Embolisation of simple PAVMs with high-volume detachable non-fibered coils: a 3-year follow-up study Authors: Daniel A F van den Heuvel, Douwe Sachumsky, Bette Zijp, Johannes J Mager, Marc van Leersum, Marco C Post, Sjors Klompmaker Published: 2026 ABSTRACT AbstractPurposeTo determine pulmonary arteriovenous malformation (PAVM) persistence and safety of high-volume detachable non-fibred (HVDNF) coil embolisation of de novo simple PAVMs using computed tomography pulmonary angiography (CTPA) at 6-month and 3-year follow-up.MethodsThis retrospective single-cohort study included patients who underwent embolisation of de novo simple PAVMs, exclusively with HVDNF coils, between January 2018 and January 2021. Primary endpoints were PAVM persistence at 6 months, cumulative persistence at 3 years and coil recanalisation at 6 months and 3 years. Secondary endpoints were technical success, safety, repeat embolisation and vein diameter reduction. Persistence and recanalisation were confirmed with digital subtraction angiography (DSA).ResultsThis study included 50 patients with 75 de novo PAVMs. PAVM persistence at 6 months was 13.2%; cumulative persistence at 3 years was 17.4%. Coil recanalisation was the persistence mechanism in 10.3% of the cases at 6 months; at 3 years, this was 13.0%. Technical success rate was 98.7%. There were no procedure- or PAVM-related serious adverse events. Repeat embolisation with HVDNF coils was performed in 4 PAVMs after 6 months and 5 PAVMs after 3 years. Mean vein diameter reduction was 38.3% (SD 29.4%) at 6 months and 44.6% (SD 36.5%) at 3 years.ConclusionsHVDNF coils are safe to use for de novo PAVM embolisation, show an excellent technical success rate and are associated with acceptable 6-month and 3-year persistence and coil recanalisation rates.Supplementary InformationThe online version contains supplementary material available at 10.1186/s42155-026-00758-9.Keywords: Lung, Arteriovenous malformations, Embolisation, High-volume coils Background Pulmonary arteriovenous malformations (PAVMs) are vascular anomalies that affect approximately 1 in 2630 individuals in the general population [1]. The most common cause of PAVMs is hereditary haemorrhagic telangiectasia (HHT) [2]. If left untreated, PAVMs can cause hypoxemia or paradoxical emboli, which can result in ischaemic strokes or abscesses [3]. To reduce the risk of these complications, the international HHT guidelines recommend selective percutaneous trans arterial embolisation of PAVMs with feeding arteries greater than 2 mm [4, 5]. However, PAVM persistence remains an issue, with pooled rates including coils and vascular plugs of around 13% [6, 7]. Although vascular plugs seem to perform superior to coils, pulmonary arterial anatomy, peripheral PAVMs and tortuous or small feeding arteries may preclude their successful deployment [8]. Coils will likely continue to play an important role in PAVM embolotherapy, at least until sufficient evidence supports the wider use of alternative embolisation techniques, including liquid embolic agents [9]. High-volume detachable non-fibred (HVDNF) coils offer promising short-term results over fibred coils, utilising mechanical occlusion to minimise the impact of clot resolution over time, but longer-term data are lacking [10]. This study assessed PAVM persistence at 6 months and 3 years following HVDNF coil embolisation of de novo, simple type PAVMs. Methods We performed a retrospective single cohort study including patients who underwent PAVM embolisation, exclusively treated with HVDNF coils (Penumbra, Alameda, CA, USA), at our institution between January 2018 and January 2021. The study was conducted in accordance with the Declaration of Helsinki and reporting based on criteria from the strengthening of the reporting of observational study in epidemiology (STROBE) statement. Based on our predefined study protocol, the medical ethical committee confirmed that the Dutch Medical Research Involving Human Subjects Act did not apply. The Institutional Review Board waived the need for informed consent. Study population Patients over 18 years old with at least one previously untreated (i.e. de novo) PAVM were included. To avoid confounding by factors other than coil type, PAVMs of the complex type were excluded. Additional exclusion criteria were pregnancy, diffuse type PAVMs (i.e. involving every subsegmental branch of at least one segment) and the use of non-HVDNF coils. PAVMs were also excluded if a completion digital subtraction angiography (DSA) or a repeat DSA for the indication of suspected PAVM persistence was missing. Screening, embolisation and follow-up protocol All patients with suspected HHT or idiopathic PAVMs received preprocedural work-up according to the international HHT guidelines [4, 5]. All patients were discussed in a multi-disciplinary team meeting with a pulmonologist, cardiologist and interventional radiologist specialised in PAVMs. Embolisation was indicated when at least one PAVM with a feeding artery ≥ 2 mm was present. Embolisation procedures were performed by one of four attending interventional radiologists, having at least 10 years of experience in treating PAVMs. All procedures were performed under local anaesthesia. After introduction of a 6-French sheath in the common femoral vein, a 6-French 145° angled pigtail catheter and hydrophobic guidewire were used to navigate through the heart and select the right or left pulmonary artery. Next, a coaxial system—consisting of a 6-French 90-cm multipurpose guiding catheter and a 4-French 125-cm multipurpose diagnostic catheter—was delivered over an 0.035″ exchange guidewire as close to the target artery as possible. After selective catheterisation of the segmental artery, super selective catheterisation of the feeding artery was performed by upgrading to a tri-axial system, using a microcatheter with a minimal inner lumen diameter of 0.025″ (Lantern Penumbra or ProGreat Terumo) and a 0.014″ hydrophilic guidewire. The aim was to embolise the feeding artery within 10 mm distance from the sac [11]. In case of a small PAVM, a sac with feeding artery embolisation approach was performed. Standard or soft coils are deployed first, and the artery is then packed with either soft or packing coils. Occasionally, a scaffold of Ruby standard coils within the sac was formed prior to embolisation of the feeding artery. This approach—in which there is no complete filling of the sac—was taken when it was anticipated that embolising the sac would require too many coils (Fig. 1). All HVDNF coil subtypes (Ruby Standard, Ruby Soft, POD system, packing coil) were available at our institution and used at the operator’s discretion. Sizing of coils was based on feeding artery or sac diameter with a 1–2 mm oversizing and aiming for maximum packing density. Complete occlusion was confirmed with a completion DSA—including a run from a proximal segmental artery and lobar artery—to rule out additional feeding arteries. All patients were scheduled for follow-up visits at 6 months and 3 years, which included a computed tomography pulmonary angiography (CTPA). In case of suspected PAVM persistence, a re-intervention was performed to define the cause of persistence and, if indicated, to perform additional embolisation. Example of a technically successful HVDNF coil embolisation procedure. Unenhanced CT showing a simple PAVM in the inferior lingula (A–C). Sac embolisation was not deemed feasible because the number of coils required to fill the sac would be too high. Instead, a 4 × 15 mm Ruby soft coil was placed as a backstop within 10 mm of the sac and subsequent embolisation was performed with 60-cm packing coils (3 ×) and 45-cm packing coil (1 ×) to ensure high packing density (D–F). PAVM, pulmonary arteriovenous malformation; HVDNF, high-volume detachable non-fibred Data collection and management Data were captured from the patients’ electronic medical records and the local picture archiving and communications system and stored in the hospital’s Research Electronic Data Capture tool (REDCap, Fort Lauderdale, FL, USA) [12, 13]. Demographic and disease-specific data included age, sex, body mass index (BMI), diagnosis of HHT, genetic testing or Curaçao criteria, past medical history, smoking status, and the use of anti-coagulant or anti-platelet medication. Pre- and post-embolisation CTPA scans were used to determine PAVM location, number of feeding arteries, diameter of the feeding arteries and draining veins within 10 mm of the sac, and short axis diameter of the aneurysm. The specific types and sizes of the used coils were recorded per feeding artery. All periprocedural symptoms and post-procedural complications were documented up to the 3-year follow-up visit. Primary and secondary outcomes The primary endpoints for this study were PAVM persistence rate at 6 months and cumulative persistence rate at 3 years. The cumulative rate is defined as persistence at 6 months plus the number of PAVMs demonstrating persistence between 6 months and 3 years relative to the total number of PAVMs available for analysis. Persistence was diagnosed if there was contrast enhancement of the sac or proximal vein (< 10 mm of the sac) on CTPA regardless of artery, sac or vein shrinkage [10]. Based on CTPA and DSA, persistent perfusion mechanisms were categorised into coil recanalisation, recruitment (pulmonary-to-pulmonary perfusion), systemic-to-pulmonary reperfusion, and missed feeding artery perfusion [14–16]. Coil recanalisation is the main mechanism of PAVM persistence and therefore also included as separate primary endpoint. It is diagnosed when there is blood flow through the device on CTPA or DSA imaging. Secondary outcomes were immediate technical success, safety, repeat embolisation rate and vein diameter reduction. Technical success was achieved when embolisation was performed ≤ 10 mm of the sac and there were no missed feeding arteries identified by the study readers. Safety was defined as freedom from any procedure- or PAVM related serious adverse event (SAE) within 3 years. Complications were graded using the recently revised CIRSE Classification System for Complications Reporting, which comprises six grades: grade 1 denotes a complication resolved during the index procedure, grade 6 corresponds to death, and grades 2–5 indicate progressively increasing severity and required interventions [17]. Statistical analysis All statistics were performed by a registered epidemiologist with advanced statistical training, using STATA version 18/BE (StataCorp LP, College Station, TX). We reported categorical data as counts with proportions and continuous data as means with standard deviations or medians with interquartile ranges (IQR), as appropriate. We applied univariable logistic regression to identify potential predictors for PAVM persistence, using alpha 0.05 as a cut-off for significance. Results During the study period, 111 patients received embolotherapy for 271 PAVMs in 151 procedures. Seventy patients had at least one de novo simple type PAVM. Of these, 20 patients were excluded because of their young age (9), other device types (8), diffuse type PAVMs (2) and pregnancy (1). The final cohort included 50 patients with 75 PAVMs that met the inclusion criteria and 18 PAVMs that were excluded based on non-HVDNF coil usage (Fig. 2). Study flow chart. PAVM, pulmonary arteriovenous malformation; HVDNF, high-volume detachable non-fibred; CTPA, computed tomography pulmonary angiography; FU, follow-up. * Both deceased subjects had 1 PAVM Median follow-up time was 42 months. Two patients (2 PAVMs) expired prior to their first follow-up visit due to atherosclerotic myocardial infarction (1) and advanced rectal cancer (1). Three patients (3 PAVMs) accidentally had an unenhanced CT, and two patients (2 PAVMs) did not have any CT scan at 6 months but all five had a CTPA at 3 years. One patient (4 PAVMs) was lost to follow-up after the 6-month CTPA. This resulted in 68 PAVMs with available 6-month and 69 PAVMs with available 3-year follow-up (Fig. 2). Baseline characteristics are outlined in Table 1. Of 75 PAVMs, 65 (86.7%) had one feeding artery and 10 (13.3%) had two feeding arteries. The mean feeding artery diameter was 3.1 mm (SD 1.3 mm). The majority of PAVMs was treated with one POD, Ruby Standard, or soft coil in combination with at least one packing coil (40%). A median of 2.0 (IQR 1.5–3.0) coils were used per PAVM correlating with a median packing volume of 102 mm3 (IQR 61–199 mm3) (Table 2). Table 1Patient baseline characteristicsMean (SD)/count (%)(N = 50)Age (years)52.5 (14.9)BMI25.1 (4.4)Gender Female32 (64.0%) Male18 (36.0%)HHT diagnosis Genetic mutation39 (78.0%) Curacao criteria9 (18.0%) No HHT2 (4.0%)Genetic mutation ENG34 (68.0%) ACVRL14 (8.0%) SMAD 41 (2.0%) Unknown/not performed*11 (22.0%)Smoking Never28 (56.0%) Past9 (18.0%) Active7 (14.0%) Unknown/not reported6 (12.0%)History of stroke (CVA/TIA) No40 (80.0%) Yes10 (20.0%)History of brain abscess No47 (94.0%) Yes3 (6.0%)History of diabetes No48 (96.0%) Yes2 (4.0%)Anti-platelet or anti-coagulation therapy No40 (80.0%) Yes10 (20.0%)Haemoglobin mmol/L8.6 (1.4)Haematocrit L/L41.8 (5.1)TTCE RLS grade Grade 210 (20.4%) Grade 329 (59.2%) Unknown/indeterminate10 (20.4%)*Including 2 patients without HHTAbbreviations: BMI body mass index, HHT haemorrhagic hereditary telangiectasia, CVA cerebrovascular accident, TIA transient ischaemic attack, TTCE transthoracic contrast echocardiography, RLS right-to-left shunt Patient baseline characteristics *Including 2 patients without HHT Abbreviations: BMI body mass index, HHT haemorrhagic hereditary telangiectasia, CVA cerebrovascular accident, TIA transient ischaemic attack, TTCE transthoracic contrast echocardiography, RLS right-to-left shunt Table 2PAVM characteristics and treatment detailsMean (SD)/median (IQR)/count (%)(N = 75)Number of feeding arteries 165 (86.7%) 210 (13.3%)Diameter artery (mm)3.1 (1.3)Diameter sac (mm)7.7 (5.0)Diameter vein (mm)3.5 (2.4)Coil type Ruby Std12 (16.0%) Ruby Soft6 (8.0%) Ruby Std + Soft8 (10.7%) POD4 (5.3%) Packing14 (18.7%) Ruby Std/Soft + POD1 (1.3%) Ruby Std/Soft/POD + packing26 (34.7%) Ruby Std/Soft + POD + packing4 (5.3%)Number of coils per PAVM2.0 (1.5–3.0)Volume of coils per PAVM (mm3)102 (61–199)Distance from device to PAVM sac (mm)0.6 (1.6)Abbreviations: POD combined anchor and packing coil, Std standard PAVM characteristics and treatment details Abbreviations: POD combined anchor and packing coil, Std standard Primary and secondary outcomes At 6 months, 59 of 68 PAVMs remained occluded and 9 showed perfusion. At 3 years, 3 additional PAVMs with perfusion were recorded. This resulted in a persistence rate of 13.2% at 6 months and a cumulative persistence rate 17.4% at 3 years. Technical success was achieved in 74 of 75 (98.7%) PAVMs. An example of a technically successful procedure is presented in Fig. 2. Coil recanalisation rates were 10.3% at 6 months and 13.0% at 3 years. According to the CIRSE complication classification, there were 4 patients (8%) with a grade 1a complication: facial tingles or visual sparks (2), angina pectoris (1), and a vasovagal state (1). Of the 9 PAVMs with persistent perfusion at 6 months, 4 received additional embolisation within 3 years, in 3 PAVMs additional embolisation was not indicated (e.g. based on vessel size or shunt grade) and in 2 PAVMs additional embolisation was performed after 3 years. The mean vein diameter reduction was 38.3% (SD 29.4%) at 6 months (persistent 33% vs occlusion 39%) and 44.6% (SD 36.5%) at 3 years (persistent 23% vs occlusion 48%). PAVM outcomes are detailed in Table 3. All PAVMs with persistent perfusion are detailed in Table 4. Table 3Short- and long-term follow-upMean (SD)/count (%)(N = 75)PAVM persistence (CTPA) at 6 months* No59/68 (86.8%) Coil recanalisation7 (10.3%) Recruitment1 (1.5%) Missed feeding artery1 (1.5%)Vein diameter reduction at 6 months (%)§38.3 (29.4)Re-embolisation after 6 months No64/68 (94.1%) Yes4/68 (5.9%)Cumulative PAVM persistence (CTPA) at 3 years* No57/69 (82.6%) Coil recanalisation9 (13.0%) Indeterminate cause of contrast2 (2.9%) Missed feeding artery1 (1.4%)Vein diameter reduction at 3 years (%)§44.6 (36.5)Re-embolisation after 3 years No64/69 (92.8%) Yes5 (7.2%)Time to additional embolisation (days)821.4 (525.4)Abbreviations: CTPA contrast-enhanced CT pulmonary angiography* CTPA was available in 68/75 PAVMs at 6 months and 69/75 PAVMs at 3 years§ Vein diameter reduction was established on non-contrast enhanced CT in 3/75 PAVMs at 6 months. Both variables were normally distributed Short- and long-term follow-up Abbreviations: CTPA contrast-enhanced CT pulmonary angiography * CTPA was available in 68/75 PAVMs at 6 months and 69/75 PAVMs at 3 years § Vein diameter reduction was established on non-contrast enhanced CT in 3/75 PAVMs at 6 months. Both variables were normally distributed Table 4Persistent PAVMs and re-interventionStudy IDDevices usedPersistent at6 monthsRe-interventionPersistent at30–42 monthsRe-intervention5Ruby stdRecanalisationNoRecanalisationNo8Ruby std + softRecanalisationYesNoNo12Ruby stdRecanalisationNoRecanalisationNo25Ruby std + softRecanalisationYesNoNo29Ruby softRecruitmentYesIndeterminate cause of contrastNo31Ruby std + packingRecanalisationNoRecanalisationYes32Ruby softRecanalisationYesRecanalisationNo41POD + packingNoNoRecanalisationYes42PODRecanalisationNoRecanalisationNo43PackingMissed feeding arteryNoMissed feeding arteryYes46POD + packingUnknown*NoIndeterminate cause of contrastYes48Ruby std + softUnknown*NoRecanalisationYesAbbreviations: std standard*No contrast-enhanced CT available Persistent PAVMs and re-intervention Abbreviations: std standard *No contrast-enhanced CT available Logistic regression analysis Univariable logistic regression revealed no significant association between persistent PAVMs and age, BMI, sex, genetic mutation, history of stroke, active smoking, anti-platelet or coagulation therapy, haemoglobin level, or haematocrit level. Logistic regression analyses for both 6-month and 3-year outcomes are detailed in Supplemental Table S1 and S2. Discussion This retrospective single cohort study shows that HVDNF coils are safe and perform well when used for embolisation of simple de novo PAVMs. The PAVM persistence rate of 13.2% at 6 months is consistent with what is reported in literature. Between 6 months and 3 years, the cumulative persistence increased to 17.4%, mainly due to coil recanalisation. Two recent systematic reviews with accompanying meta-analyses investigated the outcomes of PAVM embolisation [6, 18]. Both reviews reported on pooled persistence rates and performed sub-analyses comparing vascular plugs to coils. Sahin et al. (2024) analysed 1865 patients and reported a pooled persistence rate of 14% and no statistically significant difference in persistence between vascular plugs and coils (odds ratio 0.64; 95% CI: 0.09–1.38) [15]. More recently, Yu et al. (2025) included 2946 patients, found a comparable pooled persistence rate of 13.2% but did report a lower persistence for vascular plugs compared to coils (13.6% vs 32.7%; odds ratio, 2.07; 95% CI: 1.04–4.13; P = 0.039) [6]. The results of these meta-analyses are consistent for vascular plugs demonstrating a persistence rate of approximately 14%. However, there seems to be a substantial difference concerning the performance of coils which might be explained by the heterogeneity between included studies as well as the meta-analyses themselves. This prompted us to compare the results of the current study with a recently published historical cohort treated by van den Heuvel et al. and using identical follow-up and endpoint definitions. In a sub-cohort of that study, consisting of 61 simple PAVMs treated with Interlock-18 fibred coils (Boston Scientific, MA, USA), a 6-month persistence rate of 21% was found [11]. Whilst not a formal comparative study, these findings seem to support the findings by Mathevosian et al. who found a superiority of HVDNF coils over fibred coils. On the other hand, the reported PAVM persistence of only 3.8% (3 of 80) in PAVMs treated with HVDNF coils in their study is substantially lower compared to the current one. A potential explanation may be that Mathevosian et al. defined persistent PAVM occlusion as > 70% shrinkage of the sac or vein, which is less strict than the CTPA perfusion definition used in the present study. Also, HHT prevalence was 77% in their cohort compared to 96% in the present study. This difference in underlying disease aetiology and PAVM complexity could have contributed to the difference in persistence rates as well. The current study has several limitations. First, due to its retrospective nature, selection bias cannot be ruled out. Second, PAVM persistence is diagnosed on CTPA, with subsequent DSA confirmation. There is limited data concerning inter or intra-observer variability of CTPA and the potential influence of metal artefacts on interpretation accuracy (Fig. 3). While in this study, all PAVMs with suspected perfusion were successfully confirmed by angiography—suggesting a high positive predictive value—there is a paucity of information regarding false-negative rates. This underscores the necessity for further research to validate the reliability of measurement techniques and to assess the potential impact of artefact reduction methods on diagnostic accuracy. Third, only simple PAVMs were included in the study. Although investigating simple PAVMs probably allows for better assessment of embolic device occlusive performance, complex PAVMs are not uncommon and might respond differently to HVDNF coils. Lastly, this study does not include a cost–benefit analysis comparing HVDNF coils to other coil types. Given that HVDNF coils are generally more expensive than most 0.018″ coils, their use may be associated with higher overall costs. Assuming equivalent coil length, the volume of material of an average HVDNF coil is approximately 20% greater than that of a 0.018″ fibred coil (e.g. Interlock, Boston Scientific, MA, USA). Due to their increased volume, fewer HVDNF coils may be required to achieve a comparable packing density. Together with the lower recanalisation rate associated with HVDNF coils, and the consequent reduction in re-intervention rates, this may result in lower overall costs. This hypothesis, however, remains to be tested. CTPA follow-up after successful HVDNF coil embolisation procedure. Successfully treated simple PAVM in the left lower lobe with standard 12 mm × 60 cm (1 ×), standard 10 mm × 35 cm (1 ×), packing coil 60 cm and packing coil 45 cm (A–C). Sac embolisation was not performed because the number of coils required to fill the sac would be too high. Unenhanced pre-embolisation CT demonstrating a simple PAVM in the left lower lobe (D). Follow-up CTPA without (E) and with metal artefact reduction reconstruction (F). PAVM, pulmonary arteriovenous malformation; CTPA, computed tomography pulmonary angiography In conclusion, this study found that HVDNF coils are safe to use for de novo simple type PAVM embolisation, resulting in an excellent technical success rate and acceptable persistence rates at 6 months and 3 years. This study further highlights the need for standardised outcome definitions and prospective multicentre trials to evaluate embolic devices under uniform conditions. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgments Informed consent The need for informed consent was waived. Authors’ contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by D.A.F. van den Heuvel, D. Sachumsky and S. Klompmaker. The first draft of the manuscript was written by D.A.F. van den Heuvel and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This study was not supported by any funding. Declarations Ethics approval and consent to participate This study has obtained medical ethical approval from MEC-U (W21.132). Consent for publication For this type of study, consent for publication is not required. Competing interests The authors declare that they have no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2.