Current Expectations for Newborns Undergoing the Arterial Switch Operation

Abstract

Background:

The arterial switch operation (ASO) represents a remarkable success story in the surgical treatment of cyanotic congenital heart disease. This study is designed to assess recent outcomes after the ASO in babies presenting with transposition of the great arteries (TGA) and Taussig-Bing anomaly (TBA).

Methods:

One hundred twenty-five consecutive neonatal and infant ASOs were performed by 2 surgeons at Texas Children's Hospital between July 1, 1995 and October 1, 2003. Patients with TGA and TBA were offered ASO irrespective of patient size and associated cardiac malformations. Primary cardiac diagnoses included TGA with intact ventricular septum (TGA/IVS, n = 79, 63%), TGA with ventricular septal defect (TGA/VSD, n = 37, 30%), and Taussig Bing Anomaly (TBA, n = 9, 7%).

Results:

With complete follow-up, we observed a 30-day mortality rate of 1.6% (n = 2) with 2 late deaths (1.6%), for an overall actuarial survival rate of 96.3% at 7 years. Although there was a significant incidence of complex coronary ostial origin and branching including single coronary (n = 8, 6.4%) and intramural coronary artery (n = 8, 6.4%), this was not associated with increased operative risk. All patients are fully saturated and NYHA functional class I at latest clinic visit (0.3 to 88.4 months postoperatively). There have been no late coronary events. Of 121 survivors, 7 patients (5.8%) have required cardiovascular reoperation at an average of 15.3 ± 11.7 months postoperatively (range, 3.6 to 30.6 months) for an actuarial freedom from reoperation of 90% at 7 years.

Conclusions:

Using current methodologies, the ASO can be performed safely and with a low incidence of need for reoperation on intermediate follow-up. Recent experience indicates operative survival rates approaching 100%.

The arterial switch operation is now the procedure of choice for babies presenting with transposition of the great arteries. Recent experience documents that this operation can be performed safely for all coronary artery anatomic variations with very low operative mortality. This procedure should be offered to all babies presenting with this otherwise fatal condition. Expectations for operative survival are now approaching 100%.

The first successful arterial switch operation (ASO) was reported by Jatene et al in 1975.¹ Over the ensuing 3 decades, this procedure has come to be accepted as the preferred surgical treatment of the majority of children born with transposition of the great arteries (TGA) and forms of double outlet ventricle with subpulmonary ventricular septal defect, also known as the Taussig-Bing anomaly (TBA). Numerous publications have examined the role of the ASO in these conditions, and reported surgical outcomes have steadily improved with refinements in diagnosis, surgical technique, and perioperative management. A literature search lists over 180 articles published in the medical literature during the last 5 years focused on the subject of TGA and ASO.

The ASO benefits the patient by realigning the great vessel connections to provide ventriculoarterial concordance, an “anatomic” repair. This is in contrast to the atrial switch procedure (Mustard or Senning), which, although technically less challenging and providing “physiologic” correction, leaves the patient with ventriculoarterial discordance. The long-term detrimental effects of having a systemic right ventricle, particularly in patients with TGA with ventricular septal defect (VSD) are now well recognized and include dysrhythmia, tricuspid insufficiency and right ventricular failure.^2,3,4 In this regard, the ASO is clearly a superior option, affording the patient the benefit of a systemic left ventricle with normal circulatory pathways.

Despite the enthusiastic adoption of the ASO, controversy still exists about the short-term and long-term benefit of this operation. There is a growing body of data demonstrating late attrition due to silent coronary ischemia, semilunar valve dysfunction and dysrhythmia, leading many investigators to believe the ASO must be fairly considered a palliative procedure.⁵ Postoperative supravalvar and valvar pulmonary stenosis have become less prevalent over the last decade, but remain important short and long-term sources of hemodynamic morbidity. The operation remains technically challenging, with disparity in operative survival rates being reported in recent series.^6,7 In particular, certain anatomic substrates have been associated with increased operative risk, including patients with complex coronary origin and branching patterns, patients with associated aortic arch anomalies and left ventricular outflow tract (LVOT) obstruction.^7–10 Patients undergoing the ASO have also been observed to have heightened risk of abnormal neurologic development, believed to be at least partially associated with the widespread use of deep hypothermic circulatory arrest (DHCA), and low flow cardiopulmonary bypass (CPB) to facilitate the operation.^11,12 Furthermore, the ASO is difficult to teach and learn, and surgical outcomes have been believed related to a significant learning curve.

In 1995, changes in administrative and service organization at our hospital afforded us the opportunity to standardize the management strategy for neonates and infants presenting with TGA and TBA at our institution. Our protocols were developed and based on previously reported experience in children with TGA, with particular emphasis on well-described technical modifications of the ASO that allow application of the procedure to all anatomic coronary artery variations. This report focuses on the short and midterm results of this strategy at a single institution.

MATERIALS AND METHODS

One hundred twenty-five consecutive neonatal and infant ASOs were performed by 2 surgeons at Texas Children's Hospital between July 1, 1995 and October 1, 2003. During the study period, patients with TGA and TBA were offered the ASO irrespective of patient size and associated cardiac and extracardiac malformations. Only patients with known organic left ventricular outflow tract obstruction (LVOTO) not believed amenable to surgical resection were excluded from consideration for the ASO. Patients undergoing palliative arterial switch and switch conversion operations were excluded from analysis.

Primary cardiac diagnosis was confirmed in all cases by postnatal transthoracic 2D echocardiogram. In patients with prenatal diagnosis, an infusion of prostaglandin E1 (PGE1) was initiated in the delivery suite immediately after birth. In other patients, the PGE1 infusion was typically started when cyanosis was recognized or after the diagnostic echocardiogram. Preoperative balloon atrial septostomy (BAS) was performed before operative intervention in 109 cases (87%) to optimize the opportunity for atrial level mixing. A BAS was not performed in patients with an obviously nonrestrictive atrial septal defect. Cardiac catheterization was infrequently used to confirm intracardiac or aortic arch anatomy. Ascending aortography to determine coronary ostial origin and branching pattern was not performed. Four patients (3.2%) had undergone palliative surgery before ASO, including pulmonary artery banding (n = 2) for rapid left ventricular retraining after initial presentation at > 2 months of life, pulmonary artery banding and coarctation repair (n = 1), and placement of a right modified Blalock-Taussig shunt (n = 1).

In patients without ductal dependent descending aortic perfusion, aggressive attempts were made to discontinue PGE1 and ventilator support before surgery. Our management standard is to perform the ASO within 1 to 2 weeks of life for patients with TGA and intact ventricular septum (TGA/IVS) or TGA with a small ventricular septal defect (VSD). For patients with TGA and nonrestrictive VSD or TBA without associated aortic arch obstruction, the ASO is performed within the first 6 weeks of life, often after some intervening time period to allow for the normal fall in neonatal pulmonary vascular resistance. Patients with arch hypoplasia or interruption with ductal dependent blood flow to the lower body typically undergo repair in the first week of life. Patient size was not used as a criterion for operability, and the management strategy as outlined was not altered for small for gestational age (SGA) or premature babies. All patients underwent complete single-stage repair of all associated cardiac defects, including aortic arch obstruction at time of ASO.

Surgical Technique

The ASO was performed via median sternotomy utilizing either single atrial or bicaval cannulation. CPB was established and maintained at flow rates of at least 150 mL/kg per min. DHCA was used sparingly and avoided if possible. Low-flow CPB was not used except in cases where regional low-flow cerebral perfusion (RLFP) was used to avoid DHCA during aortic arch reconstruction (n = 3). In the latter part of the series, we began using routine noninvasive cerebral monitoring in the form of near infrared cerebral cortical spectroscopy (NIRS, INVOS 5100; Somantics Corporation, Troy, MI) and middle cerebral artery transcranial echo Doppler (TCD, EME Companion; Nicolet Biomedical Inc., Madison, WI). NIRS cortical oxygen saturation and middle cerebral artery TCD blood flow velocities were used to adjust regional cerebral blood flow rates during periods of RLFP to avoid DHCA, as we have previously reported.¹³ Myocardial protection was provided by intermittent antegrade modified crystalloid cardioplegia and cold topical cardiac irrigation. Transesophageal echocardiography was routinely used for intraoperative confirmation of anatomy and immediate postoperative assessment of the repair and cardiac function.

As noted previously, the coronary branching pattern was not evaluated before surgery, and no patient was denied the ASO on the basis of the intraoperative finding of complex branching or ostial origin. This strategy was employed on the basis of our opinion that well-described technical modifications of coronary transfer allow safe and predictable neoaortic root reconstruction for all coronary branching patterns. Thus, our commitment has been to offer all patients with TGA or TBA the presumed long-term benefits of the ASO. With regard to the critical issue of coronary ostial transfer, we believe several technical modifications allow the surgeon to perform the ASO with minimized risk of ostial malpositioning and proximal arterial torsion. At the time of initial exploration, the orientation and branching pattern of the proximal main coronary arteries is noted. It is usually readily apparent when an anomalous ostium or proximal intramural segment is present by external inspection of the character of the proximal course of the branch coronary artery. After inspection of the coronary anatomy and before cannulation for cardiopulmonary bypass, small marking stitches are placed on the main pulmonary artery to serve as guides for subsequent coronary ostial positioning.

After cardioplegic arrest, the ascending aorta is transected several millimeters cephalad to the sinotubular junction. The coronary ostia are then visualized and inspected. It is critical that the surgeon determine not only the location of the ostia, but also the proximal course of the branches. In general, we do not favor probing the coronary arteries in babies; however, this may be necessary in patients with abnormal ostial positioning, particularly in cases in which a segment of the coronary artery travels within the arterial wall (intramural course) or is in a juxtacommissural location. In such instances, it is possible for the proximal coronary artery to be inadvertently transected during ostial mobilization, as has been described by others.¹⁴ We favor a generous coronary ostial button, including the majority of the corresponding aortic sinus. In contrast, we prefer to minimize the amount of mobilization of the proximal coronary branch arteries and do not transect or sacrifice any of the proximal branches during mobilization.

We believe the most important technical modification of the ASO that allows safe and predictable coronary ostial transfer is the routine use of trapdoor flap incisions in the main pulmonary artery (neoaorta) as has been well described by Mee and others.¹⁵ The trapdoor incision allows functional lengthening of the ostium while minimizing axial torsion. In the majority of cases, we create the trapdoor incision in the location of the previously placed marking suture after transection of the main pulmonary artery just proximal to the bifurcation. In more difficult branching patterns, we occasionally reestablish ascending aortic continuity before making the trapdoor incisions, as has been previously described by.^6,16 This technique may be particularly useful in cases of a single ostium with looping of major coronary branches in front of or behind the great vessels.

In the majority of cases, the ascending aorta is brought posterior to the pulmonary artery bifurcation using the maneuver of LeCompte.¹⁷ This is not done when the main pulmonary artery is anatomically anterior (ventral) to the ascending aorta, as may be occasionally seen in patients with TBA. The defects in the aortic sinuses (neopulmonary artery) are reconstructed with liberal, individual patches of fresh autologous pericardium to minimize the potential for postoperative supravalvar neopulmonary stenosis. We use left atrial catheters for postoperative monitoring in all patients to avoid left ventricular distension and to be alerted to subtle coronary malperfusion. We believe a rising left atrial pressure may be the most sensitive indicator of myocardial ischemia, even in cases in which the electrocardiogram is normal. We have not routinely used pulmonary artery catheters, but we do prophylax against postoperative pulmonary vasospasm by moderate hyperventilation and deep sedation with pharmacologic paralysis as needed. Delayed sternal closure is rarely necessary and was not found to be associated with better outcome when performed on an elective basis.¹⁸ We have not used postoperative extracorporeal membrane oxygenation (ECMO) for support of patients after the ASO.

Postdischarge Care

Patients are seen for follow-up in the outpatient clinic within 2 weeks of discharge. Routine postoperative transthoracic echocardiography is performed within 1 month of surgery or as needed on the basis of clinical progress. Subsequent follow-up is either at the Texas Children's Heart Center for local and regional patients or at the discretion of the cardiologist for more distant referrals. We have not performed routine follow-up coronary angiography and exercise stress testing as has been advocated by some investigators.¹⁹

Data Collection and Statistical Analysis

We conducted a retrospective review of patient data from the congenital heart surgery and cardiology databases. Operative and hospital discharge data were available in all patients. The cardiology clinic database was used for the retrospective collection of postdischarge information, and the last known contact was used for final determination of status and the generation of Kaplan-Meier curves. In the case of reoperation, operative notes were obtained and reviewed. All statistical analyses were performed using the SAS statistical package (SAS Institute Inc, North Carolina) and all data are reported as mean ± standard deviation unless otherwise specified. Fisher exact test, χ² analysis, and the Student t test procedure were used for univariate analysis of factors associated with mortality, while simple linear regression analysis was used to test factors associated with hospital and ICU length of stay (LOS) and ventilator and inotrope dependence. A P value < 0.05 was required for statistical significance in all cases.

RESULTS

Demographics

There were 78 males (62%) and 47 females. Primary cardiac diagnosis included TGA/IVS (n = 79, 63%), TGA/VSD (n = 37, 30%), and TBA (n = 9, 7%). Median age at ASO was 8 days (mean age, 32.3 ± 79.6 days; range, 3 to 496 days), and median weight was 3500 g (mean weight, 3553.8 ± 1026.2 g; range, 1500 to 9700 g). Extracardiac anomalies were found in 9 patients (7.2%), including situs inversus totalis (n = 2), malrotation (n = 2), polydactaly (n = 1), Turner's syndrome (n = 1), tracheomalacia (n = 1), Grade II interventricular hemorrhage (n = 1), and 1 child with cleft plate, single umbilical artery, a sacral simple, and periauricular tags. Associated cardiac anomalies were found in 28 patients (22.4%), including coarctation with associated aortic arch hypoplasia (n = 8), right ventricular outflow tract (RVOT) obstruction (n = 6), and interrupted aortic arch (n = 2; Table 1).

TABLE 1. Associated Cardiac Anomalies

Operative Data

Mean CPB time was 229.5 ± 56.6 minutes (range, 144 to 424 minutes) and mean aortic cross-clamp time was 122.1 ± 39.6 minutes (range, 62 to 291 minutes). DHCA was used to some degree in 84 patients (67%) for a median time of 10 minutes (mean, 17.8 ± 17.9 minutes; range, 2 to 107 minutes). Along with closure of all atrial and ventricular septal defects, 33 concomitant cardiovascular procedures were performed including 10 aortic arch repairs for coarctation with aortic arch hypoplasia or interrupted aortic arch (Table 2). Coronary artery branching patterns and ostial origins as confirmed by intraoperative observation are listed in Table 3. Although we have preferred the Yacoub coronary classification system (Fig. 1) for its simplicity, cases that do not precisely fit into one of the described categories were further specified.²⁰ The most frequent coronary branching patterns were Yacoub A (66%) and D (22%). A single coronary artery system was discovered in 8 patients (6.4%), including Yacoub B (n = 5) and a single ostium from the anterior or anterior-leftward aorta (n = 3). An intramural coronary artery course was identified in 8 patients (6.4%), including 4 intramural left and 4 intramural right coronary arteries. Utilizing the alternate coronary classification system of Vouhe and associates,⁵ there were 82 low-risk, 37 intermediate, and 6 high-risk coronary patterns. The Imai modification of the ASO was used in the patient with Yacoub type B branching and an intramural RCA.¹⁵ A right internal mammary to RCA bypass graft was necessary in 1 patient with a single ostium originating from the left anterior aortic sinus when malperfusion of the anterior right ventricle became evident at the conclusion of ASO.

TABLE 2. Concomitant Procedures

TABLE 3. Coronary Artery Anatomy as Confirmed at Operation in 125 Consecutive Arterial Switch Operations

FIGURE 1. The Yacoub coronary classification system for transposition of the great arteries. Type A, circumflex from left coronary artery (LCA); type B, single ostium from a posterior facing sinus; type C, juxtacommissural ostia from posterior facing sinuses; type D, circumflex from right coronary artery (RCA); type E, circumflex from right sinus and LCA and RCA from the leftward facing sinus. Reprinted from Stark J, deLaval M, eds. Surgery for Congenital Heart Defects. Philadelphia, PA: WB Saunders; 1994. Copyright 1994 Elsevier Incorporated. Used with permission.¹⁵

Mortality

There were 2 early deaths (30-day mortality, 1.6%), both in patients with a primary cardiac diagnosis of TGA/VSD. The first was a 27-day-old, 3500-g male infant who underwent ASO, ASD, and VSD closure and aortic arch advancement for coarctation with hypoplastic aortic arch. This patient developed severe pulmonary hypertension and resulting RV failure requiring continuous inhaled nitric oxide. Left lower extremity ischemia subsequently developed as the result of a femoral arterial line. Despite removal of the line and femoral embolectomy and fasciotomy, hemodynamic instability progressed. Staphylococcal aureus sepsis was documented; despite appropriate antimicrobial therapy, the patient expired on postoperative day 10. The second case was a 15-day-old, 2500-g infant who suffered a complication related to the preoperative BAS. This was the only significant complication of BAS in our series. At the time of ASO, a pulmonary vein laceration was encountered and repaired. The patient underwent a difficult ASO but developed an intrapulmonary hemorrhage, suffered cardiac arrest approximately 12 hours after surgery, and could not be resuscitated. Univariate analysis of possible factors associated with early mortality revealed that associated cardiac anomalies (P = 0.0488), CPB time (P = 0.0015), and aortic cross-clamp time (P = 0.0013) were associated with early death (Table 4).

TABLE 4. Univariate Analysis of Early Mortality

There were 2 late deaths (1.6%) at 2.8 and 2.5 months after ASO. The first occurred in a former 32-week, 9-day-old, 2100-g female infant with bronchopulmonary dysplasia who underwent ASO, aortic arch advancement, and RVOT resection for TBA, subaortic stenosis, coarctation with aortic arch hypoplasia and single coronary ostium. After an initial difficult perioperative course, the patient recovered. The patient subsequently developed evidence of RV hypertension prompting cardiac catheterization that revealed RVOT obstruction with systemic RV pressures. Reoperation was performed, which included patch augmentation of the main and branch pulmonary arteries and RVOT resection. Unfortunately, the patient's underlying lung disease precluded successful extubation. Ultimately the patient developed sepsis and died 2.8 months after ASO. The second death occurred in a former 36-week, 8-day-old, 2800-g male infant with a primary cardiac diagnosis of TGA/IVS. He was noted to have hyaline membrane disease prior to ASO. The patient recovered from the ASO after a complicated perioperative course including early mediastinal reexploration for hemorrhage. The patient developed a late complication of superior vena caval thrombosis, presumably due to a central venous catheter. This complication progressed to severe SVC syndrome, chylous effusions, and malnutrition. Despite attempted surgical thrombectomy, the patient ultimately died 2.5 months after the ASO. Univariate analysis revealed that gestational age (P = 0.0008) and birth weight (P = 0.0238) were variables associated with mortality after 30 days (Table 5). Overall survival was determined by the Kaplan-Meier method and revealed an actuarial survival of 96.3% at 7 years postoperatively (Fig. 2).

TABLE 5. Univariate Analysis of Late Mortality

FIGURE 2. Actuarial survival calculated by the Kaplan-Meier method. Survival was 96.3% at 3, 5, and 7 years. Data are shown with 95% confidence intervals.

Postoperative Data

Median duration of mechanical ventilation was 3 days (range, 1 to 12 days), and median duration of inotropic support was 4 days (range, 2 to 20 days). Median ICU LOS was 5 days (range, 2 to 38 days), and median hospital LOS was 9 days (range, 3 to 42 days). Linear regression analysis was performed to determine possible associations of anatomic and physiologic parameters with days of ventilator and inotrope dependence and ICU and hospital LOS (Table 6). Of the factors examined, lower weight at operation, lower birth weight, and the presence of extracardiac anomalies were all associated with longer hospital and ICU LOS. Single coronary anatomy was associated with longer ventilator (P = 0.0169) and inotrope dependence (P = 0.0707), although the latter did not reach statistical significance. Although CPB time (P < 0.0001) and aortic cross-clamp time (P = 0.0015) were both associated with longer postoperative ventilator time, only CPB time was associated with longer inotrope dependence (P < 0.0011). Both the presence of associated cardiac defects and the need for preoperative intubation were highly associated with hospital and ICU LOS, longer ventilator dependence, and longer inotrope dependence.

TABLE 6. Linear Regression Analysis for Factors Associated With Postoperative Resource Utilization

Follow-up

All of the 121 survivors have been seen at the Texas Children's Hospital cardiology clinic or an outside clinic up to 7.4 years postoperatively. The most recent follow-up occurred a mean of 28.8 ± 24.2 months (range 0.3–88.4 months) postoperatively and all patients are fully saturated, NYHA class I with no symptoms referable to the cardiovascular system. Seven of the survivors (5.8%) have required cardiovascular reoperation following ASO at an average of 15.3 ± 11.7 months postoperatively (range 3.6–30.6 months). Supravalvar pulmonary stenosis was identified and treated with patch augmentation in 4 patients while concomitant pulmonary valvotomy (n=3) and RVOT resection (n = 1) were performed. One patient with TBA underwent repair of recurrent aortic coarctation via thoracotomy 3.6 months following ASO and then underwent branch pulmonary arterioplasty and RVOT resection 5 months later. The patient undergoing homograft neoaortic root replacement at ASO required repeat homograft root replacement 30.2 months postoperatively. One additional patient having undergone LVOT resection and homograft aortic root replacement at ASO underwent repeat LVOT resection and aortic root replacement 26 months postoperatively. Actuarial freedom from readmission for cardiovascular reoperation was calculated by the Kaplan-Meier method and was 90% at 7 years (Fig. 3).

FIGURE 3. Seven (5.8%) of the surviving 121 patients have required reoperation. Freedom from readmission for cardiovascular reoperation calculated by the Kaplan-Meier method was 90% at 7 years. Data are shown with 95% confidence intervals.

DISCUSSION

Our study confirms previously published series indicating that the ASO can be performed safely in the majority of patients with TGA and TBA. Our overall survival rate of 96% is gratifying and compares favorably with other large series.^6,9,21 We have observed no mortality in the past 3 years, consistent with the notion that there is some learning curve in the accurate performance of this procedure. Of the 4 deaths in the series, none were directly attributable to coronary malpositioning. Of concern is the fact the all of the deaths were in some way attributable to complications related to other aspects of the complex process required in the care of these patients. Three of the deaths were directly related to intravascular catheter complications. In 1 patient, a severe pulmonary vein laceration at BAS progressed to intrapulmonary hemorrhage. As noted previously, this was the only significant complication related to a BAS in our series, and we continue to believe that a BAS is an important component of the preoperative management strategy. The presence of an unrestrictive ASD often allows patients to be weaned from PGE1, extubated, and fed orally before surgery. Although we did not find a correlation between the need of preoperative intravenous PGE1 and outcome, we did observe increased ICU and hospital LOS and longer ventilator and inotrope dependence in patients requiring mechanical ventilation before surgery. We remain convinced that in suitable patients, it is preferable to wean PGE1 and ventilator support before ASO.

In 1 premature neonate, a percutaneous femoral arterial line caused femoral artery thrombosis, necessitating thrombectomy and fasciotomy and ultimately resulting in sepsis and death. In the other patient, a late superior vena caval syndrome developed as the consequence of a central venous line and progressive chylous effusions refractory to medical and surgical therapy ultimately caused profound malnutrition, sepsis, and death. Based on these events, we have altered our line placement protocols. We do not place superior venal caval lines in children weighing less than 4 kg. We make every effort to avoid femoral arterial lines, and when they are necessary, we often will treat the patient with a heparin infusion until the arterial line is discontinued.

Several investigators have noted that complex coronary ostial origin and branching pattern are associated with increased risk for the ASO.^8–10 Although we did observe a longer duration of mechanical ventilation in patients with single coronary artery, we did not observe an increased risk of mortality. The surgical techniques for safe transfer of all branching patterns appear to be established and reproducible. We believe that determining branching pattern prior to operation is unnecessary and that coronary anatomy should not be used as a criterion for suitability for the ASO. Perhaps a more important question relates to the long-term fate of the translocated coronary ostium, particularly in patients with complex branching or intramural segments. One must presume that such coronaries are at heightened risk of subsequent occlusion. We have not encountered late ischemic coronary events in our patients thus far, but our follow-up is relatively short. Bonhoeffer and colleagues have documented a worrisome incidence of late coronary occlusion among survivors of the ASO, a factor associated with increased risk of sudden cardiac death.²² Important questions remain unanswered regarding mode and timing of appropriate surveillance for this important potential late complication. Our incidence of complex coronary branching and ostial origin is relatively high in comparison with some series.⁷ This may be related to our position as a regional children's cardiac center in that there may be some potential for selection bias from outside referring doctors, although this was not apparent from our data analysis.

We remain hopeful that our perfusion methodology will be of long-term neurologic benefit to our patients. On the basis of the troubling data from Boston Children's Hospital and other centers regarding abnormal neurologic development in children surviving the ASO who were subjected to low-flow CPB during surgery, we make every effort to avoid this technique.^11,12 Our approach has been to use full-flow CPB, even at low temperatures. We have attempted to minimize DHCA, although this technique is still required in many cases. We are enthusiastic about the possibility that regional cerebral perfusion during aortic arch repair may offer superior neurologic protection, but there are currently no data to support this contention.

In summary, the ASO appears to be well established as the treatment of choice for TGA and TBA. Although technically challenging, recent experience confirms that the procedure can be performed with high expectations for excellent survival rates, even in patients with complex coronary artery anatomy and associated conditions. Important challenges remain, not only in terms of surgical technique but also with regard to long-term surveillance and need for late reintervention.

Discussions

Dr. Irving L. Kron (Charlottesville, Virginia): I appreciate the opportunity to discuss this excellent series by a new member and fellow fly fisherman, Dr. Fraser. The arterial switch operation has become the standard therapy for transposition. Results of this operation have become the standard regarding the quality of your program. Dr. Fraser's results are truly excellent. We have been fortunate with this operation as well with, fortunately, no hospital mortality in the last 8 years. I want to emphasize 2 points in the manuscript.

One is the marking of the coronary arteries prior to going on bypass, because that tells you exactly where the transfer should occur. We differ only slightly in that we do probe the coronary arteries. I had the unfortunate experience of transecting an intramural vessel, which was fortunately able to be repaired with an internal mammary artery bypass.

I have just 2 questions for the authors in this series. Is there anyone with transposition that they wouldn't offer a switch to at the time of diagnosis? Certainly, coronary anatomy has not been an issue, but are there issues of unbalanced ventricles?

Secondly, we ourselves have been very concerned about what happens to the neoaortic root. There is a situation where children with transposition and VSD that the neoaorta is much larger than the aorta you switched. We had thought at one time these would shrink down in size, but at late angiography we have noted in our group that these roots stay big. We are concerned will these be late aneurysms.

Dr. Fred A. Crawford, Jr. (Charleston, South Carolina): I also enjoyed this very nice paper by our new member, Dr. Fraser. I think that Dr. Kron covered most of the questions that I had and, in fact, took some of the ones that I planned to ask. But I think it is really worth emphasizing to this audience, many of whom are not cardiac surgeons, that these are truly outstanding results in a series of patients and that years ago we were unable to achieve anything like these excellent results.

Dr. Erle H. Austin, III (Louisville, Kentucky): I would like to congratulate Dr. Fraser and his colleagues in Houston for an excellent presentation of a set of enviable results for the surgical treatment of transposition of the great arteries, a condition, as he indicated, that not too long ago was fatal for essentially all infants unfortunate enough to be born with this lesion.

Dr. Fraser has indicated that the surgical approach to transposition has evolved over the past 3 decades from redirection of the venous inputs into the heart to redirection of the arterial outputs of the heart. This newer technique, of course, requires the surgical repositioning of the coronary arteries from one great vessel to the other, a key maneuver that if performed incorrectly may lead to myocardial ischemia, infarction, and death. Furthermore, especially in newborns with transposition and intact ventricular septum, the procedure is performed in the first week or 2 of life when the infant's heart is not much larger than the surgeon's thumb.

A significant issue in the application of the arterial switch is dealing with uncommon and potentially challenging coronary patterns such as single and/or intramural coronaries. In most reported series, these coronary patterns are considered significant risk factors for mortality and, in some cases, have been an indication to abandon the arterial switch in favor of an atrial switch. This has not been the approach of Dr. Fraser, as he has indicated, and he is to be commended on an exceptionally low mortality in a relatively large experience with a significant incidence of complex coronary patterns.

As in Houston, in Louisville we no longer stress a preoperative determination of coronary pattern and have applied the arterial switch to all D-transpositions with and without ventricular septal defects unless there is severe left ventricular outflow obstruction associated with a large ventricular septal defect. I do agree with Dr. Fraser's statement that the arterial switch can be performed safely for all coronary patterns, but I do think that complex coronary patterns do represent a risk factor that requires a surgeon of his experience, knowledge, and skill to neutralize. In his series the application of several coronary implantation modifications as well as one right internal mammary artery to right coronary bypass attests to the requirement of a significant surgical repertoire to achieve a successful arterial switch in all patients.

I have 2 questions for Dr. Fraser. First: What has been your approach to the infant with transposition and an intact ventricular septum that has been referred to you relatively late, that being at several weeks or even months of life? Are you confident enough to proceed directly with arterial switch in any of these patients, or do you have criteria for which you would consider retraining the left ventricle before performing the switch?

My second question: How have you dealt with the occasional transposition that exhibits severe preoperative hypoxia despite prostaglandin, a patent ductus arteriosus, a wide open atrial septum, and mechanical ventilation on 100% oxygen? Do you proceed with an emergent switch, or do you try to support them for several days with nitric oxide or ECMO before going to surgery?

Dr. Dale W. Oller (Raleigh, North Carolina): Are there dysrhythmias that can be caused by the septal balloons? And other than stenosis of the great vessels, do you anticipate any further need for surgeries in this large group of patients in the future?

Dr. Max R. Langham, Jr. (Gainesville, Florida): I enjoyed the paper and wanted to ask if there is any neurologic morbidity in your population? That is something we certainly see with ECMO and other forms of bypass.

Dr. Charles D. Fraser, Jr. (Houston, Texas): I would like to thank all of the discussants for their very nice comments, and maybe I will work backwards in answering the various questions.

The issue of neurologic morbidity in patients with congenital heart disease is still one of great concern. We have not had a significant incidence of gross neurologic morbidity in our patients. As a matter of fact, of the 121 survivors, only 2 patients had perioperative seizure activity that was documented after surgery. Nonetheless, the necessity of multiple interventions before surgery, the opportunity for atrial and ventricular level mixing, the use of hypothermic circulatory rest, all expose the patient to the long-term concern of neurologic maldevelopment. I think it is going to be another generation of surgery before we realize whether or not we are making an impact on that important concern.

With regard to reasons that patients might need to be reoperated after the arterial switch operation, certainly reconstruction of great vessels is a concern. We feel that the issue of supravalvar pulmonic stenosis has been largely alleviated by the liberal use of autologous pericardium. The long-term concern that most of us have is what is going to happen to the translocated coronary arteries? Many times we have to place our suture lines very close to the actual ostium, and it is certainly my opinion that this will expose the patient to long-term risk of ostial stenosis or occlusion.

Some series, particularly large series from Europe, have documented a late and concerning incidence of ostial occlusion. Dr. Austin asked about what we would do for a child that was referred to our institution late with transposition intact septum. What he has alluded to is that in children born with transposition and intact ventricular septum, soon after birth the left ventricle starts to involute and within several weeks the left ventricle will be incapable of sustaining systemic cardiac output. Our cut-off date has been 6 weeks after birth for one-stage switch in patients with intact septum. In a child presenting within 6 weeks of birth, we would go straight to an arterial switch operation, with the expectation that we might have to support the ones presenting late them with a ventricular assist device, although, fortunately, that has not been something we have had to do.

In children older than 6 weeks and in the absence of left ventricular outflow tract obstruction, we would probably do a rapid two-stage arterial switch where we took the baby to surgery, placed a pulmonary artery band, retrained the left ventricular. Typically you can take those patients back to surgery within a week of the banding for an arterial switch.

What about the child that presents with transposition intact septum and severe hypoxemia that is not responding to standard measures, prostaglandin, balloon atrial septostomy, hyperventilation? Certainly we have faced this situation in our experience. It is usually not the case that nitric oxide helps this situation. So our standard now is to take those patients straight to surgery. We had 2 patients in our series that had to go to surgery within 12 hours of delivery.

Dr. Kron, you asked about marking the coronary arteries. How a surgeon chooses to translocate the coronary arteries is a matter of personal style and preference, I think. In our hands we believe that we can identify the most appropriate location for ostial transfer before arresting the heart, and so we put little marking stitches on the pulmonary artery before we cannulate. Others have techniques which work well for them.

Would there be patients with transposition that we would not offer the arterial switch operation to? Patients that have severe organic left ventricular outflow tract obstruction that we could not resect, we would not offer the switch operation. There is a controversial issue about what to do in the patient that has an abnormal pulmonary valve that will become the aortic valve. We have felt justified in being aggressive about that particular issue, even to the extent that in 2 of our patients we replaced the aortic root. Clearly that is controversial, and one would be warranted in considering an atrial switch in that setting.

Likewise, in a patient with severe left or right ventricular hypoplasia we might consider not proceeding with the arterial switch. But those would be really the only settings that I can envision not offering it. Certainly no coronary branching pattern would be denied.

Finally, what happens to the neoaortic root? This is a very good and important question. The answer is not yet complete. The neoaortic root will continue to dilate in some patients, as Dr. Kron alluded to, particularly patients with transposition and ventricular septal defect who typically have a large pulmonary artery. So when we translocate the pulmonary arteries we further enlarge the aortic sinuses and that does predispose some of those patients to aortic insufficiency. In our relatively short series, we have not seen this as an important problem, although others with longer experience have seen gradually developing neoaortic insufficiency as an important problem and dilation and kinking of the proximal arteries to the extent that there are some centers that are advocating routine coronary angiography in their follow-up of their patients. We have not done that yet, but we will certainly be watching our patients very closely as they age.

Finally, Dr. Crawford, thank you for your very kind comments.