Prolongation of RV-PA Conduit Life Span by Percutaneous Stent Implantation
Background Right ventricle–to–pulmonary artery (RV-PA) homografts and bioprosthetic conduits are commonly used to palliate various types of complex congenital heart disease. These conduits frequently develop progressive obstruction and require surgical replacement. This report reviews our experience implanting balloon-expandable stents to relieve conduit obstruction and delay reoperation.
Methods and Results A retrospective review identified 44 patients who underwent placement of 48 stents in obstructed RV-PA conduits. Median patient age was 6.9 years (range, 7 months to 30 years), and median follow-up time was 14.2 months (range, 0 to 48 months). Stent implantation initially decreased the RV-PA pressure gradient from 61.0±16.9 to 29.7±11.9 mm Hg (P≤.001) and the right ventricular–to–systemic arterial pressure ratio from 0.92±0.17 to 0.63±0.20 (P≤.001). The diameter of the stenotic region expanded from 9.3±3.5 to 12.3±3.3 mm in the anteroposterior view (P≤.001) and from 6.6±2.9 to 10.9±2.5 mm in the lateral view (P≤.001). During the follow-up period, 2 patients had their stents redilated, 7 had additional conduit stents deployed, and 14 underwent surgical replacement of their conduits. Actuarial freedom from conduit reoperation was 65% at 30 months postprocedure. Seven patients were found to have fractured stents on follow-up, suggesting an important role for external compressive forces in conduit failure. Recatheterization in 16 patients a median of 11.8 months (3 to 48 months) postprocedure demonstrated hemodynamic evidence of recurrent obstruction despite sustained enlargement at the previously stented sites. Complications included stent displacement (n=1), bacterial endocarditis (n=1), and false aneurysm formation (n=1). One patient died awaiting conduit replacement surgery.
Conclusions Stent implantation in obstructed RV-PA conduits results in significant immediate hemodynamic and angiographic improvement. In a subgroup of patients, the procedure prolongs conduit life span by several years and increases the interval between conduit reoperations. Recurrent obstruction is caused by external compression and progressive stenosis outside the stented region.
RV-PA homografts and bioprosthetic conduits have been widely used to surgically palliate several forms of congenital heart disease, including truncus arteriosus, tetralogy of Fallot, and complete transposition of the great arteries with VSD. The life span of these conduits is often limited by the development of progressive lumen obstruction. The larger follow-up series for bioprosthetic conduits have observed an actuarial freedom from replacement of 68% to 95% at 5 years and 0% to 59% at 10 years.1 2 3 4 5 Early studies examining cryopreserved homografts as RV-PA conduits have reported an actuarial freedom from replacement of 55% to 94% at 5 years.4 6 7 8 9 External compression, calcification, kinking, fibrotic intimal peel formation, and homograft “contraction” are all thought to contribute to conduit deterioration. In addition, young patients, because of their size, often require smaller-diameter conduits, which they subsequently outgrow.
Surgical replacement of obstructed conduits incurs the morbidity associated with repeat sternotomy and cardiotomy and, in some patients, is technically difficult because of multiple prior operations. Percutaneous balloon angioplasty of stenosed conduits has shown marginal efficacy in diminishing pressure gradients and avoiding surgical intervention.10 11 12 Recently, balloon-expandable intravascular stents have been used to relieve stenotic lesions in various forms of congenital heart disease.13 14 15 By supporting the conduit wall after balloon dilation, stent implantation may relieve obstruction and delay surgical replacement. This article reports our experience implanting RV-PA conduit stents in 44 patients along with intermediate-term follow-up data.
A record review and computer database search identified 44 patients who underwent placement of one or more endovascular stents in an RV-PA conduit at Children’s Hospital, Boston, Mass. The procedures were performed from December 1990 through July 1994, with informed parental and/or patient consent. Patient records were reviewed retrospectively, and data from clinic visits, chest radiographs, echocardiograms, catheterizations, and surgeries were compiled. The events considered to be end points for follow-up were death, conduit replacement surgery, or most recent clinical assessment.
In general, patients were selected for RV-PA conduit stent placement if they failed balloon dilation and their RV pressure was >80% of the systemic pressure or if there was evidence of RV dysfunction. Patients whose nominal conduit diameter was deemed inadequate because of somatic growth after surgery were not candidates for the procedure.
The nonarticulated Palmaz stent (Johnson & Johnson Interventional Systems) constructed from stainless steel was used in all patients. The stent sizes used were (length×diameter) 30×3.4 mm (n=20), 18×3.4 mm (n=14), 12×3.4 mm (n=1), 20×2.5 mm (n=6), and 15×2.5 mm (n=7). Three patients required placement of two or three overlapping stents in their RV-PA conduits.
The technique of percutaneous stent implantation was described previously.13 14 Briefly, patients were given heparin intravenously to maintain an activated clotting time >200 seconds. The stenotic region of the RV-PA conduit was predilated with a balloon angioplasty catheter in the usual fashion. Balloon size, in most cases, was arbitrarily limited to 10% larger than the nominal conduit diameter to minimize the risk of false aneurysm formation. High-pressure inflations up to 21 atm were used when a residual waist was present. In some patients, the conduit was dilated by simultaneous inflation of two balloons placed side by side. A long sheath (7F to 12F according to balloon and stent size) was then advanced over a SuperStiff (MediTech) or Rosen (Cook) guide wire. The stent was mounted on a balloon catheter, and together they were advanced over the wire through the sheath and positioned at the stenotic region. The sheath was withdrawn, and angiographic confirmation of stent position was obtained. The balloon was inflated by hand, and the stent was expanded and anchored to the vessel wall. If the stent was not fully expanded, a larger-diameter balloon was used to further dilate the stent. Postimplantation pressure measurements and angiography were performed when possible. When the angiograms were reviewed, the extent of conduit calcification was qualitatively assessed and categorized as absent, mild, or severe.
Intravenous antibiotic, usually cefazolin (12.5 mg/kg), was administered at the time of stent placement and every 6 hours for a total of three doses. After catheterization, patients were placed on a continuous heparin infusion to maintain a partial thromboplastin time twofold control for a minimum of 12 hours. At discharge, most patients were prescribed either aspirin 40 to 80 mg/d (n=25) or aspirin 40 to 80 mg/d and dipyridamole 1 to 2 mg·kg−1·d−1 (n=16) for 6 to 12 months to inhibit thrombus formation. Two patients were anticoagulated with warfarin because they had a mechanical prosthetic valve or a stent in the systemic venous circulation. One patient with chronic thrombocytopenia was not placed on any anticoagulation medication.
Other Procedures Performed
In addition to RV-PA conduit stent implantation, 18 patients underwent other interventions during the same catheterization. These procedures included PA stent placement (n=7), PA balloon angioplasty (n=8), pulmonary valve balloon angioplasty (n=2), coil embolization of collateral vessels (n=1), and residual VSD closure with a clamshell device (n=1).
Preplacement and postplacement data were compared by a paired two-tailed Student’s t test. The univariate relations between patient parameters and outcome were evaluated by a one-way ANOVA, a two-sample t test, or a linear regression model when appropriate. In the analysis of reoperation and stent fracture, time to failure was the response variable of interest. Patients who did not fail were considered to be censored at the time of last follow-up. Survival estimates were obtained by the Kaplan-Meier method. Risk factor subgroups were compared by the log-rank test; a Cox proportional-hazards model was used to assess multivariate associations. Results were considered significant if P≤.05. Values are expressed as mean±SD. The stata statistical package (Computing Resource Center) supported the analysis.
Among the 44 patients who underwent stent implantation in an RV-PA conduit, there were 31 males and 13 females 7 months to 30 years old (median, 6.9 years) and weighing 5.7 to 68.8 kg (median, 22.8 kg). The patients’ primary cardiac diagnoses included tetralogy of Fallot with pulmonary atresia (n=17), d-transposition of the great arteries with VSD (n=11), truncus arteriosus (n=10), tetralogy of Fallot (n=3), aortic valve atresia with VSD (n=2), and l-transposition of the great arteries with pulmonary atresia (n=1). Seventeen of the patients had already had their RV-PA conduits surgically replaced once, and 5 more patients had undergone revision twice. The time from the most recent conduit placement to stent implantation ranged from 1 week to 13.1 years (median, 3.0 years). The RV-PA conduits were composed of cryopreserved aortic homografts (n=25), cryopreserved PA homografts (n=10), bioprosthetic conduits (n=8), and pseudointima covered by pericardium (n=1). Their nominal diameters at the time of surgical implantation measured from 6.5 to 23 mm (median, 14 mm). Among the homografts, the extent of calcification was categorized as absent (n=15), mild (n=9), or severe (n=11).
One patient was lost to follow-up after the immediate postcatheterization period. The remaining patients have been followed from 1 to 48 months (median, 14.2 months) since conduit stent implantation, a total of 59.9 patient-years.
Initial Stent Implantation
A single stent was placed in the stenotic RV-PA conduit of 41 patients, two overlapping stents in 2 patients, and three overlapping stents in 1 patient. The mean catheterization time was 233±73 minutes (range, 85 to 531 minutes). The average hospital stay for patients admitted electively and not undergoing surgery was 1.9±1.2 days (n=37). Figs 1 through 3⇓⇓⇓ illustrate examples of stent placement in RV-PA conduits.
Table 1⇓ reports the prestent and poststent mean values of the RV-PA peak-to-peak pressure gradient, the RVp/Sp ratio, and the stenosis diameter in the AP and lateral views. Stent implantation resulted in significant improvements in all of these parameters (Fig 4⇓). The number of patients with pressure gradients ≥50 mm Hg declined from 34 prestent to 3 poststent. In 5 patients, stent implantation failed to reduce the pressure gradient by >25%. The number of patients with an RVp/Sp ratio ≥0.75 fell from 38 prestent to 8 poststent. Seven patients had a <10% reduction in their RVp/Sp ratio. For those patients in whom angiographic measurements were available both before and after stent placement, the diameter of the stenotic region expanded by a mean of 2.9±2.0 mm in the AP view (n=35) and by a mean of 4.4±1.8 mm in the lateral view (n=37).
In univariate analyses, the following parameters were not significantly associated with the reduction in RV-PA pressure gradient or RVp/Sp ratio: patient age, weight, sex, cardiac diagnosis, patient age at conduit placement, conduit age, homograft versus bioprosthetic conduit, pulmonary versus aortic homograft, nominal conduit diameter, conduit calcification, and utilization of simultaneous side-by-side balloon dilation. Conduit dilation with a high-pressure balloon before or after stent placement resulted in a slightly greater reduction in the RVp/Sp ratio (P=.043) but had no significant effect on gradient relief (P=.17). Angiographic improvement in the AP view was most strongly associated with the absence of calcification (P=.032) and younger conduit age (P=.029) in a multivariate model. None of the parameters were significantly associated with enlargement in the lateral view. Finally, no “learning curve” effect on immediate efficacy was detected with later procedures.
In general, follow-up catheterization was recommended when noninvasive assessments of RV pressures suggested progressive restenosis. Sixteen of the initial 44 patients were referred for recatheterization at our center 3 to 48 months (median, 11.8 months) after stent implantation. Table 2⇓ reports the prestent, immediate poststent, and follow-up mean values of the RV-PA pressure gradient, the RVp/Sp ratio, and the stenosis diameters. Although stent implantation resulted in immediate improvement in all these measurements (P≤.001), there was a return to prestent values for the mean pressure gradient and mean RVp/Sp ratio in this subgroup (Fig 5⇓). Nevertheless, the AP and lateral-view diameters at the stent site demonstrated a significant sustained improvement compared with their prestent values. Careful examination of the angiograms revealed conduit stent fractures in 5 of the patients as described in further detail below. Even among the remaining 11 patients with intact stents, the follow-up RV-PA gradient and RVp/Sp ratio demonstrated reobstruction to prestent levels despite persistent angiographic enlargement at the stent site (data not shown).
Repeat Stent Implantation and Redilation
Seven of the 16 patients who underwent follow-up catheterization at our center had additional conduit stents placed during the procedure (Fig 5⇑). The repeat stent implantations occurred from 7 to 27 months (median, 9.7 months) after the initial placement. All 5 of the patients with fractured stents had a second stent positioned to partially overlap the original one. The other patients had either one or two additional overlapping stents deployed. As a result, the RV-PA pressure gradient decreased from 59.6±16.3 to 40.9±13.9 mm Hg (P=.002, n=7) and the RVp/Sp ratio declined from 0.98±0.14 to 0.70±0.10 (P=.006, n=7). Significant diameter enlargement was noted in the lateral view (P=.045, n=5) but not in the AP view (P=.59, n=5).
Two patients underwent balloon dilation of their intact conduit stents without implantation of additional stents 9 and 14 months after initial placement, respectively. In both patients, the RV-PA pressure gradient was decreased by <10 mm Hg. The RVp/Sp ratio declined from 0.9 to 0.7 in 1 patient and was unchanged in the other. The stent diameter in each case increased by ≈1 mm in both the AP and lateral views.
Fourteen (32%) of the 44 patients underwent surgical replacement of their RV-PA conduits from 1 to 33 months (median, 11.0 months) after stent placement. No major technical difficulties with the operations related to the stents were reported. Actuarial freedom from conduit reoperation was 65% (Greenwood 95% CI, 48% to 82%) at 30 months after stent placement (Fig 6⇓). None of the following prestent parameters had a significant effect on reoperation-free survival: patient age, weight, sex, cardiac diagnosis, patient age at conduit placement, conduit age, homograft versus bioprosthetic conduit, pulmonary versus aortic homograft, nominal conduit diameter, conduit calcification, use of simultaneous side-by-side balloon dilation, use of high-pressure balloons, RV-PA gradient, RVp/Sp ratio, and stenosis diameter. In a multivariate analysis, the strongest predictors of early surgical replacement were a greater poststent RVp/Sp ratio (P=.010) and a narrower poststent diameter in either view (P=.044).
Nine of the 16 patients who had follow-up catheterization later underwent conduit replacement (five patients within 1 week of catheterization). Among the 7 patients who had a second conduit stent implanted, 4 had surgical revision. Two of these patients had replacement performed within 3 days of the second stent placement, while the other 2 remained free of surgery for 12 and 23 months, respectively. Both patients who had only balloon dilation of their stents underwent conduit replacement within 5 months.
At the completion of the follow-up period, 22 patients had not been referred for recatheterization or conduit reoperation. Echocardiography data were available for 19 of these patients (86%), with the most recent studies performed from 1 to 32 months (median, 16.2 months) after stent implantation. One patient had evidence of an RV pressure >80% of systemic pressure, and 2 other patients had Doppler gradients >60 mm Hg across their conduits. No fractured or displaced stents were detected in this subgroup.
Seven patients (16%) were found to have fractures of their conduit stents during the follow-up period. All were asymptomatic events identified on chest radiograph, echocardiogram, or catheterization from 5 to 27 months (median, 8.4 months) after implantation. In 3 of these patients, stent fragments embolized to the pulmonary circulation. No evidence of significant pulmonary blood flow obstruction was seen in the 2 patients who had angiography, and all embolized fragments were left in place. Five patients in the fracture group had a second conduit stent placed, and no subsequent fractures were observed. Four patients in the fracture group, including 3 who had a second stent implantation, underwent surgical replacement of their conduits. No significant risk factors for stent fracture were identified with a Cox proportional-hazards model. It is noteworthy, however, that for 2 patients in the fracture group, stent expansion was asymmetric, thereby creating a shorter posterior length on initial implantation. This condition may have led to the development of unusually high stress on the supporting struts of the stent and eventual fracture.
Catheterization complications included a hemothorax (secondary to difficult vascular access into an internal jugular vein) that was percutaneously drained without reaccumulation, an episode of atrial flutter that was successfully treated by esophageal pacing, and a brachial plexus injury that self-resolved. Bacterial endocarditis developed in 1 patient 5 weeks after stent implantation. He was effectively treated with intravenous antibiotics and has not required conduit replacement. One patient developed a pseudoaneurysm and arteriovenous fistula in the subclavian vein, which had been used for venous access during catheterization. The fistula was closed by coil embolization. Balloon rupture occurred frequently, and in most instances there were no sequelae; however, in 3 patients, the balloon ruptured along the transverse axis, creating a distal fragment. In one case, the fragment was successfully withdrawn after a large sheath was placed over it; in another, the balloon became lodged in the femoral vein and was retrieved via cutdown. The third instance occurred during stent dilation and required withdrawal of the partially expanded stent to the iliac vein, in which it was deployed.
One patient, a 17-month-old boy with transposition of the great arteries, VSD, and pulmonary stenosis status post Rastelli repair with an RV-PA aortic homograft, died unexpectedly at home 5 months after conduit stent implantation. At a cardiology outpatient visit 2 weeks before his death, his physical examination and chest radiograph had been unchanged, and his ECG showed a right bundle-branch pattern with no ectopy. An echocardiogram demonstrated severe proximal conduit obstruction and a dilated RV with approximately systemic pressure and poor function. Accordingly, surgery to replace his homograft was scheduled. The patient’s death occurred at night in the setting of an acute febrile illness with vomiting. No specific cause of death was found, and at the parents’ request, no autopsy was performed.
Proximal displacement of a stent into the RV cavity was discovered in 1 patient 2 months after implantation. Because the patient had phrenic nerve paralysis, the stent had intentionally been placed just proximal to the homograft valve to avoid creating free pulmonary regurgitation and decreased systemic venous return. The patient remained clinically well and successfully underwent surgical removal of the stent and homograft replacement.
In 1 patient, a false aneurysm in the RVOT was identified 18 months after stent placement. At surgery, it appeared that the homograft had dehisced from the RV just proximal to the homograft stent. The patient underwent homograft excision and replacement and has subsequently done well. For this stent implantation, a low-pressure balloon 10 mm in diameter was used for dilation and placement into a homograft with a nominal diameter of 7 mm. Although there was no angiographic evidence of vascular injury during the procedure, the oversized balloon may have excessively dilated the homograft and weakened the vessel wall.
Prior studies on stent implantation to relieve RV-PA conduit obstruction are limited, and virtually no follow-up information is available. We previously reported significant gradient reduction and angiographic improvement after stent placement in 6 patients.14 Hosking et al15 deployed conduit stents in 5 patients and achieved similar immediate results. Doppler flow studies 48 hours after the procedure showed persistent gradient relief only in the 2 patients with the greatest angiographic improvement. On follow-up several months later, 3 of the 5 patients were scheduled to undergo surgical replacement of their conduits.
This report, with a cohort of 44 patients and almost 60 patient-years of follow-up, provides important data on a technique designed to delay conduit reoperation. Stent implantation in obstructed RV-PA conduits resulted in immediate hemodynamic and angiographic improvement that was both physiologically and statistically significant. During the follow-up period, 2 patients had their stents redilated, 7 had additional conduit stents deployed, and 14 underwent surgical replacement of their conduits. Actuarial freedom from conduit reoperation was 65% at 30 months postprocedure. On the basis of their initial hemodynamic measurements, almost all of the patients in this study would have been candidates for conduit replacement at the time of stent implantation. Thus, for a subgroup of our patients, stent implantation appears to postpone reoperation by at least several years.
Further efforts to prolong RV-PA conduit life span will require an improved understanding of their mechanisms of obstruction. Two insights into this process are suggested by our follow-up catheterization data in 16 patients with hemodynamic evidence of reobstruction. The occurrence of stent fractures in 5 of these patients supports an important role for external compressive forces in conduit failure. Stents implanted in RV-PA conduits may be predisposed to fracture because of compression between the sternum and heart, proximity to beating ventricular myocardium, or extensive calcification of the homograft vessel wall. When placing stents in pulmonary arteries, we have observed stent fracture at the time of balloon dilation but not spontaneously on follow-up evaluation. Strategies to delay conduit reobstruction must therefore take into account compressive forces that are apparently unique to the RVOT. A second important observation in the recatheterization subgroup is that we found hemodynamic evidence of recurrent obstruction despite sustained enlargement at the previously stented sites. This finding implies that reobstruction occurs all along the conduit and is multifactorial. Notably, we did not observe exuberant pseudointimal growth within the stents.
While most observed complications of stent placement could be expected in a complex interventional procedure, one deserves special comment. A false aneurysm was identified in the RVOT of 1 patient 18 months after stent implantation. The balloon size used for dilation was ≈40% larger than the nominal conduit diameter. Although other conditions may have contributed to aneurysm formation, we recommend in most cases limiting balloon size to 10% larger than the nominal conduit diameter to minimize the risk of this complication.
This study has several important limitations. Because of its retrospective design and the lack of standardized follow-up, the reported prevalence of complications and conduit reoperation should be considered lower bounds. In particular, since the stent fractures were all asymptomatic events and not all patients have had a recent radiological evaluation, additional patients may have as yet undetected fractures. Moreover, several stents that appeared intact on plain radiographs and echocardiography were found to be fractured on fluoroscopy. Our reported frequency of conduit reoperation after stent placement also merits scrutiny, since the indications for replacement were not uniform. We cannot completely exclude the possibility that stent placement raised the threshold for surgical referral during the follow-up period. Recurrent conduit obstruction warranting surgery may also have gone undetected. Ideally, all of our patients would have undergone recatheterization within the follow-up period, thereby permitting the most accurate assessment of the durability of hemodynamic and angiographic improvement. Since this was not clinically indicated, we have depended, in part, on noninvasive assessments to detect recurrent conduit obstruction. Several reports have documented the reliability of Doppler echocardiography in assessing RV-PA conduit dysfunction.16 17 18 Because no prestent parameters were significantly associated with reoperation-free survival, we were unable to refine our patient selection criteria for conduit stent placement. This failure may be attributable to the relatively small number of patients and the limited power of the study. Finally, we acknowledge that the additional procedures performed during stent placement might confound the comparison between prestent and poststent parameters. Such effects, however, would not have altered the angiographic analysis.
Patient selection for RV-PA conduit stent placement must be individualized. The most prevalent adverse effect associated with the procedure was stent fracture requiring either repeat stent placement or conduit replacement. Although we have not detected any obstruction to pulmonary blood flow by the embolized stent fragments, this risk remains. The principal benefit of stent implantation lies in prolonging conduit life span and increasing the interval between the almost inevitable reoperations. Accordingly, we would consider children whose conduits develop obstruction before being outgrown as candidates for this intervention. In this scenario, stent implantation may decrease the expected number of conduit reoperations as the patient ages from a neonate to an adult. Patients who develop significant conduit obstruction and are poor surgical candidates, typically because of multiple prior operations, may also be appropriate subjects for stent placement. After the procedure, radiological assessment for stent fractures, either with plain radiographs or preferably with fluoroscopy in multiple views, should be performed on a regular basis.
Stent implantation in obstructed RV-PA conduits results in significant immediate hemodynamic and angiographic improvement. In a subgroup of patients, the procedure prolongs conduit life span by at least several years and provides an effective means of increasing the interval between conduit reoperations. Recurrent obstruction is caused by progressive stenosis along the entire RVOT and stent fracture from external compression. Goals for further study include (1) refining patient selection, (2) proper timing of stent placement, (3) long-term follow-up to better determine outcomes and risk, (4) developing stents more resistant to fracture, and (5) further clarifying causes of conduit restenosis.
Selected Abbreviations and Acronyms
|RV-PA||=||right ventricle to pulmonary artery|
|RVOT||=||right ventricular outflow tract|
|RVp/Sp||=||right ventricular–to–systemic arterial pressure (ratio)|
|VSD||=||ventricular septal defect|
This study was supported in part by the Kobren Fund. We express our appreciation to Jane Newberger, MD, for her valuable guidance, Kimberlee Gauvreau, ScD, for her statistical advice, and Jocelyn Wise for her assistance with data collection.
Reprint requests to Stanton B. Perry, MD, Department of Cardiology, Children’s Hospital, 300 Longwood Ave, Boston, MA 02115.
- Received May 9, 1995.
- Revision received July 12, 1995.
- Accepted July 23, 1995.
- Copyright © 1995 by American Heart Association
Jonas RA, Freed MD, Mayer JE, Castaneda AR. Long-term follow-up of patients with synthetic right heart conduits. Circulation. 1985;72(suppl II):II-77-II-83.
Razzouk AJ, Williams WG, Cleveland DC, Coles JG, Rebeyka IM, Trusler GA, Freedom RM. Surgical connections from ventricle to pulmonary artery: comparison of four types of valved implants. Circulation. 1992;86(suppl II):II-154-II-158.
Cleveland DC, Williams WG, Razzouk AJ, Trusler GA, Rebeyka IM, Duffy L, Kan Z, Coles JG, Freedom RM. Failure of cryopreserved homograft valved conduits in the pulmonary circulation. Circulation. 1992;86(suppl II):II-150-II-153.
O’Laughlin MP, Perry SB, Lock JE, Mullins CE. Use of endovascular stents in congenital heart disease. Circulation. 1991;83:1923-1939.
O’Laughlin MP, Slack MC, Grifka RG, Perry SB, Lock JE, Mullins CE. Implantation and intermediate-term follow-up of stents in congenital heart disease. Circulation. 1993;88:605-614.