Active monitoring, radical prostatectomy and radical radiotherapy in PSA-detected clinically localised prostate cancer: the ProtecT three-arm RCT

Study population

Parts of this section have been reproduced with permission from Martin et al. 36 JAMA 2018;319(9):883–95. Copyright © 2018 American Medical Association. All rights reserved.

In total, 911 general practices were randomised in 99 geographical areas. Of these, 126 were subsequently excluded as ineligible. 34 Consent rates among the remaining eligible intervention (n = 398) and control (n = 387) group general practices were 68% (n = 271) and 78% (n = 302), respectively; 195,912 and 219,445 men registered with these 573 practices were eligible for the intervention and control groups, respectively. After exclusions, the main analysis was based on 189,386 men in the intervention group and 219,439 men in the control group. There were some differences between numbers of participants in the intervention group of this trial and the published ProtecT study population. 25 There were no important differences between the measured characteristics of practices that did agree to participate and those that did not agree to participate. 34 There were also no important differences in measured baseline characteristics between intervention group and control group practices or men, indicating that post-randomisation exclusions did not introduce detectable selection biases.

Among 189,386 intervention group men in the CAP study, 75,707 (40%) attended the PSA testing clinic, 67,313 (36%) had a blood sample taken and 64,436 had a valid test result. Of these 64,436 men, 6857 (11%) had a PSA level of between 3 ng/ml and 20 ng/ml (eligible for ProtecT), of whom 5850 (85%) had a prostate biopsy. Intervention group men who attended PSA testing clinics were sociodemographically similar to non-attenders. 37 Cumulative contamination (PSA testing in the control group) was indirectly estimated at ≈10–15% over 10 years, based on previously reported diagnostic referral rates and ≈20% of follow-up being subsequent to a PSA test undertaken for screening. 38–40

The results of the primary analysis have been published. 36 After an average of 10 years of follow-up, there were 8054 (4.3%) cases of cancer in the screened group and 7853 (3.6%) cases in the control group. Crucially, both groups had the same percentage of deaths (0.29%). This demonstrated that the single PSA test followed by TRUS-guided biopsy diagnostic pathway does not appear to improve disease-specific survival at a median follow-up of 10 years, but longer follow-up is needed. Critically, the testing missed the diagnosis of an important number of lethal cancers, confirming the necessity to improve our diagnostic pathway, perhaps by incorporating imaging in the form of mpMRI upstream of biopsies, in order to triage men at risk of significant disease, and targeting significant cancers.

Rationale for the ProtecT trial

The discovery and wide clinical use of serum PSA for over three decades to detect asymptomatic cancers in fit men, combined with the continuous refinement of curative radical treatments including anatomical RP, has led to a substantial increase in the detection and treatment of early prostate cancer. In countries where PSA testing has been encouraged, overdetection and overtreatment of prostate cancer has prevailed, with unnecessary adverse events caused by treatments and cost pressures for health-care providers. Unlike most developed countries, the UK’s policy on screening for prostate cancer has been not to recommend its use.

The ERSPC21 investigated the effects of screening versus no screening in a large multicentre European cohort. The study reported an advantage in favour of screening, which improved survival and reduced disease progression, but at a substantial cost of overdetection and overtreatment. 31,69 In contrast, the US screening study (PLCO22) showed no beneficial effect of screening for prostate cancer, but suffered serious limitations because of contamination in both groups. 70

In large reported series, RP conferred a 5-year disease-free survival of between 69% and 84%. 71–75 The Scandinavian SPCG-4 RCT comparing surgery and watchful waiting showed an absolute risk reduction in preventing cancer mortality within 8 years of 5% (from 14% to 9%). 76 An update has shown that this absolute difference remains unchanged with longer follow-up of 14 years. 77,78 However, the Prostate Intervention versus Observational Treatment (PIVOT) study in the USA compared RP with watchful waiting in over 700 men with localised PSA-detected prostate cancers,79 concluding that RP did not significantly reduce all-cause or prostate cancer mortality, compared with observation, through at least 12 years of follow-up. There were fewer prostate cancer deaths with RP (21/364 for RP vs. 31/367 for watchful waiting, HR 0.63, 95% CI 0.36 to 1.09; p = 0.09). In other large reported series, radiotherapy conferred a 5-year disease-free survival of 78–94% in combination with hormone therapy. 48–51

Although prostate cancer can be lethal, the majority of men who are diagnosed through PSA testing will not suffer clinically significant consequences from the disease during their lifetime, and evidence that treating such men improves survival or QoL is weak. Consequently, there have been serious concerns that increasing PSA testing in the community is resulting in overdiagnosis, overtreatment and an increasing burden to the NHS. Despite its high incidence and social and economic impact, prostate cancer continues to be under-researched and only a limited number of studies are addressing the issues of screening and long-term comparison of treatment modalities. Following a successful feasibility study, described in the following section, the NIHR HTA ProtecT study conducted by the authors of this report aimed to test the value of a single round of PSA testing in the community, and compared the effectiveness of the three conventional treatment options of AM, RP and EBRT in clinically localised disease at all risk levels. A large prostate cancer detection programme was essential to conduct the main RCT of treatment, in order to diagnose men and counsel them in advance of the diagnostic pathway, which allowed them to become well-informed participants to the treatment trial.

The ProtecT feasibility study

The ProtecT feasibility study, conducted between 1999 and 2001, was undertaken in three clinical centres to provide evidence to underpin the design and conduct of the main ProtecT trial. The methods of the study and its findings have been published in detail. 80 In brief, the feasibility study showed that men aged 50–69 years could be identified in primary care practices, invited to attend a prostate check clinic appointment to discuss having a PSA test and potentially participating in a RCT of treatment, and that sufficient numbers would attend and follow the diagnostic pathway to feed into the proposed treatment trial. By the end of the feasibility study, 8505 men from 18 primary care centres attended clinics (56% of those invited) and 7383 had a PSA test. Of these men, 861 (12%) had a raised PSA level and, after biopsy, 224 were found to have prostate cancer, with 165 clinically localised and eligible for the treatment RCT. These response rates were largely reflected in the main trial. The feasibility study also showed that men were willing to complete detailed questionnaires before the PSA test and biopsy. These questionnaires comprised patient-reported outcomes that were then used in the main trial.

As it was expected that recruitment to the trial would be extremely difficult, the feasibility study was innovatively embedded in a qualitative research study to explore a wide range of issues with the urologists and nurses undertaking recruitment and among men participating in the study. 81 In addition, the feasibility study included a RCT comparing the effectiveness and efficiency of recruiting with nurses or urologists. In total, 167 men with localised prostate cancer were identified and 150 (90%) took part. There was a 4.0% difference between nurses and surgeons in recruitment rates (67% nurses, 71% urologists, 95% CI −10.8% to +18.8%; p = 0.60). Cost minimisation analysis showed that nurses spent longer with patients but surgeon costs were higher and nurses often supported surgeon-led clinics. We concluded that nurses were as effective and more cost-effective recruiters than surgeons, and so nurses became the major recruiters in the main RCT.

The qualitative research involved in-depth interviews with 39 men before and after the PSA result, case studies of four men interviewed several times after the PSA result as they progressed through the study, 20 audio-recorded recruitment appointments and subsequent interviews, and 15 other audio-recordings of recruitment appointments with nurses or urologists attempting to recruit men to the treatment trial. Findings from these data included:

• Men perceived the offer of PSA testing as an opportunity to discover cancer early, most did not want to consider the implications of a positive result and most expected to receive a negative test result. 82

• Although most men could recall and understand the reason for the treatment RCT, they did not always find it acceptable, and most found the concept of randomisation difficult to accept. However, those who understood and were able to believe in clinical equipoise were most likely to consent to randomisation. 83 These findings formed a detailed basis for developing suitable and effective trial information in relation to clinical equipoise in the main RCT.

• In recruitment appointments, surgeon and nurse recruiters initially found it difficult to present the treatment options equivalently and many were not in equipoise, particularly in relation to the non-radical treatment group. These findings were documented and fed back to recruiters. A plan to improve recruitment was designed by Jenny L Donovan and implemented by Freddie C Hamdy: to change the order of presenting treatments to encourage greater emphasis on equivalence, avoiding terms commonly misinterpreted by patients (such as ‘trial’ and ‘watchful waiting’) and describing a redefined and reconceptualised ‘active monitoring’ group. Presentations about the plan and workshops on recruitment practice were followed by the randomisation rate increasing from 40% to 70%. In further analyses of appointments, treatments were described more clearly and with greater balance, patients became better informed and the three-group trial (including AM) became the preferred design. Embedding the trial recruitment process within this innovative and detailed qualitative research study enabled efficient recruitment that was acceptable to patients and clinicians in the three feasibility study centres. 81

The feasibility study thus showed that, against all expectations, it was possible to introduce PSA testing in primary care practices, implement a standardised diagnostic pathway, identify sufficient numbers of men with clinically localised prostate cancer and mount a full-scale three-group RCT to evaluate the major treatment options: radical surgery, radiotherapy and a reconfigured ‘active monitoring’.

Aim

The overarching aim of the main ProtecT trial was to provide robust evidence of the comparative treatment effectiveness of the three conventional treatment options in the management of clinically localised prostate cancer.

Objectives

• To evaluate the comparative treatment effectiveness of the three conventional options (AM, RP and RT) for men with clinically localised prostate cancer, with a first analysis at the 10-year median follow-up.

• To assess QoL measures and patient-reported outcomes related to the three treatment options.

• To inform patients, clinicians and policy-makers about the optimal management of patients with clinically localised prostate cancer.

• To develop a comprehensive biorepository of biobanked material donated by patients, associated with an electronic clinicopathological database for conducting effective translational prostate cancer research.

Trial design

ProtecT is a multicentre, pragmatic, parallel-group RCT that compared RP, external beam three-dimensional (3D)-conformal radiotherapy and AM for clinically localised prostate cancer detected through population-based PSA testing in primary care. The trial was conducted in nine UK centres based at hospitals. A feasibility trial conducted at three centres between June 1999 and September 2001 preceded the main trial, which recruited until January 2009, with follow-up for the primary analysis completed in November 2015. The primary outcome of prostate cancer-specific mortality was evaluated at a median of 10 years’ follow-up. Secondary outcomes analysed included disease progression and patient-reported outcomes, as well as cost-effectiveness.

Ethics approvals and research governance

The trial was approved by the NHS Multicentre Research Ethics Committee (Trent Multicentre Research Ethics Committee reference number 01/04/025). An independent Data Monitoring Committee (DMC) oversaw the trial to the primary outcome analysis and met annually. The Trial Steering Committee (TSC), with an independent chairman and seven independent members, monitored the conduct and progress of the trial annually. The DMC would have recommended changes to the TSC if clear evidence (of the order of p < 0.001) of a positive or negative balance of risks and benefits emerged for one intervention in comparison with the others. Study training programmes (described in Research nurses’ role in recruitment, randomisation and follow-up and The ProtecT recruitment story) and on-site monitoring visits were used to standardise trial conduct. 84 The main trial is registered with Controlled Clinical Trials ISRCTN20141297 (feasibility trial: ISRCTN08435261). The protocol is available at www.journalslibrary.nihr.ac.uk/programmes/hta/962099/#/.

Study setting and participants

The feasibility study recruitment was conducted in three English cities (in 24 primary care centres linked to three hospitals) from June 1999 to September 2001. The main phase of recruitment was conducted from October 2001 to January 2009 in nine cities (seven in England, one in Scotland and one in Wales) where around 100,000 men were recruited in primary care.

The CAP trial (ISRCTN92187251)34 commenced in 2001; it randomly assigned 911 primary care centres in eight UK centres to undertake either the ProtecT trial or standard UK NHS management to assess population-based screening for prostate cancer (Figure 1).

FIGURE 1.

Flow chart of the ProtecT and CAP trials.

A written invitation was sent by 337 primary care centres participating in the ProtecT trial to registered men aged 50–69 years (inclusion criterion). Trial exclusion criteria were a previous malignancy (apart from skin cancer), renal transplant or current renal dialysis, major cardiovascular or respiratory comorbidities, bilateral hip replacement or an estimated life expectancy of < 10 years. Men who responded received a ProtecT patient information sheet and an appointment with a specialist research nurse, who explained the complexities of PSA testing, assessed trial eligibility and sought written informed consent. Previous PSA test results were checked in the medical records (not an exclusion criterion). On postal receipt of a second consent form, total PSA level was analysed at hospital laboratories audited by the NHS External Quality Assessment Service.

Diagnosis of prostate cancer

Participants with a PSA concentration of ≥ 3.0 µg/l were invited to attend secondary care centres within the nine participating cities for a physical and digital rectal examination and standardised 10-core TRUS-guided prostate biopsies. Participants with an initial PSA concentration of ≥ 20.0 µg/l at diagnosis were excluded because of the high likelihood that they had more advanced cancer. Patients were staged using a combination of digital rectal examination, PSA concentration, TRUS-guided biopsies and isotope bone scanning (if their PSA level was ≥ 10 µg/l). Magnetic resonance imaging (MRI) was used for staging at the discretion of individual investigators because this imaging technique was not available in all centres during recruitment. Men diagnosed with clinically localised prostate cancer and deemed fit for radical treatment received a ProtecT treatment patient information sheet and were subsequently invited to discuss randomisation with the specialist nurses. Men with a PSA concentration of ≥ 10µg/l or a Gleason score of ≥ 7 points underwent an isotope bone scan to exclude metastatic disease. Men who initially had benign biopsy samples or who were diagnosed with locally advanced or advanced prostate cancer were managed in the NHS and excluded from the trial. Men with a benign first biopsy sample and a free-to-total PSA ratio of < 11%, or atypical small acinar proliferation or high-grade prostatic intraepithelial neoplasia, were offered further biopsies; if these repeat biopsy samples were benign, these men were managed in primary care and excluded from the trial. No further trial follow-up took place after the one round of PSA testing or identification of cancers after referral to the NHS. Histopathologists at each site reported pathology findings on standardised forms and participated in trial quality-control processes and those of the NHS Uropathology External Quality Assessment Scheme. At baseline, height and weight were measured and blood samples were taken (whole blood, plasma and serum). Biopsy and prostatectomy tissue, if relevant, were biobanked for participants who also entered the ongoing linked translational Prostate Mechanisms of Progression and Treatment study (see Chapter 7).

Interventions

Participants were randomised to one of three treatments: AM, radiotherapy or RP.

In the AM group, the aim was to avoid immediate radical treatment while assessing the disease over time, with radical treatment offered if disease progression was evident. PSA concentrations were reviewed every 3 months in the first year and twice yearly thereafter (frequency was changed as indicated). The specialist nurses also met with participants yearly to assess their overall health and discuss graphical displays of PSA results and any concerns raised, overseen by each centre’s local clinical investigator. Changes in PSA concentrations were assessed at each visit, and a rise of ≥ 50% during the previous 12 months triggered repeat testing within 6–9 weeks. If the PSA concentrations were persistently raised, or the patient had any other concerns, a review appointment was made with the centre urologist for discussion of further tests including re-biopsy and all relevant management options. A site monitoring and review team comprising trial research nurses and the trial manager visited sites annually; these visits included observation of the nurse-led AM appointments as per the protocol. 85

In radiotherapy, neoadjuvant androgen suppression was given for 3–6 months before and concomitantly with 3D-conformal radiotherapy delivered at 74 Gy in 37 fractions at each of the nine centres. Quality assurance followed RT01 trial procedures. PSA concentrations were measured every 6 months for the first year and then yearly. The study oncologist held a review appointment with participants if the PSA concentrations rose by ≥ 2.0 µg/l post nadir or if concerns were raised about disease progression. Management options were discussed, including monitoring, tests and salvage, radical or palliative treatments as indicated. 34

The predominant approach for RP was open retropubic with individual-level quality assurance to published standards at each of the nine centres. 86 Participants with a baseline PSA concentration of ≥ 10 µg/l or a biopsy Gleason score of at least 7 points received bilateral lymphadenectomy. Postoperatively, PSA concentrations were measured every 3 months for the first year, every 6 months for the subsequent 2 years and then yearly. Adjuvant radiotherapy was discussed and offered to patients with positive surgical margins or extracapsular disease. The centre urologist held a review appointment with participants if their postoperative PSA concentrations reached ≥ 0.2 µg/l to discuss adjuvant radiotherapy.

In all treatment groups, ADT was offered when serum PSA reached a concentration of 20.0 µg/l, or less if indicated. Imaging of the skeleton was recommended if the serum PSA level reached 10.0 µg/l, using isotope bone scintigraphy, plain radiographs and MRI as necessary.

Primary and secondary outcomes

Sample size

Before the start of the trial, a sample size target of 1434 randomly assigned men (478 in each group) was identified as sufficient to estimate the absolute difference in mortality probability between two treatment groups with a 95% CI of ±0.045, on the basis of an assumed prostate cancer mortality risk of 15% at 10 years, consistent with prostate cancer-specific mortality in men aged 55–69 years, with clinically detected disease managed conservatively at that time and a difference that would be deemed clinically significant by the NHS. The pilot study recruitment data were used to calculate the number of sites and duration of recruitment needed to meet the sample size target. However, more-recent data suggested that disease-specific mortality with non-radical treatment was likely to be closer to 10% at 10 years because of improvements in disease management. 34 As a result, the DMC advised in 2008 that recruitment should continue to the planned end date, with 1590 men (530 per group) expected to be randomly allocated by that point. This sample size would enable a 46% relative reduction in prostate cancer mortality to be detected with 80% power at a 5% significance level for a pairwise comparison of a radical treatment with AM. This calculation assumes a 10% prostate cancer-specific mortality at 10 years with AM, and hence a 5.4% risk with radical treatment (an absolute difference very similar to the margin of error specified in the first calculation). These sample size targets are based on differences in and ratios of risk rather than the HRs planned for the primary analysis, because the resulting calculations are simpler and more flexible. When a high survival rate is expected, calculations based on risk ratios will be a close approximation to those based on HRs.

Randomisation

Patients were randomised on an equal basis (1 : 1 : 1) to AM, surgery or radiotherapy by a remote randomisation service. Allocation concealment was assured by the treatment being assigned only after the nurse telephoned the research centre and participant identifiers and key baseline data were logged on the computerised randomisation system held at the University of Bristol. Allocations were computer generated as required for each participant, originally using Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) functions and subsequently in C++ by an independent programmer, stratified by site with stochastic minimisation (maintaining a random element to the allocation) to improve the balance across the groups in relation to age at primary care patient-identification date, Gleason sum score (< 7, 7 or 8–10 points) and mean of baseline and first biopsy PSA results (< 6.0, 6.0–9.9 or > 9.9 µg/l). The allocation was revealed after the entry of participant details and given to the participant by the nurse. Men who declined randomisation were offered trial follow-up and formed a comprehensive cohort. 95

Blinding

Clinicians, participants and researchers were not masked to group assignment. All investigators remained blinded to outcomes by group throughout recruitment and analysis. The statistical analysis plan was written by the senior statistician and was agreed by the trial team post recruitment end but prior to any statistical analysis. 96 The primary outcome was assessed by a committee independent of the trial team. 97

Statistical methods and analysis plan

The primary analyses utilised an intention-to-treat (ITT) approach comparing three treatment groups as allocated. At a median of 10 years of follow-up (November 2015), the primary outcome measure of prostate cancer mortality was compared between treatment groups using a survival analysis (Cox proportional hazards regression model) adjusted for stratification and minimisation variables. The estimated relative treatment effect for each pairwise comparison of treatments was captured as a HR, and presented with a 95% CI. Pairwise significance tests were planned if a test of an equal 10-year disease-specific mortality risk across all three groups yielded a p-value of < 0.05. 98 This was used for event-based secondary outcomes (i.e. grouped analyses of definite, probable or possible prostate cancer, all-cause mortality and metastatic disease).

Pairwise comparisons of symptom burden utilised multilevel models for repeated measures to estimate the average treatment effect over the median 10-year follow-up. Further analyses will investigate the relative burden between treatment groups over time. Prespecified subgroup analyses will investigate whether treatment effectiveness in the reduction of prostate cancer-specific mortality is modified by clinical stage, Gleason score, age or PSA concentration using stratified analyses for descriptive statistics and by formally including interaction terms in the relevant regression models. Secondary analyses will estimate the efficacy of radical treatment versus AM in the reduction of prostate cancer mortality in individuals who complied with their allocated treatment, by using a method to derive an unbiased estimate in parallel with the per-protocol analysis originally specified in the trial protocol. 99,100

Data from the recruitment, diagnostic and randomisation phases are summarised, and categorisation of continuous variables was based on either clinical thresholds (e.g. for PSA) or the aim of equal group sizes (other measures). Resident area-based material and social deprivation scores (the proportion of people living in an area of material deprivation) were derived using lower super output areas, each equating to around 1500 residents, for England, Scotland and Wales separately. Analyses were carried out in Stata^® version 10 (StataCorp LP, College Station, TX, USA).

Summary of changes to the protocol

The main changes in chronological order were:

• An alternative two-group randomisation between RP and radiotherapy was stopped in 2003 by the TSC because of limited uptake (24 participants in total).

• An additional exclusion criterion of bilateral hip replacement (precluded radiotherapy) was added in 2003.

• A pilot study of men aged 45–49 years in one centre that recruited around 1300 participants between 2003 and 2005 was discontinued because of lower recruitment and identification of mostly low-risk disease. 101

• There was a recruitment extension from June 2006 to June 2008 resulting from contractual delays in opening centres in the main trial.

• Additional PROMs were added for bowel effects [Expanded Prostate Cancer Index Composite (EPIC) added in 2005] and cancer-specific QoL (EORTC-QLQ Q30 in 2007).

• Sample size calculation changed on the advice of the DMC and TSC, as described earlier in 2008.

• The 5-year survival analysis was removed in 2009 on DMC advice.

• The chief investigator (and trial sponsorship) moved from the University of Sheffield to the University of Oxford in 2010.

The role of research nurses in recruitment, randomisation and follow-up

Recruitment in the feasibility study

Figure 2 shows the flow of participants through the ProtecT study. The recruitment process through the prospective cohort of PSA testing and prostate cancer diagnosis has been described elsewhere. 95,104 This section details the process and outcome of recruitment to the ProtecT trial from the feasibility study until completion in January 2009.

The ProtecT feasibility study was undertaken in three clinical centres from 1999 to 2001 to explore the issues that were expected to make recruitment challenging, if not impossible. The feasibility study was innovatively embedded in a qualitative (ethnographic) study of the perspectives of recruiters and potential participants, identifying obstacles and challenges to successful recruitment. Recruitment in the feasibility study was not easy, but it was eventually successful. The findings produced several recommendations for the main trial: that research nurses should undertake most recruitment activities, supported by urologists who would conduct the ‘eligibility’ assessment appointments; that the comparison should be between three treatments – surgery, radiotherapy and ‘active monitoring’ (as reconceptualised by the qualitative research); that AM should be presented first in the list of treatments; and that nurses would be trained to present the study in a carefully balanced way, avoiding terms such as ‘trial’ and ‘random’ and attempting to explain equipoise and randomisation clearly. 80,83,105 At the beginning of the feasibility study, recruitment struggled, with only 30–40% of eligible patients consenting to randomisation, and 60–70% accepting the allocation, but this rose to a randomisation rate of 70% (95% CI 62% to 77%) by the end of the feasibility study in May 2001, with around 70% accepting the allocation (Table 2).

Recruitment to the main ProtecT trial: 2001–9

The feasibility study had been undertaken in three clinical centres; the main trial required the addition of six further clinical centres. The target for accrual was to randomise 1590 men with localised prostate cancer (including the 146 recruited in the feasibility study). The clinical centres were set two simultaneous targets:

1. Achieve randomisation of at least 60% of eligible patients.

2. Ensure that participants were well informed so that at least 70% would accept the random allocation.

The recruitment intervention that had been successful in the feasibility study was further developed and extended into a complex intervention in the main trial. 102 Initially, the following actions were proposed as part of the intervention, building on the findings from the feasibility study:

• regular training for nurses

• centre reviews if study targets were not met

• provision of documents to support recruitment

• individual feedback to recruiters as required.

In addition, the lead investigators in the clinical centres were interviewed to capture their views about participating in the trial and confidence with eligibility assessments and recruitment, and their ‘eligibility’ appointments were audio-recorded if recruitment rates fell below the target rate. Nurses were asked to audio-record all of their recruitment appointments. Patient consent for audio-recording was almost universal. Recordings were selected for scrutiny when staff commenced recruitment or when rates of randomisation or acceptance of allocation fell below study targets. Some recordings were selected at random for trial monitoring, or consecutively for analysis in particular studies. Audio-recordings and interviews were analysed in accordance with methods of constant comparison based on grounded theory, with thematic methods used for interviews and a mixture of content, thematic and targeted conversation analysis methods used for audio-recordings. Centres failing to reach both study targets were selected for review, in which J Athene Lane undertook an audit of consent procedures and Jenny L Donovan completed a focused qualitative analysis of recruitment practice.

Figure 3 shows numbers of patients agreeing to randomisation over time and Table 3 shows recruitment over time according to the rates of randomisation and immediate acceptance of allocation. By the end of 2001, although the randomisation rate was ahead of the target of 60%, the rate of acceptance of allocation had slipped to 64%. The TSC advised that all efforts should be devoted to improving the acceptance rate, and suggested the need to aim to reach a rate of 75–80%. This was investigated in detail through reviews of two recruiting centres with recruitment considerably lower than the targets: centre A with 45% randomisation and 55% acceptance rates, and centre B at 50% and 65%, respectively. Details of the reviews have been reported elsewhere. 102 In brief, after these reviews, rates of randomisation and acceptance of allocation increased markedly to 86% and 78% for randomisation and 67% and 87% for acceptance, with evidence of improvement beyond chance after 12 months. These changes were sustained at 24 months, although numbers were small. In one of the centres, most of those declining randomisation chose surgery, and so training and feedback were targeted towards clearer explanation of all treatments. In the other centre, the nurse recruiters referred to prostate cancer as ‘early’, leading to men expressing a preference for AM (radical treatment was seen as using ‘a sledgehammer to crack a nut’), with nurses then accepting their preference without further discussion. These issues were dealt with by avoiding the term ‘early’ and referring instead to mostly ‘small’ or ‘slow-growing’ cancers, asking patients to keep an ‘open mind’ while they listened to information about treatments, and by acknowledging the need for nurses to explore the reasons for patients’ preferences.

FIGURE 3.

Numbers of patients agreeing to randomisation over time. Q, quarter.

Information about these issues was then included in nurse training sessions that were held in March 2002, February and June 2003, January and October 2004 and annually thereafter. In October 2002, the training was supplemented with a document of ‘tips’ for recruitment, which was sent to all nurse recruiters and urologists carrying out eligibility appointments. This document advised:

1. mentioning the purpose of the study and randomisation early in the appointment

2. avoiding the use of terms such as ‘early’ to avoid misunderstanding

3. describing the treatments succinctly and with balance, including potential advantages as well as adverse effects

4. eliciting and exploring participants’ treatment preferences, particularly if based on incorrect information (such as radiotherapy resulting in hair loss)

5. providing details about what would happen next after advising the participant of the treatment allocation.

A further document (‘Tips 2’) was produced in December 2004, focusing more on the structure and content of appointments, with the following advice:

• Explain the purpose and rationale of the study, inform patients about advantages and disadvantages of treatments so that they could perceive them as equivalent in terms of outcomes in the long term, obtain consent from randomisation only after checking that the participant was likely to accept all three treatments and ensure that they were comfortable with the conduct and outcome of the appointment.

• Conduct appointments in three basic stages: opening (points 1–3 above, and empathising with the patient’s situation), process (point 4 above) and ending (ensure that all preferences are addressed, explain purpose and advantages of randomisation, only gain consent if patient indicates willingness to accept all options – or arrange preferred treatment).

• Examples of difficult situations were given.

• Essential things to do and avoid were given.

As can be seen in Table 3 and Figure 4, the rate of acceptance of allocation increased sharply in 2002 and continued to increase, reaching 70% in 2003 and rising to 80% in 2005, before stabilising at that level to the end of recruitment. The rate of randomisation began to decline gradually from its high point of 69% in 2003 as the acceptance of allocation increased. Both original targets were thus met from 2003 onwards, as was the revised TSC target for acceptance of allocation from 2005 onwards.

FIGURE 4.

Randomisation and acceptance of allocation over time.

Years

We also investigated acceptance of allocation by group. Figure 5 shows how the low level of acceptance in 2001 was related to the radical treatments, particularly radiotherapy. Nurses were encouraged to visit oncology units and improve information provision about radiotherapy in addition to providing more balanced information about all three treatments. As can be seen in Figure 5, the acceptability of allocation to both radical treatments became similar to but remained somewhat lower than the rate for AM, reflecting the differences in the immediate consequences of these options.

FIGURE 5.

Acceptance of allocation by group over time.

Development and refinement of the ProtecT recruitment intervention

The centre reviews, nurse training, provision of ‘tips’ documents and individual confidential feedback constituted the ProtecT trial recruitment intervention. The intervention was further developed through two qualitative research studies undertaken alongside the recruitment:

1. A qualitative analysis of audio-recordings of recruitment appointments revealed that their structure was mostly either recruiter led or participant led. Recruiter-led appointments followed a more typical script of issues to be covered, whereas the participant-led appointments were driven by the questions that the participant wanted to raise. The structure was associated with how well it was possible to judge whether or not participants understood the information provided by the recruiters and whether or not they reached equipoise. In the recruiter-led appointments, patients tended to spend most of the time listening and making only occasional utterances, whereas they were able to contribute much more in participant-led appointments. The qualitative analysis suggested that informed consent was enhanced by tailored information provision including strategic use of open questions, pauses and ceding the floor to participants in the interaction. These techniques facilitated detailed and systematic exploration of each participant’s concerns and facilitated truly informed consent. In contrast, interruptions and failing to address participants’ questions were unhelpful. 106 These techniques were presented to recruiters during training and were taken up by many recruiters over time.

2. A qualitative analysis of audio-recordings of 93 consecutive recruitment appointments explored treatment preferences expressed by men, reactions to these expressed preferences by recruiters and what the outcome of the appointment was (i.e. consent to participate in the trial or selection of a treatment). This study showed that treatment preferences were more complex and dynamic than previously assumed. Most participants expressed views about treatments early in appointments, ranging on a continuum from hesitant to well-formed opinions. As recruiters explored men’s views and provided detailed evidence-based treatment and study information, some men opted for their preference, but many became uncertain and open to RCT recruitment, often accepting a different treatment from their original ‘preference’. It was clear that expressed preferences were not as strong as they originally seemed, and could be explored. Unexpectedly, we concluded that the discussion of treatment preferences did not act as a barrier to recruitment but actively enabled many men to express their concerns and reach an informed decision that often included RCT participation and acceptance of allocation, even if this was not to their originally expressed preference. 107 This nuanced understanding of expressed preferences was used in training in role plays, and there were detailed discussions about the importance of gently exploring preferences to ensure that participants were fully informed about the treatment options and could then make an informed decision about trial participation or choosing a treatment. Discussions about appropriately exploring and addressing treatment preferences took up a large proportion of training sessions, and advice on this issue formed a substantial element in the ‘tips’ documents.

ProtecT trial recruitment outcome and completion

The original projection of recruitment for ProtecT assumed that the six centres added after the feasibility study would come on stream promptly and recruit at the same level as the three feasibility centres. Unfortunately, 2001/2 was when new rules came into operation in relation to research governance in NHS organisations, and this severely delayed the opening of the new centres. At this point, recruitment fell behind target. A revised projection was made in 2005, when an extension was also applied for. Once all centres were operating, recruitment increased significantly and met and then exceeded the target (see Appendix 1, Figure 28).

By the end of the recruitment period in early 2009, there were 2664 men with clinically localised prostate cancer who were eligible for the trial and who were informed about it by ProtecT study research nurses. Of these, 1643 (62%) agreed to be randomised and 997 selected a treatment of their choice (24 had opted for the discontinued two-group trial). Among the 1643 patients who agreed to be randomised, 1336 (81%) accepted their allocation immediately. The vast majority of these men then went on to have the treatment as allocated: 1278 (78% of those randomised) commenced the allocated treatment within 9 months. The original targets of randomising at least 60% of eligible patients and for at least 70% to accept the allocation were thus achieved.

Subsequent development of the recruitment intervention for use in other randomised controlled trials

As the ProtecT recruitment intervention showed promise, its methods were codified and the transferability of the intervention was assessed in four further RCTs in the Medical Research Council Quartet (Qualitative Research to Improve Recruitment to RCTs) study54 and many more independent RCTs. Collaborations were often initiated by trial investigators who anticipated that they would encounter recruitment obstacles in the RCTs because comparisons were controversial and/or included very different treatments (e.g. current standard treatment vs. ‘less’ or ‘no’ treatment, or radiotherapy vs. surgery). These collaborations usually included some aspects of the intervention from the outset of the RCT’s recruitment, often during feasibility or pilot phases. Other collaborations began part way through the RCT recruitment period by the trial investigators or following encouragement by the funder in response to recruitment falling short of targets. Some of these RCTs were at risk of early closure, and most had already applied for funded extensions and/or revised their target sample sizes. These different contexts – the intervention integrated from the outset or inserted into the recruitment at a later stage – provided interesting and sometimes different challenges that further encouraged the development of aspects of the intervention.

The four RCTs in the Quartet study showed that the intervention could be transferred in some circumstances, particularly where aspects were integrated with the main RCT, as had been possible in ProtecT. 108–110 At this stage in the development of the intervention, involvement seemed to lead to improvements in recruitment in most RCTs. 111–114 However, there were also some RCTs in which insurmountable issues were found, leading to the early closure of the RCT. 109,115,116 These were difficult for the researchers and chief investigators, although the intervention did clearly identify the source of the insurmountable difficulties: a lack of equipoise was evident among clinicians involved in all of these trials, and there were also insufficient numbers of eligible patients or problems with the timing of intervention delivery. 109,115,116 These collaborations were certainly instrumental in refining the content of the intervention. Some collaborations at this stage also enabled further methodological development, alongside that being undertaken in ProtecT (as described previously), for example measurements of timing of recruitment appointments and content related to particular interventions or the study itself, and the identification of methods of communication. 117,118

There were considerable challenges in integrating the intervention part way through recruitment, especially in relation to research governance approvals, and opportunities to optimise recruitment were often constrained by delays and difficulties in data collection, most notably in obtaining audio-recorded recruitment appointments. 108 These audio-recorded appointments underpinned the understanding of and then feedback and training delivered in ProtecT, but a routine audio-recording process proved difficult to operationalise in other RCTs. Interviews with recruiters, scrutiny of trial documentation, focus groups and informal discussions with the TMGs were the key sources of data collection in most of these trials, and it became clear that this was limited compared with the ProtecT responsive and tailored support.

The emergence of the Quintet Recruitment Intervention

A large number of data were collected and analysed during the application of the ProtecT intervention and its translation in five further RCTs. These data were synthesised in two papers published in 2014. 119,120 The first paper showed how RCT recruiters readily identified organisational difficulties, fewer than expected eligible patients and patients’ treatment preferences as the key barriers to recruitment. 119 These were ‘clear obstacles’. As recruiters described their experiences of recruitment and data from appointments were synthesised, several otherwise ‘hidden challenges’ related to their roles as researchers and clinicians emerged, imbued with discomfort and emotion. The synthesis went on to show that doctors were most uncomfortable about aspects of patient eligibility and the effectiveness of interventions, whereas nurses were more anxious about approaching potential RCT participants and conflicts between their research and clinical responsibilities. It was notable that recruiters were unaware that their views could have an impact on recruitment. These hidden challenges were not shared with RCT chief investigators. The synthesis also showed how the feedback and training developed in the intervention could enable most recruiters to become more comfortable with key RCT concepts including equipoise, uncertainty, patient eligibility and randomisation. 119 In the second paper, the discomfort and emotion in relation to community equipoise and ‘hunches’ about particular treatments and patients was explored in greater detail. 119 Both papers concluded that recruitment to RCTs is a fragile process that is difficult for doctors and nurses intellectually and emotionally, and that they required support.

The synthesis, including and initiated by ProtecT, led directly to the formalisation of the intervention as the Qualitative Research Integrated in Trials (Quintet) Recruitment Intervention (QRI). 54 By this time, the intervention had been developed further in 13 RCTs in addition to ProtecT. The final version of the QRI uses mixed research methods in two major phases: phase I is to understand recruitment as it happens and then phase II is for developing and implementing a plan of actions to address identified difficulties and optimise informed consent and recruitment in collaboration with the RCT chief investigator and clinical trials unit. The QRI can be integrated into the feasibility/pilot or main phase of a RCT to prevent difficulties developing and to optimise recruitment from the start, or applied in a RCT experiencing shortfalls with a view to rapidly improve recruitment and informed consent or to gather evidence to justify RCT closure.

The QRIs (or their developmental precursor) have now been used in 30 RCTs and have produced a number of insights into recruitment. These include how recruiters can be trained to explore patients’ treatment preferences;121 how equipoise is conveyed during recruitment and how it can facilitate or hinder recruitment;122 the development of a measure of informed consent;123 how training can increase recruiter confidence;124 and the value of collecting data logging patients at the various stages of screening, eligibility, approach and recruitment. 125

Baseline characteristics of ineligible and eligible participants

There were over 10,000 ProtecT participants with a raised PSA level (around 11% of recruited participants), of whom 8566 in the main trial had at least one biopsy and about 3000 were diagnosed with prostate cancer; of these 3000, around 400 had advanced disease (12%) (see Figure 2). The characteristics of men with prostate cancer were similar to the characteristics of those without diagnosed cancer, except for the expected greater frequency of a family history of prostate cancer and higher initial PSA levels (Table 4).

Eligible participants with localised prostate cancer who selected their treatment had similar measured clinical and sociodemographic characteristics to the characteristics of those who accepted randomisation, except that they were less likely to reside in an area of material deprivation [odds ratio (OR) of increased deprivation in randomised vs. non-randomised participants 0.74, 95% CI 0.58 to 0.94] and may have differed in unmeasured characteristics (Table 5). These characteristics have been reported previously, including the PROMs for men with advanced prostate cancer or health-related exclusions. 95

Demographic and clinical features of randomised participants

The median age of randomly assigned participants was 62 years (range 50–69 years), and the median PSA level was 4.6 µg/l. Most randomised participants had clinical T1c disease and a Gleason score of 6 points. The distributions of age, PSA levels, Gleason scores and disease stage were well balanced across the randomised groups (Table 6).

Baseline levels of patient-reported outcomes for randomised participants

Numbers analysed and return of data-collection forms

Prostate cancer-specific and all-cause mortality

The independent CoDE Committee ascertained seven definite prostate cancer-specific deaths and one probable prostate cancer-specific death in the AM group, three definite and two probable prostate cancer-specific deaths in the RP group and four definite prostate cancer-specific deaths in the RT group (Table 8 and Figure 7). Prostate cancer-specific survival was > 98.8% in all groups, and there was no evidence of difference between the three randomised groups (log-rank test p = 0.48). The HR of prostate cancer-specific mortality for the RT group was 0.45 (95% CI 0.14 to 1.47) compared with AM and 0.80 (95% CI 0.22 to 2.99) compared with RP; for RP compared with AM, it was 0.56 (95% CI 0.19 to 1.67) (Figure 8). Subgroup analyses showed no evidence of any subgroup modifying the relative effectiveness of the three treatments in terms of prostate cancer mortality (see Table 2). All-cause deaths were evenly distributed across the treatment groups (likelihood ratio test p = 0.87). Of the patients with ascertained prostate cancer-specific deaths, 7 out of 17 had baseline features whereby most international guidelines would recommend AS, and 12 out of 17 had received AM (three out of five had been randomised to RP and two out of 45 had been randomised to RT), as shown in Table 9.

FIGURE 7.

Kaplan–Meier estimates of deaths from prostate cancer in the AM, RP and RT treatment groups.

FIGURE 8.

Kaplan–Meier estimates of death from any cause in the AM, RP and RT treatment groups.

Causes of death unrelated to prostate cancer over a 10-year median follow-up, ascertained by the CoDE Committee, are listed by allocation group in Appendix 1, Table 26.

Disease progression

A total of 204 men showed progression including distant metastases (Table 10 and Figure 9), which was higher in the AM group than in the RP and RT groups (AM, n = 112; RP, n = 46; and RT, n = 46; p < 0.001). Evidence of disease progression included the presence of metastases (AM, n = 33; RP, n = 13; and RT, n = 16; p = 0.004), clinical T3 or T4 disease (AM, n = 79; RP, n = 24; and RT, n = 21) or initiation of long-term ADT (AM, n = 47; RP, n = 26; and RT, n = 30), with evidence of more than one criterion for some men.

FIGURE 9.

Cumulative proportion of participants free of disease progression in the AM, RP and RT treatment groups.

Treatment complications

There were no deaths related to RP; nine men suffered thrombo-embolic or cardiovascular events, 14 required more than 3 units of blood transfused, one suffered a rectal injury and nine required intervention for anastomotic problems. There were three deaths unrelated to prostate cancer within 90 days of completing RT, and no cases of radiation toxicity requiring major intervention.

Numbers needed to treat

From these data, compared with AM, 178 and 137 men would need to be treated with RP and RT, respectively, to avoid one prostate cancer death; 27 and 33 men would need to be treated with RP and RT, respectively, to avoid one patient progressing to metastases; and nine men would need to be treated with either RP or RT to avoid one patient developing clinical disease progression.

Analysis of disease characteristics in progressing and non-progressing patients

We have reviewed the baseline risk classification of all ProtecT participants using the D’Amico criteria132–134 and other clinicopathological features, and compared men who developed disease progression with those whose disease remained stable to determine if there were any features associated with clinical outcomes. Furthermore, we report detailed pathological information for men who received surgery and the impact of these features on outcome. The results of these recent analyses are presented in Baseline clinicopathological features of participants with disease progression.

Men who had started a protocol treatment within 12 months were included in this analysis (n = 1585 of 1643 men). Patients were categorised as receiving AM if they had started formal treatment under this name and had at least two PSA tests within 1 year of eligibility (diagnosis). Patients were categorised as receiving RP or RT if they started treatment within 12 months and, for RT, completed it within 15 months.

Pathology

Expert histopathologists at each centre reported prostate biopsy and RP pathology findings on standardised pro formas. 135 RP specimens were whole embedded and the volume of the tumour was calculated. 135,136 Surgical margins were recorded as positive if the tumour was seen at an inked margin and classified as apical, basal, intraprostatic or extraprostatic. The ProtecT pathology group conducted internal audits of both biopsy cores and RP specimens to minimise variation in assessment.

Progression and metastasis

Metastatic prostate cancer was defined as the presence of bony, visceral, or lymph node metastases on imaging, or a PSA level of > 100 ng/ml. Clinical progression was defined as the presence of any of (1) evidence of metastases, (2) progression to clinical T3 or T4 disease, (3) long-term ADT, (4) ureteric obstruction, (5) rectal fistula or (6) need for a urinary catheter owing to local tumour growth. Primary treatment failure after RP was defined as a PSA level of ≥ 0.2 ng/ml 3 months after surgery, and primary treatment failure after RT was defined in accordance with the Phoenix Consensus Conference recommendations. 128

Statistical analysis

All analyses were carried out using Stata version 14.2. Logistic regression was used to estimate the odds of progression for baseline characteristics, adjusting for age and treatment received when appropriate. When model assumptions could not be met, the log of the exposure was used. The odds of progression were calculated for participants with baseline Gleason scores of ≤ 3 + 3 points. Gleason score values for the prostate biopsy cores were derived using the sum of the primary Gleason grade and the greater of the secondary or tertiary Gleason grades in order to reflect current practice. Aggregate and maximum lengths of tumours were measured in millimetres. The aggregate length of tumours was calculated as the summation of the lengths on the right, left and unknown sides and targeted biopsy, adding 0.5 mm for each small additional tumour (up to four). The Cox proportional hazards model and likelihood ratio test were used to test the interaction between treatment and each of baseline Gleason score (≤ 3 + 3, 3 + 4, 4 + 3 or ≥ 8 points), length of tumour (≤ 4 mm or > 4 mm) and age (< 65 or ≥ 65 years), and their impact on time to progression.

The odds of progression were compared across RP pathological features adjusted for participant age at recruitment. Logistic, ordinal logistic and linear regression were used to test the association between treatment pathway and binary, ordered categorical and continuous surgical characteristics, respectively. Upgrading was defined as an increase from a Gleason score of ≤ 3 + 3 points at baseline to > 3 + 3 points at treatment with RP, or moving from a Gleason score of 3 + 4 at baseline to ≥ 4 + 3 points at treatment with RP. Upstaging was defined as moving from T1/T2 disease at baseline to T3/T4 at RP. A positive surgical margin was categorised if any of the basal, apical, intraprostatic or extraprostatic margins were positive, and as negative if all four margins were negative or not applicable. A full statistical analysis plan is included in the protocol (see www.journalslibrary.nihr.ac.uk/programmes/hta/962099/#/).

Risk categorisation

On average, randomised participants (n = 1643) were 62 years old, of white ethnicity, with a median PSA level of 4.6 ng/ml, and their clinicopathological characteristics were reported previously. 104 These were reviewed and this showed that 1192 men (75%) had Gleason score ≤ 3 + 3 points disease and 1203 (76%) had clinical stage T1c disease. Using the D’Amico classification, 1031 participants (66%) were defined as being at low risk, 490 (31%) were defined as being at intermediate risk and 40 (3%) had features of high-risk disease (see Table 10).

Baseline clinicopathological features of participants with disease progression

A total of 199 out of 1585 (12.5%) randomised participants who commenced prostate cancer management within 12 months developed prostate cancer progression during a median follow-up of 10 years (see Table 1). There were 17 prostate cancer-specific deaths, a further 44 men developed metastatic disease and 138 men developed other clinical evidence of disease progression (see Tables 1 and 10). Older age (≥ 65 vs. < 65 years), baseline PSA level, Gleason score, cT stage, number of prostate cancer-involved biopsy cores, length of prostate cancer in any core (median 4.5 mm vs. 2.0 mm), aggregate length of tumour (median 8.0 mm vs. 4.0 mm) and presence of perineural invasion were each associated with an increased risk of subsequent disease progression (p < 0.001 for each) (see Figure 1). Thirty-nine per cent of men (n = 74) with disease progression were initially categorised with D’Amico low-risk disease, whereas 54% (n = 104) had intermediate-risk and 7% (n = 13) had high-risk disease, compared with respective values of 70% (n = 957), 28% (n = 386) and 2% (n = 27) for men without progression (p < 0.001).

Disease progression of participants with a baseline Gleason score of ≤ 3 + 3 points

A total of 101 out of 1192 men (8.5%) with a Gleason score of ≤ 3 + 3 points at baseline developed disease progression during follow-up; eight of these men died of prostate cancer, a further 20 developed metastases and 73 developed clinical progression (see Table 10). The odds of disease progression were higher among men who initially received AM than in men who initially received radical treatment (RP or RT) (p < 0.001) (Figure 10). For men with a Gleason score of ≤ 3 + 3 points at baseline, higher median PSA level (5.7 vs. 4.3 ng/ml), higher cT stage, higher D’Amico risk group, increased number of prostate cancer-involved biopsy cores, increased maximum length of prostate cancer in any core (median 3.0 vs. 2.0 mm) and increased aggregate length of tumour (median 5.5 vs. 3.0 mm) were each associated with disease progression (at least p < 0.05 for each comparison) (Figure 11).

FIGURE 10.

Cumulative risk of progression, by treatment group and age group. (a) Age < 65 years; and (b) age ≥ 65 years. a, n (n censored by time point).

FIGURE 11.

Cumulative risk of progression, by treatment group, Gleason scores and PSA levels. (a) Gleason score of < 7 points; (b) Gleason score of ≥ 7 points; (c) PSA level of < 10 ng/ml; and (d) PSA level of ≥ 10 ng/ml. a, n (n censored by time point).

Radical prostatectomy pathology characteristics

A total of 396 out of 553 men randomised to receive RP underwent this allocated treatment within 12 months. Of the 19 men who subsequently developed prostate cancer progression following surgery, none had a Gleason score of ≤ 3 + 3 points on examination of the resected prostate, whereas eight had a Gleason score of 3 + 4 points, seven had a Gleason score of 4 + 3 prostate cancer and four had a Gleason score of ≥ 8 points, which differed from men without disease progression (p < 0.001) (see Table 3). RP pathology features associated with subsequent disease progression included pathological Gleason score, pathological tumour stage, largest tumour volume, lymph node involvement, perineural and/or vascular invasion and apical or any margin positive status (p < 0.05 for each comparison). There was no evidence to suggest that the number of tumours was related to disease progression. Prostate cancer upstaging, but not upgrading, at surgery was associated with subsequent disease progression (p < 0.001).

Discussion

According to the D’Amico risk stratification criteria, 66% of ProtecT participants had low-risk prostate cancer and 34% had intermediate- or high-risk disease. The reported analysis of baseline clinical and demographic characteristics of randomised ProtecT participants demonstrates several baseline features associated with disease progression, including increased age, PSA level, Gleason score, cT stage, number of prostate cancer-involved biopsy cores, length of prostate cancer in any core, aggregate length of tumour (Figure 12) and perineural invasion.

FIGURE 12.

Cumulative risk of progression, by treatment group and length of tumour. (a) Length of tumour < 4 mm; and (b) length of tumour ≥ 4 mm. a, n (n censored by time point).

The observation that baseline Gleason score is associated with post-treatment recurrence/progression concurs with previously reported evidence,137–140 but 51% of participants with prostate cancer progression in ProtecT had a Gleason score of ≤ 3 + 3 points at baseline. It is unclear if this represents inadequate sampling or that this disease may progress over time. It is well recognised that low-risk, Gleason score ≤ 3 + 3 prostate cancer does not behave aggressively. However, the data supporting this are largely based on men who receive RP as their primary treatment, and it may mean that men with a Gleason score of ≤ 3 + 3 points are therefore cured by radical intervention even though their disease is very unlikely to progress, and thus they may not have needed radical treatment. It has been suggested that Gleason score ≤ 3 + 3 prostate tumours may no longer need to be called cancerous, but there is evidence suggesting that relatively low-grade prostate cancer foci can progress to lethal disease. 141,142 It is possible that Gleason score ≤ 3 + 3 lesions may comprise a spectrum of phenotypes and genotypes, indicating the need to delineate molecular features associated with prostate cancer progression rather than relying on conventional histological evaluation of biopsy alone.

The protocol for prostate cancer detection in ProtecT was designed in the late 1990s, before the routine use of imaging and refinement of mpMRI. Results of the recent PROMIS study suggest that mpMRI can aid prostate cancer diagnosis,20 albeit with an associated false negative rate. It is unclear whether or not new diagnostic pathways including pre-biopsy imaging and targeted biopsies, as well as existing genomic assays, will improve the prediction of outcomes in newly diagnosed patients. 20,143,144 Molecular-based risk stratification based on diagnostic samples aims to improve on the performance of current risk stratification tools. 145 Incorporation of molecular tumour profiling at baseline may possibly lead to more accurate and personalised risk stratification than is currently achievable using relatively limited clinicopathological features, but these remain to be tested prospectively in well-conducted evaluation studies.

In ProtecT, men with cT2 disease were more likely to progress than those with cT1 tumours, and a proportion of patients (29%) with cT2b tumours were found to have extraprostatic extension when operated on. The observation that an increased number of positive cores in the diagnostic biopsies and an increased maximum tumour length were associated with an increased risk of progression is also consistent with evidence from AS cohorts. 146,147 Baseline PSA levels were higher in men with disease progression in ProtecT than in men who had stable disease, consistent with studies suggesting PSA as a prognostic factor for lethal or recurrent prostate cancer following radical treatment. 73,148–150

Of the men who progressed, 101 had a Gleason score of ≤ 3 + 3 points at baseline, and eight of these participants died of prostate cancer, suggesting that transrectal 10-core prostate biopsy probably undersampled and/or underdetected high-grade prostate cancer in some patients. Among ProtecT participants with a Gleason score of ≤ 3 + 3 points at diagnosis, 87% were in the low-risk category at baseline using the D’Amico risk classification, and 13% had intermediate- or higher-risk disease.

There are limitations to our analysis and interpretation of these results. The ProtecT trial was designed in the late nineties, when the standard diagnostic pathway was a combination of digital rectal examination, serum PSA testing and TRUS-guided biopsies. It is now known that these methods lead to overdetection of indolent prostate cancer and underdetection of significant disease compared with more modern imaging techniques, such as mpMRI and targeted biopsies. The AM protocol in ProtecT has been less intensive than most contemporary AS regimes, although none of the current methods have been validated to demonstrate an improvement in outcomes. Finally, there were very few patients with high-risk disease at baseline (3%), and men from diverse ethnic origins, in particular African Caribbean men, were under-represented.

The main strengths of ProtecT are threefold: (1) its size, with over 82,000 men tested in the community, (2) the standardised diagnostic approach to detect clinically localised disease using a minimum of 10-core biopsies and (3) the high randomisation rate of men enrolled to evaluate the treatment effectiveness of RP, RT and AM. Although a reduction in the rate of metastasis and disease progression in men receiving radical treatment was demonstrated compared with AM, there was no evidence of a difference in mortality, and longer-term follow-up is needed to investigate whether or not this will change at the median 15-year follow-up.

Although the majority of ProtecT participants had a Gleason score of ≤ 3 + 3 points at diagnosis, when PSA levels and tumour volume were also taken into consideration it has become apparent that the actual prevalence of low-risk prostate cancer in the ProtecT randomised cohort was 66%, with 34% of participants having at least intermediate-risk disease.

None of the 19 ProtecT participants with a pathological Gleason score of ≤ 3 + 3 points following RP showed disease progression, suggesting that surgery cures definite low-risk Gleason score ≤ 3 + 3 disease using current criteria. These patients, if accurately identified, may benefit from AM by avoiding treatment side effects without deleterious oncological outcomes, as currently advocated in AS protocols. 151,152 The clinical dilemma is that we do not know whether or not men receiving AM or RT who subsequently developed disease progression actually had a true Gleason score of ≤ 3 + 3 points at baseline. It is recognised that around half of men with PSA-detected prostate cancer may be undergraded or overgraded by TRUS-guided needle biopsy. Although recent data evaluating pre-biopsy mpMRI suggests that this appears to improve the detection of high-grade disease compared with the diagnostic pathway used in ProtecT,20 it remains to be shown whether or not mpMRI is sufficiently reliable to avoid the overdetection of pathological Gleason score ≤ 3 + 3 prostate cancer.

The absence of any association between the number of tumour foci in the RP specimen and disease progression contrasts with published data suggesting that multifocal prostate cancer is associated with an increased risk of recurrence, although previous evidence from ProtecT suggests that solitary tumours were found in only one-fifth of RP specimens with PSA-detected localised prostate cancer.

In conclusion, baseline clinicopathological features of men with localised prostate cancer in ProtecT differed in men who developed disease progression compared with those with stable disease, but associations were not strong enough to reliably predict progression in individuals. As our understanding of the biology of prostate cancer behaviour improves, and its genomic diversity is elucidated, it is becoming clear that current methods of stratification need refinement with pre-biopsy imaging and targeted sampling, as well as through utilisation of validated genomic and other emerging biomarkers. This will need to be assessed in new, large-scale, prospective early-detection programmes. Only then will clinicians and patients be able to improve and refine the complex decision-making processes needed for the management of this ubiquitous malignancy.

General conclusions from clinical outcomes

The ProtecT trial has demonstrated no differences in survival at an average of 10 years from randomisation in clinically localised prostate cancer, irrespective of treatment allocation, in an ITT analysis. Baseline characteristics were insufficient to stratify patients and to predict outcomes of their disease. Radical treatments provided an oncological benefit compared with AM by reducing the incidence of metastases and disease progression by approximately 50%. At a median of 3 years’ follow-up, approximately 25% of men had moved out of AM to receive radical treatments, and, by 10 years, 55% of AM participants had received a radical intervention. Further analysis is under way to investigate the reasons for that change in management, and will be published and reported in due course. Over 80% of men receiving AM have no evidence of prostate cancer progression and thus have avoided side effects of radical treatments for prostate cancer.

Main patient-reported outcome measure analysis after 6 years of follow-up

Systematic reviews153–156 and large prospective cohorts7 have identified particular impacts on urinary, bowel and sexual function and little impact on generic QoL following radical treatments, but clear comparisons between contemporary treatments have been hindered by differences in outcome definitions, limited use of validated PROMs, short-term follow-up and sparse data on RT or AM/AS programmes. 157 RCT data have also been limited. The SPCG-4 trial showed greater impact on sexual and urinary function and QoL following RP than the impact following watchful waiting using the trial’s study questionnaire, and limited impact of RP on anxiety and depression. 158 PIVOT,79 comparing RP with watchful waiting for men with PSA-detected prostate cancer but with only three symptom items, reported similar results. These RCTs did not collect the full range of validated PROMs.

Methods: patient-reported outcome measures and statistical approaches

The Consolidated Standards of Reporting of Trials (CONSORT) guidelines for PROMs was used. 159 PROMs were prespecified secondary outcomes in the statistical analysis plan (see www.journalslibrary.nihr.ac.uk/programmes/hta/962099/#/), collected by validated questionnaires in four key domains:

1. urinary function and QoL impact – including urinary incontinence and LUTS (ICIQ,88 ICSmaleSF89 and EPIC160)

2. sexual function and QoL impact – including erectile function (EPIC160)

3. bowel function and QoL impact – including loose/bloody stools and incontinence (EPIC160)

4. health-related QoL – comprising generic health status (SF-1293), anxiety/depression (HADS92) and cancer-related QoL (EORTC-QLQ Q3091).

As indicated previously, study questionnaires were completed at baseline (at biopsy, before knowledge of diagnosis), at 6 and 12 months after randomisation and annually thereafter. The ICSmaleSF,89 SF-1293 and HADS92 were always included in study questionnaires, the ICIQ88 was added from 2001 and EPIC160 was added from 2005. As it concerned cancer-related QoL, EORTC-QLQ C3091 was included at year 5 only. PROMs from baseline to the complete 6 years’ follow-up were scored and analysed as recommended, with key items identified to aid interpretation of clinical relevance. Analyses were by ITT, with summary statistics and 95% CIs presented by randomised group. Multilevel models were employed to accommodate correlations between repeated measurements and test for treatment differences in follow-up assessments, and included covariates for the variables stratified by or minimised in the random allocation: age and PSA level at baseline (continuous variables), and Gleason score and study centre (dummy variables). Baseline PROMs were not included as a covariate as EPIC and ICIQ scores were not available for all men at baseline. PROMs data indicated that the allocated groups were comparable at baseline (see Chapter 3, Baseline level of patient-reported outcomes of randomised participants).

Results

Follow-up response rates were > 80% for all PROMs, without decline over time. Outcomes in the four domains are presented in the following sections and in Table 1, with selected scores and items presented graphically in Figures 13–16 (with time since randomisation on the x-axis, full measurement scale on the y-axis and scores representing ‘better’ function/health towards the top). All outcomes are reported in full in the online supplement of Donovan et al. 30

Domain A: urinary function and quality-of-life impact

All measures of urinary incontinence showed that there was greater impact in the RP group at 6 months, with some recovery, but urinary incontinence remained worse in the RP group than in the RT and AM groups at every time point (Figure 13a–c; all p < 0.001). Urinary incontinence rates were similar and little affected in the RT and AM groups, with a little worsening discernible in the AM group over time as surgery was taken up. Pad use increased from 1% at baseline to 47% in the RP group, compared with 4% in the AM group and 5% in the RT group at 6 months. By year 6, 18% of men in the RP group used pads, compared with 10% in the AM group and 3% in the RT group (see Figure 13b). There was a greater impact on urinary incontinence-related QoL in the RP group for 2 years, but this improved to become similar to that in the AM and RT groups (see Figure 13c). A similar pattern was shown for LUTS including urinary incontinence (see Figure 13d and f). Voiding LUTS were a little worse in the RT group at 6 months but then returned to close to baseline levels and similar to those in the RP and AM groups (see Figure 13e). Urinary frequency remained similar across the groups, with nocturia increasing in all groups at 6 months, particularly in the RT group, but recovering and returning closest to baseline in the RP group (see Figure 13g).

FIGURE 13.

Impact on urinary function (including urinary incontinence) and QoL. (a) ICIQ incontinence score; (b) EPIC item (one or more pads per day); (c) ICIQ incontinence problem; (d) EPIC urinary score; (e) ICSmaleSF Voiding Scale; (f) ICSmaleSF QoL item; and (g) ICSmaleSF nocturia item.

Domain B: sexual function and quality-of-life impact (including erectile function)

Erectile function reduced for all men at 6 months, with clear differences between the groups (Figure 14a; p < 0.001). At baseline, 67% of all men reported erections firm enough for intercourse, but by 6 months this reduced to 50% in the AM group, 24% in the RT group and 11% in the RP group. Erectile function remained worse in the RP group at all time points, with some recovery over 2 years but then further decline to 15% at 6 years, compared with recovery followed by decline to 29% in the RT group and a gradual year-on-year decline in the AM group (40% at year 3 to 28% by year 6). Very similar patterns of impact across the groups and over time were seen in the other measures of sexual function, bother and QoL impact of erectile dysfunction and sexual function (see Figure 14b–e).

FIGURE 14.

Impact on sexual function (including erectile function) and QoL. (a) EPIC item (erection firmness); (b) EPIC problem with erectile dysfunction; (c) EPIC sexual function score; (d) EPIC sexual bother score; and (e) EPIC sexual QoL.

Domain C: bowel function and quality-of-life impact

Bowel function and bother, and the impact of bowel habits on QoL were unchanged in the RP and AM groups, but worse in the RT group, particularly at 6 months (Figure 15a, b and f). Faecal incontinence and loose stools were similar across the groups (see Figure 15c and d), but bloody stools were experienced more in the RT group from year 2 onwards (see Figure 15e; p < 0.001). Bowel bother and QoL impact scores were a little worse in the RT group than in the other groups.

FIGURE 15.

Impact on bowel function and QoL. (a) EPIC bowel function score; (b) EPIC bowel bother score; (c) EPIC item (loose stools); (d) EPIC item (faecal incontinence); (e) EPIC item (bloody stools); and (f) EPIC item (bowel habits).

Domain D: health-related quality of life

There was no evidence of differences between the groups according to physical and mental health subscores in the generic health measure SF-12, anxiety or depression measured by the HADS or any of the symptom or function scales of the EORTC-QLQ C30 at year 5 (Figure 16a–d).

FIGURE 16.

Impact on health-related QoL. (a) SF-12 physical health score; (b) SF-12 mental health score; (c) HADS anxiety; and (d) HADS depression.

Conclusions

As described elsewhere in this report, the ProtecT RCT showed similar very high survival rates following all three allocated treatments but higher rates of metastases and disease progression in the AM group than in the radical treatment groups (metastases per 100 person-years: AM 0.63, RP 0.24, RT 0.30; p = 0.004; and disease progression per 100 person-years: AM 2.29, RP 0.89, RT 0.90; p < 0.001). In this context, understanding the profiles and levels of impact of the treatments on men’s lives becomes even more crucial for decision-making. ProtecT PROMs matched key domains recommended by international groups,155,161 and followed reporting guidelines,159 to provide definitive and unbiased comparisons of the impacts following standardised protocols for RP, RT and AM management strategies in the PSA era.

These ProtecT study findings definitively clarified the distinct impacts of prostate cancer treatments on urinary, sexual and bowel function and condition-specific QoL. RP’s impact on urinary incontinence and sexual function, particularly erectile function, was most severe at 6 months and although there was some recovery, it was worse than in the other groups over 6 years; there was no change in bowel function and some improvement in voiding LUTS. At 6 months, the impact of RT with neoadjuvant hormones on sexual function, particularly erectile function, was nearly as severe as RP and there was worse bowel function and LUTS than with either RP or AM, but there was gradual recovery, and, after 2 years, RT’s sexual and urinary impacts, including erectile function and urinary incontinence, were much less than for RP, and bowel function (apart from worsening bloody stools) was similar to AM and RP. In the AM group, impacts accumulated over time with ageing and the addition of radical treatments, with sexual and urinary function (including erectile function and urinary incontinence) gradually worsening year on year; bowel function was unchanged.

In terms of numbers needed to treat, treating four men with RP or eight men with RT rather than AM would cause one additional case of erectile dysfunction at 2 years; treating five men with RP or 143 men with RT rather than AM would cause one additional case of urinary incontinence at 2 years. By the end of follow-up at 6 years, urinary and sexual function had stabilised in the RT group after recovering for 2 or 3 years, and with the steady decline evident in the AM group, outcomes became similar for AM and RT, but remained worse in the RP group. These profiles of functional impact were mirrored in the sexual, urinary and bowel QoL items, with some evidence of accommodation to changes over time. No effects were seen on generic health status (mental or physical) or anxiety or depression in any group at any time, or in cancer-related QoL at 5 years.

The paucity of published data and the lack of consistency and variability in timing of measurements severely constrained the capacity to compare ProtecT findings directly with other RCTs or major studies of treatments at the time of publication. Direct comparisons were only possible for two specific (EPIC) items: erectile function and pad use. These showed that ProtecT findings were similar to the SPCG-4 and PIVOT studies in relation to erectile function in the RP and AM/watchful waiting groups. 79,158,162 Slightly worse results in observational cohorts7,11,163,164 would probably be related to age or selection biases. ProtecT results for pad use were considerably better than those from SPCG-4 for RP and watchful waiting, and broadly comparable to PIVOT’s watchful waiting; RT outcomes were similar at all time points in the observational cohorts. 30 Broadly comparable results were also found for bowel function and LUTS following RT155,163 and voiding after RP. EPIC scores were similar in other studies. 164,165 Further studies also found no differences in generic health or psychological/QoL assessments. 154,158,166

In terms of limitations, aspects of the ProtecT treatment policies may have affected some PROM scores. For example, the assessment at 6 months post randomisation will have captured the impact of neoadjuvant androgen suppression on sexual function at 6 months among those in the RT group. It was anticipated that the different treatments in ProtecT would lead to variation in the acceptance of allocation, and so the immediate receipt of RP by 68% and RT by 73% in those allocated groups may have attenuated some PROM comparisons. The take-up rate of AM was higher (89%), but because an integral part of the policy was the opportunity for reassessment and change of management, triggers including PSA level rise or evidence of progression led to increasing numbers of men receiving RP or RT over time. The rate of change was similar to many other AS programmes, but impacts from radical treatments combined with increased disease progression may have contributed to the gradual worsening of PROMs over time in the AM group. The AM findings also indicate a gradual decline in sexual and urinary function over time that would have occurred to some degree in the other groups.

The interventions in ProtecT remain the three most common contemporary types of treatments, but there have been considerable developments in treatment techniques since the study began, including the introduction of robot-assisted and laparoscopic surgery and brachytherapy or IMRT, and protocols for AS that exclude many of those included in ProtecT and using different strategies for monitoring and triggering change of management. Two large observational cohort studies were established in 2011–13 in the USA to evaluate short-term oncological outcomes and PROMs, explicitly expecting to show better results for PROMs with these newer treatment technologies. 55,56 The findings from these cohorts were remarkably consistent with those from ProtecT in terms of overall rates of symptoms and adverse effects. These findings, alongside randomised evidence from ProtecT, confirmed that all options carry risks of adverse effects that affect QoL, with each treatment option having a particular pattern in the short term: urinary incontinence and sexual dysfunction worst after surgery, followed by some recovery but persistent difficulties for some men; bowel problems worst after radiotherapy; and sexual dysfunction mostly related to neoadjuvant ADT. Even with AS, although adverse effects of interventions can be initially avoided, there is a natural decline in urinary and sexual function symptoms over time, and the adverse effects of radical treatments will be experienced when those treatments are received. 41

In the ProtecT RP group, 324 men received open retropubic procedures, with 23 receiving laparoscopic and 25 receiving robotic-assisted laparoscopic prostatectomy (19 were not specified), and most were nerve sparing (205 bilateral, 53 unilateral and 12 unspecified). In the two cohort studies above, most patients received minimally invasive surgery, and so these findings, combined with those of ProtecT, add to the evidence from the very short-term follow-up of a randomised trial of open surgery compared with robot-assisted surgery, also showing no evidence of differences in PROMs between open and minimally-invasive procedures. 47

There were strengths and limitations in the design and conduct of the ProtecT PROMs study. Strengths included the use of validated PROMs, well-balanced baseline data, high response rates and concordance between measures across the range of domains expected to be affected by treatments for localised prostate cancer. In addition, there was a high rate of randomisation of eligible participants – 62% (compared with 14.6% in PIVOT, for example) – here because of integrated recruitment research and enhanced team-working. The wider generalisability of ProtecT is assured by its PSA testing phase being a population-based programme embedded in a screening RCT – the CAP, which was recently published. 36,95

An important limitation was that only a small number of men of non-white ethnicity were included, although this reflected the population in the recruitment areas. 15 Other limitations relate to changes in diagnostic and treatment strategies since the study’s inception, and that populations with regular PSA testing might be different from this predominantly untested population, although on biopsy, compared with other PSA era treatment or screening RCTs, ProtecT had similar or higher numbers of participants with stage T1 (76%) and a Gleason score of 6 points (77%).

The ITT analysis undertaken as part of the ProtecT trial reflected important aspects for policy implementation; in future publications, we will present PROMs according to treatment received, and investigate reasons for management change in the AM group to further inform decision-making.

The ProtecT RCT has provided data that clarify the comparative impact of the major contemporary treatment options for localised PSA-detected prostate cancer on urinary, sexual and bowel function and QoL, including rates of recovery and outcomes up to 6 years after treatment allocation. Longer-term follow-up is needed to document the changes in PROMs and oncological outcomes that will emerge from the originally randomised groups. Although the current findings do not yet provide a complete picture, men making decisions about treatments for newly diagnosed localised prostate cancer or contemplating PSA testing can use these profiles of impact for decision-making with clinicians. As indicated above, the longer-term impacts of treatments are unknown. Follow-up for a further 5 to 10 years is required to fully inform the balance between the shorter-term impacts of treatments shown here and the longer course of prostate cancer progression, death from prostate cancer or competing causes and QoL impacts that may emerge in the longer term from these treatments or from the sequelae of treatments for metastatic or progressing prostate cancer. However, the NIHR HTA programme declined to support longer-term follow-up in ProtecT using PROMs. Although we have attempted to seek funding from other sources, as of March 2018 no such support has been found.

Methods

Urinary incontinence and its impact on QoL were assessed by the ICIQ score and QoL item, the EPIC 50-item urinary incontinence subscore and item on pad use, and the ICSmaleSF questionnaire urinary incontinence score. LUTS and their impact on QoL were assessed by the ICSmaleSF questionnaire voiding score and individual symptoms of frequency, nocturia and a QoL item, and with the EPIC urinary summary, bother and obstruction/irritation subscores. The PROMs were compared with each other.

Results

Response rates were higher than 85% for all questionnaires and did not decline over time.

All measures of urinary incontinence (ICIQ, ICSmaleSF urinary incontinence and EPIC urinary incontinence scores and EPIC item on pad use) showed that surgery had the greatest negative effect at 6 months, and that urinary incontinence remained worse in the surgery group than in the radiotherapy and AM groups at all time points over 6 years (p < 0.001 for each measure) (Figure 17). There was little difference in urinary incontinence between the radiotherapy and AM groups over 6 years. Only the ICIQ measure had a specific item assessing the QoL impact of urinary incontinence. This showed that men in the surgery group experienced most impact at 6 and 12 months after randomisation; after 24 months, the QoL impact of urinary incontinence recovered and became similar to the other groups, although the profiles were different (p < 0.001).

FIGURE 17.

Measuring urinary incontinence. (a) ICIQ incontinence score; (b) EPIC urinary incontinence; (c) ICSmaleSF incontinence; and (d) EPIC item (one or more pads per day).

The ICSmaleSF voiding score showed differences between the groups (p < 0.001), with worse voiding symptoms in the RT group at 6 months, returning to be similar to the other groups from 12 months onwards (Figure 18). The EPIC obstruction/irritative subscore did not detect these differences (p = 0.77). The ICSmaleSF illustrated differences between the groups for nocturia (p < 0.001), with radiotherapy worse at 6 months, but no differences in daytime frequency (p = 0.47). The distinct effect of voiding difficulties on QoL was unclear, however, as the question in both measures – EPIC urinary bother and ICSmaleSF QoL score – referred to the impact of all urinary symptoms including urinary incontinence. The profiles from these scores, as well as the EPIC urinary summary score, were very similar to the profiles related to urinary incontinence. The EPIC urinary bother subscore was not clearly different between the groups (p = 0.095).

FIGURE 18.

Measuring LUTS and impact on QoL. (a) ICSmaleSF voiding; (b) EPIC urinary summary score; (c) EPIC urinary irritative score; (d) ICSmaleSF nocturia item; (e) ICSmaleSF effect of urinary symptoms on QoL; and (f) EPIC urinary bother score.

Conclusion

The EPIC, ICIQ and ICSmaleSF measures produced very similar profiles for the effects of treatments on urinary incontinence. Although the magnitude of differences was a little different between the measures, they were equally effective at measuring the effects of treatments on levels of urinary incontinence. However, the only measure that assessed the impact of urinary incontinence on QoL was the ICIQ questionnaire.

The ICSmaleSF and EPIC measures did not concur over the assessment of LUTS. The main EPIC scores (urinary summary and bother) included items on urinary incontinence, and so their profiles conflated urinary incontinence with other LUTS. The EPIC measure also did not have a specific voiding score, as its subscore assessed obstructive and irritative symptoms combined (dysuria, haematuria, weak stream and frequency). ICSmaleSF had separate voiding and incontinence scores, as well as specific items for nocturia and frequency, which it recommends reporting separately. ICSmaleSF did not include items on dysuria or haematuria in its assessment. EPIC and ICSmaleSF both conflated urinary incontinence and LUTS in their QoL impact item/scores.

Thus, there are clear similarities between measures in their assessment of urinary incontinence, and so any of these validated PROMs could be used: ICIQ, EPIC or ICSmaleSF. However, the assessment of QoL related to urinary incontinence is only assessed by the ICIQ, so this should be included in studies wishing to assess this. For the assessment of LUTS, the measures have differences that need to be taken into account when deciding which measure to use and in interpreting the results. The EPIC and ICSmaleSF questionnaires include different LUTS in their assessments of voiding, urinary incontinence and obstruction/irritation. Care needs to be taken when using scores as different measures include different items within similarly named scores. Particular issues arise for the assessment of voiding or storage symptoms, which are likely to become important for men undergoing AM/AS programmes or radiotherapy. In this situation, ICSmaleSF offers a specific voiding score and items.

Methods

The aim of this substudy was to investigate this seeming contradiction by examining levels of general QoL (physical and mental health, anxiety and depression) among men reporting urinary incontinence or erectile dysfunction following treatment in ProtecT, compared with those not reporting these symptoms. Urinary incontinence was assessed by the EPIC items on absorbent pad use and erectile dysfunction by the firmness of erections for intercourse item. Physical and mental health were assessed by two domains of the SF-12 generic health measure. Anxiety and depression were assessed by the HADS. Study questionnaires were completed at baseline before the diagnosis was known, at 6 and 12 months after randomisation and annually thereafter. PROMs were scored and analysed as recommended by their authors. Means and standard deviations (SDs) were calculated for the HADS and SF-12, with p-values testing the null hypothesis of equal population means across groups without the symptom or with the symptom at 6 months, or with the symptom at 6, 12 and 24 months, assessed over the duration of the study (6 years).

Results

There were 133 men (13%) with urinary incontinence who needed to use pads at 6 months and 62 men (6%) who needed to use pads at 6, 12 and 24 months, compared with 845 (81%) who did not use pads for urinary incontinence. Men who needed to use pads for urinary incontinence had higher depression scores than those who did not need to use pads at all time points, with scores highest of all for those who needed to use pads for longer. There was strong evidence that depression scores were worse in those using pads at 6 and 12 months (p < 0.001). The pattern for anxiety scores was somewhat similar. The highest anxiety scores were evident among men needing to use pads for longer, with those using pads at only 6 months having scores more similar to those not needing pads. There was some evidence that anxiety scores were particularly worse for those using pads at 6 and 12 months (p = 0.02 at 6 months and p = 0.03 at 12 months).

The impact of urinary incontinence (pad use) on depression and physical health is shown in Appendix 1, Figures 30 and 31, and the impact of erectile dysfunction on depression and physical health is shown in Appendix 1, Figures 32 and 33.

The SF-12 scores for physical health were different at only 6 months, when there was slightly worse physical health in those who needed pads compared with those who did not (p < 0.001). After this, physical health was similar between the groups. Reflecting the HADS scores above, SF-12 scores for mental health were worse among those needing pads at 6 months (p = 0.005), with some weaker evidence of worse mental health at 12 months among those needing pads longer term (p = 0.07).

There were 278 men (27%) who had erectile dysfunction at 6 months and 454 (44%) who had erectile dysfunction at 6, 12 and 24 months, compared with 289 (28%) who did not report erectile dysfunction. HADS depression and anxiety scores were much higher in men reporting longer-term erectile dysfunction than in those with erectile dysfunction at only 6 months or with no erectile dysfunction. For depression, this evidence was very strong (p < 0.001 at each time point); it was slightly less strong for anxiety (at p = 0.001 to p = 0.07). Depression and anxiety levels did not decline over time among those with long-term erectile dysfunction, but there was some evidence that those who recovered function after 6 months experienced less anxiety and depression. These patterns were mirrored in the SF-12 mental health scores. Perhaps more surprisingly, the experience of erectile dysfunction at 6 months or longer term was associated with a small reduction in physical functioning, and this was maintained for 6 years (p < 0.001).

Conclusions

High levels of erectile dysfunction were experienced by men who underwent treatment for localised prostate cancer and, for those with these symptoms, this was associated with worse physical and mental health, including anxiety and depression. Erectile dysfunction continued to have a small but measurable impact on these aspects of general QoL for the duration of the study (6 years). Urinary incontinence affected a smaller number of men, but had a clear impact on anxiety, depression and physical and mental health at 6 months. This impact on physical and overall mental health reduced after 6 months, but urinary incontinence continued to be associated with slightly higher levels of depression and anxiety over time.

Therefore, although it has been found that treatments for localised prostate cancer do not affect overall QoL on average across the treatment groups in ITT analyses, men experiencing erectile dysfunction or urinary incontinence had worse physical and mental health, particularly anxiety and depression, than men without these side effects in a secondary analysis. Men undergoing treatment should be made aware of the likelihood of these symptoms and their potential impact, and may need additional support when they experience them.

Perspective

The primary economic evaluation is from a NHS perspective. A preliminary analysis of a cross-sectional survey of the ProtecT men (years 1–5) in which Personal Social Services costs in the form of home care were obtained showed that home care was not a main cost driver; therefore, a NHS perspective was chosen.

Time horizon

The economic evaluation compared the costs and outcomes of each group from randomisation until the closure of the ProtecT study database (November 2015): a median of 10 years’ follow-up.

Identification of relevant resource use

The scope of the economic evaluation was defined as the main NHS resources used relating to prostate cancer and the three treatments and their follow-up. Data were collected on the use of hospital inpatient, outpatient, clinic and GP services for the initial trial interventions and associated long-term follow-up.

Measurement and valuation of relevant resource use

Resource use information in relation to the initial treatments given and any subsequent RP or radiotherapy was recorded by trial research nurses onto study-specific pro formas. For surgery, information was recorded in relation to the pre-operative assessments, the actual operation itself and complications. For radiotherapy, information was recorded in relation to the pre-treatment planning of radiotherapy and relating to the actual delivery of the radiotherapy including number of fractions given. For AM, all visits for the duration of the time the patient was on the AM treatment were recorded.

Information on all follow-up outpatient visits, inpatient stays and accident and emergency attendance, and, from April 2005, primary care attendances, was recorded onto research follow-up schedules, which were completed by research nurses using information from appointments with the patient and their medical records at 12 months from randomisation and then annually thereafter. Telephone appointments were held with men unable to attend the hospital and participants were given an annual resource use log to aid recollection of events at appointments.

To ensure completeness of resource use, initial treatment pro formas and annual schedules were compared, duplicated data were dropped and then the combined sources of data were used. In addition, the mainly administrative trial clinical centres database, which contained some clinical information, was used for validation purposes to ensure that all prostate cancer treatments were recorded and took place post randomisation.

In order to value the initial treatments, in relation to surgery, information relating to the type of procedure conducted and any complications incurred was used to assign a Healthcare Resource Group (HRG) 4 code (these are groups of events that consume similar levels of resources). This was then valued using the UK NHS Reference Costs 2014 to 2015. 168

In relation to radiotherapy treatment, the neoadjuvant deprivation and hormone injections were valued for primary care delivery using the BNF. 169 Hormone injections delivered in secondary care were valued using the UK NHS Reference Costs 2014 to 2015. 168 Pre-treatment planning, including scans, was assigned a HRG4 code, as were the fractions used for each dose of radiotherapy. These were valued using the NHS Reference Costs 2014 to 2015. 168

The AM visits were valued using the average time for a face-to-face or telephone consultation given in the AM schedules and the cost of a nurse team leader (including qualifications) based on national estimates. 170

All other inpatient stays and day cases were assigned HRG codes; initially, the text cited in the questionnaires as the reason for admission was used to map the event to Office of Population Censuses and Surveys Classification of Interventions and Procedures (OPCS-4) codes. The HRG4+ 2014/15 Reference Costs Grouper Code to Group v1.0 workbook171 was then used to allocate the HRG codes. NHS Reference Costs 2014 to 2015168 were applied to the HRGs on a fixed consultant episode basis. Day-case costs were assigned when the patient did not stay in hospital overnight. In the absence of information to identify whether the admission was elective or non-elective, stays were assigned elective inpatient costs.

All outpatient visits were valued using the Department of Health and Social Care reference costs relevant to the specialty. 168

Outpatient procedures were allocated HRG codes. The ProtecT annual research follow-up was valued at half the cost of an outpatient visit to account for both the clinical and research element of this visit.

Appendix 1, Table 27, gives further details of how the resources were measured, coded and valued. All costs were valued in 2014/15 Great British pounds. Data cleaning and coding was conducted prior to the unblinding of the health economist. A number of assumptions were made in relation to missing resource use-data, which are given in Table 11.

Measurement and valuation of outcomes

The outcome for the economic analysis as recommended by the National Institute for Health and Care Excellence (NICE) is the quality-adjusted life-year (QALY) at a median of 10 years following randomisation into the trial. The utility values for each year that the participant could have been a trial participant were estimated from the EQ-5D-3L health status questionnaire and the associated published societal utility tariffs. 172 The EQ-5D-3L was completed at baseline (at the confirmatory biopsy), 6 months and 12 months following randomisation and annually thereafter. Participants who had died were assigned a zero utility value for the remaining years that they could have been trial participants (i.e. from time of death until the closure date of the database). The total number of QALYs for each participant was calculated using the area under the curve approach. For men who had missing EQ-5D-3L scores in the year prior to the year of death but available for the preceding year, the area under the curve methodology used the preceding year’s EQ-5D-3L score. Where a EQ-5D-3L time point was missing and the adjacent year’s values, other than death, were available, the mean of the adjacent year’s values was used for the analysis.

Analysis

The analysis was conducted on an ITT basis comparing the three treatment groups as randomised, considering prostate cancer or its treatment-related NHS costs in relation to the QALY. Total mean adjusted costs and outcomes were discounted at 3.5% as recommended by NICE. 173 All analyses were conducted in Stata version 14.1.

The total number of times each item of resource use was used was calculated for each participant. Participants who had died were assigned zero resource use for the remaining years that they could have been trial participants (i.e. from time of death until the closure date of the database). The total mean resource use by resource use category and by trial group was then calculated.

The cost of each item of resource use was calculated as the resource use (e.g. number of GP visits) multiplied by its relevant unit cost (e.g. cost of GP visit). The costs for each aspect of resource use were summed annually, across time and by resource use category for each participant.

Seemingly unrelated regressions, which control for the correlation between costs and QALYs, were used to estimate total adjusted mean costs and QALYs. Adjustments were made for study centre and the minimisation variables of the randomisation process: age at baseline, Gleason score (< 7, 7 or 8–10 points) and the mean of the baseline and first biopsy PSA test results together (< 6.0, 6.0–9.9 or > 9.9 ng/ml). QALYs were also adjusted for baseline EQ-5D-3L result. 174 The adjusted mean costs of the three treatment groups were compared with the adjusted mean QALYs to assess if any of the treatments were dominated (i.e. less effective and more expensive than one or both of the other two treatments). Incremental adjusted mean costs and QALYs, bias-corrected and accelerated CIs (to account for non-normal distributions) and incremental cost-effectiveness ratios (ICERs) were estimated for non-dominated treatments using seemingly unrelated regression and non-parametric bootstrapping (5000 model iterations). Regression outputs were used to estimate parametrically the net monetary benefit (NMB) statistic and associated CIs at a willingness to pay of £20,000 per QALY.

Cost-effectiveness acceptability curves (CEACs) were generated to explore uncertainty. 175 The CEAC graphs the probability that each trial group is the cost-effective option at a range of willingness to pay thresholds. A willingness to pay threshold is defined as the maximum that society would be willing to pay for an improvement in health, measured by the QALY. The CEAC was calculated as follows: NMB values were calculated for each participant at willingness to pay per QALY thresholds from £0 to £100,000 at £1000 intervals; 5000 bootstrap model iterations of the adjusted linear regression models of NMBs were performed. At each threshold, bootstrapped NMBs were compared between groups to calculate the proportion of times each group had the highest NMB. CEACs were plotted to present the probability that each trial group is the most cost-effective compared with the other two groups at a range of monetary values.

Sensitivity analyses

One-way sensitivity analyses in which the value of one of the variables was changed, and the analysis rerun, and scenario sensitivity analyses in which the values of more than one of the variables were changed simultaneously and the analysis rerun, were used to account for any methodological uncertainty or assumptions made during the course of the study and analysis.

One-way sensitivity analyses

Given the long period of follow-up, multiple imputation on total costs and total QALYs would have led to a great loss of information. The first four sensitivity analyses were therefore conducted to explore the effect of missing data:

1. The imputed mean QALYs for those who had one time point missing was replaced with a QALY score 10% lower than this mean score to take into account the possibility that non-completion of questionnaires resulted from illness and treatment.

2. An analysis excluding the cases in which the QALY had been imputed.

3. To examine the impact of not capturing information on primary care visits until 2005, the analysis was rerun for those participants who received their first annual follow-up following this date.

4. To examine whether or not administrative censoring affected the results, the analysis was rerun on the first 6 years of follow-up, the time point at which no administrative censoring had taken place. The analysis was run again on the first 6 years adjusting for the information appointment date. The results of these two analyses were compared.

5. During the feasibility period of the study, initial follow-up was not as robust as for the main trial; the men recruited during this period were therefore excluded from this sensitivity analysis.

6. The appropriate discount rate for economic evaluations in health care is the subject of debate internationally, and varies between jurisdictions. For this reason, the impact of a 1.5% discount rate on both costs and benefits was explored.

7. It was sometimes difficult to distinguish between a day-case procedure and an outpatient procedure. All day cases were therefore costed as outpatient procedures in the sensitivity analysis.

8. Prostate cancer-related resource use was used for this analysis; however, unrelated resource use was also recorded. There was some discrepancy over whether a transurethral resection of the prostate (TURP) was or was not related to the AM treatment. The increased surveillance of these men could have meant that more LUTS may have been identified, which could have meant that more TURPs were conducted. In the base-case analysis, all TURPS were included, but they were excluded from this sensitivity analysis.

9. In economic evaluations, research-based appointments are not usually costed. Following discussion with the trial research nurses, it was discovered that clinical follow-up also took place in annual research follow-up appointments. In the base-case analysis, these appointments were costed as half an outpatient visit. In the sensitivity analysis, these appointments were excluded.

10. The trial research nurses were asked to record all non-admission secondary care attendances in one section of the pro forma. It was discovered through discussion with the nurses that some nurses at certain times only recorded clinician follow-up in that section, and any outpatient procedures were recorded elsewhere in the pro forma. Although the main procedures (e.g. bone scans, biopsies, MRI and computerised tomography) were accounted for, other procedures (e.g. cystoscopy) may not have been included by some nurses in the outpatient section and were not included in the base-case analysis. This sensitivity analysis therefore includes the costs of all other procedures recorded elsewhere in the pro forma.

Scenario analyses

1. A high-cost option, whereby all trial participants without catheters were costed as day cases, an outpatient visit cost was assigned for extracting blood for PSA tests, all procedures recorded in only the follow-up schedules were costed as the highest complications and comorbidities split, annual research follow-up appointments were costed as an outpatient visit and initial and salvage radiotherapy planning were costed using technical support.

2. A current initial treatment option, whereby radiotherapy was costed as IMRT and RP was costed as a robotic RP.

Strengths

This use of individual patient data and the study’s longevity are its core strengths. The use of medical record review in conjunction with a participant visit meant that secondary care missing data were kept to a minimum and are likely to be missing completely at random; missingness did not differ by group.

Limitations

Primary care resource use was not recorded at the beginning of the trial; thus, an appropriate primary care delivery cost was applied to hormone injections prior to radiotherapy and to each PSA test. All other missing primary care data were assumed to be missing completely at random as there was no reason to believe that primary care costs collected from the annual schedules would differentially differ by group over the trial’s duration. Sensitivity analysis 3, which only included participants who were asked about primary care visits, showed that the RT group was cost-effective only at the higher, £30,000-per-QALY, threshold; however, this could be indicative of a shorter period of follow-up in keeping with other sensitivity analyses.

There were missing data in the follow-up pro formas, and reasonable assumptions, often following discussions with nurses and clinicians, were used to impute these data.

There was the potential for incomplete EQ-5D-3L data not to be missing at random, resulting from non-completion of questionnaires due to illness and treatment. Reducing the mean QALY score by 10% for those participants who had mean imputed QALY scores for 1 year of the trial did not affect the overall result (sensitivity analysis 1).

Future work

During this long-term study, different and potentially more expensive treatments (e.g. robotic prostatectomies, IMRT and brachytherapy) have become more prevalent. The most recent costs that allowed for the coding of HRGs onto NHS Reference Costs were used,168 and a sensitivity analysis in which prostatectomies were costed as robotic and radiotherapy was coded as IMRT showed the initial results to be robust; however, future modelling using these data will allow for other treatments to be compared.

Although a NHS perspective is of most interest to decision-makers, patient costs are still important. A cross-sectional analysis that aimed to capture participant costs was carried out at two different time points of the ProtecT study. Future work will include analyses using these data as well as information in relation to indirect patient costs in terms of time off work and time lost for other activities, which was collected in the annual self-completed questionnaires administered to the participants.

Background

Prostate-specific antigen testing remains controversial, with the most recent study showing no benefits of PSA screening on mortality over 10 years’ follow-up. 36 The European Association of Urology Guidelines179 and the UK National Screening Committee through the Prostate Cancer Risk Management Programme180 recommend that men who request a PSA test are given full information about the relative risks and benefits of screening to enable them to make an informed choice. An understanding of men’s beliefs and experiences about prostate cancer screening and their reasons for attending or not attending screening are needed to inform optimum support services to support decision-making by patients. 181

Methods in the ProtecT study

Men consenting to or refusing PSA testing in the ProtecT study were invited to take part in a single interview study investigating men’s decision-making about whether or not to undergo PSA testing. 116 Interviews were undertaken with men accepting a PSA test or not responding to the invitation for PSA testing. Interviews were conducted face to face or by telephone by Dr Kerry Avery (University of Bristol, ProtecT study group) and lasted for a mean of 34 minutes (range 7–70 minutes) (Table 17).

Men consenting to take part as core participants in the serial interview study were also invited to recount their reasons for responding to the invitation to take part in the study, whether at the time of the prostate check clinic (n = 7), following random allocation to a treatment (n = 31) or after choosing treatment (n = 16). The data relevant to this topic were combined.

Results

Men who declined the invitation to attend for the PSA test believed that a test was not needed and justified this belief with the view that they mostly perceived themselves to be at low risk of prostate cancer or in good health. Some also believed that prostate cancer was not life-threatening, had been advised against a PSA test (in some cases by a medical practitioner) or considered the PSA test not to be sufficiently accurate. In contrast, those who accepted the invitation to attend for PSA testing reported that they had done so because they felt that they had nothing to lose because they expected a clear result as they were in good health or lacked symptoms. Attenders indicated that a friend or family member had recommended that they attend, in some cases because of a family history of prostate cancer or previous PSA. 116

In conclusion, most men in both groups perceived themselves to be at low risk of having prostate cancer, which was accurate in that 3% were eventually diagnosed.

Background

Men’s views and experiences of undergoing the standardised 10-core TRUS-guided prostate biopsies were investigated in ProtecT. Previous research, including in ERSPC, suggested that sextant (six-core) prostate biopsy carried a risk of sepsis, pain, bleeding and, rarely, death. 182,183 There was evidence of variation in practice in the use of local anaesthesia during prostate biopsy. 184,185

The experience of pain or other side effects during or after the biopsy procedure was poorly investigated. Two studies suggested that around 18–19% of men refused to undergo repeat biopsy. 185,186 Limited qualitative research found that although most men reported biopsy to be uncomfortable rather than painful, some found it acutely stressful, exhausting or painful. 187 It was suggested in one study that pain was influenced by psychosocial factors. 188

Methods in ProtecT

Experiences of diagnostic testing were explored in two substudies to supplement the quantitative data generated in the Prostate Biopsy Effects (ProBE) study. 189 In the first study, men were purposively sampled to include those agreeing (n = 24) or refusing (n = 13) to undergo prostate biopsy in ProtecT from a range of study centres (Table 18). 116

In the second study, men taking part in the quantitative ProBE study189 were invited to participate in a single interview (n = 33), and a subset of men was sampled specifically to contribute their experience of post-biopsy infection (n = 5). These interviews were also supplemented with information about biopsy from the 47 men taking part in serial in-depth interviews (Table 19).

Results

The findings showed that most patients tolerated prostate biopsy reasonably well. However, as many as one-quarter reported problematic side effects (such as pain and bleeding) and anxiety. Side effects were more likely to be perceived as being problematic if experiences were not in line with expectations derived from the information provided in preparation for biopsy. Where men were not sufficiently prepared for the experience or severity of side effects following biopsy, they responded by contacting health professionals and asking for reassurance. They also expressed frustration that they were not adequately prepared for these side effects.

The number of patients who refused biopsy when offered it was relatively small. However, those who did refuse tended to do so because they perceived themselves to be at low risk of having cancer, or because they were anxious about the test. The perception of low risk was often associated with not having urinary symptoms.

Conclusions

Both substudies of the experience of biopsy concluded that men were insufficiently informed about the biopsy procedure, its relationship with urinary symptoms and the potential incidence and severity of side effects. 116,190 These qualitative findings were combined with quantitative data from the ProBE189 to propose an updated and evidence-based set of information for men undergoing prostate biopsy to support more informed decision-making about whether or not to attend for a biopsy, what to expect and how to respond to side effects. 190

Background

Secondary outcomes in the ProtecT trial included disease progression, treatment complications, urinary, bowel and sexual function symptoms and impact on QoL, psychological status and general health status. The study design also included qualitative assessment of these outcomes through in-depth interviews, with sampling of participants across the randomised groups and of those choosing a treatment. When the ProtecT main study started data collection in 2001, the literature reporting men’s views and experiences of undergoing testing, diagnosis and treatment for localised prostate cancer was extremely limited. The little research that had been conducted was usually in small cross-sectional studies within a short time of treatment; for example, there were small qualitative studies of men’s short-term experiences of surgery191,192 or radiotherapy. 193

As the strategy of active monitoring/surveillance was only initiated in the late 1990s, at the time of the start of the ProtecT study, studies of this option have only been published in recent years. Some have focused on levels of anxiety and depression, with some suggesting increased levels of distress,45,194 but others, including a systematic review, reported not finding adverse effects on QoL measures. 157,195,196 Qualitative research, again cross-sectional and small scale, revealed that men adopted coping strategies to deal with the uncertainty inherent in a monitoring/surveillance protocol by blocking off the diagnosis or rationalising the level of threat posed or engaging in lifestyle change as a method to counter the cancer. 188,197,198 Therefore, in relation to all outcomes from treatment options, previous qualitative studies were mostly restricted to short-term effects, usually without comparisons or long-term follow-up. In ProtecT, we were able to sample across all treatment options and explore experiences related to each of the treatments.

In a further study to evaluate the management of AM, the views of 22 of the core participants in the qualitative serial interview study were combined with data from 11 urologists and 23 nurses delivering ProtecT trial AM and 20 men, with prostate cancer managed in urology clinics elsewhere in the UK, with in-depth interviews with urologists and three specialist nurses working in these clinics. 199

Methods in ProtecT

Experiences of following AM were explored with 20 men included in the core group of patients who were seen over a period of up to 14 years, supplemented with five more who were identified by study nurses as having changed management to a radical treatment. The characteristics of the men sampled across centres, ages, cancer risk groups and who were randomised to or chose AM are given in Table 20.

Sixteen men recruited to the core serial interview study received radiotherapy treatment, and their experiences of neoadjuvant ADT and RT were explored over a period of up to 8 years post treatment. The characteristics of these men are given in Table 21.

There were 20 men who were recruited to the core serial interview study who underwent RP (see Table 21). Their experiences were explored over a period of up to 8 years. Two men who were recruited following adverse experiences of surgery were interviewed on one occasion and their data were included in this analysis (one was randomly allocated to surgery and the other chose surgery).

Findings: active monitoring

Men in the core serial interview group took part in up to six interviews over 14 years about their experiences of being assigned to or choosing AM, following the monitoring protocol and decision-making around any change to radical treatments. Their characteristics are summarised in Table 22. A manuscript is under preparation considering the multiplicity of issues for this group; in the manuscript we present the key findings.

All participants were aware of the uncertainties inherent in following an AM protocol, namely uncertainty surrounding cancer progression and whether or not monitoring would prove effective in prompting timely curative treatment if needed and desired. They experienced a key contradiction following this pathway: feeling well (many were symptom free) yet with a potentially life-threatening cancer. They employed a range of strategies to minimise the impact of living with these uncertainties and contradictions: seeking to assert control over the disease and the monitoring process, seeking reassurance from health professionals and their wider social networks and contextualising/normalising their prostate cancer. Each PSA test or review in the early stages of participation brought potential anticipation that radical treatment might be needed, and although over time this threat reduced, it did not disappear. Trust was invested in health professionals to maintain the delicate balance between anxiety and reassurance, and to recommend remaining on AM or changing to a radical treatment. The information exchanged in review appointments and the attitudes of clinical staff were pivotal in enabling men to maintain this delicate balance.

Patient-reported outcomes reported in questionnaires showed no evidence of greater anxiety among men randomised to AM than among those randomised to RP or RT at any time up to 6 years. 30 The qualitative research revealed that there was anxiety associated with the uncertainties of living with prostate cancer while on AM but it varied between men, fluctuated over time and was mostly temporary and successfully managed. In this study, men acknowledged uncertainties and most described some anxiety, but their strategies helped to maintain anxiety levels within tolerable limits. Over time, anxiety lessened to some degree as confidence in AM increased or reassurance was gained from trusted health professionals who advised whether to continue or change treatment.

ProtecT trial participants valued nurse-led AM within the study for its flexibility, its accessibility and the continuity of the service. 103 They felt confident about the quality of care. ProtecT consultant urologists and nurses also valued the nurse-led service highly, identifying continuity of care and potential resource savings as key attributes. Clinicians and patients outside the ProtecT trial believed that nurse-led AM could relieve pressure on urology clinics without compromising patient care.

Findings: radiotherapy

Sixteen participants in the core serial interview study recounted their experiences of following the ProtecT radiotherapy protocol, which involved 3 to 6 months of neoadjuvant ADT (hormone therapy) and external beam conformal radiotherapy delivered every weekday over a period of several weeks. Most men took part in three interviews. A manuscript detailing experiences is in preparation, and so only the key findings are presented here.

Men reported impacts on urinary, bowel and sexual function similar to those shown in the patient-reported questionnaires (see Chapter 4). Men considered the possibility that some loss of function was the consequence of normal ageing, and so it was difficult for them to separate out what was the effect of normal ageing and what was the effect of hormone therapy/radiotherapy, especially with regard to fatigue and sexual function. Although some experienced socially embarrassing side effects (urinary and bowel urgency, breast enlargement and hot flushes), most were able to accommodate these over the long term without significant changes to their lifestyle. On the whole, men did not seek help to regain sexual function and many reported prioritising recovery from the prostate cancer and being disease free over sexual function.

Findings: surgery

Eighteen participants in the core serial interview study and one man recruited to take part in a single interview following adverse experiences post surgery were asked about their experiences following surgery. A manuscript including experiences is in preparation, and so only the key findings are presented here. The main concern of men post surgery was urinary control. Several men reported surprise at the degree of incontinence experienced in the immediate period following removal of the catheter, although for most men the level of control improved to a level that allowed them to be able to function normally. A small number experienced significant problems that had a major impact on daily life over the longer term, with two men considering whether or not to opt to have a further procedure to implant an artificial sphincter. An important issue for men experiencing major difficulties with urinary continence was coping with the unknown trajectory of future function. They had not known in advance of surgery what level of function they would experience post surgery, and then post surgery it was also unclear what level of function would be regained.

Those who had difficulties with urinary control explained that this overshadowed other issues, such as sexual difficulties. Those who regained urinary control became more concerned about sexual function difficulties. Contextual issues (e.g. whether or not a man was in a relationship) affected men’s levels of concern about sexual function. Men who were not in relationships expressed more concern about loss of sexual function than those who were in stable relationships.

Conclusions

Appointments with health professionals (primarily nurses, supported by urologists and oncologists as required) to review PSA levels and disease progression were critical for the management of men in the AM group. These appointments enabled and supported men to achieve their particular balance between uncertainty and reassurance, and staying on AM or changing to a radical treatment. The information exchanged in review appointments and the attitudes of clinical staff were crucial in maintaining this delicate balance. The ProtecT AM service protocol is available for implementation. 190

Men undergoing radical treatments felt that they could have been better informed, particularly about the timing of when hormone or radiotherapy started and finished, and also, particularly for surgery patients, about the levels of impairment they might face and for how long it would last. Work is continuing to ensure that the ProtecT outcome data will now be presented to support decision-making.

Background

It is now standard practice to integrate PPI in trial design, conduct, reporting and dissemination. At the time that the ProtecT study was designed, PPI was not generally integrated in trial designs. In this study, PPI has not been added in the now standard way but was integrated into the design and conduct from the outset using qualitative research methods. It has been highlighted many times as an exemplar of PPI. 200–202 Men who declined randomisation in the feasibility study explained in interviews how the study should be organised, and we incorporated these into the design: the conservative study group was called ‘active monitoring’ rather than ‘watchful waiting’, there were three rather than two groups and nurses were the primary recruiters and reviewers of men’s progress. The participants in the core serial interview study have formed a PPI-like ‘sounding board’ to give views about possible changes to the study, such as e-mail contact would not be acceptable, questionnaires needed to be shorter and that they highly valued the study and contact with research nurses during the follow-up process.

Experiences of trial participation

Men’s experiences of trial participation were collected throughout the study and contributed to improvement of recruitment processes, questionnaire revision and the evaluation of nurse-led AM.

Patient and public involvement consultation in 2016

A PPI consultation was conducted when the median 10-year follow-up outcomes were published. 54 The aim of the PPI consultation was to ask men in ProtecT what information they thought should be provided to patients newly diagnosed with localised prostate cancer regarding possible treatments in order to support better-informed treatment decisions. The aim was to establish what information men are likely to want and how best to present this information.

This consultation involved ProtecT study participants and men with prostate cancer outside the study. Two patient advisory groups (PAGs) were convened in June 2017, comprising:

• nine ProtecT participants and three female partners

• two men (who were not ProtecT study participants) from the NIHR Bristol Nutrition Biomedical Research Unit (BRU) prostate cancer PPI group – a well-established local group.

The ProtecT study results newsletter (originally sent to all ProtecT study participants in September 2016) was given to PAG participants to read in advance. Group meetings began with a short presentation summarising the ProtecT study and its key findings. PAG participants then commented on study findings, presented as profiles relating to each treatment. For example, for men aged 60–69 years, profiles showed the risks of experiencing treatment harms (erectile dysfunction, urinary incontinence or bloody stools) and risk of cancer spread and risk of death from prostate cancer or other causes according to allocation group. We asked group participants to comment on how helpful these tables might be to men, in particular for men making decisions regarding treatment for newly diagnosed localised prostate cancer. We asked group participants for their views on what information should be included and how best to present it.

All participants found the study findings reassuring, particularly the number of ProtecT study participants alive 10 years after their prostate cancer diagnosis. One man was concerned that these positive findings might lead to complacency about the threat posed by prostate cancer. Both groups believed that awareness of prostate cancer had grown over the previous 10 years, partly because of media and charity campaigns. Some participants suggested that the impact of treatment side effects needed to be clearer, for example that findings need to be backed up by ‘facts and figures’ and that more information was needed on the impact across ethnic groups. One of the BRU PAG participants felt that it was important to recognise that prostate cancer was not exclusively an ‘old man’s disease’. Participants argued that it was important to be aware that individuals will interpret findings from their own perspective, for example thinking of 1 in 100 men in their own age group. They suggested that messages from the ProtecT study should make clear that findings only included men with localised prostate cancer and referred only to men between the ages of 50 and 69 years at diagnosis. They thought that showing findings relevant to a particular age group (e.g. men aged 50–59 years) would be more helpful than showing findings for all ProtecT participants overall. They agreed that headline findings should make clear that the same favourable survival rates were found for all three treatment groups. Participants discussed their understanding of the terminology used in the reports of study findings, in particular the use of the terms ‘cancer growth’ and ‘cancer spread’ to refer to cancer progression and cancer metastases. Most participants generally understood these terms and found them acceptable. One man who worked with a prostate cancer support group felt more comfortable with the terms ‘progression’ and ‘metastases’, whereas several felt that using all of the terms was useful. Some were uncertain regarding whether or not when prostate cancer spread it always spread to the bones. The numbers in the study experiencing cancer growth and spread were thought to be clear, but use of a diagram to illustrate growth within and spread beyond the prostate gland was found to be very helpful.

Some participants found that use of the word ‘bothersome’ in relation to the impact of treatment side effects diminished their experiences: ‘I think the problems you have after surgery are slightly more than bothersome’. They felt that it would be sufficient to state that side effects experienced were different between groups and to give numbers. One participant described ‘risk’ as an emotive word. Men observed that referring to ‘increased risk’ with regard to cancer growth and spread was unhelpful if not accompanied by numbers giving the difference in risk, for example stating that the risk was ‘doubled with AM and the original risk is 1%, you’ve [only] got a 2% chance’. Others argued that the figures ‘speak for themselves’. Participants highlighted that there had been changes in treatments and technology, such as the development of robotic surgery. Group members argued that this had an impact on the way that men interpreted the findings of ProtecT and the way that they considered future treatment choices for men diagnosed with localised prostate cancer, and so they thought that information in relation to these changes would be needed.

The PAG participants were also asked to indicate what they felt was the best way of providing information to men diagnosed with localised prostate cancer to support treatment decision-making. They varied considerably in their views on the format and content of the outcome profiles. Some argued that profiles were helpful, that it was important to have ‘facts and figures’ about possible side effects to help men with treatment decision-making and that presenting data on side effects ‘brings home quite dramatically what the key problems are’. They were reassured about the credibility of the data presented as data came from the ProtecT study, which was perceived as a trusted source. Others felt that there was ‘too much’ information, with one of the patients’ partners commenting that they would ‘immediately switch off, because numbers mean nothing to me’ and that a ‘picture’ would be better.

Participants commented on the presentation of side effects. One man argued that it was important to include the need for pads as this represented a significant ‘lifestyle change’ and also to include data on erectile dysfunction, but noted the lack of information on psychological impact.

In terms of the delivery of information, some participants felt that although written profiles were helpful, it was important that any such information was presented by a qualified health professional, such as a clinical nurse consultant, who could talk through the data and answer questions. Men could then make an ‘informed choice’. One participant explained that he would rather ‘leave it to the expert’ to give him advice on the best treatment for him, ‘not look at a sheet of figures’.

They considered that the timing of information-giving was important, with men likely to be very anxious following diagnosis and finding it difficult to retain information: ‘at the beginning, too much – information overload’.

The PAG participants were shown existing examples of information that used visual representations to convey messages and they were asked to comment on the formats. One BRU group member commented that using symbols such as smiley faces would be more appropriate for conveying information to children or people with learning difficulties. Some group members found it difficult to comment on examples that were not relevant to their own experience; others felt that cancer progression could be usefully explained using visual aids. There was a recognition that people have ‘different ways of absorbing data’ so that ‘a graph like this might be good for some people, straight pictures for others and others might like a good description’ and that any presentation of information should cover all of these options.

Risk stratification of prostate cancer: the unmet need

Accurate stratification of risk at the time of diagnosis is problematic for several reasons. There may be staging errors on MRI, biopsies may not always provide a reliable estimate of the risk of the cancer132–134,204–207 and PSA is not a reliable marker of metastasis, particularly at levels below 20 ng/ml. We need better methods of stratifying risk of prostate cancer, particularly for men at low and intermediate risk, so that we can advise men about appropriate treatment. As noted in earlier chapters, such methods will rely on well-annotated biorepositories derived from RCTs to provide samples for development and validation of biomarkers. Even for men at high risk, we cannot accurately identify those who require multimodal treatment, although some men benefit from early docetaxel (Taxotere^®, Sanofi-Aventis) treatment. 208

Recent translational integrative genomic studies demonstrate that prostate cancer can be categorised into subgroups209,210 that may respond to therapy in different ways. Our contention is that the ProtecT biorepository addresses the unmet need of providing samples with the right context and annotation to allow better biomarkers to be developed.

ProtecT is the largest RCT on the management of localised prostate cancer that compares surgery, radiotherapy and active monitoring/surveillance, which was designed using robust epidemiological and statistical approaches. ProtecT is unique because it has randomised men recruited from the community between three treatments and because it has generated a bioresource from asymptomatic, community-based men whose prostate cancer was diagnosed following PSA testing and whose PSA history is fully understood and documented. In the UK generally, opportunistic PSA testing is uncommon (≈8% of all men of that age). 38 As a consequence of randomisation, ProtecT differs from cohort studies used to develop and test biomarkers. A cohort study might allow analyses to show that a marker predicts mortality and outcome within treatment groups. However, ProtecT would allow investigation of the more important question of whether or not using that marker (or markers) to determine if a particular treatment would have led to better outcomes (fewer deaths and/or fewer men treated). It should provide precision genomic or protein biomarkers to assist men and their clinical advisors to choose the best individualised approach to management.

In addition, within ProtecT we have large numbers of community-based cancer-free controls who also add significant value to the repository. These controls have been flagged with the Office for National Statistics (ONS) to determine their cause of death and with the NHS Digital to determine details of any cancer diagnosis and treatment via the appropriate cancer registries. We estimate that ≈5% of these 60,000 men (≈3000) will eventually be diagnosed with prostate cancer, particularly those with a PSA level of between 1 µg/l and 3 µg/l. At the time of writing, 1391 of these men had been diagnosed with prostate cancer (126 had an initial PSA level of < 1 µg/l, 780 had an initial PSA level of 1–3 µg/l and 485 had an initial PSA level of > 3 µg/l but a negative biopsy) and 97 of them had died and had prostate cancer reported as the immediate or underlying cause of death (16 had an initial PSA level of < 1 µg/l, 62 had an initial PSA level of 1–3 µg/l and 19 had an initial PSA level of > 3 µg/l but a negative biopsy). Over 80% of this cohort of control men who later developed prostate cancer have baseline blood samples that are included in our biorepository.

Advanced cases of prostate cancer excluded from the ProtecT trial

A further 479 men in ProtecT were diagnosed with locally advanced prostate cancer and so were excluded from the randomised treatment trial. This is reported in more detail later in this chapter. These advanced cases are very important because these men were diagnosed months or years earlier than they would have been without PSA testing, being asymptomatic at the time of diagnosis. They were thus able to receive treatment earlier than those presenting with clinical symptoms. To our knowledge, no other trials have published data on such men. Figure 21 shows the outcome of this important group over 10 years, compared with a comparison group of patients with locally advanced prostate cancer from the National Cancer Registration Office (Eastern Office). Over 300 of these men have samples in the ProtecT biorepository.

FIGURE 21.

Survival in ‘advanced’ cases compared with a matched cohort from the Anglia Cancer Registry. 211 3A prostate cancer specific survival. ACN, Anglia Cancer Network.

What fluid samples do we have?

The following minimum samples were collected at baseline venepuncture, with up to 50 ml of blood taken and samples being stored at the University of Oxford:

• × 1 whole blood for germline DNA (≈2.5 ml)

• × 1 serum sample (≈4 ml)

• × 2 plasma (× 2 EDTA and × 2 heparin plasma) (≈4 ml each)

• urine for proteins and cells (50 ml) – the majority taken only during follow-up.

As noted above, baseline samples (whole blood, two plasma samples and one serum sample) were obtained for over 3000 cases in ProtecT (including the excluded men and those with advanced prostate cancer). For baseline samples, three of the four tubes still required aliquoting (serum × 1, plasma × 2) as most of the whole blood in the samples that were stored in Cambridge [by Thermo Fisher Scientific (Waltham, MA, USA)] had been converted already to germline DNA. For the 3000 cases, there are about 9000 tubes of samples.

For the 60,000 control men, ≈14,000 men have had whole blood converted to germline DNA, and there are 180,000 tubes of serum and plasma samples. There are four baseline samples on 1000 additional control men who subsequently developed prostate cancer during the 10-year follow-up period. In ProtecT, we have additional follow-up serial sample sets on the majority of men (with an average of five serial sample sets in ProtecT men; ≈37,000 plasma and serum samples over a 10-year period), which have been collected. There are also 13,500 follow-up urine samples for ProtecT cases.

In ProtecT, 2694 individual men with prostate cancer donated baseline samples that have been aliquoted. A further 417 men who donated baseline samples remain to be aliquoted. Over 8000 tubes from participants with prostate cancer and 13,000 tubes from control participants have been aliquoted. Germline DNA has been extracted from ProtecT cases (2505 men) and controls (13,844 men), and aliquoted into 96 well plates.

Method of processing of samples

The sample processing laboratory incorporates automated liquid-handling robots, a –80 °C freezer store and liquid nitrogen storage. Barcoded, standardised input serum and plasma samples are thawed as per relevant SOPs and aliquoted into final working volumes by sample-formatting robots. Each sample is barcoded with a unique Cryo-Safe number and automatically logged by the sample-formatting robot into the Prostate Cancer: Mechanisms of Progression and Treatment (ProMPT) sample tracking database so that there is a full audit trail from venepuncture at clinic to use in the laboratory. Whole-blood samples are frozen and stored in vacutainers until they are required for extraction (Figure 22).

FIGURE 22.

Method of sample processing.

The availability of the samples is coded by a ‘traffic light system’ according to value and amount:

• red (≤ 10 tubes of 110 µl)

• amber (11–20 tubes of 110 µl)

• green (≥ 21 tubes of 110 µl).

Quality control and standard operating procedures for stored materials

Each stage of sample collection, processing and storage is standardised and quality assured by an internal quality management system encompassing all relevant quality documentation, such as policies, SOPs, risk assessment and COSHH (Control of Substances Hazardous to Health Regulations). Audits of the sample collection are carried out to a predetermined schedule. In Oxford, samples derived from blood or urine are kept in cryogenic (–80 °C) freezers. Each freezer is supported by an around-the-clock temperature monitoring and reporting system, and the facility has an automatic switch system generator that is capable of running for 72 hours in the event of power failure. Biorepository staff are on hand 365 days a year to attend emergency freezer call-outs and there are dedicated emergency freezers available for the transfer of samples from failed units. In Cambridge, the samples are stored by Thermo Fisher Scientific at the Bishop’s Stortford site, but these are then transferred to Oxford in the near future. Thermo Fisher Scientific is the leading provider of biospecimen storage and biobanking services, storing over hundreds of millions of samples in more than 20 global repositories. Thermo Fisher Scientific complies with current Good Manufacturing Practices (cGMPs), Good Tissue Practices (GTPs) and Good Laboratory Practices (cGLPs). Cold storage units in the biorepository are monitored at all times. The quality assurance department certifies the temperature probes to National Institute of Standards and Technology (NIST) traceable standards every 6 months to ensure correct temperature readings. Thermo Fisher Scientific also meet US Food and Drug Administration (FDA) requirements for temperature compliance and has data to show that materials have been maintained at appropriate temperatures. Should the temperature of any unit deviate from its acceptable range, the monitoring system will automatically notify repository personnel or on-call staff, day or night. The exterior perimeters are typically monitored by video cameras. The facilities are protected from fire by sprinkler systems that automatically call the local fire station on activation.

Tissue resource

Our collaboration includes strong and long-standing input from the pathology subgroup of ProtecT (chaired initially by Dr Mary Robinson, now by Dr Jon Oxley). There was also strong input from two histopathologists in Oxford and Cambridge (Dr Clare Verrill and Dr Anne Warren). The FFPE histopathological material is being centralised. The material from the FFPE material from the RP samples can be used for both tissue microarrays and extracting DNA. We have developed methods for processing fresh tissue from radical samples, which has already been shown to be of great use in terms of next-generation sequencing approaches that have led to high-impact papers. This has been fully described previously,212,213 and is shown in brief in Figure 23. This careful approach has allowed multiple samples to be taken from each prostate and allowed preservations of margins for routine pathological assessment as well as appropriate tissue for the extraction of nucleic acids. These have been used in recent high-impact publications. 209,210

FIGURE 23.

Tissue sampling from fresh RP specimens. The prostate is shown intact (A) and inked (B). It was then sliced using a parallel-blade device (C). Multiple punch biopsies were removed and the sites of the punched cores marked on a ‘map’ diagram (D and E), leaving margins intact. The holes in the transverse slice were inked, using two coloured inks in a random pattern that was recorded on the map to enable easy identification. The cylindrical cores of tissue were placed onto individual, pre-numbered foil squares and snap frozen in liquid nitrogen before transferring immediately to pre-cooled vials and stored at −80 °C (F). The fresh slice and other portions were pinned onto a cork board to avoid warping during subsequent fixation (G). Republished with permission of Wiley Periodicals, Inc., from Method for sampling tissue for research which preserves pathological data in radical prostatectomy, Warren et al. ,212 Volume 73, edition 2, © 2012; permission conveyed through Copyright Clearance Center, Inc.

Republished with permission of Wiley Periodicals, Inc., from Method for sampling tissue for research which preserves pathological data in radical prostatectomy, Warren et al. , Volume 73, edition 2, © 2012;212 permission conveyed through Copyright Clearance Center, Inc.

Abstract

Reproduced from Johnston et al. 211 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) license. See: http://creativecommons.org/licenses/by-nc-nd/4.0/.

Background: Early detection and treatment of asymptomatic men with advanced and high-risk prostate cancer (PCa) may improve survival rates.

Objective: To determine outcomes for men diagnosed with advanced PCa following prostate-specific antigen (PSA) testing who were excluded from the ProtecT randomised trial.

Design, setting, and participants: Mortality was compared for 492 men followed up for a median of 7.4 yr to a contemporaneous cohort of men from the UK Anglia Cancer Network (ACN) and with a matched subset from the ACN. Outcome measurements and statistical analysis: PCa-specific and all-cause mortality were compared using Kaplan–Meier analysis and Cox’s proportional hazards regression.

Results and limitations: Of the 492 men excluded from the ProtecT cohort, 37 (8%) had metastases (N1, M0 = 5, M1 = 32) and 305 had locally advanced disease (62%). The median PSA was 17 mg/l. Treatments included radical prostatectomy (RP; n = 54; 11%), radiotherapy (RT; n = 245; 50%), androgen deprivation therapy (ADT; n = 122; 25%), other treatments (n = 11; 2%), and unknown (n = 60; 12%). There were 49 PCa-specific deaths (10%), of whom 14 men had received radical treatment (5%); and 129 all-cause deaths (26%). In matched ProtecT and ACN cohorts, 37 (9%) and 64 (16%), respectively, died of PCa, while 89 (22%) and 103 (26%) died of all causes. ProtecT men had a 45% lower risk of death from PCa compared to matched cases (hazard ratio 0.55, 95% confidence interval 0.38–0.83; p = 0.0037), but mortality was similar in those treated radically. The non-randomised design is a limitation.

Conclusions: Men with PSA-detected advanced PCa excluded from ProtecT and treated radically had low rates of PCa death at 7.4-yr follow-up. Among men who underwent nonradical treatment, the ProtecT group had a lower rate of PCa death. Early detection through PSA testing, leadtime bias, and group heterogeneity are possible factors in this finding. Patient summary: Prostate cancer that has spread outside the prostate gland without causing symptoms can be detected via prostate-specific antigen testing and treated, leading to low rates of death from this disease.

Death from prostate cancer occurred in two (4%) of the RP and 12 (5%) of the RT group (HR 0.95, 95% CI 0.22–4.12; p = 0.94). Death from all causes occurred in four (7%) of the RP and 37 (15%) of the RT group (HR 0.69, 95% CI 0.29–1.67; p = 0.41). A significantly greater proportion of the ADT group died from prostate cancer (n = 27, 22%) and all causes (n = 49, 40%) compared to men treated with radical therapy (p < 0.0001). RP = radical prostatectomy; RT = radical radiotherapy; ADT = androgen deprivation therapy; HR = hazard ratio; CI = confidence interval.

By the end of the study, 37 matched cases (9%) and 64 controls (16%) died from prostate cancer. Death from all causes occurred in 89 cases (22%) and 103 controls (26%). HR = hazard ratio; CI = confidence interval.

In summary then, this paper has added significantly to our understanding about what happens following a single round of PSA testing in a community which essentially had not previously been exposed to such testing. The advanced men have done extremely well from this early detection and achieved good survival rates.

Disease progression

Baseline clinicopathological features of men with localised prostate cancer within ProtecT differed in men who developed disease progression compared with those with stable disease, but associations were not strong enough to reliably predict progression in individuals. This indicates that methods of stratification that are currently employed need to be refined with pre-biopsy imaging and targeted sampling, as well as utilisation of validated genomic and other emerging biomarkers. These will, however, need to be evaluated in new large-scale prospective early-detection programmes.

Qualitative research

The qualitative research integrated into the feasibility study and main recruitment enabled a nuanced understanding of the recruitment process, and the information developed from that understanding and the training and support given to research nurses and urologists enabled the ProtecT trial recruitment to be completed successfully. The recruitment intervention was later developed into the QRI, now being applied in 30 other RCTs with difficult recruitment challenges. Other qualitative research uses a combination of single and serial (repeated) in-depth interviews to conduct cross-sectional and longitudinal research to explore ProtecT study participants’ experiences of PSA testing, biopsy and outcomes following each of the treatments. Most men taking part in PSA testing perceived themselves to be at low risk of having prostate cancer. The studies of the experience of biopsy concluded that, in future, men needed to be better informed about the biopsy procedure, the relationship with urinary symptoms and the potential incidence and severity of side effects, and a proposal for such information was developed and published. 186

The longitudinal research captured how men’s experiences evolved over up to 14 years of follow-up. Appointments with health professionals (primarily nurses, supported by urologists and oncologists as required) to review PSA levels and disease progression were found to be critical for the management of men in the AM group. These appointments enabled and supported men to achieve their particular balance between uncertainty and reassurance and staying on AM or changing to a radical treatment. The ProtecT AM service protocol is available for implementation. Men undergoing radical treatments felt that they could have been better informed about the timing of hormone therapy or radiotherapy, the levels of impairment they might face and for how long impairment might last after radical treatment. Work is continuing to include ProtecT outcome data in materials provided to men to support decision-making.

Patient and public involvement

Patient and public involvement was not undertaken in the standard way in ProtecT, but the views and opinions of men involved in the qualitative research have been integrated into the design and conduct of the study from the outset. In addition, in 2016, a PPI consultation was undertaken with ProtecT patient representatives and members of the public about the interpretation of the main ProtecT outcome findings. This consultation provided a wealth of information about the type of information about prostate cancer and its treatment that men would like to have access to for decision-making, including the timing of provision and modes of presentation, which will feed into the materials being developed.

Translational research

The ProtecT trial participants were a unique cohort of men, recruited in a community setting where PSA testing was rare. Tissue from biopsies and surgery specimens (FFPE material) were collected from 2500 men with prostate cancer, and blood samples (plasma, serum and germline DNA) were also collected from these men and from ≈60,000 men who underwent PSA testing but were below the threshold (‘controls’). These samples have been carefully stored and 14,000 have had whole blood converted to germline DNA. With full sociodemographic, lifestyle, dietary, clinical, genetic, epidemiological and outcomes data, the ProtecT biorepository represents a key resource for future testing of biomarkers and screening and treatment strategies.

Principal investigators

Freddie C Hamdy (Chief Investigator), Jenny L Donovan and David E Neal.

Trial co-ordinator

J Athene Lane.

Trial statisticians

Tim J Peters and Chris Metcalfe.

Urologists

Leads: Prasad Bollina, Andrew Doble, Alan Doherty, David Gillatt, Roger Kockelbergh, Howard Kynaston, Alan Paul, Philip Powell, Stephen Prescott, Derek Rosario and Edward Rowe.

Others: John B Anderson (deceased), Jonathan Aning, James Catto, Garett Durkan, Owen Hughes, Anthony Kouparis, Hing Leung, Param Mariappan, Alan McNeill, Edgar Paez, Raj Persad, Hartwig Schwaibold, David Tulloch and Michael Wallace.

Nurses

Leads: Susan Bonnington, Lynne Bradshaw, Deborah Cooper, Emma Elliott, Phillipa Herbert, Peter Holding, Joanne Howson, Amanda Jones, Teresa Lennon, Norma Lyons, Hilary Moody, Claire Plumb, Tricia O’Sullivan, Elizabeth Salter, Pauline Thompson, Sarah Tidball, Jan Blaikie, Catherine Gray.

Others: Tonia Adam, Sarah Askew, Sharon Atkinson, Tim Baynes, Carole Brain, Viv Breen, Sarah Brunt, Sean Bryne, Jo Bythem, Jenny Clarke, Jenny Cloete, Susan Dark, Gill Davis, Rachael De La Rue, Jane Denizot, Elspeth Dewhurst, Anna Dimes, Nicola Dixon, Penny Ebbs, Ingrid Emmerson, Jill Ferguson, Ali Gadd, Lisa Geoghegan, Alison Grant, Collette Grant, Rosemary Godfrey, Louise Goodwin, Susie Hall, Liz Hart, Andrew Harvey, Chloe Hoult, Sarah Hawkins, Sharon Holling, Alastair Innes, Sue Kilner, Fiona Marshall, Louise Mellen, Andrea Moore, Sally Napier, Julie Needham, Kevin Pearse, Anna Pisa, Mark Rees, Elliw Richards, Lindsay Robson, Janet Roxburgh, Nikki Samuel, Irene Sharkey, Michael Slater, Donna Smith, Pippa Taggart, Helen Taylor, Vicky Taylor, Ayesha Thomas, Briony Tomkies, Nicola Trewick, Claire Ward, Christy Walker, Ayesha Williams, Colin Woodhouse and Elizabeth Wyber.

Oncologists

Lead: Malcolm Mason.

Others: Amit Bahl, Richard Benson, Mark Beresford, Catherine Ferguson, John Graham, Chris Herbert, Grahame Howard, Nick James, Peter Kirkbride, Alastair Law, Cgroupel Loughrey, Duncan McClaren, Helen Patterson (deceased), Ian Pedley, Trevor Roberts (deceased), Angus Robinson, Simon Russell, John Staffurth, Paul Symonds, Narottam Thanvi, Subramaniam Vasanthan and Paula Wilson.

Histopathologists

Leads: Jon Oxley and Mary Robinson.

Others: Selina Bhattarai, Neeta Deshmukh, John Dormer, Malee Fernando, John Goepel, David Griffiths, Ken Grigor, Patricia Harnden, Nick Mayer, Murali Vgroupa and Anne Warren.

Radiologists and medical physics

Helen Appleby, Dominic Ash, Dean Aston, Steven Bolton, Graham Chalmers, John Conway, Nick Early, Tony Geater, Lynda Goddall, Claire Heymann, Deborah Hicks, Liza Jones, Susan Lamb, Geoff Lambert, Gill Lawrence, Geraint Lewis, John Lilley, Aileen MacLeod, Pauline Massey, Alison McQueen, Rollo Moore, Lynda Penketh, Janet Potterton, Neil Roberts, Helen Showler, Pam Shuttleworth, Stephen Slade, Alasdair Steele, James Swinscoe, Marie Tiffany, John Townley, Jo Treeby, Michael Weston, Joyce Wilkinson, Lorraine Williams, Lucy Wills, Owain Woodley and Sue Yarrow.

Other researchers and data managers

Lucy Brindle, Linda Davies, Michael Davis, Dan Dedman, Elizabeth Down, Kirsty Garfield, Hanan Khazragui, Richard M Martin, Sian Noble, Hilary Taylor, Marta Tazewell, Emma L Turner, Julia Wade, Eleanor Walsh and Grace Young.

Administrative support

Susan Baker, Elizabeth Bellis-Sheldon, Chantal Bougard, Joanne Bowtell, Catherine Brewer, Chris Burton, Jennie Charlton, Nicholas Christoforou, Rebecca Clark, Susan Coull, Christine Croker, Rosemary Currer, Claire Daisey, Gill Delaney, Rose Donohue, Jane Drew, Rebecca Fgrouper, Susan Fry, Jean Haddow, Alex Hale, Susan Halpin, Belle Harris, Barbara Hattrick, Sharon Holmes, Helen Hunt, Vicky Jackson, Donna Johnson, Mandy Le Butt, Jo Leworthy, Tanya Liddiatt, Alex Martin, Jainee Mauree, Susan Moore, Gill Moulam, Jackie Mutch, Kathleen Parker, Christopher Pawsey, Michelle Purdie, Teresa Robson, Lynne Smith, Carole Stenton, Tom Steuart-Feilding, Beth Stott, Chris Sully, Caroline Sutton, Carol Torrington, Zoe Wilkins, Sharon Williams, Andrea Wilson and Ashleigh Weaver.