Abstract

PURPOSE

To evaluate the efficacy and safety of Amplatzer Vascular Plug 4 (AVP4) embolization in pulmonary arteriovenous malformations (PAVMs) with small- to medium-sized feeding arteries (<6 mm) and to identify factors affecting persistence and the main persistence patterns after embolization.

METHODS

Between June 2013 and February 2023, we retrospectively reviewed 100 patients with 217 treated PAVMs. We included PAVMs with feeding arteries <6 mm, treated with AVP4 embolization, and followed adequately with computed tomography (CT). Technical success was defined as flow cessation observed on angiography. Persistence was defined as less than a 70% reduction of the venous sac on CT. We evaluated adverse events for each embolization session. Patterns of persistence were assessed using follow-up angiography. Univariate and multivariate analyses were performed to evaluate factors affecting persistence based on the 70% CT criteria.

RESULTS

Fifty-one patients (48 women, 3 men; mean age: 50.8 years; age range: 16–71 years) with 103 PAVMs met the inclusion criteria. The technical success rate was 100%. The persistence rate was 9.7% (10/103), and the overall adverse event rate was 2.9% (3/103) during a mean follow-up of 556 days (range: 181–3,542 days). In two cases, the persistence pattern confirmed by follow-up angiography involved reperfusion via adjacent pulmonary artery collaterals. The location of embolization relative to the last normal branch of the pulmonary artery was the only factor substantially affecting persistence.

CONCLUSION

Embolization with AVP4 appears to be safe and effective for small- to medium-sized PAVMs. The location of the embolization relative to the last normal branch of the pulmonary artery was found to be the main determinant of persistence.

CLINICAL SIGNIFICANCE

Given the increasing demand for the treatment of small PAVMs, AVP4 embolization could be considered a viable and effective option for managing PAVMs with feeding arteries <6 mm.

Keywords: Arteriovenous malformation, computed tomography, embolization, pulmonary, vascular plug

Main points

• Amplatzer Vascular Plug 4 embolization was performed on 103 pulmonary arteriovenous malformations (PAVMs) with small- to medium-sized feeding arteries (<6 mm). This resulted in a persistence rate of 9.7% (10/103) based on the 70% reduction criteria on computed tomography and an overall adverse event rate of 2.9% (3/103).

• Follow-up angiography conducted on 28 PAVMs identified persistence in 2 PAVMs, both of which showed reperfusion via adjacent pulmonary artery collaterals.

• The only substantial factor affecting persistence was the location of the embolization relative to the last normal branch of the pulmonary artery.

Pulmonary arteriovenous malformation (PAVM) describes a direct connection between the pulmonary artery and vein, which can lead to paradoxical embolism and result in serious complications, such as stroke and brain abscess.¹ Endovascular embolization has emerged as the preferred treatment for PAVM.² The once conventional “3 mm rule,” which recommended treating feeding arteries larger than 3 mm, no longer holds universal acceptance.^{3, 4} The current consensus now supports embolization for feeding arteries that are 2–3 mm or larger or when catheterization is feasible.⁵ Nonetheless, the choice of embolic materials for small PAVMs remains limited, and the results from coil embolization in these cases are generally less favorable.⁶

The issue of persistence following PAVM embolization is substantial, often necessitating further interventions.⁷ To address this, research has been conducted on the effectiveness of various embolic materials, including coils,^{6, 8} Amplatzer Vascular Plugs (AVPs),^{9, 10} and microvascular plugs (MVPs).^{11, 12} Despite the longstanding use of coils, their associated persistence rates are notably high.^{6, 8, 13} While venous sac embolization yields favorable outcomes, employing multiple detachable coils is costly and extends procedural times.¹⁴ More recently introduced MVPs have demonstrated promising results, although they are more expensive, and their long-term efficacy remains uncertain.¹⁵

AVPs are composed of a braided nitinol mesh and are noted for their low risk of migration in high-flow vessels or short landing zones, which permits device repositioning and provides the potential for single-device occlusion.^{16, 17} The latest generation, Amplatzer Vascular Plug 4 (AVP4), features a small-profile catheter with a 0.038-inch luminal diameter, suitable for navigating small- to medium-sized vessels and handling vascular tortuosity. Since its introduction for PAVM embolization in 2014, several studies have reported on the use of AVP4, with persistence rates ranging from 0% to 16%.^{10, 15, 18} However, many of these studies have been limited by small sample sizes or the inclusion of different generations of AVP.

Consequently, this retrospective single-center study aims to evaluate the efficacy and safety of AVP4 embolization in PAVMs with small- to medium-sized feeding arteries (<6 mm). Additionally, this study seeks to identify factors affecting persistence and to delineate the main persistence patterns following AVP4 embolization.

Methods

This retrospective study received approval from the Institutional Review Board of Kyungpook National University Hospital (KNUH 2023-12-027). All participants provided informed consent prior to the procedure.

Patient selection

The study cohort included patients who underwent endovascular embolization for PAVM from June 2013 to February 2023. The eligibility criteria for inclusion were as follows: (1) treatment-naïve PAVM with a feeding artery diameter of <6 mm; (2) embolization performed using the AVP4; and (3) availability of both initial and follow-up computed tomography (CT) scans before and after embolization. The exclusion criteria were as follows: (1) underwent additional embolization sessions for the same lesion without an intervening follow-up CT; or (2) had a follow-up period of <6 months. Data on clinical history, physical examination, and PAVM characteristics were extracted from electronic medical records. Adverse events associated with the procedure during hospitalization and outpatient follow-up were also meticulously analyzed.

Embolization procedure

Vascular access was obtained via the right femoral vein, followed by intravenous administration of a heparin sodium bolus (3,000–5,000 IU; JW Pharmaceutical, Seoul, Korea). Subsequent pulmonary angiography facilitated the selective catheterization of the juxta-sac feeding artery using a coaxial system composed of a 6-Fr guiding catheter (Flexor Shuttle Guiding Sheath; Cook Medical, Bloomington, IN, USA) and a 5-Fr diagnostic catheter (Torcon NB Advantage, MPA; Cook Medical, or Glidecath, Angled Taper; Terumo, Tokyo, Japan). In cases involving challenging navigation due to small tortuous feeders, a triaxial system was employed, which included a 1.98-Fr microcatheter (Masters Parkway Soft; Asahi Intecc, Tokyo, Japan). The procedure began with the microcatheter, followed by the advancement of a 5-Fr hydrophilic-coated catheter over it. The size of the AVP4 (Abbott, Plymouth, MN, USA) ranged from 30% to 300% oversizing, based on preprocedural CT and selective angiography findings. The AVP4 was advanced into position within the 5-Fr catheter by pushing the guidewire to the catheter tip, and then the catheter was withdrawn to deploy the device. Proper placement of the plug was verified by injecting a contrast medium through the guiding catheter; if necessary, the plug was recaptured, repositioned, and redeployed. Depending on the operator’s preference, additional coil embolization was performed occasionally to expedite flow cessation and provide reinforcement. Complete cessation of PAVM flow was confirmed in all patients through the completion of the digital subtraction angiography (DSA).

Acquisition and protocol for computed tomography and follow-up digital subtraction angiography

Initial and follow-up CT scans were primarily conducted using contrast-enhanced CT with multidetector-row scanners (Revolution EVO, Optima CT660, LightSpeed16; GE Healthcare, Chicago, IL, USA; SOMATOM Force, SOMATOM Definition Edge; Siemens Healthineers, Erlangen, Germany). For these examinations, a contrast agent (80–100 mL) was intravenously injected at a rate of 1.5–2 mL/s. CT images targeting the area of interest were reconstructed with a slice thickness of 2.5 mm in both transverse and coronal orientations. Follow-up CT scans were scheduled at 6 and 12 months post-embolization and subsequently every 2–3 years to monitor the persistence or resolution of PAVMs.¹⁹

DSA was conducted on previously treated PAVMs, particularly in cases in which multiple PAVMs were treated across separate sessions. The procedure typically began with either right or left pulmonary angiography, utilizing an injector with injection rates of 10–15 mL/s and volumes of 20–30 mL per injection. For more detailed assessments, selective angiography was performed at the segmental pulmonary artery levels, using injection rates of 3–5 mL/s and volumes of 9–15 mL. In certain instances, more precise visualization was achieved through meticulous manual injections at the distal levels of the pulmonary arteries.

Imaging analysis

All imaging obtained before, during, and following AVP4 embolization was reviewed by two experienced cardiovascular radiologists who were blinded to the outcomes of PAVM embolization. Discrepancies between radiologists were resolved by consensus.

The analysis included reviewing the location, multiplicity, complexity (categorized as simple vs. complex), and original vessel diameters, along with their changes (feeding artery, venous sac, and draining vein) between the initial and final post-procedural CT scans. Changes in vessel diameter were quantified as reduction rates and recorded separately. Additionally, the origin of the last normal branch of the pulmonary artery was documented on the initial CT and during procedural DSA,²⁰ and the embolization location relative to this branch (either proximal or distal) was confirmed on post-procedural CT. The distance from the plug to the sac was also evaluated using pre- and post-procedural CT scans.

During the procedural imaging of AVP4 embolization, the size and number of plugs, the plug oversizing ratio, the type and number of additional coils, and the procedure time were all documented. Technical success was defined as the complete cessation of flow in the PAVM upon completion of the DSA. Treatment outcomes were assessed using the widely accepted CT criteria, where occlusion was defined as a 70% reduction in the venous sac in pre- and post-procedure comparisons (referred to as the 70% CT criteria).²¹ Persistence was noted when the reduction rate of the venous sac was less than 70%. Procedure time was recorded from the femoral vein puncture to the completion of angiography, exclusively for sessions treating a single PAVM to ensure accurate assessment.

Adverse events were classified according to the Society of Interventional Radiology standards.²² Both peri-procedural and post-procedural adverse events were documented for each embolization session.

To investigate factors affecting persistence, variables such as sex, age, smoking history, use of antithrombotic agents, lobar location, complexity, multiplicity, feeding artery diameter, venous sac diameter, plug oversizing ratio, sac-to-plug distance, embolization location relative to the last normal branch, and additional coil embolization were evaluated.

Angiographically confirmed cases by follow-up DSA were analyzed to determine patterns of persistence. Persistence was classified as resulting from (a) recanalization of a previously treated feeding artery, (b) reperfusion via adjacent pulmonary artery collaterals, or (c) the presence of a previously unrecognized feeder (incomplete treatment).⁷ Reperfusion from systemic arteries was not assessed.

Statistical analysis

Continuous variables were expressed as the mean and range, whereas categorical variables were reported as the frequency (percentage). Multivariate logistic regression analysis was performed to identify factors affecting persistence using odds ratios (OR) and confidence intervals (CI). This analysis utilized the R software package (version 4.0.3, The R Foundation for Statistical Computing, Vienna, Austria). Variables that achieved a P value of <0.20 in the univariate analysis were selected as input variables for the multivariate analysis, which was conducted using a backward stepwise method. A P value of <0.05 was considered statistically significant.

Results

Fifty-one patients [48 women and 3 men; mean age: 50.8 years (range: 16–71)] with 103 PAVMs met the inclusion criteria and were included in the analysis (Figure 1). Among these patients, 9 (17.6%) exhibited symptoms of hereditary hemorrhagic telangiectasia, and 22 (43.1%) presented with symptoms attributable to PAVM. Sixteen patients (31.3%) had multiple PAVMs, averaging 2.26 lesions per patient (range: 1–10). Of the 103 PAVMs analyzed, 97 (94.2%) were classified as simple, with the remaining identified as complex. The mean diameter of the feeding arteries was 3.00 mm (range: 1.50–5.70 mm). The mean follow-up period was 556 days (range: 181–3,542 days). The characteristics of the patients and the PAVMs are summarized in Table 1.

All 103 PAVMs were successfully treated with AVP4 embolization across 59 sessions (Figure 2). On average, 1.75 PAVMs were treated per session (range: 1–8). The mean size and number of AVP4 devices used per PAVM were 6.34 mm and 1.09, respectively. Additional coils were used in 9 PAVMs (8.7%), with an average of 2.89 coils per PAVM (range: 1–5). The mean procedure time for sessions treating a single PAVM was approximately 39.62 minutes (range: 18–96 minutes). Details of the AVP4 embolization procedures are summarized in Table 2.

The technical success rate for AVP4 embolization was 100%. The persistence rate of the treated PAVMs, using the 70% CT criteria, was 9.7% (10/103). Stratified by embolization type, the persistence rates were 9.6% (9/94) for AVP4 alone and 11.1% (1/9) for AVP4 combined with coil embolization. During the 59 sessions for 103 PAVMs, three mild adverse events were reported (5.1% per session): two instances of self-limiting pleuritic chest pain and one case of transient bradycardia. There were no severe adverse events, with an overall adverse event rate of 2.9% per PAVM lesion.

Follow-up DSA was conducted for 28 (27.2%) of the 103 PAVMs at a mean interval of 436 days. Among these, occlusion was observed in 26 PAVMs, whereas the remaining 2 (7.1%) exhibited persistence due to reperfusion via adjacent pulmonary artery collaterals (Figure 3). When comparing outcomes between DSA and the 70% CT criteria, 25 out of 26 angiographically occluded PAVMs showed venous sac reductions exceeding 70% on CT, resulting in concordant findings. However, one PAVM demonstrated a reduction rate of 57.3%, leading to discordance between the two modalities. The two angiographically reperfused PAVMs showed venous sac reductions of 34.7% and 49.2%, respectively, aligning the findings across both modalities.

In both univariate and multivariate analyses, the location of embolization relative to the last normal branch of the pulmonary artery was identified as the only significant factor affecting persistence (OR: 0.18; 95% CI: 0.03–0.81; P < 0.05) (Table 3).

Discussion

The findings of this study affirm the efficacy and safety of AVP4 embolization for small- to medium-sized PAVMs with diameters of <6 mm, showing a persistence rate of 9.7% (10/103) based on the 70% CT criteria and an overall adverse event rate of 2.9% during an average follow-up period of 556 days. Follow-up DSA, conducted in 27.1% of this cohort, revealed persistence in 2 PAVMs, predominantly due to reperfusion via adjacent pulmonary artery collaterals. The location of embolization relative to the last normal branch of the pulmonary artery was identified as the only substantial factor affecting persistence according to the CT criteria.

Various generations of AVPs have been employed for PAVM embolization, with reported persistence rates ranging from 0% to 16%.^{9, 10, 15, 18, 23-25} Some studies have suggested superior outcomes with AVP compared with coils.^{26, 27} Nonetheless, there remains a scarcity of studies specifically focusing on AVP4. Rabellino et al.¹⁸ defined a successful outcome as a venous sac reduction of ≥30% in their early experience with 7 patients, achieving success across all cases over an average follow-up of 20.1 months. A more recent study in 2019¹⁰ involving 19 PAVMs reported a persistence rate of 16% using 70% CT criteria over an average follow-up of 14 months. Ratnani et al.¹⁵ specifically analyzed AVP4 and reported a persistence rate of 12.5% (1/8) over an average follow-up of 1,239 days, defining persistence based on sustained sac perfusion observed in CT angiography (CTA) or pulmonary angiography. While the outcomes of these small case series generally align with those of the current study, varying assessment criteria make precise comparisons challenging.

Pulmonary angiography is considered the gold standard, but it poses difficulties for routine use due to its invasiveness.¹⁹ The use of sac perfusion on CTA to assess persistence raises concerns about retrograde venous filling from adjacent normal branches.¹³ Presently, the 70% CT criteria are the most widely adopted, yet recent discussions highlight concerns regarding their specificity.^{13, 28, 29} Additional research and consensus are necessary to refine and agree on criteria that address these concerns effectively.

The recently introduced MVP has demonstrated superior results compared with the AVP, boasting a low persistence rate of 0%–6%.^{11, 12, 15, 30}AVP, composed of a finely braided nitinol mesh, exhibits several structural challenges in comparison to MVP, which features a polytetrafluoroethylene (PTFE)-covered nitinol cage. Particularly, introducing a 5-Fr catheter up to the juxta-sac feeding artery in cases of very small or tortuous feeding arteries can be technically challenging compared with using a 2.4- or 2.8-Fr microcatheter, as utilized with MVP. In our practice, we often overcome this challenge by using a hydrophilic-coated 5-Fr catheter (Glidecath, Angled Taper; Terumo) with appropriate angulation and advancing it over the microcatheter.

A concern exists that AVP4 may become lodged within this soft and flexible 5-Fr catheter during delivery. To address this issue, we primarily employ smaller-sized AVP4s (4–6 mm) -adequate for most small-sized feeding arteries- and advance a 6-Fr guiding catheter as distally as possible to provide support while routinely performing continuous saline flushing in the catheter to minimize friction between the plug and catheter wall. Furthermore, unlike MVP, which induces immediate flow cessation due to its PTFE cover, AVP4 relies on inducing thrombosis through its nitinol mesh, requiring patience and repeated monitoring for occlusion. The patient’s coagulation status may influence this process and raise concerns about the potential migration of in-situ thrombus on the device surface, leading to paradoxical embolism.^{16, 17} To mitigate these risks, we employ a strategy of reinforcement with several additional coils if flow cessation is not achieved within 5–10 minutes or by confirming flow cessation collectively after completing treatment for all PAVMs in cases of multiple PAVMs to save time. Consequently, we achieved a relatively short procedure time (mean: 39.62 minutes), and no procedure-related paradoxical embolisms were reported.

On the financial side, AVP4 offers a more cost-effective alternative than MVP. The mean number of AVP4 devices used in this study, 1.1 per PAVM, is comparable to the 1.1–1.3 used in previous MVP studies^{11, 12}despite the substantially higher cost of the MVP device.¹⁵ Additionally, the routine use of a microcatheter for MVP delivery adds to overall expenses. While MVP has not yet received approval for use in many countries, including ours, AVP4 remains a favorable option in centers where it is available, offering both clinical efficacy and cost-effectiveness.

In this cohort, the majority of PAVMs featured small-sized feeding arteries, with 64.1% measuring less than 3 mm and 11.7% measuring less than 2 mm. Stein et al.⁶ reported on coil embolization for 141 PAVMs with feeding arteries smaller than 3 mm; the persistence rate noted was 21%, which is higher than the 10% reported in other studies targeting PAVMs with feeding arteries of 3 mm or larger. However, in our study, there was no substantial difference in persistence rates between PAVMs with feeding arteries of 3 mm or smaller (9.4%) and those larger than 3 mm (10.3%). This outcome may highlight the advantage of AVP4 over coils, as AVP4 allows for sufficient oversizing and smooth delivery if the catheter reaches the target vessel, regardless of vessel size. In the case of the MVP, there are reports of successful treatments for feeding arteries as small as 1.3 mm;¹² however, there is a lack of studies focusing on small PAVMs or evaluating long-term outcomes. Under these circumstances, AVP4 emerges as a favorable treatment option for small PAVMs.

After coil embolization, recanalization through a previously treated feeder is the predominant persistence pattern, reported to exceed 90%.⁷ Factors such as coil packing density, the use of oversized coils, and the distance between the coil and the venous sac have been identified as substantial factors affecting persistence rates.^{6, 31, 32} In a recent study by Shimohira et al.¹³, the location of embolization relative to the last normal branch of the pulmonary artery was determined to be a substantial factor in persistence, as assessed by CT, time-resolved MR angiography, and DSA. However, similar detailed studies focusing on AVP are lacking.

In this study, reperfusion via adjacent pulmonary artery collaterals was observed in both cases where angiographically confirmed persistence occurred, specifically when proximal embolization was performed because the last normal branch was within the sac. This location was the only substantial factor affecting persistence. In this reperfusion mechanism, the shunt or feeder size is usually very small, rendering additional treatment technically challenging and generally less successful than the recanalization pattern.^{7, 8} Although this study highlighted the excellent cross-sectional occlusion capabilities of AVP4, achieving complete prevention of persistence in PAVMs where the last normal branch is located within the sac may ultimately require sac embolization.^{13, 20, 33} There are documented cases in which successful outcomes were achieved through venous sac coiling combined with feeding artery plug embolization in such scenarios.³⁴ Nonetheless, further studies involving a larger cohort are necessary to validate these findings and refine treatment protocols.

Some limitations of this study should be acknowledged. First, it was a retrospective study with a relatively small sample size. Second, owing to the widespread availability of chest CT scans and health screenings, most patients in the study presented with incidentally detected simple PAVMs. Given that treatment outcomes are less favorable for complex PAVMs, the persistence rate of these malformations may have been underestimated. Moreover, follow-up DSA was performed only in patients with multiple PAVMs, which introduced potential bias. Additionally, reperfusion via the systemic artery was not evaluated. Lastly, an important variable related to the use of AVP -occlusion time- was not measured.

In conclusion, AVP4 embolization proved to be safe and effective for treating small- to medium-sized PAVMs (<6 mm), demonstrating a low persistence rate based on the 70% CT criteria. The primary pattern observed in angiographically confirmed persistence was reperfusion via adjacent pulmonary artery collaterals. Concerning treatment outcomes based on CT criteria, the only factor affecting persistence was the location of the embolization relative to the last normal branch of the pulmonary artery.

Conflict of interest disclosure

The authors declared no conflicts of interest.