Background
The identification of patients at risk for developing left ventricular (LV) remodeling after acute myocardial infarction (AMI) has crucial prognostic implications. The aims of this study were (1) to investigate the relationship between peak subepicardial and subendocardial twist and infarct transmurality, as assessed using contrast-enhanced magnetic resonance imaging, and (2) to evaluate the association between peak subepicardial and subendocardial twist and LV remodeling 6 months after AMI.
Methods
A total of 213 patients with ST-segment elevation AMIs who underwent three-dimensional echocardiography for LV volumes and functional assessment and two-dimensional speckle-tracking analysis for the evaluation of LV twist (subendocardial vs subepicardial) were retrospectively included. A subgroup of 40 patients underwent magnetic resonance imaging within 2 months for infarct size quantification.
Results
Peak subepicardial twist was strongly related to infarct size (number of segments with transmural scar: r 2 = 0.526, P < .001; total scar score: r 2 = 0.515, P < .001) compared with peak subendocardial twist (number of segments with transmural scar: r 2 = 0.379, P < .001; total scar score: r 2 = 0.331, P < .001). In the overall population, 44 patients (21%) developed significant LV remodeling at 6-month follow-up (LV end-systolic volume increase ≥ 15%). These patients showed significantly more impaired peak subepicardial and subendocardial twist at baseline compared with patients without LV remodeling (4.5 ± 1.3° vs 9.4 ± 3.5°, P < .001; 7.0 ± 3.2° vs 12.9 ± 5.8°, P < .001, respectively). Importantly, peak subepicardial twist (odds ratio, 0.241; 95% confidence interval, 0.134–0.431; P < .001) and peak troponin T (odds ratio, 1.152; 95% confidence interval, 1.006–1.320; P = .041) were independently associated with the development of LV remodeling.
Conclusions
Peak subepicardial twist strongly reflects infarct transmurality as assessed with magnetic resonance imaging and is independently associated with LV remodeling after AMI.
Left ventricular (LV) dilatation and dysfunction after acute myocardial infarction (AMI) have crucial impacts on long-term outcomes, increasing patient morbidity and mortality. Therefore, the early identification of patients at high risk for developing LV remodeling after AMI may have important clinical implications in terms of patient monitoring and additional treatment options. In this regard, among different clinical parameters, several previous studies have demonstrated that infarct size is one of the strongest predictors of LV remodeling after AMI.
LV twist (i.e., the systolic wringing motion derived from the counter-rotating of the LV apex and base) has emerged as an accurate index of LV systolic performance. Impairment of LV twist has been described in acute and chronic ischemic heart disease, and several experimental studies have suggested that the impact of myocardial ischemia on LV rotational mechanics is dependent mainly on infarct transmural extension (from the subendocardium to the subepicardium) rather than on infarct location. Previously, cardiac mechanics were studied and quantified implanting gauges, radiopaque markers, and ultrasound crystals in the myocardium of animal experiments. Currently, two-dimensional speckle-tracking echocardiographic analysis allows a noninvasive accurate assessment of LV twist and has been validated against sonomicrometry and tagged magnetic resonance imaging (MRI) techniques. Furthermore, this technique permits separate quantification of myocardial deformation in the subendocardial and subepicardial layers. However, there is limited evidence showing the differences in LV subendocardial and subepicardial twist in patients with AMIs. In particular, the relationship between LV subendocardial and subepicardial twist and the extent of infarct transmurality has not been explored. More importantly, the potential value of LV subendocardial and subepicardial twist assessments, which may reflect differently myocardial infarction extension, to identify patients at risk for developing LV remodeling at follow-up has not been elucidated. Therefore, aim of the present evaluation was twofold: (1) to investigate the relationship between peak subepicardial and subendocardial twist and infarct transmurality as assessed with contrast-enhanced MRI (standard reference) and (2) to evaluate the association between peak subepicardial and subendocardial twist and LV remodeling at 6 months in a large group of patients with ST-segment elevation AMIs.
Methods
Patient Population and Study Protocol
A total of 245 patients admitted with first ST-segment elevation AMIs and treated with primary percutaneous coronary intervention were included. All patients underwent the institutional ST-segment elevation AMI protocol (MISSION!), which is designed to optimize the management of ST-segment elevation AMI and involves a prehospital, in-hospital, and outpatient clinical framework, as previously reported. The infarct-related artery was identified by electrocardiographic criteria and confirmed using coronary angiography. Final Thrombolysis in Myocardial Infarction flow was assessed during PCI. Patient data were prospectively collected in the departmental cardiology information system (EPD-Vision; Leiden University Medical Center, Leiden, The Netherlands) and retrospectively analyzed.
All patients underwent three-dimensional (3D) echocardiography for LV volumes and LV ejection fraction (LVEF) calculation and two-dimensional speckle-tracking echocardiography for LV twist measure within 48 hours of admission. LV twist in the subepicardial and subendocardial layers was differentiated. At 6-month follow-up, 3D echocardiography was repeated to obtain an accurate measure of LV volumes (and LVEF) and to evaluate LV remodeling. Analysis of 3D volumes at follow-up was performed blinded to baseline data.
In addition, a subgroup of 40 patients was clinically referred for contrast-enhanced MRI within 2 months from hospitalization. Myocardial infarct size was quantified, and the correlation between the extent of scarred myocardium and LV twist parameters (subendocardial vs subepicardial) was investigated.
3D Echocardiography
Patients were examined in the left lateral decubitus position using a commercially available ultrasound system (iE33; Philips Medical Systems, Bothell, WA) equipped with a 3V phased-array transducer. The apical 3D full-volume data set was derived combining seven R wave–triggered subvolumes acquired from seven consecutive cardiac cycles during one breath hold. The 3D data sets were digitally stored, and image analysis for LV volumes and LVEF quantification was performed offline using QLAB version 7.0 (Philips Medical Systems). The software automatically displays the apical four-chamber and two-chamber views and the parasternal short-axis view. Manual identification of mitral valve edges and LV apex with five reference points on both end-diastolic and end-systolic frames allows the software to automatically identify the entire endocardial border in each frame; a manual correction of the endocardial contour is possible if needed. Consequently, a 3D model is generated, providing automatic quantification of LV volumes and LVEF. LV volumes were subsequently indexed to body surface area. LV remodeling was defined as a ≥15% increase in LV end-systolic volume at follow-up.
Two-Dimensional Speckle-Tracking Echocardiography
Standard two-dimensional images were acquired from parasternal short-axis views at two different levels (basal and apical) using the same ultrasound system described above, equipped with a broadband S5-1 transducer. Frame rates ranged from 60 to 90 frames/sec. The basal short-axis view was identified by the mitral valve plane, whereas the apical short axis was recognized as the smallest cavity distal to the papillary muscles (just proximal to the level with end-systolic luminal obliteration). Images were digitally stored for offline quantitative analysis with the QLAB version 7.0 software.
LV Twist Analysis
In each short-axis view, manual delineation of the subendocardial and subepicardial border at the end-diastolic frame identified a region of interest (ROI) with high spatial resolution (multiple kernel regions of approximately 10–50 pixels depending on image resolution), which included the entire myocardial wall and was automatically divided into two separate muscular layers (subendocardial and subepicardial) of the same dimension. The software allowed manual validation of the speckle-tracking quality and adjustment of the ROI if necessary. In each short-axis view, the ROI was automatically divided into six standard segments, and LV rotation of the subendocardial and subepicardial layers was calculated as the average frame-by-frame angular displacement (referring to the ventricular centroid) of the six segments in basal and apical images. As previously described, the clockwise rotation of the LV base (as viewed from the LV apex) is defined as a negative value. Conversely, the LV apex is characterized by counterclockwise rotation, which is positive.
For the calculation of LV twist, the results of the rotation-tracking analysis were exported to a spreadsheet file. From that, three parameters were selected in both apical and basal data sets: peak subendocardial, subepicardial, and global rotation. Furthermore, LV twist was defined as the net difference (in degrees) of the apical and basal rotation at isochronal time points. Peak LV twist values were therefore calculated from apical and basal rotation data for global (entire ROI), subendocardial (subendocardial 50% of the ROI), and subepicardial (subepicardial 50% of the ROI) layers ( Figure 1 and 2 ).
Reproducibility for multilayer assessment of LV rotational parameters was evaluated.
MRI
Images were acquired during 15-sec breath holds using a 1.5-T Gyroscan ACS-NT/Intera MRI scanner (Philips Medical Systems, Eindhoven, The Netherlands) equipped with a five-element cardiac synergy coil and vector electrocardiographic gating. Contrast-enhanced images (20–24 short-axis slices) were acquired 15 min after a bolus injection of 0.15 mmol/kg gadolinium diethylenetriamine pentaacetic acid (Magnevist; Schering/Berlex, Berlin, Germany) with an inversion recovery gradient-echo sequence (field of view, 400 × 400 mm 2 ; matrix size, 256 × 256; slice thickness, 5 mm; slice gap, 5 mm; flip angle, 15°; echo time, 1.36 msec; repetition time, 4.53 msec).
Myocardial Infarct Size Quantification
Image analysis was performed using the MASS software (Medis Medical Imaging, Leiden, The Netherlands). Hyperenhanced regions were defined by selecting an ROI inside the normal myocardium and thresholding the window setting at 5 standard deviations above the mean signal intensity. Contrast-enhanced images were visually scored using a 17-segment model. Each segment was graded using a 5-point scale segmental scar score: 0 = absence of hyperenhancement, 1 = hyperenhancement extending from 1% to 25% of LV wall thickness, 2 = hyperenhancement of 26% to 50%, 3 = hyperenhancement of 51% to 75%, and 4 = hyperenhancement of 76% to 100%. Accordingly, segments with scar scores of 3 or 4 were classified as having transmural scar. The number of segments with scar scores of 3 and 4 reflected the scar transmurality in the infarct zone. The total scar score, defined as the summed segmental scar scores per patient divided by the total number of segments, was used as an index of the total scar burden.
Statistical Analysis
Normality of continuous variables was checked using Kolmogorov-Smirnov and Shapiro-Wilk tests. For reasons of uniformity, summary statistics for all continuous variables are presented as medians and interquartile ranges. Categorical variables are expressed as absolute numbers and percentages. Differences in continuous variables between two groups were evaluated using Student’s t test or the Mann-Whitney U test as appropriate. Differences in categorical variables were assessed using χ 2 tests.
The relationship between LV twist (subepicardial and subendocardial) and the number of transmural segments on the MRI analysis was nonlinear. Therefore, the r 2 value of each model was assessed after performing a logarithmic transformation of LV twist. The same procedure was used to evaluate the correlation between LV twist and total scar score.
To test the correlation between LV twist and changes in LV end-systolic volume at follow-up, a logarithmic transformation of twist was performed (given their nonlinear relationship), and the strength of the correlations was estimated from the r 2 value of the models. In addition, logistic regression analysis was performed to assess potential parameters independently related to LV remodeling. Variables with P values < .05 in univariate analysis were included as covariates in the multivariate model. A correlation coefficient of <0.7 was set to avoid multicollinearity between univariate parameters. Receiver operating characteristic curve analysis was performed to test the value of peak subendocardial and subepicardial twist to identify patients at risk for developing LV remodeling, and the optimal cutoff value of peak subepicardial and subendocardial twist for predicting LV remodeling was defined as the value providing the highest sum of sensitivity and specificity. Considering that subepicardial twist seems to better reflect infarct transmurality, the incremental value of peak subepicardial twist over clinical, conventional echocardiographic variables and peak subendocardial twist for predicting LV remodeling was evaluated by calculating the increase sin χ 2 and C-statistics of the multivariate model.
To test interobserver and intraobserver reproducibility, the data sets of 10 randomly selected patients were reanalyzed (using the first cardiac cycle if not an extrasystole) by the same investigator (≥8 weeks apart) and by a second experienced observer blinded to the former’s data. Intraclass correlation coefficients (ICC) and the absolute difference divided by the mean of the pair-repeated observations (expressed as a percentage) were calculated.
All statistical tests were two sided, and P values < .05 were regarded as statistically significant. Statistical analysis was performed using SPSS version 17.0 (SPSS, Inc, Chicago, IL).
Results
Baseline Clinical and Echocardiographic Characteristics
In 32 patients (13%), the speckle-tracking analysis was not feasible (i.e., at least one segment could not be visualized or the speckle-tracking curves, subendocardial or subepicardial, were not reliable or a true LV apical short-axis view was not achievable), and these patients were therefore excluded from further assessments. Consequently, 213 patients were included in the final analysis.
Clinical characteristics of the patient population (73% men; mean age, 61 years) are summarized in Table 1 . The left anterior descending coronary artery was the most frequently involved infarct-related artery. No patient had atrial fibrillation. Echocardiographic evaluation showed mildly reduced LVEFs (median, 49.0%; interquartile range, 44.1%–53.1%) with LV volumes within the normal range ( Table 2 ). Speckle-tracking-derived rotational parameters of the patient population are reported in Table 2 . Of interest, no significant differences were observed in subendocardial (11.3 ± 5.9° vs 12.0 ± 5.9°, P = .292) and subepicardial (8.2 ± 4.0° vs 8.4 ± 3.6°, P = .363) twist parameters between patients with and without LAD coronary artery territory infarction.
