Surgical Model

Briefly, 8- to 9-week-old male CD-1 mice (∼35 g body weight) were anesthetized with isoflurane/oxygen (2%:100%), intubated, and ventilated (Harvard Apparatus, Holliston, MA) at 150 breaths/min at a tidal volume of about 300 μL. Animals were randomly assigned to receive either MI or TAC. In the MI group, a thoracotomy was performed on the left side of each animal to expose the left anterior descending coronary artery, which was ligated with 7-0 Surgipro II polypropylene suture (Covidien, Dublin, Ireland) directly inferior to the left atrium. MI was confirmed by myocardial blanching. In the TAC group, the cartilaginous connection between the sternum and ribs 2 and 3 was separated to expose the upper mediastinum. The transverse aorta was isolated and subsequently set with 7-0 Sofsilk thread (Covidien) to the diameter of a blunted 26-gauge needle. For both surgical models, the ribs and skin were closed using 5-0 Sofsilk suture. Mice were taken off anesthetic and allowed to recover on 100% oxygen. Housing and experimental procedures were approved by the animal care committee at the University of Guelph and were in accordance with the guidelines for laboratory animal welfare set forth by the University of Guelph.

Echocardiographic Analysis

Images were obtained between 9 am and 3 pm using the Vevo2100 system (VisualSonics, Toronto, ON, Canada). All TAC mice were evaluated 18 weeks after surgery. All MI mice were evaluated 4 to 8 weeks after surgery. Mice were anesthetized with isoflurane/oxygen (1%–1.5%:100%) just below the level of a pedal reflex to ensure comparable anesthesia and maintained at 37°C throughout procedures. All LV images were acquired within 10 min of induction with the MS550D ultrasound transducer set to 40 MHz. All measurements were done using the Cardiac Package (VisualSonics). M-mode (1D) images were analyzed using the LV-trace function, and long-axis B-mode (2D) images were analyzed using the 2D area calculation. M-mode images were obtained from either long- or short-axis views of the heart at the midpapillary level. Two-dimensional images were acquired from the long-axis view of the LV, where probe placement was such that the LV chamber was visualized from the apex to the outflow tract.

PF was measured from a parasternal short-axis view of the pulmonary trunk just distal of the pulmonary semilunar valves with probe placement arranged to obtain an image as depicted in Figure 2 D, similar to the long-axis view of the LV but with the probe moved 1 to 2 mm superiorly. Aortic flow was measured from a parasternal long-axis view of the ascending aorta ( Supplemental Figure 1 , available at www.onlinejase.com ). Pulmonary and aortic flows were calculated as the product of the vessel area (2π r ² ) and the velocity-time integral (VTI) of the pulsed-wave Doppler of flow at that level ( Figure 2 C, Supplemental Figure 1 D, available at www.onlinejase.com ). Vessel diameter was measured just distal to the valve leaflets where VTI was obtained. Doppler gain was consistently set between 32 and 35 dB at a frequency of 32 MHz to limit color Doppler variability. Beam steering, a novel function of the Vevo2100 system, was used to ensure that the 2D angle between the direction of the pulmonary trunk visualized in vivo and the sound beam(s) was <45°.

Visualization of PF through the pulmonary trunk of a healthy mouse. (A) Pulmonary trunk diameter (white line) at the point at which flow was measured. (B) Color Doppler visualization through the pulmonary trunk. (C) Representative pulsed-wave tracing of the PF VTI obtained over four consecutive heartbeats. (D) An unlabeled (top) and a labeled (bottom) image of the pulmonary trunk and surrounding structures of importance. The red arrows represent the direction of blood flow.

All echocardiographic images (M-mode, 2D, and flows) were obtained in triplicate. All volumes were calculated as the mean of three images sets; the mean of each set was five consecutive heartbeats. All images were reanalyzed in a blinded manner to identify any subjectivity, which was defined as any mean exceeding 10% between repeated measures. In all cases of discrepancy, image quality was poor in part or in entirety, and values were excluded from the analysis (see Table 1 ). All images were acquired within 10 min of induction, and calculations were done after acquisition.

Flow Probe

Similar to surgical intervention, mice were intubated and ventilated (Harvard Apparatus) at 150 breaths/min at a tidal volume of about 300 μL with an isoflurane/oxygen mix (2%:100%). Mice were maintained at 37°C with a rectal probe and heating lamp throughout data collection. Ribs were separated at the cartilaginous connections of ribs 2, 3, and 4 to expose the upper mediastinum. A piece of 7-0 silk thread was looped around the ascending aorta used and to delicately place the aorta inside the 1.5-mm-diameter flow probe (Transonic, Ithaca, NY). This flow probe was connected to the TS420 Perivascular Flow Module, and CO calculations were presented as a digitized readout directly on the machine. Probe placement was adjusted to obtain peak flow, and CO was recorded when peak flow values were maintained for at about 3 consecutive seconds (approximately 30 heartbeats). In contrast to previous studies, flow probe CO calculations were made 24 hours after echocardiographic CO calculations, as flow probe placement directly impeded concurrent echocardiographic visualization of the pulmonary trunk.

CO Calculations

The following formulas were used to calculate cardiac volumes from echocardiographic images (in all cases, CO was the product of stroke volume and heart rate):

• 
  

  M-mode (1D): volume = [7.0/(2.4 + D )] D ³ , where D is the internal “diameter” at the midpapillary region of long-axis B-mode images in either systole or diastole. Stroke volume was calculated as the difference between diastolic and systolic volumes.

  
  
• 
  

  2D: volume = [(8 × A ² )/(3π L )], where A is longitudinal surface area, and L is chamber length. ² Stroke volume was calculated as the difference between diastolic and systolic volumes.

  
  
• 
  

  PF/aortic flow: volume = (VTI)(2π r ² ). Flow-derived stroke volumes were calculated as a product of the VTI and the vessel cross-sectional area (pulmonary trunk or ascending aorta) at the point of peak flow, where r is vessel radius.

Variability Assessments

Variability of echocardiography-derived CO values (M-mode, 2D, and PF) was assessed within users (intrauser variability) and between users (interuser variability) and is summarized in Table 2 . A total of nine images, three with each method, were collected per animal ( n = 11 for healthy mice, n = 9 for MI mice). All images were acquired rapidly over a period of <10 min to help ensure similar heart rates and body temperatures. CO from each image was averaged over five heartbeats (between breathes). Intrauser variability was calculated from the SD within each set of three images (M-mode, 2D, or PF) and expressed as a percentage of the average CO for that method in that particular mouse. The intraimage variability per user was averaged per mouse ( n = 3) and the reported intrauser percentage is the average variability between the two users. Interuser variability was calculated as the average absolute percentage difference between user 1’s and user 2’s calculations from the same images ( n = 30–36 per method). Intra- and interuser reliabilities were assessed using intra- and interclass correlation coefficients respectively, and values are presented in Table 3 . Correlation coefficients were calculated in Excel (Microsoft, Redmond, WA) using the “Anova: Two Factor without Replication” data analysis table (i.e., a repeated-measures analysis) and the formula {[MSRow − MSE]/[MSRow + (dfCol)(MSE) + (dfCol + 1)(MSCol − MSE)/(dfRow + 1)]} (adapted from Weir ).

Histologic Approximation of Infarct Size

A total of 10 mice were blindly selected for infarct size estimation following data collection. Briefly, hearts were perfusion-fixed with 1 × PBS, 50 mmol potassium chloride and 10% neutral buffered formalin (VWR International, Mississauga, ON, Canada) through the right common carotid artery. Hearts were processed overnight and embedded in paraffin. Cross-sectional slices (5 μm) were obtained from the approximated apical, midpapillary, and basal regions of the heart. Sections were stained with Gomori’s One Step Trichrome (Poly Scientific, Bay Shore, NY). Images were acquired using an Olympus FSX 100 light microscope and analyzed using Cell Sense software (Olympus, Tokyo, Japan). Infarct size was determined as a percentage of total LV circumference as averaged from all three sections. Average infarct size using this method was 55.2 ± 9.7% of the total LV circumference.

Bland-Altman Analysis

Bland-Altman graphs were created in Prism 6.0 (GraphPad Software, La Jolla, CA) to evaluate the agreement between reference values (e.g., flow probe) and comparator values (e.g., M-mode, echocardiography). Zero-bias lines are presented as the “ideal agreement” between the two compared methods across a range of differences. Regression lines are plotted to demonstrate the directional biases, that is, whether a bias is more likely to exist above or below a particular inflection point. The difference between the reference and comparator values was plotted against the average of these values to reduce the risk for biasing trends. Mean difference, 95% CI, and range are also presented.

Statistical Analysis

Averages are presented as ±SEM unless otherwise indicated. SD, Pearson correlation, and Bland-Altman (Tukey mean difference) plot analysis with an absolute mean/range difference were determined using Prism 6.0. All other variables were compared using either Student’s t test (for paired values) or one-way analysis of variance for group comparisons; P values < .05 were considered to indicate statistical significance in all cases. Because higher n values increase absolute inter- and intraclass correlation coefficient values, we compared a similar number of CO values for the these calculations ( n range = 32–36). inter- and intraclass correlation coefficient values were then compared against one another to determine relative reliability of each technique.