Designing a platform/adaptive randomised controlled trial for peripheral arterial disease (PAD) – The PAEDIS international platform trial development project

Patient and public involvement

Patients were involved at all stages of this development process, including two formal patient co-applicants, who were part of the research team. Patients guided the research questions and guided all of the qualitative work in this development process. They chose outcomes of interest for the future potential study, outcomes, timings of assessment, format of recruitment methodology, inclusion or primary and community care pathways and duration of follow-up.

Study and process design

A mixed methods approach was used in order to address the project objectives, in 5 separate work packages. The format and duration of each work package were proposed originally by our lay co-applicants and were further developed by our research team. Also, qualitative information from our recently completed NIHR Research for Patient Benefit (RfPB) CHABLIS study (ISRCTN registration: ISRCTN13202085) investigating barriers and enablers to provision of best medical or surgical therapy in chronic limb threatening were used. In CHABLIS¹³ we interviewed 120 patients with advanced PAD (most with tissue loss or rest pain) treated in the NHS, to understand the pathways of treatment nationally, barriers to best medical therapy, barriers to receiving prompt intervention, and what they viewed as enablers to interacting with primary or secondary care.

The 5 work packages were subsequently further developed during a series of weekly meetings involving the core research team, including national and international representation across relevant specialties and healthcare professions involved in PAD care. Methodologists from 3 clinical trials units (CTUs; Edinburgh, Imperial, Leicester) with specific expertise in surgical trials and complex research delivery were involved. Further, the Vascular Society of Great Britain and Ireland (VSGBI), Royal College of Surgeons of England, British Society or Interventional Radiology (BSIR) were represented formally during these meetings to design the work packages. International vascular research networks were included at this stage i.e. the Research Collaborative for PAD (RCPAD - formal presence in 17 vascular centres across Europe) and Vascular and Endovascular Research Network (VERN - presence in 53 countries).

Work package 1 - evidence synthesis to inform subsequent study design

The 1^st work package commenced immediately upon confirmation of funding in January 2023.

The MEDLINE, EMBASE, clinicaltrials.gov, WHO Trials Database and ISRCTN were searched systematically (since inception) using Medical Subject Headings (MeSH) terms relating to symptomatic PAD in order to identify all of the following:

NHS Library resources were used. Abstracts were screened and reference lists searched to identify potential publications or online repository information regarding the aforementioned areas. An expert and lay summary of findings was prepared.

Work package 2 - setting up lay and expert groups to guide research design

Alongside the systematic review and evidence synthesis, we set up the lay and expert groups for this research

Lay groups

We recruited to a new project-specific patient and public involvement panel, including eight members from varied backgrounds. This group took part in our future workshops and focus groups. All lay participants received training online regarding how they can take part in focus groups and we answered all their questions.

Expert groups

We instituted an international group of relevant professional stakeholders, including representation across specialties involved in PAD care as well as industry representatives. We recruited from the European Society for Vascular Surgery (ESVS), the Society for Vascular Surgery (America) and existing links with Australia and New Zealand. Industry partners from companies which manufacture products and therapies identified in work package 1 were invited. The British Society for Endovascular Therapy invited key endovascular industry partners. We advertised the project via the Society for Vascular Nursing, physiotherapy, podiatry, pharmacy, and orthopaedic networks in the UK, NHS, and abroad.

Work package 3 - define the ideal characteristics and key performance indicators (KPIs) of a PAD platform trial

This involved a one-day workshop with representation from our expert project-specific group, our lay co-applicants and those from the PPI group who wish to take part. The event was recorded and analysed. Results from work package 1 were sent to all participants beforehand.

The following areas were discussed, in order to reach consensus:

Work package 4 - gain consensus to design the research

We agreed (online meetings of the core team and lay groups), on the following:

Work package 5 - finalised the trial protocol and funding application.

Data collection & data analysis

All workshops, focus groups, and (where necessary) interviews were recorded and, if appropriate, transcribed. Thematic analysis was used to summarise the key findings of qualitative data.

Evidence review (Results from work packages 1 and 2)

We performed three scoping literature reviews, 12 workshops (lay, industry & stakeholders’), meta-analyses on PAD technologies^14,15, contributed to international guidelines¹¹, and updated a network meta-analysis on exercise therapy¹⁰. Key findings:

-There are no ongoing complex PAD trials

-Medical and exercise therapy are the key constituents of conservative PAD care and interact with all invasive PAD technologies^10,16,17

-The recent BASIL2¹⁸ and BEST-CLI PAD RCTs^19,20 did not assess most new invasive PAD technologies or their interactions^19,21, despite having jointly costed £31 million

-Despite recent attempts^22,23, there is no effectiveness-data regarding novel antithrombotics in PAD¹¹

-Home-based PAD exercise programmes have unproven cost-effectiveness²⁴.

Design and research plan for the proposed complex PAD trial (Results from work packages 3–5)

Overall, we involved 44 lay participants, 81 healthcare professionals, representatives from four research funding bodies, and methodologists from five clinical trials units, (CTUs) who took part in the development-work Work Packages 3 to 5 and co-designed PAEDIS.

Our work proposed the following PICO design with regards to the proposed trial:

Participants

People with symptomatic PAD>18 years (intermittent claudication or CLTI)

Interventions

(selected as key PAD technologies needing effectiveness-testing in our reviews & multiple PPI/expert workshops):

-Home-based exercise therapy using the Motivating Structured Walking Activity in IC (MOSAIC) regime^24,25

-Oral anticoagulation with Rivaroxaban 2.5mg & Aspirin 75mg (before surgery)

-Endovascular (minimally invasive) revascularisation of the common femoral artery

-Use of arterial stents when revascularising below the knee arteries (crural)

-Use of ultrasound-based surveillance after endovascular revascularisation

Comparator: Standards of NHS care (pragmatic design)

Outcome (primary): 1) Composite of death and/or major lower limb amputation (amputation free survival) over a minimum two year follow-up for those with CLTI & 2) Quality of life (EQ-5D-5L) for those with claudication. Our PPI and previous evidence strongly support a minimum 2-year follow-up. Secondary outcomes include mortality, various validated quality of life measures, cardiovascular events, re-admissions, amputations, and a health-economic analysis.

Our work led to the below project plan for the proposed trial:

Design

Platform multicentre prospective RCT. The design is summarised in the flowcharts (Figure 1–Figure 3); it includes two work-packages, based on symptoms (claudication or CLTI) & anatomical segment treated.

Figure 1. Final design of the proposed platform trial.

Figure 2. Work package of the development process during 2023 in order to finalise the design of the proposed platform trial.

Figure 3. Proposed management plan for the institutions involved in the work.

WP 1 (pre-surgery). Four arm randomised (1:1:1:1) study involving people with claudication randomised to: 1) Referral for supervised exercise therapy (SET) & Clopidogrel (current NHS SoC), 2) MOSAIC exercise²⁵ & Clopidogrel, 3) SET referral & Rivaroxaban & Aspirin, 4) MOSAIC exercise & Rivaroxaban & Aspirin.

WP 2 (surgery or post-surgery). Two multiarm-multistage (MAMS) randomised (1:1:1:1) studies involving people with CLTI, depending on which arterial segment is treated. Arms will be:

For those with common femoral artery disease: 1) Endovascular surgery with post-operative ultrasound surveillance, 2) endovascular surgery without surveillance, 3) open surgery with surveillance, 4) open surgery without surveillance (SoC).

For those with below the knee (crural) artery disease: 1) plain angioplasty with ultrasound surveillance, 2) plain angioplasty without surveillance (SoC), 3) primary stenting with surveillance, 4) primary stenting without surveillance.

One interim analysis will take place in each WP2 component, to terminate or replace arms when there is strong evidence either in favour or against clinical effectiveness. Recruitment will continue during the interim. If results reveal an arm is to be terminated (lack of effectiveness vs. SoC), randomisation to that arm will be stopped, with participants allocated equally between remaining arms. If an arm is superior vs. SoC, recruitment to SoC will be stopped. The superior treatment will be the new comparator. Only if all arms are terminated will recruitment be stopped.

Setting

Thirty NHS hospitals & 42 primary care sites (including the 5 NHS regions with highest amputation & diabetes rates). NIHR CRN and vascular research collaboratives who co-developed PAEDIS will support set-up. Four major funders (USA, Europe, Australia, New Zealand) co-designed PAEDIS to allow expansion to 6 European countries, USA, Australia, New Zealand via an international protocol.

For WP1, recruitment will be community-focussed; WP2 will mostly recruit in hospitals. A bespoke PAEDIS web-based toolkit and software management system developed by TCR Nottingham, who support GP practices’ RCT recruitment UK-wide, will be used for WP1, co-managed with the STAR team (Nottingham). Eligible patients will be identified at each practice using an automated system and then recruited centrally.

Population, inclusion/exclusion criteria

Adult with symptomatic PAD, regardless of protected characteristics.

Technologies assessed

Rivaroxaban with Aspirin; Home-based exercise therapy (MOSAIC); Endovascular revascularisation of common femoral artery disease; Endovascular revasularisation of crural arteries using primary stenting (drug eluting); Ultrasound surveillance after endovascular revascularisation for symptomatic PAD. All are efficacy-tested and used in the NHS already.

Measurement of costs & outcomes

A prospectively planned economic evaluation will be conducted from an NHS and personal social services perspective, as per National Institute of Care Excellence (NICE) recommendations²⁶. Use of hospital and community services will be recorded using diaries/records. Quality-of-life will be assessed using EQ-5D-5L (alongside three other claudication questionnaires), converted to health-utility scores using the UK value set recommended by NICE. Unit costs will be based on manufacturers’ prices. Days lost from work and activities will be recorded and used in a secondary analysis. The extent and pattern of missing data will be assessed and appropriate methods employed. A state-transition decision model will be constructed to estimate costs and QALYs over the lifetime of the cohort. Outcomes or states of the model will be decided based on expert consultation and previous PAD economic evaluations.

Long-term follow up

Participants will be consented to allow data to be linked to routinely collected datasets across the NHS; we have collaborated with Health Data Research UK, British Heart Foundation, and National Vascular Registry (NVR).

Sample size, recruitment

Sample sizes are calculated to provide 90% power at 2.5% significance level (one-sided). Multiple testing adjustment is applied to allow for the fact that, in each population each experimental treatment is compared to SoC twice. We adjust for multiplicity by setting the significance level for each trial arm at 1.25% (one-sided); 2.5% thus represents the total family-wise type I error-rate per treatment.

WP1 utilises a multi-arm parallel group design. The primary outcome (EQ-5D-5L at 2 years) is continuous and normally distributed, with an estimated standard deviation of 0.2²⁷. The study aims to detect a treatment effect of 0.074, found to be the minimum clinically important difference in the EQ5D-3L measure amongst people with similar characteristics²⁸. This was deemed appropriate by our PPI. The required sample size is 183 per treatment arm, or 732 in WP1, increased to 772 allowing for 5% dropout.

WP2 utilises a multi-arm multi-stage design deployed in parallel in the CFA & BTK populations. Within each parallel trial there will be a single interim analysis using asymmetrical non-binding O’Brien-Fleming efficacy and futility boundaries. The primary outcome is amputation free survival at 2 years. The 2-year probability of amputation or death among CFA & BTK patients is 0.53 based on the latest BASIL-2 & BEST-CLI RCTs and our cohort studies/meta-analyses for CLTI^{8,14,18,19,29–34}. We aim to detect a 32% reduction in amputations/death in the arm receiving the more invasive treatment. This value was set by PPI participants as the magnitude of reduction they would require to make a more invasive treatment acceptable and accounting for delays to receive the randomised surgery and has been widely used in HTA-funded PAD research/trials (all BASIL trials). We anticipate recruiting 0.84 CFA and 0.38 BTK participants monthly based on BASIL-2, BASIL-3, BEST-CLI RCTs, and our current PAD research portfolio. The required sample sizes are calculated based on a constant hazard rate in each treatment arm and allow for expected accumulated events (partial follow-up). In the CFA component, a recruitment window of 25 months with an interim at 12 months from the start of recruitment would provide sufficient power in the absence of any dropouts or missing data. In the BTK component, a recruitment window of 45 months with an interim analysis at 32 months from the start of recruitment would be sufficient. These correspond to maximum sample sizes of 882 (CFA) and 718 (BTK). To allow for dropouts, missing data, and delays to receive surgery, we extend the recruitment window to 26 months (CFA) and 48 months (BTK), leading to a maximum sample size of 918 (CFA) and 766 (BTK). If recruitment to any treatment arm is stopped at the interim analysis stage, then the actual sample size required will be less than these maximums. As a result, the expected (mean) sample size is 885 (CFA) and 713 (BTK).

Statistical analysis

In WP1 the primary outcome will be analysed using mixed effects linear regression adjusted for covariates including a fixed effect for baseline EQ-5D-5L and a random effect for site. In WP2, time to amputation/death will be analysed using mixed effects proportional hazards regression adjusted for a random effect for site. Roughly 4% of claudicants enrolled in WP1 may progress to CLTI and becoming eligible for WP2; we do not expect this to cause substantial bias. Sensitivity analyses will explore alternative analytical strategies.