To investigate the effect of infarct size and location on left ventricular (LV) functional and microstructural alterations in well-established porcine models.
To investigate the effect of infarct size and location on left ventricular (LV) functional and microstructural alterations in well-established porcine models.
Myocardium infarction was induced in mini-pigs at apical septum (Group 1, n = 6) or basal lateral wall (Group 2, n = 6) by permanent occlusion of the left anterior descending or left circumflex coronary artery, respectively. In vivo cardiac magnetic resonance (CMR) was performed 4 and 13 weeks later. Hearts were then excised and examined by ex vivo diffusion tensor imaging (DTI) for myocardium structural changes in infarct, adjacent and remote regions.
LV ejection fractions correlated negatively with infarct sizes. Between week 4 and 13, Group 1 exhibited more changes in end-systolic volume, LV mass, and ejection fraction, although it showed a smaller infarct volume percentage. Ex vivo results revealed the decreased water diffusion fractional anisotropy and increased diffusivities in infarct region in correlation with infarct size, but with no significant difference between the two groups. However, LV myocardial double-helical fiber architecture was found to alter in Group 1, shifting more towards left-handed direction as compared to controls.
Postinfarct LV functional and structural remodeling is affected by both infarct size and location. J. Magn. Reson. Imaging 2009;29:305–312. © 2009 Wiley-Liss, Inc.
LEFT VENTRICULAR (LV) remodeling after myocardial infarction (MI) is associated with alteration of LV mechanical function and myocardium structure. Several potential determinants of LV remodeling have been reported, such as infarct size and infarct location (1). Infarct size has been regarded as the primary determinant of LV remodeling (2) and associated with adverse LV function (3). However, the effect of infarct location is still controversial. Recent cardiac magnetic resonance (CMR) imaging has shown a linear relation between the infarct size with LV ejection fraction (EF) and LV volume, independent of the infarct location and transmurality. Nevertheless, several clinical studies observed that anterior MI patients usually had a marked LV dilation during early MI (4), more abnormal strain pattern (5), higher prevalence of low LV EF (6), higher mortality rate, and worse in-hospital prognosis compared to patients with MI at other locations (6). It has been suggested earlier that such different outcomes may be related to the complex heart structure (7).
As a major determinant of cardiac function, the overall myocardium structure such as a fiber structure is known to alter dramatically after MI (8). However, the impact of different MI location on the postinfarct LV myocardium structural changes remains unclear. Recently, MR diffusion tensor imaging (DTI) was demonstrated to be a powerful tool for nondestructive microstructural assessment of neural tissue and myocardium (9–16). Until now, only three studies have been reported that utilized DTI to directly examine the LV fiber structure alteration after MI (17–19). Infarct myocardium manifested the degradation of fiber integrity and alterations of myocardial fiber architecture. These studies demonstrated that DTI provides a sensitive and complementary approach to study of the degradation of myocardium structure during postinfarct LV remodeling.
In this study, we hypothesize that the location of MI may affect the degree of LV remodeling at both the functional and structural level. Both in vivo CMR and ex vivo DTI were employed to experimentally document the effect of infarct size and location on LV functional and structural remodeling in two well-established porcine models with MI in LV apical septum or basal lateral wall with similar infarct sizes.
Twelve adult mini-pigs (45–50 kg, 9–12 months old, n = 12) were divided into two groups. Distal embolization of left anterior the descending (LAD) or left circumflex (LCX) coronary arteries was performed by an incremental dose of intracoronary injection of 700 μm microspheres to induce MI in the septum near apex or lateral wall near base, respectively (19). LAD-related MI group (n = 6) was regarded as Group 1 and LCX-related MI group (n = 6) as Group 2. Six adult mini-pigs of similar weight and age (n = 6) were used as intact controls without any surgical procedures. All animal experiments were approved by the local institutional ethics committee for animal research.
All MRI experiments were conducted on a Philips 3T Achieva scanner (Philips Medical Systems, Best, Netherlands). The 12 infarct pigs in Group 1 and Group 2 underwent the same in vivo CMR imaging procedure twice after induction of MI at 4 weeks and 13 weeks, respectively. Two randomly selected intact control pigs were also imaged once as the baseline of normal LV function. During imaging, each pig was anesthetized with Propofol and the heart rates typically ranged from 70–100 bpm. Respiration was maintained by a ventilator with breath-holding by hyperventilation. A breath-hold 2D balanced fast field echo (balanced-FFE) cine sequence was first performed to acquire eight short-axis slices covering the whole heart for evaluation of LV functions. The sequence parameters were as follows: TR/TE = 5.5/2.2 msec, cardiac frames = 20, slice thickness = 8 mm without slice gap, in-plane resolution = 1.04 × 1.04 mm2, matrix size = 192 × 192, half scan factor = 0.67, and acquisition time ≈1 min (19). Then a bolus of a gadolinium contrast agent (gadopentetate dimeglumine, 0.5 mol/l, Magnevist, Germany) at a dose of 0.1 mmol/kg was injected intravenously. Breath-hold delay-enhanced T1-weighted imaging was carried out ≈10 minutes later at the same eight slice locations using an inversion recovery multiple 2D fast field echo (M2D-FFE) sequence to investigate the myocardium viability, infarct location, and volume. The parameters were TR/TE = 3.8/1.3 msec, inversion recovery time = 275 msec, matrix size = 512 × 512, half scan factor = 0.67, in-plane resolution = 0.68 × 0.68 mm2, NEX = 3, and acquisition time ≈5 minutes (19).
Immediately after in vivo CMR imaging at the second timepoint (week 13), the 14 pigs (12 infarcted and two controls) together with the other four intact controls were anesthetized and the hearts were arrested in diastole, excised rapidly, and perfused retrogradely (19). Then the hearts were fixed in formalin solution for at least 48 hours to let the possible early ventricular geometry changes (9) occur before ex vivo MR imaging. Ex vivo DTI study was performed on all 16 formalin-fixed porcine heart samples at room temperature (≈20°C) for myocardium structural analysis using a single-shot spin echo EPI (SE-EPI) DTI sequence. Imaging parameters were as follows: TR/TE = 4000/45 msec; spatial resolution = 1.13 × 1.13 × 1.13 mm3; diffusion sensitivity b values = 0 and 800 s/mm2; number of gradient directions = 15; diffusion gradient duration Δ = 10.5 msec; diffusion time δ = 26.5 msec; slice number ≈40; and number of averages = 40. The slices were planed at the same orientation as in the in vivo CMR imaging and covered the entire heart. Total scan time was ≈1 hour per sample.
LV morphology and functions were investigated from in vivo CMR images using specialized postprocessing software (Cinetool 3.9.8, General Electric Healthcare, Milwaukee, WI) (20). Endocardium and epicardium contours were manually traced. End-diastolic volume (EDV) and end-systolic volume (ESV) were measured from the cine image sets. Stroke volume was defined as (EDV-ESV), and LV EF values as [(EDV – ESV)/EDV]. LV mass was calculated as myocardium density (1.05 g/mL) × myocardium volume at end diastole. The latter was defined the volume enclosed by contours of epicardium and endocardium with exclusion of papillary muscles. Infarct was determined as pixels with signal intensity 2 × SD (standard deviation) higher than the mean value of a region in remote normal myocardium on the same slice (21, 22) in the delay-enhanced T1-weighted images. Infarct volume percentage was measured and correlated with LV EF among all 14 pigs scanned. The LV morphology and functions between the two timepoints were also examined for relative percentage changes in the two MI groups.
The 3D myocardial fiber tracking was performed from the ex vivo DTI data using PRIDE software package (Philips Medical Systems) with thresholds set to 0.2 for FA magnitude and 40° for angular change (12). DTI parameters representing myocardial fiber integrity, including fractional anisotropy (FA), mean apparent diffusion coefficient (mean ADC), three eigenvalues, and primary eigenvector were computed from DtiStudio v. 2.30 (Johns Hopkins University, Baltimore, MD) (19, 23). Axial diffusivity was calculated as the primary eigenvalue, and radial diffusivity as the average of the secondary and tertiary eigenvalues. A MatLab (MathWorks, Natick, MA) program was developed to compute the fiber helix angle from the primary eigenvector with the definitions proposed by Chen et al (17) and Scollan et al (24). For infarct groups, 10 slices with infarction were selected (Fig. 1a,c), and their corresponding controls were 10 slices at similar locations in the control group (Fig. 1b,d). On each slice, infarcted myocardium was identified as hyperintense in the T2-weighted images (ie, DWIs with b value = 0), typically exhibiting thin myocardium wall. The remaining noninfarcted myocardium was divided circumferentially into six equiangular radial segments (Fig. 2a,c). The two segments immediately adjacent to the infarct region were classified as the adjacent region and the remaining four segments were categorized as the remote region (19). Circumferential length of infarct on a slice was computed as θ × R, where θ was the infarct angle and R the radius of the epicardium on that slice. For the control group, a quarter of each slice with center of sham infarct at a location similar to that in the respective infarct group was arbitrarily regarded as sham infarct region, with sham adjacent and remote regions subsequently defined (Fig. 2b,d). Average FA, mean ADC, axial and radial diffusivities were then computed in the infarct, adjacent, and remote regions of 10 slices that covered the infarct. Their correlations with circumferential length of infarct were analyzed. To examine the myocardial fiber double-helical architecture, fibers were classified into three groups based on fiber orientation: left-handed helical fiber (LHF) in epicardium, circumferential fiber (CF) in middle myocardium, and right-handed helical fiber (RHF) in endocardium (18, 19). Their relative percentages were computed in adjacent and remote regions. Since the remote region in two infarct groups involved different anatomical locations, no statistical t-test of fiber architecture at remote region was performed. Sample-based average DTI parameters were calculated for infarct region and correlated with the infarct volume percentage measured at week 13.
Unless otherwise stated, all data are presented as means ± SD. Two-tailed unpaired Student's t-test was employed to examine the myocardial structural alterations, with P < 0.05 regarded as significant.
Figure 3 shows the typical 3D views of the heart samples with infarcts in interventricular septum near apex (Fig. 3a) and lateral wall near base (Fig. 3b). The tracked fibers were overlaid on the 3D views. As expected, the fiber continuity was clearly disrupted in the corresponding infarct region.
LV EF was found to correlate negatively with infarct size at week 13 (Fig. 4) with R = −0.90 and −0.91 for Groups 1 and 2, respectively. Values regarding postinfarct cardiac morphology and function at two timepoints are summarized in Table 1. At week 13, Group 1 was found to have a smaller infarct volume percentage but slightly lower LV EF than Group 2. Between week 4 and week 13, Group 1 exhibited more changes in ESV, LV mass, EF, and infarct volume percentage than Group 2. A significant difference in infarct volume percentages was observed between the two groups at both timepoints.
|Group 1 (n=6)||Group 2 (n=6)||Intergroup t-test|
|Week 13||Week 4||P-value||Week 13||Week 4||P-value||Week 13||Week 4|
|EDV (ml)||82.1 ± 5.1||74.6 ± 16.2||NS||93.5 ± 18.6||86.2 ± 15.1||NS||NS||NS|
|ESV (ml)||46.5 ± 6.9||34.2 ± 12.8||<0.05||49.7 ± 10.2||45.1 ± 10.1||NS||NS||NS|
|SV (ml)||35.6 ± 3.5||40.6 ± 7.3||NS||43.8 ± 13.6||41.1 ± 7.9||NS||NS||NS|
|EF (%)||43.6 ± 5.6||55.4 ± 9.0||<0.01||45.8 ± 7.1||48.2 ± 5.7||NS||NS||NS|
|LV mass (g)||74.5 ± 8.2||66.5 ± 10.8||<0.05||74.9 ± 6.5||73.5 ± 7.5||NS||NS||NS|
|IVP (%)||8.3 ± 1.4||10.0 ± 2.0||NS||13.1 ± 4.4||14.3 ± 5.1||NS||<0.05||<0.05|
Myocardial fiber integrity parameters, ie, FA, mean ADC, axial, and radial diffusivities were calculated for infarct, adjacent, and remote regions. Compared with the respective controls, Group 1 and Group 2 exhibited similar trends of alterations in DTI parameters in the infarct region, where FA values decreased significantly while mean ADC, axial, and radial diffusivities increased substantially (Fig. 5). No significant alteration was observed in adjacent and remote regions compared to corresponding controls (not shown). Between the two infarct groups, there was no statistically significant difference in FA, mean ADC, axial, and radial diffusivities even in the infarct region (Fig. 5). These findings indicated that the myocardial fiber degradation, as described by these DTI parameters, was largely independent of infarct location.
Figure 6 shows the relationship between slice-based infarct size in terms of circumferential infarct length and the DTI parameters in the infarct region. FA values correlated with infarct sizes (Fig. 6a), yielding R = −0.86 (P < 0.001) and −0.84 (P < 0.001) for the two groups, respectively. Positive correlations were observed for mean ADC (R = 0.74, P < 0.001; R = 0.77, P < 0.001 for the two groups, as shown in Fig. 6b), axial diffusivity (R = 0.63, P < 0.001; R = 0.69, P < 0.001 for the two groups in Fig. 6c), and radial diffusivity (R = 0.79, P < 0.001; R = 0.80; P < 0.001 for the two groups in Fig. 6d). The results demonstrated that the larger the infarct size, the more affected the myocardium fiber integrity, indicating that the extent of microstructural degradation within infarct region was associated with infarct size.
Figure 7 shows the percentages of RHF, CF, and LHF in the two infarct groups as compared to the controls in the adjacent and remote regions. In the adjacent region, LHF and CF in Group 1 were significantly different from the corresponding controls (P < 0.01) and Group 2 (P < 0.01) (Fig. 7a). No significant RHF difference between the controls and Group 2 was observed. In Group 2, however, all fiber groups exhibited no substantial change in the adjacent region. In the remote region, only Group 1 exhibited a significant difference in the three groups of fibers as compared to the controls (Fig. 7b). These findings indicated the greater extent of change in fiber orientation distributions in Group 1, ie, alteration of myocardial double-helical architecture depended on infarct location.
Sample-based myocardium fiber structural parameters and their correlations with infarct volume percentage measured at week 13 in the two infarct groups are summarized in Table 2. For both infarct groups, strong correlations were seen between myocardial fiber quality and infarct volume percentage in the infarct region. As for myocardial fiber orientation distribution in the adjacent and remote regions, correlations were only found in Group 1, mostly for LHF and CF. These results again indicated that the degradation of myocardial fiber integrity was primarily related to infarct size, but the alteration of myocardial fiber orientation was dependent on both infarct size and infarct location.
|Group 1 (n=6)||Group 2 (n=6)|
|Infarct region||FA||0.20 ± 0.03||−0.94||**P < 0.01||0.21 ± 0.03||−0.77||**P < 0.01|
|Mean ADC||1.01 ± 0.10||0.83||**P < 0.01||0.98 ± 0.09||0.72||**P < 0.01|
|Axial diffusivity||1.20 ± 0.11||0.70||*P < 0.05||1.19 ± 0.08||0.70||*P < 0.05|
|Radial diffusivity||0.91 ± 0.10||0.86||**P < 0.01||0.87 ± 0.10||0.73||**P < 0.01|
|Adjacent region||LHF||33.3 ± 9.7%||0.95||**P < 0.01||13.3 ± 5.3%||0.31||NS|
|CF||41.6 ± 11.8%||−0.89||**P < 0.01||67.4 ± 6.7%||0.15||NS|
|RHF||25.1 ± 8.8%||−0.22||NS||19.3 ± 5.3%||−0.42||NS|
|Remote region||LHF||24.1 ± 9.1%||0.76||**P < 0.01||25.2 ± 11.0%||0.14||NS|
|CF||55.7 ± 9.2%||−0.61||*P < 0.05||57.1 ± 6.5%||0.07||NS|
|RHF||20.2 ± 6.7%||−0.72||*P < 0.05||17.8 ± 7.1%||−0.27||NS|
Figure 8 illustrates the typical Masson's trichrome-stained views of middle myocardium tissues selected from the control and the two infarct groups around the infarct regions. Red represents muscle fiber, blue/green collagen/fibrosis, and light red cytoplasm. In the infarct groups the necrosed myocardium was replaced by fibrotic tissue that gradually extended from the infarct core into the adjacent region. Compared to the normal tissue, where myocardial fiber bundles were tightly packed, the infarcted tissue exhibited no directional coherence, while myocardium adjacent to infarct typically exhibited substantial fiber tearing. Such formation of fibrosis and fiber tearing was likely the dominant reason underlying the structural degradations as revealed by DTI.
LV remodeling is accompanied by changes in LV geometry and cardiac functions (26) and increased risk of morbidity and mortality (27). Myocardial fiber structure and functions are known to vary regionally (28–30), such as the regional difference of myocardium wall thickness (31) and the nonuniformity of the extensions and stresses across the ventricular wall (32); thus, the postinfarct LV remodeling events may vary for different MI locations.
In the current study the in vivo CMR findings show a similar trend of degradation in cardiac function and morphology in both the LAD-related (Group 1) and LCX-related (Group 2) MI groups. Between the two timepoints examined, EF was seen to continue to decrease more in Group 1 than in Group 2. Group 1 also exhibited more alterations in ESV and LV mass. although it showed a smaller infarct volume percentage. In addition, a slight infarct size decrease and LV mass increase was observed in both groups. Such changes have been attributed to the resorption of infarcted tissue and the hypertrophy of viable myocardium (33, 34).
Besides geometric and functional changes, postinfarct myocardium remodeling also involves microstructural alterations (17, 18, 35). The present study is the first to explore the postinfarct LV myocardium microstructural changes for different MI locations using DTI. The present study demonstrated that the postinfarct changes of these DTI parameters were directly related to infarct size. Compared to the intact controls, FA decreased substantially, while mean ADC, axial, and radial diffusivities increased significantly in the infarct region as reported previously (17, 18, 35). However, no significant difference was found between the two infarct groups. Therefore, our results indicated that the infarct size, not the location, primarily dictated the postinfarct myocardial fiber degradation in the infarct region. Note that FA is a normalized scalar measure of the degree of water molecule diffusion anisotropy within a voxel and reflects the degree of directional coherence of local tissue microstructure (14, 15). An FA decrease implies that the myocardial fiber structure becomes less organized in the infarct region. Axial and radial diffusivities represent the water diffusivities along and perpendicular to myocardial fiber direction. Their increases can indirectly reflect the microstructural changes along these directions. The components in myocardium (such as myocytes, fibroblasts, the extracellular matrix, and the coronary vasculature) and their structural organization are known to alter in the infarct region. The rearrangement of these components in infarct may contribute to the nonzero FA observed in the current study. A large scar leads to more LV remodeling with “stretching” at the adjacent sites and more reorientation of fibers via either cell slippage or changes in the laminar structure (36), likely causing more microsturctural degradation in the infarct region as measured by the FA decrease (Fig. 6a). However, as shown in Fig. 8, the boundary between the infarct and adjacent regions could not be perfectly separated. Thus, it is possible that certain viable myocardium in the border zone might be included in the infarct region in our quantitative analysis. Such a partial volume effect could contribute to the high average FA observed in the infarct region, particularly for small infarct size.
In the present study the LV myocardial double-helical fiber architecture in the noninfarct region was found to shift toward a more left-handed direction as compared to controls, especially in LAD-related MI. This finding implies that the endocardium fiber structure exhibited more alteration (ie, change or loss of longitudinal fiber orientation) and that such a change occurred more in the LAD-related MI group. This may likely arise from the severe loss of subendocardium, where fibers spiral right-handed and are believed to be more vulnerable to ischemia (18). It is known that myocardium deformation has a close relation with local myocardial fiber orientation. For example, the cross-fiber strain related with myocardium wall thickening has been reported to be fiber orientation-dependent (37). Thus, an increase of the LHF percentage at epicardium and decrease of RHF at endocardium may lead to a decrease of overall cross-fiber strain and influence the overall cardiac contraction. The varying extent of fiber structural alterations observed in the present study may explain the earlier finding by others that LAD-related MI was often associated with larger regional differences in intramural deformation than LCX or right coronary artery-related MI (4–6).
A number of limitations existed in this study. First, no baseline values were measured for LV morphology and functions immediately after the coronary artery occlusions. Nevertheless, their alterations at week 4 and a much later point, ie, week 13, still reveal the alteration trend and degree of LV remodeling in these two well-established porcine infarct models. Second, a broad range of infarct sizes may enable a better determination of the influence of infarct size and location. However, it was experimentally difficult to achieve in the current study because the infarct model was achieved by permanently occluding coronary arteries at similar locations, leading to similar infarct sizes in each infarct group.
In conclusion, LV EF negatively correlated with infarct volume percentage in both infarct groups at week 13. Between week 4 and week 13 the infarct volume percentage decreased and the LV mass increased in both infarct groups. More severe cardiac morphological and functional degradation was observed in the LAD-related MI group than in the LCX-related MI group between two timepoints, although the former group exhibited smaller a infarct volume percentage. Ex vivo DTI results revealed that degradation in myocardial fiber integrity measurements (ie, water diffusion FA, mean ADC, axial diffusivity, and radial diffusivity) were similar between the two MI groups, generally showing an FA decrease and diffusivity increases in the infarct region, in strong correlation with infarct size. However, the myocardial fiber orientation changes differed between the two groups. LAD-related MI led to more shift of the double-helical myocardium fiber architecture toward the left-handed direction, ie, coinciding more with the fibers in epicardium, in the noninfarct region. These results demonstrated that postinfarct LV remodeling is accompanied by temporal changes in LV function, morphology, and myocardium fiber architecture, of which both MI size and location are important determinants.
The authors thank Dr. Shuguang Zhu, Department of Medicine, University of Hong Kong, for technical assistance in animal and heart sample preparation.