Switch operation for transposition of the great arteries in neonates

J

THoRAc CARDIOVASC SURG

1988;96:354-63

Switch operation for transposition of the great arteries in neonates A study of 120 patients From March 1984 to January 1987, anatomic surgical correction was performed on 110 newborn infants (2 to 23 days old, mean 7.8 ± 3.5, standard deviation) with simple tramposition of the great arteries and 10 additional neonates (7 to 30 days old, mean 17.9 ± 8.3, standard deviation) with tramposition and a large ventricular septal defect All had preoperative catheterization. Ninety-six percent of the patients underwent balloon atrial septostomy and 90 % received prostaglandin E 1 infusion until the time of the operation. The anatomy of the coronary arteries according to the Yacoub classification was as follows: type A, 82 patients; type B, 5 patients; type C, 4 patients; type D, 23 patients; and type E, 6 patients. Continuous hypothermic bypass with no circulatory arrest was used for all patients except two. Myocardial protection was ensured by crystalloid cardioplegia. Coronary artery relocation was performed according to the Yacoub technique with some modifications, and pulmonary artery recoestrectlon was done according to the Lecompte maneuver in all patients, even when the great vessels had a side-by-side relati~hip. The proximal pulmonary artery was reconstructed with two circular patches for the first 10 patients and with a single large posterior patch for the last 110 patients. Tanned heterologous pericardium was used for the first 25 patients and autologous native pericardium for the last 95 patients. The perioperative mortality rates were 8.3% for the entire series and 5.4% for the last 110 patients, with no deaths in the group having tramposition plus ventricular septal defect Late death from acute myocardial infarction occurred in two patients in the second month after operation. No patient was lost to follow-up, which ranged from 2 to 46 months (mean 16 ± 11.2, standard deviation~ The follow-up included sequential noninvasive evaluati~ and 32 catheterizations performed 10 to 18 months postoperatively. Two patients were reoperated on for pulmonary stenosis caused by retraction of the two heterologous pericardial patches, but neither died. Six others have mild to moderate pulmonary stenosis. Two patients have trivial aortic regurgitation. None have aortic dilatation or supravalvular aortic stenosis. The lOS survivors have no cardiovascular symptoms. They all are in sinus rhythm, have normal left ventricular function, have no ischemic problems, and receive no medication.

Claude Planche, MD,a Jacqueline Bruniaux, MD,a Francois Lacour-Gayet, MD,a Jean Kachaner, MD,b Jean-Paul Binet, MD: Daniel Sidi, MD,b and Elizabeth Villain, MD,b Plessis-Robinson and Paris, France

Anatomic correction of transposition of the great arteries (TGA) I is theoretically a better therapy than atrial repair, since it does not introduce any additional From Clinique de Chirurgie Cardio-vasculaire, Hopital MarieLannelongue,' 92350 Ie Plessis-Robinson, France, and Service de Cardiologie Pediatrique, Department de Pediatric, Hopital Necker Enfants-Malades," 75734 Paris Cedex 15, France. Read at the Thirteenth Annual Meeting of The Western Thoracic Surgical Association, Colorado Springs, Colo., June 24-27, 1987. Address for reprints: Professor Claude Planche, Hopital MarieLannelongue, 133 Avenue de la Resistance, 92350 Ie PlessisRobinson, France.

354

intracardiac anomaly and it restores the left ventricle (LV) to its natural systemic function. However, anatomic correction is a delicate operation necessitating transfer of the coronary arteries and reconstruction of the pulmonary artery. The long-term fate of these vascular structures remains unknown and the perioperative mortality may be high, especially if the operation had to be performed in newborn babies.' This is the case in TGA with intact ventricular septum because of the natural underdevelopment of the subpulmonary LV.3.4 Encouraged by the successful report of the Boston group.v" in April 1984 we decided to attempt to treat simple TGA by anatomic correction in the neonatal

Volume 96 Number 3

Switch operation for TGA

September t 988

TYPE I

TYPE A : 82 PTS

~ ~

• Similar to normal distribution

TYPE II

TYPE B : 5 PTS TYPE C = 4 PTS

• -Single- coronary arteries • Coronary arteries course between the great vessels TYPE III

• Transversal course coronary arteries • With abnormal origin

355

~

~

TYPE D : 23 PTS TYPE E : 6 PTS

~ .~ .0

•

~

~I

Fig. 1. Anatomic repartition of coronary arteries by Yacoub classification. Three types are defined according to mechanism and risk of coronary relocation: Type I = similar to normal distribution; Type II = "single" coronary artery characterized by a course located between the great vessels; Type III = long transverse course of circumflex artery with an abnormal origin.

period. Well-informed parental consent was obtained in each case. We report here our results with this approach and discuss in detail the technical aspects of this operation. More recently, 10 patients with TGA associated with a single large ventricular septal defect (VSD) were also operated on in the first month after birth and are included in this report. The management of TGA with VSD is also discussed.

(j VI

c

Patients and methods Population. One hundred twenty neonates with TGA underwent an arterial switch operation from April 4, 1984, to Jan. 14, 1987, at Marie-Lannelongue Hospital, Paris, France. One hundred ten patients had simple TGA and 10 patients TGA associated with a large VSO. The mean age (± standard deviation, SO) at operation for the group with simple TGA (n = 110) was 7.8 ± 3.5 (SO) days, ranging from I to 23 days; 93% of the patients were operated on within the first 2 weeks of life. The mean age at operation for the group with TGA plus VSO (n = 10) was 17.9 ± 8.3 (SO) days; ranging from 6 to 30 days. The mean weight was 3.3 ± 0.3 (SO) kg, ranging from 1.7 to 4.8 kg, for the group with simple TGA and 3.5 ± 0.4 (SO) kg for the group with TGA plus VSO. The associated lesions included a small VSO, not closed at operation, in eight patients, tricuspid insufficiency in three patients in the simple TGA group, and coarctation of the aorta in one patient in the TGA plus VSO group. Preoperative management. All patients underwent preoperative catheterization. In the group with simple TGA (n = 110) balloon atrial septostomy was performed on all patients at the time of catheterization. Prostaglandin E, infusion (0.025Ilg/kg/min) was started before catheterization in 93 of the 110. In these, the left ventricle/right ventricular

Fig. 2. Lecompte maneuver was performed in all cases even when the vessels were side by side. ratio was 0.79/0.07 after the Rashkind procedure. Prostaglandin E 1 infusion was maintained up to the time of operation in 98 of the 110 (not started in four and discontinued in eight because of poor tolerance). Thirty-seven percent of the patients were maintained on artificial ventilation because of apnea caused by prostaglandin E 1 infusion. Delay between catheterization and operation was 6.3 ± 5 days in TGA with intact ventricular septum and 13.7 ± 7 in TGA with VSO. In the group with TGA plus VSO (n = 110), a Rashkind maneuver was performed in seven patients and prostaglandin E 1 was infused in four patients. Preoperative LV function was evaluated by two-dimensional echocardiography. LV geometry was assessed in a subxiphoid transverse view (tip of the mitral valve) by measuring in end-systole the ratio of the lateral/anterposterior diameter of the LV. This was considered a rough but reliable estimation of LV systolic pressure.'

3 5 6 Planche et al.

The Journal of Thoracic and Cardiovascular Surgery

FIRST 10 PATIENTS

Fig. 3. Relocation of coronary arteries in type I. Type I corresponds to type A of Yacoub. It is the most common form. The posterior translation is obtained in this type with a minimal risk of deformation. For the first 10 patients, the coronary anastomoses were performed after the completion of the aortic anastomosis and were completed after removal of a button of the aortic wall. In the present technique the anastomoses are done before the completion of the aortic anastomosis. A triangular incision is made on the border of the neoaorta and the ostium is then anastomosed.

Fig. 4. Reconstruction of pulmonary artery. The Valsalva sinuses, destroyed by excision of the coronary ostia, have been reconstructed with a large pantaloon patch of autologous native pericardium.

According to the geometry of the LV, 8.9 the LV pressure was judged favorable in 91 (ratio < 2), acceptable in 24 (ratio between 2 and 3), and unfavorable in five (ratio> 3). These five patients were nevertheless proposed for a switch operation because they were very young « I week), because of tricuspid insufficiency contraindicating atrial repair, or because of parental request. Anatomic findings. The anatomy of the coronary arteries according to the Yacoub classification'? was as follows: type A, 82 patients (68%); type S, 5 patients (4.3%); type C, 4 patients (3.4%); type D, 23 patients (19.3%); and type E, 6 patients (5%) (Fig. I). The origin of the left main coronary artery was

Fig. 5. Reconstruction of pulmonary vascular tract. The Valsalva sinuses have been reconstructed with a large pericardial patch. Anastomosis to the pulmonary bifurcation is done with a running suture.

tunneled in the aortic wall, with a superior location of the ostium in two patients." The circumflex coronary artery was absent in three patients. The relationship of the great arteries was anteroposterior in patients with coronary arteries of types A, S, C, and D. The aorta was strictly anterior, slightly deviated to the right or, rarely, slightly deviated to the left (three patients). For the six patients with type E arteries, the great arteries were side-by-side, the aorta being located on the right. The diameter of the aorta and pulmonary artery were almost similar in patients with simple TGA, whereas pulmonary artery diameter was always greater than the aortic diameter in TGA with VSD. The associated 10 large VSDs were perimembranous in seven and infundibular in three patients; the associated three cases of tricuspid insufficiency were related to a rupture of the chordae tendineae after a Rashkind maneuver in one patient and to elongation of the chordae in two patients. Surgical technique. The surgical technique was derived from the techniques described by Jatene."" Yacoub,'? Lecompte," and their associates, with some modifications. The technique has not been modified for the last 95 patients. All patients except two were operated on with continuous hypothermic bypass at 20° C, bicaval cannulation, and no circulatory arrest. The atrial septal defect was closed either directly or with a Dacron patch. The ducts arteriosus was divided, the aortic end being ligated with a mattress suture and the pulmonary end being directly oversewn. After aortic cross-clamping, myocardial protection was ensured by cold crystalloid cardioplegia and intermittent topical cooling. The aorta was transected above the valvular commissures. The origin of the coronary trunks was dissected, and the coronary ostia with all the adjacent sinus of Valsalva was removed. The pulmonary trunk was transected distally. The Lecompte maneuver') (Fig. 2) was performed on all patients, even those with a side-by-side relationship of the great arteries. For the first 10 patients, the coronary arteries were implanted into the wall of the proximal neoaorta (Fig. 3); for the last 110 patients, the coronary arteries were anastomosed on the edge

Volume 96 Number 3

Switch operation for TGA

September 1988

357

%

100

+

90

% survival

... % campi free

80 70 60 50 40 30 20 10 0 0

108

102

78

45

31

19

6

12

18

24

30

36

months

duration of follow-up

Fig. 6. Survival and complication-free actuarial curves. of the proximal neoaorta (Fig. 3), as described by Yacoub and Radley Smith.'? The aortic anastomosis was then performed. The anterior wall of the distal aorta was incised vertically when pulmonary artery diameter was larger than aortic diameter. The proximal neopulmonary trunk was reconstructed in the first 10 patients with two circular patches and the last 110 patients with a large single posterior patch (Fig. 4). The material used for pulmonary reconstruction was tanned heterologous pericardium for the first 25 patients and autologous native pericardium for the last 95 patients. The pulmonary anastomosis was completed during rewarming (Fig. 5). Mean aortic cross-clamp time was 67 ± 13 (SO) minutes, ranging from 49 to 118 minutes. Mean cardiopulmonary bypass time was 166 ± 34 (SO) minutes, ranging from 119 to 280 minutes. The VSO was approached through the atrium in seven patients with a perimembranous VSO and through the proximal aorta or pulmonary artery in three patients with an infundibular VSO. The patient with coarctation of the aorta was treated at 7 days of age and arterial switch was performed 10 days later. The three patients with tricuspid insufficiency were treated by valvuloplasty. Postoperative care. In the early postoperative period transient LV dysfunction is common, even in the absence of operative problems and in the presence of preoperative favorable estimated LV pressure. Left atrial pressure was monitored routinely. LV failure was treated by arterial vasodilators and inotropic support. Electrical atrial stimulation was used when necessary to induce an atrial rhythm between 160 to 180 beats/min. The LV usually recovered in a few hours to 2 to 3 days. The patient was weaned from the ventilator when the contraction of the LV was normalized according to two-dimensional echocardiography. Delayed closure of the sternum, usually at 48 hours, was required in 28 patients (23%). The sternum was left open for a mean period of 48 hours after the operation. The wound was closed with a synthetic, waterproof patch sewn at the edge of the skin incision. No infection was observed with this technique. Statistics. A x2 evaluation was performed despite the small

Fig. 7. Reimplantation of high right coronary artery. In TGA, the origin of the coronary arteries is often higher than in normal hearts. The height of the emergence must be reproduced during relocation. Therefore it is sometimes necessary to reimplant the vessel above the aortopulmonary anastomosis.

number of patients in the subgroups. For actuarial graphs we applied the method of Kaplan and Meier."

Results

The hospital mortality rate was 8.3% (10 deaths). It was 40% (four deaths) among the first 10 patients but only 5.4% (six deaths) among the last 110 patients (p < 0.(01). There were no deaths in the group having TGA plus VSD. Among the perioperative deaths, six were related to coronary artery kinking (three of these patients had type C coronary artery distribution); one was related to LV wound; one to postoperative suprasysternic pulmonary hypertension, and the last two (in the early experience)

The Journal of Thoracic and Cardiovascular Surgery

3 5 8 Planche et al.

TYPE II

Fig. 8. Relocation of "single" coronary artery. Relocation in type 8 requires turning the coronary ostium upside down, which creates a risk of torsion. were related to diffuse subendocardial ischemia attributed to insufficient myocardial protection. No deaths were definitely related to inadequate LV function. Only one of the five patients with unfavorable LV geometry, operated on beyond the age of two weeks, died; he also had tricuspid insufficiency. On x' analysis, there was no statistically significant difference in mortality rate among the three groups on the basis of LV geometry (P> 0.1). X 2 Analysis of mortality rate according to coronary artery distribution also showed no difference among groups A, B, D, and E (p > 0.1) but a significant increase in mortality in group C (p < 0.001) when this group was compared with the four others together. Survival rate is shown in Fig. 6. All 108 survivors were reviewed and no patient was lost to follow-up. The length of follow-up ranged from 2 to 46 months, mean 16 ± 11.2 (SD) months (Fig. 6). The patients were regularly followed up by the referring pediatric cardiologists, and 75 (69%) were followed up by the Departmetn of Pediatric Cardiology of the Necker Hospital, Paris. Follow-up included sequential noninvasive evaluations: two-dimensional echocardiograms, Doppler studies, electrocardiograms, x-ray studies, Holter analyses (15 patients), and right heart catheterizations (32 patients) 10 to 18 months postoperatively. Two patients underwent reoperation for pulmonary stenosis 12 to 16 months postoperatively, with excellent results. These two patients had had a pulmonary reconstruction with two patches of tanned heterologous pericardium, which were found to be retracted at operation in both cases. In addition to the two patients with pulmonary stenosis who required a reoperation, five patients had moderate to mild supravalvular pulmonary stenosis with no impairment of right ventricular function; seven

TYPE II

Fig. 9. Coronary relocation in type II (type C of Yacoub). Relocation in type C requires disinsertion of the posterior commissure, which creates potential pulmonary insufficiency associated with a risk of torsion. patients have minute pulmonary valvular insufficiency that is well tolerated. Transient myocardial ischemia was observed during the first 2 months. In several patients with repolarization abnormalities on the electrocardiogram and dysfunction of the interventricular septum; by 2 months the patients had completely recovered. One patient had a localized myocardial infarction related to the right coronary artery 2 months after operation; he completely recovered in 3 weeks and was doing well 6 months later with a normal electrocardiogram and echocardiogram and apparently normal coronary arteries on angiographic study. There were no instances of dilation of the origin of the neoaorta or aortic regurgitation, except in two patients with trivial regurgitation. LV function was normal in all patients after 6 months according to two-dimensional echocardiograms and angiograms. All patients were in sinus rhythm, and the 24-hour recordings that were done were normal. One patient had a mild residual VSD. Neurologic problems occurred in two patients. One had a minor neurologic sequelae of postoperative neurologic complication, and the other had severe brain damage as a result of child abuse 10 months after the operation. The other 106 have normal psychomotor development. The survival curve of complication-free patients is shown in Fig. 6. All survivors (108 patients) were free of cardiac symptoms, were receiving no medications, had normal LV function with no cardiomegaly on x-ray studies and no LV dilatation, normal shortening fraction (> 30%), and no dyskinetic area on echocardiograms.

Volume 96 Number 3

Switch operation for TGA

September 1988

TYPE

A

TYPE B

r>.

TYPE

359

C

I~\

~

82 PTS - 5 DEATHS TYPE

5 PTS - 1 DEATH

D

23 PTS - 0 DEATH

TYPE

6

PTS -

4 PTS - 3 DEATHS

E

1

DEATH

Fig. 10. Hospital mortality according to the anatomy of the coronary arteries. Type C carries a high risk.

Although most of the patients had a soft systolic murmur over the pulmonary outflow tract, pulmonary artery complications were rare. Discussion The infants were referred by different teams of pediatric cardiologists. The results obtained with the first 50 neonates referred by the Necker Hospital in Paris have been published elsewhere," The 16% hospital mortality rate of this initial series was higher than that of the total surgical series presented here, which is 8.3%. This disparity is related to the high mortality observed during the early learning experience. Three main groups of difficulties were encountered in switch operations in neonates: management of extracorporeal circulation, surgical technique, and hemodynamic impairments related to adaptation of the LV to the systemic circulation. Extracorporeal circulation management. The neonatal swtich operation is a difficult procedure with many vascular anastomoses performed on delicate neonatal tissue that may be a source of postoperative bleeding. The accuracy required for this operation led us to prefer continuous bypass to circulatory arrest, which is usually considered a safer method in infants. The extracorporeal circuit has been improved and reduced to accommodate the smallest possible priming volume (approximately 450 to 500 ml). This volume is made up of a mixture of fresh-frozen plasma and erythrocyte concentrates less than 8 days old. IS The gas exchanger is of the capillary fiber type. The aorta was cannulated distally, and both venae cavae were cannulated through the right atrium. Only twice in the series did severe right atrial hypoplasia lead

us to use a single atrial cannula, deep hypothermia, and circulatory arrest. No complication related to continuous hypothermic bypass was observed, even when surgical difficulties led to prolonged (>4 hours) bypass. Similarly, severe hypotrophy or prematurity was never the source of bypass-related complications. Surgical technique. Surgical difficulties are principally related to relocation of the coronary arteries and to reconstruction of the pulmonary artery. Relocation of coronary arteries. Many anatomic studies have been published dealing with the anatomy of the coronary arteries in TGA,'6-'9 and several classifications have been proposed. The categories described by Yacoub are useful in understanding the main problems encountered during the switch operation. According to the method and the risk of coronary relocation, three main types can be defined. The first type (Fig. 3) is the most common form (type A of Yacoub). The two coronary arteries originate from Valsalva sinuses contiguous to the pulmonary artery and run transversely toward the atrioventricular groove. Since the aorta is positioned anteriorly, coronary relocation results in backward translocation of the two coronary arteries. This can be obtained with a minimal risk of deformation, provided the height of the coronary arteries above the ventricles is taken into account" (Fig. 7). The second type is characterized by coronary arteries originating and coursing between the aorta and the pulmonary artery. It corresponds to the so-called single coronary artery (types Band C of Yacoub). Coronary relocation requires turning the ostial patch upside down, and this creates a risk of torsion of the branches originating from the single coronary ostium (Fig. 8).

360

The Journal of Thoracic and Cardiovascular Surgery

Planche et al.

TYPE III

Fig. 11. Coronary relocation in type III. 1. In a vessel with a posterior course, the posterior translocation creates a risk of angulation by shorteningthe course of the artery. 2. In a vessel with an anterior course, there is a risk of elongation by excessive tension. In our experience, a special anatomic configuration has given rise to problems not correctly solved in infants (Figs. 9 and 10). It consists of the emergence of two parallel coronary arteries above, but in contact with, the posterior commissure (type C of Yacoub). We obtained a good result in only one case of this type. This uncommon situation is associated with frequent postoperative coronary artery kinking. The third type corresponds to coronary arteries with abnormal origin of one artery, which runs transversely either in front of or behind the base of the great arteries (types D and E of Yacoub). The posterior vessel is usually the circumflex artery. Its originates from the right posterior sinus, together with the right coronary artery, and crosses the posterior wall of the pulmonary artery before reaching the lateral face of the LV. The posterior translocation in shortening the course of the artery, creates a risk of angulation (Fig. II). The anterior vessel is usually the right coronary artery (type E). It frequently originates from the left posterior sinus, together with the interventricular artery. It crosses the anterior aspect of the aortic root on the way to the right atrioventricular groove. The posterior translocation, increasing the course of the artery, may result in excessive tension. Care must be taken to fully dissect the artery and to evenly distribute the deformation (Fig. 11). More rarely, surgical difficulties result from another anatomic variation such as the abnormally high origin of a coronary artery or its emergence in contact with a commissure. In most cases, the collateral branches of the coronary arteries arise at some distance from the origin and do not interfere with the relocation procedure. However, some collaterals arise near the coronary ostium and branch out very early on the epicardium. In such a case, the small branches anchor the coronary artery and

interfere with its mobilization. In some instances, the phenomenon is identified only by a deformation of the initial part of the artery, when cardiac activity returns, or when bypass is discontinued. When these secondary branches are small and localized, they can be divided, but when they are rather large, they must be dissected and mobilized at the same time as the main coronary artery. 18.21 Reconstruction of pulmonary tract. Reconstruction of the pulmonary tract involves two stages: reconstruction of the Valsalva sinuses, which have been destroyed by excision of the coronary ostia, and bridging of the defect created between the aortic orifice and the pulmonary bifurcation. We strongly prefer technical solutions compatible with growth and therefore avoid the prosthetic or modified biologic conduits used by some

authors." At the beginning of our experience, in the first group of 10 patients, we used small patches of tanned heterologous pericardium to close the holes punched in the neoaortic wall. The only two patients of our series with significant pulmonary stenosis came from this early group. In the last group of 110 patients, the aortic root was reconstructed by a large pericardiaI patch, with a notched inferior border tailored to fit with the posterior commissure (Figs. 4 and 5). Tanned pericardium was used first but was promptly replaced by fresh autologous pericardium, collected at the beginning of the operation. Direct anastomosis between the aortic root and the pulmonary bifurcation was accomplished through the "Lecompte maneuver." For some authors-'" a sideby-side disposition of the two great arteries is a contraindication to the Lecompte maneuver, but we have not met with such a limitation since we use a large pericardial patch. The Lecompte maneuver has also been thought to be responsible for stenosis of the pulmonary branches," but we have not encountered this type of complication. Nevertheless, the growth of the pulmonary tracts remains a matter of concern. The absence of foreign material may allow for growth; no significant pulmonary stenosis has been observed since we have used the native pericardium of the infant.

Hemodynamic problems and selection of patients. Apart from the surgical difficulties, the switch operation in simple TGA raises the major problem of the ability of the LV to sustain systemic pressures.v' Actually, only measurement of high LV pressure preoperatively can assure satisfactory LV function. However, most of the infants were operated on at some time after catheterization and there were no objective data to assess the functional ability of the LV. In fact, LV function was

Volume 96 Number 3 September 1988

assessed on the modification of the geometry of the LV observed on two-dimensional echocardiography, which is a rough estimation of the intraventricular pressure." A qualitative evaluation of the morphologic deformation of the LV allowed us to define three classes of LV function: satisfactory, acceptable, and unfavorable.' Nevertheless, an unfavorable morphologic appearance during the first days of life does not necessarly correspond to permanent LV dysfunction. As a matter of fact, during the early days of life, opening the ductus arteriosus with prostaglandin E, usually produces a satisfactory morphologic change in LV geometry." 26 Moreover, anatomic studies have demonstrated that the ventricular wall thickness and the total LV myocardial mass were kept close to normal for several weeks after birth." It is therefore possible that we may operate later in life in patients with unfavorable geometry. In this respect, three of the five patients with unfavorable LV geometry had very simple postoperative courses, even one infant who was 23 days old. However, as the diagnosis is made early in life in the great majority of patients, we do not recommend waiting too long before operating. According to our clinical experience, the optimal age for anatomic repair in simple TGA seems to be at the end of the first week of life. At this moment, extracardiac disorders, particularly neonatal infection, are ruled out or may have been treated; resistance in the pulmonary vascular bed is low in most of the patients; and LV function is unaltered because appropriate high pressure and flow have been maintained by the patent ductus arteriosus. However, the operation can be performed earlier if the clinical status of the infant is poor, if prostaglandin E, infusion is not tolerated, or if echocardiograms show an unfavorable LV anatomy. After 2 weeks of life, it seems reasonable to perform a switch operation in simple TGA only if LV function is strictly favorable on two-dimensional echocardiograms. In a few cases, anatomic disorders of the right ventricle such as tricuspid insufficiency or right ventricular hypoplasia lead to mandatory switch operation and then may modify the management of the patients. Concerning patients with TGA and VSD, our current policy is now to avoid pulmonary artery banding to have a clear operative field without adhesions and to limit the discrepancy in the diameters of the pulmonary artery and aorta. Our tendency is also to operate early, even if congestive heart failure is controlled by medication, to avoid hyperactivity in the pulmonary vascular bed that may lead to an uncontrolled pulmonary artery hypertension. We believe that 3 weeks is the best age for correction of TGA with VSD.

Switch operation for TGA

36 1

We are grateful to Dr. Craig Miller for reviewing the manuscript. REFERENCES I. Jatene AD, Fontes VF, Souza LCB, Neger F, Galantier M, Souza JEMR. Successful anatomic correction of transposition of the great vessels: A preliminary report. Arg Bras Cardiol 1975;28:46. 2. Turley K, Mavroudis C, Ebert PA. Repair of congenital cardiac lesions during the first week of life. Circulation 1982;66(Pt 2):11214. 3. Major WK Jr, Matsuda H, Subramanian S. Failure of the Jatene procedure in a patient with d-transposition and intact ventricular septum. Ann Thorac Surg 1976; 22:386. 4. Yacoub MH, Radley Smith R, McLaurin R. Two-stage operation for anatomical correction of transposition of the great arteries with intact ventricular septum. Lancet 1977;1:1275. 5. Castaneda AR, Norwood WI, Jonas RA, Colan SO, Sanders SP, Lang P. Transposition of the great arteries and intact ventricular septum: anatomical repair in the neonate. Ann Thorac Surg 1984;38:438. 6. Hougen TJ, Colan SO, Norwood WI, et al. Hemodynamic results of arterial switch operation for transposition of the great arteries, intact ventricular septum. Circulation 1984;70(Pt 2):1111. 7. Sidi 0, Planche C, Kachaner J, et al. Anatomic correction of simple transposition of the great arteries in 50 neonates. Circulation 1987;75:429. 8. Jatene AD, Fontes VF, Souza LCB, Neger F, Galantier M, Souza JEMR. Anatomic correction of transposition of the great vessels. J THORAC CARDIOVASC SURG 1977; 72:364. 9. Huhta JC, Edwards WD, Feldt RH, Puga FJ. Left ventricular wall thickness in complete transposition of the great arteries. J THORAC CARDIOVASC SURG 1982;84:97. 10. Yacoub MH, Radley Smith R. Anatomy of the coronary arteries in transposition of the great arteries and methods for their transfer in anatomical correction. Thorax 1978; 33:418. II. Gittenberger-de Groot AC, Sauer U, Quaegebeur J. Aortic intramural coronary artery in three hearts with transposition of the great arteries. J THORAC CARDIOVASC SURG 1986;91:566. 12. Jatene AD, Fontes VF, Souza LCB, Paulista PP, Neto AC, Souza JEMR. Anatomic correction of transposition of the great vessels. J THORAC CARDIOVASC SURG 1982; 83:20. 13. Lecompte Y, Neveux JY, Zannini L, Tu TV, Duboys Y, Jarreau MM. Reconstruction of the pulmonary outflow tract without prosthetic conduit. J THORAC CARDIOVASC SURG 1982;84:727. 14. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457. 15. Nicolas F, Bruniaux J, Planche C. Recent techniques of

3 6 2 Planche et al.

16.

17.

18. 19.

20.

21.

22.

23.

24.

25.

26.

cardiopulmonary bypass in infancy. In: Hagl S, Klovenkorn WP, Mayr N, Sebening F, eds. Proceedings ofthe symposium on thirty years of extracorporeal circulation. Munich: Deutsches Herzzentrum, 1986. Elliot LP, Amplatz K, Edwards JE. Coronary arterial patterns in transposition complexes. Am J of Cardiol 1966;17:362. Gittenberger-de Groot AC, Sauer U, OppenheimerDekker A, Quaegebeur J. Coronary arterial anatomy in transposition of the great arteries: a morphologic study. Pediatr Cardiol 1983, 4(suppl 1):15. Hvass U. Coronary arteries in d-transposition: a necropsy study of reimplantation. Br Heart J 1977;39:1234. Shaher RM, Puddu GC. Coronary arterial anatomy in complete transposition of the great vessels. Am J Cardiol 1966;17:355. Goor DA, Shem-Tov A, Neufeld HN. Impeded coronary flow in anatomic correction of transposition of the great arteries: prevention, detection, and management. J THORAC CARDIOVASC SURG 1982;83:747. Sievers HH, Lange PE, Heintzen PH, Bernhard A. Surgical implications of early branching of the left coronary artery in anatomic correction of transposition of the great arteries. Thorac Cardiovasc Surg 1985;33:198. Yacoub MH, Bernhard A, Radley Smith R, Lange PE, Sievers HH, Heintzen PH. Supravalvar pulmonary stenosis after anatomic correction of transposition of the great arteries: causes and prevention. Circulation 1982;66(Pt 2):1193. Quaegebeur JM, Rohmer J, Ottenkamp J, et al. The arterial switch operation: an eight-year experience. J THORAC CARDIOVASC SURG 1986;92:361. Kanter KR, Anderson RH, Lincoln C, Rigbby ML, Shinebourne EA. Anatomic correction for complete transposition and double outlet right ventricle. Ann Thorac Surg 1985;90:690. Van Doesburg NH, Biereman FZ, Williams RG. Left ventricular geometry in infants with D transposition of the great arteries and intact ventricular septum. Circulation 1983;68:733. Lang P, Freed MD, Bierman FZ, Norwood WI Jr, Nadas AS. Use of prostaglandin E 1 in infants with d-transposition of the great arteries and intact ventricular septum. Am J Cardiol 1979;44:76.

Discussion Dr. Hillel Laks (Los Angeles. Calif). The history of the arterial switch in this country has been somewhat tortuous since the initial description by Jatene in 1975. It soon became apparent that the procedure had a high mortality and was generally reserved for the complex lesions of TGA with vso. Yacoub then attempted to approach simple TGA with pulmonary artery banding, sometimes associated with a shunt to prepare the LV. Then the Lecompte maneuver of bringing the pulmonary arteries anterior to the aorta was described. The first attempt at neonatal arteries switch in this country began with Paul Ebert's small series of about six patients in 1982. There was a relatively high mortality, and because of the

The Journal of Thoracic and Cardiovascular Surgery

superb results that he was obtaining with the atrial switch, he discontinued the series. In 1984, Aldo Castaneda, in Boston, took up the challenge and demonstrated that the procedure could be performed in the neonate with acceptable results. Recently, the Congenital Heart Surgeons Society, which consists of 20 institutions, did a prospective series, entering patients with TGA under the age of 15 days who were undergoing either arterial or atrial switches. For the arterial switch in this group the mortality rate was 19%; for the Mustard procedure it was 0% (most of these patients came from the Sick Children Hospital in Toronto); and for the Senning operation the mortality rate was 15%. However, in looking at the overall mortality, including the patients who died while waiting for a Mustard or Senning procedure, one observes that, at 1 year, with an overall survival rate of 60%, there was no statistically significant difference between patients having an arterial switch (in whom there was a higher operative mortality but no deaths in patients awaiting the operation); however, although the operative mortality was lower, there were some deaths in patients awaiting the Senning and Mustard procedures. A major associated cardiac anomaly and low birth weight increased the risk of the arterial switch. Among the I I institutions doing arterial switches, there were four in which the mortality rate averaged 67%. In light of this, the results that Professor Planche reported, with an overall mortality of 8%, are to be highly commended. Dr. Claude Planche has made some technical contributions to the operation, most important of which has been the use of the single pericardial patch on the neopulmonary artery to reconstruct the site of removal of the coronary arteries. We began to use this technique and have found this extremely useful. We have now done 16 arterial switches with two deaths. We have found that if one does not use this posterior patch, there is excessive tension on the pulmonary artery, which pulls up on the anulus of the neoaorta, to which the coronary arteries are attached. This results in coronary artery distortion owing to increased tension. Use of the posterior pericardial patch avoids this problem. I have some questions for Professor Planche. First, in view of the very high mortality for the group with posterior origin of the two coronary arteries very close to the commissure, if one is aware of that anatomy preoperatively, should one avoid an arterial switch? Dr. Planche. Yes, such findings should lead to an atrial procedure instead of an arterial switch. Dr. Laks. Do you, therefore, insist on aortograms preoperatively to look at the coronary anatomy, or do you now operate on some patients only with echocardiography? Dr. Planche. In most patients no aortogram is performed. Dr. Laks. In other words, you do not have to know the coronary anatomy. You take a chance on what you will find at operation in some of the patients. Dr. Planche. Yes. Dr. Laks. What is your approach to TGA with VSD? Do you do an arterial switch and VSD closure, do you sometimes observe these patients until an older age, or do you band the pulmonary artery in some of them? Dr. Planche. In earlier years, our approach for TGA with VSD was a two-stage procedure with preliminary banding followed by an arterial switch operation several months later. The dilatation of the pulmonary artery, which is always seen in TGA with VSD, was found to be worsened by preliminary

Volume 96 Number 3 September 1988

banding, and this carried a risk of aortic insufficiency after a switch procedure. For that reason, we decided to perform a one-stage operation, and the optimal age seemed to be at the end of the first month of life. A perimembranous VSD is closed through the right atrium; an infundibular VSD is closed through the aorta or the pulmonary artery. Dr. Laks. In patients who present rather late, for example, at 2 weeks of age, do you use any other parameters to assess the ability of the LV to take on the systemic resistance, other than echocardiography? Dr. Planche. We always perform anatomic correction

Switch operation for TGA

363

within the first 2 weeks of life, depending on the status of the LV as assessed by echocardiography. After 2 week sof life, it seems in our opinion reasonable to perform a switch operation only if LV function and anatomy are favorable. Dr. Laks, In your series, was there a difference in LV performance postoperatively in those patients who had good LV function by echocardiogram and those who had acceptable or poor function by echocardiogram? Dr. Planche. Not significantly, because only a few patients operated on had poor LV function.