Effect of Atrial Fibrillation on Atrial Thrombogenesis in Humans: Impact of Rate and Rhythm

Study population

The study comprised 55 patients with a history of AF and 15 patients with left-sided accessory pathways as a reference group who underwent catheter ablation. Consecutive patients with paroxysmal or persistent AF who were in sinus rhythm (SR) ≥48 h before the procedure (by continuous monitoring) were included. Exclusion criteria were history of valvular heart disease, left ventricular dysfunction, previous myocardial infarction, unstable angina, surgical or ablation procedure within the preceding 3 months, congenital heart disease, chronic inflammatory condition, chronic infection, chronic renal failure, chronic liver disease, and patients on antiplatelet agents.

All patients underwent baseline transthoracic echocardiography and transesophageal echocardiography within 2 days of the procedure (8). All antiarrhythmic medications were ceased 5 half-lives before the procedure. All patients underwent anticoagulation with warfarin to maintain their international normalized ratio between 2 and 3 for ≥6 weeks before the procedure. Warfarin was stopped 7 days before the procedure and substituted with enoxaparin at a dose of 1 mg/kg twice a day until ≥12 h before the procedure.

All patients provided written informed consent to the study protocol, which was approved by the Human Research Ethics Committee of the Royal Adelaide Hospital, Adelaide, Australia.

Electrophysiology study and ablation

Electrophysiological study and ablation were performed with sedation using midazolam and fentanyl. The technique used for mapping and ablation of AF in our laboratory was previously described (9). In brief, the following catheters were utilized for the procedure: 1) 10 pole catheter (Daig Electrophysiology, Minnetonka, Minnesota) positioned within the coronary sinus; 2) 10-pole circumferential catheter (Lasso; Biosense-Webster, Diamond Bar, California) to map the pulmonary veins; and 3) a 3.5-mm-tip externally irrigated ablation catheter (Thermocool, Biosense-Webster) for ablation. All patients underwent circumferential ablation of the pulmonary veins with the endpoint of electrical isolation. Additional substrate modification using either linear ablation (roofline and/or mitral isthmus) and/or ablation of complex fractionated atrial electrograms was undertaken in patients with long episodes of AF (>48 h), evidence of structural heart disease, or with a large LA (largest dimension > 57 mm).

Study protocol

For the clinical procedure, a conventional single transeptal puncture was performed using an SLO sheath (St. Jude Medical, St. Paul, Minnesota) and a BRK-1 needle (Daig Corporation). The ablation catheter was advanced through the same puncture into the LA. Following transeptal puncture, and 5 min after intravenous administration of unfractionated heparin (bolus of 100 IU/kg), blood samples were simultaneously collected from the peripheral femoral venous (FV) sheath (systemic sample), right atrial (RA) sheath, and LA sheath. Samples from the RA and LA were collected with care using a slow withdrawal technique, with the sheath positioned in the midchamber. Patients were then randomized either into the AF group, pacing group, or to serve as control patients.

Of the 55 patients with a history of AF who presented in SR, AF was induced by burst atrial pacing in 20 patients, commencing at a cycle length of 250 ms and ramping down to loss of 1:1 capture. This process was repeated up to 3 times from 3 sites, as required. Another 20 patients underwent atrial pacing at 150 beats/min. The rate of atrial pacing was limited to prevent induction of AF. To control for the effects of the transeptal puncture and procedure duration, 15 patients served as a control group, who neither underwent AF induction nor pacing. After 15 min of AF, atrial pacing, or in control patients, blood sampling was repeated from the LA, RA, and FV. No ablation was performed before the completion of the study protocol.

In addition to the control group of AF patients, to evaluate the effect of transeptal puncture on prothrombotic markers, 15 non-AF patients with left-sided accessory pathways who underwent an electrophysiological study and transeptal puncture during this period were also recruited as a reference group. Blood samples were obtained from the LA, RA, and FV following transeptal puncture at baseline. With the same exclusion criteria, consecutive patients with left-sided accessory pathways and no history of AF were evaluated.

Blood analysis

Statistical analysis

All data are expressed as mean ± SD or number (percentages) for continuous and categorical variables, respectively, unless otherwise stated. Continuous variables were compared using 1-way analysis of variance. Categorical variables were compared using Fisher's exact or Pearson's chi-square tests as appropriate. Data were tested for normality and log-transformed as appropriate. A constant of 1 was added to ADMA values before log-transformation. To compare changes in the outcome measures over time between the sites within each group, a linear mixed effects model was fitted to the data. In the model, time, sampling site, and the interaction between time and site were fitted as fixed effects, whereas individual patients were fitted as a random effect. This model takes the repeated measurements over time into account. Where the interaction was not significant, this was removed from the model so that the main effects of time and site could be interpreted. To compare between treatment groups, data were pooled across sites. A linear mixed effects model was fitted to the data. In the model, treatment group and time and the interaction between treatment group and time were fitted as fixed effects. Differences between sampling sites in the reference group were analyzed using repeated measures 1-way analysis of variance. Statistical significance was established at p < 0.05. All data were analyzed using PASW Statistics 18 (version 18.0.0, SPSS, Chicago, Illinois).

Patient characteristics

There were no significant differences among the AF, pacing, and control groups with regard to age, gender, comorbidities, medications, and echocardiographic parameters (Table 1). There was no difference in mean ventricular rate between the AF and pacing groups.

Platelet activation

Platelet P-selectin increased significantly with both AF induction (p < 0.001) and pacing (p = 0.02), taking into consideration all sites, as shown in Figures 1A and 1B. However, platelet P-selectin levels decreased in control patients over time (p < 0.001) (Fig. 1C). There was a significant difference between the sites measured (p = 0.04) in the pacing group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Platelet Activation (Platelet P-selectin)

(A) Post-AF induction: *p < 0.001 (time effect); p = 0.1 (site effect). (B) Post-pacing: *p = 0.02 (time effect); p = 0.04 (site effect); post-hoc tests at 15 min, p < 0.05 (LA vs. P). (C) Control group: *p < 0.001 (time effect); p = 0.4 (site effect). Data expressed as mean ± SEM. AF = atrial fibrillation; LA = left atrium; P = peripheral (femoral vein); RA = right atrium.

Thrombin generation

Thrombin generation (TAT) increased significantly in the LA and RA compared with peripheral samples in both the AF group (p < 0.001, site and time interaction) and the pacing group (p = 0.002), as shown in Figures 2A and 2B. Similar to the peripheral levels in the AF and pacing groups, TAT levels in the control group decreased with time (p < 0.001), with no differences in sites (p = 0.6) (Fig. 2C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Thrombin Generation (TAT)

(A) Post-AF induction: *p < 0.001 (site and time interaction); post-hoc tests at 15 min, p < 0.01 (LA vs. P), p < 0.01 (RA vs. P). (B) Post-pacing: *p = 0.002 (site and time interaction); post-hoc tests at 15 min, p < 0.01 (LA vs. P), p < 0.01 (RA vs. P). (C) Control group: *p < 0.001 (time effect); p = 0.6 (site effect). TAT = thrombin-antithrombin complex; other abbreviations as in Figure 1.

Endothelial dysfunction

ADMA levels increased significantly over time with the onset of AF (p = 0.008) (Fig. 3A). There was no change in ADMA levels with atrial pacing (p = 0.5) (Fig, 3B) or in control patients (p = 0.6) (Fig. 3C). There was no significant difference in ADMA levels between sites in the AF group (p = 0.2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

ADMA Levels

(A) Post-AF induction: *p = 0.008 (time effect); p = 0.2 (site effect). (B) Post-pacing: p = 0.5 (time effect); p = 0.3 (site effect). (C) Control group: p = 0.6 (time effect); p = 0.9 (site effect). ADMA = asymmetric dimethylarginine; other abbreviations as in Figure 1.

Platelet-derived inflammation

Soluble CD40L levels increased significantly over time with acute AF (p < 0.001), but were unchanged with atrial pacing (p = 0.8) and in control patients (p = 0.9), as shown in Figures 4A to 4C. There was no significant difference between the sites measured in the AF group (p = 0.7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Soluble CD40L Levels

(A) Post-AF induction: *p < 0.001 (time effect); p = 0.7 (site effect). (B) Post-pacing: p = 0.8 (time effect); p = 0.9 (site effect). (C) Control group: p = 0.9 (time effect); p = 1.0 (site effect). Abbreviations as in Figure 1.

AF versus pacing

We compared intracardiac levels (LA and RA) of thrombin generation and platelet activation among the AF, pacing, and control groups over time, given the difference between intracardiac and peripheral levels. For ADMA and sCD40L levels, comparison was made among the groups, taking into account all sites, given that no significant difference in sites were found.

Intracardiac platelet activation was significantly elevated in the AF and pacing groups compared with the control group (Fig. 5A). However, the difference between AF and pacing was not significant (p = 0.9). Intracardiac thrombin generation was significantly elevated in the AF and pacing groups compared with the control group (p < 0.01), but no significant difference was found between the AF and pacing groups (p = 0.4) (Fig. 5B). In contrast, endothelial dysfunction was significant in the AF group compared with the pacing group (p < 0.05) and the control group (p < 0.05) (Fig. 5C). In addition, inflammatory mediator sCD40L was raised in the AF group compared with the pacing group (p = 0.01) and the control group (p < 0.01) (Fig. 5D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Change in Prothrombotic Markers Between Groups

(A) Change in intracardiac platelet activation between groups: p < 0.001 (group and time interaction); post-hoc tests: *p < 0.01 (AF vs. control group), †p < 0.01 (pacing vs. control group), p = 0.9 (AF vs. pacing). (B) Change in intracardiac thrombin generation between groups: p < 0.001 (group and time interaction); post-hoc tests: *p < 0.01 (AF vs. control group), †p < 0.01 (pacing vs. control group), p = 0.4 (AF vs. pacing). (C) Change in ADMA levels between groups. p = 0.02 (group and time interaction); post-hoc tests: *p < 0.05 (AF vs. control group), †p < 0.05 (AF vs. pacing), p = 1.0 (pacing vs. control group). (D) Change in soluble CD40L (sCD40L) levels between groups: p = 0.001 (group and time interaction); post-hoc tests: *p < 0.01 (AF vs. control group), †p = 0.01 (AF vs. pacing), p = 0.6 (pacing vs. control group). NB: Descriptive plots shown. Statistical analyses performed using mixed effects models in which statistical significance was achieved. Abbreviations as in Figures 1 and 3.

Effect of transeptal puncture on prothrombotic markers

To eliminate transeptal puncture as a possible cause of the increased atrial prothrombotic markers observed in AF patients, a cohort of patients who underwent ablation of a left-sided accessory pathway was studied (8 men, 7 women; mean age 38 ± 7 years). Blood samples were obtained from the LA, RA, and FV following transeptal puncture. There were no significant differences in the levels of platelet P-selectin (in mean ± SEM) (log P-selectin 1.22 ± 0.07 in FV vs. 1.09 ± 0.08 in RA vs. 1.18 ± 0.07 in LA, p = 0.2), TAT (log TAT 1.29 ± 0.08 in FV vs. 1.23 ± 0.06 in RA vs. 1.22 ± 0.06 in LA, p = 0.2), ADMA (log ADMA 0.13 ± 0.01 in FV vs. 0.13 ± 0.01 in RA vs. 0.13 ± 0.01 in LA, p = 0.5), and sCD40L (log sCD40L 2.20 ± 0.05 in FV vs. 2.20 ± 0.05 in RA vs. 2.19 ± 0.05 in LA, p = 1.0) before or after transeptal puncture or between sampling sites in these patients.

Left atrial platelet activation with AF and pacing

Patients with AF are recognized to exhibit a prothrombotic state and abnormal platelet activation. This has been documented peripherally in various subsets of patients with AF (10–12). Platelets play an essential role in thrombogenesis by interacting with the endothelium, inflammatory cells, and proteins from the coagulation cascade (5,13). Platelet expression of P-selectin is commonly used as a marker of platelet activation (5). Platelet P-selectin expression has been associated with spontaneous echo contrast, presence of LA thrombus or embolic events, and silent cerebral infarction in patients with AF (6,14).

Although several studies have shown increased platelet activation in patients with AF compared with control patients (11,15,16), other studies have shown that the difference could be attributed to patient comorbidities (17). These inconsistent results could partly be explained by sampling from a heterogeneous population of AF patients, sampling from peripheral versus central sites, and by measuring at different times during AF rather than at a predefined time (5,15). In this study, the effect of AF was compared with each patient's individual baseline state, and sampling was performed in the LA at a predefined time. We showed that AF per se and rapid atrial rates resulted in elevated platelet activation.

In addition, platelet expression of P-selectin has been suggested to be linked more with acute changes in platelet activation (18). Although studies measuring patients with AF at a baseline state have yielded varying results, studies examining patients with acute episodes of AF have consistently shown the involvement of platelets (15,16,19). Our present findings are consistent with other studies in that platelet activation is enhanced in the setting of acute AF, after 3 to 12 h from peripheral sampling (16,19), and after 15 min from coronary sinus sampling (15). Platelet activation may thus play a role in the initiation of various prothrombotic pathways in the acute or paroxysmal setting.

Left atrial thrombin generation with AF and pacing

Increased thrombin generation reflected by elevated TAT levels has been found in the LA in patients with mitral stenosis and AF (4). In chronic nonvalvular AF, increased peripheral levels of coagulation markers have been demonstrated (10,12). Abnormal expression of coagulation factors in the atrial endocardium has been demonstrated in an animal model of rapidly paced atria and in human patients presenting with AF and cardiogenic thromboembolism (20,21). Akar et al. (15) found increased thrombin generation and platelet activation with AF induction from samples taken from the coronary sinus. In this study, acutely elevated thrombin generation was observed specifically at the LA and RA with the onset of AF and rapid atrial rates, which was not seen in the peripheral circulation. This study demonstrates that with the onset of AF and rapid atrial rates, the LA is significantly more thrombogenic compared with the peripheral circulation, and may explain the propensity for LA thrombus formation and cardioembolic stroke seen in these patients.

Endothelial dysfunction with AF induction

LA thrombus formation in AF has traditionally been attributed to atrial mechanical dysfunction (8,22,23). However, it is now accepted that multiple mechanisms contribute to LA thrombus formation. Abnormal endothelial function (endothelial dysfunction) has been shown in patients with AF and is a component of Virchow's triad for thrombogenesis (2,24,25). ADMA is an endogenous inhibitor of endothelial nitric oxide synthase and is known to result in endothelial dysfunction in experimental human studies (26). Nitric oxide has potent antithrombotic properties on the endothelium and inhibits platelet and monocyte adhesion (27). There is also evidence that ADMA mediates endothelial dysfunction through oxidative stress (26). Clinically, ADMA is associated with numerous cardiovascular conditions and is a predictor of mortality in cardiovascular patients (28,29).

The present study found that induction of AF was associated with increased ADMA levels both peripherally and in the human atria. The finding that induction of AF upregulates ADMA is consistent with animal models, such as the porcine AF model by Goette et al. (28). In another animal study, Cai et al. (7) demonstrated decreased atrial nitric oxide levels and endothelial nitric oxide synthase expression in a rapid atrial pacing model of AF, which was not seen in the control group of atrial pacing at 100 beats/min. Minamino et al. (6) found decreased nitric oxide levels with associated increased P-selectin expression on platelets in a canine model of AF. These findings suggest that endothelial dysfunction induced by AF, mediated by ADMA and the nitric oxide pathway, with resultant loss of its antithrombotic properties, plays an important contributory role to thrombogenesis in patients with AF.

Inflammation with AF induction

Inflammation is being increasingly recognized to play a significant role in the genesis and perpetuation of AF (24,30). C-reactive protein elevation is found in a stepwise fashion in patients with increasing AF burden (31). Studies also show that high-sensitivity C-reactive protein decreases after successful ablation for long-standing persistent AF, suggesting that AF itself may cause an inflammatory response (32).

Soluble CD40 ligand is increased in a number of cardiovascular settings (33). It is an important mediator in the pathogenesis of atherothrombotic disease and predicts mortality in patients with acute coronary syndromes (33,34). CD40 ligand on activated platelets plays a pivotal role in inflammatory responses by inducing endothelial secretion of chemokines and expression of adhesion molecules, thereby promoting leukocyte recruitment (35). The CD40/CD40 ligand system has been proposed to provide an important link between inflammation and thrombosis (34). This study showed that the onset of AF increased sCD40L levels, which was not seen with rapid atrial rates alone. This study newly demonstrates that induction of AF in humans evidently results in an increase in inflammatory signals. This provides further insight into the link between inflammation and thrombogenesis in patients with AF.

Clinical implications

This study demonstrates that AF or abnormal rhythm per se confers additional prothombotic effects in the LA beyond the patient's comorbidities. These findings point toward the benefit of maintaining SR and provide an explanation for the “on-treatment” analysis of the AFFIRM (Atrial Fibrillation Follow-up Investigation of Rhythm Management) study, which found the presence of SR associated with a lower risk of death (36). The present study also explains why increased AF burden and subclinical atrial tachyarrhythmias (from implantable devices) are associated with increased thromboembolic risk (37,38).

The finding that rapid atrial rates increase platelet and thrombotic markers in the LA provides mechanistic insight into thrombogenesis in atrial flutters, which may differ slightly from AF, but nevertheless confer increased risk of stroke (3,39,40). Furthermore, AF potentiates the thrombogenic risk over that of rate alone by activating other mechanisms, such as endothelial dysfunction and the inflammatory cascade. This highlights the importance of other therapeutic modalities that improve endothelial function and mediate the inflammatory response, and the management of concomitant cardiovascular risk factors associated with AF (30).

Study limitations

With access to the human LA, the study would have been ethically impossible without the administration of heparin. Heparin is known to possibly affect TAT levels by initially enhancing binding before causing irreversible inhibition of thrombin's activity (41). In our study, TAT levels at all sites in the control group and at peripheral sites in the AF and pacing groups decreased with time after the administration of heparin, which was consistent with previous studies (15). However, despite decreased TAT levels in control patients and in peripheral samples with heparin, atrial levels significantly increased with AF and atrial pacing.

The half-life of TAT is 15 min (42) and that of ADMA is approximately 23 min (43), close to the study sampling time. However, previous studies in humans have documented acute changes in levels of platelet P-selectin at 15 min (15), TAT at 15 min (15), and ADMA when measured at 15-min intervals (44).

Underlying patient comorbidities could have contributed to the prothrombotic state. However, the study measured the effects of AF and pacing compared with each patient's baseline state. Hence, these results indicated that the effects of AF and high atrial rates were in addition to a patient's underlying comorbidities.

Ventricular rate in both AF and pacing groups were faster than the control group and could have contributed to the increased effects. However, other studies have demonstrated that patients controlled for ventricular rate either in a paced setting or in paroxysmal supraventricular tachycardia did not show any significant difference in prothrombotic markers (11,15,19,28). The degree of rapid atrial pacing was limited to 150 beats/min to prevent inadvertent induction of AF. It is possible that the further findings observed in the AF group were due to its higher atrial rates, although this could not be explicated in the study.