Aortic valve insufficiency may be caused by abnormalities

Reconstruction of the Ascending Aorta and Aortic Root: Experience in 45 Consecutive Patients Gebrine A. El Khoury, MD, Malcolm J. Underwood, MD, David Glineur, MD, David Derouck, MD, and Robert A. Dion, MD Cliniques Universitaires Saint-Luc, Brussels, Belgium Background. Patients with aneurysms of the ascending aorta or aortic root may have associated aortic insufficiency (AI). We reviewed our experience with aortic root remodeling and reconstruction of the sino-tubular junction. Methods. Forty-five patients were operated on between July 1995 and September 1998. Transesophageal echocardiography showed AI grade III or IV in 15 patients. Twenty-seven patients had replacement of all three sinuses, 10 of one or two sinuses. Reconstruction of the sino-tubular junction alone was performed in 8 patients. Results. There was one death at 28 days. Perioperative transesophageal echocardiography showed no or discrete AI in all patients. There has been one aortic valve replacement at day 4 postoperatively for cusp repair failure. Transesophageal echocardiography in 40 patients at a mean time of 12.5 months showed no progression of AI in 38 patients, and a grade II in 2. Clinical follow-up averaged 14.5 months. There have been three late, not procedure-related deaths. Thirty-six patients are in New York Heart Association functional class I. There have been no cases of endocarditis. Conclusions. Aortic remodeling is successful in eliminating AI in patients with aortic root disease with minimal mortality and morbidity. Early echocardiography (1 year) has shown no progression of AI in 95% of cases. (Ann Thorac Surg 2000;70:1246 50) 2000 by The Society of Thoracic Surgeons Aortic valve insufficiency may be caused by abnormalities of the leaflets, the root, or a combination of both. In some patients, the disease is confined to the aorta, the leaflets remaining anatomically normal. In these patients, progressive dilatation of the aortic annulus, sinuses, and sino-tubular junction results in outward displacement of the commissures and cusps, which then results in valvular incompetence [1, 2]. Conventional treatment for these patients is composite replacement of the aortic valve and ascending aorta [3]. Despite the success of this operation, complications, including thromboembolism, endocarditis, and problems related to the long-term anticoagulation [4] required, have provided the impetus for the development of a surgical procedure that preserves the native aortic valve. Remodeling of the aortic root was originally described by Sarsam and Yacoub [5] and David [6], and variations of it have been used by other authors [7]. This article describes our experience in this field, and the early clinical and echocardiographic results obtained. Material and Methods Forty-five patients who underwent aortic root remodeling were identified from the surgical database of our unit from July 1995 to September1998. Retrospective analysis Accepted for publication April 4, 2000. Address reprint requests to Dr El Khoury, Cliniques Universitaires Saint-Luc, Avenue Hippocrate 10, B 1200 Brussels, Belgium; e-mail: elkhoury@chir.ucl.ac.be. was made from case notes of the preoperative variables, the operative procedures, and mortality and morbidity occurring during the hospital stay. Prospective follow-up of all patients was achieved by telephone questionnaire to both the patients and their referring cardiologists. Written requests were made to individual cardiologists regarding the echocardiographic follow-up. Statistical analysis of postoperative changes measured at echocardiography was made using a two-tailed paired Student s t test. Statistical significance was defined as p less than 0.05. Preoperative Data The mean age of the patients was 56.5 years (range, 13 to 76 years) and 60% were male. Twenty patients were in New York Heart Association class III or IV, 11 were in class II, and 14 patients were in functional class I. Thirteen patients were operated on for type A aortic dissection; 10 were acute and 3, chronic. One of these had previously had replacement of both the ascending and descending aorta and presented with a false aneurysm and redissection in the residual proximal aorta, and another had had a coronary artery bypass grafting operation 3 years before. Nine patients had manifestations of the Marfan syndrome. Preoperative transesophageal echocardiography showed aortic incompetence (AI) This article has been selected for the open discussion forum on the STS Web site: http://www.sts.org/section/atsdiscussion/ 2000 by The Society of Thoracic Surgeons 0003-4975/00/$20.00 Published by Elsevier Science Inc PII S0003-4975(00)01716-1

Ann Thorac Surg EL KHOURY ET AL 2000;70:1246 50 AORTIC VALVE-SPARING OPERATION 1247 Table 1. Preoperative Findings Number of patients 45 Age (y) 56.5 (13 76) Sex 27 male NYHA 3 15 patients Aortic insufficiency Grade III 15 patients Grade I and II 24 patients Non AI 6 patients Marfan syndrome 9 patients Type A dissections Acute 10 patients Chronic 3 patients AI aortic insuffiency; NYHA New York Heart Association. grade III or IV in 15 patients, grade I or II in 24 patients, and no AI in 6. The mean maximum aortic diameter was 60 mm (range, 49 to 80 mm), and 3 patients had an annular diameter of more than 28 mm. Operative Techniques A median sternotomy was performed, and cardiopulmonary bypass was instituted with ascending aorta and right atrium cannulation whenever possible. In 13 patients, ascending aorta cannulation was not feasible, and femoral artery cannulation was performed. In 1 patient (reoperative dissection), the aorta was accidentally opened at sternotomy and urgent cannulation of femoral artery and vein was required. In early operations, myocardial protection was achieved by a combination of antegrade and retrograde cold crystalloid cardioplegic solution, but in the last 25 patients (55.5% of all procedures), antegrade warm blood cardioplegic solution was used. Aortic root remodeling was performed according to the technique described by David [6] and Sarsam and Yacoub [5]. The dilated sinuses and diseased ascending aorta were excised so that only 4 to 5 mm of aortic wall was left attached to the annulus. The coronary ostia were prepared for a button reimplantation as during a conventional root-replacement procedure. The diameter of the aortic annulus was then measured using metric sizers. If it was not dilated ( 28 mm), then based on the previously described aortic root relationships, a polyethylene terephthalate fiber graft approximately 10% smaller than the size of the annulus was chosen to remodel the sinuses [8]. If the annulus itself was seen to be dilated, an annuloplasty was added to the procedure (3 patients; Fig 1). The dilatation is usually confined to the fibrous portion of the left ventricular outflow tract, and this was reduced with the annuloplasty. Sutures were placed from the inside to the outside of the aortic root in a horizontal plane below the level of the valve leaflets (largely the noncoronary leaflet), all along the fibrous portion of the outflow tract. These sutures were reinforced with a strip of polytetrafluoroethylene felt. Occasionally, performing an annuloplasty may distort the geometry of the root, but this was compensated for by Fig 1. Technique of aortic annuloplasty. (Reproduced, with the permission of the BMJ Publishing Group, from Underwood MJ, El Khoury G, Deronck D, Glineur D, Dion R. The aortic root: structure, function, and surgical reconstruction. Heart 2000;83:376 80.) performing a commissuroplasty at the commissure between the right and left coronary leaflets. To determine the most appropriate diameter graft for a patient with annular dilatation, the length of the free edges of the aortic valve leaflets were measured and averaged. A graft 10% smaller than these combined lengths was then chosen. Three equidistant marks were then made on the graft, and other authors suggest the graft should then be incised vertically along these marks for a length of approximately two thirds of the graft diameter [7]. To more accurately estimate the length of these incisions on the graft however, we find it easier to suspend the aortic valve commissures with Prolene (Ethicon, Somerville, NJ) sutures and measure their height (Fig 2). The vertical incisions in the graft were made slightly longer than this measurement, and the graft was scalloped so as to reproduce the crescent shape of the annulus and to create pseudosinuses in the supravalvular region. The Fig 2. Measurement of commissures. (Reproduced, with the permission of the BMJ Publishing Group, from Underwood MJ, El Khoury G, Deronck D, Glineur D, Dion R. The aortic root: structure, function, and surgical reconstruction. Heart 2000;83:376 80.)

1248 EL KHOURY ET AL Ann Thorac Surg AORTIC VALVE-SPARING OPERATION 2000;70:1246 50 Table 2. Operative Procedure Resection All sinuses 27 Non-coronary sinus 6 Non-coronary right sinus 3 Right sinus 1 Reconstruction of sinotubular junction 8 Aortic annuloplasty 3 Coronary artery bypass graft 3 Resection median raphe 3 Mitral valve repair 8 Tricuspid valve repair 1 Aortic valve repair 6 Elephant trunk procedure 1 Fig 3. Graft Implantation. (Reproduced, with permission of the BMJ Publishing Group, from Underwood MJ, El Khoury G, Deronck D, Glineur D, Dion R. The aortic root: structure, function, and surgical reconstruction. Heart 2000;83:376 80.) aortic commissures were then sutured to the appropriate position on the graft, and the intervening polyethylene terephthalate fiber was sutured to the remnants of the aortic sinuses (Fig 3). Because the perimeter of each scalloped area of the graft is longer than the scalloped aortic annulus, it is important that sutures are placed wider along the graft than along the annulus to create neo-sinuses of Valsalva. To achieve this, suturing was commenced at the level of the commissures. The coronary arteries were then reimplanted, and the graft was anastomosed to the distal aorta in conventional fashion (Fig 4). In the 8 patients with AI and aneurysmal dilatation of the ascending aorta, without root involvement, the ascending aorta was excised preserving the aortic sinuses, and replaced with a supracommissural conduit (neosino-tubular junction). The size of the graft to reconstruct the sino-tubular junction was determined by measuring the annulus. A graft of the same dimension was used in these patients. In patients with acute type A dissection, all diseased sinuses (dissected or dilated) were eventually excised, but the dissected area was first repaired with Cardial surgical glues (Saint Etieme, France) to render the annulus and commissures strong enough for suturing. In all acute dissections, the aortic arch was inspected during a period of circulatory arrest and replaced if necessary. Once completed, perfusion was instituted antegradely. Aortic remodeling was therefore decided on independently of any additional distal procedure required and did not have any bearing on the circulatory arrest time. Operative Procedures Twenty-seven patients underwent remodeling and excision of all three sinuses. In 6 patients, only the noncoronary sinus was replaced, in 3 patients, the noncoronary and right coronary sinuses, and in one patient, the right coronary sinus only. Reconstruction of the sino-tubular junction alone was performed in 8 patients. In 3 patients, the measured annular diameter was more than 28 mm, and an annuloplasty was performed as described. Three patients had a bicuspid aortic valve, and this was repaired at the same time by resection of the median raphe. Coronary artery bypass grafting was performed in 3 patients, mitral valve repair in 8, tricuspid valve repair in 1, aortic cusp repair in 6 (cusp resuspension, triangular resection, decalcification), and 1 patient had an elephant trunk procedure. Fig 4. Finished remodelling procedure. (Reproduced, with permission of the BMJ Publishing Group, from Underwood MJ, El Khoury G, Deronck D, Glineur D, Dion R. The aortic root: structure, function, and surgical reconstruction. Heart 2000;83:376 80.) Results Early Mortality, Morbidity, and EchocardiographicFindings There were no intraoperative deaths. The mean cardiopulmonary bypass time was 139 minutes (range, 66 to

Ann Thorac Surg EL KHOURY ET AL 2000;70:1246 50 AORTIC VALVE-SPARING OPERATION 1249 Table 3. Results Variable 177 minutes), and mean cross-clamp time was 99 minutes (range, 36 to 150 minutes). Perioperative transesophageal echocardiography showed no or discrete AI in all patients. Three patients required reoperation for bleeding. One patient died on day 28 from aspiration pneumonia. He had required urgent operation and had esophageal carcinoma previously treated by radiotherapy with residual dysphagia. One patient required aortic valve replacement at day 4 for AI grade IV because of cusp repair failure (triangular resection). Echocardiography performed at a mean of 10 days postoperatively showed no or discrete AI in 44 patients. Late Mortality, Morbidity, and Echocardiographic Findings Clinical follow-up averaged 14.5 months (range, 1 to 34 months). Ninety percent of patients were in New York Heart Association functional class I, 10% in class II. Two patients with associated mitral valve disease were treated with anticoagulants because of atrial fibrillation. There have been no reoperations after discharge nor cases of endocarditis. There have been three late deaths. One patient died 4 months postoperatively of liver failure (chronic hepatitis), 1 patient died at 12 months postoperatively because of a cerebral hemorrhage, in the absence of anticoagulation, and 1 patient died at 14 months of an unknown cause. The 40 survivors consented to repeat echocardiography at a mean postoperative interval of 12.5 months (range, 1 to 30 months). Thirty-eight patients had no or discrete AI; 2 patients had grade II. Comment No. of Patients Comments Mortality Early 1 28 days (aspiration pneumonia) Late 3 4 months (chronic hepatitis) 12 months (cerebral hemorrhage) 14 months (unknown cause) Morbidity 1 AVR (4 days postoperatively) Aortic insufficiency Postprocedure Grade 0 1 40 Late Grade 0 1 38 Grade II 2 NYHA (grade I) 36 AVR aortic valve replacement; NYHA New York Heart Association. The mechanism of aortic regurgitation in patients who have dilatation of the aortic root but normal aortic valve leaflets has been appreciated for many years [9]. Conventional treatment for these patients consists of aortic root replacement with a composite graft incorporating a prosthetic valve. This is a successful operation, and nearly all contemporary series report a 30-day mortality of less than 5% for elective operations [4]. However, these patients may be young (mean age, 56.5 years in our series), and implantation of a mechanical valve means they require life-long anticoagulation and are at risk of the many potential complications that relate to these prostheses, such as thromboembolism and endocarditis. The operative technique of remodeling the aortic root was introduced by Sarsam and Yacoub [5] and David [6] based on a better understanding of the structure and function of the aortic root and, in particular, the contribution of the sinuses of Valsalva to valvular function [10, 11]. This operation has a major theoretical advantage over conventional root replacement in that the native aortic valve is preserved, thus avoiding implantation of a mechanical prosthesis. Previously published data show that, overall, this procedure does not have a significantly higher mortality than root replacement, and the postoperative complication rates are reassuringly low. Yacoub and colleagues have performed the procedure in 151 patients since 1979 [12]. They report no operative deaths and an early (30- day) mortality of 1.3% for elective cases and 12.2% for emergency operations. Their follow-up period for this group of patients ranges from 1 to 209 months, with a mean of 79 months. The actuarial survival at 5, 10, and 15 years is 92.4%, 87.2%, and 70.4%, respectively, and, importantly, the probability of freedom from reoperation at 5 and 10 years is 95.7% and 90.7%, respectively. No patient required anticoagulation, and there were no reported instances of endocarditis or thromboembolism. Follow-up echocardiography has shown mild or no aortic regurgitation in 93% of patients, moderate in just more than 5%, and, at the time of reporting, only 1 patient was awaiting reoperation. Echocardiography also demonstrated that the reduction in left ventricular endsystolic and end-diastolic dimensions was maintained throughout the follow-up period. Gott and colleagues [4] and David [6] have published a series of 101 cases. There were two operative deaths, both caused by cardiac failure. The remaining 99 patients have been followed up for a mean of 31 months (range, 3 to 108 months). There were five late deaths: one sudden, one caused by cerebral bleed, and three unrelated to cardiovascular disease. Sixty-seven patients have none or grade I aortic regurgitation. In this study, 28 patients had Marfan syndrome. Aortic annuloplasty was performed in 11 patients who had remodeling of all three aortic sinuses, and in 28 patients who had reimplantation of the aortic valve. There has been only one failure in a young patient who had a growth spurt of 35 cm in more than 2 years; the aortic valve became stenotic and incompetent because of a relatively small polyethylene terephthalate fiber tube. After these original publications, a report by Cochran and colleagues [7] of a series of 10 patients appeared, in which they adopted a slight technical modification of the procedure to enhance the creation of neo-sinuses, which was adopted in our series. They reported no operative mortality, and follow-up averaged 13.3 months. All patients are in New York Heart Association functional class I or II. Echocardiography at 6 months postoperatively has shown no progression of regurgitation in any patient and confirmed the optimal appearance of the neo-sinuses. In the longer term, concern still exists with respect to

1250 EL KHOURY ET AL Ann Thorac Surg AORTIC VALVE-SPARING OPERATION 2000;70:1246 50 Table 4. Echocardiographic Changes in Patients Variable Number Preoperative Postoperative Follow-up LVTDD LVTSD Annulus Sinus 23 56.3 (10.7) 52.7 (9.0) 23 37.0 (8.5) 37.4 (10.2) 23.3 (4.2) 21.9 (2.8) 22.9 (4.2) n 31 n 31 p 0.012 n 27 41.7 (10.0) 33.3 (4.2) a 34.3 (4.75) a n 29 n 27 n 24 ST J 37.7 (8.9) 25.4 (3.7) a 26.7 (4.0) a n 27 n 27 n 21 Data are expressed as mean (standard error). a p 0.01. LVTDD left ventricular diastolic diameter; LVTSD left ventricular systolic diameter; STJ sino-tubular junction. the potential for late deterioration in valve function. In the remodeling operation, although attention is given to creating neo-sinuses, the operation is performed using a noncompliant tube, which may have a detrimental effect on the function of the aortic valve because of leaflet damage. Only the follow-up of increasing numbers of patients for longer periods can truly answer the question of durability. Another concern is the use of this technique in patients with the Marfan syndrome because of the potential for ongoing annular dilatation and valvular degeneration. Our results, and those of the other proponents of this technique [13], have not identified a particular problem in this group of patients. Because of these concerns, however, it may be that inclusion of an annuloplasty for all Marfan syndrome patients becomes an accepted technique to prevent subsequent annular dilatation. Other contraindications agreed on by the proponents of this technique include patients with deformed valve leaflets or bicuspid aortic valves [4]. We were able to combine resection of a median raphe with a remodeling procedure in our series, with a good result, and currently do not view this as a contraindication to aortic remodeling, as long as a good result can be obtained with the associated leaflet procedure. Acute aortic dissection must be considered as a good indication for aortic root remodeling. Because of our experience in aortic root remodeling and because of the risk of redissection of the diseased tissue left behind, our approach to acute aortic dissection type A is quite clear: if the aortic root is unaffected, a supracoronary replacement of the dissected ascending aorta is indicated. If the aortic root is involved in the dissection or dilated, and provided that aortic cusps are normal or nearly normal, remodeling by excising all diseased tissues is indicated. The only issue is the fragility of the dissected aortic tissue where the adventitia is separated from the media. GRF glue has been used for several years to preserve the aortic root as described by Guilmet and associates [14]. However, in our experience, redissection occurred in 3 of 25 patients when aortic root had been preserved by gluing. This prompted us to favor the aortic root remodeling after gluing it to reinforce the aortic annulus and commissural areas. GRF is applied in the same manner so as to preserve aortic root, but when the glue is solidified, we scallop the aortic root. We have used this technique in the last 13 dissections, and we are very satisfied with the results. If the aortic leaflets are diseased, we do replace the aortic root with a composite graft. Endocarditis has been reported as being the most common late complication of conventional root replacement [4], but so far, no author has reported a case of endocarditis after aortic remodeling. Thromboembolism has also been reported after root replacement in the setting of inadequate anticoagulation [4], a problem eliminated by the valve-sparing procedure. Surgical reconstruction of the aortic root for patients with aortic dilatation and associated valvular incompetence has been adopted as a surgical technique in preference to root replacement by a few active proponents of this procedure. Early results are encouraging and the problems of endocarditis and thromboembolism seen in patients after conventional surgery seem to have been virtually eliminated. Long-term follow-up is now required by all groups performing this procedure to establish its durability and substantiate its purported advantages over root replacement. References 1. Weaver WF, Edwards JE, Brandeburg RO. Idiopathic dilatation of the aorta with aortic valvular insufficiency: a possible forme fruste of Marfan s syndrome. Mayo Clin Proc 1959;34: 518 22. 2. Olson LJ, Subramanian R, Edwards WD. Surgical pathology of pure aortic insufficiency: a study of 225 cases. Mayo Clin Proc 1984;59:835 41. 3. Bentall HH, de Bono A. A technique for complete replacement of the ascending aorta. Thorax 1968;23:338 9. 4. Gott VL, Laschinger JC, Cameron DE, et al. The Marfan syndrome and the cardiovascular surgeon. Eur J Cardiovasc Surg 1996;10:149 58. 5. Sarsam MAI, Yacoub M. Remodeling of the aortic annulus. J Thorac Cardiovasc Surg 1993;105:435 8. 6. David TE. Aortic root aneurysms. Remodeling or composite replacement? Ann Thorac Surg 1997;64:1564 8. 7. Cochran RP, Kunzelman KS, Eddy AC, Hofer BO, Verrier ED. Modified conduit preparation creates a pseudosinus in an aortic valve-sparing procedure for aneurysm of the ascending aorta. J Thorac Cardiovasc Surg 1995;109:1049 58. 8. Kunzelman KS, Grande KJ, David TE, Cochran RP, Verrier ED. Aortic root and valve relationships. Impact on surgical repair. J Thorac Cardiovasc Surg 1994;107:162 70. 9. Bellhouse BJ, Bellhouse F, Abbot JA, et al. Mechanism of valvular incompetence in aortic sinus dilatation. Cardiovasc Res 1973;7:490 4. 10. Brewer RJ, Deck JD, Capati B, Nolan SP. The dynamic aortic root. J Thorac Cardiovasc Surg 1976;72:413 7. 11. Thubrikar MJ, Nolan SP, Aouad J, Deck JD. Stress sharing between the sinus and leaflets of the canine aortic valve. Ann Thorac Surg 1986;42:434 40. 12. Yacoub MH, Gehle P, Chandrasekaran V, Birks EJ, Child A, Radley-Smith R. Late results of a valve-preserving operation in patients with aneurysms of the ascending aorta and root. J Thorac Cardiovasc Surg 1998;115:1080 90. 13. David TE. Aortic valve repair in patients with Marfan s syndrome and ascending aorta aneurysms due to degenerative disease. J Card Surg 1994;9(Suppl):182 7. 14. Guilmet D, Bachet J, Goudot B, Dreyfus G, Martinelli GL. Aortic dissection: anatomic types and surgical approaches. J Cardiovasc Surg 1993,34:23 32.