In the acute phase of myocardial infarction (MI), infarct size and microvascular obstruction (MVO) are important prognostic factors for cardiovascular outcome. MI size is a major determinant of myocardial function, but the specific effect of MVO is less documented. The aim of this study was to evaluate the impact of MVO on longitudinal myocardial strain assessed by speckle-tracking echocardiography.
Speckle-tracking echocardiography and contrast-enhanced cardiac magnetic resonance studies were performed in 69 patients 72 hours after first acute MI. Segmental and global longitudinal systolic strain (ε L ) was measured using speckle-tracking echocardiography. Transmural extent of MI, MI size, and the presence or absence of MVO were assessed using contrast-enhanced cardiac magnetic resonance. Left ventricular (LV) ejection fraction was assessed at 6 months using echocardiography.
The mean infarct size was 23 ± 13% of LV mass. MVO was present in 64% of patients. MVO was significantly associated with ε L impairment (−7.8 ± 4.9% vs −16.3 ± 6.4%, P < .001), and ε L remained significantly worse in MVO-positive segments after adjustment for transmural extent of MI. A ε L value > −12.5% predicted the presence of MVO with 83% sensitivity and 75% specificity. On multivariate analysis, global ε L and MI size, but not MVO, were identified as independent predictors of LV ejection fraction at follow-up (β = −0.9, P = .023, and β = −0.2, P = .034, respectively).
In the acute phase of MI, segmental and global ε L is significantly altered by the presence of MVO, in addition to MI size. However, MI size but not MVO independently predicts LV ejection fraction at follow-up.
After reperfusion, the final infarct size and the presence of microvascular obstruction (MVO) are two major determinants of left ventricular (LV) remodeling and prognosis in patients with myocardial infarction (MI).
MVO is a complex and dynamic phenomenon that leads to impaired tissue perfusion at the microvascular level despite adequate restoration of epicardial vessel patency. Its presence is additional evidence of the severity of the myocardial injury caused by prolonged myocardial ischemia together with reperfusion injury. Also, MVO appears as an independent predictor of contractile recovery and adverse clinical events. There is a significant relationship between MI size and the presence of MVO, but the effect of their interaction on regional and global contractile function and recovery is not clearly defined.
Recent experimental and clinical studies have shown that myocardial deformation determined using speckle-tracking echocardiography (STE) in the acute phase of MI is closely related to MI size as quantified by contrast-enhanced cardiac magnetic resonance (CMR). Moreover, it has been shown that myocardial longitudinal systolic strain (ε L ) predicts LV remodeling and has additional prognostic value in the acute phase of MI. However, none of these studies has assessed the impact of MVO in addition to MI size on global and regional ε L in the acute phase of MI.
The main objectives of our study were to assess the impact of MVO (1) on segmental myocardial strain in addition to MI transmural extent (TME) and (2) on global myocardial strain in addition to MI size in the acute phase of MI. We also sought to determine factors associated with myocardial systolic recovery as assessed by LV ejection fraction (LVEF) at 6-month follow-up.
From December 2006 to February 2008, 69 consecutive patients were prospectively included with first acute ST-segment elevation MIs (STEMIs) according to standard electrocardiographic and enzymatic criteria. Patients presented within 24 hours of symptom onset and were referred for primary percutaneous coronary intervention or thrombolysis followed by percutaneous coronary intervention with stent implantation.
Exclusion criteria were prior MI, underlying cardiomyopathy, hemodynamic instability, atrial fibrillation, significant valvular heart disease, and contraindications to contrast-enhanced CMR. All patients received medical treatment according to guidelines.
The local ethics committee approved the study protocol. Written informed consent was obtained from all patients.
CMR imaging studies were performed <72 hours after admission using a 1.5-T whole-body scanner (Magnetom Avanto; Siemens Medical Solutions, Erlangen, Germany).
LV function at rest was assessed using retrospective electrocardiographically gated steady-state free-precession pulse cine sequences in long-axis and short-axis views (repetition time, 3.2 msec; echo time, 1.6 msec; slice thickness, 6.5 mm; temporal resolution, 35–50 ms; matrix size, 256 × 184). The short-axis scans covered the whole left ventricle with contiguous slices.
Contrast enhancement was assessed 10 min after an intravenous dose of 0.2 mmol/kg gadolinium-DOTA (Dotarem; Guerbet, Roissy, France) using a three-dimensional gradient spoiled turbo fast low-angle shot sequence with a selective 180° inversion recovery prepulse, in the short axis, covering the whole ventricle (repetition time, 1.6 msec; echo time, 1.4 ms; inversion time individually determined to null the myocardial signal, range 180–250 msec; slice thickness, 5 mm; matrix size, 256 × 192). Two or three additional long-axis views with a similar two-dimensional sequence were performed.
Contrast-Enhanced CMR Imaging Analysis
LVEF, LV end-diastolic volume, LV end-systolic volume, and absolute myocardial mass were calculated for each study using Argus postprocessing software (Siemens Healthcare, Erlangen, Germany) using short-axis volumetry.
MI was identified by contrast enhancement within the myocardium, defined quantitatively by a myocardial postcontrast signal intensity > 2 standard deviations above that within a reference region placed in the remote noninfarcted myocardium in the same slice. Infarct size was quantified on the three-dimensional data sets by manual planimetry of the hyperenhanced myocardium using the postprocessing imaging software Osirix (Osirix Foundation, Geneva, Switzerland). For all slices, absolute infarct size in grams was measured according to the following formula: infarct size (g) = Σ(hyperenhanced area [cm 2 ]) × slice thickness (cm) × myocardial specific density (1.05 g/cm 3 ).
For segmental infarct extension analysis, the myocardium was divided into 16 segments. The TME of each segment was assessed. Subendocardial infarct was defined as TME ≤ 50% of the segmental myocardial area, whereas transmural infarct was defined when TME > 50%.
The presence of MVO was defined in each segment by the presence of hypoenhancement within the hyperenhanced area on the contrast-enhanced studies. MVO size was also measured by manual planimetry of the hypoenhanced myocardium according to the same method used for the infarct size measurement. Two typical examples from our CMR study data set demonstrating a patient with MVO and another without are presented in Figure 1 .
Examinations were performed within 72 hours after reperfusion using a commercially available system (Vivid 7; GE Vingmed Ultrasound AS, Horten, Norway). Three cardiac cycles were stored for each view of the left ventricle: parasternal short-axis view at the papillary muscle level and apical four-chamber, two-chamber, and long-axis views. Particular care was taken to obtain optimal quality recordings of all LV walls. Two-dimensional grayscale images were obtained at frame rates of 76 to 90 frames/sec and stored for further strain analysis.
Echocardiographic Image Analysis
Speckle-tracking analysis was performed offline using dedicated software (EchoPAC; GE Vingmed Ultrasound AS). A 16-segment LV model was obtained from the four-chamber, two-chamber and long-axis recordings for quantitative analysis. Longitudinal systolic strain was measured from apical views. Strain was measured in the entire segment as obtained by the automatic EchoPAC segmentation during one cardiac cycle. Selection of the cardiac cycle was left to the reader’s decision (best quality of the tracking process). Longitudinal systolic strain was defined as the maximal peak value during systole or at the time of aortic valve closure (end-systole) if the peak value occurred at aortic valve closure ( Figure 2 ). The tracking process was automated from the end-systolic frame and corrected manually if necessary to obtain an optimal strain curve result. In case of poor acoustic signal, and loss of reproducibility (>10% between two consecutive measurements), segments were excluded from analysis. Peak systolic strain measurements were averaged from apical views to obtain global ε L .
All patients were scheduled for follow-up at 6 months to assess their clinical status as well as their LVEFs using transthoracic echocardiography within our institution or outside our institution by referring cardiologists.
All data were analyzed using standard statistical software (SPSS version 15.0; SPSS, Inc, Chicago, IL). Normally distributed variables are expressed as mean ± SD and were compared using Student’s unpaired t tests between groups and Student’s paired t tests within groups. Analysis of variance was applied if more than two groups were being compared. Data not normally were compared using Mann-Whitney tests for between-groups comparisons and are expressed as median (interquartile range). Categorical variables were compared using χ 2 or Fisher’s exact tests, as appropriate, and are expressed as counts and/or percentages.
In subsequent myocardial segment analysis, three groups were defined according to TME—no TME, TME ≤ 50%, and TME > 50%—and also divided into two groups according to the presence or absence of MVO. To predict infarct transmurality and MVO presence as defined by contrast-enhanced CMR, receiver operating characteristic (ROC) curves were constructed for ε L . The k-fold method was used to cross validate ROC curves.
Interobserver and intraobserver variability for ε L was assessed by two independent readers (C.B. and C.L.R.) blinded to previous measurements on complete echocardiographic data sets of 15 random patients from our study population. Reproducibility was calculated as the absolute difference of ε L between each observer’s value divided by the mean of both measurements and was expressed as a percentage. Intraobserver and interobserver intraclass correlation coefficients were calculated in a two-way mixed model with 95% confidence intervals (CIs).
To investigate the impact of MVO on regional and global strain, we performed univariate and multivariate linear regressions: (1) segmental ε L as the dependent variable and TME with MVO as independent variables and (2) global ε L as the dependent variable and baseline characteristics including age, gender, body mass index, diabetes, ischemic time, MI size, and MVO size as independent variables.
We also assessed the predictive value of baseline characteristics and 6-month LVEF. Univariate and multivariate linear regressions were performed to assess the relationship between covariates and 6-month LVEF. The covariates that were used in the univariate regression analysis included age, gender, body mass index, diabetes, ischemic time, baseline LVEF , global ε L, MI size, and MVO presence. For multivariate analysis, we performed a stepwise backward multivariate regression on all variables assessed in the univariate analysis, with a removal P value threshold of >.10. Models were tested for collinearity: eigenvalue, condition index, and variation inflation factor were assessed for each multivariate model. Otherwise, P values < .05 were considered statistically significant.
The main clinical characteristics of the study population are summarized in Table 1 .
|Age (y)||55 ± 12|
|Diabetes mellitus||13 (19%)|
|Current smoking||43 (69%)|
|MI-related data at admission|
|Killip class I/II||63 (91%)/6 (9%)|
|Systolic blood pressure (mm Hg)||125 ± 24|
|Heart rate (beats/min)||79 ± 16|
|Initial thrombolysis||12 (17%)|
|Post-PCI TIMI flow grade ≥ 2||64 (94%)|
|Peak troponin I value (μg/L)||97 ± 89|
|Peak total creatinine kinase value (UI/L)||3,082 ± 2,393|
|Infarct-related coronary artery|
|Left anterior descending||29 (42%)|
|Left circumflex||7 (11%)|
|Multivessel disease||20 (30%)|
|Treatment after PCI|
|Angiotensin-converting enzyme inhibitors||59 (85%)|
|Aldosterone antagonists||32 (46%)|
|Platelet glycoprotein IIb/IIIa inhibitors||42 (65%)|
Feasibility and Reproducibility
Contrast-enhanced CMR images were adequate for TME and MVO analysis in 1,098 (98.5%) and 1,081 (98%) segments, respectively. Analysis of ε L by STE was feasible in 1,072 segments (97%).
Interobserver variability was 11.8% (95% CI, 7.5%–18.7%) and intraobserver variability was 2.6% (95% CI, 1.3%–5.8%) for global ε L . Interobserver and intraobserver intraclass coefficients were 0.89 (95% CI, 0.70–0.96; P < .001) and 0.98 (95% CI, 0.96–0.99; P < .001), respectively.
In the whole population, average infarct size was 33 ± 20 g (23 ± 13% of LV mass). The total number of infarcted LV segments was 8.8 ± 2.6 per patient, among which 3.2 ± 2.5 had TME > 50%. Infarct size was not associated with time to reperfusion ( r = 0.15, P = .24).
MVO was present in 44 patients with MIs (64%). MVO mass was significantly correlated with MI size ( r = 0.70, P < .001). After exclusion of noninfarcted segments, MVO occurred in 10% of segments with TME ≤ 50% and in 45% of segments with TME > 50% (χ 2 = 66.5, P < .001). As shown in Table 2 , LVEF was significantly lower and LV dimensions were higher in patients with MVO than in patients without MVO. Time to reperfusion was not significantly different in both groups ( Table 2 ). As presented in Figure 3 , there were significant correlations between MI size, MVO, and LVEF.
|Global function (n = 69 patients)||44 patients||25 patients|
|Time to reperfusion (hours)||4.0 (2.5–6.0)||3.3 (2.0–11.5)||.67|
|LVEF (%)||46 ± 8||55 ± 11||.001|
|LVEDV (mL)||150 ± 35||128 ± 35||.02|
|LVESV (mL)||81 ± 25||61 ± 32||.003|
|MI mass (g)||41.4 ± 18.5||17.0 ± 12.1||<.0001|
|MI size (% of LV)||29.1 ± 12.1||13.2 ± 8.8||<.0001|
|Number of segments with TME > 50%||4.3 ± 2.1||1.3 ± 1.7||<.00001|
|MVO mass (g)||7 ± 8||0||<.00001|
|Global ε L (%)||−13.9 ± 3.5||−17.8 ± 3.3||<.001|
|Segmental function (n = 1,081 segments)|
|Contrast-enhanced CMR||118 segments||963 segments|
|TME ≤ 50%||28 (10%)||248 (90%)||<.001|
|TME > 50% (%)||90 (45%)||125 (55%)||<.001|
|STE||114 segments||942 segments|
|ε L (%)||−7.8 ± 4.9||−16.3 ± 6.4||<.001|