Improved Stratification of Autonomic Regulation for risk prediction in post-infarction patients with preserved left ventricular function (ISAR-Risk).

AIMS: To investigate the combination of heart rate turbulence (HRT) and deceleration capacity (DC) as risk predictors in post-infarction patients with left ventricular ejection fraction (LVEF) > 30%. METHODS AND
RESULTS: We enrolled 2343 consecutive survivors of acute myocardial infarction (MI) (<76 years) in sinus rhythm. HRT and DC were obtained from 24 h Holter recordings. Patients with both abnormal HRT (slope < or = 2.5 ms/RR and onset > or = 0%) and abnormal DC (< or =4.5 ms) were considered suffering from severe autonomic failure (SAF) and prospectively classified as high risk. Primary and secondary endpoints were all-cause, cardiac, and sudden cardiac mortality within the first 5 years of follow-up. During follow-up, 181 patients died; 39 deaths occurred in 120 patients with LVEF < or = 30%, and 142 in 2223 patients with LVEF>30% (cumulative 5-year mortality rates of 37.9% and 7.8%, respectively). Among patients with LVEF > 30%, SAF identified another high-risk group of 117 patients with 37 deaths (cumulative 5-year mortality rates of 38.6% and 6.1%, respectively). Merging both high-risk groups (i.e. LVEF < or = 30% and/or SAF) doubled the sensitivity of mortality prediction compared with LVEF < or = 30% alone (21.1% vs. 42.1%, P < 0.001) while preserving 5-year mortality rate (38.2%).
CONCLUSION: In post-MI patients with LVEF>30%, SAF identifies a high-risk group equivalent in size and mortality risk to patients with LVEF < or = 30%.

Introduction

Reduction of mortality by implantable cardioverter-defibrillators (ICDs) has been well documented by several trials in cardiac patients with compromised left ventricular function assessed, as a rule, by low left ventricular ejection fraction (LVEF).[1-5] Although the results of these studies have now been projected into standard clinical practice, a substantial problem remains because of the low sensitivity of the criterion of reduced LVEF.[6-8] Many cardiac patients who die from sudden cardiac death, likely preventable by prophylactic ICD implantation, do not have left ventricular performance particularly compromised.[6,8-10] According to current guidelines, these patients are considered as low risk patients and are not protected by intensive prophylactic treatment.[11,12] At present, methods for identification of high-risk patients without compromised LVEF are lacking. Consequently, prophylactic ICD implantation (or other prophylactic therapy) has never been tested in these patients. Hence, adequately potent risk stratification methods need to be first demonstrated in patients with preserved LVEF before any risk reduction prophylaxis can be attempted.

To address this need, we have conducted a large cohort follow-up study to investigate whether the assessment of abnormalities in cardiac autonomic function makes risk stratification feasible among patients without seriously compromised ventricular performance. To characterize abnormalities in cardiac autonomic function, we defined severe autonomic failure (SAF) as a combination of severely impaired baroreflex function with abnormal autonomic tone. For this purpose, we used Heart Rate Turbulence (HRT)[13] and cardiac Deceleration Capacity (DC).[14] Although the risk prediction power of these parameters was previously documented in independent patient populations, the ways of combining them have not been studied. In particular, previous studies suggested that DC might be more useful in identification of low-risk patients while the power of HRT is mainly in the selection of high-risk cases. This study therefore investigated the definition of high risk as the combination of mildly abnormal DC[14] (to exclude low-risk patients) with highly abnormal HRT[13] (to select high-risk patients).

Methods

Study design, setting, and participants

Between January 1996 and March 2005, patients were enrolled into a prospective cohort study at two centres, namely the German Heart Centre and the Klinikum Rechts der Isar, both in Munich, Germany (see Figure  for patient numbers). Eligible patients suffered from acute myocardial infarction (MI) no more than 4 weeks before enrolment, were not older than 75 years, presented in sinus rhythm, and did not meet the criteria for secondary therapy by implantable cardioverter-defibrillator (ICD)[11] before hospital discharge for the index MI. Of the 2343 study patients, 1455 were included in earlier investigations[7,14] and are reported here with extended follow-up (5 vs. 2 years). At the same time, the number of primary endpoints increased from 70 to 181.

Flow chart of patient selection.

MI was diagnosed if a patient had at least two of the following findings: chest pain for ≥20 min, creatine kinase-MB above the doubled upper normal limit of our laboratory, and ST-segment elevation of ≥0.1 mV in two or more limb leads or ≥0.2 mV in two or more contiguous precordial leads at the time of admission. Diabetes mellitus was considered present either if a patient was diagnosed and was receiving treatment (diet, tablets, or insulin) or if a blood glucose concentration of ≥11 mmol/L was found in repeated samples. The patients were followed up for a median of 4.9 years (2.8–5.0 years). The local Ethics Committee approved the collection of data and analysis of Holter recordings. Since the data obtained were non-invasive and did not exceed usual clinical management of the patients, the local Ethics Committee decided that signed informed consent was not needed. However, we did obtain oral informed consent.

Assessment of autonomic markers

Each patient of the study population (n = 2343) underwent 24 h Holter recording. These recordings were obtained at a median of 8 days after the index infarction (interquartile range 5–11 days). All recordings were routinely processed using standard commercial equipment (Oxford Excel Holter system, Oxford instruments; Pathfinder 700, Reynolds Medical; and Mortara Holter system, Mortara Instrument) to obtain the sequence of individual RR intervals together with the distinction of sinus rhythm beats and ventricular premature complexes. Using these sequences, HRT and DC were calculated using previously published technologies[13-15] that are presently available as simple add-ons to standard modern Holter systems.

Briefly, HRT quantifies the physiological short-term oscillation of cardiac cycle lengths that follows spontaneous ventricular premature complexes.[16] HRT consists typically of a brief heart rate acceleration followed by a gradual heart rate deceleration. These HRT phases are quantified by two numerical descriptors, HRT onset and HRT slope. DC is an integral measure of all deceleration-related oscillation observed over 24 h. Computation of DC is based on a novel signal processing algorithm capable of extracting periodic components out of non-stationary biological signals.

Consistent with previous reports, HRT onset was considered abnormal if ≥0%,[13] HRT slope if ≤2.5 ms per RR interval,[13] and DC if ≤4.5 ms.[14]

Patients who exhibited these pathologies in both components of HRT as well as in DC were considered to suffer from seriously abnormal cardiac autonomic modulations. The term SAF was used to describe this abnormality.

Left ventricular performance

In each patient, LVEF was assessed by left ventricular angiography (n = 1686) or biplane echocardiography (n = 657) during the second week after the index infarction. In agreement with previous reports on the utilization of prophylactic ICDs in patients with compromised left ventricular performance,[3,11] reduced LVEF was defined if ≤30%.

Endpoints of the study

Last planned follow-up assessment was 1/2007. In cases of death, the reason for death was verified from hospital and autopsy records, and from either the primary physician or those witnessing the death. An independent endpoint committee determined the mode of death. Deaths were categorized as cardiac and non-cardiac. Cardiac deaths were further categorized as sudden and non-sudden. Cardiac death was defined as sudden if it was[1] a witnessed death occurring within 60 min of the onset of new symptoms, unless there was an obvious non-cardiac cause,[2] an unwitnessed death within 24 h in the absence of pre-existing progressive circulatory failure or other causes of death, or[3] death during attempted resuscitation. The primary endpoint of the study was all-cause mortality within the first 5 years of follow-up; secondary endpoints were cardiac and sudden cardiac death also at 5 years of follow-up. The cut-off of 5 years was derived from the expected ICD longevity.

High-risk groups

Some patients who suffered from SAF also had LVEF ≤30% which made them eligible for prophylactic ICD implantation based on the presently available criteria. Thus, to investigate what risk stratification by SAF offers in addition to reduced LVEF, we considered and compared two distinct high-risk groups, that is (i) patients with LVEF ≤30% and (ii) patients with LVEF >30% and SAF. We also considered the group of all patients with SAF, and a combination of these high-risk groups, that is patients with either LVEF ≤30% or LVEF >30% and SAF.

Statistics

Continuous variables are presented as median and inter-quartile range; qualitative data are expressed as absolute numbers and percentages. Comparison of variables between high-risk groups (LVEF ≤ 30% or SAF and LVEF > 30%) was performed with either a linear or logistic model in order to adjust for differences in age and sex. Survival curves for total mortality were estimated by the Kaplan–Meier method[17] and compared using the two-sided log-rank test.[18] The influence of SAF on total mortality was estimated with the Cox proportional-hazards model adjusted for established risk factors, including age ≥65years, presence of diabetes mellitus, history of previous MI, arrhythmia signs on Holter (≥10 VPCs/h and/or non-sustained ventricular tachycardia on Holter) and sex. Mortality rates of both high risks groups (LVEF ≤ 30% and SAF and LVEF > 30%) were also compared by the Cox-model adjusted for age and sex. The proportional hazard assumption of the various parameters was investigated by using Schoenfeld residuals. Hazard ratios together with 95% confidence limits were calculated. Secondary endpoints (cardiac death and sudden death) were analysed with competing risk models[19] and compared with the procedure proposed by Gray.[20] Differences were considered statistically significant, if P < 0.05. Sensitivity and specificity were calculated for the prediction of all-cause, cardiac, and sudden cardiac death in (i) patients with LVEF ≤30%, (ii) patients with SAF, (iii) patients with SAF and LVEF >30%, and (iv) patients with either LVEF ≤ 30% or >30% and SAF (i.e. LVEF ≤ 30% and/or SAF).[21]

Statement of responsibility

The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.

Results

The clinical characteristics of recruited patients are shown in Table . Eighteen patients (i.e. 0.8% of the total population) were lost to follow-up. They were censored at the date of latest contact.

During the first 5 years of follow-up, 181 patients died (Table ). Only 39 (22%) of these deaths occurred in patients with LVEF ≤30% (n = 120; 5.1% of the total population) whereas 142 (78%) occurred in patients with LVEF > 30% (n = 2223; 94.9% of the total population). Cumulative mortality curves of patients stratified by LVEF ≤30% are shown in the left panels of Figure . After 5 years, cumulative mortality rates of patients with LVEF ≤30% and >30% were 37.9% and 7.8%, respectively (χ2 = 110.1, P < 0.001).

Kaplan–Meier curves of all-cause mortality, cardiac mortality and sudden cardiac death for all patients (n = 2343) stratified by left ventricular ejection fraction (LVEF; left panels), for all patients (n = 2343) stratified by presence of severe autonomic failure (SAF; middle panels) and for patients with LVEF >30% (n = 2223) also stratified by presence of SAF (right panels). The number of patients of the individual groups involved in the analyses at 0, 1, 2, 3, 4, and 5 years are shown under the graphs (the same numbers apply to all three end-points). The top and the bottom row corresponds to the upper and bottom Kaplan–Meier curve respectively.

Among patients with LVEF >30%, SAF identified 117 cases (i.e. 5.0% of the total population) as high-risk patients, out of whom 37 died during the first 5 years of follow-up. Cumulative mortality curves of these patients are shown in the right panels of Figure . After 5 years of follow-up, cumulative mortality rates of patients with and without SAF were 38.6% and 6.1%, respectively (χ2 = 134, P < 0.001). If one of the SAF components, i.e. either DC or HRT, was abnormal, the corresponding cumulative mortality rate was 9.5%; if neither component was abnormal, the cumulative mortality rate was 4.0%. On multivariable analysis (adjusted for age, presence of diabetes mellitus, history of previous MI, arrhythmia on Holter and sex), SAF yielded a hazard ratio of 4.6 [95% confidence interval (CI) 3.1–7.0] (P < 0.001).

There was no significant difference between the mortality rates of both high-risk groups adjusted for age and sex (SAF and LVEF > 30% vs. LVEF ≤ 30%; hazard ratio of 0.83 (95% CI 0.52–1.32). Similar figures were observed for cardiac death (relative risk = 0.88, 95% CI 0.51–1.52) and sudden cardiac death (relative risk = 0.93, 95% CI 0.42–2.08). Merging of the two high-risk groups, i.e. considering patients with either LVEF ≤ 30% or LVEF > 30% and SAF, led to an increase in sensitivity (Table ), while maintaining a high 5-year mortality rate at 38.2%.

Compared to patients with LVEF ≤ 30%, patients with SAF and LVEF > 30% were older (68 vs. 62 years) and more often of female sex (32% vs. 12%) (Table ). After adjustment for age and sex statistically significant differences between both high-risk groups were observed for prescription of diuretics (61% vs. 83%) and creatine kinase (1.257 vs. 2.269 U/L).

Discussion

In this population, SAF identifies high-risk post-MI patients among those with LVEF >30%. The size and risk-profile of the group of patients with LVEF>30% and SAF are practically identical to those of the high-risk group defined by reduced LVEF. This is not only true for total mortality but also for cardiac and sudden cardiac mortality.

Although risk stratification based purely on LVEF identified only approximately one-quarter of the patients who died during follow-up, combination of LVEF ≤ 30% with LVEF > 30% and SAF identified approximately one-half of them. At the same time, specificity was unaffected. It is thus likely that prophylactic ICD treatment in patients with LVEF > 30% and SAF would be no less efficacious than in patients with LVEF ≤ 30%. ICD therapy might actually be even more effective in patients with LVEF > 30% and SAF since these patients are less prone to deaths from pump failure which restricts the benefit offered by ICD.[22,23]

There are good reasons to combine DC with HRT when defining SAF. Although both these parameters reflect cardiac autonomic modulations, HRT expresses reflex reaction to the cardiac rhythm to distinct disturbance by isolated ventricular premature beat.[13,24,25] On the contrary, DC integrates all the regulatory processes that slow heart rate on a beat-to-beat basis thus expressing an overall status of the autonomic, predominantly vagal balance.[14,15] Both these facets of autonomic assessment are reasonably independent each of the other which facilitates making their combination such a strong risk predictor. (Note the difference between cumulative mortality rates of 38.6% and 9.5% when both HRT and DC were abnormal, and when only one factor was abnormal, respectively.).

A number of other risk factors have been reported to differentiate between high- and low-risk survivors of MI independently of LVEF (e.g. heart-rate profile of post-exercise recovery,[26,27] nonlinear dynamics, detrended fluctuation analysis,[28] T wave alternans,[29-32] electrophysiological testing,[33,34] etc.). With some of these factors, however, the reported sensitivities and specificities among patients with preserved LVEF are rather low. Other reported risk factors appear more viable. For instance, similar data to our findings were reported with T wave alternans (although with a smaller population and fewer events).[35] Possibly, combinations of T wave alternans with HRT and DC would improve risk prediction in this population even further.[36]

Moreover, in contrast to the other techniques,[37] SAF assessment requires only a Holter recording to be obtained in post-infarction patients and is thus not limited only to patients who tolerate a specific provocative investigation (e.g. patients who can exercise for the assessment of T wave alternans and/or post-exercise heart rate recovery).

The clinical efficacy of ICD implantation in patients with LVEF > 30% and SAF needs to be prospectively tested. Since the incidence and mortality rates are equal to those in patients with LVEF ≤ 30% in whom successful ICD trials exist,[1-3] such prospective studies appear fully viable.

The cumulative mortality of patients with SAF and LVEF > 30% constantly increased over the observation period and did not saturate. In other words, patients with LVEF > 30% and SAF continue to be at increased risk even if surviving the first years after MI. Thus, these patients would probably benefit from ICD prophylaxis even late after MI.

Intentionally, we did not compare primarily patients with LVEF ≤ 30% with all patients with SAF. Those patients who have both compromised LVEF and SAF are eligible for prophylactic ICD implantation according to present guidelines and we are not suggesting that SAF should compete with these guidelines. (In our population, we have not found any practically viable criterion that would allow excluding patients with LVEF≤30% from ICD prophylaxis—data not presented here.) Rather, we show that SAF identifies a group with considerable high risk among those patients who are not covered by present guidelines on primary ICD prophylaxis.

The strong predictive power of SAF was confirmed by further analyses the details of which are beyond the scope of this text. Among others, while a number of conventional measures of heart-rate variability (HRV) were significant predictors of mortality, SAF yielded the strongest relative risk both uni- and multivariably independent of the cut-off values of conventional HRV measures. Also, the SAF components, i.e. HRT and DC were significant and independent multivariable predictors.

Several limitations of our study need to be recognized. We enrolled patients younger than 76 years. Therefore, our findings cannot be extrapolated to older post-infarction patients. Moreover, SAF assessment is restricted to patients presenting in sinus rhythm and thus not applicable for instance to patients with atrial fibrillation which is known to be associated with increased mortality risk including risk of sudden death.[38] Holter recordings were obtained at median of 8 days after index infarction which might trigger a question of reproducibility. Nevertheless, previous publications on HRT[13,39,40] and DC[14] showed that the power of SAF components was maintained also in populations that collected post-infarction Holter recordings at different times. Our enrolment spanned almost 10 years during which some advances in acute treatment have been made. More detailed analysis of our data (not presented here) does not show any differences in these findings between earlier and later parts of the population. Since it is difficult to distinguish between life saving ICD therapy and other ICD interventions, we did not include ICD therapy into study endpoints. As seen in Table , the numbers of patients with effective ICD therapy were small (only 4% of patients with LVEF ≤30% experienced ICD therapy while surviving the follow-up) and including the ICD interventions into endpoints (analysis not presented here) did not change the principal results. The threshold of LVEF ≤ 30% and >30% was prospectively selected when designing this study. However, the results are not dependent on this selection. When re-analysing the data for high-risk groups of LVEF ≤35% vs. >35% and SAF, the results were practically identical. A fraction of the data in our study was included in earlier investigations.[7,14] However, the present study is based on much larger data collection with substantially increased follow-up period in which the number of endpoints was more than doubled. To test the consistency of our data, we analysed the previously used and the new data separately and confirmed that the results are fully independent of the original publications (analysis not presented here). Finally, although cohort studies of this kind characterize risk groups, they cannot replace intervention trials.

In spite of these limitations, the results of our cohort study are sufficiently robust to indicate that the autonomic markers should be strongly considered in future cardiac prophylaxis. Assessment of SAF after acute MI is presently readily available.

Supplementary material

Supplementary material is available at .

Funding

Grants from the Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie (13 N/7073/7), the Kommission für Klinische Forschung, and the Deutsche Forschungsgemeinschaft (SFB 386). Funding to Pay the Open Access Charge was provided by GS.

Conflict of interest: G.S. holds patents on Heart Rate Turbulence and Deceleration Capacity; both licensed to GE Healthcare and Medtronic. None of the other authors have any conflict of interest.

Table 1

Characteristics of the study population (n = 2343)

Characteristic
Age (years) (median, IQR)	59 (51–67)
Female sex [n (%)]	456 (20)
Diabetes mellitus [n (%)]	412(18)
History of previous MI [n (%)]	261 (11)
CKmax (U/L) (median, IQR)	1189 (598–2460)
Creatinine (µmol/L) (median, IQR)	1.1 (1.0–1.3)
LVEF (%)	55 (45–62)
VPC (count/h) (median, IQR)	0.5 (0.1–3.5)
Nonsustained VT [n (%)]	156 (7)
PCI [n (%)]	2159 (92)
Thrombolysis [n (%)]	62 (3)
CABG [n (%)]	39 (2)
Aspirin [n (%)]	2298 (98)
β-Blocker [n (%)]	2210 (94)
ACE inhibitors [n (%)]	2132 (91)
Statins [n (%)]	2057 (88)
Diuretics [n (%)]	926 (40)

ACE, angiotensin-converting enzyme; CABG, coronary artery bypass graft; CK, creatine kinase; IQR, inter-quartile range; LVEF, left ventricular ejection fraction; MI, myocardial infarction; PCI, percutaneous coronary intervention; VPC, ventricular premature complex; VT, ventricular tachycardia.

Table 2

Endpoints at 5 years

Group	Total	All-cause deaths	Cardiac deaths	Sudden cardiac deaths	Death not specified
LVEF ≤ 30%	120	39	29	12	3
SAF	152	52	40	16	3
LVEF > 30% and SAF	117	37	27	14	3
Others (=LVEF > 30% and no SAF)	2106	105	48	29	8
Total	2343	181	104	55	14

LVEF left ventricular ejection fraction; SAF severe autonomic failure.

Table 3

Positive and negative predictive accuracies, sensitivities and specificities for prediction of all-cause mortality, cardiac mortality, and sudden cardiac death by high-risk groups

	LVEF ≤ 30%	SAF	SAF (in LVEF > 30%)	LVEF ≤ 30% or LVEF > 30% and SAF
Total count	120	152	117	237
Prediction of all-cause mortality at 5 years
All-cause deaths	39	52	37	76
Positive predictive accuracy (%)	37.9	39.9	38.6	38.2
Negative predictive accuracy (%)	92.2	92.8	93.9	93.9
Sensitivity (%)	21.1	28.2	26.0	42.1
Specificity (%)	96.5	95.7	96.5	93.1
Prediction of cardiac mortality at 5 years
Cardiac deaths	29	40	27	56
Positive predictive accuracy (%)	27.6	29.9	27.1	27.3
Negative predictive accuracy (%)	95.9	96.4	97.2	97.2
Sensitivity (%)	27.1	37.2	34.8	52.9
Specificity (%)	96.1	95.2	96.0	92.2
Prediction of sudden cardiac death at 5 years
Sudden cardiac deaths	12	16	14	26
Positive predictive accuracy (%)	12.0	11.6	13.4	12.8
Negative predictive accuracy (%)	97.7	97.8	98.3	98.3
Sensitivity (%)	22.1	27.1	30.7	46.6
Specificity (%)	95.4	94.1	95.3	90.9

LVEF, left ventricular ejection fraction; SAF, severe autonomic failure.

Table 4

Comparison of high-risk groups

Characteristic	LVEF ≤ 30%	SAF and LVEF > 30%	P-value
n	120	117
LVEF (%) (median, IQR)	25 (21–28)	46 (39–56)	<0.0001
Age (years) (median, IQR)	62 (55–68)	68 (60–72)	<0.0001
Female sex [n (%)]	14 (12)	37 (32)	<0.0001
P-values adjusted for age and sex
Diabetes mellitus [n (%)]	26 (22)	37 (32)	0.349
History of previous MI [n (%)]	34 (28)	20 (17)	0.029
CKmax (U/L) (median, IQR)	2269 (952–5139)	1257 (510–2455)	<0.0001
Creatinine (mg/dL) (median, IQR)	1.3 (1.1–1.5)	1.2 (1.0–1.4)	0.316
VPC (count/h) (median, IQR)	5 (1–27)	7 (2–40)	0.700
Non-sustained VT [n (%)]	22 (18)	18 (15)	0.748
Mean HR (bpm) (median, IQR)	73 (66–82)	74 (68–81)	0.140
SDNN (ms) (median, IQR)	66 (53–85)	68 (55–89)	0.606
PCI [n (%)]	110 (92)	103 (88)	0.523
Thrombolysis [n (%)]	1 (1)	2 (2)	0.685
CABG [n (%)]	5 (4)	3 (3)	0.445
Aspirin [n (%)]	113 (94)	113 (97)	0.503
β-Blocker [n (%)]	109 (91)	102 (87)	0.475
ACE inhibitors [n (%)]	109 (91)	108 (92)	0.694
Statins [n (%)]	93 (76)	91 (78)	0.751
Diuretics [n (%)]	100 (83)	71 (61)	<0.0001
ICD implantation [n (%)]	20 (17)	6 (5)	0.013
ICD therapy in patients who survived follow-up [n (%)]	5 (4)	1(1)	0.166

ACE, angiotensin-converting enzyme; CABG, coronary artery bypass graft; CK, creatine kinase; HR, heart rate; IQR, inter-quartile range; LVEF, left ventricular ejection fraction; MI, myocardial infarction; PCI, percutaneous coronary intervention; VPC, ventricular premature complex; VT, ventricular tachycardia.