Materials and Methods

Animals—Experiments were conducted at the RVC (n = 5 dogs) and the WCPN (38). At the RVC, dogs were used to refine the experimental technique; at the WCPN, dogs were used to assess FMD responses in healthy dogs. Refinement of the experimental technique included assessment of intersonographer repeatability and off-line analysis technique but did not include assessment of between- and within-dog repeatability. Dogs used in the experiments at the RVC included 2 Labrador Retrievers, 1 Lurcher, 1 Italian Spinone, and 1 crossbreed. Median body weight of the dogs was 30.1 kg (range, 27.9 to 35.2 kg); median age was 4 years (range, 4 to 9 years). There were 3 neutered females and 2 neutered males. All dogs were considered healthy, had no history of cardiovascular disease, and were owned by members of RVC staff. Informed consent was provided by each staff member prior to his or her dog's inclusion in the study.

At the WCPN, dogs used in the experiments were owned by the WCPN and included 18 Labrador Retrievers, 13 Miniature Schnauzers, and 7 Cocker Spaniels. Median body weight of the dogs was 12.2 kg (range, 6.9 to 31.7 kg); median age was 3 years (range, 1 to 10 years). There were 7 sexually intact females, 15 neutered females, and 16 neutered males. Dogs that weighed > 15 kg (Labradors) were classified as large, and dogs that weighed ≤ 15 kg (Miniature Schnauzers and Cocker Spaniels) were classified as small. All dogs were considered healthy, and no signs of cardiovascular disease were detected via physical examination, ECG evaluation, or echocardiography. All dogs received 1 of 2 diets^a,b for 7 days prior to the experiment. Prior to experiment commencement, food was withheld from the dogs for a minimum of 6 hours and dogs were not exercised for a minimum of 4 hours. All dogs were within 10% of estimated ideal body weight. Experiments took place in a temperature-controlled environment (22° to 24°C). The dogs at the WCPN were used for between-and within-dog repeatability investigations.

All study dogs at the WCPN were habituated to the experimental procedures prior to experiment commencement. Habituation consisted of thirteen 30- to 60-minute structured training sessions that were tailored to meet the individual requirements of each dog. During habituation sessions, dogs were encouraged to lie calmly in right lateral recumbency with minimal restraint for at least 30 minutes, during which time FMD measurement was attempted. Dogs used at the RVC were not formally habituated but were selected because of their calm behavior characteristics. Studies conformed to the ethical standards of the WCPN Ethical Review Committee and the RVC Ethical Review Board.

Image acquisition—Hair over the right brachial artery of each dog was clipped prior to experiment commencement. Dogs were rested in a familiar environment for at least 30 minutes before beginning each experiment, during which time each dog was free to explore its surroundings and become acclimated to room temperature. Room temperature was recorded by use of a digital thermometer.^c

Dogs were positioned in right lateral recumbency, with the right thoracic limb extended. Dogs were encouraged to remain recumbent in a relaxed manner with minimal restraint (Figure 1). An ultrasound unit^d with a 13-MHz linear array transducer was used to acquire 2-D ultrasonographic images of the right brachial artery (from the medial aspect of the limb proximal to the right elbow joint) in each dog. Acoustic coupling gel was used to optimize images.e Images were acquired by 2 sonographers (VLF and IDJ), both of whom had previously performed at least 100 FMD scans.²¹

View Full Size

Photograph to illustrate the positioning of an inflatable cuff and an ultrasonographic probe in a dog undergoing measurement of FMD in the right brachial artery. During image acquisition, study dogs remained fully conscious, apparently comfortable, and relaxed, and minimal restraint was applied. Study dogs had been habituated to the experimental technique prior to experiment commencement.

Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1154

• Download Figure
• Download figure as PowerPoint slide

Longitudinal views of the brachial artery with clearly defined near- and far-wall interfaces were selected for continuous 2-D imaging for off-line automated edge detection and luminal diameter measurement (Figure 2). Gain settings were adjusted for optimal delineation of wall surfaces. Pre- and postprocessing settings were kept constant for all scans. Images were magnified by a resolution box function. Identification of consistent arterial segments was achieved by noting neighboring anatomic structures. Continuous 2-D image recording was triggered from a simultaneous ECG recording, so that only 1 frame was recorded for each cardiac cycle at exactly the same phase (triggered from the R wave of the ECG trace). Pulsed-wave spectral Doppler ultrasonography was used to record blood flow in the artery of interest (3-mm sample volume). Angle correction was used (angle, ≤ 70°).

View Full Size

Representative 2-D longitudinal ultrasonographic image of the right brachial artery of a healthy dog to illustrate calculation of brachial artery luminal diameter during off-line analysis The size of the ROI (dotted white line) is first specified by the operator, and its horizontal boundaries are positioned just outside the near and far arterial walls (N and F, respectively) such that it is centered on the vascular segment under investigation. The edges of the vessel walls within the ROI are then detected automatically by the software (pink lines), and the distance between them is calculated. The same ROI is then superimposed on the remaining frames to assess luminal changes observed during the entire experimental period. This system provided a fully automated method of arterial wall detection and calculation of luminal diameter

Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1154

• Download Figure
• Download figure as PowerPoint slide

For each dog, 2-D longitudinal ultrasonographic images of the artery were obtained during a 30-second period, followed by recordings of ≥ 6 consecutive arterial blood flow velocity waveforms obtained via pulsed-wave Doppler ultrasonography before cuff inflation. An additional 6 consecutive spectral Doppler ultrasonographic blood flow velocity waveforms were recorded no later than 15 seconds following cuff deflation; 2-D longitudinal ultrasonographic images of the artery were then obtained sequentially, as triggered by the R wave of the ECG during the following 3-minute period. All images were recorded digitally on the hard drive of the ultrasound machine.

Induction of reactive hyperemia—The pressure required to abolish an audible signal of flow to the distal portion of the thoracic limb was initially recorded in all study dogs at the WCPN by use of a Doppler ultrasonographic flow meter^f and an inflatable blood pressure cuff that was positioned between the elbow and carpal joints. For each dog, a cuff of appropriate size to adequately occlude flow was selected—width of the cuff used was approximately equivalent to the circumference of the distal portion of the right antebrachium. In all dogs, the pressure necessary to occlude flow was < 150 mm Hg. For induction of reactive hyperemia, the blood pressure cuff was inflated to 200 mm Hg and then deflated after 5 minutes (Figure 3).

View Full Size

Representative spectral Doppler ultrasonographic waveforms obtained from a dog before (A) and immediately following (B) brachial artery occlusion (achieved by use of a blood pressure cuff inflated to 200 mm Hg) of 5 minutes' duration. The sustained forward diastolic flow (DF) illustrated in panel B was observed immediately following cuff deflation in every dog assessed. The increase in flow likely represents the stimulus required to induce FMD.

Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1154

• Download Figure
• Download figure as PowerPoint slide

Calculation of FMD—Data were copied onto magneto-optical disks and analyzed off-line on a personal computer with commercially available software^g by a single observer (IDJ). Prior to analysis, pixel size was calibrated and calibration was repeated for every experiment. An ROI was defined manually for each image frame. By use of automated border detection software, the luminal diameter was computed as the distance between the generated near and far vessel wall edges within the ROI and a mean luminal diameter measurement along a segment of artery was calculated for each frame. The same ROI was applied to every frame during automated calculation of luminal diameter. Each image was then reviewed by the operator. Frames were rejected if automated detection of the luminal borders was not possible or inconsistent. Once analysis was complete, the numeric output was exported to a spreadsheet.^h

For each dog, mean baseline diameter was calculated from the luminal diameter measurements obtained prior to cuff inflation. Automated smoothing softwareⁱ was used to generate a spline curve from all luminal diameter measurements obtained following cuff deflation.³⁸ For each dog, the maximum luminal diameter following cuff deflation was defined as the peak of the spline curve. The FMD for each dog was calculated by use of a formula as follows: FMD (%) = ([Maximum luminal diameter − mean baseline diameter]/mean baseline diameter) × 100. Time to peak luminal diameter was also determined from the generated spline curve.

The FVI (ie, the area under the spectral pulsed-wave Doppler time-velocity curves) was measured in ≥ 6 consecutive arterial spectral Doppler ultrasonographic blood flow velocity envelopes prior to and following cuff inflation by use of the same commercially available softwareg as that used for calculation of FMD values. Calibration was repeated for every experiment, and an ROI was defined manually. The modal velocity and FVI of the spectral Doppler ultrasonographic waveforms within the ROI were calculated automatically Percentage change in FVI associated with cuff inflation was calculated by use of a formula as follows: ΔFVI (%) = (Mean postinflation FVI − mean preinflation FVI)/100.

Intersonographer repeatability of baseline brachial artery diameter measurement—Baseline brachial artery diameters in an ultrasound phantom^j (to determine intersonographer repeatability in vitro) and 5 dogs at the RVC (to determine intersonographer repeatability in vivo) were recorded by 2 sonographers (VLF and IDJ). A solid synthetic ultrasound phantom immersed in water and containing multiple channels of various sizes was used. A single channel similar in size to a brachial artery in a large dog was chosen for study. Continuous images of the ultrasound phantom were acquired for 2 minutes by each sonographer in succession; the order of sonographer assessments was randomized. The acquisitions were separated by a rest period of 2 minutes during which the probe was removed. The first 99 frames, which represented the first 30 seconds of each 2-minute scan, were used for analysis.

Continuous images of the brachial artery in each of the 5 dogs were acquired during a 30-second period by each sonographer in succession; the order of sonographer assessments was randomized. The acquisitions were separated by a rest period of 15 minutes during which the probe was removed. All frames were used for analysis.

Effect of off-line analysis technique on baseline luminal diameter calculation—Four different methods were used by the same observer (IDJ) to analyze three 2-minute baseline recordings of a brachial artery. All recordings were made in 1 dog (a 9-year-old neutered male Italian Spinone) at the RVC. The recordings were made during a 10-minute period, each separated by a 2-minute rest period. Analysis methods included use of a small (< 5 mm) ROI versus a large (> 5 mm) ROI with or without manual editing.

Between- and within-dog repeatability of FMD—For assessment of between-dog repeatability, values of FMD were calculated on at least 1 occasion in all 38 study dogs at the WCPN. Scans were performed by 1 of 2 experienced sonographers (VLF or IDJ). For assessment of within-dog repeatability, values of FMD were recorded in 28 study dogs at the WCPN on 2 occasions (24-hour interval) by 1 of the 2 sonographers (VLF or IDJ). Individual dogs were scanned by the same sonographer on each of the 2 occasions. The 28 dogs included 18 Labrador Retrievers, 7 Miniature Schnauzers, and 3 Cocker Spaniels. Median weight of the dogs was 25.5 kg (range, 6.9 to 31.2 kg); median age was 3 years (range, 11 months to 9 years). There were 7 sexually intact females, 11 neutered females, and 10 neutered males. Every attempt was made to examine the same arterial segment in each dog on each of the 2 occasions. It was assumed that different arterial segments had been recorded when baseline luminal diameter of the brachial artery differed by > 0.1 mm between the 2 scans.

Statistical analysis—Median values and range are reported for baseline luminal diameter, ΔFVI, time to peak luminal diameter, and FMD of the right brachial artery in large and small dogs. Repeatability of FMD measurement was calculated by determining CVs and ICCs. Linear associations between FMD and continuous explanatory variables were described by use of Pearson or Spearman rank correlation coefficients as appropriate. A 1-way ANOVA was used to compare multiple groups of continuous parametric data (diameter measurements acquired via 4 different off-line analysis techniques). A 2-sample t test was used to compare independent groups of normally distributed continuous data (comparisons of baseline luminal diameter measurements in large and small dogs). A Mann-Whitney U test was used to compare independent groups of nonparametric continuous data (comparisons of FMD, ΔFVI, and time to peak diameter in large and small dogs and FMD in dogs < 6 and ≥ 6 years old). Significance was defined as a value of P < 0.05. Factors found to have a significant effect on FMD were included in a standard multiple regression analysis.³⁹ All statistical analyses were carried out by use of computer software.^k

Results

Intersonographer repeatability of baseline brachial artery diameter measurement—Mean luminal diameters of the ultrasound phantom artery as measured by each of the 2 sonographers were compared. For intersonographer repeatability in vitro, the CV was 0.9% and ICC was 0.91. Mean luminal diameter of the brachial artery in the single examined dog as measured by each of the 2 sonographers was also compared. For intersonographer repeatability in vivo, the CV was 6.6% and ICC was 0.89.

Effect of off-line analysis technique on baseline luminal diameter calculation—A significant (P < 0.001) difference in brachial artery luminal diameter was recorded among the 4 analysis methods tested. Baseline luminal diameter measurement was most repeatable (CV, 2.7%) when a large ROI was assessed with manual editing. When manual editing was not performed following definition of a large ROI, the CV was 4.0%. For assessment of a small ROI, the CV was 4.3% with manual editing and 4.1% without manual editing.

Between-dog repeatability of FMD—For all dogs, the pressure required to completely occlude flow in the brachial artery was < 150 mm Hg. Because image acquisition was triggered by the R wave of the ECG, the total number of images acquired varied depending on heart rate. The median percentage of the total number of images acquired for each dog that were rejected during offline analysis was 39% (range, 4% to 49%). An increase in brachial artery luminal diameter was detected in all dogs (Figure 4). The FMD response detected in all dogs was characterized (Table 1). The between-dog CV for FMD among all 38 dogs was 99.7%. Median baseline brachial artery luminal diameter in large and small dogs was 3.27 mm (range, 2.65 to 3.57 mm) and 1.85 mm (range, 1.45 to 2.57 mm), respectively (P < 0.001). A negative correlation (r = −0.66; P < 0.001) between baseline luminal diameter and FMD was identified (Figure 5). There was no significant (P = 0.30) difference in ΔFVI and no significant (P = 0.71) difference in time to peak diameter following cuff deflation between large and small dogs. Median FMD was significantly (P = 0.02) greater in small dogs (7.7%; range, 0% to 19.3%) than it was in large dogs (2.2%; range, −2.2% to 10.6%). Median FMD was also significantly (P = 0.001) greater in dogs < 6 years old (6.2%; range, −2.2% to 19.3%) than it was in dogs ≥ 6 years old (2.9%; range, 0% to 5.6%). Standard multiple regression analysis including factors of age and weight revealed that weight was the only factor that contributed significantly to FMD values.

View Full Size

Mean ± SE percentage change in luminal diameter of the right brachial artery in 38 healthy male and female dogs of various breeds determined at intervals following release (0 seconds) of arterial flow occlusion (achieved by use of a blood pressure cuff that was inflated to 200 mm Hg and then deflated after 5 minutes). Values were calculated as the percentage change from baseline (ie, luminal diameter determined prior to cuff inflation).

Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1154

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Correlation between baseline luminal diameter and FMD of the right brachial artery in 38 healthy male and female dogs of various breeds achieved by use of a blood pressure cuff that was inflated to 200 mm Hg and then deflated after 5 minutes

Citation: American Journal of Veterinary Research 71, 10; 10.2460/ajvr.71.10.1154

• Download Figure
• Download figure as PowerPoint slide

Within-dog repeatability of FMD—The CV for within-dog repeatability derived from all FMD assessments performed 24 hours apart by the same sonographer in 28 dogs was 62.8% (range, 5.7% to 1,414.0%; ICC, 0.292). In 14 of the 28 dogs (9 Labrador Retrievers, 4 Miniature Schnauzers, and 1 Cocker Spaniel), baseline luminal diameters differed by > 0.1 mm between the 2 scans. When these dogs were excluded, CV for within-dog repeatability was 55.3% (range, 11.6% to 188.3%; ICC, 0.552).

Discussion

In the study of this report, brachial artery FMD was measured in a population of healthy, fully conscious and relaxed dogs under consistent environmental conditions. We chose to base our technique on current recommendations for measurement of FMD in humans, rather than on a previously reported method used in dogs.^21,37 Puglia et al³⁷ recorded cross-sectional images of the brachial artery in dogs and measured the vessel manually by use of electronic calipers, from which they derived an ICC of 0.38 for day-to-day diastolic baseline luminal diameter measurements. In our study, comparison of diastolic baseline luminal diameters measured 24 hours apart in 28 dogs yielded an ICC of 0.96. The method used in the present study differed from that used by Puglia et al³⁷ in that longitudinal images of the brachial artery were obtained, only end-diastolic images were stored (triggered by the ECG), and the diameter was measured off-line by use of automated edge detection software within an operator-defined ROI. This allowed images from every cardiac cycle over a continuous time period to be recorded, with an option to reject individual images if the near- and far-field vessel walls were not sufficiently clear.

By application of the technique used in our study, measurement of FMD in healthy conscious dogs of various ages and breeds was feasible, although repeatability was poor even after making every attempt to optimize conditions. Although the within-dog repeatability of FMD measurement (CV 62.8%) would be considered unacceptable for many clinical diagnostic purposes, some studies^26,37 in humans have had equally poor results. This poor repeatability is likely to reflect a combination of technical difficulties combined with environmental and biological influences.^21,40

Several practical obstacles were encountered during measurement of FMD in conscious dogs. Technical problems included difficulties in maintaining constant probe position, obtaining images of a consistent segment of the brachial artery, and obtaining true longitudinal images of the artery. To measure FMD of the brachial artery in humans, it is common practice to clamp the probe in position during image acquisition; however because conscious dogs cannot be relied upon to remain still, it was not possible to do this in our study.²¹ Without the aid of a clamp, different arterial segments may be analyzed before and after cuff occlusion. Consistent probe positioning must also be combined with appropriate probe orientation. True longitudinal images of the vessels in the study dogs were rarely achieved, which adversely affected arterial wall definition.⁴¹ If probe orientation was not consistent throughout each experiment, inaccuracies in automated edge detection were more likely to occur during off-line luminal measurement. To achieve consistent longitudinal vascular bisection, it was essential that the dogs remained immobile and relaxed. Such problems with the imaging technique are minimized as operator experience increases, if anatomic landmarks are recorded precisely during image acquisition, if dogs for which baseline luminal diameter measurements differ by > 0.1 mm between 2 scans are excluded, and if the dogs are compliant during the procedure.

Problems associated with the dogs themselves in the present study included intermittent limb movement and probable fluctuations in sympathetic tone during FMD measurement. Limb movement frequently occurred during image acquisition, even though dogs were relaxed and apparently comfortable in the recumbent position. Identification and recording of consistent arterial segments is challenging even in motionless dogs. Small limb movements also contribute to errors resulting from inconsistent probe orientation. Considering the lack of restraint used in the dogs in our study, compared with human studies, it is reassuring that the within-dog repeatability of FMD measurement was comparable with values reported for humans.²⁶ Within-dog repeatability of FMD measurement is also likely to improve as sonographer experience increases. Despite the fact that both sonographers (VLF and IDJ) in the present investigation had completed > 100 FMD assessments in dogs (following the recommendations for FMD measurement in humans²¹), their skills are likely to have improved over the study period.

Large variation in FMD values is expected given the influence of the many technical and biological factors that may affect its measurement. Individual frames of poor image quality were rejected during off-line analysis and not included in FMD calculations in the present study. Every assessment yielded an adequate number of frames for measurement of FMD; however, more rigorous image exclusion criteria (based on overall image quality) might reduce the large variation in FMD values among dogs.

Limb movement during FMD measurement in dogs is one of the most difficult factors to control. Dogs available to clinical investigators are likely to be less cooperative than the dogs used in our study. The aim of this investigation was to limit potential confounding factors and assess repeatability of FMD measurement under close to ideal conditions. Such optimal conditions would be difficult to replicate in clinical settings. It is likely that future interventional trials will require very large sample sizes to show an effect of any proposed intervention on FMD.⁴²

Problems associated with movement could be minimized by effective identification of images that were adversely affected by motion so that those images could be rejected from analysis. Sedation of the subject during image acquisition would also reduce movement, but the effect of sedation on the FMD response in either humans or dogs is not known. Sympathetic stimulation is an additional factor that affects FMD in humans and is difficult to standardize in dogs.^27,31 An objective method of assessing the magnitude of sympathetic stimulation (eg, heart rate measurement) during FMD measurement would be useful.

Despite the poor reproducibility of the FMD measurement technique in the present study, factors that affected FMD in dogs were identified. Flow-mediated vasodilation was greater in small dogs, compared with large dogs, and a negative correlation between FMD and baseline luminal arterial diameter was detected. In humans, FMD is also affected by resting baseline luminal arterial diameter.^34,43,44 A smaller baseline arterial cross-sectional area results in larger shear stress on the endothelium following cuff inflation. Because nitric oxide release following vascular shear stress is the predominant mechanism responsible for FMD, a greater dilatory response is expected if a larger shear stress is applied to the vascular endothelium.⁴⁴

In humans, FMD has been shown to diminish with age.^34,45 Although FMD was lower in older dogs in our study, age no longer had a significant influence on FMD when weight was also included in the multivariate analysis. The effect of age on FMD in dogs requires further study.

In view of the technical difficulties associated with measurement of FMD, other techniques have been proposed recently as potentially applicable screening tools for evaluating vascular function in humans, including laser Doppler velocimetry and pulse amplitude tonometry.^46,47 These techniques assess microcirculatory vascular mechanisms that diverge from the nitric oxide-dependent mechanisms responsible for conduit artery FMD. It is the vascular microcirculation itself that is responsible for the reactive hyperemic response. The increase in blood flow is triggered by microvascular vasodilation downstream from cuff occlusion. Although this may be a worthwhile area of further investigation in itself, in humans at least, FMD measurement still appears to be the most promising technique in terms of correlation of findings with cardiac disease risk factors.^48,49 Whether the same is true in dogs remains to be determined.

In the present study, determination of ΔFVI following vascular occlusion was more repeatable than FMD measurement, but the former is a measure of a different response, and ΔFVI values are not interchangeable with FMD measurements. Correlations between canine micro-vascular function and brachial artery FMD are unlikely.^50–52 Both FMD and ΔFVI determinations are useful in the comprehensive assessment of canine vascular function, but their potential clinical relevance has not been established. Studies are required to investigate the effect of naturally occurring heart disease on values of FMD and ΔFVI.

Despite the aforementioned technical difficulties, FMD measurement remains a practical option for measuring endothelial function in vivo in dogs. The results of our study suggest that peak FMD response usually occurs within 90 seconds after initiation of reactive hyperemia; therefore, it should not be necessary to record images for 3 minutes after cuff deflation. Provided that suitably calm patients are used, it may be possible to apply the technique described in this report to dogs with heart disease. Although it is not likely that FMD measurement will become a clinical test, it may be possible to identify differences in FMD between groups of dogs, given adequate sample sizes.