INTRODUCTION

Each year, approximately 565,000 Americans experience a new myocardial infarction; of these, 75% of men and 62% of women survive for at least one year (1). In addition, each year nearly 300,000 Americans experience a recurrent infarction (1). As a result, a large portion of the practice of clinical cardiology is currently devoted to management of patients with a healing or healed myocardial infarct. Excellent progress has been made, particularly in the areas of revascularization during the first hours following infarction (2, 3) and pharmacologic therapy to limit adverse geometric remodeling of the left ventricle (LV) and progression to dilated heart failure (4–6). Even more dramatic therapies are on the horizon. Direct stem cell transplantation into the healing infarct is already in use as an experimental therapy (7–10), and tissue-engineered replacement patches of myocardium may not be far behind (11, 12).

However, these therapies continue to be developed primarily on a trial-and-error basis rather than from an understanding of the mechanical properties of the healing infarct and its coupling to the LV. This trial-and-error approach has led not only to some dramatic successes but also to some catastrophic failures. For example, preliminary evidence that steroid administration limits postinfarction necrosis led to a trial of postinfarction steroid therapy in which high-dose steroid administration caused dramatic increases in infarct size and the incidence of ventricular arrhythmias, and in which 5 of 12 patients in the high-dose group died (13).

This review, therefore, has two primary goals. The first goal is to review what is known about the evolving structure and mechanics of healing myocardial infarcts. The second goal is to temporally correlate structural and mechanical information from a range of studies to formulate hypotheses about which specific structural features are the primary determinants of infarct mechanics during each temporal phase of infarct healing. It is our hope that this new analysis of the temporal course of infarct healing in terms of key structural determinants will stimulate new research on the mechanics of healing infarcts and provide a conceptual platform for improved rational design of postinfarction therapies.

This review focuses on the structure and mechanics of healing infarcts following a single, nonreperfused myocardial infarction, and is organized as follows. First, we outline the different mechanisms by which the presence of a myocardial infarct may impair ventricular function. This list includes many of the potential adverse consequences of myocardial infarction, including rupture, infarct expansion, ventricular remodeling, hypertrophy, and heart failure, the occurrence and severity of which all depend on the mechanical properties of the healing infarct. The next four sections address different temporal phases of healing and each has the same general format: a review of the composition and structure of healing infarcts at that time point, a review of available data on the mechanics of healing infarcts at that time point, hypotheses regarding structural determinants of infarct mechanics based on temporal correlation of the structural and mechanical data, and finally a brief discussion of which of the mechanisms of functional impairment are most relevant at each temporal stage of healing. General conclusions and challenges for future work are addressed in the final section.

IMPACT OF INFARCT MECHANICAL PROPERTIES ON VENTRICULAR FUNCTION

Below we list and briefly explain six different ways in which the presence of a healing myocardial infarct can impair overall pump function of the LV. In each case, the size and mechanical properties of the healing infarct determine the degree of impairment of LV function. Therefore, it follows that an understanding of the mechanical properties of the healing infarct is essential to understanding, predicting, and ultimately modifying the short- and long-term changes in ventricular function that occur following myocardial infarction.

1.	An infarct may fail catastrophically (rupture). Infarct rupture accounts for 15%–30% of deaths in the first week after infarction (14, 15). Rupture obviously represents the most catastrophic way in which the presence of an infarct can impair ventricular function. Although the exact mechanical properties most related to rupture have not been identified, the balance between the mechanical properties of the infarct and the stresses placed on it clearly determines whether rupture occurs (16–18).
2.	Infarct bulging or stretching wastes energy generated by healthy myocardium. Because lost myocardium is replaced by scar tissue rather than by regenerated muscle, clinical studies have shown that once 40% of the LV myocardium has been lost, either through a single large infarction or a combination of smaller ones, the LV is at risk of pump failure (19, 20). Although it is tempting to attribute this finding simply to a reduction in the amount of healthy myocardium contributing to ejection, model studies have found that the degree of systolic impairment is directly related to the compliance of the infarct (Figure 1). For very stiff infarcts, little systolic dysfunction is predicted (21). For compliant infarcts, much of the work of the remaining myocardium is wasted stretching the infarct, reducing systolic pump function dramatically (21–24).
3.	Infarct stiffness may limit diastolic function of the remaining healthy myocardium. Model studies have also shown an important disadvantage to an overly noncompliant infarct. Bogen et al. predicted that whereas compliant infarcts primarily disrupt systolic mechanics, the presence of a large noncompliant infarct severely limits ventricular function by impairing diastolic filling (Figure 1) (21). The presence of the very stiff infarct impairs diastolic function by increasing overall chamber stiffness (25) and limiting the ability of remaining healthy myocardium to utilize the Frank-Starling mechanism to adjust ventricular output (26).
4.	Infarct expansion and cavity dilation increase wall stress throughout the LV. One common postinfarction complication is infarct expansion, a remodeling process characterized by rearrangement of material within the infarct to yield a thinner infarct with increased endocardial surface area (27). This dilatation and thinning clearly increases the wall stress within the infarct at any cavity pressure, potentially worsening problems already mentioned, such as systolic stretching and the risk of rupture. The resulting increase in cavity size also increases wall stress in the remainder of the ventricle, forcing noninfarcted myocardium to generate higher stresses to achieve the same systolic cavity pressure (27, 28).
5.	Coupling to the infarct may limit deformation of adjacent myocardium. The arguments outlined above regarding infarct compliance appear to suggest that in terms of ventricular function, the stiffer the healing infarct the better, except in the limit of a healing infarct large and stiff enough to impair diastolic filling. However, all the reasoning to this point has been one-dimensional (infarcts are either “stiff” or “compliant”) and global (considering two-compartment models with “infarcted” and “normal” segments). In fact, healing infarcts are anisotropic (29, 30) and coupled locally to adjacent noninfarcted myocardium. During acute ischemia, coupling to the compliant infarct creates a “functional border zone” where deformation is reduced despite normal perfusion (31). Later in healing, we have argued that stiff infarcts may also restrict the deformation of adjacent noninfarcted myocardium (30). For example, high circumferential stiffness may limit systolic stretching of the infarct, but high radial stiffness would limit radial thickening of adjacent myocardium tethered to the infarct (30).
6.	The infarct sets boundary conditions for ventricular hypertrophy and remodeling. Over the long term, the presence of an infarct may also impair ventricular function indirectly by triggering adverse ventricular remodeling that increases wall stress throughout the remodeled ventricle. This remodeling has been described as a volume-overload hypertrophy of the surviving myocardium and is characterized by lengthening and thinning of the ventricular wall and overall cavity dilation (32). Although the specific stimuli that drive volume-overload hypertrophy are still incompletely understood (33), in the postinfarction setting the values of most of the likely mechanical candidates (stress, strain, work) in the noninfarcted myocardium, and hence the resulting pattern of hypertrophy and remodeling, are determined largely by the material properties and remodeling of the healing infarct. As with infarct expansion, increases in wall stress associated with cavity dilation place noninfarcted myocardium at a mechanical disadvantage and may lead to a downward spiral into dilated heart failure.

ACUTE ISCHEMIA

During the first minutes to hours after infarction, the balance between oxygen supply and demand is a dynamic one, and the final size of the infarct can be influenced by changes in loading conditions and by pharmacologic agents (34–37). During this period, the mechanics of the infarct region are dominated by the conversion of the infarcted myocardium from an active, force-generating material to a passive, viscoelastic material. Initially, the material properties of the infarct appear to change little; by 6 h after permanent coronary occlusion the infarct clearly begins to stiffen (38, 39). We therefore define acute ischemia from the point of view of infarct mechanics as beginning with the experimental or natural occlusion of the coronary artery supplying the infarct and ending when stiffening becomes evident, 4–6 h after infarction in large animal models. Reperfusion during this period may dramatically alter many or all aspects of the subsequent healing process. Owing to space limitations, we have limited the discussion throughout this review to nonreperfused infarcts, taking this as the simplest starting point for understanding the subsequent effects of a variety of interventions, including reperfusion.

Structural Changes During Acute Ischemia

Excellent descriptive studies of the time course of changes in pathologic appearance have been published for healing rat (40) and human (41, 42) infarcts. Cardiac myocytes are attached by integrins at specific sites near the Z band to an interconnected collagen network containing other mechanically and biologically active extracellular matrix (ECM) components, including glycoproteins, proteoglycans, growth factors, cytokines, and proteases (43–46). During cardiac remodeling and wound healing, any change to this network may alter mechanical properties, including changes within the myocytes, remodeling of myocyte attachments to the ECM (47, 48), changes in ECM content (49), and remodeling of ECM organization and structure (50). In general, postinfarction changes in active myocyte properties and in ECM content have received the most attention, whereas much less attention has been paid to myocyte-ECM coupling, other ECM components, and ECM organization.

Within hours after infarction, the infarcted muscle loses its striations and changes its staining properties (42). Breakdown of matrix-associated glycoproteins has been reported as early as 40 min after infarction and damage to collagen and elastin fibers has been demonstrated 2 h after coronary ligation (51, 52); one study reported a 50% drop in infarct collagen content after 3 h (53). This time course of matrix damage is consistent with the recent finding that matrix metalloproteinase (MMP) activity is significantly increased 1 h after infarction, with measurable release of soluble MMPs after 2 h (54).

Changes in Mechanical Properties During Acute Ischemia

The most important change in mechanical properties in acutely ischemic myocardium is that throughout the first few minutes of ischemia, the myocardium gradually loses its ability to generate systolic force. The ischemic myocardium then behaves as a passive elastic material throughout the cardiac cycle, displaying in-plane stretching and thinning during filling and isovolumic systole, then recoiling passively during ejection and isovolumic relaxation (55–57). The central question with regard to the mechanical properties of acutely ischemic myocardium is whether it simply behaves as passive myocardium or whether its constitutive properties are altered by ischemia. Surprisingly, although acute ischemia has received far more attention in the literature than later phases of healing, it is still not possible to definitively answer this question. As outlined below, there is wide agreement that passive pressure-segment length curves shift rightward within minutes after infarction, so that in-plane lengths at any pressure are greater than control. There is also solid evidence that by several hours after infarction, the infarct region begins to stiffen. However, the relative contributions of changes in local geometry and stresses versus changes in material properties to the reported mechanical behavior in the first hours after infarction are still largely unresolved.

The LV is more compliant than normal 1 h after experimental coronary ligation (58), but becomes less compliant than normal within a few hours after infarction (59, 60). Tracking of segment lengths in the ischemic region using strain gauges and ultrasonic crystals showed that within 30 s after experimental coronary occlusion, systolic shortening of acutely ischemic myocardium is replaced by systolic stretching (55, 56), which gradually increases in magnitude over the first 5 min (56, 61). The passive nature of the ischemic segment deformation was demonstrated convincingly by Tyberg et al., who constructed pressure-length loops throughout the cardiac cycle and showed that the ischemic segments convert from a counterclockwise loop, indicating work being performed by the segment prior to occlusion, to a clockwise loop, indicating work being performed on the segment by adjacent myocardium, 5 min after experimental occlusion (56). Akaishi showed that the ischemic region operates on a highly nonlinear tension-length curve, with the amount of systolic stretching much higher at low end-diastolic pressures (EDP), when the segment starts from a relatively flat part of the curve, than at high EDP, when the segment operates on a very steep portion of the same curve (62).

Many of these early studies also compared diastolic pressure-segment length curves before and after coronary occlusion to assess possible changes in ischemic region compliance. Although all studies agreed that the diastolic pressure-length curves shift rightward (greater segment lengths at a given diastolic pressure) (38, 39, 56, 61, 63), there was disagreement over whether the slope of the pressure-length curves increased (61, 63) or decreased (38, 56) during acute ischemia. There were a number of methodological differences among these studies, including the transmural location and orientation of the segments, the use of closed- or open-chest animals, and the definition of slope directly from the curves versus from log-transformed plots, but on close review none of these factors can completely resolve the discrepancy. In any case, it seems clear that stiffness of the ischemic region begins to increase during the next few hours after infarction. Vokonas et al. reported a gradual decrease in systolic stretching of an ischemic segment beginning 15 min after and continuing throughout the first 6 h following infarction, without concurrent changes in the EDP or end-diastolic segment length (39). Pirzada et al. reported a similar time course for diminishing systolic stretching of the ischemic segment and found that the slope of the diastolic pressure-segment length relationship increased in parallel over the same time period (38). Theroux, working in closed-chest animals at much higher diastolic pressures, saw very little systolic stretching at 5 min or 2 h, but found that the slope of the diastolic pressure-length relationship doubled between these time points (63). Analysis of regional wall motion using echocardiography revealed a slightly different time course in the same animal model. Both the circumferential extent and the severity of regional wall motion abnormalities increased during the first 30 min, then remained stable up to 6 h after coronary ligation (64).

Subsequent two- and three-dimensional analyses of the mechanics of acutely ischemic myocardium have added detail but still have not clearly resolved the question of whether the constitutive properties of ischemic myocardium differ from those of normal passive myocardium. Using a three-dimensional array of implanted markers, Villarreal confirmed that 5 min of experimental ischemia in dog converted the normal pattern of systolic circumferential and longitudinal shortening and radial thickening to circumferential and longitudinal stretching and radial thinning as expected (57). They also found that while the magnitude of normal systolic strains typically increases from epicardium to endocardium, during ischemia the systolic strains were transmurally uniform. An increase in EDP from 2.3 ± 1.5 mm Hg at control to 4.6 ± 1.0 mm Hg after 10 min of ischemia produced a small transmurally uniform stretch (remodeling strain <0.10) in the longitudinal direction but large and transmurally nonuniform circumferential stretch (remodeling strain approximately 0.10 in the epicardial third, 0.30 in the endocardial third of the wall) as well as transmurally nonuniform radial thinning (57). Consistent with the segment length data reviewed above, these findings imply a rightward shift of pressure-strain curves, but indicate a differential response in the circumferential and longitudinal directions.

Without knowledge of changes in local geometry in the infarct region, it is difficult to judge whether the differential responses observed by Villarreal in the circumferential and longitudinal directions reflected differential responses to ischemia, differences in loading, material anisotropy, or a combination of these effects. The best direct data on material anisotropy are those reported by Gupta et al. in specimens tested 4 h after experimental infarction in sheep (Figure 2) (29). They reported stresses at 15% equibiaxial stretch of full-thickness specimens in the control state and 4 h after infarction. Although none of the differences were statistically significant owing to large variability among samples, there were strong trends consistent with the in vivo strains reported by Villarreal. Stresses at 15% equibiaxial stretch were roughly threefold higher in the circumferential direction and fourfold higher in the longitudinal direction in the 4-h infarct samples compared to control; in both control and 4-h specimens, longitudinal stresses were three- to fourfold greater than circumferential stresses (29).

Determinants of Mechanics During Acute Ischemia

As outlined above, acutely ischemic myocardium behaves as a passive nonlinearly viscoelastic material, but it is unknown whether its material properties differ substantially from normal passive myocardium. Most of the reported behavior in the first hour after infarction could be explained by local geometric changes resulting in increased stresses in the infarct at any given cavity pressure, without postulating a change in material properties; this possibility is discussed first below. Then, we briefly consider mechanisms that would act to decrease infarct stiffness: strain softening, disruption of key structural proteins, and loss of coronary perfusion pressure. We omit ischemic contracture because it would shift pressure-segment length curves leftward in contrast to experimental observations and therefore does not appear to dominate the mechanics of acute ischemia. We consider it likely that the stiffening of the infarct reported to begin hours after infarction is due to edema, and therefore take this time as the break point between acute ischemia and the necrotic phase discussed later in this review.

It is clear that acutely ischemic myocardium is stretched in the circumferential and longitudinal directions and thinned in the radial direction at end-diastole compared to preinfarction control (57). Although some of this diastolic remodeling reflects increased EDP during ischemia, diastolic remodeling is more pronounced in the ischemic region than in remote regions of the same heart (39, 61, 65). This disproportionate local stretching and thinning in the ischemic region would be expected to result in higher circumferential and longitudinal stresses in the ischemic region than in remote nonischemic myocardium at any given pressure. Therefore, even if the constitutive properties of the ischemic and nonischemic regions were identical, pressure-segment length curves in the ischemic region would shift rightward, with a given cavity pressure producing an elevated stress and thereby a larger diastolic strain and segment length. This is illustrated for a simple thin-walled sphere in Figure 3, but the basic reasoning should hold for any computational or experimental model. As shown in Figure 3, a 10% increase in radius of curvature and 10% decrease in wall thickness could account for the parallel shift in pressure-length curves reported by Tyberg (56), even if material properties remained constant. Although there is disagreement among various studies regarding whether pressure-segment length slopes increase or decrease during acute ischemia, a review of the actual curves shows that the dominant effect is the rightward shift, with relatively modest changes in slope in either direction. Therefore, the simplest explanation of reported mechanical behavior of acutely ischemic myocardium is that active force generation ceases, the mechanical behavior of the ischemic myocardium is governed throughout the cardiac cycle by the normal passive constitutive properties of myocardium, and local stresses increase modestly at any given pressure owing to local geometric remodeling.

Strain softening describes a dependence of current stiffness on the maximum strain previously experienced by a material and is distinguished from viscoelasticity by a lack of recovery, even following an extended rest period (66). Emery and coworkers have demonstrated strain softening in isolated arrested rat hearts, where pressure-volume and pressure-strain curves show an increase in compliance when the arrested heart is exposed to a new maximum cavity pressure of 30 mm Hg or 120 mm Hg (67, 68). This effect certainly seems relevant to acutely ischemic myocardium, which is exposed to new maximum stresses and stretches during systole once active force generation ceases. The pressure-strain curves published by Emery closely resemble the pressure-dimension curves published by a number of investigators during acute ischemia, with a near-parallel rightward shift at pressures above 10 mm Hg, a decreased slope at lower pressures, and no change in zero-pressure lengths (67). Kirton et al. recently reported that strain softening occurs only in nonviable (i.e., incapable of generating a twitch in response to electrical stimulation) isolated cardiac trabeculae, suggesting that elevated stress and stretch alone are not sufficient to induce softening in myocardium unless other damage has occurred (66). Although this finding does not rule out a role for strain softening in acutely ischemic myocardium, at least two studies suggest that strain softening alone cannot explain observed changes in mechanics during ischemia. First, Summerour et al. could not reproduce the changes in opening angle that occur following 30 min of left coronary occlusion in the rat by inducing global strain-softening in nonischemic rat hearts (69). Second, Paulus et al. demonstrated that strain softening is not required to obtain the right-shifted passive pressure-length curves typical of ischemic myocardium. They induced relative ischemia by pacing tachycardia in dogs with coronary stenoses and found that segments with well-preserved systolic function during ischemia had left-shifted diastolic pressure-segment length curves compared to control, whereas segments with depressed systolic function had right-shifted pressure-segment length curves similar to those observed following coronary occlusion (70). Because systolic stretch was not required to produce a rightward shift of the diastolic pressure-segment length curves, strain softening was not responsible for the shift in this study.

Most of the passive stiffness of normal myocardium appears to reside in two structural proteins: titin determines stiffness at lower sarcomere lengths, whereas collagen is the primary determinant at the higher end of the working sarcomere length range (71). Therefore, disruption of either of these proteins during acute ischemia could result in changes in mechanics of the ischemic region. Titin is a particularly appealing candidate because increased compliance at low stresses (owing to titin disruption) with preserved properties at higher stresses (owing to intact collagen) would appear as a rightward shift in pressure-segment length curves at the relatively high end-diastolic pressures typical of acute ischemia. However, structural studies identifying damage to the myocardial collagen network early in ischemia suggest that collagen disruption may also play a role. Support for this idea comes from a study by MacKenna et al., in which bacterial collagenase treatment of perfused isolated arrested rat LVs caused a rightward shift in passive pressure-volume and pressure-strain curves (72). MacKenna's pressure-strain data resembled data from acute ischemia in that the rightward shifts occurred with little change in slope and the largest shifts were in circumferential strain. However, the changes in this study were faster than collagenase normally degrades collagen, and this experimental preparation becomes rapidly edematous, so it is difficult to separate changes owing to disruption of collagen or collagen-myocyte attachments from changes owing to edema. Although a discussion of myocardial stunning is beyond the scope of this review, experiments on stunning have also provided evidence linking collagen damage and increased compliance of passive myocardium (73, 74).

Perfusion of isolated arrested hearts is associated with decreased LV compliance and left-shifted passive pressure-strain curves compared to the unperfused state, raising the possibility that the increased LV compliance and rightward shift in pressure-dimension curves reported during acute ischemia could be explained, in part, by a loss of coronary perfusion pressure in the occluded vessel. Allaart et al. found that perfusion increased axial stiffness and unstressed length in papillary muscles owing to an increase in axial stiffness of the perfused blood vessels (75). From their data, loss of perfusion might be expected to decrease both stiffness and unstressed length, but changes in unstressed length have not been reported during acute ischemia. May-Newman and coworkers found that perfusion decreased longitudinal, cross-fiber, and radial strains during passive inflation of isolated arrested hearts and increased local tissue volume, especially at the endocardium (76). Because circumferential and fiber strains were not significantly altered by perfusion, loss of perfusion would not completely explain reported data for ischemic myocardium, where circumferential remodeling is prominent. However, the large radial changes reported by May-Newman could account for all of the thinning reported by Villarreal in acutely ischemic myocardium (57), and thereby for rightward shifting of pressure-dimension curves through locally increased stresses.

Ventricular Function During Acute Ischemia

Three of the mechanisms by which the presence of an infarct depresses LV function are relevant to acute ischemia: energy loss through stretching of the infarct (mechanism 2), elevated wall stresses owing to infarct and LV dilation (mechanism 4) and impaired function of adjacent myocardium owing to physical coupling with the infarct (mechanism 5). Systolic stretching of the ischemic region is apparent experimentally as a parallel rightward shift of the end-systolic pressure-volume relationship (ESPVR) (77), which can be explained using simple compartmental (22, 77) or spherical membrane (21) models (Figure 1). The key to the response is the exponential passive stress-strain behavior of ischemic myocardium. Although the ischemic region may be relatively extensible at low pressures, at the much higher pressures and wall stresses typical of end-systole, the ischemic region is stretched onto a portion of its stress-strain curve so steep it is essentially inextensible. Compared to normally activated systolic myocardium (which has contracted rather than stretched relative to its end-diastolic configuration), the ischemic region therefore contains roughly the same volume of extra blood at any physiologic systolic pressure, accounting for the rightward shift of the ESPVR. Systolic stretching of the ischemic region also depresses global ventricular function through a second mechanism not explicitly incorporated in simple compartmental models. Reduced systolic ejection eventually leads to a new steady state in which systolic and diastolic volumes are increased relative to control and ejection fraction is depressed, in other words, to global ventricular dilation (78). Dilation places the noninfarcted myocardium at a mechanical disadvantage, with higher systolic stresses required to eject against a given pressure.

In addition to the impact of systolic stretching of the infarct region, studies of regional function during acute ischemia have indicated that the extent of regional dysfunction extends beyond the region of reduced blood flow, creating a functional border zone (31). Recently, a combination of modeling and experimental studies have shown that border zone dysfunction can be explained by physical coupling to the ischemic region (79) and elevated border zone stresses (80–83), without postulating reduced contractility.

THE NECROTIC PHASE

During the first few days after infarction, the dominant pathologic processes are inflammation and necrosis. We define the necrotic phase as beginning within a few hours, when the infarct begins to stiffen, and ending when the number of fibroblasts and amount of new collagen begin to increase rapidly in the healing infarct [approximately 7 days after infarction in the human (41) and 5 days after infarction in the rat (40) (Figure 4)]. Infarct rupture is most common during this period (14, 15). Given that the infarcted muscle is dead and undergoing necrosis, and significant new collagen has not yet been deposited, it is perhaps surprising that every infarct does not rupture during this phase. Infarct mechanics during this critical period are still poorly understood. In this section, we attempt to identify structural features responsible for infarct mechanical properties and maintenance of infarct integrity during the necrotic phase.

Structural Changes During the Necrotic Phase

Within hours after infarction, the infarcted muscle loses its striations and changes its staining properties (42). Within 24 h, 94% of human infarcts have wavy fibers, indicating intercellular edema, and 90% have clear necrosis characterized by altered staining properties and nuclear pyknosis or karyolysis (41). By the fourth or fifth day, removal of dead muscle is clearly observed (41, 42). Collagenase and gelatinase activity of MMP-1, MMP-2, and MMP-9 is elevated during the necrotic phase of infarct healing (84, 85), and disruption of the collagen network continues. During the first 4 days after infarction in rats, there is a progressive decrease in the number of normally birefringent collagen fibers, and by 4 days there is a significant reduction in the number of collagen struts that laterally connect myocytes (86). As the necrotic phase concludes, deposition of new ECM components begins, forming a scaffold for the deposition of new collagen. Fibronectin (87, 88), laminin (89), and collagen type IV (89) all appear at 3–4 days in the healing rat infarct, approximately the same time that mRNA for type III (first) and I (slightly later) procollagens is first detected (90).

Changes in Mechanical Properties During the Necrotic Phase

Two changes in mechanics are apparent in the necrotic infarct. First, circumferential and longitudinal stiffness increase under multiaxial loading, whereas uniaxial tests show no change in stiffness, suggesting increased mechanical coupling between the two directions. Second, unstressed segment length increases, at least in the circumferential direction, whereas end-diastolic length does not. The net effect is an increase of segment lengths below end-diastolic pressure but a decrease in segment lengths at higher pressures.

Theroux et al. tracked the distance between pairs of circumferentially oriented sonomicrometers implanted in the subendocardium over 4 weeks following experimental infarction in dogs (63). Circumferential segmental shortening remained approximately zero in the infarct throughout the first week (0% at 1 day, 1.9 ± 0.1% at 1 week). EDP and segment length were unchanged from preinfarction control at 1 day and 1 week. However, the slope of the diastolic pressure-length relationship during filling was increased more than fivefold at 1 day, 2 days, and 1 week. Hood reported a similarly dramatic increase in the slope of the diastolic pressure-circumferential segment length relation and an increase in unstressed length in 5-day-old canine infarcts both in vivo and in isolated arrested hearts (91), whereas Lima et al. found reduced systolic principal strains in 1-week-old ovine infarcts using MRI tagging (92). Because mild thinning of the infarcted region typically occurs over the first week, stresses in the infarcts were likely similar to or slightly greater than in control regions at a given diastolic pressure. Therefore, these studies imply increased stiffness and unstressed length in healing myocardial infarcts throughout the necrotic phase.

By contrast, uniaxial tests of strips of healing infarct tissue have consistently indicated that infarct material properties do not change during the necrotic phase of healing. Laird et al. studied uniaxial strips cut from the midwall of infarcted rabbit hearts along the original myofiber direction (nearly circumferential) and found no change in stiffness during the 10 days following infarction (24). In a more detailed study at a single time point, Przyklenk et al. tested longitudinally oriented strips cut from several transmural layers of normal canine myocardium and 24-h-old infarcts. They found no differences between normal and necrotic myocardium in stiffness, tensile strength, or strain at rupture (93).

Only a single report of biaxial mechanical testing of healing infarct tissue is currently available, and the results agree better with in vivo studies. Gupta and coworkers measured circumferential and longitudinal stresses at 15% equibiaxial stretch in healing anterior ovine infarcts during each of the phases of infarct healing outlined in this review (29). At 1 week, although collagen content had increased by less than twofold, longitudinal stress at 15% equibiaxial stretch reached its peak value for the entire time course studied, roughly six times control values (Figure 2). Circumferential stress at 15% equibiaxial stretch was also increased at 1 week to more than eight times its control value, although it did not peak until 2 weeks. Although the use of only a single test protocol limits the interpretation of their data somewhat, their equibiaxial stretch data, like the in situ pressure-length curves, suggest a several-fold increase in infarct stiffness at 1 week, before the bulk of new collagen deposition occurs.

Determinants of Infarct Mechanics During the Necrotic Phase

Unfortunately, very little direct information is available regarding the determinants of mechanical properties during the necrotic phase. Most evidence is either correlative, relating pathologic observations to functional ones, or circumstantial, derived from the outcome of various experimental interventions. In addition, most of the evidence relates to the prevention or aggravation of infarct expansion. Although the degree of infarct expansion likely depends on infarct material properties, the process is not sufficiently well understood to draw conclusions about specific properties, such as infarct stiffness or tensile strength, from data on expansion. In spite of these difficulties, the evidence reviewed below strongly suggests that interstitial edema is responsible for reported increases in infarct stiffness during the necrotic phase, whereas infarct expansion is the most likely basis for the reported increase in unstressed dimensions of the necrotic infarct.

The two primary structural proteins in passive myocardium, titin and collagen, both undergo degradation during the necrotic phase. Necrotic myocytes lose their striations within hours (42), reflecting damage to the major myofibrillar proteins that compose the sarcomere, including titin. Progressive damage to collagen is also seen in the first days following infarction. Although the impact of this damage on infarct material properties has not been studied directly, the degree of damage correlates with the degree of infarct expansion (86), and selective MMP inhibition limits infarct expansion (94). If titin or collagen normally bear some tension in the stress-free state, their degradation could produce the increase in unstressed segment length reported in necrotic infarcts, but would not explain the reported increase in infarct stiffness.

Several lines of evidence support the idea that interstitial edema increases myocardial stiffness. First, studies of “iatrogenic” edema associated with cardioplegia have shown that experimentally induced global edema decreases ventricular compliance, but have not consistently found changes in unstressed chamber volume (95, 96). Second, studies of the role of edema in postischemic reperfusion injury have shown that interstitial edema increases stiffness in the ischemic region. For example, reperfusion following experimental global ischemia increased ventricular water content and diastolic pressure at a fixed volume, whereas reperfusion with a hypertonic solution returned water content to normal and diastolic pressure toward normal (97).

Although these studies demonstrate that edema could increase stiffness in necrotic myocardium, the evidence that edema actually does this in necrotic infarcts is more circumstantial. A recent MRI study by Gerber et al. found that reperfused experimental infarcts with high levels of microvascular obstruction (MO) showed less systolic stretching at 48 h postinfarction than infarcts with low levels of MO (98). Another interesting finding was that the high-MO infarcts appeared to be not only stiffer but also more isotropic than low-MO infarcts. However, although high-MO infarcts would likely have more intramyocardial hemorrhage and edema than low-MO infarcts, the degree of infarct edema was not directly verified in this study. Other studies have indicated that infarct water content is significantly increased several days after infarction, even in the absence of reperfusion (99).

The final line of evidence that edema is an important determinant of mechanical properties in the necrotic infarct is that a variety of pharmacologic agents that reduce edema and inflammation, including high-dose steroids (100, 101), ibuprofen (102), and indomethacin (103, 104), also aggravate infarct expansion in the first days following experimental infarction. One of the best of these studies, by Mannisi et al., showed that water content was significantly increased in the infarct region at 24 h in rats, steroids prevented this water increase, and steroids did not change infarct size or the prevalence of expansion but did increase the extent of infarct expansion when it occurred (101). Although the relationship between infarct material properties and infarct expansion is not well understood, these studies suggest that edema reinforces the necrotic infarct against expansion by increasing stiffness and/or tensile strength, and antiinflammatory agents promote expansion by reducing edema.

Infarct expansion does not occur in all infarcts, but when it occurs it should change two aspects of infarct mechanics. First, a dilated, thinned infarct will have higher wall stresses at any cavity pressure. This should not change the true stiffness of the infarct, but it could act to decrease the apparent stiffness of the infarct as assessed by pressure-dimension data. This is likely at most a minor effect because pressure-dimension studies consistently show apparent stiffening of the infarct. Although infarct expansion cannot explain reported increases in necrotic infarct stiffness, it is an excellent candidate for explaining the increase in unstressed segment length. Expansion proceeds by rearrangement of existing material, with slippage of bundles of cells relative to one another in planes parallel to the epicardium the primary mechanism (105). Grossly, the endocardial and epicardial area of the infarct increase and the wall thickness decreases (27, 65, 106–108). Thus, the unloaded length of a circumferential or longitudinal segment under long-term monitoring would increase as expansion proceeds. In an in vitro model of this process, Connelly and coworkers showed that during cyclic stretching of strips of 1-day rabbit infarcts, elevated peak stress induced a 4%–5% per hour permanent increase in unstressed length (109).

Ventricular Function During the Necrotic Phase

Of the mechanisms for depression of ventricular function discussed at the outset, the most concerning during the necrotic phase is infarct rupture (mechanism 1). Nearly all infarct ruptures occur during this phase (within the first week in humans) (14, 15). Otherwise, the determinants of function are similar to those discussed for acute ischemia. Moderate stiffening of the infarct through the necrotic phase increasingly limits detrimental infarct stretching (mechanism 2) (110). However, continued infarct expansion and ventricular dilation, if present, will exacerbate both local and global increases in wall stress (mechanism 5). Whether ventricular function improves or worsens will be determined by a balance between these effects. For example, Kumar showed that ventricular function curves were consistently depressed during acute ischemia in dogs but recovered substantially at 1 week, allowing the dogs to maintain control levels of cardiac output at end-diastolic pressures that were greater than control but less than those observed during acute ischemia (111).

THE FIBROTIC PHASE

Although pathologic signs of necrosis persist for weeks to months in humans (41, 42) and large animals (112), the healing infarct soon enters a phase where new collagen deposition is the primary determinant of structural and mechanical changes. Collagen content increases rapidly (29, 113, 114) and infarct stiffness roughly correlates with collagen content (29). We define the fibrotic phase as beginning when the number of fibroblasts and amount of new collagen begin to increase rapidly in the healing infarct [approximately 7 days after infarction in the human (41) and 5 days after infarction in the rat (40) (Figure 4)], and ending when collagen accumulation slows and mechanical properties decouple from collagen content. This occurs at approximately 3 weeks in large animal models (29), presumably earlier in the rat and later in humans (Figure 4).

Structural Changes During the Fibrotic Phase

Collagen content increases steadily from 1 to 6 weeks after experimental infarction in dogs (112, 114) and sheep (29). Qualitative observations at autopsy indicate a similar time course for human myocardial infarction (41, 42). In rats, the collagen content begins rising on day 4 or 5 (40) and continues to increase for at least 3 weeks (40, 115). The healing infarct contains a mixture of collagen types I, III, and other minor subtypes (115), and Whittaker et al. have suggested that an initial mesh of type III collagen forms the scaffold for subsequent deposition of large, highly aligned type I collagen fibers (116). By 3 weeks after infarction in pig, the scar is dominated by large type I collagen fibers highly aligned with one another in each transmural layer (117). The mean orientation of the collagen fibers varies with depth below the epicardium in a pattern similar to that for normal muscle fibers except that the transmural range of mean angles is smaller (30). The net result of this pattern is that the majority of large collagen fibers in the scar are oriented within 30° of the circumferential direction (118). A similar pattern of collagen fiber alignment has been reported 2 weeks after infarction in rat (119) and at 6 weeks in dog (116).

Changes in Mechanical Properties During the Fibrotic Phase

Only a few studies have evaluated mechanics during this phase of healing. The available evidence suggests that during this phase infarct stiffness peaks and the healing infarct acquires a distinctive anisotropy. Theroux reported that segment lengths changed only approximately 2% over the cardiac cycle at 1, 2, and 3 weeks after infarction in dogs, suggesting high stiffness in the healing infarcts (63). The slope of the passive pressure-segment length relationship confirmed elevated stiffness, varying from six to nine times control values depending on EDP (63). Gibbons et al. found that the circumferential extent of abnormal wall motion peaked 48 h after infarction in the dog and then decreased over the next 6 weeks (120). When we studied the three-dimensional mechanics of healing porcine infarcts, we also found that systolic strains were not different from zero at 1 week, consistent with elevated stiffness (121). However, although circumferential stretching remained minimal at 3 weeks, significant passive longitudinal shortening and radial thickening returned, suggesting developing mechanical anisotropy in the healing scar.

Connelly and Lerman both reported that uniaxial tensile strength of excised strips of 1-week-old rabbit myocardial scar tissue was roughly double that for control myocardium, but did not report stiffness values at this time point (113, 122). Gupta et al. performed equibiaxial mechanical tests of excised ovine scar tissue and found that stress at 15% equibiaxial stretch peaked at 1 week in the longitudinal direction at a value 6 times control and at 2 weeks in the circumferential direction at a value 16 times control (Figure 2) (29). Although longitudinal stresses at 15% equibiaxial stretch remained roughly twice circumferential stresses at all time points for noninfarcted myocardium, the healing scar switched from stiffer in the longitudinal direction through the first week to stiffer in the circumferential direction beyond the second week (29). We found similar anisotropy in 3-week-old porcine infarcts, which displayed little circumferential stretch in the healing scar during passive inflation of isolated arrested hearts over a physiologic range of cavity pressures, but roughly 50% greater longitudinal stretch at any pressure as remote noninfarcted myocardium (Figure 5) (123). By contrast, Omens et al. found a greater reduction in longitudinal than in circumferential epicardial strains in the scar during passive inflation of isolated arrested hearts 2 weeks after infarction in rat (119). They also found that collagen fibers in the scar straightened more rapidly with pressure but were not straighter in the unloaded state than collagen fibers in normal myocardium.

Determinants of Infarct Mechanics During the Fibrotic Phase

In this review, we define the fibrotic phase as the phase of healing dominated by new collagen deposition. During this phase, both the amount of collagen and the three-dimensional structure of the collagen fibers are important determinants of infarct mechanics. Other matrix components may also be important, but there is not yet enough information to assess their role relative to collagen.

Because infarct stiffness and collagen content increase in parallel during the fibrotic phase, it seems obvious that collagen content is one primary determinant of the mechanical properties of the healing infarct during this phase. However, the effects of alterations in collagen content and subtype ratios on scar mechanics have not been systematically studied. Lerman found that passive stiffness of the rabbit LV correlated with hydroxyproline content over the first week after infarction (113), as would be expected if hydroxyproline content correlates with stiffness of the healing infarct (21, 26).

The finding that myocardial infarcts are highly anisotropic during the fibrotic phase of healing (29, 30) implicates the highly aligned large collagen fiber structure as the second primary determinant of infarct properties during this phase. The predominance of large collagen fibers oriented in the circumferential direction (116, 117) is consistent with reports that myocardial scar is stiffer in the circumferential direction in most animal models (29, 30). However, more work is needed, particularly in the development of structural constitutive models for myocardial scar tissue (118).

Ventricular Function During the Fibrotic Phase

The two mechanisms by which infarcts in the fibrotic phase of healing may depress LV function are impaired filling owing to elevated chamber stiffness (mechanism 3) and impaired systolic function of adjacent noninfarcted myocardium owing to coupling with the infarct (mechanism 5). Janz (26) and Bogen (21) both predicted that the primary adverse effect of a very stiff infarct would be impaired filling owing to decreased LV compliance. Janz also suggested that diastolic stretch of adjacent noninfarcted myocardium would be limited during filling by tethering to the very stiff infarct, reducing systolic function via the Frank-Starling mechanism (26). We have proposed that tethering of adjacent noninfarcted myocardium to a stiff isotropic infarct would directly retard both systolic shortening parallel to the infarct border and radial thickening (30). The anisotropy we observed in 3-week-old porcine scars oriented longitudinally on the LV appears to minimize this effect: high circumferential stiffness prevents stretching of the infarct perpendicular to its border, whereas low longitudinal and radial stiffness allow the scar to deform compatibly with adjacent myocardium in these directions (30). Evidence in support of this hypothesis includes the fact that longitudinal shortening and wall thickening in the healing infarct disappear at 1 week in this animal model (when the infarct is stiff and isotropic) then reappear at 3 weeks (once infarct anisotropy is established), despite the absence of viable myocardium. Another consistent observation is that, in the study by Gerber et al. discussed above, high-MO infarcts, which appeared to be stiffer and more isotropic, reduced wall thickening in adjacent noninfarcted myocardium much more than low-MO infarcts (98).

THE REMODELING PHASE

As healing continues, the mechanical properties of the infarct decouple from collagen content. Collagen content may continue to rise for several weeks while infarct stiffness drops (29), suggesting that other factors now dominate the mechanics. We term this phase the remodeling phase, and although its onset can be defined, the healing scar tissue is a dynamic, biologically active tissue that may never reach a stable, mature configuration (healed as opposed to healing) that could be taken to mark the end of remodeling (124).

Structural Remodeling of Myocardial Scar Tissue

Remodeling of the myocardial scar occurs during this phase at both the gross and microscopic levels. On the gross level, the dominant effect is shrinkage of the scar to occupy a reduced percentage of the LV wall. In canine models, direct topographic measurements indicated a 40% shrinkage of the infarct over 6 weeks (114), whereas condensation of microspheres indicated 30% to more than 70% shrinkage (99, 125), depending on infarct size and location. At the microscopic level, the rise in collagen content slows but cross-linking continues to increase. After a tenfold increase in the first 4 weeks, Jugdutt found that hydroxyproline increased only an additional 20% from week 4 to week 6 in dogs (114). Vivaldi found a 50% increase in collagen content between 2 and 4 weeks in the rat compared to a doubling of cross-link concentration over the same period (115). Data from McCormick et al. at 13 weeks in the rat showed the same collagen content and another 50% increase in cross-linking compared to Vivaldi's 4-week data (126). Qualitative changes in collagen have also been reported. Whittaker found continued increases in molecular organization as assessed by optical retardation for at least 6 weeks in a canine model (116).

Changes in Mechanical Properties During the Remodeling Phase

There is some disagreement in the literature regarding changes in scar mechanical properties during the remodeling phase. Although Parmley found that strips of fibrous human aneurysms tested uniaxially months to years after infarction were many times stiffer than muscular aneurysms from patients who died days after infarction (23), Connelly reported only moderately (twofold) increased stiffness in samples from 3-week-old rabbit scar tissue compared to noninfarcted myocardium (122). Scar anisotropy may largely explain these differences; in 6-week-old ovine scar stretched equibiaxially by 15%, Gupta et al. reported longitudinal stresses identical to those in control myocardium, whereas circumferential stresses were tenfold greater in the scar (Figure 2) (29).

Determinants of Infarct Mechanics During the Remodeling Phase

During the early part of the remodeling phase, infarct stiffness decreases while collagen content continues to increase, indicating that collagen content and fiber structure are no longer the only important determinants of the mechanical properties of the healing infarct. Structural changes during this phase of healing have received much less attention, but one factor that does appear to correlate with infarct mechanics late in healing is the degree of cross-linking.

Connelly et al. compared uniaxial mechanics of strips of rabbit scar tissue 3 weeks after infarction in rabbits, with or without reperfusion. Late reperfusion (3 h after infarction) did not change scar collagen content or stiffness, but it did reduce cross-link density and tensile strength, suggesting that cross-linking can influence the mechanics of healing scar tissue. Similar findings have been reported in healing rabbit ligament, where reduced crosslink density is associated with reduced failure strength despite normal collagen concentrations (127). More work is needed to determine the effect of cross-linking on multiaxial mechanics of healing myocardial scar.

Ventricular Function During the Remodeling Phase

In many patients and experimental models, LV function improves as healing reaches the later stages. Clinical studies show improved hemodynamics and partially normalized LV compliance and EDP 4–6 weeks after infarction (60, 128), with few additional functional changes over the remainder of the first year (128, 129). All of the mechanisms for depression of function discussed at the beginning of this review except infarct rupture are involved to some extent in this late improvement in LV function. Scar stiffness remains higher than that of passive or acutely ischemic myocardium, limiting energy loss owing to systolic stretching (mechanism 2), and anisotropy appears to limit local tethering effects (mechanism 5). Scar shrinkage acts like infarct expansion in reverse, reducing the volume of the scar and infarct-associated cavity dilation (mechanism 4): wall motion abnormalities partially resolve (120, 125, 130), and the reduction in wall motion abnormality correlates closely with scar contraction (125). To the extent that LV dysfunction remains, it primarily reflects limitation of diastolic function owing to reduced diastolic compliance (mechanism 3, Figure 1). For example, Weisse found normal hemodynamics with mildly depressed ventricular function curves at 3–4 and 6–8 weeks following infarction in dogs. The depressed ventricular function curves were due entirely to a stiffer end-diastolic pressure-volume relationship (EDPVR), resolving when stroke work was plotted as a function of end-diastolic circumference rather than pressure (131).

However, there are some exceptions to this generally improving course. When very large infarcts are present, cavity dilation dominates other effects, such as scar shrinkage, leading to increased wall stresses and progressively depressed function (mechanism 6) (132, 133). If an aneurysm forms, the severely altered local geometry increases stresses (134) and depresses function (134–136) in the adjacent myocardium, creating a “functional border zone” analogous to that discussed for acute ischemia.

SUMMARY AND CONCLUSIONS

Based on temporal correlation of reported changes in structure and mechanics of healing myocardial infarcts, we have defined four phases of infarct healing and hypothesized the following: (a) Mechanical properties during acute ischemia (the first few hours) are essentially the normal constitutive properties of passive myocardium, (b) mechanical properties during the necrotic phase (the first 5–7 days depending on animal model) are dominated by edema, (c) mechanical properties during the fibrotic phase (up to 2–4 weeks) arise from the large collagen fiber structure, and (d) mechanical properties during the remodeling phase (the remainder of the healing process) are determined primarily by collagen cross-linking. We intend these hypotheses to stimulate further, mechanistic research on the mechanics of healing myocardial infarcts. Certainly, this review suggests many areas where more data are needed: Quantitative structural studies of the three-dimensional organization of important matrix components and determination of constitutive relations for scar tissue at multiple time points during healing would head our list.

However, the mechanics of healing infarct tissue, like those of heart tissue in general, depend both on constitutive properties and on loading conditions, which in turn are determined by hemodynamics, ventricular and local geometry, and coupling to adjacent myocardium. The individual studies reviewed here typically provide complementary, incomplete subsets of information about infarct mechanics. Studies using ventriculography and echocardiography provide information on global ventricular function and shape, plus more limited information on regional deformation in the infarct. Studies using implanted sonomicrometers or radiopaque markers provide more regional detail, with the advantages that infarct mechanics can still be related to overall ventricular function and that the deformation of an infarcted segment can be tracked not only throughout the cardiac cycle but also throughout longer-term remodeling; the primary disadvantage is that stresses cannot be measured directly but must be estimated from hemodynamic and geometric data by modeling (118). Finally, excision and mechanical testing of tissue provides the most direct characterization of tissue material properties, with the caveat that excision itself may alter those properties.

Remarkably few studies have tried to integrate these different experimental methods to obtain a complete picture of infarct mechanics even at a single time in a single animal model. The primary consequence of this lack of integration is that the wealth of information on changes in infarct deformation patterns over the course of postinfarction healing is difficult to interpret. In future studies, much more attention needs to be paid to differentiating changes in material properties (shifts of the stress-strain curve) from changes in loading (shifts along the stress-strain curve) because completely different therapeutic approaches may be appropriate to address these two different bases for altered mechanics. Multiaxial testing of infarcts at the various stages of healing is needed, but should include careful registration of these data to the in vivo working range. The other type of integration that is largely missing from the literature is direct integration of structural and mechanical data. No study we reviewed reported collagen content, cross-linking, and fiber structure along with mechanics of a healing infarct, and none of the studies directly altered tissue composition to test hypotheses about the structural basis for observed mechanical properties.

In summary, although much is known about changes in ventricular function, regional deformation, and tissue composition during the course of infarct healing, the underlying mechanics of the simplest case, permanent coronary occlusion without reperfusion, are still not sufficiently understood to predict the impact of proposed interventions or to specify the design requirements for a tissue-engineered replacement. Integrative studies combining material testing, quantitative structural analysis, and in vivo functional studies are needed, as are structural constitutive models. By allowing prediction of the changes in mechanics and function that will follow from proposed changes to healing infarct structure, these new studies would allow rational design of bioengineering therapies to improve long-term survival and quality of life for patients who suffer myocardial infarction.