Author + information
- Received June 15, 2010
- Revision received September 10, 2010
- Accepted September 17, 2010
- Published online February 22, 2011.
- Francesco Onorati, MD, PhD⁎,⁎ (, )
- Giovanni Mariscalco, MD, PhD†,
- Antonino Salvatore Rubino, MD‡,
- Filiberto Serraino, MD‡,
- Francesco Santini, MD⁎,
- Andrea Musazzi, MD†,
- Catherine Klersy, MD§,
- Andrea Sala, MD† and
- Attilio Renzulli, MD, PhD‡
- ↵⁎Reprint requests and correspondence:
Dr. Francesco Onorati, Division of Cardiac Surgery, University of Verona Medical School, Piazzale Stefani, 1, 37126 Verona, Italy
Objectives The objective of this study was to evaluate the effects of different lesion sets of ablation in patients undergoing mitral surgery plus maze.
Background The role of lesion sets on outcome after maze is poorly defined.
Methods A total of 141 patients were prospectively followed up. Two different lesion sets were prepared: 32 patients underwent a radiofrequency left atrial lesion set of maze (“limited”), and 109 had combined left and right atrial lesion sets of maze ± ganglionic plexi isolation (“extensive”). A longitudinal observational study assessed the role of “extensive” versus “limited” ablation on atrial fibrillation (AF), New York Heart Association (NYHA) functional class II/III, treatment with antiarrhythmic drugs, follow-up recovery of the ratio of E- to A-wave (E/A), and survival and time to hospitalization (overall and for heart failure).
Results The prevalence of AF over time was lower in the “extensive” arm (adjusted relative risk [RR]: 0.10; 95% confidence interval [CI]: 0.03 to 0.31; p < 0.001), with significantly lower prevalence at discharge, 3 months, and 18 months. The prevalence of patients in NYHA functional class II/III over time was lower in the “extensive” arm (adjusted RR: 0.11; 95% CI: 0.03 to 0.34; p < 0.001), with significant differences at any assessment (except the third month). The differences in E/A recovery and use of antiarrhythmic drugs were less marked, with an RR of 1.55 (95% CI: 0.99 to 2.42; p = 0.05) and RR of 0.76 (95% CI: 0.54 to 1.06; p = 0.11), respectively, with a significantly lower prevalence of antiarrhythmic drugs in the “extensive” ablation arm at 12, 18, and 24 months. Rates of hospitalization for heart failure, overall hospitalization, and the combined event death/hospitalization were lower in the “extensive” arm (p = 0.11, p = 0.003, and p = 0.002, respectively).
Conclusions The addition of right-sided ablation improves clinical and electrophysiologic results after maze procedure.
Since Cox proved that atrial fibrillation (AF) could be treated by means of surgery (1), a growing number of simplified maze procedures have been proposed (1,2). However, different percentages of recurrent AF have been reported and attributed to learning curves, lesion sets, indications, and peri-operative protocols (1–3). In particular, the role of the lesion sets has been poorly investigated (1–3).
Indeed, different lesion sets with radiofrequency have been developed at our institutions since the original maze procedure was developed (4). Therefore, it was the aim of this study to evaluate the mid-term electrophysiological and clinical results of a consecutive series of patients undergoing concomitant surgical ablation for AF and mitral valve surgery with 2 different surgical approaches—an “extensive” and a “limited” approach—during the last 5 years at 2 universities.
From April 2003 to September 2008, we enrolled 141 consecutive patients who underwent mitral surgery and concomitant surgical ablation for longstanding, persistent AF (according to the Heart Rhythm Society/European Heart Rhythm Association/European Cardiac Arrhythmia Society [HRS/EHRA/ECAS]; permanent AF according to the American College of Cardiology/American Heart Association) (3). Institutional review board approval and informed consent were obtained. Demographics are reported in Table 1.
Five different lesion sets derived from the so-called maze IV operation (1) were applied during the study time period (Fig. 1). However, a “more aggressive” surgical approach also addressing the right atrium was started in 2004 because macro–re-entrant circuits sustaining AF occur virtually anywhere in either atrium (1).
Fourteen patients (subgroup A1) underwent the left atrial (LA) lesion set of the maze procedure (Fig. 1A) by means of monopolar radiofrequency (Cardioblate Irrigated RF Monopolar Pen and Cardioblate Surgical Ablation System Generator, Medtronic Inc., Minneapolis, Minnesota). Almost all of these patients were enrolled early in the study (April 2003 to January 2004), except for 3 undergoing such a lesion set because of the recurrent nature of surgery at 10, 12, and 15 months before closure of the follow-up. Eighteen patients (subgroup A2; December 2003 to September 2007) underwent the previously mentioned LA lesion set of maze but with the bipolar clamp (Cardioblate LP Irrigated Surgical Ablation Device, Medtronic) (Fig. 1B). Because radiofrequency ablation was limited to the LA, all of these 32 patients comprised the “limited” group (group A).
Right atrial ablation was added to the subgroup A2 lesion set for 38 patients (subgroup B1; August 2004 to June 2008) (Fig. 1C), and ganglionic plexi (GP) isolation and fat pad resection was added to the subgroup B1 lesion set for 32 patients (subgroup B2; June 2005 to May 2008) (Fig. 1D) (4). Thirty-nine patients (subgroup B3; September 2006 to May 2008) underwent the subgroup B2 lesion set, but with 4 repetitions of each line of ablation. Subgroups B1, B2, and B3 also involved the right atrium and comprised the “extensive” group (n = 109; group B).
The atrial appendages were always excluded by external ligation and internal stitching.
Peri-operative care and follow-up
Twelve-lead electrocardiography (ECG) was performed pre-operatively, on admission to the intensive therapy unit, and daily thereafter until hospital discharge or whenever necessary. All patients underwent continuous electrocardiographic monitoring for the first 72 h post-operatively.
An intravenous amiodarone bolus (150 mg), followed by continuous infusion at 1 mg/kg/h for 12 h and then 0.5 mg/kg/h until patients tolerated oral intake, was administered intraoperatively. Then oral amiodarone (200 mg twice/day for 1 week and then 200 mg daily) was administered until patient discharge.
Follow-up data were collected in the outpatient clinic (at 7 days, 14 days, 1 month, 3 months, 6 months, 1 year, and then once per year or whenever requested). Follow-up telephone calls with patients and cardiologists were made whenever necessary. A 24-h Holter ECG was performed at the 1-, 3-, 6-, 12-, 18-, 24-, 36-, 48-, and 60-month follow-ups. The New York Heart Association (NYHA) functional class was collected at the same time points.
After hospital discharge, general practitioners or referral cardiologists chose to continue amiodarone, switch to propafenone or beta-blocking agents, or stop antiarrhythmic medications after the sixth month. Accordingly, use of antiarrhythmic drugs defined the use of Class Ic/II/III drugs at each follow-up interval. An initial post-ablation blanking period of 3 months was considered: the reported recurrences defined “early events” (3).
Recovery of the E- to A-wave ratio (E/A) was investigated by trans-thoracic echocardiography in patients maintaining sinus rhythm at 6-month intervals, starting from the third post-operative month. The E/A ratio measures the ratio between the passive ventricular filling (early E-wave) and active filling (atrial A-wave) at pulsed Doppler analysis. E-wave is physiologically greater than A-wave (E/A >1), but loss of atrial kick (E/A not measurable) develops with AF. Ventricular function impacts A-wave but is negligible after the third post-operative month (5). Thus, the recovery of atrial transport function is measurable from the third post-operative month by means of a recovered E/A >1 (5). Episodes of congestive heart failure (CHF) were recorded throughout the follow-up.
Follow-up was closed on March 30, 2009, and was 100% completed.
The primary efficacy end point was freedom from AF/atrial flutter (AFL)/atrial tachycardia (AT) for at least 30 s without antiarrhythmic drug therapy, according to the HRS/EHRA/ECAS (3). Secondary end points were follow-up survival; freedom from AF/AFL/AT with antiarrhythmic therapy; freedom from only AF; prevalence of “early events”; and heart rhythm at aortic declamping, discharge, and 6, 12, 18, and 24 months post-operatively, as requested by the HRS/EHRA/ECAS (3). A longitudinal observational study aimed at assessing the role of “extensive” versus “limited” ablation on survival, time to first hospitalization (overall and for CHF), and prevalence of AF during follow-up was conducted. Given the variability of the functional classes, drug regimens, and recovery of atrial transport function observed during the follow-up, different longitudinal analyses aimed at assessing the role of ablation on the prevalence of NYHA functional class II or III (defined by episodes of CHF), treatment with antiarrhythmic drugs, and recovery at E/A were conducted.
Pre- and peri-operative data are shown as mean and SD, median and 25th to 75th percentile if continuous, and counts and percentage if categoric. They were compared between treatment groups (“extensive” vs. “limited” ablation), respectively, with the Student t test, Mann-Whitney U test, and Fisher exact test. Confidence intervals (CI) for the risk difference (RD) were normally approximated. Median follow-up time was calculated according to the inverse Kaplan-Meier method.
Kaplan-Meier cumulative survival and first hospitalization-free (for heart failure, any cause, and combined event death/hospitalization) survival were computed and compared by means of a stratified Cox model. Stratification by the propensity score allowed for separate baseline hazards for homogeneous groups of patients. Hazard ratios (HRs) and 95% CI are reported. The proportional hazard assumption was tested (and satisfied) with a test based on Schoenfeld residuals. Rates of events (per 100 person-years) are reported, with 95% CI. The propensity score was obtained from a logistic model for treatment, including all pre-operative characteristics: age, sex, duration of AF, left ventricular ejection fraction, LA diameter, year of surgery, diabetes, hypertension, dyslipidemia, smoking habits, chronic obstructive pulmonary disease, and redo surgery. Continuous variables were included as linear terms. The c statistic was computed to 0.96. Six blocks were identified from the computed propensity score so that the mean propensity score was not different for the 2 treatment arms in each block and the balancing properties were satisfied.
A mixed random-effects log linear model was fitted to compare the prevalence of the outcomes over time. For this purpose, assessments up to 24 months only were included to guarantee a sufficient number of patients at each time point and avoid overfitting. Risk ratios and 95% CI were computed. All models were adjusted for the propensity score blocks (on a continuous scale) and the presence of AF at declamping. At each time point, RD and conservative 99.2% CI were calculated. Finally, conditional logistic regression (conditioned on the propensity score blocks) was used to assess the association of a series of patient characteristics with the presence of AF/AFL/AT at 6 months. Crude odds ratios (cORs) and corresponding CIs were computed. Stata version 11 (Stata Corporation, College Station, Texas) was used for computation. Two-sided p < 0.05 was considered statistically significant. Post-hoc comparisons at each time point and for subgroups were adjusted for multiplicity with the Bonferroni correction. Subgroup analyses are to be considered as descriptive only.
Pre- and peri-operative assessments
Patient characteristics were similar in the 2 groups, except for the year of surgery (p < 0.001) and recurrence (p = 0.011) (Table 1).
No patient died peri-operatively in the “limited” compared with 4 in the “extensive” ablation group (RD: 3.7%; 95% CI: 0.1% to 7.2%; Fisher exact test p = 0.57). Twenty-five patients (18%) experienced in-hospital morbidity. Again, peri-operative morbidity was comparable between the 2 groups (“limited”: n = 7 [22%] vs. “extensive”: n = 18 [16%]; RD: −8.5%; 95% CI: −24.3% to 7.2%; p = 0.27). No AF surgery–related complications were recorded.
Post-operative cardiac rhythm and freedom from AF/AFL/AT
On aortic declamping, 81 patients (58%) recovered sinus rhythm (SR), 26 (18%) were in junctional rhythm, thus requiring temporary pacemaker stimulation, and 34 (24%) displayed AF. Notably, AF at declamping was more frequent, although not significantly, in the “limited” ablation group (Table 1) (RD: 9.2%; 95% CI: 8.6% to 27.1%; p = 0.35). Five patients demonstrating junctional rhythm at declamping (4% of survivors) required definitive pacemaker implantation. All patients were treated with amiodarone post-operatively, with 119 (87%) discharged home in SR and the remaining 13 (9%) in longstanding, persistent AF. Four patients (3%) developed post-operative transient type 1 AFL during hospitalization, but all recovered with intravenous amiodarone and were discharged in SR.
Long-term event-free survival
Five patients each died in the “limited” and “extensive” treatment arms during a median follow-up of 30 months (25th to 75th: 17 to 50 months). A lower mortality (although nonsignificant) was observed in the “extensive” versus “limited” ablation arm (2.1 [95% CI: 0.9 to 5.0] vs. 3.6 deaths per 100 person-years [95% CI: 1.5 to 8.6]; stratified HR: 0.23 [95% CI: 0.02 to 3.50; p = 0.29]). Seven and 8 patients, respectively, were hospitalized for heart failure, with rates of 3.1 (95% CI: 1.5 to 6.4) versus 6.8 hospitalizations per 100 person-years (95% CI: 3.4 to 13.7) and a stratified HR of 0.23 (95% CI: 0.04 to 1.42; p = 0.11); 13 and 14 patients, respectively, were hospitalized for any cause, with rates of 6.0 (95% CI: 3.5 to 10.3) and 14.0 hospitalizations per 100 person-years (95% CI: 8.3 to 23.7) and a stratified HR of 0.11 (95% CI: 0.02 to 0.53; p = 0.003). For the combined event death or hospitalization, the stratified HR was 0.12 (95% CI: 0.03 to 0.49; p = 0.002). Kaplan-Meier survival and event-free survival are shown in Figure 2.
The success of the “extensive” procedure on the combined event, in a bivariable stratified Cox model, was independent of the presence of AF at discharge (p = 0.87), and the effect of the surgical technique was not modified by the presence of AF at discharge (p for interaction = 0.26) (Fig. 3).
Overall, the prevalence of AF over time was lower in the “extensive” ablation arm (adjusted relative risk [RR]: 0.10; 95% CI: 0.03 to 0.31; p < 0.001) (Table 2). Post-hoc comparisons showed significantly lower prevalence (after Bonferroni correction) at discharge and at 3 and 18 months. Similarly, the prevalence of patients in NYHA functional class II/III over time was lower in the “extensive” ablation arm (adjusted RR: 0.11; 95% CI: 0.03 to 0.34; p < 0.001), with significant differences at any time point (except for the third month) (Table 2). However, manifest episodes of CHF were recorded only rarely in both the “extensive” and “limited” ablation arms, with no significant differences between them (5 [5%] vs. 4 [13%], RD: −8% [99.2% CI: −25% to 9%] at 12 months; 0 vs. 3 [10%)], RD: −10% [99.2% CI: −25% to 5%] at 18 months; and 3 (6%) vs. 1 (4%), RD: 2% [99.2% CI: −11% to 15%] at 24 months). The differences in E/A recovery and use of antiarrhythmic drugs were less marked, with an adjusted RR of 1.55 (95% CI: 0.99 to 2.42; p = 0.05) and 0.76 (95% CI: 0.54 to 1.06; p = 0.11), respectively. However, although all patients were taking antiarrhythmic drugs at discharge and at 3 and 6 months, a significantly lower use (after Bonferroni correction) of antiarrhythmic drugs was registered in the “extensive” ablation arm at 12, 18, and 24 months post-operatively (Table 2). As shown in Figure 4 (for description only), at the 12-, 18-, and 24-month assessments, patients with AF tended to be taking antiarrhythmic drugs more frequently than patients without AF. A similar behavior was seen for both the limited and extensive approaches (Figs. 5A and 5B).
Rates of the combined event (death and hospitalization) per 100 person-years were 25 (95% CI: 13 to 47), 11 (95% CI: 5 to 23), 7 (95% CI: 3 to 14), 15 (95% CI: 8 to 29), and 4 (95% CI: 1 to 15) in the A1, A2, B1, B2, and B3 ablation subgroups, respectively (Fig. 3) (stratified Cox model p < 0.001). The introduction of GP ablation (B2 vs. B1) did not significantly modify the risk of the combined event (HR: 1.01; 97.5% CI: 0.32 to 3.48; p = 0.93), whereas the introduction of 4 linear ablations (B3 vs. B2) was able to reduce it (HR: 0.19; 97.5% CI: 0.03 to 1.13; p = 0.017).
The prevalence of AF at 6 months was 61% (95% CI: 31% to 92%), 47% (95% CI: 21% to 73%), 26% (95% CI: 11% to 42%), 23% (95% CI: 7% to 39%), and 13% (95% CI: 2% to 24%) in the A1, A2, B1, B2, and B3 ablation subgroups, respectively (conditional logistic model p = 0.001). Neither the introduction of GP ablation (B2 vs. B1) nor 4 linear ablations (B3 vs. B2) significantly modified this prevalence (cOR: 0.56; 97.5% CI: 0.13 to 2.35; p = 0.36 and cOR: 0.48; 97.5% CI: 0.11 to 2.03; p = 0.25, respectively).
To explore the issue of a “learning curve” effect, the prevalence of AF at 6 months in the last 10 patients undergoing “limited” and the last 10 patients undergoing “extensive” approaches were computed; the prevalence was 50% (95% CI: 12% to 88%) and 30% (98% CI: 0% to 64%), respectively, favoring a protective effect of the “extensive” approach. Of note, all models comparing the 2 approaches did account (indirectly through the propensity score) for the year of surgery and thus for the used approaches, these being sequential in time.
In this study, we collected our experience with the surgical ablation procedures for AF developed over the last 5 years, during which different surgical lesion sets were used. Although a 7-day continuous ECG monitor would have possibly shown a lower success rate than Holter, in this study, we found a significant benefit—as shown by the lower prevalence of AF over time (adjusted RR: 0.18; 95% CI: 0.07 to 0.44; p < 0.001)—when the “extensive” approach was used.
Interestingly enough, a recent study by Albrecht et al. (6) reported that a “limited” approach with isolation of the pulmonary veins achieved comparable results to those of maze III. The study, however, appears underpowered to define the best treatment strategy for a dichotomic outcome variable, such as the recovery of SR; furthermore, no patients underwent Holter monitoring during follow-up either. Moreover, from an anatomic point of view, Chauvin et al. (7) observed in explanted hearts striated muscle around the coronary sinus connecting the inferior right atrium to the LA, which may account for the inconsistent results reported whenever the isolated LA lesion set is chosen. From an electrophysiologic point of view, Lammers et al. (8) first showed that the LA refractory period is shorter than the right. Given that AF can persist in an enlarged right atrium (9), it can be speculated that an isolated left-sided procedure may suffice to prevent recurrence in patients with normal atria, owing to the shorter refractory periods of the LA. On the other hand, additional right atrial ablation lines will be required in patients with enlarged atria (8), as is often encountered in the cardiac surgical population. Furthermore, Lin et al. (10) have recently demonstrated with a map-guided approach that paroxysmal AF may be induced in specific right atrial “trigger zones,” where ablation eliminates AF. Therefore, it is not surprising that previous surgical experiences have reported superior results with the adjunct of right atrial ablation procedures (3,11,12). More recently, the results from the International Registry of Atrial Fibrillation Surgery reported a biatrial approach to be predictive of SR restoration at 5 years' follow-up (2). We were not able to confirm an improved electrophysiologic outcome at 6 months with the “extended” approach associated with GP isolation or when introducing 4 linear ablations. A better clinical outcome (combined event of death and hospitalization) was shown with the addition of the 4 linear ablation approaches but not with the addition of GP isolation. These findings, originating from subgroup analyses, are to be considered with due caution.
It is recognized that unipolar ablation carries the risk for incomplete “nontransmural” lesions (1,3), mostly because none of the unipolar devices provide the surgeon with feedback relative to transmurality (3). In 2008, Benussi et al. (13) demonstrated the safety and efficacy of a complete bipolar radiofrequency ablation, with superior results compared with those of the unipolar devices. Accordingly, we report here a significant improvement of electrophysiological results during the follow-up by simple switching from unipolar to bipolar LA ablation. Despite bipolar clamping integrated by the measurement of conductance solving the shortcomings of the unipolar device (3), it has been shown that not all of the lesions produce conduction block (14). Studies have shown that gaps even smaller than 1 mm can allow the propagation of AF (15). Moreover, although isolation of the pulmonary veins can be tested intraoperatively, the remaining lesions cannot be verified, thus leaving the potential for incomplete transmurality (14). Finally, the so-called maze procedures with alternative energy sources all leave less isolated LA tissue compared with the Cox-maze procedures (14). The critical role of transmurality has been recently stressed by Albrecht et al. (6), showing significantly better results with surgical pulmonary vein isolation compared with transcatheter ablation. All of these data led us to empirically move toward the 4-time repetition of each line of ablation. Indeed, subgroup B3 showed the highest freedom from AF/AFL/AT, although the study was not sufficiently powered for sound conclusions from subgroup analyses.
The 2 groups also showed significant differences in terms of follow-up prevalence of NYHA functional class II/III; use of antiarrhythmic drugs; and freedom from overall hospitalization, hospitalization for CHF, and combined death/hospitalization. In particular, we found a significant reduction in follow-up NYHA functional class II/III at all time points (except for the third post-operative month, possibly because of a still ongoing recovery), together with a lower freedom from CHF in the “limited” approach. In addition, it is noteworthy that the use of antiarrhythmic drugs was less common in the “extensive” lesion set in the late follow-up period, thus suggesting a more “stable” restoration of SR. On the other hand, it has to be kept in mind that the decision to stop antiarrhythmic drugs in our series was left to the treating physician (cardiologists or general practitioners), therefore precluding the possibility to understand whether the antiarrhythmic treatment was stopped because patients were in SR or SR was present because patients were still on treatment.
Different studies have demonstrated that conversion to SR resulted in reverse remodeling of atrial diameters, volumes, and function (2,4). These data, together with the demonstration of an improved shuttle walk distance and a reduction in the levels of plasma brain natriuretic peptide (15), further confirmed those of Stulak et al. (16) and demonstrated a better functional status whenever SR was restored, regardless of the underlying valve disease. Therefore, because of the higher prevalence of SR and the correlated better functional status, it is not surprising that the “extensive” group also showed better freedom from overall hospitalization, better freedom from hospitalization for heart failure, and significantly better freedom from combined death/hospitalization.
Finally, the risk has been recently claimed for reduced atrial contractility after maze procedures (5,17). Yuda et al. (5) reported a reduced LA filling fraction 12 months after modified maze surgery. Incisions of the LA wall and excision of the LA appendage may contribute to the decreased LA contractility (5). Nonphysiological coupling of atrioventricular contractions caused by delayed LA activation may also contribute to these results (17). Moreover, after Cox-maze III surgery for paroxysmal AF, Lönnerholm et al. (18) demonstrated that LA and right atrial contractility were significantly reduced after 6 and 56 months versus baseline. The initial reduction in the atrial areas was proposed to depend on atrial scarring owing to extensive surgery, but the later deterioration of atrial dimensions and function was unexpected in patients with SR (18). Despite these reports contrasting with others showing functional atrial recovery (4,16,19), we found comparable results in terms of recovery of E/A at longitudinal analysis between the 2 groups, thus suggesting a comparable recovery of atrial kick between the “extensive” and “limited” groups, regardless of the extension of atrial scars.
The main limitation of the study is the fact that patients were accrued sequentially and the lesion set evolved over time, thus limiting the follow-up length of the last ablative strategy. The lack of randomization to treatment is another. The use of analyses stratified by the propensity score partially obviate this, however, because unmeasured confounders were not accounted for. Moreover, different associated cardiac diseases affected the study population. Therefore, randomized controlled trials are necessary to further clarify the best achievable lesion set. Also, the study was not powered for sound subgroup comparisons, and all findings derived from these are to be considered as descriptive only. Nonetheless, they might provide indications for future research protocols. Further, the low event rate (although a favorable outcome for the patient) hampers extensive multivariable modeling. In this series, however, the low event rate was in part overcome by performing most analyses while adjusting for AF at declamp and stratifying on the propensity score. The propensity score was constructed as a measure of the propensity to use either the “limited” or “extensive” approach based on all patients' pre- and peri-operative characteristics.
All authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- atrial flutter
- atrial tachycardia
- congestive heart failure
- ratio of E- to A-wave
- ganglionic plexi
- left atrial/atrium
- New York Heart Association
- sinus rhythm
- Received June 15, 2010.
- Revision received September 10, 2010.
- Accepted September 17, 2010.
- American College of Cardiology Foundation
- Calkins H.,
- Cappato R.,
- Crijns H.J.G.,
- et al.
- Yuda S.,
- Nakatani S.,
- Kosakai Y.,
- Yamagishi M.,
- Miyatake K.
- Chauvin M.,
- Shah D.,
- Haissaguerre M.,
- et al.
- Lammers W.,
- Schalij M.,
- Kirchhof C.,
- Allessie M.A.
- Lin Y.-J.,
- Tai C.-T.,
- Kao T.,
- et al.
- Deneke T.,
- Khargi K.,
- Grewe P.H.,
- et al.
- Benussi S.,
- Nascimbene S.,
- Galanti A.,
- et al.
- Melby S.J.,
- Lee A.M.,
- Schuessler R.B.,
- Damiano R.J.
- Stulak J.M.,
- Dearani J.A.,
- Daly R.C.,
- Zehr K.J.,
- Sundt T.M. III.,
- Schaff H.V.