Composite valve graft replacement has become

A 23-Year Experience With Composite Valve Graft Replacement of the Aortic Root Karl M. Dossche, MD, Marc A. A. M. Schepens, MD, PhD, Wim J. Morshuis, MD, PhD, Aart Brutel de la Rivière, MD, PhD, Paul J. Knaepen, MD, and Freddy E. E. Vermeulen, MD Department of Cardiothoracic Surgery, Sint-Antonius Hospital, Nieuwegein, the Netherlands Background. This is a retrospective study of early and long-term results of composite valve graft replacement of the aortic root. Methods and Results. Between July 1974 and July 1997, 244 patients underwent aortic root replacement with a composite valve graft. Mean age was 54 15 years. The inclusion technique was used in 178 patients (73.0%), the open technique in 65 (26.5%), and the Cabrol II technique in 1 patient (0.5%). Hospital mortality was 7.8% (70% confidence limit, 6.1% to 9.5%). Independent determinants of hospital mortality were preoperative creatinine level more than 150 mol/l (p 0.04), prolonged cardiopulmonary bypass time (p 0.006), intraoperative technical problems (p 0.048), and year of operation (p 0.015). Follow-up was 99.6% complete, median 96 months (range, 2 to 256 months). Fifty-seven patients (25.3%; 70% confidence limit, 22.4% to 28.2%) died during follow-up. Cumulative survival at 5, 10, and 20 years was 76%, 62%, and 33%. Independent risk factors for late death were postoperative complications (p 0.027), technique for coronary reattachment (p 0.028), and concomitant aortic arch operation (p 0.01). Twenty patients (8.8%; 70% confidence limit, 7.0% to 10.6%) underwent reoperation on the aortic root. Estimated freedom from reoperation for pseudoaneurysms at 3 years was 96% in the inclusion group and 94% in the open group (p 0.236). Conclusions. Aortic root replacement with a composite valve graft can be performed with low hospital mortality and morbidity. Pseudoaneurysms did occur in the inclusion group, but also in the open group. (Ann Thorac Surg 1999;67:1070 7) 1999 by The Society of Thoracic Surgeons Composite valve graft replacement has become widely accepted as the preferred treatment for annuloaortic ectasia and other complicated aneurysmal or destructive diseases involving the aortic root [1 7]. The original wrap/inclusion technique described by Bentall and DeBono [8] in 1968 has been associated with delayed false aneurysm formation at the coronary ostial anastomoses [1, 3, 9]. Therefore, this technique has been largely abandoned in favor of an open technique [1 3, 5 7, 9] or the Cabrol II technique [10, 11]. Several studies reported favorable early results with the different techniques, with a hospital mortality between 4% and 10% [1 7, 9 11]. However, few overviews of the procedure including risk factor analysis and examination of long-term results have been published [1 4]. In this report, we present our entire experience with composite valve graft replacement of the aortic root in 244 patients during a 23-year interval. Patients and Methods Study Patients Between July 1974 and July 1997, 244 consecutive patients underwent aortic root replacement with a composite valve graft. Eight surgeons were involved during the Accepted for publication Oct 15, 1998. Address reprint requests to Dr Dossche, Department of Cardiothoracic Surgery, Sint-Antonius Hospital, Koekoekslaan 1, 3435 CM Nieuwegein, the Netherlands. entire period. Data were collected retrospectively from operative records, hospital charts, and clinic follow-up notes. The mean age of the patients was 54 15 years (range, 16 to 80 years); 165 patients (67.6%) were men. Indications are listed in Table 1. Thirty-nine (16%) operations were urgent and performed within 24 hours after onset of symptoms; 33 of them were for acute aortic type A dissection, 6 patients had an aneurysm rupture with hemodynamic shock, a tear in the aortic wall at operation, and a hemopericardium. Thirty-six patients (14.8%) had Marfan s syndrome of which 18 presented with annuloaortic ectasia and 8 with acute type A dissection. Forty-four patients (18%) had undergone 57 previous surgical cardiac interventions; in 40 of them at least one component of the aortic root was operated on previously. Significant coronary artery occlusive disease ( 70% stenosis of at least one major coronary artery) was present in 43 patients (17.6%); 17 patients (7%) had grade III or IV mitral insufficiency. Global left ventricular function was estimated from echocardiography or left ventriculography; a normal ejection fraction (EF) of 50% or more was present in 180 patients (73.8%), a moderately depressed EF between 30% and 50% in 20 patients (8.2%), and a severely depressed EF less than 30% in 2 patients (0.8%). In 42 patients (17.2%), no data were available. Five patients with acute type A dissection had neurologic symptoms preoperatively: coma in 3 patients (1.2%), paraparesis in 1 (0.4%), and left arm paresis in 1 patient 1999 by The Society of Thoracic Surgeons 0003-4975/99/$20.00 Published by Elsevier Science Inc PII S0003-4975(99)00162-9

Ann Thorac Surg DOSSCHE ET AL 1999;67:1070 7 AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1071 Table 1. Indications for Aortic Root Replacement in 244 Patients (July 1974 July 1997) Operative Procedure Patients % Primary operations 200 82.0 Degenerative aneurysm 72 29.5 Annuloaortic ectasia 68 27.9 Acute type A dissection 33 13.5 Postdissection aneurysm 17 7.0 Poststenotic dilatation 9 3.7 Aneurysm of sinus of Valsalva 1 0.4 Reoperations 44 18.0 Aortic sinus aneurysms after aortic valve replacement and/or supracoronary graft replacement of aorta for: Aortic valve disease 20 8.2 Acute type A dissection 8 3.2 Congenital aortic valve disease 6 2.5 Endocarditis 6 2.5 Other 4 1.6 Total 244 100 (0.4%). Other variables related to patient, disease, and operation are listed in the Appendix. Operative Technique All 244 patients undergoing aortic root replacement had a composite valve graft placed by means of the Bentall technique or its modifications [8 10]. In general, standard hypothermic (28 to 32 C) cardiopulmonary bypass techniques were used. Myocardial protection consisted of antegrade administration of cold, low sodium, normopotassic oxygenated cardioplegic solution and maintenance of myocardial hypothermia by continuous topical 4 C saline solution in the pericardial well [12]. From 1974 to 1994, the original inclusion/wrap procedure was used in 178 patients (73%) [8]. The Cabrol II technique was used in only 1 patient (0.5%) [10]. The original inclusion/ wrap technique was abandoned in 1994 in favor of the open or button technique [9]. The open technique [6, 7, 9] has been used since then in 65 patients (26.5%) [6, 7, 9]. Reinforcement of the coronary ostial suture lines with felt or autologous pericardium was not used in any of the patients. If aortic dissection was present, the continuity between the separated layers of the aorta was restored using gelatin-resorcinol formaldehyde biologic glue (GRF; F.I.I., Saint-Just Malmont, France) [13]. This cuff was then sewn to the graft with a continuous 4-0 or 5-0 polypropylene suture. Fifty-eight patients (23.7%) required profound hypothermic circulatory arrest to facilitate anastomosis of the composite graft to the proximal arch or to permit partial or total arch replacement. In another 44 patients (18%) patients, in whom a circulatory arrest period of more than 30 minutes was anticipated, unilateral or bilateral antegrade selective cerebral perfusion was used during circulatory arrest [14, 15]. Concomitant procedures included planned coronary artery bypass grafting in 39 patients (16%), coronary artery bypass grafting due to perioperative technical problems in 3 (1.2%), and mitral valve repair or replacement in 13 patients (5.3%). Mean cardiopulmonary bypass time was 194 71 minutes (range, 90 to 530 minutes) and the mean aortic cross-clamp time was 130 43 minutes (range, 61 to 343 minutes) for the entire cohort of patients. In 97 patients (40%), vascular tube grafts were sutured intraoperatively to various porcine xenografts (29 patients) or mechanical devices (68 patients). Composite valve grafts manufactured by Shiley Laboratories (Shiley Inc, Irvine, CA), St. Jude Medical (St. Jude Medical Inc, St. Paul, MN), Sorin (Sorin Biomedica S.P.A., Saluggia, Italy), or Carbomedics (Carbomedics Inc, Austin, TX) were used in the remaining 147 patients (60%). Knitted vascular grafts were used in 137 patients (56%) and woven vascular grafts in 107 patients (44%). Statistical Analysis Variables related to patients, disease, and operation are listed in Appendix 1. Variables omitted from the univariate and multivariate analysis because of incomplete data collection ( 5%) during the earlier years were preoperative New York Heart Association functional class, left ventricular EF, and the surgeon. The types of aortic disease were divided into three major categories: degenerative aneurysm (including chronic dissection or postdissection aneurysm), annuloaortic ectasia, and acute type A dissection. All continuous values were expressed as mean standard deviation. Differences between categorical variables were tested using a 2 test, or Fisher s exact test when appropriate; differences between continuous variables were tested using the Student s t test. Results with p values less than 0.05 were considered statistically significant. Multivariate analysis of factors of influence on hospital mortality, postoperative renal dysfunction, or central neurologic damage was performed by stepwise logistic regression analysis and the relative risk of the covariates was calculated [16]. Cumulative survival curves were made using the Kaplan-Meier product-limit method [17]. Comparison of the survival groups was made using the Tarone-Ware test [18]. The independent predictors of long-term survival were identified using the Cox proportional hazards regression analysis [19]. All computations were performed with the aid of BMDP (BMDP Software, Los Angeles, CA) and the SAS (SAS Institute, Cary, NC) statistical software packages. Results Hospital Mortality Hospital mortality rate (defined as all patients who died within the first 30 days after operation or during the initial hospitalization) was 7.8% (n 19; 70% confidence limit [CL], 6.1% to 9.5%). Four patients (1.6%; 70% CL, 0.8% to 2.4%) were considered as operative deaths. Cause of death was heart failure in 6 patients, severe central

1072 DOSSCHE ET AL Ann Thorac Surg AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1999;67:1070 7 neurologic damage in 4, rupture of a distant aneurysm in another 4, intraoperative exsanguination, intestinal ischemia, and rupture of a suture line in 1 patient each. Mortality for patients with acute type A dissection was 12.1% (70% CL, 6.5% to 17.7%), for patients without acute dissection 7.1% (70% CL, 5.4% to 8.8%) (p 0.151). Hospital mortality decreased steadily over time; in the era before 1988, the hospital mortality rate was 13.4% (13 of 97 patients; 70% CL, 10.0% to 16.8%), decreasing to 4.1% (6 of 147 patients; 70% CL, 2.5% to 6.6%) since 1988 (p 0.008). Result of the univariate analysis of risk factors is given in the Appendix. Variables entered into a stepwise logistic regression to identify risk factors for early mortality in all patients (including operative deaths) were preoperative creatinine level more than 150 mol/l, operative indication, cardiopulmonary bypass time, cross-clamp time, intraoperative technical problems, year of operation, emergency operation, and method to reattach the coronary ostia. Factors showing a significant association with hospital mortality were preoperative creatinine level more than 150 mol/l, prolonged cardiopulmonary bypass time, intraoperative technical problems, and operation before 1988. The probability of dying in the hospital was 0.9% in the absence and 62% in the presence of all incremental risk factors (Table 2). To study the additional effects of postoperative complications on hospital mortality, stepwise logistic regression was performed with exclusion of the intraoperative deaths. Variables included in the analysis were the same as in the first model, with addition of postoperative complications (central neurologic damage, need for hemodialysis, respiratory insufficiency). Preoperative creatinine level more than 150 mol/l, intraoperative technical problems, and postoperative complications affected significantly hospital outcome (Table 3). Hospital Morbidity Early ( 24 hours) reintervention for excessive bleeding was necessary in 41 patients (16.8%; 70% CL, 14.4% to 19.2%); in none of these patients was cardiopulmonary bypass required. Patients who underwent an extended Table 2. Multivariate Prediction (stepwise logistic regression analysis) of Hospital Mortality, Including Operative Deaths and Excluding the Influence of Postoperative Complications in 244 Operations a Risk Factor Log Regression Coefficient Standard Error Preoperative creatinine 1.663 0.776 0.032 150 mol/l Intraoperative technical 1.731 0.844 0.043 problems Prolonged CPB time b 1.713 0.689 0.013 Operation after 1988 1.335 0.564 0.018 Constant 3.285 0.611 0.000 b CPB cardiopulmo- a Global 2 of final model: 28.57 ( p 0.000). nary bypass time. Table 3. Multivariate Prediction (stepwise logistic regression analysis) of Hospital Mortality in 225 Operations, Excluding Operative Deaths and Including Postoperative Complications a Risk Factor Log Regression Coefficient Standard Error Preoperative creatinine 1.980 0.787 0.012 150 mol/l Intraoperative technical 2.781 0.842 0.001 problems Postoperative 1.581 0.638 0.013 complications b Constant 3.619 0.419 0.000 a Global 2 of final model: 26.97 ( p 0.000). b See text. replacement had a higher early reintervention rate than patients in whom simple aortic root replacement was performed (p 0.025). Perioperative myocardial damage (serum creatine kinase level 300 IU/L, with a creatine kinase MB isoenzyme fraction 3%) occurred in 7 patients (2.8%; 70% CL, 1.7% to 3.9%). Eight patients (3.3%; 70% Cl, 2.2% to 4.4%) had a complete heart block after the intervention and needed a permanent pacemaker. Respiratory insufficiency requiring prolonged mechanical ventilation for more than 5 days occurred in 35 patients (14.3%); 7 of them needed a tracheotomy. Postoperative stable elevated creatinine levels ( 150 mol/l) were present in 8 patients (3.3%; 70% CL, 2.2% to 4.4%). Another 10 patients (4.1%; 70% CL, 2.8% to 5.4%) required temporary hemodialysis postoperatively. Five of them died in the hospital, the others recovered without the need for long-term dialysis. Preoperative creatinine level more than 150 mol/l (p 0.013), previous aortic valve or aortic operation (p 0.029), and concomitant aortic arch operation (p 0.014) were independent determinants of postoperative need for hemodialysis by multivariate analysis. Nine patients (3.7%; 70% CL, 2.5% to 4.9%) sustained postoperative central neurologic damage: coma in 3 patients, hemiplegia in 3, brainstem dysfunction in 2, and ischemia of the basal ganglia in 1 patient. Four of them died in the hospital. All 5 patients with a neurologic symptom preoperatively were free of neurologic symptomatology postoperatively, all were patients with acute type A dissection. Multivariate analysis marked urgent operation (p 0.000) and operation before 1988 (p 0.033) as independent determinants of postoperative central neurologic damage. Technical Problems In 8 patients (3.3%; 70% CL, 2.2% to 4.4%), intraoperative problems required repeat cross-clamp of the aorta and cardioplegic arrest. In 3 patients, this was attributable to massive bleeding without obvious cause; in 2 patients problems occurred with a reattached coronary artery requiring unforeseen coronary artery bypass grafting; in another 2 patients, the initial operation consisted of aortic valve replacement and supracoronary ascending aorta

Ann Thorac Surg DOSSCHE ET AL 1999;67:1070 7 AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1073 Table 4. Causes of Late Death in 57 Patients During Follow-up (median 96 months) Causes of Late Death (N 57) Patients Prosthesis related 7 Pseudoaneurysm 3 Endocarditis early a 2 late b 2 Cerebral 8 Stroke 3 Cerebral hemorrhage 4 Intracranial malignancy 1 Cardiac 15 Heart failure 9 Arrhythmias 5 Ischemic heart disease 1 Cancer 4 Rupture of other aneurysm 9 Respiratory 6 Other 4 Unknown 4 Total 57 a 1 year after operation. b 1 year after operation. replacement. Due to tears in the native aortic tissue (supracoronary segment) and uncorrectable bleeding, a complete root replacement had to be done. Finally, in 1 patient an unforeseen coronary artery bypass grafting was necessary for low cardiac output supposedly attributable to coronary malperfusion. Late Results Late follow-up data were obtained from clinic records and direct contact with the patients and their primary physicians. All data were collected between August and September 1997 by one investigator. Follow-up of hospital survivors was 99.6% complete, only 1 patient was lost to follow-up. Median follow-up was 96 months (range, 2 to 256 months). Table 5. Multivariate Prediction (proportional hazards model) of Late Mortality in 225 Hospital Survivors a Risk Factor Log Regression Coefficient Standard Error Postoperative complications b 0.825 0.374 0.027 Technique for reattachment 2.251 1.028 0.0286 of coronary arteries Concomitant aortic arch operation 0.937 0.364 0.011 a Global 2 of final Cox model 16.281 ( p 0.001). b See text. MORTALITY RATE. There have been 57 late deaths (25.3%; 70% CL, 22.4% to 28.2%). Seven of these deaths (12.3%) were related to the composite valve graft operation. Other causes of late death are listed in Table 4. Figure 1 depicts estimated survival for the overall group of 244 patients (including hospital deaths). Survival was 88% (70% CL, 86% to 90%) at 1 year, 76% (70% CL, 73% to 79%) at 5 years, 62% (70% CL, 58% to 66%) at 10 years, and 33% (70% CL, 23% to 43%) at 20 years. Eight patient- and procedure-related variables were screened as potential risk factors for late death ( 30 days): age at operation, gender, operative indication, preoperative creatinine level, Marfan syndrome, technique used for reattachment of coronaries, extent of replacement, and postoperative complications. Not included because of more than 5% missing data were preoperative functional class, peripheral vascular disease, and left ventricular function. Postoperative complications (p 0.027), technique used for reattachment of coronary arteries (p 0.028), and concomitant aortic arch operation (p 0.01) were independent risk factors for late death, using Cox proportional hazards regression analysis (Table 5) [19]. Patients operated on for annuloaortic ectasia had a slightly better long-term survival rate compared to patients operated for degenerative aneurysm (p 0.144) or acute type A dissection (p 0.451) (Fig 2). The difference in long-term survival of Marfan and non-marfan patients was not significant (p 0.736). REOPERATIONS. Reoperation due to complications of the composite valve graft procedure was necessary in 20 patients (8.8%; 70% CL, 7.0% to 10.6%), 2 to 102 months after primary operation. Pseudoaneurysms at the aortic or coronary ostial suture lines were identified in 17 patients (7.5%). In 15 of them, the inclusion/wrap technique was used (9.2%), in 2 patients the open technique (3.3%). Fourteen of these patients underwent reoperation, 3 died of rupture of the false aneurysm (confirmed by autopsy). The 2 patients with the open technique had a pseudoaneurysm at the coronary ostial suture lines. One had Takayashu s disease, the other had early prosthetic valve endocarditis after composite valve graft replacement for acute type A dissection. Three patients underwent reoperation for failure of a porcine aortic bioprosthesis. Two other patients required reoperation Fig 1. Overall survival curve (time 0 being the operation). Dotted lines represent 70% confidence limits.

1074 DOSSCHE ET AL Ann Thorac Surg AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1999;67:1070 7 for prosthetic valve endocarditis. The composite valve graft was replaced with an aortic allograft root. One patient underwent replacement of a mechanical aortic prosthesis by a stentless porcine bioprosthesis because of anticoagulant-related complications. All but 1 patient survived reoperation. Freedom from reoperation on the aortic root for any cause was 94.8% (70% CL, 93.3% to 96.3%) at 5 years, 88.9% (70% CL, 86.7% to 92.1%) at 10 years, and 76.2% (70% CL, 68.9% to 83.5%) at 20 years. At 3 years, 96.2% (70% CL, 94.7% to 97.7%) of patients in whom the inclusion/wrap technique was used and 94.6% (70% CL, 91.6% to 97.6%) of patients in whom the open technique was used, were free from reoperation for pseudoaneurysms (p 0.236) (Fig 3). The use of knitted or woven vascular grafts did not influence pseudoaneurysm formation (p 0.643), nor did the use of homemade or commercially available composite grafts (p 0.432). OPERATIONS ON THE REMAINING AORTA. Fifteen patients (6.6%; 70% CL, 5.0% to 8.2%) underwent subsequent operation for aneurysmal disease of the remaining thoracic or abdominal aorta 3 to 132 months after initial operation. Freedom from operation on the remaining aorta was 91.2% (70% CL, 88.5% to 93.9%) at 10 years and 81.5% (70% CL, 74.4% to 88.6%) at 20 years. PROSTHETIC ENDOCARDITIS. Nine patients (4%; 70% CL, 2.7% to 5.3%) developed prosthetic valve endocarditis (4 early prosthetic valve endocarditis; 5 late prosthetic valve endocarditis). Three patients followed at our cardiology department underwent operation. The infected valved conduit was replaced by an allograft aortic root. Six patients were not presented for operation by other cardiologists and were treated medically, 5 of them died. Freedom from prosthetic valve endocarditis was 92.8% (70% CL, 90.2% to 95.4%) at 10 years. Type of vascular prosthesis (woven versus knitted, homemade versus commercially available composite graft) was not significant (p 0.375). Fig 3. Freedom from reoperation for pseudoaneurysm for the inclusion/wrap technique and the open technique. THROMBOEMBOLISM. Thromboembolic events occurred in 5 patients (2.2%; 70% CL, 1.2% to 3.2%). Three patients had a mechanical prosthesis, 2 a porcine xenograft. Four of them had a stroke, 1 had a mesenterial thrombosis. All patients died. The rate of thromboembolic events was 0.39 events/100 patient-years (5 events/1,298 patientyears of follow-up). Freedom from thromboembolic events was 97.7% (70% CL, 96.0% to 99.4%) at 10 years, and 95.7% (70% CL, 93.1% to 98.3%) at 20 years. Five other patients experienced transient visual disturbances, including diplopia, focal blind spots that resolved within minutes, and blurring of vision. Because of the transient and reversible character, these disturbances were not considered as thromboembolic events. ANTICOAGULANT-RELATED COMPLICATIONS. Sixteen patients (7%; 70% CL, 5.3% to 8.7%) had complications related to anticoagulant therapy. Five of them died from cerebral hemorrhage, 1 from severe gastrointestinal bleeding. Ten patients had nonfatal anticoagulant-related complications including retroperitoneal hematoma and intestinal bleeding. Freedom from anticoagulant-related complications was 87.5% (70% CL, 84.2% to 90.8%) at 10 years and 84.6% (70% CL, 80.2% to 89.0%) at 20 years. Fig 2. Survival curve for different pathologic conditions (including hospital deaths). Comment The natural history of patients with thoracic aneurysms is well established [20]. Composite valve graft replacement has been widely accepted as the preferred treatment for atherosclerotic aneurysms or dissections involving the aortic root. The results of this study confirm the observations of other investigators that composite valve graft replacement can be performed with low operative risk [1 7]. Period of operation, preoperative elevated serum creatinine level, and intraoperative technical problems were strong independent risk factors for hospital mortality. Since 1988 aortic root procedures are performed by a limited number of surgeons at our department. Although a reduction in hospital mortality was observed throughout the entire period of this study, this reduction was

Ann Thorac Surg DOSSCHE ET AL 1999;67:1070 7 AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1075 more consistent since 1988. Other factors, such as the availability of zero porosity vascular grafts, better cardiopulmonary bypass techniques, and improved anesthesiologic management, may have enhanced this reduction in mortality during the long period but are difficult to quantify. The technique for reattachment of the coronary arteries and acute type A dissection did not emerge as a predictor of hospital death. This, in general, is in agreement with other recent reports [1 4, 7]. Preoperative elevated serum creatinine level requires additional attention. Higher perfusion pressures during cardiopulmonary bypass, mild inotropic support, and prompt recognition of hemodynamic instability during the perioperative period all play a major role. Based on reports on late complications of the inclusion/wrap technique, we abandoned this technique in favor of the open technique [3, 9]. We now use the open technique in every pathologic condition. We do not share the opinion of Svensson and associates [1] that the button technique is technically more demanding and more time consuming. In our experience, detachment and mobilization of the coronary arteries was never a problem and could easily be performed within a few minutes. Especially in cases of a fragile aortic wall or acute aortic dissection, where it may be difficult to take the whole thickness of the aortic wall into the suture, isolation of the coronary artery buttons offers the best opportunity for tension-free and thus trouble-free anastomosis. Technique used for reattachment of the coronary arteries emerged as a predictor of late death. Despite a significant p value, our results do not present valid evidence that the new technique is better than the older inclusion/wrap approach. Patients operated on more recently may have had a better prognosis at the same time that the more recently used reattachment technique had a better prognosis. There can be no simple way to answer the above question unless there would be a period of overlap during which both approaches were used with a large enough number of patients to produce statistically meaningful conclusions. With this type of retrospective analysis, this is not possible. A randomized prospective study is the only valid option. During follow-up, we have observed pseudoaneurysm formation at each of the different suture lines in 15 of 163 hospital survivors with the inclusion/wrap technique. In other series, the incidence of pseudoaneurysms with this technique has ranged from 8% to 15% [3, 21, 22]. In 2 patients with the open technique, pseudoaneurysms were identified at the coronary ostial suture lines. In a series of 110 consecutive patients having composite aortic root replacement with direct coronary artery reimplantation, Hilgenberg and associates [7] reported no reoperation for coronary ostial pseudoaneurysms. Depending on the friability of the tissues, they used either a continuous suture or interrupted mattress sutures with small felt or a combination of the two methods to reattach the coronary artery buttons. Miller and Mitchell [5] describe the use of a life-saver or doughnut of Teflon felt or autologous pericardium (tanned in 0.625% glutaraldehyde solution for 10 to 15 minutes) placed around the coronary ostium on the adventitional aspect to prevent tearing of tissue. We did not use any reinforcement of the coronary ostial sutures in any of the patients with the open technique. In our report, differences in the incidence of reoperation for pseudoaneurysms between the inclusion/wrap and open technique were not statistically significant at 3 years. In the series of Kouchoukos and colleagues [3], the reoperation rate for pseudoaneurysms in both groups was still not statistically significant at 8 years, although none of the patients in the open group needed reoperation for pseudoaneurysm formation. Thirty-three patients (13.5%) with acute type A dissection underwent aortic root replacement. This represents 16% of all patients operated on for acute type A dissection during the same period. Although aortic valvepreserving operation can be performed in the majority of patients with acute type A dissection and aortic valve involvement [23, 24], we agree with Jex and colleagues [25] that in patients with acute type A dissection and annuloaortic ectasia and for selected patients with severe destruction of the proximal aorta below the level of the valve commissures, a complete root replacement is the only option. Although some researchers have found long-term survival to be statistically less favorable after root replacement for acute dissection, this was not a predictor of late mortality in our series [26]. Infection of the composite valve graft requires an aggressive surgical approach. It is our standard policy to replace an infected prosthesis or composite graft with an aortic allograft root. Conservative treatment failed to eradicate infection in all but 1 patient treated as such in our series. Finally, some limitations inherent to the current investigation should be mentioned. First, many factors changed during this long time interval that could not be accounted for with the multivariate statistical techniques. Second, due to incomplete data collection during the earlier years of the study, some important variables such as preoperative functional class, peripheral vascular disease, and left ventricular function were not included in the analysis. Therefore, the influence of these variables on hospital and late mortality could not be studied. In summary, composite valve graft replacement of the aortic root can be performed with low mortality and morbidity. At present, we believe that the open technique offers some technical advantages and should be used whenever possible, although we could not confirm the absence of pseuodoaneurysms in the open technique, in contrast to most other reports on this issue. References 1. Svensson LG, Crawford ES, Hess KR, Coselli JS, Safi HJ. Composite valve graft replacement of the proximal aorta: comparison of techniques in 348 patients. Ann Thorac Surg 1992;54:427 39. 2. Gott VL, Gillinov AM, Pyeritz RE, et al. Aortic root replacement. Risk factor analysis of a seventeen-year experience. J Thorac Cardiovasc Surg 1995;109:536 45.

1076 DOSSCHE ET AL Ann Thorac Surg AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1999;67:1070 7 3. Kouchoukos NT, Wareing TH, Murphy SF, Perrillo JB. Sixteen-year experience with aortic root replacement. Result of 172 operations. Ann Surg 1991;214:308 20. 4. Lewis CTP, Cooley DA, Murphy MC, Talledo O, Vega D. Surgical repair of aortic root aneurysms in 280 patients. Ann Thorac Surg 1992;53:38 46. 5. Miller DC, Mitchell RS. Composite aortic valve replacement and graft replacement of the ascending aorta plus coronary ostial reimplantation: how I do it. Semin Thorac Cardiovasc Surg 1993;5:74 83. 6. Yun KL, Miller DC. Ascending aortic aneurysm and aortic valve disease: what is the most optimal surgical technique? Semin Thorac Cardiovasc Surg 1997;9:233 45. 7. Hilgenberg AD, Akins CW, Logan DL, et al. Composite aortic root replacement with direct coronary artery implantation. Ann Thorac Surg 1996;62:1090 5. 8. Bentall HH, DeBono A. A technique for complete replacement of the ascending aorta. Thorax 1968;23:338 9. 9. Kouchoukos NT, Marshall WG, Wedige-Stecker TA. Elevenyear experience with composite graft replacement of the ascending aorta and aortic valve. J Thorac Cardiovasc Surg 1986;92:691 705. 10. Cabrol C, Pavie A, Gandjbakhch I, et al. Complete replacement of the ascending aorta with reimplantation of the coronary arteries. New surgical approach. J Thorac Cardiovasc Surg 1981;81:309 15. 11. Cabrol C, Pavie A, Mesnildrey P, et al. Long-term results with total replacement of the ascending aorta with reimplantation of the coronary arteries. J Thorac Cardiovasc Surg 1986;91:17 25. 12. Bleese N, Doring V, Kalmar P, Krebber HJ, Pokar H, Rodewald G. Clinical application of cardioplegia in aortic crossclamping periods longer than 150 minutes. J Thorac Cardiovasc Surg 1979;27:390 9. 13. Guilmet D, Bachet J, Goudot B, et al. Use of biological glue in acute aortic dissection. Preliminary clinical results with a new surgical technique. J Thorac Cardiovasc Surg 1979;77: 516 21. 14. Frist WH, Baldwin JC, Starnes VA, et al. A reconsideration of cerebral perfusion in aortic arch replacement. Ann Thorac Surg 1986;42:273 81. 15. Kazui T, Kimura N, Yamada O, Komatsu S. Surgical outcome of aortic arch aneurysm using selective cerebral perfusion. Ann Thorac Surg 1994;57:904 11. 16. Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967;54:167 79. 17. Kaplan EL, Meier P. Non-parametric estimation for incomplete observations. J Am Stat Assoc 1958;53:457 81. 18. Tarone RE, Ware J. On distribution-free tests for equality of survival distributions. Biometrika 1977;64:156 60. 19. Cox DR. Regression models and life-tables. J R Stat Soc B 1972:34:187 220. 20. Pressler VBA, McNamara JJ. Thoracic aortic aneurysm. Natural history and treatment. J Thorac Cardiovasc Surg 1980; 79:489 98. 21. Taniguchi K, Nakano S, Matsuda H, et al. Long-term survival and complication after composite graft replacement for ascending aortic regurgitation. Circulation 1991;84(suppl 3): 31 9. 22. Aoyagi S, Kosuga K, Akashi H, Oryagi A, Oishi K. Aortic root replacement with a composite graft: results of 69 operations in 66 patients. Ann Thorac Surg 1994;58:1469 75. 23. Weinschelbaum EE, Schamun C, Caramutti V, Tacchi H, Cors J, Favaloro RG. Surgical treatment of acute type A dissecting aneurysm, with preservation of the native aortic valve and use of biologic glue. J Thorac Cardiovasc Surg 1992;103:369 74. 24. Fann JL, Glower DD, Miller DC, et al. Preservation of the aortic valve in patients with type A dissection complicated by aortic valvular regurgitation. J Thorac Cardiovasc Surg 1991;102:62 75. 25. Jex RK, Schaff HV, Piehler JM, et al. Repair of ascending aortic dissection. J Thorac Cardiovasc Surg 1987;93:375 84. 26. Lytle BW, Mahfood SS, Cosgrove DM, Loop FD. Replacement of the ascending aorta. Early and late results. J Thorac Cardiovasc Surg 1990;99:651 8. Appendix. Univariate Analysis of Patient-, Disease-, and Operation-related Variables Variable Patients Hospital Mortality n % (univar.) Sex Male 165 11 6.6 Female 79 8 10.1 0.346 Age (y) 55 108 7 6.5 55 136 12 8.8 0.459 Preop central neurologic deficit Yes 5 0 0.0 No 239 19 7.9 0.512 Preop creatinine ( mol/l) 150 233 15 6.4 150 11 4 36.0 0.000 Coronary artery disease No 201 14 6.9 Yes 43 5 11.6 Aortic incompetence ( grade II/IV) Yes 182 13 7.2 No 62 6 9.6 0.583 Marfan s syndrome Yes 36 3 8.3 No 208 16 7.7 0.895 Operative indication Degenerative Aneurysm 142 14 9.8 Annuloaortic Ectasia 68 1 1.5 0.068 Acute type A Dissection 34 4 11.7 Previous aortic valve or aortic operation Yes 44 4 9.1 No 200 15 7.6 0.722 Emergency Yes 39 6 15.4 No 205 13 6.4 0.061 Pump time (min) 180 118 3 2.5 180 126 16 12.7 0.003 Myocardial ischemic time (min) 120 119 4 3.3 120 125 15 12.0 0.012

Ann Thorac Surg DOSSCHE ET AL 1999;67:1070 7 AORTIC ROOT REPLACMENT WITH COMPOSITE GRAFT 1077 Hospital Mortality Hospital Mortality Variable Patients n % (univar.) Variable Patients n % (univar.) Operative technique Distal clamp 142 9 6.3 DHCA 58 6 10.3 0.418 ASCP 44 4 9.0 Extend of replacement Ascending 183 12 6.5 Ascending and 61 7 11.5 0.267 arch Concurrent valve Yes 13 2 15.3 No 231 17 7.3 0.293 Concurrent CABG Yes 42 5 11.9 No 202 14 6.9 0.275 Method to reattach coronary ostia Bentall 178 15 8.4 Open 65 4 6.1 Cabrol II 1 0 0.0 0.807 Year of operation 1988 98 13 13.4 1988 146 6 4.1 0.008 Technical problems Yes 8 4 50.0 No 236 15 6.3 0.000 Postop myocardial infarction Yes 7 1 14.2 No 233 14 6.0 0.000 Postop respiratory insufficiency Yes 35 8 22.8 No 205 7 3.4 0.085 Postop bleeding Yes 41 5 12.2 No 199 10 5.0 0.090 Postop neurologic deficit Yes 9 4 44.4 No 231 11 4.7 0.000 Postop hemodialysis Yes 10 5 50.0 No 230 10 4.3 0.000 ASCP antegrade selective cerebral perfusion; CABG coronary artery bypass grafting; DHCA deep hypothermic circulatory arrest.