Download Paradoxical effect of respiration on ventricular rate in atrial fibrillation

Clinical Science (1989) 76, 109-112
109
Paradoxical effect of respiration on ventricular rate in atrial
fibrillation
JOHN M. RAWLES, G. R. PAl
AND
SUSAN R. REID
Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, U.K.
(Received 30 December 1987/3 May 1988; accepted 10 June 1988)
SUMMARY
1. In 50 subjects with atrial fibrillation we have
attempted to demonstrate variation of ventricular rate
with respiration, as evidence of cardioregulatory reflex
activity.
2. The electrocardiogram was recorded for 3 min
during spontaneous respiration. The presence of respiratory variation of R-R intervals was analysed by multiple
regression against a cosine function (cosinor analysis),
making it possible to determine the phase of respiration
when the intervals were longest.
3. Variation in ventricular rate with respect to respiration was demonstrated (P< 0.05) in seven (14%) cases.
On average, R-R intervals were longest at the end of
inspiration; this contrasts with sinus rhythm where P-P,
P-R and R-R intervals are longest around the time of
end-expiration.
4. These results suggest that in atrial fibrillation the
beat-to-beat ventricular rate may be under the influence
of cardioregulatory reflexes, but the effect of respiration is
weak and paradoxical.
Key words: atrial fibrillation, heart conduction system,
vagus nerve.
INTRODUCTION
Although the pulse in atrial fibrillation is commonly
described as being irregularly irregular, we have shown
that in 30% of cases the ventricular rhythm is nonrandom [1]. This could be due to beat-to-beat operation
of cardiovascular reflexes modulating ventricular rate by
altering conduction through the atrioventricular node.
In the previous paper we describe a method of quantifying respiratory variation of electrocardiogram intervals
and show that, in sinus rhythm, there is respiratory varia-
tion of the P-R interval [2]. In this paper the method is
applied to R-R intervals in patients with atrial fibrillation.
If the atrioventricular node behaves in the same way as it
does in sinus rhythm, respiratory variation of R-R intervals might be expected, with the longest duration in expiration. However, because of the large irregular
beat-to-beat changes of ventricular rate that characterize
this rhythm, detection of small, short-term changes of rate
associated with respiration presents formidable statistical
difficulties.
METHODS
Subjects
Fifty subjects, with a mean age of 72.5 (SD 10.9) years
(range 31-87 years), in established atrial fibrillation were
studied. Those taking drugs known to affect the heart rate
or interfere with autonomic function were excluded, in
particular those on anticholinergic drugs, including ipratropium by inhaler, ,B-adrenoceptor blockers, or antiarrhythmic agents, except digoxin, which was taken by 33
patients. None of the patients was diabetic or thought to
have autonomic neuropathy. Their diagnoses were as
follows: idiopathic atrial fibrillation, 21; ischaemic heart
disease, 14; congestive cardiomyopathy, nine; mitral valve
disease, three; obstructive airways disease, two; viral pericarditis, one.
Electrocardiogram recording
The end-expiratory point and the electrocardiogram
were observed and recorded as previously described for
sinus rhythm [2], except that a paper speed of 25 mm/s
was used. The R-R intervals were measured together with
their relationship to end-expiratory points using a microcomputer and digitizing pad; the results were stored on
disc.
Statistical analysis
Correspondence: Dr John M. Rawles, Department of Medicine, University of Aberdeen, Foresterhill, Aberdeen AB9 2ZD,
UK
The best fit of a cosine function curve to the cyclical
variation of R-R intervals around their mean value is
J. M. Rawles et al.
110
found by multiple regression analysis. The duration of
each respiratory cycle is taken as 360 degrees, both 0 and
360 signifying end-expiratory points. The timing of each
R-wave is calculated as its angular position (t, in degrees)
in the respiratory cycle that commenced with the previous
end-expiratory point. The duration of the associated
R-R interval is expressed as a percentage of the
mean R-R interval for the whole 3 min sequence and
regressed against costr) and sin(t) to give the amplitude,
phase angle and statistical significance of respiratory
variation of R-R intervals. Phase angle is defined as the
angular time in the respiratory cycle when the fitted
cosine function curve has its maximum positive value, that
is, when R-R intervals are longest. Fig. 1 shows an
example of cosinor analysis against respiration in a
64-year-old man with idiopathic atrial fibrillation.
In order to compare the distribution of phase angles in
respiration with those that might be obtained just by
chance, cosinor analyses were then repeated with respect
to an arbitrary time mark inserted every 1.5 s through the
recordings.
around the end-inspiratory point (Fig. 2a). The distribution of phase angles between six sectors of the respiratory
cycle was non-random (X 2 = 11.4, df= 5, 1'< 0.05), and
was different from that previously reported for P-P, P-R
and R-R intervals in sinus rhythm (X 2=46.4, df=5,
1'<0.001; X2=27.7, df=5, 1'<0.001; X2=52.4, df=5,
1'< 0.001), where phase angles clustered around endexpiration (Fig. 3) [21. The distribution of phase angles
derived from cosinor analysis against a time mark (Fig.
2b) did not differ from random (X2 = 4.2, df = 5, P> 0.5).
Demonstrable respiratory variation of ventricular rate
was not related to taking digoxin. Neither did the distribution of phase angles through the respiratory cycle differ
between patients taking digoxin and those not taking this
drug (X 2 = 4.2, df= 5, NS).
Table 1. Number of patients at various probability levels
with variation of R-R intervals with respect to respiration
and to a time mark
Significance
No.of patients
RESULTS
Patients with atrial fibrillation were older than subjects
with sinus rhythm studied previously (73 vs 45 years) and
had a higher ventricular rate (85 vs 72 beats/min) and a
higher respiratory rate (18 vs 14 breaths/min) [2].
Seven patients had significant (1'< 0.05) variation of
R-R intervals with respect to respiration, compared with
two in relation to a time-mark (Table 1).
The modal phase angle, when R-R intervals were
longest, was 150 degrees, phase angles tending to cluster
P>0.05
0.01 < P< 0.05
0.001 <P<O.OI
Total
25
Respiration
Time-mark
43 (86%)
48 (96%)
7(14%)
0(0%)
50(100%)
1 (2%)
1 (2%)
50 (100%)
(a)
20
+10
~
~
15
+5
Q)
S
S
0
....
......
10
'"
t)
0
Q)
'E'
Q)
u
;:l
t::
~
5
'"
Q)
'-
....
0
-5
ci
Z
is
0
00
(b)
10
-10
0
180
90
Inspiration
270
360
Expiration
t
(degrees)
Fig. 1. Result of cosinor analysis of R-R intervals. Ordinate: mean duration of R-R intervals expressed as the
percentage difference from the mean for the 3 min
sequence. Abscissa: angular time in respiratory cycle. R
(multiple correlation coefficient) = 0.15; I' (probability of
respiratory variation) < 0.05; amplitude, (amplitude of
respiratory variation) = 4.2%; phase (angular time in
respiration when respiratory variation has maximum positive value) = 163 degrees.
5
0
1-60
D DO
D
61-120 1 1 2 1 - 1 8 0 1 8 1 - 2 4 0 2 4 1 - 3 0 0 3 0 1 - 3 6 0
Phase angle (degrees)
Fig. 2. Histogram of distribution of phase angles of R-R
intervals against respiration (a) and against an arbitrary
time mark (b). D, Not significant by cosinor analysis; _,
significant by cosinor analysis.
Respiratory variation of heart rate in atrial fibrillation
35
30
25
~
20
u
Q)
:.s
;:l
....0
Vl
15
0
Z
10
5
0
1-60
I
0
iii
61-120 121-1801181-240241-300301-360
Phase angle (degrees)
Fig. 3. Histogram of distribution of phase angles derived
from cosinor analysis of R-R intervals in sinus rhythm [2].
Compare with Fig. 2(a). 0, Not significant by cosinor
analysis; ., significant by cosinor analysis.
DISCUSSION
In atrial fibrillation ventricular rate increases with atropine and exercise [3], and a possible mechanism is withdrawal of vagal tone which may be responsible for the
maintenance of a submaximal ventricular rate at rest. In
sinus rhythm resting vagal tone fluctuates in phase with
respiration, resulting in the phenomenon of respiratory
sinus arrhythmia which is associated with parallel changes
in atrioventricular nodal function [2]. Demonstration of
respiratory variation on ventricular rate at rest in atrial
fibrillation would therefore provide circumstantial evidence of autonomic control of heart rate by the vagus
nerve acting on the atrioventricular node.
In atrial fibrillation there is a differential of approximately five to one between the rate of arrival of stimuli at
the atrioventricular node from the atria, and the rate of
egress of stimuli to the ventricles. The conducting properties of the atrioventricular node would therefore be
expected to have a marked effect on ventricular rate, and
it would be surprising if ventricular rate did not alter
phasically with respiration. However, because of the
widely variable duration of R-R intervals in atrial fibrillation, the background variance is high and the respiratory
variation of ventricular rate may be difficult to demonstrate with statistical conviction.
Using cosinor analysis we have demonstrated
(P< 0.05) respiratory variation of R-R intervals in atrial
fibrillation in seven out of 50 cases, compared with two to
three which would be expected by chance. When cosinor
111
analyses were performed against an arbitrary time mark 2,
positive (P< 0.05) results were obtained, as expected. The
greater frequency of positive tests related to respiration
rather than to an arbitrary time mark suggests the presence
of respiratory variation of ventricular rate.
The method of cosinor analysis assumes that the variation of R-R intervals around their mean value is sinusoidal in form, which is not necessarily the case. However,
it is valuable in indicating the phase of respiration when
R-R intervals are longest (phase angle), taking into
account all the data points.
The non-random distribution of phase angles calculated by cosinor analyses against respiration, compared
with the random distribution against a time-mark, also
suggests that respiratory variation of ventricular rate is
genuinely present in atrial fibrillation.
If the changes in ventricular rate during respiration are
brought about by altered conductivity of the atrioventricular node, such as we have demonstrated in sinus
rhythm, an increase in ventricular rate in inspiration
would be expected. However, we have shown that in atrial
fibrillation maximum ventricular rate occurs in expiration,
the difference in phase angle distribution from that of P-P,
P-R and R-R intervals in sinus rhythm being statistically
highly significant.
In 1920, Kilgore [4] described 'Respiratory variations
of heart rate in the presence of auricular fibrillation'. 'Out
of nine cases of auricular fibrillation studied, six show a
fairly consistent tendency for shorter heart intervals
during late expiration or early inspiration'. He comments,
'It will be noticed that the tendency to acceleration during
late expiration and inspiration is the reverse of the
customary relation to respiration in cases of sinus
arrhythmia.'
There are several possible explanations for the difference in phase of respiratory variation of heart rate in atrial
fibrillation and sinus rhythm. Patients in atrial fibrillation
were older, and had a greater mean heart rate and respiratory rate than those in sinus rhythm and in many cases
had underlying cardiac disease. However, in sinus rhythm
neither age, heart rate nor respiratory rate were correlated with phase angle, and a complete inversion of phase
angle seems unlikely to be explained by these differences
even though the factors that determine the phase angle
of sinus arrhythmia are not fully known.
Inversion of the phase of the respiratory variation of
heart rate in atrial fibrillation could be due to changes in
the afferent or the efferent arm of the reflex mechanism.
The exact afferent stimulus for sinus arrhythmia is
unknown but a possible contribution arises from volume
receptors on the right side of the heart, where
volume-time relationships are undoubtedly different in
atrial fibrillation compared with sinus rhythm. However,
the transition to atrial fibrillation from sinus rhythm is
unlikely to lead to a reversal of the effect of respiration on
venous return and right-sided chamber volumes.
In the effector arm, the vagus nerve may not always
have parallel effects on the sinoatrial and atrioventricular
nodes, and it is possible that in atrial fibrillation there is a
paradoxical action on the atrioventricular node, vagal
112
1. M. Rawles et al.
stimulation causing an apparent increase in conductivity,
a situation that has been demonstrated experimentally in
dogs [5].
In the light of the expectation that the effect of respiration on ventricular rate in atrial fibrillation would be an
exaggeration of that seen in sinus rhythm, the paucity of
reports of respiratory variation on ventricular rate in
atrial fibrillation in man, the difficulty of proving its
presence, and the paradoxical nature of the effect
reported by Kilgore [4] and ourselves, are all noteworthy.
If the atrioventricular node is considered as a slow but
direct route for the transmission of stimuli from atria to
ventricles, it is difficult to see why variation of vagal tone
during respiration does not have a pronounced effect on
ventricular rate in atrial fibrillation; a paradoxical action
is even more puzzling. Also, such a through conductor
model of the atrioventricular node, even invoking concealed conduction, does not explain the distribution of
R-R intervals in atrial fibrillation [6].
The explanation for these observations may lie in a different model of the atrioventricular node, that of a periodic biological oscillator, as described by Guevara &
Glass [7]. Such a model is capable of explaining many
complex varieties of atrioventricular block, and it closely
simulates the behaviour of the canine atrioventricular
node during electrophysiological study [8]. The phases of
the atrioventricular oscillator may be retarded by early
arrival of supraventricular stimuli, the physiological
mechanism being hyperpolarization. This property results
in seemingly paradoxical slowing of the ventricular rate
with increased supraventricular rate or improved atrioventricular conductivity once the ratio of supraventricular
and atrioventricular rates exceeds a certain value, set at
2: 1 in the mathematical model.
Thus, increased vagal tone might alter ventricular rate
in three ways. By its action on the atria, the rate of fibrillation might be increased [9] and the ventricular rate
reduced. By depressing conductivity of the upper part of
the atrioventricular node, the rate of arrival of stimuli at
the nodal oscillator would fall, and ventricular rate would
be increased. By reducing the intrinsic frequency of the
nodal oscillator, thought to be in the junctional region,
ventricular rate would be reduced. Because of the opposing effects on ventricular rate of vagal action at different
sites, the net effect is weak, and if the effect on conductivity predominates, paradoxical.
The weak, paradoxical effect of respiration on ventricular rate therefore challenges the view of the atrioventricular node as a through conductor, but gives support to
the idea of the atrioventricular node as a biological oscillator.
ACKNOWLEDGMENTS
We are indebted to Mr Alan Anderson, Senior Lecturer
in Statistics in the University of Aberdeen, for constructive criticism of statistical methods, and to Professor Cecil
Kidd for helpful comments on the manuscript. A research
grant from the Grampian Health Board is gratefully
acknowledged.
REFERENCES
1. Rawles, J.M. & Rowland, E. (1986) Is the puse in atrial
fibrillation irregularly irregular? British Heart Journal, 56,
4-11.
2. Rawles, J.M., Pai, G.R. & Reid, S.R. (1989) A method of
quantifying sinus arrhythmia: parallel effect of respiration
on P-P and P-R intervals. Clinical Science, 76, 103-108.
3. Beasley, R, Smith, D.A. & McHaffie, D.J. (1985) Exercise
heart rates at different serum digoxin concentrations in
patients with atrial fibrillation. British Medical Journal, 290,
9-11.
4. Kilgore, E.S. (1920) Time relations of heart beats. Respiratory variations of heart rate in the presence of auricular
fibrillation. Heart, 7, 81-104.
5. Martin, P. (1977) Paradoxical dynamic interaction of heart
period and vagal activity on atrioventricular conduction in
the dog. Circulation Research, 40, 81-89.
6. Cohen, RJ. & Berger, RD (1983) A quantitative model for
the ventricular response during atrial fibrillation. IEEE
Transactions on Biomedical Engineering, 30, 769-781.
7. Guevara, M.R. & Glass, L. (1982) Phase locking, period
doubling bifurcations and chaos in a mathematical model of
a periodically driven oscillator: a theory for the entrainment
of biological oscillators and the generation of cardiac
dysrhythmias. Journal ofMathematical Biology, 14, 1-23.
8. van der Tweel, I., Herbschleb, J.N., Borst, C. & Miejler, EL.
(1986) Deterministic model of the canine atrio-ventricular
node as a periodically perturbed, biological oscillator.
Journal ofApplied Cardiology, I, 157-173.
9. Smeets, J.L.RM., Allessie, M.A., Lammers, W.J.E.P.,
Bonke, ELM. & Hollen, J. (1986) The wavelength of the
cardiac impulse and reentrant arrhythmias in isolated rabbit
atrium. Circulation Research, 58, 96-108.