We retrospectively reviewed the clinical files of 326 consecutive patients which were part of a larger study on the effects of cardiac rehabilitation (CR) after AMI. All had suffered a complicated AMI (208 STEMI, 118 NSTEMI) and had been admitted to our CR unit for a period of residential, exercise-based rehabilitation, a median of 13.5 d (95%CI of the mean 15.2-17.3) after the index event. All patients had undergone coronary angiography on initial admission to the Intensive Coronary Care Unit, within 24 h from beginning of AMI symptoms: 194 (94%) STEMI underwent successful PCI of the culprit coronary artery within 6 h of symptoms[13], while only a small percentage of them (14 cases, 6%) could not be revascularized due to unfavourable coronary anatomy and had to be placed on medical therapy; NSTEMI patients underwent coronary angiography within 24 h[14], with successful PCI in 86 cases (73%) and medical therapy in the remaining 32 cases. Of the primary-PCI STEMI patients, 67 (32%) had also been immediately treated on non-culprit coronary lesions; 42 (36%) NSTEMI patients were similarly also immediately treated on non-culprit coronary lesions. Eight (4%) STEMI and 4 (3%) NSTEMI patients received further elective percutaneous revascularization during the initial stay in the Cardiology Department. Overall, at the time of transferral to the CR unit, 121 (58%) STEMI patients and 54 (46%) NSTEMI patients had a complete revascularization, while 87 STEMI and 64 NSTEMI patients were still incompletely revascularized.

Patients were selected for referral to our CR program if they suffered a complicated AMI (cardiogenic shock or pulmonary edema, episode of cardiac arrest, complex ventricular arrhythmias), or if they had incomplete revascularization (because of unfavorable coronary anatomy or technical failure)[15]. Low risk patients were referred as out-patients to a CR program in a different centre and excluded from this study.

Echo- or cardiac MRI- documented intracavitary thrombosis, extreme thinning or intra-myocardial bleeding and/or suspected rupture of the ventricular wall were other exclusion criteria from referral to the CR program.

The following clinical variables were recorded for each patient: Age, gender, body mass index, cardiovascular risk factors, site of infarction, culprit coronary artery vessel, number of diseased coronary artery vessels (defined as presence of diameter stenosis > 50%), history of previous PCI or coronary or valvular surgery, presence of ancillary diseases (renal failure, thyroid dysfunction, known diabetes or abnormal glucose metabolism, pulmonary diseases, history or presence of neoplastic diseases, carotid and peripheral vascular disease) and previous and concurrent drug therapy. During their hospitalisation in CR, all patients without previous diagnosis of diabetes underwent an oral glucose tolerance test to identify subclinical abnormal glucose metabolism. Left ventricular ejection fraction (LVEF) was measured before discharge by 2-D echocardiography, following the Simpson method.

On the day of admittance to CR, all patients underwent 24-h ECG Holter recording, using 3-channel digital recorders (Lifecard CF, Del Mar Reynolds, Irvine, CA, United States), monitoring chest leads CM5, CM3 and modified aVF. Recordings were analyzed using a commercial Holter device system (Del Mar-Reynolds Impresario Holter Analysis System, vers. 2.8.0024; Time-domain HRV Analyzer, vers. 1.0.8.4, CENTUM and Del Mar Reynolds Medical Inc., Irvine, CA, United States; sampling rate of 128 Hz).

After cleansing of arrhythmias and artefacts, the usual time domain HRV variables were assessed including: Standard deviation of all normal-to-normal (NN) intervals (SDNN), standard deviation of all 5-min mean NN intervals (SDANN), root mean square of successive differences (RMSSD), and mean of the standard deviations of all RR intervals of all 5-min segments in the 24 h (SDNN-i). For the purposes of this study, the main variable that was considered in the correlations with other clinical parameters was the SDNN, as it is usually considered a measure of total variance in heart rate; it is also the variable most widely used in previous studies[10] and more strongly associated with mortality compared to other variables[1]. SDNN parameters were analyzed for the entire 24 h period; analysis of day and night hours was also done separately. “Day” was defined as the time period between 06:00 and 22:59 and “night” as the period between 23:00 and 05:59.

Patients with atrial fibrillation, or rhythm disturbances that could interfere with accurate HRV analysis (e.g., frequent ectopic beats, rhythm induced by pacemaker) were excluded from the study, as were patients with inadequate/inaccurate recordings.

The primary outcome measure was the occurrence of cardiac death; the secondary end-point was occurrence of major clinical events (MCE), defined as death (all-cause mortality, cardiac mortality) or readmission for a new AMI, new revascularization, episodes of heart failure or stroke. At the time of follow-up, the clinical status of the patients was assessed by telephonic interviews, performed either by a doctor or a trained team nurse. In case of clinical events, detailed information was obtained from the patient or his/her relatives. Outcomes were analyzed by intention to treat.

Continuous variables were expressed as a mean ± standard deviation (SD) and compared using an unpaired t test; otherwise, variables were expressed with median and interquartile range (IQR) and compared using a Wilcoxon-Mann-Withney test. Categorical variables were expressed as frequencies and percentages and were compared between groups by a χ² test. The relationships between continuous variables were evaluated by Pearson’s correlation coefficient. A Cox regression multivariate analysis was also performed to determine the influence of different factors on HRV parameters, including in the multivariable model only variables with a P value ≤ 0.1 at univariate analysis. HRV variables were initially analyzed as continuous variables; subsequently, HRV variables that showed a significant association with other factors at multivariate analysis were dichotomized and analyzed according to the lowest quartile value. Kaplan-Meier estimates of the distribution of times from baseline to death were computed, and Mantel-Cox Log-Rank analysis was performed to compare the survival curves between the groups. All reported probability values are two-tailed and the significance level was set at 0.05. Statistical analyses were performed using SPSS 18 software package (SPSS Inc, Chicago, IL, United States).

During CR hospitalization, all participants had been fully informed on the procedures they were undergoing; a written consent was obtained from all patients before performance of the medical procedures. The routine diagnostic examinations and follow-up protocol for CR were applied; no special tests or treatments were performed. The research was conducted in accordance with the ethical guidelines of the 1975 Declaration of Helsinki. This study is part of a larger follow-up study on patients admitted to CR; approval of the Provincial Ethics Committee (Provincial Health Directorate, Belluno, Italy) was obtained for the main research.

Main patients’ characteristics are presented in Table 1, together with the medical therapy prescribed at the time of discharge from hospital. Patients with NSTEMI were older than those with STEMI, and presented more often history of hypertension, previous MI and coronary revascularization procedures, and clinical signs of metabolic syndrome. Patients with NSTEMI had greater number of critical coronary stenoses, revascularization was more often incomplete, and such patients presented more often with symptoms of heart failure on initial admission to the coronary care unit.

In the same Table 1, time-domain HRV parameters are also reported. In spite of the above described clinical differences, all main HRV parameters did not show significant differences between STEMI and NSTEMI patients, except for mean heart rate that was lower in NSTEMI cases.

In a total of 52 patients (16% of the whole group; 16% of STEMI and 15% of NSTEMI: χ² = 0.067, P = 0.796), SDNN was < 70 ms, and in 13 (4% of the whole group; STEMI 5.3%, NSTEMI 1.7%: χ² = 2.539, P = 0.111) it was < 50 ms. When subdivided into 4 quartiles according to the value of SDNN, the 81 patients in the lowest quartile presented a mean SDNN of 63.7 ± 11.8 ms (vs a mean of 119.4 ± 35.0 ms of the other quartiles; P < 0.001). Patients with STEMI or NSTEMI were equally distributed between quartiles of SDNN (χ² = 1.536, P = 0.674).

On average, SDNN was higher during night hours than during day-time (P < 0.001), although in 109 patients the day-night variation was insignificant or negative; patients with STEMI or NSTEMI behaved in the same way as regards day vs night SDNN.

Female patients presented on average lower values of SDNN than male patients (97.1 ± 42.2 ms vs 107.9 ± 38.1 ms; P = 0.046), and in 24 out of 65 cases they presented SDNN values in the lower quartile (females 37% vs males 17%; χ² = 6.723, P = 0.010).

SDNN values in the lower quartile were present more frequently in patients older than 65 years (χ² = 4.478, P = 0.034), as well as in patients with known diabetes (χ² = 10.859, P = 0.001) but not in patients with abnormal glucose metabolism detected during the rehabilitation period (χ² = 0.762, P = 0.383). Patients with known diabetes presented also absence of the day/night variation of SDNN (SDNN day: 85.2 ± 32.2 ms in diabetic patients vs 99.5 ± 35.7 ms in non-diabetics, P = 0.002; SDNN night: 85.3 ± 31.0 ms in diabetic patients vs 109.8 ± 42.2 ms in non-diabetic patients, P < 0.001; SDNN day vs night: P = 0.975 in diabetic vs P < 0.001 in non-diabetic patients).

Patients with history of previous MI, or previous CABG or previous PCI were equally distributed among quartiles of SDNN (respectively: χ² = 0.017, P = 0.999; χ² = 1.306, P = 0.728; χ² = 1.729, P = 0.631).

No correlation was found between number of diseased coronary arteries and quartiles of SDNN (whole group: ρ -0.044, P = 0.428; STEMI: ρ 0.001, P = 0.985; NSTEMI: ρ -0.120, P = 0.199). Patients with complete or incomplete coronary revascularization did not differ as regards distribution among quartile of SDNN (χ² = 0.059, P = 0.807).

Similarly, no correlation was found between quartile of SDNN and successful vs unsuccessful primary PCI (whole group: χ² = 0.158, P = 0.691; STEMI: χ² = 0.031, P = 0.861; NSTEMI: χ² = 0.684, P = 0.408). Overall, patients with unsuccessful primary PCI presented markedly reduced variation of SDNN values between day and night (SDNN day 94.7 ± 39.3 ms; SDNN night 102.9 ± 37.9 ms; P = 0.092); by the contrary, such variations persisted in patients with successful primary PCI (SDNN day 96.2 ± 34.6 ms; SDNN night 103.9 ± 41.5 ms; P < 0.001).

In the whole group of patients, a linear correlation was found between LVEF and the values of some HRV parameters (SDNN: ρ 0.168, P = 0.002; SDANN: ρ 0.225, P < 0.001), but not with other HRV parameters such as RMSSD, SDNN Index, Triangular Index and pNN50. Patients with lower LVEF (< 40%) presented more often values of SDNN in the lowest quartile (χ² = 12.668; P < 0.001). Mean heart rate during the 24 h of Holter recording was lower in patients with higher LVEF (ρ -0.310, P < 0.001).

The 37 patients that presented symptoms of heart failure during the acute phase of AMI showed SDNN values in the lower quartile more often than the remaining patients (χ² = 7.884; P = 0.005), with SDNN < 70 ms in 32% of cases (vs 14% of the patients without initial symptoms of heart failure; χ² = 8.457; P = 0.004).

At multivariate Cox regression analysis, a significant correlation with the lowest quartile of SDNN was maintained only by female sex, history of diabetes mellitus and LVEF < 40% (respectively: β = -0.143, P = 0.008; β = 0.146, P = 0.008; β = -0.179, P =0.001).

Ten (3.1%) patients were lost to follow up, which occurred a median of 25.0 mo after the index event (95%CI of the mean: 23.3-28.0).

Of the 316 patients which could be interviewed, MCE occurred in 56 (17.2%) of which 20 deaths (6.3%; 14 cardiac deaths), 9 cases of new non fatal AMI (3.0%), 5 patients with stroke (1.6%) and 17 cases of successful elective revascularization (5.4%: 4 CABG, 13 elective PCI); 21 (6.6%) patients had one or more hospital readmissions for heart failure.

No significant difference in overall incidence of MCE at follow-up was evident between cases with STEMI vs NSTEMI (χ² = 2.166, P = 0.141), although all-cause mortality was higher among NSTEMI patients (14% vs 2%; χ² = 17.863, P < 0.001) as well as cardiac mortality (10% vs 1.5%; χ² = 11.471, P = 0.001). Only after one and a half year of follow-up, the Kaplan-Meier MCE-free survival curves begin to diverge, with NSTEMI patients presenting worse outcomes (Mantel-Cox Log-Rank: χ² = 5.525, P = 0.019); Figure 1A.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Kaplan-Meier major clinical events-free survival curves for: Patients with recent ST-elevation myocardial infarction vs patients with recent non-ST-elevation myocardial infarction (A), patients with preserved left ventricle ejection fraction (> 40%) vs patients with depressed left ventricle ejection fraction (≤ 40%) (B). STEMI: ST-elevation myocardial infarction; NSTEMI: Non-ST-elevation myocardial infarction; MCE: Major clinical events.

Patients with a 3-vessel disease and those who received an incomplete revascularization had a greater incidence of MCE during the period of follow-up (respectively: χ² = 14.369, P = 0.006; and χ² = 6.987, P = 0.008). A significant correlation was evident between all-cause deaths or cardiovascular deaths and number of diseased vessels (respectively: χ² = 19.218, P = 0.001; and χ² = 13.077, P = 0.011). Incomplete revascularization showed a correlation with all-cause deaths (χ² = 4.732, P = 0.030) but not with cardiac deaths (χ² = 1.859, P = 0.173) at follow-up.

The 234 patients that maintained a better preserved myocardial function, as suggested by LVEF > 40%, had lower number of MCE (32) and cardiovascular deaths (3) in the follow-up, than patients with more compromised LVEF, that suffered 24 MCEs and 11 cardiovascular deaths among 82 patients (for MCE: χ² = 10.126, P = 0.001, and OR = 0.383, 95%CI: 0.209-0.701; for cardiovascular deaths: χ² = 21.110, P < 0.001, and OR = 0.084, 95%CI: 0.023-0.309).

At multivariate Cox regression analysis, the variables that showed predictive value for MCE were presence of a three-vessel disease (β = 0.062, P = 0.013), elective PCI (β = 0.250, P = 0.021), known diabetes mellitus (β = 0.124, P = 0.013) and LVEF < 40% (β = -0.114, P = 0.017), while no significant correlation was found with age, sex, number of vessels treated by PCI (β = -0.032, P = 0.229), history of incomplete revascularization, STEMI vs NSTEMI (β = -0.002, P = 0.970), site of infarction or presence of signs of heart failure during initial admission.

In Figure 1B, the Kaplan-Meier events-free survival curves are presented for patients stratified according to LVEF ≤ 40% vs LVEF > 40%; the Mantel-Cox Log-Rank demonstrated statistically significant differences between the curves (χ² = 10.896, P = 0.001).

Even in the subgroup in the lower quartile of SDNN, no difference was found in incidence of MCE (14/76 cases) in comparison with other subgroups (42/240 cases), (χ² = 0.034, P = 0.855); this finding was similar for STEMI and NSTEMI patients (respectively: 10 MCE among 52 STEMI patients with lower quartile of SDNN vs 21/150 of the other quartiles, χ² = 0.813, P = 0.367; and 4 MCE among 24 NSTEMI patients with lower quartile of SDNN vs 21/90 of the other quartiles, χ² = 0.492, P = 0.483).

Patients with negative day-night variations of SDNN presented long-term events similar to patients with positive SDNN day-night variations (χ² = 2.107, P = 0.147).

The Kaplan-Meier MCE-free survival curves were similar between the group of patients with SDNN in the lowest quartile vs the patients of the other SDNN quartiles (Log-Rank χ² = 0.376, P = 0.540; Figure 2A), with no difference for STEMI (Log-Rank χ² = 1.801, P = 0.180) or NSTEMI patients (Log-Rank χ² = 0.373, P = 0.541). In particular, no correlation was found between quartile of SDNN and recurrence of MI during the follow-up period (χ² = 0.489, P = 0.484).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Kaplan-Meier survival curves for patients with lowest quartile of standard deviation of all normal-to-normal intervals vs patients in the other standard deviation of all normal-to-normal intervals quartiles: Major clinical events-free survival curves (A), all-cause mortality curves (B). HRV: Heart rate variability; SDNN: Standard deviation of all normal-to-normal intervals; MCE: Major clinical events.

As regards to death for all causes, 7 out of 20 deaths occurred among patients with lowest quartile of SDNN (χ² = 1.401, P = 0.236); analyzing the whole group of AMI patients, the Kaplan-Meier survival curves for all-cause mortality (Figure 2B) and for cardiac deaths did not evidence significant differences between patients with SDNN in the lowest quartile and other quartiles of SDNN (Mantel-Cox log rank, respectively: χ² = 2.207, P = 0.137; and χ² = 0.399, P = 0.527). After separating STEMI from NSTEMI patients, different survival curves have been observed only for all-cause mortality (Figure 3), but not for cardiac mortality: STEMI patients with SDNN in the lowest quartile presented 3 out of 4 all-cause deaths (Mantel-Cox log rank χ² = 6.591, P = 0.010) and 2 out of 3 cardiac deaths (Log-Rank χ² = 3.685, P = 0.055), while among NSTEMI patients no difference in the survival curves was observed between patients with SDNN in the lowest quartile vs the other quartiles (all-cause mortality: Log-Rank χ² = 0.195, P = 0.659; cardiac deaths: Log rank χ² = 0.040, P = 0.842).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Kaplan-Meier all-cause mortality curves for ST-elevation myocardial infarction patients (A) and non-ST-elevation myocardial infarction patients (B) divided between cases with lowest quartile of standard deviation of all normal-to-normal intervals vs cases in the other standard deviation of all normal-to-normal intervals quartiles. HRV: Heart rate variability; SDNN: Standard deviation of all normal-to-normal intervals; STEMI: ST-elevation myocardial infarction; NSTEMI: Non-ST-elevation myocardial infarction.

Analysis of the long term outcomes of patients with low values of RMSSD gave similar results as those described for low SDNN values: Kaplan-Meier events free survival did not differ between patients with RSDNN in lowest quartile vs the other quartiles, both for incidence of MCE (χ² = 0.849, P = 0.357) and all-cause or cardiac mortality (respectively: χ² = 0.060, P = 0.806; and χ² = 0.245, P = 0.621).

Patients’ main clinical parameters (age, sex, time from STEMI to CR, site of infarction, number of diseased vessels, fasting glucose, HbA1c, haemoglobin level at admission, LVEF) and HRV values (SDNN, RMSSD, SDNNi, SDANN) have been compared between the 10 cases lost to follow-up and the 314 patients that completed the study. Patients lost to follow-up presented more elevated average values of HbA1c (7.2% ± 2.0% vs 6.4% ± 1.1%; P = 0.022) but not of fasting glucose (109.5 ± 37.9 mg/dL vs 95.2 ± 22.5 mg/dL; P = 0.085) at admission. All the other clinical and HRV parameters were not significantly different in comparison to followed-up patients.

Criteria of exclusion from this study included presence of atrial fibrillation or flutter, a rhythm induced by the pacemaker, or inadequate Holter recordings. Although such patients may have had significant autonomic dysfunction, HRV could not be measured in them. Consequently it is not known if a depressed HRV could have had any long-term impact on their prognosis.

The degrees of autonomic derangement presented by our patients, that may have been conditioned by the kind of complications suffered during the initial phase of their disease, may possibly not be generalized to the other complicated or uncomplicated STEMI patients.

HRV was performed only in time domain; no analysis was available in the frequency domain. However, it is already known that time-domain HRV indices measured over a 24-h period are well correlated with frequency domain indices in coronary artery disease patients[34]. Among time-domain parameters, SDNN was identified as having the same high predictive value as the frequency-domain LF amplitude in post-AMI patients[16].

In conclusion, in our group of patients with a recent complicated MI, abnormal autonomic parameters (evidenced by low HRV) have been found with a prevalence that was similar for STEMI and NSTEMI cases, and substantially unchanged in comparison to what reported in the pre-primary-PCI era.

Long-term outcomes in PCI-treated STEMI patients were more favorable than in old cohorts of patients. They did not correlate with the level of depression of HRV parameters recorded in the subacute phase of the disease, both in STEMI and in NSTEMI patients. Traditional HRV parameters seem to have lost their prognostic significance in the present era of immediate coronary reperfusion.

After an acute myocardial infarction (AMI), overactivation of the sympathetic component and relative suppression of the parasympathetic component of the autonomic nervous system occur, leading to a marked imbalance of cardiac autonomic regulation. Such sympatho-vagal imbalance is usually considered to be linked to increased short and long-term mortality. Clinically, cardiac autonomic modulation can be evaluated observing the fluctuations of instantaneous heart rate (intervals between normal RR complexes) during variable periods of time (often 24 h). The most simple and widely used parameter for evaluation of heart rate variability (HRV) is the SDNN (standard deviation of all normal RR intervals), that gives an overall estimate of cardiac autonomic control. In the last 30 years, various studies confirmed that depressed values of SDNN indicate poor prognosis after an AMI. Unfortunately, most of these studies on the prognostic value of depressed HRV in post-AMI patients have been conducted before the primary-PCI era: So, their results may not be easily applied to present day patients, in which early reperfusion (that may lead to salvage of cardiac muscle and preservation of better heart function), multiple pharmacological therapies and cardiac rehabilitation may impact on autonomic balance, quality of life and long-term survival.

A few small-scale studies have been conducted so far on AMI patients treated by primary PCI, investigating if depressed HRV maintains a correlation with long-term prognosis. Such studies did not reach clear conclusions. With the present work, the authors aimed at contributing and clarifying if parameters of HRV (in particular SDNN) could still be considered reliable prognostic indicators in primary-PCI-treated AMI patients.

While in the past depressed HRV had been identified as a reliable marker of poor prognosis after an AMI, this paper demonstrates that nowadays in AMI patients treated by early revascularization HRV has lost its prognostic significance. In addition, the paper is the first to report HRV parameters separately for patients with ST-elevation and Non-ST-elevation myocardial infarction.

The evaluation of simple HRV parameters does not seem to hold any more prognostic significance in AMI patients treated by early revascularization. Further studies are needed in these patients to identify reliable long-term prognostic indicators.

HRV refers to the evaluation of the oscillations in the intervals between consecutive heart beats. They are determined by cyclic variations of sympathetic and parasympathetic autonomic influences on cardiac pace-maker cells, modulated by central and peripheral mechanisms (respiratory and vasomotor centres, fluctuations in arterial pressure, humoral factors). After an AMI, HRV usually presents various degrees of depression, linked to a marked increase of the sympathetic activity (probably due to abnormal geometry of the beating heart and consequent distortion of autonomic nervous endings). The most simple method of evaluating HRV is by statistical analysis of the time intervals between consecutive normal RR beats (“time domain” method); the standard deviation of these intervals (SDNN) has been used in our study. SDNN reflects all the cyclic components responsible for variability in the period of recording; for this reason, it has been used in various studies for assessment of risk after AMI.

The authros analysed the potential prognosis of HRV in patients treated with primary PCI.