S (UVH) remains problematic. Whether naturally occurring. Operations for Subaortic Stenosis in Univentricular Hearts

ORIGINAL ARTICLES Operations for Subaortic Stenosis in Univentricular Hearts Tom R. Karl, MS, MD, Kevin G. Watterson, FRACS, Shunji Sano, MD, PhD, and Roger B. B. Mee, ChB, FRACS Victorian Paediatric Cardiac Surgical Unit, Royal Children s Hospital, Melbourne, Australia Optimal prevention and treatment of subaortic stenosis (SAS) in the univentricular heart with subaortic outlet chamber and high pulmonary blood flow remains controversial, especially when complicated by aortic arch obstruction. Herein we analyze our surgical results. Group 1 consisted of 11 infants (mean age, 10 days) with univentricular heart and SAS. Ten required repair of interrupted aortic arch (n = 7) or coarctation with hypoplastic arch (n = 7). Four patients had relief of SAS by either Damus-Kaye-Stansel connection (n = 2) or aortopulmonary window (n = 2), with three operative deaths and one late death. Six had one-stage arterial switch and atrial septectomy with arch repair (5/6) with one operative death and one late death. Two survivors have progressed to bidirectional cavopulmonary shunt, a third has had a Fontan operation, and a fourth awaits Fontan. In group 2, 11 children required operation for acquired SAS after pulmonary artery banding. Nine have progressed to Fontan operation with either staged (n = 3) or concurrent (n = 6) relief of SAS by Damus-Kaye-Stansel connection or subaortic resection. Fontan mortality was 11% (70% confidence limits, 2% to 32%). Group 3 consisted of 3 patients without pulmonary artery banding who had SAS diagnosed at Fontan evaluation. All 3 survived Fontan operation and relief of SAS by Damus- Kaye-Stansel connection or subaortic resection. Group 4 consisted of 1 patient with previous pulmonary artery banding (no SAS) who underwent Fontan operation but required Damus-Kaye-Stansel connection 30 months later for SAS. We conclude that arterial switch, arch repair, and atrial septectomy effectively palliate newborns with univentricular heart and SAS, although ultimate Fontan suitability is yet to be established. Both Damus-Kaye-Stansel connection and subaortic resection provide good long-term relief of SAS in select patients, and previous or concurrent treatment of SAS does not preclude successful Fontan operation in otherwise suitable patients. (Ann Tkorac Surg 1991;52:420-8) ubaortic stenosis (SAS) in the univentricular heart S (UVH) remains problematic. Whether naturally occurring or secondary, SAS and its frequent sequelae, ventricular hypertrophy and dysfunction, have been identified as incremental risk factors for poor outcome after Fontan operations [l-71. Although it is now widely accepted that pulmonary artery banding (PAB) can cause or accelerate the development of SAS in various types of UVH with discordant ventriculoarterial connections, a universally acceptable alternative for control of pulmo- For editorial comment, see page 415. nary blood flow in infancy has not been proposed [5, &lo]. In this study we analyze our own experience with various palliative and definitive operations for SAS in the UVH. Long-term and short-term outcome, effect of palliation on Fontan suitability, and risks relative to patients with UVH without SAS are examined retrospectively. We also report our initial experience with a new form of palliation for newborns with UVH, SAS, and aortic arch Presented at the Twentyseventh Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Feb 18-20, 1991. Address reprint requests to Dr Karl, Royal Children s Hospital, Flemington Rd, Parkville, Victoria, Australia 3052. obstruction, the subset of patients in which severe SAS is most likely to develop after PAB (4, 111. Material and Methods All patients undergoing surgical procedures for SAS in the setting of UVH at the Royal Children s Hospital were identified using our database (1979 to present). Hospital and departmental records were examined by us, including operative notes, inpatientloutpatient entries, reports of cardiac catheterization and echocardiographic studies, and where appropriate the echocardiographic video tapes and angiograms. Autopsy data were either obtained from pathologist s report or when possible from personal examination of specimens by one of us. Follow-up information was obtained by us directly and from our referring cardiologists. No survivors were lost to follow-up. Analysis of data was done using standard statistical methods (when population size warranted). Seventy percent confidence intervals were calculated using continuity correction for upper and lower limits. Patients and Operative Results Twenty-six patients are included in this study, mean age at initial operation for SAS being 45 months (median, 13 0 1991 by The Society of Thoracic Surgeons 0003-4975/91/$3.50

~ ~ ~ 1991;52:42&8 KARLETAL 421 Table 1. Clinical Profile of 22 Newborns With UVH Who Required Surgical Treatment of SAS Patient No. Diagnosis SAS Criteria Operation (Age) Outcome 1 DILV, TGA, type A IAA 2 3 4 5 6 7 8 9 10 Left AVV atresia, TGA, CoA, hypoplastic arch DILV, TGA, CoA, hypoplastic arch DILV, CoA, hypoplastic arch, duct-dependent circulation Tricuspid atresia, TGA, CoA + hypoplastic arch, ductdependent circulation Left AVV atresia, TGA, type B IAA Tricuspid atresia, TGA, CoA, hypoplastic arch, ductdependent circulation DILV, TGA, type B IAA, ductdependent circulation, single coronary DILV, TGA, situs inversus, bilateral SVC DILV, TGA, CoA, hypoplastic arch 11 DILV, TGA, hypoplastic arch Subaortic muscle bands, small BVF (echo) Subaortic muscle bands + small BVF (echo) Subaortic muscle bands + small BVF (echo) a. Switch + atrial septectomy + arch repair + PAB (2 d) b. Fontan (25 mo) Arch repair + atrial septectomy + PAB + APW (10 d) Arch repair + enlargement BVF, PAB (1 d) a. Switch + arch repair + atrial septectomy (15 d) b. Right MBTS (5 wk) a. Switch + arch repair + atrial septectomy (4 d) b. Right MBTS (3 mo) Arch repair + atrial septectomy + APW + right MBTS (5 d) Switch + arch repair + atrial septectomy (2 wk) Arch repair + DKS + right MBTS (2 d) Switch + atrial septectomoy + PAB a. Switch + arch repair + atrial septectomy (4 wk) b. MBTS (4 mo) c. BCPS + division MPA (19 mo) Arch repair + atrial septectomy + PAB (6 d) 3 mo post Fontan, free of SAS Well at 38 mo, awaits Fontan, free of SAS Well at age 13 mo, awaits Fontan. free of SAS Late death (3% mo, probable recoarctation) Late death (3 mo, 50- mm Hg at APW) (neoaortic valve incompetence) (? coronary compression by DKS) Well at 30 mo, awaits Fontan, free of SAS Well at 25 mo, awaits Fontan, free of SAS ~ ~ ~ ~ ~ ~ ~ ~ _ APW = aortopulmonary window; AVV = atrioventricular valve; BCPS = bidirectional cavopulmonary shunt; BVF = bulboventricular foramen; CoA = coarctation of the aorta; DILV = double-inlet left ventricle; DKS = Darnus-Kaye-Stansel connection; echo = echocardiography; IAA = interrupted aortic arch; MBTS = modified Blalock-Taussig shunt; MPA = main pulmonary artery; NYHA = New York Heart Association; PAB = pulmonary artery banding; SAS = subaortic stenosis; SVC = superior vena cava; TGA = transposition of the great arteries; UVH = univentricular heart. months). The initial anatomical diagnoses are listed in Tables 1 through 3. The majority of patients had doubleinlet left ventricle (or atrioventricular valve atresia) with a subaortic outlet chamber, discordant ventriculoarterial connection, and restrictive bulboventricular foramen (BVF). Patients were grouped according to age when first seen, type of operation for relief of SAS, and relation to previous operation, if any. Group 1 Eleven patients had SAS recognized and treated surgically in infancy at a mean age of 10 days. The diagnosis was established by echocardiography in all. The criteria for diagnosis of SAS were a BVF significantly smaller (less than half diameter) than the aortic valve annulus or a measurable gradient from dominant ventricle to aorta (Doppler estimated), or both. Ten patients in group 1 had associated aortic arch obstruction consisting of inter- rupted aortic arch (n = 3) or coarctation plus hypoplastic transverse arch (n = 7). Several surgical procedures were employed in this group (see Table 1). Four patients seen early in the series had one-stage arch repair through a median sternotomy plus atrial septectomy and either creation of aortopulmonary window plus PAB (n = 2) or Damus-Kaye-Stansel connection (DKS) plus modified Blalock-Taussig shunt (n = 2). There were three operative deaths, and the fourth patient died 3 months after operation. This fourth patient had recurrent/persistent SAS with a 50-mm Hg gradient across the aortopulmonary window at catheterization just before his death. A fifth patient in group 1, with a right-sided outlet chamber and double-inlet left ventricle, underwent arch repair plus enlargement of BVF and PAB. He is currently awaiting Fontan operation and is free of SAS. More recently, 6 patients have been treated with one-

422 KARLETAL 1991;52:42&8 Table 2. Clinical Profiles of 11 Children With UVH in Whom SAS Developed After PAB Patient Operation No. Diagnosis SAS Criteria (age) Outcome 1 2 3 4 5 6 7 8 9 10 11 AVSD, hypoplastic hypoplastic arch DILV, TGA LAVV atresia, TGA DILV, TGA, CoA DILV, TGA LV, CoA, Left AVV atresia, TGA, CoA Single ventricle (indeterminate morphology); common AVV, LA isomerism, interrupted IVC, CoA Tricuspid atresia, TGA Large VSD, hypoplastic right ventricle straddling tricuspid valve DILV, TGA DILV, TGA 30-mm Hg cath gradient 10-mm Hg cath gradient 100-mm Hg cath gradient + 15-mm Hg cath gradient 40-mm Hg cath gradient (Vent-Ao) 20-mm Hg cath gradient 40-mm Hg cath gradient 70-mm Hg cath gradient induced with isoproterenol 15-mm Hg cath gradient a. Arch repair + PAB (1 mo) b. DKS + BCPS (14 mo) c. AVV replacement (22 mo) a. PAB (10 d) b. DKS + Fontan (40 mo) a. PAB + septostomy (30 d) b. DKS + Fontan (28 mo) a. PAB + CoA repair (3 d) b. Subaortic resection (trans- RV) (2 mo) c. Repair RV aneurysm (3 mo) d. Fontan (34 mo) a. PAB (10 d) b. PA reconstruction + central shunt (71 mo) c. Fontan (41 mo), subaortic resection (trans-ao) (81 mo) a. PAB + CoA repair (7 d) b. DKS + BCPS septectomy (13 mo) c. Fontan (35 mo) a. PAB + CoA repair (5 d) b. DKS + BCPS + common AVV replacement (19 mo) a. PAB (8 d) b. Subaortic resection (trans- Ao) + BCPS (37 mo) c. Fontan (45 mo) a. PAB + septectomy (5 mo) b. Fontan + subaortic resection (trans-rv) (62 mo) c. Subaortic resection (trans- Ao) (15 Y) a. PAB (3 y) b. Fontan + DKS (82 mo) a. PAB (5 mo) b. Fontan + subaortic resection (trans-ao) (7 y) Probably Fontan unsuitable 15 mo postop 16 mo postop, free of SAS 19 mo postop, free of SAS 46 mo postop, free of SAS 4 mo postop, 15 to 30- mm Hg SAS (echo) 25 mo postop, free of SAS NYHA 8 mo I postop, free of SAS 18 mo postop, Free of SAS AO = aorta; cath = catheterization; IVC = inferior vena cava; LA = left atrial; LV = left ventricle; PA = pulmonary artery; RV = right ventricle; VSD = ventricular septa1 defect; other abbreviations as in Table 1. stage arterial switch, aortic arch repair (5/6), and atrial Group 2 septectomy, with one operative death (17%; 70% confi- Eleven children, median age 10 days, who did not have dence limits, 2% to 46%). Three of 5 survivors required SAS underwent PAB. All subsequently required surgical modified Blalock-Taussig shunt at a mean interval of 10 weeks from the time of arterial switch. There has been one treatment for SAS 2 to 199 months after PAB (mean, 52 sudden late death in this latter group, occurring at 5 months; standard deviation, 56 months). Four of 11 (36%) months, probably related to recoarctation. required a coarctation repair at the time of PAB. There Of the 4 long-term survivors, 2 have undergone suc- were no operative deaths related to PAB. The diagnosis of cessful bidirectional cardiopulmonary shunts (BCPSs) SAS after PAB was established by two-dimensional Dopwith division of the main pulmonary artery (PA). A third pler echocardiography or cardiac catheterization or both. patient has had a Fontan operation, and a fourth awaits Echocardiographic criteria were the same as in group 1, BCPS or Fontan. All survivors are free of SAS and and any directly measured ventricular-aortic gradient was expected to be Fontan suitable. considered clinically significant.

~ ~~~ ~ 1991;52:420-8 KARLETAL 423 Table 3. Clinical Profiles of 3 Children With UVH and SAS Who Had Concurrent Fontan Operation and Relief of SAS and 1 Child With UVH Who Required Surgical Treatment of SAS After Fontan Operation Patient No. Diagnosis SAS Criteria Operation (age) Outcome Group 3 1 DILV, TGA, mild PS and 30-mm Hg cath gradient Fontan + DKS (137 mo) NYHA 1 Sub-PS 15 mo postop, 10- mm Hg SAS by Doppler echo 2 DILV, TGA, PS 50-mm Hg gradient (LV- Fontan + resection SAS Ao) intraop (240 mo) 17 mo postop, free of SAS 3 DILV, TGA, PS 10-mm Hg cath gradient Fontan + DKS + pulmonary and small BVF (echo + angio) valvotomy (34 mo) 9 mo postop, free of SAS, mild A1 Group 4 1 DILV, TGA, left-sided 50-mm Hg cath gradient a. CoA repair + PAB NYHA 1 outlet chamber (3ASV), (newborn) 30 mo after DKS, free of CoA b. Fontan (34 mo) SAS c. DKS (64 mo) A1 = aortic insufficiency; angio = angiography; Ao = aorta; cath = catheterization; echo = echocardiography; LV= left ventricle; PS = pulmonary stenosis; 3ASV = type 3A single ventricle (Van Praagh); other abbreviations as in Table 1. Treatment of SAS in this group consisted of either DKS or subaortic resection. In general, the former procedure was favored when the subaortic chamber was left sided (as in levotransposition of the great arteries), whereas direct resection was used for patients with right-sided outlet chamber. The fate of patients in whom SAS developed after banding is outlined in Table 2. Nine of 11 ultimately progressed to Fontan operations. Of these 9 patients the SAS was dealt with at staging procedures in 3 (subaortic resection with or without BCPS, n = 2; DKS + BCPS, n = 1). The remaining 6 patients had Fontan operation plus concurrent subaortic resection (n = 3) or Fontan operation plus concurrent DKS (n = 3). There was one operative death in the Fontan group (11%; 70% confidence limits, 2% to 32%), occurring in a patient who had concomitant DKS. Postmortem examination showed severe myocardial hypertrophy and interstitial fibrosis. Long-term follow-up of the patients who underwent Fontan operation revealed recurrence of SAS in 1 patient 10 years after transventricular subaortic resection. She underwent reoperation to relieve a 40-mm Hg ventricularaortic gradient and could not be separated from cardiopulmonary bypass. She was bridged to cardiac transplantation with a ventricular assist device but died postoperatively of gram-negative sepsis. Postmortem examination revealed severe myocardial hypertrophy. The remaining 7 patients who underwent Fontan operation are well 12 to 46 months (mean, 21 months) after relief of SAS, although 1 has a detectable ventricular-aortic gradient by two-dimensional Doppler echocardiography. Two children in group 2 have had DKS plus BCPS and common atrioventricular valve replacement, with one operative death due to pulmonary vascular disease. The survivor has no SAS 23 months after operation but may not be a Fontan candidate due to elevated upper body systemic venous and PA pressures. Group 3 Three patients aged 34 months to 20 years, with no previous PAB, were discovered to have SAS at the time of preoperative evaluation for Fontan operation (n = 2) or during the operation (n = 1) (see Table 3). Two patients (with leftward outlet chambers) had Fontan + DKS and the third (with rightward outlet chamber) had transaortic subaortic myectomy. There were no operative deaths, and all 3 patients are in New York Heart Association class I at 9 to 17 months follow-up. One has a Doppler signal compatible with a 10-mm Hg ventricular to aortic gradient, as compared with a resting 25-mm Hg mean gradient at preoperative catheterization, at which time the heart appeared volume loaded with a dynamic component of obstruction. Group 4 A single patient who underwent PAB and coarctectomy in the newborn period underwent Fontan operation at 34 months of age. Neither catheterization nor two-dimensional Doppler echocardiography suggested SAS at that time. Thirty months later, however, SAS was documented with a 50-mm Hg ventricular-aortic gradient at catheterization. A DKS was successfully constructed at that time using the previously transected and oversewn PA. The patient remains free of SAS and is in New York Heart Association class I at 30 months follow-up (see Table 3).

424 KARLETAL 1991;52:420-8 Surgical Techniques Pulmonary Artery Banding Our newborn banding technique employed median sternotomy whenever possible, the exception usually being a requirement for concomitant coarctation repair. In this latter case, the PAB was applied through a left thoracotomy. Through a sternotomy, the upper pericardium was incised in the midline and the PA encircled with silicone elastomer-impregnated Teflon tape. The initial circumference of the band was usually 25 mm for a 3-kg baby, and adjustment was made to ensure a 10-mm Hg increase in mean arterial pressure with banding. Proximal and distal PA pressures were not measured. An arterial hemoglobin saturation of greater than 0.75 with satisfactory appearance of the ventricle and no acute increase in central venous pressure were additional requirements for a satisfactory band. The pericardium was closed whenever possible after the band was secured to the proximal PA with multiple sutures. If a future DKS procedure was a possibility, then high fixation of the band was employed to preserve as much length of PA as possible. The sternotomy approach allowed a more precise placement of the PAB and in our opinion reduced the risk of late problems with the branch pulmonary arteries. Resternotomy after newborn PAB generally was not difficult, and the siliconized bands were easily separated from PA and left atrial appendage. Damus-Kaye-Stansel Connection This technique, originally described as part of an operation for transposition of the great arteries and later adapted for relief of SAS, involves end-to-side connection of the proximal transected PA to the proximal aorta [12-161. Using hypothermic cardiopulmonary bypass and cardioplegic arrest, we routinely excised a triangular portion of aortic wall large enough to create an unobstructed PA-aortic connection. In most cases the connection was made without the addition of prosthetic material, even after PAB. We have used this connection in a patient with valvar pulmonary stenosis (combined with pulmonary valvotomy) and after previous division and oversewing of the proximal PA, with good relief of stenosis in both cases. Bidirectional Cavopulmonary Shunt All cavopulmonary shunts were performed during cardiopulmonary bypass with a beating heart. If concomitant Fontan operation was not performed, the superior vena cava was not routinely cannulated, and superior vena caval return was collected with a pump sucker during anastomosis. Cavopulmonary anastomoses were performed with continuous and interrupted fine polypropylene sutures, using a pulmonary arteriotomy of greater diameter than the superior vena cava to avoid narrowing the anastomosis. Su baortic Resect ion Enlargement of the BVF in hearts with right-sided outlet chambers was performed during hypothermic cardiopul- monary bypass with cardioplegic arrest, through either aorta (n = 4) or anterior ventricle (n = 2). In either case the resection was performed to enlarge the BVF in a superior and anterior direction. The end point was to create a BVF or subaortic channel equal in size to the aortic valve. There were no episodes of heart block using either the transaortic or transventricular approach. An aneurysm at the ventriculotomy developed in 1 patient and required early reoperation. Arterial Switch Plus Aortic Arch Reconstruction Repair of hypoplastic or interrupted aortic arch was performed from an anterior approach using profound hypothermia and circulatory arrest, in 3 cases with concurrent isolated myocardial perfusion [17]. All ductal tissue was excised and the coarctation and the hypoplastic arch segments (if present) were resected. Oblique end-to-end or end-to-side (interrupted aortic arch) anastomoses were then performed with running polypropylene sutures, after which pump flow was resumed. Arterial switch was then performed using the Royal Children s Hospital technique previously described, with atrial septectomy performed during a brief second period of circulatory arrest [18]. The PA was transected as high as possible to minimize any size mismatch with the native aortic root. Fontan Operation The type of Fontan connection used was atriopulmonary (n = 10) or cavopulmonary (n = 3). In either case, atrial neoseptation was performed with a Gore-Tex baffle, leaving atrioventricular valves and coronary sinus to the left. Comment Subaortic stenosis in the UVH has been recognized worldwide as a clinical entity since the early years of the Fontan experience [lo, 121. The key anatomical feature in SAS in the UVH is the BVF, which may be a slitlike opening with a completely muscular rim, somewhat smaller than the aortic annulus itself [2]. As such, there is a tendency for spontaneous closure, a process that is undoubtedly accelerated by the immediate reduction in diastolic heart volume after an effective PAB. In the longer term there is further reduction in BVF size due to PAB-related myocardial hypertrophy [19, 201. Other anatomical causes of SAS may also exist in the UVH and may or may not be related to the presence of PAB [12, 211. The time course for development of SAS after PAB for patients in whom it is not recognized at birth appears to be variable. Subaortic stenosis can become apparent very early in the postoperative period and is said to nearly always be evident by 2.5 years [5, 11, 201. The mean time from PAB to surgical treatment of SAS in our series was 52 months (standard deviation, 56 months), although it is possible that the problem might have been present (but clinically silent) earlier in some patients, skewing the distribution. Furthermore, some newborns had open palliation of SAS, eliminating part of the cohort in which more severe SAS probably would have developed early after PAB. It is also likely that the diagnosis of SAS would have been more

1991;52420-8 KARLETAL 425 frequently made in the latter years of this study, with wide availability of two-dimensional Doppler echocardiography. The actual incidence of SAS after banding in patients with UVH has varied from series to series and relates to a certain extent to the underlying anatomy. The classic case is the double-inlet left ventricle (or tricuspid atresia) with aorta arising from a right ventricular outlet chamber (discordant ventriculoarterial connection) and high pulmonary blood flow [5, 9, 111. The incidence of SAS after PAB in this anatomical subset has been estimated to be as high as 42% to 84.4% in carefully followed up infant cohorts [5, 9, 111. Our own experience supports this. The majority of our UVH patients coming forward for treatment of SAS have likewise had left ventricular morphology (24/26) and discordant ventriculoarterial connection (23/26). In our series, patients with any type of aortic arch obstruction were 7 times more likely to require surgical treatment of SAS in the newborn period (p < 0.005). The association of arch obstruction and early SAS in the UVH has been documented previously [4, 111. it has been suggested that the presence of SAS in utero can result in subnormal aortic flow, a situation favoring development of coarctation or other types of arch obstruction [20]. The diagnosis of SAS in the UVH requires echocardiographic demonstration of a BVF that is significantly smaller than the aortic root, or direct measurement of a ventricular aortic gradient. The majority of newborns at the Royal Children s Hospital currently are diagnosed with two-dimensional Doppler echocardiography (see Tables 1-3). Because this group tends to have arch obstruction and high pulmonary blood flow, a ventricular aortic pressure gradient is not always generated before PAB and an increase in aortic flow [4, 5, 221. Our arbitrary echocardiographic criterion for diagnosis of SAS over the period of this study has been BVF size less than half the diameter of the aortic root. For patients undergoing cardiac catheterization, any measured ventricular aortic gradient was considered to be clinically significant. We have also used isoproterenol challenge to induce a gradient in 1 patient, as recommended by others [4, 101. Even with catheterization and echocardiographic studies, the diagnosis of SAS may be difficult to establish. One patient in our series, who was 20 years old and had double-inlet left ventricle, transposition of the great arteries, pulmonary stenosis, and no previous PAB had no gradient at catheterization and an adequate size BVF by two-dimensional Doppler echocardiography and cineangiography. At the completion of Fontan operation, hemodynamics were poor and a harsh thrill was noted at the base of the heart. A 60-mm Hg gradient was measured between ventricle and aorta. This was successfully relieved with transaortic subaortic myectomy. it is thus possible that the Fontan operation itself, by volume unloading the single ventricle can, under some anatomical circumstances, initiate or increase SAS. This further strengthens the argument that any systemic gradient must be considered to be important in such patients. The development of SAS, though manageable in some cases with complex surgical procedures, is certainly an undesirable situation. Progressive ventricular hypertrophy, subendocardial ischemia, and reduced ventricular compliance are predictable consequences. Although not mentioned per se in Fontan s original exclusion criteria, ventricular hypertrophy has been identified by Kirklin and associates [7] as a risk factor for early and late mortality after Fontan operation. A reduction in ventricular compliance may primarily affect diastolic function, with elevation of ventricular end-diastolic pressures as well as PA pressures distal to a band [6]. This results in a quite unfavorable post-fontan hemodynamic situation, and may in some cases result in Fontan unsuitability. Prevention of SAS in patients with the anatomical substrate would therefore be highly desirable, although at present it is not always possible. Alternatives to PAB to control pulmonary blood flow in the newborn period may generate problems worse than the band itself. Procedures involving disconnection of the PA from the heart require an alternate source of pulmonary blood flow. The BCPS is not appropriate in newborns, and systemic to PA shunts may create an unstable postoperative hemodynamic situation after cardiopulmonary bypass without guarantee of preventing ventricular hypertrophy and SAS. In our view, PAB still provides the best palliation. Because the development of SAS after PAB may ultimately exclude some patients from Fontan operation, early further interim palliation may be required to prevent progression or allow regression of ventricular hypertrophy. In our institute, of 178 patients undergoing Fontan operation, 31 had previous PAB, and 10 of 31 required surgical relief of SAS before or concurrent to Fontan operation. We have not identified previous PAB or previouskoncurrent treatment of SAS as a risk factor for early mortality after Fontan operation (p < 0.32 and p < 0.89, respectively). We recognize, however, that the numbers are relatively small and that the question of Fontan suitability in remaining palliated patients remains unanswered. Optimal management of established SAS remains controversial. The newborn group presents the most difficult problem, as the majority of these infants also have aortic arch obstruction requiring urgent treatment. in the past, various forms of palliation have been employed by ourselves and others. in most centers, including our own, results with operations employing a DKS or aortopulmonary window have had disappointing results in newborns [l, 2, 20, 23, 241. The limiting factor appears to be the unstable post-cardiopulmonary bypass state created by a systemic shunt-dependent pulmonary circulation. The problem of maintaining balanced pulmonary and systemic flows is similar to that encountered after operation for the hypoplastic left heart syndrome. Also, survivors of aortopulmonary window may experience late problems with stenosis of the connection or distortion of semilunar valves, which would make subsequent Fontan suitability less likely [4, 18, 241. Because of these problems, our preferred treatment for newborns with UVH and established SAS is now arterial switch plus atrial septectomy, usually with concomitant arch reconstruction. Pulmonary blood flow after arterial

426 KARLETAL 1991;52:420-8 switch was effectively limited by the restrictive BVF in all but 1 of our patients (who had no arch obstruction and required PAB after switch + septectomy). This is similar to the situation of many newborns with tricuspid atresia or double-inlet left ventricle plus subpulmonic outlet chamber and normally related great arteries, a group in which SAS rarely develops. The requirement for modified Blalock-Taussig shunt after arterial switch is consistent with the natural history of the latter anatomical situation. The initial regulation of pulmonary blood flow by a restrictive BVF avoids the problems associated with a systemic shunt after a long period of cardiopulmonary bypass with fluctuating pulmonary and systemic vascular resistance. We believe that the presence of a valve in the pulmonary circuit may be important in this respect, preventing continuous flow. The initial operative mortality of 116 (17%; 70% confidence limits, 2% to 46%) appears to compare favorably with experience with other SAS procedures in the newborn, although the small number of patients so treated does not allow us to generate a meaningful nonparametric comparison with alternative operations. Also, there has been one late sudden death at 5 months postoperatively, probably related to recurrent arch obstruction (autopsy not performed). None of the survivors has a catheterization or two-dimensional Doppler echocardiographically detectable ventricular-aortic gradient or severe ventricular hypertrophy. Arterial switch has been employed previously in the UVH. Trusler [2] has used arterial switch combined with right atrium to right ventricle Fontan connection in older patients with rudimentary subaortic right ventricle large enough to contribute stroke work to the pulmonary circulation. This would appear to be an excellent option in this older group, although we have not yet employed it in our own unit. Treatment of acquired SAS in older patients, whether after PAB or de novo, has varied according to age at detection, severity, and Fontan suitability by other criteria (see Table 2). The DKS is an effective way to bypass SAS and appears to provide sustained relief of obstruction. We have used this connection successfully even in the presence of pulmonary stenosis (with pulmonary valvotomy) or after previous PA division. Combined with BCPS and division of main PA as interim palliation, it is possible to allow regression of hypertrophy and growth of the patient before Fontan operation. For otherwise suitable patients (without severe hypertrophy), DKS can be performed concurrent to Fontan operation with acceptable results [9, 12, 231. Our own experience is limited to 3 patients undergoing DKS plus BCPS (1 hospital death; 33%; 70% confidence limits, 4% to 75%), 1 of whom has since had a successful Fontan operation. An additional 5 patients had DKS plus Fontan operation without interim BCPS, with one death (20%; 70% confidence limits, 3% to 52%). Finally, 1 patient in whom SAS developed late after a Fontan operation had successful relief with a DKS constructed 64 months postoperatively. Another surgical option that we have employed for SAS is enlargement of the BVF, which sometimes must be combined with resection of additional subaortic muscle bands. Despite potential complications of heart block, coronary injury, and recurrent SAS this approach has been used extensively in many centers [l, 2, 4, 6, 21, 231. It has also been combined with right ventricular outflow patch with good results [25]. We prefer a transaortic approach whenever possible, as aneurysm formation at the systemic pressure ventriculotomy is possible using a transright ventricular approach, as we have seen in 1 newborn. The main risk appears to be heart block due to proximity of the conduction axis to the inferior rim of the BVF [12, 211. Fortunately, we have not encountered this complication in any of our patients. Obtaining adequate relief of SAS with direct resection is more difficult in patients with a left-sided subaortic outlet chamber, and we consider these patients to be generally better candidates for a DKS-type connection if the pulmonary valve and artery are adequate. Resection is probably also a better option in larger patients. Relief of SAS with resection has been good, with only 1 patient requiring reoperation 10 years after the initial procedure. We and others have used resection both as a palliative and definitive (with Fontan) procedure, with no operative deaths. Other technical options for relief of SAS in univentricular heart include ventricular-aortic valved conduits and PA-descending aortic conduit with distal PAB, with which we have no experience [4, 20, 22, 231. Cardiac transplantation may also be a reasonable option in severe cases. Newborns with UVH and SAS usually have coexisting aortic arch obstruction and are often prostaglandin El dependent with low cardiac output. Under the current protocol at the Royal Children s Hospital, these babies undergo resuscitation and two-dimensional Doppler echocardiographic diagnosis and are then treated with arterial switch, arch reconstruction, and atrial septectomy. A right modified Blalock-Taussig shunt is used if pulmonary blood flow becomes inadequate in the postoperative period. If further palliation is required and the child is unsuitable for Fontan operation due to age or other factors, we would consider BCPS and division of main pulmonary artery; otherwise, we would proceed directly to Fontan operation. For infants with UVH, high pulmonary blood flow, and no SAS, transsternotomy PAB is our procedure of choice. If isolated coarctation is present (without hypoplastic arch or SAS) then a left thoracotomy is used for concomitant coarctation repair and PAB. Patients in whom SAS develops beyond infancy, with or without previous PAB, are treated with DKS plus BCPS or subaortic myectomy plus BCPS if unsuitable for Fontan operation (due to young age, severity of SAS, ventricular hypertrophy or other factors). Older children with mild SAS and without severe hypertrophy would proceed directly to Fontan plus DKS or Fontan plus transaortic subaortic myectomy, depending on anatomy (suitability of pulmonary root, position of outlet chamber and aorta). In conclusion, arterial switch plus arch repair and atrial septectomy provides effective palliation for newborns with UVH, arch obstruction, and SAS. The operative risk

1991;52:42&8 KARLETAL 427 will probably compare favorably with other forms of palliation. The ultimate suitability of newborns thus treated for Fontan operation appears to be good, although the small number of patients precludes firm conclusions at this time. The DKS operation and subaortic resection both provide good relief of SAS in select patients and are useful either as staging procedures or concurrent to a Fontan operation. Pulmonary artery banding, although associated with development of SAS in the UVH, does not preclude successful Fontan operation. Close follow-up is advisable after PAB to allow early detection of SAS and possible interim palliation before Fontan operation. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Barber G, Hagler DJ, Edwards WD, et al. Surgical repair of univentricular heart (double inlet left ventricle) with obstruction of anterior subaortic outlet chamber. J Am Coll Cardiol 1984;4:771-8. Trusler GA, Freedom RM. Management of subaortic stenosis in the univentricular heart. 1989;47:64M. OLeary PW, Driscoll DJ, Connor A, Puga FJ, Danielson GK. The univentricular heart with subaortic stenosis: results of a staged approach. Circulation 1990;82(Suppl 3):76. Penkoske PA, Freedom RM, Williams WG, Trusler GA, Rowe RD. Surgical palliation of subaortic stenosis in the univentricular heart. J Thorac Cardiovasc Surg 1984;87: 767-81. Freedom RM, Lee NB, Smallhorn JF, Williams WG, Trusler GA, Rowe RD. Subaortic stenosis, the univentricular heart and banding of the pulmonary artery: an analysis of the courses of 43 patients with univentricular heart palliated by pulmonary artery banding. Circulation 1986;73:75%64. Newfield EA, Nikaidoh H. Surgical management of subaortic stenosis in patients with single ventricle and transposition of the great vessels. Circulation 1987;76(Suppl 3):29. Kirklin JK, Blackstone EH, Kirklin JW, Pacific0 AD, Bargeron LM. The Fontan operation: ventricular hypertrophy, age and date of operation as risk factors. J Thorac Cardiovasc Surg 1986;92: 1049-64. Freedom RM, Sondheimer H, Dische R, Rowe RD. Development of "subaortic stenosis" after pulmonary artery banding for common ventricle. Am J Cardiol 1977;39:7%83. Waldman JD, Lamberti JJ, Kirkpatrick GL, et al. Experience with the Damus procedure. Circulation 1988;78(Suppl 3): 32-9. Somerville J, Becu L, Ross D. Common ventricle with acquired subaortic stenosis. Am J Cardiol 1974;34:20&14. Franklin RCG, Sullivan ID, Anderson RH, Shinebourne E, Deanfield JE. Is banding of the pulmonary trunk obsolete for infants with tricuspid atresia and double inlet ventricle with a discordant ventriculoarterial connection? Role of aortic arch obstruction and subaortic stenosis. J Am Coll Cardiol 1990; 16:145564. 12. Yacoub MH, Radley-Smith R. Use of a valved conduit from RA to PA for "correction" of single ventricle. Circulation 1976;54(Suppl 3):63-70. 13. Doty DB, Marvin WJ, Lauer RM. Single ventricle with aortic outflow obstruction. Operative repair by creation of double outlet to the aorta and application of the Fontan principle. J Thorac Cardiovasc Surg 1981;81:636-40. 14. Damus R. Correspondence. 1975;20:724. 15. Stansel HC. A new operation for d-loop transposition of the great arteries. 1975;19:56.57. 16. Kaye MP. Anatomic correction for transposition of the great arteries. Mayo Clinic Proc 1975;50:638-40. 17. Sano 5, Mee RBB. Isolated myocardial perfusion during arch repair. 1990;49:970-2. 18. Brawn WJ, Mee RBB. Early results for anatomic correction of transposition of the great arteries and for double-outlet right ventricle with subpulmonary ventricular septal defect. J Thorac Cardiovasc Surg 1988;95:230-8. 19. Rao SP. Further observations on the spontaneous closure of physiologically advantageous ventricular septal defects in tricuspid atresia: surgical implications. 1983;35:121-9. 20. Jonas RA, Castaneda AR, Lang P. Single ventricle (single or double inlet) complicated by subaortic stenosis: surgical options in infancy. 1985;39:3614. 21. Cheung HHC, Lincoln C, Anderson RH, Ho SY, Shinebourne E, Rigby M. Options for surgical repair in patients with univentricular atrioventricular connection and subaortic stenosis. J Thorac Cardiovasc Surg 1990;100:672-81. 22. Bethea MC, Reynolds JL. Treatment of bulboventricular foramen stenosis by ventricular-ascending aorta valved conduit bypass. 1989;47:765-6. 23. Lin AE, Laks H, Barber G, Chin AJ, Williams RG. Subaortic stenosis in complex congenital heart disease: management by proximal pulmonary artery to ascending aorta end to side anastomosis. J Am Coll Cardiol 1986;7617-24. 24. Sen D, Gidding SS, Bucker C, Benson DW, Mavroudis C. Surgically created aortopulmonary window for subaortic stenosis in complex congenital cardiac defects: a 12 year experience. Circulation 1990;82(Suppl 3):716. 25. Crupi G, Parenzan L. Subaortic outflow tract reconstruction for relief of subaortic stenosis in double inlet left ventricle and tricuspid atresa [Abstract]. Ninth Biennial Asian Congress on Thoracic and Cardiovascular Surgery, Taipei, ROC, 1989. DISCUSSION DR MICHEL ILBAWI (Oak Brook, IL): I would like to congratulate Dr Karl and associates on very beautiful work. I think this is a very complex problem and the results are very outstanding. Recently we have adopted a totally different approach to the problem. At the present time we prefer to palliate these patients very early, before the development of a pressure gradient. The procedure is pulmonary artery to ascending aorta anastomosis. In cases of aortic arch anomalies, we connect the proximal pulmonary artery directly to the arch. We use a piece of Gore-Tex patch to augment the anastomosis and prevent tension and distortion of the semilunar valves. Out of the 13 patients we have treated in the last 8 years, 9 had the initial banding in the neonatal period, and subsequent to that, around 2 to 6 years of age, had relief of the stenosis whenever the pressure gradient became apparent. In the 4 patients who were treated more recently, the relief of stenosis was done in the neonatal period, around 10 days of age. The evaluation is very impressive and shows a marked differ-

428 KARLETAL 1991;52:420-8 ence between the two groups when it comes to pulmonary vascular resistance, for example. The pulmonary vascular resistance was much higher in group I, who had the delayed relief of the subaortic stenosis, in contrast to group 11, where the relief of stenosis was done in the neonatal period. The ventricular muscle mass, which is inversely proportional to ventricular compliance, was also markedly increased in the delayed group, and the ventricular mass to volume ratio, which reflects both the pathological hypertrophy and the contractile state of the ventricle, was markedly increased in group I, who had delayed relief of the obstruction. The post-fontan results again were markedly different. The central venous pressure was higher, takedown of the Fontan operation was necessary in 2 patients, and the hospital mortality was higher in group I because of the severe ventricular hypertrophy and noncompliant ventricle. I wonder whether Karl and associates have any similar experience when it comes to timing of the procedure. The other question relates to techniques. Most of these patients have levotransposition of the aorta, and you are recommending arterial switch. Would you like to comment on the technical details of how you would do an arterial switch on levotransposition of the aorta? DR KARL: To answer the second question, our switch technique in univentricular heart would be identical to our switch technique in simple transposition, using medially based trapdoor flaps for translocation of the coronary arteries. We would transect the pulmonary artery somewhat higher than usual to minimize the size mismatch between the neoanastomosed vessels. Regarding the timing of operation for subaortic stenosis, I may not have made this clear in my talk, but we would treat subaortic stenosis at the time of diagnosis, especially in the newborn group. We would not, for example, band a patient who we thought had a small bulboventricular foramen. We would opt for the treatment course that I have outlined. In children in whom subaortic stenosis develops after banding, we would intervene early, probably with a Damus-Kaye-Stansel operation or direct enlargement of the bulboventricular foramen, depending on the position of the subaortic chamber and aorta. In a young patient who might not be otherwise suitable for Fontan operation, we would perform a bidirectional cavopulmonary shunt concurrently. DR ADNAN COBANOGLU (Portland, OR): Dr Karl, how often do you perform pulmonary artery banding now in patients with univentricular heart and unrestricted pulmonary blood flow? How often do you perform pulmonary artery banding; is it a very uncommon procedure now on your service? DR KARL: I can say that we would perform pulmonary artery banding in such patients when we are happy about the size of the bulboventricular foramen and convinced by the Doppler echocardiographic studies that there is no subaortic stenosis. Otherwise we would proceed with arterial switch and atrial septectomy. Now, in the newborn group this is almost always a child with an obstructed aortic arch who needs an open procedure for that reason as well. DR COBANOGLU: The problem that we have seen is the difficulty at times to decide if the bulboventricular foramen is truly restrictive or not. The morphology of the foramen is not a circle, it is more like a buttonhole, and at times it is difficult by echocardiography to decide whether you have a restrictive foramen or not. DR KARL: Right. I think that question might better be posed to our cardiologists, but I think if the bulboventricular foramen is small, it is not a subtle thing. I mentioned that our criterion was a foramen less than half the diameter of the aortic root, and if there is an associated arch obstruction, I think one should have an even lower threshold to treat it as subaortic stenosis. In our current protocol for newborns, if subaortic stenosis is present, we will perform the arterial switch as outlined. If subaortic stenosis is not present, then we would apply a pulmonary artery band through a sternotomy unless there is an isolated coarctation present, in which case it could be done through a left thoracotomy. For the older child with acquired subaortic stenosis, a Damus- Kaye-Stansel connection is used when there is a left-sided outlet chamber, and subaortic resection, which is usually transaortic, is employed for a right-sided outlet chamber. Depending on other anatomical and physiologic features, a bidirectional cavopulmonary shunt or Fontan operation would be performed concurrently.