Atrioventricular Septal Defects: Effect of BridgingLeaflet Division on Early Valve FunctionRandall S. Fortuna, MD, David A. Ashburn, MD, Nilto Carias De Oliveira, MD,Harold M. Burkhart, MD, Igor E. Konstantinov, MD, John G. Coles, MD,Jeffery F. Smallhorn, MD, William G. Williams, MD, and Glen S. Van Arsdell, MDDepartment of Surgery, Division of Cardiovascular Surgery, and Department of Pediatrics, Division of Cardiology, Hospital forSick Children, University of Toronto Faculty of Medicine, Toronto, Ontario, Canada

Background. Bridging leaflet division may facilitaterepair of atrioventricular septal defects (AVSD). How-ever, the consequences of bridging leaflet division onearly valve function and mortality are not well defined.

Methods. Records of children undergoing AVSD repairbetween January 1995 and January 2002 were reviewed.Multivariable analysis defined risk factors for moderateor greater atrioventricular valve regurgitation (AVVR)and death/reoperation within 1 year of repair.

Results. A total of 209 children (median age 5 months,median weight 5 kg) had defects whose repair includedthe possibility of bridging leaflet division. Bridgingleaflets divided were both (n � 119, 58%), one (n � 30,15%), or none (n � 55, 27%). Freedom from AVVR(moderate or greater) is 84%, 80%, and 78% at 1, 6, and 12months. Risk factors include technical factors: number ofbridging leaflets divided, longer cross-clamp time, andright-sided annuloplasty. Other risk factors include pre-operative AVVR (moderate or greater), double-orifice or

parachute left AV valve, and younger age. Freedom fromdeath/reoperation for AVVR is 96%, 92%, and 90% at 1, 6,and 12 months. Risk factors are preoperative AVVR(moderate or greater) and parachute left AV valve. Find-ings at reoperation (n � 15, 7.2%) were cleft dehiscence ortear along cleft closure (n � 10), dehiscence of dividedleaflet from septation patch (n � 1), or other (n � 4).Operative mortality (n � 6, 2.9%) included failed reop-erations for AVVR (n � 4), dehiscence of divided leafletfrom septation patch (n � 1), and sepsis (n � 1).

Conclusions. Division of bridging leaflets is a riskfactor for AVVR (moderate or greater) during the firstyear after repair. Preservation of bridging leaflet integritymay improve valve competency, decrease the need forfuture reoperation, and eliminate some causes of opera-tive mortality.

(Ann Thorac Surg 2004;77:895–902)© 2004 by The Society of Thoracic Surgeons

In 1975 Trusler and associates [1] presented early re-sults from our institution for repair of atrioventricular

septal defects (AVSD) using a novel two-patch technique,sandwiching the undivided bridging leaflets betweenpatches, and complete left atrioventricular valve cleftclosure. Techniques of preserving bridging leaflet integ-rity can be time consuming and have not been univer-sally applied. More recently Najm and associates [2]reported a series, also from our institution, of 363 chil-dren repaired between July 1982 and February 1995. Twopatch repair was used in most children with divison ofone or more bridging leaflets in only 114 (31%). Divisionof bridging leaflets was not a risk factor for reoperation ormortality by multivariable analysis. They concluded thatleaflet division is generally preferred because it mayprovide better visualization of the ventricular septaldefect and subaortic region without increasing the risk ofreoperation or mortality. The conclusion that leaflet di-

vision is safe may be less applicable to children repairedat age less than 6 months when leaflets are more fragile.

Recent early reoperation and sudden death frombridging leaflet dehiscence in a few children promptedus to reevaluate our methods of repair and return to thetechniques originally described by Trusler and col-leagues [1]. Herein we analyze our current experiencesince adopting a more liberal philosophy of dividing thebridging leaflets followed by a shift back to the Truslersandwich repair.

Patients and Methods

Patient SelectionA computer search of the Congenital CardiovascularSurgery Database at the Hospital for Sick Children iden-tified 256 children with AVSDs repaired between January1995 and January 2002. In all, 209 children undergoingseptation of a common atrioventricular valve orifice andpatch closure of a ventricular septal defect are includedin this report. Forty-seven children were excluded be-cause they had distinct left and right sided atrioventric-ular valves, absent or trivial ventricular septal defectsclosed primarily, associated transposition of the great

Presented at the Thirty-ninth Annual Meeting of The Society of ThoracicSurgeons, San Diego, CA, Jan 31–Feb 2, 2003.

Address reprint requests to Dr Van Arsdell, Division of CardiovascularSurgery, Hospital for Sick Children, 555 University Ave, Toronto, ON,Canada M5G 1X8; e-mail: glen.vanarsdell@sickkids.ca.

© 2004 by The Society of Thoracic Surgeons 0003-4975/04/$30.00Published by Elsevier Inc doi:10.1016/S0003-4975(03)01066-X

CA

RD

IOV

AS

CU

LA

R

arteries, associated total anomalous pulmonary venousconnection, or unbalanced defects requiring single ven-tricle repair. All children included in this report havedefects whose repair included the possibility of bridgingleaflet division to facilitate patch closure of the ventricu-lar septal defect. The decision to divide or not dividebridging leaflets was made by the staff surgeon. Foursurgeons repaired 202 defects (97%) using techniqueswith and without bridging leaflet division. The fifthsurgeon did not divide bridging leaflets in any of theseven defects repaired.

Data CollectionData were collected through the first year after repair.Operative notes, echocardiography reports, perfusionrecords, and hospital cardiology charts were reviewed.AV valve morphology data were obtained from operativenotes and preoperative echocardiograms. AV valve func-tion data were obtained from echocardiograms per-formed before repair (n � 209, 100%), immediately afterrepair or before hospital discharge (n � 202, 97%), andwithin one year of repair (n � 128, 61%). Reoperationsand mortality during the first year after repair wereidentified by the Congenital Cardiovascular Surgery Da-tabase and confirmed by operative notes and hospitalcardiology charts. Additional follow-up was obtained byphone contact with referring physicians as needed. Fol-low-up was complete at 30 days for 206 (99%), 6 monthsfor 197 (94%), and one year for 183 children (88%).Approval for this study was obtained from the Hospitalfor Sick Children Research Ethics Board.

DefinitionsAtrioventricular valve regurgitation (AVVR) is defined asmoderate or worse regurgitation by echocardiography.Catastrophic AV valve failure is defined as hemodynamicinstability or collapse as a result of sudden developmentof AVVR. Operative mortality is defined as mortalitywithin 30 days of repair or during the same admission.All cause mortality is used for analysis of risk factors formortality. Reoperation for AVVR is defined as any reop-eration where repair of an AV valve was the primaryobjective of the procedure.

Patient CharacteristicsTwo hundred nine children (116 female, 56%) underwentcomplete repair at a median age of 5.0 months (range, 6days to 15 years) and a median weight of 5.0 kg (range, 1.9to 37 kg). 132 children (63%) were repaired at less thanage 6 months (our current practice). Down’s syndromewas identified in 167 children (80%). A small ventricle(unbalanced defect) was noted by echocardiography orthe operating surgeon at the time of repair in 30 children(14%, 25 small right, 5 small left) but was not hypoplasticenough to preclude a two-ventricle repair.

Associated cardiac malformations in addition to patentductus arteriosus and secundum atrial septal defectswere identified in 49 children (23%). The associateddefects include tetralogy of Fallot (n � 12, 5.7%), aorticarch hypoplasia (n � 8, 3.8%, with left ventricular outflow

tract obstruction in 1), heterotaxy syndrome (n � 6, 2.9%),coarctation (n � 6, 2.9%, with heterotaxy syndrome in 1),pulmonary artery or valvar stenosis (n � 4, 1.9%), sub-aortic membrane (n � 3, 1.4%), double outlet rightventricle with or without right ventricular outflow tractobstruction (n � 3, 1.4%), multiple ventricular septaldefects (n � 2, 1%), and other (n � 6, 2.9%).

Previous palliative procedures were performed in 17children (8.1%). Thirteen children (6.2%) had pulmonaryartery banding including 6 with unbalanced ventricles, 6with aortic coarctation repairs (associated with unbal-anced ventricles in 2), 1 with complex left ventricularoutflow tract obstruction due to anomalous position of ananterolateral papillary muscle, and 2 for unclear reasons(referred from outside hospitals). Tetralogy of Fallot wasinitially treated with modified Blalock-Taussig Shunts in4 children (1.6%).

Atrioventricular Valve Morphology and FunctionRastelli’s classification of bridging leaflet morphology(182 children) is type A in 112 children (62%), B in 10children (5.5%), and C in 60 children (33%). Fourteenhave a parachute left AV valve. Nine have a doubleorifice left AV valve. Preoperative left AVVR (184 chil-dren) was absent in 31 (17%), trace to mild in 146 (79%),and moderate to severe in 7 (3.8%). Preoperative rightAVVR (178 children) was absent in 13 children (7.3%),trace to mild in 159 children (89%), and moderate tosevere in 6 children (3.4%).

Surgical TechniqueModerate hypothermia was used in 187 children (89%),including 4 with selective cerebral perfusion and briefperiods of intermittent circulatory arrest (8 to 22 minutes’duration) for repair of aortic arch hypoplasia or coarcta-tion. Profound hypothermia was used in 22 children(11%) combined with circulatory arrest for arch recon-struction (n � 1) or intracardiac repair (n � 21). Since1996 profound hypothermia and circulatory arrest forintracardiac repair was used in only one child due tosmall size (2.2 kg). Cold blood cardioplegia was used inall children. Aortic cross-clamp times averaged 96 � 29minutes (n � 209). Circulatory arrest times for intracar-diac repairs averaged 67 � 6.0 minutes (n � 21). Totalcardiopulmonary bypass times, including periods of cir-culatory arrest, averaged 139 � 43 minutes (n � 209).Details of the surgical technique are summarized inTable 1. Patch material used for closure of ventricularseptal defects (VSD) included Gore-Tex (W. L. Gore &Assoc, Flagstaff, AZ), Dacron (C. R. Bard, Haverhill, PA),or glutaraldehyde-treated autologous pericardium. Atrialseptal defects (ASD) were closed with autologous peri-cardial patches (either treated or untreated). Most VSDand ASD patches were secured using running suture.Interrupted horizontal mattress sutures were commonlyused to resuspend bridging leaflets or sandwich thembetween the two septation patches. Techniques for treat-ment of subaortic stenosis associated with AVSD havebeen previously described [3].

896 FORTUNA ET AL Ann Thorac SurgEARLY VALVE FUNCTION AFTER AVSD REPAIR 2004;77:895–902

CA

RD

IOV

AS

CU

LA

R

Statistical AnalysisAll analyses were performed using SAS statistical soft-ware, version 8 (SAS Institute, Cary, NC). Data arepresented as frequencies, means with standard devia-tions, and medians with ranges as appropriate.

Analyses were performed to determine the prevalenceof and risk factors for two primary outcomes: (1) death orreoperation and (2) development of moderate or worseAVVR. Nonrisk-adjusted estimates of freedom from eachoutcome were calculated using Kaplan-Meier methods.Modeling the hazard function, searching for variousphases of risk, and determining the characteristic equa-tion of each phase allowed calculation of parametricestimates of time-related freedom from each outcome [4].

Demographic, morphologic, and surgical factors asso-ciated with each outcome after AVSD repair were soughtby multivariable regression of the parametric hazardmodels with criterion of p less than 0.15 for variable entryand p less than 0.10 for retention. The analysis fordevelopment of AVVR was further explored to determinerisk factors separately for left and right AVVR. In the caseof missing data missing value indicator variables werecreated and the mean value of available informationimputed noninformatively. In the multivariable analysessignificant missing value indicators were carried to adjustfor the possibility that children with a given missingvalue differ with respect to outcome from those in whomthe value is not missing.

Reported variable estimates represent the contributionof a variable to the overall model. Variable increment andmathematical transformation affect interpretation of thecontribution and are specified where applicable. Themultivariable models were solved for single or combina-tion of risk factors to demonstrate the magnitude of effectof these factors on outcomes.

Results

Moderate or Worse AVVR After RepairAtrioventricular valve regurgitation (moderate or worse)was present in 45 children (22%; left valve, 33 children[16%]; right valve, 6 children [2.9%]; both valves, 6 chil-dren [2.9%]). Degree of AVVR was moderate in 39 valves,moderate to severe in 8 valves, and severe in 6 valves.Freedom from moderate or worse AVVR in either valveafter repair is 84%, 80%, and 78% at 1, 6, and 12 months,respectively (Fig 1).

RISK FACTORS FOR MODERATE OR WORSE AVVR. Incremental riskfactors identified by multivariable analysis for either leftor right AVVR during the first year after repair are inTable 2. Inherent patient-related incremental risk factorsinclude preoperative AVVR and double-orifice or para-chute left AV valve. Younger age at operation adds some

Table 1. Details of Surgical Technique

Variable N (%)

Number of patches1 29 (14%)2 180 (86%)

Bridging leaflet division (n � 204)None 55 (27%)Superior only 11 (5.4%)Inferior only 19 (9.3%)Both superior and inferior 119 (58%)

Left AV valve cleft closure (n � 205)None 14 (6.8%)a

Partial 21 (10%)Complete 170 (83%)

Annuloplasty/commissuroplastyLeft AV valve 67 (32%)Right AV valve 16 (7.7%)Left and right AV valve 10 (4.8%)

Primary valve replacementLeft AV valve 1 (0.5%)b

a Parachute left atrioventricular (AV) valve in 8 children. b Anomalousposition of anterolateral papillary muscle with thickened chordal com-plexes obstructed the left ventricular outflow tract.

Fig 1. (A) Nonrisk-adjusted freedom (Free) from moderate or worseatrioventricular valve regurgitation (AVVR), left or right, after atrio-ventricular septal defect (AVSD) repair in 209 children. The solidline represents parametric estimates surrounded by the 90% confi-dence interval (dashed lines) overlying nonparametric estimates(circles with error bars) obtained by Kaplan-Meier method. Thenumbers overlying the x-axis indicate the number of children re-maining at risk at various time points. (B) Hazard function for de-veloping moderate or worse AVVR after AVSD repair.

897Ann Thorac Surg FORTUNA ET AL2004;77:895–902 EARLY VALVE FUNCTION AFTER AVSD REPAIR

CA

RD

IOV

AS

CU

LA

R

additional risk but the effect is small. Technique-relatedincremental risk factors include the number of bridgingleaflets divided, longer duration of aortic cross-clamp,and need for right AV valve annuloplasty. When consid-ered separately, similar risk factors were found for leftAVVR; however, only right AV valve annuloplasty andleft AVVR were associated with right AVVR (Table 3).Left AV valve annuloplasy is not a risk factor. Weight,gender, Down’s syndrome, associated cardiac defects,Rastelli classification, unbalanced ventricles, one patchrepair, two patch repair, and cleft closure at the time ofrepair were not identified as independent risk factors.

EFFECT OF BRIDGING LEAFLET DIVISION ON FREEDOM FROM AVVR.

The predicted freedom from AVVR based on the numberof bridging leaflets divided is shown in Figure 2. There isan incremental decrease in freedom from AVVR witheach leaflet division (94% with none divided, 91% withone leaflet divided, 84% with both leaflets divided, at 12months). The additional influence of preoperative AVVRon this effect is seen in Figure 3. The presence ofpreoperative AVVR accentuates the incremental de-crease in freedom from regurgitation with division ofbridging leaflets. The effect of bridging leaflet division on

Table 2. Incremental Risk Factors for Developing Moderateor Worse AVVR After Repair of AVSD

VariableParameter

Estimate � SEa p Value

Patient-relatedModerate or worse preoperative

AVVR2.22 � 0.50 �0.001

Double-orifice or parachute leftAV valve

1.05 � 0.39 0.008

Younger age at operation (per1 month of age)

0.048 � 0.026 0.059

TechnicalIncreasing number of bridging

leaflets divided (per leaflet)0.55 � 0.20 0.007

Right AV valve annuloplasty 0.98 � 0.49 0.046Longer aortic cross clamp time

(per 10 min)0.12 � 0.04 0.006

a For a single early hazard phase.

AV � atrioventricular; AVSD � atrioventricular septal defect;AVVR � atrioventricular valve repair.

Table 3. Incremental Risk Factors for Developing Moderateor Worse Regurgitation for Left and Right AV Valves AfterRepair of AVSD

VariableParameter

Estimate � SEa p Value

Left AV valveModerate or worse preoperative

left AVVR1.76 � 0.63 0.006

Double-orifice or parachute leftAV valve

0.99 � 0.42 0.019

Younger age at operation (per 1month of age)b

0.39 � 0.17 0.026

Increasing number of bridgingleaflets divided (per leaflet)

0.53 � 0.21 0.012

Longer aortic cross clamp time(per 10 min)

0.09 � 0.04 0.035

Right AV valveRight AV valve annuloplasty 1.62 � 0.67 0.015Left AV valve failure 1.52 � 0.58 0.009

a For a single early hazard phase. b Entered after logarithmic trans-formation.

AV � atrioventricular; AVSD � atrioventricular septal defect;AVVR � atrioventricular valve repair.

Fig 2. Impact of bridging leaflet division on development of atrio-ventricular valve regurgitation (AVVR), moderate or worse AVVR,within the first year after atrioventricular septal defect repair. Thecurves represent risk-adjusted estimates obtained from solution ofthe parametric model based on division of none, one, or both bridg-ing leaflets (indicated by numbers rightward of graph) with otherpredictor variables set to their mean. (Free � freedom.)

Fig 3. Impact of moderate or worse preoperative atrioventricularvalve regurgitation (AVVR) and bridging leaflet division on devel-opment of AVVR (moderate or worse) within 1 year after atrioven-tricular septal defect repair. The curves represent risk-adjusted esti-mates obtained from solution of the parametric model based onpreoperative AVVR (solid lines � mild or less; dashed lines �moderate or worse) and division of bridging leaflets (none, one, orboth indicated by numbers rightward of the graph) with otherpredictor variables set to their mean. (Free � freedom.)

898 FORTUNA ET AL Ann Thorac SurgEARLY VALVE FUNCTION AFTER AVSD REPAIR 2004;77:895–902

CA

RD

IOV

AS

CU

LA

R

immediate postoperative AVVR is seen in Table 4. Themechanical consequnces of bridging leaftet divison andresuspension are compared with bridging leaflet preser-vation in Figure 4.

Reoperation and MortalityFreedom from death or reoperation after repair is 96%,92%, and 90% at 1, 6, and 12 months, respectively (Fig 5).By multivariable analysis, incremental risk factors fordeath or reoperation (for AVVR) during the first yearafter repair include parachute left AV valve and preop-erative AVVR (moderate or worse; Table 5). Age, weight,sex, Down’s syndrome, associated cardiac defects, Ras-telli classification, unbalanced ventricles, aortic cross-clamp time, one patch repair, two patch repair, cleftclosure, need for valvuloplasty at the time of repair, anddivision of bridging leaflets at the time of repair were notidentified as independent risk factors.

REOPERATIONS. Twenty children (9.6%) required reopera-tion during the first year after repair. Eleven children(5.3%) had reoperations during the same hospital admis-sion. The most common indication for reoperation was

Table 4. Change in Atrioventricular Valve (AVV)Regurgitation Immediately After Repair

Number of BridgingLeaflets Divided

PreoperativeModerate orWorse AVV

Regurgitation

ImmediatePostoperativea

Moderate orWorse AVV

Regurgitation

Two (n � 119) 5 (4.2%) 20 (17.9%)b

One (n � 30) 1 (3.3%) 2 (6.7%)None (n � 55) 4 (7.3%) 4 (7.3%)

a Prior to discharge. b Only 112 echocardiograms available for review.

Fig 4. Mechanical consequences of division and resuspension ofbridging leaflets compared with preservation of bridging leaflets. (A)Trusler sandwich technique. Sandwiching an undivided leaflet be-tween the atrial and ventricular septation patches preserves bridgingleaflets with the net effect of pledgets on either side of the repair. (B)Division and resuspension of bridging leaflets. Division of bridgingleaflets obligates a portion of leaflet tissue to the repair, whether us-ing one or two patches. Loss of leaflet tissue may lead to distortionof the remaining valve tissue and increased tension on the cleft clo-sure, leading to regurgitation and increased risk of cleft dehiscence.(LAVV � left atrioventricular valve; RAVV � right atrioventricularvalve; VSD � ventricular septal defect.)

Fig 5. (A) Nonrisk-adjusted freedom from death or reoperation afteratrioventricular septal defect (AVSD) repair in 209 children. Thesolid line represents parametric estimates surrounded by the 90%confidence interval (dashed lines) overlying nonparametric esti-mates (circles with error bars) obtained by Kaplan-Meier method.The numbers overlying the x-axis indicate the number of childrenremaining at risk at various time points. (B) There is a single, rap-idly declining early phase of hazard for death or reoperation. (Free� freedom.)

Table 5. Incremental Risk Factors for Death or ReoperationAfter Repair of AVSD

VariableParameter

Estimate � SEa p Value

Moderate or worsepreoperative AVVR

1.34 � 0.64 0.037

Parachute mitral valve 1.52 � 0.57 0.008

a For a single early hazard phase adjusted for missing information forpreoperative AVV.

AVSD � atrioventricular septal defect; AVVR � atrioventricular valverepair.

899Ann Thorac Surg FORTUNA ET AL2004;77:895–902 EARLY VALVE FUNCTION AFTER AVSD REPAIR

CA

RD

IOV

AS

CU

LA

R

left AVVR (n � 14, 6.7%) with associated residual VSD(n � 5), hemolysis (n � 2), left AV valve stenosis (n � 1),right AVVR (n � 1), pulmonary vein stenosis (n � 1), orresidual patent ductus arteriosus (n � 1). Other indica-tions for reoperation included: isolated residual VSD (n� 1), right AVVR with associated ASD patch dehiscence(n � 1), superior vena caval stenosis (n � 1), left dia-phragm paralysis with respiratory compromise (n � 1),complete heart block (n � 1), and sternal dehiscence (n �1). Left diaphragm plication and permanent pacemakerinsertion were also required in one child with reopera-tion for AVVR. Two children (0.96%) in this series devel-oped complete heart block requiring a permanentpacemaker.

REOPERATIONS FOR AVVR. Reoperation for AVVR (14 leftsided, 2 right sided) was required in 15 children (7.2%).The most common finding at reoperation was cleft de-hiscence or adjacent tear (n � 10, 4.8%) involving distalportion (n � 7), basal portion (n � 2), or entire length (n� 1) of cleft. Associated findings with cleft dehiscencewere leaflet tear along cleft (n � 3), residual VSD (n � 2),dilated left AV valve annulus (n � 1), double orifice leftAV valve (n � 1), right AVVR (n � 1), and bridging leaflettears at previous repair sites (n � 1). Left-sided bridgingleaflet dehiscence from the VSD patch with associatedresidual VSD at the crest of the ventricular septum wasfound in 1 child. Ruptured primary chordae to the leftsuperior leaflet after resection of accessory valve tissue atthe initial operation were found in 1 child. A prolapsedmural leaflet with residual VSD was found in 1 child. Aseverely dysplastic left AV valve associated with leftinferior pulmonary vein stenosis (left atrial endocardialfibrosis) and a residual VSD at the infundibular septumwas found in 1 child. Poor coaptation of right-sidedbridging leaftlets in association with atrial septal patchdehiscence was found in 1 child. In total residual ventric-ular septal defects were repaired in 5 children at time ofAV valve reoperation.

REOPERATION TECHNIQUES FOR ATRIOVENTRICULAR VALVE REGUR-

GITATION. Valve repair (n � 14) was attempted in allchildren except 1 with a severely dysplastic left AV valve.Cleft reclosure and leaflet repair techniques frequentlyutilized autologous pericardium for pledgets andpatches. Annuloplasty or commussuroplasty sutureswere used in the majority of redo repairs. Mechanicalvalve replacement after reoperation was performed in 4children at 3, 13, 306, and 311 days after reoperation. Atotal of 6 children (2.9%) in this series had mechanicalvalve replacement during the first year.

MORTALITY. Operative mortality was 2.9% (n � 6, within 30days or the same hospital admission) with an additional3 deaths during the first year (total 1-year mortality 4.3%,n � 9). Catastrophic left-sided AV valve failure accountsfor 5 of the 6 operative mortalities (83%). Both superiorand inferior bridging leaflets were divided in 4 of 5children with catastrophic left-sided AV valve failure.One child (4.4 months of age) presented to the emer-

gency room in cardiac arrest within 24 hours of dis-charge. At autopsy he had leaflet dehiscence of previ-ously divided bridging leaflets from the septation patch.Two children with division of both bridging leaflets wereresuscitated with extracorporeal membrane oxygenation(3 and 7 days after repair), underwent emergency reop-eration for ruptured chordae or cleft dehiscence, andboth died. Two additional children underwent unsuc-cessful reoperations for either cleft dehiscence or leaflettear along cleft closure (1 of the 2 had division of bothbridging leaflets). Two children died of infectious com-plications (Staphylococcus organism sepsis or fulminanthepatitis). Two children died outside of the hospital morethan 5 months after repair for unknown reasons. One ormore bridging leaflets were divided in 6 of 9 childrenwho died during the first year after repair.

Comment

This report is unique because it follows a contemporaryevolution in surgical technique from bridging leafletdivision to preservation of bridging leaflets at a singleinstitution. This gradual shift in operative strategy cre-ated groups of children who received relatively homog-enous treatment during the same period, with or withoutdivision of bridging leaflets.

Risk of Dividing Bridging LeafletsFor the first time we have been able to identify thenumber of bridging leaflets divided during repair ofAVSD as an incremental risk factor for postoperativeAVVR by multivariable analysis. Bogers and associates[5] found a negative effect of division of the inferiorbridging leaflet on reoperation. Others have reported nosignificant influence [6, 7]. Bridging leaflet division andresuspention may cause distortion of leaflet tissue [8].Loss of leaflet tissue necessary for reattachment of di-vided bridging leaflets may result in subtle, complexgeometric changes that increase tension on the cleftclosure and increase the risk of cleft dehiscence (the mostcommon site of dehiscence at reoperation in this series,and others) [9, 10]. Greater loss of valve tissue by divisionof both bridging leaflets may explain why division of twoleaflets is a stronger risk factor than division of only oneleaflet.

Competent AV valves before repair suggest the pres-ence of adequate leaflet tissue capable of sufficient coap-tation and should remain competent in the immediatepostoperative period after an adequate repair. The inci-dence of immediate postoperative AVVR compared withpreoperative AVVR is greater however after repair withdivision of both bridging leaflets (17.9% after repaircompared with 4.2% preoperatively). Preservation ofbridging leaflet integrity may be most likely to maintainvalve competency.

Atrioventricular Valve RegurgitationThe strongest risk factor for postoperative AVVR in thisseries is preoperative AVVR. When division of bothbridging leaflets is added to preoperative AVVR the

900 FORTUNA ET AL Ann Thorac SurgEARLY VALVE FUNCTION AFTER AVSD REPAIR 2004;77:895–902

CA

RD

IOV

AS

CU

LA

R

outcome is dismal with 25% freedom from AV valveregurgitation at 12 months. Preoperative AVVR suggestsan intrinsically pathologic valve where even minor dis-tortion, changes in leaflet tension, or loss of leaflet tissuemay worsen regurgitation. Children identified as being athigh risk for postoperative AVVR due to the presence ofpreoperative AVVR may benefit most from preservationof bridging leaflet integrity (increase from 25% to 63%freedom from AVVR at 12 months). Right AV valveannuloplasty as a risk factor for postoperative AVVR mayindicate leaflet tissue deficiency. Younger age at repairhas less effect on postoperative AVVR but its significanceas a risk factor suggests caution in managing leaflets.Rastelli type is not a risk factor for postoperative AVVR,suggesting a lack of correlation between Rastelli type andbridging leaflet division. The Rastelli classification sys-tem does not take into account either the location orextent of naturally occurring divisions of the bridgingleaflets that may preclude surgical division of a leaflet.

Catastrophic AVVRPreserving bridging leaflet integrity may also protectagainst catastrophic AVVR. Suture disruption at theatrioventricular septal patch junction is more likely tocause shunting than sudden regurgitation (as wouldoccur with division and resuspension of bridging leaf-lets). Two children in this series had leaflet dehiscencefrom a septation patch after bridging leaflet division anddied despite extracorporeal membrane oxygenation re-suscitation in 1. These 2 children may have been prevent-able deaths.

Residual VSD After Preservation of Bridging LeafletsDecreased visibility of the VSD may be a disadvantage ofbridging leaflet preservation in comparison with bridg-ing leaflet division [2]. Reoperation for residual VSDs inour previous series (69% bridging leaflet preservation)occurred in 2.5% of children (n � 9, 2 in association withleft AV valve repair). In the current series (27% bridgingleaflet preservation) reoperations for residual VSDs oc-curred in 2.9% of children (n � 6, 4 with small residualdefects closed at time of left AV valve repair). Theincidence of residual VSDs is nearly identical in bothseries. Only 1 child in the current series with bridgingleaflet preservation had a residual VSD. (The remaining5 children had division of both bridging leaflets.) Withmeticulous and somewhat time-consuming techniquesbridging leaflets may be preserved without an increasedincidence of residual ventricular septal defects.

Operative MortalityOperative mortality for repair of AVSDs has declinedover the past several decades to between 2.8% and 6% inrecent reports [2, 5, 7, 8, 11–13]. Several factors accountfor improved survival including better understanding ofanatomy, improved myocardial protection, improvedsurgical techniques, and better postoperative care basedon early repair [11].

Early repair has obviated pulmonary hypertensive cri-sis, but may have uncovered new technical problems.

Operative mortality in this series (2.9%) is the result oftechnical problems in all but 1 child who died of over-whelming sepsis. The elimination of technical problemsin this series would result in a theoretical operativemortality for repair of AVSDs of less than 1%. Leaflettears, cleft disruption, ruptured chordae, and leaflet de-hiscences suggest the need for further refinement insurgical technique. Bridging leaflets were divided in 5 of6 children who died (although the numbers are too smallto reach statistical significance). Technical problems aftercomplete repair of AVSDs are universal [2, 5, 7–17].Further reduction of operative mortality in the currentera will require solutions to technical problems.

Weaknesses and LimitationsSeveral weaknesses and limitations of this analysis areidentified. The status of repair and AV valve function in3 children who died after discharge is unknown. Deathfrom all causes is used in the analysis to exclude thiseffect. No patients required reoperation for isolated AVvalve stenosis. However we did not include the evalua-tion of AV valve stenosis in this analysis. Five surgeonsperformed operations with inherent differences in tech-nique. Echocardiograms were not independently re-viewed. Late echocardiogram follow-up is only 61%.

ConclusionsDivision of bridging leaflets is a risk factor for develop-ment of moderate or worse AVVR during the first yearafter repair of AVSDs. Preoperative AVVR magnifies therisk of dividing bridging leaflets. Dehiscence of dividedleaflets is a source of early reoperation and mortality.Surgical techniques emphasizing preservation of bridg-ing leaflet integrity may improve valve competency,decrease the need for future reoperation, and eliminatethe risk of divided leaflet dehiscence.

Doctor Ashburn gratefully acknowledges financial support ofthe Research Training Center, Hospital for Sick Children, andthe Department of Surgery, Wake Forest University School ofMedicine.

References

1. Mills NL, Ochsner JL, King TD. Correction of type c com-plete atrioventricular canal. J Thorac Cardiovasc Surg 1976;71:27–8.

2. Najm HK, Coles JG, Endo M, et al. Complete AVSDs. resultsof repair, risk factors, and freedom from reoperation. Circu-lation 1997;96(Suppl 2):311–5.

3. Van Arsdell GS, Williams WG, Boutin C, et al. Subaorticstenosis in the spectrum of AVSDs. J Thorac CardiovascSurg 1995;110:1534–42.

4. Blackstone EH, Naftel DC, Turner ME. The decomposition oftime-varying hazard into phases, each incorporating a sep-arate stream of concomitant information. J Am Stat Assoc1986;81:615–24.

5. Bogers AJJC, Akkersdijk GP, de Jong PL, et al. Results ofprimary two-patch repair of complete AVSD. Eur J Cardio-thorac Surg 2000;18:473–9.

6. Meisner H, Guenther T. AVSD. Pediatr Cardiol 1998;19:276–81.

7. Gunther T, Mazzitelli D, Haehnel CJ, Holper K, Sebening F,

901Ann Thorac Surg FORTUNA ET AL2004;77:895–902 EARLY VALVE FUNCTION AFTER AVSD REPAIR

CA

RD

IOV

AS

CU

LA

R

Meisner H. Long-term results after repair of completeAVSDs: analysis of risk factors. Ann Thorac Surg 1998;65:754–60.

8. Bando K, Turrentine MW, Sun K, et al. Surgical manage-ment of complete AVSDs. J Thorac Cardiovasc Surg 1995;110:1543–54.

9. Michielon G, Stellin G, Rizzoli G, Casarotto DC. Repair ofcomplete common atrioventricular canal defects in patientsyounger than four months of age. Circulation 1997;96(Suppl2):316–22.

10. Backer CL, Mavroudis C, Alboliras ET, Zales VR. Repair ofcomplete atrioventricular canal defects: results with the twopatch technique. Ann Thorac Surg 1995;60:530–7.

11. Crawford FA, Stroud MR. Surgical repair of complete AVSD.Ann Thorac Surg 2001;72:1621–9.

12. Hanley FL, Fenton KN, Jonas RA, et al. Surgical repair of

complete atrioventricular canal defects in infancy. J ThoracCardiovasc Surg 1993;106:387–97.

13. Tweddell JS, Litwin SB, Berger S, et al. Twenty-year experi-ence with repair of complete AVSDs. Ann Thorac Surg1996;62:419–24.

14. Capouya ER, Laks H, Drinkwater DC, et al. Management ofthe left atrioventricular valve in the repair of completeAVSDs. J Thorac Cardiovasc Surg 1992;104:196–203.

15. Merrill WH, Hammon JW Jr, Graham TP, Bender HW.Complete repair of AVSD. Ann Thorac Surg 1991;52:29–32.

16. Alexi-Meskishvili V, Ishino K, Dahnert I, et al. Correction ofcomplete AVSDs with the double-patch technique and cleftclosure. Ann Thorac Surg 1996;62:519–25.

17. Michielon G, Stellin G, Rizzoli G, et al. Left atrioventricularvalve incompetence after repair of common atrioventricularcanal defects. Ann Thorac Surg 1995;60:S604–9.

DISCUSSION

DR JOSEPH J. AMATO (Chicago, IL): I wish to congratulate youon an excellent presentation of an interesting subject. HoweverI was confused by the artist’s rendition in your drawings of theleaflets of the mitral and tricuspid valves? Perhaps I have to goback to my embryology books. But the construction of theleaflets as drawn by your illustrator is incorrect. I use theone-patch technique and I’ve always divided the commonposterior leaflet and common anterior leaflet, unequally. I leavemost of the tissue on the mitral side and not on the tricuspidside. The tricuspid septal leaflet that is reformed serves a smallamount of the surface area of the tricuspid valve. Thus in leavingmore tissue on the mitral side, I’ve not found regurgitation maynot be a major issue. Again I wish to congratulate you on thepresentation of a very complex type of surgery.

DR FORTUNA: There are series in the literature where bridgingleaflets are divided without reports of increased regurgitation,reoperation, or mortality. Certainly those results are enviable.Our series is somewhat unique because it shows a shift insurgical technique at a single institution. We did find that whenwe changed our technique to dividing bridging leaflets we hadmore problems and have since shifted our technique back.

DR PEDRO J. DEL NIDO (Boston, MA): This is a very interest-ing presentation on a very difficult problem. One of the ques-tions I had is your interpretation that the cause of the regurgi-tation when you divide a leaflet being purely on the basis of thefact that you take up more leaflet when you reattach it to thepatch. Is it not fair to say that probably the single biggest factorthat will affect coaptation of the leaflets is how much surface areaactually overlaps, which will be governed by the width of yourpatch and to some degree by the distance from the crest of theseptum to where the leaflets meet?

Another explanation for your findings, and I don’t disagreewith your findings, would be that in fact when you divide theleaflets it’s too easy to make the patch a bit too wide. And thatactually pulls the leaflets apart at the cleft because you’re pullingthe aortic end and the AV node end of the valve annulus apart.I notice that the regurgitation was through the cleft and not thatthe leaflets had dehisced. If you’re pulling the superior andinferior leaflets apart, in fact at the cleft, your sutures are goingto have to withstand the forces of systole with every heartbeat.It’s going to let go eventually. So that another interpretation ofyour data would be that one needs to remember how toconstruct the patch particularly with respect to its width. I don’tthink it matters so much whether you have one patch or two.

DR FORTUNA: Yes, I would agree, there are multiple factorsthat go into creating a competent valve, including those whichyou have discussed. Unfortunately those concerning the patchare something that it would be very difficult, if not impossible,for us to take into account in this type of a study.

DR WINFIELD J. WELLS (Los Angeles, CA): One of the argu-ments for dividing the leaflets would be to reduce the incidenceof residual VSD. So the question is, did you have a higherincidence of residual VSD in patients where the leaflets were notdivided?

DR FORTUNA: The incidence of residual VSDs in this series is2.9%. And as you know, in this series 73% of patients haddivided leaflets. If I compare that historically to our previousseries reported, where 31% of the leaflets were divided, theincidence of residual VSDs is 2.5%. So we don’t see much of adifference in the incidence of residual VSDs between these twoseries.

902 FORTUNA ET AL Ann Thorac SurgEARLY VALVE FUNCTION AFTER AVSD REPAIR 2004;77:895–902

CA

RD

IOV

AS

CU

LA

R