Electrocardiogram (ECG) for the Prediction of Incident Atrial Fibrillation: An Overview

Abstract

Electrocardiograms (ECGs) have been employed to medically evaluate participants in population-based studies, and ECG-derived predictors have been reported for incident atrial fibrillation (AF). Here, we reviewed the status of ECG in predicting new-onset AF. We surveyed population-based studies and revealed ECG variables to be risk factors for incident AF. When available, the predictive values of each ECG risk marker were calculated.

Both the atrium-related and ventricle-related ECG variables were risk factors for incident AF, with significant hazard risks (HRs) even after multivariate adjustments. The risk factors included P-wave indices (maximum P-wave duration, its dispersion or variation and P-wave morphology) and premature atrial contractions (PACs) or runs. In addition, left ventricular hypertrophy (LVH), ST-T abnormalities, intraventricular conduction delay, QTc interval and premature ventricular contractions (PVCs) or runs were a risk of incident AF. An HR of greater than 2.0 was observed in the upper 5th percentile of the P-wave durations, P-wave durations greater than 130 ms, P-wave morpholyg, PACs (PVCs) or runs, LVH, QTc and left anterior fascicular blocks. The sensitivity , specificity and the positive and negative predictive values were 3.6-53.8%, 61.7-97.9%, 2.9-61.7% and 77.4-97.7%, respectively.

ECG variables are risk factors for incident AF. The correlation between the ECG-derived AF predictors, especially P-wave indices, and underlying diseases and the effects of the reversal of the ECG-derived predictors on incident AF by treatment of comorbidities require further study.

Introduction

Atrial fibrillation (AF) is a common arrhythmia, and the number of patients with AF is increasing in many developed countries. AF is associated with increased morbidity and mortality [1], and the prediction of AF development is expected to improve both health and clinical outcomes at the population level. Thus far, demographic and clinical variables have been extensively studied, and some have been shown to be risk factors for AF development [2]-[6].

Electrocardiograms (ECGs) are essential to diagnose AF and are used in population-based health examinations because of their ease of use and low cost. Recent advancements in high-quality signal acquisition and the availability of automated hardware ECG setups have facilitated the use of ECGs in mass examinations, and many ECG-derived markers have been confirmed as risk factors for incident AF. Here, we reviewed community-based cohorts that have confirmed ECG-derived risks for AF development.

Studies treating incident AF

Mainly, studies treating incident AF in community-based cohorts were collected from the literature. Small studies of less than 100 participants were not included, except for those measuring P-wave duration or dispersion by signal averaged ECGs (SAECGs). SAECGs may present more accurate P-wave durations or dispersions in spite of the difficulty of using them in population-based examinations.

AF was diagnosed using ECG and Holter monitors. New AF was diagnosed when the participants had no AF at baseline ECG but had AF on an ECG during the follow-up period. Then, we included some paroxysmal AF (PAF), whereas some cases of PAF might have been undetected if the sinus rhythm had resumed.

Most ECG variables were automatically measured by the hardware ECG set-up, but some were confirmed by cardiologists in most studies.

Atrium-related ECG-derived predictors [Table 1]

Table 1. Atrium-related ECG-derived predictors for new-onset AF.

PAC: premature atrial contraction. FU: follow-up. HR: hazard risk. mo: months. yrs: years. Ref. No.: reference number.

Author	Number of	FU	Annual	ECG finding	HR (95% CI, P-value)
(Rf. No.)	participants	time	incidence
P-wave duration
Perez [13]	42,751	5.3 yrs	0.45%	>120 ms	1.6 (1.3-1.8, P<0.0001)
				per 20 ms	1.1 4 (1.02-1.27, P=0.022)
Soliman[14]	15,429	6.97 yrs	0.11%	>120 ms	1.94 (1.66-2.28) Model 1
					1.79 (1.51-2.14) Model 2
				Upper 5%	5.23 (3.33-8.22) Model 1
					4.07 (2.55 - 6.51) Model 2
Magnani[15]	1,550	15.8 yrs	1.46%	Upper 5%	2.51 (1.13-5.57)
				per 1 SD	1.1 (0.90-1.47, P=0.27)
Nielsen[28]	285,933	6.7 yrs	0.49%	>89 ms	1.60 (1.41−1.81 P<0.001)
				100-105 ms	1.00
				112-119 ms	1.22
				120-129 ms	1.50
				≥130	2.06
P-wave dispersion
Perez [13]	-	-	-	>80 ms	1.95 (1.7-2.3, P<0.0001)
				SD of P duration >35	1.7 (1.3-2.1, P<0.0001)
P-morphology
Soliman[14]	-	-	-	95th% of PTFV1	1.23 (1.04-1.46)
Magnani[16]	-	-	-	PTFV1 >0.04 V•s	1.56 (1.24-2.00)
Enriquez[35]	122	30 mo	46.7%	P duration ≥120 ms	2.9 (1.02-8.6, P<0.04)
				Biphasic P in II,I II, aVF
Abn. P axis
Perez [13]	-	-	-	>95th%	1.21 (0.69-2.12, P<0.0001)
				>74° and <24°	1.52 (1.05-2.21, P>0.0001
PAC/runs
Watanabe [51]	63,368	10 yrs	0.14%	Standard ECG, ≥1/10 sec	2.89 (1.80-3.35)
Perez [13]	-	-	-	Presence	2.1 (1.6-2.7, P<0.0001)
Murakoshi [52]	63,197	14 yrs	0.105%	Presence	4.87 (6.61-6.57) for men
					3.87 (2.69-5.57) for women
Acharya[56]	1,357	7.5 yrs	11.4%	≥100/day	2.97 (1.85-4.80)
				couplets ≥50/day	3.11
				bigeminy	3.67
				3 PACs run ≥20 /day	2.94
				runs (≥10 beats/run)	1.73
PR interval
Perez [13]	-	-	-	>200 ms	1.3 (1.1-1.6, P=0.003)
Nielsen[69]	288,181	6.7 yrs	3.8%	≥196 ms men	1.30 (1.17-1.44)
				≥204 ms women	1.18 (1.06-1.530)
				<129 ms men	1.09 (0.92-1.29) ‡
				<121 ms women	1. 32 (1.12-1.56)
Alonso[25]	262,288	5 yrs	6.8%	per 30 ms	1.03 (0.85-1.26, <0.0001)
					to 1.22 (1.01-1.42, 0.027)
Macfarlane[68]	5,680	3.2 yrs	9.4%	per 30 ms	1.19 (1.09-1.30, P<0.0001)

P-rate (heart rate) and P-wave indices and PR intervals are considered to represent alterations of electrophysiological and/or structural remodeling of the atrium in association with or without demographic and clinical variables that are risk factors for incident AF. The appearance of premature atria contractions (PACs) or runs may trigger AF and can be another risk for incident AF.

P-rate

Bradycardia is known to be associated with AF, especially in athletes, but heart rate was not a significant predictor for incident AF after multivariable analysis [2],[5].

P-wave duration

The P-wave duration represents the time required for a sinus impulse to propagate from the sinus node to the entire atrium. A prolonged P-wave duration correlates with a slowed conduction velocity within the atria [7],[8], and fragmented local electrograms suggest the presence of fibrosis [9],[10]. P-wave duration (ms) was measured from the P-wave onset to its offset, and the maximum P-wave duration, defined as the longest duration observed on a standard 12-lead ECG, is most often used to predict incident AF in population-based studies and can be an intermediate ECG phenotype of AF [11],[12]. Perez [13] analyzed the data of 42,751 patients who were followed for 5 .3 years. New AF developed in 1,050 patients at an annual rate of 0.45%/yr. After correcting for age and sex, the maximum P-wave duration of >120 ms was predictive of AF, with a hazard ratio (HR) of 1.6 (95% confidence interval (CI): 1.3-1.8, P <0.0001). The ARIC study (15,429 participants), which had an annual rate of AF development of 0.11%/yr ([14]), and the Framingham Heart Study (FHS) (1,550 participants aged ≥60 years), which had an annual rate of AF development of 0.46%/yr ([15]), showed similar results. The HR per 20 ms or per 1 SD of the P-wave duration was also significant [13],[16]. In studies with smaller sample sizes, P-wave duration was a predictor of PAF in patients without [17],[21] or with heart diseases [22],[24].

The sensitivity, specificity and positive and negative predictive values for a P-wave duration ≥120 ms were 53.8%, 61.7%, 49.0% and 97.3%, respectively [25], and they were 7.2%, 94.5%, 34.6% and 76.4%, respectively, for a P-wave duration above the 95th percentile [16].

In studies with smaller sample sizes (n=50 to 371), a shorter P-wave duration was associated with AF development [26],[27]. In the Copenhagen ECG study, 285,933 participants were followed for 6.7 years [28]. The ECGs were digitally analyzed with clinically validated software (Marquette 12SL algorithm, General Electric Co., Fairfield, CT). AF developed at a rate of 0.49%/yr in this study. Compared to subjects who had a P-wave duration of 100-105 ms, both shorter and longer P-wave durations were associated with an increased risk of AF development [Table 1].

Using SAECG, the filtered P-wave duration (FPD) ≥125 ms as well as the root mean square value of the last 20 ms (RMS20) of ≤3.3 μV were independent predictors for incident AF [29],[32].

P-wave dispersion and variation

P-wave dispersion (PWD) is defined as the difference between the longest and shortest P-wave durations across 12-leads of the surface ECG. Although a precise electrophysiological role for AF development is only speculative, a P-wave dispersion of >80 ms was associated with an AF history ([13],[15]-[18],[26],[30],[33]) or new AF detection on Holter ECG recordings [23]. The sensitivity, specificity and positive and negative predictive values for a P-wave dispersion above the 95th percentile were 3.6%, 94.5%, 16.7% and 76.4%, respectively [15]. The standard deviation (SD) of the P-wave duration across 12 leads >35 ms was another risk factor for incident AF [13].

P-wave morphology and P-wave axis

P-wave duration ≥120 ms with and without inversion of the terminal part of the P-wave was defined as an inter-atrial block (IAB) [34], and an advanced IAB was defined as a P-wave duration ≥120 ms, with a biphasic morphology in the inferior leads shown to represent a block in the Bachman bundle with caudo-cranial activation of the left atrium. Association of an IAB with new onset of AF or recurrence of AF after cardioversion has been observed [35]-[38]. Following cavo-tricuspid isthmus ablation in 122 patients with typical flutter and no history of AF (mean age 63 years), advanced IAB was observed in 23% [35]. After a mean follow-up of 30 months, 46.7% developed new-onset AF. The incidence of AF was significantly higher in patients with advanced IAB than in those without: 71.4% vs. 39.4% (P=0.003) and a risk of new AF ([Table 1]). A recent meta-analysis included 18,204 patients (mean age 56±13, 48% male) with a mean follow-up period of 15.1 years [39]; advanced IAB was a significant predictor of new-onset AF with a pooled HR of 2.58 (95% CI: 1.35 to 4.96; P< 0.01), but the risk of new-onset AF for partial IAB (=P-wave duration ≥120 msec) did not reach statistical significance (HR: 1.42, 95% CI: 0.85 to 2.34; P=0.18)

The P-wave terminal force (μV•ms) was determined as the product of the negative P-wave deflection in the lead V1 (μV) and the duration (ms) from the onset of the negative deflection to its nadir (PTFV1). PTFV1 is considered to represent a volume or pressure overload of the left atrium [40], inter-atrial conduction delay [7] and fibrosis of the left ventricle as measured by MRI [41]. PTFV1 >0.04 mV•s and the upper 5th percentile of PTFV1 are risk factors for developing AF ([13],[14],[52]). Notching or deflection of the P-wave [30], P-pulmonale [43], the area of the initial portion of the P-wave [14] and pre-specified P-wave morphology were risk factors for incident AF [32]. The P-wave axis represents the electrical wave front propagation and the geometry of the atria, and it is altered under a volume or pressure overload of the atrium. A P-wave axis outside of 24°-74° [13] or <74° in the frontal plane was another risk factor for new AF [43].

Premature atrial contractions

The prevalence of PACs increases with advancing age and comorbidities, and PAC becomes a risk factor for AF. PAC may originate within pulmonary veins and precipitate PAF [44]. Prevention is difficult when it progresses to chronic AF [45]-[49], and catheter ablation is the only effective therapeutic modality thus far [50]. Watanabe analyzed data from a mass examination of a general population of 63,386 participants aged ≥50 years in the Niigata Preventive Medicine Study [51]. During a 10-year follow-up, new AF developed at a rate of 0.14%/yr. The presence of PACs in 10-second recordings of standard 12-lead ECGs was associated with AF, with an odds ratio (OR) of 2.89 (1.80-3.35). This result was confirmed later by Perez [13] and Murakoshi [52].

In Holter ECGs, supraventricular tachyarrhythmias have been shown to be a predictor of incident AF [53]-[59]. Frequent PACs, defined as ≥100 beats /day [56]-[57], ≥30 beats/hr [58] or PACs ≥102 beats/day [59], were a risk factor for incident AF. It has been found that subclinical atrial tachyarrhythmias detected by implanted devices within 3 months are a predictor of clinical AF, with an HR of 5.56 (95% CI: 3.78-8.17, P<0.001) [60]. The CHADS2 score and the count of PACs can predict incident AF independently and synergistically [59]. PACs have been shown to aid the detection of AF in patients with cryptogenic stroke or transient ischemic attack who had no AF at baseline [61]-[68].

The sensitivity was in the range of 4.1-29.0%, with a high specificity of >90%. The positive and negative predictive values ranged from 2.9-9.1% and 88.7-97.7%, respectively ([13],[51],[52]).

PR interval

The PR interval (ms) was defined by automatic measurement from the onset of the P-wave to the initiation of the QRS segment. It is composed of atrial conduction, conduction through the atrio-ventricular node and the His-Purkinje system, and it is affected by autonomic tones. A PR interval >200 ms or PR interval of ≥95th percentile (≥196 ms for women and ≥204 ms for men) was a risk factor for incident AF ([3], [66]-[71]). A significant HR was obtained for a 30-ms increment of the PR intervals [3],[25]. In a subgroup of patients with P-pulmonale (n=591), a PQ interval >150 ms was associated with a higher HR of 6.89 (95% CI: 2.39-29.15, P<0.0001) [43]. A short PR interval (≤121 ms for women and ≤129 ms for men) [69] or PR interval variation (maximum PR interval minus minimum PR interval) >36.5 ms [71] was another risk for AF development.

The sensitivity, specificity and positive and negative predictive values were 7.5%, 95.8%, 5.2% and 77.4%, respectively, for PR intervals >95th percentile in FHS [15] and were 13.2%, 90.1%, 4.8% and 96.6%, respectively, for PR intervals >200 ms compared with those ≤200 ms [25].

However, there was a study showing that a prolonged PR interval was not associated with increased incident AF and that a PR interval >200 ms could normalize to ≤200 ms in 30% of patients during follow-up [72]. The underlying mechanism of how a prolonged PR interval induces AF is not well understood

Ventricle-related ECG-derived predictors ([Table 2])

Table 2. Ventricle-related ECG-derived predictors for new-onset AF.

†: From the pooled data of the Framingham study, the ARIC study3) and the Cardiovascular Health study40) for derivation and the Rotterdam study41) for validation in CHARGE-AF Consortium.25) BBB: bundle branch block or hemi-block. FU: follow-up. HR: hazard risk. LAFB: left anterior fascicular block. LBBB: left bundle branch block. LVH: left ventricular hypertrophy. NSVT: non-sustained ventricular tachycardia ≥3 beats. RBBB: right bundle branch block. PVC: ventricular premature contraction. Ref. No.: reference number. ST-T abn. : ST-T abnormality

Author	Number of	FU	Annual	ECG finding	HR (95% CI, P-value)
(Ref No.)	participants	time	incidence
LVH
Watanabe[51]	-	-	-	Sokolow-Lyon	1.43 (1.13-1.80)
Perez[13]	-	-	-	Romhilt-Estes	1.3 (1.0-1.7, P=0046)
Chrispin[75]	4942	-	-	Sokolow-Lyon	1.83 (1.06-3.14)
Alonso[25]	-	-	-	-	1.04 (0.99-1.10) to 2.96 (1.08-8.01)
Macfarlane[68]	117	-	-	Minnesota code, definite	2.20 (1.44-3.37, 0.0003)
	138	-	-	probable	2.33 (1.58-3.44, <0.0001)
	709	-	-	possible	1.35 (1.08-1.70, 0.010)
PVC/runs
Watanabe[51]	-	-	-	presence	3.49 (2.40-5.08)
Perez[13]	-	-	-	presence	1.5(1.2-1.9, P=0.004)
Agarwal[76]	14,783	15-17yrs	-	presence in 2 min	1.56 (1.30-1.87)
Lin[77]	3751	10.1 yrs	0.68	multiform NSVT	1.546 (1.058-2.258)
Lin[78]	3367	10.1 yrs	0.68	NSVT (≥3 PVCs)	1.546 (1.058-2.258)
					1.716(1.243-2.368)
ST-T abn
Watanabe[51]	-	-	-	without LVH	1.89 (1.34-2.13)
Macfarlane[68]	-	-	-	Minnesota code 5-1/5-2	1.7 6(1.41-2.21, P<0.0001)
BBB
Perez[13]	-	-	-	LBBB	1.7(1.2-2.5, P<0.0001)
Watanabe[51]	-	-	-	LBBB	0.96 (-)
				RBBB	0.84 (-)1.17(1.08-1.27, P<0.001)
	-	-	-	QRS width per 20 ms	1.17(1.08-1.27, P<0.001)
Macfarlane [68]	4,696	12.3 yrs	2.16%	LAFB	2.1 (1.1- 3.9, P=0.023)
QTc
Macfarlane[68]	-	-	-	per 30 ms increment	1.31(1.20-1.42, P<0.0001)
Mandyam[83]	14,538	-	0.51%	per 10 ms increment	1.11(1.07-1.14, P<0.001)
				≥460ms (women)/≥450 ms (men)	1.99 (1.37-2.89, P<0.001)
Nielsen[84]	-	-	-	≤372 ms	1.45 (1.14-1.84; P=0.002)
				≥458 ms	2.32 (1.52- 3.54, P<0.001)
Nguyen[55]	-	-	-	≥460 ms (women) or≥450 ms (men)	2.5 (1.4-4.3, P=0.002)

Left ventricular hypertrophy (LVH), premature ventricular contraction (PVC) or non-sustained ventricular tachycardia (NSVT), ST-T abnormalities and the QT (QTc) interval or bundle branch block (BBB) have been examined in large studies.

LVH

LVH is usually diagnosed from ECG criteria, and ECG-derived LVH has been used in mass examinations [73],[74]. The Sokolow-Lyon, Romhilt-Estes, Cornell voltage criteria or Minnesota code were used. LVH reduces the compliance of the left ventricle and imposes a pressure overload on the atrium, leading to dilatation and remodeling.

LVH was a risk factor for AF development during a 10-year follow-up period in Japanese people [51] (OR: 1.43; 95% CI: 1.13-1.80). This was confirmed by subsequent studies ([13],[25],[67],[68]). The ECG-derived LVH measure of the Sokolow-Lyon voltage product had a risk similar to that of the cardiac magnetic resonance-derived LVH: 1.83 (95% CI: 1.06-3.14) and 2.04 (95% CI: 1.15-3.62), respectively [75]. The calculated sensitivity, specificity and positive and negative predictive values for LVH were 5.1-11.5%, 94.7-97.9%, 5.2-8.3% and 97%, respectively ([13], [15], [25]).

PVC and runs

Watanabe [51] and later, Perez [13], showed that the presence of a PVC on surface ECG recordings during examination was associated with incident AF. Any PVCs recorded for 2 minutes were risk factors for incident AF [76]. In studies from Taiwan [77],[78], non-sustained ventricular tachycardia was a risk for new-onset AF.

ST-T abnormalities

ST-T abnormalities reflect the presence of hypertrophy, fibrosis or conduction abnormalities of the ventricle and ischemia of the ventricle, and these conditions can lead to diastolic dysfunction of the ventricle and atrial remodeling. In the general population, Watanabe [51] showed that ST-T abnormalities unrelated to LVH were risk factors for new-onset AF, and this was confirmed by the PROSPER study [68].

Bundle branch block (BBB)

Perez [13] has shown that left BBB (LBBB) is a risk factor for AF development, but neither right BBB nor left BBB was a risk factor for new-onset AF in the Niigata Preventive Medicine study [51]. In the PROSPER study [68], widening of the QRS complex was a risk factor for incident AF. A left anterior fascicular block (LAFB) was a risk for AF development [55], but cases with LAFB were found in only 2.3% of the 1,664 participants [79].

QT interval

An inherited abnormality of the QT interval is known to be associated with a high incidence of AF ([80]-[82]), and its pathogenesis is related to the ion channelopathy of the atrium. QT intervals were corrected by either the Bazett, Framingham study, Hodges or Fridericia formula. A QTc ≥460 ms for women or QTc ≥450 ms for men was a risk factor for incident AF in the ARIC studies [83]. The risk of incident AF increases for every 40 ms [62] or 10 ms [1]increment of QTc. A QTc interval lower than the first percentile (≤372 ms) was also associated with a significant HR for incident AF compared with that in reference individuals (411 to 419 ms) ([84]).

Discussion

In population-based studies, some ECG-derived variables are risk factors for AF development. P wave abnormality, such as prolongation of the P-wave width, its dispersion, and morphology, and the presence of PACs/runs are the ECG phenotype of electrical remodeling of the left atrium and triggers for AF initiation, respectively. The ECG markers of the ventricle were also predictors for incident AF. Most of these ECG predictors are associated with low sensitivity and low positive predictive values, which leads to difficulty in identifying patients who are likely to develop AF. However, the high negative sensitivity can be used to identify those who are unlikely to develop AF.The ECG-derived risk findings can be a result of the underlying diseases that cause remodeling of the ventricle and the atrium

Measurements of ECG-derived variables

The ECG-derived predictors for incident AF can be easily and simply obtained in a population-based study and measured automatically by recent ECG hardware setups, with the exception of P-wave index measurements. P-wave durations were initially measured manually and more recently by a computer-assisted system, but it may still require confirmation by cardiologists. P-wave dispersion is defined as the difference between the longest and the shortest P-wave durations across 12-leads, but its electrophysiological significance and role in AF development need to be clarified. P-wave durations may be measured precisely using SAECG, but its use is impractical for screening mass examinations.

ECG-based LVH is associated with incident AF, and it can be measured easily and automatically. However, it is known that the ventricular mass is more accurately determined by echocardiography [85] or magnetic resonance imaging [86], although the use of such techniques is apparently limited in population-based screening. The number and runs of PACs and PVCs can be counted precisely, but we were unable to locate their origin from the surface ECG alone. The majority of PACs that trigger paroxysmal AF is known to occur within the pulmonary vein [44]. Of other ECG variables, the QT interval has been shown to be a risk for new-onset AF, but it was corrected differently by the formulas of Bazett, Framingham, Hodges and Fridercia. Uniformed measurements may be desired in screening candidates for incident AF in a population-based study.

Clinical risk variables and ECG-derived risks

There are well-known clinical predictors for new-onset AF (age, gender, hypertension, obesity, diabetes mellitus and heart failure) ([1]-[6],[79]), and all of these variables induce electrical and structural remodeling of the atrium and the ventricle. With advancing age, the heart increases in weight, ventricular wall thickness and valve circumferences, and the myocardium contains increased fat, collagen, elastin and lipofuscin [14]; these abnormalities all involve the LA. The age-related LA remodeling is accelerated by comorbidities, such as hypertension, diabetes mellitus or obesity. On the surface ECG, the maximal P-wave duration, P-wave dispersion and standard deviation of P-wave duration were observed to correlate with advancing age ([87]-[89]). LV dysfunction imposes a pressure overload on the left atrium and leads to electrical and structural remodeling of the atrium. P-wave durations showed a correlation with diastolic dimensions, ejection fractions and diastolic functions of LV ([28], [72]-[74]) and interstitial fibrosis as determined by MRI [75].

In hypertension, the conduction time is prolonged ([90]-[92]), and the P-wave duration is prolonged in the surface ECG [18],[93]. Furthermore, P-wave indices can be a marker of target organ damage [94]. P-wave indices are abnormal in type 2 diabetes ([95],[96]) and obesity ([93]) compared to those in healthy control individuals

Reversal of ECG-derived risks

Treatment of hypertension induces reversed remodeling of the heart [97]. With the use of angiotensin-2 receptor blockers or angiotensin converting enzyme inhibitors, prolonged P-wave durations or dispersions were observed to normalize ([98]-[102]), but this was not the case with calcium channel blockers [99].

These findings are compatible with the results of our experiments. In spontaneously hypertensive rats (SHRs), olmesartan (OM), an angiotensin 2 type 1 receptor blocker, reversed hypertrophy and fibrosis of the atrium more effectively than azelnidipine, a calcium channel blocker [103]. OM decreased atrial oxidative stress and activation of Rac1. The structural remodeling of the atrium was prevented by OM and a mineral corticoid antagonist eplerenone in Dahl salt-sensitive rats [104]. They induced an attenuation of oxidative stress and an attenuation of the Rac1-oxidative stress/inflammatory axis. The regression was associated with prevention of AF induction by electrical stimulation [104].

Through intensified treatment of blood pressure in patients with type 2 diabetes mellitus, a composite endpoint of new AF and prolonged P-wave indices and incident abnormal P-waves were shown to decrease (P=0.02 for both) in the ACCORD Blood Pressure Trials [105].

A significant decrease of P-wave duration or dispersion was observed in patients with loss of their original weight ([106],[107]). Severely obese patients (n=40) who underwent bariatric surgery showed a significant decrease of P-wave dispersion after surgery [108]. Treatment of underlying diseases can correct the abnormal P-wave indices, but their prophylactic efficacy for incident AF is yet to be determined.

Limitations

In the population-based studies, AF was diagnosed on ECGs recorded for approximately 10 seconds, and this may lead to the under diagnosis of PAF. The correlation between the ECG-derived predictors and clinical predictors for incident AF was studied in a small number of patients. It is unknown whether the reversal of ECG risk variables can prevent new-onset AF. Future study of a larger number of participants is needed. Recently, a genetic analysis has highlighted candidate genes or SNPs associated with AF development [109], but their relevance to ECG-derived predictors remains unknown

Acknowledgments:

This study was supported by the Practical Research Project for Life-Style-Related Diseases, including Cardiovascular Diseases and Diabetes Mellitus from the Japan Agency for Medical Research and Development (AMED) (17930285).

Conclusion

ECG-derived variables have been confirmed as a risk for incident AF from large population-based studies. The correlation between these markers with underlying disease or clinical predictors of incident AF needs to be shown in a study with a larger number of participants. Whether AF can be prevented by treatment of comorbidities requires prospective study.