| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Third Department of Internal Medicine, Nagoya City University Medical School, Nagoya, Japan.
Correspondence to: Jun-ichiro Hayano, MD, Third Department of Internal Medicine, Nagoya City University Medical School, 1 Kawasumi Mizuho-cho Mizuho-ku, Nagoya 467-8601, Japan; e-mail: hayano{at}med.nagoya-cu.ac.jp
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Design and setting: Retrospective cohort study in medium-sized university general hospital.
Measurements and results: In a
cohort of 250 patients with CAD who were undergoing elective coronary
angiography, we analyzed HRV during standardized HUTT under paced
breathing with discontinuation of treatment with all medications.
During a subsequent mean follow-up period of 99 months, there were 13
cardiac deaths and 12 noncardiac deaths. Cox regression analysis
adjusted for cardiovascular risks revealed that increased postural
change (supine to upright) in the power of low-frequency component (LF)
power predicted an increased risk for cardiac death (relative risk
[per 1-ln ms2 increment], 4.36; 95% confidence interval,
1.64 to 11.6), while neither the high-frequency component nor its
response to HUTT predicted any form of death. When the patients
were trichotomized by the level of postural LF change (large drop,
- 0.6 ln[ms2]; small drop and rise, > 0
ln[ms2]), the three groups did not differ in terms of
clinical features or CAD severity at baseline or coronary interventions
during the follow-up period; however, the 8-year cardiac mortality
rates were 0%, 6%, and 12%, respectively (p = 0.008 [log rank
test]). Additionally, the difference was enhanced when analyzed
excluding 64 patients who had been treated with a ß-blocker during
the follow-up period (0%, 7%, and 15%, respectively; p = 0.006
[log rank test]).
Conclusions: The postural response of HRV predicts the risk for death in patients with CAD. Postural LF increase (LF rise), in particular, is an independent risk factor for cardiac death.
Key Words: autonomic nervous system • coronary disease • heart rate variability • mortality • posture • spectrum analysis
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
In the present study, we aimed at examining whether HRV assessment under well-controlled conditions predicts the risk for death in stable patients with CAD during a long-term follow-up. We investigated 8-year survival data for a cohort of patients who had undergone a standardized autonomic function test by HRV analysis during head-up tilt testing (HUTT) when they had been admitted to the hospital for elective coronary angiography. The HUTT was performed after the discontinuation of treatment with all medications and with the standardization of conditions, including the time of day, the temperature of the laboratory, and the food and beverage intake of the patients from the previous night. Also, HRV and the patients’ responses to HUTT were assessed under paced breathing. There is much evidence supporting that the power of the high-frequency component (HF) measured under these conditions provides an accurate assessment of cardiac vagal function4 7 and that the postural increase in the power of the low-frequency component (LF) reflects, at least in part, postural ß-adrenergic sympathetic activation.8 9 10 These features of the present study allowed us to evaluate the independent associations between the intrinsic autonomic functions and long-term survival in stable patients with CAD.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
At baseline, the mean (± SD) age of the patients was 57 ± 9 years (age range, 40 to 75 years). Ninety-eight patients (39%) had a history of myocardial infarction, but none had a history of coronary angioplasty or bypass graft surgery, 52 (21%) had non-insulin-dependent diabetes mellitus, and 136 (54%) were smokers. The mean left-ventricular ejection fraction was 64 ± 11%. The protocol of this study was approved by the Nagoya City University Medical School Institutional Review Board.
Baseline Measurements and Follow-up
Baseline clinical characteristics including medical history,
lifestyle, plasma lipid concentration, and glucose concentration after
a 14-h fast were available for all patients from their medical records
during the index hospital admission. The number of patients with
coronary artery stenosis (which was defined as a luminal narrowing of
75% in a major coronary artery or branch) and the values for left
ventricular ejection fraction and end-diastolic pressure were obtained
from the cardiac catheterization data recorded at baseline. For the
purposes of the autonomic assessment, treatment with all medications
but sublingual nitroglycerin had been discontinued for at least 1 week
before hospital admission. None of the patients had been receiving a
long-acting ß-blocker during the preceding month. Additionally, no
patients developed any symptoms of syncope or presyncope during the
HUTT.
After discharge from the hospital, all patients were observed and medically treated by cardiologists in the Nagoya City University Hospital or by their family doctors. In 1999, the medical records were reviewed for cardiovascular events, intervention therapies, and medications. Furthermore, after obtaining written informed consent, the patients or their families were interviewed by telephone about cardiovascular and noncardiovascular events by a cardiologist who was blinded to the baseline HRV measurements. The interview was not possible in 7 of 250 patients (2.8%) for whom the follow-up period terminated at the last date on their available medical records. The mean follow-up period duration for the seven patients was 29 ± 9 months (range, 14 to 47 months).
We used only death as the end point of the present study. The causes of death were classified as follows: (1) cardiac death (ie, death from myocardial infarction, heart failure, fatal arrhythmia, or sudden cardiac death, which was defined as death within 1 h after the onset of a new symptom); and (2) noncardiac death.
Control Subjects
To estimate the normal ranges for stationary-state levels of HRV
and the responses of patients to the HUTT, we studied an age-matched
and gender-matched group of 90 healthy subjects (65 men and 25 women;
mean age, 58 ± 10 years; range, 45 to 70 years). The subjects had
been screened for latent disorders through medical history, physical
examinations, laboratory examinations, and ECG. Elderly subjects
(ie, those subjects 65 years old) also had been screened
for occult cardiovascular disease by exercise tolerance testing. None
of the subjects had received any medications for > 2 weeks preceding
the study, but 32 subjects (35%) were smokers. These subjects
underwent HUTT with the same protocol as that used in the patients,
although they did not undergo coronary angiography.
HUTT and Analysis of HRV
The protocol of HUTT and the method of HRV analysis at baseline
have been reported previously.7
11
Briefly, patients were
asked to avoid cigarette smoking and beverages containing caffeine
after 8:00 PM the day preceding the HUTT. The test was
performed in an air-conditioned room (temperature range, 23°C to
24°C) between 2:30 and 3:30 PM and at least 2 h
after consuming a meal. After a 30-min rest in the sitting position,
patients were placed on a tilt table. ECG electrodes (CM5 lead) and a
respiration sensor (node-tip thermistor) were attached and connected to
a polygraph system and a frequency modulation tape recorder (model
MR30; Teac; Tokyo, Japan).
To control the effects of respiration on HRV measures, the HUTT was performed under paced breathing. After a 10-min stabilization period, patients were instructed to breathe quietly in synchrony with a metronome signal (for 15 breaths/min; 0.25 Hz) until the end of the HUTT. Data were collected with the patient in the supine position for 5 to 10 min and in the 70° head-up tilt position for 6 to 10 min after the table was moved to the head-up tilt position for > 15 s. After the end of the data collection in each position, the cuff BP was measured on the right upper arm with a sphygmomanometer.
HRV was analyzed off-line on a personal computer. The ECGs were played back from the frequency modulation tape and were digitized to 12-bit data at a sampling frequency of 1 KHz. All R-R intervals were measured with a fast-peak detection algorithm, and all errors in the detection of R waves were edited manually. R-R interval time series of 256 s in a stationary state and including < 1% of ectopic beats were selected for both supine and head-up tilt periods (excluding the first minute after tilting).
The power spectral density of the R-R interval time series was estimated by fast-Fourier transformation. Normal-to-normal R-R intervals were interpolated by cubic spline function, were resampled at 1 Hz, were detrended with linear regression, and were filtered with a Hanning window. Power spectral density was computed by a 256-point fast-Fourier transformation, was corrected for loss of variance resulting from the sampling and filtering processes described earlier, and was integrated by > 0.04 to 0.15 Hz, 0.20 to 0.30 Hz, and 0.00 to 0.50 Hz, respectively, for obtaining the LF, the HF, and the total power. The powers of these frequency components were expressed as the natural logarithm of the absolute value, and the power of the LF also was expressed as a normalized unit (LFnu), which was calculated by dividing the power by the total power minus power below 0.03 Hz.9 Heart rate was calculated from the mean of normal-to-normal R-R intervals in each position. The postural response of each measure was evaluated as the difference between measurements made in the supine and the tilt positions (value during tilt - value during supine). Figure 1 shows the time series and power spectra of the R-R interval during the HUTT in representative patients.
|
2 tests with Yates correction
were used for between-group comparisons of quantitative and categoric
variables, respectively. The Bonferroni method was used for multiple
comparisons to guard against an increase in the type-I error level.
A Cox proportional hazards regression model was used for
survival analysis. The associations of the baseline demographic and
clinical variables with a risk for death were evaluated by the
univariate Cox model. Prognostic associations of HRV measures were
determined by both univariate models and multivariate models adjusted
for the other clinical variables. For the graphic display of survival
probabilities, Kaplan-Meier survival curves were used. Quantitative
data were presented as the mean ± SD, and risks for death were
presented as risk ratio (RR) with the 95% confidence interval (CI).
For all statistical analyses, p < 0.05 was considered to be
significance. |
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Baseline Characteristics of Patients Grouped by Subsequent Survival
State
When the patients were grouped by subsequent survival state, some
baseline clinical data showed a significant group difference (Table 1
). Compared with surviving patients, patients who subsequently died from
cardiac causes were older, had lower body mass index levels, had lower
left ventricular ejection fractions, and were more likely to have
significant stenosis in multiple coronary arteries at baseline, while
patients who subsequently died from noncardiac causes were older but
were comparable for other clinical variables. However, the fractions of
patients undergoing coronary interventions and those receiving regular
medications of ß-blockers and angiotensin-converting enzyme
inhibitors during the follow-up period did not differ among the groups.
|
|
|
|
|
|
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Earlier Observations and Strength of the Present Study
Much clinical evidence has been accumulated for the association of
reduced HRV with adverse prognosis in patients after acute myocardial
infarction1
2
3
and in those patients with congestive heart
failure.14
15
Because experimental studies7
8
have demonstrated that HRV is almost abolished by atropine, cardiac
vagal dysfunction and/or the resulting relative sympathetic
overactivity have been hypothesized as the mechanisms mediating adverse
prognosis in patients with reduced HRV.1
2
3
16
However, convincing clinical evidence for this concept seems limited. In most of the earlier studies17 18 on the prognostic value of HRV, even those analyzing short-term HRV, data were collected by ambulatory ECG monitoring, mostly in patients receiving medications. Under these conditions, the autonomic neural activities are influenced by posture, food intake, wake-sleep cycle, and other physical and mental activities. The relationships between autonomic neural activities and HRV also could be influenced by medications and respiratory parameters. An increase in breathing frequency and a decrease in tidal volume reduce the HF without changing the mean cardiac vagal tone.4 19 Indeed, in earlier clinical studies1 3 15 with ambulatory monitoring, the HRV measures showing the strongest predictive power were those reflecting global or long-term variability, such as 24-h SD, triangular index,2 and ultra-low and very-low frequency components.16 The underlying mechanisms of these measures are unclear. Basic physiologic research reported that HRV measures such as the successive difference in R-R intervals and HF reflect more specifically the cardiac vagal function, but these measures had only moderate prognostic power in the clinical studies4 using ambulatory monitoring. Thus, the prognostic significance of intrinsic autonomic dysfunction in patients with CAD is not clear from these earlier observations.
From this point of view, the present study has strength. It is unique compared to earlier studies concerning the following points: (1) the use of HUTT under a controlled environment and physiologic state; (2) the withdrawal of treatment with medications that might influence autonomic functions and their assessment by HRV; and (3) the use of paced breathing to control the effects of respiration on autonomic assessment by HRV. Much evidence from basic physiologic research of HRV supports the notion that the analysis of HRV under these conditions provides a reliable autonomic functional assessment.4 7 8 13 The observations of the present study seem to be useful for evaluating the associations between intrinsic autonomic functions and prognosis in patients with CAD.
Prognostic Value of Reduced HF
The present study partly supports the concept that decreased
cardiac vagal activity results in an increased risk for death in
patients with CAD. We observed that a decreased HF in both the supine
position and during a head-up tilt showed a significant
univariate association with an increased risk for noncardiac death. The
HF was assessed with the same method and the same conditions as were
used in a previous study7
in which we observed the strong
correlation between the HF and the level of cardiac vagal tone as
assessed by pharmacologic autonomic blockades. In the present study,
however, we also observed that a decreased HF had no significant
association with the risk for cardiac death and that its predictive
power for noncardiac death disappeared after adjustment for
cardiovascular risk factors and CAD severity. Cardiac vagal dysfunction
may be associated with adverse medical conditions that have a poor
prognosis, but this association could be the consequence of, at least
in part, coexisting cardiovascular risks and/or disease
severity.5
20
21
Possible Mechanisms for Postural LF Rise
As to the autonomic neural mechanisms underlying the postural
response of LF (ie, LF rise), we need to consider complex
interactions between the following two factors: ß-adrenergic
activation; and cardiac vagal withdrawal.
There is convincing evidence indicating the important involvement of ß-adrenergic sympathetic activation in the postural increase in LF.8 9 Studies have reported that the LF, particularly its normalized power, increases with the patient in the upright position and that the increase is suppressed by ß-blockade. In the present study, we observed that when the patients were grouped by postural LF response at baseline, those showing an LF rise (group R) were more frequently treated by ß-blockers during the follow-up (Table 5) and that the difference in cardiac mortality among the groups was enhanced after excluding those patients regularly treated by ß-blockers. Although the mean R-R interval at baseline was not short in group R, one may speculate that patients showing an LF rise may have presented clinical conditions in which ß-blocking therapy was beneficial and that ß-blocker therapy may have moderated the harmful effects of the LF rise. These speculations are consistent with the hypothesis that the LF rise reflects the exaggerated postural ß-adrenergic activation, which could lead to an adverse prognosis in stable patients with CAD.
However, there is also much evidence suggesting vagal involvement in LF. The LF decreases with strenuous exercise22 23 and vagal blockade with atropine,8 and it increases with sleep at night.24 The LF is decreased in patients with severe congestive heart failure, a state known as the loss of vagal-cardiac restraint.25 26 27 Additionally, the LF may result from Mayer waves in BP through baroreflex control of the heart rate, which is mainly vagally mediated.28 In the present study, we observed that the postural decrease in the HF was smallest in patients who died from cardiac causes (difference not significant; Table 2 ) who showed the greatest LF rise. Given that postural vagal withdrawal itself could be a factor reducing the LF in the upright position, LF rise may partly reflect impaired postural vagal response due to a decreased vagal reserve for postural withdrawal.
It is important to note that a postural increase in the LF has been observed in less than half of subjects.29 30 In the present study, we found that the distribution of postural LF response was similar between patients with CAD and age-matched and gender-matched healthy control subjects. In both groups, an LF rise was observed in about one third of subjects (Fig 2) . A similar observation was reported in an earlier study10 with healthy subjects. The study10 also reported that an LF rise is characteristic among subjects susceptible to neurally mediated syncope, a condition in which ß-adrenergic overactivation plays an important pathophysiologic role in its early stage.31 These data indicate that the LF rise is not a consequence of disease but an individual characteristic of postural autonomic response.
If the postural response of the LF is determined by the interactions between increased ß-adrenergic activation and impaired cardiac vagal response, the patients showing LF rise may reflect those patients with greater dependence on ß-adrenergic activation than on vagal withdrawal in the autonomic neural regulation of the postural heart rate response. Such characteristics of autonomic function may adversely affect the prognosis of stable patients with CAD.
Limitations
Because this is a retrospective cohort study of prognosis, it is
subject to all of the limitations of such studies. Particularly, the
prognosis of CAD could depend on therapeutic strategies selected by
physicians. We observed, however, that there were no group differences
in the frequency of coronary interventions performed during the
follow-up (Table 3)
and that the frequency of regular ß-blocking
therapy was greater in group R. Thus, the adverse prognosis in patients
with LF rise was not attributable to the insufficient implementation of
effective therapies.
It is unclear whether LF rise is involved in the pathophysiologic mechanism of death in patients with CAD or is merely characteristic of patients with poor prognosis. We observed that the LF rise is associated only with an increased risk for cardiac death but not for noncardiac death. Also, a survival analysis only for stroke death (n = 6) revealed no association of LF rise (data are not shown). Given that the causes of cardiac death in this study were acute myocardial infarction and sudden cardiac death, the LF rise may be associated with the progression of CAD or an increased susceptibility to fatal arrhythmia.
In the present study, we assessed HRV and its postural response under paced breathing, which may have affected the LF and its postural response.19 The use of paced breathing improves the accuracy of the assessment of cardiac vagal modulation through better separation of the LF and HF in frequency domain and through controlling the nonautonomic effects of respiratory frequency and tidal volume on the HF.19 32 However, in a previous study19 we observed that paced breathing reduced the LF with patients in the supine and tilt positions by 31% and 36%, respectively, and hence the postural LF response by a similar percentage. Thus, the results of our study, particularly the definition of groups by postural LF response, may be applicable only to HRV measures assessed under paced breathing.
Finally, we did not perform ambulatory electrocardiography in our patients. Although we observed that LF rise is a powerful predictor of cardiac mortality in patients with CAD, we were unable to examine whether our method is advantageous over long-term HRV analysis with ambulatory electrocardiography.
|
Conclusion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
|
Footnotes |
|---|
This work was supported by Grant-in-Aid for Scientific Research (C) from the Ministry of Education, Science, Sports and Culture, Tokyo, Japan (No. 11670698; JH), a Grant-in-Aid for Research in
Nagoya City University, Nagoya, Japan (1999; JH), and a research grant from Pfizer Health Research Foundation, Tokyo, Japan (No. 99A052; JH).
Received for publication January 26, 2001. Accepted for publication May 17, 2001.
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
