Author + information
- Received October 25, 2005
- Revision received March 7, 2006
- Accepted March 16, 2006
- Published online August 1, 2006.
- Alan W. Nugent, MBBS⁎ (, )
- Amy Britt, MA,
- Kimberlee Gauvreau, SCD,
- Gary E. Piercey, BS,
- James E. Lock, MD, FACC and
- Kathy J. Jenkins, MD, MPH, FACC
- ↵⁎Reprint requests and correspondence:
Dr. Alan W. Nugent, Children’s Hospital Heart Center, 6621 Fannin MC 19345-C, Houston, Texas 77030-2399.
Objectives This study sought to review the outcomes of 3 generations of closure devices for secundum atrial septal defects (ASDs) at a single institution.
Background Transcatheter closure of ASDs is now increasingly performed with devices that have been modified over time to improve performance.
Methods A review of prospective clinical trials of Clamshell (C.R. Bard Inc., Murray Hill, New Jersey), CardioSEAL (NMT Medical Inc., Boston, Massachusetts), and STARFlex (NMT Medical Inc.) closure devices for simple ASDs was performed. The entry criteria for these trials were age ≥2 years, isolated secundum ASD, evidence of right ventricular volume overload, and maximum stretched diameter varying from 20 to 25 mm. Successful outcomes were defined as no more than small residual leak (≤2 mm) with absence of a severe complication or the need for an additional device or surgery.
Results A total of 72 Clamshell, 30 CardioSEAL, and 42 STARFlex devices were placed in uncomplicated ASDs. Each device modification resulted in improved closure rates, with the STARFlex device achieving a 93% complete closure rate with a device/stretched diameter ratio significantly smaller than with the other devices (p = 0.001). Severe complications occurred in 5 Clamshell, 1 CardioSEAL, and 0 STARFlex cases. Successful outcomes increased for each generation of device (79%, 93%, and 98% respectively, p = 0.009). There have been no long-term complications for either the CardioSEAL or the STARFlex devices.
Conclusions Modifications in 3 generations of devices have resulted in improved results for percutaneous ASD closure.
Transcatheter closure of secundum atrial septal defects (ASDs) is an attractive alternative to cardiac surgery now increasingly performed and offered as primary therapy. Advantages include the avoidance of sternotomy, cardiopulmonary bypass, right atriotomy, postpericardiotomy syndrome, and longer hospitalization. Since first described (1), various devices have been designed and manufactured for ASD closure, with modifications over time aimed at improving device deliverability and closure rates.
The Clamshell device (C.R. Bard Inc., Murray Hill, New Jersey) was studied in the U.S. and Canada from 1989 to 1991 with initial favorable reports (2), but a high rate of late device arm fractures led to cessation of clinical trials. After redesign and extensive preclinical evaluation, the second-generation CardioSEAL device (NMT Medical Inc., Boston, Massachusetts) entered clinical trials in 1997, again with encouraging results (3). The third-generation STARFlex modification (NMT Medical Inc.) has a self-centering mechanism consisting of flexible nitinol microsprings, and entered clinical trials in 1999.
The recent descriptions of several dozen late catastrophic cardiac erosions after Amplatzer (AGA Medical Corp., Golden Valley, Minnesota) ASD closure (4,5) have rekindled interest in alternate devices and techniques for closing holes in the atrial septum. We have therefore reviewed the outcomes and complications of 3 generations of closure devices for simple ASDs at a single institution.
Children’s Hospital Boston has over 15 years (6) of continuous experience with transcatheter device closure of ASDs and other atrial defects. A review of prospective clinical trials of device closures for simple ASDs was performed. In the Clamshell trial, patients defined with simple ASD and in the CardioSEAL and STARFlex trials the low-risk patients were analyzed. Internal review board-approved informed consent was obtained for all cases.
Inclusion criteria were consistent for all trials: patient age ≥2 years, isolated secundum ASD, evidence of right ventricular volume overload, evidence of shunt on transthoracic echocardiogram (TTE), patients who were surgical candidates for ASD closure, and a defect that was remote from venous or valvar structures to permit safe seating of the device. The maximum stretched diameter permitted was 25 mm for Clamshell, 20 mm for CardioSEAL, and 22 mm for STARFlex devices.
Exclusion criteria were multiple atrial level shunts, unrepaired associated cardiac defect, pregnancy, patients with previously implanted devices, intra-atrial thrombus, left heart failure, concurrent atrial arrhythmias or antiarrhythmia medication, aspirin intolerance, presence of significant pulmonary hypertension (Wood units >5), endocarditis, or signs of active infection/bacteremia.
Procedures were performed under general anesthesia with transesophageal echocardiogram (TEE) guidance of device placement. All patients were treated with 100 U/kg heparin at the start of the case. Sufficient heparin was given to maintain activated clotting times >200 s (measured hourly). Prophylactic Kefzol (vancomycin, if allergy) was given during the case, and a second dose was given 6 h later. Endocarditis prophylaxis was recommended for 6 months after the procedure. Antiplatelet therapy of aspirin 1 to 3 mg/kg/day was prescribed for 6 months after the device was placed for all patients.
The individual cardiologist performing the catheterization was ultimately responsible for deciding the device size. A device-stretched diameter ratio of 1.7 to 2:1 was recommended for the Clamshell and CardioSEAL devices. With the addition of the self-centering mechanism a device-stretched diameter ratio of 1.6 to 1.7:1 was recommended for the STARFlex device. To deliver the device, an 11-F sheath was required for the Clamshell, a 10-F or 11-F for the CardioSEAL, and a 10-F for the STARFlex device.
Data were collected from the prospective device database and study files, procedural catheterization and echocardiography reports, and computerized department database. The procedural TEE assessed defect size and degree of surrounding rims and the stretched diameter was measured by cine angiograms of the sizing balloon waist. Patient age, body weight, device type and size, and device position were recorded. Malposition was defined as any device placed in an incorrect position and includes devices that were removed during the procedure (unreleased/released/embolized), removed at surgery, or left in situ (device deemed stable and classified as incorrect final location).
Catheter-based complications were recorded and classified into minor, moderate, and severe categories. A minor event was defined as one that resulted in minimal or transient impairment that did not require any intervention, e.g., brief self-limiting arrhythmia. A moderate adverse event was defined as an event that caused a transient impairment that required an intervention to treat or correct/prevent permanent damage, e.g., medication, cardioversion, or device retrieval. A severe event was defined as life threatening, or causing permanent impairment or necessitating significant intervention to prevent permanent damage, e.g., surgical management. Device fractures were recorded separately unless related to an adverse event.
Follow-up echocardiography was mandated by prospective protocols the day after and at a minimum either at the 6-month or the 1-year follow-up. Clinical review was required at 1 month, 6 months, and 1 year (and 2 years for the STARFlex cohort). The latest available follow-up was included when available. Fractures to the device at follow-up were examined by fluoroscopy and chest radiograph.
Residual flow was based on the most recently available echocardiogram. No residual flow was defined as nil detectable residual color flow through the defect. Trivial residual flow was defined as a single color flow jet width <1 mm. Small residual flow was a single color flow jet 1 to 2 mm and not of hemodynamic significance. Greater than small was any residual flow >2 mm. Multiple leaks were defined as above, i.e., “multiple, trivial” leaks was more than 1 leak, each of which had a color jet width of <1 mm. The residual flow was recorded as uncertain if no follow-up echocardiography was available or the acoustic windows did not allow adequate interrogation of the atrial septum.
Successful outcomes were defined as a procedure that resulted in no more than a small residual leak (≤2 mm), had no severe complication related to device or procedure, and had no need for an additional device or surgery. A procedure with >2 mm residual leak, a severe adverse event, or the need for a second device or surgical procedure at any future time was defined as unsuccessful.
Patient and device characteristics were compared for the 3 groups of subjects receiving Clamshell, CardioSEAL, and STARFlex devices. A Fisher exact test for categorical variables and a Kruskal-Wallis test for continuous variables were performed. Proportions of successful procedures were compared for STARFlex versus the Amplatzer Septal Occluder and surgical outcomes using the Fisher exact test.
A total of 72 Clamshell, 30 CardioSEAL, and 42 STARFlex devices were placed in uncomplicated ASDs. Table 1shows the characteristics of the study sample; there is no significant difference in gender, age, or weight at procedure by device type. As expected, the earlier-generation device has a longer period of follow-up.
The defect and device characteristics are shown in Table 2.There is no significant difference between the sizes of defects closed with each device. Because of the differing guide for device sizing, as expected, the STARFlex device-stretched diameter ratio was significantly smaller than the other 2 devices, with 60% of the STARFlex insertions having a device-stretched diameter ratio <2.
There is an increase in closure rates with each device modification. None or trivial residual flow with each generation improves from 72% to 80% to 93% (Table 3).
Severe complications occurred in 5 Clamshell, 1 CardioSEAL, and 0 STARFlex patients; 4 Clamshell patients had surgery: 2 embolizations, 1 mitral valve interference, and one late thrombus formation related to a fracture. The other severe complication was an arm fracture with embolization to the right ventricular free wall that was managed conservatively. The CardioSEAL severe complication was a device embolization to the right pulmonary artery noted on postprocedure chest radiograph requiring surgical removal. A full list of complications is given in Table 4.
A summary of successful and unsuccessful outcomes is shown in Table 5.The increasing percentage of successful outcomes for each generation of device is 79%, 93%, and 98%, respectively (p = 0.009). Differences in success rates for Clamshell versus CardioSEAL versus STARFlex remain after controlling for gender and age.
Fifteen (21%) patients in the Clamshell group had unsuccessful outcomes, 12 because of the need of a second procedure. Ten patients had surgery, 4 for device-related complications, and 3 patients had a second device placed (1 patient had both a second device and surgery). Of the 9 patients with greater than small residual defect, 6 had a second procedure, and only 3 were medically managed.
Only 2 (7%) in the CardioSEAL group had an unsuccessful outcome. The first has been described in the previous text, and the second was a greater than small residual flow in a malpositioned device that was managed with a second device. In the STARFlex group, only 1 (2%) had a greater than small residual leak shown by echocardiography, and no patients had a second procedure or surgery. This case was a 33-mm device that was placed in a defect with deficient rims and was positioned with 5 of the 8 arms on the right-hand side of the septum. No further intervention has been required.
Of the 144 patients in this study, there is a total of 596 person-years of follow-up, with the Clamshell group having the most follow-up. As noted previously, long-term complications led to the cessation of clinical trials with the Clamshell device. The prospective trials of the CardioSEAL and STARFlex devices have 137 person-years of follow-up, with the latest complication (minor, transient arrhythmia) occurring 1 month after device insertion. No late complications have been noted.
Deficient rims were not recorded for the Clamshell group, but procedural TEE reports show deficient rims in 27% (8 of 30) of CardioSEAL and 33% (14 of 42) of STARFlex procedures. All were deficiencies of the retroaortic (anterior) rim, with 1 case also having a deficient superior rim. Of these patients, 91% (20 of 22) had successful outcomes with only 1 patient in each group, 40-mm CardioSEAL embolization and STARFlex with 3- to 4-mm residual shunt, having unsuccessful results.
Fractures are more common in the earlier-generation device, with fractures detected in 68% (47 of 69) of Clamshell, 41% (12 of 29) of CardioSEAL, and 34% (13 of 38) of STARFlex devices (p = 0.001). Fluoroscopy of 67 devices was available in 29 of 30 CardioSEAL and 38 of 42 STARFlex devices, enabling further analysis. In this group, 25 fractures occurred, with 59% (19 of 32) of fractures in the 33/40-mm devices compared with 17% (6 of 35) for the smaller devices (p < 0.001). When a fracture did occur, successful outcomes still occurred in 92% (23 of 25) of cases.
The outcome of percutaneous device closure of uncomplicated ASDs in a single institution is reported with improved outcomes for each generational modification. The latest-generation STARFlex device has excellent outcomes with almost universal success.
The first-generation Clamshell device was made of a stainless steel frame with knitted Dacron fabric attached in an umbrella-like fashion. There was only 1 joint per arm, and the device was available in 5 sizes: 17, 23, 28, 33, and 40 mm (Fig. 1).The second-generation CardioSEAL device entered clinical trials in 1997, after extensive redesign to improve fatigue fracture resistance and spring-back. Although the sizes were identical to those of the Clamshell device, the arms were made of a new metal alloy, different wire diameter, and 2 joints per arm (Fig. 2).The third-generation STARFlex device entered clinical trials in 1999. The major enhancements were a self-centering mechanism, improved pin-to-pin delivery, and a smaller delivery profile (Fig. 3).There were initially 4 sizes: 23, 28, 33, and 40 mm. Although there were no issues with the 40-mm device in this trial of simple ASD closure, because of increased embolizations noted in other patients it was no longer available after 2001.
The initial Clamshell trial from 1989 to 1991 was discontinued because of a high rate of fractures. Closure rates and successful outcomes improved with the CardioSEAL device, which could be partly explained by superior operator performance after the initial learning curve. Successful outcomes between the CardioSEAL and STARFlex devices improved from 93% to 98% and were most likely attributable to device modifications, although by 1999 improved experience among multiple operators also may have played a role. The different closure rates do not reach statistical significance, with the potential for a type 2 error with a small number of cases from a single institution. The latest-generation device, with the addition of the self-centering mechanism, not only provides the best closure rates but also attains this result with significantly smaller device sizes. The prospective protocol recommended a device-balloon stretched diameter ratio of at least 1.7:1 for the STARFlex device. This has resulted in excellent outcomes and is recommended for future use.
The STARFlex modification is not approved or available in the U.S. currently, and this report supports the availability of this device for simple ASD when inserted by practitioners adequately trained and familiar with the device. This is particularly pertinent with the recent reports of late perforations associated with the Amplatzer Septal Occluder (4,5). The data sets from all CardioSEAL and STARFlex trials at this institution were reviewed, specifically looking for erosions, perforations, effusions, or explants. We found 3 erosions or perforations after more than 900 device implantations: 2 in apical ventricular septal defects and 1 in an ASD. In all 3 cases, the perforation of a single arm was clinically silent and was only discovered at the time of cardiac surgery for a different indication.
Comparison with previous reports
It is difficult to compare the Clamshell closure rates of this study with previously published results because of different patient selection. The earlier reports included defects associated with cyanotic heart disease, patent foramen ovale, and Fontan leaks. Rome et al. (2) reported a 63% closure rate (12 of 19) with TTE 6.5 months after placement; however, only 7 patients in this series had an uncomplicated ASD. Boutin et al. (7) with 10 months of follow-up found a closure rate of only 47% in predominantly ASDs.
For CardioSEAL or STARFlex devices, these results compare favorably with the other published reports. A single-institution report of the first 50 patients with a CardioSEAL implant shows, at a mean of 9.9 months’ follow-up, a small shunt in 46% of cases but right ventricular dimensions returned to normal (8). There are limited reports of the STARFlex device (9), but 1 study compares the CardioSEAL and STARFlex devices. A total of 117 ASD patients had 79 CardioSEAL and 38 STARFlex devices placed, with significantly lower rates of residual shunts with the STARFlex device (10).
The published outcomes for different percutaneous closure devices (11–14) and surgery (11,15–17) are summarized in Table 6.Du et al. (11) use the same definition for residual leak on follow-up for the Amplatzer device and provide a complication list, enabling classification using the same criteria for both device and surgical patients. The Amplatzer Septal Occluder had severe complications in 5 patients of 442 (1.1%) who had a device attempt including surgery for 4 embolizations and 1 heart block requiring pacemaker; 331 patients completed the 12-month follow-up, with a successful closure rate of 98.5% (326 of 331). However, in the text there are 3 further spontaneous closures; therefore, to be consistent with the present study, the successful closure rate would be 99% (329 of 331) with 1 large residual shunt and 1 patient undergoing a second device placement. The surgical cohort has a 5.2% (8 of 154) severe complication rate and follow-up on 149 patients with a quoted 100% closure rate, although only 72% of cases were confirmed by echocardiography and 28% were “assumed closed.”
Comparison with the Amplatzer septal occluder was performed because it is the only device approved for use in the U.S., and the same surgical group was used for both devices. It is recognized that a direct comparison is not able to be performed, especially for surgery, and also for device insertion because of the capability of the Amplatzer device to close larger defects and to use smaller delivery sheaths. In any event, the comparison shows remarkable similarities between STARFlex and Amplatzer septal occluder devices (Table 7).
Device fractures were found to be less frequent in the newer generations of devices and also in the smaller devices. More frequent fractures in 40-mm CardioSEAL devices previously have been reported (10). The presence of a fracture does not seem to impact on the success of the procedure except in rare cases. Presumably once the device is endothelialized, the integrity of the repair no longer requires the support of the arms of the device.
The major limitation of the present study is that the cohort of STARFlex patients is relatively small and cases were from a single institution, compared with the larger, multicenter Amplatzer and surgical cohorts (11). Residual leaks are documented by TTE, which is not as accurate as TEE for documenting residual defects (7). Some patients were referred from outside institutions, and data after completion of trial protocols were not available, which may underestimate closure rates because of ongoing spontaneous resolution with time (7).
Modifications in 3 generations of devices and increased procedural experience have resulted in significantly improved outcomes for percutaneous ASD closure. The STARFlex device provides excellent outcomes, with significantly smaller device sizes than the earlier-generation devices, and compares favorably to published results of other devices and surgery.
Children’s Hospital Boston received support for clinical trials from C.R. Bard, Inc. and NMT Medical, Inc., per patient enrollment, receives royalties for sales of CardioSEAL and STARFlex devices based on a licensing agreement with NMT Medical, Inc. Drs. Gauvreau and Jenkins have received honoraria for presenting data on NMT’s behalf.
- Abbreviations and Acronyms
- atrial septal defect
- transesophageal echocardiogram
- transthoracic echocardiogram
- Received October 25, 2005.
- Revision received March 7, 2006.
- Accepted March 16, 2006.
- American College of Cardiology Foundation
- Rome J.J.,
- Keane J.F.,
- Perry S.B.,
- Spevak P.J.,
- Lock J.E.
- Divekar A.,
- Gaamangwe T.,
- Shaikh N.,
- Raabe M.,
- Ducas J.
- Amin Z.,
- Hijazi Z.M.,
- Bass J.,
- Cheatham J.P.,
- Hellenbrand W.E.,
- Kleinman C.S.
- Lock J.,
- Cockerham J.,
- Keane J.,
- Finley J.,
- Wakely P.,
- Fellows K.
- Boutin C.,
- Musewe N.N.,
- Smallhorn J.F.,
- Dyck J.D.,
- Kobayashi T.,
- Benson L.N.
- Pedra C.A.C.,
- Pihkala J.,
- Lee K.J.,
- et al.
- Du Z.D.,
- Hijazi Z.M.,
- Kleinman C.S.,
- Silverman N.H.,
- Larntz K.,
- Amplatz Investigators
- Rao P.S.,
- Berger F.,
- Rey C.,
- et al.,
- International Buttoned Device Trial Group
- Galal M.O.,
- Wobst A.,
- Halees Z.,
- et al.