An evaluation of the effect of an angiotensin-converting enzyme inhibitor on the growth rate of small abdominal aortic aneurysms: a randomised placebo-controlled trial (AARDVARK)

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 08/109/02. The contractual start date was in April 2011. The draft report began editorial review in September 2015 and was accepted for publication in April 2016. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Neil Poulter receives grants and personal fees from Servier, outside the submitted work. Janet Powell receives grants from the National Institute for Health Research and the Camelia Botnar Research Foundation and personal fees from the American Heart Association, outside the submitted work. Colin Bicknell receives personal fees from Hansen Medical, Medtronic and Bolton Medical, outside the submitted work.

Copyright statement

© Queen’s Printer and Controller of HMSO 2016. This work was produced by Kiru et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Introduction

An abdominal aortic aneurysm (AAA) is a ballooning of the infrarenal aorta to either 1.5 times its normal anteroposterior (AP) diameter or to an absolute value of ≥ 3 cm. 1 Small AAAs can be defined as those between 3.0 cm and 5.4 cm in diameter. These small AAAs have a low risk of rupture, whereas the operation to repair them is fatal in approximately 2–3% of patients. 2 Small AAAs are generally managed by optimising cardiovascular health and placing the patient on a surveillance programme to measure the AAA diameter at regular intervals. Once AAAs reach 5.5 cm (or if initially detected at a larger size), they are often repaired as the risk of rupture rises exponentially above this size. If rupture does occur, the results of emergency aneurysm repair are not good, with only about 40% of patients surviving. Without repair few survive, so that overall the survival of AAA rupture is probably < 20%. Although recent reports have suggested that the incidence of aneurysms appears to be in decline,3,4 AAA remains a significant health risk in the older population, with around 4000 deaths each year in England and Wales attributed to AAA rupture. 5

Except when they rupture, most AAAs are usually asymptomatic and so, until recently, they were detected as an incidental finding on clinical examination or ultrasonography, abdominal computerised tomography (CT) or magnetic resonance imaging performed for other purposes. However, in the UK, the NHS Abdominal Aortic Aneurysm Screening Programme (NAAASP) was introduced in 2009 and so many more small AAAs are now being detected early. The programme has been very successful and screened its millionth man in autumn 2015. There is an opportunity to reduce the number of patients needing AAA repair if AAA growth can be slowed or prevented in this growing cohort of patients.

Although data on the effects of angiotensin-converting enzyme inhibitors (ACE-Is) in this context are not consistent, ACE-Is have been associated with a reduced incidence of AAA rupture in analysis of administrative databases. 6 Previous trials of some other drugs to slow AAA growth have been hindered by poor patient compliance. 7 Therefore, this pilot trial was undertaken to assess whether or not ACE-Is could potentially slow AAA growth and are well tolerated in doing so. We are unaware of any other completed randomised controlled trials (RCTs) designed to examine the efficacy of ACE-Is or angiotensin receptor blockers (ARBs) in limiting or inhibiting AAA progression, although two trials of the impact of ARBs on the growth of AAAs are in progress.

Risk factors

A wide variety of risk factors have been attributed to the formation and progression of AAAs. The single most important risk factor for AAAs has consistently been found to be smoking,8–10 although other risk factors including male sex, age, high blood pressure (BP) [particularly raised diastolic BP (DBP)] and family history of AAA are frequently linked with the aetiology of AAA. 11 Low prevalence rates have been observed among African12 and Asian13 men compared with Caucasian men. Several studies have also found a strong coexistence of localised and generalised atherosclerosis and AAA,14,15 an underlying disturbed connective tissue metabolism16 and an increased risk for AAA with increasing alcohol consumption. 17

There are many genetic syndromes that are associated with aortic aneurysmal disease affecting patients often at a very early age, including Marfan syndrome, Ehlers–Danlos syndrome, Loeys–Dietz syndrome and familial thoracic aortic aneurysms and dissections. 18

The NHS Abdominal Aortic Aneurysm Screening Programme

Phased implementation of the NAAASP began in July 2009. It was introduced after data from a number of studies and existing local screening programmes in England showed a reduction in aneurysm-related mortality when men aged ≥ 65 years were offered ultrasound screening.

The Multicentre Aneurysm Screening Study (MASS) was designed to assess whether or not AAA screening would be beneficial. 19 This study enrolled men aged 65–74 years who were randomised to receive screening or not. Extended follow-up of patients confirmed that screening resulted in a reduction in all-cause mortality. Over 13 years there were 224 AAA-related deaths out of the 33,883 participants in the invited group and 381 AAA-related deaths out of the 33,887 participants in the control group, a 42% relative reduction. 20

In 2005 this evidence was assessed by the UK National Screening Committee, which concluded that ultrasound screening should be offered to men in their 65th year, with men aged ≥ 65 years being able to self-refer within the NHS. 5

The initial outcomes from the NAAASP in England identified a lower prevalence of AAA than reported in the MASS (1.4% vs. 4.7%). However, in the MASS, men aged 65–74 years were included, whereas the NAAASP invites men in their 65th year only for screening.

Between 2009 and 31 March 2014 the NAAASP had scanned > 700,000 men and referred > 1000 men with a large AAA for surgery. In the period 2013–14, 491 of the screened men had an elective AAA repair and four of these men died (an elective repair mortality rate of 0.8%). In addition, seven of the 10 screened men who suffered aneurysmal rupture died (a rupture mortality rate of 70%). 21

Because of the NAAASP a greater number of patients with a small AAA are being detected and, if there were effective treatments to slow AAA growth, this could provide an opportunity to intervene before the AAAs expand significantly and rupture. Also, the NAAASP potentially provides a useful pool of patients for research purposes, not only for logistical reasons but also because this group of patients (who were previously unaware of their AAA) may be receiving less clinical/medical intervention than patients who are already receiving monitoring for their AAA. They are therefore of particular interest for interventional studies.

Current guidelines for the management of small abdominal aortic aneurysms

Given the variability in aneurysm expansion rates,22 the optimal interval between surveillance scans remains uncertain. Meta-analysis of small AAA growth rates has demonstrated that the screening interval should be dependent on diameter and that long intervals between screening may be safe for the majority of patients. 23 However, guidelines must balance the need to reduce the cost of surveillance programmes and the need to ensure safety, as well as increasing the face-to-face time for direct cardiovascular risk factor education.

The UK guidelines recommend that rescreening intervals should shorten as the aneurysm enlarges and these guidelines are expected to be updated in 2017. 24 Usual clinical practice in the UK, and for those patients in the NAAASP, involves follow-up surveillance imaging at 12-monthly intervals for patients with an AAA of 3.0–4.4 cm in diameter and at 3-monthly intervals for those patients with an AAA diameter between 4.5 cm and 5.4 cm.

Abdominal aortic aneurysm treatment

In both the USA and the UK, elective surgery by either open or endovascular repair is undertaken to prevent AAA rupture and this is generally recommended for patients with an AAA of ≥ 5.5 cm in diameter, for symptomatic aneurysms or for aneurysms that have increased by > 0.5 cm in the past 6 months. The UK Small Aneurysm Trial (UKSAT) demonstrated that the overall mortality of patients with an aneurysm of < 5.5 cm in diameter who received surveillance was similar to that in patients who received early open surgery. 22 Furthermore, surveillance was the more cost-effective option. Subsequent studies that have randomised patients to surveillance or endovascular treatment of AAAs have corroborated this finding. 25,26 There has been an increasing trend in the proportion of repairs performed as endovascular aneurysm repair (EVAR) procedures, increasing from 54% in 2009 to 66% in 2013. 27

Studies indicate that without surgery the 5-year survival rate for patients with an aneurysm of diameter > 5 cm is about 20%. 28 Surgery to replace the aneurysmal segment or endovascular placement of a covered stent graft excluding the aneurysm is recommended if the risk of aneurysm rupture is high enough to justify the risk of surgery. The rate of rupture of an aneurysm rises exponentially after it reaches 5.5 cm in size, justifying the need for repair in most patients as aneurysm rupture is associated with a high mortality rate. Approximately half of the patients with a ruptured AAA fail to reach hospital and, of those patients who undergo emergency surgery, there is a 35–40% mortality rate at 30 days. 29

Open surgical repair carries a significant risk of perioperative morbidity and mortality. The 2014 National Vascular Registry report stated that, over the 5 years between 2009 and 2013, in-hospital mortality for open repairs was 3.6%. 27 The less-invasive EVAR technique has significantly lowered perioperative morbidity and mortality rates27 but not all patients are anatomically suitable for EVAR and there is still debate regarding the long-term benefits of EVAR. Several trials, including the EVAR,30 DREAM (Dutch Randomized Endovascular Aneurysm Repair)31 and OVER (Open Versus Endovascular Repair)32 trials, have found that the advantage of EVAR over open repair is lost during mid-term follow-up, with survival rates beyond 2 years being similar in both groups.

Given the risks involved with AAA repair, strategies to reduce the need for surgery are needed. Currently, there are no clear recommendations on pharmacological treatment approaches to prevent aneurysm progression or reduce the risk of rupture. 33

Growth rates of small abdominal aortic aneurysms

The growth rate of AAAs is highly variable both between patients and in the same patient over time. Average growth rates increase as the aneurysm enlarges. The average growth rate for a 3.5-cm aneurysm is estimated at 1.90 mm per year, whereas that for a 4.5-cm aneurysm is 3.52 mm/year. Therefore, given an exponentially increasing aneurysm diameter, it would take an average of 6.2 years for a 3.5-cm aneurysm to grow to 5.5 cm, whereas a 4.5-cm aneurysm would grow to 5.5 cm in 2.3 years. 34 These growth rates highlight the need for very accurate measurements of AAA to be obtained in a trial setting.

In the UKSAT, AAA growth was most strongly associated with diameter at baseline22 and smoking was associated with an incrementally increased growth rate of 0.4 mm per year. Multivariate analysis of other potential risk factors demonstrated that the presence of peripheral arterial disease (adversely) or diabetes (beneficially) influenced AAA growth.

The risk factor profile for aneurysm growth has been reproduced in other studies, with AAA growth being increased in smokers8,35 and decreased in patients with diabetes. 36

Average baseline diameters and growth rates reported in the Western Australia screening study,37 MASS,19 Propranolol Aneurysm Trial,7 UKSAT22 and Second Manifestation of ARTerial disease study38 are shown in Table 1.

Rupture rates of small abdominal aortic aneurysms

Aneurysm size is one of the strongest predictors of the risk of rupture, with risk increasing considerably for aneurysm diameters of > 5.5 cm. The UKSAT reported the risk of rupture for AAAs up to 5.5 cm in diameter to be < 1 per 100 person-years in men and 3 per 100 person-years in women. 39

Similarly, the 5-year overall cumulative rupture rate of incidentally diagnosed AAAs in population-based samples is 25–40% for aneurysms of > 5.0 cm diameter and 1–7% for aneurysms of 4.0–5.0 cm in diameter. 40,41

Rupture rates have been found to be doubled in current smokers compared with ex-smokers or non-smokers (p = 0.001) and to increase with mean arterial pressure (per 10 mmHg) (p = 0.001). 42

Blood pressure and abdominal aortic aneurysms

Raised BP was the leading risk factor contributing to the overall global burden of disease in 2010. 43 The recent decrease in case fatality rates associated with acute cardiovascular events in high-income countries has been associated with a rise in the numbers of patients living with cardiovascular disease and the wider use of preventative drugs in the context of primary and secondary prevention.

An association between hypertension and the incidence of AAA is frequently cited. 12,44 The CALIBER (CArdiovascular research using LInked Bespoke studies and Electronic health Records) study used linked electronic health records to assemble a cohort of > 1 million patients aged ≥ 30 years and initially free from cardiovascular disease, one-fifth of whom received BP-lowering treatments. 36

Of all cardiovascular diseases, AAA had the strongest association with DBP [hazard ratio (HR) per 10 mmHg 1.45, 95% confidence interval (CI) 1.34 to 1.56] and mean arterial pressure (HR 1.61, 95% CI 1.48 to 1.75) and the weakest association with systolic BP (SBP) (HR 1.08, 95% CI 1.00 to 1.17). Furthermore, it was the only cardiovascular outcome for which the association with higher pulse pressure was reversed (HR 0.91, 95% CI 0.86 to 0.98). 36

However, mean baseline BP levels reported in several large AAA surveillance studies7,19,22,37 (Table 2) are all above what is currently considered as controlled (< 140/90 mmHg45) and the AAA growth rates observed in these studies (see Table 1) may at least in part be related to these higher BPs.

Despite the relatively strong association between hypertension and the prevalence of AAA, the association between increased BP and the rate of AAA growth or incidence of rupture is not clear and the evidence supporting increased growth as a result of hypertension is lacking.

Non-pharmacological treatments to reduce the growth and rupture rate of abdominal aortic aneurysms

There is clear observational evidence that smoking increases the likelihood of developing an AAA. 11,12,46 For example, in the large (n = 114,567) Aneurysm Detection and Management (ADAM) screening study, a history of ever smoking was associated with an odds ratio of 2.97 (95% CI 2.65 to 3.32) for 3.0- to 3.9-cm AAAs and 5.07 (95% CI 4.13 to 6.21) for ≥ 4-cm AAAs. 46 In addition, a recent prospective population-based study of 92,728 men in Oxfordshire found that men aged 65–74 years who were current smokers had a 3% 10-year risk of acute AAA, highlighting the need for screening campaigns to reach this high-risk group. 47

Furthermore, several studies8,15,17 and meta-analyses48 have found higher growth rates in current smokers than in past smokers.

Consequently, the standard non-pharmacological treatment for AAA is smoking cessation. However, it has been suggested that smoking cessation may lose some of its importance once significant aortic dilatation has occurred. 35

Pharmacological treatments to reduce the growth and rupture rate of abdominal aortic aneurysms

There remains a significant need to find medical therapies that could reduce the growth and rupture rates of small and medium-sized AAAs.

As well as interest in the development of new AAA-specific treatments, there has also been interest in assessing the impact of treatments already in use for other indications. Early evidence often arises from animal studies but a small number of RCTs in humans have been carried out to assess the efficacy of some of the currently available drugs.

The role of the renin–angiotensin system

The renin–angiotensin system (RAS) is a peptidergic hormone system that has been recognised to be highly involved in disturbances of the cardiovascular system. RAS blockade by ACE-Is and ARBs has been found to not only decrease arterial pressure but also prevent or reverse endothelial dysfunction and aspects of the atherosclerotic process, which results in a reduction in cardiovascular mortality and morbidity. 65,66

In experimental studies both angiotensinogen and angiotensin type 1 receptors have been found to be upregulated by approximately twofold in the walls of AAAs compared with the walls of atherosclerotic aortas, although the expression of angiotensin type 2 receptors was similar. 67 In hypercholesterolaemic mice, angiotensin II infusion induces medial dissection of the aorta proximal to aortic branch points, with subsequent formation of suprarenal aortic aneurysms, which can be prevented with the use of ACE-Is. 68

Furthermore, in a recent study, perindopril (Coversyl arginine, Servier) inhibited aortic degeneration and AAA formation in the experimental AAA model induced by elastase and calcium chloride. 69

Designing the trial

To date we are unaware (based on a recent literature review) of any other completed RCTs designed to examine the efficacy of ACE-I or ARBs in limiting or inhibiting AAA progression. This report describes the AARDVARK (Aortic Aneurysmal Regression of Dilation: Value of ACE-Inhibition on RisK) trial, which was commissioned by the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) programme to address this need.

Objectives

Final study design

This randomised, single-blind, placebo-controlled study took place at 14 investigational sites in England. The trial consisted of three parallel randomised arms, with patients receiving 10 mg of perindopril daily (arginine salt), placebo daily (primary comparison) or 5 mg of amlodipine daily (secondary comparison). The perindopril and amlodipine doses used were estimated to have similar effects on BP reduction73–75 and hence the secondary comparison was included to assess whether or not any benefits of perindopril were independent of BP reduction.

Trial participants

The following factors were taken into consideration when deciding on the inclusion and exclusion criteria for the trial:

1. Abdominal aortic aneurysm size. Patients with an AAA of ≥ 5.5 cm diameter would be considered for surgical intervention as per the current clinical guidelines. Therefore, patients with an AAA of ≤ 5.4 cm diameter were included to minimise the rate of patient withdrawal from the study.

2. Sex. Although a lower prevalence of AAAs has been found in women, there is no evidence to suggest that the trial medications would have a different mechanism of action between the sexes. Therefore, both men and women were included.

3. Age. Assuming potentially differential benefits of ACE-Is for patients with AAAs related to genetic syndromes, a lower age limit (initially 60 years and then amended to 55 years) was set to more effectively screen out this group.

4. Ethnicity. Although a higher prevalence of AAAs has been found in Caucasian men than in black and Asian men, there is no evidence to suggest that the trial medications would have a different mechanism of action between races, specifically on AAA growth. Therefore, patients from all ethnicities were included.

5. Blood pressure. Regarding the treatment of hypertension, the 2011 National Institute for Health and Care Excellence (NICE) guidelines45 state:

   • aim for a target clinic BP of < 140/90 mmHg in people aged < 80 years with treated hypertension

   • aim for a target clinic BP of < 150/90 mmHg in people aged ≥ 80 years with treated hypertension.

However, the Quality and Outcomes Framework target for general practitioners (GPs) is a SBP of < 150 mmHg in people aged < 80 years. It seemed reasonable, therefore, to set an inclusion criterion of a SBP of < 150 mmHg in line with this Quality and Outcomes Framework target. Otherwise, eligible patients who had a SBP of ≥ 150 mmHg could be subsequently recruited into the trial but only after suitable BP medication had been supplied and the SBP was controlled to < 150 mmHg (see Planned drug interventions).

1. Medical history. Patients with any medical conditions that would interfere with their participation in the trial or who would be at an increased risk of adverse effects by taking the study medications were excluded from the trial.

2. Concomitant medications. Patients already receiving an ACE-I, a CCB or an ARB could not participate in the trial. The only exception was patients receiving 5 mg of amlodipine because the maximum dosage of amlodipine is 10 mg, thereby allowing the in-trial allocation to a further 5 mg.

The final inclusion and exclusion criteria are listed in the following sections. Only patients who met these criteria were considered for inclusion in the trial.

Recruiting centres

Participants were recruited from 14 centres across England (Figure 1). Initially, patients were recruited from five centres:

• Imperial College Healthcare NHS Trust (St Mary’s Hospital)

• Imperial College Healthcare NHS Trust (Charing Cross Hospital)

• Guy’s and St Thomas’ NHS Foundation Trust (St Thomas’ Hospital)

• Royal Free London NHS Foundation Trust (Royal Free Hospital)

• University Hospitals Coventry and Warwickshire NHS Trust (University Hospital Coventry).

FIGURE 1.

Organisation of recruiting sites. DAR, Dartford Hospital; ICCH, International Centre for Circulatory Health; KCH, King’s College Hospital; LEW, Lewisham Hospital; MTW, Maidstone and Tunbridge Wells; QE, Queen Elizabeth Hospital, Woolwich; WMX, West Middlesex University Hospital.

The original arrangement was for there to be only two sonographers in the study, one to perform all of the scans on patients at the London sites with the same mobile ultrasound scanner and one to perform all of the scans on patients from University Hospital Coventry on the same model of ultrasound scanner located in Coventry. This was principally to reduce intersonographer variability.

However, before the first patient was recruited into the study it was decided that patients in London should be screened and recruited at their local research site but that all visits and measurements from baseline onwards would take place at a central hub, the International Centre for Circulatory Health (ICCH), Imperial College London. The advantages of this were that:

1. Patients would have complete flexibility in the days/times of their study visits.

2. In the case of cancellations or missed appointments, patients could be rebooked without restriction, making it less likely for them to fall out of their protocol-defined visit window.

3. Experts in hypertension and antihypertensive medications and their side effects are based on site at ICCH and were available to see at short notice patients who had experienced any AEs.

4. The issue of being able to identify available clinic space to conduct the patient visits at each of the research sites was overcome.

5. In the case that the sonographer was unable to conduct patient visits (because of sickness, annual leave, etc.), the back-up sonographer (who was based at ICCH) was available to conduct visits at short notice with minimal disruption to her other duties.

Nine further sites were later added to enhance recruitment:

1. Hull and East Yorkshire Hospitals NHS Trust (Hull Royal Infirmary)

2. Royal Bournemouth and Christchurch Hospitals NHS Foundation Trust (Royal Bournemouth Hospital)

3. Colchester Hospital University NHS Foundation Trust (Colchester General Hospital)

4. Newcastle upon Tyne Hospitals NHS Foundation Trust (Freeman Hospital)

5. City Hospitals Sunderland NHS Foundation Trust (Sunderland Royal Hospital)

6. York Teaching Hospitals NHS Foundation Trust (York Hospital)

7. Norfolk and Norwich University Hospitals NHS Foundation Trust (Norfolk and Norwich University Hospital)

8. Central Manchester University Hospitals NHS Foundation Trust (Manchester Royal Infirmary)

9. Sheffield Teaching Hospitals NHS Foundation Trust (Northern General Hospital).

Patient identification centres

Several patient identification centres (PICs) were also added to the study to further enhance recruitment (see Figure 1). The PICs identified potential participants for the trial at their sites and referred them to the associated research site for recruitment into the trial. The following PICs were approved for the study:

• West Middlesex University Hospital NHS Trust (West Middlesex University Hospital)

• King’s College Hospital NHS Foundation Trust (King’s College Hospital)

• Maidstone and Tunbridge Wells NHS Trust (Tunbridge Wells Hospital)

• South London Healthcare NHS Trust (Queen Elizabeth Hospital, Woolwich)

• Lewisham and Greenwich NHS Trust (Lewisham Hospital)

• Dartford and Gravesham NHS Trust (Dartford Hospital).

Recruitment

The clinical registries and the NAAASP databases (when relevant) at the study sites were used to identify patients with an AAA. Permission to recruit NAAASP patients was obtained from the NAAASP research committee. All patients were then prescreened against the inclusion and exclusion criteria. Sites were asked to capture the patient initials for each identified patient and to record the reasons for ineligibility or for non-participation to help inform study progress. These patients were entered onto the Consolidated Standards of Reporting Trials (CONSORT) flow chart. Eligible patients were then subsequently invited to consider entry into the trial.

It was the investigators’ responsibility to obtain written informed consent from patients after adequate explanation of the aims, methods, anticipated benefits and potential hazards of the study, and before any study procedures commenced. Potential participants were given a copy of the patient information sheet (PIS) and informed consent form (ICF) (see Appendix 1). The original copy of the signed and dated ICF was retained by the site.

Patients were given at least 24 hours to read the PIS and consider their participation.

Planned drug interventions

The primary comparison was the effect on AAA growth of the ACE-I compared with placebo; one-third of randomised patients received 10 mg of perindopril arginine daily and one-third received placebo daily. To evaluate the BP-independent effects of the ACE-I, one-third of patients were randomised to a CCB (5 mg of amlodipine daily). It was estimated that, at these doses and in this population, the two drugs would produce similar average BP-lowering effects of approximately 6/4 mmHg. This protocol also allowed both drugs to be compared with placebo to evaluate any BP-lowering impact on AAA growth.

If the SBP of potential trial recruits was > 150 mmHg at screening, sites were asked to arrange for these patients to receive 1.5 mg of slow-release indapamide daily (either prescribed locally or through their GP) or, if this was not appropriate, 5 mg of amlodipine for a 6-week period. Such patients then had their BP measurements repeated after 6 weeks and if their SBP had fallen to < 150 mmHg they were eligible to proceed to randomisation.

The most common side effect of ACE-Is is cough, which affects about 15% of those treated. 76 However, exclusion of those with a pretrial history of ACE-I intolerance was enforced to reduce the incidence of this problem. During the trial, this side effect was monitored, particularly as we anticipated (based on past research) that the majority of patients in the trial would be smokers or ex-smokers who tend to tolerate ACE-Is less well. When in-trial cough was persistent and intolerable, patients stopped medication for 2 weeks and if the cough resolved they were changed to the ARB losartan (100 mg per day). If the cough continued (and hence was deemed to be unrelated to the trial medication), perindopril was restarted.

For all patients recruited into the trial who were not currently receiving a statin, sites were advised to request that the their GP prescribe a drug in this class as per current guidelines.

Visit schedule

The study visit schedule for the AARDVARK trial is shown in Table 3.

Each patient had a maximum of 10 planned study visits. The visit schedule and interventions were as follows:

1. screening – informed consent, collection of demographic information, past medical history and current medical therapies, review of most recent AAA ultrasound measurement, BP, blood for measurement of creatinine and electrolytes

2. baseline – review of informed consent, review of demographic information, review and checking of past medical history, review and checking of current medical therapies, ultrasound of AAA as per study protocol, BP readings in triplicate, review of screening blood test results, collection of urine and blood for the biomarker study (in a subset of patients), confirmation of patient eligibility, randomisation and dispensing of study medication

3. 3 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, collection of blood for measurement of creatinine and electrolytes, dispensing of study medication and pill count

4. 6 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, dispensing of study medication and pill count

5. 9 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, dispensing of study medication and pill count

6. 12 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, collection of blood for measurement of creatinine and electrolytes, dispensing of study medication and pill count

7. 15 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, dispensing of study medication and pill count

8. 18 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, dispensing of study medication and pill count

9. 21 months – review and check medical history, review and check current medical therapies, collect details of any AEs, ultrasound of AAA as per study protocol, BP readings in triplicate, dispensing of study medication and pill count

10. 24 months – review and check medical history, review and check current medical therapies, ultrasound of AAA as per study protocol, BP readings in triplicate, collection of urine and blood for biomarker study (in a subset of patients), collection of European Quality of Life-5 Dimensions (EQ-5D) questionnaire, health resource use questionnaire and pill count.

Blood pressure protocol

At each visit three BP recordings were taken in the sitting position using a validated semiautomated device after at least 10 minutes rest. The mean of the second and third readings was used in analyses. Smoking was not permitted in the 30 minutes before BP measurement. Omron 705CP-II machines (OMRON Healthcare, Milton Keynes, UK) were distributed for collection of BP measurements at all except five sites, where BP Plus devices (USCOM, Sydney, NSW, Australia) were used. The purchase of six BP Plus devices was funded by the Foundation for Circulatory Health, Imperial College London. These machines were distributed to the five sites (one machine was kept as a backup) where we expected the highest recruitment levels (St Mary’s Hospital, Royal Bournemouth Hospital, Hull Royal Infirmary, Norfolk and Norwich University Hospital and York Hospital). At these sites both peripheral and central BP recordings were collected from the baseline visit onwards for all patients. At ICCH, 20 patients were already participating in the trial prior to the BP Plus machine arriving; these patients had all of their BP measurements taken using an OMRON machine.

Clinical laboratory samples

Angiotensin-converting enzyme inhibitors commonly cause mild increases in serum creatinine as part of the desired result of reducing intraglomerular pressure. This slight rise in serum creatinine is to be expected and is acceptable after starting ACE-Is. 77 Blood tests for measurement of creatinine and electrolytes were therefore carried out at screening and 3, 12 and 24 months (in keeping with best recommended practice for the management of hypertension with ACE-Is) and results were reviewed regularly by the study team and the Data Safety Monitoring Committee (DSMC). If the serum creatinine level rose > 30% above baseline or progressively increased over time, investigators were advised to discontinue the study medication. Lesser increases in serum creatinine were monitored as required more frequent blood tests.

Quality of life

The EQ-5D is a standardised measure of health status developed by the EuroQol group78 to provide a simple, generic measure of health for clinical and economic appraisal. It is applicable to a wide range of health conditions and treatments, and provides a simple descriptive profile and a single index value for health status that can be used in the clinical and economic evaluation of health care.

The EQ-5D consists of two pages: a descriptive system and a visual analogue scale. The descriptive system has five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each has three levels of severity (no problems, some problems, extreme problems) in the three-level version and five levels of severity (no problems, slight problems, moderate problems, severe problems and extreme problems) in the five-level version. The three-level version was utilised for the AARDVARK trial (see Appendix 2).

The visual analogue scale used to value EQ-5D health states is presented as a 20-cm vertical line calibrated from zero (‘worst imaginable health state’) to 100 (‘best imaginable health state’). It asks respondents to ‘mark an ✗ on the scale to indicate how your health is today’.

This questionnaire was administered after 12 and 24 months of follow-up to allow quality of life to be compared among the three randomised groups during treatment as part of the safety analyses.

Health resource use questionnaire

The health resource use questionnaire (see Appendix 3) was created specifically for this study to evaluate the associated costs and burden of AAA patients to the NHS if there was a significant effect of ACE-Is on aneurysm growth.

The questionnaire collects information relating to patients’ health service use (e.g. visits to their doctor, nurse or GP) for reasons related to their aneurysm, use of social services, the amount of help they receive from their family and carers and service use for reasons not related to their aneurysm.

This questionnaire was administered after 12 and 24 months of follow-up.

Data collection

Data were collected by the study team onto paper case report forms (CRFs) (see Appendix 4) and were then entered onto corresponding electronic forms on the InForm^™ ITM (Integrated Trial Management) system (Oracle Corporation UK Ltd, Reading, UK). This is a web-based data entry system that builds an Oracle database for each individual clinical trial. Bespoke web-based electronic CRFs with built-in validation rules were designed to identify data entry errors in real time and provide a full audit trail of data entry and changes. All persons entering data were trained prior to start-up and given personal login details with access to forms restricted according to site and role. The electronic CRFs were designed in accordance with the requirements of the trial protocol and complied with regulatory requirements. An automated audit trail was recorded when (and by which user) records were created, updated or deleted. Error reports were generated when data clarification was required and data queries were resolved by research nurses at trial sites.

Trial reporting

The trial team was required to submit annual reports on trial progress, data completion rates and safety and protocol compliance to the Medicines and Healthcare products Regulatory Agency (MHRA) and the Research Ethics Committee (REC). Reports were also prepared for all trial-oversight committees.

In addition, monthly key figures and 6-monthly reports were prepared for the funding body (the NIHR HTA programme). Monitoring meetings to discuss trial progress were also held with the NIHR at its request.

Trial monitoring

The data collected for the study were monitored on a regular basis to ensure that the integrity of the data and the rights and well-being of participants were protected. Monitoring was completed at a central level and a site level to check for any data errors, deviations or protocol non-compliance.

Validation checks were built into the InForm system, which enabled validation reports to be generated. Any missing values, values out of range or spurious values within the data set were flagged. Queries were sent to sites and followed up for resolution prior to data lock.

All sites were visited prior to opening to recruitment, within 2 weeks of randomising their first patient and then as required to achieve the following:

• Source data verification of 100% of ICFs signed since previous visit to ensure correct completion.

• Drug accountability for 100% of patients.

• Verification that all serious adverse events (SAEs) had been reported correctly.

• Source data verification of at least 50% of subjects randomised (a list of random patient numbers was provided by the statistician in advance) should have been performed by the end of the study for the following:

  • Eligibility – the data on the screening worksheets should have been checked against the inclusion/exclusion checklist and the subjects’ medical notes; for scanning sites, this eligibility should also have been entered correctly on InForm.

  • Existence – verification that subjects’ name and date of birth on the study worksheets and ICFs match the details in the medical notes.

  • Aneurysm measurements – by comparison of data on InForm against the study worksheets to ensure that the measurements had been entered correctly.

  • EQ-5D – by comparison of data on InForm against the hard copy of the questionnaire completed by the patient.

  • Health resource use questionnaire – by comparison of data on InForm against the hard copy of the questionnaire completed by the patient.

  • AEs and SAEs – by comparison of data on InForm against worksheets and medical notes.

  • Blood pressure – by comparison of data on InForm against the data recorded on the study worksheets. For BP measurements recorded using an OMRON machine with a printer, a printout of the results should have been attached to the worksheets. If a BP Plus machine was used, the measurements entered on InForm should have been checked against the study worksheets and, if available, against the electronic BP Plus database.

Randomisation

Randomisation was performed using a 1 : 1 : 1 ratio stratified by centre and by baseline size of aneurysm stratified into two size ranges: 3.0–4.5 cm and 4.51–5.40 cm. The randomisation code was generated by an independent statistician using randomly permutated blocks of varying sizes using SAS computer software (SAS Institute, Cary, NC, USA).

Any subjects successfully screened for the study and found to be eligible to proceed in the study were then randomised via the InForm system by a trained member of the research team after appropriate consent. Patients were allocated a unique study number for use in all future data collection.

Investigational medicinal products

Perindopril, amlodipine and placebo were produced in accordance with good manufacturing practice79 and packaged and labelled by the Royal Free Hospital Pharmacy Manufacturing Unit. Overencapsulation of the tablets for this pilot study proved too costly; therefore, the amlodipine, perindopril and placebo tablets were not identical. The three investigational medicinal products (IMPs) were randomly allocated to either bottle A, bottle B or bottle C by a statistician and this information was provided to the Royal Free Manufacturing Unit to allow the IMPs to be deblistered and packed into the corresponding bottles.

Perindopril arginine (10 mg) was provided by Servier at no cost. All products were checked by a qualified person at the Royal Free Manufacturing Unit prior to release.

The Royal Free Manufacturing Unit held a manufacturer’s authorisation for IMPs (called the Manufacturing and Importation Authorisation at the time of this study). Blinded treatment kits were labelled as per European Commission Good Manufacturing Practice Guidelines, Annex 1379 to enable the treatment to be identified and the batch source of the materials traced.

The IMPs were supplied to the participating sites on a demand basis with minimal wastage of materials. Bottle accountability logs were maintained by all parties to allow full reconciliation of IMPs, including assignment to patients.

Although the study was classed as single blind because the tablets were not identical, the following measures were taken to avoid site staff becoming aware of treatment allocation:

• At no point during the study were the site staff informed of the contents of bottles A, B and C.

• Bottles A, B and C were the same colour and opaque so that the tablets could not be seen by participating staff.

• Sites were advised that returned tablets should be counted only by the pharmacy staff. This was to avoid any members of the research team identifying any of the tablets by their appearance.

Consequently, for all practical purposes the trial might reasonably be considered double blind.

Slow-release indapamide (1.5 mg) or amlodipine (5 mg) for use in the treatment of BP in those with a SBP > 150 mmHg following the initial screening visit was supplied by the site pharmacy or patients’ GPs in blister packs (not blinded). Losartan (100 mg) (an ARB) for use in patients who developed a cough during the trial was supplied by the site pharmacy or patients’ GP in blister packs (not blinded).

Reporting of protocol violations and deviations

Sites were requested to report all deviations from the study protocol. The site research staff were responsible for ensuring that procedures were undertaken and treatment given in accordance with the protocol. Any suspected protocol violation or deviation was reported on a protocol deviation form and also recorded on the InForm system. The AARDVARK management team was responsible for reviewing all protocol deviations and informing the sponsor as appropriate. Participants continued to participate in the trial except if they had been randomised in error or if they requested to be totally withdrawn from the study. Fully consented patients enrolled on the trial were followed up and analysed as per ITT analysis.

Subject confidentiality

All study staff were responsible for ensuring that participant confidentiality was maintained at all times. On the study worksheets or any other documents submitted to the sponsor, subjects were identified by a subject ID number and initials only.

The chief investigator and PIs were permitted direct access to subjects’ records and source documents only for the purposes of monitoring, auditing or inspection by the sponsor, authorised representatives of the sponsor, regulatory authorities and REC.

Retention of trial documents

This trial was coordinated by the Imperial Clinical Trials Unit (ICTU), which has well-established protocols for the protection of data and facilities for the retention of documents in place. Data will be stored for a minimum of 10 years (or according to changes in regulatory requirements) following completion of this trial. Data generated by this work will be processed in accordance with the Data Protection Act 1998. 80 The ICTU adheres to the Imperial College Code of Practice, drawn up in association with the College’s data protection policy, relating to the collection, holding and disclosure of data relating to individuals. The principal applicant and co-applicants act as custodians of the data and are responsible for its security. The chief investigator or delegate will ensure the continued storage of the documents, even if they leave the clinic/practice or retire before the end of the required storage period. Delegation will be documented in writing. The PI at each site is responsible for the archiving of all of the essential trial documents, including the Investigator Site File, in accordance with regulatory requirements.

The chief investigator and PIs were expected to retain a comprehensive and centralised filing system of all study-related documentation that was suitable for inspection by the sponsor and representatives of regulatory authorities.

Primary efficacy variables

The primary outcome measure was the growth of AAA external diameter (measured in the longitudinal plane) as per the results of the intra-/inter-variability studies (see Chapter 3).

In the absence of any convincing evidence regarding the best method of measuring AAAs, it was decided to inspect the within-trial assessments of measurement repeatability and base the primary outcome on the modality (internal or external) with the greatest within-trial repeatability for the AP diameter in the longitudinal plane.

Secondary efficacy variables

Secondary outcome measures included:

• time taken for the aneurysm to reach the threshold for intervention (diameter of 5.5 cm)

• aneurysm rupture

• aneurysm repair/referral for repair

• repeatability of internal and external aneurysm diameters

• quality of life (EQ-5D) and a health resource use questionnaire (12 and 24 months)

• intolerance of ACE-Is

• drug compliance

• reduction in BP.

Safety variables

Safety was assessed during the trial by:

• assessment of AEs and SAEs

• monitoring changes in serum creatinine levels.

Statistical methods

Research governance and management

This IMP trial was conducted in accordance with Medical Research Council (MRC) Guidelines for Good Clinical Practice81 and the Medicines for Human Use (Clinical Trials) Regulations 2004. 82 Imperial College London acted as the trial sponsor. A clinical trial authorisation was applied for and received from the MHRA. A site agreement between Imperial College London and the participating sites outlined the responsibilities of all parties and was signed prior to commencement of recruitment at the participating sites.

Three committees were established to govern the conduct of the trial: the TSC, the independent DSMC and the TMG. These committees functioned in accordance with ICTU standard operating procedures.

Patient and public involvement

Several patients with small AAAs were consulted to seek their opinions about the design and running of the trial. All of these patients reported that they would feel reassured by the increased surveillance of their AAA. In addition, a close relative of one patient approached emphasised the painful and stressful nature of the surgery that his relative had undergone and how he wished that there had been an alternative to that surgery. Therefore, anything that could be done to slow the growth of the AAA, to prevent or delay the need for distressing major surgery, was seen as positive.

A patient representative was present on the TMG and was involved in the design of the trial as well as review of protocol amendments and changes to the patient literature.

Protocol changes

All changes to the trial protocol and study conduct were reviewed by the sponsor and submitted to the REC and MHRA for approval as appropriate.

A summary of the amendments made prior to participant recruitment, during recruitment and after recruitment are provided in Tables 4–6, respectively.

Obtaining abdominal aortic aneurysm measurements

Accurate initial and repeat measurements of AAAs are essential to correctly ascertain growth rates in a trial setting and also to direct clinical management. Ultrasonography is a widely used, non-invasive and effective method for screening and obtaining measurements of AAA diameter. Static images are produced by freezing the image obtained using ultrasound and then accurate measurements can be taken using callipers.

Ultrasonography has been found to be accurate when used to assess AAA diameters by trained observers when compared with CT, and 95% of the differences between the two measurements can be expected to be < 3.5 mm. 83 Similar accuracy was obtained for trained screeners in the UKSAT22 and other aneurysm screening19 and observation84 studies.

In a clinical setting, ultrasound studies of AAAs are most often undertaken by vascular scientists and sonographers who have been trained extensively to undertake such studies. However, staff with no or limited previous experience of ultrasonography can be trained rapidly to identify the presence of AAAs reliably. 85–87 This model of training is used extensively in the NAAASP.

The visualisation of the abdominal aorta may be dependent on the sonographer’s experience and the extent of the patient’s bowel gas and body mass index. However, in a study by Hoffman et al. ,88 despite the level of difficulty reported, the accuracy of aortic measurements by the novices was independent of obesity and central adiposity.

Other common difficulties when scanning include ‘eccentric’ aneurysms and tortuous aortas, as well as the incorrect identification of other structures that appear tubular in cross-section (e.g. the inferior vena cava, the superior mesenteric artery and the gall bladder). The accurate selection of the boundaries of the aorta is also clearly important. 89 It is important that these factors are addressed and that quality improvement strategies are appropriately implemented in trials using ultrasound measurements.

When using ultrasound it is generally accepted that AP abdominal aortic diameters are recorded as standard because of their superior repeatability compared with transverse diameters. 90 However, the most reliable method of measuring AAA diameter remains debatable. The two main methods are measuring the internal diameters [i.e. inner anterior wall to the inner posterior wall (inner to inner or ITI) or intima to intima] or measuring the external diameters [i.e. from the outer anterior to the outer posterior wall (OTO) or adventitia to adventitia] (Figure 2). Some studies also use leading edge to leading edge measurements.

FIGURE 2.

An image of an AAA in the transverse plane showing internal and external AP measurements. The crosses (×) indicate the position of the inner anterior and inner posterior wall and the pluses (+) indicate the position of the outer anterior and outer posterior wall. The asterisk indicates the inner border of an area of thrombus on the posterior wall and it is important that the posterior inner wall calliper is not placed on the inner border of the thrombus.

Screening technicians employed by NAAASP are trained to correctly identify and take ITI measurements of AAAs by undertaking a 3-month accredited training programme. For trained screening technicians, the ITI method has been found to have better reliability and reproducibility than OTO measurement. 91 Conversely, when comparing the reliability and reproducibility of measurements taken by vascular scientists, it has been suggested that the OTO method is superior, with the ITI method underestimating the aortic size92 and having greater variability because of difficulty identifying the internal wall because of thrombus93 (see Figure 2), that is, aneurysm growth rates measured using internal diameters have greater noise or scatter than growth rates measured using external diameters. 42

The purpose of the trial quality assurance (QA) process was to:

1. ensure that there was adequate completeness of data collection by all sites

2. ensure the use of a standard protocol in the ultrasound scans carried out across 11 scanning sites by different sonographers on different ultrasound scanning systems

3. ensure the reliability of the results in terms of inter- and intra-observer variability

4. evaluate the reliability of AP AAA measurements taken using internal and external methods in both transverse and longitudinal planes to determine which provided the most repeatable measurements.

Abdominal aortic aneurysm measurement terminology

To ensure accurate comparison with both past and present research it is important to define the terminology used in this report.

Planes

In this study, AP measurements of the AAAs were taken on a transverse plane (referred to as transverse measurements within this report) and on a longitudinal plane (referred to as longitudinal measurements within this report). Figure 3 shows the transverse and longitudinal body planes.

FIGURE 3.

Transverse and longitudinal body planes.

Direction of the ultrasound probe

Figure 4 shows the direction of the probe required to obtain an AP diameter in a longitudinal plane. An example of an image that would be obtained in this view in shown in Figure 5.

FIGURE 4.

Direction of probe required for an image in the longitudinal plane. Reproduced with permission from Fpnotebook.com.

FIGURE 5.

Example of an AP diameter taken in the longitudinal plane.

Figure 6 shows the direction of the probe required to obtain an AP diameter in a transverse plane. An example of an image that would be obtained in this view is shown in Figure 7.

FIGURE 6.

Direction of probe required for an image in the transverse plane. Reproduced with permission from Fpnotebook.com.

FIGURE 7.

Example of an AP diameter taken in the transverse plane.

An example of a lateral wall to lateral wall diameter taken in a transverse plane is shown in Figure 8. Lateral wall to lateral wall diameters were not collected in this trial. This is because tortuosity of the aorta is common in aneurysmal disease as the aorta grows in length as well as width and these diameters can be considerably inaccurate.

FIGURE 8.

Example of an AAA image in the transverse plane with the measurements taken from lateral wall to lateral wall.

Ultrasound scanning protocol

At the time of this study there still remained debate over the best method of obtaining AAA measurements, as highlighted earlier. Therefore, four measurements in total – both maximum AP internal and external measurements in the transverse and longitudinal planes – were recorded on all patients, to enable a reliable comparison and to inform future practice.

The study was initially designed to have just two dedicated scanning staff between five sites (four in London and one in Coventry) to enhance the accuracy and repeatability of measurements. After 6 months it became clear that fewer patients than anticipated were eligible for the trial at these sites and therefore we were required to expand the study to recruit at 14 sites to meet the recruitment target, with scanning taking place at 11 sites (patients from all four London recruiting sites were scanned at ICCH, St Mary’s Hospital; see Figure 1).

This required more rigorous scanning and quality control protocols to be implemented to ensure maximum accuracy in the measurement of AAAs at all sites. A scanning protocol was provided to all participating sites to ensure the highest possible standard and accuracy in the ultrasound scans carried out across the 11 scanning sites. The final version of the scanning protocol is provided in Box 1.

BOX 1 - Ultrasound scanning protocol

• At each site it is requested that a single fully trained sonographer, vascular scientist or ultrasound screening technician performs all scans when possible.

• When alternative arrangements for the sonographers are made at some sites (e.g. vacation, illness), sonographers should follow identical protocols and should ensure that there is internal consistency in recorded measurements from the site.

• The same ultrasound system and probes should be used for each subject on each occasion.

• The patient should be brought into the room and allowed to rest on the couch in the semi-recumbent position for a period of 5 minutes for acclimatisation (while details are entered onto the duplex system).

• Once scanning is to be performed, the patient should be laid flat on the examination couch with abdomen exposed from xiphisternum to pubic symphysis.

• The room should be appropriately lit to facilitate accurate scanning in a standard fashion (i.e. similar on all occasions).

• Review previous image to determine the point from which to begin measurement.

• Perform an initial scan, checking for compressibility, pulsation and anterior branches, to enable the aorta to be correctly identified.

• The aorta should then be scanned in both longitudinal and transverse planes from the renal arteries to the bifurcation of the iliac arteries.

• The point of maximum dilatation in both transverse and longitudinal planes should be identified.

• Image optimisation should be achieved using gain/depth etc. control to produce an image with the clearest wall appearance.

• All images must be taken with the highest quality and accuracy.

• Following maximum point identification, the longitudinal image should be frozen on screen.

• Four images in total must be saved.

Longitudinal measurement

• Measurements are now taken using the calipers from the anterior side of the aorta to the posterior side. The measurements taken should be at 90° to the wall.

• An ITI measurement should be made by placing the calipers on the inside of the aortic wall, taking care not to misidentify plaque or thrombus as in the inner wall.

• This image should be saved after labelling the scan image with LS (for longitudinal view), ITI and the time of the scan within the study (baseline, 3 months, 6 months, etc.) and adding the sonographer’s initials.

• An OTO measurement should be made by placing the calipers on the outer walls, taking care not to misidentify the surface of the spine or other structures as the posterior wall.

• This image should be saved after labelling the scan image with LS, OTO and the time of the scan within the study (baseline, 3 months, 6 months, etc.) and adding the sonographer’s initials.

Transverse measurements

• The ultrasound probe should be pivoted 90° at the same point (of maximum size) on the aorta. This image in the transverse plane should be frozen.

• An ITI measurement should be made by placing the calipers on the inside of the aortic wall, taking care not to misidentify plaque or thrombus as in the inner wall.

• This image should be saved after labelling the scan image with TS (for transverse view), ITI and the time of the scan within the study (baseline, 3 months, 6 months, etc.) and adding the sonographer’s initials.

• Check the results that have been obtained. If the ITI and OTO measurements in TS and LS are not similar then the patient should be rescanned and checked – if there is a reason for disagreement such as tortuosity of the vessel, make a comment on the record.

• Problematic measurements: if you are experiencing difficulty defining and measuring the aorta and its aneurysm, please consult a more senior sonographer for assistance and make a note on the ‘Relevant comments’ section of the scanning log.

Storage of images

Images, with measurements and labelled as described in the scanning protocol, will be stored on the hard drive of the scanner and subsequently downloaded onto the Picture Archiving and Communication System or other storage devices at the co-ordinating centre. All sites will be required to download all images onto a CD to be sent to the co-ordinating centre for QA purposes.

Longitudinal anteroposterior inner-to-inner measurements

Longitudinal anteroposterior outer-to-outer measurements

Transverse anteroposterior inner-to-inner measurements

Transverse anteroposterior outer-to-outer measurements

Performance in data collection

Recording of the four AAA measurements was required for each patient at each visit. A key performance indicator for sites was that of completeness of data recording for all four required measurements. The completeness of data collection at each visit was between 94% and 100%. Table 7 describes the percentages of patients who attended who had all four required measurements recorded at the end of the study visit.

Quality assurance events

Inter- and intraobserver QA scanning events were organised to ensure consistency between sonographers, screening technicians and others responsible for aortic measurements and between measurements by the same observers. Specific aims of the QA events were:

1. to ensure the reliability of the results in terms of inter- and intra-observer variability.

2. to evaluate which was the most accurate and repeatable of AP AAA measurements taken using ITI and OTO methods in both transverse and longitudinal planes.

Three QA days were arranged, one each in York, Hull and London. A total of 19 observers scanned patients over the course of the trial (Table 8). Of these, 12 (63%) attended the QA days. At least one observer from each of the 11 scanning sites attended one of the QA events. The numbers of observers attending and scanning centre locations are documented in Table 8. Six volunteers were scanned on 8 July 2013, five volunteers were scanned on 13 September 2013 and five volunteers were scanned on 8 November 2013.

At each event the following took place:

• review of the scanning requirements as documented in the protocol

• discussion of example problem cases as a teaching prompt

• highlighting of known improvement strategies to reduce variability as needed

• scanning of volunteer patients for inter- and intraobserver variability studies.

Patients with an AAA who were part of the trial were invited to attend the QA days. Each was informed that they were to have their AAA measured on at least two occasions by each sonographer who was able to attend that QA session.

The scanning protocol was reviewed by all of the clinicians. Teaching on problem cases and strategies for best practice were presented and the structure for the day outlined in full.

All of the sonographers who had scanned trial patients were observed scanning at least one subject following the scanning protocol until the trial sonographer was convinced that appropriate protocols were being followed and adequate images were being obtained consistently. Each of the sonographers from the sites then scanned each of the AAA patients using the scanning protocol. All sonographers were blinded to the clinical details, size and anatomical configurations of the aneurysms of patients as far as possible. Volunteers were not named but were allocated an identifier for anonymity in an attempt to minimise bias. All four required measurements were recorded and handed to the trial manager. All patients were measured on at least two occasions by each sonographer. The trial clinical vascular scientist also performed a set of measurements for each patient.

The mean intraobserver variability is documented in Table 9. The mean intraobserver variability repeatability coefficient is documented for each of the four measurements obtained by observers in the study along with the SD and range. The repeatability coefficient is the value below which the absolute difference between repeated test results is expected to lie with a probability of 95%.

Clearly, measurements in the longitudinal plane were more repeatable than measurements in the transverse plane. The differences in the mean (range) ITI and OTO intrasonographer repeatability coefficients for longitudinal measurements were similar at 0.32 (0.24–0.39) cm for ITI measurements and 0.33 (0.21–0.45) cm for OTO measurements. This was compared with 0.48 (0.23–1.07) cm and 0.50 (0.20–1.24) cm for transverse ITI and OTO measurements, respectively.

As most sites used more than one observer to follow up the enrolled trial patients, interobserver variability was of great importance.

The differences between the AAA measurements taken by the sonographers and those taken by the senior clinical vascular scientist in the same patients were used to calculate the intersonographer variability by using Bland–Altman 95% limits of agreement (Table 10). The limits of agreement define the range within which 95% of the differences between two measurements made by two observers are likely to fall. Figure 9 shows the variability in the measurements of external longitudinal diameter compared with the measurements of the senior clinical vascular scientist (expert). Thirteen pairs disagreed by > ±4 mm for ITI measurements, whereas seven pairs disagreed by > ±4 mm for OTO measurements. These data suggest that the interobserver repeatability is better for longitudinal OTO measurements than for longitudinal ITI measurements.

FIGURE 9.

Interobserver repeatability graphs (longitudinal diameters). (a) Longitudinal internal diameter; and (b) longitudinal external diameter. Each symbol on the graphs represents a different AARDVARK scanner (three from the July event, three from the November event and six from the September event). Five or more patients are shown for each event, so 10 vertical scatters have only three points and five vertical scatters have six points. Each vertical scatter is a separate patient.

Quality control of ultrasound scans

The quality of the ultrasound images obtained in trial was assessed centrally to ensure that a reliable standard of ultrasound scans was obtained across the 11 scanning sites by different sonographers on different ultrasound scanning systems. The following protocol was used:

• The baseline images for all of the patients in the trial were reviewed as soon as possible after the first monitoring visit.

• All patients at a site or a minimum of 10 patients at a site (for sites with > 10 patients) had each of their 3-monthly images reviewed. Review was undertaken by the senior clinical vascular scientist. Up to 10 random patients were selected by the trial statistician for quality control analysis. If one of the selected patients reached an end point or withdrew from the study or did not attend the visit, the number of images checked might be less than baseline in certain cases. In addition, we asked for additional images to be checked if poor image quality was found at sites. Hence, > 10 images were checked at certain visits. The numbers of recorded images reviewed from each site by the senior clinical vascular scientist are provided in Table 11.

• All sites were informed of the scans that needed to be forwarded to the trial centre.

• Each site was asked to save images from longitudinal and transverse planes at the point of maximum dilation to include ITI and OTO calliper placement records.

• The senior clinical vascular scientist reviewed all images sent to the trial centre. All of the images had patient-identifiable information removed and the senior clinical vascular scientist was blinded to treatment allocations.

• Each image was assessed on clarity (looking at depth, gain, focus and good wall definition) and calliper placement (ITI and OTO). The following criteria were used to judge an image as acceptable:

  • the image was of the highest image quality possible

  • the callipers were placed correctly for both ITI and OTO measurements in relation to the wall

  • measurements were taken at right angles to the aneurysm sac wall

  • the image was labelled correctly

  • the image was judged to be taken in a true transverse and true longitudinal plane.

Each image that was reviewed was classified as acceptable or unacceptable.

This protocol was based on the standard operating procedures for the NAAASP [see www.gov.uk/government/collections/aaa-screening-supporting-documents (accessed 3 June 2016)], with the addition of the following assessment criteria: measurements to be taken at right angles to the aneurysmal sac and judged to be in the transverse and longitudinal planes.

If an image was deemed unacceptable, the site was contacted and asked to review the image to ensure that the correct image had been sent, the callipers had been placed correctly and the AARDVARK scanning protocol had been followed. If required, the senior clinical vascular scientist responsible for quality control and the relevant sonographer liaised directly until a resolution was reached.

Sites/observers with unacceptable images were contacted. When a single image taken by an observer was unacceptable, the observer was contacted to review the image. If more than two images were unacceptable then the site was visited and all observers underwent further dedicated training sessions. In these training sessions the site was visited by the lead sonographer for the study and (1) the scanning protocol was reviewed and relevant teaching given and (2) each of the images that was deemed unacceptable was reviewed. This was required at only one site.

Quality control of the final ultrasound data

In addition to quality control of the ultrasound images and the QA events, the ultrasound measurement database as a whole was monitored for any anomalies or spurious results. Sites were requested to measure the maximum AP diameter of the AAA at each visit. Therefore, the ITI transverse and longitudinal measurements and the OTO transverse and longitudinal measurements at each visit were expected to be within ±3 mm of each other.

If the longitudinal and transverse measurements varied by > ±3 mm, sites were contacted and asked to recheck their scans and measurements to ensure that the maximum AP diameter had been accurately measured. Cases were accepted or rejected accordingly and remeasured when an obvious error had been made.

In total, 107 ITI measurements and 35 OTO measurements had differences > ±3 mm. This, together with the results of the interobserver variability study and the use of multiple observers for the trial, suggested that the longitudinal OTO measurement was the most reliable measurement on which to base aneurysm growth rate. Therefore, longitudinal OTO measurements were used for the evaluation of the trial end points.

The 35 OTO measurements were reconsidered as this measurement was the primary measurement to be used in the main trial analyses. Four measurements were amended after obvious measurement errors were made. Adjusted measurements were used for the analyses in the trial.

Justification for the choice of diameter used for primary end point

The mean (range) intrasonographer repeatability coefficient for longitudinal measurements was 0.33 (0.21–0.45) cm for OTO measurements and 0.32 (0.24–0.39) cm for ITI measurements compared with 0.48 (0.23–1.07) cm and 0.50 (0.20–1.24) cm for transverse ITI and OTO measurements, respectively. Longitudinal plane measurements were therefore used for the primary analyses.

However, possibly reflecting the small number of patients included (n = 11), there was no meaningful difference between the variability of OTO and ITI measurements reported in intraobserver variability studies that could determine the superiority of OTO or ITI measurements for use as the primary end point in the study. Considering ITI interobserver variability measurements, 13 pairs of measurements (by an observer and the senior clinical vascular scientist) differed by > ±4 mm, whereas for OTO measurements seven pairs of measurements differed by > ±4 mm. These data suggest that interobserver repeatability was better for longitudinal OTO measurements than for longitudinal ITI measurements

When comparing transverse and longitudinal maximum diameters in the same patient at each time point there was less variation between OTO measurements than between ITI measurements (35 vs. 107 measurements were different by > 0.3 mm, respectively). In addition, there was less variation in the SDs of AAA measurements for most time points in the study of OTO measurements (see Table 19).

Therefore, OTO diameters measured from longitudinal images were used for evaluation of the primary outcome measure.

Site recruitment

Recruitment was originally to take place at five sites: Imperial College Healthcare NHS Trust (St Mary’s and Charing Cross Hospitals), Royal Free London NHS Foundation Trust, Guy’s and St Thomas’ NHS Foundation Trust and University Hospitals Coventry and Warwickshire NHS Trust. We identified a pool of approximately 860 patients potentially eligible for the trial from existing databases from these four hospital trusts (200, 90, 380 and 189, respectively). We also predicted that expansion of the NAAASP was likely to increase this patient pool before trial recruitment started, with approximately 4% of men screened expected to have an AAA. Against this, we estimated that 30% of patients would be ineligible for the trial because they were already receiving ACE-I therapy or because they had an unsuitable BP, that is, SBP could not be controlled to < 150 mmHg before randomisation. Assuming a 15% refusal rate in the remaining 262 patients, the target of 225 patients seemed achievable given a steady influx of newly screened and eligible patients over the recruitment period.

The first trial site was opened to recruitment in September 2011, with recruitment due to end after 12 months. However, to meet the recruitment target a further nine sites were added during the trial. The four sites in London became recruitment and screening sites, with patients having all other study visits at the ICCH, Imperial College London. The remaining 10 scanning sites were located across the country and performed all study visits locally.

Participant recruitment

A recruitment extension of 7 months was required before the recruitment target was successfully met. The reasons for this were as follows:

• The number of patients taking ACE-Is was higher than anticipated and this was the main barrier to recruitment. At the time of the HTA programme commissioned call in 2009, approximately 32.3% of the population aged > 60 years received ACE-Is; by 2012 this had increased to 52.9% (Health Survey for England). 94

• The proportion of patients with an AAA picked up through the NAAASP was less than half the proportion expected: 4% of patients screened were expected to have an AAA, whereas in reality this proportion was 1.2%.

• Progress was slow in securing NHS research governance approval at some of the participating sites. The quickest research and development approval was granted in 33 working days (approximately 1.5 months), whereas the longest approval took 147 working days (6 months). After waiting 168 working days (7 months) with approval still not granted in one further site, attempts to gain approval were abandoned.

To try and overcome some of the barriers to recruitment, the following strategies and actions were implemented approximately 6 months into the study:

• Review of methods of approaching patients – sites were requested to review their methods of contacting patients and try different strategies (i.e. speaking to patients at their next visit instead of cold-calling) in an attempt to increase uptake into the trial.

• Breakdowns of site recruitment were requested monthly and were reviewed at the monthly management meetings to try and identify and address any specific issues.

• Addition of PICs – some centres had PIC sites set up to widen the pool of potential participants.

• The age range was expanded to include those aged ≥ 55 years (originally set at ≥ 60 years).

• Patient invitation letter – this was created because some site investigators felt that the PIS was too lengthy and daunting as a first introduction to the trial.

• Trial poster – this was created and put on the noticeboards in relevant clinical areas (e.g. in the vascular ultrasound laboratories) at some sites to create awareness of the trial.

• AARDVARK trial web page – this was set up via the ICTU website and was available as a resource for both patients and site staff. The site contained general information about the trial and was updated regularly with current recruitment numbers.

• Update/simplify the PIS – some sites felt that the original PIS was written in such a way that the side effects and risks of the study were unduly worrying to potential participants. Therefore, the PIS was reworded and a list of side effects tabulated as an appendix.

• Additional research sites – by far the most successful strategy for increasing recruitment was increasing the number of research sites, specifically targeting those with large populations and time/resources to conduct the study.

• Chief investigator presence at site initiation visits – the chief investigator visited all sites as part of the site initiation visits to stress the importance of the trial and answer any clinical queries.

In total, 227 patients were initially randomised to the trial. However, the randomisation of three patients was subsequently discovered to be a protocol violation because these patients did not meet all of the entry criteria for the trial. Table 12 shows the final recruitment numbers by site. The Royal Bournemouth Hospital and Hull Royal Infirmary were the top recruiting sites, with 50 and 40 patients randomised, respectively. The first patient was randomised to the study on 16 December 2011 and the last patient was randomised to the study on 19 April 2013 (Figure 10).

FIGURE 10.

Participant recruitment by month.

Table 13 shows that the number of patients randomised to each treatment arm was well balanced, with 79 patients randomised to placebo, 73 randomised to perindopril and 72 randomised to amlodipine.

Non-recruited patients

During the trial recruitment period we requested that the sites capture details of reasons why patients with an AAA were not recruited into the trial. These data were collected by prospective pre-screening by recruiting staff, mainly by case note and clinical database review. Unfortunately, limited research nurse availability and clinical service pressures prevented reliable collection of these data at some centres. However, data were collected from a total of 1912 non-recruited patients. The patients were excluded on the basis of ineligibility (Figure 11) or because they declined to participate (Figure 12). The main reason for ineligibility was already taking ACE-Is; this excluded 643 patients from participating in the trial. Unfortunately, a specific reason for declining to participate was not recorded for 107 patients; however, 56 patients declined because they did not want to participate in a clinical trial.

FIGURE 11.

Reasons for trial ineligibility, n (%) (N = 1595).

FIGURE 12.

Reasons for declining the trial, n (%) (N = 317).

Status of patients in the study

Table 14 shows the numbers of patients who completed each study visit across the three groups. In total, 161 patients (71%) successfully completed all study visits.

As previously described, a 10% attrition rate was included in the power calculations for the trial. Participants were included in the attrition rate if data from fewer than two study visits were available for analysis. A total of 13 patients counted towards the attrition rate: three patients who were randomised in error and 10 patients who withdrew from the study before completing at least two study visits. The patients randomised in error were not included in the final analyses; however, the remaining 10 patients were included on an ITT basis. The final attrition rate for the study was therefore 6%. The CONSORT diagram (Figure 13) shows the status of patients at the end of the trial. Six patients withdrew from the trial because of AEs attributed to the study medications (n = 2 perindopril, n = 4 amlodipine). Four patients (n = 3 perindopril, n = 1 amlodipine) switched to losartan because of cough. No patients suffered AAA rupture and 26 underwent elective surgery (n = 9 placebo, n = 10 perindopril, n = 7 amlodipine) during the trial period.

FIGURE 13.

The AARDVARK trial CONSORT diagram. Reproduced from Bicknell et al. 2016. 95 © Bicknell et al. 2016. Published by Oxford University Press on behalf of the European Society of Cardiology. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.

Baseline comparability of randomised groups

The baseline characteristics for all patients in the study are shown in Table 15. Although groups were randomised and generally well matched, there were some small imbalances between the groups that may influence aneurysm growth, for example presence of diabetes [n = 8 (10.1%) placebo, n = 2 (2.7%) perindopril, n = 6 (8.3%) amlodipine] and use of statins [n = 48 (61%) placebo, n = 53 (73%) perindopril, n = 45 (63%) amlodipine]. All patients but one were Caucasian.

Abdominal aortic aneurysm growth

The changes in AAA diameter over the duration of the trial are shown in Table 16 (summary longitudinal AAA measurements) and Table 17 (differences in longitudinal AAA measurements). See Appendix 5 for a summary of the transverse AAA measurement data.

There was a small increase in the mean internal and external AAA diameter across all three treatment groups. The mean difference between month 24 and baseline for the longitudinal external diameter was 0.27 cm for the three groups combined. Similar results were observed for the mean external transverse measurements. Comparisons of 2-year changes across the three randomised groups are invalid because of the variable reduction in numbers in each group over time. The SDs of the diameters recorded in the longitudinal plane for the ITI measures (see Table 16) were generally systematically higher than those for the OTO measures at each time point, as were the SDs of the differences between baseline and each time point (see Table 17).

The AAA growth trajectories in each patient by group are shown in Figure 14 (placebo), Figure 15 (perindopril) and Figure 16 (amlodipine). Most of the patients had slow growth rates but there were a small number of patients with faster growth rates in each randomised group.

FIGURE 14.

Abdominal aortic aneurysm growth trajectories by patients in the placebo group.

FIGURE 15.

Abdominal aortic aneurysm growth trajectories by patients in the perindopril group.

FIGURE 16.

Abdominal aortic aneurysm growth trajectories by patients in the amlodipine group.

Figure 17 shows the observed mean AAA diameter for each treatment group at each visit time point. The numbers of patients were different across each visit, with lower numbers of patients in the later visits. There were no significant differences between the groups over time and any apparent differences between the groups must be considered in the context of the changing numbers of participants. This is especially relevant when patients with an AAA of ≥ 5.5 cm who are taken for surgery and do not complete any further measurements are considered.

FIGURE 17.

Mean longitudinal external diameter over time by randomised group.

Histograms of change in longitudinal external AAA diameter from baseline to 3, 12 and 24 months are shown in Figures 18–20, respectively (for the histograms at 6 and 18 months see Appendix 6). These histograms show the distribution of the difference in the data between the two specified time points. Overall, no obvious differences were apparent across groups at any of the three time points, whereas in all three histograms growth was apparent overall at 12 and 24 months reflected by the shift of bars to the right of the graphs.

FIGURE 18.

Histograms of change in AAA longitudinal external diameter from baseline to month 3. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

FIGURE 19.

Histograms of change in AAA longitudinal external diameter from baseline to month 12. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

FIGURE 20.

Histograms of change in AAA longitudinal external diameter from baseline to month 24. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

Counting those AAAs that exhibit fast growth rates (as defined by growth of > 0.5 cm per year) in these histograms (n = 14 placebo, n = 8 perindopril, n = 14 amlodipine) suggested possible differential effects among the three treatment groups. Formal analyses of the number of patients who exhibit fast growth rates in each of the three groups (taking into account aneurysm size and adjusting for confounders) was not undertaken as part of the main analyses of this trial but will be considered as part of future research plans.

Box plots of change in longitudinal external AAA diameter from baseline to 3, 12 and 24 months are shown in Figures 21–23, respectively (the box plots for change from baseline to 6 and 18 months can be found in Appendix 7). These figures confirm the findings shown in Figures 18–20 in that there are no apparent differences in change in diameter at 3, 12 or 24 months across the three groups and growth of a similar magnitude is apparent in all three groups at 12 and 24 months.

FIGURE 21.

Box plot of change in AAA longitudinal external diameter from baseline to month 3. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

FIGURE 22.

Box plot of change in AAA longitudinal external diameter from baseline to month 12. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

FIGURE 23.

Box plot of change in AAA longitudinal external diameter from baseline to month 24. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

Histograms and box plots of change in diameter for longitudinal internal diameters and transverse external and internal diameters are provided in Appendices 8–13.

Primary outcome

Multilevel modelling was used to determine the maximum likelihood estimates for AAA diameter growth. Table 18 provides these results for the longitudinal external AAA diameter. The estimates for annual diameter growth were as follows:

• 1.68 mm [standard error (SE) 0.02 mm] in the placebo group

• 1.77 mm (SE 0.02 mm) in the perindopril group

• 1.81 mm (SE 0.02 mm) in the amlodipine group.

The differences in the slopes of modelled growth over time were not significant between perindopril and placebo (p = 0.78) or between perindopril and amlodipine (p = 0.89). The difference in slope between perindopril and placebo plus amlodipine combined was also not significant (p = 0.92).

Blood pressure reduction

Summary BP statistics for the duration of the trial are shown in Table 21 and BP differences from baseline are shown in Table 22. There was an increase in mean SBP in the placebo group from baseline to month 24 (+2.5 mmHg) and a decrease in in mean SBP in both the perindopril (–5.0 mmHg) and amlodipine (–2.8 mmHg) groups from baseline to month 24. There was little change in mean DBP between baseline and month 24 in the placebo group (–0.7 mmHg) but reductions were seen in the perindopril and amlodipine groups (–5.2 mmHg and –3.3 mmHg, respectively). Mean SBP and DBP over the duration of the trial are shown in Figures 24 and 25, respectively.

FIGURE 24.

Mean SBP over the duration of the trial by randomised group. (a) Placebo; (b) perindopril; and (c) amlodipine. Vertical bars represent the 95% CIs for the means. Reproduced from Bicknell et al. 2016. 95 © Bicknell et al. 2016. Published by Oxford University Press on behalf of the European Society of Cardiology. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.

FIGURE 25.

Mean DBP over the duration of the trial by randomised group. (a) Placebo; (b) perindopril; and (c) amlodipine. Vertical bars represent the 95% CIs for the means. Reproduced from Bicknell et al. 2016. 95 © Bicknell et al. 2016. Published by Oxford University Press on behalf of the European Society of Cardiology. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.

Figures 26–28 show the histograms of change in SBP from baseline to 3, 12 and 24 months, respectively, by randomised group and combined. These histograms show the distribution of the difference in the data between the two specified time points. Negative values indicate a decrease in BP and positive values indicate an increase in BP. The histograms for the placebo group generally follow a normal distribution around zero, whereas reductions in SBP in the perindopril and amlodipine groups are shown by the increased density of bars on the left-hand side of each histogram, particularly at 12 and 24 months. For histograms of the change in SBP from baseline to 6 and 18 months see Appendix 14.

FIGURE 26.

Histograms of change in SBP from baseline to month 3. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

FIGURE 27.

Histograms of change in SBP from baseline to month 12. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

FIGURE 28.

Histograms of change in SBP from baseline to month 24. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

Box plots of change in SBP from baseline to 3, 12 and 24 months are shown in Figures 29–31, respectively. These show that there was little change in SBP in the placebo group but a reduction in SBP in the perindopril and amlodipine groups. For box plots of change in SBP from baseline to 6 and 18 months see Appendix 15.

FIGURE 29.

Box plot of change in SBP from baseline to month 3. p-value from regression of SBP at 3 months on treatment adjusted for SBP at baseline. a, perindopril vs. placebo; b, amlodipine vs. placebo; c, amlodipine vs. perindopril. Circle is outside value. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

FIGURE 30.

Box plot of change in SBP from baseline to month 12. p-value from regression of SBP at 3 months on treatment adjusted for SBP at baseline. a, perindopril vs. placebo; b, amlodipine vs. placebo; c, amlodipine vs. perindopril. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

FIGURE 31.

Box plot of change in SBP from baseline to month 24. p-value from regression of SBP at 3 months on treatment adjusted for SBP at baseline. a, perindopril vs. placebo; b, amlodipine vs. placebo; c, amlodipine vs. perindopril. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

Figures 32–34 show the histograms of change in DBP from baseline to 3, 12 and 24 months, respectively. In line with the changes seen in SBP, the histograms for the placebo group generally follow a normal distribution, whereas the reductions in DBP in the perindopril and amlodipine group are shown by the increased density of bars to the left-hand side of each histogram. For histograms of change in DBP from baseline to 6 and 18 months see Appendix 16.

FIGURE 32.

Histograms of change in DBP from baseline to month 3. (a) Placebo; (b) perindopril; (c) amlodipine; and and (d) total.

FIGURE 33.

Histograms of change in DBP from baseline to month 12. (a) Placebo; (b) perindopril; (c) amlodipine; (d) total.

FIGURE 34.

Histograms of change in DBP from baseline to month 24. (a) Placebo; (b) perindopril; (c) amlodipine; and (d) total.

Box plots of change in DBP from baseline to 3, 12 and 24 months are shown in Figures 35–37, respectively. These show that there was little change in DBP in the placebo group but a reduction in DBP in both the perindopril group and the amlodipine group. For the box plots of change in DBP from baseline to 6 and 18 months see Appendix 17.

FIGURE 35.

Box plot of change in DBP from baseline to month 3. p-value from regression of DBP at 3 months on treatment adjusted for DBP at baseline. a, perindopril vs. placebo; b, amlodipine vs. placebo; c, amlodipine vs. perindopril. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

FIGURE 36.

Box plot of change in DBP from baseline to month 12. p-value from regression of DBP at 3 months on treatment adjusted for DBP at baseline. a, perindopril vs. placebo; b, amlodipine vs. placebo; c, amlodipine vs. perindopril. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

FIGURE 37.

Box plot of change in DBP from baseline to month 24. p-value from regression of DBP at 3 months on treatment adjusted for DBP at baseline. a, perindopril vs. placebo; b, amlodipine vs. placebo; c, amlodipine vs. perindopril. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

Secondary outcomes

Time to reach an AAA diameter of 5.5 cm, being referred to or having AAA surgery, or AAA rupture

The end points for this analysis were reaching a diameter of 5.5 cm in any of the four measurements (i.e. the first visit when this happened even if the patient continued in follow-up) or having/being referred to surgery. A total of 26 patients reached an AAA diameter of 5.5 cm during the course of the trial, with the same number being referred to/having AAA surgery (Tables 23 and 24). There were no AAA ruptures reported. For patients who reached both end points, the date of reaching an AAA diameter of 5.5 cm was used for analysis. Two cases proceeded to surgery before an AAA diameter of 5.5 cm was recorded in-trial and for these two patients the date of surgery was used. No significant differences were found between the three treatment groups for this combined secondary end point (Figure 38).

TABLE 23 - Numbers of patients reaching study end points by randomised group and combined

Study end point	Placebo	Perindopril	Amlodipine	Total
Reaching 5.5 cm	11	6	9	26
Being referred to or having AAA surgery	9	10	7	26
Combined end point	11	10	11	32
Surgery only	0	4	2	6

TABLE 24 - Numbers of patients reaching an AAA diameter of 5.5 cm at each study visit by randomised group and combined

Visit	Placebo	Perindopril	Amlodipine	Total
Baseline	1	0	1	2
Month 3	4	0	1	5
Month 6	0	1	0	1
Month 9	1	1	2	4
Month 12	4	3	2	9
Month 15	0	1	0	1
Month 18	0	0	0	0
Month 21	0	0	1	1
Month 24	1	0	2	3
Total	11	6	9	26

FIGURE 38.

Kaplan–Meier estimates of the proportions of patients reaching secondary end points. a, 11 patients randomised are not included as they were seen only at baseline; b, apparent disparity in numbers attending their 24-month visit largely because of this visit occurring before 720 days. Reproduced from Bicknell et al. 2016. 95 © Bicknell et al. 2016. Published by Oxford University Press on behalf of the European Society of Cardiology. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.

Quality-of-life outcomes

The overall EQ-5D scores at 12 and 24 months are shown in Table 25 (see Appendix 18 for the scores for each of the five domains: mobility, self-care, usual activities, pain/discomfort and anxiety/depression). Overall, mean scores were similar across all three treatment groups at 12 months but a reduction in the mean score from 12 to 24 months was observed in the placebo group (from 71.4 to 64.0), representing a reduction in quality of life. A similar reduction in score was not observed in the perindopril or amlodipine group.

Drug compliance

Compliance was evaluated using pill count data (number dispensed minus number returned) for each time period (Table 26). Mean compliance combining all three groups was > 80% for each 3-month period evaluated.

Safety

Serum creatinine

Table 28 and Figure 39 show the distribution of serum creatinine levels at 3, 12 and 24 months for the three treatment arms and combined. Table 29 shows the serum creatinine differences between each of the study visits and screening. As expected, there was little difference in mean serum creatinine levels in the placebo and amlodipine groups across the duration of the trial but a small increase was observed in the perindopril group (6% at 3 months). However, this increase was not statistically significant and no patients were withdrawn from the trial because of concerns regarding renal function.

TABLE 28 - Serum creatinine levels (µmol/l) at screening and 3, 12 and 24 months by randomised group and combined

P25, 25th percentile; P75, 75th percentile.

Visit	n	Mean	SD	Median	P25	P75	Minimum	Maximum
Placebo
Screening	73	86.6	20.5	79	73	96	56	154
Month 3	72	87.5	19.7	84	72	100	56	153
Month 12	63	88.2	19.8	85	73	99	55	154
Month 24	52	85.5	16.5	83	74	97	58	126
Perindopril
Screening	69	87.0	18.7	86	75	95	54	156
Month 3	66	92.1	19.1	88	78	101	66	169
Month 12	54	93.0	21.9	89	75	105	64	161
Month 24	49	90.3	20.7	86	75	101	53	137
Amlodipine
Screening	67	90.0	20.0	87	78	96	55	178
Month 3	64	90.0	20.3	88	74	100	61	161
Month 12	50	90.3	18.8	87	78	96	54	136
Month 24	43	89.8	18.7	89	75	95	65	143
Total
Screening	209	87.8	19.7	86	75	96	54	178
Month 3	202	89.8	19.7	87	76	100	56	169
Month 12	167	90.4	20.2	87	77	100	54	161
Month 24	144	88.4	18.7	85	75	98	53	143

FIGURE 39.

Median serum creatinine levels at screening and 3, 12 and 24 months by randomised group. (a) Placebo; (b) perindopril; and (c) amlopidine. Circles are outside values. The box represents the 25th and 75th percentiles and the whiskers are the upper and lower adjacent values.

TABLE 29 - Differences in mean serum creatinine levels (µmol/l) by randomised group at 3, 12 and 24 months compared with screening levels

P25, 25th percentile; P75, 75th percentile.

Study period	n	Mean	SD	Median	P25	P75	Minimum	Maximum
Placebo
Month 3 – screening	67	0.6	12.9	0	–4	5	–64	49
Month 12 – screening	61	0.5	14.6	1	–5	6	–61	30
Month 24 – screening	50	–0.4	13.4	–1	–7	8	–51	31
Perindopril
Month 3 – screening	62	5.6	10.4	4	–1	11	–9	54
Month 12 – screening	51	4.8	10.6	4	–2	10	–19	37
Month 24 – screening	47	3.8	11.2	1	–4	11	–17	44
Amlodipine
Month 3 – screening	60	–0.3	10.3	–1	–7	6	–23	35
Month 12 – screening	46	1.7	11.5	1	–4	6	–21	49
Month 24 – screening	41	–0.1	8.2	–2	–5	6	–18	18

Strengths

To our knowledge this study is unique in being the only placebo-controlled RCT to have completed an evaluation of the impact of ACE-Is on the growth rate of small AAAs. However, two other RCTs of the ARBs valsartan (NCT01904981) and telmisartan99 are currently in progress. Interestingly, the large Canadian case–control study6 that reported the protective effects of ACE-Is on the rupture of AAAs did not show similar benefits for ARBs.

Although a significant proportion of trial participants did not complete 2 years of follow-up in the AARDVARK trial because of reaching a censoring trial end point (e.g. referral for surgical intervention or death), the great majority continued regular follow-up and the overall attrition rate was only 6% (which was less than the target attrition rate of 10% used in power calculations). Hence, 94% of patients did contribute at least two sequential AAA measurements to facilitate an evaluation of growth.

Drug compliance among attendees was excellent as evaluated by pill counts and data collection was efficient with only a small number of key measurements or investigations (e.g. BP levels) missing in the trial overall.

The trial included a system for quality control. The standardised procedures used for AAA measurements as outlined in the scanning protocol were designed to reduce measurement variability associated with using 19 sonographers from 11 sites. This protocol was taught at specifically designed QA days. Image quality was also robustly assessed and sites with poor-quality scans were reassessed. Lastly, the measurements taken for all time points were audited by the trial senior clinical vascular scientist and, when obvious differences between measurements in the same subject at the same time point existed, a review took place.

Limitations

The profile of the trial participants – largely white men aged ≥ 55 years with a heavy smoking history – is typical of that for UK patients with an AAA but this should be set in the context of those ineligible for the trial or who did not take part (see Figures 11 and 12). Indeed, the limitations of the AARDVARK trial include the potentially restricted generalisability of the trial recruits. Trial inclusion and exclusion criteria predetermined that, although the overall profile of the recruits might at first sight be considered typical of patients from the UK with an AAA, the age range was limited to ≥ 55 years and SBP had to be < 150 mmHg.

Furthermore, it is clear from Figure 11 that most patients considered for the trial were already taking an antihypertensive agent that made them ineligible. Allied with the other reasons why patients declined to join the trial (see Figure 12), it is reasonable to assume that the trial population may not be representative of the true AAA population in the UK as a whole. This conclusion is supported by the mean SBP and DBP levels recorded in other AAA studies (see Table 2), with SBP levels ranging from 143 to 157 mmHg and DBP levels ranging from 81 to 91 mmHg. These levels appear to be significantly higher than the mean BPs found at baseline in all three treatment arms in the AARDVARK trial (131.7/77.9 mmHg, 130.9/76.7 mmHg and 131.9/78.0 mmHg for placebo, perindopril and amlodipine, respectively).

At a more global level, the findings of a study predominantly including UK-based white men may clearly not be generalisable to non-UK populations, women or those of different ethnicities.

The doses of ACE-I and CCB chosen for the trial were expected to cause similar reductions in BP;73–75 however, this was not found to be the case. Although both treatments were effective in lowering BP, the mean differences in SBP from baseline to 24 months in the perindopril and amlodipine groups were –5.0 (SD 16.3) mmHg and –2.8 (SD 11.7) mmHg, respectively, compared with 2.5 (SD 16.5) mmHg in the placebo group. With similar withdrawal rates observed among the three treatment groups and no differences in relation to compliance, the reason for the difference in BP reduction between the perindopril group and the amlodipine group remains unclear. Although ACE-Is are known to be particularly effective in treating hypertension in patients with renal artery stenosis, patients with known renal artery stenosis of > 50% were excluded from the trial. However, no further investigations were undertaken to explore or exclude lesser degrees of renal artery stenosis in the trial population, making it difficult to draw any further conclusions about the observed BP differences in relation to the role of renal artery stenosis.

A potentially crucial limitation of the AARDVARK trial relates to the statistical power. To calculate the sample size of the trial we estimated that the trial would have 90% power at the 5% level to detect a 38% (1-mm) difference in growth rate between those receiving perindopril and those on placebo. This was based on an estimated annual growth in AAA diameter of 2.6 (SD 1.8) mm, as reported in the UKSAT trial. 22 However, given the actual average annual growth rate observed of 1.7 (SD 3.0) mm, 190 patients per group would have been required to detect a 1-mm difference in growth with a power of 90%. Based on the actual sample size (75 per group), we had 51% power to detect a 1-mm difference in growth (between two groups) and 85% power to detect a difference of 1.5 mm (close to the annual growth rate). However, our estimated difference in the slopes of regression between perindopril and placebo was 0.08 mm with a 95% CI of –0.50 mm to 0.65 mm, which statistically excludes a likely reduction of 1 mm. This implies that an important clinical difference in growth rate of small AAAs is unlikely to occur in this population in association with the use of perindopril compared with placebo.

Given the very small changes in AAA size that were expected over the 2-year follow-up period, numerous quality control measures were put in place to obtain accurate AAA measurements. Furthermore, sites were requested to have the same observer scanning the same patient on the same machine for the duration of the study. All sites reported that they adhered to this request to the greatest extent possible, bearing in mind clinical service pressures and staffing changes. Despite this, the mean (range) intrasonographer repeatability coefficient was 0.33 (0.21–0.45) cm for OTO measurements. Although ultrasonography is a cost-effective tool for the identification of AAAs and sufficient to direct clinical management of patients, the accuracy of measurements made in a RCT becomes critical because of its high operator dependence. During the site initiation visits for the trial it became clear that the routine methods for taking ultrasound measurements varied among sites. Despite the protocol and training that were provided for the trial, unfamiliarity with taking some of the AAA measurements required by the trial may explain the greater than expected range in measurement variability.

The greater than expected intra- and inter-observer variability and measurement error add uncertainty to the interpretation of the results. For future studies involving observers at multiple sites it would be pertinent to undertake detailed and practical training sessions prior to the recruitment of patients. Furthermore, measurement variability might be reduced by using more accurate methods of obtaining images, such as performing an electrocardiogram simultaneously with recording ultrasound videos, allowing for electrocardiogram gating in the reading process. 100,101

Implications for health care

Despite some earlier evidence suggesting that the rupture rates of AAAs may be lower in patients taking ACE-Is, this trial found no evidence that the growing cohort of patients with a small AAA under surveillance should be prescribed an ACE-I to slow AAA growth. The QA studies undertaken as well as the comparison of various aspects of variability of internal and external measurements provide support for the use of external rather than internal AAA diameter measurements taken in the longitudinal plane as the routine measurement of choice for the screening and follow-up of AAAs.

Recommendations for future research

The results of this pilot trial do not provide strong support for a larger trial to evaluate the effects of ACE-Is on small AAA growth rate.

Other than the apparently negative findings of the trial (albeit with some caveats around the issue of power), the large and increasing numbers of elderly patients with an AAA already receiving some form of RAS blockade (the most common cause of ineligibility for this trial) would make randomisation to an ACE-I or an ARB significantly more difficult, with associated concerns about the generalisability of the findings of any such trial. Nevertheless, patients recruited through the NAAASP were less likely to be already receiving a RAS blocker than those already in a vascular database and could therefore offset some of those concerns.

Patients considered unfit for surgery but with an AAA of larger diameter than those included in the AARDVARK trial may, by virtue of their larger AAA growth rates, provide a further source of eligible patients for a more powerful evaluation of medical interventions with regard to AAA size and/or rupture.

Importantly, whichever groups of patients are included in future trials of AAA, strict protocols around ultrasound measurements and training of trial staff are critical if AAA measurements are key outcomes.

The following research recommendations are made as a consequence of the conduct and findings of the AARDVARK trial:

• Further work related to the data already collected in the AARDVARK trial:

  • A multivariate analysis of determinants of AAA growth in the AARDVARK trial.

  • Potential differences were observed between the three treatment groups in relation to the number of patients whose AAA grew at a fast rate during the trial (as defined by a growth rate of > 5 mm per year). However formal analyses are still required.

  • An evaluation of the incremental predictive power of baseline and changing central BPs and BP variability with regard to AAA growth rates.

• Further work potentially arising from the AARDVARK trial:

  • An evaluation of currently available data regarding AAA growth rates in those with SBP of < 150 mmHg and ≥ 150 mmHg to investigate whether growth rates could be critically affected by this systolic threshold or other systolic and diastolic thresholds.

  • An evaluation of whether the BP-lowering effect of perindopril and amlodipine are affected by the presence or absence of an AAA.

  • The strong protective effect of type 2 diabetes on the development of AAAs observed in large observational databases merits further investigation.

  • A large measurement variability study to optimise training and standardisation.

  • A trial to evaluate the impact of ACE-Is on the rupture of larger AAAs.

Grant holders

Professor Neil Poulter, Professor Deborah Ashby, Professor Janet Powell, Mr Colin Bicknell, Professor Christopher Imray, Mr Matthew Waltham and Ms Meryl Davies.

Trial Steering Committee

Mr Jonothan Earnshaw (chairperson), Professor Cliff Shearman and Mr Barrie Newton.

Data Safety Monitoring Committee

Professor Simon Thompson (chairperson), Professor Gareth Beevers and Mr Teun Wilmilk.

AARDVARK trial collaborators

• Imperial College Healthcare NHS Trust (St Mary’s Hospital): Mr Colin Bicknell (PI), Ms Louise Brown, Ms Jennifer Scriven, Mr Alex Rolls, Ms Candice Coghlan, Ms Rebecca James and Mr Andrew Whitehouse.

• Imperial College Healthcare NHS Trust (Charing Cross Hospital): Professor Janet Powell (PI) and Ms Angela Williams.

• Guy’s and St Thomas’ NHS Foundation Trust (St Thomas’ Hospital): Mr Bijan Iodarai (PI), Mr Matthew Waltham (PI) and Ms Susan Clark.

• Royal Free London NHS Foundation Trust (Royal Free Hospital): Ms Meryl Davis (PI).

• University Hospitals Coventry and Warwickshire NHS Trust (University Hospital Coventry): Professor Christopher Imray (PI) and Ms Alison Kite.

• Hull and East Yorkshire Hospitals NHS Trust (Hull Royal Infirmary): Professor Ian Chetter (PI), Ms Emma Clarke and Ms Anna Firth.

• Royal Bournemouth and Christchurch Hospitals NHS Foundation Trust (Royal Bournemouth Hospital): Mr Dynesh Rittoo (PI) and Ms Sara Baker.

• Colchester Hospital University NHS Foundation Trust (Colchester General Hospital): Mr Sohail Choksy (PI), Ms Marianne Morgan and Ms Nyasha Nygo.

• Newcastle Hospitals NHS Foundation Trust (Freeman Hospital): Mr Tim Lees (PI), Ms Lesley Wilson, Ms Vera Wealleans and Ms Noala Parr.

• City Hospitals Sunderland NHS Foundation Trust (Sunderland Royal Hospital): Mr Andrew Brown (PI), Ms Florrie Mowatt and Ms Ruth Chipp.

• York Teaching Hospitals NHS Foundation Trust (York Hospital): Mr Andrew Thompson (PI), Mr Andy Gibson, Ms Zoe Coleman and Mr Thomas Smith.

• Norfolk and Norwich University Hospitals NHS Foundation Trust (Norfolk and Norwich University Hospital): Ms Felicity Meyer (PI) and Ms Mandy Burrows.

• Central Manchester University Hospitals NHS Foundation Trust (Manchester Royal Infirmary): Mr Vince Smyth (PI), Mr Andrew Schiro and Ms Helna Edlin.

• Sheffield Teaching Hospitals NHS Foundation Trust (Northern General Hospital): Mr Shah Nawaz (PI), Rebecca Warren, Diane Swift, Craig King, Cheryl Bailey and Faith Okhuoya.

Thank you also to all of the pharmacy staff and scanning staff/sonographers at all of the above sites.

Contributions of authors

Ms Gaia Kiru (Trial Manager) contributed to the design of the study, the day-to-day management and delivery of the trial and interpretation of the results, and was the principal author responsible for the writing and delivery of the final report.

Mr Colin Bicknell (Vascular Surgeon and coinvestigator) contributed his expertise to the design of the study, recruitment, interpretation of the trial findings, and writing and delivery of the final report.

Ms Emanuela Falaschetti (Statistician) was responsible for the interim and final statistical analyses for the study, and contributed to the writing of the statistical methods and results sections.

Professor Janet Powell (Professor of Vascular Medicine and coinvestigator) contributed her expertise to the design of the study, recruitment, interpretation of the trial findings, and writing and delivery of the final report.

Professor Neil Poulter (Professor of Preventative Cardiovascular Medicine and chief investigator) contributed his expertise to the design of the study, recruitment, interpretation of the trial findings and writing and delivery of the final report, and held overall oversight and responsibility for the study.

Publication

Bicknell CD, Kiru G, Falaschetti E, Powell JT, Poulter NR, on behalf of the AARDVARK collaborators. An evaluation of the effect of an angiotensin converting enzyme inhibitor on the growth rate of small abdominal aortic aneurysms: a randomized placebo-controlled trial (AARDVARK) [published online ahead of print July 1 2016]. Eur J Heart 2016.

Data sharing statement

Requests for available data can be made by contacting the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.