Professional Documents
Culture Documents
2010 - ClimentAM-Role of The Atrial Rate As A Factor Modulating VR - PACE
2010 - ClimentAM-Role of The Atrial Rate As A Factor Modulating VR - PACE
Background: During atrial fibrillation (AF), RR interval histograms show different populations of
predominant RR (pRR) intervals. These pRR intervals have been suggested to be multiples of the refractory
period of the atrioventricular (AV) node or caused by the existence of a dual AV node physiology. In this
study, the hypothesis that pRR intervals are related to the dominant atrial fibrillatory rate is tested.
Methods: In this study, Holter electrocardiogram signals from 55 patients with persistent AF were
analyzed. Number and position of pRR intervals were detected and compared with mean and standard
deviation of the dominant atrial cycle length (DACL). In addition, effects of an enhancement of vagal
activity and rate-control treatments (β-blockers and verapamil) were evaluated.
Results: In all patients with more than one pRR interval and in 47% with one pRR interval, RR interval
populations were statistically related with multiples of the DACL. During night activities and during β-
blockers treatment, mean ventricular rate was decreased (P < 0.01). This change was associated with a
variation in the percentage of occurrences of each pRR (P < 0.01), whereas no statistical differences were
present in the mean DACL or in the position of pRR intervals. A variation of the DACL due to verapamil
was associated with a consistent modification in the position of the pRR intervals.
Conclusion: The relation between pRR and multiples of the DACL during AF suggests that more probable
RR intervals are caused by different conduction ratios of the atrial rate. (PACE 2010; 33:1510–1517)
atrial fibrillation, ventricular response, atrial cycle length, AV node conduction, rate control
Table I.
Patient Characteristics
Age, years 65 ± 11 67 ± 10 63 ± 11 NS
Male, n (%) 43 (78) 17 (81) 26 (76) NS
AF duration, months 19 ± 20 23 ± 17 18 ± 21 NS
Left atrial diameter, mm 50 ± 8 52 ± 7 50 ± 9 NS
Left ventricular ejection fraction, % 50 ± 14 46 ± 17 53 ± 10 NS
Heart disease* NS
None, n (%) 3 (5) 0 (0) 3 (9)
Hypertension, n (%) 23 (42) 7 (33) 16 (47)
Coronary artery disease, n (%) 10 (18) 4 (19) 6 (18)
Valvular heart disease, n (%) 6 (11) 4 (19) 2 (6)
Cardiomyopathy, n (%) 6 (11) 3 (14) 3 (9)
Medication* NS
β-blockers, n (%) 38 (69) 14 (66) 24 (71)
Calcium channel blockers, n (%) 15 (27) 3 (14) 12 (35)
Digitalis, n (%) 24 (44) 9 (42) 15 (44)
Ventricular rate, beats per minute 81 ± 22 79 ± 13 82 ± 74 NS
*Treatment received during recording, more than one possible; NS = not significant.
and classified using Getemed software. Results defined as the dominant peak in the power
of automatic analysis were visually inspected spectrum between 3 and 10 Hz. The DACL of
and each error in the detection or labeling was each segment was calculated as the inverse of the
corrected manually. Only RR intervals between dominant frequency. The main limitations in the
normal (sinus) beats were used in the subsequent estimation of the AF rate were the low signal-
analysis. Holter recordings with a high degree of to-noise ratio of Holter signals and the possible
noise or with a percentage of ventricular ectopic QRS complex residue. In order to reject segments
and aberrant beats higher than 20% were excluded in which a precise estimation of the DACL was
(n = 10). The remaining 55 patients comprise the not possible, the spectral concentration around the
study population (Table I). dominant frequency (±0.5 Hz) was measured. In
In order to evaluate the effect of rate-control cases where this spectral concentration was lower
treatments, two Holter recordings were acquired than 40%, DACL estimation was discarded. For
for six patients (three under β-blockers and each 30-minute segment, 180 values of DACL and
three under verapamil). In each subgroup (β- spectral concentration were calculated (one value
blockers and verapamil), baseline recordings were every 10 seconds for each parameter). If more
compared to drug recordings for two patients, and than 70% of the 10-second segments presented a
recordings obtained under different dosages of the spectral concentration higher than 40%, mean and
same drug were compared for one patient. standard deviation of DACL of segments with high
spectral concentration were computed. In addition
to these parameters, mean DACL during day (7–
Analysis of the Atrial Activity 23 hours) and night (23–7 hours) activity periods
In order to estimate the mean atrial fibrillatory and during the whole day were compared.
rate, the dominant frequency of atrial fibrillatory
signals was estimated from the surface Holter
ECG lead. Ventricular activity was reduced by Analysis of the VR
subtracting a matching QRS-T template. This The analysis of pRR during AF was based on
template was computed for each beat as a the bidimensional histogram profile (BHP) method
linear combination of the principal components that has been presented in detail elsewhere.12
previously obtained from the analysis of all beats In brief, this methodology allows analyzing the
in the ECG segment.11 For every 10 seconds dynamic behavior of the VR during AF by means
of ECG recording, the dominant frequency was of the analysis of only near-equal consecutive RR
intervals derived from a Poincaré plot. Short-term between day and night. Similar comparisons were
variations in the VR were analyzed by computing a done between baseline and rate-control Holters.
30-minute BHP every 15 minutes, with 50% over- A P-value < 0.05 was considered statistically
lapping (i.e., for a Holter recording of 24 hours, significant.
95 BHP from overlapping sequential segments can
be generated), so that the number and position of Results
pRR during the recording was monitored. For each
30-minute signal, an algorithm to identify pRR on Atrial Cycle Length and pRR Intervals
the BHP was used in order to automatically detect Short-term variations of BHP and multiples of
different RR populations.12 For each 30-minute DACL of two representative patients are depicted
segment, one, two, or three pRR intervals were in Figure 1. In the recording shown in panel 1, one
detected and patients were grouped according to peak was detected in the BHP during the whole
their VR pattern. Although the existence of two day. The detected pRR laid inside the window of
or more pRR in at least one 30-minute segment the fourth multiple of the DACL in 90% of the
may suffice to assess the existence of a multimodal segments (P < 0.01). The recording shown in panel
VR pattern, in order to guarantee robustness to 2 was classified as multimodal due to the existence
erroneous pRR detections, a total percentage of of two pRR intervals in 25% BHP segments. In
BHP segments with two or more pRR greater than this patient, the long pRR laid inside the window
10% was required for assessing multimodality. of the fifth multiple of the DACL in 100% of the
In addition, mean ventricular rate during the 91 segments (P < 0.01). Similarly, the short pRR
day (7–23 hours) and night (23–7 hours) activity interval appeared inside the window of the third
periods and during the whole day were compared. multiple of the DACL in 100% of the 23 segments
Moreover, position and probability of occurrence (P < 0.01).
of each pRR were analyzed for day (7–23 hours) By analyzing short-time variations of BHPs
and night (23–7 hours) activity periods in all over 24 hours for the whole database, 21 patients
patients with more than one pRR. Specifically, presented a unimodal VR pattern and 34 patients a
for each activity period, mean position and mean multimodal VR pattern (23 with two pRR intervals
percentage of RR interval occurrences around and 11 with three pRR intervals). There was
50 ms of each pRR were measured. no difference in clinical characteristics between
patient groups (Table I). Of those, two patients
Comparison between Atrial Cycle Length with unimodal VR patterns and 10 patients
and pRR Intervals with multimodal VR patterns were excluded
A pRR was considered to be an nth multiple from further analysis due to QRS-T cancellation
of DACL if it was located within a window artifacts and estimation of DACL. Summarizing, 19
at n · (DACL ± std DACL ). For instance, if during patients with unimodal VR pattern and 24 patients
30 minutes the measured DACL was 150 ± 5 ms, with multimodal VR pattern were analyzed.
a pRR was considered a second multiple if its Notice that a thorough estimation of the DACL
duration ranged from 290 to 310 ms, a third was not possible for all 30-minute segments of
multiple if its duration ranged from 435 to 465 ms, these 43 patients. Thirty-minute segments with
etc. In order to avoid identifying relationships an insufficient spectral concentration or with a
between DACL and pRR due to wide tolerance standard deviation of the DACL larger than 15 ms
ranges, segments that presented a std DACL greater were rejected. Patients with less than 50 segments
than 15 ms were rejected. In addition, mean DACL fulfilling these criteria were not considered for
and mean ventricular rate during the whole day further analysis. The mean number of measurable
and during day and night activity periods were segments and the percentage of them that satisfy
compared. the relation between atrial cycle length and pRR
interval are shown in Table II.
Statistical Analysis For all patients with a multimodal VR pattern
In order to evaluate the relationship between and for nine of the 19 patients with a unimodal
multiples of the DACL and the different RR VR pattern, it was statistically demonstrated that
populations, one tailed t-test was performed to pRR intervals were harmonics of the DACL. For
evaluate the hypothesis that pRR-n · DACL was these nine patients in 75 ± 15% of measurable
lower than n · std DACL for each patient. In addition, segments, pRR was statistically related to DACL
paired Student’s t-tests were performed in order (P < 0.05). In the other 10 patients with a unimodal
to compare (1) DACL between day and night, VR pattern, 56 ± 21% of segments presented
(2) ventricular rate between day and night, (3) a pRR-n·DACL lower than std DACL (P = ns).
percentage of occurrences of each pRR between Notice that the standard deviation of the position
day and night, and (4) position of each pRR of the pRR interval in these 10 patients was
Figure 1. In panels 1A and 2A, 24 hours bidimensional histogram profile (BHP) are depicted. In
panels 1B and 2B, short-time variations of the BHP during the day are represented according to
a red color scale. Crosses represent local maxima of each BHP segment automatically detected.
DACL multiples are represented as the mean and standard deviation of each measurable 30-
minute segment. In panels 1A and 1B, a unimodal VR is depicted in which a single pRR was
found during the whole day together with the fourth multiple of DACL that was statistically
related with the pRR. In panels 2A and 2B, a multimodal patient is depicted in which a bimodal
BHP was detected during several segments of the day; the third and fifth multiples of DACL were
statistically related with pRR intervals.
90 ± 50 ms, significantly higher than in the rate was significantly lower for night periods. This
other nine patients with a unimodal pattern (50 ± decrease was associated with an increase in the
23 ms, P < 0.05). These patients may have been proportion of long pRR in comparison to short
incorrectly classified into the unimodal group and pRR. The number of patients with a third pRR was
present two nearby pRR populations leading to an too small to detect statistically significant trends.
inaccurate location of the pRR at some segments An example of the stability of the position of pRR
and thus inability to demonstrate its relation with and their day-to-night variation of occurrences is
the DACL. illustrated in panel 2B of Figure 1. In this patient,
during the night, number of occurrences of the
Autonomic Tone Effects
Comparison of day and night periods for all
patients with more than one pRR is summarized Table III.
in Table III. No statistical differences were found
neither in the mean DACL nor in the position of Comparison between Day and Night Activity Periods
each pRR interval, although the mean ventricular
Day Night
(7–23 hours) (23–7 hours)
Table II.
RR mean ms 711 ± 131 832 ± 158*
Measurable Segments and Percentage of Them DACL mean ms 160 ± 20 161 ± 21
Statistically Related with Atrial Cycle Length 1◦ pRR ms 505 ± 142 489 ± 166
% oc. 29 ± 14% 14 ± 12%*
Measured pRR-n · DACL < 2◦ pRR ms 782 ± 162 792 ± 170
N Segments n · std DACL % oc. 24 ± 15% 30 ± 13%*
3◦ pRR ms 1116 ± 251 1123 ± 257
Unimodal BHP 19 76 ± 43 65 ± 22% % oc. 11 ± 13% 12 ± 5%
Multimodal BHP 24 79 ± 36 85 ± 11%
BHP = bidimensional histogram profile; DACL = dominant atrial *P < 0.001. DACL = dominant atrial cycle length; pRR =
cycle length; pRR = predominant RR interval. predominant RR interval; oc. = ocurrences.
harm. % oc.
49%
22%
36%
pRR corresponding to the third atrial harmonic
harm.
6th
9th
decreased markedly, whereas the number of
3◦ pRR
occurrences of the pRR corresponding to the fifth
5th
6th
5th
2◦ pRR
910 ± 30
1208 ± 38
atrial harmonic increased.
(ms)
For the nine patients who presented unimodal
759 ± 23
19% 1018 ± 27
839 ± 36
VR pattern and pRR was demonstrated to be a
(ms)
multiple of the DACL, mean RR interval was
longer during night periods (767 ± 181 ms
harm.
5th
7th
5th
vs 849 ± 187 ms, P < 0.05). However, two
2◦ pRR
β-Blockers Recordings
different mechanisms produce these reductions in
harm. % oc.
781 ± 28
931 ± 38
797 ± 39
29%
Verapamil Recordings
4%
the ventricular rate. In six patients, a significant
(ms)
variation of the mean DACL was detected between
day and night (157 ± 14 ms vs 165 ± 16 ms, P <
1◦ pRR
3rd
3rd
4th
0.05). In these patients, both day and night periods
harm.
4th
5th
4th
presented a pRR similar to the fifth multiple
464 ± 23
706 ± 53
497 ± 31
of the DACL and so the position of the pRR
1◦ pRR
(ms)
(ms)
In three patients, no significant differences were
measurable for the mean of the DACL between day
154 ± 12
171 ± 14
168 ± 16
DACL = dominant atrial cycle length; pRR = predominant RR interval; harm. = harmonic order respect to mean DACL.
Mean (ms)
was the fourth multiple of the mean DACL during
(ms)
DACL
the day, whereas during the night pRR was the
fifth multiple of the mean DACL.
Rate-Control Treatment Effects
RR Mean
Rate-Control Effects
(ms)
(ms)
Comparisons between Holter recordings dur-
ing baseline and rate-control treatment for the
Table IV.
harm.
4th
and for the three patients treated with verapamil
are summarized in Table IV.
3 pRR
On one hand, patients treated with β-blockers
711 ± 21
5th
6th
5th
2 pRR
(ms)
presented a multimodal VR pattern with two ◦
populations during both recordings. No significant
◦
harm.
(ms)
539 ± 21
1017 ± 30
associated with an increase in the percentage of 672 ± 43
harm. % oc.
Baseline Recordings
24%
39%
(ms)
◦
3rd
3rd
4th
2nd
3rd
5th
1 pRR
(ms)
Mean (ms)
DACL
(ms)
Figure 2. Panels 1A and 1B represent 24 hours BHP and short-time variations of the BHP together
with the third and fifth atrial rate multiples (panels 1B and 2B) during a baseline Holter (panels 1A
and 1B) and during rate-control treatment with β-blockers (panels 2A and 2B).
increased due to an increase in the harmonic order lated that pRR conductions should be multiples of
(Table IV). the refractory periods of the AV node.3 Although
it was a reasonable hypothesis, it does not explain
Discussion satisfactorily the variations of pRR during day and
Main Findings night or β-blockers. Since AV nodal refractory
period is modified by the vagal tone,16 position
In this study, a relation between the atrial of pRR should gradually change between day and
cycle length and the VR during AF has been found. night, following the vagal tone. Rather than this,
For the first time during AF, the position of pRR the reduction of the mean RR during the night
intervals has been shown to be correlated with was caused by an increase in the number of
multiples of the DACL. It has been illustrated occurrences of long RR interval populations and
how the reduction of the ventricular rate due to a decrease in the number of occurrences of short
changes in the autonomic tone (e.g., days vs night) RR interval populations, whereas the position of
does not imply variations in the position of pRR populations did not change. A similar behavior
intervals, but rather variation in the probability of was also observed in the three patients treated
occurrence of each pRR intervals. Similar behavior with β-blockers.
was found when the conduction properties of the In 1956, Moe et al. presented evidence of
AV node were modified by β-blockers. In addition, the existence of an AV node dual physiology.17
it has been illustrated how a modification in the It was demonstrated that the so-called slow
atrial rate due to verapamil produce shifts in pathway has a shorter refractory period but
the position of the pRR consistent with modified lower conduction velocity than the so-called fast
DACL values. pathway. The concept of two pathways with
separated conduction properties was used later
AV during AF by Olsson et al., Ingemansson et al.,5 and Cai
The VR during AF is usually assumed et al.18 for suggesting that pRR intervals during
to be a random phenomenon.13 However, the AF are caused by conduction through different
existence of a certain relationship between each AV node pathways. However, even when the
RR interval and the preceding RR series was slow pathway presents a slightly shorter refractory
demonstrated by means of bidimensional his- period than the fast pathway, both refractory
tograms and autocorrelation functions,14,15 where periods range from 250 to 350 ms,19 differences
different populations of predominant RR intervals that do not exceed 100 ms, far from the positions
can be found. Despite the importance of these of the pRR intervals observed in our results and
phenomena in rate-controlling therapies for the previous studies.5,8 According to our results, short
AF, mechanisms responsible of this observation and long RR distributions, or even multiple RR
remain controversial. In 1950, Söderström postu- populations, do not necessarily imply conduction
References
1. Bollmann A, Husser D, Mainardi L, Lombardi F, Langley P, with chronic atrial fibrillation – Role of morphology of RR interval
Murray A, Rieta JJ, et al. Analysis of surface electrocardiograms in variation. Circulation 2001; 103:2942–2948.
atrial fibrillation: Techniques, research, and clinical applications. 7. Oka T, Nakatsu T, Kusachi S, Tominaga Y, Toyonaga S, Ohnishi
Europace 2006; 8:911–926. H, Nakahama M, et al. Double-sector Lorenz plot scattering in an
2. Bollmann A, Lombardi F. Electrocardiology of atrial fibrillation. R-R interval analysis of patients with chronic atrial fibrillation:
Current knowledge and future challenges. IEEE Eng Med Biol Mag Incidence and characteristics of vertices of the double-sector
2006; 25:15–23. scattering. J Electrocardiol 1998; 31:227–235.
3. Söderström N. What is the reason for the ventricular arrhythmia in 8. Rokas S, Gaitanidou S, Chatzidou S, Agrios N, Stamatelopoulos S.
cases of auricular fibrillation? Am Heart J 1950; 40:212–223. A noninvasive method for the detection of dual atrioventricular
4. Ingemansson MP, Carlson J, Olsson SB. Modification of intrinsic node physiology in chronic atrial fibrillation. Am J Cardiol 1999;
AV-nodal properties by magnesium in combination with glucose, 84:1442–1445.
insulin, and potassium (GIK) during chronic atrial fibrillation. J 9. Tebbenjohanns J, Schumacher B, Korte T, Niehaus M, Pfeiffer
Electrocardiol 1998; 31:281–292. D. Bimodal RR interval distribution in chronic atrial fibril-
5. Olsson SB, Cai N, Dohnal M, Talwar KK. Noninvasive support for lation. Impact of dual atrioventricular nodal physiology on
and characterization of multiple intranodal pathways in patients longterm rate control after catheter ablation of the posterior
with mitral-valve disease and atrial-fibrillation. Eur Heart J 1986; atrionodal input. J Cardiovasc Electrophysiol 2000; 11:497–
7:320–333. 503.
6. Rokas S, Gaitanidou S, Chatzidou S, Pamboucas C, Achtipis D, 10. Weismuller P, Braunss C, Ranke C, Trappe HJ. Multiple AV nodal
Stamatelopoulos S. Atrioventricular node modification in patients pathways with multiple peaks in the RR interval histogram of
the holter monitoring ECG during atrial fibrillation. Pacing Clin 18. Cai N, Dohnal M, Olsson SB. Methodological aspects of the use
Electrophysiol 2000; 23:1921–1924. of heart-rate stratified RR interval histograms in the analysis of
11. Castells F, Mora C, Rieta JJ, Moratal-Perez D, Millet J. Estimation atrioventricular-conduction during atrial fibrillation. Cardiovascu-
of atrial fibrillatory wave from single-lead atrial fibrillation lar Research 1987; 21:455–462.
electrocardiograms using principal component analysis concepts. 19. Hegbom F, Orning OM, Heldal M, Gjesdal K. Effects of ablation,
Med Biol Eng Comput 2005; 43:557–560. digitalis, and beta-blocker on dual atrioventricular nodal pathways
12. Climent AM, Guillem MS, Husser D, Castells F, Millet J, Bollmann and conduction during atrial fibrillation. J Cardiovasc Electrophys-
A. Poincare surface profiles of RR intervals. A novel noninvasive iol 2004; 15:1141–1146.
method for the evaluation of preferential AV nodal conduction 20. Garrigue S, Tchou PJ, Mazgalev TN. Role of the differential
during atrial fibrillation. IEEE Trans Biomed Eng 2009; 56:433– bombardment of atrial inputs to the atrioventricular node as a factor
442. influencing ventricular rate during high atrial rate. Cardiovascular
13. Meijler FJ, Jalife J. AV node function during atrial fibrillation. In: Research 1999; 44:344–355.
Mazgalev TN and Tchou PJ (eds.): Atrial-AV Nodal Electrophysiol- 21. Garrigue S, Mowrey KA, Fahy G, Tchou PJ, Mazgalev TN.
ogy: A View from the Millennium. Armonk, NY, Futura Publishing Atrioventricular nodal conduction during atrial fibrillation: Role
Co., 2000, pp. 251–268. of atrial input modification. Circulation 1999; 99:2323–2333.
14. Gelzer AR, Moise NS, Vaidya D, Wagner KA, Jalife J. Temporal 22. Chorro FJ, Kirchhof CJHJ, Brugada J, Allessie MA. Ventricular
organization of atrial activity and irregular ventricular rhythm response during irregular atrial pacing and atrial-fibrillation. Am
during spontaneous atrial fibrillation: An in vivo study in the horse. J Physiol 1990; 259:H1015–H1021.
J Cardiovasc Electrophysiol 2000; 11:773–784. 23. Kirsh JA, Sahakian AV, Baerman JM, Swiryn S. Ventricular response
15. Van Den Berg MP, Van Noord T, Brouwer J, Haaksma J, Van to atrial fibrillation: Role of atrioventricular conduction pathways.
Veldhuisen DJ, Crijns HJ, Van Gelder IC. Clustering of RR intervals J Am Coll Cardiol 1988; 12:1265–1272.
predicts effective electrical cardioversion for atrial fibrillation. J 24. Roithinger FX, SippensGroenewegen A, Karch MR, Steiner PR, Ellis
Cardiovasc Electrophysiol 2004; 15:1027–1033. WS, Lesh MD. Organized activation during atrial fibrillation in man:
16. Nayebpour M, Talajic M, Villemaire C, Nattel S. Vagal modulation Endocardial and electrocardiographic manifestations. J Cardiovasc
of the rate-dependent properties of the atrioventricular node. Circ Electrophysiol 1998; 9:451–461.
Res 1990; 67:1152–1166. 25. Zhang Y, Mazgalev TN. Achieving regular slow rhythm during atrial
17. Moe GK, Preston JB, Burlington H. Physiologic evidence for a dual fibrillation without atrioventricular nodal ablation: Selective vagal
A-V transmission system. Circ Res 1956; 4:357–375. stimulation plus ventricular pacing. Heart Rhythm 2004; 1:469–475.