Analyzing the Failures of Percutaneous Closure of the Patent Ductus Arteriosus in Patients Over 5 kg

Abstract: Percutaneous closure of the patent ductus is the gold standard therapy. Our aim was to analyze our failures between 2001 and 2010. Methods. All patients over 5 kg benefited from a transcatheter attempt at duct closure. Coils and Amplatzer duct occluder (ADO) I were used before 2008, and ADO I and ADO II afterward. The failure was recovered when another percutaneous attempt was successful and definite when surgery was needed. Results. There were 138 patients. Coils were used in 22 patients (16%), ADO I in 74 (54%), and ADO II in 42 (30%). Immediate and 6-month closure rates were 55% and 100% for coils, 40% and 96% for the ADO I, and 74% and 93% for the ADO II, respectively. There were no failures in the coil group, and 3 failures in each of the ADO I and ADO II groups. Among the 3 ADO I failures, 1 was recovered after device migration into the abdominal aorta. The 2 other failures were definite, due to immediate device protrusion, once in the aorta and once in the pulmonary artery. One of the 3 ADO II failures was definite, due to protrusion into the aorta, 10 days following the procedure. The two other failures were due to immediate device migration into the pulmonary artery, and were both recovered. Conclusions. 97% of ducts can be closed percutaneously. The combination of coil and ADO I gives excellent results. Failed attempts with the ADO II were bailed out by the ADO I.

J INVASIVE CARDIOL 2012;24(9):434-438

Key words: Amplatzer duct occluder I, patent ductus

_____________________________________________________

The percutaneous closure of almost any patent ductus arteriosus (PDA) beyond the early infancy, using detachable coils and Amplatzer duct occluders (ADO) in a complementary way, has been the gold standard therapy over the last decade.^1-3 Nevertheless, some large PDAs and/or unusual anatomical PDA types, particularly in small infants, still challenge the interventional cardiologist.^4,5 Choosing the adequate device is based on several important factors including anatomy of the PDA, the diameter variations during the cardiac cycle, and the physical characteristics of all available devices. Despite extensive experience with a large variety of new devices, surgery is still a reasonable option in rare selected cases. We hereby report, and analyze, our own “avoidable” and “probably unavoidable” failures of percutaneous closure of PDA.

Methods

Between December 2001 and December 2010, all patients weighing more than 5 kg referred to our institution for PDA closure benefited from a transcatheter attempt and were enrolled in this retrospective study.

Before June 2008, coils were used for the closure of PDA <2 mm in diameter and ADO I for larger ones. With the advent of the ADO II in June 2008, our protocol was modified; coils were not used anymore, and the ADO II was used for ducts <5.5 mm in diameter and the ADO I for larger ones. Surgery was offered only after the failure of percutaneous attempts. Failures were considered to be “recovered failures” when the duct could be closed during a second percutaneous attempt. The failure was considered “definite” when the patient was referred for surgical closure. Informed written consent was obtained from all the adult patients and the children’s parents.

Technical considerations. All interventions were performed by the same operator (ZS). The technical details with the different devices used were reported earlier. Briefly, under general anesthesia in children, following the intravenous administration of 50 IU/kg heparin and 30 mg/kg cephazolin, the femoral artery was catheterized by a 4 Fr or 5 Fr catheter. An aortogram was then obtained in the lateral projection to determine the shape and diameter of the PDA, which is defined as the minimal restricted zone distance. In large PDA (>3 mm), the diameter variation in systole and diastole was carefully studied to define the maximal diameter during the cardiac cycle. In some patients, supplemental projections were needed for a more precise delineation of the PDA. In 4 adults, balloon sizing was necessary for accurate measuring.

Arterial access was used in all cases of coil closure. Venous delivery was used in all cases of ADO I. Most ADO II (83%) were delivered through an arterial approach. The size of the ADO I was 2 to 3 mm larger than the minimal ductal diameter, except in 3 patients in whom the ADO I was generously oversized on purpose and totally deployed in the PDA body.

The size of the ADO II was also chosen according to the size of the duct. The ducts were anatomically classified according to the Krichenko classification.1 For non-tubular ducts, the device was oversized in comparison to the narrowest pulmonary diameter in order to place the connecting waist on the aortic side of the duct. In type C ducts, the diameter of the tubular part imposed the choice of an ADO II waist diameter from 3 mm (for ducts narrower than 2.5 mm) to 6 mm (for ducts up to 5.5 mm in diameter).  The length of the ADO II was chosen depending on the PDA’s length: 4 mm for PDA shorter than 5 mm, and 6 mm for PDA between 5 and 12 mm in length. In the PDA with an aortic ampulla (type A and E ducts), the length of the device was calculated according to the supposed point of arrest of the aortic disc into the duct’s ampulla.

Before detaching the device, an aortogram was obtained to check for its position. Additionally, if the device seemed protruding in the pulmonary artery or the aorta, a pulmonary or aortic pull-back pressure tracing and angiogram were obtained to rule out any significant obstruction. Once in a correct position, the device was released by rotating the delivery cable in a counter-clockwise manner. One last aortogram was obtained 5 minutes following the release to confirm the final position of the device, and check for any residual shunt or aortic obstruction. All patients were discharged 24 hours following the procedure. Before discharge, a chest x-ray and Doppler echocardiography were obtained. Special attention was paid to residual shunts and left pulmonary or aortic isthmic obstruction. Urinalysis was performed only in case of residual shunting. Patients were scheduled for cardiology consultation and echocardiography at 10 days, then at 1, 3, 6, and 12 months. Results are expressed as percentages and mean values with standard deviation.

Results

There were 138 patients, with 84 females (61%) and 21 adults (15%). Patient characteristics are listed in Table 1. The mean age was 6.4 ± 10.2 years. The mean PDA minimal diameter was 3.11 ± 1.41 mm (range, 1-7.4 mm). According to Krichenko classification, the PDA was type A in 103 patients (75%), type B in 4 patients (3%), type C in 14 patients (10%), type D in 4 patients (3%), and type E in 13 patients (9%).

Coils were used in 22 patients (16%), ADO I in 74 patients (54%), and ADO II in 42 patients (30%). Mean fluoroscopy time was 3.68 ± 1.14 minutes (range, 2-5 minutes) for detachable coils, 7.08 ± 2.27 minutes (range, 3.5-12.8 minutes) for the ADO I, and 4.80 ± 2.52 minutes (range, 1.6-11 minutes) for the ADO II. Mean procedural time (from the moment vascular access was obtained until it was withdrawn) was 23.75 ± 11.39 minutes (range, 10-35 minutes) for detachable coils, 50.30 ± 12.29 minutes (range, 30-100 minutes) for the ADO I, and 35.89 ± 12.10 minutes (range, 20-60 minutes) for the ADO II.

A 1-month follow-up assessment was completed scrupulously and systematically in all patients. Follow-up duration ranged from 1 to 24 months, with a mean follow-up duration of 20.6 months.

Failures (Table 2; Figures 1 and 2). Overall, there were 6 failures in 6 patients (4.3%): 3 recovered failures (2.1%), and 3 definite failures (2.1%). There were no failures in the coil group, 3 failures in the ADO I group (1 recovered), and 3 failures in the ADO II group (2 recovered).

For detachable coils, immediate closure was achieved in 55% of the cases (12/22 patients), rising to 82% at 24 hours, and to 100% at 6 months.

For the ADO I, immediate closure was achieved in 40% of the cases (30/74 patients), rising to 94.6% (70/74 patients) at 24 hours, and to 96% (71/74 patients) at 6 months. Thus, the ADO I failed to close the PDA in 3 patients. The first patient (Patient 1 in Table 2 and Figure 1) was a 3-year-old, 17 kg boy with a 4 mm type D PDA which was initially closed with an 8-6 ADO I. The Amplatzer asymptomatically migrated to the abdominal aorta before discharge from the hospital. The shunt seen on the follow-up serial echocardiograms was thought to be a residual shunt, and the migration was diagnosed only 1 year later. We believe that this migration occurred on the second day, because the predischarge chest x-ray examined retrospectively 1 year later did not show the radio-opaque markers of the ADO in the usual position. The device was embedded asymptomatically in the abdominal aorta, facing the celiac artery (Figure 3A, black arrow). Attempts at its percutaneous removal 1 year later failed due to strong adhesions to the aortic wall; hence, it was left in place since it wasn’t obstructive. This failure was considered recovered, as the PDA was percutaneously closed using a 14-12 ADO I (Figure 3B); the device was oversized to achieve incomplete deployment of the aortic disc inside the PDA (Figure 4). The first ADO is still in the abdominal aorta with no gradient on the serial Doppler studies for 6 years now. The second patient (Patient 2 in Table 2 and Figure 1) was a 7-month-old who weighed 6.5 kg, with a 2.2-mm wide and 9-mm long type C PDA. When it was advanced and correctly positioned in the PDA, the 6-4 ADO I became severely kinked with a very narrow angle, and thus kept sliding backward and protruding into the pulmonary artery during every release attempt. The device wasn’t released, the procedure was abandoned, and the patient was referred for surgery 3 months later. This happened before the ADO II era. The third ADO I failure (Patient 3 in Table 2 and Figure 1) was a 6-month-old, 7 kg baby girl with a type B, 6.6 mm duct; the 10-8 ADO I totally obstructed the aortic flow and was retrieved before releasing. The patient was also referred for surgery.

For the ADO II, immediate closure was achieved in 73.8% of the cases (31/42 patients), rising to 95.2% (40/42 patients) at 24 hours, but decreasing again at 10 days to 92.8% (39/42) and remaining like that at 6 months. The ADO II device that failed secondarily at 10 days was in a 5-month-old, 6.6 kg baby girl, with a 3 mm type A PDA (Patient 4 in Table 2 and Figure 2). It was successfully closed with a 6-6 device with no residual shunt or aortic obstruction. However, the 10-day follow-up showed a residual shunt with severe aortic obstruction due to a secondary kinking of the aortic retention disc (Figure 5). The device was surgically retrieved and the PDA ligated. There were also 2 recovered failures in the ADO II group (Patients 5 and 6 in Table 2 and Figure 2); a 2.5-year-old, 10 kg girl with a 4.6 mm type A PDA, and a 3-year-old, 23 kg boy with a 4.8 mm type A PDA. Both were initially closed with 6-6 ADO II, and both devices migrated to the left pulmonary artery, a few minutes following their detachment. The devices were immediately retrieved percutaneously in both patients. In the first patient, the duct was closed with a 12-10 ADO I during the same procedure. In the second patient, the duct was closed with a 10-8 ADO I 6 months later.

No other major complications were noted on follow-up. Urinalysis at 24 hours, for the detection of hemolysis, was negative in all patients presenting a residual shunt at the end of the procedure. All patients were discharged 24 hours following the procedure without treatment.

Discussion

Studies have shown that percutaneous closure of almost any PDA beyond early infancy can be efficiently performed using detachable coils (Cook Cardiology) for small type A ducts, and the ADO I and II (AGA Medical) for almost all duct types and sizes.^1-3 However, despite the availability of multiple devices, and the vast experience of interventional cardiologists, failures have been encountered in closing some large ducts, particularly in small infants. These include procedural failure, significant residual shunts, or device migration.^4-8 Some reported failures may have been avoidable, and have been secondarily recovered through immediate or delayed percutaneous closure.^9-11 Some failures may ultimately need surgery. In order to learn how to avoid surgery beyond the neonatal period, we reviewed our failures for the last 10 years.

Interestingly, all 3 recovered failures were attributed to migration of the device, and all 3 definite failures referred to surgery were due to the protrusion of the device, and obstruction of the aorta or pulmonary artery. Concerning migration, one may assume that if an ADO is completely deployed, secondary migration should be obviously due to the small size of the device relative to the diameter of the PDA; this assumption was confirmed in our series by the successful secondary occlusion of the ducts using larger devices in all 3 patients. The patient illustrating best the advantages of device oversizing was Patient 1. In this patient, the measured minimal diameter of the 20-mm long type D PDA was around 4 mm. An 8-6 ADO I was placed in the aortic restriction, which resulted in the migration of the device. This failure was recovered by modifying the deployment technique of the ADO I; the device was oversized on purpose (14-12 ADO I), to obtain an incomplete deployment of the aortic disc inside the duct. The excessive length of this PDA prevented protrusion of the device into the pulmonary artery after its release.³ This technique was successfully reiterated in other patients with type C ducts measuring more than 5 mm. However, care must be taken with device oversizing; generous oversizing may tear the duct wall. Oversizing needs also a small restriction area somewhere in the duct. In a 30 kg, 14-year-old patient with severe pulmonary hypertension and a short type A PDA, the ADO I was largely oversized, but normally deployed, so that the pulmonary narrowing transformed the pulmonary part of the device in a retention disc. This prevented aortic migration of the ADO that might have been facilitated by pulmonary hypertension. This technique is not applicable in small infants.

The absence of restriction, taken together with small weight and a very sharp angle at the pulmonary insertion, resulted in another kind of failure in Patient 2, a 7-month-old, 6.5 kg patient with a 2.2-mm wide, 9-mm long, type C PDA. The 6-4 ADO I couldn’t be detached in the PDA. At every attempt at deployment, the device was pulled away, protruding into the pulmonary artery, most probably due to the heaviness and the rigidity of the delivery cable used in the first-generation ADO. We suppose that this long type C duct would have been easily occluded with an ADO II, unavailable at that time. Finally, the device was retrieved and the PDA was surgically ligated 3 months later.

Since its introduction in our practice, the ADO II proved to be a useful additional tool for PDA closure. It should be noted, however, that the ADO II is not yet available in the United States, and that phase 2 multicenter trials have just been completed. Complete definitive PDA closure was more rapidly achieved with the ADO II in our series — immediate closure in 74% of the patients, versus 40% with the ADO I. However, this finding proved to be irrelevant, since both devices were equally successful at 24 hours (95%). The ADO II has a 72 or 144 inner wire mesh, and most importantly, a 144 outer wire mesh, which promotes better occlusion, despite the lack of fabric that decreases the profile of the device. It is the device of choice for type C PDAs and ducts smaller than 5.5 mm.¹⁰ It shows a superior stability after deployment due to its symmetrical design.¹⁰ Nevertheless, the rate of migration in our series, as well as in the literature, was higher compared to ADO I.^3,7,11 All of our failures with the ADO II have occurred in large, type A ducts with ampullas, in particular when the size is chosen to fit the minimal pulmonary diameter rather than the aortic ampulla. These ADO II failures may be also related to both its physical features, such as flexibility and malleability, and the morphology of the ducts. In our opinion, one of the big advantages of the ADO II, flexibility, accounted for its failure in Patient 4. Flexibility, due to the presence of 2 articulations, became a disadvantage in this relatively young 6.6 kg baby. The aortic disc was not well applied to the aortic wall, and was continuously hit with the high-pressure stream of the aortic isthmic blood flow; this resulted in the secondary kinking and elongation of the device, with a hemodynamically significant aortic obstruction.

Other theoretical advantages of the ADO II may be incriminated in failures and act like double-edged swords. The symmetrical design of the device makes the arterial delivery more rapid and tempting, despite less control on the aortic disc. The multiple available sizes make it difficult to choose the correct size. A shorter device may lead to disc elongation and a loss in central disc diameter, thus causing a residual leak, or even dislodgment and migration. On the other hand, a longer device may lead to disc protrusion in either the aorta or pulmonary artery. Finally, malleability of the device may be linked to a higher rate of migration.¹² In Patient 5 (10 kg, 4.6-mm wide, 11.5-mm long, type A PDA), we probably underestimated the duct diameter and didn’t take into account the cyclic variations (systole and diastole) which are significant in large ducts: a 30% variation for a 4.6 mm duct results in a diameter >5.5 mm, reaching the device’s superior limit. The ADO II migrated to the left pulmonary artery and was immediately retrieved. The duct was secondarily closed with an ADO I. In large, non-tubular ducts, and in short ones with a large aortic ampulla and fast tapering, the ADO I should probably be the first choice. The second ADO II migration occurred in a 23 kg patient with a very unusually shaped 19-mm long PDA (Figure 2; Patient 6); it had a large pulmonary ampulla, an uncommon tubular part, a 4.8 mm restriction, and then an aortic ampulla. The pulmonary ampulla was considered to be an aneurysm of the pulmonary artery and was excluded from the duct; hence, the rest of this duct could be classified as a classical type A PDA with fast tapering. With this strategic approach, the virtual reconstruction with a 6-6 ADO II seemed appropriate. Once again, the ADO II migrated to the left pulmonary artery a few minutes following its release. In this case, too, it was retrieved and the PDA was closed with an ADO I.

Conclusion

More than 97% of PDAs in patients over 5 kg can be closed percutaneously. Failures were encountered equally with the ADO I, and the relatively recent ADO II. However, we feel that the ADO II was less successful in large, type A PDAs; oversizing in this case is recommended, and the central stent should be placed to the right of the pulmonary restriction and the aortic disc should be firmly applied to the aortic wall through venous insertion whenever possible. The ADO I is still an excellent device, bailing out failed attempts with the ADO II. The combination of coil and ADO I in type A PDAs gives excellent results. Surgery is still needed in a few patients, particularly those less than 7 kg, with large ducts.

References

1. Krichenko A, Benson LN, Burrows P, et al. Angiographic classification of the isolated, persistently patent ductus arteriosus and implications for percutaneous catheter occlusion. Am J Cardiol. 1989;63(12):877-880.
2. Gudausky TM, Hirsch R, Khoury PR, et al. Comparison of two transcatheter device strategies for occlusion of the patent ductus arteriosus. Catheter Cardiovasc Interv. 2008;72(5):675-680.
3. Saliba Z, El-Rassi I, Helou D, et al. Development of catheter-based treatment of patent ductus arteriosus: a medium-sized centre experience. Arch Cardiovasc Dis. 2009;102(2):111-118.
4. Saliba Z, El-Rassi I, Abi-Warde MT, et al.  The Amplatzer duct occluder II: a new device for percutaneous ductus arteriosus closure. J Intervent Cardiol. 2009;22(6):496-502.
5. Morgan G, Tometzki AJ, Martin RP. Transcatheter closure of long tubular patent arterial ducts: The Amplatzer duct occluder II — a new and valuable tool. Catheter Cardiovasc Interv. 2009;73(4):576-580.
6. Thanopoulos BV, Eleftherakis N, Tzannos K, Stefanadis C, Giannopoulos A. Further experience with catheter closure of patent ductus arteriosus using the new Amplatzer duct occluder in children. Am J Cardiol. 2010;105(7):1005-1009. 
7. Forsey J, Kenny D, Morgan G, et al. Early clinical experience with the new Amplatzer ductal occluder II for closure of the persistent arterial duct. Catheter Cardiovasc Interv. 2009;74(4):615-623.
8. Abadir S, Boudjemline Y, Rey C, et al. Significant persistent ductus arteriosus in infants less or equal to 6 kg: percutaneous closure or surgery? Arch Cardiovasc Dis. 2009;102(6-7):533-540.
9. Bilkis AA, Alwi M, Hasri S, et al. The Amplatzer duct occluder: experience in 209 patients. J Am Coll Cardiol. 2001;37(1):258-261.
10. Bhole V, Miller P, Mehta C, Stumper O, Reinhardt Z, De Giovanni JV. Clinical evaluation of the new Amplatzer duct occluder II for patent arterial duct occlusion. Catheter Cardiovasc Interv. 2009;74(5):762-769.
11. Barcin C, Kursaklioglu H. Editorial: late-term displacement of Amplatzer duct occluder II device following percutaneous closure of patent ductus arteriosus in an adult. J Interv Cardiol. 2010;23(4):368-369. 
12. Beck C, Laser KT, Haas NA. Failure of the Amplatzer ductal occluder II: kinking of the aortic retention disk at 24 hours. Catheter Cardiovasc Interv. 2010;75(7):1100-1103.

_____________________________________________________

From the Departments of Pediatric Cardiology & Cardiac Surgery, Saint Joseph University, Hotel-Dieu de France Hospital, Beirut, Lebanon.
Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. The authors report no conflicts of interest regarding the content herein.
Manuscript submitted March 8, 2012, provisional acceptance given March 27, 2012, final version accepted April 23, 2012.
Address for correspondence: Issam El-Rassi, MD, Hotel-Dieu de France Hospital, PO Box  166830, Beirut, Lebanon. Email: issam.rassi@gmail.com