End-stage renal disease (ESRD) requiring haemodialysis (HD) is a growing global problem, impacting 554,038 US and 325,180 EU patients in 2018 [1, 2]. HD adequacy is best achieved by creating an arteriovenous access circuit, i.e. an autologous arteriovenous fistula (AVF) or a prosthetic arteriovenous graft (AVG). However, once created the access circuit is prone to complications, leading to problems with dialysis and even access abandonment.

The leading cause of dysfunction is stenosis associated with neointimal hyperplasia and adverse vascular remodelling [3]. Stenosis occurs in all access circuits. Both AVFs and AVGs develop stenoses at key anatomical points including in the draining veins, e.g. the cephalic arch in AVFs, or near the venous anastomosis in AVGs [3]. Both patients with AVFs and AVGs develop central vein stenosis (CVS). While CVS has largely been attributed to prior device use in the central veins, CVS may also develop in patients without prior intervention [4, 5].

Current stenosis treatments are sub-optimal with 12-month patency rates for percutaneous transluminal angioplasty (PTA) below 25% and limited benefit provided by additional bare metal stenting (BMS) [6,7,8,9,10,11,12,13,14]. Initial trials with drug-coated balloons (DCB) showed promise, though recent trials have questioned their impact [15,16,17,18]. Additionally, both PTA and DCB are ineffective when the stenosis is too elastic and recoils, or fibrotic and significant dissection occurs.

Stent grafts (i.e. covered stents or endoprostheses) are recommended for recurrent stenosis due to superior patency rates, reduced reintervention, and financial savings vs. PTA [8,9,10, 14]. Nevertheless, up to 70% of patients require reintervention within 12 months [9, 19]. Stent graft failure is multi-factorial. The two most established mechanisms are blood interactions with the device surface and edge-stenosis at the device’s inflow and outflow ends [20,21,22]. A third parallel mechanism was recently identified, namely transmural cell growth through the porous polymer covering leading to neointimal hyperplasia [23].

This article reports the results of a prospective, observational study investigating the safety and effectiveness of WRAPSODY Cell-impermeable Endoprosthesis for the treatment of access circuit stenosis. The study device has been designed with a multilayer, cell-impermeable polymer covering designed to limit restenosis and thrombus formation. The potential benefits of this design were demonstrated in a chronic ovine arterial model, where the study device exhibited reduced stenosis vs. a conventional stent graft [23].

Materials and Methods

Study Design

A multicentre, prospective, non-randomised, clinical trial was performed. Forty-six patients with clinically relevant access circuit stenosis were recruited from January 2019 through January 2020. The study was undertaken at three centres experienced with complex vascular access (Churchill Hospital, Oxford, UK; Queen Elizabeth University Hospital, Glasgow, Scotland; General Hospital of Athens “G. Gennimatas”, Athens, Greece). The trial is registered at (identifier: NCT03644017). The study followed good clinical practice guidelines and local regulatory requirements. An independent data monitoring/clinical event committee (DMC) was formed with three individuals with expertise in clinical trials and safety evaluations. The DMC reviewed and adjudicated reinterventions, access abandonments, as well as device and procedure-relatedness of adverse events.

Study Population

The eligible population comprised patients undergoing HD through an AV circuit with clinical and radiological evidence of stenosis in the peripheral or central veins. Eligible patients required a mature AVF in the arm with ≥ 1 successful dialysis session or an AVG created ≥ 30 days prior. The target lesion was ≤ 9 cm in length with at least 50% stenosis. Patients with secondary lesions were excluded. Table 1 lists all eligibility criteria.

Table 1 Inclusion and exclusion criteria

All patients provided written informed consent and were informed that they had the right to withdraw from the study at any time for any reason. Eighty-three patients were consented with 46 meeting the eligibility criteria and receiving treatment (consecutive enrolment). Enrolled patients included 16 patients with AVF and 10 with AVG and stenosis of the peripheral outflow veins (e.g. cephalic arch, basilic vein swing point); 9 with stenosis at the graft-vein anastomosis; and 11 with CVS. Figure 1 summarizes patient accountability. Tables 2 and 3 detail patient demographics and access circuit/target lesion characteristics. Tables S1 and S2 list patient’s concurrent medical conditions and anticoagulant/antiplatelet regimens.

Fig. 1
figure 1

Flow diagram showing patient enrolment and retention

Table 2 Baseline demographics of haemodialysis patients by access type and lesion location
Table 3 Access circuit and target lesion characteristics of haemodialysis patients

Study Device

WRAPSODY Endoprosthesis (Merit Medical Systems, Inc.; South Jordan, Utah, USA) is a self-expanding, cell-impermeable endoprosthesis. The endoprosthesis is composed of a helically wound nitinol wire stent fully encapsulated in a multilayer fluoropolymer covering (Fig. 2). The covering’s luminal layer is composed of a spun polytetrafluoroethylene (spun PTFE) designed to limit fibrin deposition and thrombus formation. The cell-impermeable middle layer is designed to prevent transmural cell growth. The outer layer is made of a traditional expanded polytetrafluoroethylene (ePTFE), permitting cell-ingrowth for device anchoring. The device is offered in various sizes, ranging in diameter from 6.0–16.0 mm (allowing treatment of vessels from 4.6–14.4 mm in diameter) and ranging in length from 30−125 mm.

Fig. 2
figure 2

Study device. (A) Representative image of a 14 mm diameter x 30 mm device. (B) Cross-sectional schematic surrounding a single nitinol wire illustrating the organization of the multilayer graft covering. Similar to other stent grafts, the abluminal layer is composed of ePTFE to facilitate cell-ingrowth to anchor the device in place. Directly adjacent to the nitinol wire are layers of cell-impermeable fluoropolymer to prevent transmural cell migration through the graft covering. Below the innermost cell-impermeable layer is a second layer of porous ePTFE. The luminal surface of the graft is composed of a spun PTFE designed to reduce fibrin deposition and subsequent thrombus formation

Study Treatment and Follow-Up

Procedures were undertaken by four operators across the three sites, all with experience in interventional procedures. An uncoated standard angioplasty balloon (type and size left to site discretion) was used for pre-dilation. Full expansion of the pre-dilation balloon was required. Following pre-dilation, reassessment, and eligibility confirmation, patients were treated with the study device. To ensure adequate fixation and wall vessel apposition, devices were deployed in a 10–25% oversized configuration (vs. adjacent healthy vessel) with at least 1 cm overlap with healthy vessel or the synthetic graft. Devices were post-dilated with a balloon no greater than the endoprosthesis’ diameter. If multiple devices were deployed in an overlapped configuration, post-dilation of the first device occurred before deploying the second. If different sized devices were overlapped to accommodate vessel diameter changes, the smaller device was deployed first. Patients were treated and discharged according to each centre’s standard of care, including anticoagulant or antiplatelet administration. Table S4 summarizes device disposition.

An in-person physical examination was scheduled at 1, 3, 6, and 12 months post-procedure with additional visits as needed for assessment and treatment of access circuit dysfunction. Follow-up times were chosen to facilitate analysis over the critical period when current interventions fail. Physical examinations included at minimum an assessment of hand, arm, neck, and trunk oedema; pain related to dialysis circuit; respiratory and neurological symptoms; skin changes; and a review of AEs and interventions. The access circuit was also assessed for quality of flow.

Outcome Measures

The primary safety measure was the proportion of patients without any localised or systemic safety events affecting the access circuit during the first 30 days that resulted in surgery, hospitalization, or death. The primary effectiveness measure was the target lesion primary patency rate (TLPP) at 30 days. TLPP was defined as the interval of uninterrupted patency from initial study procedure to the next intervention performed on the target lesion or uncorrectable target lesion occlusion, whichever occurs first.

Secondary outcome measures included clinical (resumption of successful dialysis for at least one session), anatomical (less than 30% residual stenosis), and procedural success (achievement of both clinical and anatomical success). Device- or procedure-related adverse events and TLPP were further assessed at 3, 6, and 12 months. Assisted TLPP, access circuit primary patency (ACPP), and access circuit secondary patency were evaluated at 1, 3, 6, and 12 months. Table S4 contains a definition for each outcome measure.

Statistical Methods

Statistics and Data Corporation (SDC) and Clinlogix provided statistical analysis services for the study. Statistics were calculated using SAS (v9.4, SAS Institute Inc.). As this was a single-arm study, statistical analysis only included descriptive statistics (e.g. mean and standard deviation). Patency values were calculated using the Kaplan–Meier method. An interim analysis was preformed when all patients completed the 6-month visit.

Role of the Study Sponsor

Study devices were supplied by the sponsor. The sponsor was involved in establishing the study design; data collection, analysis, and interpretation; as well as the preparation of and decision to submit the manuscript. All authors confirm that they had full access to the study data and accept responsibility for publication submission.


Procedural Outcomes

Anatomical, clinical, and procedural success were achieved in all cases (100% [46/46]). Figure 3 highlights the procedural outcomes of an example case.

Fig. 3
figure 3

Example procedural outcomes. Treatment of a patient with a left brachiocephalic AVF that presented with a stenosis at the cephalic arch extending into the distal subclavian vein. (A & B) Fistulograms taken (A) preintervention and (B) after pre-dilitation of the stenosis. (C) Fluoroscopy of fully deployed study device. (D) Completion fistulogram

Safety Outcomes

One patient required surgery within 30 days of the study procedure following AVF circuit thrombosis (primary safety of 97.8% [45/46]). As reported by the site and confirmed by the DMC, this event was related to a pre-existing perianastomotic seroma causing extrinsic compression near the inflow end of the circuit and not related to the study procedure or device. With this event, there were 47 AEs in a total of 16 patients and 17 serious adverse events (SAEs) in 10 patients. None were unanticipated adverse device effects. All events followed expectations for the patients enrolled.

There was one AE, a thrombosed fistula, where the access circuit was not salvageable and there was no imaging to assess device-relatedness. Therefore, this event was listed as possibly device- and procedure-related (Table 4). Four other AEs were adjudicated by the DMC to be only procedure-related. Of note, one AVG peripheral patient experienced brachial vein rupture at the vascular access site. The rupture was successfully treated via balloon tamponade and placement of additional non-study stent grafts. One central patient experienced migration of two overlapped devices during the index procedure. The investigator, DMC, and an independent physician review (following DMC recommendation) adjudicated that migration was the result of inappropriate vessel sizing during pre-dilation (oversized pre-dilation balloon was used to assess vessel diameter) leading to the placement of inappropriately sized study devices. Specifically, the devices were excessively oversized for the vessel at the peripheral aspect of the stents, in contraindication to the IFU. This in combination with malapposition to the vessel wall along the central aspect of the device led to the migration. The devices were repositioned to the inferior vena cava where they were believed to be well seated and two additional, appropriately sized devices were placed to treat the original stenosis. Upon further assessment at the 30-day visit, it was observed that the initial two devices had migrated again (likely due to inadequate wall apposition during Valsalva manoeuver). To ensure against any further migration, the devices were pulled to the iliac vein and secured with an additional high radial force bare metal stent. No further migration issues occurred in either the repositioned devices or the replacement devices placed at the target lesion.

Table 4 Device-and procedure-related events as adjudicated by DMC

Performance Outcomes

One patient underwent kidney transplant 7 days after the index procedure and was excluded from all patency calculations. No reinterventions were required to maintain target lesion patency through 30 days (primary effectives of 100% [45/45], Fig. 4a). Six-month TLPP was 97.7% (42/43) and 84.6% (33/39) at 12 months (Fig. 4a). TLPP was similar irrespective of access type and lesion location (Fig. 4b) or whether patients were treated with a single device or multiple overlapped devices (Table 5). Reintervention at the target lesion was successful in maintaining patency in all cases (12-month assisted TLPP: 100% [39/39]). ACPP was 84.4% (38/45) at 6 months and 65.9% (29/44) at 12 months (Fig. 5). Access circuit secondary patency was 88.6% (39/44) at 12 months (Table 6).

Table 5 Target lesion primary patency (TLPP) for patients treated with a single or multiple overlapped devices
Fig. 4
figure 4

TLPP through 12 months. Kaplan–Meier (KM) curves for (A) all patients and (B) sub-cohorts based on access type and lesion location. The tables below each graph list the number of patients at risk, those censored (i.e. exited the study for renal transplant or loss to follow-up), and those who experienced an event leading to the loss of TLPP. Triangles on the KM curves represent either a patient’s date of censoring or final follow-up visit

Fig. 5
figure 5

Kaplan–Meier curves for access circuit primary patency (ACPP). ACPP through 12 months. Kaplan–Meier (KM) curves for (A) all patients and (B) sub-cohorts based on access type and lesion location. The tables below each graph list the number of patients at risk, those censored (i.e. exited the study for renal transplant or loss to follow-up), and those who experienced an event leading to the loss of ACPP. Triangles on the KM curves represent either a patient’s date of censoring or final follow-up visit

Table 6 Secondary patency of the access circuit


Results indicate that the study device can safely and effectively treat stenoses in the peripheral outflow and central veins of AV access circuits. Only one safety event (unrelated to the study device or procedure) occurred within 30 days of treatment and only one AE was adjudicated as possibly study device-related. The 6- and 12-month TLPP and ACPP rates were unexpectedly high. While needing further confirmation, the study device’s design may be a contributing factor in limiting restenosis. If confirmed in larger studies, this degree of patency preservation should reduce revision frequency and access abandonment rates, leading to less hospitalisation and lower healthcare costs [24, 25].

The high patency rates were seen across all access types and lesion locations. While not a comparative study, performance in patients with a peripheral stenosis in their AVF access circuit and in patients with stenosis of the graft-vein anastomosis compares favourably to the results from previous large-scale studies that have investigated other stent grafts for these uses [7, 9, 19, 26,27,28]. Furthermore, current results suggest that the device can safely treat CVS in the subclavian and brachiocephalic veins. Though care should always be given when using a stent graft in the central veins to avoid damaging the device due to external compression from the clavicle or jailing important collaterals.

Patency rates also compare favourably to historic results for other treatments (i.e. PTA, BMS, and DCB). However, it must be remembered that stent graft use may not be universally appropriate. One should consider the potential for and impact of jailing collateral vessels, restricting future access sites, or compromising device integrity (via compression, flexion, or puncture). Furthermore, appropriate vessel sizing is critical to avoid vessel damage and to ensure against device migration.

The high patency results vs. historical results for other devices may be attributed to a variety of factors. A previous preclinical study showed that the study device had a reduced stenosis rate than a conventional stent graft [23]. Therefore, these findings may be due to the study device’s underlying stent mechanics and multilayer graft covering. Beyond device design, differences in eligibility criteria may also underlie the current high patency results. This study excluded patients with thrombus or secondary lesions in the access circuit and other studies have noted a marked patency rate reduction when patients present with either condition [26,27,28]. This exclusion was intentional in order to avoid other variables confounding the assessment of safety and access circuit outcomes that might be associated with the novel aspects of the study device. Nevertheless, the lack of patients with thrombus or secondary lesions does not appear to fully explain patency rate differences with previous studies. Specifically, the respective 6-month TLPP rates for patients without thrombus or secondary lesions from these previous studies ranged from 64.6–76.9% and 57.0– 79.3% while the ACPP rates were 42.3–49.7% and 45.1–59.8% [26,27,28].

Beyond the stringent eligibility criteria, another limitation is the study’s size and lack of a control. To address these limitations, a larger, multicentre study, The Merit WRAPSODY AV Access Efficacy Study (NCT04540302), has been initiated. The study also enrolled patients with AVF and AVG access with a variety of commonly occurring lesions impacting these access circuits. While this design facilitated an initial screening of potential issues across the array of haemodialysis patients, case heterogeneity also limited the conclusions that can be drawn regarding specific situations. Other limitations include lack of a standardized measurement procedure and image assessment by an independent angiographic core laboratory. The current study also left use of anticoagulants and antiplatelets to the discretion of each site, creating the potential for site-to-site variability.


Current results indicate that the study device may represent a safe and effective treatment for AV access circuit stenoses. The high patency rates are promising and may offer patients with access circuit dysfunction improved treatment with more durable outcomes. Additional follow-up in larger RCTs is needed to verify these results.