Intravenous Plus Oral Amiodarone, Atrial Septal Pacing, or Both Strategies to Prevent Post-Cardiothoracic Surgery Atrial Fibrillation: The Atrial Fibrillation Suppression Trial II (AFIST II)
Background— The effect of a hybrid intravenous and oral prophylactic amiodarone regimen on postcardiothoracic surgery (CTS) atrial fibrillation (AF) is unknown. The impact of active atrial septal pacing on post-CTS AF has not been well characterized. In addition, the effect of using both amiodarone and atrial septal pacing together to prevent atrial fibrillation is unknown.
Methods and Results— Patients (n=160) were randomized to amiodarone or placebo and then to pacing or no pacing using a 2×2 factorial design. All therapies began within 6 hours post-CTS. Amiodarone was given by intravenous infusion for the first 24 hours (1050 mg total) followed by oral therapy for 4 postoperative days (4800 mg total). Atrial septal pacing was given for 96 hours. Amiodarone reduced the risk of AF by 43% and the risk of symptomatic AF by 68% (P=0.037 and P=0.019) versus placebo. Atrial septal pacing did not reduce AF or symptomatic AF incidence versus no pacing. The risk of post-CTS AF in the patients receiving amiodarone+pacing was lower than the placebo+no pacing and the placebo+pacing groups (57.9% and 60.5% reductions, P=0.047 and P=0.040, respectively).
Conclusions— Amiodarone given as both an intravenous and oral regimen is effective at reducing post-CTS AF but atrial septal pacing is ineffective. Combining amiodarone and pacing is better than placebo with or without pacing but not amiodarone alone.
Over 750,000 cardiothoracic surgeries (CTS; ie, bypass and heart valve surgery) are performed annually in the United States.1 Without prophylaxis, atrial fibrillation (AF) develops in up to 65% of post-CTS patients with two-thirds of cases arising on postoperative days 2 and 3.2 Clinical consequences of AF can include hemodynamic instability, ventricular arrhythmias, and stroke.2
Beta-blocker prophylaxis reduces the incidence of AF post-CTS.3,4 Despite beta-blocker use, the incidence of AF remains approximately 30% and could be due to inadequate dosing clinically and the withdrawal of beta-blockade before postoperative day 3.2,5–7 Amiodarone, sotalol, and biatrial pacing are acceptable alternative strategies.3
In the Atrial Fibrillation Suppression Trial (AFIST), we compared oral amiodarone to placebo.7 Amiodarone was given in either a 5- or 1-day preoperative loading regimen followed by therapy for 5 postoperative days (delivering 7g or 6g total, respectively). The risk of atrial fibrillation was reduced by 41% and the risk of ventricular arrhythmia or stroke was reduced by 76% in amiodarone treated patients. When each amiodarone dosing strategy was evaluated separately, there were qualitatively better effects in the 5-day loading group as compared with the 1-day group in atrial fibrillation parameters.8 Whether the differences reflect the higher dose or longer duration of therapy is not known.
Intravenous (IV) amiodarone has better bioavailability than oral (100% versus 50%) amiodarone, has a rapid onset of action, and does not produce nausea in higher doses.9 In a small study of 77 patients undergoing CTS, IV amiodarone reduced the incidence of post-CTS AF (5% versus 21%, respectively, P<0.05).10 Whether an IV and oral hybrid regimen begun after CTS and delivering the equivalent to 7 g of oral amiodarone in our AFIST trial would provide efficacy among patients already receiving beta-blockade as part of a clinical pathway is unknown.
Seven trials evaluated prophylactic temporary epicardial bi-atrial pacing to prevent post-CTS atrial fibrillation.3 In a meta-analysis of these trials, bi-atrial pacing reduced the risk of atrial fibrillation significantly (OR 0.46, 95% CI 0.30 to 0.71). However, therapy required the placement of two epicardial pacing wires rather than the one.3 In 1 trial, there was no benefit from right atrial or bi-atrial pacing in the total population but in the subset on beta-blockers, there was a trend toward lower post-CTS atrial fibrillation in post-hoc analysis.11 The reason for this effect was not known and if it relates to other antiadrenergic drugs such as amiodarone is not known. Whether pacing through a single epicardial wire placed in the atrial septum, the electrical connection between the atria, can prevent post-CTS atrial fibrillation and the impact of concurrent adjunctive amiodarone on efficacy is unknown. Preliminary investigations in patients with atrial fibrillation unrelated to CTS found 74% of patients free from chronic atrial fibrillation post-implantation versus only 47% in those receiving right atrial appendage pacing (P=0.01).12
The Atrial Fibrillation Suppression Trial II was designed to evaluate the prophylactic use of a hybrid intravenous and oral amiodarone regimen, atrial septal pacing, or both strategies in post-CTS patients.
This was a 2×2 factorial design study evaluating two active treatments, amiodarone and atrial septal pacing in post-CTS patients (n=160). Patients were randomized to amiodarone or placebo and were then randomized to active atrial septal pacing or no active atrial septal pacing. Using stratified allocation, CABG, and valve surgery patients were randomized separately to ensure equal distribution between groups. This study was designed as a management trial in that the recommended treatment regimens were established by study investigators and available to clinicians but the patient’s physician determined whether to adjust regimen intensity or to discontinue therapy without investigator consultation. The Institutional Review Boards of Hartford Hospital and the University of Connecticut approved the study with written informed consent being obtained prior to surgery.
Patients over 50 years of age scheduled to undergo coronary artery bypass grafting or heart valve surgery at Hartford Hospital were screened. Patients could be excluded for the following reasons: (1) chronic atrial fibrillation or flutter; (2) known amiodarone hypersensitivity; (3) current use of class I or class III antiarrhythmics; (4) current use of Implantable Cardioverter Defibrillators (ICD) or implantable pacemakers; (5) cardiogenic shock or advanced congestive heart failure (NYHA class IV); (6) marked sinus bradycardia (HR <50 bpm) or second or third degree AV block; (7) patients with moderate to severe liver disease; or (8) current use of cyclosporine, cimetidine, phenytoin, or cholestyramine.
Overall, 671 patients were screened and 511 patients were excluded from participating for the following reasons: age (n=91), amiodarone drug interaction (n=3), antiarrhythmic use (n=42), atrial fibrillation (n=58), heart block/bradycardia (n=11), pacmaker/ICD implantated (n=24), patient declined (n=102), physician refusal (n=32), already enrolled in a competing study (n=129), or shock/Class IV heart failure (n=3).
Experimental Regimens/Patient Management
Patients received a 1050 mg IV loading dose of amiodarone (Cordarone, Wyeth Ayerst, Philadelphia, PA) as a continuous infusion over 24 hours beginning within 6 hours postsurgery or matching placebo. This was followed by 400 mg of oral amiodarone (Pacerone, Upsher-Smith, Minneapolis, MN) 3 times a day on postoperative days 1 to 4 (4800 mg total oral drug) or matching placebo. This IV and oral regimen delivers the equivalent of 6900 mg of oral amiodarone.
All patients in the study had epicardial pacemaker wires placed in the atrial septum at Bachmann’s Bundle, the anatomical electrical connection between the left and right atrium. A temporary PM generator (TPM dual chamber, Medtronic, Minneapolis, MN) was programmed to AAI mode and set up according to the protocol in Figure 1. Atrial septal pacing was initiated within 6 hours of CTS and continued for 96 hours.
As part of the institution’s CTS critical pathway, the preprinted admission order sheet includes beta-blockers. Patients initially were cared for within the cardiothoracic intensive care unit before transfer to a monitored bed. The hospital telemetry equipment (Marquette, Milwaukee, WI) saves all abnormal rhythms in the previous 24-hours and a certified technician continuously monitored the alarm-triggered equipment printing out abnormal strips. A study investigator obtained a 12-lead ECG daily (MAC500, Marquette, Wisconsin, WI) on and off pacing, evaluated pacing threshold measurements, and reviewed the previous 24-hours of recording with the study physician for the occurrence of arrhythmias. Patient care was provided by the patient’s hospital clinicians including the management of atrial fibrillation, other arrhythmias, and hemodynamic issues.
The primary endpoint was the development of atrial fibrillation within 30 days of CTS. For the purposes of this study the following definitions were used: postoperative atrial fibrillation, any documented atrial fibrillation of more than 5 minutes duration; symptomatic atrial fibrillation, associated with hemodynamic compromise (hypotension, heart failure) requiring treatment or feelings of subjective discomfort (palpitations, chest pain, shortness of breath, syncope); recurrent atrial fibrillation, atrial fibrillation occurring more than 24 hours after resolution of a previous episode; and cerebrovascular accident, development of a transient ischemic attack (documented focal neurological deficit lasting less than 24 hours) or stroke (documented focal neurological deficit lasting more than 24 hours with cerebral infarction confirmation via computed tomography or magnetic resonance imaging). Patients received a phone call from a study investigator at 1 week and 1 month postdischarge to determine if an event occurred. If a post-discharge event was suggested, the hospital record or the outpatient medical record from the patient’s physician was garnered to confirm the event.
Post-CTS, side effects reported in the medical chart and side effects elucidated by study investigators on discussion were recorded.
Length of Stay/Cost Analysis
This trial was not designed to be powered to detect significant changes in length of stay or total costs. For each patient enrolled in the trial, the total costs of care were determined from the time of randomization through hospital discharge.
Power analysis was for a 2×2 fixed effects analysis of variance. The alpha-value was set at 0.05 and the analysis was set as 2 tails. With 40 cases per cell (4 cells total) and with an estimated effect size of (f) 0.25, the power is 0.88.
Data are presented as means±SD or proportions and a probability value <0.05 is considered significant for all comparisons. Chi-square analysis was used to compare categorical data. Student t-tests were used for two group comparisons of continuous parametric data while Mann-Whitney tests were used for nonparametric two group comparisons. One way ANOVA with Bonferroni corrected t-tests were used for multiple comparisons of parametric data.
Amiodarone versus Placebo
The AFIST II population was 65.8±8.7 years of age, 75.6% male, and 21.3% had valvular surgery. Baseline characteristics in the amiodarone and placebo groups (regardless of pacing randomization) were similar for all evaluated parameters (Table 1). Intraoperative parameters were similar except for a higher incidence of perioperative hypotension in the placebo group (Table 1).
The overall risk of atrial fibrillation was reduced by 42.7%, the mean ventricular response rate was 7.0% slower, and symptomatic atrial fibrillation was reduced by 68.3% in amiodarone treated patients versus placebo (P=0.037, P<0.0001, and P=0.019, respectively). This occurred even though there was a trend toward higher beta-blocker dosing in the placebo group (P=0.07) (Table 2). Therapy was well tolerated with no differences in IV or oral consumption of study drug (Table 2) and no significant differences in safety endpoints were noted between groups in the trial (Table 3).
Pacing versus No Pacing
Baseline characteristics in the pacing and no pacing groups (regardless of amiodarone randomization) were similar except for greater preoperative beta-blocker dosing in the pacing group (P=0.011) (Table 1). Surgical parameters were similar except the pacing group had a higher cross clamp time (P=0.049) (Table 1). Pacing was ineffective at preventing post-CTS atrial fibrillation with no differences in safety parameters as well (Tables 2 and 3⇑).
Overall, 53.8% of patients in the atrial septal pacing group discontinued active pacing for a portion of the postoperative period. Thirty-nine percent of patients stopped active pacing because their native heart rates exceeded 100 beats per minute as suggested in our pacing protocol (Figure 1). Deterioration of clinical status, development of atrial fibrillation, and technical difficulties such as loss of capture or problems with pacing wire placement accounted for 28.5%, 26.8%, and 16.7% of the discontinuations, respectively. A few patients had more than 1 reason to discontinue active pacing.
In a subgroup analysis, patients in the pacing group were divided in to those with premature discontinuation and those continuing pacing until postoperative day 4. There were no differences in atrial fibrillation between the two subgroups (34.7% versus 34.3%, P=0.838), respectively.
Amiodarone+No Pacing, Placebo+No Pacing, Both Therapies, and Placebo+Pacing
Baseline characteristics were similar except for the history of myocardial infarction which was higher in the placebo+pacing group versus the placebo+no pacing group (P=0.033). Surgical characteristics were similar between groups. Table 4 contains selected baseline and surgical characteristics for the 4 groups.
The risk of post-CTS atrial fibrillation in the amiodarone+pacing group was lower than the placebo+no pacing and the placebo+pacing groups (57.9% reduction, P=0.047 and 60.5% reduction, P=0.040, respectively). Amiodarone+pacing reduced the risk of symptomatic atrial fibrillation by 89.9% as compared with the placebo+pacing group (P=0.038) and exhibited a trend toward lower symptomatic atrial fibrillation versus the placebo+no pacing group (P=0.071) (Table 5). Safety endpoints were similar in the 4 groups for all parameters. A trend toward less symptomatic bradycardia was noted in the placebo+pacing group versus the amiodarone+pacing group (P=0.078) (Table 6).
No difference in any efficacy parameter was noted between the amiodarone+pacing and amiodarone+no pacing groups. However, the patients receiving amiodarone+pacing showed a trend toward receiving a higher dose of amiodarone than those receiving amiodarone without pacing (oral dosing equivalent [(2*IV dose)+oral dose] of 5871.3±1600.4 versus 5145.5±2205.2, P=0.10, respectively).
Length of Stay and Hospital Costs
The index hospitalization length of stay trended lower among patients receiving amiodarone than those receiving placebo (7.88±6.16 versus 11.36±16.83 days, P=0.08) but pacing was not different than no pacing (10.01±16.86 versus 9.41±8.43 days, P=0.77). Similarly, there was a trend toward lower costs in the amiodarone treated patients versus placebo treated patients ($20 737±13 878 versus $29 911±45 203, P=0.08) but no difference in the pacing versus no pacing group ($25 970±42 546 versus $25 098±25 333, P=0.87).
Among the four groups, no significant differences in length of stay (P=0.16) or total costs (P=0.27) were found. Qualitatively, amiodarone+pacing had the lowest length of stay and hospital costs (6.76±3.79 days, $18 697±8174) followed by: amiodarone + no pacing (8.97±7.71 days, $22 725±17 661), placebo+no pacing (9.77±9.04, $27 026±30 226), and finally placebo+pacing (13.54±23.7 days, $33 868±60 309).
We found that a hybrid intravenous and oral amiodarone regimen was effective at reducing post-CTS atrial fibrillation and symptomatic atrial fibrillation but atrial septal pacing was ineffective. Patients receiving both amiodarone and pacing had significantly lower atrial fibrillation rates than those receiving placebo with or without pacing. The groups with the lowest rate of atrial fibrillation had the lowest length of stay and total hospital costs. These experimental interventions were done in addition to the hospital’s standard of care therapy, beta-blockade. Beta-blockade was utilized post-CTS in over 80% of study patients.
We evaluated a hybrid regimen with intravenous amiodarone being given initially after surgery followed by oral therapy for post-CTS atrial fibrillation prevention. Studies have previously evaluated solely intravenous or oral amiodarone regimens. Using intravenous therapy for the first 24-hours post-CTS has several advantages over oral dosing regimens. Patients are transferred to the surgical intensive care unit post-CTS and initially receive drugs via nasogastric tube; the percent absorption of oral amiodarone from the gut of a post-CTS patient is unknown; and there is a limit to how much oral amiodarone you can load a patient with because of the risk of nausea. This is why previous oral amiodarone trials have utilized preoperative loading. Although preoperative loading reduces the amount of drug needing to be delivered immediately after CTS, it excludes patients who do not have the specified number of days before CTS for oral loading. Having multiple oral amiodarone regimens for patients with varying numbers of days from the day the surgery is scheduled until it takes place is cumbersome clinically.
If the a hybrid intravenous and oral regimen provides similar efficacy as an all-intravenous regimen, there would be cost advantages to the hybrid regimen. For example, supplying all of the amiodarone utilized in this trial intravenously would have an average wholesale cost of $2277.00 versus only $748.00 with our hybrid regimen. In addition, we utilized a bioequivalent and less expensive brand of amiodarone tablets (oral Pacerone®) rather than Cordarone®.
Previous trials of intravenous amiodarone and our original AFIST trial suggest that the dose of amiodarone delivered and not the duration of therapy is the important factor in preventing post-CTS atrial fibrillation. A 4-day intravenous amiodarone regimen was conducted with 2700 mg being given over the first 2 days postcoronary artery bypass surgery and 1800 mg being given over the next 2 days (equivalent to 9 g of oral drug).10 In that trial, patients experienced a 76% reduction in the risk of atrial fibrillation in the amiodarone group versus placebo. In another trial, patients undergoing either coronary artery bypass or valvular surgery received 2 g of intravenous amiodarone over 2 days (equivalent to 4 g orally). The atrial fibrillation rate among amiodarone treated patients was reduced by 25% versus placebo group.13 Our AFIST II regimen delivered the equivalent of 7 g of oral amiodarone with 43% reductions in atrial fibrillation falling between those of the 2 previous IV amiodarone studies. These results are also consistent with the 48% reduction in atrial fibrillation from AFIST where 7 g of oral amiodarone was given over 10-days beginning 5 days before CTS.8
We utilized atrial septal pacing to prevent post-CTS atrial fibrillation. Historically, epicardial pacing wires were placed on the right atrium during CTS and used only in cases of symptomatic bradycardia.3 Most of the clinical trial experience using active pacing to prevent atrial fibrillation is with biatrial pacing in which an epicardial atrial pacing wire is placed on each atria.3 However, placing 2 epicardial wires requires additional work on behalf of the surgeons. We had hoped that atrial septal pacing would provide efficacy similar to that seen with biatrial pacing in post-CTS patients and would provide efficacy as was seen previously in the chronic atrial fibrillation setting.12
Even though atrial septal pacing was ineffective overall, it was effective at preventing post-CTS atrial fibrillation in combination with amiodarone. This is similar to one study of right atrial and biatrial pacing where therapy was ineffective overall but efficacious in the subset of patients receiving beta-blockade.11 In our trial the patients receiving amiodarone with pacing received the oral equivalent of 726 mg in additional amiodarone versus those receiving amiodarone without pacing. This suggests that pacing may prevent clinicians from reducing the dose or discontinuing anti-adrenergic therapy.
With a larger sample size, the impact of amiodarone, pacing, and the combination of the two therapies on length of hospital stay, total hospital costs, and cerebrovascular accident risk could have been adequately explored. Greater investigator control over the use of standard of care beta-blockade dosing, post-CTS treatment of episodes of atrial fibrillation, and when and whether to discontinue or postpone investigational therapy would have increased the internal validity of the study but would have done so at the expense of external validity. We sought to explore the impact of these therapies as they would be employed in the “real world” and feel we achieved this aim. However, this meant that a large proportion of patients (53.8%) did not receive continuous atrial septal pacing which could have impacted the success of this treatment. In our subgroup analysis of the patients receiving pacing, premature discontinuation of pacing did not seem to impact results versus sustained pacing for four postoperative days.
A hybrid regimen with intravenous and oral amiodarone given to post-CTS patients reduces the risk of atrial fibrillation. Atrial septal pacing is ineffective as monotherapy but when combined with amiodarone substantially reduces the risk of post-CTS atrial fibrillation versus placebo based regimens with or without pacing. However, combination therapy with amiodarone and atrial septal pacing is not substantially better than amiodarone alone and may add unnecessary complexity to postoperative management.
This study was sponsored by the Hartford Hospital Research Foundation in Hartford, CT. Pacerone and matching oral placebo tablets were donated by Upsher-Smith Pharmaceuticals, Minneapolis, MN, while external pacemaker boxes were donated by Medtronic Corporation, Minneapolis, MN.
Am Heart Association. 2002 Heart and Stroke Statistical Update. Dallas, Texas: Am Heart Association, 2001 (accessed August 7, 2002).
Crystal E, Connolly SJ, Sliek K, Ginger TJ, Yusuf S. Interventions on prevention of postoperative atrial fibrillation in patients undergoing heart surgery: a meta-analysis. Circulation. 2002; 106: 75–80.
Laurer MS, Eagle KA, Buckly MJ, et al. Atrial Fibrillation following coronary artery bypass surgery. Prog Cardiol Dis. 1989; 31: 367–378.
United States Pharmacopoeia Convention. USPDI Drug Information for the Health Care Professional. Micromedex, Inc, Englewood, CO, 2002.