Assessment of Myocardial Contractile Function Using Global and Segmental Circumferential Strain following Intracoronary Stem Cell Infusion after Myocardial Infarction: MRI Feature Tracking Feasibility Study

Background. Magnetic resonance imaging (MRI) strain analysis is a sensitive method to assess myocardial function. Our objective was to define the feasibility of MRI circumferential strain (ε cc) analysis in assessing subtle changes in myocardial function following stem cell therapy. Methods and Results. Patients in the Amorcyte Phase I trial were randomly assigned to treatment with either autologous bone-marrow-derived stem cells infused into the infarct-related artery 5 to 11 days following primary PCI or control. MRI studies were obtained at baseline, 3, and 6 months. ε cc was measured in the short axis views at the base, mid and apical slices of the left ventricle (LV) for each patient (13 treatments and 10 controls). Mid-anterior LV ε cc improved between baseline −18.5 ± 8.6 and 3 months −22.6 ± 7.0, P = 0.03. There were no significant changes in ε cc at 3 months and 6 months compared to baseline for other segments. There was excellent intraobserver and interobserver agreement for basal and mid circumferential strain. Conclusion. MRI segmental strain analysis is feasible in assessment of regional myocardial function following cell therapy with excellent intra- and inter-observer variability's. Using this method, a modest interval change in segmental ε cc was detected in treatment group.


Introduction
Circumferential strain ( cc ) analysis is an established method to assess myocardial function. cc has been demonstrated to detect changes in myocardial contractility across a variety of cardiac conditions including hypertensive [1] or hypertrophic cardiomyopathy [2] and Duchenne muscular dystrophy [3] before changes in le ventricular ejection fraction (LVEF) are observed. Cell therapy offers a promising approach for regeneration of damaged vascular and cardiac tissue aer acute myocardial infarction (MI) [4][5][6][7][8]. e Amorcyte trial evaluated effect of autologous bone marrow derived CD34+ cell therapy on LVEF and myocardial perfusion [9]. Signi�cant improvement in myocardial perfusion and a trend towards improvement in LV ejection fraction were reported. Because the changes seen in systolic function are modest and by design regional, the more sensitive myocardial strain techniques offer an attractive option for analyzing these results. Accordingly, we evaluated the feasibility of MRI-derived segmental cc analysis in patients treated with cell therapy following primary intervention for ST-segment elevation myocardial infarction (STEMI), using the feature tracking (FT) technique. e FT technique was previously validated for cc assessment against harmonic phase imaging (HARP) [10] and subsequently utilized for assessment of cardiac involvement in Churg Strauss disease [11] and in dobutamine cardiac MRI stress testing [12]. One of the major advantages of this technique is that additional imaging is not needed, as analysis is performed on cine (steady-state free precession, SSFP) images and does not require tagging.

Study Population and Protocol in the Amorcyte Study.
Subjects with STEMI who were treated with intracoronary stent implantation within 3 days of infarction and an LVEF ≤50% by echocardiography demonstrating regional wall motion abnormality in the distribution of the infarct-related artery ≥4 days aer stenting were enrolled [9]. Brie�y, the treatment groups patients were treated with escalating dose of stem cells (5, 10, and then 15 million CD34+ cells) and compared with control. At 24-48 hours aer bone marrow harvest, CD34+ cells were infused into the infarct-related artery through an over-the-wire balloon catheter placed within the previously stented segment using a stop �ow technique [13].

MRI Imaging.
All studies were performed using 1.5 Tesla MRI scanners. Cardiac functional imaging was performed with retrospective electrocardiogram (ECG) gating, utilizing a standard SSFP technique, LV end-systolic and end-diastolic volumes, LVEF, and infarct size at baseline, 3 months, and 6 months were assessed as described previously [9].

Image Analysis.
Using Diogenes CMR FT soware (TomTec Imaging Systems, Munich, Germany), global cc was measured in le ventricular short axis steady-state free precession (SSFP) views at base, mid, and apex for each patient at baseline, 3 months, and 6 months by 2 independent readers who were blinded to treatment assignment. In addition, segmental cc was measured for all individual midventricular segments (segments 13 through 18), Segmental strain analysis was not undertaken in apical and basal segments due to low reproducibility (for apical segments) and no infarcted tissue present (basal segments). irteen treatments and 10 controls underwent strain analysis.

Diogenes CMR FT Soware.
Diogenes CMR FT soware (TomTec Imaging Systems, Munich, Germany) is a vector-based analysis tool based on a hierarchical algorithm that operates at multiple levels using a combination of 1-dimensional and 2-dimensional tracking techniques. Based on a contour manually drawn by an expert reader along the LV endocardial border frame, the soware automatically propagated the contour and followed its features throughout the remainder of the cardiac cycle. Features tracked in each voxel (the 3-dimensional analogue of a pixel) by the soware include brightness gradient at the tissue-cavity interface in homogeneities of the tissue (with respect to a 256-level gray scale), geometrical "roughness" of the tissue edges, and additional speci�c anatomical elements (papillary muscles and septum). Tracking is based on optical �ow algorithms applied on sequentially reduced windows to ensure resolving large amplitude displacements and still discern local differences, and also implicitly accounts for spatial coherence and timeperiodic motion. e tracking over all frames of a set of individual material points along the border permits to quantify their frame by frame displacement. e circumferential strain is de�ned as the instantaneous local border lengthening or shortening: if indicates the distance between two neighboring points along the border, the strain is computed as / , where indicates such length at the ECG R wave. e systolic strain is therefore / . is represents the Lagrangian strain and it is the most frequently used in cardiac imaging [14][15][16]. Strain values are computed at 48 points along the border, the entire border is then automatically subdivided into six segments of same length and segmental strain is computed as the average strain of all the points falling in that segment. e global strain is then de�ned as the average of all segmental strain values. Analysis can be accomplished very quickly, similar to speckle tracking in echocardiography [17] (this reference includes video of actual FT procedure).

Statistical Analysis.
Global circumferential strain data were summarized as mean ± SD as well as the median. Due to small sample size, comparisons within each group in change from baseline for the global circumferential strain at 3 months and 6 months were conducted using the Wilcoxon signed-rank test. Similarly, change from baseline in circumferential strain at mid-anterior wall at 3 months and 6 months was conducted using the Wilcoxon signed-rank test.
Bland and Altman [18] analysis was used to test the interobserver and intraobserver agreement for global circumferential strain at base, mid, and apex. e test statistic [19] for testing the readings agreement between observer 1 and observer 2 was based on √ / √ 1 , where follows the distribution with k-2 degrees of freedom, is the correlation between the average readings ([observer 1 + observer 2]/2) and the difference (observer 1 -observer 2) with zero or small indicating an agreement between methods (i.e., no systematic bias between readers), and is the paired sample size.
Infarct size as a function of dose and changes in mid anterior strain at baseline, month 3, and month 6 was analyzed using analysis of variance (ANOVA).
Statistical signi�cance was set at . Statistical analysis was performed using SAS soware version 9.2 (SAS Institute, Inc., Cary, NC).

Results
Most patients (9/10 in the control group and 12/13 in the treatment group) had evidence of anterior wall myocardial infarction by late gadolinium enhancement (LGE). ere was one lateral infarct in the control group and one inferior in the treatment group.
No difference was detected in the treatment or control groups in global cc at any level, except for the basal slice in the treatment group, the change from baseline and 3 months was statistically signi�cant (−18.4 ± 14.4 versus −24.4 ± 4.8, . , Table 1). Mid-anterior cc in the treatment group showed improvement between baseline −18.5 ± 8.6 and 3 months −22.6 ± 7. , . with no further improvement at 6 months −2 . ± 12.4, . 4 ( Figure 1). ere was no statistically signi�cant difference in mid-anterior segments of the control group (−19.7 ± 7.3 to −18.6 ± 10.7, . 1) at 3 months and no changes were observed at 6 months (−19.7 ± 7.3 to −23.0 ± 8.4, . 6). No changes were noted in any other segments in the mid-ventricular short axis between any time points.
Bland and Altman analyses showed an excellent interobserver agreement for global cc at basal, mid, and mid segmental with (Pearson's correlation between average and difference readings ) − . , .8 , .1 , .44, and . 8, .14, respectively. However, there was a poor interobserver agreement on apex with − .4 , .
. In addition, there was an excellent intraobserver agreement for global cc at basal, mid, and apex with − . 7, .6 , .1 , .2 , and .26, .14, respectively. In the treatment group, ANOVA analysis showed no signi�cant difference between the doses of stem cell, size of infarct at baseline, and changes in mid-anterior cc at baseline, Month 3, and Month 6. As previously reported, there were no signi�cant differences in the change in infarct size or LVEF in treatment versus control groups aer 3 or 6 months.
Example of patient in the treatment group is presented in Figure 2.

Discussion
e ma�or �nding of this substudy from the Amorcyterandomized trial was feasibility of detection of segmental cc changes using FT technique in patients treated with intracoronary bone-marrow-derived CD34+ cell infusion aer acute ST elevation MI. is is of importance as FT analysis does not require additional imaging (such as tagging) and it can be used retrospectively as cine sequences are obtained in all studies while tagging is typically not a part of the standard protocol. FT technique is speci�cally designed for MRI analysis and differs from recently reported velocity vector imaging (VVI) soware (which was designed for echocardiographic speckle tracking) [20]. In addition, we demonstrated that patients who received cell therapy had a statistically signi�cant improvement in the mid-anterior cc without signi�cant changes in LVEF at 3 but not at 6 months. As most of the analyzed cohort suffered anterior wall MI, the improvement in cc within this infarcted region was likely driven by the bene�cial effects of cell therapy. Finally, both inter-and intraobserver agreement, for strain analysis were excellent for mid and basal segments. Interestingly, intraobserver (but not interobserver) variability was poor for apical segment; possible explanation is presence of apical F 2: Late gadolinium enhancement and mid-anterior cc for baseline, 3, and 6 months. ere was initial improvement at 3 months followup, but deterioration at 6 months, due to le ventricular remodeling.
trabeculations, which have been de�ned as endocardium by one reader versus compact myocardium used by other reader. Improvement in 2D echo-derived LV strain has been reported by Herbots et al. [21] who demonstrated significant improvement in the global longitudinal strain rate in patients undergoing cell therapy following acute MI. Strain rate improved in infarcted areas and appeared to be a more sensitive marker of response to therapy than was change in global le ventricular EF. Similarly, Plewka et al. [22] reported improvement in systolic strain measurements involving infarcted segments in patients who received cell therapy. Karatasakis et al. [23] reported improvement in longitudinal strain in patients with prior anterior wall MI following cell therapy in the absence of concomitant revascularization. e current study is the �rst, to our knowledge, to demonstrate the utility of MRI FT-derived segmental cc in the settings of stem cell therapy. Furthermore, the broad spectrum of data available from MRI (valve assessment, LVEF, infarct mass by LGE, and strain) may obviate the need for echocardiography following cell therapy aer MI. Finally a recent study by Hopp et al. [24] which is a substudy of Autologous Stem Cell Transplantation in acute myocardial infarction trial (ASTAMI) explored global and regional myocardial function aer intracoronary injection of autologous mononuclear bone marrow cells in acute anterior wall myocardial infarction treated with percutaneous coronary intervention using MRI tagging. e authors found that intracoronary injection of autologous mononuclear bone marrow cells did not in�uence regional or global myocardial function in this substudy.
Segmental cc may be an attractive and sensitive surrogate endpoint for evaluation of bene�ts of stem cell therapy. A recent meta-analysis of postinfarct stem cell trials concluded that changes in LVEF were very modest and strain analysis may be amore useful and sensitive endpoint [25]. Indeed, in the present analysis, infarcted (and treated) mid-anterior segmental cc improved signi�cantly in the absence of an appreciable change in LVEF. Although the clinical signi�cance of improvement in segmental cc in the absence of a more global increase in LV contractility is yet to be determined; such changes may predict a bene�cial response to therapy with respect to LV remodeling and the degree of infarct zone electrical heterogeneity (which may be associated with increased risk for ventricular reentrant arrhythmias) [26]. Further studies examining the relationship of cc to both global measures of LV volume or function over time and to meaningful clinical outcomes are clearly warranted.

Limitations
Small study population also limited the evaluation of a relationship between infarct transmurality, heterogeneity, presence of microvascular obstruction, myocardial hemorrhage, and the subsequent likelihood of response to stem cell therapy. Further improvement in FT soware resolution will be needed to assess changes in cc in the peri-infarct zone ("gray zone") which is more likely to provide a favorable environment for infused stem cells than the fully infarcted tissue. Additionally, only mid-myocardial strain was analyzed; it is possible that additional changes would be detected with analysis of endocardial strain. As this method relies on "perfect alignment" of analyzed SSFP images at baseline and followup, a small misalignment may miss the edge of infarct and affect the results. In addition, LGE images need to be obtained at exactly the same level as SSFP images, in order to assess changes in contractility in the infarct and peri-infarct regions.

Conclusions
In conclusion, FT method can detect regional changes in cc following intracoronary stem cell infusion aer myocardial infarction. e relationship between these variable and clinical outcomes requires further study.