Prognostic utility of quantifying evolutionary ST-segment depression on early follow-up electrocardiogram in patients with non-ST-segment elevation acute coronary syndromes

Abstract

Introduction

The standard 12-lead electrocardiogram (ECG) is a well-established clinical tool that provides immediate prognostic information for early risk stratification of patients with non-ST-segment elevation acute coronary syndromes (NSTE-ACS). In particular, the presence of ST-segment depression (STD) on the admission ECG has been well recognized to be a powerful adverse prognosticator.^1–6 Accordingly, it has been incorporated into various validated risk models.^7–10 Several clinical trial investigations have further suggested that more extensive STD on admission predicts a greater risk of unfavourable cardiovascular events.^11–13

Notwithstanding the practical and robust prognostic utility of the admission ECG, it provides only a single initial assessment that does not reflect persistence or resolution of myocardial ischaemia. Indeed, silent dynamic STD captured by continuous ST-segment monitoring soon after admission has been shown to be independently associated with adverse short- and long-term outcomes.^14–19 Since STD detected on continuous ST-segment monitoring and standard 12-lead ECG may reflect the same underlying pathophysiology,^1,19–21 it is conceivable that examination for ST-segment deviation on early post-admission follow-up ECG could also provide important prognostic insights. Although ECG is inexpensive, widely available, and often repeated routinely during the standard in-hospital management of patients admitted with NSTE-ACS in contemporary clinical practice, the prognostic utility of these readily available follow-up ECG has not been adequately defined. In particular, very little is known about the incremental prognostic value of STD on an early follow-up ECG performed within 12–24 h of admission. In addition, it remains unclear whether incorporating quantitative evaluation of the magnitude of STD on early follow-up ECG could improve risk assessment.

Accordingly, using the ECG data acquired in the Superior Yield of the New Strategy of Enoxaparin, Revascularization, and Glycoprotein IIb/IIIa Inhibitor (SYNERGY) trial, we sought to: (i) describe the evolution of STD between admission and routine 12–24 h follow-up ECG among contemporary NSTE-ACS patients; (ii) determine the prognostic value of STD on early follow-up ECG beyond that of STD on admission; (iii) examine whether incorporating quantitative evaluation of STD on early follow-up ECG in conjunction with the admission ECG can further refine risk stratification of patients with NSTE-ACS.

Methods

Study setting and population

This investigation was a pre-specified objective of the SYNERGY ECG substudy. Details of the SYNERGY trial^22,23 have been previously published. Briefly, SYNERGY was a prospective, randomized trial conducted at 467 centres in 12 countries to evaluate the efficacy of open-label enoxaparin vs. unfractionated heparin in 9978 high-risk patients with NSTE-ACS managed with contemporary guideline-recommended medical therapies and an intended early invasive strategy. Patients with ischaemic symptoms of ≥10 min within the preceding 24 h were eligible if they met two or more of the following criteria: age ≥60 years, troponin or creatinine kinase-MB elevation above the upper limit of normal for the local laboratory, or STD ≥1 mm or transient ST-elevation <30 min in two contiguous leads on ECG. This substudy examined 4292 patients enrolled from 85 North American SYNERGY sites who had baseline admission and early in-hospital follow-up ECG collected and analysed centrally at the core laboratory.²⁴ We excluded those with ECGs that were incomplete or performed at an unknown time or outside of the protocol-specified time window, with left bundle branch block or ventricular-paced rhythm, and those with recurrent myocardial infarction (MI) within 24 h of admission. Thus, the present study cohort consisted of the remaining 3877 patients (Figure 1). Institutional review board or Ethics Committee approval was obtained at each enrolling site, and all patients provided written informed consent.

Figure 1

SYNERGY electrocardiogram substudy population categorized according to the status of ST-segment depression on the admission and 12–24 h follow-up electrocardiograms. ECG, electrocardiogram; STD, ST-segment depression.

Open in new tabDownload slide

Electrocardiogram interpretation

The admission ECG at enrolment and the follow-up ECG performed within 12–24 h of randomization irrespective of timing for in-hospital revascularization as stipulated a priori in the SYNERGY ECG substudy protocol were labelled and forwarded to the ECG core laboratory at the Canadian Heart Research Centre for evaluation by one of the three full-time trained physicians, who were blinded to patient clinical data and outcome. All ECGs were recorded in standard 12-lead format at a paper speed of 25 mm/s calibrated at 1.0 mV/cm. ST-segment deviation was measured at 80 ms after the J-point in all leads, except aVR. ST-segment depression was considered significant and be present if one or more leads (excluding aVR) exhibited STD ≥ 1 mm. Quantitative measure of STD was defined as the maximum magnitude of STD on any one lead (except aVR) with a significant STD of ≥1 mm and was determined to the nearest 0.5 mm using a magnified calliper. The magnitude of T-wave inversion was abstracted but not incorporated into this present analysis. Incomplete ECG and those with poor quality, paced ventricular rhythm or left bundle branch block precluding accurate evaluation of ST-segment were excluded. This core laboratory had extensive experience in systematic ECG interpretation (interobserver agreement for quantitative ST-segment elevation of 93%) as reported previously.²⁵

Follow-up and clinical outcome

Consistent with the pre-specified primary and secondary outcome measures of the SYNERGY trial, the primary outcome of this study was the composite endpoint of all-cause mortality and/or non-fatal MI at 30 days after randomization. Secondary outcome measures were all-cause mortality at 30 days and the combined incidence of all-cause death and/or non-fatal MI at 6 months.

Patients with index MI at enrolment were not considered in the composite outcome to have MI unless the definition of recurrent MI was fulfilled.²² Because adjudication of recurrent MI was based in part on abnormal cardiac biomarker (re)elevation, there is inherent difficulty in defining its precise timing and in distinguishing early recurrent MI from index MI. As our objectives are to critically evaluate the predictive value of the follow-up ECG at 12–24 h for future cardiovascular events, recurrent MI adjudicated as occurring within 24 h of enrolment was excluded. All cardiovascular outcomes within 30 days of randomization were adjudicated by the blinded clinical events committee. Extended follow-up at 6 months was conducted by telephone contact and clinic visit. Medical records were reviewed to ascertain clinical events including death and MI between 30 days and 6 months. Patient follow-up for vital status was 99.9% complete at 6 months.

Statistical analysis

Continuous variables are reported as medians with inter-quartile ranges (IQR) and compared using the Kruskal–Wallis test. Categorical data are presented as frequencies and percentages with comparisons conducted by the Pearson χ² test. The study population was stratified into four mutually exclusive subgroups according to the status of STD on the admission and follow-up ECGs: Group 1, no STD on either ECG; Group 2, the absence of STD on admission and development of de novo STD at follow-up; Group 3, STD on admission that normalized at follow-up; and Group 4, persistent STD on admission and follow-up. Unadjusted differences in clinical events were first compared across the four subgroups. Selected pair-wise comparisons between subgroups with no STD (Group 1) vs. de novo STD (Group 2) and subgroups with normalized (Group 3) vs. persistent (Group 4) STD were then performed specifically to examine the incremental prognostic significance of STD on the follow-up ECG. The Kaplan–Meier survival curves were generated for each subgroup, and intergroup differences in time to first clinical event were tested by the log-rank method.

To determine the independent prognostic value of STD on each of the admission and follow-up ECGs, Cox's proportional hazards models were constructed with forced entry of the magnitude of STD on the two respective ECGs simultaneously as continuous variables together with clinical parameters of established prognostic significance in this study population as reported in prior studies and SYNERGY risk prediction models.^26,27 For the composite endpoint of death/MI at both 30 days and 6 months, we adjusted for age, gender, ethnicity, height, heart rate, diabetes, smoking status, prior MI, creatinine, treatment randomization, and region of enrolment, in addition to peak cardiac troponin within the first 24 h of admission.²⁷ The model for 30-day mortality was adjusted for age, weight, smoking status, prior MI, heart failure and percutaneous coronary intervention, heart rate, blood pressure, rales on admission, treatment randomization, and creatinine and peak cardiac troponin level.²⁷ Further, we performed sensitivity analyses which (i) additionally incorporated the presence of Q-wave on each of the admission and follow-up ECGs as covariates, given that Q-wave is a clinically relevant ECG feature of potential independent prognostic value, and (ii) additional exclusion of MI occurring within 24 h after revascularizations undertaken before the follow-up ECG was acquired, as it cannot be precisely ascertained whether these events were not procedural-related, or occurred subsequent to the follow-up ECG. We employed the smoothing spline-based score tests to verify proportionality assumption.²⁸ Linearity assumptions were tested for all continuous variables using restricted cubic spline transformations, and violations were handled by appropriate transformations, such as piece-wise linear splines. Statistical analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC, USA). Statistical significance was defined as two-sided P-value of <0.05.

Results

The study population comprising 3877 patients was similar to the rest of the SYNERGY trial population with respect to all primary and secondary outcome measures as well as major clinical characteristics, with the exception of a higher prevalence of smoking (62 vs. 57%) and previous coronary bypass surgery (18 vs. 16%), and a lower likelihood of having hypertension (66 vs. 69%) or prior angina (42 vs. 48%).

At baseline, STD was present on the admission ECG in 1110 (28.6%) patients. Of these patients, 534 (48.1%) sustained persistent STD on the follow-up ECG, whereas the remaining 576 (51.9%) had normalization of STD. Among the 2767 (71.4%) patients without STD on admission, a minority (n = 174, 6.3%) subsequently developed de novo STD on the follow-up ECG (Figure 1). The median time elapse from randomization to acquisition of follow-up ECG was 16.5 h (IQR = 10.6, 20.7). Overall, 16.4% of the follow-up ECGs were acquired after in-hospital revascularization, the proportion of which and timing of acquisition were not significantly different across subgroups (Table 1).

Baseline characteristics of the study population stratified by the status of STD on the admission and follow-up ECGs are illustrated in Table 1. Compared with their counterparts without STD on the follow-up ECG, patients with persistent STD on the follow-up ECG had a more adverse clinical profile, including advanced age with greater prevalence of hypertension, diabetes, and prior heart failure, in addition to higher heart rate on presentation as well as more extensive coronary artery disease on angiography and lower left ventricular ejection fraction on echocardiography (Table 1).

Table 2 displays the unadjusted incidence of death, MI, and the composite endpoint of death/MI at 30 days and 6 months according to the status of STD on the admission and follow-up ECGs. The respective rates of death and death/MI were consistently lowest in patients without any STD, intermediate among those with STD on either the admission or follow-up ECG, and highest in those with persistent STD (Table 2). Corresponding cumulative probabilities of death, MI, and the composite endpoint of death/MI over time for the four subgroups are shown in Figure 2A–C, respectively. In patients without STD on admission, those developed de novo STD compared with their counterparts without this interval change sustained significantly higher mortality at 30 days (P = 0.027) and 6 months (P < 0.001). Among patients exhibiting STD on admission, those with persistent STD on follow-up ECG had significantly higher incidences of all individual and composite endpoints at 30 days (death, P = 0.01; MI, P = 0.025; and death/MI, P = 0.004) and 6 months (death, P < 0.001; MI, P = 0.014; and death/MI, P = 0.0005) compared with those whose STD normalized (Figure 2).

Figure 2

Kaplan-Meier estimates of cumulative probability of (A) death, (B) myocardial (re)infarction, (C) death and/or myocardial (re)infarction, up to 6 months based on status of ST-segment depression on admission and follow-up electrocardiograms. For death: no STD vs. de-novo STD (log-rank tests): P = 0.027 at 30-day, P < 0.001 at 6-month; normalized STD vs. persistent STD (log-rank tests): P = 0.01 at 30-day, P < 0.001 at 6-month. For myocardial (re)infarction: no STD vs. de-novo STD (log-rank tests): P = 0.28 at 30- day, P = 0.38 at 6-month; normalized STD vs. persistent STD (log-rank tests): P = 0.025 at 30-days, P = 0.014 at 6-month. For death and/or myocardial (re)infarction: no STD vs. de-novo STD (log-rank tests): P = 0.27 at 30-day, P = 0.19 at 6-month; normalized STD vs. persistent STD (log-rank tests): P = 0.004 at 30-day, P = 0.0005 at 6-month.

Open in new tabDownload slide

Table 3 summarizes the complementary and independent prognostic value of quantitative STD on each of the admission and follow-up ECGs as derived from the entire study population. After adjustment for established clinical prognosticators specific to this population, the magnitude of STD on both the admission and follow-up ECGs was each a significant and independent predictor of the primary endpoint of death/MI at 30 days, as well as the secondary outcomes of 6-month death/MI and 30-day mortality. There were no significant interactions between STD on the admission and follow-up ECGs for all outcomes (P-value for interaction: 30-day death/MI = 0.32; 6-month death/MI = 0.27; 30-day mortality = 0.94). Specifically, beyond established clinical prognosticators and the extent of STD on the admission ECG of retained independent prognostic significance, every 1 mm increment in the magnitude of STD on the follow-up ECG was further independently associated with a graded additional 1.60 [95% confidence interval (CI) = 1.29–1.98], 1.19 (95% CI = 1.03–1.36), and 1.17 (95% CI = 1.03–1.32) fold increased risk of 30-day mortality, 30-day death/MI, and 6-month death/MI, respectively. Furthermore, in our sensitivity analyses, these associations remained significant and largely unchanged even after additional adjustment for the presence of Q-wave on both ECGs and after exclusion of MI occurring within 24 h of revascularizations undertaken before the follow-up ECG (data not shown).

Discussion

On the basis of the contemporary management of NSTE-ACS in the SYNERGY trial, the results of this substudy demonstrate that an early routine follow-up ECG affords easily interpretable and important prognostic information beyond that provided by the admission ECG. Our principal finding is that independent of established clinical prognosticators and the extent of STD on the admission ECG, STD on routine follow-up ECG performed during the first 12–24 h after admission is a significant adverse prognosticator, and its magnitude confers an independent and graded risk for clinical events at 30 days and 6 months. To our knowledge, the present study is the first to establish the incremental prognostic utility of quantitative analysis of STD on an early follow-up ECG. Our results underscore the potential of incorporating quantitative ST-segment analysis of an early follow-up ECG in conjunction with the admission ECG to better risk stratify patients with NSTE-ACS.

Extensive data from randomized controlled trials and observational studies have established that STD on admission is a powerful independent predictor of unfavourable outcome in patients with NSTE-ACS.^1–6 Accordingly, evaluation of the admission ECG specifically for the presence of STD has been widely endorsed by contemporary practice guidelines^29,30 as a critical component of the integrated approach to patient risk assessment, and such qualitative ECG finding has also been incorporated as an independent dichotomous prognosticator into various validated ACS risk models.^7–10 Beyond the well-accepted and robust prognostic importance of recognizing the presence of baseline STD, several clinical trial investigations have identified an independent and graded relationship between the quantitative extent of STD on admission and the risk for subsequent cardiovascular events.^11–13 Although the generalizability of this finding may not extend to routine clinical practice,^5,31 our findings nevertheless corroborate these prior seminal studies by demonstrating that the magnitude of STD on the admission ECG is predictive of short- and long-term death/MI in a graded manner, in a large contemporary (albeit selected) NSTE-ACS population in a clinical trial.

Although acquisition of repeat ECGs has become a routine part of the ongoing surveillance of patients admitted with NSTE-ACS, limited data are available to support their ability to improve risk assessment. Prior studies by Schechtman et al.,³² the PARAGON-B troponin substudy³³ and the Canadian ACS Registry,³⁴ all of which reported STD on post-admission ECG in relation to clinical outcome, had not explicitly demonstrated incremental prognostic utility of STD on repeat ECG and had reached different conclusions, in part attributed to inadequate power and inherent limitations in their analytical approach which had not directly evaluated the added prognostic value of STD on follow-up ECG beyond that on admission.^32–34 In contradistinction to these seminal investigations, the present study imparts new insights into the evolution of ST-segment changes based on an early follow-up ECG at 12–24 h and their independent prognostic value. First, we performed stratified analysis by the status of STD on admission and demonstrated that the presence of STD on follow-up ECG compared with its absence portends significantly higher unadjusted mortality, both among patients with and without admission STD. Secondly, we explicitly examined the independent prognostic value of STD on follow-up ECG by directly modelling its quantitative magnitude and demonstrated that it conferred a graded increase in risk independent of that predicted by established clinical prognosticators and extent of STD on the admission ECG. Our results provide direct evidence that evaluating STD on follow-up ECGs compliments, but by no means substitutes, the assessment of STD on admission in further refining risk stratification. Thirdly, our follow-up ECGs were acquired as per protocol-defined algorithm at an early prospectively determined time frame of 12–24 h after admission. Its incremental prognostic information available upfront within 24 h of admission is of more practical potential to facilitate early identification of high-risk individuals, thereby guiding risk-tailored, downstream in-hospital management strategies.

It is noteworthy that continuous ST-segment monitoring for dynamic STD early after admission has been demonstrated to carry important prognostic information.^13–17 Nevertheless, its widespread application to ACS patients in routine clinical practice is in part hindered by the lack of an accurate and reliable automated ischaemia detection system and limited experience outside of research settings. Although conceivably devoid of under-sampling errors and thus potentially more sensitive than repeat ECGs in capturing more transient STD, it is anticipated that continuous STD monitoring can complement and even supersede standard 12-lead follow-up ECG in capturing recurrent or persistent ischaemia. Its practical value and robustness are expected to increase with ongoing improvement in its signal fidelity and more universal availability of multilead systems. The cost-effectiveness and practicality of continuous ST-segment monitoring as a clinical tool, as well as its incremental prognostic value and practical utility to refine risk stratification relative to that afforded by standard follow-up ECGs, should be examined in future studies.

Our study has several limitations. First, our analyses were performed on a selected cohort enrolled in a randomized trial. Whether our results can be generalized to patients in routine clinical practice awaits evaluation in future prospective studies of less selected patients. Secondly, it was unknown whether symptomatic ischaemia was present during ECG acquisitions. Nevertheless, previous studies have shown that STD is a more robust predictor than chest pain for adverse cardiovascular events.¹⁹ Furthermore, these ECGs were routinely performed as specified by the study protocol and were not predicated upon subjective clinical assessment. Thirdly, one-sixth of our follow-up ECGs were acquired after in-hospital revascularization. Although it remains to be definitively established whether revascularization and its relative timing with repeat ECG acquisition may modulate the prognostic value of STD on the follow-up ECG, our sensitivity analysis confirms that it is a robust prognosticator for incident events and not a marker of concurrent peri-procedural-related event. Fourthly, our analysis has not addressed the prognostic implications of temporal evolutions in other ECG features such as T-wave inversion, which are important topics for future investigations. Although we had previously confirmed good reproducibility in quantifying ST-segment elevation in our core laboratory, we did not evaluate intra/interobserver agreement in quantitative STD for this study population. Fifthly, site investigators were not blinded to the follow-up ECGs and we cannot exclude the possibility that STD on follow-up ECG would have triggered treatment intensification which in turn could have favourably affected outcome. However, there were no systematic group differences in use of non-randomized medical treatments and revascularization (data not shown). Further, any such potential confounding effect would have only abrogated the prognostic value of STD at follow-up. Finally, future studies should examine whether the prognostic information afforded by follow-up ECG can guide strategic decisions in patient management and ultimately improve outcome.

Conclusion

In patients presenting with NSTE-ACS, the magnitude of STD on a routine follow-up ECG at 12–24 h provides incremental prognostic information independent of established clinical prognosticators and the extent of STD on admission. Incorporating a routine follow-up ECG and its quantitative evaluation for early evolutionary STD may enhance risk stratification of patients with NSTE-ACS.

Funding

The SYNERGY trial was supported by grants from Sanofi-Aventis. The sponsors had no involvement in the study conception or design; collection, analysis, and interpretation of data; in the writing, review, or approval of the manuscript; and in the decision to submit the manuscript for publication.

Conflict of interest: none declared.

Acknowledgement

R.T.Y. is supported by Fellowship Award from the Canadian Institutes of Health Research and the Detweiler Travelling Fellowship Award from the Royal College of Physicians and Surgeons of Canada. A.T.Y. is supported by a New Investigator Award from the Heart and Stroke Foundation of Canada.

References