SSARaDD

ocn

t(setift

ttFrd

Rbcqsodcdeernrrsco

A

AHS

©P

CA

RD

IOV

ASC

ULA

R

ORIGINAL ARTICLES: CARDIOVASCULAR

pontaneous Closure of Small Residual Ventriculareptal Defects After Surgical Repair

li Dodge-Khatami, MD, PhD, Walter Knirsch, MD, Maren Tomaske, MD,ené Prêtre, MD, Dominique Bettex, MD, Valentin Rousson, PhD,nd Urs Bauersfeld, MDivisions of Congenital Cardiovascular Surgery, Pediatric Cardiology, and Anesthesiology, University Children’s Hospital;

epartment of Biostatistics, Institute for Social and Preventive Medicine, University of Zürich, Zürich, Switzerland

msqltg

mDndrmndia

Background. Residual shunts may be detected by intra-perative or postoperative echocardiography after surgicallosure of a ventricular septal defect (VSD). The hemody-amic relevance and rate of late closure are unknown.Methods. Between 1994 and 2005, 198 consecutive pa-

ients underwent surgical correction of an isolated VSDn � 100), tetralogy of Fallot (n � 52) or atrioventriculareptal defect (n � 46). Intraoperative transesophagealchocardiography (TEE) was routine, and postoperativeransthoracic echocardiography was performed in thentensive care unit, at hospital discharge, and duringollow-up. Residual defects were graded as absent, be-ween 1 and 2 mm, or greater than 2 mm.

Results. Shunt-related discrepancy was observed be-ween intraoperative TEE and intensive care unit trans-horacic echocardiographic findings; significantly so afterallot repair (p < 0.0001). After discharge, 83% of allesidual defects less than 2 mm closed. Of nine residual

efects greater than 2 mm, only three closed after a

Vp

M

TtrpciTsmcpsaasswcp

ospital, University of Zürich, Steinwiesstrasse 75, CH-8032 Zürich,witzerland; e-mail: ali.dodge-khatami@kispi.unizh.ch.

2007 by The Society of Thoracic Surgeonsublished by Elsevier Inc

edian follow-up of 3.1 years. In patients with residualhunts, they were hemodynamically insignificant, re-uired no medication, and no endocarditis was noted. At

ast follow-up, there was no significant difference be-ween the percentage of residual shunts among the threeroups (p � 0.135).Conclusions. Postsurgical residual VSDs less than 2m closed spontaneously in the majority within a year.efects greater than 2 mm are unlikely to close sponta-eously. Residual shunts after atrioventricular septalefect repair almost always close, whereas one third willemain open after Fallot or isolated VSD repair. Atidterm follow-up, residual shunts remained hemody-

amically and clinically irrelevant. Revision of a residualefect greater than 2 mm on cardiopulmonary bypass at

nitial repair, guided by TEE, may spare late redo surgerynd lifelong antibiotic prophylaxis.

(Ann Thorac Surg 2007;83:902–6)

© 2007 by The Society of Thoracic Surgeons

esults after surgical closure of an isolated ventricularseptal defect (VSD) are excellent, with regard to

oth in-hospital mortality, which reaches zero in manyenters [1], and to the long-term follow-up with regard touality of life and exercise capacity [2]. A ventriculareptal defect is also a consistent component of tetralogyf Fallot (TOF) and of complete atrioventricular septalefect (AVSD), and is routinely closed during surgicalorrection of these more complex defects. Small residualefects are frequently described on intraoperative trans-sophageal echocardiography (TEE) and also on postop-rative transthoracic echocardiography (TTE), but theate of possible spontaneous closure, or the hemody-amic and clinical significance in case of a persistentesidual shunt, are sparsely documented [3–5]. In aecent era of surgery we evaluated our results afterurgical closure of a VSD with regard to residual defects,ompared the findings of TEE with TTE in the diagnosisf residual defects, and looked at the rate of spontaneous

ccepted for publication Sept 25, 2006.

ddress correspondence to Dr Dodge-Khatami, University Children’s

SD closure in time with its eventual implications for theatient in the midterm follow-up.

aterial and Methods

he Institutional Review Board of our hospital approvedhe study. Retrospective chart and echocardiographyeport analysis were anonymously performed, wherebyatient consent was waived. Between 1994 and 2005, 198onsecutive patients underwent surgery for closure of ansolated perimembranous VSD (n � 100), a malalignmentOF VSD (n � 52), or an inlet AVSD (n � 46). Standardurgical technique using cardiopulmonary bypass withoderate hypothermia, cross-clamping, and cold blood

ardioplegia were used. Patch material included xeno-ericardium (Edwards Xenopericardium, Edwards Life-ciences Services GmbH, Unterschleissheim, Germany,nd Shelhigh Pericardial Patch, Shelhigh Inc, Union, NJ),nd a running suture technique with a nonabsorbableuture was routinely employed. Exposure for VSD clo-ure was routinely through the right atrium. For patientsith complete AVSD, patch closure of the ventricular

omponent of the defect was routine, except in two

atients with very small defects in which a one-patch

0003-4975/07/$32.00doi:10.1016/j.athoracsur.2006.09.086

(ftcstcrSssibadfSua1alcr

gpsa

R

MaaicdI242aerttfas

trwdpacf

T

A

F

I

T

T

A EE �

903Ann Thorac Surg DODGE-KHATAMI ET AL2007;83:902–6 RESIDUAL SHUNTS AFTER CLOSURE OF VSD

CA

RD

IOV

ASC

ULA

R

atrial septal defect � VSD closure) technique was per-ormed. In patients with TOF and isolated VSD repair,he choice of direct closure or xenopericardial patchlosure of the VSD was left to the discretion of theurgeon. Direct VSD closure was performed in five pa-ients during correction of TOF, and in 13 patients duringlosure of an isolated VSD. Intraoperatively, TEE wasoutine for all cases (Philips Sonos 5000, Philips Medicalystems, The Netherlands). In case of a large residualhunt (larger than 3 mm, or a ratio of pulmonary toystemic blood flow [Qp:Qs] � 1.5:1 as measured byntraoperative right heart saturations), cardiopulmonaryypass was resumed to achieve defect closure. Postoper-tively, in the intensive care unit (ICU), at hospitalischarge, and during follow-up, TTE was routinely per-

ormed (Philips Sonos 2000 or 5000, Philips Medicalystems). Echocardiographic measurements of resid-al VSD size were made with two-dimensional imagingnd residual defects were graded as absent, betweenand 2 mm, or larger than 2 mm. After repair of tetralogynd AVSD, patients were recommended to receive life-ong antibiotic endocarditis prophylaxis. After successfullosure of an isolated VSD, endocarditis prophylaxis wasecommended for six months and stopped thereafter.

Statistical analysis included a sign test to compare therade of residual defects at different follow-up timeoints. To compare different diagnostic groups with re-pect to a binary endpoint (percentage of residual defects

able 1. Rate and Size of Residual VSD

Diagnosis/Age at Surgery Size TEE ICU

VSD (n � 46):Mean age 10.2 months no VSD 37 32(range, 0.6–72.4) � 2 mm 9 13

� 2 mm 0 1Total residual VSD% 20% 30%

allot (n � 52):mean age 11.6 months no VSD 44 28(range, 0.2–48.1) � 2 mm 6 19

� 2 mm 2 5Total residual VSD% 15% 46%

solated VSD (n � 100):Mean age 38.0 months no VSD 75 70(range, 0.2–192.4) � 2 mm 19 27

� 2 mm 6 3Total residual VSD% 25% 30%

otal group (n � 198):no VSD 156 130� 2 mm 34 59� 2 mm 8 9

otal residual VSD (%) 21% 34%

VSD � atrioventricular septal defect; ICU � intensive care unit; T

t any given follow-up point), a �2 test was used. s

esults

edian follow-up was 3.1 years (range, 0.5 to 9.7 years),nd was similar in all three diagnostic groups. Discrep-ncy was observed between the intraoperative TEE find-ngs and those of the first TTE performed in the intensiveare unit, either on the same day as the operation or oneay afterward. As seen in Table 1, the first TTE on the

CU disclosed a 30% rate of residual VSD versus only0% on intraoperative TEE in patients after AVSD repair;6% versus 15% after repair of tetralogy, and 30% versus5% after isolated VSD closure, respectively. For the groups a whole (n � 198), this discrepancy between intraop-rative TEE and the first TTE on the ICU with regard toesidual VSD detection was significant (p � 0.002). Whenhe analysis was performed separately for each diagnos-ic group these discrepancies were no longer significantor patients after AVSD repair (p � 0.27), nor for patientsfter isolated VSD closure (p � 0.85), while they weretrongly significant for the TOF group (p � 0.0001).

Taking the whole group into consideration the evolu-ion of residual defects was followed over time, withegard to either spontaneous closure or on the contraryidening of the defect, independent of the size at initialetection. When comparing the first postoperative TTEerformed on the ICU to the TTE at last follow-up, 75% ofll residual defects closed and 2% became larger. Whenomparing the TTE before hospital discharge and at lastollow-up, 76% closed and no defects reopened. The

HospitalDischarge

Ambulatory(Median, 0.8 Years;

Range, 0.03–5.4)

Last Follow-up(Median, 3.1 Years;

Range, 0.5–9.7)

29 45 4516 1 11 0 0

37% 2% 2%

32 45 4516 4 44 3 3

38% 13% 13%

65 87 9131 9 64 4 3

35% 13% 9%

126 177 18163 14 119 7 6

36% 11% 9%

transesophageal echocardiography; VSD � ventricular septal defect.

ame analysis was performed in the three diagnosis

spcasttrtt

wpqaTgp

tt3dcdsrAdn

C

SfnapWdts1ctbpohoyr

13eohca

t(onrac[oddiTbnmrsttbsssbtc

ldshVppwwsdiwfdcrtiwdsspa

tdlar

904 DODGE-KHATAMI ET AL Ann Thorac SurgRESIDUAL SHUNTS AFTER CLOSURE OF VSD 2007;83:902–6

CA

RD

IOV

ASC

ULA

R

ubgroups. From hospital discharge to last follow-up inatients with AVSD repair, 94% of residual defectslosed. No defects became larger or reopened. In patientsfter tetralogy repair or isolated VSD closure, defectspontaneously closed in 65% and 74% of cases, respec-ively. Although there was a stronger trend toward spon-aneous residual defect closure in patients after AVSDepair, there was no statistically significant difference inhe evolution toward spontaneous closure among thehree diagnostic groups (p � 0.59) in time.

Patch closure of the VSD was performed in 178 cases,hereas direct VSD closure was judged feasible in 20atients. A residual defect was significantly more fre-uent after direct closure (5 of 20; 25%) as compared withfter-patch closure of a VSD (12 of 178; 6.7%) (p � 0.006).he small numbers of residual defects in each diagnosticroup with regard to closure technique were too small toerform a meaningful subgroup analysis.After hospital discharge, 83% of all residual defects less

han 2 mm closed. Of the nine residual defects greaterhan 2 mm, only three closed after a median follow-up of.1 years. In all other patients with persistent residualefects at last follow-up the shunts were hemodynami-ally insignificant, as documented by TTE and electrocar-iogram (neither enlargement of cardiac chambers norigns of ventricular strain, respectively), the patientsequired no medication, and no endocarditis was noted.t last follow-up, there was no statistically significantifference between residual VSDs among the three diag-ostic groups (p � 0.135).

omment

urgical closure of an isolated VSD is a routinely per-ormed operation, using standardized surgical tech-iques and postoperative care management. Outcomesre excellent, with mortality and morbidity rates ap-roaching zero in almost all centers internationally [1].hen a VSD is part of a more complex congenital heart

isease, such as tetralogy of Fallot or complete atrioven-ricular septal defect, they are routinely closed, withurgical mortalities ranging from 1% to 5% and 3% to6%, respectively [6, 7]. In the more modern era, electivelosure is performed in infancy or early childhood,hereby reducing the potential residual long-term mor-idity after successful surgical repair from long-standingreoperative pulmonary hypertension or volume overloadf the ventricle. Therefore, the focus on the postsurgicalistory of these defects lies more in the long-term qualityf life and functional status of the growing children andoung adults, which will be influenced by eventualesidual lesions, one of which is a VSD shunt.

We found a global residual shunt rate ranging between5% and 25% on intraoperative TEE, compared with a0% to 46% rate of residual defects detected by postop-rative TTE in the ICU, and a rate between 35% and 38%n TTE by hospital discharge. This incidence may seemigh but includes a majority of 1 mm defects, which arelearly hemodynamically insignificant. After subgroup

nalysis, the discrepancy between TEE and first TTE in a

he ICU was significant only for patients after TOF repairp � 0.0001). There are two possible interpretations of thisbservation: (1) either the TEE disclosed a “false-egative” finding, whereby a residual VSD was incor-ectly missed in the operating room; or (2) TEE provided“true-negative” observation, whereby an accurate TEE

orrectly excluded a residual shunt in the operating room4, 5, 8] but which unfortunately “reopened” by the timef the subsequent TTE in the ICU. An explanation for theefects incorrectly missed by intraoperative TEE could beue to the tachycardia and hypovolemia that are frequent

mmediately after coming off cardiopulmonary bypass.here is often turbulence in the various cardiac cham-ers, owing to indwelling cannulae of the cardiopulmo-ary bypass system, to jets against intracardiac patchaterial, to jets in the resected muscle bundles of the

ight outflow tract, or to jets created by a newly recon-tructed pulmonary valve, which may make the distinc-ion with a residual shunt more difficult. Furthermore,ransiently elevated right ventricular pressures, shadowingy the anteriorly positioned patch material, or peripatchwelling in the operating room may lead to unmasking ofmall residual defects only days after surgery [3]. In oureries, this is illustrated by the significant discrepancyetween TEE and TTE findings after complete repair of

etralogy, as compared with that after isolated VSDlosure or AVSD repair.

Similar to our study and findings, Yang and colleagues [3]ooked at the frequency and significance of residualefects in 294 patients undergoing surgical closure of aimple VSD, or that with associated complex congenitaleart disease. The intraoperative TEE detected a residualSD in 96 of 294 patients (33%), being slightly higher inatients after closure of an isolated VSD (41%) as com-ared with that after correction of more complex defectsith a VSD component (30%), although the differenceas not significant. From the 96 of 294 residual defects

een by intraoperative TEE, only 33 had a residual VSDetected by TTE at hospital discharge (34%). Interest-

ngly, 15 patients who had no VSD in the operating roomere noted to have a residual defect within 18 months of

ollow-up after surgery, although these were all smallefects less than 2 mm. Yang and colleagues [3] con-luded as to the utility of TEE as an assessment tool ofesidual defects in the operating room, stating that onehird of patients will have a residual defect detectedmmediately after surgery but that two thirds of theseill spontaneously close by hospital discharge. For them,efects larger than 4 mm should undergo immediateurgical revision, and those with tetralogy of Fallot with amall residual defect at the superior aspect of the VSDatch should be followed closely for defect enlargementnd a late significant shunt [3].In our study population, 83% of all residual shunts less

han 2 mm closed spontaneously, and the vast majorityid so within a year after the initial operation. Those

arger than 2 mm closed in only three of nine patients,fter a median follow-up of 3.1 years. Patients withesidual defects were asymptomatic, medication-free,

nd had no episodes of endocarditis at last follow-up, so

tS1mfaiswpTchaBpwDctth

smmpautVc

actsnttvssoDad

roswl

adIpatfd

aarbpliiasTa

R

1

2

3

4

5

6

7

8

9

905Ann Thorac Surg DODGE-KHATAMI ET AL2007;83:902–6 RESIDUAL SHUNTS AFTER CLOSURE OF VSD

CA

RD

IOV

ASC

ULA

R

hat their quality of life was judged normal to excellent.imilar to our findings, in a retrospective study including09 patients operated for closure of an isolated VSD after aean follow-up of 14.5 years, Meijboom and colleagues [2]

ound 84% of patients to subjectively assess their healths good or very good. Mean exercise capacity was normaln 84% of patients and 89% were free of any medical orurgical intervention since the operation. Residual VSDsere found in 7% of patients and none had signs ofulmonary hypertension or symptomatic arrhythmias.he incidence of late death was 2%. The authors con-luded that, despite the presence of many anatomic,emodynamic, or electrophysiologic sequelae, virtuallyll patients are asymptomatic [2]. In a more recent study,ol-Raap and colleagues [9] studied 188 consecutiveatients undergoing surgical closure of an isolated VSD,ith a mean follow-up of 2.6 years (range, 0.1 to 9.4).uring follow-up, no reoperations were necessary for

losing a residual VSD, and 37 of 73 (51%) of the residualrivial shunts disappeared spontaneously at a medianime of 3.9 years. By actuarial analysis, all trivial shuntingad disappeared by 8.4 years.We found a significant difference between residual

hunts after direct or patch closure of a VSD, namely aore than threefold incidence after direct closure. Thisay intuitively be explained by the excessive tension

ulling on the suture lines at the edges of the VSD, aftern attempt at direct closure. Currently, this is onlyndertaken in our unit with small defects in older pa-

ients who have sufficiently fibrosed margins of theirSD, which allows for a secure and permanent defect

losure. Patch closure is otherwise the norm.Residual shunts closed spontaneously more frequently

fter complete repair of AVSD, as compared with afterorrection of TOF and closure of an isolated VSD, al-hough this did not reach statistical significance. Wepeculate that the increased “surgical traffic,” namely theeo-crux cordis of the heart, consisting of the interface of

wo patches for VSD and atrial septal defect closure, andhe suture abundance with reconstruction of two atrio-entricular valves, will lead to more inflammation andubsequent fibrosis, which may allow for eventual clo-ure of a residual shunt. The use of xenopericardium inur series, rather than more inert material such asacron or Gore-Tex, may have induced more postoper-

tive inflammation leading to closure of small residualefects, although this remains speculative.Limitations of this study are inherent foremost to its

etrospective nature. Although not influencing the post-perative results with regard to spontaneous defect clo-ure or being a focus of this study, there were instanceshere we went back on cardiopulmonary bypass to close

arger residual shunts, as detected by TEE or intraoper-

tive saturation measurements, although the exact inci-ence cannot be quantified due to missing observations.

nterobserver variability may have biased the quality andrecision of the findings as our team consists of twonesthesiologists specialized in pediatric cardiac anes-hesia and TEE, and seven pediatric cardiologists per-orming the postoperative TTE in the intensive care unit,uring the in-hospital stay, and at out-patient follow-up.In conclusion, residual shunts smaller than 2 mm will

lmost always close spontaneously, and more surely softer AVSD repair, most often by one year after surgicalepair. Defects greater than 2 mm are unlikely to closeut are hemodynamically and clinically irrelevant for theatient. Even after a perfect repair of AVSD and TOF,

ifelong endocarditis prophylaxis is indicated. At ournstitution, for patients after successful closure of ansolated VSD, endocarditis prophylaxis is discontinuedfter six months. In this subgroup, revision of a residualhunt greater than 2 mm, as detected by intraoperativeEE, may spare lifelong endocarditis prophylaxis and(or)void subsequent reoperation.

eferences

. Mavroudis C, Backer CL, Jacobs JP. Ventricular septal defect.In: Mavroudis C, Backer CL, eds. Pediatric cardiac surgery.3rd ed. Philadelphia, PA: Mosby; 2003:298–320.

. Meijboom F, Szatmari A, Utens E, et al. Long-term follow-upafter surgical closure of ventricular septal defect in infancyand childhood. J Am Coll Cardiol 1994;24:1358–64.

. Yang S-G, Novello R, Nicolson S, et al. Evaluation of ventric-ular septal defect repair using intraoperative transesophagealechocardiography: frequency and significance of residualdefects in infants and children. Echocardiography 2000;17:681–4.

. Rosenfeld HM, Gentles TL, Wernovsky G, et al. Utility ofintraoperative transesophageal echocardiography in the as-sessment of residual cardiac defects. Pediatr Cardiol 1998;19:346–51.

. Muhiudeen IA, Robertson DA, Silverman NH, Haas GS,Turley K, Cahalan MK. Intraoperative echocardiography forevaluation of congenital heart defects in infants and children.Anesthesiology 1992;76:165–72.

. Hirsch JC, Bove EL. Tetralogy of Fallot. In: Mavroudis C,Backer CL, eds. Pediatric cardiac surgery. 3rd ed. Philadel-phia, PA: Mosby; 2003:383–97.

. Backer CL, Mavroudis C. Atrioventricular canal defects. In:Mavroudis C, Backer CL, eds. Pediatric cardiac surgery. 3rded. Philadelphia, PA: Mosby; 2003:321–38.

. Wienecke M, Fyfe DA, Kline CH, et al. Comparison ofintraoperative transesophageal echocardiography to epicar-dial imaging in children undergoing ventricular septal defectrepair. J Am Soc Echocardiogr 1991;4:607–14.

. Bol-Raap G, Weerheim J, Kappetein AP, Witsenburg M,Bogers AJJC. Follow-up after surgical closure of congenitalventricular septal defect. Eur J Cardiothorac Surg 2003;24:

511–5.