• Open Access

Systemic Inflammation and Endothelial Dysfunction in Dogs with Congestive Heart Failure


  • Partial results of this study were presented as a Research Report at the 2010 ACVIM Forum in Anaheim, CA.

Corresponding author: Suzanne M. Cunningham, Tufts Cummings School of Veterinary Medicine, Department of Clinical Sciences, 200 Westboro Road, North Grafton, MA 01536; e-mail: suzanne.cunningham@tufts.edu.



Congestive heart failure (CHF) is associated with endothelial dysfunction in people and in dogs with experimentally induced CHF, but this is not well characterized in dogs with naturally occurring CHF.


To evaluate endothelial function via assessment of reactive hyperemia (RH) in healthy dogs and dogs with CHF, and to assess for relationships with plasma biomarkers of vascular function and clinical markers of disease severity.


Twenty client-owned animals with CHF due to myxomatous mitral valve disease (n = 15) or dilated cardiomyopathy (n = 5) and 17 healthy control dogs.


Prospective case-controlled observational study. Dogs underwent blood sampling, echocardiography, and Doppler assessment of brachial artery velocity (VTI) at baseline and during reactive hyperemia (RH-VTI). RH-VTIs between control dogs and dogs with CHF were compared, and the relationships between RH-VTI, clinical parameters, and plasma biomarkers were assessed.


Dogs with CHF (96.5 ± 51.7%) had an attenuated % increase in VTI during RH compared to healthy controls (134.8 ± 58.7%; = .04). Increasing ISACHC class (R2 = 0.24; = .004), plasma NT-proBNP (R2 = 0.15; = .03) and CRP (R2 = 0.2; = .02) were associated with reduced RH-VTI. Increased plasma CRP, NOx, and NT-proBNP concentrations were found in dogs with CHF (< .02 for all). No differences were detected in other plasma markers.

Conclusions and Clinical Importance

Dogs with CHF have an attenuated RH response, and increased plasma CRP and NOx concentrations. Doppler assessment of RH velocity could represent a novel noninvasive method of evaluating endothelial function in the dog.


asymmetric dimethylarginine


blood pressure


congestive heart failure


C-reactive protein


coefficient of variation


flow-mediated vasodilation


fractional shortening


heart rate


International Small Animal Cardiac Health Council



LA : Ao

left atrial to aortic root ratio


mitral regurgitation


myxomatous mitral valve disease


nitric oxide


nitrate and nitrite


N-terminal pro B-type natriuretic peptide


reactive hyperemia


right parasternal


symmetric dimethylarginine


vertebral heart score


von Willebrand factor antigen


velocity time integral

Cardiovascular disease is a significant cause of morbidity in dogs, estimated to affect at least 11% of the total canine population.[1] Both myxomatous mitral valve disease (MMVD) and dilated cardiomyopathy (DCM) are progressive diseases, sharing a common endpoint of congestive heart failure (CHF) in a substantial proportion of affected dogs.[2, 3] Once CHF develops, the long-term prognosis is poor; treatment remains limited to symptomatic management, and survival times are often < 1 year.[2] Additional research into the etiopathogenesis of CHF is needed to facilitate identification of novel therapeutic targets.

Impaired function of the vascular endothelium is well documented in human CHF patients[4-7] and is associated with increased cardiac afterload, diminished tissue perfusion, reduced exercise tolerance, and disease progression.[4, 5, 7-10] Endothelial dysfunction has also been identified in dogs with experimentally induced CHF[11-13] and might be similarly implicated in the pathogenesis and progression of naturally occurring CHF in dogs. However, investigations of endothelial function in this population have been limited.[14] Noninvasive evaluation of endothelial function can be accomplished via measurement of plasma biomarkers,[6, 15-20] endothelium-dependent flow-mediated vasodilation (FMD),[21, 22] or assessment of reactive hyperemia (RH).[23-26] Potential plasma markers of endothelial function include markers of systemic inflammation such as C-reactive protein (CRP)[27-30]; markers of nitric oxide (NO) bioavailability such as nitrate and nitrite (NOx),[15, 16, 31]l-arginine (l-Arg), asymmetric dimethylarginine (ADMA), and symmetric dimethylarginine (SDMA)[18, 19, 32, 33]; and indicators of endothelial damage or repair such as von Willebrand factor (vWF)[17, 34] or circulating endothelial progenitor cells,[35] respectively. In people, FMD is a well-validated marker of endothelial function[21, 22] and a good predictor of adverse outcomes in chronic CHF.[5, 8-10] Measurement of FMD is also feasible in dogs;[36, 37] however, this technique can be limited by body size dependence, high variability, and difficulty in accurately measuring small arterial diameters.[36, 37] FMD can also be impacted by factors such as the age and sex of the patient and prior cardiac therapies.[38, 39]

Flow-mediated vasodilation reflects the ability of the vascular endothelium to release NO and other vasodilators in response to arterial occlusion-induced tissue hypoxia and the subsequent increase in endothelial shear stress. Dilation of resistance arterioles during the period of tissue hypoxia results in a transient period of increased blood flow upon release of arterial occlusion, termed RH. The magnitude of the RH response is an indicator of microvascular endothelial function[25, 26, 40-42] that can be measured relatively easily and noninvasively via Doppler ultrasound. Doppler-derived hyperemic flow velocities are strongly correlated to FMD measures in people, and may be similarly impacted by factors such as patient age, sex, and cardiovascular therapies.[24-26] However, recent evidence suggests that brachial artery flow velocity during RH might be a more sensitive predictor of adverse cardiovascular outcomes in people than is FMD.[24-26] The aim of this study was to determine the feasibility of measuring brachial artery RH responses in a diverse population of healthy dogs and to assess for alterations in RH in dogs with CHF. In addition, potential relationships between RH, plasma biomarkers of endothelial function, and clinical markers of cardiac disease severity were investigated.

Materials and Methods

Study Population

Healthy dogs owned by veterinary students, faculty, and staff of Tufts Cummings School of Veterinary Medicine (TCSVM), and client-owned dogs with CHF were enrolled in the study between July 2009 and July 2010. The study was approved by the TCSVM Clinical Studies Review Committee. All owners provided written informed consent prior to study enrollment.

Inclusion Criteria and Diagnostic Screening

Healthy dogs were recruited for participation in the study and enrolled on the basis of normal history, physical examination findings, NT-proBNP, and echocardiogram. Dogs with CHF due to MMVD or DCM were eligible for this study if they were free of other systemic diseases that could affect endothelial function, such as obvious infection or neoplasia, primary renal failure, or uncontrolled endocrine disease. Dogs with atrial fibrillation or sustained ventricular tachycardia at the time of the exam were also excluded. International Small Animal Cardiac Health Council (ISACHC) classification was recorded on each dog with CHF.[43] Echocardiography, including standard M-mode, 2-dimensional, color-flow, spectral Doppler and tissue Doppler imaging was performed by a single experienced examiner (SMC) in all dogs.1 M-mode dimensions were measured in right parasternal (RPS) short axis views using 2-dimensional guidance in accordance with the guidelines established by the American Society of Echocardiography.[44] The 2-dimensional left atrial to aortic ratio (2D LA : Ao) was obtained in the RPS short axis view during end-systole. Severity of mitral regurgitation (MR) was quantified in the RPS long-axis 4-chamber and left apical 4-chamber views as follows: 1 + MR defines a regurgitant jet comprising less than 20% of the left atrial (LA) area; 2 + MR describes a jet comprising 20–40% of the LA area; and 3 + MR describes a jet occupying >40% of the LA area. Mitral inflow E and A wave velocities were evaluated using pulsed wave Doppler in the left apical view by placing a 2 mm sample volume at the tips of the mitral valve leaflets and the ratio of mitral E wave to A wave velocity (mitral E : A) was calculated. Pulsed wave tissue Doppler E' and A' velocities were recorded from the lateral mitral annulus and the ratio of mitral E wave velocity to E' velocity (E/E') was calculated. Continuous ECG monitoring was implemented during the echocardiographic examination.

All dogs in the CHF group underwent thoracic radiography and determination of vertebral heart score (VHS). Congestive heart failure was diagnosed on the basis of radiographic evidence of cardiogenic pulmonary infiltrates or cavitary effusions, with accompanying severe atrial enlargement and clinical signs compatible with CHF. A diagnosis of MMVD was made if there was echocardiographic evidence of mitral valve thickening and prolapse with concomitant severe mitral or tricuspid regurgitation. Dilated cardiomyopathy was diagnosed based on echocardiographic evidence of LV wall thinning, chamber dilation, and reduced LV contractile function (defined as fractional shortening [FS] of <25%). Indirect Doppler blood pressure (BP) was measured in all dogs; the mean of 3 systolic measurements was used.

Circulating Biomarkers

Dogs in both the CHF and control groups were fasted for 8–12 hours prior to blood collection. Blood was collected from each dog in a sitting position via jugular venipuncture using the vacutainer system, and placed in appropriate serum separator (for serum biochemistry analysis), EDTA (for complete blood count, CRP, NOx, dimethylarginine, and l-arg analysis), citrate (for vWF analysis), or NT-proBNP transport tubes. A complete blood count and serum biochemistry analysis were immediately performed on all dogs. Other tubes were centrifuged, and plasma was separated within 30 minutes of collection and stored at −80°C until batched analysis. Plasma for NT-proBNP was shipped to the assay laboratory on ice according to the guidelines established by the manufacturer and was analyzed using a commercially available assay for the quantitative determination of canine NT-proBNP.2 Plasma CRP3 and NOx4 concentrations were analyzed in duplicate by means of commercially available ELISA according to the manufacturer's instructions. The interassay coefficient of variation for the CRP ELISA was 8.2% at low concentrations and 7.8% at high concentrations whereas the interassay coefficient of variation for the NOx kit was 3.4%. Circulating vWF antigen concentrations were assessed using a latex-enhanced immunoturbidimetric assay.5 Plasma concentrations of l-arg, ADMA and symmetric dimethylarginine (SDMA) were determined simultaneously using high-performance liquid chromatography (HPLC) validated for canine plasma.6 Briefly, plasma samples were diluted 1 : 1 with high purity water and applied to a preconditioned carboxylic acid solid phase extraction cartridge.7 ADMA and SDMA were eluted with ammonia: methanol and taken to dryness under nitrogen. The dried samples were derivatized with a fluorescent tag and injected onto an HPLC with fluorescent detector.8 ADMA and SDMA concentrations were quantified by determining peak areas which were read off the standard curve; the inter-assay coefficient of variation for these analyses was 4.1%.

Brachial Artery Reactive Hyperemia

Dogs were positioned on an echocardiography table in right lateral recumbency in a dark, temperature-controlled room and allowed to acclimate for 10 minutes prior to the ultrasound examination. None of the dogs were sedated for the examination. An appropriately sized sphygmomanometric cuff (width of the cuff approximating 40% of the circumference of the antebrachium) was secured in place on the right antebrachium, just distal to the elbow. An ultrasound unit1 equipped with a 10 MHz variable frequency probe was used to image the brachial artery proximal to the elbow. The same experienced examiner (SMC) performed all brachial artery ultrasound examinations. The artery was located using 2-D and color Doppler, using the anatomic landmarks previously described.[36] The vessel was imaged longitudinally following rotation and angulation of the transducer so that the ultrasound beam was as parallel as feasible to the length of the vessel. Pulsed wave (PW) Doppler was used to record blood flow in the brachial artery using a constant 2 mm sample volume. Angle correction was utilized and the angle of insonation was held constant at a value of less than 60o throughout the exam. Baseline PW brachial artery velocity waveforms were recorded. The sphygmomanometric cuff was then inflated to a pressure of 220 mm Hg, or at least 50 mm Hg greater than the dog's systolic BP, for a 3-minute period of occlusion. The cuff was then rapidly deflated and PW velocities were subsequently recorded during and after RH at 10, 15, 20, 30, 45, and 60 second time points following cuff deflation. Continuous ECG monitoring was implemented throughout the exam and used to confirm the timing of Doppler waveforms. Peak and mean velocities and velocity time integrals (VTIs) were manually traced for the positive early systolic wave (S), the subsequent first negative late systolic reversal wave (M), and the second positive diastolic wave (D; Fig 1). For each pre and postocclusion time interval, the average VTI over 3 cardiac cycles was measured. The absolute change (delta) in mean VTI from baseline to each of the RH intervals was recorded and the% increase in VTI was calculated using the following formula: ([peak hyperemic VTI − baseline VTI]/baseline VTI)*100).

Figure 1.

Representative pulsed wave Doppler waveforms recorded from the brachial artery of a dog before (A) and during peak reactive hyperemia (RH; B) 10 seconds after release of a 3-minute period of cuff occlusion. The increase in flow during RH results in increased forward flow velocity and velocity time integral (VTI) and spectral broadening during systole and diastole; the loss of the negative reversal wave can also be appreciated. S, systolic wave; M, negative (minimum) reversal wave; D, diastolic wave.

Brachial Artery Hyperemic Velocity Repeatability Cohort

To evaluate intraobserver repeatability of baseline and hyperemic brachial artery VTI, 9 dogs were imaged by the same observer (SMC) twice during the same day with a 10 minute washout period between examinations. Intra- and interexamination coefficients of variation (CV) were calculated for baseline systolic and diastolic VTI and for the increase in VTI during RH at 10 seconds post occlusion.

Statistical Analysis

Data are presented as mean ± SD or as median and interquartile range when the data were not normally distributed. Data were graphically inspected, and normality was assessed using the Kolmogrov-Smirnov test. All data that were not normally distributed were logarithmically transformed prior to analysis. Continuous variables were compared between groups using independent t-tests. The post occlusion changes in systolic VTI over time were compared between the CHF and controls using analysis of variance with repeated measures. Associations of baseline and delta VTI during RH with clinical and echocardiographic variables (age, weight, heart rate [HR], BP, ISACHC Class, VHS, 2D LA : Ao, FS%, mitral E wave velocity, mitral E : A and E : E', NT-proBNP) and plasma markers of endothelial function (ie, CRP, NOx, vWF, l-arginine, ADMA, SDMA) were tested using Spearman's rank-correlation coefficient. To account for the differences in age and body weight between groups and further characterize the effect of significant disease-related clinical variables and plasma markers (ISACHC, 2D LA : Ao, mitral E wave velocity, mitral E : A, NT-proBNP, CRP and NOx) on the dependent variables of systolic RH-VTI and % change in VTI during RH, multiple linear regression analysis was performed with age and body weight included as covariates. Multiple linear regression was also performed to further characterize the relationship of the independent disease-related variables on the dependent study variables of plasma CRP and NOx concentrations with age and body weight as covariates. A P-value < .05 was considered statistically significant. Statistical analysis was performed using commercially available software.9


Intraobserver Repeatability of Baseline and Hyperemic Brachial Artery VTI

Nine healthy dogs were included in the intraobserver repeatability cohort. Breeds represented in this cohort included mixed breed (n = 5), Doberman (n = 1), Rottweiler (n = 1) and Golden Retriever (n = 2). The mean age of these dogs was 7.6 ± 2.7 years, and the mean body weight was 30.8 ± 8.0 kg. At baseline, the intraexamination CV for systolic VTI was 5.9%, whereas the CV for diastolic VTI was 10.6%. The interexamination CV at baseline was 6.9% for systolic VTI and 17.4% for diastolic VTI. During RH, the CV was 19.4% for the change in systolic VTI from baseline and 24.3% for the change in diastolic VTI.

Study Population

A total of 20 dogs with CHF and 23 healthy controls were screened for the study. Of these, 5 control dogs were excluded after the echocardiogram on the basis of occult MR, and 1 control dog was excluded due to inability to lie still during the echocardiogram. Thus, 37 dogs were enrolled in the study, including 17 healthy control dogs and 20 dogs with CHF. Breeds represented in the control group included mixed breed (n = 7), Doberman Pinscher (n = 4), Golden Retriever (n = 2), and 1 each of the following: Australian Shepherd, Boxer, Rottweiler, and Standard Poodle. Breeds in the CHF group due to MMVD included Cavalier King Charles Spaniel (CKCS; n = 3), Shetland Sheepdog (n = 2), Beagle (n = 2), and one each of the following: Australian Cattle Dog, Chihuahua, Dalmatian, Greyhound, Pointer, Shih Tzu, Standard Poodle, and Weimeraner. Breeds in the CHF group due to DCM included 3 Doberman Pinschers, 1 Flat Coated Retriever and 1 Pit Bull Terrier. ISACHC classifications for the dogs with CHF were II (n = 15), IIIa (n = 2), and IIIb (n = 3). Clinical characteristics and echocardiographic data for the dogs are presented in Table 1. Heart rate was not significantly different between groups (= .51) but systolic BP was significantly lower (= .002) and 2D LA : Ao and NT-proBNP were significantly higher (both < .001) in dogs with CHF compared to controls. Dogs with CHF were significantly older (= .001) and weighed less (= .005) than the control dogs. Brachial artery diameters were smaller in dogs with CHF compared to controls in systole (0.24 ± 0.06 cm versus 0.30 ± 0.05 cm; = .008) and diastole (0.22 ± 0.05 cm versus 0.25 ± 0.05 cm; = .006). Cardiac medications in dogs with CHF were variable and included: furosemide (n = 20), pimobendan (n = 19), enalapril (n = 16), lisinopril (n = 1), amlodipine (n = 1), digoxin (n = 1), mexiletine (n = 2), amiodarone (n = 1), carvedilol (n = 2), metoprolol (n = 1), theophylline (n = 1), lidocaine (n = 1), and dobutamine (n = 1). Lidocaine and dobutamine infusions were stopped for at least 30 minutes prior to the ultrasound examination.

Table 1. Baseline characteristics in healthy control dogs (n = 17) and dogs with congestive heart failure (CHF; n = 20)
ParameterUnitsHealthy ControlsCHFP
  1. Data are presented as mean and standard deviation or median and range as appropriate. P values in bold are significant. BP, blood pressure; VHS, vertebral heart score; 2D LA : Ao, 2-dimensional left atrial to aortic ratio; NT-proBNP, N-terminal pro B-type natriuretic peptide; BA, brachial artery.

Ageyears5.8 ± 2.99.4 ± 3.0.001
SexMale9 (all castrated)12 (11 castrated) 
 Female8 (6 spayed)8 (7 spayed).41
Weight kg30.0 ± 6.920.7 ± 11.5.005
Heart ratebpm104 ± 22109 ± 25.51
Systolic BPmm Hg 136.3 ± 20.2110.6 ± 24.2.002
VHSNAND12.1 ± 0.9NA
2D LA : AoNA1.54 ± 0.172.33 ± 0.40<.001
Mitral E velocitym/sec0.58 ± 0.111.03 ± 0.30<.001
NT-proBNPng/mL 462 (38–1210)2837 (407–≥3000)<.001
Serum Creatininemg/dL 1.0 (0.7–1.4) 1.0 (0.5–4.7).81

Plasma Markers

Plasma CRP concentration was measured in 19/20 dogs with CHF and 16/17 healthy dogs. Other plasma biomarkers were measured in all dogs (n = 37). In the CHF group the CRP concentration was above the range of the assay in 1 dog and below the range of the assay in 2 dogs; these values were not included in statistical analysis. Dogs with CHF had higher plasma CRP (= .02) and NOx (= .005) concentrations compared to the healthy controls (Table 2). Age and body weight were not significantly correlated to plasma CRP; when CRP was assessed as the dependent variable in multiple linear regression analysis including age and weight as covariates, increased ISACHC class, NT-proBNP, 2D LA : Ao, MR severity, mitral E wave velocity, and decreased BP were found to be predictive of plasma CRP concentration. Age and body weight were not predictive of NOx. Although a positive correlation was noted between plasma NOx and NT-proBNP concentrations (= 0.40; = .02), NT-proBNP was not predictive of NOx in regression analysis. No differences were detected in concentrations of vWF (= .09), l-Arg (= .17), ADMA (= .77), the ratio of l-Arg to ADMA (= .29), SDMA (= .68), or the ratio of ADMA to SDMA (= .89) in dogs with CHF compared to controls. Age, body weight, Cavalier King Charles Spaniel or Doberman breed, and serum creatinine concentration were not significantly related to ADMA concentrations. Modest negative correlations were seen between systolic RH-VTI and plasma concentrations of NT-proBNP (= −0.35; = .03), CRP (= −0.43; = .01), and NOx (= −0.34; = .04) (Table 4). Using multiple linear regression, NT-proBNP (R2 = 0.20; = .045), and CRP (R2 = 0.31; = .007) were found to be predictive of systolic RH-VTI, whereas age, weight, and NOx were not significantly associated with systolic RH-VTI.

Table 2. Circulating markers of vascular function in healthy controls and dogs with congestive heart failure
ParameterUnitHealthy ControlsCHFP
  1. Data are presented as mean and standard deviation or median and range as appropriate. Bolded P values are significant. CRP, C-reactive protein; vWF, von Willebrand factor antigen activity; NOx, nitrate and nitrite; l-Arg, l-arginine; ADMA, asymmetric dimethylarginine; SDMA, symmetric dimethylarginine.

CRPµg/mL0.88 (0–19.87)4.22 (0.43–45.44).02
vWF%74.5 ± 24.559.4 ± 27.8.09
NOxµM5.30 (1.33–41.98)11.60 (2.82–103.44).005
l-ArgµM95.0 ± 25.083.1 ± 25.6.17
ADMAµM1.62 ± 0.401.68 ± 0.61.77
SDMAµM0.39 ± 0.050.40 ± 0.07.68
l-Arg/ADMANA59.9 ± 14.153.7 ± 19.6.29
ADMA/SDMANA4.15 ± 0.914.22 ± 1.87.89

Brachial Artery Velocities

At baseline, no differences were seen in pre occlusion systolic VTI (= .14) in CHF dogs compared to healthy controls (Table 3) but diastolic VTI was significantly lower in dogs with CHF (= .008). A negative correlation was found between age and baseline systolic VTI (= −0.42; = .01) and diastolic VTI (= −0.34; = .04), whereas positive correlations were found between baseline systolic and diastolic VTI and body weight (= 0.39; = .02 and = 0.35; = .03, respectively), as well as systolic arterial BP (= .44; = .007 and = 0.40; = .02, respectively). The peak RH response occurred at 10 seconds post occlusion in both healthy dogs and dogs with CHF. During peak RH, dogs with CHF had a lower systolic RH-VTI (= .002; Table 3) as well as a reduced absolute change (= .001; Fig 2) and % change (= .04) in systolic RH-VTI relative to baseline systolic VTI. The diastolic RH-VTI at 10 seconds was also numerically lower in dogs with CHF (21.01 ± 7.79 cm) compared to the healthy dogs (27.26 ± 11.26 cm), but this difference did not reach statistical significance (= .055).

Figure 2.

Percent change in systolic reactive hyperemic VTI relative to baseline systolic VTI in normal dogs (n = 17) and dogs with congestive heart failure (CHF; n = 20) following release of a 3-minute period of cuff occlusion. The % increase in systolic RH-VTI relative to baseline systolic VTI is depicted at specified time intervals during the 60-second period following cuff release. The peak RH response occurred at 10 seconds in both groups; however, dogs with CHF had a less pronounced peak response compared to the normal dogs. VTI, velocity time integral, % Delta VTI, change in systolic VTI relative to baseline; CHF, congestive heart failure; *< .05.

Table 3. Brachial artery velocity time integrals at baseline and during peak reactive hyperemia
Doppler ParameterUnitHealthy Controls (n = 17)CHF (n = 20)P
  1. Data are presented as mean and standard deviation. P-values in bold are significant. RH, reactive hyperemia; VTI, velocity time integral; RH-VTI-VTI during peak RH at 10 seconds post cuff occlusion.

Systolic Indices
Systolic VTI basecm8.7 ± 2.17.5 ± 2.6.14
Systolic RH-VTIcm19.7 ± 3.914.4 ± 5.2.002
Absolute change systolic VTIcm10.9 ± 3.26.9 ± 3.6.001
% change systolic VTI%134.8 ± 58.796.5 ± 51.7.04
Diastolic Indices
Diastolic VTI basecm4.4 ± 2.42.6 ± 1.5.008
Diastolic RH-VTIcm27.3 ± 11.321.0 ± 7.8.06
Absolute change diastolic VTIcm22.8 ± 11.418.4 ± 7.6.17
% change diastolic VTI%722.8 ± 786.21024.5 ± 877.6.28

There were no significant associations detected between absolute or % change in post occlusion systolic VTI during peak reactive hyperemia (RH-VTI) and either age, weight, or BP. There was a negative association between heart rate and diastolic RH-VTI (= −0.38; = .03), but systolic RH-VTI was not significantly correlated with heart rate (= −0.08; = .64). There were negative correlations between the change in systolic RH-VTI from baseline and ISACHC Class, 2-D LA : Ao, and mitral E wave velocity. The % change in systolic RH-VTI relative to baseline VTI was inversely correlated with ISACHC Class and mitral E velocity, but not 2-D LA : Ao. (Table 4). ISACHC Class (R2 = 0.24; = .004), mitral E wave velocity (R2 = 0.25; = .003), and NT-proBNP (R2 = 0.15; = .03), but not 2D LA : Ao (= .15) were predictive of the % increase in systolic VTI during peak RH using multiple linear regression.

Table 4. Correlation coefficients (r) of brachial artery systolic velocity time integral during peak reactive hyperemia (RH-VTI); change in RH-VTI from baseline (Delta VTI); and % change in RH-VTI from baseline (% Delta VTI) with selected clinical and echocardiographic parameters and plasma biomarkers (n = 37)
 RH-VTIDelta VTI% Delta VTI
  1. P values in bold are significant. RH-VTI, velocity time integral during peak reactive hyperemia; HR, heart rate; BP, blood pressure; ISACHC Class, International Small Animal Cardiac Health Council Classification; 2-D LA : Ao, 2-dimensional echocardiographic ratio of left atrial to aortic size; FS%, fractional shortening %; Mitral E : A, ratio of mitral inflow E wave to A wave velocity; Mitral E : E', ratio of mitral E wave velocity to tissue Doppler derived E'; NT-proBNP, N-terminal pro-brain natriuretic peptide; CRP, C-reactive protein; NOx, nitric oxide metabolites; vWF, von Willebrand factor antigen; l-Arg, l-arginine; ADMA, asymmetric dimethylarginine.

Heart rate−0.08.64−0.08.66−0.03.86
ISACHC Class−0.48.003−0.47.003−0.38.02
2D LA : Ao−0.41.02−0.34.04−0.20.24
Mitral E velocity−0.61<.001−0.60<.001−0.39.02
Mitral E : A−0.50.002−0.54.001−0.42.01
Mitral E : E'−0.34.05−0.20.25−0.05.77

When comparing post occlusion systolic VTI to baseline VTI, there was a significant difference in VTI over time in all dogs (< .001) and a significant difference in the magnitude of the change between CHF and control dogs (= .04). The systolic VTI of dogs with CHF took longer to return to baseline values compared to the control dogs. Although the controls were no longer different from baseline at 45 and 60 seconds post occlusion, the CHF dogs were no longer different from baseline only at 60 seconds.

Effect of DCM versus MMVD

No differences were noted in absolute systolic or diastolic VTI measurements at baseline or during RH in dogs with MMVD versus those with DCM; however, the percent increase in systolic VTI during peak RH relative to baseline was significantly lower in the DCM dogs (60.5 ± 35.2% versus 108.5 ± 55.6%; = .04). Dogs with DCM also had higher CRP concentrations (n = 4; median = 11.53 μg/mL [0.43–45.44 μg/mL]) than dogs with MMVD (n = 12; median = 2.91 μg/mL [0.52–8.21 μg/mL]; = .048). Although NOx concentration was numerically lower in dogs with DCM (median = 6.88 μM [4.97–11.31 μM] versus 14.89 μM [2.82–103.44 μM]) this failed to reach statistical significance (= .06). Concentrations of other plasma biomarkers did not differ significantly between dogs with DCM and MMVD dogs.


The results of this study suggest that the brachial artery reactive hyperemic response, a marker of microvascular endothelial function, is impaired in this population of dogs with naturally occurring CHF. Moreover, circulating CRP and NOx concentrations were elevated in dogs with CHF relative to healthy control dogs, and an inverse relationship was noted between plasma concentrations of these markers and systolic hyperemic VTI (RH-VTI). A modest association was also noted between declining RH-VTI and increases in plasma NT-proBNP, ISACHC class, LA size, and mitral E wave velocity, markers of more advanced cardiac disease. Thus, RH-VTI may prove valuable as a novel, relatively simple marker of endothelial function in the dog.

The normal endothelium is a metabolically active tissue responsible for production of myriad paracrine and autocrine factors that dictate vascular tone and maintain a vasodilatory, anti-inflammatory and antiproliferative homeostatic state. Systemic vasomotor dysfunction accompanying CHF is multifactorial, resulting from a combination of reduced shear stress, reduced NO bioavailability, increased vascular oxidative stress from activation of the renin-angiotensin system, and increased levels of proinflammatory mediators.[7, 45-47] Impaired function of the vascular endothelium and the resultant imbalance between mediators of vasoconstriction and vasodilation in CHF can result in inappropriate vasoconstriction, diminished tissue perfusion, reduced exercise tolerance, and progression of cardiac disease.[7, 45] Identification of endothelial dysfunction might thus portend a poorer prognosis[8, 22, 24] and has become an important adjunctive therapeutic target in human patients with CHF.[4, 7, 31, 47]

The most commonly reported noninvasive marker of endothelial function in people is flow-mediated vasodilation (FMD). Although recent studies in dogs have shown the feasibility of FMD measurement in dogs,[36, 37] the technique is technically demanding and limited by high variability, body size dependence, and the need for high-resolution ultrasound equipment.[37] In this study, measurement of brachial artery VTI was well tolerated and possible in all but 1 dog that was excluded from the study before the brachial artery ultrasound for its inability to lie still during the echocardiogram. Furthermore, the intra observer variability of RH-VTI was found to be lower than prior studies of FMD in dogs.[36, 37] It should be noted that measurement of FMD and RH-VTI evaluate different aspects of endothelial function. Reactive hyperemia serves as the microvascular stimulus for FMD,[41, 48] which is more reflective of conduit artery vasodilatory capacity in response to increased hyperemic shear stress. Recent studies in people have established the value of brachial artery shear stress and VTI during RH as novel, noninvasive markers of microvascular endothelial function[25, 26, 41, 42] that may outperform or provide additive prognostic value to FMD.[26, 40, 41] In the presence of cardiovascular risk factors, impaired RH-VTI and shear stress can be detected long before an abnormal FMD response is noted.[25, 26] Factors that have been shown to impact the RH response in outwardly healthy individuals include age, systolic blood pressure, fasting blood glucose, low-density lipoprotein and C-reactive protein levels, and body mass index.[25] In a study of Framingham Heart Study Offspring participants, prevalent cardiovascular disease, sex, treatment with antihypertensive and hormone replacement therapies, and baseline brachial artery flow were also associated with the magnitude of the brachial artery flow response during RH.[24] Brachial artery flow variables were strongly correlated to brachial artery FMD in this study, raising the possibility that the impaired FMD response associated with cardiovascular risk factors may be partially explained by microvascular dysfunction and reduced hyperemic stimulus for dilation, rather than an intrinsic abnormality of conduit artery endothelial function.[24] Although NO bioavailability plays a role in determining microvascular tone,[49] the RH response is a complex measure of microvascular function that is only partly mediated by endothelium-derived NO release.[26, 49-51] Additional contributors to RH include adenosine, endothelium-derived hyperpolarizing factor, ATP-sensitive potassium channels, and prostaglandins, among others.[41, 51]

Although the optimal way to measure brachial artery VTI has yet to be determined,[26, 48] we chose to measure systolic and diastolic waveforms separately and calculated the average VTI over 3 consecutive cardiac cycles to minimize the effect of respiratory sinus arrhythmia on waveform variability. Brachial artery VTI was found to correlate weakly with age and body weight at baseline; however, no associations among age, weight, and VTI were noted during RH. If additional studies confirm the lack of association of hyperemic VTI and body weight in dogs, this could have significant value as a body size-independent marker of endothelial function. Systolic VTIs were more repeatable than were diastolic VTIs at baseline and during RH; this finding likely results in part or whole from the preferential effect that changes in heart rate (HR) have on the diastolic interval compared to changes seen in the systolic interval. At faster HR the shorter diastolic interval results in a smaller diastolic VTI, with the converse seen at slower HR. As expected, a negative correlation was noted between HR and diastolic VTI, whereas no significant associations were found between systolic VTIs and HR. Thus, changing R-R intervals associated with sinus arrhythmia had a marked impact on the beat-to-beat variability of diastolic brachial artery VTI, whereas the systolic VTI remained relatively unchanged and was generally more repeatable.

A modest but significant negative correlation was seen between declining systolic RH-VTI and increasing plasma concentration of CRP. Moreover, CRP concentration was found to be predictive of RH-VTI after controlling for the potential confounders of age and body weight using multiple regression, suggesting a possible link between systemic inflammation and impaired microvascular endothelial function in this population of dogs with CHF, as has been demonstrated in people with cardiovascular disease.[25, 52, 53] Although the mechanisms belying endothelial dysfunction in heart failure have not been fully elucidated, systemic inflammation and increased oxidative stress are known to play a role in both the pathogenesis of endothelial dysfunction and the progression of CHF.[28-30, 46, 54, 55] C-reactive protein serves as both a marker and mediator of systemic inflammation[27, 28, 30] and endothelial dysfunction,[27, 28, 30, 54, 56] and in several studies CRP values have been shown to increase incrementally with CHF severity and stage of decompensation.[29, 57]

Previous studies evaluating CRP levels in dogs with MMVD have yielded somewhat divergent results.[58-60] In a prior study from this institution, Rush et al demonstrated increased plasma CRP in dogs with MMVD compared to healthy control dogs.[58] Tarnow et al later reported a markedly increased CRP level of >10 mg/L in only 1 of 35 dogs with CHF secondary to MMVD[59]; however, the median values and statistical comparison of CRP concentration in dogs with CHF compared to their healthy controls was not reported. A recent study by Ljungvall et al[60] found that circulating CRP concentration was not associated with severity of MMVD; however, unlike the study reported herein, that study population was comprised predominantly of CKCS and CRP levels were found to be higher in non-CKCS dogs. Ljungvall et al also reported significant associations between plasma CRP and cardiac TnI concentrations, breed, and systolic BP.[60] An association between CRP level and systolic BP was also noted in the study reported herein; this association was also noted in the multivariate analysis, as were associations between plasma CRP and NT-proBNP, mitral E wave velocity, severity of MR, and LA size. No associations were found between CRP and age, body weight or breed. The results of this study suggest that CRP concentration may increase with increasing severity of CHF in dogs, as has been seen in people[57] and additional research is warranted to evaluate whether or not systemic inflammation plays a role in the pathogenesis and progression of spontaneous cardiac disease in dogs.

Nitric oxide is an endogenous free radical gas produced by endothelial cells; it plays vital roles in the regulation of endothelial cell function, vascular tone, and myocardial contractility and remodeling. In this study, plasma concentrations of the metabolites of NO (nitrate and nitrite; NOx) were found to be higher in dogs with CHF, with no significant difference in NOx noted between dogs with DCM and dogs with MMVD. The finding of increased NOx in dogs with CHF is consistent with the results of prior studies describing increased NO production in people with CHF[32, 61]; although NO concentrations may vary based on the severity and stage of heart failure.[62] Several studies have reported that the inducible form of nitric oxide synthase (iNOS) is elevated in CHF[62-64] and that inflammatory mediators play a major role in inducing increased iNOS activity, while decreasing the constitutive production of NOS within the endothelium.[63, 65-67] This phenomenon may explain the seemingly paradoxical impairment of microvascular endothelial function in the face of increased total NO production seen in this study. Plasma NOx have previously been assessed in dogs with spontaneous cardiac disease, with disparate results, and plasma NOx may also be influenced by factors such as diet, renal function, and recent exercise.[15, 16, 19, 68] A possible role of NO in the pathogenesis of MMVD has been suggested by the results of prior studies reporting increased NADPH activity in myxomatous canine mitral valves[68]; and increased NOS activity and NO production in diseased regions of myxomatous porcine mitral valves.[69]

Asymmetric dimethylarginine (ADMA) is an endogenous inhibitor of NO synthase. Elevated levels of ADMA may accompany increased levels of NOx and inflammatory cytokines in human heart failure patients,[32] and have been associated with adverse cardiovascular outcomes and endothelial dysfunction.[33] The results of this study failed to show any difference in ADMA concentrations in dogs with CHF; likewise, no relationship was noted between l-arginine, ADMA, SDMA or vWF and any measure of CHF severity or brachial artery flow. Pedersen et al also failed to demonstrate any effect of worsening disease severity on dimethylarginine concentrations in dogs with asymptomatic mitral regurgitation.[18] In that study, body weight and creatinine were found to influence levels of ADMA and SDMA, respectively. Moesgaard et al similarly demonstrated an effect of body weight, as well as variable effects of breed, sex, exercise, and “white coat effect” on circulating levels of dimethylarginines, NOx, and vWF.[19] More recently, Moesgaard et al found a reduced l-arg/ADMA ratio in dogs with increasingly severe MMVD; however, this seemed to relate mostly to the effect of increasing age.[20] The study reported herein also identified a sex effect on ADMA but failed to find any effect of age, body weight or creatinine on dimethylarginines in this diverse population of dogs. Moreover, the only breed effect on circulating markers that was noted was a lower vWF% in Dobermans, which is likely attributed to the high incidence of carriers of von Willebrand disease in the breed.

Limitations of this study include the potential for underpowering due to the relatively small sample size, and the fact that the groups were not matched with respect to age, breed, and body weight. Due to the prevalence and natural history of MMVD in dogs, it is difficult to find older, small breed dogs with no evidence of valvular or systemic disease. The CHF group was comprised of a mixed population of dogs with DCM and MMVD, two diseases which may have different associations with changes in vascular function. Multiple-linear regression taking these differences in age, CKCS or Doberman breed, body weight, and underlying cardiac disease type into account failed to show any effect of these variables on RH velocities or plasma biomarkers. However, future studies should investigate the role that these factors might play in altering markers of endothelial function in the dog. Other practical limitations of the study were the inability to blind the study observer (SMC) to the group designation of the dogs, and the inability to standardize cardiac medications received by dogs in the CHF group. Medical interventions such as pimobendan, statins, and ACE-inhibitors may exert effects on endothelial function via inhibition of inflammatory mediators and iNOS expression, or through direct effects on vasomotor tone.[39, 70, 71] Additional studies are indicated to evaluate the effects of these drugs on markers of systemic inflammation and endothelial function in dogs with naturally occurring cardiac disease.

In conclusion, the results of this study provide evidence of systemic inflammation and microvascular endothelial dysfunction in this heterogeneous population of dogs with CHF. Both increased CRP concentration and impaired RH response were associated with markers of increasing heart failure severity and RH-VTI may prove valuable as a novel, relatively simple marker of endothelial function in the dog. Additional research is warranted to further characterize brachial artery flow profiles and to examine the relationship among inflammatory mediators, impaired endothelial function, and progression of cardiac disease in the dog.


The authors acknowledge the invaluable assistance of Dawn Meola, Dr Daniel Hall, Barbara Brewer, and Nicole Elie in data collection for this study, and Lori Lyn Price and Jennifer Bassett Midle for statistical support. We also thank Katie Cyr and IDEXX Laboratories for support of shipping and analysis of NT-proBNP samples.

The study was supported in part by the Barkley Fund.


  1. 1

    GE Vivid 7 echocardiograph, GE-Medical, Milwaukee, WI

  2. 2

    Cardiopet proBNP test, IDEXX Laboratories, Inc, Westbrook, ME

  3. 3

    TriDelta Phase canine CRP assay, Tri-Delta Diagnostic Inc, Boonton Township, NJ

  4. 4

    Cayman medical nitrate/nitrite colorimetric assay kit, Cayman Chemical, Ann Arbor, MI

  5. 5

    ACL ELITE Instrumentation Laboratory, Beckman Coulter, Brea, CA

  6. 6

    Oxonon Bioanalysis Laboratory, Oakland, CA

  7. 7

    Bond Elut, Agilent Technologies, Santa Clara, CA

  8. 8

    Agilent 1100 series, Agilent Technologies

  9. 9

    SPSS version 17.0; SPSS, Inc, Chicago, IL