Regional Myocardial Three-Dimensional PrincipalStrains During Postinfarction RemodelingJames J. Pilla, PhD, Kevin J. Koomalsingh, MD, Jeremy R. McGarvey, MD,Walter R. T. Witschey, PhD, Larry Dougherty, PhD, Joseph H. Gorman, III, MD, andRobert C. Gorman, MDDepartment of Radiology and Gorman Cardiovascular Research Group, Department of Surgery, University of Pennsylvania,Philadelphia, Pennsylvania

Background. The purpose of this study was to quantifymyocardial three-dimensional (3D) principal strains asthe left ventricle (LV) remodels after myocardial infarc-tion (MI). Serial quantification of myocardial strains isimportant for understanding the mechanical response ofthe LV to MI. Principal strains convert the 3D LV wall-based strain matrix with three normal and three shearelements, to a matrix with three nonzero normal ele-ments, thereby eliminating the shear elements, which aredifficult to physically interpret.

Methods. The study was designed to measure principalstrains of the remote, border zone, and infarct regions in aporcine model of post-MI LV remodeling. Magneticresonance imaging was used to measure function andstrain at baseline, 1 week, and 4 weeks after infarct.Principal strain was measured using 3D acquisition andthe optical flow method for displacement tracking.

Accepted for publication Oct 31, 2014.

Address correspondence to Dr R. C. Gorman, Perelman School of Medi-cine, University of Pennsylvania, 3400 Civic Center Blvd, Philadelphia,PA 19104-5156; e-mail: robert.gorman@uphs.upenn.edu.

� 2015 by The Society of Thoracic SurgeonsPublished by Elsevier

Results. Principal strains were altered as the LV remod-eled. Maximum principal strain magnitude decreased inall regions, including the noninfarcted remote, whilemaximum principal strain angles rotated away from theradial direction in the border zone and infarct. Minimumprincipal strain magnitude followed a similar pattern;however, strain angleswere altered in all regions. Evolutionof principal strains correlated with adverse LV remodeling.Conclusions. Using a state-of-the-art imaging and

optical flow method technique, 3D principal strains canbe measured serially after MI in pigs. Results areconsistent with progressive infarct stretching as well aswith decreased contractile function in the border zoneand remote myocardial regions.

(Ann Thorac Surg 2015;-:-–-)� 2015 by The Society of Thoracic Surgeons

fter a myocardial infarction (MI), the left ventricle

A(LV) is at risk for remodeling. Infarct expansion hasbeen implicated in sustaining LV remodeling after MI [1].Immediately after ischemia, the infarct region ceases tocontract and is subjected to mechanical loads producedby cavity pressure and the noninfarcted remainder of theventricle. This abnormal loading results in stretching ofthe infarct and increased stress in the border zone (BZ)region adjacent to the infarct [2]. This dysfunctionalBZ becomes more hypocontractile and progresses toinvolve additional perfused myocardium as remodelingcontinues.

Tagged magnetic resonance imaging (MRI) is a methodof tracking the myocardial displacement using noninva-sive markers [3]. The development of three-dimensional(3D) tagging in a single acquisition makes regional LVmeasurements of the 3D strain possible [4, 5]. Principalstrains derived from the strain tensor provide informationon the magnitude and direction of the deformation thatare more amenable to physical interpretation than LVwall-based strains. For LV wall-based strains, the strain

matrix is oriented with respect to the left ventriculargeometry. The coordinate system is rotated so that the x,y, and z directions align with the radial, circumferential,and longitudinal directions of the LV, respectively.Components of the normal and shear strains areexpressed as magnitudes values in the given wall-baseddirection. Changes in cardiac strain during remodelingare reported as normal and shear values with constantorientation. Normal strains are easily interpreted whereasshear strains are more difficult to comprehend because oftheir definition and complex orientation (Appendix 1).Studies using invasive methods and tagging have

suggested that mechanical changes in the BZ and infarctregions are associated with remodeling [6–8]. Thesestudies have shown that wall-based circumferential andlongitudinal strain magnitudes are altered in the infarctand BZ regions. However, a complete understanding ofthe mechanical alterations in these regions has beenlimited by the inability to measure 3D strains duringremodeling. The purpose of this study was to serially

The Appendices can be viewed in the online version ofthis article [http://dx.doi.org/10.1016/j.athoracsur.2014.10.067] on http://www.annalsthoracicsurgery.org.

0003-4975/$36.00http://dx.doi.org/10.1016/j.athoracsur.2014.10.067

2 PILLA ET AL Ann Thorac SurgSTRAINS DURING POSTINFARCTION REMODELING 2015;-:-–-

quantify myocardial 3D principal strains after MI to betterunderstand the mechanism of post-MI remodeling.

Material and Methods

The study was designed to quantify the changes inregional principal strains in a porcine model of post-MIremodeling. Animals were treated under an experi-mental protocol in compliance with National Institutes ofHealth “Guide for the Care and Use of Laboratory Ani-mals” (NIH publication 85-23, revised 1996) and approvedby the University of Pennsylvania Institutional AnimalCare and Use Committee. Five animals received a base-line MRI to access LV volume and regional principalstrains before infarction. Subsequently, a posterolateralinfarct was created by ligating the left circumflex arterydistal to the first obtuse marginal artery branch, andmarkers were placed along the boundary of the infarct.The animals were recovered, and an MRI was performedat 1 week and 4 weeks after infarction.

Magnetic resonance imaging was used to measure LVend-systolic volume (ESV) and end-diastolic volume (EDV),and regional principal strain at baseline, 1week, and 4weeksafterMI using a 3T Scanner (Siemens,Malvern, PA). The LVvolume imaging consisted of a cine cardiac acquisition fol-lowed by strain imaging using a 3D tag sequence [4].

Data AnalysisThe LV volume data were obtained from the cine MRIscans. The endocardial contours were drawn for eachslice and phase then imported into a custom program tocalculate volumes. Systolic LV regional strain wasassessed from tagged images using a method previouslydescribed and validated [7]. An optimized 3D optical flowmapping algorithm tracked the myocardium motion andproduced displacement fields in the x, y, and z direction(Appendix 2). The Lagrangian Green’s strain tensor wascalculated from the displacement fields between theinitial state of end diastole and the deformed state of endsystole, and principal strains were determined from thestrain tensor by solving the eigenvalue problem. Principalstrains represent the magnitude (eigenvalue) and direc-tion (eigenvector) of the maximum stretch (E1), maximumshortening (E3), and mutually orthogonal difference of thestretch and shortening (E2) of the myocardium. They

Fig 1. Coordinate system of principal strainvectors. (A) Radial-circumferential angle(qRC) is the principal strain vector projectedon the radial-circumferential plane. Anal-ogous descriptions are used for the (B)radial-longitudinal (qRL) and (C) longitu-dinal circumferential (qCL) planes. (Circ ¼circumferential; Long ¼ longitudinal.)

differ from wall-based strains by providing a means fortracking not only strain magnitude but also strain orien-tation as the heart remodels (Appendix 1).A local wall-based coordinate system was established

for eigenvector orientation [9]. The circumferential (c)direction was defined by a vector tangent to epicardialcontour. Radial direction (r) was defined by a vector in-ward and normal to the local epicardial wall. The longi-tudinal direction (l) was defined to be in the direction ofthe vector cross product of c and r, tangent to theepicardial surface (Fig 1).Three midventricular slices were selected for quanti-

tative analysis for each animal. To investigate alteration ofthe principal strains during postinfarction ventricularremodeling, each slice was divided into three segments:infarct, BZ, and remote. Infarct regions were delineatedusing the markers. The BZ region was determined to bethe myocardium encompassed by a 20-degree arc be-tween the marker and the remote region [10].

StatisticsData are presented as mean � SEM. The LV volume datawere assessed using one-way repeated measures analysisof variance with Tukey post-hoc evaluation. Two-wayrepeated measures analysis of variance with Tukey mul-tiple comparisons was used to analyze the principal strainfor differences between regions and timepoints. Pearsonand Spearman’s ranked-order correlation were used toanalyze the relationship between the change in regionalstrain and LV volume. Strain magnitudes and directionsin each segment were compared by paired Student’s ttest. A value of p less than 0.05 was considered statisticallysignificant for all comparisons.

Results

Global LV RemodelingGlobal LV remodeling occurred in all animals throughoutthe study period. Statistically significant and progressiveincreases in EDV and ESV were documented, as was aconcurrent decrease in ejection fraction (Table 1).

Principal StrainsRepresentative 3D tagged images are shown in Fig 2.Altered deformation of the tags in the infarct and BZ

Table 1. Change in Left Ventricle Function With Remodeling

Function Baseline 1 Week 4 Weeks

EDV, mL 61.4 � 4.7 90.4 � 8.7a 119.9 � 11.8b

ESV, mL 34.7 � 3.6 58.9 � 5.1a 84.1 � 8.3b

EF, % 43.3 � 2.5 34.7 � 1.4a 29.7 � 3.2b

a p < 0.05 versus baseline. b p < 0.05 versus 1 week.

Values are mean � SEM.

EDV ¼ end-diastolic volume; EF ¼ ejection fraction; ESV ¼ end-systolic volume.

3Ann Thorac Surg PILLA ET AL2015;-:-–- STRAINS DURING POSTINFARCTION REMODELING

regions is evident at end systole. Maximum, minimum,and intermediate principal strains are presented belowfor each region and timepoint.

Maximum Principal StrainThe magnitude (eigenvalue – E1Mag) and direction(eigenvectors – E1qRC, E1qRL, E1qCL) of the maximumprincipal strain (E1) for the cohort of animals are pre-sented in Figure 3. At baseline, the magnitude of the E1 issimilar in all regions (Fig 3A). The E1qRC (Fig 3B) andE1qRL (Fig 3C) measure the angles between local radialdirection and the projection of E1 onto transverse andlongitudinal planes, respectively. Both E1qRC (Fig 3B) andE1qRL (Fig 3C) before infarction are less than 10 degreeswhereas E1qCL is near 45 degrees (Fig 3D) for all regions,demonstrating that before infarction, E1 is directedpredominately in the radial direction, indicative ofnormal wall thickening.

One week and 4 weeks after MI, the magnitude of E1

(ie, E1Mag) is significantly reduced in all myocardialregions (Fig 3A). Interpretation of this reduced magnitudeis highly dependent on the associated changes in eigen-vector directions: E1qRC, E1qRL and E1qCL. Reduction inE1Mag can be interpreted as a decrease in radial wallthickening if the associated eigenvector directions angles

are not significantly changed from the baseline values(ie, E1qRC and E1qRL near zero; E1qCL near 45�). However,changes in the associated eigenvector directions, espe-cially increases in E1qRC and E1qRL, are indicative of somedegree of regional myocardial stretching.One week after MI, the eigenvector of E1 in the BZ and

infarct regions deviate from the original radial direction,resulting in significant increase in E1qRC, with the infarcthaving the largest change. The E1qCL decreases in boththe BZ and infarct regions. These changes are consistentwith a shift from radial thickening to circumferentialstretching. The E1qRL is unchanged in the BZ butincreases in the infarct, indicating a shift from radialthickening to longitudinal stretch in this region. Allangles in the remote region are unchanged at 1 week afterMI; however, E1Mag in this region is significantlyreduced, indicating reduced wall thickening.Four weeks after MI, the eigenvector for E1 in the

infarct continues to deviate from the radial direction, asindicated by the continued increases in E1qRC and E1qRL.These data are consistent with progressive circumferen-tial and longitudinal infarct stretching as well as completeloss of wall thickening in the infarct region. In the BZ,E1Mag remains decreased and all three angles areincreased, which is consistent with decreased wall thick-ening and increased circumferential and longitudinalstretching. In the remote region, all angles remainpreserved, but there is a persistent reduction in E1Mag,indicating that remote wall thickening remains impairedat 4 weeks after MI.

Minimum Principal StrainThe magnitude (eigenvalue – E3Mag) and direction(eigenvectors – E3qRC, E3qRL and E3qCL) of the minimumprincipal strain (E3) for the cohort of animals are pre-sented in Figure 4. Before infarction, E3Mag is similar inall regions. The E3qRC and E3qRL approach 90 degrees at

Fig 2. Tagged images of left ventricularremodeling after myocardial infarction.Short-axis midventricular images acquiredat baseline (left), 1 week (middle), and 4weeks (right) after infarct. (Top row) Enddiastole (ED) images show the change inleft ventricle size and geometry with infarctprogression. (Bottom row) Regional tagdeformation at end systole (ES) are shownfor the three timepoints. Infarct region isdelineated using markers, which createsignal voids (arrows).

Fig 3. Regional maximum principal strain magnitude and angles during left ventricle remodeling. (A) Maximum principal strain magnitude(E1Mag) is altered in all regions as the ventricle remodels. Infarct (INF) has the greatest change in E1Mag, decreasing at 1 week, then increasing at4 weeks. Maximum principal strain angles (B) E1qRC, (C) E1qRL, and (D) E1qCL were unchanged in the remote (REM) region. Infarct and borderzone (BZ) angles rotated away from the radial direction to a more circumferential-longitudinal direction, indicating stretch. þp < 0.05 versusbaseline; zp< 0.05 versus 1 week; *p< 0.05 versus remote; �p< 0.05 versus BZ. (Black bars¼ baseline; dark gray bars¼ 1 week; light gray bars¼4 weeks.)

4 PILLA ET AL Ann Thorac SurgSTRAINS DURING POSTINFARCTION REMODELING 2015;-:-–-

baseline while E3qCL is near 45 degrees (Figs 4B through4D). These eigenvectors indicate that the majority ofmyocardial shortening occurs in the circumferential andlongitudinal directions. The orientation of E3qCL is alsoindicative of a spiraling or torsional contraction from apexto base (Fig 4D). The physical interpretation of thesefindings is that, before infarction, the minimum principalstrain (E3) represents the maximum shortening orcontraction in the LV wall. During systole, the normal LVsimultaneously shortens in both the longitudinal andcircumferential directions, which contributes to the well-described torsional movements of the LV [11, 12].

One week after MI, all regions demonstrate decreasedE3, with the infarct region having the lowest strain(Fig 4A). The infarct region also exhibits a significantdecrease in E3qRC, indicating a shifting from circumfer-ential shortening to radial shortening. The E3qRL issignificantly reduced in all regions owing to a change inshortening orientation away from the longitudinal direc-tion. At 4 weeks after MI, E3Mag in all regions increasesrelative to 1 week, but fails to reach preinfarction levels.The increased E3Mag in the remote region may be jointlyattributed to contractile compensation as well as toincreased infarct compliance, which theoretically coulddecrease initial resistance to contraction. Whereas in the

infarct, the increase in E3Mag is indicative of increasedstretch, in the BZ, it is likely a combination of both con-tractile compensation and increased stretch. Moreover, at4 weeks, E3qRC and E3qRL in the infarct remain signifi-cantly reduced relative to preinfarction, indicative ofpersistent atypical shortening in the radial direction (ie,continued infarct systolic thinning). The BZ region con-tinues to have altered strain in both E3qRL and E3qCL. Theremote region returns to baseline E3qRL values while theE3qCL continued to be elevated, signifying altered torsion.Bulls-eye plots of the minimum and maximum principalstrain eigenvalues and eigenvector components for arepresentative animal are shown in Figures 5A and 5B.

Intermediate Principal StrainsThe physical interpretation of intermediate principalstrain (E2) is more difficult to determine. Whereas E1 andE3 represent the thickening and shortening, respectively,E2 is the mutually perpendicular strain to E1 and E3. Afterinfarction, significant changes occur in E2 (Fig 6A). TheE2qRC is altered at 4 weeks after infarction, with littlechange in the remote and the BZ (Fig 6B). The E2qRL isdecreased at 1 week and 4 weeks after MI (Fig 6C). E2qCLdecreases in REM and BZ as LV remodels while INF

Fig 4. Regional minimum principal strain magnitude and angles during left ventricle remodeling. (A) At 1 week, minimum principal strainmagnitude (E3Mag) decreases in all regions but increases as remodeling continues. Remote (REM) improvement could be attributed to left ventriclecompensation whereas border zone (BZ) and infarct (INF) increase is probably a consequence of tissue stretch. Angles (B) E3qRC, (C) E3qRL, and (D)E3qCL are altered in all regions, which may be attributed to both changes in infarct and BZ tissue properties and decreased left ventriculartorsion. þp < 0.05 versus baseline; zp < 0.05 versus 1 week; *p < 0.05 versus remote; �p < 0.05 versus BZ. (Black bars ¼ baseline; dark graybars ¼ 1 week; light gray bars ¼ 4 weeks.)

5Ann Thorac Surg PILLA ET AL2015;-:-–- STRAINS DURING POSTINFARCTION REMODELING

remains elevated above baseline at 1 week and 4 weeks(Fig 6D).

Correlation of Principal Strain With LV RemodelingOne week after MI, the change in EDV from baseline wascorrelated with changes in several of the principal straincomponents. The E1 components in both the remote andinfarct regions were highly correlated whereas compo-nents in the BZ and infarct were better correlated for E3.The ESV change was also correlated with components ofboth E1 and E3 (Table 2).

The E3 was the only principal strain correlated withremodeling at 4 weeks after infarct (Table 3). The EDVwas correlated with eigenvector angles for both theremote and the BZ whereas changes in ESV were asso-ciated with angles in the infarct and BZ regions.

Principal strains at 1 week were compared with thechange in LV remodeling from 1 week to 4 weeks as apotential predictor of remodeling (Table 4). The E3 infarcthad a strong positive correlation with increasing EDV asthe LV continued to remodel. Angular components of E1

in the remote region also demonstrated an influence ondilation. The ESV dilation was strongly negatively corre-lated to E2 infarct and E1 remote and positively correlated

a maximum principal strain angle in the infarct region.Ejection fraction was correlated to the magnitudes of E1 inthe BZ and E2 in the remote region as well as to anangular component of E1 infarct.

Comment

In this study, we used an optical flow method [4] as wellas an acquisition scheme that enables direct measure-ment of 3D myocardial displacement before and after aposterolateral MI in pigs. Using this combined acquisi-tion and tracking method, we quantified the temporalchanges in the 3D strain tensor to gain an improved un-derstanding of the evolution of postinfarct remodeling.Our results indicated that during LV remodeling, strain

was altered in all regions and at all timepoints after MI,with the most pronounced changes occurring in the BZand infarct regions. Previous ex vivo tissue experimentshave demonstrated that infarct passive mechanical stiff-ness is greatest at 1 week after coronary occlusion anddecreases progressively with time, leading to increasedtissue stretch [13]. This temporal change in infarct mate-rial properties is confirmed by the magnitude andorientation of the eigenvectors presented in this study.

Fig 5. Bulls-eye plots of (A) maximumprincipal strain (E1Mag) and (B) minimumprincipal strain (E3Mag) at baseline, 1 week(1-Wk), and 4 weeks (4-Wk) after infarct(INF) for a representative animal. Plotsdisplay three midventricular slices fromapex (inner) to base (outer) for remote(REM), border zone (BZ), and infarct (INF)regions. The E1Mag decreases in all regionsat 1 week but increases at 4 weeks in theinfarct. Infarct E1 vectors (E1qRC, E1qRL)shift from radial direction at baseline tomore circumferential-longitudinal, signi-fying decreased wall thickening andincreased tissue stretch. Circumferentialshortening (E3Mag) decreases in all regionsas the left ventricle remodels, whereasinfarct E3 vectors (E3qRC, E3qRL) shift fromcircumferential-longitudinal to a moreradial direction, indicating tissue stretch.

6 PILLA ET AL Ann Thorac SurgSTRAINS DURING POSTINFARCTION REMODELING 2015;-:-–-

Fig 6. Regional intermediate principal strain magnitude and angles during left ventricle remodeling. (A) Contrasting maximum strain magnitude(E1Mag) and minimum strain magnitude (E3Mag), the regional magnitude of E2 (E2Mag) varies significantly with location before infarction (INF).(B) After infarction, E2qRC is altered at 4 weeks, with little change in the remote (REM) and border zone (BZ). (C) The E2qRL is decreased at 1 weekand 4 weeks after myocardial infarction. (D) E2qCL decreases in REM and BZ as LV remodels while INF remains elevated above baseline at 1 weekand 4 weeks. þp < 0.05 versus baseline; *p < 0.05 versus REM; �p < 0.05 versus BZ. (Black bars ¼ baseline; dark gray bars ¼ 1 week; light graybars ¼ 4 weeks.)

7Ann Thorac Surg PILLA ET AL2015;-:-–- STRAINS DURING POSTINFARCTION REMODELING

One week after MI, the orientation of the E1 eigenvectorsdemonstrate that the infarct is stretching in the longitu-dinal and circumferential direction rather than thickeningin the radial direction. Four weeks after MI, the orienta-tion of the E1 eigenvectors indicate progressive stretch inlongitudinal and circumferential directions consistentwith a continued decrease in infarct stiffness. Concomi-tant postinfarction changes in the orientation of the E3

Table 2. Correlation Between Function and Principal Strain atOne Week

Function Principal Strain R Value p Value

EDV E1qRC remote 0.85 0.06E1qRC infarct 0.90 0.03E3 border zone 0.82 0.08E3qRC border zone �1 0.01E3qCL infarct �1 0.01

ESV E1qRC remote 0.81 0.09E1qRC infarct �0.86 0.05E1 border zone 0.92 0.02E3 border zone 0.90 0.08E3qRC border zone �1 0.01

EDV ¼ end-diastolic volume; ESV ¼ end-systolic volume.

eigenvectors are indicative of progressive systolic wallthinning and confirm progressive loss of infarct stiffnessduring post-MI LV remodeling.The changes in principal strain magnitude and angles

over time are indicative of the active role the infarct me-chanical properties play in LV remodeling and function.Rapid loss of contractile function after coronary occlusionimmediately reduces the systolic compliance of theinfarct. Over time, the infarct stiffness decreases further,which increases infarct strain (stretching) and wall stressas the infarct becomesmore compliant. Progressive infarctcompliance increases the workload of the remote

Table 3. Correlation Between Function and Principal Strain atFour Weeks

Function Principal Strain R Value p Value

EDV E3qRL remote 0.90 0.08E3qCL border zone �0.90 0.08

ESV E3qRL border zone 0.85 0.07E3qCL border zone �0.81 0.09E3qRL infarct 0.90 0.08E3qCL infarct �0.84 0.07

EDV ¼ end-diastolic volume; ESV ¼ end-systolic volume.

Table 4. Correlation Between Principal Strain at One Weekand Function From One Week to Four Weeks

Function Principal Strain R Value p Value

EDV E1qRL remote �0.88 0.04E1qCL remote �0.81 0.09E3 infarct 0.92 0.02

ESV E1 remote �0.94 0.01E1qCL infarct 0.94 0.01E2 infarct �0.98 0.003

EF E1 border zone 0.83 0.08E1qRL infarct �0.86 0.06E2 remote �0.90 0.02

EDV ¼ end-diastolic volume; EF ¼ ejection fraction; ESV ¼ end-systolic volume.

8 PILLA ET AL Ann Thorac SurgSTRAINS DURING POSTINFARCTION REMODELING 2015;-:-–-

myocardium by increasing the work lost to stretching inthe infarct region [14]. Increased LV wall stress has alsobeen associated with maladaptive biologic sequelae,particularly matrix metalloproteinase activation [15].

The BZ, which is classified as the hypokinetic regionadjacent to the infarct, is under unique stress due to itsposition between the normally contracting remotemyocardium and the noncontractile infarct [2]. Thisregion undergoes a comparable temporal evolution ofstrain magnitude and angles as the infarct region eventhough it is not part of the initial ischemic insult. TheE1Mag decreases early and remains depressed at the latetimepoint. Angles are rotated away from the preinfarctradial direction but to a lesser degrees than the infarctregion, suggesting there is a component of BZ thickeningin the radial direction and that a portion is stretching inboth the longitudinal and circumferential directions.These observed temporal changes in BZ strain can beattributed to the altered stress distribution due the adja-cent infarct region and are potentially a stimulus forcontinued remodeling [16, 17].

Remote principal strainmagnitudes decreased at 1 weekafter MI, and then subsequently increased at 4 weeks.Improved strain in the remote region may be jointlyattributed to contractile compensation and increasedinfarct compliance. Infarct stretch lessens initial contractileload, increasing the remote strain; however, that results inincreased LV workload for a given stroke volume [14].The orientation E1 eigenvectors (E1qRC, E1qRL, and E1qCL)in the remote region changed little after MI, indicatingmyocardial thickening continued to be predominantly inthe radial direction.Minimumprincipal strain angles werealtered in the remote with remodeling, signifying thatshortening had rotated away from the longitudinal direc-tion to a more circumferential orientation. This rotationwas likely a consequence of reduced LV torsion [18, 19].Decreased torsion due to the noncontracting infarct regionwould impair the ability of the LV to shorten in the longi-tudinal direction, altering the longitudinal components(qRL and qCL) of E3. This decreased shortening would in-crease myocardial stress and oxygen demand byincreasing LV workload and could contribute to continueremodeling and decreased function [14, 20].

An appreciation of the temporal changes in LV wallprincipal strains after MI is crucial to understanding themechanism of LV remodeling as well as to developingnew strategies to halt or reverse the process. Altering theprincipal strains in the infarct and BZ has become apotential therapeutic target to prevent the progression ofLV remodeling after MI. Mechanical and biologicalmethods of modifying stress/strains in the infarct havebeen proposed. These include restraint devices [21, 22]placed over the infarct and stiffening agents injected[23, 24] into the infarct tissue. The successful applicationof these novel therapeutic approaches will depend on abetter understanding of the time-dependent changes thatoccur in healing and mature MI.This study also examined the prognostic value of

principal strains in predicting remodeling. Correlationstudies between the evolution of principal strain andadverse LV remodeling demonstrate that regionalmagnitude and angles have prognostic significance,indicating that principal strain changes precede ventric-ular dilation and performance decline. Previous studiesusing two-dimensional echocardiography measuringnormal strains have concluded that global circumferentialand longitudinal strains were predictive of adverseremodeling after MI [25–27]. These global strain changesare a consequence of the regional alterations. Infarctand BZ principal strains will influence the strains overthe entire LV by modifying systolic and diastolic stressdistribution and blood flow patterns. Regional principalstrain may be a more powerful predictor becauseremodeling after MI likely begins on the regional level,then affects the ventricle globally.Principal strain orientation maybe less affected by

preload and afterload, especially in the infarct and BZregions. Strain orientation in these regions is more afunction of material properties than of contractile load.Accordingly, regional principal strains may be a betterpredictor of which patients will benefit from emerginginfarct modification therapies as well as determining theoptimal timing for such interventions.

References

1. Jackson BM, Gorman JH, Moainie SL, et al. Extension ofborderzone myocardium in post infarction dilated cardio-myopathy. J Am Coll Cardiol 2002;40:1160–7.

2. Jackson BM, Gorman JH, Salgo IS, et al. Border zone ge-ometry increases wall stress after myocardial infarction:contrast echocardiographic assessment. Am J Physiol HeartCirc Physiol 2003;284:H475–9.

3. Axel L, Dougherty L. MR imaging of motion with spatialmodulation of magnetization. Radiology 1989;171:841–5.

4. Xu C, Pilla JJ, Isaac G, et al. Deformation analysis of 3Dtagged cardiac images using an optical flow method. Car-diovasc Magn Reson 2010;30:12–9.

5. Zhong X, Spottiswoode BS, Meyer CH, et al. Imaging three-dimensional myocardial mechanics using navigator-gatedvolumetric spiral cine DENSE MRI. Magn Reson Med2010;64:1089–97.

6. Zimmerman SD, Criscione J, Covell JW. Remodelingin myocardium adjacent to an infarction in the pigleft ventricle. Am J Physiol Heart Circ Physiol 2004;287:H2697–704.

9Ann Thorac Surg PILLA ET AL2015;-:-–- STRAINS DURING POSTINFARCTION REMODELING

7. Cupps BP, Pomerantz BJ, Krock MD, et al. Principal strainorientation in the normal human left ventricle. Ann ThoracSurg 2005;79:1338–43.

8. Blom AS, Pilla JJ, Arkles J, et al. Ventricular restraint pre-vents infarct expansion and improves borderzone functionafter myocardial infarction: a study using magnetic reso-nance imaging, three-dimensional surface modeling, andmyocardial tagging. Ann Thorac Surg 2007;84:2004–10.

9. Moore CC, Lugo-Olivieri CH, McVeigh ER, et al. Three-dimensional systolic strain patterns in the normal human leftventricle: characterization with tagged MR imaging. Radi-ology 2000;214:453–66.

10. Pilla JJ, Blom AS, Gorman JH, et al. Early postinfarctionventricular restraint improves borderzone wall thickening dy-namics during remodeling. Ann Thorac Surg 2005;80:2257–62.

11. R€ussel IK, G€otte MJ, Bronzwaer JG, et al. Left ventriculartorsion: an expanding role in the analysis of myocardialdysfunction. J Am Coll Cardiol Img 2009;2:648–55.

12. Buchalter MB, Weiss JL, Rogers WJ, et al. Noninvasivequantification of left ventricular rotational deformation innormal humans using magnetic resonance imagingmyocardial tagging. Circulation 1990;81:1236–44.

13. Gupta KB, Ratcliffe MB, Fallert MA, et al. Changes in passivemechanical stiffness of myocardial tissue with aneurysmformation. Circulation 1994;89:2315–26.

14. Pilla JJ, Gorman JH, Gorman RC. Theoretic impact of infarctcompliance on left ventricular function. Ann Thorac Surg2009;87:803–10.

15. Wilson EM, Moainie SL, Baskin JM, et al. Region and typespecific induction of matrix metalloproteinases in postmyocardial infarction remodeling. Circulation 2003;107:2857–63.

16. Guccione JM, Moonly SM, Moustakidis P, et al. Mechanismunderlying mechanical dysfunction in the border zone of leftventricular aneurysm: a finite element model study. AnnThorac Surg 2001;71:654–62.

17. Moustakidis P, Maniar HS, Cupps BP, et al. Altered leftventricular geometry changes the border zone temporaldistribution of stress in an experimental model of left

ventricular aneurysm: a finite element model study. Circu-lation 2002;106(Suppl 1):I168–75.

18. Nucifora G, Marsan NA, Bertini M, et al. Reduced left ven-tricular torsion early after myocardial infarction is related toleft ventricular remodeling. Circ Cardiovasc Imaging 2010;3:433–42.

19. Jang JY, Woo JS, Kim WS, et al. Serial assessment of leftventricular remodeling by measurement of left ventriculartorsion using speckle tracking echocardiography in patientswith acute myocardial infarction. Am J Cardiol 2010;106:917–23.

20. Beyar R, Sideman S. Left ventricular mechanics related to thelocal distribution of oxygen demand throughout the wall.Circ Res 1986;58:664–77.

21. Koomalsingh KJ, Witschey WR, McGarvey JR, et al.Optimized local infarct restraint improves left ventricularfunction and limits remodeling. Ann Thorac Surg 2013;95:155–62.

22. Pilla JJ, Blom AS, Brockman DJ, et al. Ventricular constraintusing the Acorn cardiac support device reduces myocardialakinetic area in an ovine model of acute infarction. Circula-tion 2002;106(Suppl 1):I207–11.

23. Ryan LP, Matsuzaki K, Noma M, et al. Dermal filler injection:a novel approach for limiting infarct expansion. Ann ThoracSurg 2009;87:148–55.

24. Ifkovits JL, Tous E, Minakawa M, et al. Injectable hydrogelproperties influence extent of post-infarction left ventricularremodeling in an ovine model. Proc Natl Acad Sci 2010;107:11507–12.

25. Hung CL, Verma A, Uno H, et al. Longitudinal andcircumferential strain rate, left ventricular remodeling, andprognosis after myocardial infarction. J Am Coll Cardiol2010;56:1812–22.

26. Cho GY, Marwick TH, Kim HS, et al. Global 2-dimensionalstrain as a new prognosticator in patients with heart failure.J Am Coll Cardiol 2009;54:618–24.

27. Su HM, Lin TH, Hsu PC. Global left ventricular longitudinalsystolic strain as a major predictor of cardiovascular events inpatients with atrial fibrillation. Heart 2013;99:1588–96.