (0%)5/5 (100%)
LA/Ao ratio2.00 ± 0.641.71 ± 0.582.49 ± 0.47*1.73 ± 0.672.19 ± 0.66
Regurgitant fraction (%)47.9 ± 13.545.5 ± 13.854.0 ± 11.741.5 ± 8.653.9 ± 15.7
Estimated SPAP (mmHg)61.2 ± 34.556.6 ± 39.071.0 ± 26.641.1 ± 14.677.4 ± 42.8
Fractional shortening
Mean ± SD48.9 ± 7.947.3 ± 8.651.3 ± 6.644.8 ± 7.951.6 ± 10.0
Teichholz formula method (GM)
 GM-EDVI (mL/m2)
 Mean ± SD158.1 ± 53.2135.4 ± 34.3196.1 ± 63.4*131.8 ± 33.2175.6 ± 40.1
 GM-ESVI (mL/m2)
 Mean ± SD30.9 ± 15.628.6 ± 14.835.7 ± 17.831.5 ± 16.630.7 ± 17.9
 GM-EF (%)
 Mean ± SD80.3 ± 7.978.9 ± 9.082.4 ± 5.676.5 ± 9.182.9 ± 10.6
Planimetric methods
 Simpson's derived method (SDM)
  SDM-EDVI (mL/m2)
  Mean ± SD86.5 ± 34.270.6 ± 21.4119.0 ± 33.4*69.7 ± 25.0101.0 ± 24.4
  SDM-ESVI (mL/m2)
  Mean ± SD20.2 ± 7.718.2 ± 6.425.0 ± 8.4*18.2 ± 7.622.4 ± 6.9
  SDM-EF (%)
  Mean ± SD75.8 ± 6.474.0 ± 6.778.6 ± 4.973.8 ± 7.077.0 ± 6.3
 Length area method (LAM)
  LAM-EDVI (mL/m2)
Mean ± SD87.7 ± 32.172.9 ± 21.5117.4 ± 30.8*69.8 ± 25.2102.6 ± 25.2
  LAM-ESVI (mL/m2)
  Mean ± SD20.6 ± 7.418.0 ± 5.526.3 ± 7.7*18.8 ± 6.122.5 ± 6.4
  LAM-EF (%)
  Mean ± SD75.3 ± 6.473.9 ± 7.177.1 ± 4.771.2 ± 6.878.0 ± 7.2


Few studies have focused on systolic function in dogs with MVD.12,31 We report the intraobserver variability of markers for quantifying LV shape, volumes, and function calculated by different echocardiographic methods (GM and PM) and association in dogs with MVD with regard to heart failure severity, several echocardiographic and Doppler variables (LA/Ao, RF, and SPAP), and short-term prognosis.

The within- and between-day variability of LV volumes and EF assessed by 3 different echocardiographic techniques (ie, Teichholz formula method, SDM, LAM) were good and comparable in the awake standing dog. The within- and between-day CV values for all indices (ESVI, EDVI, and EF) were <11% (2–10.6%) without any interaction between dog and day for the 2 PM (SDM, LAM). Similar results were obtained for LVSI, which combines 1 M-mode variable (LV end-diastolic diameter) and 1 2D variable (LV end-diastolic length measured on the left apical 4-chamber view). These results are similar to those obtained in humans.32

Study 2 results demonstrate that MVD progression is associated with a change in LV shape. This shape alteration is characterized by an increased sphericity as shown by the significant decrease in LVSI with ISACHC heart failure class. Because the Teichholz formula does not take into account this progressive change in LV shape during the time course of the disease,33 the GM is inaccurate for assessing LV volumes and the subsequent EF in dogs with MVD. This inadequacy of GM compared with the 2 other “anatomic-based” methods tested here (SDM and LAM) was confirmed by the Bland-Altman results, which indicate that GM overestimates ESVI in a nonlinear way. This bias is of major importance because it means that GM leads to an overestimation of LV size that becomes more marked with disease progression. These results are consistent with a comparative echographic and catheter-based study performed on dogs with drugs-induced changes in loading conditions.b Moreover, there were tight limits of agreement between the 2 PM. From a practical point of view, this finding is of importance because LAM can be easily performed with any commercially available echocardiographic system, without requiring a specific software (which is not the case for SDM).

The results of Study 2 also confirm the presence of increased ESVI in dogs with symptomatic MVD, as already reported in large- and small-breed dogs.12 This confirms that a “global” SyD becomes apparent during the course of the disease. However, although our study included more diseased small-breed dogs than in the Borgarelli et al report12 (77 versus 18, respectively), an increased ESVI could not be identified in ISACHC class 2 dogs (subgroup 2-2). This lack of statistical significance was probably related to the particularly wide range of ESVI observed in both healthy and diseased dogs, with all methods performed. The use of indices indexed to body surface area should theoretically have limited the influence of a weight effect. However, in the present study a marked individual variability exists, leading to an important overlap between healthy and diseased dogs, therefore, limiting the possibility to determine a clear pathologic threshold. For example, the use of a GM-ESVI higher than 30 mL/m2, a threshold commonly described as indicative of SyD in humans (but never validated in the dog),7,12 would lead to the diagnosis of SyD in 37.5% (9/24) and 53% (8/15) of symptomatic dogs (ISACHC classes 2 and 3, respectively) from the present study. However, the same threshold would also “detect” SyD in 34% (13/38) of ISACHC class 1 dogs and 29% (7/24) of healthy control dogs. This wide range of ESVI observed in healthy dogs may be related to interbreed differences in systolic function, presence of occult primary myocardial disease, athletic ability, or differences in autonomic nervous system tone. The same factors of interindividual variation may also be found in diseased dogs, because of the potential interaction between hyperkinesia secondary to MR and SyD intrinsically related to MVD. Finally, one may hypothesize that right ventricular alteration related to PH may also be responsible for a change in LV volume. In symptomatic dogs, the estimated SPAP was frequently markedly increased, reaching a similar level in certain cases to systemic pressure. This increased SPAP could exert an influence on LV volume by both decreasing LV preload and “mechanically” limiting LV expansion with flattening of the interventricular septum. This probably explains why some dogs with severe MVD in our study showed decreased LV volumes when assessed by GM. These limitations may be overcome by new promising noninvasive imaging techniques such as tissue Doppler imaging derived techniques. Recently, the diagnostic interest of myocardial strain and strain rate imaging has been described in dogs with MVD,f alterations in these parameters being observed in dogs with severe MR compared with control dogs.

Indirect assessment of MR incompetence (assessed by RF) and its consequences on pulmonary vascular tone (assessed by SPAP) have already been described by our group to be significantly correlated to heart failure classes, left atrial size, and also to each other.5,23 The present study confirms these previous results, but fails to show a significant correlation between ESVI and both RF and SPAP. This lack of correlation between ESVI and these 2 markers of MR severity may be explained by the fact that in symptomatic dogs, RF and SPAP were markedly increased in almost all dogs, whereas ESVI showed greater variability.

Nevertheless, although PM-derived ESVI, RF, and SPAP appear to be relatively independent of each other, the present results also indicate that all these variables represent prognostic factors for survival or disease worsening of symptomatic MVD. To date, the prognostic interest of ESVI has only been retrospectively studied in dogs with asymptomatic MVD, and was found to exert no influence over probability of the disease worsening.g

The current studies have several limitations. First, the validation protocol was performed in healthy dogs of various sizes, and the results are not necessarily transposable to small breed diseased dogs. Additionally, this validation protocol did not include assessment of interobserver variability. Second, in our study, as is commonly the case in humans,18 PM relied on the left apical window, which may truncate the apex.b The right parasternal long axis 4-chamber view might have been a useful alternative.b Additionally, the dogs in Group 1 were significantly younger and with a different sex ratio than those in Group 2, and an influence of age or sex on systolic function may be hypothesized. However, ESVI has already been demonstrated to be independent of age when young and old dogs of the same breed were compared.34 More than 5 dog breeds were used in the present study, and this represents another limitation. A clear breed effect on systolic function has been demonstrated using the tissue Doppler imaging technique.27 Therefore, a comparison of PM and GM systolic variables to catheter-based measurements within a single breed of diseased dogs, together with the assessment of inter- and intraobserver variabilities, would have been the ideal protocol design.

In conclusion, according to our results, the use of GM based on the Teichholz formula should not be recommended for the assessment of LV volume in dogs with MVD, because this method does not take into account the increasing sphericity that occurs during the disease progression. The 2 PM may be used instead with acceptable repeatability, reproducibility, and good limits of agreement. However, a pathologic threshold for PM-ESVI allowing diagnosis of SyD in dogs with MVD could not be fixed from the present results, owing to an important overlap of values between healthy dogs and dogs with MVD from different ISACHC classes. Further studies should, therefore, be carried out to establish breed-specific reference ranges of PM variables. Until such studies are completed, the detection of SyD using PM-ESVI should rely on longitudinal follow-up on the same animal. The long-term prognostic value of PM systolic variables also needs to be assessed in further prospective studies.


aO'Sullivan ML, O'Grady MR, Walker C, et al. Frequency of ventricular ectopy in dogs with chronic mitral valve disease and congestive heart failure treated with pimobendan or benazepril. Presented at the 25th Annual ACVIM Forum Seattle, WA, June 6–9, 2007. J Vet Intern Med 2007;21:587 (abstract)

bDe Morais HAS, Bonagura JD, Muir III WW, Nakade T. Left ventricular volumes obtained by echocardiography in intact dogs: Validation using the conductance catheter. Presented at the 15th Annual ACVIM Forum, Lake Buena Vista, FL, May 22–25, 1997. J Vet Intern Med 1997;11:140 (abstract)

cVivid 7, General Electric Medical System, Waukesha, WI

dEchopac Dimension, General Electric Medical System

eSystat version 10.0, SPSS Inc, Chicago, IL

fWess G, Javornik A, Keller K, et al. Differences in systolic function between small and large breed dogs with myxomatous mitral valve degeneration assessed by myocardial strain and strain rate measurements. Presented at the 25th Annual ACVIM Forum, Seattle, WA, June 6–9, 2007. J Vet Intern Med 2007;21:590 (abstract)

gCrosara S, Borgarelli M, La Rosa G, et al. Natural history and risk predictors of chronic degenerative mitral valve disease in asymptomatic dogs. Presented at the 17th Annual ECVIM Forum, Budapest, Hungary, September 13–15, 2007 (abstract)


This study was supported by the resident grant program of Vetoquinol Pharmaceutical Laboratory, Lure cedex, France. The Vivid 7 ultrasound system was sponsored by Novartis Animal Health, Rueil Malmaison, France.