Targeted Imaging of the Spatial and Temporal Variation of Matrix Metalloproteinase Activity in Porcine Model of Post-Infarct Remodeling: Relationship to Myocardial Dysfunction

Abstract

Background

Matrix metalloproteinases (MMPs) are known to modulate left ventricular (LV) remodeling after a myocardial infarction (MI). However, the temporal and spatial variation of MMP activation and their relationship to mechanical dysfunction post MI remains undefined.

Methods and Results

MI was surgically induced in pigs (n=23) and cine MR and dual isotope hybrid SPECT/CT imaging obtained using thallium-201 (²⁰¹Tl) and a technetium-99m labeled MMP targeted tracer (^99mTc-RP805) at 1, 2 and 4 weeks post MI along with controls (n=5). Regional myocardial strain was computed from MR images and related to MMP zymography and ex vivo myocardial ^99mTc-RP805 retention. MMP activation as assessed by in vivo and ex vivo ^99mTc-RP805 imaging/retention studies was increased nearly 5-fold within the infarct region at 1 week post-MI and remained elevated up to 1 month post-MI. The post-MI change in LV end-diastolic volumes was correlated with MMP activity (y=31.34e^0.48x, p=0.04). MMP activity was increased within the border and remote regions early post-MI, but declined over 1 month. There was a high concordance between regional ^99mTc-RP805 uptake and ex vivo MMP-2 activity.

Conclusions

A novel, multimodality non-invasive hybrid SPECT/CT imaging approach was validated and applied for in vivo evaluation of MMP activation in combination with cine MR analysis of LV deformation. Increased ^99mTc-RP805 retention was seen throughout the heart early post-MI and was not purely a reciprocal of ²⁰¹Tl perfusion. ^99mTc-RP805 SPECT/CT imaging may provide unique information regarding regional myocardial MMP activation and predict late post-MI LV remodeling.

Left ventricular (LV) remodeling following myocardia l infarction (MI) is characterized by infarct expansion, progressive LV dilation, hypertrophy of non-infarct zones, and overall global ventricular remodeling ^1-4. Clinical and animal studies have demonstrated an association between the development of adverse LV remodeling after MI and poor outcomes ^{1, 2, 5, 6}. The changes in anatomic structure or function associated with post-MI LV remodeling are routinely evaluated using imaging modalities including echocardiography, magnetic resonance (MR) imaging, cine x-ray computed tomography (CT), or even conventional nuclear based approaches, like single photon emission computed tomography (SPECT). Although changes in LV size and shape appear to be a direct measurement of remodeling, these indices do not provide information on the underlying molecular processes, which may then be used as surrogate markers of LV remodeling ¹.

A cause-effect relationship has been established between LV remodeling and activation of the matrix metalloproteinases (MMPs) ^7-12. However, defining the spatial and temporal variation of MMP activation post MI in relationship to LV remodeling has been difficult because of the lack of a direct in vivo approach to assess MMP activity. Past studies demonstrated the feasibility of targeted non-invasive imaging with a ^99mTc-labeled MMP-targeted radiotracer (^99mTc-RP805) for the in vivo detection of regional myocardial MMP activation following MI ^{13, 14}. The current study extended these initial findings to a relevant pre-clinical post MI porcine model in order to test the central hypothesis that enhanced MMP activation occurs following MI, and is associated with altered regional myocardial deformation. The rationale for these studies is that defining the relationship between regional MMP activity and LV geometry and mechanics will be critical to better understand the mechanism underlying LV remodeling post-MI. Accordingly, the aims of the current study were two-fold: First, validate an MMP-targeted radiotracer approach for in vivo evaluation of MMP activation post-MI, comparing radiotracer derived indices of regional MMP activation with quantitative MMP zymography, and second, employ hybrid SPECT/CT imaging of MMP activation in combination with MR image derived indices of regional myocardial deformation to relate temporal changes in regional MMP activation with changes in regional myocardial deformation.

Methods

Surgical Preparation

MI was surgically induced in Yorkshire pigs (n = 23; male, 25kg) as previously described ^{9, 10}. Briefly, MI was induced through the ligation of the first two obtuse marginal branches of the circumflex artery, which results in a uniform MI size involving ~22% of the LV in a region positioned on the posterior-lateral wall^{9, 10}. Following a 7-10 day recovery period, pigs were randomly assigned to undergo cine 3-dimensional cardiac MR imaging and in vivo SPECT/CT imaging with euthanasia at 1 week (n= 9), 2 weeks (n = 6) or 4 weeks (n = 8) after MI. An additional group of non-operated controls (n=5) were used for reference imaging and biochemical studies. All experimental protocols were approved by Institutional Animal Care and Use Committees at the Medical University of South Carolina and the Yale University School of Medicine. Studies were performed according to the NIH guidelines for Care and Use of Laboratory Animals (1996).

Imaging Protocol

MR Imaging

Images were acquired on a Siemens Sonata 1.5T magnet using a multiple phase ECG triggered TrueFISP sequence (TR = 47.2ms, TE = 1.58ms, field of view = 340 mm, slice thickness = 5 mm, interslice gap = 1 mm, flip angle = 60°) ¹⁵. After orienting the acquisition plane along both long axes of the heart, contiguous 5 mm thick short axis images were obtained with an end-expiratory breath hold, achieved by transient disconnection of ventilator. The final image resolution was 1.5mm × 1.5mm × 5mm. The raw images were then converted to a 4D DICOM format.

MR Image Analysis

The MR images were semi-automatically segmented from below the membranous septum down to the apex from end diastole to end systole using a modified version of ITK-SNAP ¹⁶ and Bioimagesuite ¹⁷. Local shape properties of the endocardial and epicardial surfaces were used to derive 3-dimensional trajectories for a dense field of points on the surface of the heart over the cardiac cycle. These trajectories were then used to deform a mesh that represented the LV myocardial volume using a continuum bio-mechanics model. This integrated shape-based approach for tracking regional myocardial deformation has been previously validated in both 2- and 3-dimensions ¹⁷.

In vivo SPECT/CT Imaging and reconstruction

SPECT/CT images were acquired (GE Infinia, Hawkeye) ungated with a 360 degree acquisition orbit (32 stops/head, 20 seconds/stop), zoom factor was 1.3, and 128x128 matrix size. In vivo SPECT/CT imaging was performed at 112±14 minutes after intravenous injection of ^99mTc-RP805 (25 mCi), a macrocyclic compound that binds to the catalytic domain on MMPs ¹³, thus being able to provide for localization of regions where MMPs are activated. This was followed by dual isotope imaging 15 minutes after intravenous injection of ²⁰¹Tl (3 mCi). Images were obtained using 15% windows centered on the photopeaks of ²⁰¹Tl (75±5% keV) and ^99mTc (140±5% keV).

SPECT/CT transaxial images were reconstructed as described previously ¹⁴. The ²⁰¹Tl images were used for reorientation of the heart into standard short and long axis views. ²⁰¹Tl and ^99mTc-RP805 images were batch reconstructed to assure exact orientation.

Ex vivo Imaging and Sampling

Immediately after in vivo SPECT/CT imaging, pigs were euthanized and the heart extracted to measure ²⁰¹Tl and ^99mTc radioactivity as previously described ¹⁴. Briefly, the hearts were filled with an alginate material and then cut into 5mm thick short axis slices. One slice containing the infarct region was divided into 8 epicardial and endocardial radial sectors and flash frozen in liquid nitrogen for subsequent MMP zymography. The remaining LV slices were placed on the collimator of a gamma camera (GE Millenium VG) and planar images acquired using the same energy peaks of ²⁰¹Tl and ^99mTc. These LV slices were also divided into 8 epicardial and endocardial radial sectors (16 total sectors) for gamma well counting. All reported values for regional myocardial ²⁰¹Tl and ^99mTc-RP805 retention were derived from gamma well counting of tissue samples. Absolute myocardial uptake was computed as a percentage of the injected dose (ID) per gram (%ID/gm) tissue after correcting for background activity and radioactive decay ¹⁴.The radial segments from the slice used for zymographic analysis were compared with the corresponding segments above and below those analyzed for radiotracer activity.

MMP Zymography, Immunoblotting, and Activity Assay

The radial sections were maintained at −80°C to allow for radioactive decay (~30 days), and then LV extracts prepared for MMP zymographic analysis ^{9, 10, 13, 18}. Briefly, LV homogenates (2 μg protein) were subjected to electrophoresis using gels containing denatured collagen, and the proteolytic bands corresponding to proMMP-9 (92 kDa), proMMP-2 (72 kDa) and the active form of MMP-2 (64 kDa) were digitized. MMP-7 (primary antibody: Ab38996, Abcam) and MT1-MMP (Open Bio systems) levels were det ermined by immunoblotting ^{9, 18}.

A specific global MMP fluorogenic substrate (Enzo Life Sciences BML-P126; excitation 330 nm; emission 405 nm) was utilized to measure MMP activity within the tissue homogenates ^{19, 20}. Briefly, 30 μM of the fluorogenic MMP substrate was incubated with myocardial tissue extract (25 μg) and fluorescence of the cleaved substrate recorded (FLUOstar, BMG Labtech, Durham, NC). The fluorescence signal was converted to MMP activity (ng/g wet weight) based on a calibration curve determined using active MMP-2 catalytic domain (BML-SE237, Enzo Life Sciences) and the initial wet tissue weight of the homogenate.

Data Analysis

End-diastolic volumes were indexed to body mass and values at each post-MI time point were compared using analysis of variance (ANOVA) using time as the independent variable and adjusted for multiple comparisons using Dunnett’s test with the non-MI pigs being designated as the comparison control group. Selected mid-ventricular MR images were manually aligned and registered with the postmortem slices using structural features. The LV radial segments analyzed by zymography were aligned by sector with adjacent segments from the well counting. In order to develop a relationship between perfusion, MMP tracer uptake and MMP zymography, a 2-dimensional histogram was constructed. For each LV section, the MMP tracer values and zymographic values were normalized on a scale from 0-100 and these values assigned to each of the 16 sectors, and displayed in a concentric circle format for the endocardial and epicardial regions using a colorimetric registration. Regional ²⁰¹Tl and ^99mTc-RP805 uptake and MMP expression/activity were subjected to 2-way ANOVA, using time post-MI and LV region as independent factors. Given that the myocardial samples for each region were obtained from the same pigs, an adjustment for a within group effect was performed by including each pig as a random effect variable within the ANOVA design. Post-hoc separation of means was performed using Bonferroni-adjusted pairwise comparison of means (module prcomp, STATA). Measured ^99mTc-RP805 activity was correlated with quantitative estimates of total MMP activity for each MMP subspecies derived from the zymography using standard curve fitting.

Average radial strain was the computed for the infarct, border and remote regions for each animal and compared between groups using ANOVA, which included animal as a random effect variable. Means for each region were analyzed using Welch’s t-test. Statistical analyses were performed using STATA (v8.0, College Station, TX) or BMDP (Statistical Solutions, Saugas, MA). Values are presented as Mean ± SEM. Values of p<0.05 were considered to be statistically significant.

Results

Post-MI Model

Animals demonstrated comparable hemodynamics during the serial MR and SPECT imaging. LV end-diastolic volumes indexed for body mass in the control pigs was 1.43±0.10 mL/kg. Consistent with LV remodeling post-MI ^{9, 10}, LV end-diastolic volume index was increased by 59±8% over control values by 1 week post-MI (2.17±0.12 mL/kg, p<0.05), remained elevated over control values at 2 weeks post-MI (2.09±0.14 mL/kg, p<0.05), and was further increased by 4 weeks post-MI (2.24±0.11 mL/kg, p<0.05 vs. controls).

Dual-Isotope In Vivo SPECT/CT Imaging and Ex Vivo Planar Imaging

In vivo dual isotope (^99mTc-RP805 and ²⁰¹Tl) SPECT-CT images from representative MI pigs at one week, two weeks and four weeks post MI show a ²⁰¹Tl perfusion defect in the posterolateral wall with corresponding focal ^99mTc-RP805 uptake within the perfusion defect (Figure 1A). The maximal ^99mTc-RP805 uptake was observed at 2 week post-MI. Ex vivo planar dual energy imaging of LV slices confirmed the co-localization of the perfusion defect and the region of maximal increase in MMP tracer uptake (Figure 1B), although ^99mTc-RP805 uptake was also seen in remote regions at 1 and 2 weeks post-MI. There was an excellent correspondence between the in vivo and ex vivo images (Figure 1C).

Figure 1.

Dual-isotope in vivo SPECT/CT imaging and ex vivo planar imaging. (A) In vivo ²⁰¹Tl and ^99mTc-RP805 SPECT-CT images of pigs at 1 week, 2 weeks and 4 weeks post MI are shown in transaxial, coronal and sagittal views. Note the perfusion defect in the lateral wall and the time dependent changes in the intensity of ^99mTc-RP805 retention in the same regions (green arrows). The yellow double arrows point to ^99mTc-RP805 activity in the surgical sternal wound. A known point source (orange arrow) can be used to quantify hotspot uptake. Scale bars: 2cm. (B) Postmortem short axis myocardial slices filled with alginate (top row) are shown from a representative pig at one week post-MI. Slices are oriented with the anterior wall on top. Below are the corresponding ex vivo short axis ²⁰¹Tl perfusion images (second row), and ^99mTc-RP805 images (third row), and color coded fused images (²⁰¹Tl – green; ^99mTc-RP805 – red). The infarct can be seen as a thinned fibrotic area in the lateral wall of the four most basal slices. The fusion image clearly demonstrates maximal uptake of ^99mTc-RP805 in the infarct region. Scale bars: 2cm. (C) In vivo SPECT short axis images and ex vivo dual isotope planar slice images from a pig at 2 weeks post MI. There is excellent correlation between the ²⁰¹Tl perfusion defect and the ^99mTc-RP805 retention, which is seen on both in vivo and ex vivo images. Scale bars: 2cm.

Gamma Well Counting of ²⁰¹Tl and ^99mTc-RP805 Radioactivity

Myocardial segments were segregated into infarct, border, and remote regions based on the regional quantitative ²⁰¹Tl activity from well counting as illustrated in Figure 2. Specifically, myocardial segments in which ²⁰¹Tl uptake was lower than 60% of the remote region were designated as the infarct region and the two myocardial segments immediately adjacent to the infarct region (on either side of the infarct region) were designated as the border region. Time-dependent changes in ²⁰¹Tl and ^99mTc-RP805 retention are summarized in Table 1 and the relationship between ²⁰¹Tl and ^99mTc-RP805 activity in the infarct, border, and remote regions is shown in Figure 3. At one and two weeks post-MI, there was increased in ^99mTc-RP805 retention in all myocardial regions, with over a 4-fold increase in ^99mTc-RP805 activity within the infarct region compared to control pigs. By four weeks post MI, ^99mTc-RP805 retention returned to control levels in the remote region, but remained higher than control levels in the infarct and border regions.

Figure 2.

Analysis of ²⁰¹Tl and ^99mTc-RP805 radioactivity. (A) Short axis slice through the infarct region of a pig at 1 week post MI showing sections (numbered) used for determination of ²⁰¹Tl and ^99mTc-RP805 uptake. B- border region, I – infarct region. (B) Ex vivo planar image of the same slice shows increased ^99mTc-RP805 (red) retention in the ²⁰¹Tl perfusion (green) defect in the infarct region located in the thinned inferolateral wall. Scale bar: 2cm. (C) Circumferential count profiles of corresponding well counting data shows increased ^99mTc-RP805 activity in the region of diminished ²⁰¹Tl activity. B- border region (light gray), I – infarct region (dark gray).

Table 1.

Regional and time-dependent changes in gamma well counting following myocardial infarction (MI). All data corrected for tissue weight (mg)

	Control	1 Week MI			2 Week MI			4 Week MI
		Remote	Border	MI	Remote	Border	MI	Remote	Border	MI
Perfusion (201TI, ×10−2)
ENDO	107.3±2.9	110.0 ±1.8	100.7±2.1R	46.9±4.8* RB	109.4±1.9	94.0±5.2* R	42.6±4.7* RB	108.8±3.0	93.5±2.2* R	41.9±3.4* RB
EPI	95.7±2.1	88.9±2.2	83.2±3.1*	40.0±3.3* RB	86.9±2.8	94.6±6.9#	46.7±6.3* RB	94.6±2.4	83.7±2.7* R	38.9±4.4* RB
TOTAL	101.5±2.5	100.0 ±1.4	91.8±2.3* R	43.2±2.3* RB	98.2±2.2	92.3±3.2*	47.2±5.4* RB	101.7±1.4	90.9±1.1* R	40.7±3.9* RB
MMP Tracer (99mTc-RP805, ×10−4)
ENDO	11.3±2.5	13.7±1.3	16.6±2.1	40.0 ±5.3* RB	15.6±1.9	17.9±2.4	52.8±8.5*# RB	6.8±1.5	10.1±1.9	26.1±3.7*#$ RB
EPI	11.6±2.1	14.9±1.6	19.4±2.2	38.6±4.7* RB	17.7±2.1	21.3±1.7	42.5±7.7* RB	7.6±1.5$	10.4±1.8#$	24.5±3.2*#$ RB
TOTAL	11.5±2.3	13.8 ±1.5	18.3±2.0	39.7±4.8* RB	16.7±1.9	21.3±1.7	42.5±7.8* RB	7.2±1.5$	9.8±1.8#$	24.5±2.7*#$ RB
Sample Size (n)	5	9	9	9	6	6	6	8	8	8

Values presented as Mean ± SEM.

^*p<0.05 vs.Control,

^#p<0.05 vs. 1 Week Post MI (Same region),

^$p<0.05 vs. 2 Week Post MI (Same region),

^Rp<0.05 vs. Remote(same week post-MI),

^Bp<0.05 vs. Border (same week post-MI)

Figure 3.

Gamma well counting of ²⁰¹Tl and ^99mTc-RP805 radioactivity. Correlation of the average relative myocardial ²⁰¹Tl activity and ^99mTc-RP805 activity for infarct, border and remote myocardial regions in pigs at 1, 2 and 4 weeks post-MI as determined by gamma well counting. Notice that the relative ²⁰¹Tl perfusion deficit does not change over time, although there are dramatic time dependent changes in ^99mTc-RP805 activity. These data suggest that ^99mTc-RP805 imaging may provide unique information following myocardial infarction. *p<0.05 vs. remote. +p<0.05 vs. Border region. #p<0.05 vs. 1 week post-MI. ^p<0.05 vs. 2 week post-MI.

Comparison of ²⁰¹Tl and ^99mTc-RP805 Activity with MMP Zymography

Changes in the levels of MMP2, MMP-7, MMP-9, and MT1-MMP and MMP activity are summarized in Table 2. There were distinct spatial and temporal patterns in the expression of the various MMP types in the post-MI period. For example, MMP-9 levels within the MI region were elevated in the earlier post-MI time points and then were lower than one week post-MI values at 4 weeks post-MI. MMP-7 and MT1-MMP levels within the border region were elevated at one week post-MI but were similar to control levels by 4 weeks post-MI. MMP activity, which was measured as a function of cleavage of a MMP substrate ²⁰, was increased within the MI region at 1 week post-MI and was further elevated at 4 weeks post-MI. In order to provide a representation of the spatial-temporal changes in MMP targeted radiotracer retention and MMP zymographic levels, the normalized distribution of average values were computed for the endocardial and epicardial regions for each LV sector and are pictorially depicted in Figure 4. A clear spatial and temporal concordance was observed between the perfusion defect, ^99mTc-RP805 retention, and MMP levels, particularly for the active form of MMP-2 (Figure 5). The segment by segment correlation coefficient between the active form of MMP-2 and absolute ^99mTc-RP805 retention was excellent at one week post-MI (r = 0.78, p<0.01), although this relationship continued to declined between two weeks (r = 0.72) and 4 weeks (r = 0.59) post MI. There was an exponential relationship between the post-MI change in LV end-diastolic volume and MMP activity (y=31.34e^0.48x, r: 0.38, p=0.04; Figure 6).

Table 2.

Regional and time-dependent changes in levels of matrix metalloproteinases (MMPs) following myocardial infarction (MI). All data corrected for tissue weight (mg)

	Control	1 Week MI			2 Week MI			4 Week MI
		Remote	Border	MI	Remote	Border	MI	Remote	Border	MI
MMP-9
ENDO	90±17	222 ±107	510 ±135	3543 ±1130* RB	41±11	68±31	2210±854* RB	142±78	344±148	750±465#
EPI	98±40	326±154	574±203	2097±505* RB	42±17	62±18	1704±1111* RB	108±51	208±79	657±238#
TOTAL	94±28	318±144	539±144	2596±596* RB	42±13	200±86	1604±789*# RB	125±63	213±93	617±271#$
MMP-2
ENDO	43±9	107±13	462±129	3190±629* RB	73±30	150±48	2541±651* RB	209±124	735±284	1743±419*# RB
EPI	63±25	132±23	571±217	3387±717* RB	71±9	385±168	2674±1200* RB	131±67	367±100	2550±626* RB
TOTAL	53±16	119±15	535±145	3146±454* RB	72±14	362±99	2275±918* RB	170±95	434±126	2157±389*# RB
MMP-7 (×10-1)
ENDO	1.5±0.2	8.1±3.0*	15.9±9.6*R	12.5±6.8*	6.1±2.5*	9.8±5.1*	8.2±7.4*	3.0±0.9	4.6±1.3	23.5±11.0RB
EPI	1.7±0.2	4.1±2.2	4.5±1.4	3.2±1.3	3.0±1.3	3.9±1.0	5.3±2.4*	2.2±0.7	1.2±0.7#	14.2±8.6*#$ RB
TOTAL	1.6±0.2	5.1±2.9*	9.2±3.7*	6.5±4.5*	4.6±1.9	9.9±3.6*	5.9±3.1*	2.6±0.7	3.0±0.5	14.2±9.1*$ RB
MT1-MMP
ENDO	11.1±1.2	22.5±4.0*	27.1±5.3*	34.0±6.4*R	20.8±5.3*	36.5±7.7*R	36.2±8.5*R	14.0±2.5	16.2±2.8	30.9±10.2*RB
EPI	8.6±0.7	14.0±1.5*	15.0±2.8*	15.7±4.4*	13.6±2.1	17.3±3.2*	25.8±4.7*#R	11.3±1.2	11.1±2.6	12.0±3.5$
TOTAL	9.9±0.7	18.0±2.1*	21.5±3.1*	26.2±4.2*R	17.2±2.9*	28.7±6.1*	30.9±5.3*R	12.6±1.4	13.7±1.9	19.5±5.0*R
MMP Activity (ng/mg/hr, ×10−1)
ENDO	2.3±0.4	5.5 ±0.7	4.9 ±0.9	6.3±0.8*	4.8±0.5	5.1±0.5	6.2±1.1	3.3±0.5	4.7±1.0	10.4±2.0*#$ RB
EPI	2.0±0.3	4.2±0.2	3.9±0.4	5.2±0.6	3.4±0.4	4.3±0.5	7.4±1.3* R	2.9±0.3	3.4±0.4	8.2±1.8*#$ RB
TOTAL	2.2±0.3	4.8±0.4*	4.3±0.6	5.6±0.6*	4.1±0.3	4.8±0.4	6.6±1.0*	3.1±0.2	3.9±0.6	9.3±1.7*#$ RB
Sample Size (n)	5	9	9	9	6	6	6	8	8	8

Values presented as Mean ± SEM.

^*p<0.05 vs.Control,

^#p<0.05 vs. 1 Week Post MI (Same region),

^$p<0.05 vs. 2 Week Post MI (Same region),

^Rp<0.05 vs. Remote(same week post-MI),

^Bp<0.05 vs. Border (same week post-MI)

Figure 4.

Comparison of ²⁰¹Tl and ^99mTc-RP805 Activity with MMP Zymography. Shown are color coded spatial maps of relative myocardial ²⁰¹Tl activity (top row), ^99mTc-RP805 activity (second row), MMP-9 activity assessed by zymography (third row), total MMP-2 activity assessed by zymography (fourth row), MMP-7 levels (fifth row), MT1-MMP levels (sixth row), and MMP activity determined as a function of cleavage of a fluorescent substrate (seventh row). Images are oriented with the lateral wall on the right. Note the time dependent changes in regional ^99mTc-RP805 activity correlate with spatial and temporal changes in MMP levels/activity as assessed by zymography. Reference color bar and an orientation grid are at the bottom.

Figure 5.

Relationship between in vivo and in vitro determination of MMP activity. In vivo MMP activity, determined as retention of ^99mTc-RP805, was significantly related to in vitro changes in levels of active MMP-2 (y = 1687e^0.033x, r=0.89, p<0.05).

Figure 6.

There was a significant relationship between the change in body mass indexed LV end-diastolic volume relative to control values and MMP activity within the MI region (y=31.34e0.48x, r=0.38, p=0.04).

Comparison of ^99mTc-RP805 Retention with Regional Myocardial Strain

The interaction between altered LV mechanics and MMP activation was assessed by comparing regional myocardial ^99mTc-RP805 retention with regional myocardial thickening (radial strain) at different time points post-MI (Figure 7). At one week post MI, there was approximately a 4-fold increase in ^99mTc-RP805 retention in the dyskinetic infarct region compared to referent control values. Although the border region initially demonstrated a similar degree of dysfunction to the infarct region with near zero radial strain, ^99mTc-RP805 retention was much lower than in the central infarct region. There was also increased ^99mTc-RP805 retention in the remote regions of the heart suggesting global activation of MMPs early post-MI associated with the global remodeling process. At two weeks post-MI, the infarct and border regions remained dysfunctional, and again a significant increase in ^99mTc-RP805 retention was seen in all three regions. By four weeks post-MI, regional ^99mTc-RP805 retention remained increased only in the infarct region which remained akinetic/dyskinetic.

Figure 7.

Comparison of ^99mTc-RP805 retention with regional myocardial strain. The correlation of changes in regional ^99mTc-RP805 activity and MR-derived regional radial strain for infarct, border, and remote regions for pigs at 1, 2, and 4 weeks post-MI is illustrated. There were complex time dependent changes in regional myocardial strain and MMP activation. Regional MMP activation appears to be dissociated from regional strain. *p< 0.05 strain vs. control and remote. #p < 0.05 RP805 vs. control. #p < 0.05 RP805 vs. 1 wk infarct.

Discussion

Animal models have demonstrated a clear cause and effect relationship between matrix metalloproteinase (MMP) activation and LV remodeling post MI ^{7-10, 12}. In addition, clear changes in plasma MMP profiles occur in patients post-MI, which were related to indices of post-MI outcomes^21-23. However, plasma profiling is a surrogate marker, and a direct relationship to myocardial MMP activation post-MI and to the LV remodeling process remains to be established. Thus, a critical translational research development would be to develop an approach for noninvasively detecting and quantifying in vivo MMP activity following MI. Accordingly, this study employed a combined molecular-mechanical approach to define the relationship between regional changes in MMP activity and regional mechanics in a well established and clinically relevant porcine model of post-MI remodeling. The significant and unique findings of this study were 2-fold. First, a validated MMP targeted radiotracer was utilized for in vivo evaluation of MMP activation, and allowed for comparing quantitative radiotracer-based indices of regional MMP activation with quantitative MMP zymography. Second, a hybrid SPECT/CT imaging approach for quantifying MMP activation was utilized in combination with MR derived myocardial strains to quantify the temporal and spatial variation of MMP activation with regional myocardial strain during the initial four weeks following MI. This non-invasive multi-modality approach demonstrated that early global activation of MMPs was associated with severe infarct and peri-infarct mechanical dysfunction and significant post-MI remodeling within the first month post-MI. Moreover, this study demonstrated that clinically applicable radiotracer-based imaging can be used to visualize and quantify myocardial MMP activation in the post-MI context.

Spatial and Temporal changes in ^99mTc-RP805 retention in relationship to regional MMP activation

A past study from this laboratory using a rodent model of MI provided proof-of-concept that ^99mTc-RP805 could be used to provide in vivo localization of myocardial MMP activation in the setting of LV remodeling post-MI ¹³. However, it must be recognized that significant differences exist with respect to processes that contribute to LV remodeling post-MI in rodents compared to humans ^{8, 11, 24}. First, the portfolio of MMPs expressed in mice is notably different from that in larger mammalian species, including humans ²⁴. Second, MI sizes in mice are often considerably larger than that in other larger mammalian species ^{7, 8, 12}, which may activate a different set of MMPs during the progression of LV remodeling post-MI. Finally, given the high intrinsic heart rate in mice, it is technically difficult to determine relationships bet ween regional MMP act ivation and mechanical fu nction post-MI. Accordingly, the findings of the present study build upon those of the past rodent study by demonstrating that ^99mTcRP805 may be effectively used to localize MMP activation in a clinically-relevant, large animal MI model, using a conventional clinical SPECT imaging system.

Changes in the myocardial levels of several MMP types occurs with distinct temporal trajectories in the post-MI period ^{24, 25}. For example, MMP-9 levels achieve peak values early in the post-MI period and decline over longer durations ²⁵. In contrast, the increase in MMP-2 levels is sustained over longer post-MI periods ²⁵. Consistent with these past findings, levels of both MMP-2 and MMP-9 were increased within the infarct region at all post-MI time points. However, while the increase in MMP-2 levels was sustained through 4 weeks post-MI, there was a time-dependent decline in MMP-9 levels at the later post-MI time points. This observation is in agreement with a previous study that demonstrated temporal and spatial changes of MMP-2, MMP-8, MMP-13 and MT1-MMP levels in border, remote and infarct regions ¹⁸. In the present study, the greatest retention of the MMP-targeted radiotracer, ^99mTc-RP805, was observed within the infarct region at one week post MI, which was correlated to changes in the levels of active MMP-2, MMP-7, MMP-9, and MT1-MMP at this early post-MI time point. Myocardial ^99mTc-RP805 retention remained elevated within the infarct and border regions at two and four weeks post-MI and correlated with MMP-2 zymographic levels. The significance of these findings is two-fold: First, gamma well counting of ^99mTc-RP805 demonstrates a complex spatial and temporal variation of MMP activity post MI. Second, the correlation between in vivo ^99mTc-RP805 retention and ex vivo measurements of MMP levels provides validation for using targeted radiotracers as a means to longitudinally track fundamental molecular processes that contribute of LV remodeling post-MI.

Relationship between regional deformation and MMP activation

Mechanical dysfunction post MI has been hypothesized to cause increased MMP activation, potentially leading to further infarct expansion and exacerbation of LV dilation post-MI ^{2, 9}. Some in vitro models have shown that fibroblasts grown on mechanically relaxed collagen I produced high levels of MMP-2 and MT1-MMP ²⁶, while others have shown that cyclic stretching of myofibroblasts causes increased MT-MMP activity ²⁷. Altering the mechanical activation pattern of a region of the LV free wall in vivo increased MMP-9 levels, interstitial MMP activity, and increased collagen degradation ^{19, 28}. Rohde et. al. reported that increased wall stress in the border zones was associated with increased MMP-9 activity ⁸. The present study builds upon these previous investigations by directly correlating regional LV strain with MMP activation at selected time points post MI. Specifically, at one week post MI there was a significant correlation between MMP activation and mechanical dysfunction in the infarct and border regions, although there was also MMP activation in remote regions with normal radial strains. In addition, MMP activation became dissociated from radial strain in the border region over time. The effects of mechanical dysfunction on MMP activation post MI are likely multifactorial, although our analysis of the relationship between changes in regional MMP activation and mechanical functioning post-MI was restricted to systolic strain. One may hypothesize that other mechanical events including localized shear forces and diastolic dysfunction may contribute more to MMP activation and LV remodeling. Nevertheless, our results support a complex relationship between regional LV mechanics and MMP activation.

Prediction of LV remodeling

LV remodeling following MI can vary on a patient-to-patient basis in terms of LV dilation and changes in pump function ². Past studies have provided evidence that several factors including, but not limited to, infarct size, the transmurality of injury, extent of myocardial viability, and post MI wall stress are correlated to the degree of LV dilation post-MI ^2-4. Targeted imaging of MMPs, which are central to LV remodeling, provides an attractive approach to follow the process of LV remodeling in vivo, and a potential means to predict late outcome following MI. Our earlier work demonstrated the feasibility of ^99mTc-RP805 for target imaging of MMP activation in vivo in a mouse model of MI ¹³. The present study extends those observations to a more clinically relevant porcine model of transmural infarction and defines the regional spatial and temporal changes in ^99mTc-RP805 retention in relation to changes in myocardial strain and global LV geometry. These dynamic changes in ^99mTc-RP805 retention were not directly associated with specific regional changes in ²⁰¹Tl perfusion or myocardial strain, suggesting the ^99mTc-RP805 imaging might provide unique information about the LV remodeling process following MI. The retention of ^99mTc-RP805 was not purely a reciprocal of ²⁰¹Tl activity, and may provide unique information about regional MMP activation that can be derived in vivo with noninvasive SPECT imaging. Perhaps more importantly, a significant increase in ^99mTc-RP805 retention was observed in the remote regions, which may provide a better predictor of late global remodeling than other estimates of MI size or molecular events within the infarct region.

Experimental limitations

In the current study, the MI was created surgically via a left lateral thoracotomy that resulted in significant chest wall inflammation and increased focal retention of ^99mTc-RP805 in the chest wall immediately over the injured region of the heart, complicating the evaluation of myocardial radiotracer retention. Therefore, clinical ^99mTc-RP805 SPECT/CT images are likely to be of higher quality than in the pigs. In our correlation of ^99mTc-RP805 retention with MMP activity, we were restricted to analyzing radial segments from a single short axis slice of the heart with segments from immediately adjacent slices that were used for gamma well counting of myocardial ^99mTc-RP805 activity. Additional analyses at even earlier time points (one day post MI and three days post MI) or in the setting of ischemia/reperfusion and non-transmural infarction may further help elucidate the factors involved in post infarct LV remodeling. In the present study, changes in LV geometry and MMP activation were determined during terminal examination at each of the designated post-MI time points. Therefore, longitudinal post-MI changes in LV geometry and MMP activation could not be determined. Finally, the dosage of ²⁰¹Tl and ^99mTc-RP805 that were employed were selected based on standard clinical dosing for imaging with our clinical SPECT/CT camera, and ratios of the two radiotracers w selected to optimize in vivo dual isotope imaging and facilitate reliable gamma well counting. However, this dosage would represent approximately 3 times the standard clinical dose used (based on subject weight). These dosages per subject weight could be easily reduced for clinical translation, particularly with application of newer solid state multi-detector imaging technology that provides ~5 times the sensitivity of a conventional camera.

Conclusions

A novel multimodality non-invasive hybrid SPECT/CT imaging approach was validated and applied for in vivo evaluation of MMP activation in combination with cine MR analysis of LV deformation in a clinically relevant porcine model to quantify the temporal and spatial variation of MMP activation and regional myocardial strain during the initial four weeks following MI. Past studies have provided evidence of cause-effect relationships between MMP activation and LV remodeling post-MI ^7-9. Moreover, plasma profiling of MMPs has been suggested to hold prognostic value in terms of predicting LV remodeling post-MI ^21-23. However, plasma levels of MMP can represent spillover from multiple compartments and may not directly reflect net myocardial MMP activity. The present study provides proof-of-concept that direct high quality in vivo imaging of myocardial MMP activation is feasible on a standard clinical hybrid SPECT/CT system and may provide a clinically applicable means to determine the effects of MMP activation within the context of LV remodeling post-MI.

POTENTIAL CLINICAL IMPACT.

Left ventricular (LV) remodeling following myocardial infarction (MI) remains an important cause of morbidity and mortality. While changes in LV geometry can be clinically assessed using a variety of imaging modalities, these images do not address fundamental underlying mechanisms that contribute to LV dilation post-MI. A number of animal studies have demonstrated a causal relationship between LV dilation post-MI and activation of the matrix metalloproteinases (MMPs). In the present study, a technetium-99m labeled compound (^99mTc-RP805) designed to bind to the catalytic domain of a number of MMPs was used to image in vivo MMP activation post-MI. The myocardial spatial distribution and specificity of this MMP targeted compound was validated through direct comparisons with ex vivo immunoblotting and zymography for the determination of regional MMP levels/activity. Regional in vivo MMP activation was also correlated with temporal changes in regional myocardial strain and post-MI remodeling. The post-MI change in LV end-diastolic volumes correlated with MMP activity. High quality in vivo images of ^99mTc-RP805 retention were obtained on a standard clinical hybrid SPECT/CT system, suggesting that this approach holds potential for extension to imaging MMP activation in patients following MI with potential value for risk stratification and directing therapy. Thus, 99mTc-RP805 SPECT/CT imaging may provide unique information regarding regional myocardial MMP activation and predict late post-MI LV remodeling.