Author + information
- Received July 1, 2005
- Revision received August 29, 2005
- Accepted September 8, 2005
- Published online February 7, 2006.
- Toshiyuki Yano, MD⁎,
- Tetsuji Miura, MD, PhD, FACC⁎,⁎ (, )
- Peter Whittaker, PhD†,
- Takayuki Miki, MD, PhD⁎,
- Jun Sakamoto, MD, PhD⁎,
- Yuichi Nakamura, MD⁎,
- Yoshihiko Ichikawa, MD, PhD⁎,
- Yoshihiro Ikeda, MD⁎,
- Hironori Kobayashi, MD⁎,
- Katsuhiko Ohori, MD⁎ and
- Kazuaki Shimamoto, MD, PhD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Tetsuji Miura, Second Department of Internal Medicine, Sapporo Medical University School of Medicine, South-1 West-16, Chuo-ku, Sapporo 060-8543, Japan
Objectives We aimed to determine the effects of macrophage colony-stimulating factor (M-CSF) and granulocyte colony-stimulating factor (G-CSF) treatment on both the repair process and ventricular function after myocardial infarction (MI).
Background The M-CSF and G-CSF have multiple potential effects on cells involved in wound repair.
Methods Myocardial infarction was induced by 45- or 90-min coronary occlusion and reperfusion in rats with or without subsequent injection of M-CSF (106IU/kg/day) or G-CSF (50 μg/kg/day) for five days. We examined histology and messenger ribonucleic acid (mRNA), and assessed left ventricular function in situ using a conductance catheter.
Results Five days after MI, M-CSF increased the number of ED-1–positive cells, mRNA levels of transforming growth factor-β-1, collagen I and III, and collagen fibers within the infarct. Fourteen days after MI, induced by 45-min ischemia, left ventricular end-systolic elastance (Ees) was reduced (1,191 ± 87 mm Hg/ml vs. 1,812 ± 150 mm Hg/ml) and both isovolumic relaxation time constant (τ) (11.9 ± 0.9 ms vs. 8.5 ± 0.4 ms) and left ventricular end-diastolic volume (LVEDV) (0.225 ± 0.014 ml vs. 0.172 ± 0.011 ml) increased versus sham-operated rats. These alterations after MI were attenuated by M-CSF (Ees = 1,650 ± 146, τ = 9.7 ± 0.7, LVEDV = 0.199 ± 0.012) but not by G-CSF. This beneficial effect of M-CSF on Ees was also detected in hearts with MI induced by 90-min ischemia. Furthermore, M-CSF increased collagen content within infarcts and reduced the proportion of thin collagen fibers 14 days after MI. The Ees significantly correlated with infarct collagen content. Nevertheless, neither M-CSF nor G-CSF modified infarct size.
Conclusions The M-CSF treatment attenuates deterioration of left ventricular function after MI by accelerating infarct repair.
Myocardial salvage by reperfusion of ischemic tissue is the primary strategy to prevent heart failure after acute myocardial infarction (MI). Nonetheless, recent studies (1–3) demonstrate that modification of the infarct’s repair process also profoundly influences subsequent ventricular function. One possible manipulation of infarct repair is to mobilize inflammatory cells and bone marrow stem cells by cytokine administration. In fact, granulocyte colony-stimulating factor (G-CSF) administration has been reported to enhance myocyte regeneration and angiogenesis in peri-infarct regions (2–4) and to increase collagen deposition in infarcted areas (3). We also showed that a combination of G-CSF and macrophage colony-stimulating factor (M-CSF) attenuated remodeling of the border zone of infarcted hearts (5). Nevertheless, these earlier studies (2–5), except for one using rabbits (1), employed permanent coronary artery occlusion to induce MI, and hence the effects of G-CSF on reperfused infarcts remain poorly characterized and the influence of M-CSF alone has not been examined.
Thus, because the majority of acute MI patients receive reperfusion therapy, we employed a rat model of ischemia-reperfusion to evaluate the effects of G-CSF and M-CSF on left ventricular (LV) pressure-volume relationships in situ and their association with structural changes within the myocardium.
The study was approved by the Animal Use Committee of Sapporo Medical University.
Surgical preparation and study groups
Male Sprague-Dawley rats (8 to 10 weeks old; 280 to 380 g) were anesthetized with an intraperitoneal injection of ketamine (90 mg/kg) and xylazine (10 mg/kg), intubated, and ventilated (Harvard Apparatus respirator, model 683; South Natick, Massachusetts) with supplemented oxygen. After a left thoracotomy, a snare was placed around the left coronary artery. In the first protocol, rats underwent either sham operation or 45 min of coronary occlusion followed by reperfusion and all rats received antibiotics after surgery (10 mg ampicillin and 10 mg cloxacillin, intramuscularly). Twenty percent of the MI rats died within the first 24 h after the operation (none died thereafter), and surviving rats were randomized to MI, MI + M-CSF, and MI + G-CSF groups 24 h after surgery. In the MI, MI + M-CSF, and MI + G-CSF groups, rats received intraperitoneal injection of saline, human recombinant M-CSF (106IU/kg/day) and G-CSF (50 μg/kg/day), respectively, from day 1 to day 5 after MI. Macrophage colony-stimulating factor and G-CSF were provided by Kyowa Hakko Kogyo Co. (Tokyo, Japan) and Chugai Pharmaceutical Co. (Tokyo, Japan), respectively. These dosing protocols were selected on the basis of our previous study (5), which revealed the ability of combined M-CSF and G-CSF treatment to attenuate border zone remodeling in a permanent occlusion model. Blood samples were collected 5 days after MI, and analysis of cardiac function was performed 14 days after MI. Hearts were excised under anesthesia for histological or messenger ribonucleic acid (mRNA) analyses 5 or 14 days after MI. In the second protocol, rats were assigned to sham operation or 90-min coronary occlusion and reperfusion. Surviving rats were randomized to either the MI or MI + M-CSF groups; M-CSF was administered as in the first protocol, and LV function was assessed 14 days later.
Histological analysis of inflammation and regenerating cells: five days after MI
Pilot experiments revealed that the number of Ki-67–positive cells peaked between three and five days after MI; hence the presence of regenerating cells was investigated five days after MI. Excised hearts were immersion fixed in 10% buffered formalin, and four cross-sectional, 2-mm-thick slices were cut. Histology slides were prepared from each slice. We measured infarct area and the central core of necrosis (Fig. 1)in hematoxylin and eosin (HE)-stained sections using Scion Image (Scion Co., Frederick, Maryland) and summed the area data from each slide for each heart. The number of neutrophils within infarct regions was counted in 10 fields (each 250 × 250 μm; ×400 magnification), and the data were averaged for each heart. The sections were also incubated with antibodies against monocyte/macrophage (ED-1, DAKO, Glostrup, Denmark), osteopontin (OPN, IBL, Takasaki, Japan), Ki-67 (DAKO), a marker of cell proliferation, and α-sarcomeric actin (Zymed, South San Francisco, California). Immunostaining was performed using a peroxidase-based technique with a DAKO EnVision+kit and detected using diaminobenzidine (DAKO). ED1- and OPN-positive cells were counted in 10 fields (250 × 250 μm; ×400 magnification) randomly selected from the central infarct region and from non-infarcted tissue. We also examined sections (four per heart) co-stained with anti-Ki-67 antibodies and anti–α-sarcomeric actin antibodies, fluorescein isothiocyanate-conjugated secondary antibodies and tetramethylrhodamine isothiocyanate-conjugated secondary antibodies (used for detection of Ki-67 and α-sarcomeric actin, respectively) with a confocal laser microscope.
Transforming growth factor (TGF)-β-1 and collagen gene expression: five days after MI
Levels of TGF-β-1, collagen I, collagen III, and glyceraldehydes phosphate dehydrogenase (GAPDH) mRNAs in infarcted and non-infarcted regions were determined as previously described (5). The level of GAPDH mRNA was used to normalize levels of the other target gene mRNAs.
Histological analysis of fibrosis and capillary density: 14 days after MI
After measurement of ventricular function 14 days after MI (see the following text), hearts were fixed and sectioned as described for the 5-day samples. We measured infarct area and LV cross-sectional area in HE-stained sections and expressed infarct size as a percentage of LV area. To calculate infarct transmurality, total LV wall thickness and infarct thickness were measured at the center and both lateral border zones in each slide, and the ratio of infarct thickness to total wall thickness in all slides was averaged for each heart.
Using sections stained with picrosirius red (PSR), areas occupied by collagen (stained red) in infarcted and non-infarcted regions were determined (×40 magnification) using brightfield illumination. Images were converted to red, green, blue format by PhotoShop (Adobe Systems Inc., San Jose, California), and areas positive for collagen staining in both regions were determined after conversion to binary images (Scion Image). The total area of infarcted and non-infarcted regions was also determined to normalize the collagen content measurements. We assessed, in six rats per group, three additional structural parameters in PSR-stained sections viewed with circularly polarized light: collagen fiber color, the retardation of light by collagen, and two-dimensional fiber orientation. As thickness increases, fiber color changes from green to yellow to orange to red; the relative distribution of these different colors was measured in three regions (one from the scar’s center and one from each lateral edge; each ∼0.05 mm2) using image-analysis software (SigmaScan Pro 5.0, Chicago, Illinois) and previously published methodology (6). The retardation (γ) of polarized light is given by the equation γ = t (ne− no), where t is sample thickness and the term in parentheses is the difference in refractive index between the directions parallel and perpendicular to the fiber (7). The latter depends on collagen’s molecular anisotropy, which increases during maturation. Retardation, which has also been shown to increase progressively in scars (8), was measured using an automated analysis system (PolScope, CRI Inc., Woburn, Massachusetts); 50 measurements (locations selected by construction of a 10 × 5 point grid over each image) were taken in each of the three regions and averaged. Fiber orientation was also assessed with the PolScope system and 50 measurements recorded at the same locations as the retardation measurements. The degree of fiber alignment was represented by calculation of the angular deviation of each orientation distribution; the smaller the angular deviation, the more aligned the fibers (7). The accuracy of the automated methods has been validated for both of these parameters (P. Whittaker, unpublished data, 2004).
To determine capillary density, we employed immunostaining with an antibody against Factor VIII (DAKO), an endothelial marker. The number of factor VIII-positive vessels was counted in 10 randomly selected fields (250 × 250 μm; ×400 magnification) in each slide, and the data were averaged for each heart.
In situ ventricular function 14 days after MI
We assessed in situ LV function using a miniature conductance catheter as previously described (9). Rats, anesthetized with pentobarbital sodium (80 mg/kg), were ventilated with oxygen-supplemented room air. A PE-50 catheter was placed in the right carotid artery to monitor pressure. The chest was opened via a midline sternotomy, a conductance catheter with a 2-mm interelectrode distance (Unique Medical, Tokyo, Japan) was inserted into the LV cavity through the apex, and the distal driving electrode and proximal electrode were positioned at the level of the aortic valve and near the apex, respectively. A 3-F catheter-tip micromanometer (Miller Instruments, Houston, Texas) was inserted into the LV through the apex. To obtain LV pressure-volume loops, we used a string-occluder to gradually occlude the ascending aorta until left ventricular end-diastolic volume (LVEDV) slightly increased. The respirator was stopped during data acquisition to avoid respiratory-mediated hemodynamic fluctuation. For measurement of a parallel conductance volume, 0.02 ml of 5% NaCl solution was injected into the pulmonary artery, which transiently changed LV blood conductivity. The LV conductance volume was calculated by subtraction of the parallel conductance volume from the actual measured volume. Finally, blood sample conductivity was measured using a small cuvette. Since hypotension (<50 mm Hg) may cause myocardial hypoperfusion, rats that failed to maintain diastolic pressure above this level were excluded. Integral 3 (Unique Medical, Tokyo, Japan) was used to store data from the conductance catheter and tip-manometer and to calculate the systolic and diastolic functional parameters, including left ventricular end-systolic elastance (Ees). Left ventricular end-systolic elastance was determined from the slope of the LV end-systolic pressure-volume relationships.
All data are presented as means ± SEM. Intergroup differences were examined by one-way analysis of variance (ANOVA), and when the p value for the overall difference was <0.05, pairwise comparisons were performed by Student-Newman-Keuls post hoc test. The p values presented in the Results are those obtained for pairwise comparisons unless otherwise stated. Differences between regression lines were tested by analysis of covariance.
The number of peripheral leukocytes was significantly higher in the MI group than in the sham group (9,443 ± 535/mm3vs. 6,167 ± 921/mm3). A further increase was observed in the MI + G-CSF group (19,300 ± 2,981/mm3, p < 0.05 vs. the MI group), but not in the MI + M-CSF group (10,962 ± 1,046/mm3). The number of red blood cells and platelets were similar in all study groups (data not shown).
Histological analysis five days after MI
Five days after infarction in the control hearts, the peripheral zone of necrotic myocytes had been absorbed and replaced by granulation tissue (Fig. 1A). Collagen content in the infarct was increased from 11.0 ± 1.7% to 17.6 ± 1.7% by M-CSF (p < 0.05), but not by G-CSF (12.2 ± 2.3%). Macrophage colony-stimulating factor increased the number of ED1-positive cells in infarct regions, whereas no significant effect was detected in non-infarct regions (Fig. 1D). OPN-positive macrophages were concentrated around the central core of necrotic myocytes, but were only sporadically present in granulation tissue. These infarct-region OPN-positive cells tended to be increased by M-CSF (100 ± 9/mm2), but not by G-CSF (70 ± 13/mm2) versus MI controls (77 ± 9/mm2); however, the difference was not significant (p = 0.124 for overall ANOVA). The coagulative core necrosis volume did not differ between groups; MI, 13.8 ± 4.2%; MI + M-CSF, 9.0 ± 2.4%; MI + G-CSF, 16.1 ± 3.6%; p = 0.386 for overall ANOVA. Despite significant differences in peripheral neutrophil count, the numbers of neutrophils within the infarcts were similar: MI, 142 ± 18; MI + M-CSF, 123 ± 17; MI + G-CSF, 153 ± 23/mm2. The Ki-67–positive cells found around necrotic myocytes and in the granulation tissue appeared to be macrophages, endothelial cells, and spindle-shaped fibroblasts (Fig. 2Ato 2C). In addition, a few Ki-67–positive cells were co-labeled with anti-α-sarcomeric actin in the MI + G-CSF group (Fig. 2D), but not in the MI and MI + M-CSF groups.
TGF-β-1 and collagen mRNA expression five days after MI
The mRNA levels of TGF-β-1 and collagen I and III were significantly elevated in the infarct (Fig. 3).The TGF-β-1 and collagen III expression in the non-infarct region were similar to the levels in sham-operated hearts. Macrophage colony-stimulating factor enhanced elevation of mRNA levels of TGF-β-1 and collagen I and III in the infarct, but not in non-infarcted tissue. Granulocyte colony-stimulating factor did not significantly increase TGF-β-1, collagen I, or collagen III expression levels in infarct areas.
Histological analysis 14 days after MI
Because of a technical error, one MI control-group heart was improperly fixed and was not included in the histological analysis. Fourteen days after MI, heart mass and heart-to-body mass ratio were significantly larger in the MI group than in the shams, indicating MI-associated ventricular remodeling (Table 1).Most of the infarcted area was composed of granulation and fibrous tissue (Figs. 4Ato 4C), while myocytes in non-infarcted regions appeared hypertrophic. Neither M-CSF nor G-CSF affected heart mass after MI, infarct size, infarct transmurality, or numbers of Factor VIII-positive vessels in infarcted and non-infarcted regions. Macrophage colony-stimulating factor significantly increased collagen, which had replaced granulation tissue, in the infarct region without any effect on collagen level in the non-infarcted areas. We found an inverse correlation between infarct size and infarct collagen content in the MI group (Fig. 5),consistent with our earlier finding that the rate of infarct repair depends upon infarct size (10). The regression line for the MI + M-CSF group was shifted upward versus the MI group, consistent with M-CSF-accelerated collagen deposition (Fig. 5). Granulocyte colony-stimulating factor had no significant effect on collagen content in infarcted and non-infarcted areas or on the infarct size-collagen content relationship.
Myocardial infarction group scars contained a higher proportion of thin, green fibers than was found with M-CSF (23.3 ± 2.4% vs. 13.4 ± 3.6%, p < 0.05). Conversely, although the proportion of thick, orange fibers was higher with M-CSF, the difference was not significant (29.5 ± 6.5% vs. 21.5 ± 4.9% MI group). There were no statistical differences in fiber color between the MI + G-CSF and MI groups. Collagen retardation was higher with M-CSF treatment (61.5 ± 4.9 nm) than in MI controls (45.1 ± 3.5 nm; p < 0.05). This 30% increase is illustrated in the pseudo-color images of representative regions shown in Figure 6;higher values are denoted by yellow, orange, and red, while lower values are indicated by green and blue. In contrast, G-CSF treatment did not increase retardation versus MI control (51.6 ± 3.2 nm; p = NS). There was no significant inter-group difference in angular deviation, indicating no differences in fiber orientation (data not shown).
Ventricular function 14 days after MI
Table 2shows the results of the 45-min occlusion protocol. Heart rate and LV end-systolic and end-diastolic pressures did not differ between groups. The volume axis intercept of end-systolic pressure volume relation (V0) and LVEDV in the MI group were larger than in the sham group. Although this increase in V0was not attenuated by M-CSF or by G-CSF, M-CSF significantly reduced the increase in LVEDV after MI. Furthermore, reduction of Ees after MI was significantly attenuated by M-CSF, an effect not seen with G-CSF (Table 2). Left ventricular end-systolic elastance was positively correlated with infarct collagen content (Fig. 7A)and negatively correlated with infarct transmurality (Fig. 7B). Left ventricular end-systolic elastance tended to negatively correlate with infarct size (y = −16.1x + 1832, r = −0.29), but did not achieve statistical significance. Tau (τ), isovolumic relaxation time constant, significantly increased after MI; however, this change was blunted by M-CSF treatment.
Table 3shows the results of the 90-min occlusion protocol. In these experiments, infarcts were almost all transmural and were considerably larger than in the 45-min protocol. Macrophage colony-stimulating factor attenuated both the elevation of LVEDV and the fall in Ees after MI, although the beneficial effect on τ was not detected. In additional experiments, we confirmed that M-CSF treatment did not affect Ees in normal hearts; Ees and τ were similar in sham-operated rats with and without M-CSF (1,791 ± 305 vs. 1,931 ± 332 mm Hg/ml, 9.3 ± 0.9 vs. 9.5 ± 0.8 ms, n = 4 each group).
Effects of M-CSF on ventricular function after MI
We found that M-CSF treatment attenuated the usual deterioration of left ventricular function that occurs after MI. Furthermore, this benefit was associated with an acceleration of infarct healing; specifically, increased scar collagen content and increased collagen fiber thickness within the scar. Nevertheless, M-CSF failed to reduce infarct transmurality; that is, we found no appreciable muscle regeneration.
In this rat model of ischemia-reperfusion, cardiac output and LVEDP were unchanged in infarcted controls, but LVEDV increased and both Ees, a load-insensitive index of LV contractility, and LV dP/dtmax were substantially reduced versus non-infarcted shams (Tables 2 and 3). These features indicate that our ischemia-reperfusion preparation is a model of compensated LV dysfunction. In our study, M-CSF mitigated deterioration of Ees, a parameter determined by contractility of non-infarcted muscle, infarct transmurality, and collagen content (Fig. 7), presumably via the observed increase in the latter. A plausible explanation for the correlation between Ees and infarct collagen content is that stiffening of the infarcted region by the fibrotic replacement of necrotic muscle increases the Ees of the entire heart (the sum of elastances in infarcted and non-infarcted regions). This explanation is consistent with earlier findings that application of stiff polyester meshes on the infarct during the acute phase of infarction improved ventricular function (11,12). The enhanced stiffness of the infarcted region also explains the smaller τ after MI in the M-CSF–treated versus untreated rats. Although we cannot completely exclude the contribution of factors besides increased collagen content to explain the beneficial effect, a direct M-CSF–mediated effect on viable myocytes appears unlikely because M-CSF did not affect LVESE or τ in sham-operated animals.
The association of increased collagen deposition within the infarct with improved ventricular function (Fig. 7) appears to contradict reported beneficial effects of collagen suppression after MI. Angiotensin-converting enzyme inhibitors and matrix metalloprotease inhibitors suppress both post-infarct ventricular dysfunction and collagen deposition (13,14). Nevertheless, it is likely that cardiac fibrosis leads to different outcomes determined not only by its extent, but also by its location (infarcted vs. viable regions) and timing (acute vs. chronic phase after infarction). A GM-CSF inducer, romurtide, suppressed collagen deposition in the infarcted region during the early phase of MI, resulting in expansion of the infarcted zone (15). Suppression of fibrosis after infarction by neutralization of interleukin-1β (16) or by osteopontin gene knockout (17) actually augmented ventricular remodeling and contractile dysfunction. These findings suggest that timely increases in scar collagen deposition can be beneficial for ventricular function.
Effects of M-CSF on infarct repair
Macrophage colony-stimulating factor increased collagen deposition within the infarct; particularly in large infarcts (Fig. 5), which is potentially advantageous because large infarcts heal more slowly than small infarcts (10). Furthermore, although fiber orientation was unchanged, the observed reduction in the proportion of thin, green fibers within the scars and the increase in retardation are consistent with M-CSF–enhancing healing. For example, increased fiber thickness has been associated with increased scar stiffness (18) and increased retardation with increased scar maturation (8,18).
The mechanism for this M-CSF–mediated enhancement is presumably via enhanced macrophage infiltration (Fig. 1D). Macrophage colony-stimulating factor is a lineage-specific cytokine that promotes the survival, proliferation, and differentiation of mononuclear phagocytes (19). Monocyte migration is also enhanced by M-CSF through enhanced expression of MCP-1 in endothelial cells (20). Macrophages in the infarcted myocardium express c-Fms, an M-CSF receptor (21), and are responsible not only for removal of necrotic myocytes, but also for promoting fibrosis by secreting TGF-β1 (22,23). Significantly higher levels of TGF-β-1 and collagen mRNAs in infarcted regions of M-CSF–treated MI hearts versus those in untreated MI hearts suggest enhanced production of TGF-β-1 and collagen in these regions (Fig. 3). Nonetheless, we cannot exclude the possibility that M-CSF also affected collagen degradation.
At the time we assessed LV function (14 days after MI), fibrosis of control infarcted regions was incomplete. Thus, it is possible that the impact of infarct repair acceleration by M-CSF on LV remodeling and function would be less apparent if assessed later. Nevertheless, accelerated healing could reduce the time when the infarct is vulnerable to mechanically mediated expansion. Another potential confounding factor is that the effects of M-CSF found in the young rats that we used may not necessarily be expected in older subjects in which the function of bone marrow and circulating progenitor cells are reduced.
Effects of G-CSF on ventricular function and infarct repair after MI
The effect of G-CSF on infarcted hearts is controversial. Although initial studies using mice and rabbits (1,4) suggested that G-CSF augmented myocyte regeneration in infarcted hearts by recruiting bone marrow-derived stem cells, this finding was not confirmed in subsequent studies using primates, mice, and rats (2,3,5,24). In these negative studies, using non-reperfused MI, improvement in LV function after MI, which was associated with increased angiogenesis or collagen deposition in infarcted regions, was reported for mice (2) and rats (3,5) but not for primates (24). In our study, G-CSF neither induced substantial myocyte regeneration nor improved LV function in a rat model of reperfused MI. Although our method for detecting myocyte regeneration may have underestimated its contribution, it is unlikely that the amount of such cardiac regeneration, if any, was consequential, because neither infarct size nor infarct transmurality was reduced by G-CSF. These conflicting reports (1–5,24) cannot be readily reconciled; however, differences in protocols for G-CSF treatment, infarct size, severity of heart failure, and animal species might be responsible. In addition, the presence or absence of reperfusion may play a role because reperfusion results in profound differences in the timing (min vs. h) and location (generally throughout the infarct vs. border zones alone) of infiltration of circulating cells.
A close relationship between Ees and infarct transmurality confirmed the importance of myocardial salvage by reperfusion in preservation of ventricular function after myocardial infarction. These data extend and refine, in a more relevant clinical model, our earlier finding (5) that combined M/G-CSF therapy promoted healing and in vitro LV function after permanent coronary occlusion. Furthermore, our current study indicated that favorable modulation of infarct repair and ventricular function after MI can be achieved by M-CSF treatment alone.
Supported by grants #08670812 and #14770322 from Japan Society for the Promotion of Science and Hokkaido Heart Association Grant for Research.
- Abbreviations and Acronyms
- analysis of variance
- left ventricular end-systolic elastance
- end-systolic pressure volume relation
- glyceraldehydes 3-phosphate dehydrogenase
- granulocyte colony-stimulating factor
- hematoxylin and eosin
- left ventricular end-diastolic pressure
- left ventricular end-diastolic volume
- macrophage colony-stimulating factor
- myocardial infarction
- messenger ribonucleic acid
- picrosirius red
- transforming growth factor
- Received July 1, 2005.
- Revision received August 29, 2005.
- Accepted September 8, 2005.
- American College of Cardiology Foundation
- Minatoguchi S.,
- Takemura G.,
- Chen X.H.,
- et al.
- Ohtuka M.,
- Takano H.,
- Zou Y.,
- et al.
- Sugano Y.,
- Anzai T.,
- Yoshikawa T.,
- et al.
- Rich L.,
- Whittaker P.
- Itou T.,
- Takaki M.,
- Yamaguchi H.,
- et al.
- Kelley S.T.,
- Malekan R.,
- Gorman J.H. III.,
- et al.
- Jugdutt B.I.
- Creemers E.,
- Cleutjens J.,
- Smits J.,
- et al.
- Maekawa Y.,
- Anzai T.,
- Yoshikawa T.,
- et al.
- Hwang M.-W.,
- Matsumori A.,
- Furukawa Y.,
- et al.
- Trueblood N.A.,
- Xie Z.,
- Communal C.,
- et al.
- Frangogiannis N.G.,
- Mendoza L.H.,
- Ren G.,
- et al.
- Frangogiannis N.G.,
- Smith C.W.,
- Entman M.L.
- Norol F.,
- Merlet P.,
- Isnard R.,
- et al.