Published by Oxford University Press on behalf of the European Society of Cardiology 2007
Quantifying fractionation and rate in human atrial fibrillation using monophasic action potentials: implications for substrate mapping
1 University of California and Veterans Affairs Medical Centers, San Diego, CA, USA; 2 Veterans Affairs Medical Center, Georgetown University, Washington, DC, USA
* Corresponding author. Tel: +1 202 745 8389; fax: +1 202 745 8472. E-mail address: michael.r.franz{at}verizon.net
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
Aims: To use monophasic action potentials (MAPs) to better assess the rate and the presence of fractionated electrograms during the mapping of atrial fibrillation (AF). Substrate mapping is increasingly central to AF ablation. However, traditional bipolar signals poorly represent waveform shape, making it unclear whether fractionation reflects local waveform variations, true electrogram fragmentation, or noise, and raising issues on whether their spectral dominant frequencies (DFs) accurately estimate AF rate.
Methods and results: In 28 patients with paroxysmal or persistent AF (left atrial diameters 44 ± 8 mm), we studied 49 epochs of right atrial MAPs during AF. We compared fractionation, spectral and time-domain AF rate estimates using MAPs and bipolar electrograms obtained by filtering the MAPs. Fractionation was overestimated in bipolar rather than MAP electrograms (P = 0.005) and often reflected artefacts on the MAPs. Conversely, local waveform variability in the MAPs, including alternans or fractionation, was often uniform in the bipolar electrograms. The measured AF cycle length (CL) was accurately represented by the DF of the MAPs (r = 0.73, P < 0.001) but, due to double counting, not by the DF of bipolar signals (r = 0.29, P = 0.07). Spectral CL estimates were therefore accurate (
20 ms from measured CL) for 77% of MAPs but for 45% of bipolar signals only. A novel autocorrelation method better estimated CL in MAPs (r = 0.92; P < 0.001) and bipoles (r = 0.82; P < 0.001), with 89 and 77% accuracy, respectively (P < 0.01).
Conclusion: Atrial fibrillation organization and rate are better represented by MAPs, which portray fibrillatory waveform shape, than by bipolar recordings. This approach may more reliably portray electrogram variability, fragmentation, and rate for the mapping of AF substrates.
Key Words: Atrial fibrillation, Substrate mapping, Ablation, Monophasic action potentials, Fourier analysis, Autocorrelation
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
Mapping the organization (‘substrates’) of atrial fibrillation (AF) for fractionated electrograms1
,2 or rapid rate3 is a promising approach to ablation. Regions of electrogram fractionation may be critical for AF perpetuation,1,2 whereas intra-atrial AF rate gradients may separate paroxysmal from persistent AF.4 Nevertheless, some recent reports have questioned whether ablation at sites of fractionation or rapid rate are effective3 and whether AF rate gradients are robust indicators of AF substrates.3–5
This controversy partly reflects the methods used to map AF substrates.6 Because routine bipolar electrograms obscure atrial waveform morphology, it is unclear whether fractionation represents organized yet fluctuating waveforms such as alternans,7 true fragmentation that reflects wavefront collision, slow conduction, pivot points8 or autonomic innervation,9 or other substrates. Technically, subtraction used to create bipolar electrograms may cause ‘double-counting’ that influences spectral dominant frequencies (DFs) and may overestimate AF rate.3
Monophasic action potentials (MAPs) usefully portray atrial waveform morphology in sinus rhythm, pacing,10 atrial flutter11 and transitions to AF.7,12 We hypothesized that MAPs would portray local AF electrogram fractionation and variability not apparent in bipolar electrograms, and thus better estimate AF rate. We tested this hypothesis using a filtering strategy to compare fibrillatory MAPs and bipolar electrograms simultaneously at the same site, avoiding problems inherent in using adjacent catheters that examine slightly different myocardial regions, in patients with paroxysmal and persistent AF.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
Patient recruitment
We studied 28 consecutive male patients (age 69 ± 10 years) with paroxysmal or persistent AF referred to electrophysiological study (EPS) at the Veterans' Affairs Medical Centers (VAMC) in Washington, DC and San Diego, CA, USA. The study was approved by the Institutional Review Board of both institutions, and all patients provided written informed consent. All patients had been anticoagulated or did not show thrombus formation on transoesophageal echocardiography (Table 1).
|
Data collection
Electrophysiological study was performed in the fasted state, at least five half-lives after discontinuing all anti-arrhythmic medications except amiodarone (n = 4 patients; Table 1). A quadrapolar MAP-pacing catheter (EPT, Sunnyvale, CA, USA)10
,13 was advanced transvenously to the high right atrium (RA), in addition to clinically indicated catheters. In each patient, AF was recorded for two 1 min epochs separated by 5–10 min. The MAPs were bandpass filtered routinely between 0.05 and 775 Hz. Similarly, the bipolar electrograms were bandpass filtered between 30 and 500 Hz. The surface ECG was recorded prior to the clinical procedure and bandpass filtered between 0.05 and 100 Hz.
Electrogram signal processing
Signals were digitized at 1 kHz to 16-bit resolution and exported for analysis from the physiological recorder (Bard, Billerica, MA, USA, at both performance sites) to a PC running custom software written in Labview (National Instruments, Austin, TX, USA) written by the first author. Recordings with excessive baseline wander, artefact, or noise were excluded from analysis.
To compare MAP and bipolar signal characteristics at identical sites, we derived bipolar electrograms from the ‘unfiltered’ quadrapolar MAP catheter by 30–500 Hz filtering using a second order Butterworth filter. We reasoned that these are also bipolar signals, obtained by tip to nearby-electrode subtraction. This was validated by comparing these derived bipolar electrograms with ‘true’ bipolar electrograms recorded from a juxtaposed 6 F quadrapolar catheter (filtered at 30–500 Hz) in a patient subset (Figure 1, top panels). When catheters were apposed without noise, they correlated well in the time and frequency domains (Figure 1A).
|
Grading fractionation in fibrillatory electrograms
We compared waveform shape in the simultaneous MAP and derived bipolar electrograms at each site for >2 s. This duration has been selected previously,4
based on the fact that AF is stable for <6 s only.14 Morphological variability was graded as 0 (regular: minimal variability; Figure 1A), 1 (variable, but fractionated for <20% of tracing; Figure 2), 2 (fractionated for 20–50% of tracing; Figure 2) and 3 (continuous fractionation for >50% of tracing; Figure 2). Monophasic action potential fractionation was defined as 
30% variation in plateau amplitude or duration at 70% repolarization, unrelated to noise. Bipolar fractionation was defined as high-frequency electrograms with duration 60 ms.
|
Gold standard manual measurement of atrial fibrillation cycle lengths
Atrial fibrillation cycle lengths (CLs) were measured manually for >2.048 s (typically 10–15 intervals) using digital calipers at a timescale of 200 mm/s (Figure 1, top panels). We excluded cycles
70 ms (extending recent reports that excluded <100 ms15 or <120 ms1) (Figure 1). Reproducibility was assessed by repeating each measurement; Pearson's coefficient for inter-observer measurement was r = 0.96 (P < 0.0001) with no difference between means.
Spectral quantification of fibrillatory electrograms
Spectral analysis was blinded to CL measurement and patient demographics. We analysed electrograms over 2.048 s using a Hanning window to taper MAP and bipolar electrograms to zero at their edges.4 Each signal was rectified, and a 2048-point fast Fourier transformation was used to compute its power spectrum (spectral resolution 0.48 Hz). Dominant frequency was the frequency of the largest amplitude in the 3–15 Hz bandwidth (CL 333 to 67 ms; Figure 1, bottom panels).
Spectral AF organization, measured as DF narrowness, was computed as the area of the DF peak (1 Hz width) to the encompassing area of width 2.5 Hz.16 Peak area ratios nearer 1 suggest the narrow spectral DF of organized AF, whereas values nearer zero reflect the broader spectral envelope of less organized AF.16
Autocorrelation estimate of cycle length
We also estimated CL using the autocorrelation technique, extending our recent ECG work.16–19 Briefly, the electrogram series is correlated to itself using the Pearson function after progressive 1 ms timeshifts, generating an autocorrelation function exemplified in Figure 3 (same patient as shown in Figure 1A). The largest peak in the range 67–300 ms reflects the time-shift that optimally re-registers the time series, i.e. estimated CL (Figure 3: autocorrelation estimate 174 ms, measured CL 174 ms).
|
Statistical analysis
Continuous data are represented as mean ± standard deviation (SD). The two-tailed t-test was used to compare continuous variables. The Wilcoxon matched-pairs signed-ranks test was used to determine the significance of the association of paired morphological grades for MAP and bipolar series. Paired continuous variables were compared using linear regression. The
2 test was applied to contingency tables. A probability <5% was considered statistically significant. |
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
The baseline clinical characteristics are summarized in Table 1. After excluding 7 noisy recording segments, 49 segments were available for analysis. Manual measurements of AF CL from MAPs and derived bipolar electrograms correlated well (r = 0.96; slope =0.98; P < 10–5). Recording from the high RA helped to minimize far-field ventricular activity.
Fibrillatory fractionation vs. rate
There was a poor relationship between the presence (grades II and III) or the absence (grades 0 and I) of fractionation in MAP and bipolar electrograms (P > 0.20 for each). Figure 2 shows organized regular MAPs that appeared fractionated in bipoles (Figure 2A), and vice versa (Figure 2B). The MAPs also revealed variations in waveform shape (e.g. alternans) obscured in bipolar electrograms (Figure 2D), distinct activations that resembled artefact in the bipoles (Figure 2D) and, conversely, noise or far-field signals that appeared as distinct activations in the bipoles (Figure 2E).
Fractionation (grades II and III) was also overestimated in bipolar compared with MAP signals (Figure 4A: 42.6 vs. 21.2%, P = 0.005; 2). As a corollary, spectral CL estimates differed in accuracy for MAP vs. bipolar electrograms (Figure 4B). Compared with the gold standard of the manual measurements, CL estimates were ‘accurate’ ( 20 ms error) in 44.7% of the spectra of the bipolar signals, due to double counting, vs. 74.5% of MAPs. Notably, the autocorrelation method improved the accuracy of CL estimates, yielding 76.6% for the bipolar and 89.4% for the MAP signals (P < 0.05).
|
Using morphology assessment, less organized AF correlated with shorter measured AF CL in MAPs (r = 0.391; 0.014) and bipolar (r = 0.514; P < 0.001) series, as expected.8
However, using spectra, the less organized types of AF (broader envelope, lower peak area ratio16) did not reflect shorter CL in the MAP or bipolar series.
Spectral atrial fibrillation cycle length estimates
The spectral DF on the bipolar signals (8.43 ± 1.84 Hz) significantly underestimated CL at 125 ± 30 ms (P < 10–8 vs. measured 165 ± 25 ms, Table 1), due to double counting. Conversely, the spectral DF on the MAPs (5.98 ± 1.03 Hz) more accurately estimated CL at 172 ± 28 ms (P = 0.44 vs. measured 167 ± 25 ms, Table 1).
Figure 5A shows a case of accurate spectral MAP and bipolar-based CL estimates, with DF = 6.91 Hz and estimated CL = 145 ms (=1000/6.91) for each. The autocorrelation-based CL estimates were 145 and 146 ms, and the measured CLs were 144 and 146 ms, respectively. Conversely, Figure 5B shows inaccurate bipolar CL estimates due to fractionation that likely reflects noise on the MAPs (arrow). The bipolar-based DF was 10.3 Hz, vs. a MAP-based value 4.95 Hz, yielding CL estimates of 97 ms (vs. measured CL 205 ms) and 202 ms (vs. measured CL 201 ms). The bipolar and MAP spectral CL estimates are summarized in Figure 6.
|
|
Bipolar spectral cycle length estimates may be inaccurate even if spectra are narrow
Bipolar spectral CL estimates could be inaccurate even if the spectra were narrow (Figure 6A). However, in the MAP spectra, accurate CL estimates (
20 ms error) had narrower DF (peak area ratio 0.50 ± 0.16) than the (inaccurate) CL estimates (peak area ratio 0.37 ± 0.14; P = 0.001; Figure 6B). In general, the autocorrelation-based CL estimates were more accurate than the spectral-based DF, both for bipolar (Figure 6C) and MAP (Figure 6D) CLs, as also shown in Figures 4 and 5. |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
We used MAPs to quantify organization in human AF. Notably, fractionated electrograms in MAPs were often inseparable from noise or other artefact in bipolar signals. Conversely, organized fluctuations such as MAP alternans, with possible mechanistic significance to AF,7
were often obscured in bipolar signals. As a result, spectral DF estimated AF CL more accurately in MAPs than in bipolar electrograms. Further work is required to determine how assessing fibrillatory wave shape may improve AF substrate mapping and ablation outcome in AF.
Fibrillatory MAP vs. bipolar signals
Monophasic action potential recordings represent the time courses of tissue depolarization and repolarization as distinct events even in AF, unlike bipolar signals. Monophasic action potentials also exhibit an intimate relation between repolarization and refractoriness (point of re-excitability).20 Although local repolarization as a function of time is slower than depolarization, typical bipolar electrogram filtering largely obscures repolarization. Unipolar electrograms also reflect local waveform shape, yet the MAP is more localized10 and may be better suited to map AF substrates for ablation.
We observed surprising stability in MAP shape and amplitude in many AF recordings (Figures 1, 3B and D, and 5A and B) that may reflect organized wavelets, which, however, were not evident in the bipolar recordings. Conversely, the MAP signals revealed AF variability (Figures 1A and 3D), including alternans (Figure 3D) that may precede wavefront fractionation,7 which was not seen in bipolar electrograms.
Complex fractionated electrograms in atrial fibrillation
The presence of fractionation correlated poorly between bipolar and MAP electrograms (Figures 3 and 4), largely because bipolar signals less effectively separated fractionation from artefact (Figures 4 and 6).
Substrate mapping currently groups sites of complex fractionated electrograms (CFAE) and rapid CL as ablation targets.1,2 Our studies using MAP morphology support this grouping, since fractionated electrograms co-migrated with shorter AF CL as did a seminal study using unipolar electrograms.8 Conversely, spectral indices of organization (narrow vs. broad spectra) did not predict shorter CL during AF.
Cl estimates using spectral or autocorrelation methods
Mounting evidence emphasizes the importance of AF CL measured manually15,21 or from spectral DF3 for guiding ablation. Manual CL measurement is tedious, yet spectral DF of bipolar electrograms may be a poor surrogate (Figures 4 and 6). The limitations of spectral DF CL estimates have recently been reported.6,22 However, few studies have compared23 manual15,21 with spectral3,4 CL estimates. Further work should determine whether the correlation with ablation outcome is improved if AF CL is estimated using time-domain methods (either manual or via autocorrelation).
Double counting occurred in dedicated and derived bipolar spectra (Figure 1A and B), even when electrograms appeared regular (Figure 1B). This likely reflects the sequence of irregularity within the electrogram analysed by FFT.6 Recent studies minimized double counting using a 3–15 Hz bandpass filter,3 yet our results show that peaks from double counting may lie within this 3–15 Hz bandpass. Previous studies have also increased the signal-to-noise ratio by analysing only ‘organized’ segments with narrow spectral DF.3 However, in our results, narrow DF predicted accurate CL estimates only in MAP spectra, but not in bipolar spectra. One simple method of avoiding double counting may be to manually verify any estimated CL 100 ms (i.e. 10 Hz).
In addition, this study further validates autocorrelation as a time method to estimate AF CL. We recently used autocorrelation to estimate AF CL while examining the time course AF fluctuations.22 In addition to its avoidance of double counting and harmonics, autocorrelation has a temporal resolution <1 ms, whereas spectra typically introduce errors >3–5 ms (based on 0.12 Hz resolution at 5–8 Hz).
Clinical implications: accuracy of fractionation and fibrillatory rate
The present study suggests that bipolar CFAE may not always reliably indicate local fractionation in AF.1,2 Further work should thus define whether targeting electrogram morphology, using an MAP or unipolar signals, improves AF ablation outcome. Our results also suggest caution in using spectral DF of bipolar signals for automated AF CL estimates. Since very short CL estimates may reflect double counting, one simple ‘check’ may be to verify any CL 100 ms (DF 10 Hz) manually or by autocorrelation.
Limitations
We recorded only from the RA that is less critical to AF maintenance than the left atrium (LA).24 However, our aim was to apply quantification methods to MAP vs. bipolar signals that do not require LA recordings. Second, we analysed MAPs for only 2 s, although longer epochs may be required, as reported previously.4
A third limitation of the present study is that the bipolar electrograms were derived from MAPs. However, this method allowed us to compare signal characteristics at exactly the same myocardial site. Moreover, MAP recordings are also bipolar—obtained from tip-to-nearby electrode subtraction. The primary difference is that the MAP side-pole is not in contact with tissue, although clinical bipoles are occasionally recorded in this orientation if the catheter lies perpendicular to tissue.
Finally, we studied older, male patients with a high prevalence of left ventricular dysfunction, reflecting our Veterans Affairs population. Thus, our results should be reconfirmed in younger and female patients with preserved left ventricular function.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
In human AF, MAPs portrayed local electrogram morphology, and thus separated fractionation from far-field artefacts or noise that were indistinguishable in bipolar signals. As a result, AF CL estimates were more accurate using MAPs than bipolar signals. Assessments of local AF electrogram morphology may help in mapping AF substrates during ablation.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
This work was supported, in part, by grants from the United States National Institutes of Health (K23 HL70529) to S.M.N.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
We thank Murat Celebi, MD, for retrieving the digital data for our analysis and for helping to collate patient demographics and Paul Clopton, MS, for help with the statistical analyses.
Conflict of interest: none declared.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsFundingAcknowledgementsReferences |
|---|
[1] Nademanee K, McKenzie J, Kosar E, et al. A new approach for catheter ablation of atrial fibrillation: mapping of the electrophysiologic substrate. J Am Coll Cardiol (2004) 43:2044–53.
[2] Oral H, Chugh A, Lemola K, et al. Noninducibility of atrial fibrillation as an end point of left atrial circumferential ablation for paroxysmal atrial fibrillation: a randomized study. Circulation (2004) 110:2797–801.
[3] Sanders P, Berenfeld O, Hocini M, et al. Spectral analysis identifies sites of high-frequency activity maintaining atrial fibrillation in humans. Circulation (2005) 112:789–97.
[4] Lazar S, Dixit S, Marchlinski FE, Callans DJ, Gerstenfeld EP. Presence of left-to-right atrial frequency gradient in paroxysmal but not persistent atrial fibrillation in humans. Circulation (2004) 110:3181–6.
[5] Wu T-J, Doshi RN, Huang H-LA, et al. Simultaneous biatrial computerized mapping during permanent atrial fibrillation in patients with organic heart disease. J Cardiovasc Electrophysiol (2002) 13:571–7.[CrossRef][Web of Science][Medline]
[6] Ng J, Kadish AH, Goldberger JJ. Effect of electrogram characteristics on the relationship of dominant frequency to atrial activation rate in atrial fibrillation. Heart Rhythm (2006) 3:1295–305.[CrossRef][Web of Science][Medline]
[7] Narayan SM, Bode F, Karasik PL, Franz MR. Alternans of atrial action potentials as a precursor of atrial fibrillation. Circulation (2002) 106:1968–73.
[8] Konings K, Smeets J, Penn O, Wellens H, Allessie M. Configuration of unipolar atrial electrograms during electrically induced atrial fibrillation in humans. Circulation (1997) 95:1231–41.
[9] Pappone C, Santinelli V, Manguso F, et al. Pulmonary vein denervation enhances long-term benefit after circumferential ablation for paroxysmal atrial fibrillation. Circulation (2004) 109:327–34.
[10] Franz MR, Chin MC, Sharkey HR, Griffin JC, Scheinman MM. A new, single-catheter technique for simultaneous measurement of action potential duration and refractory period in vivo. J Am Coll Cardiol (1990) 16:878–86.[Abstract]
[11] Franz MR, Karasik PL, Li C, Moubarak J, Chavez M. Electrical remodeling of the human atrium: similar effects in patients with chronic atrial fibrillation and atrial flutter. J Am Coll Cardiol (1997) 30:1785–92.[Abstract]
[12] Kim B-S, Kim Y-H, Hwang G-S, et al. Action potential duration restitution kinetics in human atrial fibrillation. J Am Coll Cardiol (2002) 39:1329–36.
[13] Franz MR. The role of monophasic action potential recordng in atrial fibrillation. In: Atrial Fibrillation: Mechanisms and Therapeutic Strategies—Olsson S, Alessie M, Campbell RW, eds. (1994) Armonk, NY: Futura Publishing Company. p109–25.
[14] Everett TH, Kok L-C, Vaughn RH, Moorman JR, Haines DE. Frequency domain algorithm for quantifying atrial fibrillation organization to increase defibrillation efficacy. IEEE Trans Biom Eng (2001) 48:969–78.[CrossRef]
[15] Haissaguerre M, Sanders P, Hocini M, et al. Changes in atrial fibrillation cycle length and inducibility during catheter ablation and their relation to outcome. Circulation (2004) 109:3007–13.
[16] Hoppe BL, Kahn AM, Feld GK, Hassankhani A, Narayan SM. Separating atrial flutter from atrial fibrillation with apparent ECG organization using dominant and narrow F-wave spectra. J Am Coll Cardiol (2005) 46:2079–87.
[17] Narayan SM, Feld GK, Hassankhani A, Bhargava V. Quantifying intra-cardiac organization of atrial arrhythmias using temporospatial phase of the electrocardiogram. J Cardiovasc Electrophysiol (2003) 14:971–81.[CrossRef][Medline]
[18] Narayan SM, Hassankhani A, Feld GK, Bhargava V. Separating non-isthmus from isthmus dependent atrial flutter using wavefront variability. J Am Coll Cardiol (2005) 45:1269–79.
[19] Brown JP, Krummen DE, Feld GK, Narayan SM. Using electrocardiographic activation time and diastolic intervals to separate focal from macroreentrant atrial tachycardias. J Am Coll Cardiol (2007) 49:1965–73.
[20] Bode F, Kilborn M, Karasik P, Franz MR. The repolarization-excitability relationship in the human right atrium is unaffected by cycle length, recording site and prior arrhythmias. J Am Coll Cardiol (2001) 37:920–5.
[21] Haissaguerre M, Sanders P, Hocini M, et al. Catheter ablation of long-lasting persistent atrial fibrillation: critical structures for termination. J Cardiovasc Electrophysiol (2005) 16:1125–37.[CrossRef][Web of Science][Medline]
[22] Narayan SM, Krummen DE, Kahn AM, Karasik PL, Franz MR. Evaluating fluctuations in human atrial fibrillatory cycle length using monophasic action potentials. Pacing Clin Electrophysiol (2006) 29:1209–18.[CrossRef][Medline]
[23] Sahadevan J, Ryu K, Peltz L, et al. Epicardial mapping of chronic atrial fibrillation in patients: preliminary observations. Circulation (2004) 110:3293–9.
[24] Ravelli F, Faes L, Sandrini L, et al. Wave similarity mapping shows the spatiotemporal distribution of fibrillatory wave complexity in the human right atrium during paroxysmal and chronic atrial fibrillation. J Cardiovasc Electrophysiol (2005) 16:1071–6.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||