Quantification of murine myocardial infarct size using 2-D and 4-D high-frequency ultrasound

Abstract

Ischemic heart disease is the leading cause of death in the United States, Canada, and worldwide. Severe disease is characterized by coronary artery occlusion, loss of blood flow to the myocardium, and necrosis of tissue, with subsequent remodeling of the heart wall, including fibrotic scarring. The current study aims to demonstrate the efficacy of quantitating infarct size via two-dimensional (2-D) echocardiographic akinetic length and four-dimensional (4-D) echocardiographic infarct volume and surface area as in vivo analysis techniques. We further describe and evaluate a new surface area strain analysis technique for estimating myocardial infarction (MI) size after ischemic injury. Experimental MI was induced in mice via left coronary artery ligation. Ejection fraction and infarct size were measured through 2-D and 4-D echocardiography. Infarct size established via histology was compared with ultrasound-based metrics via linear regression analysis. Two-dimensional echocardiographic akinetic length (r = 0.76, P = 0.03), 4-D echocardiographic infarct volume (r = 0.85, P = 0.008), and surface area (r = 0.90, P = 0.002) correlate well with histology. Although both 2-D and 4-D echocardiography were reliable measurement techniques to assess infarct, 4-D analysis is superior in assessing asymmetry of the left ventricle and the infarct. Strain analysis performed on 4-D data also provides additional infarct sizing techniques, which correlate with histology (surface strain: r = 0.94, P < 0.001, transmural thickness: r = 0.76, P = 0.001). Two-dimensional echocardiographic akinetic length, 4-D echocardiography ultrasound, and strain provide effective in vivo methods for measuring fibrotic scarring after MI.

NEW & NOTEWORTHY Our study supports that both 2-D and 4-D echocardiographic analysis techniques are reliable in quantifying infarct size though 4-D ultrasound provides a more holistic image of LV function and structure, especially after myocardial infarction. Furthermore, 4-D strain analysis correctly identifies infarct size and regional LV dysfunction after MI. Therefore, these techniques can improve functional insight into the impact of pharmacological interventions on the pathophysiology of cardiac disease.

INTRODUCTION

Ischemic heart disease is the leading cause of death in the United States, Canada, and worldwide (1–3). This category of disease includes myocardial infarction (MI), which results from occlusion of the coronary arteries, preventing blood flow and myocardial oxygenation, leading to necrosis of a region of the myocardium (4). To prevent further damage, inflammatory signaling pathways become activated, triggering the release of cellular mediators and cytokines. The most significant of these factors with regard to remodeling are fibroblasts, which trigger the release of profibrotic factors (5). This results in necrotic tissue being replaced by a fibrotic scar, leading to global cardiac remodeling, and often cardiac dysfunction (6). Current preclinical strategies to assess cardiac function in murine models use a combination of conventional echocardiography, such as motion mode (M-mode) or two-dimensional (2-D) brightness mode (B-mode), for assessment of left ventricular function, along with histological analysis to measure cardiomyocyte diameter, infarct size, and extent of fibrosis (7, 8). However, these methods present several limitations, such as the lack of in vivo infarct size measurement and acquisition angle-dependent analysis, which must be considered.

Because of its accessibility, echocardiography is a preferred method of cardiac function assessment (7). However, the conventional 2-D analysis completed with echocardiography cannot accurately determine cardiac function after MI, since infarct results in irregular geometry of the left ventricle (LV), making the symmetrical assumptions of 2-D analysis invalid (9, 10). In addition, histological analysis, a reliable ex vivo infarct assessment method, is commonly used in combination with 2-D echocardiography to complement functional data (11). However, histological analysis is an end-point measurement; therefore, long-term follow-up and further investigations cannot be performed (11). This precludes important analysis since cardiac remodeling after an infarction will significantly vary at different time points, and there is a need to complete a longitudinal assessment (11).

Echocardiography has been used to assess infarct size as an in vivo measurement technique. Since echocardiographic imaging in a parasternal long-axis view (PLAX) allows visualization of the LV from base to apex, it can potentially also be used for infarct characterization through the measurement of akinetic length. Similar parameters to the akinetic length have been assessed in previous studies, primarily by using several short-axis views (SAX) (9, 12). In fact, correlations have been established between echocardiographic SAX infarct measurements and histological infarct size values (9). However, analysis of akinetic length to determine infarct size using 2-D echocardiographic PLAX images in B-mode has not yet been assessed (13–15).

Nonetheless, 2-D echocardiography does not provide sufficient information regarding the irregular geometry of the LV after MI (9, 10). Therefore, the recent development of 4-D ultrasound (4DUS) has gained attention owing to its ability to assess infarct size and LV function after MI regardless of irregular geometry (16, 17). This mode of imaging allows visualization of a three-dimensional (3-D) representation of the LV while also considering the dynamic movement of the heart during the cardiac cycle. Although this technique is more complex and time-consuming, it provides accurate tertiary structural information (8). In addition, improvements in imaging hardware and analysis techniques, such as machine learning, are expected to make this imaging modality more accessible in the near future (18). Importantly, 4DUS imaging has made it possible to complete volumetric analysis and to use contouring of the heart walls to create 4-D strain maps (Supplemental Video S1; see https://doi.org/10.6084/m9.figshare.17227115.v1). Therefore, these new analysis techniques can be used to better determine infarct size and LV function, including regional strain estimates.

In this article, we evaluated the reliability and accuracy of a new and fast 2-D akinetic length analysis technique and the reliability of 4-D mode echocardiography to define 3-D infarct size, using a volumetric approach and a surface area estimate. We also aimed to validate a new 4DUS strain analysis software for the assessment of regional function as well as for the estimate of infarct size using transmural thickness and surface area strain. We observed that 2-D akinetic length and both 4-D infarct volume and surface area correctly measure infarct percentage when compared with diagnostic gold standard histology. Also, we found that 4-D strain analysis provides significant insight into both function and infarct size. Taken together, our study suggests that these novel echocardiographic techniques allow for the longitudinal assessment of infarct development and can be used in future preclinical work, ideally for the assessment of therapies for MI.

MATERIALS AND METHODS

Experimental Animals and Myocardial Infarction

Animals were cared for in accordance with the Canadian Guide for the Care and Use of Laboratory animals (CCAC). All experimental procedures were performed based on the approved animal utilization (Protocol Nos. 2909 and 2920) by the University of Ottawa Animal Care and Veterinary Service, or studies were conducted in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals and approved Protocol No. 2002002016 by the Institutional Animal Care and Use Committee at Purdue University. Mice were housed under a 12-h:12-h light/dark cycle and maintained on regular chow (RC; Harlan Teklad, Mississauga, ON at Ottawa or Envigo 2018S; Envigo, Indianapolis, IN at Purdue). Before thoracotomy and left coronary artery ligation, three female and 22 male mice (age = 6–16 wk) were anesthetized with isoflurane (2%–3%) and intubated with an endotracheal tube. Ventilation was maintained at 200 breaths/min with the assistance of a small animal ventilator (SomnoSuite, Kent Scientific, Torrington, CT), or pure oxygen (UOHI). An incision was made in the left thorax at the level of the third or fourth intercostal space. The ribs were subsequently retracted, the pericardium dissected away, and LV exposed. An 8-0 nylon suture was passed under the left coronary artery at approximately midway down the LV and tied off to induce infarct. Following completion of the surgery, the chest was closed with a 6-0 visorb suture (CP Medical, Norcross, GA) and the skin closed with a 6-0 prolene suture (Ethicon, Rariton, NJ) using a simple interrupted pattern (Purdue) or purse-string suture (UOHI). Mice were recovered with supplemental heat and allowed to self-extubate. As necessary, some mice received supplemental oxygen to aid in recovery. Buprenorphine (0.05 mg/kg ip) was used for postoperative analgesia.

Echocardiography: Image Acquisition

Echocardiographic images were acquired using either the Vevo 3100 (n = 20) or Vevo 2100 (n = 5) high-frequency small animal ultrasound system (FUJIFILM VisualSonics, Toronto, Canada), using either the MX400, MX550D (Vevo 3100), or MS550D (Vevo 2100) linear array transducers. Images of the LV were taken at baseline, before thoracotomy and left coronary artery ligation, and at 4-wk postsurgery using PLAX in 2-D B-mode for a subset of mice (n = 20). M-mode images of the heart were also obtained in a SAX view for a subset of mice (n = 12). At 4-wk postsurgery, another subset of mice (n = 23) was used to acquire 4-D mode images for 4-D analysis with Vevo LAB (FUJIFILM VisualSonics, Toronto, Canada), which will be referred to as Vevo LAB 4-D Visualization and/or a custom-built 4-D strain analysis software using a MATLAB graphical user interface, referred to as a 4-D Strain Toolbox (MathWorks, Natick, MA). Finally, in another smaller subset of mice (n = 5), 4-D mode images were taken at baseline, and 1 day, 3 days, 1 wk, 2 wk, and 4 wk post-MI for longitudinal 4-D analysis with the 4-D Strain Toolbox. To acquire 4-D mode images, a step motor was used to collect serial, ECG- and respiratory-gated images along the SAX, with a step size of 80–130 μm and a frame rate of 300–400 Hz.

Echocardiography: Infarct Size Quantification

Using 2-D PLAX B-mode images in the Vevo LAB software (v. 5.6.1), 2-D akinetic length analysis was completed on a subset of mice (n = 12) using the manual distance tracing function. An initial outline was made of the entire endocardial border at end-diastole starting and ending at the aortic root. This was defined as the total distance. Edge tracing was performed a second time along the endocardial border at end diastole in the region of the wall demonstrating no movement or contractility (Fig. 1B). This segment was defined as the akinetic segment. The akinetic length percentage was then calculated as such:

$$\mathrm{A}\mathrm{k}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{t}\mathrm{i}\mathrm{c}\mathrm{ }\mathrm{l}\mathrm{e}\mathrm{n}\mathrm{g}\mathrm{t}\mathrm{h}\mathrm{ }\left(\mathrm{\%}\right)=\frac{\mathrm{A}\mathrm{k}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{t}\mathrm{i}\mathrm{c}\mathrm{ }\mathrm{s}\mathrm{e}\mathrm{g}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{ }\left(\mathrm{m}\mathrm{m}\right)}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{ }\left(\mathrm{m}\mathrm{m}\right)}\mathrm{ }\times \mathrm{ }100$$

Figure 1.

Representative images of various infarct sizing techniques. A: histological scans stained with Masson’s Trichrome. Scale bar = 1 mm. B: 2-D B-mode echocardiographic images in PLAX view with akinetic segment indicated in red, at end diastole. Scale bar = 1 mm. C: SAX view of infarct, indicated in red, using Vevo LAB 4-D visualization, at end diastole. D: 3-D mesh of infarct and LV using Vevo LAB 4-D visualization, at end diastole. E: PLAX view of LV using 4-D Strain Toolbox for strain and infarct analysis. 4-D, four-dimensional; LV, left ventricle; PLAX, parasternal long-axis view; SAX, several short-axis; 3-D, three-dimensional; 2-D, two-dimensional.

Images in 4-D mode were used to volumetrically assess infarct size. Using the Vevo LAB 4-D visualization tools, the images of the ventricle obtained with 4DUS were segmented into three volumes to calculate infarct size (n = 8 at baseline, n = 12 postsurgery). First, endocardial and epicardial boundaries were traced in multiple SAX images from the apex to the base with interpolating the tracings for the remaining slices at three time points, including end systole, end diastole, and a middle time point between end systole and end diastole. This was completed to obtain the volume of the lumen region and the volume of the entire ventricle (Fig. 1C). A third tracing was completed to quantify the infarct volume, again in multiple slices from the apex to the base of the heart at the same time points. Wall thinning and wall motility were considered when defining the infarct area in the SAX frames. Global cardiac function data, such as ejection fraction (EF), can be extracted from the volume tracings. Once tracings were completed, the segmented volumes were reconstructed into a 3-D mesh of the LV (Fig. 1D). Infarct size, expressed as a percentage, was calculated as such:

$$\text{4}\text{-}\text{D volume infarct size }\left(\text{\%}\right)\text{ = }\frac{\text{Infarct volume }\left({\text{mm}}^{\text{3}}\right)}{\left[\text{Epi volume }\left({\text{mm}}^{\text{3}}\right)\mathrm{-}\text{Endo volume }\left({\text{mm}}^{\text{3}}\right)\right]}\text{ × 100}$$

Using the volume segmentation, surface area measurements were also extracted, using a specialty build-in of Vevo LAB (v. 5.6.1, B2401), to calculate infarct size as a proportion of the total LV surface area (n = 8 at baseline, n = 12 postsurgery). The surface area extracted from the infarct volume was divided by 2, as this area included both the surface at the epicardial and endocardial borders. This method assumes a negligible width in the infarct zone due to wall thinning. The infarct surface area is then divided by the surface area at the epicardial border. Since a closed volume segmentation was obtained using Vevo LAB 4-D analysis tools, the 2-D area representing the base of the heart was subtracted from the epicardial volume-extracted surface area to better approximate the ventricle surface area. The following equation was used:

$$\text{4}\text{-}\text{D Surface area size }\left(\text{\%}\right)\text{ = }\frac{\left[\text{Infarct surface area }\left({\text{mm}}^{\text{2}}\right)\text{/2}\right]}{\left[{\text{Epi surface area (mm}}^{\text{2}}\text{) }\mathrm{-}\text{ 2}\text{-}\text{D base area}\right]}\text{ × 100}$$

Echocardiography: 4-D Strain Toolbox Analysis

Strain analysis was completed using a 4-D Strain Toolbox, a customized α software that can be used to quantify echocardiographic images (19). This software allows visualization of 4DUS data and uses border tracings to create a 3-D map of strain. First, the user oriented and centered the images corresponding to the SAX, PLAX, and coronal views of the LV. The apical endocardial and base boundaries were set at peak systole and end diastole. Automated epicardial and endocardial contours were then added to the images. Manual adjustments can be made regarding the width of the endocardial contour and the myocardial thickness. Detailed contouring of the epicardial and endocardial borders can then be completed, using four SAX and three PLAX planes. For strain analysis, circumferential strain was obtained for regions corresponding to the base, mid, and apical slices of the LV. Longitudinal strain was obtained for the anterior free wall, anterior, anterior septum, posterior septum, posterior, and posterior free wall sections of the LV. Surface area and transmural strain were obtained according to the American Heart Association 17-segment model of the LV (20). Peak strain, systolic strain rate, and early and late diastolic strain rates were obtained for each indicated region. This software also allows for the measurement of transmural length, between corresponding endocardial and epicardial boundary points along the LV at different time points of the cardiac cycle. Regions with transmural length values of less than 0.5 mm were defined as regions of infarct as used previously by others (13, 21). Infarct size was then defined as the percentage of the total LV at systole with a myocardial thickness of less than 0.5 mm. In addition, we examined a novel estimate of infarct size from 4DUS by identifying regions of low surface area strain (<20%) on the 4-D segmentation of the endocardium based on our previous studies (13). We then estimated infarct size by calculating the percentage of the LV where surface area strain was below 20%, indicating areas of dyskinesis.

Echocardiography: Functional Analysis

With the use of PLAX B-mode and SAX M-mode images in Vevo LAB, automated LV tracings, using the autoLV function, were completed to assess LV function at both baseline and 4-wk post-MI. Manual LV contours were also completed in the PLAX B-mode images by tracing the endocardium in both end systole and end diastole using Vevo LAB’s operator defined LV trace function. Four-dimensional- mode functional measurements were obtained directly from volumetric analysis after tracing the contour of the endocardial boundary using Vevo LAB 4-D visualization. Finally, analysis in a 4-D Strain Toolbox also allowed the acquisition of global cardiac functional metrics.

Histology and Assessment of LV Infarct Scar Formation

Animals were anesthetized and maintained on 4% isoflurane, whereas the chest was opened and an apical injection of 1 M KCl arrested the heart in diastole. Excised hearts were perfused briefly with cold PBS, then perfusion-fixed and postfixed with 4% paraformaldehyde, embedded in paraffin, sectioned at 6 μm every 300–600 μm, stained with Masson’s Trichrome (MTC), and scanned with a Leica Aperio Versa 8 with a ×20 objective. MTC was used to distinguish between myocardial fibers (red) and scar tissue (blue). Infarcted tissue was identified as loss of myofibers with replacement by collagen on MTC-stained slides. We then calculated infarct size using a midline length approach by analyzing three to five equally spaced histological cross sections of the LV (22, 23). Using Aperio Image Scope software (Leica Biosystems, Sausalito, CA), the LV myocardial midline was traced by identifying the midpoint between the endocardial and epicardial boundaries. Infarct length was then determined by tracing the midline length of the myocardium, where the collagen scar encompassed at least 50% of the wall thickness (22). We then calculated infarct size for each cross section by dividing the midline infarct length estimates by the total midline lengths from each slice. These estimates were then averaged to obtain an estimate of the total infarct. In some cases where infarct was not seen to extend beyond the mid-LV, a histological section was not taken, and a 0% basal infarct was assumed. For mild infarcts, confirmatory qualitative analysis of infarct presence and severity was performed by a board-certified veterinary pathologist based on standard histopathological morphology (22–24).

Statistical Analysis

Absolute data are expressed as means ± SE (%). Depending on the number of groups being compared, significant differences were determined using ANOVA or t test. Linear regression analysis was performed to compare the scar size as measured through various methods. Correlations were graded as follows: poor (0.0–0.5), moderate (0.5–0.7), strong (0.7–0.9), or very strong (0.9–1.0). Bland–Altman analysis was also performed to determine the agreement between novel and conventional methods. Intraclass correlation (ICC) was used to evaluate the interuser reliability of various infarct sizing techniques between three users of different experience levels (Beginner and Experienced). For all statistical tests, P ≤ 0.05 was considered significant. All correlation analysis and statistical tests were performed using GraphPad Prism 9 software, except for ICC, which was calculated using Microsoft Excel.

RESULTS

Relative Consistency in Global Cardiac Function

Cardiac function metrics in infarcted mice were mostly consistent among various imaging modalities, including 2-D B-mode, M-mode, and 4-D mode imaging at baseline and 4-wk post-MI surgery (Supplemental Tables S1 and S2). Specifically, 4-D mode images, analyzed through both Vevo LAB 4-D visualization and 4-D Strain Toolbox, demonstrated a similar average decrease in EF after MI (Vevo LAB: 30.1 ± 5.0%; 4-D Strain Toolbox: 30.8 ± 6.1%; Supplemental Table S1). EF and stroke volume (SV) analyzed through the 4-D Strain Toolbox correlated very strongly with the values acquired through Vevo LAB 4-D visualization (r = 0.95 and r = 0.92, respectively; Supplemental Fig. S1).

Echocardiographic Akinetic Length Is Comparable with Histology

Akinetic length assessment provides a quick estimate of infarct as it only requires analysis of one PLAX frame (Fig. 1B). There is no significant difference between 2-D akinetic length and midline length histological analysis (Fig. 2). A correlation analysis completed between these two techniques demonstrated a strong and significant correlation (r = 0.76, P = 0.03; Fig. 3A). Similarly, a very strong negative correlation was found between 2-D akinetic length and EF (r = −0.98, P < 0.001; Fig. 4A). Finally, Bland–Altman analysis was completed to compare the agreement between these two techniques and a mean difference of −2.38% (95% confidence interval, from –26.08 to 21.33%) was obtained (Fig. 3D). These results indicate that 2-D PLAX echocardiographic akinetic length analysis provides comparable results with histology in terms of infarct size values and that akinetic length is a valid alternative method to histology.

Figure 2.

Comparison of infarct size, expressed as a percentage (%), as measured by histology, 2-D echocardiographic akinetic length, and 4-D echocardiographic volume and surface area; n = 8–12 (all males); analyzed by mixed-effects one-way ANOVA with post hoc Tukey’s test. **P ≤ 0.01, ***P ≤ 0.001. 4-D, four-dimensional; 2-D, two-dimensional.

Figure 3.

Novel 2-D and 4-D Vevo LAB infarct sizing techniques strongly agree with gold standard histology. A–C: correlation analysis between various echocardiographic infarct estimates and midline length-based histological analysis. D–F: Bland–Altman analysis between various echocardiographic infarct estimates and midline length-based histological analysis. n = 8 (all males); analyzed by Pearson correlation coefficient r and P values and Bland–Altman analysis; AL, akinetic length; 4-D, four-dimensional; IS, infarct size; 2-D, two-dimensional.

Figure 4.

Various infarct sizing techniques correlate with cardiac function. A–C: correlation analysis between 2-D and 4-D Vevo LAB infarct sizing techniques and EF (%) (n = 12; all males). D and E: correlation analysis between various 4-D Strain Toolbox infarct estimates and EF (%) (n = 14; n = 11 males and n = 3 females). Analyzed by Pearson correlation coefficient r and P values. EF, ejection fraction; 4-D, four-dimensional; 2-D, two-dimensional.

Echocardiographic Infarct Volume Correlates with Histology

Our results demonstrate that although 4-D echocardiographic volume assessment of infarct size as a percentage of the total LV correlates with histology, it provides conservative infarct size values when compared with histology (Figs. 2 and 3). There was a significant difference between the values of infarct size as measured by histology and 4-D volume analysis with Vevo LAB 4 D visualization (P = 0.006, Fig. 2), as well as between 4-D volume and 2-D akinetic length (P < 0.001, Fig. 2). Bland–Altman analysis of these techniques found a mean difference of −17.57% (95% confidence intervals, −36.90%–1.77%), which confirms this difference in infarct size using 4-D volume analysis compared with histology (Fig. 3E). Despite this difference, a strong correlation was found between techniques (r = 0.85, P = 0.008; Fig. 3B). A strong negative correlation was also found between this technique and EF (r = −0.87, P < 0.001, Fig. 4B). These results suggest that 4-D echocardiographic volumetric infarct sizing provides infarct size values that are smaller than the conventional method of histology; however, this method might be used to evaluate the relative sizes of infarct scars 4 wk after MI.

Surface Area Estimates of Infarct Size Are Consistent with Histology

There was no significant difference between the values of infarct surface area and histological midline analysis (Fig. 2). Four-dimensional infarct surface area was significantly different compared with the previously described 4-D infarct volume (P < 0.001, Fig. 2), but this metric was comparable to 2-D akinetic length (Fig. 2). A very strong and significant correlation was found between surface area and histology (r = 0.90, P = 0.002; Fig. 3C) as well as a strong correlation between surface area and EF (r = −0.87, P < 0.001; Fig. 4C). Finally, a Bland–Altman analysis of these techniques found a mean difference of −8.70% (95% confidence intervals, −24.57%–7.16%; Fig. 3F). These results indicate that 4-D echocardiographic infarct surface area provides similar infarct size values to the gold standard technique of histology. Furthermore, this technique exceeds histology as it allows for 3-D visualization of infarct.

Strain Analysis Correctly Identifies Regional Dysfunction and Infarct Sizing Methods Correlate with Histology

We validated the ability of the 4-D Strain Toolbox software to determine regional dysfunction with strain on a segmental basis in a subset of mice (n = 15). All segments showed a trend toward decreased strain rates, taken during systole, and peak strain after MI, with more significant reductions in circumferential strain rate in the mid-LV and apical regions compared with the basal region (Fig. 5). Also, the average of all segments for each of the four strain parameters was correlated against histological infarct size (Fig. 6). Correlation coefficients were respectively –0.82 (P = 0.003) for transmural peak strain, 0.84 (P = 0.002) for longitudinal peak strain, 0.92 (P < 0.001) for circumferential peak strain, and 0.95 (P < 0.001) for surface area peak strain. These correlations are all stronger than that of EF, determined in 2-D B-mode, and histological analysis (r = −0.78, P = 0.008, Fig. 6A). As well as providing functional insight into global and regional strain, the data that this strain analysis tool provides regarding transmural thickness and surface area strain were used to estimate infarct size. Transmural thickness, used to define infarct where the ventricle wall had a width of less than 0.5 mm at systole, demonstrated similar results to gold standard histology, however with slightly more conservative values (P = 0.02, Fig. 7). Despite this difference in values, which was also confirmed using a Bland–Altman analysis (Fig. 8C), this technique showed moderate to strong correlations with conventional methods. Specifically, the transmural thickness estimate of infarct displayed a correlation score of 0.76 (P = 0.001) with histology and a negative correlation of –0.62 (P = 0.02) with EF (Figs. 8A and 4D). Although using reduction in surface area strain, with a threshold of –20%, to complete infarct sizing, infarct size estimates corresponded very well with histology, with no significant difference between techniques (Fig. 7) and a bias of 1.67% (95% confidence intervals, −27.69 –31.04; Fig. 8D). A correlation coefficient of 0.94 (P < 0.001) was found with histology and of –0.89 (P < 0.001) with EF (Figs. 8B and 4E). These different techniques can be visually represented using colored polar plots and 3-D meshes of the LV (Fig. 9).

Figure 5.

Four-dimensional regional strain analysis identifies decrease in myocardial deformation after MI. A: circumferential segmental peak strain and strain rate. B: longitudinal segmental peak strain and strain rate. C: transmural segmental peak strain and strain rate. D: surface area segmental peak strain and strain rate. n = 15 (n = 12 males and n = 3 females); analyzed by repeated-measures two-way ANOVA with post hoc Sidak test; Ant, anterior; FW, free wall; MI, myocardial infarction; Post, posterior; S, septum (**P ≤ 0.01, ***P ≤ 0.001).

Figure 6.

Strain parameters have strong correlations with histology. A: correlation analysis between conventional 2-D EF (%) and histology (%). B–E: correlation analysis between various strain parameters and histology (%). n = 10 (n = 7 males and n = 3 females); analyzed by Pearson correlation coefficient r and P values; EF, ejection fraction; IS, infarct size; ; 2-D, two-dimensional.

Figure 7.

Comparison of infarct size, expressed as a percentage (%), as estimated by transmural thickness and surface area strain, using a 4-D Strain Toolbox. n = 15 (n = 12 males and n =3 females); analyzed by repeated-measures one-way ANOVA with post hoc Tukey’s test; *P ≤ 0.05. 4-D, four-dimensional.

Figure 8.

Novel 4-D strain-derived infarct sizing techniques strongly agree with gold standard histology. A and B: correlation analysis between various 4-D Strain Toolbox infarct estimates and midline length-based histological analysis. C and D: Bland–Altman analysis between various 4-D Strain Toolbox infarct estimates and midline length-based histological analysis. n = 15 (n = 12 males and n = 3 females); analyzed by Pearson correlation coefficient r and P values and Bland–Altman analysis; 4-D, four-dimensional; IS, infarct size.

Figure 9.

Infarct location, outlined in black, identified on colored polar plots using various 4-D Strain Toolbox infarct sizing techniques. Colored polar plots (A) and 3-D meshes of the LV volume at the endocardial wall for a mild, moderate, and large (left to right) sized infarct, with infarct location identified using a transmural thickness approach (dotted line) (B). Colored polar plots (C) and 3-D meshes of the LV volume at the endocardial wall for a mild, moderate, and large (left to right) sized infarct, with infarct location identified using a surface area strain estimate (solid line) (D). Scale bar = 1 mm. 4-D, four-dimensional; LV, left ventricle; 3-D, three-dimensional.

Strain Predicts Eventual Infarct Size Earlier than Wall Thinning

We examined longitudinal changes in surface area strain and wall thinning in a subset of our mice (n = 5) using the 4-D Strain Toolbox. We found that regions of the myocardium with low surface area strain (<20%) were statistically similar to day 28 infarct size as early as day 1 post-MI (Fig. 10D; P = 0.85), whereas areas of myocardial thinning (<0.5 mm) were similar to day 28 infarct size after day 7 post-MI (P = 0.17). Strain-estimated infarct size was also larger than wall-thinned infarct size estimation at days 1 (P = 0.004), 3 (P = 0.001), 7 (P = 0.06), 14 (P = 0.01), and 28 (P = 0.001) although both wall-thinning and strain-estimated infarct size at day 28 were statistically similar to histology (P = 0.08 and P = 0.07, respectively). Average left ventricular ejection fraction (LVEF) decreased from 66% ± 5.1 at baseline to 43% ± 4.5 (P = 0.008) at day 1 postprocedure and 34% ± 3.5 (P < 0.001) 28 days following the procedure (Fig. 10C).

Figure 10.

Longitudinal infarct analysis. A: progression of infarct size (dotted line) from baseline to day 28 post procedure using a wall-thinned estimate, demonstrated with a polar plot (left) and 3-D model (right). B: progression of infarct size (solid line) from baseline to day 28 post procedure using a strain-estimated infarct size, demonstrated with a polar plot (left) and 3-D model (right). C: ejection fraction calculated from 4DUS data decreased significantly over time compared with baseline values in infarcted mice. D: infarct size estimations as a percentage of total left ventricle using wall-thinned (dotted line) and strain-estimated (solid line). Surface area strain results are statistically similar to day 28 as early as day 1 post-MI, whereas myocardial thinning results are similar to day 28 infarct size after day 7 post-MI. n = 5; significance determined using ordinary one-way ANOVAs with multiple comparisons; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Scale bar = 1 mm. 4-D, four-dimensional; 4DUS, 4-D ultrasound; MI, myocardial infarction; 3-D, three-dimensional.

Infarct Size and Functional Metrics Display High Interuser Reliability

Infarct analysis using Vevo LAB was completed by three independent observers to determine the interuser reliability of these metrics. In a subset of mice, all users demonstrated the same trend between the various echocardiographic techniques compared with histology, which was completed by only one user (Supplemental Fig. S2). Two-dimensional akinetic length assessment demonstrated the highest ICC for infarct estimation, with volume and surface area analysis also displaying high correlations (2-D akinetic length: 0.97, 4-D volume: 0.79, 4-D surface area: 0.83; Supplemental Table S3). Four-dimensional volumetric assessment of function was also completed by various users and demonstrated a high correlation of 0.84 for EF (Supplemental Table S2). Finally, 2-D functional analysis was completed manually by multiple users, demonstrating a positive correlation with EF (r = 0.95). This 2-D manual assessment of function provided relatively consistent values with an automated analysis of the same 2-D data (Supplemental Table S2).

DISCUSSION

Quantitative analysis of MI in rodent models for biomedical research is an important step in understanding the pathophysiology of MI, the formation of collagen-rich infarcted tissue, and the progression of cardiac remodeling. With the arrival of new regenerative therapies to treat infarction, it has become important to have accessible and reliable methods to evaluate infarct size. Conventionally, infarct size is determined through terminal histological analysis (5, 7, 11, 22). This is combined with the assessment of standard cardiac function metrics (7). This study sought to establish robust and reliable preclinical in vivo echocardiographic imaging and analysis techniques to quantify infarct size and LV function after MI (8–13). Specifically, we explored 2-D echocardiographic akinetic length measurement from PLAX and 4-D echocardiographic volumetric and surface area measurements as well as 4-D strain parameters, using 4DUS images, as possible methods to reliably quantify infarct size and cardiac function. To validate their reliability, we compared functional and infarct measurement data between echocardiography techniques versus gold standard histology.

The present study demonstrates that infarct size as measured by 2-D echocardiographic akinetic length and by various 4-D echocardiographic infarct sizing techniques strongly correlate with a histological approach. Because of no significant differences found between techniques, we can conclude that 2-D and 4-D surface area echocardiography are reliable methods to determine infarct size 4-wk post-MI surgery. Moreover, high interuser reliability of the various echocardiographic techniques was obtained when performed independently by three different users. Although both 2-D and 4-D displayed reliability in assessing the extent of the infarct, 4DUS presents several additional benefits, such as the ability to visualize the 3-D shape of the infarct. As well, multiple infarct sizing and functional data can be extracted simultaneously from the same analysis and used accordingly, depending on the purpose of the study being conducted. Finally, data acquired in 4DUS can be further assessed with strain analysis to evaluate regional dysfunction and several other 3-D infarct sizing techniques. Therefore, 2-D analysis of akinesis in PLAX provides a quick and efficient method of infarct sizing, whereas 4DUS provides an in-depth analysis of infarct, global function, and regional strain data.

Previous literature using 2-D echocardiography for infarct quantification has demonstrated similar results. Strong correlations have been made with histology using the average of a series of transverse, or SAX, frames to measure the extent of infarct at the endocardial border (9, 12, 25–27). However, many of these studies have concluded that 2-D echocardiographic measurement of infarct overestimates the extent of infarction compared with conventional pathological evaluation (25, 26). These studies offer several possible explanations. Some studies note that overestimation of infarct size could stem from the dysfunction of nonischemic myocardium adjacent to the infarcted tissue, called tethering, or the temporary loss of function in viable tissue with normal blood flow (26). These regions of dysfunction are therefore considered in echocardiography but not histology, thus rendering 2-D echocardiographic infarct size values larger. Other studies will also consider akinetic, dyskinetic, and hypokinetic tissue in their echocardiographic evaluation of infarct, whereas only akinetic tissue is considered to be truly infarcted with histology (25). In the present study, we did not observe an overestimation of the extent of infarct with 2-D echocardiography (Fig. 2), similar to studies conducted by Yao et al. (28) and Kanno et al. (12). Specifically, these studies used strict inclusion criteria to define the dysfunctional region and used more SAX slices to better understand the LV structure (12, 28). Similarly, we only included akinetic regions in our 2-D echocardiographic analysis and we used one PLAX plane (Fig. 1B), which we believe provides a comparable understanding of LV structure with many SAX slices, while also requiring less analysis time. Overall, we can conclude that this 2-D method provides a quick and reliable assessment of infarct size.

Four-dimensional echocardiographic infarct sizing is a technique that uses 4DUS for image acquisition. This method allows for the acquisition of high-resolution images and minimizes the effects of through-plane motion that often occurs when imaging the heart (19). Previous studies have used 4DUS imaging to assess infarct size, such as with a wall motion index score and a strain estimate (8, 13). However, 4DUS had not yet been used to extract infarct size as a percentage of the LV volume or surface area, both of which we have demonstrated here from animals with different infarct sizes and between multiple cohorts. Heterogeneity in infarct size was obtained as a result of surgery being performed by different surgeons at different institutions and ranged from <10% to >60% infarct size as determined by histological analysis. Despite strong correlations between novel and conventional techniques in our current study, we observed conservative volume estimates of infarct compared with histology. Interestingly and similar to our results, a previous study that used a 3-D reconstruction of the heart also displayed underestimation of infarct size (10). This observation has also been made with cardiac MRI techniques (10). A suggested explanation for this observation is that the dehydration process during tissue preparation for histology shrinks the healthy myocardial tissue more than the infarcted tissue, thus rendering the percentage of infarction higher with histological analysis (10). However, we believe the most plausible explanation for the lower values obtained using a volume quantification of infarct is the wall thinning, and therefore decrease in volume, that occurs in the infarcted regions, combined with potential compensatory thickening in remote regions.

Following this volume analysis, we used a specialty build of the Vevo LAB 4-D visualization (v5.6.1, B2401) to continue analyzing infarct in a 3-D view, however, without considering the extent of wall thinning or thickening which can occur after MI, which can vary significantly depending on the transmural extent of the infarct. Our gold standard analysis made use of a midline approach in multiple histological sections from base to apex, which has previously been shown to be a reliable and straightforward method of infarct assessment (22, 23). We, therefore, replicated this technique as closely as possible with echocardiography. This surface area technique overcame the underestimation of infarct size that occurs when using a volume assessment, as it displayed a significant difference compared with 4-D infarct volume, but no difference when compared with histology (Fig. 2). This technique also had the strongest correlation with histology (r = 0.90) compared with the other Vevo LAB analysis techniques (Fig. 3). Therefore, although both 4-D infarct volume and surface area provide quantification that reliably determines the correct order of relative infarct sizes and correlates with histology, the surface area estimate replicates gold standard histology to a closer degree than volume.

A final aim of our study was to use a novel 4-D strain analysis tool to evaluate global function, regional reductions in strain, and potentially identify strain-derived infarct sizing techniques. We acquired global cardiac function metrics using various echocardiographic modalities, including Vevo LAB 4-D visualization and the 4-D Strain Toolbox. Specifically, when completing segmentations in the 4-D Strain Toolbox, global cardiac function parameters such as EF and SV were automatically calculated by the software. These metrics correlated very well, r = 0.95 and 0.92 for EF and SV, respectively, with the values obtained through Vevo LAB 4-D visualization, which has previously been validated against cardiac MRI for functional analysis after MI (8, 16). These previous studies have also observed that SAX M-mode analysis provides values that are larger than both 4DUS and MRI. The current study presented similar trends. At baseline, SV measured in SAX M-mode images is larger than both 4-D analysis tools and both EF and SV are larger than 4-D techniques after MI (Supplemental Table S2). Building upon previous research, we can therefore suggest that the 4-D Strain Toolbox correctly estimates global cardiac function metrics and further confirms that M-mode is not an appropriate analysis for myocardial function, as it only provides a one-dimensional (1-D) measurement.

When comparing pre- and post-MI regional strain parameters, most studies have demonstrated a significant reduction after MI surgery (13–15). This is consistent with previous reports of reduction in regional segmental longitudinal and radial strain and strain rates between sham-operated and infarcted mice (29). Circumferential strain rate was less significantly impaired in the basal region compared with the mid-LV and apical regions, validating the apical location of the infarct (Fig. 5). Interestingly, regional transmural peak strain and strain rate did not significantly decrease after MI. Specifically, despite trends toward reduced strain and strain rate, these parameters did not significantly decrease in the mid-LV and apical regions either. This could be explained by the difficulty of capturing the extent of wall thinning in these regions using echocardiographic imaging.

Not only did regional strain provide important functional data, the average of all regional segments for each of the four strain parameters displays a stronger correlation with histology than EF (EF: r = −0.78, transmural: r = −0.82, longitudinal: r = 0.84, circumferential: r = 0.92, surface area: r = 0.95, Fig. 6). This confirms that these various strain parameters provide a more sensitive overview of LV function after MI than EF. Indeed, reduction in EF reflects the overall dysfunction of the heart without the ability to distinguish the different directions of this dysfunction. This is important to note as EF may be perceived to be within a normal range while there is impairment of certain strain orientations, such as longitudinal strain, but this is being compensated for in another orientation, such as with increased circumferential function (30).

Four-dimensional strain analysis is also beneficial because of the ability to estimate surface area strain, which is not possible using 2-D techniques. Indeed, using 4DUS and subsequent image analysis, we estimated infarct area in regions of low surface area strain (<20%). Our longitudinal data also suggest that regions with low surface area strain (−20%) are predictive of eventual extent of infarct size as early as day 1 following infarction (Fig. 10D), whereas our wall-thinning estimate (<0.5 mm) was predictive after 7 days which was similar to findings by Soepriatna et al. (13). Similar to our 2-D akinetic area analysis, the surface area strain infarct size estimate revealed areas of the myocardium exhibiting dyskinesis which correlated well with histological values of infarct size. Although this correlation was strong, a few limitations of this method should be noted. First, this infarct area estimate relies on dyskinesia and is not necessarily a direct measure of infarcted myocardium. Indeed, the Bland–Altman plot in Fig. 8D revealed that in some cases surface area strain may be overestimating infarct size for larger infarcts. Although the number of infarcts is limited, infarct size may be underestimated for smaller infarcts, suggesting that these smaller infarcts with less myocardial involvement may not exhibit a proportionate level of dyskinesis. Despite these limitations, this surface area strain analysis provides a novel and potentially useful metric for infarct estimation from 4-D imaging, though further work may be needed to fully validate its use.

Another benefit of 4-D image analysis is the rich geometric information that can be used for analysis. Using 4-D mode acquired images, analysis is no longer impacted by through-plane motion, allowing for more robust parameter estimation with fewer assumptions. The acquired images are also less dependent on imaging windows and angle of acquisition, which is a constant consideration for 2-D imaging. This rich geometric information from 4-D imaging allowed us to estimate infarct size and geometry through a wall thinning approach similar to that used by others (13, 21). We observed a strong correlation of our wall thinning infarct size estimations with histology though the following limitations to this method should be noted. In this work, we examined our wall thinning metric in the majority of our mice 28 days postischemic injury after cardiac remodeling had already occurred. In a smaller subset of our mice (n = 5), however, we did find that wall thinning infarct size estimations underestimated true infarct extent during the early stages of cardiac remodeling following ischemic injury which was also similar to findings by Soepriatna et al. (13). In addition, in mild infarcts, such as subendocardial infarcts, wall thinning may not always be appreciated, thus limiting this method’s use and accuracy for all infarct sizes. This may partially explain that although our wall thinning estimate from our 4-D Strain Toolbox correlates well with histology, overall infarct size is underestimated (Fig. 7). Overall, we found that wall thinning could be used to predict infarct size, though relative infarct size and timing after surgery should be considered when applying this method.

LIMITATIONS

Although we have some preliminary data showing 4-D assessment of the progression of infarct size, further validation will be needed for 2-D and 4-D echocardiographic in the early stages of cardiac remodeling. To do so, the addition of echocardiographic measurements and replicate time points is necessary during the acute phase, typically considered to be within 1-wk post-MI. This is important because remodeling and scarring vary significantly during the week following the onset of MI and can influence the delineation of boundaries.

Another limitation is the time required for 4-D image acquisition and analysis. Four-dimensional mode imaging requires between 5 and 15 min for acquisition compared with 1–5 s for a 2-D B-mode image or 30 s to 1 min for a 2-D ECG-gated kilohertz visualization (EKV). Four-dimensional analysis also necessitates the analysis of more slices, in both SAX and PLAX view. Specifically, the 4-D Strain Toolbox analysis was typically completed in ∼30 min per data set for a trained user and up to 2 h for a new user. Four-dimensional infarct sizing using Vevo LAB typically took even longer, up to several hours, depending on the user’s level of training. Conversely, 2-D infarct sizing using the akinetic length strategy takes 10 to 15 min to complete. Several groups are considering ways to decrease 4-D image processing times through the use of machine learning techniques for image segmentation (18). However, even despite this limitation, 4-D ultrasound provides a more holistic image of LV function and structure, especially after infarct.

Conclusions

In summary, this study has described and validated 2-D echocardiographic akinetic length analysis and 4-D echocardiographic volume, surface area, and strain analysis for the assessment of infarct size and LV function by comparing these novel methods to the conventional method of serial histology. Although our study presented some limitations, we can conclude that 1) both 2-D and 4-D echocardiographic analysis techniques are reliable in quantifying infarct size, with 4-D analysis providing additional benefits; 2) histology-estimated infarct size is slightly larger than that of 4-D echocardiographic volume; and 3) 4-D strain analysis through the 4-D Strain Toolbox correctly identifies regional LV dysfunction after MI and infarct size. We propose that these techniques can be used in future preclinical research aiming to assess therapies and other strategies for reducing infarct size and therefore, increasing recovery after MI.

SUPPLEMENTAL DATA

Supplemental Video S1, Supplemental Tables S1–S3, and Supplemental Figs. S1 and S2: https://doi.org/10.6084/m9.figshare.17227115.v1.

GRANTS

The infrastructure is supported by Canadian Foundation for Innovation and a Diabetes New Investigator Grant (to E.E.M.). Fellowship support was provided from the Indiana University Medical Student Program for Research and Scholarship and the Indiana Clinical and Translational Sciences Institute, which is funded in part by Grants TL1TR002531 and UL1TR002529 (to C.C.E.; S. Moe and S. Wiehe, co-PIs) from the National Institutes of Health, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award.

DISCLAIMERS

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

DISCLOSURES

C.J.G. is a member of the Scientific Advisory Board and paid consultant of FUJIFILM VisualSonics, Inc. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

K.-H.K., C.J.G., and E.E.M. conceived and designed research; M.M.D., S.Q.C., N.A.T., C.C.E., L.E.S., S.M.P., E.N.L., K.A., J.M.G., I.L.S., R.S., C.J.G., and E.E.M. performed experiments; M.M.D., S.Q.C., C.C.E., L.E.S., S.M.P., E.N.L., K.A., J.M.G., A.D.C., C.J.G., and E.E.M. analyzed data; M.M.D., C.C.E., C.J.G., and E.E.M. interpreted results of experiments; M.M.D., S.Q.C., C.C.E., C.J.G., and E.E.M., prepared figures; M.M.D., S.Q.C., C.C.E., C.J.G., and E.E.M. drafted manuscript; M.M.D., S.Q.C., N.A.T., C.C.E., L.E.S., S.M.P., E.N.L., K.A., J.M.G., I.L.-S., K.-H.K., C.J.G., and E.E.M. edited and revised manuscript; M.M.D., S.Q.C., N.A.T., C.C.E., L.E.S., S.M.P., E.N.L., K.A., J.M.G., A.D.C., I.L.-S., R.S., K.-H.K., C.J.G., and E.E.M. approved final version of manuscript.