Background
The subendocardium is highly vulnerable to damage and is thus affected even in subclinical disease stages. Therefore, methods reflecting subendocardial status are of great clinical relevance for the early detection of cardiac damage and the prevention of functional impairment. The aim of this study was to investigate the potential ability of myocardial strain parameters to evaluate changes within the subendocardium.
Methods
Male 129/Sv mice were injected with isoproterenol (ISO; n = 32) to induce isolated subendocardial fibrotic lesions or saline as appropriate control ( n = 15). Transthoracic echocardiography was performed using a 30-MHz linear-frequency transducer coupled to a high-resolution imaging system, and acquired images were analyzed for conventional and strain parameters. The degree of collagen content within the different cardiac layers was quantified by histologic analysis and serum levels of tissue inhibitor of metalloproteinase–1, a biomarker for fibrosis, were assessed.
Results
ISO treatment induced a marked increase in subendocardial collagen content in response to cell loss (control vs ISO, 0.6 ± 0.3% vs 5.8 ± 0.9%; P < .001) and resulted in a moderate increase in left ventricular wall thickness with preserved systolic function. Global longitudinal peak strain (LS) and longitudinal strain rate were significantly decreased in ISO-treated animals (LS, −15.49% vs −11.49% [ P = .001]; longitudinal strain rate, −4.81 vs −3.88 sec −1 [ P < .05]), whereas radial and circumferential strain values remained unchanged. Global LS was associated with subendocardial collagen content ( r = 0.46, P = .01) and tissue inhibitor of metalloproteinase–1 serum level ( r = 0.52, P < .05). Further statistical analyses identified global LS as a superior predictor for the presence of subendocardial fibrosis (sensitivity, 84%; specificity, 80%; cutoff value, −14.4%).
Conclusion
Assessment of LS may provide a noninvasive method for the detection of subendocardial damage and may consequently improve early diagnosis of cardiac diseases.
Highlights
- •
STE was performed in an experimental model of isolated subendocardial fibrosis, and comprehensive morphologic-functional correlation analyses were performed.
- •
Functional impairment was present only in decreased global LS and global LSR values, whereas global systolic function and other strain parameters remained unaffected.
- •
Global LS was associated with subendocardial collagen content and serum level of TIMP-1.
- •
Global LS was identified as a superior predictor for the presence of subendocardial damage.
Two-dimensional speckle-tracking echocardiography (STE) provides a rapid, noninvasive, and reproducible diagnostic method of assessing cardiac function. STE-based parameters reflecting tissue deformation (strain) during the cardiac cycle, such as longitudinal peak strain (LS), have been shown to be superior prognostic factors and predictors for early cardiac impairment in subclinical disease development. On the basis of this cumulative evidence, LS assessment has recently been included in guideline recommendations, especially to detect early and minor changes in ventricular systolic function.
Strain is caused by contraction and relaxation of muscle fibers during systole and diastole. Within the left ventricle, these myofibers are organized in layers, whose orientation changes from a right-handed helix in the subendocardium to circumferential in the mesocardium, back to a left-handed helix in the subepicardial layer ( Figure 1 A). Because the longitudinal mechanics of the left ventricle are determined mainly by contraction of subendocardial myofibers, LS may be used as an indicator of subendocardial status ( Figure 1 B).
The subendocardium is highly vulnerable to damage and is thus affected even in subclinical disease stages. Therefore, it is generally assumed that early alterations in LS are due to the reflection of (impaired) subendocardial function. Several studies have suggested this association in different cardiovascular pathologies. However, most of them were limited to either indirect measures of subendocardial damage or histology derived from biopsies, which may not reflect the morphology of the whole myocardium.
In this study, we aimed to evaluate the accuracy of the various strain parameters with regard to their ability to assess changes within the subendocardium. We carried out comprehensive STE-based strain analyses combined with detailed histologic studies and biomarker assessment in a well-established animal model of subendocardial damage and fibrosis and performed functional-morphologic correlation analyses. We hypothesized that in particular, longitudinal strain parameters may be able to detect subendocardial lesions and allow estimation of the degree of damage.
Methods
All animal procedures were performed in accordance with the guidelines of the German Law on the Protection of Animals. The experimental protocols ( Figures 1 C and 1D) were reviewed and approved by the authorities (Landesamt für Gesundheit und Soziales, Berlin, Germany). Animals used in this study served as placebo-treated control groups in ongoing projects and were kept under identical housing conditions (12-hour light/dark cycle, standard diet ad libitum). Data sets, histologic sections, and blood samples were acquired prospectively and analyzed for this study in a retrospective manner.
Study Protocols
Subendocardial damage was induced by administration of isoproterenol (ISO; Sigma-Aldrich, St. Louis, MO), as described previously. Male 129/Sv mice (6–8 weeks old; Janvier Labs, Le Genest-Saint-Isle, France) were injected subcutaneously with ISO (25 mg/kg body weight, dissolved in saline; n = 32) for 4 consecutive days or saline as appropriate control ( n = 15). Animals were assigned to two different study protocols.
The first cohort ( n = 32) was investigated 2 weeks after final application of ISO or saline, when short-term treatment effects were considered to be resolved ( Figure 1 C). These animals underwent baseline echocardiographic examination before the first administration of ISO or saline. Final echocardiography and measurements of systolic blood pressure (SBP) were performed 12 to 13 days after the completion of treatment. During necropsy on day 14 after final application, blood samples and tissues were collected for further analyses. A small subset of this cohort ( n = 3) was used for histologic analyses of longitudinal sections of the heart only and did not undergo any further examination.
Short-term treatment effects were assessed in the second cohort ( n = 12), which was sacrificed 2 days after the final application of ISO or saline ( Figure 1 D).
Echocardiographic Examination
Echocardiographic studies were performed by trained investigators using an MX400 ultra-high-frequency linear-array transducer (18–38 MHz; center transmit, 30 MHz; axial resolution, 50 μm) coupled to a Vevo 3100 Imaging System (both FUJIFILM VisualSonics, Toronto, ON, Canada).
Initially, mice were weighed, anesthetized with 3% isoflurane (in 80% oxygen; Abbott Laboratories, Abbott Park, IL), and fixed in a supine position on a heated pad (37°C). The chest was depilated, and prewarmed ultrasound gel (Parker Laboratories, Fairfield, NJ) was applied. For examination, isoflurane concentration was reduced to a minimum (1%–2%) to obtain constant and comparable heart rates, continuously monitored by electrocardiographic recording.
All animals were scanned in parasternal long- and short axis views, according to a standard operating protocol ( Supplemental Material : Standard Operating Protocol, available at www.onlinejase.com ). In contrast to clinical examination, the small size of the rodent heart allows visualization of the long axis of the left ventricle in its maximum dimension from apex to base ( Figure 2 A, [CR] ). Moreover, the apical two- and four-chamber views are challenging to obtain in small animals. Therefore, evaluation of cardiac systolic function in mice is preferably performed in images in the parasternal long-axis view, rather than in apical views. Before data storage, images were reviewed in slow motion to verify reproducibility (e.g., to verify absence of apical foreshortening).
Images in short-axis view were taken at the midpapillary level of the left ventricle, in which both papillary muscles served as anatomic landmarks for standardized views ( Figure 2 D, [CR] ). Additionally, M-mode images were assessed in the short-axis view to determine wall thicknesses.
All acquired images were digitally stored in raw format for further offline analyses.
Image Analysis
All image analyses were performed by a single observer using Vevo LAB (FUJIFILM VisualSonics). The software displays acquired high-resolution images in adjustable slow-motion loops, allowing appropriate analysis despite high heart rates. Echocardiographic parameters were calculated according to current American Society of Echocardiography formulas.
Parameters of global systolic function were assessed semiautomatically in B-mode images derived from the parasternal long-axis view using the monoplane Simpson method of disks. M-mode images from the midpapillary region of the left ventricle in the short-axis view were analyzed for left ventricular (LV) wall thickness and diameter.
STE-based analyses were performed using a dedicated software package (VevoStrain [based on an algorithm by TomTec Imaging, Unterschleissheim, Germany]; FUJIFILM VisualSonics), as described previously. Strain and strain rate parameters were assessed from grayscale B-mode images during three consecutive cardiac cycles (electrocardiographically controlled; the interval between two R waves was considered one cardiac cycle). Care was taken to select cycles in which no deep breathing occurred, to avoid plane movement of the myocardium. Semiautomatic tracing of endo- and epicardial borders was repeated three times with manual correction if necessary.
Longitudinal and radial strain parameters were generated from images taken in the parasternal long-axis view (frame rate, 227 ± 1 frames/sec; heart rate/frame rate ratio, 1.8 (beats/min)/(frames/sec); depth, 12.9 ± 0.1 mm; width, 10.7 ± 0.1 mm). Myocardial boundaries were traced beginning at the midbasal level ( Figures 2 B and 2C, Videos 3 and 4 ).
Circumferential strain parameters were assessed in images taken in the short-axis view (frame rate, 277 ± 8 frames/sec; heart rate/frame rate ratio: 1.5 ± 0.1 (beats/min)/(frames/sec); depth, 11.7 ± 0.1 mm; width, 9.2 ± 0.3 mm), in which papillary muscles were excluded from tracing ( Figures 2 D and 2E, Videos 5 and 6 ).
All acquired images in the long-axis view reached appropriate quality for speckle-tracking echocardiographic analysis, proved by visual evaluation of myocardial contours and high frame rates. Image quality required for adequate STE from the short-axis view was reached in 25 animals (86%) at baseline and in 28 animals (97%) on final echocardiography. LV length (base-apex distance) was assessed in the parasternal long-axis view as a surrogate for putative apical foreshortening.
For determination of longitudinal and circumferential strain parameters, the software virtually places 48 adjacent points along the traced endocardial border, whose deformation is analyzed over the cardiac cycle. Similarly, the deformation of endocardial points and corresponding ones along the epicardial border is analyzed to determine radial strain parameters. No standardized thickness for the regions of interest is assumed by the software to allow detection of small changes in myocardial strain and strain rate.
The myocardium was divided into six segments for images derived from long- and short-axis views, respectively, similar to the definition of the American Heart Association. Segmental strain curves were averaged for the calculation of global strain and strain rate parameters.
All analyses were performed blinded to results from histology, blood sample analysis, and necropsy. Assessment of intra- and interobserver variability was performed in randomly selected animals from both treatment groups ( n = 18). The same investigator and a second observer analyzed identical images and heart cycles several months later for determination of intra- and interobserver variability, respectively.
Physiologic Data
SBP was measured in a subset of animals ( n = 19) using a noninvasive blood pressure controller. Briefly, conscious mice were strained in a plastic tube, and a pressure cuff with a pulse electrode (MLT125/M; ADInstruments) was placed around the tail. To reduce stress during final examination, animals were pretrained by exposing them to the procedure and allowed to acclimatize for several minutes before measurements were taken. SBP was calculated as the mean of multiple measurements per animal.
Reported heart rates were assessed during baseline and final echocardiographic examinations via electrocardiographic recording.
Blood Sample Analyses
Blood samples were collected from the retrobulbar venous plexus before animals were sacrificed. Blood glucose levels were assessed using a clinical glucose meter (Contour XT; Bayer, Leverkusen, Germany). Serum levels of tissue inhibitor of metalloproteinase–1 (TIMP-1) were quantified in a subset of animals ( n = 18) using an enzyme-linked immunosorbent assay according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN). Values are reported on a logarithmic scale as TIMP-1 serum concentrations in picograms per microliter.
Histologic Analysis
During necropsy, hearts were excised and weighed gravimetrically after being rinsed and the atria removed. Cross-sections of the cardiac midregion and longitudinal sections of the heart were fixed in formalin and paraffin embedded. Histologic slides were stained with Picrosirius Red (Morphisto, Frankfurt am Main, Germany) and Hematoxylin and Eosin (H&E; Carl Roth, Karlsruhe, Germany) for determination of collagen content, tissue degeneration and regeneration, and cardiomyocyte hypertrophy. Distinction between subendocardium and subepicardium was made by analyzing 11 or 12 digitalized microscopic images randomly taken from both border regions of LV cross-sections (magnification 20×).
Collagen content was defined as percentage fraction of Picrosirius Red–stained collagen fibers from total image area, automatically assessed and calculated using ImageJ (National Institutes of Health, Bethesda, MD; Supplemental Material : ImageJ Macro, available at www.onlinejase.com ).
A blinded expert in veterinary pathology (R.K.) was asked to classify the type of tissue degeneration or regeneration in acquired images of H&E-stained subendocardium and subepicardium: 0 = no abnormality; 1 = well-differentiated granulation tissue as a sign of resolved damage (mostly fibrocytes with elongated, spindle-shaped nuclei; mostly well-definable collagen fibers arranged in parallel); 2 = less differentiated granulation tissue as a sign of subacute damage (fibrocytes with both elongated, spindle-shaped and more blastic, round nuclei; some collagen fibers arranged in parallel); and 3 = low-differentiated granulation tissue as a sign of acute damage (mostly fibrocytes with plump, blastic, round nuclei; more homogenous extracellular matrix).
Similarly, the degree of cardiomyocyte hypertrophy was graded blinded in H&E-stained images as follows: 0 = all cardiomyocytes with regular size in transverse sections, 1 = few cardiomyocytes with increased size in transverse sections, 2 = some cardiomyocytes with increased size in transverse sections, and 3 = majority of cardiomyocytes with increased size in transverse sections.
Statistical Analysis
Results are shown as mean ± SE. Normal distribution was proved visually by using box plots, histograms, and Q-Q plots. Statistical analysis was performed by using unpaired Student’s t tests or analysis of variance for multiple comparisons, followed by the Bonferroni posttest. Correlation of linearly related variables was studied using Pearson’s correlation coefficient ( r ). Receiver operating characteristic analyses were performed to determine the ability of various parameters to predict ISO-induced subendocardial fibrosis. Inter- and intraobserver variability were expressed as intraclass correlation and coefficient of variation, as described previously. All analyses were performed using GraphPad Prism (GraphPad Software, La Jolla, CA) and R (R Foundation for Statistical Computing, Vienna, Austria). A two-tailed P value of <.05 was considered to indicate statistical significance.
Results
Study Population
Animals were randomly assigned to both study protocol and treatment regime. Exclusively males were used in this study, aiming for better comparability and exclusion of hormonal fluctuation as a possible confounder.
Four animals ( n = 3 in the first cohort, n = 1 in the second cohort) died in the course of ISO injections for reasons not further investigated (overall mortality under ISO treatment 12.5%; Supplemental Figure 1 , available at www.onlinejase.com ). Data from these animals were excluded from study results. All control animals finished the protocols. Both treatment groups showed comparable development of body weight and had similar blood glucose levels during final assessment ( Tables 1 and 2 ).
| Control | ISO | P | |
|---|---|---|---|
| Male | 5 (100) | 6 (100) | — |
| Physiologic data | |||
| BW (g) | 25.7 ± 0.7 | 27.2 ± 0.9 | .24 |
| BW change (g [%]) | +0.5 ± 0.2 (+2.0 ± 0.6) | +0.5 ± 0.6 (+1.8 ± 2.1) | .98 |
| Blood sample analyses | |||
| Glucose (mg/dL) | 152 ± 10 | 150 ± 10 | .89 |
| Necropsy data | |||
| HW (mg) | 109 ± 3 | 136 ± 5 | .001 ∗∗ |
| HW/BW (mg/g) | 4.22 ± 0.08 | 5.01 ± 0.09 | <.001 ∗∗∗ |
| Control | ISO | P | |
|---|---|---|---|
| Male | 10 (100) | 19 (100) | — |
| Physiological data | |||
| BW (g) | 26.8 ± 0.5 | 28.0 ± 0.5 | .18 |
| BW change (g [%]) | +0.9 ± 0.2 (+3.5 ± 0.9) | +1.3 ± 0.2 (+4.9 ± 0.7) | .19 |
| SBP (mm Hg) | 126 ± 3 | 119 ± 3 | .06 |
| Heart rate (beats/min) | 408 ± 9 | 387 ± 11 | .22 |
| Blood sample analyses | |||
| Glucose (mg/dL) | 156 ± 8 | 145 ± 6 | .27 |
| TIMP-1 (log[pg/μL]) | 3.11 ± 0.03 | 3.25 ± 0.04 | .01* |
| Necropsy data | |||
| HW (mg) | 114 ± 3 | 116 ± 2 | .51 |
| HW/BW (mg/g) | 4.24 ± 0.07 | 4.16 ± 0.06 | .47 |
| Conventional echocardiography | |||
| EF (%) | 50 ± 1 | 47 ± 1 | .11 |
| FS (%) | 25 ± 1 | 24 ± 1 | .53 |
| FAC (%) | 38 ± 2 | 36 ± 1 | .40 |
| LVAW (mm) | 0.42 ± 0.01 | 0.55 ± 0.02 | .001** |
| LVID (mm) | 3.99 ± 0.03 | 4.04 ± 0.05 | .52 |
| LVPW (mm) | 0.42 ± 0.01 | 0.5 ± 0.02 | .002** |
| Base-apex distance (mm) | 7.29 ± 0.12 | 7.34 ± 0.08 | .72 |
| STE | |||
| GLS (%) | −15.5 ± 0.7 | −11.5 ± 0.7 | .001** |
| GRS (%) | 23.9 ± 2.5 | 25.2 ± 1.4 | .61 |
| GCS (%) | −19.1 ± 0.9 | −18.0 ± 0.7 | .36 |
| GLSR (sec −1 ) | −4.8 ± 0.3 | −3.9 ± 0.2 | .02** |
| GRSR (sec −1 ) | 5.5 ± 0.4 | 5.9 ± 0.3 | .40 |
| GCSR (sec −1 ) | −5.3 ± 0.3 | −5.2 ± 0.2 | .81 |
Because differences in loading conditions and heart rate might affect STE, possible β-adrenergic effects by ISO treatment were monitored before final echocardiography was performed. As expected, no significant long-term effects of ISO on SBP and heart rate were observed 12 to 13 days after completion of treatment ( Table 2 ).
Necropsy revealed transient cardiac hypertrophy due to ISO treatment ( Table 1 ) that was no longer detectable 2 weeks later when final echocardiography was performed ( Table 2 ).
Histologic Findings
Histopathologic analysis showed distinct fibrotic lesions 14 days after final ISO treatment, indicated by red-stained collagen fibers ( Figures 3 A and 3B). The amount of collagen content within the subendocardial and subepicardial layer was assessed to determine detailed local distribution of fibrosis ( Figure 3 C). Quantification revealed a 10-fold increase of subendocardial collagen content in ISO treated animals (control vs ISO, 0.6 ± 0.3% vs 5.8 ± 0.8%; P < .001), whereas subepicardial collagen differed insignificantly between both groups (0.4 ± 0.1% vs 0.9 ± 0.1%, P = n.s.; Figure 3 D). Evaluation of longitudinal sections of the heart indicated that fibrotic lesions occurred predominantly in apical rather than in basal segments of the left ventricle ( Figure 3 E).
Preceding pathologic changes were assessed 2 days after final application of ISO or saline in H&E-stained cardiac cross-sections ( Figures 3 E and 3F). Under short-term effects, ISO led to cell loss accompanied by a disorganized, homogenous, extracellular matrix and formation of granulation tissue, predominantly in the subendocardial layer (type 2 lesion; Figure 3 G). After 2 weeks, acute damage was mostly resolved, and scar formation was apparent as well-differentiated, reorganized extracellular matrix and low-grade granulation (type 1 lesion; Figures 3 H and 3I). Type 3 lesions as described in the “Methods” section were not detected.
Consistent with observed cardiac hypertrophy, ISO induced a transient increase in cardiomyocyte cross-sections 2 days after final application ( Figure 3 G). This effect was present in both myocardial layers but appeared slightly more pronounced next to subendocardial lesions ( Figures 3 G and 3J). No signs of cardiomyocyte hypertrophy were detected 2 weeks later ( Figure 3 H).
TIMP-1 Serum Levels
TIMP-1 serum levels were significantly increased in ISO treated animals ( Table 2 ). Further statistical analyses revealed that serum concentrations were in good agreement with degree of subendocardial collagen content, as indicated by a strong correlation between both parameters ( Table 3 ).
| Pearson’s r | Collagen Endo | Collagen Epi | TIMP-1 | HW/BW |
|---|---|---|---|---|
| Collagen Endo | — | — | 0.68** | — |
| GLS | 0.46* | 0.14 | 0.52* | 0.14 |
| GLSR | 0.16 | −0.01 | 0.45 | 0.20 |
Conventional Echocardiography
Both treatment groups showed comparable systolic function on baseline echocardiography ( Supplemental Table 1 , available at www.onlinejase.com ).
ISO treatment resulted in moderate thickening of LV walls but unaltered LV inner diameters ( Table 2 ). Global systolic function did not differ between both groups on final examination, indicated by preserved LV ejection fraction and fractional shortening in ISO-treated animals ( Table 2 ).
Speckle-tracking Echocardiography
All peak strain parameters, as well as radial and circumferential strain rates, showed excellent intra- and interobserver agreement (intraclass correlation coefficient ≥ 0.80; Supplemental Table 2 , available at www.onlinejase.com ). Intraclass correlation coefficients for global longitudinal strain rate (LSR) were 0.61 and 0.56 for inter- and intraobserver variability, respectively ( Supplemental Table 2 , available at www.onlinejase.com ).
At baseline, both treatment groups showed comparable values for strain and strain rate parameters ( Supplemental Table 1 , available at www.onlinejase.com ).
ISO treatment led to a marked decrease in global LS, whereas global radial and circumferential peak strain remained unaffected ( Table 2 ). Similarly, differences between both treatment groups regarding tissue deformation over time were present only in decreased global LSR after ISO treatment, not in radial and circumferential strain rates ( Table 2 ).
Changes in speckle-tracking echocardiographic parameters from baseline to final assessment were evaluated for each individual animal. Interestingly, almost all parameters decreased over time to at least a slight extent ( Figure 4 ). However, a statistically significant reduction due to treatment was reached only in the ISO group in global LS and LSR ( Figures 4 A and 4D).
