of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  Biatrial maze procedure versus pulmonary vein isolationfor atrial fibrillation during mitral valve surgery: Newanalytical approaches and end points Eugene H. Blackstone, MD, a,b Helena L. Chang, MS, c Jeevanantham Rajeswaran, PhD, b Michael K. Parides, PhD, c Hemant Ishwaran, PhD, d Liang Li, PhD, e John Ehrlinger, PhD, b AnnetineC.Gelijns,PhD, c AlanJ.Moskowitz,MD, c MichaelArgenziano,MD, f  JosephJ.DeRose,Jr,MD, g Jean-Phillipe Couderc, PhD, MBA, h Dan Balda, MD, i Franc¸ois Dagenais, MD,  j Michael J. Mack, MD, k  Gorav Ailawadi, MD, l Peter K. Smith, MD, m Michael A. Acker, MD, n Patrick T. O’Gara, MD, o andA. Marc Gillinov, MD, a for the Cardiothoracic Surgical Trials Network Investigators ABSTRACTObjective:  To use novel statistical methods for analyzing the effect of lesion seton (long-standing) persistent atrial fibrillation(AF)inthe Cardiothoracic SurgicalTrials Network trial of surgical ablation during mitral valve surgery (MVS). Methods:  Two hundred sixty such patients were randomized to MVS  þ  surgicalablation or MVS alone. Ablation was randomized between pulmonary veinisolation and biatrial maze. During 12 months postsurgery, 228 patients (88 % )submitted 7949 transtelephonic monitoring (TTM) recordings, analyzed for AF,atrial flutter (AFL), or atrial tachycardia (AT). As previously reported, moreablation than MVS-alone patients were free of AF or AF/AFL at 6 and 12 months(63 %  vs 29 % ;  P < .001) by 72-hour Holter monitoring, without evidentdifference between lesion sets (for which the trial was underpowered). Results: EstimatedfreedomfromAF/AFL/ATonanytransmissiontrendedhigherafter biatrial maze than pulmonary vein isolation (odds ratio, 2.31; 95 % confi-dence interval, 0.95-5.65;  P  ¼  .07) 3 to 12 months postsurgery; estimated AF/ AFL/AT load (ie, proportion of TTM strips recording AF/AFL/AT) was similar(odds ratio, 0.90; 95 % confidence interval, 0.57-1.43;  P  ¼  .6). Within 12 months,estimated prevalence of AF/AFL/AT by TTM was 58 % after MVS alone, and36 % versus 23 % after pulmonary vein isolation versus biatrial maze ( P < .02). Conclusions:  Statistical modeling using TTM recordings after MVS in patientswith (long-standing) persistent AF suggests that a biatrial maze is associatedwith lower AF/AFL/AT prevalence, but not a lower load, compared with pulmo-nary vein isolation. The discrepancy between AF/AFL/AT prevalence assessed at 10090807060504030201000123456 Months after Surgery    P  r  e  v  a   l  e  n  c  e  o   f   A   t  r   i  a   l   A  r  r   h  y   t   h  m   i  a   (   %   ) 789101112Mitral valve surgery aloneMitral valve surgery + biatrial mazeMitral valve surgery + pulmonaryvein isolation Estimated prevalence of atrial arrhythmia after mitral valve surgerywith or without atrial ablation. Central Message A biatrial maze may be associated with lower prevalence of atrial arrhythmias than pulmonary vein isolation for atrialfibrillation. These findings and analytic approach requireconfirmation. Perspective Thirty percent to 50 % of patients undergoing mitral valvesurgeryhaveatrialfibrillation.Ablationincreaseslikelihoodof return to normalsinus rhythm; however, lesionsetchoiceis controversial. A novel analysis of weekly transtelephonicrecordings suggests biatrial maze may be associated withfewer atrial arrhythmias than pulmonary vein isolation.This inference and analytic approach require confirmation. SeeEditorial Commentariespages 244and 246.See Editorial page 231. From the  a Department of Thoracic and Cardiovascular Surgery, Cleveland Clinic,Cleveland, Ohio;  b Department of Quantitative Health Sciences, Cleveland Clinic,Cleveland,Ohio; c Department of PopulationHealth Science and Policy, IcahnSchoolof Medicine at Mount Sinai, New York, NY;  d Department of Public Health Sciences,Division of Biostatistics, University of Miami, Miami, Fla;  e Department of Biostatis-tics,TheUniversityofTexasMDAndersonCancerCenter,Houston,Tex; f  Divisionof Cardiothoracic Surgery, Department of Surgery, College of Physicians and Surgeons,Columbia University, New York, NY;  g Department of Cardiovascular and ThoracicSurgery, Montefiore-Einstein Heart Center, Bronx, NY;  h Heart Research Follow-UpProgram, Cardiology Department, University of Rochester Medical Center, Roches-ter, NY;  i Medicomp, Inc, Melbourne, Fla;  j Department of Cardiac Surgery, InstitutUniversitaire de Cardiologie de Qu  ebec, H ^ opital Laval, Qu  ebec City, Qu  ebec, Can-ada;  k  Department of Cardiothoracic Surgery, Baylor Research Institute, BaylorHealth Care System, Plano, Tex;  l Division of Thoracic and Cardiovascular Surgery,University of Virginia School of Medicine, Charlottesville, Va;  m Division of Cardio-vascular and Thoracic Surgery, Department of Surgery, Duke University MedicalCenter, Durham, NC;  n Department of Surgery, Division of Cardiovascular Surgery,University of Pennsylvania School of Medicine, Philadelphia, Pa; and  o Cardiovascu-lar Medicine Division, Brigham and Women’s Hospital, Boston, Mass. Supported by grant No. R01 HL103552 and by a cooperative agreement (No. U01HL088942) with the National Heart, Lung, and Blood Institute, including fundingby the National Institute of Neurological Disorders and Stroke and the CanadianInstitutes of Health Research.Dr Parides is currently affiliated with the Department of Cardiovascular and ThoracicSurgery, Montefiore-Einstein Heart Center, Bronx, NY.A complete list of investigators in the Cardiothoracic Surgical Trials Network is pro-vided in Appendix E1.  Read at the 97th Annual Meeting of The American Association for Thoracic Sur- gery, Boston, Massachusetts, April 29-May 3, 2017. Received for publication May 1, 2017; revisions received June 7, 2018; accepted forpublication June 27, 2018. Address for reprints: Eugene H. Blackstone, MD, Department of Thoracic and Car-diovascular Surgery, Cleveland Clinic, 9500 Euclid Ave, Desk JJ-40, Cleveland,OH 44195 (E-mail: 0022-5223/$36.00 Copyright  2018 by The American Association for Thoracic Surgery 234 The Journal of Thoracic and Cardiovascular Surgery  c January 2019 AD UL T  ADULT: ARRHYTHMIAS  The Cardiothoracic Surgical Trials Network (CTSN),sponsored by the National Institutes of Health andCanadian Institutes of Health ( identifier:NCT00903370), randomized 260 patients 1:1 withpersistent or long-standing persistent atrial fibrillation(AF) associated with mitral valve disease to either mitralvalve surgery (MVS) with left atrial appendage closure orMVS with an ablation procedure for AF. 1 At both 6 and12 months, 63 % of patients in the ablation arm were freeof AF as assessed by 72-hour continuous Holter monitoringversus29 % of patients in the nonablation arm. The ablationarm was further randomized into 2 different ablative lesionsets: pulmonary-vein isolation alone (PVI) or biatrial maze(Appendix E2 and Figure E1), which had similar apparent efficacy for the primary 1-year trial end point (61 %  vs66 %  free of AF, respectively). 2 However, the trial wasnot powered to detect a difference in AF prevalencebetween ablative lesion sets when assessed with Holtermonitoring.A secondary objective of the trial was to compare 2techniques for postablation heart rhythm monitoring toguide follow-up strategies for future studies of rhythmcontrol in AF patients: 72-hour Holter monitoring at6 and 12 months and weekly and symptom-related transte-lephonic monitoring (TTM) recordings. 1 The present studyfocuses on these TTM recordings and the application of 2statistical models to estimate occurrence of AF, atrial flutter(AFL), and atrial tachycardia (AT). Because TTM data arecomposed of periodically sampled, short binary rhythmsnapshots, such data present a statistical challenge. There-fore, The National Heart, Lung, and Blood Institute fundeda multicenter ancillary R01 (HI, JE, JR, MKP; Co-PIs EHBand LL) with the aim of developing alternative methods toanalyze these types of longitudinal data. We hypothesizedthat more frequent rhythm assessment by TTM would in-crease power to detect outcome differences between the 2ablative lesion sets. Objectives of this study were to usethese new statistical methods to identify the temporalpattern of AF/AFL/AT for each of the 3 trial arms, comparethesepatterns,and estimatewith increased precision the po-tential magnitude of differences in indices of AF/AFL/ATbetween PVI and biatrial lesion sets. PATIENTS AND METHODSTrial Design Between 2010 and 2013, 260 patients with (long-standing) persistentAF scheduled for MVS at 1 of 20 CTSN centers were randomized 1:1 toMVS and concomitant left atrial appendage closure alone (MVS alone)(n  ¼  127) or MVS and surgical ablation with left atrial appendage closure(MVS  þ  ablation) (n  ¼  133), stratified by site using a random permutedblockdesign,withblocksizeof4or8chosen atrandom. 1 Patientsrandom-ized to ablation were further randomized in this same fashion 1:1 to PVI(n  ¼  67) or biatrial maze (n  ¼  66).CTSN investigators designed the trial, which was approved by institu-tional review boards at each participating center, and all patients providedwritten informed consent. A data coordinating center, an independentevents adjudication committee, and a data and safety monitoring board ap-pointed by the NIH oversaw trial progress. Because devices were notapprovedforAFtreatment,thetrialwasconductedunderaninvestigationaldevice exemption. Patients Patientsweregenerallyelderly,witha highprevalenceofdiabetes;mostwere taking anticoagulants; and concomitant coronary artery bypass graft-ing was performed in 20 % , a tricuspid valve procedure in 38 % , and aorticvalve replacement in 13 % (Table E1). End Points The endpointforthis analysiswasAF /AFL/ATlasting  30 secondsonweekly postdischarge TTM recordings. 3 During the index hospitalization,patients were given a TTM device (Medicomp, Melbourne, Fla) capable of recording 90-second rhythm strips, and instructed to transmit telephoni-cally to the Medicomp central monitoring facility 1 strip weekly for12 months. To ensure the monitoring schedule was independent of con-founders, patients were instructed to transmit the strip on a specified dayand time. In addition, patients with symptoms suggestive of an arrhythmiawere instructed to transmit a rhythmstrip. TTM recordings were submittedby 228 patients (88 % ); 547 of 8496 recordings (6.4 % ) were not evaluablebecause of missing or poor quality transmissions, or indeterminate atrialrhythms (Figure E2). Medicomp, the data coordinating center, and site in-vestigators interacted with patients to encourage compliance. Medicompanalyzed the TTM recordings, blinded to randomization group, to assessatrial rhythm.Medicomp classified rhythms as 1 or more of AF, AFL, AT, sinusrhythm, junctional rhythm, heart block, atrial pacing, or atrioventricular Abbreviations and Acronyms AF  ¼  atrial fibrillationAFL  ¼  atrial flutterAT  ¼  atrial tachycardiaCTSN  ¼  Cardiothoracic Surgical Trials Network MVS  ¼  mitral valve surgeryNIH  ¼  National Institutes of HealthPVI  ¼  pulmonary vein isolationTTM  ¼  transtelephonic monitoringScanning this QR code willtake you to supplemental ta-bles for this article. To viewthe AATS Annual MeetingWebcast, see the URL next tothe webcast thumbnail.2timepointsbyHoltermonitoringversusweeklyTTMsug-gests the need for a confirmatory trial, reassessment of def-initions for failure after ablation, and validation of statistical methods for assessing atrial rhythms longitudi-nally. (J Thorac Cardiovasc Surg 2019;157:234-43) The Journal of Thoracic and Cardiovascular Surgery  c Volume 157, Number 1 235Blackstone et al Adult: Arrhythmias       A     D     U     L     T  sequential pacing. An electrocardiogram core laboratory overread 20 % of rhythm strips. Of the 7949 recordings evaluated (Figure E3 and Appendix E3),3668showedAF,235AFL,290AFandAFL,and21AT;3655showedsinusrhythm,63junctionalrhythm,and17heartblock.Bothscheduledandunscheduled TTMs were included for all analyses. Analysis End Point Definitions FreedomfromAF/AFL/ATisdefinedasnoAF/AFL/ATdetectedonanyTTM transmission after a 3-month blanking period. The 3-month blankingperiod may be called the therapy stabilization period 4,5 based on theassumption that temporary inflammatory processes following an ablationor surgical procedure may cause AF/AFL/AT independent of the ablationprocedure effectiveness.AF/AFL/AT load is defined as the proportion of each patient’s TTMrecordings documenting AF/AFL/AT after a 3-month blanking period. 6 AF/AFL/AT prevalence is defined as the proportion of patients in agroup that is in AF/AFL/AT at any given timepoint. 6-13 Data Analysis Two complementary analytical methods were used: a 2-part zero-inflated negative binomial model to estimate freedom from AF/AFL/ATand AF/AFL/AT load, and a temporal decomposition parametric nonlinearmixed-effects modelto estimateAF/AFL/AT prevalence.The zero-inflatednegative binomial model provided a broad picture of AF/AFL/AT occur-rence and load from 3 to 12 months after surgery, whereas the temporaldecomposition model added finer details to the overall temporal patternof prevalence of AF/AFL/AT from surgery to 12 months, incorporatingprocedure- and time-specific features. Analyses used SAS version 9.4(SAS Inc, Cary, NC) and R version 3.3.1 (R Foundation for StatisticalComputing, Vienna, Austria). Freedom from AF/AFL/ATand AF/AFL/AT load.  Aftera3-month blanking period, 6127 TTMs from 191 patients (95 MVS alone,50 MVS  þ  PVI, and 46 MVS  þ  biatrial maze) were available to assessAF/AFL/AT load. A zero-inflated negative binomial model was used toaccount for the large frequency of zeros (29 % ), meaning no documentedAF/AFL/AT on any TTM rhythm strip (freedom from AF/AFL/AT), andthe substantialheterogeneityofAF/AFL/ATamongtheremainingpatients.The 2-part modeling of the zero-inflated negative binomial regression pro-vided the effect of randomized assignment on freedom from AF/AFL/ATand the relative frequency of AF/AFL/AT occurrence over time amongthose who developed it. Logistic regression was used to model the proba-bility of being free of AF/AFL/AT; negative binomial regression was usedto model mean number ofTTM recordingsdocumenting AF/AFL/AT from3 through 12 months. In both regression models, randomized assignmentwas evaluated. The negative binomial model also included a covariatefor the (log) number of TTM strips each patient transmitted to accountfor variation in number of interpretable TTM transmissions submitted.The regression coefficient of this covariate—an offset—was set to 1.0. AF/AFL/AT prevalence.  A nonlinear logistic mixed-modeltemporal decomposition model was used to resolve a number of temporalphases(orperiods)intheoddsdomainandestimatetheshapingparametersfor each (Appendix E4). Each phase was modulated by a function of timeand a patient-specific random effect, along with a log-linear model of randomization group. 7 It was implemented using PROC NLMIXED(in SAS). Prevalence of AF/AFL/ATover time was estimated by averagingindividual patient-specific profiles generated by the resulting model.To illustrate the method, Figure E4 shows 2 temporal componentsmaking up the temporal decomposition model for AF/AFL/AT. The firstis a peaking phase representing early occurrence of AF/AFL/AT aftersurgery. The second appears later after surgery and reaches a plateau.Figure E5 shows modeled AF/AFL/AT profiles for each patient. Noticethe grouping of profiles near 100 %  (continuous AF/AFL/AT) and 0 % (no AF/AFL/AT), with many profiles in between. The ensemble averageof these individual profiles reveals a peaking period over the first fewmonths, after which prevalence stabilizes at a constant value (Figure E6). RESULTSFreedom From AF/AFL/AT After a 3-month blanking period, more patients in theablation group were free from AF/AFL/AT on any TTMrecording during months 3 through 12 after surgery thanin the MVS-alone group (43 %  vs 16 % , respectively;odds ratio [OR], 4.04; 95 %  confidence interval [CI],1.89-8.64;  P < .001) (Table 1 and Figure 1,  A ). Whencompared with MVS alone, patients after either PVI(34 % ) or a biatrial maze (52 % ) were more likely to befree from AF/AFL/AT (PVI vs MVS alone: OR, 2.64;95 % CI, 1.07-6.51;  P  ¼  .04; biatrial maze vs MVS alone:OR, 6.11; 95 %  CI, 2.56-14.61;  P < .001) (Table 1 andFigure 1,  B ). Freedom from AF/AFL/AT trended higheramong patients randomized to a biatrial maze than thoserandomized toPVI (OR, 2.31;95 % CI,0.95-5.65; P ¼ .07). AF/AFL/AT Load After the 3-month blanking period, the proportion of TTM recordings documenting AF/AFL/AT (ie, AF/AFL/ AT load) was lower in the ablation groups than in theMVS-alone group (average percentage of recordings withAF/AFL/AT among patients with detected AF/AFL/ATrecurrence 56 %  vs 77 % , respectively; rate ratio [RR],0.71; 95 %  CI, 0.53-0.95;  P  ¼  .02) (Table 1). The PVIand biatrial maze groups also had a lower proportion of TTM recordings exhibiting AF/AFL/AT than thoseundergoing MVS alone (PVI vs MVS alone: RR, 0.74;95 % CI, 0.52-1.03;  P  ¼  .08; biatrial maze vs MVS alone:RR, 0.66; 95 %  CI, 0.44-0.99;  P  ¼  .05). However, theproportion of TTM recordings with AF/AFL/AT wassimilar between the 2 lesion-set groups (OR, 0.90; 95 % CI, 0.57-1.43;  P  ¼  .6). Prevalence of AF/AFL/AT In both the MVS-alone and ablation procedure groups,estimatedprevalenceofAF/AFL/ATpeakedwithin2weeksof surgery, declined for several months, then stabilized(Figure 2,  A  and Figure E7,  A ). This pattern was evidentin the raw frequency data that did not account for repeatedmeasurements in a given patient. Estimated prevalence waslower in the ablation group at months 3 (34 % vs 67 % ), 6(31 % vs 59 % ), 9 (30 % vs 58 % ), and 12 (29 % vs 58 % )compared with the MVS-alone group (Table 2 andAppendix E4). This prevalence pattern was also seen inthe 2 lesion sets (Figure 2,  B  and Figure E7,  B ), with anestimated prevalence of AF/AFL/AT in the PVI group thatwas higher than that in the biatrial maze group at months3 (41 %  vs 28 % ), 6 (37 %  vs 24 % ), 9 (36 %  vs 23 % ),and 12 (36 % vs 23 % ) (Table 2 and Appendix E4), as was also apparent in the raw frequency data. 236 The Journal of Thoracic and Cardiovascular Surgery  c January 2019Adult: Arrhythmias Blackstone et al AD UL T   In the early peaking period of estimated AF/AFL/ATprevalence, surgical ablation by either lesion set wassuperior to no ablation, although PVI appeared lesseffective than a biatrial maze (Table 2). In the later plateauperiod, prevalence of AF/AFL/ATafter PVI was lower thanafter no ablation ( P  ¼  .01) and the biatrial maze wassuperior to PVI ( P  ¼  .02) (Table 2). DISCUSSIONPrincipal Findings Statistical modeling of weekly TTM recordings in MVSpatients with persistent or long-standing persistent AFsuggests that when compared to PVI, a biatrial mazeprocedure may be associated with greater freedom fromAF/AFL/AT and a lower estimated prevalence of AF/AFL/AT during the first 12 months after surgery.However, there was no statistically significant differencebetween the 2 lesion sets with respect to freedom fromAF/AFL/AT or AF/AFL/AT load. AF in MVS Patients AF is present in 30 % to 50 % of patients who present forMVS. 14 Left untreated, the arrhythmia is associated withreduced long-term survival and increased risk of stroke. 15 Randomized clinical trials have shown that surgicalablation at the time of MVS reduces the likelihood of AFpersisting by 50 % or more. 16-30 In addition, observationalstudies suggest that restoration of sinus rhythm in thesepatients improves long-term survival and quality of lifewhile reducing the risk of late stroke. 31 For these reasons,multiple guidelines and expert consensus reportsrecommend the addition of  surgical ablation to MVS inpatients with coexisting AF. 32-34 The use of surgical ablation in such MVS patients varieswidely. 35 Many of these patients receive neither ablationnor management of their left atrial appendage duringMVS surgery. Furthermore, when ablation is performed,choice of lesion set is variable. Results presented hereinmay inform that choice within the context of this statisticalmodeling of weekly TTM data. The CTSN Trial The CTSN trial of surgical ablation for AF during MVSrandomized 260 patients with persistent or long-standingpersistent AF requiring MVS to undergo either surgicalablation or no ablation. 2 The largest randomized clinicaltrial in MVS patients with AF, the trial employed the2012 Heart Rhythm Society definitions of persistent andlong-standing persistent AF in place at the time of trialdesign. 1,36 The majority of patients had long-standingpersistent AF and would be classified as such according tothe revised 2017 multisociety definitions. 3 The Food and Drug Administration–specified primaryend point of the trial was freedom from AF or AF/AFL atboth 6 and 12 months as assessed by 72-hour Holtermonitoring. Patients who died before the 12-monthassessment, were too ill to undergo heart-rhythmassessment, or underwent subsequent ablation therapy forAFwereconsideredtreatmentfailures.Agreaterproportionof patients in the ablation group than in the MVS-alonegroup were free of AF or AF/AFL at both 6 and 12 months(63 %  vs 29 % ;  P  < .001), confirming the early-termeffectiveness of surgical ablation for (long-standing)persistent AF. 2 Nested within this trial was a comparison of 2 lesion sets:PVI and biatrial maze. 1 The trial was not powered to detecta difference between these lesion sets using the primary,Holter-monitor–based end point. In fact, analysis of Holterresults revealed no apparent difference between the lesionsets. Analysis of weekly TTM recordings enabled anadditional comparison of the 2 lesion sets and potentialidentification of differences in heart rhythm outcomes thatmight emerge if more events were identified withalternative rhythm monitoring. 37,38 Monitoring After Surgical Ablation Participants in the CTSN trial were given a TTM deviceand instructed to transmit weekly rhythm recordings andadditional information when they experienced symptomsthat might be attributable to a heart rhythm abnormality. TABLE 1. Effect of surgical ablation on prevalence of atrialfibrillation, flutter, or atrial tachycardia: Zero-inflated and negativebinomial modelsModel Result  P  value Analysis of surgical ablation vs no ablationZero-inflated*Surgical ablation vs MVSalone4.04 (1.89-8.64)  < .001Negative binomial y Surgical ablation vs MVSalone0.71 (0.53-0.95) .02Analysis of type of surgical ablation and no ablationZero-inflated*PVI vs MVS alone 2.64 (1.07-6.51) .04Biatrial vs MVS alone 6.11 (2.56-14.61)  < .001Biatrial vs PVI 2.31 (0.95-5.65) .07Negative binomial y PVI vs MVS alone 0.74 (0.52-1.03) .08Biatrial vs MVS alone 0.66 (0.44-0.99) .05Biatrial vs PVI 0.90 (0.57-1.43) .7 Values for the zero-inflated model are presented as odds ratio (95 %  confidenceinterval), whereas values for the negative binomial model are presented as rate ratio(95 % confidence interval).  MVS  , Mitral valve surgery with concomitant left atrialappendage closure alone;  PVI  , pulmonary vein isolation. *The zero-inflated modelis a logistic regression that models the probability of freedom from atrialfibrillation/atrialflutter/atrialtachycardia. y Thenegativebinomialmodelisanegativebinomial regression that models the mean count of transtelephonic monitoringrecordings that show atrial fibrillation/atrial flutter/atrial tachycardia. The modelincludes an offset for the (log) number of transtelephonic monitoring recordingssubmitted by each patient. The Journal of Thoracic and Cardiovascular Surgery  c Volume 157, Number 1 237Blackstone et al Adult: Arrhythmias       A     D     U     L     T
Similar documents
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!