Validating Left Atrial Low Voltage Areas During Atrial Fibrillation and Atrial Flutter Using Multielectrode Automated Electroanatomic Mapping

JACC: CLINICAL ELECTROPHYSIOLOGY

VOL.

-, NO. -, 2018

ª 2018 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER

Validating Left Atrial Low Voltage Areas During Atrial Fibrillation and Atrial Flutter Using Multielectrode Automated Electroanatomic Mapping Moisés Rodríguez-Mañero, MD, PHD,a,b Miguel Valderrábano, MD,c Aurora Baluja, MD, PHD,d Omar Kreidieh, MD,e Jose Luis Martínez-Sande, MD, PHD,a,b Javier García-Seara, MD, PHD,a,b Johan Saenen, MD, PHD,f Diego Iglesias-Álvarez, MD,a,b Wim Bories,f Luis Miguel Villamayor-Blanco,a María Pereira-Vázquez,a Ricardo Lage, MD, PHD,a,b Julián Álvarez-Escudero, MD, PHD,d Hein Heidbuchel, MD, PHD,f José Ramón González-Juanatey, MD, PHD,a,b Andrea Sarkozy, MD, PHDf

ABSTRACT OBJECTIVES This study aimed: 1) to determine the voltage correlation between sinus rhythm (SR) and atrial fibrillation (AF)/atrial flutter (AFL) using multielectrode fast automated mapping; 2) to identify a bipolar voltage cutoff for scar and/ or low voltage areas (LVAs); and 3) to examine the reproducibility of voltage mapping in AF. BACKGROUND It is unclear if bipolar voltage cutoffs should be adjusted depending on the rhythm and/or area being mapped. METHODS High-density mapping was performed first in SR and afterward in induced AF/AFL. In some patients, 2 maps were performed during AF. Maps were combined to create a new one. Points of <1 mm difference were analyzed. Correlation was explored with scatterplots and agreement analysis was assessed with Bland-Altman plots. The generalized additive model was also applied. RESULTS A total of 2,002 paired-points were obtained. A cutoff of 0.35 mV in AFL predicted a sinus voltage of 0.5 mV (95% confidence interval [CI]: 0.12 to 2.02) and of 0.24 mV in AF (95% CI: 0.11 to 2.18; specificity [SP]: 0.94 and 0.96; sensitivity [SE]: 0.85 and 0.75, respectively). When generalized additive models were used, a cutoff of 0.38 mV was used for AFL for predicting a minimum value of 0.5 mV in SR (95% CI: 0.5 to 1.6; SP: 0.94, SE: 0.88) and of 0.31 mV for AF (95% CI: 0.5 to 1.2; SP: 0.95, SE: 0.82). With regard to AF maps, there was no change in the classification of any left atrial region other than the roof. CONCLUSIONS It is possible to establish new cutoffs for AFL and/or AF with acceptable validity in predicting a sinus voltage of <0.5 mV. Multielectrode fast automated mapping in AFL and/or AF seems to be reliable and reproducible when classifying LVAs. These observations have clinical implications for left atrial voltage distribution and in procedures in which scar distribution is used to guide pulmonary vein isolation and/or re-isolation. (J Am Coll Cardiol EP 2018;-:-–-) © 2018 by the American College of Cardiology Foundation.

From the aCardiology Department, Hospital Universitario Santiago de Compostela, Santiago de Compostela, IDIS, Spain; bCentro de Investigación Biomédica en Red de Enfermedades Cardiovasculares (CIBERCV CB16/11/00226 - CB16/11/00420), Madrid, Spain; c

Division of Cardiac Electrophysiology, Department of Cardiology Houston Methodist Hospital, Houston, Texas; dCritical Patient

Translational Research Group, Department of Anesthesiology, Intensive Care and Pain Management, Hospital Clínico Universitario, Santiago de Compostela, Spain; eCardiology Department, Newark Beth Israel Medical Center, Newark, New Jersey; and the fCardiology Department, Cardiac Electrophysiology Section, University Hospital of Antwerp, Antwerp, Belgium. Dr. Valderrábano has received research support from Biosense Webster. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page. Manuscript received May 18, 2018; revised manuscript received July 17, 2018, accepted August 16, 2018.

ISSN 2405-500X/$36.00

https://doi.org/10.1016/j.jacep.2018.08.015

2

Rodríguez-Mañero et al.

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

T

ABBREVIATIONS AND ACRONYMS AF = atrial fibrillation

he pulmonary veins (PVs) are impor-

patient records. All participants provided written,

tant trigger sites of paroxysmal atrial

informed consent for both the ablation procedure and

fibrillation (AF), and their electrical

inclusion

isolation from the left atrium (LA) is associ-

AFL = atrial flutter

-, NO. -, 2018 - 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

in

medical

research

at

the

time

of

procedure.

ated with a high rate of freedom from AF

CI = confidence interval

(1,2). In persistent AF, pulmonary vein isola-

HIGH-DENSITY ATRIAL MAPPING. Patients with AF

LA = left atrium

tion (PVI) is less successful (3), possibly

who underwent first or repeat ablation also under-

LVA = low voltage area

because additional arrhythmogenic atrial

went sequential mapping of the LA using a PentaRay

PV = pulmonary vein

sites beyond the PVs are responsible for AF

catheter (Biosense Webster, Diamond Bar, California),

PVI = pulmonary vein isolation

maintenance. Atrial fibrosis and scar tissue

which contains 5 splines with 4 electrodes each (1 mm,

SR = sinus rhythm

serve as an important substrate for focal

spaced 2-6-2 mm apart). In SR, atrial electrograms

and re-entrant activity involved in persistent

were captured by setting the window of interest

AF (4–7). Therefore, electroanatomic mapping has

from
50 to
350 ms, preceding the sharp component

been suggested to delineate areas of scar tissue for

of each ventricular QRS complex as a reference. High-

targeting. Areas with low-voltage amplitude on map-

density bipolar voltage mapping of the LA was per-

ping correspond to unexcitable tissue and delayed

formed with an equal distribution of points using a fill

gadolinium enhancement on magnetic resonance

threshold of 5 to 7 mm, with a minimum of 1,000

imaging

Therefore,

points in each LA map. Low-voltage zones were

voltage-guided AF substrate modification by ablation

defined as 0.1 to 0.49 mV (peak-to-peak bipolar

targeting low-voltage areas (LVAs) has been per-

voltage) and transitional zones were considered $0.5

formed to improve long-term AF ablation efficacy

to 1.4 mV. If the patient was in AF, the patient was

(13–20). Most of these procedures have performed

cardioverted at the beginning of the study to obtain

mapping using established cutoffs for voltage during

the first map in SR (hence, avoiding map displacement

sinus rhythm (SR), and 1 study performed mapping

after the cardioversion). Afterward, AF and/or AFL

during AF (15).

was induced by atrial burst pacing from the distal or

in

multiple

studies

(8–12).

There remain multiple unresolved issues with

mid-coronary sinus at a cycle of 250 to 180 ms. In those

LVA-guided substrate modification. Although some

patients in whom the 2 maps were performed during

studies have showed that measured voltages are

AF, a 10-min waiting period was performed (2 ob-

higher during SR than during AF, none have studied

tained maps called AF1 and AF2).

voltages in other atrial arrhythmias, such as atrial

In addition, high-density mapping was added at

flutter (AFL) (11,21). Furthermore, it is unclear if bi-

sites where LVAs were recorded to exactly delineate

polar voltage cutoffs should be adjusted depending

the extent of the LVA. Stability was selected at 5 mm.

on the rhythm, catheter type (22) (electrode size and

To avoid poor contact points, we set the interior and

interelectrode distance), and/or anatomic area being

exterior projection distance filtering to 5 mm from the

mapped. Most studies have used cutoffs in SR using

geometry surface. In SR, points that did not conform

an ablation catheter (3.5-mm tip), but the corre-

to the surface electrocardiogram P-wave morphology

sponding thresholds in AF are not known. Finally,

or 75% of the maximum voltage of the preceding

there are no data regarding the reproducibility of LVA

electrogram were excluded. Signals were filtered at

mapping in AF.

30 to 400 Hz and displayed at 100 mm/s. The elec-

We aimed to: 1) determine the bipolar voltage

trogram at each point acquired on the LA shell was

cutoff for scar and/or LVAs using small sized, closely

manually reviewed to exclude noise or a pacing

spaced

mapping

artefact before being accepted. We included 7 seg-

(MFAM) during AF and AFL that corresponds to a si-

ments (septum, anterior wall, floor inferior, lateral

nus bipolar voltage of 0.5 mV; 2) determine the

wall, posterior wall, roof, and PVs). Each LA
PV

multielectrode

fast

automated

voltage correlation between SR and AF and/or AFL;

junction was defined as the region that extended 5

and 3) to examine the reproducibility of LA voltage

mm proximal to the PV ostia circumferentially and

mapping in AF.

that documented an impedance rise of 10 ohms compared with the LA.

METHODS

Two separate LA shells (bipolar voltage maps) were created for each patient in 2 different clinical rhythms

This was a multicenter prospective study, performed

or in AF at different times. To compare local bipolar

in 3 hospitals with experience in the field of AF

voltages from each site in different rhythms, each

mapping and ablation. Patient demographics, clinical

map was tagged in a separate color before being

characteristics, and medications were exported from

combined on a new map (Figure 1). Complete

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018

Rodríguez-Mañero et al.

- 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

F I G U R E 1 Schematic Demonstrating How Maps Were Superimposed

Schematic demonstrating how maps were superimposed on each other to allow for comparison of points in corresponding regions during different times and/or rhythms. (A.1, B.1, and C.1) The first map (posterior, anterior, and inferior views, respectively) in sinus rhythm. Subsequently, after atrial fibrillation and/or atrial flutter induction, another map is taken in atrial fibrillation and/or atrial flutter (A.2, B.2, and C.2, same projections). In D1 (sinus rhythm) and in D2 (atrial fibrillation), multielectrode fast automated electroanatomic maps have been made transparent. The second map performed in atrial fibrillation is copied and merged with the first one in sinus rhythm (D3). Afterwards, careful point-by-point inspection and voltage comparison is done (yellow dots).

transparency was used to manually review points

The bias and mean limits of agreement from the

obtained in the same location on each mapping pro-

Bland-Altman-Tukey analyses were used to predict

cedure and/or LA shell. Only those point pairs with a

the SR voltage that would be equivalent to a new

distance of <1 mm between them were analyzed.

simulated set of AFL and AF voltages. A peak-to-peak

Once the 2 points were compared, they were tagged in

voltage of 0.5 mV in SR was defined as a reference

a different color to avoid duplication.

threshold value that defined scarred tissue (10–12,23).

Following the maps, PVI was performed with

The AFL and/or AF thresholds were defined as the

circumferential ablation around both ipsilateral PVs

peak-to-peak voltage equivalent to a sinus amplitude

using a dragging technique.

of 0.5 mV and were set to be the corresponding values when the predicted sinus value matched 0.5 mV.

STATISTICAL ANALYSIS. Numerical data were tested

To evaluate the performance of this method, pre-

for normality using the Shapiro-Wilk test, and for

dicted SR values were obtained from the measured

homoskedasticity with Levene’s test, then summa-

AFL and/or AF values, and then compared with the

rized with mean, median, SD, and interquartile range

actual sinus voltages. Using the proportion of pre-

values where appropriate. Voltages by rhythm and

dicted sinus values that fell into the same category

region were represented by a ridgeline plot shown in

(either <0.5 mV or $0.5 mV) as their measured sinus

Figure 2, ordered by median voltage. Correlation was

counterparts, we were able to calculate sensitivity,

explored with scatterplots and agreement analysis

specificity, false positive and negative ratios, and

with Bland-Altman (Tukey mean difference) plots.

accuracy.

3

4

Rodríguez-Mañero et al.

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018 - 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

F I G U R E 2 Ridgeline Plots Showing Regional Bipolar Voltage Distribution in AF

Ridgeline plots showing regional bipolar voltage distribution in atrial fibrillation (AF) in (A) sinus rhythm, (B) atrial flutter, (C) AF1, and (D) AF2.

To find the regional AFL and/or AF peak-to-peak

given predicted mean sinus voltage. To evaluate the

voltages equivalent to 0.5 mV in SR, we adjusted a

performance of this method, we predicted SR volt-

generalized additive model (GAM), taking into account

ages using the generalized additive model, and then

the patient as an independent random effect. Flexible

compared them with the actual sinus voltages. Using

penalized splines were used to model the continuous

the proportion of predicted sinus values that fell into

covariate with the following formulas (simplified):

the same category (either <0.5 or $0.5mV) as their

ðPredicted Sinus mVÞ z ðmeasured Flutter mVÞ þ ðRegionÞ þ ðPatient random effectÞ ðPredicted Sinus mVÞ z ðmeasured Atrial Fibrillation mVÞ

measured sinus counterparts, we were able to calculate sensitivity, specificity, false positive, negative ratios, and accuracy. Estimates of thresholds were calculated both for the global model and for every region. All analyses were performed in R (R Core Team,

þ ðRegionÞ

Vienna, Austria), using the packages ggplot (24),

þ ðPatient random effectÞ:

dplyr (Wickham 2017), ggridges (Wilke 2017), BlandAltmanLeh (Lehnert, 2015), and mgcv (24).

Similar to the prediction that used the BlandAltman analyses, we calculated the threshold in

RESULTS

either AFL or AF voltages that corresponded to a minimal sinus voltage of 0.5 mV when using the

PATIENT AND MAP CHARACTERISTICS. A total of 31

lower band of the 95% confidence interval (CI) for a

patients were included in the study. Twenty-one of

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018

Rodríguez-Mañero et al.

- 2018:-–-

them underwent repeat procedures, and the rest were de novo procedures. Mean age was 67  12 years; 21

Identification of Low-Voltage Areas During Atrial Fibrillation

T A B L E 1 Mean Regional Left Atrial Voltage Distribution in Sinus Rhythm,

Atrial Flutter, AF 1, and AF 2

patients were men (68%), 21 had nonparoxysmal AF (68%), and the mean LA diameter was 43 mm. All patients underwent sequential maps in SR and AF. Ten patients also underwent mapping during AFL. A

Region

Sinus (mV)

Atrial Flutter (mV)

Posterior

2.45  2.16

1.52  1.26

AF 1 (mV)

AF 2 (mV)

Anterior

2.27  1.87

1.78  1.4

1.29  2.04

1.37  1.23

Roof

2.17  1.91

1.52  1.24

1.45  4.1

1.85  6.14 0.82  0.65

1.2  0.95

1.26  0.76

total of 2,002 paired points (4,004 points) were ob-

Septum

1.64  1.74

0.83  0.79

0.82  0.63

tained from the 31 patients’ maps: 962 paired points

Lateral

2.26  2.3

0.96  1.07

1.11  1.03

1  0.84

(48.1%) that compared sinus SR versus AF, 545

Inferior

2.39  1.95

1.77  1.8

0.9  0.54

1.24  0.64

(27.2%) that compared SR versus AFL, and finally 495

Appendage

4.78  2.59

3.47  1.92

2.96  2.04

2.68  1.36

(24.7%) that compared AF versus AF. Online Table 1

Veins

0.28  0.38

0.3  0.61

0.26  0.36

0.34  0.39

describes the number of points and their locations. Briefly, the most sampled region corresponded to the

Values are mean  SD. The same map is used within 10-min difference. AF ¼ atrial fibrillation.

posterior wall, with 609 (30.4%) paired points. The LA appendage was the least sampled to minimize the risks of the procedure (63 paired points; 3.1%).

VOLTAGE CORRELATION BETWEEN SR-AF AND/OR

GLOBAL VALUES. Mean regional LA voltage distri-

AFL

bution in SR, AFL, AF1, and AF2 are shown in Table 1.

LVA. As previously explained, 2 different statistical

Values are significantly higher in SR than in AFL and

analyses were performed to define a cutoff value be-

AF. Moreover, we observed distinct regional differ-

tween 1) AFL and SR, and 2) AF and SR.

WITH

A

POTENTIAL

CUTOFF

VALUE

FOR

ences in voltage values even when patients were in the same rhythm. Table 2 and Figure 3 show regional differences for each rhythm. The Kendall rank correlation coefficient (Kendall’s tau coefficient), which

T A B L E 2 Regional Differences (in Voltage Map) Obtained in

Sinus Rhythm, Atrial Flutter, and Atrial Fibrillation (AF)

was used to measure the ordinal association between 2 measured quantities, is shown in Figures 3A, 3C, and 3E. Although the Kendall’s tau coefficient is modest,

Region

Sinus

Anterior: appendage

<0.001

Atrial Flutter

0.001

AF

<0.001

as seen graphically, it could be due to greater varia-

Appendage: inferior

<0.001

<0.001

<0.001

Appendage: lateral

<0.001

<0.001

<0.001

tion with the higher voltages. It seems to show a good

Appendage: posterior

<0.001

0.23

<0.001

relationship for values <0.5 mV, but this is lost at

Appendage: roof

<0.001

<0.001

<0.001

higher voltages. Figures 3B, 3D, and 3F show a graphic

Appendage: septum

<0.001

<0.001

<0.001

comparison of regional bipolar voltage correlation.

Anterior: veins

<0.001

<0.001

<0.001

Because bipolar voltages within the PVs are low during SR and AF, in we offer additional detail in Online Figure 1 into the correlation pattern between

Appendage: veins

<0.001

<0.001

<0.001

Inferior: veins

<0.001

<0.001

<0.001

Lateral: veins

<0.001

<0.001

<0.001

Posterior: veins

<0.001

<0.001

<0.001

sinus (A1), AFL (A2), and AF (A3) in the PV region.

Roof: veins

<0.001

<0.001

<0.001

Graphs corresponding to the correlation between si-

Septum: veins

<0.001

<0.001

<0.001 <0.001

nus and AF voltages, scaled at 0 to 0.25, 0 to 0.5, and

Posterior: septum

<0.001

<0.001

0.5 to 1.5, are shown in Online Figures 1B1 to 1B3.

Inferior: septum

<0.001

0.002

Online Figures 1B1 to 1B3 also shows grouped volt-

Anterior: septum

<0.001

<0.001

<0.001

ages, whereas Figures 2A to 2C show voltages by re-

Roof: septum

0.02

0.008

0.008

Lateral: septum

0.07

0.86

0.12

Inferior: lateral

0.31

0.03

0.87

Lateral: posterior

0.37

0.03

0.13

gion. In Online Figures 2B and 2C, data are insufficient to correctly draw 95% CI bands for the

0.09

local regressions. In the appendage region, there was

Anterior: lateral

0.43

0.006

0.11

only 1 point <0.5, and there were no voltages in the

Inferior: roof

0.47

0.95

0.27

0.5 and 1.5 ranges. The positive correlation between

Posterior: roof

0.57

0.88

0.51

both voltages was maintained across all ranges.

Anterior: roof

0.63

0.51

0.44

However, the dispersion also consistently increased

Inferior: posterior

0.76

0.82

0.02

Lateral: roof

0.77

0.05

0.45

Anterior: inferior

0.79

0.48

0.02

Anterior: posterior

0.09

0.24

0.81

with the increase in voltages, making log transformation of the data necessary to correct this finding.

5

6

Rodríguez-Mañero et al. Identification of Low-Voltage Areas During Atrial Fibrillation

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018 - 2018:-–-

F I G U R E 3 Linear and Graphic Bipolar Voltage Correlation Between the Different Rhythms

(A) Linear bipolar voltage correlation between atrial flutter and sinus rhythm. (B) Comparison of regional bipolar voltage correlation. (C) Linear bipolar voltage correlation between AF and sinus rhythm. (D) Comparison of regional bipolar voltage correlation. (E) Linear bipolar voltage correlation between AF and the same patient with a second map in AF. (F) Comparison of regional bipolar voltage correlation. Additional lines are added in all figures at 0.5 mV. Abbreviation as in Figure 2.

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018

Rodríguez-Mañero et al.

- 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

F I G U R E 4 Bland-Altman Plots

Bland-Altman plots showing the agreement between voltages in sinus rhythm and either (A) AF or (B) atrial flutter, original scale. (C and D) Natural logarithmic scale is shown. Abbreviation as in Figure 2.

After correlation and Bland-Altman analyses, the

(95% CI: 0.11 to 2.18) (negative predicted value: 0.94,

hetero-

positive predicted value: 0.85, specificity: 0.94,

skedastic funnel-like structure (Figure 4), which was

sensitivity: 0.85, and accuracy: 0.92; negative pre-

corrected after natural-logarithm transformation of

dicted value: 0.93, positive predicted value: 0.83,

the data, followed by exponentiation of the final es-

specificity: 0.96, sensitivity: 0.75, and accuracy: 0.91,

timates. Final voltage thresholds and their prediction

respectively).

corresponding

plots

showed

a

heavily

assessment values were calculated according to the

In contrast, if generalized additive models were

methods described by Bland-Altman (25) and Euser

used, the best model gave rise to a cutoff of 0.31 mV

et al. (26) for logarithm transformed data, and are

in AF for predicting a minimum value of 0.5 mV in SR

shown in their respective tables.

(0.5 to 1.2), with a specificity of 0.95 and a sensitivity

In Figure 5 and Online Table 2, using Bland-

of 0.82. For AFL, a value of 0.38 mV predicted a

Altman model, we showed the predicted voltage for

minimum value of 0.5 mV in SR of 0.55 (0.5 to 1.6),

SR (with its 95% CI) for each specific point obtained

with a specificity of 0.94 and a sensitivity of 0.88.

in AFL and in AF. As shown, 0.35 mV in AFL pre-

Negative and positive predictive values, and accuracy

dicted a sinus voltage of 0.5 (95% CI: 0.12 to 2.02),

are shown in Table 3. This correlation can be seen in a

and 0.25 mV in AF predicted a sinus voltage of 0.5

dynamic way in the MASH-AF 1 study (27).

7

Rodríguez-Mañero et al.

8

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018 - 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

F I G U R E 5 Predicted Voltage in Sinus Rhythm

A

B

Predicted voltage in sinus rhythm (mean, minimum, and maximum) for each value obtained in (A) atrial flutter and (B) AF. As shown, 0.35 mV in atrial flutter and 0.25 mV in AF would correspond with 0.5 mV in sinus rhythm (95% confidence intervals: 0.12 to 2.02 and 0.11 to 2.18, respectively). Abbreviation as in Figure 2.

RELIABILITY OF LA VOLTAGE MAP IN AF. As previ-

normal tissue (1.45 to 4.1 to 1.85 to 6.14). Further-

ously mentioned, there were differences in mean

more, when we examined the 545 points for concor-

regional LA voltage between AF1 and AF2. The Ken-

dance in classifying a segment as LVA or not (<0.5 mV

dall’s tau coefficient is shown in Figure 2D. However,

or $0.5 mV), 24 points were $0.5 mV in AF1 and <0.5

although there are significant differences between

in AF2 map, and 26 points were vice versa. Concor-

the 2 AF maps, these values were clinically irrelevant;

dance was far more prevalent with 93 points

if we used the conventional division of low-voltage

being <0.5 mV in both maps and 352 points $0.5 mV

zones (0.1 to 0.49 mV), transitional zones (0.5 to

in AF1 and AF2.

1.49 mV), and healthy areas (1.5), there was no change

Finally, the mean value matched well with the

in the classification of any LA region other than the

value obtained in the first map, although the CI

roof, which changed from a transitional zone to

showed a wide distribution (Online Table 2C). For

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018

Rodríguez-Mañero et al.

- 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

T A B L E 3 Global and Regional Final Generalized Additive Models

Voltage (mV)

Min <0.5

Mean <0.5

Max <0.5

NPV

PPV

Specificity

Sensitivity

Accuracy

Overall Sinus-AF

0.36

0.5

1,23

1.95

0.94

0.86

0.96

0.8

0.93

Sinus-Atrial flutter

0.26

0.5

1.02

1.55

0.94

0.84

0.94

0.85

0.92

Sinus-AF Appendage

0.15

0.5

1.23

1.95

1.0

1.0

1.0

1.0

1.0

Anterior

0.29

0.5

0.81

1.11

0.97

0.88

0.99

0.75

0.96

Posterior

0.16

0.5

0.79

1.07

0.93

1.0

1.0

0.61

0.94

Inferior

0.17

0.5

0.84

1.19

0.95

0.82

0.98

0.56

0.94

Roof

0.29

0.5

0.95

1.41

0.96

0.75

0.94

0.82

0.92

Veins

0.36

0.5

0.86

1.23

0.71

0.94

0.67

0.95

0.91

Septum

0.29

0.5

0.83

1.15

0.96

0.60

0.88

0.82

0.87

Lateral

0.25

0.5

0.96

1.41

0.92

0.64

0.92

0.64

0.87

Sinus-Atrial flutter Appendage Anterior

NA

NA

NA

NA

1.0

NA

1.0

NA

1.0

0.48

0.5

0.87

1.25

0.98

0.78

0.96

0.88

0.95

Lateral

0.12

0.5

1.02

1.55

0.91

1.0

1.0

0.67

0.93

Veins

0.49

0.5

0.85

1.21

0.92

0.93

0.69

0.98

0.93

Posterior

0.22

0.5

0.76

1.03

0.93

0.86

0.97

0.71

0.92

Inferior

0.26

0.5

0.85

1.20

0.91

0.88

0.98

0.54

0.90

Roof

0.43

0.5

0.94

1.38

1.0

0.38

0.86

1.0

0.88

Septum

0.49

0.5

0.85

1.20

0.94

0.70

0.83

0.89

0.85

For the predictive global capacity of the suggested cutoff with their respective negative (NPV) and positive predictive value (PPV), specificity, sensitivity and accuracy. Other abbreviation as in Table 1.

instance, 2.15 mV in AF1 corresponded to a mean

ATRIAL FIBROSIS AND AF. Atrial fibrosis has been

value of 2.27 mV in the second map, but with a 95% CI

observed in greater frequency in patients with AF

of 0.498 to 9.286.

(5,6). At the same time, evidence for extensive fibrosis is also associated with a longer standing his-

DISCUSSION

tory of the arrhythmia as well as a lower success rate for PVI (3). These findings suggest a possible mecha-

This was a prospective study that examined differ-

nistic relationship between atrial fibrosis and/or

ences in LA voltage between SR and AF or AFL in

scarring and AF. Although reparative fibrosis has been

the same patient using point-by-point comparison

suggested to replace electrically active conductive

by high-density, closely spaced small electrode bi-

tissue, thus causing anisotropy and enhancing re-

polar voltage mapping. We also assessed the LA

entry, interstitial fibrosis may be more electrically

voltage consistency in patients with AF. Our main

inert (28). However, substrate manipulation in areas

findings were: 1) there were significant differences

of fibrosis has been suggested to help improve out-

in global and regional voltage distribution when we

comes for selected patients (29). Wang et al. (17)

compared different areas during the same rhythm

performed a randomized controlled trial of substrate

or different rhythms in the same area; 2) despite a

modification versus a more traditional stepwise

difference in voltage between rhythms, it was

ablation approach for patients with long-standing

possible to establish new cutoffs for AF and AFL

persistent AF. The authors found improved effec-

with

tiveness after a first procedure in the substrate

acceptable

validity

in

predicting

a

sinus

voltage <0.5 mV; and 3) multielectrode fast auto-

modification groups,

mated electroanatomic bipolar mapping with closely

recurrence rate, and a shorter procedure time, but the

a lower atrial tachycardia

spaced (2 mm) small electrodes (1 mm) in AF seems

benefits faded after repeat procedures. Lin et al. (9)

to be reliable and reproducible when classifying low

showed that rotors involved in AF maintenance

voltage zones.

exhibited lower voltage than sites without rotors

These observations had clinical implications for LA

(0.68  0.44 vs. 0.71  0.63). A meta-analysis of

voltage distribution, particularly in procedures in

similar studies showed a benefit of adding substrate

which scar distribution was used to guide PV isolation

modification targeting areas of scar identified by low

and/or re-isolation.

voltage compared with traditional PVI only (15,30).

9

10

Rodríguez-Mañero et al.

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018 - 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

It seems that absence of atrial low voltage may

noise levels in early electroanatomic mapping sys-

identify patients in whom PVI alone is likely to be

tems and was only later validated in imaging studies

sufficient, whereas the presence of atrial low voltage

(23,35). Therefore, the ideal clinical cutoff value was

may indicate scar tissue and potential alternative

arguably still unknown, but our study findings sug-

sources of the arrhythmia related to a slow con-

gested that such a threshold was likely to be

ducting substrate (29).

different depending on the rhythm during mapping.

IDENTIFYING FIBROTIC TISSUE BY MEANS OF ABNORMAL ELECTROCARDIOGRAM. Electrophysiologically,

atrial

fibrosis produces low-amplitude electrograms (10), electrogram fractionation, conduction heterogeneity, and manifests as abnormal signals that can be identified using electroanatomic mapping during SR (31,32). A 3-dimensional map may thus show areas of fibrosis to help guide diagnosis or ablation procedures. Several authors already showed that LVAs identified in SR on such electroanatomic maps of the atria correlated well with late gadolinium enhancement seen on cardiac magnetic resonance imaging (8,10,11,33,34). Nevertheless, some limitations to those studies should be highlighted. First, the use of current magnetic resonance imaging technology to diagnose atrial scarring has several limitations and is not well validated against histological evidence (30). Second, a low voltage on bipolar electrodes may be related to several factors independent of fibrosis, including direction of conduction vector, amount of contact, and electrode size and interelectrode distance characteristics (28). VOLTAGE

DIFFERENCES

Furthermore, only a few studies validated this conventional cutoff when it was used in atrial mapping during AF (7). Even more important, most clinical studies of substrate modification procedures performed mapping mostly during SR (15). Jadidi et al. (19) performed mapping in AF with different types of catheters and evaluated the effects of ablating at points with certain electrogram characteristics lying in or near LVAs after PVI. The authors successfully demonstrated

improved

effectiveness

compared

with a conventional PVI-only strategy for persistent AF. Although values changed slightly according to the statistical analysis used, we found that a 0.38mV cutoff in AFL and 0.31-mV cutoff in AF provided a good negative predicted value, positive predicted value, specificity, sensitivity, and accuracy for predicting a voltage of at least 0.5 mV in SR. Moreover, as indicated in Table 3, these cutoffs could be even further adjusted according to the sampled region. In the study by Yagashita et al. (11), the number of points with LVA in AF, using a bipolar voltage cutoff of 0.5 mV (4.0  2.4) (using a 3.5-mm tip ablation

DURING

DIFFERENT

catheter), was significantly higher than those used in

RHYTHMS. Our finding of higher voltages in SR than

SR using the same cutoff (1.9  2.1; p < 0.001). An

AF is in agreement with previous studies in the

adjusted bipolar voltage of 1.5 mV in SR produced a

literature. Yagashita et al. (11) found a linear voltage

similar area of fibrosis to that of 0.5 mV in AF (p ¼

correlation between SR and AF using an ablation

0.125). In our opinion, there are several limitations to

catheter (3.5-mm tip) (11), whereas Masuda et al. (21)

that approach for defining cutoffs. First, because a

found that the correlation was only present if elec-

sinus voltage of 0.5 mV is the most studied, we

trograms did not become fractionated during AF.

believe it is most important to define the thresholds

There was a possible mechanistic explanation for the

in AF and AFL that predict such a voltage in SR and

observation of progressively lower voltages in AFL

not the inverse. Second, the total area of low voltage

and AF compared with SR. During arrhythmias with

depends on the number of points sampled in each

shorter cycle lengths, a substantial amount of tissue

particular segment. Derivations based on the area of

might not depolarize when it is still refractory, or

low voltage may thus be biased by the interpolation

small pieces of underlying or neighboring tissues

created at the time of the electroanatomical map. In

might depolarize nonsimultaneously and in opposite

our study, we derived thresholds from an analysis

directions, which results in low bipolar voltage am-

based on a point-by-point comparison and were able

plitudes

to prove great accuracy for predicting a sinus

(21).

In

addition, bipolar

voltage

was

dependent on the direction of wavefront propaga-

voltage <0.5 mV. Our study also derived threshold

tion. In AFL, the more organized activation seen in

values for AFL and studied the reproducibility of

macro
re-entries compared with micro
re-entries in

LVAs in AF patients.

AF could give rise to less beat-to-beat variation and a more steady peak-to-peak deflection.

RELIABILITY OF MAPPING IN AF. Bipolar voltage is

measured in a single window as the maximum peak-

IDEAL CUTOFF FOR ABLATING. The bipolar voltage

to-peak voltage of 2 to 3 consecutive AF beats, and

of <0.5 mV cutoff was originally based on baseline

may be subject to temporal variation and quality of

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018

Rodríguez-Mañero et al.

- 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

contact. For this reason, we attempted to use a

In some patients, we might have underestimated

widened window of 400 ms. Remarkably, despite the

the voltage due to possible stunned myocardium. We

beat-to-beat variation seen in AF, we proved good

used CARTO3 for electroanatomic voltage mapping

reproducibility of the AF voltage maps and confirmed

and 1-mm electrodes with 2-mm interelectrode

the reliability of LVA mapping during this rhythm.

spacing; therefore, our findings might not be applicable to other mapping systems and catheters.

CLINICAL IMPLICATIONS. Voltage mapping in the LA

Because we do not currently perform pre-ablation

to guide AF ablation depends on the accuracy of each

cardiac magnetic resonance routinely on patients

mapping technique and the underlying rhythm. This

undergoing repeat AF ablation, we were unable to

is particularly important, because recent publications

make

that described success with substrate-based AF

mapping-derived scar versus LA scar as seen on car-

ablation approaches were highly dependent on ac-

diac magnetic resonance in this patient series.

comparisons

between

electroanatomic

curate identification and interpretation of LA scar (14–19). Our present study suggested that there might

CONCLUSIONS

be differences in ideal voltage thresholds according to the underlying rhythm. This must be taken into

There are further implications in non-AF studies as

account when mapping during nonsinus clinical

well. For example, “fine tune” scar recognition and

rhythms or when designing future studies on the

subsequent ablation is extremely important during

subject. For example, this is critical awareness of low

AFL ablation. In this situation, interruption of AFL

voltage distribution is critical for instance when

might be undesirable to avoid difficulty with inducing

conversion of AF to sinus rhythm fails or when it is

it afterwards. Hence, activation mapping and elec-

not desired to perform a cardioversion in those cases

troanatomical mapping, both performed during on-

when mechanism based AF mapping is attempted.

going AFL or AF, could help delineate the re-entry

We showed that mapping using an adjusted lower

or focal site, and the substrate without interrupting

LVA detection cutoff might produce valid and reli-

the rhythm.

able results that could identify areas with a voltage of <0.5 mV in SR. STUDY LIMITATIONS. There were several limitations

for our study. Our analysis could have been affected by undetected map shifts. We attempted to minimize this by performing well-distributed electroanatomic voltage mapping using CARTO3 system (Biosense Webster, Diamond Bar, California) and strict analysis, including only adjacent points that were <1 mm apart. Furthermore, we cardioverted patients with AF at the beginning of the study, so that the first mapping procedure was obtained in SR, thus avoiding any significant map shifts during cardioversion. We had nearly identical LA volumes in SR and AF maps that we believe add credence to the reliability of the mapping obtained. However, despite the fact of having nearly identical LA volumes in SR and AF maps, as it is well known that LA volume can be different between SR and AF because volume loading is different with different rhythms. We tried to overcome this limitation by taking stables references and comparing them along the course of the procedure (coronary sinus, His location, and the PV antrum) and by excluding those cases in which a cardioversion was needed. For all these reasons, this is the reason why it should not have significantly altered the conclusions of the present study.

ADDRESS

FOR

CORRESPONDENCE:

Dr.

Moisés

Rodríguez-Mañero, Cardiology Department, Complejo Hospital Universitario de Santiago, Santiago de Compostela, Spain, CIBERCV, Travesía da Choupana s/n, Santiago de Compostela, 15706 A Coruña, Spain. E-mail: [email protected]. PERSPECTIVES COMPETENCY IN MEDICAL KNOWLEDGE: Atrial fibrosis and/or scar tissue serves as an important substrate for focal and re-entrant activity. Electroanatomic mapping has been suggested to delineate areas of scar. However, there remain multiple unresolved issues, such as the voltage correlation between sinus and atrial AF or AFL and the reproducibility of voltage mapping in AF. TRANSLATIONAL OUTLOOK: It is possible to establish new cutoffs for AF and/or AFL with acceptable validity in predicting a sinus voltage <0.5 mV. Moreover, electroanatomic mapping in AF seems to be reliable and reproducible when classifying LVAs. Investigators are requested to consider the sufficiency of this data and to determine whether it has clinical implications in those procedures in which scar distribution is used to guide AF ablation.

11

12

Rodríguez-Mañero et al.

JACC: CLINICAL ELECTROPHYSIOLOGY VOL.

-, NO. -, 2018 - 2018:-–-

Identification of Low-Voltage Areas During Atrial Fibrillation

REFERENCES 1. Wilber DJ, Pappone C, Neuzil P, et al. Comparison of antiarrhythmic drug therapy and radiofrequency catheter ablation in patients with paroxysmal atrial fibrillation: a randomized controlled trial. JAMA 2010;303:333–40. 2. Haissaguerre M, Jais P, Shah DC, et al. Spontaneous initiation of atrial fibrillation by ectopic beats originating in the pulmonary veins. N Engl J Med 1998;339:659–66. 3. Verma A, Wazni OM, Marrouche NF, et al. Preexistent left atrial scarring in patients undergoing pulmonary vein antrum isolation: an independent predictor of procedural failure. J Am Coll Cardiol 2005;45:285–92. 4. Rostock T, Salukhe TV, Steven D, et al. Longterm single- and multiple-procedure outcome and predictors of success after catheter ablation for persistent atrial fibrillation. Heart Rhythm 2011;8: 1391–7. 5. Platonov PG, Mitrofanova LB, Orshanskaya V, Ho SY. Structural abnormalities in atrial walls are associated with presence and persistency of atrial fibrillation but not with age. J Am Coll Cardiol 2011;58:2225–32. 6. Oakes RS, Badger TJ, Kholmovski EG, et al. Detection and quantification of left atrial structural remodeling with delayed-enhancement magnetic resonance imaging in patients with atrial fibrillation. Circulation 2009;119:1758–67. 7. Jadidi AS, Cochet H, Shah AJ, et al. Inverse relationship between fractionated electrograms and atrial fibrosis in persistent atrial fibrillation: combined magnetic resonance imaging and highdensity mapping. J Am Coll Cardiol 2013;62: 802–12. 8. Anter E, McElderry TH, Contreras-Valdes FM, et al. Evaluation of a novel high-resolution mapping technology for ablation of recurrent scarrelated atrial tachycardias. Heart Rhythm 2016; 13:2048–55. 9. Lin YJ, Lo MT, Lin C, et al. Prevalence, characteristics, mapping, and catheter ablation of potential rotors in nonparoxysmal atrial fibrillation. Circ Arrhythm Electrophysiol 2013;6:851–8. 10. Anter E, Tschabrunn CM, Josephson ME. High-resolution mapping of scar-related atrial arrhythmias using smaller electrodes with closer interelectrode spacing. Circ Arrhythm Electrophysiol 2015;8:537–45. 11. Yagishita A, S DEO, Cakulev I, et al. Correlation of left atrial voltage distribution between sinus rhythm and atrial fibrillation: identifying structural remodeling by 3-d electroanatomic mapping irre-

new end point for enhancing pulmonary vein isolation. Heart Rhythm 2014;11:190–11. 13. Jadidi A, Cochet H, Miyazaki A, et al. Impact of atrial fibrosis on wave propagation in human atrial fibrillation - evidence for regional channel / pivot / slow conduction at delayed enhanced sites. Heart Rhythm 2012;9:S79. 14. 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. 15. Blandino A, Bianchi F, Grossi S, et al. Left atrial substrate modification targeting low-voltage areas for catheter ablation of atrial fibrillation: a systematic review and meta-analysis. Pacing Clin Electrophysiol 2017;40:199–212. 16. Rolf S, Kircher S, Arya A, et al. Tailored atrial substrate modification based on low-voltage areas in catheter ablation of atrial fibrillation. Circ Arrhythm Electrophysiol 2014;7:825–33. 17. Wang XH, Li Z, Mao JL, He B. A novel individualized substrate modification approach for the treatment of long-standing persistent atrial fibrillation: preliminary results. Int J Cardiol 2014; 175:162–8. 18. Cutler MJ, Johnson J, Abozguia K, et al. Impact of voltage mapping to guide whether to perform ablation of the posterior wall in patients with persistent atrial fibrillation. J Cardiovasc Electrophysiol 2016;27:13–21. 19. Jadidi AS, Lehrmann H, Keyl C, et al. Ablation of persistent atrial fibrillation targeting lowvoltage areas with selective activation characteristics. Circ Arrhythm Electrophysiol 2016;9. 20. Yamaguchi T, Tsuchiya T, Nakahara S, et al. Efficacy of left atrial voltage-based catheter ablation of persistent atrial fibrillation. J Cardiovasc Electrophysiol 2016;27:1055–63. 21. Masuda M, Fujita M, Iida O, et al. Comparison of left atrial voltage between sinus rhythm and atrial fibrillation in association with electrogram waveform. Pacing Clin Electrophysiol 2017;40:559–67. 22. Teh AW, Kistler PM, Lee G, et al. The relationship between complex fractionated electrograms and atrial low-voltage zones during atrial fibrillation and paced rhythm. Europace 2011;13: 1709–16. 23. Sanders P, Morton JB, Kistler PM, et al. Electrophysiological and electroanatomic characterization of the atria in sinus node disease: evidence of diffuse atrial remodeling. Circulation 2004;109: 1514–22.

26. Euser AM, Dekker FW, le Cessie S. A practical approach to Bland-Altman plots and variation coefficients for log transformed variables. J Clin Epidemiol 2008;61:978–82. 27. MASH-AF1 Study. Available at: https://aurora. shinyapps.io/mashaf. Accessed September 13, 2018. 28. Anter E, Josephson ME. Bipolar voltage amplitude: what does it really mean? Heart Rhythm 2016;13:326–7. 29. Hunter RJ, Diab I, Tayebjee M, et al. Characterization of fractionated atrial electrograms critical for maintenance of atrial fibrillation: a randomized, controlled trial of ablation strategies (the CFAE AF trial). Circ Arrhythm Electrophysiol 2011;4:622–9. 30. Yang B, Jiang C, Lin Y, et al. STABLE-SR (Electrophysiological Substrate Ablation in the Left Atrium During Sinus Rhythm) for the treatment of nonparoxysmal atrial fibrillation: a prospective, multicenter randomized clinical trial. Circ Arrhythm Electrophysiol 2017;10. 31. Arentz T, Muller-Edenborn B, Jadidi A. Arrhythmogenic atrial substrate in persistent atrial fibrillation: relationship to heart failure. J Am Coll Cardiol EP 2018;4:97–8. 32. Prabhu S, Voskoboinik A, McLellan AJA, et al. Biatrial electrical and structural atrial changes in heart failure: electroanatomic mapping in persistent atrial fibrillation in humans. J Am Coll Cardiol EP 2018;4:87–96. 33. Hwang SH, Oh YW, Lee DI, Shim J, Park SW, Kim YH. Relation between left atrial wall composition by late gadolinium enhancement and complex fractionated atrial electrograms in patients with persistent atrial fibrillation: influence of non-fibrotic substrate in the left atrium. Int J 1191–9.

Cardiovasc

Imaging

2015;31:

34. Spragg DD, Khurram I, Zimmerman SL, et al. Initial experience with magnetic resonance imaging of atrial scar and co-registration with electroanatomic voltage mapping during atrial fibrillation: success and limitations. Heart Rhythm 2012;9:2003–9. 35. Koutalas E, Rolf S, Dinov B, et al. Contemporary mapping techniques of complex cardiac arrhythmias - identifying and modifying the arrhythmogenic substrate. Arrhythm Electrophysiol Rev 2015;4:19–27.

KEY WORDS atrial fibrillation, high-density mapping, low voltage

spective of the rhythm. J Cardiovasc Electrophysiol 2016;27:905–12.

24. Wickham H. ggplot2: Elegant graphics for data analysis. New York: Springer-Verlag, 2009.

12. Squara F, Frankel DS, Schaller R, et al. Voltage mapping for delineating inexcitable dense scar in

25. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of

A PPE NDI X For supplemental tables and figures, please see the online version of this

patients undergoing atrial fibrillation ablation: a

clinical measurement. Lancet 1986;1:307–10.

paper.