Association Between Interatrial Block, Left Atrial Fibrosis and Mechanical Dyssynchrony: Electrocardiography-Magnetic Resonance Imaging Correlation

STRUCTURED ABSTRACT

Introduction:

Advanced interatrial block (IAB) on a 12-lead electrocardiogram (ECG) is a predictor of stroke, incident atrial fibrillation (AF), and AF recurrence after catheter ablation. The objective of this study was to determine which features of IAB structural remodeling is associated with LA MRI structure and function.

Methods/Results:

We included 152 consecutive patients (23% nonparoxysmal AF) who underwent preprocedural ECG and cardiac magnetic resonance (CMR) in sinus rhythm prior to catheter ablation of AF. IAB was defined as P-wave duration ≥120 ms, and was considered partial if P-wave was positive and advanced if P-wave had a biphasic morphology in inferior leads. From cine CMR and late gadolinium enhancement, we derived LA maximum and minimum volume indices, strain, LA fibrosis, and LA dyssynchrony. A total of 77 patients (50.7% paroxysmal) had normal P-wave, 52 (34.2%) partial IAB, and 23 (15.1%) advanced IAB. Patients with advanced IAB had significantly higher LA minimum volume index (25.7 vs. 19.9 ml/m2, p=0.010), more LA fibrosis (21.9 vs. 13.1%, p=0.020), and lower LA maximum strain rate (0.99 vs. 1.18, p=0.007) than those without. Advanced IAB was independently associated with LA minimum [p= 0.032] and fibrosis [p= 0.009]). P-wave duration was also independently associated with LA fibrosis (β: 0.33, p = 0.049) and LA mechanical dyssynchrony (β: 2.01, p = 0.007).

Conclusion:

Advanced IAB is associated with larger LA volumes, lower emptying fraction, and more fibrosis. Longer P-wave duration is also associated with more LA fibrosis and higher LA mechanical dyssynchrony.

INTRODUCTION

Interatrial block (IAB), defined as prolonged P-wave duration (>120 ms) on a 12-lead electrocardiogram (ECG), reflects delayed conduction between the right and the left atrium (LA)(1). IAB is divided into partial (P-IAB) and advanced IAB (A-IAB). In P-IAB, the stimulus is delayed between the two atria, but can cross the septum via the Bachmann’s bundle region(2) and is represented by a prolonged positive P-wave in inferior leads. A-IAB is characterized by a prolonged P-wave with biphasic morphology in inferior leads, due to a complete degree of interatrial delay or blockade at the Bachmann’s bundle that results in caudocranial activation of the left atrium(2).

IAB and associated atrial arrhythmias have been named Bayés syndrome(3). A-IAB is a predictor of stroke(4), incident atrial fibrillation (AF)(5), and AF recurrence after catheter ablation(6),(7). Moreover, in individuals with A-IAB, low LA strain rate during atrial contraction (Sra) by speckle tracking echocardiography predicts incident AF and ischemic stroke(8). LA volumes, strain, dyssynchrony and fibrosis are also predictors of stroke(9),(10), incident AF(11), and AF recurrence after catheter ablation(12),(13),(14). Understanding which features of LA structural remodeling are associated with A-IAB would improve the diagnostic and prognostic value of 12-lead ECG. To determine this association, we analyzed preprocedural 12-lead ECG and cardiac magnetic resonance (CMR) imaging during sinus rhythm in patients referred for catheter ablation of AF. We aimed to evaluate whether IAB is a predictor of LA MRI features.

METHODS

Study population.

Consecutive patients with symptomatic, drug-refractory AF referred for catheter ablation of AF at the Johns Hopkins Hospital between June 2010 and December 2015 who underwent pre-procedural CMR and ECG during sinus rhythm were included. All patients undergoing AF ablation had pre-procedural imaging, either CMR or cardiac CT. The decision to choose CMR over CT is affected by several factors, including patient preference (e.g. claustrophobia), presence of pacemakers/defibrillators, physician preference, and scheduling logistics. Patients who had prior AF ablation or surgical procedure in the LA were excluded. Patients who were in AF at the time of CMR or ECG were also excluded (CMR: n=37 and ECG: n=30). The protocol was approved by the Institutional Review Board and all the patients provided written informed consent.

Cardiac magnetic resonance protocol.

CMR was performed with a 1.5‐Tesla scanner (Avanto; Siemens Medical Systems, Erlangen, Germany), a 6‐channel phased array body coil in combination with a 6‐channel spine matrix coil. An ECG-gated, breath‐holding cine CMR images were acquired in the long axis two- and four-chamber views by True Fast Imaging with Steady‐State Precession (TrueFISP) sequence with the following parameters: TE/TR 3.0/1.5 ms; flip angle 78°; in‐plane pixel size 1.5×1.5 mm²; slice thickness 8 mm; 30 frames per ECG R-R interval with a temporal resolution of 20–40 ms. A fraction of patients also underwent respiratory-navigated, ECG-gated late gadolinium enhancement (LGE) to quantify LA fibrosis. LGE images were acquired within 15–25 minutes following the injection of gadopentetate dimeglumine (0.2 mmol/kg, Bayer Healthcare Pharmaceuticals, Montville, NJ) using a fat‐saturated 3‐D inversion recovery‐prepared fast spoiled gradient‐recalled echo sequence with the following parameters: TE/TR 1.52/3.8 ms; flip angle 10°; in‐plane pixel size 1.3×1.3 mm²; slice thickness 2.0 mm. The trigger time for 3‐D LGE images was optimized to acquire imaging data during LA diastole as determined by the cine CMR images. The optimal inversion time was determined by an inversion time scout scan (median 270 ms; range 240–290 ms) to maximize nulling of the LA myocardium. The endocardial and epicardial contours were manually draw around the LA myocardium (Figure 1). The image intensity ratio (IIR)(15) was measured to quantify LA fibrosis using Qmass MR (version 7.2; Leiden University Medical Center, Leiden, The Netherlands) on axial images from 3‐D axial image data (Figure 1). The IIR threshold of 1.22 that corresponds to bipolar voltage 0.3 mV on intracardiac electrogram was used to define LA fibrosis(16).

Figure 1. Left atrial late gadolinium enhancement CMR (A-D) and Quantification of left atrial regional function using cine cardiac magnetic resonance (E -F).

A – B, LA shell view with areas of enhancement (red). C - D quantification of LA enhancement by CMR using IIR, areas of enhancement (red). E, Two-chamber view with six equi-length segments; F, Four-chamber view with six equi-length segments. Left side (A, C) individual with low enhancement – right side (B, D) individual with high enhancement. CMR, cardiac magnetic resonance; IIR, image intensity ratio.

LA Volumes, Strain and Strain Rate.

Multimodality Tissue tracking software (MTT; version 6.1, Toshiba, Japan) is an automated frame-to frame template matching software. Initially, an experienced operator defines the LA endocardial and epicardial borders at the reference frame – ventricular end-systolic frame identified just before mitral valve opening, when the LA is at its biggest dimension. The confluence of the pulmonary veins and LA appendage are not included in the segmentation. The software then propagated these borders across the cardiac cycle automatically using a template matching algorithm. The software recorded a characteristic pixel pattern of each 10 × 10 mm square area in the reference frame; an area with identical pixel pattern was recognized in the next frame that maximized the similarity evaluated by cross-correlation between the square areas. This procedure was repeated for all pixels in each image and for each frame to track the borders throughout the whole cardiac cycle. Finally, the operator verified the quality of the tracking generated by the software. Based on untagged long-axis 2-chamber and 4-chamber projections MTT provides: Maximum LA volume (Vmax): LA volume at end-systole, immediately before mitral valve opening; Minimum LA volume (Vmin): LA volume at end-diastole, immediately before mitral valve closure; Peak global longitudinal strain (Smax): Indirect measurement of atrial relaxation during LV systole. LA strain rate at maximum (Srmax): Time derivate peak strain rate during ventricular systole. Left atrial emptying fraction (LAEF %) was calculated as: (Vmax-Vmin)/Vmax × 100. LA volumes were adjusted by body habitus by dividing the respective values by the body surface area.

LA mechanical dyssynchrony analysis.

On cine CMR images, the LA endocardial and epicardial borders at the LA end diastole were defined by an operator with more than 3 years of experience in performing LA measurements by CMR, who was also blinded to the group assignment (Figure 1). The MTT software automatically divides the LA in six equi-length segments in each of the two- and four-chamber views, creating a total of 12 segments (Figure 1). Longitudinal strain was calculated within each of the 12 segments. Based on those time series we defined the standard deviation (SD) of the time to the peak longitudinal strain (TPS). We then corrected SD-TPS by the cycle length to derive SD-TPS as a percentage of the cycle length $\left(\frac{\mathit{\text{SD}}-\mathit{\text{TPS}}\left(ms\right)}{\mathit{\text{Cycle}}\mathit{\text{length}}\left(ms\right)}\chi 100\right)$ higher values reflect greater degrees of LA mechanical dyssynchrony.

IAB analysis.

The duration of the P wave and IAB were manually analyzed (Figure 2). The duration of the P wave was measured from the digitalized ECG images with Geogebra Software version 4.2. The images were amplified by 20 times to define the beginning of the earliest and the latest depolarization on the precordial leads, which is by definition the positive or negative deflection, respectively. P-wave duration was defined as the interval in milliseconds between the earliest and the latest depolarization (onset and offset of P wave) in the frontal leads. P wave measurements were performed by two independent observers. If there was a difference in two independent measurements greater than 1 millisecond, a third independent observer determined the final measurement.

Figure 2.

Examples of partial interatrial block (left) and advanced interatrial block (right).

Statistical analysis.

Baseline patient demographics are presented as mean+SD or percentage. Comparison between groups was using Student’s t-test, X² and, Fisher exact test as appropriate. The “No IAB (W/O IAB)” group excluded patients with P-IAB/A-IAB. Multivariable linear regression was used to assess the association between LA structure and function (independent variable) on IAB (dependent variable). Results are presented as beta coefficients. Model 1: unadjusted, Model 2: the model was adjusted for clinical characteristics (sex, type of AF, body mass index (BMI), hypertension, and obstructive sleep apnea [OSA]). Moreover, multivariable linear regression was used to evaluate the association between LA dyssynchrony/fibrosis and P-wave duration. Model 1: unadjusted, Model 2: adjusted for clinical characteristics (sex, type of AF, body mass index (BMI), hypertension, and obstructive sleep apnea [OSA]). The clinical variables included into multivariable models were included if p in univariable analysis < 0.5. Correlation between LA mechanical dyssynchrony (SD-TPS), and % LA fibrosis with P-wave duration was defined by the Pearson correlation coefficient. We also evaluated the specificity and sensitivity for cardiac remodeling of IAB, based on previous published cut-offs of LA dyssynchrony (> 2.86 %)(12) and LA-LGE (> 35%)(14) in predicting AF recurrence after AF ablation. The significance level of 0.05 was used for all hypothesis tests, and all t-tests were 2-sided. The statistical computations were performed using the STATA (version 15.1, StataCorp).

RESULTS

Patient demographics.

A total of 152 patients were included in the final analysis. Table 1 summarizes patients’ clinical characteristics. Mean age was 59.5+10.6 years, and 50 were women (32.9%). A total of 66 patients (43.4%) had recurrence after index ablation. A total of 107patients (70.4%) underwent respiratory-navigated, ECG-gated LGE to quantify LA fibrosis.

Table 1. Baseline characteristics, ECG and cardiac magnetic resonance (CMR) variables.

Data are presented as median (IQR - interquartile range), or mean ± standard deviation (SD). ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blockers; Ca, calcium. LA, left atrium; LAEF, LA emptying fraction; Smax, peak longitudinal LA strain; SRmax, peak longitudinal strain rate; Sre, early diastolic strain rate; Sra, late diastolic strain rate; TPS, time to peak strain. * n=107 subjects.

Characteristics	Values
Age, years	59.5 + 10.6
Body mass index, Kg/m2	28.6 + 5.6
Female, n (%)	50 (32.9)
Persistent AF, n (%)	36 (23.8)
Heart failure, n (%)	16 (10.5)
Coronary artery disease/vascular disease, n (%)	13 (8.6)
Diabetes, n (%)	14 (9.3)
Hypertension, n (%)	70 (46.1)
History of Stroke/TIA, n (%)	10 (6.6)
CHA2DS2 -VASC	1.7 + 1.6
Obstructive sleep apnea, n (%)	22 (14.6)
Ablation strategy (Cryoablation), n (%)	31 (20.5)
LVEF, %	57.1 + 8.0
Recurrence at 12 months, n (%)	66 (43.4)
Medication
ACE inhibitor/ARB, n (%)	47 (31.1)
Beta-Blockers, n (%)	76 (52.1)
Ca-channel blockers, n (%)	30 (20.1)
Anticoagulant use, total (%)	152 (100)
Number of antiarrhythmic drugs	1.3 + 0.8
CMR LA Structure/Function	Values
Minimum LA volume index, ml/m2	20.8+8.5
Maximum LA volume index, ml/m2	39.4+10.8
Total LAEF, %	48.6+10.7
Smax, %	28.7+9.4
SRmax	1.15+0.45
SD-TPS %,	2.7 (1.95)
LA fibrosis, % total LA area *	12.4 (12.46)
ECG Variables	Values
Partial IAB, n (%)	52 (34.2)
Advanced IAB, n (%)	23 (15.1)
P-wave duration, ms	124.0±18.4
CMR LA Structure/Function	Values
Minimum LA volume index, ml/m2	20.8+8.5
Maximum LA volume index, ml/m2	39.4+10.8
Total LAEF, %	48.6+10.7

IAB types and LA structural remodeling.

Out of 1,824 LA segments analyzed in a total of 152 patients, a total of 18 segments (0.98%) were excluded from the final analysis because they lacked well-defined peaks in the strain curves. Table 1 summarizes the CMR and ECG variables among subjects included in this study. A total of 77 patients (50.7%) had normal P-wave, 52 (34.2%) P-IAB, and 23 (15.1%) A-IAB. A-IAB was significantly associated with higher LA minimum volume index (25.7 vs. 19.9 ml/m2, p=0.010), more LA fibrosis (21.9 vs. 13.1%, p=0.020), and lower Srmax (0.99 vs. 1.18, p=0.007) (Table 2). In this study cohort, P wave measurements by two independent observers did not match in 5 cases (3.3%), which were finalized by a third independent observer. The inter/intra- reader reproducibility (measured by ICC [Intra-class correlation coefficient]) were published before by our group(17),(18),(15). LA strain(17) (Inter reader ICC: 0.96; Intra reader ICC: 0.92), strain rate(17) (Inter reader ICC: 0.91; Intra reader ICC: 0.89), dyssynchrony(18) (Inter reader ICC: 0.86; Intra reader ICC: 0.85) and scar quantification by LGE(15) (Inter reader ICC: 0.97; Intra reader ICC: 0.98). The A-IAB sensitivity was 18.8% and the specificity was 87.9% for LA dyssynchrony, and 23.1% and 89.0%, respectively, for LA-LGE.

Table 2. Interatrial block (IAB) and left atrial (LA) structure and function.

Data are presented as median/mean as appropriate. LAEF, LA emptying fraction; Smax, peak longitudinal LA strain; SRmax, peak longitudinal strain rate; TPS, time to peak strain. W/O: without IAB, W: with IAB. *p<0.05.

	Partial IAB			Advanced IAB
	W n: 52	W/O n: 77	p	W n: 23	W/O n: 77	p
Minimum LA volume index, mm3/m2	19.0	21.6	0.131	25.7	19.9	0.010*
Maximum LA volume index, mm3/m2	37.8	40.3	0.216	38.6	44.2	0.134
Total LAEF, %	50.4	47.7	0.057	43.4	49.5	0.090
Smax, %	29.6	28.2	0.185	25.1	29.3	0.125
SRmax, s^-1	1.21	1.12	0.080	0.99	1.18	0.007*
SD-TPS %,	3.3	3.1	0.216	4.0	3.2	0.081
LA fibrosis, % total LA area (n=107)	15.8	13.4	0.786	21.9	13.1	0.020*

Linear regression univariable and multivariable analyses.

In Model 1, a univariable (unadjusted) analysis identified minimum/maximum LA volume, peak longitudinal LA strain (S_max,), LAEF (%), and LA fibrosis (% total LA area) as significant associated with A-IAB. After adjusting for clinical risk factors, minimum (β: 0.18, p= 0.032) and LA fibrosis (β: 0.25, p= 0.0009) remained significant (Table 3, Model 2).

Table 3. Univariable and multivariable linear regression analysis.

β; beta coefficient, 95% confidence interval (CI) LA, left atrium; LAEF, LA emptying fraction; Smax, peak longitudinal LA strain; SRmax, peak longitudinal strain rate; TPS, time to peak strain. Model 1: unadjusted; Model 2 adjusted for sex, type of AF, body mass index (BMI), hypertension, and obstructive sleep apnea [OSA]. IAB: interatrial block.

	Model 1 Unadjusted		Model 2 Adjusted for clinical variables
	β	p	β	p
Partial IAB (n=52)
Minimum LA volume, mm3/m2	−0.14	0.073	−0.11	0.180
Maximum LA volume, mm3/m2	−0.11	0.171	−0.08	0.328
Total LAEF, %	0.12	0.141	0.09	0.260
Smax, %	0.07	0.382	0.04	0.584
SRmax s^-1	0.09	0.256	0.05	0.571
SD-TPS %,	0.007	0.993	0.02	0.754
LA fibrosis, % total LA area (n=107)	0.10	0.282	0.11	0.238
Advanced IAB (n=23)
Minimum LA volume, mm3/m2	0.23	0.004*	0.18	0.032*
Maximum LA volume, mm3/m2	0.19	0.021*	0.16	0.055
Total LAEF, %	−0.20	0.011*	−0.14	0.084
Smax, %	−0.16	0.047*	−0.12	0.144
SRmax s^-1	−0.14	0.070	−0.09	0.287
SD-TPS %,	0.15	0.071	0.09	0.265
LA fibrosis, % total LA area (n=107)	0.28	0.004*	0.25	0.009*

P-wave duration and LA structural remodeling.

The correlation coefficient between LA mechanical dyssynchrony and P-wave duration was 0.26 (p= 0.001). The correlation coefficient between the LA fibrosis and P-wave duration was 0.24 (p = 0.010). Univariate analysis between both LA mechanical dyssynchrony (β: 0.45, p = 0.007)/LA fibrosis (β: 2.56, p < 0.001) and p-wave duration were both significant. Multivariable analysis adjusting for clinical cardiovascular risk factors (sex, body mass index, hypertension, diabetes, obstructive sleep apnea, AF type) showed P-wave duration was independently associated with LA fibrosis (β: 0.33, p = 0.049) and LA mechanical dyssynchrony (β: 2.01, p = 0.007).

DISCUSSION

Main findings.

Our main findings are summarized as follows: 1) A-IAB is an independent marker of LA minimal volume, and LA fibrosis; 2) P-wave duration is associated with LA fibrosis and LA mechanical dyssynchrony. The association between A-IAB and atrial fibrosis quantified by LGE MRI has been suggested qualitatively in a case report (19) . This is the first study to demonstrate the quantitative association between A-IAB and not only atrial fibrosis, but also atrial structure and function in a series of more than 100 patients in a cross-sectional fashion.

Interatrial block and left atrial structural remodeling.

It has been hypothesized that IAB results from fibrotic atrial remodeling due to reduction of the blood flow to the Bachmann bundle and posterior fibrosis(20). Common manifestations of atrial structural remodeling include reduced atrial function, fibrosis and dilatation. Perpetuation of those clinical risk factors seems to contribute to disease progression from P-IAB to A-IAB.

Previous studies using speckle-tracking echocardiography studied the association between IAB and LA structure/function (21)(22) (23)(1). Our investigation represents the first large study to demonstrate the association between IAB and LA structure/function by cardiac MRI. Most importantly, we demonstrate the quantitative association between IAB and atrial fibrosis, which has never done before. LA fibrosis quantified by LGE is a marker of atrial fibrotic remodeling, and we previously reported that LA fibrosis predicts AF recurrence after catheter ablation of AF(14).

Clinical implications.

Remodeling processes precede arrhythmic substrate maturation. For example, episodes of scar-related ventricular tachycardia (VT) associated with healed myocardial infarction (MI) can occur after many years of post-MI remodeling processes.(24) In addition, previous studies(25), (26) strongly suggest that AF simply is a marker of increased thromboembolic risk, but is not mechanistically responsible for the thromboembolic events(27). Therefore, alterations in LA structure, function and synchrony even prior to development of AF may still confer an increased thromboembolic risk. Furthermore, not all individuals with LA remodeling may develop arrhythmia. For example, VT late after MI occurs in only 1 percent of patients who had MI(28). Therefore, assessment of A-IAB in sinus rhythm may identify a high-risk subgroup without clinically recognized AF for stroke. This subgroup may benefit from prophylactic oral anticoagulation or LA appendage closure to reduce the risk of stroke and cognitive decline. Further prospective studies are needed to address the utility of A-IAB in identifying the high-risk subgroup without clinically recognized AF.

Limitations.

This study represents a single-center analysis of patients referred for catheter ablation of AF. Therefore, there is a non-negligible chance of selection bias. We also did not perform adjustments in our statistical significance based on multiple comparisons of our multivariable models. For the LA functional analysis, we used only two- and four-chamber cine CMR, which was included in a routine image-acquisition protocol. Therefore, it is possible that our analysis underestimated the degree of LA dysfunction by missing regions that were not covered by those two views. Since the strain was two-dimensional (2-D) and was obtained only in the in-plane direction, SD-TPS may have been underestimated compared with those obtained from 3-D strains. Despite those potential causes of underestimation, our analysis demonstrated a significant association of A-IAB and P-wave duration with LA structure and function and AF recurrence. Therefore, we believe that the advantage of our approach outweighs the disadvantage of including more views to assess the whole LA deformation, which would increase the scan time and post-processing burden. The correlation between p-wave duration and LA mechanical dyssynchrony or LA fibrosis, although significant, is small and thus further investigation is necessary to confirm our findings. There is a possibiity that cardioversion-induced atrial stunning could have confounded our findings. However, only a small proportion of patients in our cohort received elective cardioversion(9). In addition, cardiac MRI was performed ~8 weeks after cardioversion on average, while cardioversion-induced atrial stunning usually recovers within 4 weeks (29). Therefore, we believe that the possibility is low. We also excluded subjects that were in AF either at the time of CMR or ECG acquisition which is also a limitation of our approach and source of bias. Another potential source of bias in our study was that a relatively small portion of the patients referred to our institution for catheter ablation of AF undergo pre-procedural cardiac MRI, whereas most patients undergo pre-procedural cardiac CT mainly for logistic reasons. Lastly, although significant, the magnitude of the correlation between LA structure/function indices and A-IAB is relatively weak.

Conclusions.

A-IAB is associated with larger LA volumes, lower emptying fraction, and more fibrosis. Longer P-wave duration is also associated with more LA fibrosis and higher LA mechanical dyssynchrony.

Figure 3. Correlation between LA mechanical dyssynchrony, left atrial fibrosis and P-wave duration.

A, LA mechanical dyssynchrony (x) and P-wave duration (y). B, Left atrial fibrosis (x) and P-wave duration (y).