Atrial fibrillation (AF) is the most commonly sustained atrial arrhythmia and increases in prevalence with advancing age. It is detected in 3% to 4% of the population over age 60 years and has a prevalence of almost 9% in octogenarians (1). United Nations sources estimate that 19% of the world's population is over 60 years of age with this predicted to rise to 33% by 2050. Aging is the most frequently attributable risk factor for AF. By the ninth decade of life, AF is the most important risk factor for stroke. In addition, AF has a deleterious effect on longevity with a doubling of all-cause mortality.

Despite the importance of the association between AF and aging, little is known of the underlying atrial substrate that develops with senescence. Potential triggers for AF have been shown to increase with advancing age with almost 50% over 65 years of age having episodes of supraventricular tachycardia detected on 24-h monitor (2).

There is a relative paucity of data detailing the age-related changes in atrial substrate. Prior animal studies of aging-related changes in the atrium have demonstrated the presence of interstitial fibrosis associated with variable conduction slowing (3- 4). Similarly, in isolated human preparations, Spach et al. (5) demonstrated the age-related development of extensive collagenous septa that separated small groups of atrial fibers, the consequence of which was a decrease in transverse conduction velocities. However, detailed human electrophysiologic studies describing the nature of age-related atrial remodeling are lacking. The aim of the present study was to characterize the pattern of atrial remodeling seen with human aging. We performed detailed prospective electrophysiologic and electroanatomic studies to describe these changes.

We prospectively studied three groups of patients:

In all groups, the study population consisted of patients undergoing electrophysiology study with a view to radiofrequency ablation (RFA) of supraventricular tachycardia. Patients were excluded from the study if there was a history of hypertension, structural heart disease, or previous use of amiodarone. In order to clearly characterize the type of atrial electrophysiologic and electroanatomic changes seen with aging without the confounding effects of atrial arrhythmias, patients with a prior history of such arrhythmias were excluded. Patients with clinical evidence of sinus node dysfunction were also excluded. All patients underwent two-dimensional echocardiography to exclude structural heart disease. The baseline characteristics of the study cohort are presented in (Table le1). All antiarrhythmic medications were ceased at least five half-lives before the study. All patients gave written informed consent according to a protocol approved by the Melbourne Health Clinical Research and Ethics Committee.

After the diagnostic electrophysiology study and RFA procedure, autonomic blockade was administered (atropine 0.04 mg/kg and propranolol 0.2 mg/kg over 10 min) before commencement of the research protocol. Multipolar catheters were positioned as previously described 6. In brief, catheters were positioned in the following manner: 1) coronary sinus (CS) catheter (10-pole, 2-5-2 mm interelectrode spacing, positioned with the proximal bipole at the ostium of the CS by best septal left anterior oblique projection; 2) crista terminalis (CT) catheter (20-pole, 1-3-1 mm interelectrode spacing, positioned along the CT with the aid of a long sheath to ensure stability and facilitate its close apposition to the CT; where necessary, positioning of this catheter was aided with the use of intracardiac echocardiography (9 MHz) as previously described 7; 3) lateral right atrial (LRA) catheter (20 pole, 2-5-2 mm interelectrode spacing was positioned such that the first 10 electrodes were in a linear arrangement along the LRA border in the left anterior oblique projection and in an anterior position in the right anterior oblique projection; and 5) high septal right atrial (RA) catheter (8 pole, 2.5-5-2.5 mm interelectrode spacing) along the high septal RA.

Bipolar intracardiac electrograms filtered between 30 and 500 Hz were recorded and stored digitally on a computerized system simultaneously with 12-lead surface electrocardiograms (ECG). Off-line analysis was performed using on-screen digital calipers at 400 mm/s speed.

Atrial ERP was measured from four sites: distal CS, low LRA, high lateral RA, and high septal RA. Atrial ERP was measured at twice diastolic threshold (for a pacing threshold of <2 mA) at cycle lengths of 600, 500, and 400 ms. At each site, the ERP was measured 3 × during each cycle length, and, if the maximum and minimum amounts differed by >10 ms, two additional measurements were taken and the total averaged. Heterogeneity of atrial ERP was measured as the coefficient of variation of ERP at each cycle length. The coefficient of variation was calculated as the standard deviation as a percentage of the mean ERP at each cycle length (standard deviation/mean × 100%) (8).

P-wave duration in sinus rhythm was measured from lead II on the surface ECG as an average of 10 consecutive beats.

Conduction time was measured in both atria at pacing cycle lengths of 600, 500, and 400 ms during stable capture. Local conduction time was measured along the CS by pacing from the distal bipole (1-2) and measuring time to activation at the proximal bipole (9-10). This was repeated along the LRA border pacing from the distal bipole (1-2) and measuring time to activation at bipole 9-10. Conduction time was measured from the pacing artefact to the onset of the first initial deflection recorded by the proximal bipole of the same catheter. This was calculated as the average of 10 consecutive measurements.

Conduction delay at the CT was assessed during pacing at cycle lengths of 600, 500, and 400 ms and during the earliest extrastimulus that conducted to the atrium from the distal CS, low and high LRA.

Conduction delay at the CT was analyzed by the presence of discrete double potentials. Double potentials were defined as potentials separated by an isoelectric interval of ≥50 ms (9).

The corrected sinus node recovery time was assessed at cycle lengths of 600, 500, and 400 ms after a 30-s pacing train. The longest time interval from the last paced atrial beat to the first spontaneous atrial depolarization was recorded as the sinus node recovery time_max (SNRT_max).

Right atrial electroanatomic maps were created during constant pacing. The distal bipole of a 10 pole catheter positioned in the CS was selected to ensure pacing stability. A long vascular sheath was used to ensure stability of the mapping catheter.

Endocardial contact during mapping was ensured by:

Atrial points were then acquired if the stability criteria in space (≤6 mm) and in local activation time (≤5 ms) were met. Editing of points was performed offline. Local activation time was manually annotated to the beginning of the first deflection from the isoelectric line on bipolar electrograms. Points were deleted if the P-wave morphology on 12-lead ECG did not match the majority of points or if the annotated electrogram was <75% of the maximum voltage of the preceding electrogram.

On the electroanatomic maps, fractionated signals and double potentials, as previously defined, were tagged with brown dots and blue dots, respectively. Fractionated signals were defined as complex activity of ≥50 ms duration (9). Electrically silent areas (scar) were defined as the absence of recordable activity or a bipolar voltage amplitude of ≤0.05 mV (6). Low voltage areas were defined as contiguous areas ≤0.5 mV (6).

Atrial bipolar voltage was analyzed as the mean of 10 representative points at each of the following six RA sites: high and low septal RA, high and low LRA, high and low posterior RA. The probability density graph of surface area to bipolar voltage is presented as previously described (6).

The electroanatomic mapping system calculates the conduction velocity between two points by expressing the linear distance between the points as a function of the difference in the local activation times. Isochronal maps of the atria were created at 5-ms intervals in local activation times to determine regional conduction velocity. Conduction was determined as the mean of the conduction velocity between five pairs of points along the activation front through regions of least isochronal crowding at the six RA sites.

An index of heterogeneity for both voltage and conduction velocity were obtained by calculating the coefficient (standard deviation/mean × 100%) of the entire RA.

Data are expressed as mean ± SEM. Statistical analysis was performed utilizing GB-Stat software (Version 6.5, Dynamic Microsystems Inc., Silver Spring, Maryland). Comparisons between groups were performed by analysis of variance, with post-hoc analysis using the Newman-Keuls procedure. Pearson's correlation coefficient was used to evaluate the relation between atrial variables and aging. Statistical significance was assumed at p < 0.05.

Thirteen patients age ≥60 years (nine males, four females; group A), 13 patients age 31 to 59 years (five males, eight females; group B), and 15 patients age ≤30 years (six males, nine females; group C) were enrolled in the study. Patient characteristics are presented in (Table le1). The mean age in group A was 66.4 ± 1.7 years, in group B was 50 ± 2.1 years, and in group C was 24.7 ± 1.0 years (p < 0.0001). Echocardiographic parameters were well matched between the three groups. In group A, nine patients underwent RFA for either atrioventricular nodal re-entrant tachycardia or atrioventricular reciprocating tachycardia with the remaining four having normal electrophysiologic studies for syncope. In group B, 10 patients underwent RFA for either atrioventricular nodal re-entrant tachycardia or atrioventricular reciprocating tachycardia with the remaining three having normal electrophysiologic studies for syncope. In group C, 14 patients had successful RFA for either atrioventricular nodal re-entrant tachycardia or atrioventricular reciprocating tachycardia with one patient having successful RFA for right ventricular outflow tract ventricular tachycardia.

At all atrial sites and at all cycle lengths, the ERP was longer in group A compared with group C, with group B being intermediate. These differences in ERP were statistically significant using a repeated measures analysis of variance at the distal CS at two cycle lengths (600 ms and 500 ms; p < 0.01), at the low LRA at one cycle length (400 ms; p < 0.05) and at the high septal right atrium at all three cycle lengths (p < 0.01). There was a positive correlation between the ERP and age at each of three cycle lengths (r = 0.56 at 600 ms, r = 0.55 at 500 ms, r = 0.58 at 400 ms; p < 0.001).

There was no significant change in the heterogeneity of atrial ERP at any cycle length as determined by the coefficient of variation between the three groups.

Rate adaptation of refractoriness (ERP₆₀₀ − ERP₄₀₀) was maintained at each site with aging (LLRA: group A, 14 ± 9.3 ms; group B, 12.3 ± 0.8 ms; and group C, 14.6 ± 4.3 ms; p = NS).

Corrected SNRT_max demonstrated a significant increase with age as demonstrated in (Figure 1).

There was a significant increase in P-wave duration with age (group A, 103.5 ± 1.9 ms; group B, 96.8 ± 3.9 ms; group C, 91.4 ± 2.9 ms; p < 0.01). A positive correlation was evident between P-wave duration and aging (r = 0.60, p < 0.01).

Conduction time increased with age at all three cycle lengths tested when pacing the distal coronary sinus and the LRA (Table le2). This was significant at all cycle lengths in the CS and at a cycle length of 400 ms in the LRA.

During the electrophysiologic study, there was significant conduction delay at the CT in group A compared with groups B and C as evidenced by the increased number of double potentials recorded along this structure (Table le2). This increase was observed during the basic drive cycle, but was significantly amplified with the tight coupled extra-stimulus (Table le2) when pacing from high and low lateral RA and distal CS.

Electroanatomic activation mapping also demonstrated more extensive areas of double potentials and fractionated signals in groups A and B compared with group C. There was a strong correlation between advancing age and increasing double potentials and fractionated signals (p < 0.01, r = 0.67). These are represented in (Figure 2) and are seen to particularly cluster along the posterior RA.

The total percentage of points demonstrating abnormal conduction (fractionated or double potentials) was significantly greater in group A (16.0 ± 2.9%) and group B (12.3 ± 3.5%) versus group C (3.1 ± 1.0%) (p < 0.01 and p < 0.05, respectively).

Regional atrial conduction velocity was slower in all regions assessed in the RA in groups A and B compared with C (Figure 3) (p < 0.05 for groups A and B compared with C).

The differences in conduction velocity were statistically significant in all of the predefined regions: the high septum, low septum, high lateral, low lateral, high posterior, and low posterior (Figure 3). There was no significant difference in the heterogeneity of conduction.

A total of 238.4 ± 12.1 points were analyzed in group A, 230.7 ± 20.2 in group B, and 256.3 ± 39.2 in group C (p = NS). There was no significant difference in RA volumes between the three groups of patients (group A, 84.5 ± 4.4 ml; group B, 67.8 ± 4.2 ml; and group C, 66.6 ± 7.7 ml).

There was a significant reduction in the mean total RA bipolar voltage amplitude in group A (1.9 ± 0.1) and group B (1.6 ± 0.1) compared with group C (3.0 ± 0.3; p < 0.01). A strong correlation was seen between advancing age and decreasing RA voltage (p < 0.01, r = −0.62).

In addition, bipolar voltage mapping revealed increased low voltage as a percentage of the total points (≤0.5 mV) in the group A and group B patients compared with group C patients (Figure 2). Areas of electrical silence (≤0.05 mV) or scar were not seen in any patients.

The probability density plot of voltage surface area demonstrates the peak density of surface area in groups A and B in a lower voltage range than in group C (Figure 4). In group C, the voltage surface area did not show an early peak but rather a plateau between 1.0 mV and 3.5 mV.

Regional atrial bipolar voltage amplitude was significantly lower at all RA sites in group A compared with group C and significantly lower at four of six sites in group B relative to group C (Figure 5).

In addition, the heterogeneity of voltage as determined by the coefficient of variation was significantly greater in group A versus group C (70.1 ± 3.0% vs. 56.6% ± 2.2%; p < 0.05). Voltage dispersion in group B was intermediate at 65.9 ± 3.4% but not significantly different from either group A or C.

This study presents detailed prospective information on the atrial electrophysiologic and electroanatomic changes associated with aging. The following changes were seen with increased age:

Several prior studies have evaluated the specific effects of increased age on the electrophysiologic properties of the atrium. There have been conflicting results regarding the effect of aging on atrial ERP with some studies showing an increase (12- 13), while others have suggested no change (14). Studies to date have been limited with restriction to single RA sites only. Early studies by Dubrow et al. (12) examined functional and ERPs at varying cycle lengths in children at single atrial sites and compared them with adult controls from an earlier study. In children, observed refractory periods were shorter at cycle lengths ranging between 280 and 600 ms. Sakabe et al. (13) reported an increase in the “atrial vulnerability index” in humans with aging, which included prolongation of atrial ERP at a single site and cycle length. This cohort included patients with atrioventricular nodal disease who may have had more extensive conduction disease. In a retrospective study of 354 patients who had undergone clinically indicated electrophysiologic testing, Taneja et al. (14) found no prolongation in atrial ERP with aging.

In the present study in patients with no prior history of atrial arrhythmia, we found an increase in atrial ERPs with aging. An increase in ERP was seen at all sites tested and at all cycle lengths. Using a rigorous between-group analysis of variance, this increase was significant at three of the four sites and three cycle lengths. There was no increase in the heterogeneity of refractoriness with aging.

Prior animal studies of aging-related changes in electrophysiology have demonstrated development of conduction slowing, frequently associated with the presence of interstitial fibrosis. In a study of the cellular electrophysiology of old canine atria, Anyukhovsky et al. (4) showed a reduction in the conduction velocity with an increased window of conduction slowing in response to premature stimuli but not during normal beats. In that study, there was a twofold increase in the amount of fibrous tissue detected in old atria. Similarly, Hayashi et al. (3) observed that interatrial conduction time and P-wave duration were both prolonged in association with increased inducibility of AF in the Langendorff perfused aged rat heart. Histologic evaluation of these atria again demonstrated an increase in atrial fibrosis together with cellular hypertrophy, which was presumed to be the cause of conduction slowing.

Clinical studies of the effect of aging on atrial conduction are limited. Spach et al. (5) demonstrated anisotropic conduction at a microscopic level in isolated human preparations. Electrical uncoupling of the side-to-side connections between small groups of fibers with aging produced a “zigzag” course of transverse impulse propagation. This correlated with the development of extensive collagenous septa that separated small groups of fibers, the consequence of which was an age-related decrease in transverse conduction velocities. Sakabe et al. (13) reported a positive correlation between percentage atrial fragmentation index as a marker of conduction and advancing age in 73 patients with no history of AF or structural heart disease. P-wave duration on signal-averaged ECG has been shown to prolong with advancing age (15).

In the current study, we observed evidence of generalized conduction slowing in the atria both on standard multipolar electrophysiology catheters deployed along the CS and lateral RA and also during electroanatomic mapping. In addition, with advancing age, there was development of increasing numbers of fractionated and double potentials. These were observed generally in the atrium but also particularly along the CT. This may be important because functional conduction delay at this structure has been implicated in the development of a range of atrial arrhythmias. In this study, while we did not have access to pathology samples of the atrium, electroanatomic voltage maps demonstrated significant reduction in voltage amplitude. We hypothesize that this reduction in voltage may reflect development of atrial fibrosis. Recent studies suggest that this loss of voltage may also relate to development of atrial amyloidosis, which has been observed in more than 90% of octogenarians (16). In the absence of histologic data, the relationship between RA voltage decrease and structural change remains speculative.

Advancing age was associated with a significant increase in sinus node recovery time at all cycle lengths in the current study. An increase in sinus node recovery time has been previously correlated with age in a review of individuals undergoing electrophysiologic testing (14). Histopathology of the sinus node complex reveals an increase in elastic fibers along with fatty infiltration and irregularity in the distribution and size of muscle fibers with aging (17).

The changes associated with increased age seen in patients in the current study might be expected to provide the substrate for multiple wavelet re-entry. Generalized conduction slowing and regional functional conduction delay together with structural abnormalities are important factors in establishing the basis for re-entry. While an increase in ERP might potentially be protective, it is probable that this effect is overshadowed by the conduction changes.

Indeed, prior studies of different atrial substrates associated with an increased incidence of AF have observed similar findings (18- 19). Conduction delay has been previously reported in animal models of rapid atrial pacing with a delayed onset compared with changes in refractoriness (20). Cardiac failure has been shown to increase propensity to AF both in animal (19) and human studies (6) due to conduction slowing and structural changes and despite an increase in refractoriness. Similarly, chronic atrial stretch due to the volume overload of an atrial septal defect causes regional conduction slowing and increased atrial refractoriness (18). Similar findings have been observed when chronic atrial stretch occurs in paced patients due to loss of atrioventricular synchrony (21). However, the effects of acute atrial stretch on refractoriness have been inconsistent with shortening of ERP seen in several isolated cardiac preparations (22- 23).

Importantly, conduction delay at the CT has been shown to play an important part in the development of both typical and atypical forms of atrial flutter and of AF (24). In the present study, fractionated signals and double potentials became far more evident with increasing age, particularly in the region of the CT.

Finally, the presence of areas of low voltage has been associated with chronic atrial arrhythmias (9). The importance of lines of block and silent areas in the stabilization of left atrial flutter circuits has been previously documented on electroanatomic mapping (9). In the current study, aging was associated with not only a significant reduction in voltage but an increase in the heterogeneity of voltage. It was initially noted in the landmark study by Wijffels et al. (11) that the changes in atrial refractoriness did not parallel the stabilization of AF. This led Allessie et al. (25) to identify the presence of a “second factor” involved in the perpetuation of chronic AF, which they hypothesized to be structural remodeling. In a study in goats with chronic AF, Ausma et al. (26) observed changes in cellular ultrastructure without any alteration in tissue structure. In a subsequent study, they demonstrated that these changes had not completely resolved as late as four months after return to sinus rhythm (27). In dogs with atrial remodeling due to chronic heart failure, Shinagawa et al. (28) reported the persistence of local conduction abnormalities and increased AF duration associated with atrial fibrosis as late as five weeks after resolution of left ventricular dysfunction. We hypothesize that the changes in voltage and conduction seen with aging represent the structural remodeling, which may be important in the development of AF.

The development of AF depends not only on substrate but also, importantly, on pulmonary vein triggers, which were not addressed in the present study. Whether the abnormalities observed in patients with aging are responsible for the increased incidence of AF in older people remains speculative, as those with atrial arrhythmias were necessarily excluded from the study. In this clinical study, detailed evaluation was predominantly of the RA with assessment of ERP and conduction in the CS. There is increasing evidence, however, that findings in the CS may not always reflect those of the left atrium (29). Nevertheless, it is reasonable to expect that both atria would be affected in a similar manner.

This study demonstrates that aging is associated with atrial remodeling characterized by: anatomical and structural changes, reductions in atrial voltage with discrete areas of low voltage, widespread conduction slowing as well as anatomically determined functional conduction delay and block, and sinus node dysfunction. These changes may, in part, be responsible for the increased propensity to atrial arrhythmias observed with increasing age.