• Open Access

Occlusion Devices and Approaches in Canine Patent Ductus Arteriosus: Comparison of Outcomes


  • This abstract was presented at the 2011 ACVIM Forum, Denver, Colorado.
  • This study was done at the William R. Pritchard Veterinary Medical Teaching Hospital, University of California, Davis, Dr Singh's present address is the Small Animal Specialist Hospital, Level 1, 1 Richardson Place, North Ryde, Australia, 2113.

Corresponding author: Mark D. Kittleson, Department of Medicine and Epidemiology, One Shields Avenue, University of California, Davis, CA; e-mail: mdkittleson@ucdavis.edu.



A comparison of transvascular occlusion methods for closing patent ductus arteriosus (PDA) in dogs has not been done.


To determine if clinically important differences exist between the approaches and devices currently used.


A total of 112 client-owned dogs with left-to-right shunting PDA.


Retrospective study. Records from dogs that underwent attempted transvascular PDA occlusion from January 2006 to December 2009 were examined. Dogs were placed into 4 groups: Group 1: Amplatz Canine Duct Occluder (ACDO) (transarterial) — 36 dogs; Group 2: Gianturco or MReye Flipper Detachable Embolization (Flipper) coil (transarterial) — 38 dogs; Group 3: Amplatzer Vascular Plug (AVP) (transarterial) — 23 dogs; Group 4: Flipper coil (transvenous) — 15 dogs.


The overall success rate of the procedures was high (92%) with comparable success rates among groups (87–97%). There were significantly fewer complications (< .0001) in dogs receiving an ACDO than in the remaining groups (3% for ACDO versus 26–33% for the other groups). Fluoroscopy time for the transvenous method was significantly longer (median, 13 minutes) than for the other groups (median, 6 minutes) (< .0001). Severity of residual flow 24 hours postprocedure was significantly less in the ACDO group than in the remaining groups (P = .0001–.05).


The ACDO appears superior in ease of use, complication rate, and completeness of occlusion. The remaining limiting factor with this device is patient size. Until a smaller ACDO device is marketed, coils remain the only choice for interventional closure in very small dogs.


Amplatz canine duct occluder






Amplatzer vascular plug


left atrium


minimal ductal diameter


patent ductus arteriosus


Veterinary Medical Teaching Hospital


pulmonary artery

The 1st transvascular patent ductus arteriosus (PDA) closure was performed in a dog in 1994.1 Since then, numerous reports have described minimally invasive transvascular PDA occlusion in dogs, and it now has become the treatment method of choice for most veterinary cardiologists.[1-13] The procedure historically has encompassed a variety of occlusion devices intended for human use, including thrombogenic coils,[1, 4, 10-12, 14, 15] the AVP[7, 9] and the Amplatzer Duct Occluder.[2, 3] Both transvenous and transarterial routes for deployment of occlusion devices have been used.[1, 3, 4, 9-12, 15] All techniques and devices have potential limitations and complications and no single technique appears suitable for all dogs.[1, 3, 4, 7-9] In recent years, the Amplatz canine duct occluder (ACDO),[6, 8] a nitinol mesh device specifically developed to fit the size and shape of the canine PDA, appears to have surpassed previous devices with regard to ease of use, degree of closure, and decreased complication rate.[6, 8, 13] Although there are individual reports on the use of these devices in dogs, the indications, complications, and long-term outcomes of the various devices has not been compared in 1 study. This study aims to compare the signalment, fluoroscopy time, outcomes, and complications among the 4 devices placed by 2 different approaches that were used to noninvasively close PDAs in dogs at our institution from January 2006 (when the ACDO became available at our facility) to December 2009.

Materials and Methods

Initial and follow-up records from all dogs that underwent attempted transvascular occlusion to correct PDA at the University of California, Davis, William R. Pritchard Veterinary Medical Teaching Hospital (VMTH) from January 2006 to December 2009 were retrospectively reviewed. Information extracted from the records included age, sex, body weight, presence of congestive heart failure, presence of other cardiac abnormalities, PDA morphology, minimal ductal diameter (MDD), ampulla diameter (amp), aortic diameter (Ao), type of occlusion device, vascular approach, fluoroscopy time, residual flow at time 0 (within 10–20 minutes after device deployment) and 24 hours, left atrial:aorta ratio (LA:Ao), left ventricular internal diameter in diastole:aorta ratio (LVIDd:Ao) before occlusion and at 24 hours, complications, and success rate. In addition, the MDD:Ao and amp:Ao ratios were calculated. Absence of clinically relevant pulmonary hypertension was verified by applying the modified Bernoulli equation to the ductal jet velocity.

The presence of a PDA was confirmed and severity assessed by echocardiography,2, 3 which was performed in all dogs as previously described.[16] Assessment of severity was made based on the extent of LA enlargement, left ventricular volume overload, and presence or absence of systolic dysfunction. General anesthesia was induced using a variety of sedative and anesthetic agents as deemed appropriate by the Anesthesia Service at the VMTH. Transesophageal echocardiography4 was performed if the patient was large enough in order to visualize the PDA and to obtain maximum ampulla and MDD for subsequent device selection. If the patient was not large enough, these variables were measured transthoracically and, in some, by angiography. The transvascular occlusion procedure was performed by the transarterial or transvenous route depending on the preference of the faculty clinician at the time. These procedures have been described previously in detail.[1, 4, 6, 8, 12, 15] Briefly, dogs were positioned in right lateral recumbency and the right inguinal region prepared for isolation of the proximal right femoral artery or vein. For the transarterial method, the artery was surgically isolated and the largest end-hole catheter (3–8 F)5 that could be inserted into the artery was advanced to the transverse or proximal descending aorta under fluoroscopic guidance.6 Iodinated contrast material,7 at a dosage of approximately 1 mL/kg, was rapidly injected manually to identify the position of the ductal ampulla and to measure ductal diameter in some instances.

For the transvenous method, vascular access was attempted percutaneously using a modified Seldinger technique.[17] If this was not successful after multiple attempts, the femoral vein was surgically isolated. A 4 or 5 French introducer sheath then was inserted.8 A balloon wedge catheter9 was inserted under fluoroscopic guidance into the main pulmonary artery (PA) and then passed retrograde through the PDA and into the aorta. If passage of the balloon wedge catheter retrograde through the PDA was not successful, this was facilitated by first passing a 0.018″ guide wire.11 The balloon wedge catheter then was fed over the guide wire and the wire removed. An exchange guide wire10 was passed through the balloon catheter into the descending aorta and the balloon catheter exchanged for a 4 or 5 Fr angiographic catheter.12 An angiogram was then performed as described previously. Angiograms were reviewed and ductal shape identified as type I, IIa, IIb, or III as previously described.[18]

Occlusion devices used included detachable and nondetachable Gianturco or MReye Flipper Detachable Embolization coils,13 AVP,14 and ACDO.15 The device used was chosen based on the largest gauge catheter that could be inserted in the vessel or introducer, the size of the MDD and ampulla and faculty clinician preference. The ACDO waist size was chosen to be 1.5–2.7× the MDD,[6, 8] coil diameters were generally chosen to be at least twice the MDD[1, 4, 15] and the AVP diameter was chosen to be approximately 1.3× the maximal ampulla diameter.[7, 9]

Coils were deposited within the ampulla of the PDA. If the transvenous method was used, most of the coil was placed in the ampulla with 1–2 loops spanning the MDD into the PA.[10, 12] If the transarterial method was used, the spanning method was not performed and all coil loops were placed in the ductal ampulla. The AVP was deposited in the ampulla of the PDA using the transarterial method.[9] The ACDO was always deployed by the transarterial method as previously described.[6, 8] Devices that did not appear stable or resulted in ongoing marked continuous flow were removed and replaced with a larger device. For the majority of cases, direct systolic and diastolic blood pressures were monitored during and immediately after deployment of the device and the dog was auscultated for a continuous murmur. If direct blood pressure monitoring was not possible, systolic blood pressure was measured indirectly by Doppler methodology. When closure was performed by the transarterial route, an angiogram, and occasionally transthoracic or transesophageal echocardiography, was performed to identify residual flow. For the transvenous approach, residual flow after deployment was assessed using transthoracic echocardiography.

Cefazolin16 was administered at a dosage of 30 mg/kg IV q2h during the procedure in some patients depending on clinician preference. Dogs were monitored continuously until complete recovery from anesthesia. Within 24 hours of recovery, thoracic radiographs (lateral and dorsoventral) were taken to confirm device location. Echocardiography also was performed the day after occlusion to identify any residual flow and to obtain LA and left ventricular measurements. A recheck echocardiogram was recommended 3 months later to look for delayed device embolization and residual flow, and to again obtain LA and left ventricular measurements.

Data Analysis

Patient records were retrospectively reviewed. When objective data were absent or subjectively reported, the original echocardiographic examination was reviewed by a single investigator (MKS) and objective measurements obtained. LA, LVIDd, MDD, and ampulla were indexed to the aortic diameter (LA:Ao, LVIDd:Ao, MDD:Ao, Amp:Ao) to take into account the various body sizes.[19, 20] Residual ductal flow was classified using a modified scale of 0–4 by color flow Doppler echocardiography as previously described.[21] Briefly this was as follows: grade 0 — no residual flow in the PA; grade 1 — trivial flow at the entrance of the PDA to the main PA; grade 2 — mild flow, where the jet of continuous flow passes through the PDA along the main PA but does not reach the PV; grade 3 — moderate flow, defined as a broad turbulent jet in the main PA that reaches the PV.[21] In addition, dogs in which no closure was obtained by the device (ie, the jet of turbulent flow was similar in size to before the procedure), were classified as grade 4 (severe). Dogs were arranged into 4 groups depending on the device and type of vascular access. Group 1: ACDO (transarterial), Group 2: thrombogenic coil (transarterial), Group 3: AVP (transarterial), and Group 4: thrombogenic coil (transvenous).

Statistical Analysis

The data were not normally distributed, and the Kruskal–Wallis test was used to identify differences in clinical measurements among the 4 groups. Dogs that had transvascular closure attempted using more than 1 device or route were grouped according to the initial procedure attempted for the purposes of statistical analysis. Average degree of residual flow was compared among groups using the previously described grading scheme. For categories that were significantly different, posthoc analysis using a Mann–Whitney test was done to determine which groups within each category caused the significant difference. Holm's sequentially rejective method of multiple comparison adjustment was used to control Type I errors in posthoc analyses.[22] For paired data (ie, comparison of LA:Ao and LVIDd:Ao before the procedure and 24 hours later within each group), an exact Wilcoxon signed-ranks test was used. To look for differences in the distribution of PDA morphology among groups an exact chi-square test was used. For all tests, < .05 was considered significant. Exact logistic regression models initially were created to perform univariate analysis of putative risk factors for complications, and variables with P-values < .15 were candidates for inclusion into a multiple logistic regression model. Group (with the ACDO group as reference) was included in all models as the variable of interest, and age, weight, and ductal diameter were included in all models as confounders. Only variables with improvement-in-fit likelihood ratio P-values < .05 were considered for inclusion in the final model. Results are presented as odds ratios (OR) and 95% confidence intervals (95% CI). Any individual with missing data for a particular variable was excluded from analysis for that variable. Statistical software was used.17


Transvascular occlusion was attempted in 112 dogs over the 4-year period. All dogs had a continuous murmur best heard in the left axillary region. Grade of heart murmur ranged from II to VI and there was no significant difference among groups. The number of dogs in each group were Group 1 (ACDO) — 36 dogs, Group 2 (transarterial thrombogenic coil) — 38 dogs, Group 3 (AVP) — 23 dogs, Group 4 (transvenous thrombogenic coil) — 15 dogs. Of the 15 dogs in Group 4, 7 (47%) had the femoral vein surgically isolated for catheter introduction due to unsuccessful attempts to use the modified Seldinger technique. These dogs all weighed <3.5 kg. Transesophageal echocardiography was performed in dogs large enough to accommodate an adult-sized probe. Five dogs weighed <1.4 kg and could not have transesophageal echocardiography performed.

Age ranged from 2–84 months (median, 7 months). The median age for group 1 was 10 months (range, 2–84 months), for group 2 was 6.5 months (range, 2–84 months), for group 3 was 10 months (range, 2–48 months), and for group 4 was 6 months (range, 3–36 months). Body weight ranged from 1 to 34 kg (median, 4.7 kg). The median body weight for group 1 was 12.5 kg (range, 2.6–34 kg), group 2 was 2.2 kg (range, 0.8–8.6 kg), group 3 was 6.1 kg (range, 2.3–27 kg), and group 4 was 2 kg (range, 1.2–6 kg). There was no significant difference in age among the 4 groups (Table 1). There was a significant difference in body weight among the 4 groups (< .0001). Posthoc analysis showed that groups 1 and 3 had significantly higher body weight than groups 2 and 4 (< .0001) but that groups 1 and 3 were not significantly different from each other and groups 2 and 4 were not significantly different from each other (Table 1).

Table 1. Demographic, echocardiographic, and fluoroscopic data (median and range).
 Entire population (n = 112)Group 1 (n = 36)Group 2 (n = 38)Group 3 (n = 23)Group 4 (n = 15)P
  1. a significantly different from b and c.

  2. MDD, minimal ductal diameter; LVIDd/Ao, left ventricular internal diameter in diastole:aorta ratio; LA:Ao, left atrial:aorta ratio; PDA, patent ductus arteriosus.

Age (months)7 (2–84)10 (2–84)6.5 (2–84)10 (2–48)6 (3–36).30
Body weight (kg)4.7 (1–34)12.5a (2.6–34)2.2b (0.8–8.6)6.1a (2.3–27)2.0b (1.2–6.0)<.0001
PDA MDD (mm)2.2 (1–8)2.9a (1.2–8)1.8b (1–3.1)2.3a (1.1–5)1.7b (1.1–4)<.0001
MDD/Ao0.16 (0.02–0.32)0.16 (0.02–0.33)0.16 (0.08–0.32)0.14 (0.08–0.28)0.16 (0.09–0.24).35
PDA amp (mm)6.2 (2.5–23)8a (3–23)4.6b (3–8)8.1a (3–12.1)4.4b (3–12)<.0001
Amp/Ao0.44 (0.18–0.82)0.44 (0.18–0.82)0.42 (0.25–0.7)0.44 (0.25–0.68)0.39 (0.18–0.75).9
LA:Ao (prior)1.5 (1.0–3.6)1.6 (1.1–3.6)1.5 (1.0–2.2)1.5 (1.1–3.0)1.4 (1.2–2.4).25
LA:Ao (24 hours)1.3 (1.0–2.4)1.3 (1.0–2.0)1.3 (1.0–1.9)1.3 (1.0–3.0)1.3 (1.1–2.4).98
LVIDd:Ao (prior)2.2 (1.4–3.9)2.2 (1.6–3.0)2.2 (1.6–3.2)2.1 (1.4–3.9)2.3 (1.5–3.2).4
LVIDd/Ao (24 hours)1.9 (1.5–3.5)1.8 (1.5–2.8)1.9 (1.5–3.0)1.8 (1.2–3.5)2.1 (1.5–2.5).36
Fluoroscopy time (minutes)8.0 (3–78)5.1a (2.8–20.1)8.0c (3.4–33.5)5.3a (3.2–15.2)13.0b (10.2–78.1)<.0001

Eleven dogs had evidence of congestive heart failure before the occlusion, 6 (17%) from group 1, 2 (5%) from group 2, 3 (13%) from group 3, and 1 (7%) from group 4. Ten dogs had evidence of other structural heart disease based on echocardiography. These included 4 dogs in group 1 (2 with mild subaortic stenosis, 1 with mild mitral valve dysplasia, and 1 with both mild mitral and mild tricuspid valve dysplasia), 3 in group 2 (1 with severe valvular pulmonic stenosis, 1 with mild supravalvular pulmonic stenosis, and 1 with mild mitral regurgitation due to myxomatous mitral valve degeneration), 2 in group 3 (both with mild to moderate mitral valve dysplasia), and 1 in group 4 (with mild subaortic stenosis). Five dogs had a concurrent arrhythmia on the resting electrocardiogram before the occlusion procedure. All 5 dogs were in group 1 and consisted of 2 with atrial fibrillation, 1 with atrial fibrillation and ventricular premature complexes, 1 with ventricular premature complexes only, and 1 with both atrial and ventricular premature complexes.

There was no evidence of clinically relevant pulmonary arterial hypertension in any dog, with pressure gradients across the PDA ranging from 80 to 140 mmHg (calculated from ductal jet velocity using the modified Bernoulli equation). Morphologically, the PDA in 3 dogs was classified as type I, in 86 as type IIa, in 20 as type IIb, and in none as type III. Three dogs could not be classified due to a lack of an angiogram in 2 (the femoral artery was too small to pass the catheter in 1 whereas the femoral artery tore in the other) and inability to visualize the PDA clearly in the 3rd. The morphology was evenly distributed among groups (P = .16).

The PDA MDD ranged from 1.0 to 8 mm (median, 2.2 mm). The median PDA MDD for group 1 dogs was 2.9 mm (range, 1.2–8 mm), for group 2 was 1.8 mm (range, 1.0–3.1 mm), for group 3 was 2.3 mm (range, 1.1–5.0 mm), and for group 4 was 1.7 mm (range, 1.1–4.0 mm). There was a significant difference in MDD among the 4 groups (< .0001) (Table 1). Posthoc analysis showed that groups 1 and 3 were significantly different from groups 2 and 4 (< .0001), but that groups 1 and 3 were not significantly different from each other (= .036) and groups 2 and 4 were not significantly different from each other (= .77). When the MDD was indexed to the aorta to take into account the various body sizes (MDD/Ao) this value ranged from 0.02 to 0.32 (median 0.16). For group 1, the median MDD/Ao was 0.16 (range, 0.02–0.33), for group 2 was 0.16 (range, 0.08–0.32), for group 3 was 0.14 (range, 0.08–0.28), and for group 4 was 0.16 (range, 0.09–0.24). There was no significant difference among groups (Table 1). The PDA ampulla diameter (amp) ranged from 2.5 to 23 mm (median, 6.2 mm). For group 1, the median amp was 8 mm (range, 3–23 mm), for group 2 was 4.6 mm (range, 3–8 mm), for group 3 was 8.1 mm (range, 3–12.1 mm) and for group 4 was 4.4 mm (range, 3–12 mm). There was a significant difference in amp diameter among groups (< .0001). Posthoc analysis showed that groups 1 and 3 were significantly different from group 2 and 4 (< .0001) but that groups 1 and 3 were not significantly different from each other (P = .45) and groups 2 and 4 were not significantly different from each other (P = .35). When the ampulla was indexed to the aorta in each dog (amp:Ao) to take into account the various body sizes, the range was 0.18–0.82 (median, 0.44). For group 1, the median amp:Ao was 0.44 (range, 0.18–0.82), for group 2 was 0.42 (range, 0.25–0.7), for group 3 was 0.44 (range, 0.25–0.68), and for group 4 was 0.39 (range, 0.18–0.75). There was no significant difference among groups (Table 1).

The LA:Ao ratio before the procedure ranged from 1 to 3.6 (median, 1.5). For group 1, the median LA:Ao ratio was 1.6 (range, 1.1–3.6), for group 2 was 1.5 (range, 1.0–2.2), for group 3 was 1.5 (range, 1.1–3.0), and for group 4 was 1.4 (range, 1.2–2.4). There was no significant difference in LA:Ao ratio among groups before the procedure. The LA:Ao ratio 24 hours after the procedure ranged from 1 to 2.4 (median, 1.3). For group 1, the median LA:Ao at 24 hours was 1.3 (range, 1.0–2.0), for group 2 was 1.3 (range, 1.0–1.9), for group 3 was 1.3 (range, 1.0–3.0), and for group 4 was 1.3 (range, 1.1–2.4). There was no significant difference in LA:Ao ratio among groups 24 hours after the procedure. The LVIDd:Ao ratio ranged from 1.4 to 3.9 (median, 2.2). For group 1, the median LVIDd:Ao ratio before to the procedure was 2.2 (range, 1.6–3.0), for group 2 was 2.2 (range, 1.6–3.2), for group 3 was 2.1 (range, 1.4–3.9), and for group 4 was 2.3 (range, 1.5–3.2). There was no significant difference in LVIDd:Ao ratio among groups before the procedure. The LVIDd:Ao 24 hours after the procedure ranged from 1.5 to 3.5 (median, 1.9). The median LVIDd:Ao 24 hours after the procedure for group 1 was 1.8 (range, 1.5–2.8), for group 2 was 1.9 (range, 1.5–3.0), for group 3 was 1.8 (range, 1.2–3.5), and for group 4 was 2.1 (range, 1.5–2.5). There was no significant difference in LVIDd:Ao among groups 24 hours after the procedure (Table 1). When the LA:Ao and LVIDd:Ao before the procedure was compared to that at 24 hours within each group, there was a significant reduction present for all groups and both parameters (P < .0001).


Complications occurred in 22/112 dogs (20%). Only 1 dog (3%) in group 1 (n = 36; ACDO) suffered a complication (no device large enough to close the PDA). Ten dogs (26%) in group 2 (n = 38; transarterial coil) suffered a complication. Five of these had pulmonary arterial embolization of a coil. In 4, the femoral artery tore, necessitating surgical closure of the PDA in 1 whereas the catheter procedure could be completed in the remaining 3. None of these dogs suffered clinically relevant bleeding. All 4 of these dogs were small (≤3 kg). In another dog, the artery was too small to advance a 4 Fr catheter necessitating surgical closure of the PDA. This dog weighed 0.8 kg. Six dogs (26%) in group 3 (n = 23; vascular plug) suffered a complication. These included immediate rotation of the device resulting in a marked increase in flow in 2 dogs, delayed recanalization or movement resulting in clinically relevant recrudescence of flow in 3 dogs and lack of thrombosis of the device due to Von Willebrand disease in 1 dog. Five dogs (33%) in group 4 (n = 15; transvenous coil) suffered complications. These included severe femoral vein hemorrhage requiring overnight monitoring in the intensive care unit in 1 dog in which the procedure was performed percutaneously, arterial laceration in 1 dog (this occurred during surgical isolation of the vein because the modified Seldinger technique was not successful), caudal vena caval perforation with bleeding into the retroperitoneal space in another dog in which the procedure was performed percutaneously, and inability to close the PDA in 2 dogs. There was no significant difference in complication rates among groups 2, 3, and 4 after controlling for age, weight, and ductal diameter. However, compared to group 1, complication proportions were significantly higher in group 2 (adjusted OR = 59.3, 95% CI = 2.1–1,704, = .017), group 3 (adjusted OR = 38.8, 95% CI = 2.1–700.5, = .013), and group 4 (adjusted OR = 92.2, 95% CI = 2.9–2,925, = .010).

Success for transvascular PDA occlusion was defined as a reduction of flow through the PDA to the point where it was considered hemodynamically insignificant. This was assessed by grade of residual flow using angiography in the 10–20 minutes after deployment, an increase in diastolic blood pressure in those dogs that had direct arterial blood pressure monitoring performed, disappearance of the continuous heart murmur, and echocardiographic measurements at 24 hours and 3 months. If further intervention was deemed unnecessary, the procedure was considered successful. Transvascular PDA occlusion was successful in 103/112 dogs (92%). It was not successful in 1/36 group 1 dogs (3%). This was due to the PDA MDD being too big for the largest available ACDO occlusion device (14 mm). The ACDO device in this dog was unstable, and with gentle predetachment manipulation, continued to pass through the MDD into the PDA ampulla. In this dog, a 16 mm AVP also was unstable. This dog's PDA was successfully closed by surgery. PDA closure was not successful in 3/38 (8%) dogs in group 2. In 1 dog multiple coils embolized into the PA. In the 2 remaining dogs, very small femoral arteries precluded placement of a sufficiently sized catheter in 1, whereas the artery tore in the other. The PDA in all 3 dogs was successfully closed by surgery. Transvascular occlusion was not successful in 3/23 (13%) dogs in group 3. In 1 dog, the device recanalized 2 months postplacement. A 2nd larger AVP was placed that also resulted in ongoing marked residual flow. This dog's PDA eventually was surgically ligated. In another dog, the AVP was found to have rotated within the ductus, resulting in moderate residual flow with recurrence of volume overload to the left heart at the 3-month recheck. This dog's PDA also was successfully surgically ligated. The 3rd dog was a Doberman Pinscher that was later discovered to have Von Willebrand disease. This dog's PDA was not closed completely by the AVP resulting in ongoing moderate flow and volume overload. This dog was lost to follow-up. The procedure was not successful in 2/15 (13%) dogs in group 4. In both dogs, the Flipper coils were unstable in the PDA and resulted in severe ongoing residual flow or migration into the PA and were removed. One dog's PDA was successfully closed 1 month later using an ACDO. In the other dog, 10 mm and 12 mm AVPs were used but also were unsuccessful. The dog eventually was euthanized due to financial constraints.

Total fluoroscopy time ranged from 3 to 78 minutes (median, 8 minutes). Median fluoroscopy time in group 4 (transvenous coil; 13 minutes), was significantly longer than in group 1 (ACDO; 5.1 minutes), group 2 (transarterial coil; 8 minutes), and group 3 (AVP; 5.3 minutes) (< .0001). Fluoroscopy time in group 1 was significantly shorter than in group 2 (= .0013) and group 4 (P < .0001) but not group 3 (= .43) (Table 1).

Residual flow was evaluated within 10–20 minutes after device deployment by angiography (transarterial route) or echocardiography (transvenous route). All dogs had residual flow reevaluated the next morning by echocardiography. Overall, 48/110 (44%) of dogs had some residual flow immediately after deployment (dogs in which device deployment was never attempted were excluded). In group 1, 9/36 dogs (25%) had some immediate residual flow. Of these, 7 had grade 1 flow, 1 had grade 2 flow, and 1 had grade 4 flow (the device was removed and the dog underwent surgery). After 24 hours, 1/35 dogs (2.8%) had ongoing grade 1 residual flow. In group 2, 23/36 (64%) dogs had residual flow immediately after deployment. Of these, 10 had grade 1 flow, 10 had grade 2 flow, and 3 had grade 3 flow. After 24 hours, 16/35 (45%) of group 2 dogs had ongoing residual flow. Of these, 7 had grade 1 flow, 6 had grade 2 flow, 2 had grade 3 flow, and 1 had grade 4 flow. The dog with grade 4 residual flow suffered embolization of both coils to the lungs sometime during the recovery period. Its PDA was closed successfully by surgical ligation. In group 3, 11/23 (48%) dogs had some residual flow in the immediate period after deployment. Of these, 7 had grade 1, 3 grade 2, and 1 grade 3 flow. After 24 hours 9/23 (39%) had ongoing residual flow with 5 having grade 1, 3 grade 2, and 1 grade 3. Of the dogs in group 4, 5/15 (33%) had residual flow in the immediate post-deployment period. Of these 1 was grade 1, 2 were grade 2, and 2 were grade 4. The 2 dogs with grade 4 flow had the coils removed and other devices attempted (ACDO in one and AVP in the other). After 24 hours, 3/13 (23%) continued to have ongoing residual flow with 2 being grade 1 and 1 grade 2. Inadequate numbers of dogs returned for the 3-month recheck to assess residual flow at this time. Statistical analysis of residual flow data at 24 hours showed that group 1 had significantly less ongoing residual flow at this time than the other 3 groups (P = .0001–.05). There was no significant difference in average degree of residual flow at 24 hours among groups 2, 3 and 4 (P = .15–.58).


Transvascular occlusion was successful in most dogs with an overall success rate of 92%. This result is similar to previous reports with success rates of 86% being reported for transarterial coil methods,[1, 4] 93–100% reported for AVP methods,[7, 9] 98–100% for the ACDO,[8, 13] and 91–100% for transvenous occlusion methods.[3, 10, 12] Since the development of the ACDO and commercial availability in 2007, this device has gained popularity among veterinary cardiologists for its ease of use, low potential for device migration, and completeness of ductal occlusion.[6, 8, 13] The present study confirmed previous preliminary findings in support of the ACDO as the current device of choice for closure of PDA. There were no instances of ACDO migration in this study. In contrast, coil migration into the PA when deployed by the transarterial route occurred in 13% of cases in our study. In a previous retrospective study examining the use of thrombogenic coils in 125 dogs with PDA at the same facility there was a 22% incidence of coil migration.[4] Thus, pulmonary embolization appears to be a common complication that can occur with the use of thrombogenic coils.

Previous studies have reported angiographic or echocardiographic persistence of ductal flow at conclusion of the procedure in 0–31% of dogs receiving an ACDO,[8, 13] 28–40% of dogs receiving a thrombogenic coil[1, 4, 11, 12, 15] and in 25–33% after placement of an AVP or Amplatzer duct occluder.[3, 7, 9] Color flow Doppler done 24 hours postprocedure showed 29 dogs (27%) with residual flow, but the percentage of cases varied widely depending on the procedure. Only 3% (1 dog) had residual flow after placement of an ACDO, whereas 45% of dogs occluded transarterially with a thrombogenic coil, 39% with a vascular plug and 23% with a transvenously placed thrombogenic coil had residual flow. Although most dogs had trivial to mild persistent flow, 2 dogs from group 2 and 3 dogs from group 3 had moderate persistent flow and 1 dog from group 2 had severe persistent flow. These findings support previous reports that the ACDO is not only easy to use and highly successful, it also more frequently produces complete ductal occlusion than other devices. However, in all groups of dogs, there was a significant reduction in LA and left ventricular size (as shown by the significant reduction in LA:Ao and LVIDd:Ao), indicating considerable improvement in the severity of volume overload to the left heart. Unfortunately, not enough dogs returned for the recommended 3-month recheck to statistically assess the presence of persistent residual flow among groups at this time. Previous studies indicated delayed complete closure in a substantial number of cases that have undergone coil occlusion with reported cumulative closure rates ranging from 60–90%.[4, 10, 11]

Complications that have been described for transvascular occlusion of PDA include pulmonary or systemic arterial embolism, hemodynamically relevant residual shunt, femoral artery tear, mechanical hemolysis as a result of incomplete ductal closure, left PA branch stenosis, and death. Such limitations have contributed to procedure abandonment in 11–27% of coil embolization attempts.[1, 4] The complication rate for ACDO placement was lower (3%) than the other 3 methods (26–33%) in the present study. This finding was similar to that of other reports where complication rates of 5–7% with the ACDO were seen.[8, 13] Although the AVP has a lower risk of embolization and may provide more complete ductal occlusion than thrombogenic coils, movement (rotation) appears to be a complicating factor[5] and occurred in 3 dogs in this study. Postplacement movement did not occur with any ACDO in the present study. The presence of congestive heart failure associated with a large PDA (large MDD resulting in high flow) has been shown to confer a 3-fold increase in risk of coil embolization.[4] Embolization did not occur with the use of ACDO in this study and procedural abandonment was rare even in the presence of a large PDA and congestive heart failure. One German Shepherd did have the ACDO procedure abandoned in favor of surgery due to a very large PDA (MDD, 7.5 mm; ampulla, 18 mm) being too large for the largest ACDO (14 mm) or AVP device (16 mm).

Fluoroscopy time is a consideration when contemplating any interventional procedure because limiting excessive radiation exposure is an important workplace health consideration. In this study, the transvenous route was associated with significantly longer fluoroscopy time with a median of 13 minutes versus 5–8 minutes for transarterial methods. A recent study assessing the transvenous method of transvascular closure also suggested that fluoroscopy times were prolonged, but the actual numbers were not recorded.[10] One study in humans comparing a small number of babies who underwent transarterial closure to the transvenous approach at different institutions also found increased fluoroscopy times associated with the transvenous approach.[23] However, this finding did not reach statistical significance. Nevertheless, this may be a important workplace health consideration when considering the type of transvascular occlusion to be performed.

The transarterial approach has been the most common method used for transvascular closure of PDA in veterinary medicine. In human medicine, the transvenous approach usually is preferred due to apparent lower complication rates.[24] However, this has not been proven to be the case in several previous studies comparing the different approaches in humans.[23, 25] The transarterial approach avoids right heart catheterization that can be associated with difficult retrograde PDA access, kinking of the device sheath and cardiac arrhythmias.[3, 10, 15] In 1 previous study, the author encountered several difficulties during cardiac catheterization and deployment of the Amplatzer duct occluder transvenously.[3] In some dogs, it was difficult to pass the catheter or guide wire across the pulmonic orifice of the ductus arteriosus. This increased fluoroscopic and procedural time. In small dogs, the introducing sheath was reported to have a tendency to kink as it crossed the pulmonic valve.[3] In addition, if it is found during the procedure that more than 1 coil is required occlusion would be difficult to accomplish by the transvenous approach and the arterial approach then would be required.[12]

Hemorrhage from the catheter site was a substantial problem in previous studies utilizing the percutaneous transvenous method,[3, 12] and mild bruising and bleeding also were noted to occur in this study in some dogs that had the modified Seldinger venous approach despite prolonged digital occlusion. In 1 dog, this complication was severe, resulted in a 30% decrease in packed cell volume and required transfer to the intensive care unit for monitoring. The shape of the hind limb and close proximity to the inguinal region precludes application of an effective occlusive bandage. Thus, it previously has been suggested that one needs to anticipate the need to ligate the femoral vein if substantial bleeding occurs.[3] On the other hand, the transarterial approach resulted in femoral artery laceration in 4 dogs in this study. All dogs were small (≤3 kg). Although no dog suffering femoral artery laceration experienced clinically relevant hemorrhage, this is a potential life-threatening complication that needs to be considered when contemplating this approach in very small dogs.

There was a significant difference in MDD and amp among the 4 groups with groups 1 and 3 being significantly larger than groups 2 and 4 but with groups 1 and 3 not significantly different from each other and groups 2 and 4 not significantly different from each other. However, when MDD was indexed to the aorta to take into account the various body sizes, there was no significant difference among groups.[19, 26] This finding suggested that the larger MMD and amp in groups 1 and 3 was due to the larger body size of these dogs when compared to groups 2 and 4 and, when taking into account body size, there was no significant difference in the size of the PDA among groups.

Both the ACDO and the AVP were used more successfully than coil methods in larger dogs and those with larger MDD, but the ACDO was associated with a lower complication rate and more complete closure than the AVP. However, the smallest dog in which the decision to place an ACDO was made and was successful was 2.6 kg whereas the PDA of dogs ≤2.5 kg were closed successfully with thrombogenic coils by both the transarterial and transvenous routes. The smallest dog successfully receiving a coil in the transarterial group was 1.0 kg and in the transvenous group was 1.2 kg. Despite recent studies suggesting that the transvenous approach may be more useful in very small dogs,[10-12] there was no significant difference between body weights in dogs receiving transarterial or transvenous coils in this study. However we did find that very small dogs (<3 kg) were more likely to have complications such as a femoral arterial laceration or inability to place a catheter in the smaller diameter and less distensible femoral artery (5 dogs in this study) warranting consideration of the transvenous approach in very small dogs. If the currently developed low profile ACDO for dogs ≤3 kg18 becomes commercially available, this limiting factor may be circumvented.

Several limitations need to be considered when interpreting the results of this study. Being a retrospective study, patients were not randomized to treatment groups introducing substantial bias in comparisons. In addition, results are limited by the accuracy of patient records. All procedures in this study were performed by a resident supervised by a faculty member. Device selection was not randomized but assigned by faculty clinician preference. It was not possible to perform clinician analysis to look for bias involving clinician preference because the faculty member involved usually was not recorded. Despite recommendations for 3-month recheck, a substantial number of dogs were not returned, resulting in inability to perform meaningful statistical analysis at this time. It is highly likely that a number of coil cases would have had more complete closure by this recheck.[4, 10, 11]

Determining appropriate treatment is important in assuring optimal outcome. Successful PDA closure depends on device selection and typically is based on ductal morphology and dimensions as well as patient size. The results of this study support the ACDO as the device of choice for the majority of PDA occlusions although, currently, a small subset of dogs (≤2.5 kg) still may require the transvenous or transarterial coil approach for transvascular closure because the currently available ACDO device is too large for transarterial delivery in these dogs.


This study was not funded by a grant.


  1. 1

    Miller MW, Meurs KM, Boswood A. Echocardiographic assessment of PDA after occlusion. In: Proceedings of the 14th Annual Veterinary Forum, ACVIM, San Francisco, CA; 1994:305–306 (abstract)

  2. 2

    Phillips iE33. Phillips Medical Systems, NA, Bothell, WA

  3. 3

    Acuson 128 XP/10. HP sonos 5500, Hewlett Packard, Andover, MA

  4. 4

    Phillips transesophageal ultrasound transducer. Phillips Medical Systems

  5. 5

    Cook MPA non tapered end hole catheter, Cook Veterinary Products Inc, Bloomingdale, IN

  6. 6

    GE OEC 9800, OEC Medical Systems Inc, Salt Lake City, UT

  7. 7

    Hypaque, Aversham Health Inc, Princeton, NJ

  8. 8

    Fast-Cath, St. Jude Medical, Minnetonka, MN, or Cook Mullins introducer set, Cook Veterinary Products Inc

  9. 9

    Balloon Wedge Pressure Catheter, Arrow International Inc, Reading, PA

  10. 10

    Fixed Core Wire Guide, Cook Medical Inc

  11. 11

    Glidewire, Angled Flexible Tip, Terumo Medical Corporation, Somerset, NJ

  12. 12

    Beacon, Straight Tip Royal Flush Plus Angiographic Catheter, Cook Medical Inc.

  13. 13

    Cook embolization coils or MReye Flipper Detachable Embolization Coil, Cook Medical Inc.

  14. 14

    Amplatzer Vascular Plug, AGA Medical Corp., Plymouth, MN

  15. 15

    Amplatz Canine Duct Occluder, Infiniti Medical, LLC, West Hollywood, CA

  16. 16

    Ancef, Cefazolin Sodium, Bristol-Myers Squibb Co, Princeton, NJ

  17. 17

    StatXact 8.0, Cytel Software Corp., Cambridge, MA

  18. 18

    Olson JLC, Tobias AH, Stauthammer CD, et al. Minimally invasive per-catheter patent ductus arteriosus occlusion in small dogs (</=3 kg): Preliminary results. In: ACVIM Forum Proceedings, Anaheim 2010 (abstract)