This study was approved by the Institutional Review Board of Anam Hospital of Korea University. All patients provided written informed consent. Eighty-one patients with PeAF (male:female=63:18, mean age 59.1±10.5 years old) who underwent external electrical CV were included in the study. We excluded patients with a history of any previous CV, significant mitral valvular heart disease, a huge left atrium (LA; > 55 mm), a recent infection, surgery or acute coronary syndrome in the 2 months prior to the collection of blood samples. All patients maintained optimal anticoagulation, had been taking anti-arrhythmic drugs for at least 1 month, and maintained them after CV. Transesophageal echocardiography was done to exclude an intra-cardiac thrombus on the same day of CV in every patient. The blood samples for the serologic assays were drawn before sedation for external CV.

After obtaining written informed consent, electrical CV was performed under sedation with intravenous midazolam (0.05 mg/kg) and thiopental sodium (60 mg/kg). A biphasic R wave synchronized shock (Lifepak12, Physiocontrol Ltd. Redmond, WA, USA) was applied to the patients via self-adhesive skin electrodes (TZ Medical Inc., Portland, OR, USA) in an anteriorposterior position. We delivered an initial CV with 70 J. If the initial shock failed to terminate AF, the biphasic shock energy was gradually increased to 100 J, 150 J, and then 200 J serially (5 minutes intervals). If CV terminated AF successfully, the patient's cardiac rhythm was monitored for 15 minutes to detect an atrial premature beat (APC) or recurrence of AF. If AF returned within 15 minutes of termination of CV, amiodarone 150 mg was administered intravenously and the same energy shock was repeated. Patients in whom AF remained even after being given a 200 J CV or who exhibited repeated immediate recurrence of AF in spite of amiodarone were defined as failed CVs.

We acquired a 5 mL sample of peripheral blood immediately before CV to measure the plasma levels of several protein markers or chemokines by enzyme linked immuno-sorbent assay: pro-atrial natriuretic peptide (ANP; Biomedica, Antony, France), matrix metalloproteinase (MMP)-9, transforming growth factor (TGF)-β, and stromal cell derived factor (SDF)-1α (R&D Systems, Minneapolis, MN, USA). High sensitivity C-reactive protein (hs-CRP) was measured on a Hitachi 912 assay system (Roche Diagnostics, Indianapolis, IN, USA) using a Kamiya K-assay (Kamiya Biomedical Corp., Seattle, WA), which quantitatively determines CRP by a latex particle-enhanced immuno-turbidimetric assay.

After CV, all patients were prospectively followed up at an outpatient clinic at 1, 2, 4, 8 weeks, and then every 3 months thereafter. Elecrocardiography (ECG) was performed at every visit or anytime the patient reported palpitations. A Holter ECG (24 hours or 48 hours) and/or event recorder was evaluated at 2 months in patients who did not experience recurrence. We classified patients into 4 different groups according to clinical outcome as follows: 1) failed CV, 2) early recurrence of AF (ERAF; within 2 weeks), 3) late recurrence of AF (LRAF; after 2 weeks), and 4) no recurrence (NR) after 2 months.

We evaluated the clinical factors {e.g., age, sex, LA size, ejection fraction (EF), spontaneous echo contrast (SEC)} and serologic factors (MMP-9, SDF-1α, TGF-β, pro-ANP, and hs-CRP) in terms of success or failure of CV and the timing of recurrence. Comparisons between groups were analyzed by the Mann-Whitney test or a t-test. In order to identify the predictors of failed CV or the recurrence of AF after successful CV, uni-variate and multivariate logistic regression analyses were performed. AF free rates were compared in terms of clinical, electrophysiological, and serological parameters by utilizing Kaplan-Meier curves. We chose cutoff numbers that were similar to the mean that divided 2 groups with similar numbers of the patients. All values are expressed as mean±SD. All statistical analyses were performed using Statistical Package for the Social Sciences (SPSS) version 12.0 and a p<0.05 was considered statistically significant.

We performed external electrical CV in 81 patients (59.1±10.5 years old, 63 males) with PeAF or permanent AF. The mean LA size was 45.5±6.0 mm, and the mean EF of the LV was 48.7±15.1%. All patients had been taking antiarrhythmic drugs for at least 1 month before and after CV (amiodarone 60.5%, other class IC anti-arrhythmic drugs 39.5%). The mean number of electrical shocks performed was 1.9±1.4, the final successful CV energy was 96.7±32.3 J, and the cumulative CV energy was 170.4±178.5 J. Among the 81 patients, 8.6% (7/81) showed failed CV and 91.4% (74/81) underwent successful CV. We followed up the 74 patients after their successful CV for an average of 13.2±11.0 months. Of the 74, 22 (29.7%) showed AF recurrence within 2 weeks (ERAF), 26 (35.1%) after 2 weeks (LRAF), and 26 (35.1%) remained in SR and had NR during the follow-up period.

Table 1 summarizes comparisons of the clinical and biochemical parameters in patients with successful and failed electrical CV. There were no significant differences in age, sex, LA size, EF, the existence of SEC, or medications. However, the prescription rate of amiodarone was lower in patients with failed CV compared to those with successful CV (28.6% vs. 64.9%, p=0.0301). Both the final energy (121.4±26.7 J vs. 93.7±31.8 J, p=0.0154) and the incidence of the requirement of intravenous amiodarone (100% vs. 14.9%, p<0.0001) were higher in patients with failed CV. The pre-CV plasma level of TGF-β, which reflects the degree of fibrosis, was significantly higher in patients with failed CV (22.7±22.9 ng/mL) than in those with successful CV (13.5±9.8 ng/mL, p=0.0260).

The recurrence rate of AF after successful CV was 64.9% (48/74) during the 13.2±11.0 months follow-up. Table 2 summarizes comparisons of clinical, electrophysiological, and serological parameters. The patients who showed recurring AF after CV were older than those without recurrence (60.4±9.0 years old vs. 55.3±12.5 years old, p=0.0220). There was no significant difference in either LA size or EF. Pre-CV plasma levels of SDF-1α (3.2±1.2 ng/mL vs. 3.9±0.8 ng/mL, p=0.0105) were significantly lower in patients with recurrent AF than in those without recurrence. In univariate logistic regression analyses, it was found that old age {odds ratio (OR) 1.053, 95% confidence interval (CI) 1.004-1.053, p=0.035} and lower SDF-1α (OR 0.603, 95% CI 0.379-0.960, p=0.033) may predict the recurrence of AF after successful CV. In multivariate logistic regression analysis, low SDF-1α (OR 0.537, 95% CI 0.302-0.954, p=0.034) was an independent risk factor for AF recurrence after successful CV. In Kaplan-Meier analyses, the AF recurrence rate was higher in patients who were older than 60 (p=0.0295) and had plasma SDF-1α <3.0 ng/mL (p=0.0114) (Fig. 1).

Among 48 patients whose AF recurred during the follow-up period after successful CV, 45.8% (22/48) recurred within 2 weeks (ERAF), while 54.2% (26/48) recurred after 2 weeks (LRAF) (Table 3). The mean time to LRAF was 85.5±134.7 days. However, we failed to find a significant difference in clinical and serological parameters between ERAF and LRAF groups (Table 3).

We compared patients with LA ≥45 mm to those with LA <45 mm. In patients with LA ≥45 mm, the incidence of SEC (69.8% vs. 45.5%, p=0.0164) was higher. However, the failure rate of CV and the recurrence rate of AF after successful CV were not affected by LA size. Although the serological parameters were not different between the two groups, MMP-9 (139.6±49.97 ng/mL vs. 107.78±65.15 ng/mL, p=0.0285) and SDF-1α (3.76±1.26 ng/mL vs. 3.24±0.99 ng/mL, p=0.0261) were significantly higher in patients with LA ≥48 mm than in those with LA <48 mm.

In this study, a low plasma level of SDF-1α was an independent risk factor for AF recurrence after successful CV. The chemokine SDF-1α has been shown to play a key role in hematopoietic or endothelial progenitor cell trafficking in the myocardial ischemia model,13-15) and the non-ischemic titrated cardiac injury model during catheter ablation of AF.16) However, the role of SDF-1α has been poorly explored in the pathophysiology of AF. Goette et al.17) report that SDF-1α levels are higher in PeAF than PAF or controls, and SDF-1α plays some role in the restitution of hematopoietic progenitor cells after failed CV. The reason for the high recurrence of AF in patients with low plasma levels of SDF-1α remains to be studied. Although it has been reported that high levels of hs-CRP are associated with an increased risk of recurrence of AF,12)18) our data and data of Conway et al.19) failed to prove its predictive value in the clinic. The plasma level of hs-CRP may be associated with the permanence of AF related to systemic inflammation or prothrombotic status.20)

The pathophysiology of AF is heterogeneous, and, like heart failure, includes various kinds of heart disease.21) Like the diversity of AF pathophysiology, the factors related to failed CV or recurrence after successful CV might be widespread. An enlarged LA may reflect the degree of structural remodeling and LA pressure, and is one of the predictors for recurrence after RFCA of AF.22) However, LA diameter was not related to the failure of CV or recurrence of AF in this study or in other previous studies.10)23) With this in mind, one may ask why the meaning of LA diameter is different after electrical CV and RFCA. First, anti-arrhythmic drugs were maintained after CV, but stopped within the 3rd month after RFCA in most institutes.22) Second, CV does not change atrial critical mass, but RFCA reduces it, which contains multiple reentries.21) Third, the risk of recurrence after AF ablation is high in patients with an enlarged LA because of the potential conduction gap on the long distance of linear ablation or non-pulmonary vein foci.25) Fourth, patient selection bias also contributes. In contrast, the mean LA diameter was 45.5 mm in this study, whereas LA dimensions >65 mm have been reported to be associated with AF recurrence after CV.24) We previously reported that poor mechanical reserve of the LA appendage predicts AF recurrence after CV,11) and prolonged atrial stunning after electrical CV is also related to poor clinical outcomes.26) Therefore, mechanical function of the atria may be more important in clinical outcomes after CV than morphological changes in the atria.

Although approximately 50% of patients who are successfully cardioverted initially experience AF recurrence within the first month after CV,5)6) anti-arrhythmic drugs24) and upstream medical therapy are effective in preventing this. Nakashima et al.27) first suggested the potential role of angiotensin II inhibitors for preventing atrial electrical and structural remodeling, and Madrid et al.28) proved its effect in a prospective randomized clinical study in patients with AF after electrical CV. ACE inhibitors29) and statins30) also play roles in the inhibition of AF recurrence after CV. These upstream therapies may reduce the recurrence rate of AF after electrical CV by preventing atrial structural remodeling.

The patients included in this study were a highly selected group chosen for their rhythm control; and the number of patients was limited. The exclusion of patients with large atria (bigger than 55 mm) may influence the outcomes related to LA size. The peripheral blood samples may only partially reflect the remote process in the atria. Although we used anti-arrhythmic drugs in all patients before and after CV, the proportion taking amiodarone was 60.5%. In this study, mean age was higher in patients with recurrence after CV, and old age with low regenerating power may partially affect the low plasma level of SDF-1α. However, the low plasma level of SDF-1α was still a statistically significant predictor for AF recurrence even after adjustment for age by Cox regression analysis {hazard ratio (HR) 0.443, CI: 0.245-0.803, p=0.007}. Although SDF-1α has been known to be related to stem cell homing and inflammatory processes, the mechanism underlying it effect on the recurrence of AF is beyond our purpose and remains to be studied.