Comparison of Human Tissues and Mechanical Prostheses for Aortic Valve Replacement in Children
Background Aortic valve replacement in children is problematic because of complications of mechanical valves and uncertain outcomes associated with human valves. The results of pediatric aortic valve replacements over 5 years were reviewed.
Methods and Results Mechanical valves were used exclusively during the first part of this series (n=26). Thereafter, 25 consecutive aortic valve replacements were performed with autografts (n=19) or allografts (n=6). Allografts were used for Marfan’s syndrome patients or those with unusable pulmonary valves. Among autograft/allograft recipients, 16 patients underwent 27 prior operations. In the mechanical group, 18 patients underwent 19 previous operations. Three patients in each group underwent a previous mechanical aortic valve replacement. Operative complications included two mild strokes and one pacemaker in the autograft/allograft group and three deaths and two pacemakers in the mechanical group. One autograft recipient required reoperation for pulmonary allograft stenosis. In the mechanical group, late complications included six cases of nonstructural degeneration and two cases of endocarditis, with three reoperations. Reoperation-free survival was 96% at 2 years in the autograft/allograft group and 80% at 2 years and 75% at 3 years in the mechanical group. Event-free survival was 96% at 2 years in the autograft/allograft group compared with 67% at 2 years and 49% at 3 years in the mechanical group (P<.05).
Conclusions The frequency of reoperations for mechanical aortic valve replacement has been surprisingly high. Aortic valve replacement in children with only autografts or allografts achieves good early results.
Aortic valve replacement in children has been a procedure of last resort, performed only when multiple other treatments—including medication, catheter intervention, and reconstructive surgery—have failed or proven impractical. One of the main reasons that aortic valve replacement may be delayed or not performed has been the unsatisfactory performance of valve substitutes. Porcine and pericardial bioprostheses used in children are notorious for their rapid calcification and requirement for early reoperation. Mechanical prostheses, while undergoing rare structural degeneration, are susceptible to other complications, including hemolysis, infection, and thromboembolism.1 2 The use of anticoagulation in children presents its own risks and problems.3 4 5 6 7 8
The difficulties presented by prosthetic and bioprosthetic valves in pediatric patients have motivated a number of institutions to use human valve substitutes. This approach to aortic valve replacement makes use of aortic or pulmonary valve allografts and the pulmonary valve autograft, or the Ross procedure. These operations have been performed in adults requiring aortic valve replacement for decades but only recently have found widespread acceptance in children. Recent results of human valve use for aortic valve replacement in children have been highly favorable.9 10 11 12 An obvious question raised by these observations is how the results of aortic valve replacement with autografts and allografts in children compare with those of mechanical valve prostheses. To address this question, the current institutional policy of using only human tissues for aortic valve replacement in children was reviewed. The results of this approach were analyzed and compared with the previous practice in this hospital of the exclusive use of mechanical prosthetic valves.
All aortic valve replacements performed at Children’s Hospital and Medical Center from January 1, 1991, through December 31, 1995, were reviewed. Demographic information, operative details, and late postoperative data were obtained from medical records. Late follow-up was established from clinical records, including echocardiographic data and catheterization findings when available, and from personal communication with the patients, families, and referring physicians. The design of this study was approved by the Institutional Review Board of Children’s Hospital and Medical Center.
A total of 51 operations were performed in 49 patients by three different surgeons. Two patients each underwent two aortic valve replacements during this period, and each operation was included as a separate event. A mechanical aortic valve replacement was performed in 26 patients by two surgeons, and a human valve replacement was performed in 25 patients by a third surgeon. Table 1⇓ lists the previous operations performed on these patients. Three patients in each group had undergone aortic valve replacement with mechanical valves. Only a third of the patients had undergone no previous operation. In the autograft/allograft group, 4 of the 9 patients who had not undergone a previous surgical procedure had received balloon valvotomy for aortic stenosis. None of these patients required urgent or emergency valve replacement. Indications for operation were classified by the predominant valvar pathology. Among the mechanical valve recipients, 19 had aortic valve stenosis as the primary indication for operation, 6 had aortic valve insufficiency, and 1 had endocarditis. This was similar to the distribution of pathology among the human valve recipients: 16 had aortic stenosis, 8 had aortic insufficiency, and 1 had endocarditis.
During the first portion of this series ending in 1994, mechanical aortic valves were used exclusively. Twenty-two of the valves were manufactured by St Jude Medical (St Paul, Minn) and three by CarboMedics (Austin, Tex). The three CarboMedics valves were all 18 mm. Of the St. Jude valves, 1 was 17 mm, 5 were 19 mm, 6 were 21 mm, 6 were 23 mm, 3 were 25 mm, and 2 were 27 mm. Ages of these patients ranged from 2 to 18 years (mean, 11.6 years), and weights were 12 to 110 kg (mean, 46.4 kg). There were 17 operations performed on boys and 9 on girls. All patients were anticoagulated with warfarin, with a goal of maintaining the patient’s international normalized ratio between 2.5 and 3.5.
Beginning in 1994, 25 consecutive operations were performed with pulmonary valve autografts (Ross procedure, n=19) or aortic or pulmonary valve allografts (n=6). The allografts used for aortic valve replacement, four aortic valves and two pulmonary valves, were cryopreserved and obtained from CryoLife, Inc (Marietta, Ga) or the Northwest Tissue Bank (Seattle, Wash). Among the 19 patients undergoing the Ross procedure, right ventricular outflow reconstruction was performed with a pulmonary allograft in 18 patients and with an aortic allograft in 1. All patients undergoing implantation of a human valve received a pulmonary valve autograft whenever possible. Indications for allograft use were Marfan’s syndrome in 2 patients, congenital absence of the pulmonary valve in 3 patients (2 after truncus repair and 1 after repair of complete AV septal defect and tetralogy of Fallot), and inability to use a pulmonary valve in 1 patient with a previous Konno procedure whose resulting left ventricular outflow was excessively large for the pulmonary valve autograft. The human valve recipients were 2 to 18 years old (mean, 11.6 years) and weighed 11 to 66 kg (mean, 46.4 kg). There were 21 boys and 4 girls. There were no significant differences in weights, ages, or sex distribution between the human valve group and the mechanical valve group.
All autograft and allograft procedures were performed as root replacements with implantation of the coronary arteries on the graft by use of standard techniques. Moderate systemic hypothermia (26°C) was used with intermittent cold blood/potassium cardioplegia administered antegrade and retrograde. For three patients undergoing a pulmonary autograft, an aortoventriculoplasty (Konno) was required. Reconstruction of the right ventricular outflow tract was performed with cryopreserved pulmonary valve allografts in all but one patient, when size limitations required the use of an aortic valve allograft. Right-sided reconstruction was performed during myocardial reperfusion and rewarming to limit ischemic injury.
All autograft and allograft recipients underwent early postoperative echocardiographic studies of the neoaortic valves before hospital discharge, and all patients received follow-up echocardiograms as outpatients. Left ventricular outflow tract maximum velocity (LVOT Vmax) was recorded, and valve insufficiency was graded on a 0 to 4+ scale.
For each group, mean values, SDs, and 70% confidence limits (CLs) were calculated. Standard definitions were used for reporting postoperative valve-related complications.13 Comparisons of echocardiographic studies were performed by use of Student’s t test for paired samples and χ2 analysis. Survival analysis was performed with the S-plus statistical package. A pair of Cox curves, one for each group, was fit for the following analysis outcomes: survival, survival free of operation, and survival free of all valve-related complications. The Cox proportional hazards regression included covariates for age, weight, valve replacement type, and number of previous operations. The proportional hazards assumption was checked by analysis of residual plots. A value of P<.05 was considered statistically significant.
There were no deaths in the autograft/allograft group (0%; 70% CL, 0% to 7.4%) and three deaths in the mechanical valve group (11.5%; 70% CL, 5.1% to 21.9%). Two of the deaths occurred intraoperatively and resulted from technical problems. The third death occurred in a patient with fungal endocarditis who died on postoperative day 11 because of a cerebral hemorrhage. There were two strokes in the autograft/allograft group. Both of these patients had had previous mechanical valves, and both patients had previous histories of stroke. In both patients, the strokes were mild and left them with mild residua, and the patients have responded to rehabilitation. Three patients developed complete heart block after valve replacement and required insertion of permanent pacemakers; two of these patients had received a mechanical valve, and one had received an autograft.
Follow-up was 100% complete in the autograft/allograft group. Follow-up duration ranged from 3 to 24 months, with a mean follow-up of 13 months. There was one late complication in the this group (4%; 70% CL, 0.5% to 13.0%). This occurred in a patient who had undergone a Ross procedure after a previous mechanical aortic valve replacement. The patient’s pulmonary valve allograft used in right ventricular outflow reconstruction became stenotic in its midportion, with a systolic pressure gradient measured at catheterization of 69 mm Hg. The patient underwent reoperation for pulmonary allograft replacement 6 months after the autograft procedure. This patient again developed a moderate recurrent stenosis in the midportion of the allograft, which was partially relieved by balloon dilatation. All aortic valves, both autografts and allografts, and all remaining right-sided allograft valves have shown excellent function, with no stenosis and little or no valvar insufficiency. Some allograft valves, in both the aortic and pulmonary positions, have some radiographic evidence of calcification in the arterial wall, although there has not been any impairment of valve leaflet motion in these grafts. In the autograft/allograft group, there were no late deaths and no cases of structural neoaortic valve failure, endocarditis, hemolysis, or thromboembolism.
Echocardiography was performed in the immediate postoperative period (range, 2 to 11 days after surgery; mean, 4 days) in all 25 patients in the autograft/allograft group. Subsequent postoperative echocardiographic examinations were performed 1 to 20 months after surgery (mean, 8.5 months). The LVOT Vmax at the immediate postoperative study ranged from 0.8 to 3.0 m/s (mean, 1.9 m/s). The LVOT Vmax at the latest postoperative study ranged from 0.8 to 2.8 m/s (mean, 1.4 m/s). Comparison of the immediate and later postoperative LVOT Vmax among the 24 patients with complete velocity measurements made at both examinations demonstrated a significant decrease in LVOT Vmax from 1.8±0.6 to 1.4±0.4 m/s (P=.003). At the early postoperative study, insufficiency of the neoaortic valve was graded 0 or trivial in 15, 1+ in 9, 2+ in 1, and 3+ or 4+ in 0 patients. At the most recent postoperative study, insufficiency of the neoartic valve was graded 0 or trivial in 12, 1+ in 12, 2+ in 1, and 3+ or 4+ in 0 patients. χ2 Analysis demonstrated no significant differences in the early and later echocardiographic grading of valve insufficiency.
Follow-up in the mechanical valve group was 94% complete, with 2 patients lost to follow-up. Follow-up ranged from 1 to 63 months, with a mean follow-up of 33 months. Late complications occurred in 8 patients who had received mechanical valves (34.8%; 70% CL, 23.5% to 47.8%). Five of these complications were a result of pannus formation in the subvalvar region of the left ventricle. All of these patients exhibited substantial increases in left ventricular outflow obstruction by echocardiographic studies. Maximal systolic velocities across the left ventricular outflow tract ranged from 3.3 to 5.0 m/s. Progressive obstruction with velocities >4.0 m/s and increasing left ventricular hypertrophy were evident in 1 of these patients, who required reoperation and additional aortic valve replacement. In this patient, the mechanical valve had no abnormality in its structure or function; rather, the exuberant proliferation of subaortic fibrous tissue appeared to be responsible for the stenosis. The other 4 patients who have not undergone an additional operation are being followed closely and may yet require another valve replacement. Two cases of endocarditis occurred after mechanical aortic valve replacement. One was successfully treated medically, whereas the other required another valve replacement. One patient required reoperation for severe hemolysis caused by a paravalvular leak that required multiple transfusions before reoperation. This patient underwent repair of the leak and has since done well with his original prosthetic valve. In the mechanical valve group, there were no late deaths and no cases of structural valve failure, thromboembolism, or anticoagulant-related hemorrhage.
Survival outcomes are displayed in Figs 1 through 3⇓⇓⇓. All deaths were perioperative, and there was no significant difference in overall survival between the two groups. Examination of reoperation-free survival showed a trend toward improved results in the autograft/allograft group, but this did not reach statistical significance. In autograft/allograft group, reoperation-free survival was 96% at 2 years. In the mechanical valve group, reoperation-free survival was 80% at 2 years and 75% at 3 years and beyond. Table 2⇓ shows a multivariate analysis of survival free of all valve-related events with a Cox proportional hazards model. By single variable analysis, there was a significant difference between the two groups (P=.02). The relative risk in the mechanical group was 6.05 times (95% confidence interval, 1.3 to 28.7) greater than in the autograft/allograft group. Performance of a likelihood ratio test was performed as a test of the hypothesis that all variables in the model have zero coefficient, ie, have no effect on the hazard rate. There remained a statistically significant difference between the groups (P=.0438). The overall model χ2 result was statistically significant (P<.05). When the variable “year of operation” was added to this model, the results remained the same, and the new variable was not statistically significant.
This investigation demonstrated that aortic valve replacement in children can be performed safely and effectively with the routine use of the pulmonary autograft or aortic or pulmonary allograft. This departure from the established policy in this institution of routine mechanical valve use was made largely because of the problems with mechanical valves in children. Despite the consistent results obtained with mechanical valves in adults, pediatric patients pose considerably more difficult challenges in long-term follow-up.
The record of aortic valve replacement with mechanical valves in children is characterized by good early results but multiple late complications.2 14 15 16 In effect, implantation of a mechanical valve is tantamount to imposing on the patient the new burden of “prosthetic valve disease.” As in the adult population, risks of endocarditis, hemolysis, and rare but catastrophic structural failure are present. Problems with thrombosis, embolism, and anticoagulant-related complications may be particularly difficult in children. Compliance, intrinsic differences in pediatric metabolism, and issues of lifestyle limit the control of the pediatrician and pediatric cardiologist caring for the young mechanical valve recipient. For this reason, several trials have been conducted to evaluate the use of antiplatelet agents as the primary means of prophylaxis. The results of these efforts have been mixed. Some institutions have reported good results with mechanical valves in children without warfarin anticoagulation,3 4 whereas others have had a high incidence of thromboembolic complications when warfarin was not used.1 5 6 One large center was initially impressed by positive results without anticoagulation7 but subsequently found the late thromboembolic rate to be disappointingly high.8 Although it is encouraging that the mechanical valve recipients in the present series had no hemorrhagic or thromboembolic complications, it would be hazardous to conclude that this group of patients will remain free of these problems over the long term.
It is noteworthy that the most common complication in this group of mechanical valve recipients was the emergence of subvalvar stenosis. This was the major factor in the difference between the two groups with respect to event-free survival. Although this complication may also arise because growth of the heart exceeds the capacity of the valve to function appropriately, all such cases in this series were attributed to pannus formation in the subvalvar area. The cause of this late complication is unknown, and its frequency is not discussed in the available literature. Nor is it clear whether avoidance of this problem is possible. The formation of this material may be overlooked if a high outflow tract gradient is attributed solely to growth of the heart. In our experience, the development of such a subvalvar fibrous formation is not exclusive to pediatric patients, but it may be more common in this population. Thus, as this study shows, it is likely that young recipients of mechanical aortic valves are at considerable risk for reoperation.16
The Ross procedure has more recently become well established as safe and effective in children. The first report focusing on pediatric patients undergoing this procedure described four deaths in 34 patients from 1967 through 1988, with all deaths occurring before 1971.9 At 15 years, actuarial survival was 77%, and actuarial freedom from reoperation on the left ventricular outflow tract for operative survivors was 74%. These investigators speculated that growth of the autograft would be an important issue for future success of this procedure but did demonstrate such growth in their patients. Santangelo and associates11 provided the first unequivocal demonstration of growth of the pulmonary autograft. These investigators showed that beyond 1 year after the Ross procedure, autograft diameter increased by 3 to 5 mm, depending on the site of measurement and the method of implantation. This increase was not accompanied by insufficiency of the valve, implying that true growth of the leaflet structures occurs. A subsequent study by Elkins et al12 showed that growth of the autograft paralleled somatic growth. Although autografts implanted as root replacements showed increased diameter above the normal range, this was not accompanied by valve insufficiency. The benefits of the Ross procedure have been extended to smaller children, infants, and neonates.17 18 There were no valve replacements performed in children younger than 2 years of age during this study period. The absence of smaller children and infants undergoing valve replacement is largely attributable to the success and low rate of complications associated with balloon valvotomy at this institution and the aggressive use of aortic valve repair techniques as an alternative to valve replacement.
The results in children of aortic valve replacement with an aortic allograft compared with the Ross procedure have been described by Gerosa and colleagues.10 This, the largest pediatric study of its kind, found no significant differences between the two groups with respect to operative mortality; late mortality; or freedom from reoperation, endocarditis, and all complications. While recognizing that their results suggested the acceptability of either valve for aortic valve replacement, the authors regarded the finite incidence of primary tissue failure of the allografts as a strong argument for autografts, which had no such failures in their series. This recommendation appears to be sound. Nevertheless, it is certain that a proportion of the pediatric population requiring aortic valve replacement will not have a usable pulmonary valve. In most cases, this will be a consequence of congenital anatomic abnormality. In other cases, prior surgical interventions may compromise the pulmonary valve. The use of the pulmonary autograft for patient’s with Marfan’s disease and aortic valve insufficiency is questionable because of the diffuse nature of the connective tissue abnormality in these patients. Because the autograft of patients without this condition is known to undergo dilatation,12 it would appear prohibitively dangerous to use the pulmonary valve of a Marfan’s patient. The autograft would likely exhibit a similar dilatation and insufficiency, possibly at an accelerated rate, although the empirical evidence of this is lacking. Thus, the aortic or pulmonary valve allograft must remain an option for the child with irreparable aortic valve pathology and an unusable pulmonary valve.
The question of tissue valve durability will strongly affect the decision of cardiac surgeons and cardiologists to undertake a policy of human valve implantation such as the one developed at this institution. Clearly, the pioneering efforts in this area have demonstrated a high degree of structural integrity with longer periods of follow-up. Gerosa et al9 observed 77% actuarial survival at 16 years, with 74% freedom from reoperation on the left ventricular outflow tract and 80% freedom from reoperation on the right ventricular outflow tract.
The comparison described in this report has certain limitations. There was no randomization of the patients, who were operated on in a sequential rather than a contemporaneous fashion, and the surgeons performing the operations were different. This series therefore reflects a complete change in approach to the problem of severe aortic valve disease in children. Although lacking some of the opportunities for comparison inherent in a tightly controlled study, this investigation does demonstrate that this complete change to uniform human valve use can be successfully accomplished.
In summary, the routine use of autografts and allografts can be carried out with good immediate results. Moreover, even with a short period of follow-up, the advantages over mechanical valves begin to emerge. Perhaps the main argument for the routine application of autografts and allografts in children is one that does not lend itself easily to statistical analysis: the quality of life enjoyed by these patients. The freedom from anticoagulant therapy, frequent blood testing, and risk of hemorrhage allows a more normal childhood that is not possible with a mechanical valve. This advantage of human valves will become even more striking as female patients approach child-bearing age and make decisions regarding pregnancy. The medical and lifestyle advantages thus justify the policy of this institution to abandon the mechanical valve for pediatric aortic valve replacement. This policy will be continued, and the longer-term results will be followed carefully.
- Received August 26, 1996.
- Revision received February 3, 1997.
- Accepted February 10, 1997.
- Copyright © 1997 by American Heart Association
Verrier ED, Tranbaugh RF, Soifer SJ, Yee ES, Turley K, Ebert PA. Aspirin anticoagulation in children with mechanical aortic valves. J Thorac Cardiovasc Surg. 1986;92:12013-1020.
Cabalka AK, Emery RW, Petersen RJ, Helseth HK, Jakkula M, Arom KV, Nicoloff DM. Long-term follow-up of the St Jude Medical prosthesis in pediatric patients. Ann Thorac Surg. 1995;60:S618-S623.
Calhoun JH, Bolton JWR. Ross/Konno procedure for critical aortic stenosis in infancy. Ann Thorac Surg. 1995;60:S597-S599.
Reddy VM, Rajasinghe HA, McElhinney DB, van Son JAM, Black MD, Silverman NH, Hanley FL. Extending the limits of the Ross procedure. Ann Thorac Surg. 1995;60:S600-S603.