Result of biventricular repair for double-outlet right ventricle

Surgery for Congenital Heart Disease

Result of biventricular repair for double-outlet right ventricle The choice of optimal repair for many patients with double-outlet right ventricle continues to challenge the heart surgeon. We present the results of a 10-year surgical experience with the biventricular repair for double-outlet right ventricle with situs solitus and atrioventricular concordance. Preoperative anatomic findings within this population of 73 patients are detailed. These morphologic features are correlated with type of anatomic repair and clinical outcome. Patients were classified by ventricular septal defect location. Normal coronary anatomy was found in the majority of patients with subaortic and doubly-committed ventricular septal defects. Patients with subpulmonary and noncommitted ventricular septal defects had a wide variety of coronary anatomy. Patients with subpulmonary and noncommitted ventricular septal defects also had a considerably higher prevalence of aortic arch obstruction. A tricuspid-to-pulmonary annular distance equal to or greater than the diameter of the aortic anulus was found to indicate the possibility of achieving a conventional ventricular septal defect-to-aorta intraventricular tunnel repair. Tricuspid-to-pulmonary annular distance sufficient for intraventricular tunnel repair predominates in those patients with a right posterior or right side-by-side aorta. Five types of repair were used during the study period: intraventricular tunnel repair, arterial switch with ventricular septal defect-to-pulmonary artery baffle, Rastelli-type extracardiac conduit repair, Damus-Kaye-Stansel repair, and atrial inversion with ventricular septal defect-to-pulmonary artery baffle. Overall actuarial survival estimate at 8 years is 81 %. The presence of multiple ventricular septal defects and patient weight lower than the median were nearly significant risk factors for early mortality (p < 0.06). Nineteen patients (26 %) required 24 reoperations. Patients with subaortic ventricular septal defects were significantly reoperation free (p < 0.05). Patients with noncommitted ventricular septal defects were at significantly higher risk for reoperation during the study period (p < 0.05). The prevalence of late right or left ventricular outflow obstruction in the nonsubaortic groups is concerning. The median age at repair in this series was 0.76 years, and there was a nonsignificant trend (p = 0.13) for early mortality in patients younger than 1 year of age. These patients tended to have other serious cardiac anomalies associated with double-outlet right ventricle that necessitated early operation. On the basis of these data, we favor early repair for double-outlet right ventricle if possible. (J THoRAe CARDIOVASC SURG 1994;107:338-50)

Mitsuru Aoki, MD (by invitation), Joseph M. Forbess, MD (by invitation), Richard A. Jonas, MD, John E. Mayer, Jr., MD, and Aldo R. Castaneda, MD, PhD, Boston, Mass.

From the Department of Cardiovascular Surgery, Children's Hospital, and the Department of Surgery, Harvard Medical School, Boston, Mass.

Address for reprints: Aldo R. Castaneda, MD, PhD, Department of Cardiovascular Surgery, Children's Hospital, 300 Longwood Ave., Boston, MA 02115.

Read at the Seventy-third Annual Meeting of The American Association for Thoracic Surgery, Chicago, Ill., April 25-28, 1993.

0022-5223/94 $1.00 + .10

338

Copyright

I:

1994 by Mosby-Year Book, Inc. 12/6/51197

The Journal of Thoracic and Cardiovascular Surgery Volume 107, Number 2

The first repair of double-outlet right ventricle (DORV) with a subaortic ventricular septal defect (VSD) and atrioventricular concordance was performed by Kirklin in 1957. 1 Since then the broad spectrum of pathophysiologic presentations and great anatomic variability within this anomaly has been recognized. The surgeon intent on biventricular repair for DORV is therefore presented with a number of surgical options. An earlier report presented the early and late results for repair of DORV at this institution.? The purpose of this review is to present the operative results from a subsequent IO-year surgical experience with biventricular repair for DORV with situs solitus and atrioventricular concordance. Early and late mortality, as well as need for reoperation and functional status, are correlated with both preoperative surgical anatomy and type of repair.

Materials and methods Seventy-three patients with situs solitus and atrioventricular concordance underwent biventricular repair for DORV at The Children's Hospital, Boston, Mass., between October 1981 and December 1991. The diagnosis of DORV was based on angiographic and two-dimensional echocardiographic evidence that both great arteries originated predominantly from the right ventricle. Specifically, we used the "50% rule," which requires that one great artery arise fully over the right ventricle and that more than 50% of the other great artery also originate from the right ventricle. Five types of anatomic repair were performed: ( I) intraventricular repair with baffle from the VSD to the aorta (IVR), (2) arterial switch operation (ASO) with baffle from VSD-to-pulmonary artery, (3) Rastelli-type right ventricleto-pulmonary artery conduit with bafflefrom VSD to aorta, (4) Damus-Kaye-Stansel procedure, (5) atrial inversion with the Senning or Mustard operation and baffle from the VSD to the pulmonary artery. Ten patients met the above anatomic criteria for inclusion but underwent a Fontan-type univentricular repair. Uniform findings in this group included the presence of a noncommitted atrioventricular septal defect and abnormal atrioventricular valvechordal attachments. Appropriate closure or baffle of the VSD in these patients was therefore deemed impossible. These patients were excluded from subsequent analyses. A bilateral subarterial conus was documented before the operation in 51 patients (70%). There were 40 male patients (55%) and 33 female patients (45%). Age at operation ranged from 4 days to 18.9 years, with a median age of 0.76 years and a mean age of 1.9 years (standard deviation [SD] = 2.9 years). Patient weight at operation ranged from 2.0 to 58 kg with a median of 6.4 kg and a mean weight at repair of 8.7 kg (SD = 7.6 kg). Thirty-three patients had undergone previous palliative procedures (Table I). Patient data were compiled by review of clinical records, including operative reports, and preoperative angiographic and two-dimensional echocardiographic studies. Tricuspid-to-pulmonary valve annular distance (TPD) was obtained from review of these echocardiographic results or intraoperative observations. Follow-up information was collected either from the patient or parentts) directly or from referring physicians. Follow-up information was obtained for 58 of 65 (89%) operative survivors.Time to follow-upranged

Aoki et al.

339

Table I. Previous palliative procedures Procedure

No.

BTS BTS then 2nd BTS with pectus repair

9

PAB

9

PAB with ectopia cordis repair PAB then BTS AAR Simultaneous AAR/PAB AAR followed by PAB Simultaneous AAR/PAB then BTS Blalock-Hanlon procedure Repair of vascular ring Total (%)

13 6 I 2 I I 33 (45)

BTS, Blalock-Taussig shunt; PAB, pulmonary artery band; AAR, aortic arch repair.

from 0.3 to II years, with a median follow-upof 4.2 years. This compiled a total of 264 patient-years. Early survival was assessed by the ability to leave the hospital after repair. Late results were evaluated by prevalence of death and reoperation, as well as postoperative functional status and subjective assessment of quality of life at the time of follow-up. Potential risk factors for early mortality were analyzed in contingency tables with X2 or Fisher's exact tests. Multivariate analysis of early survival and reoperation-free survival were performed with stepwise logistic regression and Cox regression methods. The relationship of discrete variables to reoperation-free survivalwas examined with the log-rank test. Survivorship estimates were made with the Kaplan-Meier method. All analyses were performed with the use of a standard commercially available software package (SAS Institute, Inc., Cary, N.C.). Anatomic findings VSD. Patients were categorized by the relationship of the VSD to the arterial outflow tracts. This basic anatomic distinction within DORV has been used to classify patient subgroups by other investigators as initially proposed by Lev and associates.' The VSD was subaortic in 31 patients (42%). Four of these patients also had an additional noncommitted VSD, and the subaortic defect extended to the inlet portion of the septum in one patient. Three patients had a single subaortic defect that was restrictive. Twenty-seven patients (37%) had a subpulmonary VSD. Five patients with a subpulmonary VSD also had some form of pulmonary stenosis. The remaining 22 cases of DORV with subpulmonary VSD and without pulmonary stenosis could be loosely defined as Taussig-Bing hearts. One patient in this group had an additional inlet septal defect. The subpulmonary defect extended to the inlet septum in three patients and was restrictive in two patients. Five patients had a VSD equally committed to both great arterial outflow tracts. These were all unrestrictive defects, and one patient had an additional defect of the muscular septum. The 10 remaining patients (14%) had noncommitted VSDs: five (50%) of these were atrioventricular canal-type (inlet) defects, one (10%) was muscular, and four (40%) were conoventricular, but distant from both arterial outflow tracts. Three of these VSDs (one from each region) were restrictive. Associated arterial outflow tract obstruction. Fifty-two patients (71 %) had pulmonary or aortic outflow obstruction (Fig. 1); 27 (37%) patients were had a preoperative diagnosis of

340

The Journal of Thoracic and Cardiovascular Surgery February 1994

Aoki et af.

SubAo

Sub P

31 (42%)

27 (37%)

15 (48%)

5 (19%)

4 (13%)

4 (15%)

(3%)

14 (52%)

2 (6%)

14 (52%)

N=73 n(%N)

ePulmonary Stenosis

VSD

e Aortic Stenosis eAoArch Obstruction eTPD
VSD

Subpulmonary VSD

Subaortic VSD

N=73

NonDoublycommitted committed 10 (14%)

5 (7%)

ePulmonary Stenosis

5 (50%)

2 (40%)

e Aortic Stenosis

3 (30%)

3(60%)

• AoArch Obstruction

2 (20%)

2(40%)

eTPD
4 (40%)

o

n (%N)

Noncommitted

Do.ubly-

Fig. 1. Important anatomic features of VSD groups. Ao, Aortic; P, pulmonary. pulmonary outflow tract obstruction. In patients with a subaortic VSD, 15 (48%) had valvular or subvalvular pulmonary stenosis (or both). Two patients (40%) with doubly-committed VSDs and four patients (40%) with noncommitted VSDs had pulmonary stenosis. In contrast, only five patients (19%) with subpulmonary VSD had associated pulmonary stenosis. Only two patients (3%) had both aortic and pulmonary stenosis. Aortic stenosis at the subvalvular or valvular level (or both) was seen in 14 of the 73 patients (19%) in this series. Four of 31 patients (13%) with a subaortic VSD and four of27 patients (15%) with a subpulmonary VSD had aortic stenosis. Aortic stenosis was present in three of 10 patients (30%) with a noncommitted VSD and three of five patients (60%) with a doubly-committed VSD. Eight of these 14 patients (57%) with aortic stenosis also had aortic arch obstruction in the form of coarctation or variable degrees of arch hypoplasia. Aortic arch obstruction was present in 19 patients (26%). This associated lesion was relatively predominant in the subpulmonary VSD group, where 14 of the 27 patients (52%) had arch obstruction: 12 patients with coarctation, three patients with arch hypoplasia, and one with an arch interruption. Only one patient (3%) with subaortic VSD also had aortic arch obstruction (coarctation). Two patients (40%) with a doubly committed VSD had associated arch obstruction (coarctation), where-

as two patients (20%) with a noncommitted VSD had this additional lesion (hypoplastic arch and interrupted arch). Great artery relationship. The relationship of the aorta relative to the pulmonary trunk within each of the VSD subgroups is shown in Fig. 2. The subaortic VSD group had a right posterior-oblique aorta in 16 cases (52%), a right side-by-side aorta in 11 cases (35%), and two cases each with right anterior-oblique and left anterior-oblique aortic position. Patients with a subpulmonary VSD had a right side-by-side aorta in 16 cases (59%), a right anterior-oblique aorta in seven cases (26%), a left anterior-oblique aorta in two cases (7%), and one case each with right posterior-oblique and directly anterior aortic position. Tricuspid-pulmonary valve annular distance. Tricuspid-topulmonary valve annular distance (TPD) is currently a morphologiclandmark that features prominently in the preoperative evaluation of patients with DORV. The importance ofTPDin this setting was first proposed by Sakata and associates." TPD greater than the diameter of the aortic anulus is believed to indicate that conventional intraventricular rerouting (IVR) with blood flow baffled directly from the VSD to the aortic outflow tract and preservation of native right ventricle-to-pulmonary arterial continuity is feasible.?Twenty-nine patients (94%) with subaortic VSD had sufficient TPD. The two patients with insufficent TPD (6%) were the only patients in this subgroup

The Journal of Thoracic and Cardiovascular Surgery Volume 107, Number 2

Aoki et al.

VSD

n

SA SP NC

31 27 10

DC

5

VSD

n

SA SP

31 27 10

NC DC

5

~(8

RPO

16 1 1 0

AP

~ (4%) (10%)

11 (35%) 16 (59%) 5 (40%) 3 (60%)

~[E

LAO

2 2 1 0

~

RAO~

RSS

(52%) (4%) (10%)

0 1 1 0

~

34 1

(6.5%» (7%) (10%)

2 7 2 1

(7%) (26%) (20%) (20%)

"+ Post

Ant

Fig. 2. Great artery relationships by VSD location. RPO, Right posterior oblique aorta; RSS, right side-by-side aorta; RAO, right anterior-oblique aorta; AP, aorta directly anterior; LAO, left anterior-oblique aorta; SA, subaortic; SP, subpulmonary; NC, noncommitted; DC, doubly-committed.

with the aorta anterior to the pulmonary trunk. Similarly, all patients with doubly-committed VSD were judged to have sufficient TPD. Only 13 (48%) patients with subpulmonary VSD and six (60%) with noncommitted VSD had sufficient TPD. Table II shows that if the aorta is posterior and right oblique to the pulmonary trunk, the TPD is always adequate, but, if the aorta is directly anterior or left anterior oblique to the pulmonary trunk, the TPD is never sufficient. Conus and conal septum. The presence of a bilateral subarterial conus was found in 51 patients (70%) in this series. The conal septum was noted to be long and prominent in 28 patients. Partial resection was performed in a total of 19 patients. Five patients had tricuspid chordae inserted on the conal septum. This feature necessitated abandonment ofIVR in three patients. Seven patients (25%) with a prominent conal septum had subaortic stenosis reported before the operation. Similarly, nine patients (32%) with a long conus had a diagnosis of subpulmonary stenosis. This prevalence of subaortic and subpulmonary stenosis is not markedly higher than those for the full series. Coronary anatomy. Coronary anatomy was classified according to the relationship of the great arteries as normal (left coronary artery from right-posterior facing sinus), usual type for transposition of the great arteries (left coronary artery from left-anterior facing sinus), single coronary, and other variants (Fig. 3). Normal coronary patterns predominate in the subaortic VSD (77%) and doubly-committed VSD (60%) subgroups. In contrast, patients with subpulmonary and noncommitted VSDs tended to have diverse coronary patterns, with only 30% and 40% of each group, respectively, possessingthe normal coronary pattern. Table III shows the strong correlation between great artery relationships and coronary anatomy. Fifteen (83%) of the 18 patients with the aorta in a right posterior oblique positiondisplayed a normal coronary pattern. All seven patients

Table II. TPD versus great artery relationship Aortic position versus PA

RPO RSS RAO AP LAO Unde! Total

No.

TPD>Ao

TPD
18 35 12 2 5 I 73

18 (1000/0) 29 (83%) 5 (42%) o (0%) 0(0%) I (100%) 53 (73%)

0(0%) 6 (17%) 7 (58%) 2 (100%) 5 (100%) 0(0%) 20 (27%)

PA, Pulmonary artery; Ao, aortic anulus; RPO, right posterior oblique; RSS, right side-by-side; RAO, right anterior oblique; AP, anterior-posterior; LAO, left anterior oblique; Under, undetermined.

with the aorta in a directly anterior or left anterior oblique position had a coronary anatomy like that with transposition of the great arteries. A majority (63%) of patients with the side-by-side great vessels still had the normal coronary pattern, but those patients with the aorta in a right anterior oblique position had a wide variety of coronary anatomy. Surgical treatment Surgical strategy. Conventional IVR was the preferred surgical approach whenever possible;43 patients (59%) underwent IVR in the present series (Table IV). Ten of these (23%) were performed via a right atrial approach and required no right ventricular outflow patch. Of the remaining 33 patients, 30 (91 %) received an outflow patch at the time of IVR through a right ventriculotomy. Six of these patches (25%) were transannular. Although the prevalence of pulmonary stenosis known before operation in this group was appreciable, 15 (50%) of these patches were placed in patients without the preoperative

The Journal of Thoracic and

342

Cardiovascular Surgery

Aoki et al.

February 1994

VSD Location Coronary Pattern

~

W Other Variants

DC

Total

SubAo

Sub P (0=27)

(0=10)

(0=5)

(n=73)

24 (77%)

8(30%)

4 (40%)

3 (60%)

39 (53%)

2

(7%)

8(30%)

3 (30%)

0

13 (18%)

2

(7%)

3 (11%)

1 (10%)

1 (20%)

3 (10%)

7(26%)

1 (10%)

0

1 (4%)

1 (10%)

1 (10%)

(0=31)

Unrecorded

NC

7 (10%)

11 (15%) 3

Fig. 3. Coronary anatomy by VSD location. Ao, Aortic; P, pulmonary; NC, noncommitted; DC, doubly-committed.

Table III. Coronary anatomy versus great artery relationship Coronary pattern

RPO

RSS

RAO

Normal TGA-type Single from posterior sinus Other singular pattern Other variants Unrecorded Total

15

22

2

2

4

3 3

3

2

4

1

18

I 35

12

AP

LAO

2

5

2

5

1 1

Undet

Total 39 (53%) 13 (18%) 7 (10%) 4 (5%) 7 (10%) 3 (4%)

73

RPO, Aorta right posterior oblique; RSS, aorta right side-by-side; RAO, aorta right anterior oblique; AP, aorta directly anterior; LAO, aorta left anterior oblique;

Undet, could not be determined from available records; Normal. left coronary artery from posterior sinus; TGA-type, transposition of the great arteries-type with left coronary from anterior sinus.

diagnosis of pulmonary stenosis. As mentioned previously, only two IVRs were performed in patients with a TPD less than the diameter of the aortic anulus. In both cases, the baffle was placed inferior to the pulmonary valve. Anatomy precluding IVR in the remaining 30 patients included insufficient TPD in 17 (57%), abnormal tricuspid valve insertion on the conal septum in three (10%), aortic valve hypoplasia in three (10%), pulmonary stenosis with subpulmonary coronary in four (13%), proximal pulmonary trunk distortion after pulmonary artery banding with critical subaortic stenosis in one (3%), coronary anatomy precluding right ventriculotomy in one (3%), and a severely restrictive subpulmonary VSD in one (3%). When IVR was inappropriate for these reasons, ASO with VSD-to-pulmonary artery baffle was considered and performed in 10 (33%) of the remaining 30 patients in this series. Nine (90%) of these patients had a subpulmonary VSD, with the remaining patient having a noncommitted conoventricular VSD. The Lecompte maneuver- 6 was performed in all but one of the ASO group. The median age at time of ASO was 0.19 years, well below that for the series as a whole. A VSD-to-aorta baffle accompanied by external right ventri-

cle-to-pulmonary artery conduit (conduit repair) was performed in eight patients. ASO was contraindicated in these patients by the presence of significant subpulmonary stenosis, with a subpulmonary coronary in four patients (50%), pulmonary valvular hypoplasia or atresia in two (25%), and coronary anatomy unsuitable for ASO in two (25%). A Darnus-KayeStansel operation':" was performed in seven patients. Indications for this repair included a hypoplastic aortic anulus in three (43%), coronary anatomy unsuitable for ASO in two (29%), abnormal coronary sinus drainage precluding the Lecompte maneuver in one (14%), and severe proximal pulmonary trunk distortion after pulmonary banding accompanied by marked subaortic stenosis in one (14%). Five patients in this DamusKaye-Stansel group had the VSD baffled to both great arteries, whereas the remaining two had a VSD-to-aorta baffle placed. Atrial level switch with VSD-to-pulmonary artery baffle was performed in five patients. Coronary anatomy precluded right ventriculotomy or ASO in two patients, and three patients underwent repair before our use of ASO for this patient population. The VSD was determined to be restrictive before the opera-

The Journal of Thoracic and Cardiovascular Surgery Volume 107, Number 2

Aoki et at. 3 4 3

Table IV. Type of repair versus VSD location Location of VSD

IVR

Subaortic Subpulmonary Doubly-committed Noncommitted Total (n = 73)

29 5 5 4 43 (59%)

ASO

Conduit

DKS

SenfMus

2 9

3

6

4

1

3 9 (12%)

1 7 (10%)

1 5 (7%)

10 (14%)

Conduit. Extracardiac conduit repair; DKS. Damus-Kaye-Stansel procedure; SenfMus, atrial inversion with Senning or Mustard procedure.

tion in eight patients (11%). VSD enlargement with resection of muscular septum was performed in a total 14 cases (19%) to improve baffle geometry or reduce the potential for residual transseptal pressure gradient. As mentioned previously, the conal septum was partially resected in 19 patients (26%) to reduce the potential for subarterial stenosis and improve baffle geometry by allowing a more direct route from the VSD to the semilunar valve.

Results There were eight hospital deaths (11 %) in the present series. Age at operation, type of repair, location of VSD, and cause of death for these cases are shown in Table V. Eighteen potential risk factors for early mortality were analyzed as shown in Table VI. The presence of multiple VSDs emerged from both univariate and multivariate analyses with nearly significant p values of <0.06. Six patients had multiple VSDs, and two hospital deaths occurred in this group. One death was related to a residual VSD, and the second resulted from residual aortic arch obstruction. There was a statistically nonsignificant trend identifying age at repair at less than I year (p = 0.132) and weight at less than median (p = 0.055) as possible risk factors for early mortality in univariate analysis. Neither of these, however, was significant in subsequent multivariate analysis. Resection of conus was significant (p = 0.03) in univariate analysis but also did not reach statistical significance in multivariate analysis. Four late deaths (5.5%) occurred during the follow-up period of this study. These cases are presented in Table VII. It is noteworthy that were no late cardiac deaths in patients with a subaortic or doubly-committed YSD. One late death was not cardiac related (varicella with superimposed bacterial sepsis). One death occurred at reaperation for a residual VSD, and the causes of the remaining two were unknown. Autopsy results available from one of these patients revealed biventricular hypertrophy and an intact IVR. Overall actuarial survival is shown in Fig. 1. The actuarial survival estimate for all patients is 81 % at 8 years. Nineteen patients (26%) underwent 24 reaperations during the follow-up period of this study. Actuarial reaperation-free survival by VSD location is shown in Fig. 5. Indications for reoperation by type of repair and VSD

location are presented in Tables VIII and IX. Nine reaperations (38%) were for subaortic stenosis. Eight reaperations (33%) were for pulmonary stenosis. Four of these eight reoperations for pulmonary stenosis were for extracardiac conduit obstruction. Three patients underwent five reoperations (21%) for a significant residual VSD, and two of the seven patients who underwent DamusKaye-Stansel repair had significant aortic valve regurgitation into the right ventricle, which required patch closure of the native aortic valve. Log-rank tests were performed, and 18 potential risk factors for reaperation were examined, as shown in Table X. Only the presence of a noncommitted VSD was a significant risk factor for reaperation (p = 0.045). In contrast, patients with a subaortic VSD were significantly free from reaperation (p = 0.007). The relative risk for reaperation in this subgroup was 10.5% over the follow-up period as determined with the Cox regression model. Forty-six (79%) of operative survivors had no restrictions on their physical activity as a result of impaired cardiac status. Forty-eight (83%) patients required no cardiac medications other than endocarditis antibiotic prophylaxis where appropriate. Discussion This study emphasizes the careful definition of important anatomic features when planning and performing a biventricular repair for DORV. Classification of DORV by the location of VSD remains a useful surgicalanatomic distinction. These data confirm that patients with DORV and a subaortic VSD may safely undergo IVR with a low overall mortality and low prevalence of reaperation. Although patients younger than I year of age did comprise 65% (20 patients) of this IVR subgroup, all of the hospital deaths (three patients) occurred in patients younger than 1 year of age. These deaths were all before 1985. Two infants had sudden, unexplained cardiac failure, and one patient died of a significant residual VSD which was eventually approached via a left ventriculotomy. In view of the lower mortality and need for reaperation since 1985, we favor repair in infancy for these patients because we also believe it is preferable to avoid

344

The Journal of Thoracic and Cardiovascular Surgery February 1994

Aoki et al.

Table V. Hospital deaths Age

Procedure

VSD location

Cause of death

2 mo 5 mo 5 mo 8 days 3 mo 3 mo 28 days 2.7 yr

IVR IVR IVR ASO Conduit OKS IVR Conduit

Subaortic Subaortic Subaortic* Subpulmonary* Subpulmonary Subpulmonary Noncommitted Noncommitted

Unexplained Unexplained Low output, residual VSO Residual aortic arch obstruction Unexplained Low output, residual VSO Ventilator dependence Low output, resid AS and CHB

Conduit, Extracardiac conduit repair; DKS, Damus-Kaye-Stansel procedure; AS, aortic stenosis; CHB, complete heart block.

'Multiple VSDs

Table VI. Risk factor analysis: hospital mortality Risk factor

p Value*

Risk factor

p Value*

Age at operation Type of repair Resection of conus Location of VSO Coronary origin Great artery relationship Prominence of conus Aortic arch obstruction Previous palliation

NS NS 0.03t NS NS NS NS NS NS

Weight at repair < median Pulmonary outflow procedure VSO enlargement Multiple VSOs Coronary course Tricuspid-pulmonary distance Pulmonary stenosis Aortic stenosis Qp/Qs > median

0.055t NS NS 0.056+ NS NS NS NS NS

Qp, Pulmonary arterial blood flow; Qs, systemic blood flow. *Univariate analysis.

tNot significant in multivariate analysis. tSignificant in multivariate analysis.

Table VII. Late deaths Age at repair

Time after repair

Procedure

VSD location

Cause of death

5 mo I yr 1.3 yr 19 yr

4mo 5 mo 7.5 yr 6mo

IVR Conduit OKS Conduit

Subpulmonary Subpulmonary Subpulmonary Noncommitted

Unknown Unknown Varicella with sepsis Low output (residual VSO)

Conduit, Extracardiac conduit repair; DKS, Damus-Kaye-Stansel procedure.

the sequelae of long-term palliation. Similarly, we favor early IVR for patients with DORV with doubly-committed YSDs. DORY with a subpulmonary or noncommitted YSD has challenged surgeons throughout the modern era of congenital heart surgery. Since the initial reports of successful repair of DORY with subaortic YSD with the use of the intraventricular tunnel, I numerous reports have described the application of IYR to patients with subpulmonaryor noncommitted YSDs. I0-12 Late results of these repairs may be less than satisfactory.l-'!" Because the morbidity and mortality of the arterial switch is now low,I4-16 we favor this repair for those patients with DORY in whom the TPD is insufficient for conventional IYR. We performed 10 ASOs for such an indication with

a single hospital death in a patient who also had significant subaortic stenosis, aortic arch hypoplasia, and coarctation of the aorta. We were unable to wean this patient from cardiopulmonary bypass after early reoperation for residual aortic arch obstruction. This case is illustrative of the fact that this population of patients with DORY has a prevalence of significant associated anomalies that may complicate the repair and result in higher ASO-related mortality when compared with patients with simple TGA. Despite this, we believe that ASO with an intraventricular tunnel to the pulmonary artery is the optimal alternative for biventricular repair of DORY if conventional IYR with baffle directly traversing the TPD is not an option. The Rastelli-type right ventricle-to-pulmonary artery

The Journal of Thoracic and Cardiovascular Surgery Volume 107. Number 2

Aoki et at. 3 4 5

1.,.----------------, 0.75 Probability of Survival

4.Il-----~----___.j (51)

T 1

(25 )

81%

(9 )

0.5 0.25 0+----,------.-----,-----1 o 2 4 6 8 Years ofFollow-up

Fig. 4. Actuarial survival curve with 70% confidence intervals after biventricular repair for DORY. Numbers in parentheses indicate number of patients available to follow-up.

Prob

~

0.5

VSDLocation Subaortic Subpulmonary

0.25

Noncommitted Doubly-committed

0+----,..----=-------.--------.--

o

2

5

8

Years of Follow-up

Fig. 5. Actuarial reoperation-free survival after biventricular repair for DORY with 70% confidence intervals. external conduit repair or the Damus-Kaye-Stansel repairs proved useful in those patients who had a subaortic VSD with coronary anatomy precluding transannular pulmonary outflowpatch or those with subpulmonary or noncommitted VSDs who were not suited for the ASO. The combined overall mortality for the 15 patients undergoingconduit repair and Damus-Kaye-Stansel was 40%(sixdeaths). The rate of reoperation in this group will necessarily become 100% as growth-related conduit obstructionis unavoidable.Although these results are less than desirable, extracardiac conduit obstruction may be corrected with close to zero mortality. Moreover, we believe that preservation of the morphologically left ventricle as the systemic ventricle will avoid the serious late complications expected for atrial inversion with VSDto-pulmonary artery baffle. 17-23 Indeed, we have not performed a repair of DORV with an atrial switch for more than 5 years. The only patient who underwent repair with this technique during this decade had abnormal cor-

onary anatomy which precluded a right ventriculotomy and ASO. Regarding the correlation between great artery position and TPD, as mentioned before, Table II shows that if the aorta is in a right posterior-obliqueor right side-byside position,the TPD was sufficientfor IVR in the vast majority of cases. In contrast, if the aorta is either directly anterior or in a left anterior-oblique position the TPD was insufficient for IVR in all cases. For patients with these configurations,then, great artery relationship may also serve to direct preoperative planning. Patients with the aorta in a right anterior-oblique position are only slightlymore likelyto have insufficient TPD in this series. The abovegeneralization,therefore,doesnot hold true for these patients. These conclusions do not conflict with those previously drawn by Serraf''" or Sakata" and their associates, because it was in precisely this subgroup of patients that they found poor correlation between great artery position and TPD.

346

The Journal of Thoracic and Cardiovascular Surgery February 1994

Aoki et al.

Table VIII. Reoperations by type of repair Indication for reoperation

No.

IVR

ASO

Subaortic stenosis Membrane Baffle Restrictive VSD MV tissue Insufficient AV growth Pulmonary stenosis Conduit obstruction TV tissue Insufficient PV growth Peripheral PA stenosis Residual VSD Aortic regurgitation into RV Right ventricular failure Total (n = 19 patients)

9 3 I

5 I I I

3

2 I 2 8 4 I

Conduit

Sen/Mus

I

2 2

2

DKS

3 3

2

I

5 2

2 2

I 24

4

8

4

3

Conduit. Extracardiac conduit repair; DKS, Damus-Kaye-Stansel; SenfMus, atrial inversion with Senning or Mustard procedure; MV, mitral valve; AV, aortic valve; TV, tricuspid valve; PV, pulmonary valve; PA, pulmonary artery; RV, right ventricle.

Table IX. Reoperations by VSD location Indication/or reoperation

No.

Subaortic stenosis Membrane Baffle Restrictive VSD MV tissue Insufficient AV growth Pulmonary stenosis Conduit obstruction TV tissue Insufficient PV growth Peripheral PA stenosis Residual VSD Aortic regurgitation into RV Right heart failure Total (n = 19 patients)

9 3 I 2 I 2 8 4 I

SA

4 I

I

2 5 2 I I I

2 I 5

3

2

2

2 I 24

SP

3

I 12

6

3

SA, Subaortic; SP, subpulmonary; NC, noncommitted; DC, doubly-committed; MV, mitral valve; AV, aortic valve; TV, tricuspid valve; PV, pulmonary valve; PA, pulmonary artery; RV, right ventricle.

Analysis of hospital mortality revealed only multiple VSDs as a possibly significant risk factor for early mortality in both univariate and multivariate analyses (p < 0.06). Conal resection was significant (p = 0.03) only in univariate analysis. Examination of the causes of death and autopsy findings (when available) in this conal resection group revealed a wide range of causes of death of these patients. No anatomic correlation between partial resection of conus and operative mortality could be drawn. Damage to the first perforating branch ofthe left anterior descending coronary artery was ruled out by angiography in two of these patients who died. We continue to believe that partial conal resection to enable IVR is an important component of the surgical approach to

DORV. 25 With regard to multiple VSDs and the risk for significant postoperative residual defects, we are currently using transcatheter device closure of such VSDs before, during and early after the operation if surgical closure was unsuccessful.e-?? In our general experience, the importance of multiple VSDs as a risk factor for hospital mortality may be greatly reduced by such techniques. The trend toward higher overall early mortality in the younger patients and those with lower body weight although statistically nonsignificant, is nevertheless concerning. These patients in the subaortic VSD subgroup have been discussed previously. Those same arguments apply for the series as a whole, but additional operative risk was encountered in a number of the neonates with

The Journal of Thoracic and Cardiovascular Surgery Volume 107, Number 2

Aoki et al.

347

Table X. Risk factor analysis: need for reoperation Risk/actor

p Value*

Risk/actor

p Value*

Age at operation Type of repair Resection of conus Location of VSD Subaortic Subpulmonary Noncommitted Doubly-committed TPD Aortic arch obstruction Previous palliation

NS NS O.034t

Weight at repair Pulmonary outflow procedure VSD enlargement Multiple VSDs Coronary course Pulmonary stenosis Aortic stenosis Qp/Qs> median Great artery relationship Coronary origin Prominence of conus

NS NS NS NS NS

O.(107t§ NS

O.045t NS NS NS NS

O.Ol7t§ NS NS NS NS NS

Qp, Pulmonary arterial blood flow; Qs, systemic blood flow. *Univariate analysis. tNot significant in multivariate analysis. :j:Also significant in multivariate analysis. §Significant for freedom from reoperation.

subpulmonary and noncommitted VSDs because they had a higher prevalence of severe associated lesions, particularly subaortic stenosis and aortic arch obstruction, which were addressed in the neonatal period at the time of repair for DORV. We therefore feel that the somewhat higher hospital mortality for neonates and infants within this series, while disturbing, must be compared with the unsatisfactory results of long-term palliation for these serious defects. The prevalence of postrepair subaortic stenosis requiring reoperation was not insignificant; nine reoperations were performed in eight (11 %) patients during the follow-up period. This prevalence is similar to that previously reported. 14 Tables VIII and IX show that this complication occurs predominantly in the nonsubaortic VSD subgroups for a variety of reasons. The cause of baffle related-subaortic stenosis either in the form of a discrete membrane or diffuse fibrous proliferation is poorly understood. It may result from turbulent flow within the intraventricular tunnel and therefore develop in proportion to the complexity of the baffle course. It is also noteworthy that a relatively high postoperative prevalence (60%) of discrete subaortic stenosis exists in the small group of patients with doubly-committed VSDs. This has been documented in a previously reported seriesl' and may be a result of suture placement in the immediate subvalvar region in these patients, which should be avoided during baffle placement if at all possible. Six patients (8%) underwent eight reoperations for pulmonary stenosis, four (50%) for anticipated extracardiac conduit obstructions. Despite the high prevalence of prerepair pulmonary stenosis in patients with a subaortic VSD, only one patient, who had undergone valvotomy and outflow patch placement for valvular pulmonary

stenosis at the time of IVR, had recurrence of significant pulmonary stenosis. At the time of reoperation, the stenosis was found to be a result of persistent pulmonary annular hypoplasia. Important information is provided by the seven patients in the Damus-Kaye-Stansel subgroup. The two patients who had intraventricular baffle placed to the pulmonary artery alone both had significant aortic regurgitation into the right ventricle, requiring patch closure of the aortic valve. The five patients with baffle placement to both great arteries had no significant semilunar valve regurgitation Damus-Kaye-Stansel at the time this article was written. Consideration should be given to patch closure of the native aortic valve at the time of DamusKaye-Stansel if the intraventricular tunnel is directed only to the pulmonary valve. Finally, there are two limitations to this present study. First, the use of "the 50% rule," although clinically practical and popular with surgeons, is at odds with the more exclusive pathologic definition of DORV.28 Van Praagh and associates.f in particular, require that all cases of DORV must have both great arteries originate predominantly from the right ventricle and possess bilateral subarterial coni or, if a unilateral conus is present, the semilunar valve of that great artery with absent conus must be in fibrous continuity with the tricuspid valve, the right ventricular aspect of a straddling mitral valve, or the right ventricular aspect of a common atrioventricular valve. The integrity of this definition is unassailable. However, in a retrospective series such as this, the issue of semilunar-to-atrioventricular valve fibrous continuity is difficult to resolve without direct mention of these features in the operative record. At this point we can only note that of the 31 patients with subaortic VSD in this

348

Aoki et al.

series, nine had only subpulmonary conus and could be considered to have tetralogy of Fallot with more than 50% aortic override ifthere was indeed aortic-to-mitral fibrous continuity in these patients. Similarly, of the 27 patients with a subpulmonary VSD, six had only a subaortic conus. Pulmonary-to-mitral fibrous continuity here could be considered transposition of the great arteries with VSD and pulmonary arterial override. All patients with noncommitted and doubly-committed VSDs had bilateral conus in this series. A second limitation of this series is the failure to include those patients who underwent univentricular repair for DORV during the same study period. As the morbidity and mortality of Fontan-type procedures continues to decrease, increasing numbers of patients deemed inappropriate for biventricular repair have undergone this type of univentricular repair. 29• 30 As mentioned previously, 10 patients with DORV, situs solitus, and atrioventricular concordance underwent Fontan-type cavopulmonary anastomoses during this study period. A complete analysis of the preoperative features and postoperative outcome of this specific group should be a topic for further study. The anatomic features and postoperative results of a l Osyear experience with the biventricular repair for DORV have been presented. From these data, we conclude that patients with a subaortic VSD may undergo conventional IVR at low risk for mortality and significantly reduced risk for reoperation compared with other DORV subsets. We therefore advocate that the repair be performed in these patients when they are infants or neonates. Because of its correlation with TPD, great artery relationship may be used as a rough guide for IVR feasibility in all patients with DORV except those with a right anterior aorta. Subpulmonary VSDs continue to challenge surgeons, but the application of ASO here should continue to improve early and late results for this difficult group of difficult conditions. 12, 13 Surgical options, as position of VSD, are variable in the noncommitted VSD group. The range of associated lesions and intraventricular relationships in these patients requires extreme flexibility from the surgeon because no single type of repair is superior for this group as a whole. An increased need for reoperation was observed in all three nonsubaortic VSD subgroups. Patients with subpulmonary, noncommitted, and doubly-committed VSDs should be monitored closely for the development of late right or left ventricular outflow obstruction. Finally, early repair for patients with subpulmonary and noncommitted VSDs remains a surgical challenge. A heightened awareness of the associated anatomic features here, particularly aortic arch obstruction and the TPD, will hope-

The Journal of Thoracic and Cardiovascular Surgery February 1994

fully lead to more complete preoperative diagnoses and improved survival with these complex conditions. We thank Nancy Cook, DSc, for her assistance with statistical analysis. REFERENCES I. Kirklin JW, Harp RA, McGoon De. Surgicaltreatmentof origin of both vessels from the right ventricle, including cases ofpulmonary stenosis. J THORAC CARDIOVASC SURG 1964;48:1026-36. 2. Luber JL, Castaneda A, Lang P, Norwood W. Repair of double-outlet right ventricle: early and late results. Circulation I983;68(Suppl):IlI44-7. 3. LevM, BharatiS, MengCCL, Liberthson RR, Paul MH, IdrissF. A concept of double outlet right ventricle. J THORAC CARDIOVASC SURG 1972;64:271-81. 4. Sakata R, Lecompte Y, Batisse A, Borromee L, Durandy Y. Anatomic repairof anomalies of ventriculoarterial connection associated withventricular septaldefect. 1. Criteria of surgical decision. J THORAC CARDIOVASC SURG 1988; 95:90-5. 5. Lecompte Y, Neveux JY, Leca F, et aI. Reconstruction of the pulmonary outflow tract without prosthetic conduit. J THORAC CARDIOVASC SURG 1982;84:727-33. 6. Lecompte Y, ZanniniL, Hazan E, et aI.Anatomic correction of transposition of the great arteries: new technique without the useofa prosthetic conduit. J THORAC CARDIOVASC SURG 1981 ;82:629-3 I. 7. StanselHe. A new operation ford-loop transposition ofthe great vessels. Ann Thorac Surg 1975;19:565-7. 8. KayeMP.Anatomic correction oftransposition ofthe great arteries. Mayo Clin Proc 1975;50:638-40. 9. DamusPS, Thomson N, Mcloughlin TG. Arterial repair without coronary relocation for complete transposition of the great vessels withventricular septaldefect: report of a case. J THORAC CARDIOVASC SURG 1982;83:316-8. 10. Patrick DL, McGoon DC. Operation for double-outlet right ventricle with transposition of the great arteries. J THORAC CARDIOVASC SURG 1968;9:537-42. I I. Kawashima Y, Fujita T, Miyamoto T, Manabe H. Intraventricular rerouting of blood forthe correction ofTaussigBingmalformation. J THORAC CARDIOVASC SURG 1971; 62:825-9. 12. Pacifico AD, Kirklin JK, Colvin EV, Bargeron LM Jr. Intraventricular tunnel repair for Taussig-Bing heart and related cardiac anomalies. Circulation 1986;74(Suppl): 153-60. 13. Musumeci F, Shumway S, Lincoln C, Anderson RH. Surgicaltreatmentfor double-outlet rightventricle at the BromptonHospital, 1973-1986. J THORACCARDIOVASCSURG 1988;96:278-87. 14. Kirklin JW, Pacifico AD, Blackstone EH, Kirklin JK, Bargeron LM Jr. Currentrisks and protocols foroperations fordouble-outlet rightventricle: derivation froman 18-year experience. J THORAC CARDIOVASC SURG 1986;92:913-30. 15. Quaegebeur JM, RohmerJ, OttenkampJ, et aI.The arte-

The Journal of Thoracic and Cardiovascular Surgery Volume 107, Number 2

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

rial switch operation: an eight-year experience. J THORAC CARDIOVASC SURG 1986;92:361-84. Lupinetti FM, Bove EL, Minich LL, et al. Intermediateterm survival and functional results after arterial repair for transposition of the great arteries. J THORAC CARDIOVASC SURG 1992;103:421-7. Harvey JC, Sondheimer HM, Williams WG, Olley PM, Trusler GA. Repair of double outlet-right ventricle. J THORAC CARDIOVASC SURG 1977;73:611-5. Martin RP, Qureshi SA, Ettedgui JA, et al. An evaluation of right and left ventricular function after anatomical correction and intr-atrial repair operations for complete transposition of the great arteries. Circulation 1990;82:808-16. BenderHW Jr,StewartJR,Merrill WH, Hammon JW Jr, Graham TP Jr. Ten years' experience with the Senning operation for transposition of the great arteries: physiologic results and late follow-up. Ann Thorac Surg 1989;47:21823. Turina MI, Siebenmann R, von Segesser L, Schonbeck M, Senning A. Late functional deterioration after atrial correction for transposition of the great arteries. Circulation I989;80(Suppl):I 162-167. Arciniegas E, Farooki ZQ, Hakimi M, Perry BL, Green EW. Results ofthe Mustard operation for dextroposition of the great arteries. J THORAC CARDIOVASC SURG 1981; 81:580-7. AshrafMH,CotroneoJ, DiMarcoD, Subramanian S. Fate of long term survivors of the Mustard procedure (inflow repair) for simple and complex transposition of the great arteries. Ann Thorac Surg 1986;42:385-9. Kato H, Nakano S, Matsuda H, Hirose H, Shimazaki Y, Kawashima Y. Right ventricular myocardial function after atrial switch operation for transposition of the great arteries. Am J Cardiol 1989;63:226-30. Serraf A, Lacour-Gayet F, Bruniaux J, et al. Anatomic repair of Taussig-Bing hearts. Circulation 1991;84(Suppl):III200-5. Smolinsky A, Castaneda AR, Van Praagh R. Infundibular resection: surgical anatomy of the superior approach. J THORAC CARDIOVASC SURG 1988;95:486-94. Lock JE, Block PC, McKay RG, Keane JF. Transcatheter closure of ventricular septal defects. Circulation 1988;78:361-8. Bridges ND, Perry SB, Keane JF, et al. Preoperative transcatheter closure of congenital muscular ventricular septal defects. N Engl J Med 1991;324:1312-7. Van Praagh S, Davidoff A, Chin A, Shiel FS, Reynolds J, Van Praagh R. Double outlet right ventricles: anatomic types and developmental implications based on a study of 101 autopsied cases. Coeur 1982;13:389-440. Mayer JE Jr, Helgason H, Jonas RA, et al. Extending the limits for modified Fontan procedures. J THORAC CARDIOVASC SURG 1986;92:1021-8. Russo P, Danielson G K, Puga FJ, McGoon DC, Humes R. Modified Fontan procedure for biventricular hearts with complex forms of double-outlet right ventricle. Circulation 1988;78(Suppl):III20-5.

Aoki et al. 3 4 9

Discussion Dr. Yasunaru Kawashima (Osaka, Japan). I am particularly interested in the subpulmonic VSD-type operation, and you have done a rather large number of Rastelli-type and DarnusKaye-Stansel-type of operations. I would like to ask you whether you selected these procedures before the operation or changed from the intraventricular rerouting to those kinds of operations after opening the heart? Although I prefer intraventricular rerouting to ASO for this type of DORV, there are controversies in the selection of the operative procedure. One disadvantage to intraventricular rerouting in comparison with switch operation is considered to be the necessity of right ventriculotomy, which may impair the right ventricular function or cause ventricular ectopic beat late after the operation. My colleague, Dr. Yagihara, has performed this procedure without ventriculotomy through the right atrium and the pulmonary artery in a l-year-old boy, and neither subaortic nor subpulmonic stenosis was significant after the operation. The patient was doing quite well 8 months after the operation. I believe it is possible to select this procedure for some patients with DaRV and subpulmonic VSD because we can now understand the intraventricular structure precisely with the use of the advanced technique of echocardiography. Dr. Forbess. The primary indication for performing a Rastelli-type conduit repair was the presence of pulmonary stenosis, atresia, or both with an important coronary artery crossing the infundibulum. Damus-Kaye-Stansel repairs were performed in patients with either aortic annular hypoplasia or coronary anatomy unsuitable for the ASO. These anatomic features were diagnosed before the operation, and operative plans were made accordingly. There were unusual cases in which right ventricle-to-pulmonary artery conduits were placed because of insufficient pulmonary blood flow immediately after patch augmentation. Dr. Serafin Y. DeLeon (Maywood, Ill.). What is your current philosophy when you do the Damus-Kaye-Stansel procedure? Do you routinely close the aortic valve? We found that because the aortic valve faces the right ventricle, which eventually will become thin with low pressure, the aortic valve, which is subjected to high systolic pressure and should remain closed, eventually prolapses and becomes incompetent. On the other hand, when we use the main pulmonary artery as an outflow to the aorta in the presence of subaortic stenosis in single ventricles or univentricular hearts, we do not often see aortic prolapse and insufficiency because the aorta is still connected to a small, thick outlet chamber. When we do the Damus-Kaye-Stansel procedure, because of the prevalence of aortic insufficiency, we routinely close the aortic valve. Dr. Forbess. The two patients who required closure of the aortic valve after a Damus-Kaye-Stansel repair were the only patients who did not have the VSD baffled to both great arteries. Although the total number of patients undergoing DamusKaye-Stansel repair was small (seven), this anecdotal experience suggests that the risk of significant semilunar valve regurgitation is reduced if both great arteries receive some degree of antegrade flow. This technique is now used preferentially at our institution. My coauthors may have additional comments regarding those cases in which this type of baffle is not feasible. Dr. Mayer. I think we have had some concerns about both semilunar valves with the Damus-Kaye-Stansel procedure. This

350

Aoki et al.

issue was recently reviewed and published in Circulation, and there is actually a prevalence of valvular insufficiencywhenever the pulmonary valve is placed into the systemic circulation, including such situations as hypoplastic left heart stage I palliation, as well as the Damus-Kaye-Stansel type of operation. I think that whether that original aortic valve opens and closes is important. It seems that it is the group of patients in whom the semilunar valve is connected to a low-pressure ventricle such that it will never open (because the aortic pressure is higher than the ventricular pressure) who are at risk. I think under those circumstances one has a couple of options, one of which is to try to baffle the VSD to both great vessels; even if it is going to be a restrictive pathway from the VSD to the original aorta, at least some blood will come from the left ventricle out that semilunar valve. It seems, and this is only anecdotal experience because only a few of these patients have had to have their semilunar valve patched, that somehow the leaflets almost start deteriorating, presumably because they are not being subjected to the usual forces that open and close the semilunar valve. That is a long answer to say that I am not sure whether, on a routine basis, the semilunar valve needs to be closed when a pulmonary artery-to-aorta connection is performed, but I would not do it in all cases and would wait to reoperate if it became a significant problem. Dr. Hillel Laks, (Los Angeles, Calif). We have actually seen a similar thing with the pulmonary valve on the low-pressure

The Journal of Thoracic and Cardiovascular Surgery February 1994

side or the aortic valve becoming regurgitative, and we would now close it at the time of the initial Damus- Kaye-Stansel procedure. I wonder about the prevalence of late subaortic obstruction, which is disturbing and seems to relate to a high prevalence of reoperation. Do you have any thoughts about how to prevent this and would you consider the unmentionable word in Boston, "palliation," in some of those cases? Any ideas about the technical aspects of preventing that? Dr. Forbess. Of the 24 reoperations performed during this study period, nine were for late subaortic stenosis. The exact nature of this late stenosis was variable. Three of these nine cases involved the development of a discrete membrane within the course of the baffle. Two cases were a result of persistently restrictive VSDs, and one case was caused by mitral valve tissue obstructing the baffle inlet. Two cases resulted from insufficient aortic valve growth possibly because of sutures placed close to the valve leaflets. One additional case was caused by restrictive baffle geometry. This range of causes of late subaortic stenosis makes the preoperative determination of those patients who are likely to experience this late complication difficult, but not impossible, to address. The presence of a restrictive VSD, overall baffle geometry, and suture placement in the area of the aortic anulus all deserve special attention at th e time of repair.