Patterns of beat-to-beat heart rate variability in advanced heart failure

Patterns of beat-to-beat advanced heart failure

heart rate variability

in

Diminished heart rate variability is associated with high sympathetic tone and an increased mortality rate in heart failure cases. We constructed Polncare plots of each sinus R-R interval plotted against the subsequent R-R interval from 24hour Halter recordings of 24 healthy subjects (control group) and 24 patients with heart failure. Every subject ln the control group had a comet-shaped Poincare plot resulting from an increase in beat-to-beat dispersion as heart rate slowed. No patient with heart failure had this comet-shaped pattern. Instead, three distinctive patterns were identified: (1) a torpedo-shaped pattern resulting from low R-R interval dispersion over the entire range of heart rates, (2) a fanshaped pattern resulting from restriction of overall R-R interval ranges with enhanced dispersion, and (3) complex patterns with clusters of points characteristic of stepwise changes in R-R intervals. Poincare pattern could not be predicted from standard deviations of R-R intervals. This first use of Poincare plots in heart rate variability analysis reveals a complexity not readily perceived from standard deviation inform&on. Further study is warranted to determine if this method will allow refined assessment of cardiac-autonomic integrity in heart failure, which could help identify patients at highest risk for sudden death. (AM HEART J lgg2;123:704.)

Mary A. Woo, MN, RN, William G. Stevenson, MD, Debra K. Moser, MN, RN, Robert B. Trelease, PhD, and Ronald M. Harper, PhD. Los Angeles, Calif.

Heart rate variability provides a noninvasive indication of autonomic nervous system t0ne.l Increased sympathetic activity decreases heart rate variability, and increased parasympathetic activity increases it.2 Analysis of R-R intervals with the use of standard deviations3s 4 histograms59 ’ and spectral techniques7, 8 provides an assessment of overall variability but obscures instantaneous beat-to-beat changes.g,lo Recently, Goldberger et al.,li, l2 Glass and Zeng,13, l4 and others l5 have provided evidence that sinus node depolarization modulated by autonomic fluctuations can behave as a nonlinear process. Beatto-beat variation can be easily displayed for visual assessment by the graphing of each R-R interval against the subsequent R-R interval. This is known as a Poincare plot and is often used to detect patterns resulting from nonlinear processes that may not be

From the Division of Cardiology, UCLA School of Medicine, of Nursing, the Department of Anatomy and Cell Biology, Research Institute, University of California, Los Angeles. Supported Tau and Nursing. Received Reprint 47-123 4/I/34403

704

in part by a grant by the Audrienne for publication

July

and the School and the Brain

from the Gamma Tau Chapter of Sigma Theta H. Moseley Scholar Award, UCLA School of 11, 1991,

accepted

Sept.

3, 1991.

requests: William G. Stevenson, MD, UCLA Division of Cardiology, CHS, 10833 Le Conte Ave., Los Angeles, CA 90024-1679.

observable by other methods of analysis.g* lo Poincare plots have not been previously applied to the examination of heart rate variability. Heart failure is characterized by increased sympathetic and depressed parasympathetic nervous system activity. I’, I7 Increased sympathetic nervous system activity is associated with increased mortality in patients with heart failure.ls Furthermore, diminished heart rate variability is associated with increased mortality after myocardial infarction.“, I9 We hypothesize that alterations and interactions of autonomic and cardiovascular systems accompanying heart failure could produce complex heart rate variability patterns detectable in Poincare plots, which may add to information derived from analysis of standard deviation of R-R intervals. The purpose of this study was (1) to establish the characteristics of Poincare plots of R-R intervals in healthy individuals, (2) to determine if Poincare plots from patients with advanced heart failure differ from those of healthy individuals, and (3) to determine if Poincare plots from patients with heart failure who subsequently die suddenly are different from those of patients who survive. METHODS

Patients referred to the UCLA Cardiomyopathy Center for cardiac transplant evaluation and hospitalized between

Volume Number

Table

123 3

Beat-to-beat

variability

in CHF

705

I. Group demographicdata Variable

Group 1

Group 2

No. of patients Age (yd Male Vasodilator Hydralazine Captopril Cause Idiopathic CM CAD LVEF Serum sodium (mEq/L) Antiarrhythmic Quinidine Procainamide Digoxin Cardiac index (L/min/m2) RA pressure (mm Hg) Mean PA pressure (mm Hg) PCW pressure (mm Hg)

12 46 + 12 9

12 46 k 13 8

7 5 6 6 0.17 * 0.05 134 * 5 1 2 7 2.6 it 0.6 6~4 31 + 7 18 k 5

BA

7 5

RRn

6 6 0.17 + 0.04 133 + 5 1 2 7 2.6 f 0.4 7k2 27 + 4 16 k 4

CM, Cardiomyopathy, CAD, coronary artery disease; LVEF, left ventricular ejection fraction; RA, right atrial; PA, pulmonary artery; PCW, pulmonary capillary wedge.

Fig. 1. Poincarbplot construction is schematicallyshown. For the first point on the plot, A is RRn (x axis) and B is RRn + 1 (y axis). For the next point, B is RRn and C is RRn + 1.

II. Standard deviation values of sinusR-R intervals for &minute epochs Table

Control Group 1

May 1,1987, and September30,1990, wereincluded in the study (approved by UCLA Human Subject Protection Committee) if they met the following criteria: (1) advanced heart failure requiring treatment with vasodilators,diuretics, and hemodynamic monitoring with a thermodilution pulmonary artery catheter20;(2) New York Heart Association functional class III or IV; (3) sinus rhythm; (4) 24-hour Holter electrocardiogramobtained before hospital discharge; (5) informed consent for study participation. Exclusion criteria included therapy with P-blockers, calcium-channel blockers, amiodarone,or mechanical ventilatory or circulatory support; renal failure; diabetesmellitus; myocardial infarction within the preceding 6 months; presence of atrioventricular block; and history of cerebrovascular accident. Sudden death wasdefined as death occurring within 15 minutes of a change in symptomsor during sleep. Of the patients who met study criteria, 12 patients who died suddenly in less than 6 months after the 24-hour Holter recording were identified. These suddendeath victims formed group 1. Group 2 consistedof 12 patients who alsomet study criteria but lived longer than 12 months after the 24-hour Holter recording and who were matched to group 1 subjectsfor age,vasodilator therapy, causeof heart disease,left ventricular ejection fraction, serum sodium levels, antiarrhythmic medication, digoxin, cardiac index, right atria1pressure,and pulmonary capillary wedgepressure.All patients had advancedheart failure, with average left ventricular ejection fraction of 0.17 f 0.05, cardiac index of 2.02 f 0.32L/min/m2, right atria1pressureof 11 2 6 mm Hg, and pulmonary capillary wedgepressureof 25 f 6 mm Hg before adjustment of therapy. Demographic data

No. of patients Heart rate variability

group

-

12 24 62 + 35 msec 52 + 33 msec 137 + 43 msec

mean+ SD <50msec >49, lOO

Group 2

msec

12

5 (42%) 4 (33%) 3 (25%)

8 (67%) 3 (25%) 1 (8%)

0 (0%) 5 (21%) 19 (79%)

and final hemodynamicvaluesfor groups1 and 2 are shown in Table I. The control group consistedof 24healthy volunteers who were matched by ageand sexto the advanced heart failure group. Each had a normal result on a 12-lead electrocardiogram, a normal physical examination, and no history of heart disease. Holter analysis. Holter tapeswerescannedon a Del Mar 750scanner(Del Mar Avionics, Irvine, Calif.) with semiautomatic technique. Data obtained from the tapesprovided the time of beat, type of beat, serial R-R intervals with a resolution of + 2.4 msec,and standard deviation of normal R-R intervals in 5-minute epochsfor the 24-hour recording period.21 Poincarb plots of each R-R interval plotted against the next interval (RRn vs RRn + 1) were created with custom-designed

software

(Fig. 1). Only sinus beats

were used to generate the PoincarB plots, and all ectopic atria1 and ventricular beats and sinus beats immediately before and after the ectopic beats were removed. During long-term electrocardiogram recording, missed triggers

are a major source of artifacts.

In plots of 80,000 to

145,000sinusbeatsrecordedover 24 hours,periodic failure

706

Woo et al.

L !01

c

200

American

RR (msecl

RR (msecl

2000

2000

B

D

200

2ko

RR (msec)

RR (msec)

March 1992 Heart Journal

2000

mm

Fig. 2. Poincare plots from four healthy volunteers. Panel A is a 28-year-old man (heart rate variability standard deviation = 148 msec). Panel B is a 45year-old man (heart rate variability standard deviation = 167 msec). Panel C is a 54-year-old man (heart rate variability standard deviation = 90 msec). Panel D is a 52-year-old woman (heart rate variability standard deviation of 104 msec).

to detect even 0.01 S’, of beats produces a characteristic pattern of an R-R interval preceded or followed by an interval approximately twice as long. When these patterns occurred the Holter tapes were rescanned. If visual inspection confirmed the presence of undetected QRS complexes, these intervals and sinus beats immediately preceding and following the missed beats were excluded from analysis. Poincare plots were inspected by four observers blinded to the clinical characteristics of the subjects. Each observer categorized the observed pattern according to one of four classes (Figs. 2 and 3). Agreement between observers was 96 % . The mean of the standard deviations of sinus R-R intervals was calculated for 5-minute epochs over the 24hour recording period. 21 As with the Poincare plots, only R-R intervals between sinus beats were used, and ectopic beats were deleted from the analysis. Statistics. Chi square analyses were used to compare proportions. Analysis of variance and posthoc t tests were

used to compare continuous variables. 0.05 were considered significant. RESULTS Healthy

p Values less than

control subjects. All healthy volunteers shared a characteristic pattern in their Poincare plots (Fig. 2). These plots showed a wide range of R-R intervals, typically extending from approximately 500 to 1000 msec along both axes. In addition, as the R-R interval increased, the dispersion of subsequent R-R intervals increased. As heart rate slowed, the interbeat variation increased, creating a distinctive comet-shaped pattern. Patients with heart failure. Patients with advanced heart failure exhibited Poincare plots that differed markedly from those of healthy volunteers (Fig. 3). The comet-shaped pattern was not observed in any

Volume Number

123 3

Beat-to-beat

I

A

variability

in CHF

2000

2:”

RR

(msec)

0

2000

RR

(msec)

2000

.. .,.’.:. .I’ ‘@. .,:-

E

1

200

RR

(msec)

2000

RR

(msec)

2000

D

F

200

RR

(msec)

2000

I

200

RR

(msec)

2000

Fig. 3. Poincare plots from six patients with advanced heart failure. Panels A and B have examples of the torpedo pattern. In panels A and B heart rate variability standard deviations were 41 and 43 msec, respectively. Panels C and D have examples of the fan pattern. In panels C and D heart rate variability standard deviations were 26 and 59 msec, respectively. Panels E and F have examples of complex patterns. In panels E and F heart rate variability standard deviations were 104 and 44 msec respectively.

707

708

woo

et al.

Amerlcan

March 1992 Heart Journal

III. Patient characteristics by PoincarB patterns

Table

FAN No. of patients Heart rate variability mean Mean HR (beats/min) Mean BP (mm Hg) CI (L/min/m2) RA (mm Hg) Mean PA (mm Hg) PCW (mm Hg) Ventricular ectopy/hr Serum sodium (mEq/L) LVEF

f SD (msec)

3 26 + 108 + 75 k 2.7 t 9+4 30 k 18 k 316 i130 * 0.17 *

4** 26 6 0.5 5 5 542 2 0.04

TORPEDO

COMPLEX

12 71 + 42** 93 t 17 78 + 8 2.7 f 0.4 6+2 28 i 5 16 +- 5 286 + 774 135 * 5 0.17 k 0.03

9 51 + 99 k 76 f 2.5 + 7-t2 30 + 17 + 220 + 134 + 0.17 f

COMET

25 13 6 0.5 8 4 355 5 0.05

24 137 + 43 msec* 77 -+ 10*

1 f 3*

Average heart rate; BP, blood pressure; CJ, cardiac index; RA, right atria1 pressure; PA, pulmonary artery pressure; PCW, pulmonary capillary wedge pressure; LVEF, left ventricular ejection fraction.*Significant differences between comet and other patterns (p < 0.001); **p = 0.06 between fan and torpedo patterns; no significant differences between fan, torpedo, and complex patterns.

HR,

Table

IV. Distribution of Poincarb patterns Heart Sudden

Torpedo Fan Complex Comet

death 4 1 7 0

failure No sudden 8 2 2 0

death

Healthy volunteers 0 0 0 24

*By chi square analysis the patterns of healthy volunteers differed from those of the heart failure group (p < 0.001). The distribution of patterns between patients with heart failure who later died suddenly and patients who survived more than 1 year after Holter monitoring did not reach statistical significance (p = 0.06f

patient with advanced heart failure. Patterns from the heart failure group could be categorized into three types. Twelve patients (50%) exhibited a torpedo-shaped (Fig. 3, A and B) pattern resulting from little or no increase in R-R interval dispersion as heart rate slowed. In three patients (12 % ) the plots formed a triangular or fan-shaped plot (Fig. 3, C and D) in which points radiated in a relatively symmetric manner from a narrow base at the lower R-R intervals. In nine patients (38%) the patterns were complex, consisting of several clusters of points, often with some of the features of both the fan and torpedo formations (Fig. 3, E and F). Comparison with R-R interval standard deviation. The means of the R-R interval standard deviations, calculated from successive 5-minute epochs over the entire 24-hour recording for each group, are shown in Table II. All healthy volunteers had values greater than 50 msec. The standard deviation of R-R intervals in the healthy volunteers was significantly greater than in groups 1 and 2 (p < 0.001). However, the standard deviation values of R-R intervals between the patients in group 1, who died suddenly, and the

patients in group 2, who survived, were not significantly different. Table III shows the standard deviation of R-R intervals, hemodynamics, and mean number of hourly ventricular ectopic beats associated with the four different Poincare patterns (torpedo, fan, complex, and comet). The standard deviation of R-R intervals was greater for the normal comet pattern than for the abnormal patterns (p < 0.001). Although the mean heart rate and R-R interval standard deviation tended to be reduced for the fan pattern, the difference did not reach statistical significance, possibly because of the small number of fan patterns observed. The R-R interval standard deviations and mean heart rate were similar for the torpedo and complex Poincarb patterns. Some patients with similar values for the standard deviation of R-R intervals had markedly different Poincari! patterns (Fig. 3). Heart failure severity and sudden death. In the heart failure group, hemodynamics, ejection fraction, and serum sodium concentration were similar regardless of Poincare pattern (Table III). In the advanced heart failure group, therefore, hemodynamic severity of heart failure did not predict Poincari! pattern. There was no correlation between the mean hourly number of ventricular ectopic beats, digoxin or antiarrhythmic use, and Poincarit pattern. The distribution of patterns in patients who died suddenly (group 1) or survived (group 2) is shown in Table IV. Although a complex pattern was frequently present in the patients who died suddenly, the distribution of patterns between groups 1 and 2 did not reach statistical significance (p = 0.06), possibly because of the small number of patients in the study. DISCUSSION

Heart rate normally fluctuates throughout the day in response to physiologic demands. Some sources of

Volume

123

Number

3

variation are near-periodic, such as the cyclic contribution from respiration (respiratory sinus arrhythmia). Such near-periodic sources of variation may be assessedby mathematic procedures optimized for analysis of cyclic activity, such as spectral techniques. However, nonperiodic influences, such as momentary changes in baroreceptor activity, psychologic state, and physical activity, can also change heart rate. Moreover, recent studies suggest that some of the factors causing heart rate variability can be modeled as nonlinear processesz2 Nonlinear behavior may produce patterns that are not detected from analysis of R-R intervals with standard deviations, R-R interval histograms, or spectral analysis, all of which provide summary evaluations of variation in R-R intervals. Poincare plots provide a beatto-beat visual display that can reveal patterns resulting from nonlinear processes and nonperiodic fluctuations.iO The technique does not require data in a continuous-time series or a normal distribution of R-R intervals.lO This study demonstrates that Poincare plots of R-R intervals reveal distinctive patterns that distinguish healthy volunteers from patients with advanced heart failure. Furthermore, Poincare plots provide additional information on moment-to-moment changes in heart rate independent of the R-R interval standard deviations. In the healthy control group, beat-to-beat variability increased as R-R intervals lengthened, creating a characteristic comet-shaped appearance. This strikingly consistent pattern was observed in all subjects in the control group. Distinctively different patterns were observed in the patients with advanced heart failure. The torpedo-shaped pattern indicated that for any R-R interval, the next R-R interval was likely to deviate only minimally. However, this minimal deviation did not indicate a fixed heart rate but was suggestive of a heart rate pattern that changed gradually, yet maintained small beat-to-beat R-R variability. As heart rate slowed, the expected increase in dispersion found in healthy subjects was minimal or absent. Other patterns suggested very different relationships in beat-to-beat variability. In the fanshaped and complex patterns the overall range of R-R intervals was diminished, but the beat-to-beat R-R interval dispersion was greater than in the healthy volunteers. When compared with the cometshaped pattern, the fan-shaped patterns had small increases in R-R interval length that were associated with greater subsequent R-R interval dispersion over a much shorter overall range. The complex patterns lacked the graded relationship between successive R-R intervals found in the comet-shaped formation. Instead, the complex configurations exhibited dis-

Beat-to-beat

variability

in CHF

709

crete stepwise clusters of points, with distinct gaps between the clusters (Fig. 3, E and F). The complex patterns may result from stepwise changes in R-R intervals, consistent with nonlinear behavi0r.i’ The mechanisms of these strikingly different patterns are not entirely clear. Multiple factors could cause such diverse R-R interval configurations and would be of potential interest for future investigations. For example, the “stem” or “core” of the patterns could be the result of sympathetic nervous system influence, whereas the increased dispersion, noticed at longer R-R intervals, may be reflective of parasympathetic activity, respiratory sinus arrhythmia, or sleep state. Although victims of sudden death tended to have complex patterns, this difference did not reach statistical significance, possibly because of the small sample size. We speculate that the torpedoshaped pattern is closer to normal and that complex and fan-shaped patterns reflect a greater disturbance in autonomic-cardiac regulation. However, hemodynamics and serum sodium concentration, previously shown to be related to heart failure severity and total mortality,20~ 23were similar in patients with complex or fan patterns. Limitations. Our rigorous entry criteria restricted our sample size to a small number of patients with advanced heart failure. It is possible that the lack of statistical difference in pattern distribution between sudden death and non-sudden death groups was the result of this small sample size. It must be noted that the qualitative separation of the plots into pattern categories was somewhat arbitrary. It is likely that other patterns occur. Sleep states influence heart rate and may exert effects on the beat-to-beat pattern development.24 We did not record sleep state information but used 24-hour recordings in all patients. Gathering such sleep state information may assist in determining the particular sources of variability on patterns associated with specific behavioral conditions. Conclusions. Poincare plots of R-R intervals over a 24-hour period reveal a degree of complexity in beatto-beat relationships that is not appreciated from measurement of the standard deviation of R-R intervals. Qualitatively distinctive patterns that separate normal heart rates from those of patients with advanced heart failure are observed. Further study is warranted to determine if refinement for sudden death risk stratification and characterization of autonomic nervous system abnormalities in heart failure can be achieved with this method of heart rate variability analysis. We thank RN, UCLA

Lynne W. Stevenson, MD, and Julie Creaser, MN, Cardiomyopathy Center; Juan A. Tan, BS, UCLA

710

Woo et

al.

School of Nursing Computer Lab; Alan Garfinkel, UCLA Department of Kinesiology; and Connie Wright, RN, UCLA Holter Lab, for their assistance.

American

12. Goldberger 13.

REFERENCES

1. Wolf MW, Varigos 2.

3.

4.

GA, Hunt D, Sloman JG. Sinus arrhythmia in acute myocardial infarction. Med J Aust 1978;2:52-3. Bieaer JT. Kleiaer RE. Fleiss JL. Rolnitzkv LM. Steinman RCrMiller’JP. &mponents of heart rate variability measured during healing of acute myocardial infarction. Am J Cardiol 1988;61:208-15. Lombardi F, Sandrone G, Pernpruner S, Sala R, Garimoldi M, Cerutti S, Baselli G, Pagani M, Malliani A. Heart rate variability as an index of sympathovagal interaction after acute myocardial infarction. Am J Cardiol 1987;60:1239-45. Bennett T, Farquhar IK, Hosking DJ, Hampton JR. Assessment of methods for estimating autonomic nervous control of the heart in natients with diabetes mellitus. Diabetes 1978;

27:1167-74. 5. Casolo G, Balli E, Taddei 6. 7.

T, Amuhasi J, Gori C. Decreased spontaneous heart rate variability in congestive heart failure. Am J Cardiol 1989:64:1162-7. Law HF, Epstein RA, Epstein MAF. High resolution determination of the R-R interval. Am J Phvsiol 1979:236:H894-8. Hayano J, Yamada M, Sakakibara YiFujinami’T, Yokoyama K, Watanabe Y, Takata K. Shortand long-term effects of cigarette smoking on heart rate variability. Am J Cardioll990;

14. 15. 16. 17.

18. 19.

9. 10. 11 --.

T, Bryg RJ, Maddens ME, Bhasin SS, Cronin S, Boden WE, Lehmann MH. Effects of transdermal scopolamine on heart rate variability in normal subjects. Am J Cardiol 1990,65:604-8. Denton TA, Diamond GA, Helfant RH, Khan S, Karagueuzian H. Fascinating rhythm: a primer on chaos theory and its application to cardiology. AM HEART J 1990;120:1419-40. Abraham R, Shaw E. Dynamics: the geometry of behaviorpart II: chaotic behavior. Santa Cruz: Aerial Press, 1983:137. Goldberger AL, Findley LJ, Blackburn MR, Mandell AJ. Nonlinear dynamics in heart failure: implications of longwavelength cardiopulmonary oscillations. AM HEART J 1984;

107:612-5.

AL. Nonlinear dynamics, fractals. and chaos: anplications to cardiac electrophysiology. Ann Biomed Eng 1990;18:195-8. Glass L, Zeng W. Complex bifurcations and chaos in simple theoretical models of cardiac oscillations. Ann NY Acad Sci 1989:316-27. Glass L. There is a theory of heart. Physica 1990;43:1-16. Chialvo DR, Gilmour RF Jr, Jalife J. Low dimensional chaos in cardiac tissue. Nature 1990;343:653-7. Porter TR, Eckberg DL, Fritsch JM, Rea RF, Beightol LA, Schmedtje JF, Mohanty PK. Autonomic pathophysiology in heart failure patients. J Clin Invest 1990;85:1362-71. Leimbach WN, Wallin BG, Victor RG, Aylward PE, Sundlof G, Mark AI. Direct evidence from intraneural recordings for increased central sympathetic outflow in patients withheart failure. Circulation 1986;73:913-9. Cohn JN. Abnormalities of peripheral symaathetic nervous system control in congestive heart failure. Circulation 1990; 82(suppl I):I-59-67. Kleiger RE, Miller JP, Bigger JT, Moss AJ. Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am J Cardiol 1987;59:256-

62. 20. Stevenson

21.

65:84-8. 8. Vybiral

March 1992 Heart Journal

22. 23.

24.

LW, Dracup KA, Tillisch JH. Efficacy of medical therapy tailored for severe congestive heart failure in patients transferred for urgent cardiac transplantation. Am J Cardiol 1989;63:461-4. Martin GJ, Magid NM, Valentini V. Heart rate variability and the evaluation of risk of sudden cardiac death [Abstract]. Clin Res 1987;35:302a. Rigney DR, Mietus JE, Goldberger AL. Is normal sinus rhythm “chaotic”? Measurement of Lyapunov exponents [Abstract]. Circulation 1990;82(suppl 111):236. Lee WH, Packer M. Prognostic importance of serum sodium concentration and its modification by converting-enzyme inhibition in patients with severe chronic heart failure. Circulation 1986;73:257-67. Martin RJ. History and comments. In: Martin RJ, ed. Cardiorespiratory disorders during sleep. 2nd ed. Mount Kisco: Futura Publishing Company, 1990:1-14.