Abstract
The objective of this study was to determine the combination of left ventricular ejection fraction (LVEF) and individual electrocardiographic parameters related to abnormal depolarization/repolarization or baroreceptor sensitivity that had the best predictive value for major adverse cardiac events (MACE) in patients with acute coronary syndrome (ACS). Patients with ACS who underwent coronary angiography and percutaneous coronary intervention (PCI) were included in this prospective study. Ventricular late potential (VLP), heart rate turbulence (HRT), heart rate variability (HRV), and T wave alternans (TWA) parameters were measured using 24 h Holter monitoring 2-4 weeks after onset of ACS. Initial and follow-up LVEF was measured by ultrasound. Patients were followed for at least 6 months to record the occurrence of MACE. Models using combinations of the individual independent prognostic factors found by multivariate analysis were then constructed to use for estimation of risk of MACE. In multivariate analysis, VLP measured as QRS duration, HRV measured as standard deviation of normal RR intervals, and followup LVEF, but none of the other parameters studied, were independent risk factors for MACE. Areas under ROC curve (AUCs) for combinations of 2 or all 3 factors ranged from 0.73 to 0.76. Combinations of any of the three independent risk factors for MACE in ACS patients with PCI improved prediction and, because these risk factors were obtained non-invasively, may have future clinical usefulness.
Keywords: Acute coronary syndrome, ventricular late potential, heart rate turbulence, heart rate variability, LVEF
Introduction
Acute coronary syndrome (ACS) patients have a high risk for major adverse cardiac events (MACE), including sudden coronary death (SCD). Common, non-invasively determined parameters used for risk assessment of these patients include left ventricular ejection fraction (LVEF) [1], ventricular late potential (VLP) [2], T-wave alternans (TWA) [3], heart rate variability (HRV) [4-7], and heart rate turbulence (HRT) [8].
LVEF, for example, has recently been reported to be an independent predictor of in-hospital death in patients with ACS undergoing percutaneous coronary intervention (PCI) [9]. VLP and TWA, markers of abnormal ventricular depolarization and repolarization, have been reported to predict life-threatening ventricular arrhythmias in ACS patients during coronary reperfusion therapy [10,11]. HRV and HRT are indicators of baroreceptor sensitivity and autonomic nervous system input to the sinus node [12]. HRV has been reported to be a predictor of mortality [13] and HRT has been reported to be a predictor of sudden cardiac death [14,15]. The indicators mentioned above vary in odds ratio (OR), sensitivity, and specificity. For example, low LVEF is a good predictor of over-all mortality, but whether it is specific for SCD is not completely clear [16]. The negative predictive value of VLP for SCD is very high, but its sensitivity for prediction of SCD is low [17].
Because ACS patients have a high rate of later cardiac events, it would be important to develop a non-invasive, rapid, simple, and inexpensive method for risk stratification that can be used widely in clinical practice. These non-invasive electrocardiographic prognostic indicators have been studied individually, but data are lacking on how they all compare with each other in a single defined clinical cardiac condition, and whether a combination of these risk factors might increase the reliability of risk prediction.
In the current study, we evaluated the association with MACE of LVEF and 8 electrocardiographic parameters (HRV, TWA, VLP and its three constituent markers, and the two markers for HRT) in ACS patients who had received percutaneous coronary intervention (PCI), and constructed receiver operating curves (ROC) to determine the area under the curve (AUC), sensitivity, and specificity for MACE of each parameter. We then performed univariate and multivariate regression analysis on those factors that had shown statistically significant association with MACE. Finally, we constructed ROC curves using different combinations of the three factors found to be independent risk factors in multivariate regression in order to determine the combination with the highest sensitivity and specificity for prediction of MACE.
Patients and methods
Patient selection
ACS patients undergoing PCI and standard drug therapy were enrolled as subjects. Inclusion criteria: 1. Fulfillment of the ACS diagnostic criteria (acute ST-elevation myocardial infarction, acute non ST-elevation myocardial infarction and unstable angina); 2. No more than 90 years old; 3. Baseline rhythm was sinus rhythm. 4. Ventricular premature contraction (VPC) more than twice within 24 h (to enable HRT calculation). Exclusion criteria: 1. Sinoatrial block; 2. Atrioventricular block; 3. Temporary and permanent pacemaker implantation; 4. Long-term use of anti-arrhythmic drugs; 5. Malignant tumor; 6. Serious electrolyte disorder interfering with 24 h Holter monitoring; 7. Part or all of the Holter monitor data missing; 8. No myocardial injuries with unstable angina pectoris.
Two hundred eighty-six of the total of 517 ACS patients were selected for the study, based on the inclusion and exclusion criteria. Forty-seven of these patients declined to participate, leaving a total of 239 patients. Of these patients, 37 did not complete the study either because they did not complete the diagnostic procedures or because they were lost to follow-up, leaving a final enrollment of 202 patients. Of these patients, 99 provided TWA results, 150 provided VLP results, 158 provided initial and follow-up LVEF measurements and 48 provided complete results for all 5 markers: LVEF, HRT, HRV, VLP, and TWA.
Institutional review board (IRB) approval for the study was obtained from Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai, China. Informed consent was obtained in advance from all patients.
Study protocol
During hospitalization of patients identified with ACS onset, an ECG workstation (GE, USA) was used to test patients in a resting state, and a DELMAR Avionics DCG Holter monitoring system (USA) was used to perform 24 h dynamic electrocardiography. HRT and HRV analysis systems were also used. During outpatient visits 2-4 weeks after exit from hospitalization, routine ECG examinations and Holter monitoring were given to patients, and follow-up records were established. Follow-up was continued for at least 6 months, and the occurrence of major adverse cardiac events (MACE) was recorded. Initial and follow-up ECG were also recorded using ultrasound.
Definition of MACE
MACE was defined as: 1. All-cause mortality, including sudden cardiac death (SCD); 2. Serious ventricular arrhythmia, including sustained ventricular tachycardia and ventricular fibrillation; 3. The recurrence of non-fatal ACS. 4. Rehospitalization due to reasons 2 and 3 was also defined as MACE if the events due to these reasons were not recorded as MACE elsewhere.
VLP measurement and diagnostic criteria
The duration of the QRS complex (fQRSd) was obtained from an average ECG at high frequency [18]. The root mean square value of the voltage in the last 40 mSec of the QRS complex (RMS (40)) and the duration below 40 µV in the last 40 mSec period (LAS (40)) were also calculated. Technically, the noise ratio was required l to be < 1 µV, with a frequency response of 40-250 Hz. The diagnostic criteria for VLP were fQRSd> 110 ms; RMS (40) < 20 µV; LAS (40)> 38 ms, and if VLP met 2 of these criteria, it was considered positive.
Definition and measurement of TWA
T wave alternans (TWA) refers to the alternation of amplitude and shape of the T wave with every beat in the ECG. According to ACC/AHA/ESC 2006 Guidelines for Management of Patients with Ventricular Arrhythmias and the Prevention of Sudden Cardiac Death, TWA is defined as class IIa of fatal ventricular arrhythmia risks [19]. The electrocardiographic signal was pretreated (removal of ECG interference, baseline drift, power frequency interference, etc.) before TWA calculation. The heart beat was then divided into an odd and an even group. According to the update strategy, heartbeats of the two groups are corrected with asymptotic increment. T wave alternans was calculated as the maximum absolute difference between the two groups, which is the quantitative index of TWA [20].
HRV
HRV, defined as the standard deviation of normal RR intervals (SDNN) during a continuous 24 h period, was calculated through the HRV analysis system.
HRT
Turbulence onset (T) and turbulence slope (TS) were obtained through the HRT analysis system. The neutral point for TO was defined as 0; TO < 0 referred to an accelerated initial heart rate after the premature ventricular contraction; and TO ≥ 0 referred to a decelerated initial heart rate after the premature ventricular contraction; the neutral point for TS was defined as 2.5, TS> 2.5 referred to the deceleration after the acceleration of sinus rhythm, while TS ≤ 2.5 referred to the absence of deceleration [14].
Statistical analysis
Continuous variables without normal distribution were presented as median and inter-quartile range (IQR), and Mann-Whitney U tests used for group comparisons; continuous variables with normal distribution were presented as mean and standard deviation (SD), and independent t tests were used for group comparisons. Categorical variables were presented as counts and percentages. Chi-square tests or Fisher exact tests were applied for group comparisons. Receiver operating characteristic (ROC) analyses were performed to find the optimal cutoff value for each continuous variable and used to detect the effects of different combinations of risk factors of MACE. Youden indexes, defined as (sensitivity + specificity-1), were set to determine the cut-off values for optimal MACE prediction. Area under ROC curve (AUC) with null hypothesis (AUC=0.5) was analyzed by using the Wilcoxon rank sum test. Univariate and multivariate logistic regression analyses were performed to investigate risk factors for MACE. If statistically significant differences between those with MACE and those without MACE were found, they were included in univariate logistic regression analyses. Factors that reached statistical significance in univariate analysis were included in multivariate logistic regression analysis. Factors included in multivariate logistic regression were as follows: gender, age, hypertension, VLP, HRV (SDNN), VLP-fQRSd, VLP-LAS (40), LVEF (follow-up). The stepwise model selection method was used to investigate the influence of factors for MACE diagnosis during multivariate logistic regression analysis. Statistical analyses were performed with SPSS software version 17 (SPSS Inc, Chicago, IL, US), and two-tailed P < 0.05 indicated statistical significance.
Results
Baseline demographic characteristics
Baseline characteristics of subjects with and without MACE are shown in Table 1. Hypertension and use of β-blockers were significantly more common in those with MACE (P=0.022, P=0.034, respectively). The MACE and no MACE groups showed no significant difference in any other baseline demographic characteristic.
Table 1.
Characteristics distribution of subjects
| Total valid N | No MACE (n=171) | MACE (n=31) | P-value | |
|---|---|---|---|---|
| Age (years) | 202 | 62.87±10.81 | 64.29±10.34 | 0.499 |
| Gender | 202 | 0.366 | ||
| Female | 37 (21.64%) | 9 (29.03%) | ||
| Male | 134 (78.36%) | 22 (70.97%) | ||
| proBNP | 195 | 161.25 (67.71-524.23) | 234.1 (82.18-634.1) | 0.119 |
| LDL | 182 | 2.76 (2.14-3.47) | 2.64 (2.22-3.02) | 0.389 |
| SBP (mmHg) | 202 | 130 (117-140) | 128 (116-135) | 0.554 |
| DBP (mmHg) | 202 | 75 (67-82) | 78 (64-89) | 0.76 |
| HR | 202 | 75 (68-80) | 78 (68-83) | 0.294 |
| BMI (kg/m2) | 202 | 24.42±2.86 | 25.44±3.81 | 0.164 |
| Anterior MI | 201 | 0.376 | ||
| No | 133 (78.24%) | 22 (70.97%) | ||
| Yes | 37 (21.76%) | 9 (29.03%) | ||
| Killip gradea | 202 | 0.061 | ||
| I | 160 (93.57%) | 26 (83.87%) | ||
| II | 8 (4.68%) | 3 (9.68%) | ||
| III | 3 (1.75%) | 1 (3.23%) | ||
| IV | 0 (0%) | 1 (3.23%) | ||
| Smoker | 200 | 0.801 | ||
| No | 75 (44.38%) | 13 (41.94%) | ||
| Yes | 94 (55.62%) | 18 (58.06%) | ||
| ACEI/ARB | 202 | 0.198 | ||
| No | 37 (21.64%) | 10 (32.26%) | ||
| Yes | 134 (78.36%) | 21 (67.74%) | ||
| β-blocker | 202 | 0.034 | ||
| No | 22 (12.87%) | 0 (0%) | ||
| Yes | 149 (87.13%) | 31 (100%) | ||
| Statinsa | 202 | 1 | ||
| No | 3 (1.75%) | 0 (0%) | ||
| Yes | 168 (98.25%) | 31 (100%) | ||
| Hypertension | 202 | 0.022 | ||
| No | 70 (40.94%) | 6 (19.35%) | ||
| Yes | 101 (59.06%) | 25 (80.65%) | ||
| Diabetes | 202 | 0.378 | ||
| No | 119 (69.59%) | 24 (77.42%) | ||
| Yes | 52 (30.41%) | 7 (22.58%) | ||
| HRT-TO | 110 | -1.08 (-2.48-0) | -0.84 (-1.24-1.04) | 0.2 |
| HRT-TS | 110 | 5.05 (2.45-8.08) | 5.3 (2.35-8.03) | 0.943 |
| HRV (SDNN) | 167 | 108.64±30.54 | 93.68±31.17 | 0.034 |
| TWA | 99 | 45.5 (39-66.75) | 59 (48-64) | 0.184 |
| VLP | 151 | < 0.001 | ||
| Negative | 121 (70.76%) | 8 (25.81%) | ||
| Positive | 14 (8.19%) | 8 (25.81%) | ||
| VLP-fQRSd | 151 | 87 (80-92) | 92.5 (82-117.75) | 0.031 |
| VLP-RMS (40) | 151 | 28.9 (23.7-39.85) | 27.95 (18.03-36.1) | 0.269 |
| VLP-LAS (40) | 151 | 30 (24-34) | 32.5 (26.75-45.5) | 0.039 |
| LVEF (admission status) | 178 | 61 (57-64) | 59 (55.75-63) | 0.24 |
| LVEF (follow-up) | 180 | 62 (59-64) | 56 (51-64) | 0.002 |
Analyzed by Fisher exact test for group comparison.
Continuous variables without normal distribution are presented as median and inter-quartile range (IQR), and Mann-Whitney tests used for group comparisons; continuous variables with normal distribution are presented as mean and standard deviation (SD) and independent t tests used for group comparisons. Categorical variables are presented as counts and percentages. Chi-square tests or Fisher exact tests were applied for comparisons.
LVEF, fHRV (SDNN), VLP, and VLP-fQRSd were potential risk factors of MACE
Table 1 also shows values for LVEF and the 8 electrocardiographic parameters investigated as possible risk factors of MACE. LVEF, although not significantly different between the two groups at hospital admission, was significantly smaller at follow-up in those with MACE (MACE vs. no MACE, 56% vs. 62%, P=0.002). Beat-to-beat variability in rate (fHRV (SDNN)) was lower in those with MACE (MACE vs. no MACE, 93.68 vs 108.64, P=0.034), and those with MACE were more likely to have a positive VLP (P < 0.001), and a longer ventricular depolarization period (VLP-fQRSd 92.5 vs. 87, P=0.031; VLP-LAS (40) 32.5 vs, 30, P=0.039). The other parameters studied, VLP-RMS (40), HRT-TO, HRT-TS, and TWA, had no significant association with the occurrence of MACE.
ROC results showed significant predictive power for VLP-fQRSd, VLP-LAS (40), HRV (SDNN) and follow-up LVEF
ROC analyses for the 8 potential parameters are shown in Table 2, and the individual ROC for each parameter is shown in Figure 1. Significant predictive power and similar AUCs (0.65 to 0.68) were seen for VLP-fQRSd, VLP-LAS (40), HRV (SDNN), and LVEF (follow-up). The other parameters, HRT-TO, HRT-TS, TWA, and VLP-RMS (40), had no significant discriminatory power.
Table 2.
ROC results for optimal cutoff value and MACE discrimination
| Optimal cutoff value | AUC (95% CI) | Sensitivity | Specificity | P-value | |
|---|---|---|---|---|---|
| HRT-T0 | ≥ -1.17 | 0.61 (0.46-0.75) | 78.57 | 48.96 | 0.200 |
| HRT-TS | ≤ 12.1 | 0.51 (0.35-0.66) | 12.50 | 100 | 0.943 |
| HRV(SDNN) | ≤ 74.5 | 0.64 (0.51-0.77) | 91.03 | 36.36 | 0.036 |
| TWA | ≥ 47.5 | 0.61 (0.45-0.76) | 80.00 | 54.76 | 0.184 |
| VLP-fQRSd | ≥ 97.5 | 0.64 (0.50-0.79) | 45.45 | 89.92 | 0.031 |
| VLP-RMS(40) | ≤ 21.9 | 0.57 (0.43-0.72) | 86.82 | 36.36 | 0.269 |
| VLP-LAS(40) | ≥ 39 | 0.64 (0.50-0.78) | 36.36 | 93.80 | 0.039 |
| LVEF(follow-up) | ≤ 56.5 | 0.68 (0.56-0.81) | 87.42 | 55.17 | 0.002 |
Receiver operating characteristic (ROC) analyses were performed to find the optimal cutoff values for each continuous variable. Youden indexes, defined as (sensitivity + specificity-1) were set to determine the cutoff values for optimal MACE prediction. Area under ROC curve (AUC) with null hypothesis (AUC=0.5) was analyzed by using the Wilcoxon rank sum test.
Figure 1.
ROC for Electrocardiographic Parameters and Follow-up LVEF for Prediction of MACE. Receiver operating characteristic (ROC) analyses were performed to find the optimal cutoff value for each continuous variable. Youden indexes, defined as (sensitivity + specificity -1) were set to determine the cutoff values for optimal MACE prediction. Area under ROC operating curve (AUC) with null hypothesis (AUC=0.5) was analyzed by the Wilcoxon rank sum test.
Values for HRV (SDNN) below the optimal value of 74.5 identified those with MACE with a sensitivity of 91.03%, but a specificity of only 36.3%. Values above than the optimal VLP-fQRSd cut-off value of 97.5 identified patients with MACE were with a sensitivity of only 45.45%, but with a specificity of 89.92%. Values higher than the optimal VLP-LAS (40) value of 39 identified patients with MACE with a sensitivity of 36.36%, but a high specificity (93.8%). LVEF values smaller than the optimal cut-off value of 56.5% identified those with MACE with a sensitivity and specificity of 87.42% and 55.17% respectively.
Six parameters in univariate analysis and three in multivariate analysis were significantly related to MACE
In univariate analysis (Table 3), six parameters, hypertension, VLP, VLP-fQRSd, VLP-LAS (40), HRV (SDNN), and LVEF (follow-up), were significantly related to diagnosis of MACE. Step-wise multivariate analysis was performed with the six parameters plus gender and age. Table 4 shows only 3 of these parameters, HRV (SDNN), VLP-fQRSd, and LVEF (follow-up) to be significantly and independently related to MACE. With adjustment for the other 2 parameters, the odds for MACE were significantly higher in those with HRV (SDNN) ≤ 74.5 (adjusted OR=6.41, P=0.017), in those with VLP-fQRSd ≥ 97.5 (adjusted OR=5.99, P=0.007), and in those with LVEF (follow-up) ≤ 56.5 (adjusted OR=4.85, P=0.021). The highest odds for MACE were observed with HRV (SDNN) ≤ 74.5.
Table 3.
Results of univariate logistic regression to detect risk factors
| β±SE | OR (95% CI) | p-value | |
|---|---|---|---|
| β-blocker | 12.41±230.9 | NA | 0.957 |
| Hypertension | 1.06±0.48 | 2.89 (1.13-7.41) | 0.027 |
| VLP (ref=negative) | 2.16±0.57 | 8.64 (2.81-26.63) | < 0.001 |
| HRV(SDNN) (ref ≥ 74.5) | 1.76±0.53 | 5.8 (2.05-16.4) | 0.001 |
| VLP-fQRSd (ref ≤ 7.5) | 2.01±0.52 | 7.44 (2.69-20.55) | < 0.001 |
| VLP-LAS (40) (ref ≤ 39) | 2.16±0.57 | 8.64 (2.81-26.63) | < 0.001 |
| LVEF (follow-up) (ref ≥ 56.5) | 2.15±0.45 | 8.55 (3.56-20.53) | < 0.001 |
NA: non-available for unstable estimates. Univariate logistic regression analyses were performed to investigate risk factors for MACE. If statistically significant differences between those with MACE and those without MACE were found, they were included in the univariate logistic regression analyses.
Table 4.
Results of multivariate logistic regression to detect risk factors
| β±SE | OR (95% CI) | p-value | |
|---|---|---|---|
| HRV (SDNN) (ref ≥ 74.5) | 1.86±0.78 | 6.41 (1.4-29.35) | 0.017 |
| VLP-fQRSd (ref ≤ 97.5) | 1.79±0.66 | 5.99 (1.64-21.84) | 0.007 |
| LVEF (follow-up) (ref ≥ 56.5) | 1.58±0.69 | 4.85 (1.27-18.59) | 0.021 |
Multivariate logistic regression with stepwise model selection method were performed to investigate risk factors for MACE. Factors that reached statistical significance in univariate analysis were included in multivariate regression analysis. Factors included in multivariate logistic regression were as follows: gender, age, hypertension, VLP, HRV (SDNN), VLP-fQRSd, VLP-LAS (40), LVEF (follow-up). This final result was conducted by stepwise model selection.
The combination of HRV (SDNN), VLP-fQRSd, and follow-up LVEF produced the highest AUC
The AUC results for combinations of the 3 parameters found through multivariate analysis to be independent risk factors for MACE are shown in Table 5. Combinations of any 2 of these parameters produced the same AUC, 0.73. The combination of all 3 factors produced a somewhat higher AUC, 0.76. The logit equation for the 3 parameter combination is shown below:
Table 5.
Area under ROC curve in different combinations of risk factors
| Combined factors | AUC | 95% CI |
|---|---|---|
| HRV (SDNN) & VLP-fQRSd | 0.73 | 0.59-0.87 |
| HRV (SDNN) & LVEF (follow-up) | 0.73 | 0.60-0.87 |
| VLP-fQRSd & LVEF (follow-up) | 0.73 | 0.59-0.87 |
| HRV (SDNN) & VLP-fQRSd & LVEF (follow-up) | 0.76 | 0.62-0.91 |
CI: Confidence Interval. Receiver operating characteristic (ROC) analyses were used to detect the effects of different combinations of risk factors for MACE. The larger area under the ROC curve (AUC) indicates the better combination.
Logit (MACE/No MACE) = -0.49 + 1.86 × HRV (SDNN) + 1.79 × VLP-fQRSd + 1.58 × LVEF (follow-up)
All combinations show good discriminatory power compared to the discriminatory power (AUC 0.64) seen when either of the three is used alone.
Discussion
In the current study, three non-invasively determined, significant, independent risk factors for MACE were identified in ACS patients who had undergone PCI. These were an indicator of cardiac contractile function (LVEF), an indicator of the sensitivity of cardiac ANS regulation (HRV (SDNN)), and an indicator of the speed of action potential conduction in the ventricles (fQRSd).
Continuation of the QRS signal beyond its normal termination is a sign of slow action potential conduction at some location in the ventricle and of arrhythmogenic potential. We studied four parameters by which the lengthening of the QRS can be quantified: VLP, VLP-LAS (40), VLP-RMS (40), and VLP-fQRSd. Of these parameters, only the total duration of the QRS complex, VLP-fQRSd, was a significant and independent predictor of MACE. VLP (the combination of at least 2 of the 3 other markers) and the markers using information from the last 40 ms of the QRS complex (that is, VLP-LAS (40) and VLP-RAM (40)) were not independent related factors in multivariate analysis. A prolonged QRS has previously been reported to be predictive of cardiac death in suspected coronary artery disease patients and in post-MI patients, but to have no prognostic significance in non-ischemic cardiomyopathy [11,21-23].
Changes in ventricular repolarization (that is, TWA) were not related to of MACE in any of the statistical analyses performed here. In other studies, TWA, the beat-to-beat fluctuation in the T wave, has been shown to be a long term predictor of arrhythmia [10,24,25], and a predictor for post-MI cardiac death [8,26]. However, the mechanistic link between repolarization and arrhythmia is less direct than the link between depolarization and arrhythmia. Further studies are needed to clarify the use of TWA as a predictor.
The sensitivity of the heart to ANS control decreases with age and in the failing heart [14,27]. Heart rate variability, that is, the oscillation in the interval between heart beats, and the two markers of heart rate turbulence are indicators of this sensitivity. In our study HRV was within the range reported by others for the elderly and sick [15,28-30] and was an independent marker for MACE, but the two turbulence markers, HRV-TO and HRV-TS, showed no significant association with MACE in any of our statistical analyses. In previous studies, decreased HRV has consistently been associated with increased risk of death and cardiac death, results consistent with ours. However, turbulence markers have also been reported to predict cardiac events [31]. In one previous study, HRV and HRV-TS predicted non-fatal cardiac events [32], and in another study, if both HRV-TO and HRV-TS were abnormal, this combination was found to be as good a predictor of post-MI death as decreased LVEF [14,32]. These results do not correspond to ours.
Multivariate analysis yielded 3 significant and independent predictive factors for MACE. Of these factors, the marker for poor ANS sensitivity (that is, HRV (SDNN)) had the highest OR for MACE and showed high sensitivity and low specificity in MACE analysis. The marker for abnormal ventricular depolarization speed (VLP-fQRSd) had very high specificity for MACE, but low sensitivity, and an OR of 5.99. LVEF, the marker for abnormal pumping function, showed intermediate sensitivity and specificity compared to the other 2 parameters. And combinations of any 2 or of all three parameters gave higher ROC values than those for any individual parameter.
This study has several limitations. It was limited by the relatively small number of samples. The sample size and number of events included in the multivariate regression model was even lower due to missing data. The relatively mild severity of the patients’ cardiovascular disease, and thus the overall low occurrence rate of MACE may have caused a lack of baseline differences in some variables between the MACE and non-MACE patient groups. Additional limitations were that no filtering method was described for the time series, and there was no detailed explanation of the clinical characteristics of the cardiovascular disease of individual patients (we felt that inclusion of individual clinical characteristics would detract from the focus of the study).
In conclusion, identification of patients with MACE using a combination of any two or all three risk factors (VLP-fRQSd, HRV (SDNN) and LVEF) in ACS patients with PCI was more powerful (AUC 0.73 to 0.76) than identification using only one of the risk factors (each AUC 0.64). It improves predictability and has the potential to become a useful clinical tool.
Acknowledgements
This study was funded by Science and Technology Commission of Shanghai Municipality (fund number: 114119a8900).
Disclosure of conflict of interest
None.
References
- 1.Passman R, Goldberger JJ. Predicting the future: risk stratification for sudden cardiac death in patients with left ventricular dysfunction. Circulation. 2012;125:3031–3037. doi: 10.1161/CIRCULATIONAHA.111.023879. [DOI] [PubMed] [Google Scholar]
- 2.Simson MB. Use of signals in the terminal QRS complex to identify patients with ventricular tachycardia after myocardial infarction. Circulation. 1981;64:235–242. doi: 10.1161/01.cir.64.2.235. [DOI] [PubMed] [Google Scholar]
- 3.Adachi K, Ohnishi Y, Yokoyama M. Risk stratification for sudden cardiac death in dilated cardiomyopathy using microvolt-level T-wave alternans. Jpn Circ J. 2001;65:76–80. doi: 10.1253/jcj.65.76. [DOI] [PubMed] [Google Scholar]
- 4.Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation. 1996;93:1043–1065. [PubMed] [Google Scholar]
- 5.Huikuri HV, Stein PK. Heart rate variability in risk stratification of cardiac patients. Prog Cardiovasc Dis. 2013;56:153–159. doi: 10.1016/j.pcad.2013.07.003. [DOI] [PubMed] [Google Scholar]
- 6.Thayer JF, Yamamoto SS, Brosschot JF. The relationship of autonomic imbalance, heart rate variability, and cardiovascular disease risk factors. Int J Cardiol. 2010;141:122–131. doi: 10.1016/j.ijcard.2009.09.543. [DOI] [PubMed] [Google Scholar]
- 7.Liu Y, Syed Z, Scirica BM, Morrow DA, Guttag JV, Stulz CM. ECG morphological variability in beat space for risk stratification after acute coronary syndrome. J Am Heart Assoc. 2014;3:e000981. doi: 10.1161/JAHA.114.000981. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sulimov V, Okisheva E, Tsaregorodtsev D. Non-invasive risk stratification for sudden cardiac death by heart rate turbulence and microvolt T-wave alternans in patients after myocardial infarction. Europace. 2012;14:1786–1792. doi: 10.1093/europace/eus238. [DOI] [PubMed] [Google Scholar]
- 9.Sladojevic M, Sladovejic S, Culibrk D, Tadic S, Jung R. Echocardiographic parameters as predictors of in-hospital mortality in patients with acute coronary syndrome undergoing ercutaneous coronary intervention. ScientificWorldJournal. 2014;2014:818365. doi: 10.1155/2014/818365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Takasugi N, Kubota T, Nishigake K, Verrier RL, Kawasaki M, Takasugi M, Ushikoshi H, Hattori A, Ojio S, Aoyama T, Takemura G, Minatoguchi S. Continuous T-wave alternans monitoring tonpredict impending life-threatening cardiac arrhythmias during emergent coronary reperfusion therapy in patients with acute coronary syndrome. Europace. 2011;13:708–715. doi: 10.1093/europace/euq512. [DOI] [PubMed] [Google Scholar]
- 11.Santangeli P, Infusino F, Sgueglia GA, Sestito A, Lanza GA. Ventricular late potentials: a critical overview and current applications. J Electrocardiol. 2008;41:318–324. doi: 10.1016/j.jelectrocard.2008.03.001. [DOI] [PubMed] [Google Scholar]
- 12.Vanderlei LC, Pastre CM, Hoshi RA, Carvalho TD, Godoy MF. Basic notions of heart rate variability and its clinical applicability. Rev Bras Cir Cardiovasc. 2009;24:205–217. doi: 10.1590/s0102-76382009000200018. [DOI] [PubMed] [Google Scholar]
- 13.Harris PRE, Stein PK, Fung GL, Drew BJ. Heart rate variability measured early in patients with evolving acute coronary syndrome and 1-year outcomes of rehospitalization and mortality. Vasc Health Risk Manag. 2014;10:451–464. doi: 10.2147/VHRM.S57524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Francis J, Watanabe MA, Schmidt G. Heart rate turbulence: A new predictor for risk of sudden cardiac death. ANE. 2005;10:102–109. doi: 10.1111/j.1542-474X.2005.10102.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Marynissen T, Flore V, Heidbuchel H, Nuyens D, Ector J, Willems R. Heart rate turbulence predicts ICD-resistant mortality in ischaemic heart disease. Clin Res. 2014;16:1069–1077. doi: 10.1093/europace/eut303. [DOI] [PubMed] [Google Scholar]
- 16.Dagres N, Hindricks G. Risk stratification after myocardial infarction: is left ventricular ejection fraction enough to prevent sudden cardiac death? Eur Heart J. 2013;34:1964–1971. doi: 10.1093/eurheartj/eht109. [DOI] [PubMed] [Google Scholar]
- 17.Buxton AE. Risk stratification for sudden death in patients with coronary artery disease. Heart Rhythm. 2009;6:836–847. doi: 10.1016/j.hrthm.2009.02.016. [DOI] [PubMed] [Google Scholar]
- 18.Gadaleta M, Giorgio A. A method for ventricular late pote ntials detection using timefrequency representation and wave denoising. ISRN Cardiol. 2012;2012:258759. doi: 10.5402/2012/258769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.European Heart Rhythm Association; Heart Rhythm Society; Zipes DP, Camm AJ, Borggrefe M, Buxton AE, Chaitman B, Fromer M, Gregoratos G, Klein G, Moss AJ, Myerburg RJ, Priori SG, Quinones MA, Roden DM, Silka MJ, Tracy C, Smith SC Jr, Jacobs AK, Adams CD, Antman EM, Anderson JL, Hunt SA, Halperin JL, Nishimura R, Ornato JP, Page RL, Riegel B, Priori SG, Blanc JJ, Budaj A, Camm AJ, Dean V, Deckers JW, Despres C, Dickstein K, Lekakis J, McGregor K, Metra M, Morais J, Osterspey A, Tamargo JL, Zamorano JL American College of Cardiology; American Heart Association Task Force and European Society of Cardiology Committee for Practice Guidelines. ACC/AHA/ESC 2006 guidelines for the management of a) patients with ventricular arrhythmias and the prevention of sudden death: a report of the American College of Cardiology/American Heart Association Task Force and the European Society of Cardiology Committee for Practice Guidelines. J Am Coll Cardiol. 2006;48:e247–e346. doi: 10.1016/j.jacc.2006.07.010. [DOI] [PubMed] [Google Scholar]
- 20.Cuesta-Frau D, Mico-Tormos P, Aboy M, Biagetti MO, Austin D, Quinteiro RA. Enhanced modified moving average analysis of T-wave alternans using a curve matching method: a simulation study. Med Biol Eng Comput. 2009;47:323–331. doi: 10.1007/s11517-008-0415-y. [DOI] [PubMed] [Google Scholar]
- 21.Brenyo A, Zareba W. Prognostic significance of QRS duration and morphology. Cardiol J. 2011;18:8–17. [PubMed] [Google Scholar]
- 22.Schinkel AF, Elhendy A, van Domburg RT, Biagini E, Rizzello V, Veltman CE, Ten Kate GL, Sijbrands EJ, Akkerhuis KM, Geleijnse ML, Ten Cate FJ, Simoons ML, Bax JJ, Poldermans D. Prognostic significance of QRS duration in patients with suspected coronary disease referred for noninvasive evaluation of myocardial ischemia. Am J Cardiol. 2009;104:1490–1493. doi: 10.1016/j.amjcard.2009.07.012. [DOI] [PubMed] [Google Scholar]
- 23.Elhendy A, Hammil SC, Mahoney DW, Pelikka PA. Relation of QRS duration on the surface 12-lead electrocardiogram with mortality in patients with known or suspected coronary artery disease. Am J Cardiol. 2005;96:1490–1493. doi: 10.1016/j.amjcard.2005.06.038. [DOI] [PubMed] [Google Scholar]
- 24.Sagaki K, Ikeda T, Miwa I, Miyakoshi M, Abe A, Tsukada T, Ishiguro H, Mera H, Yusu S, Yoshino H. Time-domain T-wave alternans neasured from Holter electrocardiograms Predicts cardiac mortality in patients with left ventricular dysfunction: a prospective study. Heart Rhythm. 2009;8:332–337. doi: 10.1016/j.hrthm.2008.12.011. [DOI] [PubMed] [Google Scholar]
- 25.Gehi AK, Stein RH, Metz LD, Gomes JA. Microlt T-wave alternans for the risk stratification of ventricular tachyarrhythmic events: a meta-analysis. J Am Coll Cardiol. 2005;46:75–82. doi: 10.1016/j.jacc.2005.03.059. [DOI] [PubMed] [Google Scholar]
- 26.Wita K, Filipecki A, Szydio K, Turski M, Tabor Z, Wróbel W, Elzbieciak M, Lelek M, Bochenek T, Trusz-Gluza M. Prediction of long-term outcome after primary percutaneous coronary intervention for acute anterior myocardial infarction. Kardiol Pol. 2010;68:393–400. [PubMed] [Google Scholar]
- 27.Reed MJ, Robertson CE, Addison PS. Heart rate variability measurements and the prediction of ventricular arrhythmias. QJM. 2005;98:87–95. doi: 10.1093/qjmed/hci018. [DOI] [PubMed] [Google Scholar]
- 28.Karjaleinen JJ, Kiviniemi AM, Hautala AJ, Piira OP, Lepojarvi ES, Peltola MA, Ukkala OH, Hedberg PS, Huikuri HV, Tulppo MP. Determinants and prognostic value of cardiovascular autonomic function in coronary artery disease patients with and without type 2 diabetes. Diabetes Care. 2014;37:286–294. doi: 10.2337/dc13-1072. [DOI] [PubMed] [Google Scholar]
- 29.Song T, Qu XF, Zhang YT, Cao W, Han BH, Li Y, Piao JY, Yin LL, Cheng HD. Usefulness of the heart rate variability complex for predicting cardiac mortality after acute myocardial infarction. BMC Cardiovascular Disorders. 2014;14:59. doi: 10.1186/1471-2261-14-59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Karason K, Molgaard H, Wikstrand J, Sjostrom L. Heart rate variability in obesity and the effect of weight loss. Am J Cardiol. 1999;83:1242–1247. doi: 10.1016/s0002-9149(99)00066-1. [DOI] [PubMed] [Google Scholar]
- 31.Yin DC, Wang ZJ, Gui S, Xie HW, Sun L, Qiu W, Qu XF. Prognostic significance of heart rate turbulence parameters in patients with chronic heart failure. BMC Cardiovasc Disorders. 2014;14:50. doi: 10.1186/1471-2261-14-50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Perkiomaki JS, Jokinen V, Tapanainen J, Airaksinen KE, Huikuri HV. Autonomic markers as predictors of nonfataln acute coronary events after myocardial infarction. Ann Noninvasive Electrocardiol. 2008;13:120–12. doi: 10.1111/j.1542-474X.2008.00211.x. [DOI] [PMC free article] [PubMed] [Google Scholar]