Computational modeling of acute myocardial infarction

Abstract

Myocardial infarction, commonly known as heart attack, is caused by reduced blood supply and damages the heart muscle because of a lack of oxygen. Myocardial infarction initiates a cascade of biochemical and mechanical events. In the early stages, cardiomyocytes death, wall thinning, collagen degradation, and ventricular dilation are the immediate consequences of myocardial infarction. In the later stages, collagenous scar formation in the infarcted zone and hypertrophy of the non-infarcted zone are auto-regulatory mechanisms to partly correct for these events. Here we propose a computational model for the short-term adaptation after myocardial infarction using the continuum theory of multiplicative growth. Our model captures the effects of cell death initiating wall thinning, and collagen degradation initiating ventricular dilation. Our simulations agree well with clinical observations in early myocardial infarction. They represent a first step towards simulating the progression of myocardial infarction with the ultimate goal to predict the propensity toward heart failure as a function of infarct intensity, location, and size.

1 Introduction

Heart disease is the number one cause of death worldwide. This high death rate has lead to intense research in many fields of biology and engineering. Treatment of myocardial infarction has been evolving rapidly with the understanding of the mechanisms underpinning its progression. There is a clear correlation between infarct size and ejection fraction. Therefore many clinical therapies have been devoted to maintain normal ejection fractions mechanically, pharmacologically, intervention ally and surgically [7, 35, 41].

Cardiac tissue is composed of two major components that determine its mechanical behavior, cardiac cells - primarily cardiomyocytes, and extracellular matrix - primarily collagen. Cardiomyocytes are terminally differentiated from myoblasts. They comprise approximately 25% of all cells in the heart and, in their healthy state, they exert tension by contraction to pump blood to the rest of the body. Cardiomyocytes contraction is tightly regulated by ion concentrations inside and outside the cell. Their action potential is a result of sodium, potassium, and calcium concentration differences across the cell membrane.

The extracellular matrix provides a scaffold-type structure that provides the mechanical strength. Collagen I and III are the most important types of collagen in the extracellular matrix and the amount and distribution of the collagen fibers critically influence cardiac performance. Collagen I and III are synthesized mainly by myofibroblasts. Collagen I, which is associated with thick fibers, constitutes 85% of all the collagen in the heart; collagen III is associated with thin fibers. Collagen degradation and deposition is mainly influence by growth factors including TGF-β [3] and the MMP/TIMP ratio [42].

The healthy heart is conditioned by a continuous supply of oxygen and a balance between nutrient supply and demand. Myocardial infarction occurs when blood supply is interrupted and nutrient supply is compromised. During myocardial infarction, blood supply is interrupted by plaque formation in the coronary arteries [15] and the infarcted myocardium loses its contractile features. If the damaged region is large enough, myocardial infarction will trigger chronic heart failure. With increasing tissue hypoxia Adenosine Triphosphate (ATP) depletes, which causes failure of cardiac contraction. Two main types of cell deaths have been observed in hypertensive cardiomyopathy [20] and in ischemic heart disease [6]: oncosis and apoptosis [25]. Oncosis results in cell swelling and is associated with ischemic damage. Oncosis is the dominating process in the later stages, from 4 to 12 days after myocardial infarction. Apoptosis is an energy related, ATP-dependent process. Apoptosis is associated with cell shrinkage and programmed cell death [20, 31]. Apoptosis is involved in early stages of cell death, 4–8 hours after infarction, on ischemic injury and during reperfusion.

Morphological and mechanical changes in the heart after myocardial ischemia follow a well-defined pattern in time [2, 24]. Similar to wound healing [4], post-infarction remodeling has a short and a long time scale and establishes a threshold 72 hours post infarction [40]. The first stage, cardiomyocyte death, initiates a cascade of biochemical signaling processes in the form of inflammatory reactions in the infarct zone. Cell death and collagen degradation trigger wall thinning. Besides reducing structural support, this process enable cell migration into the infarcted region and neutrophils release matrix metalloproteinase into the infarct zone [9]. Collagen degradation begin 30 minutes post infarction and continues throughout the entire first week [13, 26]. Collagen degradation is a rather local process: it is observed only in the infarcted region, not in the surrounding tissue.

In the short term, the most important morphological and mechanical consequence is wall thinning, associated with collagen degradation and expansion of the ventricle. Figure 1 displays the immediate consequences of myocardial infarction: Cardiomyocyte death and collagen degradation. Within hours, the reduction of extracellular matrix can lead to ventricular expansion [11], which can eventually result in wall rupture [10]. The digestion of intercellular collagen fibers promotes the slippage of dead myofilaments that cause expansion of the infarct zone [34]. This expansion of the ventricles alters the stroke volume. To preserve the stroke volume and optimize function in the sense of the Frank Starling curves, a number of mechanical processes take place in the infarct region, the border zone, and the healthy myocardium.

Fig. 1.

Acute myocardial infarction at the macroscopic and microscopic scales. Histochemical staining reveals cardiomyocyte death in the posterolateral left ventricle, left. Histopathologic changes manifest themselves in wavy fibers at the periphery of the infarct, left, compared to the normal fibers in the healthy myocardium, right. The widened spaces between the dead fibers contain edema fluid and facilitate cardiomyocyte slippage, right. Adopted with permission from [27].

In the long term, myofibroblast invade the infarct zone. Myofibroblasts acquire contractile properties that promote wound contraction. At the same time, the formation of new collagen begins to compensate cardiac expansion and stabilize the infarcted heart [12].

In the following sections, we will address these molecular processes in view of a systems-biology based approach and their mechanical consequences [5]. Although important efforts have been made towards simulating infarct-induced remodeling processes in the heart [17, 29], the multiscale coupling between molecular events and organ function has received little attention to date. In the first section we will discuss a simple model of cardiomyocyte death, which will serve as a trigger variable for subsequent processes. This will allow us to model ventricular wall thinning in response to myocardial infarction. We explore collagen degradation and characterize its impact on ventricular expansion. Finally we suggest a computational mechanics model of the heart for the different stages of cardiac adaptation and we compare the results with observations reported in the literature.

2 Methods: Model problem of infarcted heart

Myocardial ischemia is characterized by a decreased blood supply. Figure 2 illustrates myocardial infarction as a result of plaque formation in the coronary arteries. Let us recall the basic kinematics of finite deformation. The deformation gradient F = ∇_Xφ, tangent to the motion φ, represents a two-point linear map over the reference configuration. Within the framework of finite growth, the key kinematic assumption is the multiplicative decomposition of the deformation gradient F into an elastic part F^e and a growth part F^g [37],

$$\mathbf{F}={\nabla }_{\mathit{X}}\mathit{\phi }={\mathbf{F}}^{\mathrm{e}}·{\mathbf{F}}^{\mathrm{g}}.$$ (1)

The underlying concept follows the multiplicative decomposition in finite elastoplasticity [28]. We can think of the growth tensor F^g as a second-order internal variable to characterize arbitrary forms of isotropic or anisotropic growth [33]. Here we consider two types of growth and parameterize the two growth tensors exclusively in terms of two single scalar-valued internal variables, the growth multipliers ϑ^⊥ and ϑ^‖, which characterize growth in the muscle sheet direction n^⊥ and in the muscle fiber direction n^‖ [17,39]. The Jacobian of the deformation gradient J = det(F) characterizes changes in volume, which we can decompose into reversible elastic volume changes J^e = det(F^e) and volume changes attributed to growth, J^g = det(F^g),

$$\mathrm{J}=\text{det}(\mathbf{F})={\mathrm{J}}^{\mathrm{e}}{\mathrm{J}}^{\mathrm{g}}.$$ (2)

We further introduce the right Cauchy Green tensor,

$$\mathbf{C}={\mathbf{F}}^{\mathrm{t}}·\mathbf{F}={({\mathbf{F}}^{\mathrm{g}})}^{\mathrm{t}}·{\mathbf{C}}^{\mathrm{e}}·{\mathbf{F}}^{\mathrm{g}}\hspace{1em}\text{with}\hspace{1em}{\mathbf{C}}^{\mathrm{e}}={({\mathbf{F}}^{\mathrm{e}})}^{\mathrm{t}}·{\mathbf{F}}^{\mathrm{e}},$$ (3)

as the covariant pull back of the elastic right Cauchy Green tensor C^e. It remains to specify the mechanisms of growth and with them the kinematics and kinetics of the growth tensor F^g.

Fig. 2.

Schematic drawing of the left anterior coronary artery and the obstruction of blood flow in the detailed view during myocardial infarction.

2.1 Cardiomyocyte death and ventricular wall thinning

Diminished oxygen supply causes cell death via necrosis or apoptosis. We model cell death by means of diffusion phenomena originating from several locations across the heart wall. These points represent ramifications of the left anterior descending coronary artery into arterioles. We characterize the reduction in the oxygen concentration of the cardiomyocytes as a result of impaired blood supply as

$${ċ}_{{\mathrm{O}}_2}=\text{Div}({\mathbf{q}}_{{\mathrm{O}}_2})+{𝒬}_{{\mathrm{O}}_2},$$ (4)

where c_O2 is the normalized oxygen concentration, q_O2 = −D_O2 ∇c_O2 is the diffusion of oxygen in the myocardial tissue and 𝒬_O2 is the oxygen source. For the particular problem of the ventricular wall thinning, we define the growth tensor as

$${\mathbf{F}}^{\mathrm{g}}=\mathbf{I}+[{\vartheta }^{\perp }-1]{\mathbf{n}}^{\perp }\otimes {\mathbf{n}}^{\perp },$$ (5)

where ϑ^⊥ is the scalar-valued growth multiplier that defines the amount of growth or shrinkage along the sheet vector n^⊥ [17, 36]. With the sheet vector pointing radially outward, positive growth, ϑ^⊥ > 1, results in a thickening of the myocardial wall, while negative growth, ϑ^⊥ < 1, results in wall thinning [18]. The function ϕ(c_O2) = c_O2 − 1 represents the growth criterion. We define the evolution of the growth multiplier ϑ^⊥ it in terms of an oxygen concentration function ϕ(c_O2), and a threshold function κ(ϑ^⊥) for ϕ(c_O2) < 0,

$${\dot{\vartheta }}^{\perp }=\kappa ({\vartheta }^{\perp })\varphi (c_{{\mathrm{O}}_2}),$$ (6)

where

$$\kappa ({\vartheta }^{\perp })=\frac{1}{\tau }{\left[\frac{{\vartheta }^{\perp }-{\vartheta }_{\text{min}}^{\perp }}{1-{\vartheta }_{\text{min}}^{\perp }}\right]}^{\gamma }.$$ (7)

${\vartheta }_{\text{min}}^{\perp }$ sets a growth threshold while τ and γ control the evolution of the growth process [21].

Unlike previous mechanically-driven growth, our model is oxygen-driven, which is not a mechanical property. Figure 3 illustrates the mechanism by which wall thinning occurs: Due to a reduction in oxygen supply, cell death is initiated and results in a loss of volume. This volume loss manifests itself in negative cardiomyocyte growth and ventricular wall thinning.

Fig. 3.

Healthy and infarcted myocardium. Loss of oxygen causes cardiomyocyte death, which results in a loss of volume and ventricular wall thinning modeled through negative growth in the thickness direction n^⊥.

2.2 Collagen degradation and ventricular dilation

During the early stages of the remodeling process, dysfunctional cardiomyocytes release matrix metalloproteinase, which causes increased collagen degradation. We model changes in matrix metalloproteinase, c_MMP, through the following rate equation [38],

$${ċ}_{\mathrm{M}\mathrm{M}\mathrm{P}}={𝒬}_{\mathrm{M}\mathrm{M}\mathrm{P}}\hspace{1em}\text{with}\hspace{1em}{𝒬}_{\mathrm{M}\mathrm{M}\mathrm{P}}={\left[\frac{c_{\mathrm{M}\mathrm{M}\mathrm{P}}}{c_{\mathrm{M}\mathrm{M}\mathrm{P}}^{*}}\right]}^{-m}[1-c_{{\mathrm{O}}_2}],$$ (8)

where c_O2 is the normalized oxygen concentration, c_MMP is the density of matrix metalloproteinases, $c_{\mathrm{M}\mathrm{M}\mathrm{P}}^{*}$ is its initial value, and the exponent m controls its evolution. We assume that the collagen concentration c_col is correlated to the intensity of matrix metalloproteinase c_MMP through the scaling coefficient γ_MMP,

$${ċ}_{\text{col}}={\gamma }_{\mathrm{M}\mathrm{M}\mathrm{P}}{ċ}_{\mathrm{M}\mathrm{M}\mathrm{P}}.$$ (9)

The continuing collagen turnover and increase in collagen degradation makes the ECM to lost its ability to serve as a scaffold of the cardiomyocytes. This degradation makes the ventricle to expand along time with the cardiac cycles. This process also softens the ventricular wall. We model ventricular dilation mathematically similar as wall thinning. We introduce the growth tensor F^g as

$${\mathbf{F}}^{\mathrm{g}}=\mathbf{I}+[{\vartheta }^{\Vert }-1]{\mathbf{n}}^{\Vert }\otimes {\mathbf{n}}^{\Vert },$$ (10)

where ϑ^‖ is the scalar-valued growth multiplier that defines the amount of growth or shrinkage along the cardiomyocyte’s long axis n^‖ [17]. With the cardiomycyte vector pointing helically around the heart, positive growth, ϑ^‖ > 1, results in a dilation of the ventricles [18]. Ventricular dilation is governed through the scalar-valued parameter ϑ^‖. We model the evolution of ventricular dilation as

$${\dot{\vartheta }}^{\Vert }=\kappa ({\vartheta }^{\Vert })\varphi (c_{\text{col}}),$$ (11)

where the dilation function, ϕ(c_col) = 1 − c_col, is parameterized in terms of the collagen concentration c_col and

$$\kappa ({\vartheta }^{\Vert })\frac{1}{\tau }{\left[\frac{{\vartheta }^{\text{max}}-{\vartheta }^{\Vert }}{{\vartheta }^{\text{max}}-1}\right]}^{\gamma },$$ (12)

for ϕ(c_col) > 0, controls dilation through the threshold ϑ^max.

2.3 Mechanical model

Last, we define the mechanical problem and incorporate the characteristic features of quasi-incompressibility, non-linearity, and finite strains. We model quasi-incompressiblity–mainly attributed to the water content–through the volumetric-isochoric decomposition of the deformation [14]. We decouple the elastic part of the deformation gradient F^e into dilatational and isochoric parts as

$${\mathbf{F}}^{\mathrm{e}}={({\mathrm{J}}^{\mathrm{e}})}^{1/3}{\overline{\mathbf{F}}}^{\mathrm{e}}\hspace{1em}\text{with}\hspace{1em}{\mathrm{J}}^{\mathrm{e}}=\text{det}({\mathbf{F}}^{\mathrm{e}})\hspace{1em}\text{and}\hspace{1em}{\overline{\mathbf{F}}}^{\mathrm{e}}={({\mathrm{J}}^{\mathrm{e}})}^{1/3}{\mathbf{F}}^{\mathrm{e}}.$$ (13)

We adopt a finite hyperelastic framework to capture the non-linear behavior and introduce the following strain energy density function [23],

$$\Psi ={\Psi }_{\text{vol}}(J^{\mathrm{e}})+{\Psi }_{\text{iso}}({Ī}_1^{\mathrm{e}},{Ī}_4^{\mathrm{e}})\hspace{1em}\text{with}\hspace{1em}{\Psi }_{\text{iso}}({Ī}_1^{\mathrm{e}},{Ī}_4^{\mathrm{e}})={\Psi }_1({Ī}_1^{\mathrm{e}})+{\Psi }_4({Ī}_4^{\mathrm{e}}),$$ (14)

where Ψ_vol is the volumetric part parameterized in terms of the elastic Jacobian J^e, and Ψ_iso is the isochoric part parameterized in terms of the first and forth invariants ${Ī}_1^{\mathrm{e}}$ and ${Ī}_4^{\mathrm{e}}$ of the isochoric elastic right Cauchy Green deformation tensor C̄^e, = (F̄^e)^t · F̄^e

$${Ī}_1^{\mathrm{e}}={\overline{\mathbf{C}}}^{\mathrm{e}}:\mathbf{I}\hspace{1em}\text{and}\hspace{1em}{Ī}_4^{\mathrm{e}}={\overline{\mathbf{C}}}^{\mathrm{e}}:[{\mathbf{n}}^{\Vert }\otimes {\mathbf{n}}^{\Vert }].$$ (15)

Here we have further decomposed the isochoric energy to independently characterize the isotropic and anisotropic response, ${\Psi }_{\text{iso}}({Ī}_1^{\mathrm{e}},{Ī}_4^{\mathrm{e}})={\Psi }_1({Ī}_1^{\mathrm{e}})+{\Psi }_4({Ī}_4^{\mathrm{e}})$. For the isotropic contribution, we adopt an exponential form,

$${\Psi }_1({Ī}_1^{\mathrm{e}})=\frac{a}{2b}[\text{exp}(b{[{Ī}_1^{\mathrm{e}}-3]}^2)-1],$$ (16)

where a and b are material parameters. For the anisotropic contribution, we adopt the classical Holzapfel model [22],

$${\Psi }_4({Ī}_4^{\mathrm{e}})=\frac{k_1}{2k_2}[\text{exp}(k_2{[{Ī}_4^{\mathrm{e}}-1]}^2)-1]$$ (17)

with parameters a=0.5, b=7.2, k₁ = 15.2 and k₂ = 20.5 [19]. The above decomposition of the strain energy density function naturally affects the elastic second Piola-Kirchhoff stress [22, 32],

$${\mathbf{S}}^{\mathrm{e}}=2{\partial }_{{\mathbf{C}}^{\mathrm{e}}}{\Psi }_{\text{vol}}(J^{\mathrm{e}})+2{\partial }_{{\mathbf{C}}^{\mathrm{e}}}{\Psi }_{\text{iso}}({Ī}_1^{\mathrm{e}},{Ī}_4^{\mathrm{e}})={\mathbf{S}}_{\text{vol}}^{\mathrm{e}}+{\mathbf{S}}_{\text{iso}}^{\mathrm{e}},$$ (18)

which also consists of volumetric and an isochoric contributions,

$${\mathbf{S}}_{\text{vol}}^{\mathrm{e}}=2{\partial }_{{\mathbf{C}}^{\mathrm{e}}}{\Psi }_{\text{vol}}\hspace{1em}\text{and}\hspace{1em}{\mathbf{S}}_{\text{iso}}^{\mathrm{e}}=2{\partial }_{{\overline{\mathbf{C}}}^{\mathrm{e}}}{\Psi }_{\text{iso}}:{\partial }_{{\mathbf{C}}^{\mathrm{e}}}{\overline{\mathbf{C}}}^{\mathrm{e}}={({\mathrm{J}}^{\mathrm{e}})}^{-2/3}{\mathbb{P}}^{\mathrm{e}}:{\overline{\mathbf{S}}}^{\mathrm{e}}.$$ (19)

Here S̄^e = ∂_C̄^eΨ_iso(C̄^e) is the fictitious second Piola-Kirchhoff stress ${\mathbb{P}}^{\mathrm{e}}=𝕀-\frac{1}{3}{({\mathbf{C}}^{\mathrm{e}})}^{-1}\otimes {\mathbf{C}}^{\mathrm{e}}$ is the fourth order projection tensor [22]. The total second Piola-Kirchhoff stress

$$\mathbf{S}=2{\partial }_{\mathbf{C}}\Psi =2{\partial }_{{\mathbf{C}}^{\mathrm{e}}}\Psi :2{\partial }_{\mathbf{C}}{\mathbf{C}}^{\mathrm{e}}={\mathbf{F}}^{\mathrm{g}-1}·{\mathbf{S}}^{\mathrm{e}}·{\mathbf{F}}^{\mathrm{g}-\mathrm{t}},$$ (20)

then follows from the contravariant pull back of the elastic second Piola-Kirchhoff stress S^e of Eq. (18).

2.4 Numerical implementation

The diffusion problem is solved, with standard elements and pre-implemented constitutive equations, with the commercial software ABAQUS (Dassault Systemes Simulia Corp., Providence, RI, U.S.A.). In the mechanical part, we introduce the growth multipliers, ϑ^⊥ for ventricular wall thinning and ϑ^‖ for ventricular dilation, as internal variables on the integration point level. To solve the growth kinetics in time, we adapt an implicit Euler backward scheme,

$$\dot{\vartheta }=[\vartheta -{\vartheta }^{\mathrm{n}}]/\Delta t,$$ (21)

where Δt = t − tⁿ denotes the current time increment, with the understanding that we omit the superscript (◦)ⁿ⁺¹ for ease of notation. To solve for the evolution of growth, we apply a local Newton iteration with the discrete local residual,

$$ℛ=\vartheta -{\vartheta }^{\mathrm{n}}-\varphi \kappa (\vartheta )\Delta t,$$ (22)

and its local tangent,

$$𝒦={\mathrm{d}}_{\vartheta }ℛ=1-\varphi {\partial }_{\vartheta }\kappa \Delta t.$$ (23)

For ventricular wall thinning, ${\partial }_{\vartheta }{\kappa }^{\perp }=\gamma {\kappa }^{\perp }/[{\vartheta }^{\perp }-{\vartheta }_{\text{min}}^{\perp }]$; for ventricular dilation, ${\partial }_{\vartheta }{\kappa }^{\Vert }=-\gamma {\kappa }^{\Vert }/[{\vartheta }_{\text{max}}^{\Vert }-{\vartheta }^{\Vert }]$. The incremental update of the growth multiplier takes the following format,

$$\vartheta \leftarrow {\vartheta }^{\mathrm{n}}-ℛ/𝒦,$$ (24)

which we use to iterate until the absolute value of the residual reaches a user-defined threshold level. At local equilibrium, we store the growth multiplier ϑ^⊥ or ϑ^‖, and successively calculate the growth tensor F^g using Eq. (5) or (10), calculate the elastic tensor F^e and its Jacobian J^e using Eq. (1), calculate the isochoric elastic tensor F̄^e using Eq. (13), and calculate the isochoric elastic Cauchy Green deformation tensor C̄^e along with its invariants ${Ī}_1^{\mathrm{e}}$ and ${Ī}_4^{\mathrm{e}}$ using eq. (15). Last, we calculate the elastic second Piola Kirchhoff stress S^e using eq. (18) and the total second Piola Kirchhoff stress S using eq. (20). To solve the set of finite element equations within a global Newton iteration, we finally calculate the fourth order tensor of tangent moduli 𝖢 = 2d_CS, which relate increments in stress and strain and are essential for an efficient finite element implementation.

3 Results

3.1 Cardiomyocyte death and wall thinning

Figure 4 illustrates the progression of cell death due to a 90% decrease in the blood supply from the left anterior descending coronary artery. As initial conditions, we reduce the local oxygen concentration in the affected ischemic area to c_O2 (t₀) = 0.1 and assume that the remaining tissue is initially unaffected by the ischemic event with c_O2 (t₀) = 1.0. Figure 4 documents the gradual progression of the cell death in regions of reduced oxygen supply. Cell necrosis is a trigger for several processes that subsequently occur across the heart.

Fig. 4.

Evolution of oxygen supply as a trigger for cardiomyocyte death following left anterior descending arterial occlusion.

Figure 5 illustrates the resulting thinning of the left ventricular wall. Wall thinning naturally takes largest values in regions of largest cardiomyocyte death. The maximum relative reduction in the wall thickness occurred in the center of the infarct and was 23%.

Fig. 5.

Evolution of the wall thinning in the left ventricle due to progressive cardiomyocyte death.

3.2 Collagen degradation and ventricular dilation

Figure 6 shows the evolution of collagen degradation after partial occlusion of the left anterior descending artery.

Fig. 6.

Evolution of collagen degradation following left anterior descending arterial occlusion.

Increased collagen degradation triggers progressive cardiomyocyte slippage, which leads to continued ventricular dilation. Figure 7 illustrates the progression of ventricular dilation with a maximum value of 40%. Figure 8 illustrates the wall thinning and ventricular dilation in a cross-sectional cut and the evolution of the ventricle thickness and the expansion area in the location of maximum ischemia.

Fig. 7.

Evolution of ventricular dilation in response to collagen degradation.

Fig. 8.

Wall thinning (left) and ventricular dilation (right). Translucent color shows the original geometry and solid colors shows the final thinned and dilated geometry. In the right, figure displays the gradual thinning of the wall in time (red line) and the area increase of the internal ventricular surface in the region of maximum expansion (blue line).

3.3 Mechanical model

To illustrate the effects of wall thinning and ventricular dilation, we simulate three different scenarios: a healthy heart, a wall thinned heart, and a dilated heart. To fix the heart in space, we impose Dirichlet boundary conditions in the basal plane: We fix the basal region of the septum in all three directions in space, and fix the displacements of the remaining basal regions only in the perpendicular direction. We prescribe the left and right ventricular pressures in both ventricles at the isovolumetric contraction phase according to the pressure curves in [19].

Figure 9 summarizes the spatial evolution of the ventricular wall stresses at the beginning of the isovolumetric contraction stage (≈ 80–100kPa). In comparison to the healthy heart, top, the heart, which had undergone cardiomyocyte death and wall thinning, middle, displays a significant increase in the stress. In comparison to the healthy heart, top, the heart, which had undergone collagen degradation and ventricular dilation as can be observed in the figure, bottom, experiences a larger increase in the stress.

Fig. 9.

Spatial evolution of stresses in the healthy heart (top), after wall thinning (middle), and after ventricular dilation (bottom) at four equidistant time points across the cardiac cycle.

4 Discussion

We have introduced a new model to study the thinning and dilation of the ventricle after myocardial infarction. Previous models of cardiac growth and remodeling are driven by mechanical variables such as strain and stress [17, 36]. Although these rather phenomenological models compare well to clinical observations, they lack a sound biological description of the underlying biochemical phenomena in the affected tissue regions. In several pathologies, stress and strain are often viewed as triggers of cardiac remodeling [16]; yet, in myocardial infarction, it is a cascade of biochemical events that initiates morphological alterations in the affected tissue regions [8,24]. Here we have adapted previous computational models of volumetric growth [17, 39] to include biochemical effects as the driving forces for growth and remodeling. We first studied the ischemic tissue and showed how reduced oxygen supply initiates progressive cardiomyocyte death. This has, in turn, lead to thinning of the ventricular wall. At the same time, biochemical phenomena, including synthesis of matrix metalloproteinase, initiate changes in collagen content [9, 24]. These phenomena triggers cardiomyocyte slippage and results in progressive ventricular dilation [11]. We have proposed evolution equations for these two cascades of events and simulated both in a finite element setting. Our results agree well with the clinical observations in response to myocardial infarction. Wall thinning and subsequent ventricular dilation are two well-know consequences of infarction. They are collectively referred to as short-term adaptation.

Our model has some limitations that should be mentioned. In this first step, we have neglected the electro-mechanical behavior of the heart [1]. We have focused exclusively on the passive material behavior. However, in our current model, growth is not driven by stress–for which we would need to account for passive and active contributions–but rather by biochemical events. Moreover, the mechanical effects of wall thinning and ventricular dilation are mainly relevant during the passive filling phase of the cycle. Nevertheless, these two effects can also affect the pump function during systole. For example, wall thinning can alter the end-diastolic strain in the ventricles and because of the Frank-Starling mechanism, this geometrical change would have an impact on the active contraction. While the contracting phase would be mainly affected by the loss of viable cardyomyocytes. This is a active mechanism of the process that we do not consider here. Therefore, we exclusively focus on the passive constitutive response. This lets us to recover some mechanical variables, including stress profiles, and allows us to directly compare alterations in response to cardiomyocyte death and collagen degradation. This model, as in many others, has a number of material parameters that are difficult to calibrate [30].

In our future work, we would like to overcome these limitations by performing controlled experimental studies of the interaction of biochemical fields and sarcomere number on the cellular scale, cardiomyocyte density and collagen content on the tissue scale, wall thickness and chamber size on the organ scale, and ejection fraction and ventricular pressure on the system scale. We will perform a series of experiments to quantify and model the interaction of these characteristics in both space and time. We hope to create a calibrated computational model of the time line of myocardial infarction with the ultimate goal to predict the propensity toward heart failure as a function of infarct intensity, location, and size.