The prognostic value of global longitudinal strain in patients with myocardial infarction and preserved ejection fraction: a prespecified substudy of the REDUCE-AMI trial

Abstract

Aims

The REDUCE-AMI trial showed that beta-blockers in patients with preserved left ventricular ejection fraction (LVEF) after acute myocardial infarction (AMI) had no effect on mortality or cardiovascular outcomes. The aim of this substudy was to evaluate whether global longitudinal strain (GLS) is a better prognostic marker than LVEF, and if beta-blockers have a beneficial effect in patients with decreased GLS.

Methods and results

REDUCE-AMI was a registry-based randomized clinical trial. Conventional echocardiographic parameters and GLS were obtained and a likelihood ratio test between models adjusted for age, sex, hypertension, smoking, diabetes, previous AMI, and multi-vessel disease was used to compare LVEF and GLS as prognostic methods. A Cox regression model evaluated the impact of beta-blocker treatment on the composite endpoint of death from any cause or new AMI. A total of 1436 patients (28.6% of the total population) were included in this substudy. Due to poor image quality or incompatible equipment, 324 (22.6%) patients were excluded from the analysis of GLS. The median GLS was 17.3%. The likelihood ratio test resulted in no difference (P = 0.56) when comparing the combination of GLS to LVEF. The results were robust when adding beta-blocker randomization status as an independent variable.

Conclusion

In patients after AMI with preserved LVEF, GLS did not add prognostic value regarding death from any cause or new AMI. In addition, beta-blocker treatment did not alter the prognostic information obtained from GLS. Consequently, this study does not support an additive value of GLS compared with standard echocardiographic measurement in this patient population.

Graphical Abstract

Open in new tabDownload slide

See the editorial comment for this article ‘Global longitudinal strain and beta-blockers: rethinking their roles in post-AMI patients with preserved ejection fraction’, by I. Lechner  et al., https://doi.org/10.1093/ehjci/jeaf039.

Introduction

For several decades, patients surviving acute myocardial infarction (AMI) have been treated with beta-blockers for secondary prevention irrespective of left ventricular (LV) systolic function.¹ Randomized controlled trials, conducted during the 1980s, provided evidence of the beneficial effect of beta-blockers in lowering mortality by approximately 20%.^2–4 Contemporary guideline-directed therapies, including percutaneous coronary intervention (PCI), were not available at that time, resulting in larger infarcts with subsequent heart failure and malignant arrhythmias. Echocardiography was not widely available, and evaluation of LV function was not routinely performed.

Today, the assessment of LV systolic function by left ventricular ejection fraction (LVEF) with echocardiography has become a cornerstone of cardiac diagnostics in clinical practice and research, and established guidelines provide well-defined LVEF cut-off values for risk stratification and therapeutic decision-making.^1,5 However, LVEF has several limitations and previous studies have demonstrated that only LVEF values <40% correlate with a negative prognosis, and above this threshold, LVEF shows no association with mortality and other outcome measures.^6,7 Furthermore, the assessment of LVEF is subject to considerable interobserver variability.⁸

Global longitudinal strain (GLS), obtained by two-dimensional speckle tracking echocardiography, measures LV function by analyzing the deformation of the myocardium in the longitudinal plane.^9,10 The added value of GLS for early diagnosis and refined prognostication compared with standard LVEF has been demonstrated in several studies,^11–13 notably also in patients with LVEF >40% after AMI, where a GLS worse than −14% was associated with a three-fold increase in adverse events.¹¹ Similarly, GLS by cardiac magnetic resonance imaging has shown improved risk stratification in AMI,¹⁴ even in patients with EF ≥50%.¹⁵ A limitation of LVEF is that the biplane Simpson’s method assumes a somewhat symmetric LV geometry. AMI typically results in regional dysfunction altering LV geometry, limiting the ability of LVEF to detect smaller disturbances. In contrast, GLS relies less on geometric assumptions and also provides measurements of regional myocardial function.¹⁶ Moreover, the subendocardial layer with its longitudinal fibres is more vulnerable due to the anatomy of the coronary circulation and impaired in coronary artery disease. Consequently, under increased pressure or compromised coronary blood flow, peak GLS may more accurately reflect LV systolic function than LVEF.¹⁷

Furthermore, LV mechanical dispersion (MD), calculated as a part of GLS analysis, assesses the heterogeneity of the time to peak segmental myocardial shortening. A prolonged MD after AMI has been associated with adverse events such as sudden cardiac death and ventricular arrhythmia.¹⁸ However, to our knowledge, no study has evaluated the prognostic value of GLS and MD after AMI in patients with preserved LVEF.

Recently, the Randomized Evaluation of Decreased Use of betabloCkErs after Acute Myocardial Infarction (REDUCE-AMI) trial found no evidence of a beneficial effect of beta-blockers compared with no beta-blockers in patients with AMI and preserved LVEF when evaluating the primary composite endpoint of death from any cause or new AMI.¹⁹ However, there is still a debate about whether a subgroup of patients with preserved LVEF may benefit from beta-blocker treatment. The present substudy aimed to evaluate whether GLS and MD are superior prognostic markers compared with LVEF, and if beta-blockers may have a beneficial effect in patients with decreased GLS.

Method

Study design

REDUCE-AMI was a prospective, multicenter, open-label, parallel-group, registry-based randomized controlled trial evaluating routine initiation of beta-blockers in patients after AMI with preserved ejection fraction (LVEF ≥50%). The design and rationale of the main trial and its primary results have been published previously.^19,20 The trial was conducted in three countries: Sweden (38 centres), Estonia (1 centre), and New Zealand (6 centres). This prespecified substudy included four centres in Sweden with routine follow-up in the Swedish Web System for Enhancement and Development of Evidence-based Care in Heart Disease Evaluated According to Recommended Therapies (SWEDEHEART) registry.²¹

The trial was approved by the Swedish Ethical Review Authority (2016/1707-31/4 and 2018/2499-32).

Patient population

The trial included adult patients who provided written informed consent up to 7 days after AMI with preserved LVEF (≥50%) by echocardiography at the time of inclusion and significant obstructive coronary artery disease by invasive coronary angiography at any time point before randomization. Exclusion criteria included any condition that could influence patients’ ability to comply with the study protocol or an indication or contraindication for beta-blocker treatment other than as secondary prevention after AMI (see Supplementary data online, Table S1).

For pragmatic reasons, only sites in Sweden that had enrolled >100 patients into the REDUCE-AMI trial and were using General Electric (GE) Healthcare scanners and EchoPac software for echocardiographic evaluation were invited to participate in this substudy. No additional informed consent was required.

Study procedure and data collection

After providing written informed consent, patients were randomly assigned to either initiation of beta-blocker treatment (Metoprolol or Bisoprolol) or no beta-blocker treatment in a 1:1 manner with permuted blocks using an internet-based randomization module. Patients who were already on beta-blockers and randomized to no beta-blockers received a tapering schedule over 2–4 weeks.

This study utilized data from several mandatory Swedish registries. Baseline data for this patient population, as well as data on readmission due to a myocardial infarction were collected from the SWEDEHEART register, which encompasses data from the hospital stay and at two routine follow-up visits (6–10 weeks and 11–13 months after AMI) and has been described in detail earlier.²¹ Information about the date of death, emigration, and cause was obtained from the Swedish Population Register and the National Cause of Death Register. The incidence of heart failure was obtained from the National Patient Register.

Adherence data were collected from SWEDEHEART (details in Supplementary data online, Appendix). For patients randomized to beta-blockers, non-adherence was defined as the non-use of beta-blockers at follow-up visits; for patients randomized to no beta-blocker use, non-adherence was defined as the reported use of beta-blockers during follow-up visits.

Echocardiography analysis

Echocardiography was performed during the hospital stay for AMI. Standard echocardiographic images were recorded according to clinical routine and established guidelines. Basic data analysis, including LVEF assessed with biplane Simpson’s method, was performed by experienced echocardiographers.

All GLS analyses were performed by the first author and were measured by manual delineation of the endocardial border with careful adjustment after assessment of myocardial tracking. One cycle from each apical view was used, with a frame rate of at least 40 frames per second. The end of systole was defined by the automatically identified aortic valve closure. Image quality was deemed poor if two or more adjacent segments of the endocardial border were not clearly visualized in all three apical views, resulting in the exclusion of GLS measurements for that patient.

Although GLS is mathematically a negative value, for the sake of simplicity, it is presented as a positive value in this manuscript.

All echocardiographic examinations were performed using GE Healthcare E95 scanners, using phased array 1.5–4.6 MHz probes, and EchoPac software version 204 for post-processing.

Statistical analysis

No separate sample size calculation for this substudy was performed.

Patients randomized to beta-blockers were compared in terms of background factors and echocardiographic findings to those not receiving beta-blockers. Continuous variables are presented as medians and interquartile ranges (IQRs) (Q1, Q3). Categorical variables are presented as counts with corresponding percentages.

The main analysis in this study was to assess the prognostic value of GLS vs. LVEF among patients with AMI and preserved LVEF. The variables GLS and LVEF were treated as natural splines with 3 degrees of freedom and analysed using Cox proportional hazard regression modelling. One model used LVEF as an independent variable, and the other model used LVEF and GLS. The models were adjusted for age, sex, hypertension, smoking, diabetes mellitus, previous AMI, impaired renal function, and multi-vessel disease. These models were compared using a likelihood ratio (LR) test. Furthermore, as a secondary analysis, we performed Cox proportional hazard analyses adding beta-blocker randomization status to the aforementioned models. Secondary analyses also included the assessment of MD as a prognostic marker for death from any cause or new AMI, and the assessment of GLS as a prognostic marker for readmission due to heart failure. All analyses were performed using the intention-to-treat principle.

Missing data were handled as missing completely at random and no imputation was applied. Data processing and analysis were performed in R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Patients

For this substudy, 1436 patients (28.6% of all included in REDUCE-AMI) were included from 4 different sites in Sweden, 719 patients (50.1%) were randomized to beta-blocker treatment, and 717 patients (49.9%) were randomized to no-beta-blockers. A total of 3352 patients were not eligible (1243 from low-volume centres, 2109 due to different equipment). The baseline characteristics (Table 1) were well balanced between the two groups and did not differ significantly from the main trial, with the exception of a higher proportion of prior hypertension and diabetes in this substudy (49.7% vs. 46.2%, P = 0.02, and 17.3% vs. 13.8%, P = 0.001, respectively). The median age was 65 years and 21.7% were females. At the time of AMI, 32.5% had an ST elevation myocardial infarction, and 44.4% had multi-vessel disease. At discharge, patients were well treated according to guidelines, 95.5% were treated with PCI and 4.8% with coronary artery bypass grafting.

Echocardiographic characteristics are presented in Table 2. After excluding 225 patients (15.7%) due to poor image quality and 99 patients (7.9%) for use of non-GE equipment (some sites utilize multiple vendors), GLS analysis was feasible in 1112 patients (77.4%). The median GLS was 17.3% [intraobserver variability with intraclass correlation coefficient 0.928 [95% confidence interval (CI), 0.855—0.966]. No patient had more than mild valvular dysfunction.

Treatment adherence

Of those randomized to beta-blockers 92.3% and 82.5% still reported beta-blocker use after 6–10 weeks and 11–13 months, respectively. Of those randomized to no beta-blockers, 11.8% and 13.0% reported the use of beta-blockers after 6–10 weeks and after 11–13 months, respectively.

Outcomes

During a median follow-up time of 2.9 years (IQR 1.7–4.3), 59 out of 719 patients (8.2%) in the beta-blocker group reached the primary composite endpoint of death from any cause or new AMI, compared with 51 out of 717 patients (7.1%) in the no-beta-blocker group [hazard ratio (HR) 1.07; 95% CI, 0.73 to 1.57; P = 0.7] (Table 3, Figure 1). The addition of GLS to LVEF compared with LVEF alone did not result in any significant increase in prognostic value, as assessed using the LR test (P = 0.56). The relationship between the HR for the primary outcome and the respective levels of GLS and LVEF is demonstrated in Figure 2. Including beta-blocker randomization status as an independent variable did not change the result (P = 0.31). Figure 3 demonstrates the same relationship as Figure 2, by beta-blocker randomization status. The addition of MD and beta-blocker randomization status did not improve the risk prediction (P = 0.6) (see Supplementary data online, Figure S1). The occurrence of rehospitalization due to heart failure (5 out of 719, 0.7% in the beta-blocker group, and 7 out of 717 patients, 1.0% in the no-beta-blocker group) (Table 3) was too few to justify statistical analysis.

Discussion

In the present substudy of the REDUCE-AMI trial, the addition of GLS to LVEF did not add prognostic value regarding death from any cause or new AMI. These findings were consistent regardless of the treatment group. Similarly, the addition of MD did not add prognostic value. To the best of our knowledge, this is the first evaluation of GLS as a prognostic marker compared with LVEF in patients with AMI and preserved LVEF in the setting of a randomized trial, as well as the first to assess the potential influence of long-term beta-blocker therapy on the relationship between these imaging parameters and hard outcomes.

Previous evidence

The randomized landmark trials from the 1980s are not directly comparable to our study, as they did not routinely assess LV function and were conducted before the advent of widespread use of reperfusion therapy and effective secondary preventive strategies. Consequently, those studies included higher-risk patient populations, leading to higher event rates, e.g. death occurred in 13.9% in the placebo group and 7.77% in the beta-blocker group (P < 0.001) with a cumulative reinfarction rate of 20.1% in the placebo group and 14.4% in the beta-blocker group (P < 0.001).² In contrast, patients in this study likely had much smaller infarcts and lower cumulative risk, and death occurred in 3.3% in the no-beta-blocker group and 4.2% in beta-blocker (P = 0.7), and similarly for reinfarction; 4.0% in no-beta-blocker group and 4.6% in beta-blocker group (P = 0.9).

A few recent observational studies evaluating GLS as a prognostic marker in patients with AMI and preserved or mildly reduced LVEF indicated that GLS has added prognostic value compared with LVEF.^11,22 However, this was not demonstrated in the present trial. Possible explanations for this difference may be that previous observational studies included higher-risk patient populations (e.g. approximately 10% of patients had previous AMI,¹¹ 20% were diabetic,²² and 60% were current smokers^11,22), they also had smaller sample sizes, and had less optimal use of reperfusion therapy (approximately 80% of patients received reperfusion therapy¹¹) as compared with our cohort.

Beta-blocker therapy

The neutral effect of beta-blocker treatment compared with no-beta-blocker treatment observed in our study aligns with the finding of the main trial,¹⁹ as well as recent observational studies and meta-analysis of such studies.^23–25 The patient population in the present substudy comprises patients representative of both the overall REDUCE-AMI patient population as well as the underlying population,^19,26 having a generally low risk and receiving guideline-recommended therapy to a high degree. The low levels of MD in both groups also indicate that the enrolled patients had low risk of ventricular arrhythmia.

Global longitudinal strain vs. LVEF

The neutral result regarding the additional value of GLS compared with LVEF in this trial may reflect the healthier population enrolled. Previous studies enrolled sicker patients with higher event rates.^11–15 With most patients having normal systolic function and subsequently low event rates, it appears the superior diagnostic capacity of GLS to LVEF does not materialize in this population. The pragmatic design of this trial did not include detailed collection of reasons for screening failures. Consequently, we do not know if patients with preserved LVEF but larger infarct sizes were excluded from participation by the treating physicians. However, as described above, it appears the REDUCE-AMI enrolled a population representative of contemporary AMI patients,²⁶ in which neither GLS nor MD seem to add incremental value.

Clinical consequences of trial results

Current guidelines strongly recommend the use of beta-blockers after AMI, primarily based on evidence from older landmark trials.¹ The neutral clinical results from the REDUCE-AMI trial were recently complemented with findings from a prespecified substudy that did not demonstrate any impact of routine beta-blocker treatment on quality of life and well-being.^27. The present study suggests that despite using more comprehensive diagnostic methods combining LVEF, GLS, or MD, it is not possible to identify any subset of patients with preserved LVEF after AMI who draw benefit from beta-blocker treatment. Such a minority of post-AMI patients may still exist, but future research needs to provide and evaluate novel prognostic markers in order to identify individuals likely to benefit. At this time point, there are other large, ongoing trials examining long-term treatment with beta-blockers in similar AMI patients in order to evaluate their effect in lowering mortality and cardiovascular risk. These studies enrol patients with LVEF ≥40% which constitutes a great opportunity to further assess GLS as a prognostic marker.^28,29

Strengths and limitations

The strengths of this trial include the large sample size, the representativeness of the patient population, and the use of national registries to ensure a comprehensive follow-up. Another important strength is that all GLS measurements were performed by a single operator, eliminating interobserver variability. Limitations include the relatively healthy, low-risk population in this study, which makes it difficult to compare the study with other observational studies that generally evaluated sicker populations. There is also a risk of selection bias in this study population due to the pre-requisite of high-volume centres and use of specific equipment to participate in this substudy, and the additional exclusion of patients due to poor image quality, which left only around one-quarter of the initial study population. Another limitation includes a lack of comprehensive beta-blocker dose and adherence information, which may potentially affect the results when evaluating the beta-blocker effect on GLS as a prognostic marker.

Conclusion

In patients after AMI with preserved LVEF, GLS did not add prognostic value regarding death from any cause or new AMI compared with LVEF. In addition, beta-blocker treatment did not alter the prognostic information obtained from GLS compared with no-beta-blocker treatment, indicating that the lack of beta-blocker effect on outcomes is consistent across the spectrum of GLS and LVEF when LVEF is preserved. Consequently, this study does not support an additive value of GLS compared with standard echocardiographic measurements. Overall, in contrast to routine initiation, it appears reasonable to adopt a more restrictive use of beta-blockers in this patient population.

Supplementary data

Supplementary data are available at the European Heart Journal—Cardiovascular Imaging online.

Acknowledgements

We want to thank the hospital staff at all involved hospitals for recruiting patients and the patients for participating and contributing to the research.

Funding

This work was supported by grants from the Swedish Heart Lung Foundation (20210216, to K.M.; and 20180187 and 20210273 to R.H.) and Region Stockholm (2018-0490 and 2021-0932 to R.H.).

Data availability

The data underlying this article cannot be shared publicly due to the General Data Protection Regulation (2016/679). The data will be shared on reasonable request to the corresponding author.

References

Author notes