Study Population

This was a prospective historical cohort study. Data were obtained from patients with STEMI who were treated with primary PCI in the Department of Cardiology at Aarhus University Hospital Skejby in Denmark between November 2002 and November 2008. During this time span, an estimated 6,000 patients with STEMI underwent primary PCI at our institution, with a median age of 65 years (interquartile range, 55–74 years); 27% were women, and 9% were in Killip class ≥ 2 on arrival. Routine echocardiography on patients with STEMI in this period was performed using bedside ultrasound equipment with low frame rates and limited storage capabilities, rendering speckle-tracking analysis impossible. Therefore, in this analysis, we decided to include only patients from one of three previous studies, in which echocardiography was performed per protocol with the consistent use of ultrasound equipment, allowing speckle-tracking strain analysis. Patients from these source studies, with verified STEMI diagnoses, who had undergone per protocol echocardiographic examinations within the first day after revascularization, were eligible for inclusion. The number of patients figuring in our statistical models varied depending on the availability of the variables included in each model. When echocardiograms were acquired, special focus was given to LV systolic function, with the secondary objective of evaluating the prognostic utility of systolic indices, including strain. The study by Terkelsen et al. was descriptive of periprocedural ST-segment resolution and enrolled patients with tentative STEMI diagnoses in which continuous ST-segment monitoring was initiated in the prehospital phase ( n = 122). Patients were excluded if a diagnosis of STEMI was not confirmed, if no PCI was performed, or if ST-segment monitoring data were incomplete. The second study investigated the effect of routine thrombectomy on myocardial salvage. Patients were randomized at the catheterization laboratory when presenting with angiographically verified STEMI ( n = 215). In the Remote Ischemic Conditioning in Primary PCI trial, randomization took place in the prehospital phase, during emergency vehicle transport. Patients were assigned to receive conventional primary PCI treatment (control group) or consecutive episodes of upper limb ischemia applied by the ambulance staff using a standard blood pressure cuff as an adjunct to primary PCI (intervention group) ( n = 333). Inclusion criteria were the same in the studies, including symptoms consistent with STEMI lasting ≥30 min but ≤12 hours and cumulative ST-segment elevation of ≥2 mV in two or more contiguous leads. Descriptions of exclusion from analysis criteria are given in each report, but assuming that the involved interventions have not altered the association between postintervention echocardiographic indices and the risk for adverse outcomes, in this study, we compare the prognostic value of LV function indices.

The local ethics committee approved each of the source studies, and participants gave written informed consent.

Echocardiography

Resting echocardiography was performed at a median time span of 12 hours (interquartile range, 6–18 hours) after revascularization using standard equipment (Vivid 7; GE Vingmed Ultrasound AS, Horten, Norway) with a 3.5-MHz probe. LV volumes and LVEF were measured by one observer (K.M.) using Simpson’s biplane method from apical two-chamber and four-chamber views. LV end-systolic volumes were indexed to body surface area, calculated using the Dubois formula (0.007184 × weight ^0.425 × height ^0.725 ).

Using a 16-segment model with six basal, six midventricular, and four apical segments, each myocardial segment was attributed a wall motion score reflecting dyskinesia (−1), akinesia (0), hypokinesia (1), normokinesia (2), or hyperkinesia (3), from which the average WMSI was calculated. Wall motion scoring patients from the first source study was made by C.J.T., by N.H.A. for patients in the second source study, and by K.M. in the third source study.

GLS

Speckle-tracking analysis was performed by one observer (K.M.) blinded to clinical data, using standard software (EchoPAC PC SW-Only version 7.0.0; GE Healthcare, Milwaukee, WI). Figure 1 shows an example of two-dimensional strain analysis. From cine loops with three consecutive cardiac cycles, measurements of longitudinal strain by speckle tracking were obtained from one representative cycle, not as an average of three cycles, avoiding premature and post extrasystolic contractions. The timing of systole was determined from continuous-wave Doppler recording through the aortic valve or alternatively from the identification of aortic valve closure in the apical long-axis view. In each apical view, we marked the U-shaped endocardium, and the software then defined the myocardium. When the speckle region of interest suggested by the software covered the LV myocardium inadequately, this was manually adjusted. The software has an algorithm assessing tracking quality. When according to this algorithm, tracking of a segment was inadequate, the segment would be excluded per default. In rare cases (estimated at <5%), a segment excluded by the software algorithm was noticed by the observer to be tracked sufficiently and was approved manually. Likewise, when a segment initially qualified by the software algorithm was clearly insufficiently tracked, this was disqualified by the observer. Peak systolic longitudinal strain was obtained in each standard apical view as the strain of the entire U-shaped myocardium in each view. GLS was measured by the software as the peak systolic strain of the entire left ventricle, considered as one big segment consisting of all segments at that specific time in systole when the value was at its peak. The software allowed the calculation of GLS only when tracking quality was adequate in at least five of six segments in each apical view. In individuals ( n = 32) in whom peak systolic longitudinal strain could only be assessed in two of three apical views, we calculated GLS as the mean peak longitudinal strain from two views.

Speckle-tracking strain analysis from day 1 after STEMI with culprit lesion in the left anterior descending coronary artery, displaying (A) peak systolic longitudinal strain in the apical long-axis view of −8.7%, (B) peak systolic longitudinal strain in the four-chamber view of −10.0%, (C) peak systolic longitudinal strain in the two-chamber view of −11.8%, and (D) the corresponding strain map with systolic strain values provided in each myocardial segment and with a parametric code, with strong red colors representing systolic shortening and blue colors depicting systolic lengthening of a segment. GLS was 10.2% in this patient.

We recently published intraobserver and interobserver reproducibility data for the involved echocardiographic methods. In brief, for the intraobserver analysis, we found intraclass coefficients as follows: for GLS, 0.96 (95% confidence interval [CI], 0.90 to 0.98); for LVEF, 0.80 (95% CI, 0.61 to 0.91); for WMSI, 0.91 (95% CI, 0.81 to 0.96); and for end-systolic volume, 0.95 (95% CI, 0.90 to 0.98). For the interobserver analysis, the low and high limits of agreement representing the mean difference between measurements made by two observers ± 2 SDs on a relative scale were as follows: for GLS, −0.14 (95% CI, −0.21 to −0.06) and 0.28 (95% CI, 0.21 to 0.35); for LVEF, −0.38 (95% CI, −0.48 to −0.27) and 0.20 (95% CI, 0.10 to 0.30); for WMSI, −0.17 (95% CI, −0.24 to −0.10) and 0.23 (95% CI, 0.16 to 0.30); and for end-systolic volume, −0.70 (95% CI, −0.86 to −0.54) and 0.21 (95% CI, 0.05 to 0.37).

Follow-Up and End Points

The primary end point was the composite of all-cause mortality and hospitalization with myocardial infarction, stroke, or hospitalization or outpatient clinic visit because of congestive heart failure, whichever occurred first. As a secondary end point, we used crude mortality. Ideally, crude mortality should have been the primary end point, but in planning the analysis, we realized that the small number of mortality events ( n = 39) would preclude multivariate adjustment for confounders. On these grounds, the crude mortality analysis is confined to univariate models. Information about the outcomes was obtained by record linkage with population-based Danish registries using the civil registration number, a unique personal identification code that is assigned to all Danish citizens and non-Danish residents.

Information on all-cause mortality was obtained from the Danish Civil Registration System. Information on readmissions was obtained from the Danish National Patient Registry, which holds data on all hospitalizations from all Danish nonpsychiatric hospitals since 1977, including discharge diagnoses assigned by treating physicians and coded according to the International Classification of Diseases, Eighth Revision, until the end of 1993 and the International Classification of Diseases, Tenth Revision, thereafter.

Follow-up began on the date of the index myocardial infarction and ended on the date of death, admission with myocardial infarction, stroke, or heart failure, whichever occurred first, or otherwise September 2009. Follow-up data were available for all patients until this date.

Statistical Analyses

Data were analyzed using standard statistical software (Stata/IC version 10.1; StataCorp LP, College Station, TX). P values < .05 were considered statistically significant.

Inspection of histograms and quantiles of normal distribution plots (Q-Q plots) were used to check for normality. Continuous, normally distributed data are presented as mean ± SD or as medians and interquartile ranges when not conforming to a normal distribution. Categorical data are summarized as absolute counts with percentages. Comparison of continuous variables in patients with and without events in the follow-up period was done using unpaired t tests and Mann-Whitney U tests for variables that were not normally distributed. Categorical data were compared using χ ² tests or Fisher’s exact tests when tabled numbers were <10.

Univariate Cox regression models were used to examine relations between potential predictors and the end point, and Kaplan-Meier event-free survival curves were constructed for GLS. The log-rank test was used to test for differences in event-free survival according to arbitrarily chosen cutoff values of GLS.

For assessment of adjusted associations, we used multiple Cox regression with the investigator variable (GLS) and clinical confounders as forced entry variables. Clinical confounders were determined before the analysis. Besides gender and age, we included Killip class ≥ 2 on presentation, systolic blood pressure ≤ 100 mm Hg or heart rate ≥ 100 beats/min on presentation, anterior infarct location (culprit lesion in the left anterior descending coronary artery), and diabetes and/or hypertension, which are the most important established determinants of outcome after STEMI. Besides the forced-entry variables, survival models included conventional measures of systolic function separately and in selected combinations.

The validity of Cox survival models used was assessed by inspection of log-log plots for categorical variables. For measures of LV function, as a model validation, we included the LV function measure with an associated continuous time-varying coefficient. This indicated a time dependency of the hazard ratio (HR) associated with baseline LV function, implying that acute-phase LV function contained less information on the risk for experiencing new events as time passed after the index event. Because of the time dependency, we chose to report estimates of models assuming constant HRs during the first year and constant HRs at another level after the first year of the index event.

All reported estimates are results of complete case analyses. As a robustness test to ensure that estimates were not biased by covariates with missing values, we imputed missing values by multiple imputations. This was done using potential predictors of outcome along with the Nelson-Aalen estimates of the cumulative hazard and the event indicator in the multiple regressions. Analyses using imputed values did not change our conclusions regarding GLS.

Receiver operating characteristic (ROC) curves with comparison of areas under the curves (AUCs) were used to illustrate the sensitivity and specificity for LV function indices in predicting 1-year event-free survival. Furthermore, comparison of ROC curves adjusted for clinical confounders was made, with covariates assumed to act linearly on the control marker distribution, which was estimated empirically conditional on the covariates.

Study Population

Table 1 shows the availability of echocardiograms and the feasibility of GLS measurement by source study. In total, the source studies initially enrolled 670 patients with tentative STEMI diagnoses. Among these, the STEMI diagnoses were verified in 626 patients, among whom day 1 echocardiograms were available for 576, making up the study population. Patients without baseline echocardiography were older (mean age, 69 ± 13 vs 63 ± 12 years; P < .001), more frequently women (19 of 50 [38%] vs 130 of 576 [23%], P = .01), and had worse prognoses than those who completed echocardiography (HR adjusted for clinical data, 2.2; 95% CI, 1.4–3.6; P = .001; conditioned on no events on the first day after the index event). Of the 576 patients with baseline echocardiograms, GLS was obtained in 425 patients (74%). GLS data could not be obtained because of poor or inadequate image quality and/or low frame rates for speckle tracking in 129 patients and because the software could not read the data for technical reasons in 22. Event rates did not differ between patients with GLS and patients in whom GLS could not be assessed (HR adjusted for clinical data, 0.79; 95% CI, 0.57–1.11; P = .18). Characteristics of the study population ( n = 576) divided according to individuals with ( n = 162) and individuals without ( n = 414) events during follow-up are outlined in Table 2 .

Events during Follow-Up

During a median follow-up period of 24 months (interquartile range, 13–61), 162 patents (28%) experienced the composite end point, comprising 39 deaths and 85 hospitalizations with reinfarction, 29 with heart failure, and nine with stroke. Two thirds ( n = 107 [66%]) of these events happened within the first year, and approximately one third (60 [37%]) occurred within the first month.

Univariate Analysis

Table 2 provides univariate predictors of the composite end point. Besides clinical characteristics on presentation such as age, heart rate, Killip class ≥ 2, anterior infarct location, multivessel disease, history of hypertension, diabetes, and prior infarction, GLS and the other echocardiographic indices of systolic function were each strongly associated with the composite end point. Furthermore, GLS (unadjusted HR, 1.20; 95% CI, 1.08–1.32, P < .001), WMSI (unadjusted HR, 0.42; 95% CI, 0.19–0.92; P = .029), and LVEF (unadjusted HR, 0.97; 95% CI, 0.95–0.99; P = .02) were predictors of crude mortality, while ESVI (unadjusted HR, 1.02; 95% CI, 1.00–1.05; P = .08) showed a borderline significant association with mortality.

Figure 2 displays the event-free survival of patients divided arbitrarily into three groups of severely affected (GLS ≥ −10%), moderately affected (−10% > GLS ≥ −15%), and mildly affected (GLS ≤ −15%). Patients ( n = 62) with severely (less negative) affected values of GLS (≥ −10%) constituted a high-risk group (HR, 4.6; 95% CI, 2.8–7.7; P < .001), while patients in the moderately affected group ( n = 182) had borderline significantly increased risk for the composite end point (HR, 1.61; 95% CI, 1.0–2.6; P = .05) compared with patients with mildly affected GLS (< −15%) ( n = 181).

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Appropriate Use Criteria for Echocardiography: New Applications for a New Era of Utilization
2. ECMO and ECHO: The Evolving Role of Quantitative Echocardiography in the Management of Patients Requiring Extracorporeal Membrane Oxygenation
3. Altered Central Aortic Elastic Properties in Kawasaki Disease are Related to Changes in Left Ventricular Geometry and Coronary Artery Aneurysm Formation
4. Teaching TEE for Use in the Operating Room: Where Are Things Now…and Where Are We Going?

Stay updated, free articles. Join our Telegram channel

Join