Optimal trial design for vascular access studies went under the microscope at the Vascular Access Society of Britain and Ireland (VASBI) 2025 annual scientific meeting (25–26 September, Bournemouth, UK), where speakers highlighted some of the challenges facing researchers in conducting randomised controlled trials within this space.
Speakers at the two-day conference—which brings together specialists in vascular access surgery, interventional radiology, nephrology and other professionals involved in dialysis access from the UK and further afield—discussed whether they could be best served using more data from retrospective studies or real-world registries to guide practice, as a way of overcoming some of the limitations of randomised trials which are seen as the gold standard of scientific research, but can be costly and complicated to administer.
Michael Robson, a consultant nephrologist at King’s College London (London, UK) set out the extent of the challenge facing trialists in the vascular access space, describing it as a “difficult job” to design and execute a trial. Offering details of the ongoing PAVE-2 trial, a multicentre double-blind randomised controlled trial to determine the efficacy of additional paclitaxel-coated or sirolimus-coated balloons for treating stenosis in arteriovenous fistulas (AVF)—for which he is the lead investigator—Robson outlined the scale of the task of bringing the trial from conception in late 2021 to its first patient enrolment in June 2024, months later than anticipated.
The trial, which has been funded by the National Institute for Health and Care Research (NIHR), is a three-armed trial aiming to include over 600 patients from 20 centres around the UK, assessing both the IN.PACT AV (Medtronic) and MagicTouch (Concept Medical) drug-coated balloons (DCBs) compared to a control group, with a primary endpoint of time to end of treatment segment primary patency.
“For PAVE-2, we wanted to try and confirm the findings of the IN.PACT AV access trials which showed that the Medtronic IN.PACT paclitaxel-coated balloon was effective, whereas other trials, the PAVE trial and the Lutonix trials using a different balloon had given a negative result,” Robson detailed. “We also wanted to look at sirolimus coated balloons in the same trial and we really wanted to provide a definitive answer as to whether drug-coated balloons were a good thing, so we wanted a big sample.”
Recruiting the size of population needed for the trial to provide meaningful evidence has been a particular challenge, as has organisation of the supply and payment for the devices used in the study. These were major factors pushing back the commencement of recruitment beyond both the anticipated mid-2023 start date, and the officially planned January 2024 start date. Robson said that the trial will need to enrol in the region of 20 patients per month to reach its recruitment targets, and NIHR are actively reviewing progress.
“Trials, studies and research come with different levels of evidence; systematic reviews are at the top of the pyramid, and randomised controlled trials are basically the gold standard,” commented Nicholas Inston (University Hospitals Birmingham, Birmingham, UK), a renal transplant and vascular access surgeon, offering a view on the ideal trial design for vascular access surgery research. Much of the evidence that underpins practice in the vascular access field, he said, is derived from what is perceived to be lower quality evidence which includes cohort studies and case series. “If you look at the guidelines there is very little that is led by randomised controlled trials and a lot of that is expert opinion,” he noted.
Inston offered an appraisal on the design, delivery and interpretation of randomised trials and questioned whether this level of evidence should necessarily be viewed as the default when it comes to shaping care that is most relevant to patients requiring vascular access interventions.
Factors such as operator variability can make direct statistical comparison challenging within a surgical setting, Inston commented, noting that issues such as a surgeon’s experience with certain types of procedure may also colour outcomes. “Different surgeons will do different procedures, and so it is very difficult to standardise this stuff,” he said. “There is a learning curve. Often these will be new devices, so what do you do with the first five cases? Do you do a training set so that people do five new grafts before we start collecting data?”
Inston also questioned whether, with a heavily selected population of patients, randomised trials can be considered truly representative of real-world practice. “Trials are going to have inclusion and exclusion criteria; many trials will exclude old patients, younger patients, patients with previous access, patients with certain comorbidities. This doesn’t apply to the real world, and the control group is actually pushed up because these patients are in a trial and are being looked at more rigorously and picked out,” he said, commenting that, instead, greater utilisation of real-world data in the form of registries is something that “shouldn’t be dismissed”.
Following Inston, interventional radiologist Robert Jones (Queen Elizabeth Hospital, Birmingham, UK) detailed what he saw as the key studies needed in the endovascular management of vascular access. “There are many unanswered questions both within the AV [arteriovenous] access maintenance space, and more recently within the AV access creation space with the advent of endoAVF,” he said.
According to Jones, an area that has been well studied is the application of covered stents for the treatment of AV access stenosis, which has level I evidence for both grafts and fistulas demonstrating superiority over angioplasty alone. Jones, however, questioned whether trials in this area adopt a “one-size-fits-all” approach, which, he said may miss some of the nuance of patient-specific factors that could impact outcomes.
“Covered stent may not always be appropriate, for example in the inflow segment. We know stenosis is the enemy of vascular access, and each stenosis is, to some extent, fairly unique, both in terms of its pathophysiology and its response to angioplasty. Lots of trials and studies that are being designed tend to adopt a one-size-fits all approach to stenosis,” he said.
Jones called for a “lesion specific” approach to answer questions in this arena, in order to draw out “the right treatment for the right lesion”.
“As a community we are very committed to work on improvement of the treatment of our patients, but if we compare ourselves with the people of cardiology, we do a rather poor job in terms of performing trials,” Joris Rotmans (Leiden University Medical Centre, Leiden, the Netherlands), chair of the session, commented during discussion that followed the presentations. Rotmans said that patients are not solely undergoing an intervention, but there is also very intensive use of the AV access afterwards, for which there is a lot of practice variation. “That is, to some extent, hampering the execution of these kinds of trials,” he commented.
Rotmans asked the panelists how best to deal with practice variation, in light of their comments, and how to minimise its impact on the outcome of trials.
“All surgeons will do something differently unless you have someone peering over their shoulder,” responded Inston, who commented that results can sometimes differ between trials depending on how they are funded. “Sometimes there is influence where you have got that industry rep behind you saying ‘do it like this’,” he said.
Jones commented that a lot of variation in practice can be attributed to how aware clinicians are of latest data. “Even in my own centre not everyone is aware of the latest stent graft data for example—a lot of it boils down to lack of awareness of what the data say,” he noted.
Robson offered the view that trials can be designed to minimise the impact of bias and to lessen the influence of operator variability on results. “The design of a trial to try and minimise bias is really important and blinding is part of that; how the patients are allocated and randomised is another thing,” he said. “In terms of practice variation, that is part of life, and you can account for that if you stratify or minimise according to centre, so that you make sure every centre has the same number of patients in each treatment arm.”
