Liver resection surgery compared with thermal ablation in high surgical risk patients with colorectal liver metastases: the LAVA international RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/153/04. The contractual start date was in January 2016. The draft report began editorial review in January 2019 and was accepted for publication in December 2019. 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

Julia Brown is a member of Health Technology Assessment (HTA) Mental, Psychological and Occupational Health Methods Group, HTA Clinical Trials Committee, HTA Prioritisation Committee Methods Group and HTA Funding Committee Policy Group. Stephen Morris was formerly a member of the National Institute for Health Research (NIHR) Health Services and Delivery Research (HSDR) Research Funding Board, the NIHR HSDR Commissioned Board, the NIHR HSDR Evidence Synthesis Sub Board, the NIHR Unmet Need Sub Board, the NIHR HTA Clinical Evaluation and Trials Board, the NIHR HTA Commissioning Board, the NIHR Public Health Research (PHR) Research Funding Board and the NIHR Programme Grants for Applied Research (PGfAR) expert subpanel. Maureen Twiddy is a member of the Research for Patient Benefit North East and Yorkshire Advisory Panel. Brian Davidson is chairperson of the London NIHR Research for Patient Benefit panel. Daniel Hochhauser reports Medical Research Council CASE studentship with Merck Serono (Darmstadt, Germany). Kurinchi Gurusamy reports grants from NIHR, Cancer Research UK Multidisciplinary Award, UK Oncology Nursing Society, University College London and Wellcome Trust/Department of Health and Social Care – Health Innovation Challenge Fund 4 – Smart Surgery, during the conduct of the study.

Copyright statement

© Queen’s Printer and Controller of HMSO 2020. This work was produced by Davidson et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. 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.

2020 Queen’s Printer and Controller of HMSO

Background: pilot study of 1 year

Recruitment to RCTs with very different treatment arms can be difficult, and recruitment to surgical trials is particularly challenging. 12 Qualitative research can identify aspects of the trial design that hinder recruitment and detect possible solutions,13,14 so qualitative work was embedded into the LAVA (Liver Resection Surgery Versus Thermal Ablation for Colorectal LiVer MetAstases) pilot study. 1 Studies show that patient-related (e.g. difficulties of informed consent, understanding randomisation and preference for certain treatments) and clinician-related (e.g. concern about impact on the doctor–patient relationship, clinical equipoise and how the trial is presented to patient) factors affect recruitment. 13–16 To anticipate these issues, a training manual, based on published guidance and supplemented by research findings, was developed. 13,17,18 This provided advice about presenting the trial, the evidence base for the LAVA trial and a step-by-step guide to help recruiters present the information in an unbiased way. The manual was presented at the study launch meeting, a copy was given to each local principal investigator (PI) at site initiation, and this was followed up with telephone-based support. The qualitative substudy aimed to understand barriers to recruitment from the perspective of patients and staff.

Objectives

The purpose of the main trial was to compare the clinical effectiveness and cost-effectiveness of thermal ablation with liver resection surgery in high surgical risk patients who were eligible for liver resection.

An internal pilot study was run in the first year of recruitment and the objectives were to assess the feasibility of recruitment, the quality of ablations and the quality of liver resection surgery to determine acceptable standards for the main trial and to centrally review the reporting of computed tomography (CT) scan findings and recurrence (in both arms) to ensure quality of reporting.

The aims of the qualitative study were to:

• explore patient and clinician acceptability of the trial and recruitment processes/study procedures to assist in optimisation of recruitment

• explore clinical equipoise in the liver surgery community

• understand how information is presented to recruiters, in particular explore the content and style of delivery and feed this back promptly to recruiters to improve practice.

Patient and public involvement aims

Two patient and public involvement (PPI) representatives were identified to inform key aspects of the pilot study, in particular:

• to ensure that the design and wording of all information sheets and consent forms were presented in a clear and accessible manner

• to provide input into the topic guide for the participant interviews (pilot study)

• interpretation of the qualitative results

• to ensure that the content and the format of the dissemination of trial findings were relevant and appropriate to patients

• to ensure that the patient representatives were supported to provide input into the pilot study.

Trial design

The LAVA trial was a prospective, international (UK and the Netherlands), multisite, open, pragmatic, parallel-group, RCT design with an internal pilot to investigate the clinical effectiveness and cost-effectiveness of thermal ablation (RFA or MWA) compared with liver resection for the treatment of patients with resectable CLM who would be considered as high risk for surgery and with a low chance of cure.

The target was 330 participants (165 in each arm) to be recruited into the trial over a planned 48-month recruitment period.

Participants

Participants who met the eligibility criteria below were recruited from tertiary hepatopancreatobiliary (HPB) centres in the UK and in the Netherlands (see Appendix 1).

Summary of protocol amendments

The summary of protocol amendments is shown in Table 1.

Randomisation

Participants were randomised by minimisation incorporating a random element (1 : 1 allocation ratio) to either liver resection surgery or thermal ablation (MWA or RFA). Minimisation incorporated a random element, stratified by research site, synchronous/metachronous disease, primary cancer in situ, size of largest tumour, prognostic factors for inclusion, planned surgical resection and planned ablative treatment. Randomisation was performed by the Clinical Trials Research Unit, Leeds.

Interventions

Participants were randomised to either liver resection carried out within regional centres in accordance with individual protocols or thermal ablation, RFA or MWA, depending on local availability and expertise. Centres were permitted to perform ablation laparoscopically or at open surgery if the percutaneous approach was contra-indicated. Minimum ablation and surgical standards were detailed in the LAVA Site Standard Operating Procedures, with ablation criteria based on the CIRSE guidance. 19

Follow-up

Follow-up of participants was planned at 3, 6, 12, 18 and 24 months post randomisation. All participants were planned to be followed up for 2 years post randomisation. Longer-term survival data were to be obtained from the Office for National Statistics at 5 years post randomisation.

Sample size

A total of 330 patients were required to demonstrate non-inferiority of thermal ablation with respect to resection in terms of the primary end point [2-year disease-free survival (DFS)] with 80% power at the 2.5% one-sided level of significance, assuming a median time to event of 14 months in the resection group and a non-inferiority margin for thermal ablation of 4 months, and allowing for a 5% drop-out rate.

The qualitative substudy had a planned sample size of ≈20 patients, comprising consenters, decliners and those withdrawing post randomisation, and 15 recruiters. Participants were to be recruited from across all centres open during the pilot study. Up to 60 hours of recruitment encounters were to be audio-recorded.

Analysis methods: statistical

Owing to the small sample size and early closure of the trial, it was not possible to perform any of the original planned analysis for the main trial. The intention for the main trial was to assess the difference in DFS. The non-inferiority hypothesis would have been tested using an appropriate survival model to incorporate random effects with respect to research sites, and including adjustment for the stratification factors. Differences in rates, such as complication rates and recurrence rates, would have been analysed using multilevel logistic regression incorporating random effects with respect to research site, and would have included adjustment for the stratification factors.

Analysis methods: qualitative study

In-depth, semistructured telephone and face-to-face interviews were conducted with recruiters (consultants, interventional radiologists and nursing staff), and face-to-face interviews were conducted with patients. Interviews were audio-recorded, transcribed verbatim, anonymised and analysed thematically. 20 This allowed the interpretation of key issues faced by recruiters, and reasons for non-consent from the perspective of patients and staff to be examined. Participant information sessions (recruiter consultation encounters) were audio-recorded (with permission) to examine how information was presented by recruiters and received and understood by patients. An iterative approach to interviewing was taken with earlier interviews guiding later discussions. Data saturation was reached for interviews with recruiters but not with patients because of the small number of patients who consented to be interviewed.

Central scan review

Some of the other objectives of the pilot study were to assess the quality of ablation and the CT scan reporting through a central review. Central review of imaging was carried out for baseline scans, post-ablation scans and 3-month imaging review to assess the consistency between local centre reporting and central review of imaging and whether or not the local centre expertise in ablation was adequate for trial inclusion. The aspects checked in the baseline scans included whether the lesions were unilobar or bilobar, the number of lesions, the size of the lesions, the presence of extrahepatic metastases or extrahepatic nodal involvement, and whether or not the lesions were within the Cardiovascular and Interventional Radiological Society of Europe (CIRSE) criteria. The aspects checked in the post-ablation scans were whether or not complete ablation was achieved in all lesions and whether the treatment margin was ≥ 1 cm or < 1 cm. The aspects checked in the follow-up review scan were locoregional and distant recurrences.

End points

Delays during site set-up

The LAVA trial encountered numerous challenges and obstacles that delayed site set-up. These delays were experienced at centres in both the UK and the Netherlands.

Interest in taking part in the trial was high as 12 centres in the UK registered an interest, with a further seven expressions of interest received. However, the target number was 15 sites.

The sites were asked to complete a Site Expression of Interest and Feasibility Form to confirm their eligibility to take part in the trial as defined in the protocol. To be eligible, sites had to meet the research site eligibility and surgeon and radiologist eligibility below.

Research site eligibility:

• The site must be a tertiary liver, pancreatic and biliary (HPB) centre.

• The site must have experience in the provision of liver resection surgery and thermal ablation for CLM.

• The site must have a shared specialised sMDT with representatives from surgery and radiology.

Surgeon and radiologist eligibility:

• To perform liver resections within the trial – prior experience of performing liver resection procedures for a minimum of 20 patients with liver cancer.

• To perform thermal ablation within the trial – prior experience of performing ablative procedures for a minimum of 20 patients with liver cancer.

The sites that met the above criteria were selected and set-up commenced. The trial document packages were circulated to selected sites at the beginning of August 2016. The first UK site opening to recruitment was King’s College London on 30 November 2016, but there were significant delays in opening other UK sites for recruitment. It became apparent during site set-up that not all CT scans required for trial assessments were being conducted as part of standard care. There was a delay to site set-up and recruitment while approval was sought from the Research Ethics Committee (REC) and Health Research Authority for exposure to this additional ionising radiation. Following approval of the amendment on 22 March 2017, the first UK site re-opened to recruitment on 28 March 2017 (The Royal Free Hospital, London), followed closely by another four sites in the UK (i.e. King’s College London, Cambridge, Oxford and Newcastle).

It was difficult to identify local research infrastructure to support the trial. Links to those with research governance responsibility within the radiology departments and who had a key role in approving the study at the site set-up stage also caused significant delays. Two sites were keen to take part but were delayed because research nurses left, which resulted in a lack of local capacity. Contact was maintained with these sites to facilitate set-up once the local network provided additional support.

Another contributing factor was the radiology approvals. Although the study had received national radiology approval, each department conducted an internal radiology review. This led to further delays in receiving confirmation of capability and capacity from the research and development and hospital finance departments.

At some sites, the trial interventions were delivered across two centres (i.e. surgery in one centre and ablation in the other). This caused delays during set-up while the complexities of the process and accrual allocation were discussed.

Set-up at the sites in the Netherlands also took longer than anticipated. Each site had to submit the study and contract template for review to their local board to obtain approval before they could be open to recruitment. This caused major delays because of contract negotiations and the need to meet Dutch law requirements for participant insurance cover.

Obtaining participant insurance cover that was acceptable to both the sponsor (University College London) and the Dutch REC presented a major obstacle.

Although the UK uses the model non-commercial agreement for research sites in which the main body of the agreement is used without modification or negotiation, this was not the case for the international sites. All changes had to undergo review and negotiation between the sponsor and site lawyers, which took several months as the contract went back and forth between both parties.

As a result of the delays experienced in site set-up and the trial amendment, the pilot phase assessment was reset to calculate the 12-month period from March 2017 until the end of February 2018. During this period, the trial opened in 17 sites (13 UK and 4 Dutch sites).

Screening and recruitment challenges

Owing to the delays in site openings, not all sites were opened for the full duration of the pilot study. Overall, only eight sites were opened for > 6 months during the internal pilot.

Once opened to recruitment, the trial teams at the sites were asked to send screening logs summarising reasons for ineligibility or non-randomisation. We received 366 screening logs for potential LAVA patients from 13 of the 17 sites that had been opened for recruitment. A total of five sites did not contribute any screening data over the course of the study. This includes the site that opened but did not reach the point of screening patients at the time of completion of the pilot.

Of the 366 screened patients with CLM discussed in the sMDT, 59 were eligible (16%) and 307 were ineligible. Of the 59 eligible patients, 14 were not approached, 36 were approached but not consented and nine were consented and randomised.

Of these nine participants, five were randomised to the thermal ablation group and four were randomised to the liver resection group. All but one patient (who was allocated to the thermal ablation group) had metachronous disease, and no patients had their primary colorectal cancer in situ at the time of randomisation.

All of the patients who were allocated to the surgical resection group and three out of the five who were allocated to the thermal ablation group were considered high risk because of their age, and four of the patients allocated to the thermal ablation group had major comorbidities present. No patients were considered high risk because of liver metastases with poor prognosis or high risk for surgery because of tumour burden. In no patient was the largest lesion > 5 cm.

Seven patients completed the allocated intervention. One participant allocated to the thermal ablation group did not receive their allocated intervention because of new clinical findings found prior to the ablation (see Appendix 2). Another patient switched from their surgery allocation to ablation as their intraoperative ultrasound showed a more complex anatomical location than previously seen on preoperative imaging. This resulted in the surgeon opting for ablation instead of the planned surgical resection. At 3 months post randomisation, six patients had no hepatic or new extrahepatic disease, one patient had withdrawn, one patient had evidence of new liver and extrahepatic disease and one had not returned their 3-month follow-up form.

Of the screened patients, 16% (59/366) of all patients were considered to be suitable for the LAVA trial. They were appropriate for either ablation or liver resection and were considered to be at high surgical risk because of participant or tumour characteristics. The median age of eligible participants was 75 years.

The trial was stopped early because of poor recruitment as the internal pilot for 1 year recruited only nine participants (UK, n = 3; the Netherlands, n = 6) out of the recruitment target of 45. No formal statistical analyses will be conducted as there are insufficient patient data to provide any significant conclusions. The health economic analysis will not be performed because of the small numbers of participants who were randomised into the study.

The reasons for ineligibility and non-randomisation can be found in Tables 2–4.

Assessing barriers to recruitment and actions taken to increase recruitment

Emerging issues related to recruitment were discussed as some centres reported fewer potentially eligible patients than expected. This was reported to the Independent Steering Committee, which requested that the qualitative team shift the focus of their work from understanding the experiences of patients and staff taking part in the trial to understanding the challenges of identifying eligible patients so that these may be addressed by intervention from the research team.

It was reported that there were significantly fewer eligible patients at the sites than expected. Closer examination identified that some sites were misapplying the inclusion and exclusion criteria, and the chief investigator visited each site and met with clinical colleagues to remedy these misunderstandings. Patients suitable for trial inclusion were identified at all sMDT meetings visited. However, this did not lead to improvements in recruitment numbers. Interviews with staff revealed that even when potentially eligible patients were identified in the sMDT meetings, usually chaired by a surgeon, they were often denied the opportunity for discussion of the trial. Few patients consented to be interviewed and so little can be gleaned from their perspective; however, there was some evidence from staff interviews that recruiters struggled to maintain equipoise when presenting the trial.

Prior to study launch, a training manual was developed to help sites with study introduction, and this was discussed at each site initiation meeting. A follow-up teleconference with recruiting staff was also arranged following site set-up to identify early challenges and to go through the training manual with recruiting staff. However, it was many months between training and the identification of potentially eligible patients, so much of the initial learning was lost. During interviews with staff, we explored training needs; however, most centres refused training as they considered that they had a strong track record of recruiting to studies, and so felt that they did not require training. Alternatively, some centres had not identified any eligible patients and so felt that it was premature to judge if additional training were needed.

Thirteen out of the 15 originally planned staff interviews were conducted and these were analysed as intended. The 13 interviews generated 268 minutes of rich audio data. All interviews were transcribed verbatim and managed using NVivo 11 (QSR International, Warrington, UK). The data were analysed using thematic analysis. 20,24 A coding frame was established (MT and JaB) and emerging themes were explored iteratively in the remaining interviews, with additional codes added, and then applied to all transcripts.

The patient recruitment encounters (audio-recordings of recruitment encounters) were examined initially using the Q-QAT (Quanti-Qualitative Appointment Timing) technique25 to assess the time taken to present each treatment as well as the study itself. Content of the discussion about the surgical and ablation options was then examined, as well as how recruiters demonstrated equipoise and how they described randomisation using thematic analysis. 20 All of the recordings were listened to, detailed notes were made using the Q-QAT technique and then a thematic analysis of the content of the discussions was performed.

Key issues inhibiting recruitment from the perspective of local teams were:

• Fewer than anticipated eligible patients.

• Clinician judgement about the suitability of the patient for both treatments led to fewer than expected patients identified.

• Patient preferences for individual therapies prevented randomisation.

• Lack of a dedicated research nurse for the trial and so inability to attend sMDT meetings and keep the study in the minds of the clinicians.

• Research nurses working across studies and so lacking confidence when talking about the trial or interventions (as few patients approached).

• Fewer than half of the clinicians could recall receiving a training manual, but did not think that they required training to improve recruitment of patients.

• The trial was not well embedded into clinical pathways and so patients had to wait until their second appointment before being consented.

Interviews with clinical staff (both face to face and telephone) were conducted by Jaqueline Birtwistle, an experienced researcher with a background in psychology. Interviewees were consultant surgeons (n = 5), a specialist registrar (n = 1), an interventional radiologist (n = 1), research nurses (n = 5) and a research assistant (n = 1). All were actively involved in trial delivery and recruitment (Table 5).

Are there enough eligible high-risk patients?

Opinion was strongly divided on the question of whether or not there were sufficient eligible patients for this trial. Some surgeons believed that the inclusion criteria meant that there were few eligible patients, whereas others maintained that eligible patients exist and used their own records to support this, but believed that clinician preferences meant that suitable patients were not being considered, or were being excluded by the sMDT, which was usually chaired by a surgeon:

No I just don’t think there’s enough patients.

Consultant 2

I think the numbers you’ve got are actually quite small, who would be appropriate, and then, yes, and then there’s the issue of equipoise amongst the clinicians that again cuts it down even further.

Consultant 6

We went through our sMDT minutes for last year to estimate the number of potential patients we would have, so I was thinking roughly 15 to 20 a year who would be eligible, and I thought they would be straight-forwardly accepting to go on the trial. I was wrong.

Consultant 4

One of the biggest challenges from the perspective of recruiters was the involvement of the sMDT, usually chaired by a surgeon, in screening eligible participants. Conflicting views about the eligibility criteria and clinician preferences meant that eligible patients were excluded from the trial. The role of the sMDT in the LAVA trial was to determine patient eligibility, and recruiters described how some colleagues acted as gatekeepers, screening outpatients who the participating clinicians believed would be good candidates. The dominant discourse was that, if the surgeon caring for the patient did not feel that the patient would benefit from trial involvement, then the patient was not provided with trial information or an opportunity to discuss the trial. There was a strongly held view by local PIs that many colleagues believe that surgery is better than ablation, and ‘their fragile equipoise evaporated when faced with individual cases’, making it difficult to get patients through the sMDT. Two surgeons also made the case that a surgeon is rated on their ability to undertake complex surgeries, so there is a strong preference for surgery and an imperative to undertake complex cases:

I think we’ve all slightly been taken by surprise by the fact that in many instances in the sMDT, when a patient is considered to be a potential candidate we find that for one reason or another, there is a very clear preference for one treatment over the other. [There is] a strong feeling that there isn’t equipoise in the treatment option, so ethically we don’t feel if it was for ourselves or a relative, we just don’t feel that offering the trial is acceptable, and that’s happening more frequently I think than we anticipated.

Consultant 2

Surgeons are not rated on their ability to cure patients, surgeons are rated on their ability to perform more complex operations. So ultimately the more complex operations that you perform technically well – the better surgeon you are – that’s what people consider.

Registrar 1b

In interviews, some put clinician reluctance to put patients forward for the trial down to a difference in the threshold for surgery between surgeons. Other common reasons for refusing to put patients forward were a belief that the tumour was inoperable and a belief that the patient would not want to take part, so there was no point in offering the trial:

. . . today’s patient, both myself and [interventional radiologist] would have been perfectly happy to randomise, but the colleague whose patient it was felt that they had a preference for ablation, and that can be a problem because where equipoise lays varies between consultants, so the main problem is finding patients. Everybody in the sMDT is screened for LAVA, we just, we have that as a routine, for every single patient, and we don’t really miss any, so it’s about equipoise in the staff in the senior staff about whether they are or are not suitable for LAVA.

Consultant 8a

Clinical equipoise

Most clinicians had seen few eligible patients at time of interview. Only two clinicians provided audio-recordings of the consultation or recruitment interview, so evidence of how the trial was presented to patients is very limited. The audio-recordings were from consultants who had attended the recruitment training, were involved in the study design and were very experienced in presenting studies to patients. The recordings showed that these recruiters had a generally good understanding of how to present the trial in a balanced way. Although the discussion was driven by the clinician, the terminology used did not overtly favour one treatment over the other. Both clinicians took time to check the patients’ understanding of the treatment options in the trial and answered questions as they went along. However, neither actively challenged patient preferences when these were expressed.

In the interviews, we asked clinicians about equipoise in the clinical community, and found that although clinicians felt that at a community level there was equipoise, clinicians found themselves not to be in equipoise when faced with individual patients:

He [surgeon] pushed the patient towards ablation and not in for randomisation, he just said ‘comorbidities are against resection here’ and pushed that particular patient towards ablation.

Interventional radiologist 8b

I think as a whole, the whole HPB surgery community probably rather underestimated the number of times where there actually isn’t equipoise, I think that’s what I perceive as a main issue.

Consultant 2a

We asked clinicians to run through what a recruitment encounter might look like. All said that they would start by discussing the diagnosis with the patient before introducing the trial. The trial would be mentioned mid-way through the appointment, but not discussed in detail until after the treatment options had been presented. When asked about their own treatment biases, all mentioned that they felt comfortable saying that they did not know which treatment was best, because they had the opinion of the sMDT to back up this decision. If the sMDT thought that the patient was suitable for the trial, the consensus was that the participating surgeon should be in equipoise, because the sMDT was in equipoise:

I do a lot of randomised trials so I would hope that if there is true equipoise such that we’ve decided to offer them the trial, they would not for me pick up bias because essentially I wouldn’t have it, I wouldn’t be asking them to participate.

Consultant 8a

Although many surgeons claimed to be in equipoise, the terminology used in the research interviews indicated that there was often some unconscious bias towards surgery, with surgeons often presenting surgery first, and describing it more positively than ablation. Surgery was described as ‘standard treatment’ and, although it was common for the risks of surgery to be clearly described, the risk of cancer recurrence was mentioned less often. In contrast, ablation was described in a balanced way, or recruiters expressed less confidence in the treatment:

If you say well you know, the results from ablation have never been proven, you know, effectively compared to surgery then they say ‘well why would I want to have something that’s inferior?’.

Consultant 6

I would say to the patient that in our practice of standard care in the absence of the trial, it’s surgical resection, that is our policy. In someone in whom both treatments could be done unless there’s a specific reason for them to have ablation, and that would be either related to prior surgery or related to their poor state of health, in general terms they would be offered surgical treatment in the absence of the trial.

Consultant 8a

When asked, clinicians admitted that if a patient asked for their professional opinion they would still guide the patient towards surgery, and the interventional radiologist and nurses also believed that this would be the case:

The feeling among the surgeons, be it correct or incorrect, or evidence based or not, is that surgery is preferential. And so, if the patient asks the surgeon, which they quite often do, you know, ‘what would you do?’ or ‘what would you do it if it was your mother or father?’. And they will, they do all prefer surgery.

Nurse 3a

Both clinicians and nurses talked about the complexity of the decisions patients needed to make, and the importance of ensuring that patients understand their choices. This complexity could have an impact on how the trial was presented, the order in which information was discussed, and whether or not the study was presented at all:

I would talk them through what the procedures would involve. Then I would explain the question of equipoise, of uncertainty as to which, whether or not one or other treatment is better, and then I would try and work out whether or not the patient understood the question about the different treatment options and how there might be uncertainty as to whether or not one is better or worse. If I thought they had understood that question, that there might be two treatments that are similar with some advantages and some disadvantages to both arms, then I would now introduce the concept of the trial.

Consultant 1a

. . . equivalent ablation might be a percutaneous one, so what that means is that in counselling the patients about the two options although from a cancer perspective the two may be equivalent, from a post-operative recovery perspective it’s very different.

Consultant 2a

Three clinicians sought to reduce the variability in how the trial was presented by seeing all potentially eligible patients themselves. However, this had unexpected negative consequences for the clinician involved because many of the patients did not prove to be suitable for the trial.

. . . that group of high-risks patients are all seeing me which is maybe a disadvantage but maybe from a point of view of trying to get the trial running it’s ideal.

So, what makes it a disadvantage?

Well I’ve got operating lists to fill and if I see only the patients that are extremely high risk, I spend the vast amount of my clinic time seeing patients that may not have surgical treatment – so, in a way, it’s just practical difficulties.

Consultant 1

What is patient acceptability of the trial?

Of the four patients initially consenting to interview, two subsequently declined for health reasons. Both interviewees were decliners. One patient thought that their clinician was able to maintain a good level of equipoise, but the other reported that, when asked, the clinician had expressed a preference for ablation, ‘in her case’.

Understanding and addressing patient preferences

At the time of their interview, four out of five surgeons had identified and approached at least one eligible patient. There was a strong belief among all of them that patient preferences existed and recruiting staff reported that they did not feel comfortable challenging patient preferences. For this reason, telephone-based training using a six-step recruitment process to help direct discussions17 had been provided to all sites to help them present the trial in a balanced way and to help them gently probe patient preferences. However, delays in opening sites meant that several months had passed before they approached any eligible patients. This meant that the learning accrued during the training had been lost.

Patient decision-making is known to be affected by the initial clinician encounter, and there was a belief that preferences were often already set before the patient met the liver team. Referrals came from colorectal colleagues, often based in other hospitals, and almost all of the clinicians mentioned that patients had already been made aware that they had been referred for surgery (rather than to discuss treatment). The surgeons felt that this made it difficult for them to introduce the possibility of the two different treatments in the study and proposed that communication with referring sites needed to be improved:

We do find sometimes that the treating clinicians before referring to us haven’t actually explained everything, you know, or imply the surgery is something very minor without risk you just cut here. [. . .] So they don’t go in to the depths, we kind of start afresh pretty much, making sure they understand the diagnosis, the prognosis, the treatment options, the complications, risks, etc.

Consultant 6

A further complication was that patients were coming to see a surgeon at a surgical clinic, which primed patients for discussion of surgical intervention. A key patient eligibility criterion was ‘high risk for surgery’, but most patients had already undergone surgery for their colorectal cancer, and interviewees at one site reported that many patients they met with could not reconcile themselves to the idea that they were high risk:

You’re talking about a non-surgical intervention in a surgical clinic, and patients know they’re coming to see a surgeon, and surgery has been mentioned to them all along their investigations. These are patients with metastatic disease so therefore they’ve already undergone surgery for bowel cancer, they know the drill, [. . .] if they’ve undergone a bowel operation then to them they’re fit for a liver operation.

Nurse 5a

The problem with the referring hospitals is they don’t necessarily have the expertise to know the options. Usually when they refer patients in they refer them in to a liver surgical clinic, so the patient expects liver surgery.

Consultant 5

Staff interviewees described how some patients were adamant they did not want further surgery if they could avoid it:

Most of our patients have had a bowel operation and chemotherapy and have had enough of surgery. They prefer ablation. They know they can always come back and we will take it [cancer] out. That is a plan for them as well, this way they can have a better quality of life. [. . .] They keep coming with such strong opinions of their own, which is very, very unusual in our region, we are very surprised it happened. As we do so many trials, they come mentally prepared for that, most of them are, which is why I was so taken aback when a patient said, ‘No, I want RFA’.

Consultant 4

Clinicians also described how, in other cases, patients came to their first appointment with strong preferences, and these were often based on personal situation as much as clinical factors:

It depends quite a lot actually on their social situation and so people that might have a lot of caring responsibilities or they want to get back . . . you know perhaps to work or something like that, they all really prefer ablation. People that are perhaps looking at the longer-term benefits, then they’re quite, you know they’re not worried by surgery. That sort of thing, they, they tend to go for surgery.

Nurse 3

Although recruiters did ask patients why they had a particular preference, they tended to accept the reasons given at face value if these preferences appeared informed in the light of the patient’s own circumstances:

I am uncomfortable doing that [challenging preferences], for example because according to, let’s say the Helsinki declaration if they have a strong preference then their interests go ahead of any other interests.

Specialist registrar 1b

All of the ones that I’ve seen have very quickly said ‘I would prefer . . . one or the other’.

Consultant 3

A common discourse from nurses was that they were reluctant to challenge patient preferences, arguing that it was not their role to coerce patients into taking part in the trial:

We got another lady and she preferred ablation because [. . .] She was going to have, go for surgery eventually. But she was treating the liver lesions first and so . . . she didn’t want two surgeries – one for the liver and one for the colorectal afterwards, so she preferred ablation [. . .] they consider their situation, what they can bear I think . . . that would fit within their lives.

Nurse 1c

In contrast, when patients expressed misconceptions about the two treatments, staff felt able to manage these much more easily, as such misconceptions are common in clinical practice:

I think the fact that they’ve all been discussed at the sMDT before we see them in clinic helps because we can back our discussion up with the fact that it’s been discussed by a group of experts. I don’t think that’s an issue as much for us. Now there may be some misconceptions about the technical aspects of ablation or the surgery or whatever, and that’s something we can address.

Consultant 6

To get anything out of a consultation they need to be in complete understanding of the facts [. . .] and so I am happy, if they have a misconception, then I’ll say, ‘well, it might not be like that . . .’ and explain and give them the evidence and the facts. I would never let anyone walk out of the room with a misunderstanding or they will base their choices on that.

Nurse 3a

Integration of the trial into clinical practice

The study was not well integrated into clinical practice and some recruiters felt that the structure of the recruitment process was weighted against recruitment. Recruiters described how some patients appeared open to the study initially, but changed their mind by the second appointment. The study and the issue of clinical uncertainty were introduced to patients during the first meeting. In the face of this uncertainty, patients used the time between appointments to decide which treatment was best for them, often talking to family and friends, and returned to the second appointment having made a treatment decision:

In one case the patient had made up their mind they wanted surgery and they wouldn’t countenance ablation, and the reverse was another case, you know, so it is difficult.

Consultant 6

Some centres thought that recruitment would improve if this criterion was changed, which was a view that was not universally supported because others felt that the information provided was too complex for one visit:

The problem was the consenting procedure. The protocol had the patient returning to clinic a week later, and that was a problem as patients travel a great distance. This was amended, so now we see them the same way we do other trial patients, we see them once and give them the information sheet and then they post it back to us, if they consent. [. . .] most patients are receptive to trials here, we are a very research-oriented institute and we have not had any problems with other research we have. When people come here, they know they are going to be asked about trials.

Consultant 4

To explain the complexities of the differences and give them time to take that on board and then introduce all the issues about different treatment options and all the pros and cons of the different treatments, it is a lot to take on board. [. . .] How somebody that has just been introduced to this [trial] could possibly take it all on board, I think it is a lot and as you know some of the centres are saying well let’s do it in just in one visit, and I don’t know.

Consultant 1a

A further problem with the recruitment process is that patients did not get to see the interventional radiologist until the surgeon had seen the patient, and so some patients were screened out before the interventional radiologist got to see them:

I’m not seeing the patients, the surgeon is, and therein lays a problem, possibly on some sites, you know, it’s, it remains the historical case that the surgeon sees them in outpatient clinic when the trial is really of ablation [. . .] so there is a lack of a balanced process.

Interventional radiologist 8b

All nurses worked over multiple trials and some reported a lack of experience when talking about trial options with markedly different interventions and acknowledged that they still needed time to get to know the project. Familiarity with the trial was hindered by the small numbers of patients being approached, as nurses did not get to sit in on consultations and so improve their knowledge through regular and repeated exposure to discussions about the interventions:

I don’t know much about the ablation to be honest, and that something I think as part of the study that we’re learning as we go, because the pathway for patients is different [. . .] that’s something we are trying to get our heads around as well.

Nurse 5a

Patient and public involvement input to the LAVA trial

The PPI representative in the Trial Management Group was actively involved during the grant application stage of the trial. The input and guidance received for the patient information was valuable during early trial development. However, once the trial moved to the set-up and recruitment stage, it was difficult to allocate a specific task to the PPI representative as there were only a few patients recruited to the trial. We met with the PPI representative to discuss where their input would be more valuable. A plan was put in place for the PPI representative to assist with the qualitative study, review the patient recruitment encounters and discuss the patient interviews. Unfortunately, this was not achieved as the PPI representative had to resign because of poor health.

Computed tomography scan central review

Baseline review was carried out on six patients. There was consistency between local centre reporting and central review as to whether the lesions were unilobar or bilobar and as to the number of lesions. The segments were consistent in three patients, over two segments rather than one in two patients and in segment 5 rather than segment 6 in one patient. The difference in the size of the lesions varied from 2 mm to 6 mm and was not available for two lesions. There was no extrahepatic nodal involvement noted in either the local or the central review. In one patient, a lung metastasis was noted on central review and not on the local team report.

Central review showed minimal size differences in the lesions, all of which were within CIRSE criteria. The stage of the disease was considered different at central review (possible lung metastasis) in one patient only.

Post-treatment scan

Central review of the post-treatment images suggested that there was complete ablation of all of the lesions, as suggested by local team reports. The treatment margin was > 1 cm in three out of the six lesions but was felt to be < 1 cm in the other three. Overall central review supported local centre assessment of successful treatment.

The scan review suggested that all local centre expertise was appropriate.

Three-month follow-up

The 3-month follow-up scan was available for five patients. One patient had developed liver recurrence and two developed lung metastases.

Central review raised uncertainty as to whether or not the patient with a 5-mm lung metastasis had metastatic disease. CT scanning has a low diagnostic accuracy for small and solitary lung lesions and serial review would be required to confirm or refute this finding.

The follow-up imaging review supported local review in all except one patient, for whom follow-up imaging would be required for a definitive result regarding the presence or absence of recurrence. Therefore, differences are minor and suggest a high standard of local team radiology reporting.

Main trial

We have reported the results of the internal pilot of a RCT comparing liver resection surgery with thermal ablation for people with CLM and who are considered high risk because of patient or tumour characteristics. We have highlighted the difficulty in recruiting into a RCT of this nature and the reasons for non-randomisation.

The delays in site set-up during the 1-year pilot led to reduced opportunities for recruitment despite our efforts to reduce the site opening times. In addition, the screening data received showed that only 16% of patients with CLM were eligible for the trial. The lack of a uniform definition of high surgical risk may have been a problem for identifying suitable patients at the sMDT meetings.

The lack of equipoise at site, patient preference for either ablation or surgery, and the dislike of randomisation were the main reasons for non-randomisation. It should be noted, however, that there was no evidence that more patients preferred ablation over surgery or vice versa.

Although hypothesis testing was considered to be inappropriate because neither the hypothesis nor the sample sizes were calculated a priori, the numbers of participants who expressed a strong preference for surgery or a strong preference for ablation were quite similar. This is surprising given that surgeons might be predicted to have a bias towards liver resection even in those who are high risk.

Qualitative study

During interviews with clinicians, they all described episodes where fellow members of the sMDT, usually chaired by a surgeon, declined to provide information on the trial for suitable patients or to allow patients an opportunity for trial discussion. This resulted in fewer than one-third of patients who fulfilled the criteria for the trial having a face-to-face discussion about recruitment into the LAVA trial. One solution would have been to screen referred patients and highlight their suitability for the trial prior to a discussion at the sMDT meetings.

Trial eligibility criteria involved the application of clinical judgement, and surgeons tended to view patients simply as having resectable metastases, or not, rather than weighing up the option of two treatment modalities.

The original idea behind this study was to identify those patients who were at high risk physiologically or oncologically. Several interviewees acknowledged that some surgical colleagues did not have a good grasp of the clinical research evidence and so were convinced that surgery was better than ablation and, therefore, were reluctant to have their patients involved in the trial. If patients were at high risk physiologically for surgery, the risk of mortality was higher. This led some clinicians to ask ‘Why would I even randomise this patient? They are too high risk for surgery, I would send them for ablation’.

With a few exceptions, centres also cited a lack of suitable patients as a major reason for poor recruitment. One centre had compared its anticipated numbers with its actual number of eligible patients but could not find an explanation for the lower than expected numbers. The difference in numbers of patients screened (screening log data) suggests that some sites had difficulty with clinical equipoise and so did not screen some patients. Surgical trials are known to be difficult to recruit to26–28 and, for this reason, time was dedicated during site set-up to train trial staff about recruiting and the importance of having clinical equipoise. However, the sMDT, usually chaired by a surgeon, many of whom were not actively involved or supportive of the trial, made this training process less effective, as these clinicians did not take part in training. The findings of the study suggest that the LAVA trial suffered from a lack of clinical equipoise at the community level. Although all of the local PIs wanted the trial to succeed, the role of the sMDT in clinical decision-making undermined the equipoise necessary for the LAVA trial and reduced any influence that the local PIs had over trial recruitment. More time and input was needed at site set-up to get referring centres and all members of the sMDT on board for the trial, and ensure that staff understand, and buy into, the rationale for the trial. Basic training on the importance of equipoise was delivered at the study launch and site initiation meetings, but this was not disseminated to the many members of the sMDT. A manual was also provided with a six-step recruitment plan to help guide discussions with patients. However, this training focused on how to present the trial to patients in a balanced way. It did not consider how to educate the sMDT members and ensure equipoise within the wider clinical community.

There is some evidence from interviews with patients and staff that clinicians did not always present the interventions in a balanced way, with surgery often painted in a more positive light than ablation, and the risks of cancer recurrence after surgery were not always mentioned in the consultation episode. Such biases were anticipated and a series of learning opportunities were built into the study to address this. The chief investigator gave a presentation of the research evidence at the study launch, and repeated this information in the study protocol, at site initiation meetings and as centre sMDT attendances. However, in most sites several months elapsed between site initiation and first patient being approached, and learning accrued from exposure to the research evidence was potentially lost. It may also be the case that the clinicians were not convinced by the evidence and training did not change long-held views. The site leads, who were engaged and may have been well informed regarding the equipoise in the patient group to be recruited, were small in number compared with the many members of the sMDT who may have lacked understanding of the background evidence and hence may have been disinclined to support recruitment.

Similarly, some of the nursing staff involved in recruitment lacked confidence in describing the two treatments, and this may also have contributed to poor recruitment. Interview data from across the sites suggest that the structure of the recruitment process had the potential to introduce a bias towards surgery as ‘a patient seeing a surgeon expects surgery’. Although this was a trial of ablation, there was no interventional radiologist input at the recruitment interview, and the interventional radiologist did not appear well engaged at most sites, as evidenced by the fact that only one interventional radiologist agreed to be interviewed. This may have biased some patients away from the trial as they did not know enough about ablation to make an informed decision.

There is significant research evidence to show that the informed consent process is a point at which many patients are lost to surgical trials. 29 To mitigate this, the current study provided training on how to present the trial and interventions in a balanced way. The written information provided details of a technique that had worked in previous clinical trials, but many interviewees were unaware of the manual, and most refused additional training because at the time of the interview they believed themselves to be competent recruiters or had still not approached a patient. We also planned to record recruitment encounters to monitor discussions about the interventions and study processes to identify good practice and areas where support was needed. However, despite multiple reminders, recordings were received from only two clinicians. These recordings provided no evidence of systematic bias in describing the trial or interventions to patients, but both were experienced researchers.

Our ability to develop a strong plan of action to improve recruitment was hampered by a lack of data. We were able to interview only six surgeons who were involved in recruiting to the trial from 13 UK sites, and received details of only four potential UK patients to interview. This limited our ability to fully comprehend the challenges to the recruitment process at all sites. Aspects of the Quintet Recruitment Intervention30 were embedded into the pilot study, specifically interviews with recruiting staff and audio-recordings of appointments between recruiters and potentially eligible patients. However, if we had undertaken interviews with staff prior to the study opening, we may have identified some of the hidden challenges to recruitment, in particular the influence of the sMDT in decision-making and the misconceptions about the trial inclusion and exclusion criteria, which we could have addressed prior to sites opening. Similarly, if we had undertaken a feasibility study, we would have been better placed to revise our study protocol to address the issues identified. Owing to funding considerations, a decision was taken not to undertake recruitment training at sites during the pilot study and, with hindsight, the difficulties that recruiters experienced with challenging patient preferences could have been addressed if staff had received explicit training to do this. 27

Lessons for the future

From the above discussion, it is evident that the uncertainty about the relative benefits and harms of thermal ablation versus liver resection in high surgical risk patients with CLM will remain for the next decade.

Surgeons and other members of the sMDT should be aware of the ongoing lack of evidence to guide the choice of liver resection or thermal ablation in patients with colorectal liver metastases, especially in those who are in a high surgical risk group. A change in current clinical practice may be necessary. It would be good practice for treating clinicians to provide detailed information about the uncertainty that surrounds the issue including the risks of the interventions and the success of cancer eradication while considering patient preferences.

If we were to set up the LAVA study again, we would ideally conduct interviews with the entire sMDT at each centre before the trial began and identify how we could better integrate the trial into clinical practice, as many of the challenges identified during the pilot study could have been addressed if they had been more clearly understood at an earlier stage. Similarly, earlier patient engagement to understand why many patients had a strong treatment preference despite a lack of evidence to support individual treatments could have helped to develop clinician expertise in understanding and questioning patient preferences. This would also have allowed us to explore possible patient preferences for treatment, and if we had widened the PPI group to incorporate more members we may also have picked this message up from them.

The use of the sMDT to identify eligible patients highlights the importance of getting the whole clinical team on board, and indeed ensuring that patients feeding into the trial have not been primed to receive surgery. We would undertake more intensive work with sites prior to opening to ensure that they understand the trial and that the entire sMDT is supportive.

The pathway for patients through the eligibility assessment and consent was complex. The need for a second visit to determine eligibility for ablation meant that many patients were lost to the study. We would work with centres more closely to see if there are ways of simplifying this process. We would also explore the feasibility of either training nursing staff to counsel and recruit patients or having both a surgeon and an interventional radiologist present when discussing the two interventions.

At the point that sites opened to recruitment we would undertake face-to-face training at each site with recruiters to give them experience of exploring patient preferences. Recruiters were often content to accept these at face value, if they appeared ‘sensible’, and so missed opportunities to engage with the patient in a discussion about the study.

Although we incorporated some aspects of the Quintet Recruitment Intervention approach into the pilot study, this was not well supported by sites and so its usefulness was curtailed. In future, we would ensure that all sites agreed to support the efforts to optimise the recruitment process. Finally, as recruitment at the Dutch sites was better than at the UK sites, with hindsight it would have been informative if we had gained ethics approval to undertake interviews with international colleagues to see if lessons could be learned from their experience.

These lessons are applicable in multicentre RCTs comparing major surgery with non-surgical treatments where the anticipated number of eligible patients per centre is low. It is also useful to get regulatory approvals for qualitative research in other countries in international multicentre RCTs to gain insight into the way that trials are conducted in different countries with a view to increase UK trial recruitment.

Conclusions

The uncertainty remains about the relative benefits and harms of thermal ablation versus liver resection in high surgical risk patients with CLM. Surgeons and other members of the sMDT should be aware of the ongoing lack of evidence to guide the choice of liver resection or thermal ablation in patients with CLM, especially in those who are in a high surgical risk group.

Contributions of authors

Brian Davidson (https://orcid.org/0000-0002-9152-5907) (HPB Surgeon) conceived the study, participated in its design, secured the funding for the study as the principal investigator and commented on the manuscript.

Kurinchi Gurusamy (https://orcid.org/0000-0002-0313-9134) (Evidence-based Specialist) conceived the study, participated in its design, was part of the team that secured the funding and drafted the manuscript.

Neil Corrigan (https://orcid.org/0000-0002-1424-9830) (Statistician) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Julie Croft (https://orcid.org/0000-0001-7586-3394) (Senior Trial Co-ordinator) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Sharon Ruddock (Trial Manager of the study) collected all of the data from the centres, drafted some aspects of the manuscript and commented on the manuscript.

Alison Pullan (https://orcid.org/0000-0001-6060-9123) (Statistician) participated in the study design and commented on the manuscript.

Julia Brown (https://orcid.org/0000-0002-2719-7064) (Statistician and Trial Methodologist) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Maureen Twiddy (https://orcid.org/0000-0002-3794-1598) (Qualitative Researcher) participated in the study design, was part of the team that secured the funding, drafted some aspects of the manuscript and commented on the manuscript.

Jaqueline Birtwistle (https://orcid.org/0000-0002-6083-589X) (Qualitative Researcher) collected qualitative data, drafted some aspects of the manuscript and commented on the manuscript.

Stephen Morris (https://orcid.org/0000-0002-5828-3563) (Health Economist) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Nick Woodward (https://orcid.org/0000-0002-9177-4292) (Radiologist) participated in the study design, was part of the team that secured the funding, drafted some aspects of the manuscript and commented on the manuscript.

Steve Bandula (https://orcid.org/0000-0002-4558-288X) (Radiologist) participated in the study design and commented on the manuscript.

Daniel Hochhauser (https://orcid.org/0000-0001-5522-9281) (Oncologist) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Raj Prasad (https://orcid.org/0000-0001-6061-001X) (HPB Surgeon) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Steven Olde Damink (https://orcid.org/0000-0002-5202-9345) (HPB Surgeon) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Marielle Coolson (https://orcid.org/0000-0002-1608-6668) (HPB Surgeon) assisted with site set-up and organisation, identified and recruited study participants, and commented on the manuscript.

K van Laarhoven (https://orcid.org/0000-0003-2166-7931) (HPB Surgeon) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

Johannes HW de Wilt (https://orcid.org/0000-0001-6773-9668) (Colorectal Surgeon) participated in the study design, was part of the team that secured the funding and commented on the manuscript.

All authors read and approved the final manuscript.

Data-sharing statement

The major aspect of this study is a qualitative study and, therefore, the data generated are not suitable for sharing beyond those contained within the report. Further information can be obtained from the corresponding author.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

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 and Social Care. 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 and Social Care.