In this observational multicenter study, a retrospective chart review of all children aged less than 10 years, having undergone a cardiac, vascular, or conduit occlusion procedure using AVP II or IV between January 2007 and March 2020, was conducted in 6 academic centers from the same European country. All cardiovascular abnormalities were potential candidates for AVP II or IV implantation. All patients were treated and followed-up through a dedicated nationwide network. All vascular abnormalities were potential candidates for AVP device implantation whenever another “on-label” device could not be selected, such as the one used during some PDA closures. All procedures contributing to this work comply with the ethical standards of the relevant national guidelines on human experimentation, and with the Helsinki Declaration of 1975, as revised in 2008. Informed consent was obtained from each study participant’s parents or legal guardians, and the Institutional Review Board of one of the study centers approved all study procedures (IRB 2020/053).

The tri-lobe AVP II (4 to 22 mm, 5 to 9 Fr delivery catheter) was first delivered in 2007. A few years later, the highly flexible AVP IV was developed for the adult peripheral vascular structure. The AVP IV (4 to 8 mm; length of 10 to 13.5 mm delivery catheter, 4 to 5 Fr) is adequate for highly tortuous vessels of less than 6 mm in diameter [16,20,21]. Both AVP II and IV are easily re-sheathable and repositionable after partial or complete deployment until best possible position is obtained for optimal release. The main AVP II and IV characteristics are summarized in Fig. 1.

Each device was chosen by the interventional paediatric cardiologists in order to obtain adequate occlusion without abnormal protrusion into nearby structures. Device selection was based on multiple parameters such as lesion anatomy (type, size, length), blood flow crossing the defect, vascular access, best technical approach, and finally, the availability of an appropriate landing zone. Due to its property of rapid occlusion and wide availability of sizes, AVP II was selected more frequently for the occlusion of large vessels (diameter > 6 mm), even in very young children. AVP IV was more often selected for closing smaller but longer and tortuous vessels, as well as small surgical conduits.

Overall, AVP was usually chosen 20% to 50% larger than the arterial native vessel size measured on initial angiography [12–14] considering the narrowest landing zone, as well as immediate upstream and downstream diameters. For vein and surgical conduits, AVP was more often upsized to 140%–150%. For non-tubular PDA, the chosen AVP size was between 120% and 150% of the minimum diameter, most often localized at the pulmonary end [14]. An over sizing ≥150% was always considered in case of severe pulmonary hypertension. Regarding large tubular PDAs, device size and selection was defined according to both maximal diameter and vessel length in lateral and right anterior oblique projections, with a device measuring 120%–160% of the minimal diameter and at least 1–2 mm more than the diameter of the largest PDA portion. Furthermore, our learning curve experience with AVP II in closure of very large tubular PDA included a cautious “pull through” with immediate device upsizing whenever needed [14]. It is to be noted that half of the participating centers always considered AVP II in infants for the occlusion of tubular PDAs > 6 mm, which would not have been possible with other commercially available devices, while the remaining centers referred these small patients for surgical PDA ligation [14].

Following parent/legal guardian consent, interventions were performed in a digital catheterization laboratory under conscious sedation or general anesthesia, according to age, procedure type, and team practice. All patients received IV heparin and standard antibiotic prophylaxis. Anterograde and retrograde route were used according either to the choice of the operators, the lesion to occlude, type and size of the selected device, findings during the procedures, or requested size of the delivery sheath. A few patients with either venous fistulae, partial cavo-pulmonary connection, single ventricle circulation, or interruption of the inferior vena cava, requested exclusive internal jugular vein access. All devices were deployed through a delivery sheath (AVP II) or a diagnostic catheter (AVP IV) using previously described techniques [6,10,12,21]. An additional echocardiographic guidance was performed whenever indicated, mainly in small infants [21]. For occlusion of large PDA using AVP II, the usual implantation technique was to contain 2 or 3 of the disks inside the body of the PDA whenever possible. Congenital artery fistulae (CAF) were closed with AVP IV only, predominantly in neonates and infants, using either a retrograde or antegrade approach, with or without arterio-venous loop. In all cases, once adequate position and occlusion without significant residual shunt was demonstrated, the AVP was released by unscrewing counter-clockwise. A final angiogram was usually performed to assess any residual shunt.

Demographic data and procedure characteristics such as vascular access, occlusion type, and implanted AVP types and number were recorded. PDA morphologies were defined according to Krinchenko’s classification [22]. To determine the real oversizing of the AVP plug, the ratio of plug diameter to either minimal vessel diameter or to patient weight was measured. In addition, considering the occlusion of very large PDAs in low weight patients, the ratio of plug size to minimal PDA diameter, as well as the ratio of minimal PDA diameter to patient weight, were calculated. Residual shunt presence was described. Additional interventions performed during the same procedure were listed.

Immediate technical success was defined as successful plug delivery in an adequate position with no more than mild intra-prosthetic residual shunt and without residual para-prosthetic shunt around the device or compression of nearby structures. Occlusion rate was monitored immediately after the procedure, at day 1 and 6 months later using echocardiography or CT scan whenever indicated. Adverse events (severity level ≥3) were defined according to the previously published CHARM (Catheterization for congenital Heart Disease Adjustment for Risk Method) model (Supplemental Table 1) [23,24]. An adverse event of severity level 3 refers to transient changes in condition requiring treatment, with a condition returned to baseline after monitoring or moderate intervention (reversal agent, additional medication, and transfer to the intensive care unit for monitoring). Major Adverse Events (MAE, severity level ≥4) included any need for a new percutaneous procedure or surgical intervention. Cardiovascular event occurrence or death was monitored during at least 6-months post procedure.

Baseline patient and procedure characteristics were described and compared according to AVP type and MAE occurrence. Quantitative variables were described using mean ± standard deviation or median (minimum-maximum range). Categorical variables were presented as absolute values and percentages. Group comparisons were conducted with Student t-test, Wilcoxon-Mann-Whitney test, Chi-square test, or Fisher’s exact test. Univariate and multivariate logistic regression models were fitted to assess the independent effect of predictors on MAE risk. Receiver Operating Curve (ROC) analysis was used to establish optimal cut-off points for MAE prediction by the following ratios: (a) AVP plug diameter to patient weight; (b) AVP diameter to minimal vessel diameter; (c) AVP plug diameter to minimal PDA diameter; (d) minimal PDA diameter to patient weight. Variables with significant association in univariate analysis (p < 0.25) were retained for multivariate models. All statistical analyses were performed using SAS software (version 9.4, Cary NC, USA), with p-values < 0.05 considered statistically significant.

Over a 14 year-period, 125 children under 10 years (57.6% female, mean age: 2.8 ± 3.1 years, mean weight 13.9 ± 13.7 kg) underwent 136 procedures using either AVP II and/or AVP IV for the percutaneous treatment of 147 different lesions. Overall, 169 plugs (60 AVP II and 109 AVP IV) were successfully implanted out of the 171 initially engaged. As such, two procedures failed due to poor patient tolerance requiring withdrawal before release. The first procedure failure involved an infant who developed immediate left pulmonary artery (LPA) stenosis (Fig. 2) during closure of a large PDA (AVP II). The second failure concerned another patient in whom the obliteration (AVP IV) of a lobar sequestration collateral artery induced an immediate acute chest pain syndrome.

A wide variety of procedures were carried out throughout the study period (Fig. 3).

Overall, 55 PDA occlusions were performed with a minimal diameter of 5.5 ± 1.6 mm and different morphologies [22]: 47.3% type C (tubular morphology, Fig. 4), 7.3% type A, 21.8% type D, 3.6% type D/E, and 20% type E.

Furthermore, 80% of PDA closure procedures included both arterial and venous femoral access with arterio-venous loop, while 7 AVP II were implanted into PDA by venous route only, and 4 plugs (3 AVP II) by a pure arterial approach. Five AVP II, as well as 2 AVP IV, were implanted to replace “on-label” PDA devices. Patient and procedure characteristics according to AVP II and IV “Off-label” use in children are further summarized in Table 1.

Successful closures of anomalous or aberrant systemic arterial supply to a lung segment was achieved using 18 AVP IV and 2 AVP II. Six uncomplicated closures of single or multiple congenital artery fistulae (CAF) using 6 to 8 mm AVP IV in infants and neonates were also completed, for which delayed post-procedure angiograms demonstrated the absence of residual shunt or coronary thrombosis. The effective closures of 2 peri-membranous Ventricular Septal Defects (VSD) with aneurysm (AVP II) as well as 1 postoperative VSD (AVP IV) and 2 muscular VSDs (AVP IV, Fig. 3L) were also achieved. The miscellaneous group included single procedures such as closure of fenestration (Fontan circulation), closure of congenital fistula of the left aortic sinus to the left atrium (CAF “like”), ductus venosus closure, or left hepatic artery occlusion.

Overall, the median oversizing of devices was 130%, but was higher with AVP II than with AVP IV (1.4 vs. 1.3, p < 0.001), while the ratio of plug size to patient weight was also higher for AVP II implantations (1.3 vs. 0.5, p < 0.001). Moreover, in comparison to AVP IV, the use of AVP II was characterized by larger devices (mean diameter: 8.0 (4.0–20.0) mm vs. 6.0 (4.0–8.0) mm; p < 0.001) in younger children (53.5% of infants ≤1year vs. 45.3%, p = 0.37) often weighing less than 5 kg (37.9% vs. 17.2%, p = 0.01) and concerned more PDA closures (71.2% vs. 14.9%, p < 0.001) predominantly of type C (39.0% vs. 2.7%, p < 0.001) as well as the occlusion of larger vessels (6 mm vs. 5 mm, p < 0.001).

In contrast, AVP IV was more widely used for the occlusion of single or multiple AP collaterals (25.7% vs. 3.4%, p < 0.001) as well as for the occlusion of the arterial supply of pulmonary sequestration (16% vs. 0%, p < 0.001) or scimitar vein occlusion (1 case). AVP IV was also preferred for CAF closure (8.1% vs. 0%, p = 0.03). Overall, because of the number of lesions treated simultaneously during the same procedure, the number of AVP IV implanted per procedure was higher, compared to AVP II (min-max range: 1–5 for AVP IV vs. 0–1 for AVP II, p < 0.001). Finally, AVP II was more predominantly delivered using a venous route compared to AVP IV in this study (p < 0.001).

The positioning and release of at least one device was possible in 134 procedures, accounting for an immediate procedure success rate of 98.5%. Full occlusion rate observed at day 1 and 6-month post procedure were 94.8% and 98.5%, respectively. No difference was observed according to AVP type. Procedure efficacy and safety of AVP II and IV are detailed in Table 2.

There was no procedure related death. Thirty-day in-hospital mortality occurred only in one patient who died due to hypoxemic pneumonia complicating a severe bronchiolitis, unrelated to procedure. Significant adverse events of severity level ≥3 (including 4-transient femoral artery thrombosis) occurred in a total of 8.1% of procedures. Patient and procedure characteristics according to occurrence of Major Adverse Events (MAE, grade ≥ 4) in children are summarized in Supplemental Table 2.

In total, 5.2% of patients underwent MAE, solely observed during catheter closure of large PDAs using AVP II. Observed MAE were predominantly device embolization or the need for urgent or delayed surgery. MAE occurred more frequently in younger patients (mean age 1.6 ± 3.5 years, vs. 2.8 ± 3.1 years in patients with no MAE, p = 0.03) with a lower mean weight (9.6 ± 14.8 kg vs. 14.2 ± 13.6 kg, p = 0.05). Furthermore, 5 (71.4%) patients with MAE weighed 5 kg or less. MAE prevalence was also higher with larger AVP plug size (mean diameter of 9.1 ± 2.3 mm vs. 7.7 ± 3.3 mm in patients with no MAE, p = 0.03) and a higher ratio of the AVP to vessel diameter (1.4 vs. 1.3, p = 0.03).

In all, there were a total of 4 device embolization, always observed within the first 24 h (Fig. 5), 2 of which were recovered by snaring technique in the catheter laboratory, and 2 by surgical removal.

Additional surgical procedures were performed in 2 patients presenting delayed severe LPA stenosis. The latter were documented 2 weeks after PDA closure in a 3.2 kg baby (10 mm AVP II) and 8 weeks after the procedure in another infant weighing 4.9 kg (10 mm PDA, 14 mm AVP II). This patient presented with a device “attraction” of the descending aorta and severe plication of the LPA. The last MAE involved a 3.6 kg infant already on inotropic support at the entrance to the catheterization room who presented a cardiac arrest during the first aortography. He was successfully resuscitated. There was no EKG modification, with unremarkable immediate coronary angiogram. PDA closure was successfully achieved using a 6 mm AVP II. Three other mild LPA stenosis did not require treatment, with uneventful follow-ups. There was no haemolysis, severe bleeding, cardiac arrhythmia, or late embolization.

Using univariate logistic regression analysis (Table 3), MAE was significantly associated with a weight ≤5 kg (Odds ratio, OR: 7.9, 95% Confidence Interval, CI: 1.5–42.8; p = 0.02), type C PDA (OR: 6.5, 95% CI: 1.36–31.04, p = 0.02), greater device oversizing illustrated by the ratio of AVP diameter to minimum vessel diameter ≥1.38 (OR: 5.56, 95% CI: 1.04–29.90, p = 0.046). We also observed a potential association between MAE and higher values of either AVP diameter to patient weight ratio (OR: 2.3, 95% CI: 1.3–4.0, p = 0.003; optimal cut-off: 1.45) or PDA diameter to patient weight ratio (OR: 2.4, 95% CI: (1.0–5.8, p = 0.048; optimal cut-off: 1.25).

Multivariate analysis (Table 3) retained a high ratio of AVP diameter to patient weight as an independent predictor (OR: 2.33%, 95% CI: 1.3–4.1, p = 0.004), with an optimal cut-off value set at 1.45. As such, a ratio ≥1.45 predicted a significant 9-fold increase of MAE risk (OR: 9.02, 95% CI: 1.67–49.00, p = 0.01). Furthermore, an association between MAE and PDA of tubular morphology was also supposed, with an OR of 5.35 (95% CI: 0.99–28.97, p = 0.052).

Our retrospective study design might have introduced potential interpretation bias. Firstly, we did not take into account center heterogeneity among the 6 participating institutions, in terms of patient volume, procedure type, and operator-biased device selection. Secondly, large tubular PDAs in infants of low weight were not considered for AVP catheter occlusion in half of the participating centers, which preferred surgical procedures for this high-risk patient group [14]. Finally, even in skilled hands, procedure safety, efficacy, and feasibility remain intricately linked to operator experience.