Abstract
Background:
Three-dimensional (3D) speckle tracking echocardiography (STE) after ST-elevation acute myocardial infarction (STEMI) is associated with left ventricular (LV) remodeling and 1-year prognosis. This study investigated the clinical significance of 3D-STE in predicting the long-term prognosis of patients with STEMI.
Methods and Results:
A total of 270 patients (mean age 64.6 years) with first-time STEMI treated with reperfusion therapy were enrolled. At 24 h after admission, standard 2D echocardiography and 3D full-volume imaging were performed, and 2D-STE and 3D-STE were calculated. Patients were followed up for a median of 119 months (interquartile range: 96–129 months). The primary endpoint was occurrence of a major adverse cardiac event (MACE: cardiac death, heart failure with hospitalization), and 64 patients experienced MACEs. Receiver operating characteristic curves and Cox hazard multivariate analysis showed that the 3D-STE indices were stronger predictors of MACE compared with those of 2D-STE. Additionally, 3D-global longitudinal strain (GLS) was the strongest predictor for MACE followed by 3D-global circumferential strain (GCS). The Kaplan-Meier curve demonstrated that 3D-GLS >−11.0 was an independent predictor for MACE (log-rank χ2=132.2, P<0.0001). When combined with 3D-GCS >−18.3, patients with higher values of 3D-GLS and 3D-GCS were found to be at extremely high risk for MACE.
Conclusions:
Global strain measured by 3D-STE immediately after the onset of STEMI is a clinically significant predictor of 10-year prognosis.
Two-dimensional speckle tracking strain echocardiography (2D-STE) imaging quantifies even subtle left ventricular (LV) dysfunctions in patients with various cardiac diseases including ST-elevation acute myocardial infarction (STEMI).1
A previous study has shown that 3D-STE can estimate the changes in the LV region, thereby identifying it as a promising tool for the accurate evaluation of regional wall motions.2
We have reported the superior utility of 3D-global longitudinal strain (GLS) in patients with STEMI compared to 2D-GLS in predicting LV remodeling and 1-year prognosis.3
Furthermore, 3D-STE is a useful tool for evaluating cardiac function.4
However, the clinical significance of 3D-STE in patients with STEMI for predicting long-term prognosis remains unclear due to its limited use. Therefore, we conducted a pioneering study involving long-term follow-up examinations to investigate the clinical significance of 3D-STE as a novel predictor of prognosis of patients with STEMI.
Editorial p ????
Methods
Patients
We screened 334 STEMI patients between April 2008 and July 2012 at the Yokohama City University Medical Centre, Yokohama, Japan. All patients who had undergone reperfusion therapy using percutaneous coronary intervention (PCI) within 12 h after the onset of symptoms were enrolled. STEMI was defined as: (1) typical chest pain lasting >30 min; (2) ST-segment elevation >0.1 mV within 2 contiguous leads on initial ECG; and (3) elevation of creatinine phosphokinase (CPK) to twice the upper limit of the normal range. The protocol for patient selection is shown in
Figure 1. The inclusion criteria were as follows: (1) age >20 years; and (2) those who underwent successful revascularization of the infarct-related artery using primary PCI within 12 h of onset of chest pain. The exclusion criteria were as follows: (1) previous myocardial infarction (MI; n=15); (2) significant valvular heart disease (n=8); (3) chronic atrial fibrillation (n=19); and (4) inadequate myocardial tracking by 2D-STE or 3D-STE due to the fair image quality (n=20). The final study population comprised 272 patients who underwent echocardiography approximately 24 h after admission (within 48 h). Reperfusion time was defined as the time from symptom onset to reperfusion (TIMI2). 99 mTc-sestamibi single-photon emission computed tomography (SPECT) was performed 7–14 days after PCI to estimate the final infarct size in 264 patients. Patients were administered an intravenous injection of 16 mCi (600 MBq) 99 mTc-sestamibi, and SPECT was performed 1 h after injection of the radio-active agent. Infarct size was defined as <50% of uptake area, which was measured by the experienced RT (N.T.) and >20% was defined as a large infarction. Highest levels of creatinine and plasma brain natriuretic peptide (BNP) were also measured. Renal function was assessed based on the estimated glomerular filtration rate (eGFR). CPK level and myocardial band (MB) were first determined on admission, then at 3-h intervals during the first 24 h, at 6-h intervals for the next 2 days, and then daily until discharge. A 12-lead ECG was recorded at 24 h after onset at a paper speed of 25 mm/s and an amplification of 10 mm/mV. In addition to ST-segment measurements, we calculated the 32-point QRS score according to the previous study.5
The patients were followed up for a median of 119 months (interquartile range [IQR], 96–129 months; follow-up rate=100%) on regular visits to their attending physicians or by telephone interviews. The primary endpoint was occurrence of a major adverse cardiac event (MACE: incidence of cardiac death and heart failure [HF]). The secondary endpoint was all-cause mortality. HF was defined according to the Framingham criteria for congestive HF with hospitalization. The study was approved by the institutional ethics committee and the provisions of the Declaration of Helsinki. This study was registered in the University Hospital Medical Information Network (UMIN) Clinical Trial Registry (UMIN 000001917).
Both 2D and 3D echocardiography were performed using iE33 (Philips Medical System, Andover, Massachusetts) by an experienced cardiologist (identified as N.I.). Patients were examined in the left lateral or supine position based on the results of the precordial 2D and 3D echocardiography. LV volumes (end-diastolic volume [EDV], end-systolic volume [ESV]), and ejection fraction (EF) were calculated using 3D echocardiography. Left atrial (LA) volume was calculated using the area length method. The mitral inflow peak early velocity (E)/mitral annular peak early velocity (e′) or the E/e′ ratio and severity of mitral regurgitation were visually assessed. Based on previous guidelines, 3D full-volume data were acquired from the apical view while the patients held their breath using a matrix array transducer (X3-1; Philips Medical Systems). To ensure the inclusion of the entire LV region within the pyramidal scan volume with a relatively high volume rate, data sets from 4 cardiac cycles were acquired using the wide-angle mode, wherein multiple wedge-shaped sub-volumes were acquired with electrocardiographic gating for at least 5- to 7-s single breath holds (stitched data).
2D-Speckle Tracking Echocardiography
The 2D-STE was performed using vendor-independent 2D speckle-tracking software (2D Cardiac Performance Analysis Ver 1.3; TomTec Imaging Systems, Germany) based on the appropriate method. Longitudinal strain was measured by manual tracing of the endocardial border in 3 apical views (4-chamber, 2-chamber and long-axis views). Circumferential strain was determined by endocardial tracing at 3 levels of the short-axis views. After a frame-by-frame speckle-tracking analysis of the LV endocardium that averaged over 2 cardiac cycles, the software provided regional strain curves of 6 segments (4 segments in the apical view) in each view, from which the peak regional strain value was calculated. Global strain was calculated as the peak strain value from the average of 16 segmental strain curves (2D-GCS and 2D-GLS). Adequacy of tracking was visually verified, and if tracking was considered suboptimal, the endocardial border was manually adjusted.
3D-Speckle Tracking Echocardiography
Three-dimensional strain measurements of the LV were performed using 3D-STE. Three-dimensional full-volume datasets (stitched data) were analyzed using vendor-independent 3D speckle-tracking software (4D LV Analysis, version 3.1; TomTec Imaging Systems, Germany) by an experienced investigator (N.I.). After importing 3D full-volume data sets, the apical 4-chamber, 2-chamber, and long-axis views and 3 short-axis views at end-diastole were automatically extracted. Non-foreshortened apical views were identified to select the point of the apex and the center of the mitral annular line connecting both sides of the mitral annulus with the largest LV long-axis dimensions, after which the 3D endocardial surface was automatically reconstructed. Manual adjustments of the endocardial surface were performed when necessary. The same procedure was performed at the end-systolic frame. Subsequently, the software performed 3D speckle-tracking analysis throughout the cardiac cycle. For 3D strain analysis, LV was automatically divided into 16 segments using standard segmentation schemes. The software provided segmental longitudinal, circumferential, radial, and principal strain time curves, from which peak global strain and average peak strain at 3 LV levels (basal, midventricular, and apical) were determined (3D-GLS, 3D-GCS, 3D-global radial strain [GRS], and 3D-global principal strain [GPS]).6
When tracking was not satisfactory, the endocardial surface was manually readjusted. If tracking was still inaccurate, participants were excluded from the analysis.
Supplementary Figure 1
shows a representative case of the 3D speckle-tracking analysis. The left panels show the endocardial tracking, whereas the right panels show the longitudinal strain curves (A), the circumferential strain curves (B), the radial strain curves (C), and the principal strain curves (D).
Reproducibility
Intra-observer and inter-observer (N.I. and J.K. [second investigator]) variability in 2D-STE and 3D-STE measurements were assessed in 20 randomly selected participants in this study, and the agreement between 3D-GLS and 3D-GCS was assessed using the Bland-Altman analysis,7
with predefined accuracy set as 95% limits of agreement±2SD.
Follow up
The patients were followed up after discharge. For patients experiencing >1 acute event, only the first event was considered in the primary endpoint analysis. All events were followed up by a hospital visit or telephonic interview with an experienced cardiovascular physician blinded to the clinical details and outcomes. Telephonic contact with patients, physicians, and the next of kin was made if the patients had been treated at another hospital.
Statistical Analysis
Most continuous variables were not normally distributed (as evaluated by Kolmogorov-Smirnov tests). For uniformity, summary statistics for all continuous variables are presented as medians with the 25th and 75th percentiles (IQR). Categorical data are summarized as frequencies and percentages. We used the t-test for continuous variables and the Chi-squared test for categorical variables to compare positive and negative variables for MACE. Inter-observer and intra-observer reproducibility was assessed in 20 randomly selected patients using the Bland-Altman analysis.7
To determine the optimal threshold of 2D- and 3D-STE for the prediction of endpoints, receiver operating characteristic (ROC) curve analysis was applied. We compared the area under curves (AUCs) of 3D- and 2D-STE using the DeLong Method.8
In addition, we obtained cut-off values for the LV strain with 3D-GLS and 3D-GCS. We draw Kaplan-Meier curves for MACE for the 4 groups categorized based on the cut-off values of 3D-GLS and 3D-GCS, as determined by ROC curves. The log-rank test was used to evaluate differences among the groups. Next, pairwise models were examined. To control for effects of confounding factors, we adopted 4 stepwise, Cox proportional hazard models for the prediction of MACE with all independent variables. We selected the most significant variables from each category below: (1) clinical variables included age, sex, BSA, sBP, dBP, HR, hypertension and diabetes mellitus; (2) infarct-related indices included anterior MI, MVD, Killip>1, reperfusion time, CPK, infarct size; (3) ECG-related variables included QRS score; (4) biochemical markers included HbA1c, admission and maximum BNP; (5) renal function variables included eGFR, βblocker, ACE inhibitor/ARB, statins; (6) conventional echocardiography variables included EF and E/e′; and (7) speckle tracking-related variables.
Due to the non-normal distribution of BNP, we used log10-transformed BNP in all analyses. In addition, the incremental effect of STE on strong indices (age, infarct size by 99 mTc Methoxy-isobutyl-isonitrile [MIBI], and log BNP) in the prediction of future MACEs was evaluated using the net classification index (NRI).
All statistical tests were 2-sided. For all tests, a P value <0.05 was considered statistically significant. Statistical analyses were performed using the JMP Pro 15 (SAS Institute Inc., Cary, NC, USA) and EZR, which is a graphical version of R (version 3.1.2; The R founding, Vienne, Austria).
Results
Baseline Data of the Study Population
Table 1
shows the baseline characteristics of the patients. We examined a second echocardiography for 251 patients. There are significant differences between the first echocardiography and the second echocardiography for many variables.
Table 1.
Patient Characteristics According to the Presence of MACE
| Category |
All patients |
MACE (+) |
MACE (−) |
P value* |
| Number |
270 |
64 |
206 |
|
| Clinical variables |
| Age (years) |
65 (57~73) |
72 (63~78) |
62 (55~71) |
<0.0001 |
| Sex male, n (%) |
222 (82) |
50 (78) |
172 (83) |
0.32 |
| BSA (m2) |
1.71 (1.58~1.83) |
1.6 (1.54~1.79) |
1.72 (1.61~1.84) |
0.03 |
| sBP (mmHg) |
110 (100~120) |
104 (96~121) |
110 (100~121) |
0.15 |
| dBP (mmHg) |
60 (56~68) |
60 (51~64) |
60 (56~68) |
0.33 |
| HR (beats/min) |
72 (66~80) |
76 (66~83) |
72 (66~80) |
0.08 |
| Risk factors |
| Hypertension (yes) |
166 (62) |
45 (70) |
121 (59) |
0.11 |
| Diabetes mellitus (yes) |
73 (27) |
18 (28) |
55 (30) |
0.8 |
| Infarction-related indices |
| Anterior MI (yes) |
144 (53) |
42 (66) |
102 (49) |
0.02 |
| MVD, n (%) |
87 (34) |
29 (45) |
58 (28) |
0.01 |
| Killip >1 |
73 (27) |
27 (42) |
46 (22) |
0.002 |
| Reperfusion time, min |
135 (99~220) |
156 (107~255) |
128 (95~209) |
0.08 |
| Peak CPK (IU/L) |
3,026 (817~3,841) |
3,219 (1,117~6,371) |
1,924 (787~3,353) |
0.005 |
| Peak CPK-MB (IU/L) |
274 (88~366) |
302 (124~582) |
176 (74~321) |
0.007 |
| MIBI SPECT |
| Infarct size (IS) |
6.3 (0~32) |
16.5 (0.5~53.6) |
5 (0~25) |
0.008 |
| ECG-related index |
| QRS score at 24 h |
2.98±3.02 |
4.44±3.89 |
2.23±2.56 |
0.0002 |
| Biochemical markers |
| HbA1c (%) |
5.7 (5.3~6.2) |
5.7 (5.3~6.2) |
5.7 (5.4~6.4) |
0.7 |
| Admission BNP (pg/mL) |
26.8 (11.8~78.2) |
63.5 (26.7~189.7) |
22.3 (9.9~57.3) |
<0.0001 |
| Maximum BNP (pg/mL) |
91.7 (42.5~204.3) |
219.5 (90.65~522.9) |
73.7 (38.3~159.6) |
<0.0001 |
| Renal function |
| eGFR (mL/min/1.73 m2) |
63.1 (52.1~77.7) |
53.1 (45.1~68.9) |
64.9 (54.8~79.5) |
<0.0001 |
| Mediation at discharge |
| β-blocker, n (%) |
228 (84) |
54 (84) |
174 (84) |
0.98 |
| ACE inhibitor/ARB, n (%) |
259 (96) |
61 (95) |
198 (96) |
0.77 |
| Statins, n (%) |
267 (99) |
64 (100) |
203 (99) |
0.33 |
| Echocardiography |
| Conventional echocardiography |
| LVEDVI (mL/m2) |
61 (49~73) |
64.6 (51.9~77.7) |
60.8 (47.6~72.3) |
0.08 |
| LVESVI (mL/m2) |
27 (20~38) |
32.4 (24.6~48.1) |
26.6 (19.3~36.5) |
0.007 |
| LVEF (%) |
52 (44~62) |
45 (35~61) |
53 (45~62) |
0.03 |
| LAVI (mL/m2) |
39 (30~48) |
39 (31.3~51.5) |
33.1 (26.8~44.1) |
0.02 |
| E/A |
0.85 (0.69~1.13) |
10.89 (0.68~1.31) |
0.84 (0.68~1.1) |
0.23 |
| Dct (ms) |
204 (169~254) |
187 (155~256) |
204 (173~253) |
0.14 |
| E/e′ |
12.5 (9.7~15.6) |
15.3 (12.7~18.1) |
11.9 (9.3~15) |
<0.0001 |
| MR >moderate/severe |
21 (8) |
10 (17) |
11 (5) |
0.001 |
| Speckle tracking indices |
| 2D-GLS (%) |
−12.9 (−15.0~−10.0) |
−10.0 (−13.0~−8.0) |
−13.1 (−15.1~−11.6) |
<0.0001 |
| 2D-GCS (%) |
−16.0 (−22.0~13.0) |
−13.1 (−16.0~−10.1) |
−17.8 (−20.5~−13.9) |
<0.0001 |
| 3D-GLS (%) |
−12.5 (−14.7~−10.3) |
−8.8 (−152~−11.6) |
−13.2 (−15.2~−11.6) |
<0.0001 |
| 3D-GCS (%) |
−20.1 (−24.1~−16.3) |
−15.0 (−18.1~−11.6) |
−21.5 (−24.7~−18.6) |
<0.0001 |
| 3D-GRS (%) |
40.2 (33.1~55.4) |
31.2 (25.4~41.2) |
44.1 (33.7~55.4). |
<0.0001 |
| 3D-GPS (%) |
−30.9 (−35.1~−26.0) |
−25.0 (−29.1~−17.4) |
−32.1 (−34.0~−28.3) |
<0.0001 |
| 3D-volume rate (/s) |
22 (20~24) |
– |
– |
– |
*P values were for patients with MACE vs. without MACE. A, late diastolic; ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker; BSA, body surface area; BNP, brain natriuretic peptide; CPK, creatinine phosphokinase; Dct, deceleration time; dBP, diastolic blood pressure; E, early diastolic wave velocity; e′, early diastolic velocity of mitral annulus; ECG, electrogardiogram; EDVI, end-diastolic volume index; eGFR, estimated glomerular filtration rate; ESVI, end-systolic volume index; GCS, global circumferential strain; GLS, global longitudinal strain; GPS, global principal strain; GRS, global radial strain; HBA1c, hemoglobin A1c; HR, heart rate; LAVI, left atrial volume index; LV, left ventricular; LVEF, LV ejection fraction; MACE, major adverse cardiac events; MB, myocardial band; MI, myocardial infarction; MIBI, 99mTc Methoxy-isobutyl-isonitrile; MR, mitral regurgitation; MVD, multivessel disease; sBP, systolic blood pressure; SPECT, single photon emission computed tomography.
During the follow-up period, 64 patients experienced MACE, and 54 patients died from all causes (24 cardiovascular deaths, 30 non-cardiovascular deaths). Next, we determined the predictors for MACE.
Figure 2A
shows ROC curves to predict MACE for 3D-GLS (red), 3D-GCS (green), 2D-GLS (blue) and 2D-GCS (orange) in all 270 patients. There were significant differences among the 4 curves (P=0.0001). Furthermore, a significant difference was found between 3D-GLS and 2D-GLS (P=0.002) and between 3D-GCS and 2D-GCS (P<0.0001).
Figure 2B
shows plots for patients with large infarctions (>20%, n=85). The AUC for 3D-GLS was 0.939 (95% CI: 0.872–0.973) and that of 2D-GCS was 0.872 (95% CI, 0.776–0.931).
Figure 2C
shows plots for patients with non-large infarctions (≤20%; n=179). The AUC of 3D-GLS was 0.812 (95% CI: 0.696–0.891) and 2D-GLS was 0.737 (95% CI: 0.629–0.823). Thus, there was a significant difference between the 2 curves (P=0.04). The AUC for 3D-GCS was 0.756 (95% CI: 0.640–0.845) and 2D-GCS was 0.630 (95% CI: 0.530–0.720), indicating a significant difference between the 2 curves (P=0.03).
Table 2
shows Cox proportional hazard models using 2D-GLS, 2D-GCS, 3D-GLS, and 3-GCS. Models using 3D-GLS and 3D-GCS were stronger than those using 2D-GLS and 2D-GCS. Furthermore, the inclusion of 2D- and 3D-STE indices for age, log BNP, and infarct size by MIBI are shown in
Table 3. This table shows that the inclusion of 3D-GLS was associated with a NRI of 0.5519% (P=0.012), and 3D-GCS was associated with a NRI of 0.4599% (P=0.037), suggesting an effective reclassification. Accordingly, 3D-GLS and 3D-GCS had a significant incremental effect, but 2D-GLS and 2D-GCS did not. Age was the strongest predictor of the secondary endpoint (all-cause death).
Table 2.
Multivariable Cox Proportional Hazard Analysis for MACE
| Variables |
2D-GLS |
2D-GCS |
| HR |
95% CI |
P value |
HR |
95% CI |
P value |
| Age (per 1 year) |
1.05 |
1.02~1.09 |
0.0005 |
1.06 |
1.02~1.09 |
0.0004 |
| Infarct size, MIBI (%) |
0.99 |
0.98~1.01 |
0.56 |
0.99 |
0.97~1.01 |
0.54 |
| Maximum Log BNP (per/1 pg/mL) |
0.89 |
0.35~2.22 |
0.82 |
1.17 |
0.46~2.99 |
0.74 |
| QRS score |
1.12 |
1.01~1.25 |
0.01 |
1.21 |
1.09~1.33 |
0.0001 |
| E/e′ |
1.05 |
0.99~1.09 |
0.05 |
1.06 |
1.02~1.99 |
0.01 |
| eGFR (mL/min/1.73 m2) |
0.98 |
0.96~1.00 |
0.06 |
0.98 |
0.96~0.99 |
0.02 |
| 2D-GLS (%) |
1.39 |
1.23~1.59 |
<0.0001 |
– |
– |
– |
| 2D-GCS (%) |
– |
– |
– |
1.06 |
1.00~1.13 |
0.03 |
| 3D-GLS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GCS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GPS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GRS (%) |
– |
– |
– |
– |
– |
– |
| |
3D-GLS |
3D-GCS |
| HR |
95% CI |
P value |
HR |
95% CI |
P value |
| Age (per 1 year) |
1.03 |
1.01~1.07 |
0.01 |
1.04 |
1.01~1.08 |
0.002 |
| Infarct size, MIBI (%) |
0.98 |
0.97~1.00 |
0.09 |
0.99 |
0.97~1.01 |
0.74 |
| Maximum Log BNP (per/1 pg/mL) |
1.46 |
0.57~3.56 |
0.41 |
0.96 |
0.37~2.39 |
0.93 |
| QRS score |
1.08 |
0.97~1.21 |
0.14 |
1.14 |
1.02~1.28 |
0.01 |
| E/e′ |
1.06 |
1.01~1.11 |
0.01 |
1.03 |
0.99~1.08 |
0.14 |
| eGFR (mL/min/1.73 m2) |
0.99 |
0.97~1.01 |
0.31 |
0.98 |
0.95~0.99 |
0.01 |
| 2D-GLS (%) |
– |
– |
– |
– |
– |
– |
| 2D-GCS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GLS (%) |
1.56 |
1.39~1.77 |
<0.0001 |
– |
– |
– |
| 3D-GCS (%) |
– |
– |
– |
1.25 |
1.15~1.35 |
<0.0001 |
| 3D-GPS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GRS (%) |
– |
– |
– |
– |
– |
– |
| |
3D-GPS |
3D-GRS |
| HR |
95% CI |
P value |
HR |
95% CI |
P value |
| Age (per 1 year) |
1.04 |
1.01~1.08 |
0.01 |
1.05 |
1.02~1.09 |
0.01 |
| Infarct size, MIBI (%) |
0.99 |
0.97~1.01 |
0.31 |
0.99 |
0.97~1.01 |
0.41 |
| Maximum Log BNP (per/1 pg/mL) |
1.46 |
0.56~3.68 |
0.42 |
1.45 |
0.54~3.79 |
0.44 |
| QRS score |
1.05 |
0.95~1.17 |
0.29 |
1.17 |
1.05~1.29 |
0.001 |
| E/e′ |
1.06 |
1.01~1.11 |
0.008 |
1.05 |
1.01~1.09 |
0.01 |
| eGFR (mL/min/1.73 m2) |
0.98 |
0.96~1.01 |
0.06 |
0.98 |
0.96~1.00 |
0.07 |
| 2D-GLS (%) |
– |
– |
– |
– |
– |
– |
| 2D-GCS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GLS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GCS (%) |
– |
– |
– |
– |
– |
– |
| 3D-GPS (%) |
1.11 |
1.05~1.17 |
<0.0001 |
– |
– |
– |
| 3D-GRS (%) |
– |
– |
– |
0.95 |
0.93~0.99 |
0.01 |
CI, confidence interval; HR, hazard ratio; other abbreviations are as per Table 1.
Table 3.
Incremental Effect of Each STE to the Base Model
| STE variable* |
NRI |
95% CI |
P value |
| 2D-GLS |
0.0004 |
(−0.4337~0.4329) |
0.9986 |
| 2D-GCS |
0.3491 |
(−0.0575~0.7587) |
0.0924 |
| 3D-GLS |
0.5519 |
(0.1194~0.9844) |
0.0124 |
| 3D-GCS |
0.4599 |
(0.0264~0.8933) |
0.0376 |
*Each variable was added to the base model, which was identified in Table 2 and consisted of Age+Log BNP+Infarct size by MIBI. CI, confidence interval; 2D, two-dimensional; 3D, three-dimensional; NRI, net reclassification index; STE, speckle-tracking echocardiography; other abbreviations are as per Table 1.
The Kaplan-Meier curves for MACE based on each optimal cut-off value are shown in
Supplementary Figure 2A
(3D-GLS) and
Supplementary Figure 2B
(3D-GCS) with the cut-off values (3D-GLS=−11.0, 3D-GCS=−18.3) used. Both 3D-GLS and 3D-GCS were significant predictors.
Figure 3
shows the 3D-GLS and 3D-GCS plots, with the red plot indicating the patients with MACE and the blue plot indicating the patients without MACE. There were significant differences in the frequency of MACE among the 4 groups (Pearson, P<0.0001). The patients were divided using cut-off values into the 4 groups: group A (red line: 3D-GLS <−11.0 and 3D-GCS <−18.3), group B (blue line: 3D-GLS <−11.0 and 3D-GCS ≥−18.3), group C (green line: 3D-GLS ≥−11.0 and 3D-GCS <−18.3) and group D (yellow line: 3D-GLS ≥−11.0 and 3D-GCS ≥−18.3). Next, we used pairwise models for analysis. Group D had the worst outcome compared to the outcomes of all 3 groups (P<0.0001). Groups B and C had significantly poorer outcomes compared to those of group A (P=0.001 and P=0.007, respectively). There were significant differences among the 4 groups, except between groups B and C. This result suggests that patients with both higher values (lower absolute values) in both 3D-GLS and 3D-GCS were at high risk for MACE (Figure 4).
Supplementary Figure 3
shows ROC curves by 4 types of 3D-STE.
The intra-observer variabilities for the 3D strain were 9.5% for 3D-GLS, 10.0% for 3D-GCS, 15.8% for 3D-GRS and 12.5% for 3D-GPS. The corresponding inter-observer variabilities for 3D-GLS, 3D-GCS, 3D-GRS and 3D-GPS were 9.8%, 10.8%, 17.3%, and 13.6%, respectively. These values were similar to those reported in a previous study.9
Bland-Altman plot analysis revealed a bias of 0.0537 for 3D-GLS calculated by 2 observers, with 95% limits of agreement ranging from 0.518 to −0.735. This was also observed in the 3D-GCS analysis, which revealed a bias of 0.190, as calculated by 2 observers, with 95% limits of agreement ranging from 0.611 to −0.198. The correlations and Bland-Altman plot analyses7
are calculated. The intra-observer variabilities for the 2D strain were 7.6% for 2D-GLS and 11.8% for 2D-GCS. The corresponding inter-observer variabilities for 2D-GLS and 2D-GCS were 8.6% and 14.8%, respectively. Bland-Altman plot analysis revealed a bias of 0.225 for 2D-GLS calculated by 2 observers, with 95% limits of agreement ranging from 0.694 to −0.245. This was also observed in the 2D-GCS analysis, which revealed a bias of 0.609, as calculated by 2 observers, with 95% limits of agreement ranging from 1.612 to −0.916.
Discussion
This study demonstrated that 3D-STE immediately after reperfusion therapy can be a strong predictor of patients’ 10-year prognosis after STEMI. Additionally, 3D-STE was superior to 2D-STE in predicting long-term prognosis. The addition of 3D-STE to the known strong predictors (age, infarct size, and log BNP) improved risk stratification in STEMI patients. Although 3D-GLS alone was a strong predictor for MACE, using it in combination with 3D-GCS improved the prediction of prognosis after STEMI.
Novel Aspects of this Study
Some exceptional findings distinguish our study from existing literature. First, we performed 3D echocardiography at 24 h after the onset of STEMI. We think strain analysis even at the bed side enables us to evaluate LV damage and function precisely, leading us to predict the long-term prognosis. Second, to reduce the effects of confounding factors, we studied only those patients who had their first STEMI with reperfusion within 12 h of symptom onset. Third, we analyzed cardiac function and estimated infarct size using SPECT. There are significant differences among strain values calculated by different vendor software;10
however, we analyzed not only 3D-STE but also 2D-STE findings using vendor-independent software. Finally, the patients were followed up for a median of approximately 10 years, which is longer than previously reported follow-up time.11,12
Thus, we were able to conclude that 3D-STE is a good prognostic indicator that can be easily performed at the bedside.
Multidirectional Assessment of LV Strain After MI
Previous guidelines recommend GLS as a reproducible index that can be used to identify improvement or worsening of patients’ conditions.13,14
As demonstrated previously, GLS indicates infarct size after MI, which partially contributes to its prognostic value. Sub-endocardial longitudinal fibers are sensitive to hypoperfusion during ischemia. Therefore, GLS may also indicate the area at risk. As longitudinally arranged sub-endocardial fibers are at the maximum risk of injury in MI, there is a risk of adverse LV remodeling.15
Although GLS is a reliable predictor, circumferential movement is also important, and a recent study suggested that GCS indicates advanced myocardial damage more effectively than GLS in patients with aortic stenosis.16
The risk of transmural necrosis depends on the duration of ischemia.17
Therefore, we think that the deterioration of GCS may represent advanced transmural myocardial damage and deterioration of GLS may represent both early ischemic damage and LV fibrosis. Furthermore, 3D-STE enables simultaneous identification of 3D-GCS and 3D-GLS values, which allows simultaneous analysis of LV geometry and function. Although GCS is reported to be susceptible to out-of-plane motion, use of 3D-STE may overcome this problem. GPS is a unique parameter that might be useful for detecting subclinical damage, and the corresponding AUC in our study was close to those of 3D-GLS and 3D-GCS (Supplementary Figure 3). The evaluation of principal and secondary strain provides an integrated assessment of LV deformation that is intrinsically related to cardiac myofiber geometry. Shin et al reported the usefulness of area strain after STEMI; however, they did not compare area strain with the other 3D-STE indices.18
Moreover, we did not calculate area-strain in our study; therefore, we believe that we would examine area strain in a future study. However, we could not compare the 2D values. We found that 3D-STE was more advantageous than 2D-STE; hence, we think this difference may be due to the whole heart examination using 3D-STE.
Clinical Implications
Our study has confirmed that 3D global strain, especially 3D-GLS, can predict long-term prognosis after STEMI. The prognosis after STEMI can be precisely predicted in the acute phase by using 3D-STE, in addition to the other indices. Furthermore, accuracy of prediction increases when 3D-GLS is combined with 3D-GCS, and both indices can be calculated simultaneously. All examinations could be performed using 3D-STE in 1-image measurements. Thus, patients can be effectively assessed by simple bedside echocardiographic examinations. We believe that 3D-STE can reveal multidirectional wall motion and LV function.
Study Limitations
This study has several limitations. First, this was a single-center study. Second, the analyses by 2D- and 3D-STE were performed using offline software. We believe that 3D-STE requires additional progression, which will allow easy bedside examination. Third, because echocardiographic examinations were performed at the bedside, scanning the patients was sometimes difficult. There is a learning curve in obtaining appropriate data from full-volume 3D images, and the investigators underwent thorough training and careful examination daily.19
Fourth, infarct size estimation in all patients was performed by scintigraphy; however, magnetic resonance imaging is considered the standard technique used for estimation of infarct size. Finally, iE33 is a relatively classical 3D machine; however, our study was a pioneering study to demonstrate the clinical usefulness of 3D-STE in patients with STEMI. Nevertheless, these limitations maybe considered small, and this newly developed method definitely facilitates improved treatment options for cardiac patients.
Conclusions
The present study shows that 3D global strain is a useful tool for the prediction of long-term prognosis in patients with STEMI who have undergone reperfusion therapy. We recommend 3D-STE in patients with STEMI because it facilitates the examination of not only the longitudinal wall motion, but also the circumferential wall motion. Further studies are needed to confirm the usefulness of 3D-STE in determining STEMI prognosis.
Authors’ Contributions
All authors read and approved the final manuscript.
Sources of Funding
This study did not receive any funding.
Disclosures
K.K., M.K. are Editorial Board members for
Circulation Journal.
Data Availability
1. The individual deidentified participant data will be shared.
2. All analyzable data set related to the study will be shared.
3. Study protocol, statistical analysis plan will be available.
4. The data will be available immediately following publication, and ending 10 years after the publication.
5. Anyone who wishes to access the data will be granted access.
6. The data will be shared as Excel files via E-mail.
Supplementary Files
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-21-0183
References
- 1.
Cimino S, Canali E, Petronilli V, Cicogna F, De Luca L, Francone M, et al. Global and regional longitudinal strain assessed by two-dimensional speckle tracking echocardiography identifies early myocardial dysfunction and transmural extent of myocardial scar in patients with acute ST elevation myocardial infarction and relatively preserved LV function. Eur Heart J Cardiovasc Imaging 2013; 14: 805–811.
- 2.
Seo Y, Ishizu T, Enomoto Y, Sugimori H, Aonuma K. Endocardial surface area tracking for assessment of regional LV wall deformation with 3D speckle tracking imaging. JACC Cardiovasc Imaging 2011; 4: 358–365.
- 3.
Iwahashi N. Impact of three-dimensional global longitudinal strain for patients with acute myocardial infarction. Eur Heart J Cardiovasc Imaging, https://doi.org/10.1093/ehjci/jeaa241.
- 4.
Abate E, Hoogslag GE, Antoni ML, Nucifora G, Delgado V, Holman ER, et al. Value of three-dimensional speckle-tracking longitudinal strain for predicting improvement of left ventricular function after acute myocardial infarction. Am J Cardiol 2012; 110: 961–967.
- 5.
Iwahashi N, Gohbara M, Kirigaya J, Abe T, Horii M, Takahashi H, et al. Prognostic significance of a combination of QRS Score and E/e’ obtained 2 weeks after the onset of ST-elevation myocardial infarction. Circ J 2020; 84: 1965–1973.
- 6.
Pedrizzetti G, Sengupta S, Caracciolo G, Park CS, Amaki M, Goliasch G, et al. Three-dimensional principal strain analysis for characterizing subclinical changes in left ventricular function. J Am Soc Echocardiogr 2014; 27: 1041–1050.e1041.
- 7.
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–310.
- 8.
DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 1988; 44: 837–845.
- 9.
Nagata Y, Takeuchi M, Wu VC, Izumo M, Suzuki K, Sato K, et al. Prognostic value of LV deformation parameters using 2D and 3D speckle-tracking echocardiography in asymptomatic patients with severe aortic stenosis and preserved LV ejection fraction. JACC Cardiovasc Imaging 2015; 8: 235–245.
- 10.
Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al. Intervendor differences in the accuracy of detecting regional functional abnormalities: A report from the EACVI-ASE strain standardization task force. JACC Cardiovasc Imaging 2018; 11: 25–34.
- 11.
Antoni ML, Mollema SA, Delgado V, Atary JZ, Borleffs CJ, Boersma E, et al. Prognostic importance of strain and strain rate after acute myocardial infarction. Eur Heart J 2010; 31: 1640–1647.
- 12.
Ersboll M, Valeur N, Mogensen UM, Andersen MJ, Moller JE, Velazquez EJ, et al. Prediction of all-cause mortality and heart failure admissions from global left ventricular longitudinal strain in patients with acute myocardial infarction and preserved left ventricular ejection fraction. J Am Coll Cardiol 2013; 61: 2365–2373.
- 13.
Galderisi M, Cosyns B, Edvardsen T, Cardim N, Delgado V, Di Salvo G, et al. Standardization of adult transthoracic echocardiography reporting in agreement with recent chamber quantification, diastolic function, and heart valve disease recommendations: An expert consensus document of the European Association of Cardiovascular Imaging. Eur Heart J Cardiovasc Imaging 2017; 18: 1301–1310.
- 14.
Haugaa KH, Dejgaard LA. Global longitudinal strain: Ready for clinical use and guideline implementation. J Am Coll Cardiol 2018; 71: 1958–1959.
- 15.
Joyce E, Hoogslag GE, Leong DP, Debonnaire P, Katsanos S, Boden H, et al. Association between left ventricular global longitudinal strain and adverse left ventricular dilatation after ST-segment-elevation myocardial infarction. Circ Cardiovasc Imaging 2014; 7: 74–81.
- 16.
Fung MJ, Thomas L, Leung DY. Alterations in layer-specific left ventricular global longitudinal and circumferential strain in patients with aortic stenosis: A comparison of aortic valve replacement versus conservative management over a 12-month period. J Am Soc Echocardiogr 2019; 32: 92–101.
- 17.
Tarantini G, Cacciavillani L, Corbetti F, Ramondo A, Marra MP, Bacchiega E, et al. Duration of ischemia is a major determinant of transmurality and severe microvascular obstruction after primary angioplasty: A study performed with contrast-enhanced magnetic resonance. J Am Coll Cardiol 2005; 46: 1229–1235.
- 18.
Shin SH, Suh YJ, Baek YS, Lee MJ, Park SD, Kwon SW, et al. Impact of area strain by 3D speckle tracking on clinical outcome in patients after acute myocardial infarction. Echocardiography 2016; 33: 1854–1859.
- 19.
Tanabe K, Yamaguchi K. Incorporating three-dimensional echocardiography into clinical practice. J Echocardiogr 2019; 17: 169–176.
