Development, validation and evaluation of an instrument for active monitoring of men with clinically localised prostate cancer: systematic review, cohort studies and qualitative study

Prostate cancer

For men in the UK, PCa accounts for 24% of cancers diagnosed and 14% of cancer mortality;1 yet, currently, its natural history is not fully understood. Autopsy results have shown that approximately half of men over 50 years of age have some form of PCa. 2 However, based on data from the USA, there is a 16.5% lifetime chance of being diagnosed with, and a 3% chance of dying from, PCa. 3,4

Screening for prostate cancer

Prostate cancer is one of the most common newly detected cancers in men worldwide, primarily because of the increasing use of PSA as a screening test. 5 PCa screening is, however, a controversial issue in contemporary health care. 6 Screening using PSA can detect cancers that are still confined within the prostate7 when radical prostatectomy or radiotherapy could possibly achieve a cure. However, current tests cannot differentiate between tumours with biological potential for progression and the majority of slow-growing tumours, which will not cause clinical disease in a man’s lifetime. The recent publication of the results of screening trials in Europe and the USA have further fuelled the controversies in this area, showing, for example, that 1410 men needed to be screened to find 48 cancers to prevent one cancer death. 8

Treatment for clinically localised prostate cancer

Men diagnosed with clinically localised PCa are often treated by either radical prostatectomy (surgery to remove the prostate) or some form of radiotherapy. Radical treatments can cause serious side effects, particularly incontinence and impotence. 9 There is currently a lack of robust evidence about the comparative effectiveness of treatment for PCa. A recent systematic review10 sought to compare monitoring with radical treatment but found no comparative trials. Two randomised trials have evaluated the effectiveness of a passive strategy called watchful waiting (WW). WW historically involved palliative treatment once symptoms appeared and was aimed at older men who were not suitable for surgery or radiotherapy. The Prostate cancer Intervention Versus Observation Trial (PIVOT) recently found no difference between WW and radical prostatectomy for either all-cause mortality or PCa-specific mortality (PCSM) after at least 10 years of follow-up11 for men with PSA-detected PCa. The Scandinavian Prostate Cancer Group study number 4 (SPCG4) trial12 found that radical prostatectomy reduced PCSM compared with WW among men who had been diagnosed clinically (rather than screen detected as in PIVOT). This trial started recruitment prior to the use of PSA and, consequently, the majority of participants had clinically detected disease. 13 The Prostate testing for cancer and Treatment (ProtecT) trial is comparing AM with radical treatment (surgery and radiotherapy) in men with PSA-detected clinically localised PCa, but the long periods of follow-up required for clinically relevant outcomes to be observed means that it will not report for at least another 2 years. 14,15

Interest in the safety and acceptability of monitoring for clinically localised PCa, which avoids unnecessary intervention, has grown over the past decade. Although it is accepted that AM protocols involve closer monitoring than that which is employed in the WW arm of SPCG4, recent systematic reviews have concluded that there is little evidence or expert consensus over the most effective monitoring protocol. 10,16–18 Serial measures of PSA level are used consistently, but various aspects of PSA kinetics are used to trigger further clinical review, including PSA doubling time [(PSADT) length of time for PSA to double] and PSA velocity [(PSAv) rate of PSA change per year]. Hence, there is increasing interest in using PSA levels to monitor men with low-risk clinically localised tumours and in using strategies that would indicate when further clinical review and radical treatment would be appropriate and when men could remain monitored without intervention. 19

Active monitoring or surveillance

Active monitoring and active surveillance (AS) have developed as management strategies for men with clinically localised PCa to avoid overtreatment of low-risk disease. 20 AS involves regular follow-up by PSA testing, digital rectal examination (DRE), review of symptoms and scheduled repeat biopsy, whereas AM only uses a rebiopsy in the presence of changing PSA or DRE findings, because a rising PSA can be indicative of worsening PCa. Under an AM protocol, men with stable PSA remain on monitoring, whereas those with rapidly rising PSA are recommended for clinical review. At clinical review, a biopsy of the prostate can be performed and, depending on the specific protocol, a patient may be recommended for or suggested to consider radical intervention (e.g. radiotherapy or prostatectomy) or continuation on AM. The key tenet of AM/AS is to balance the risk of harm of intervention in men with non-lethal cancer, with the risk of not intervening where cancer is progressing. A previous systematic review16 based on five AM/AS cohorts20–24 found little consensus on the eligibility criteria, optimum protocol for monitoring, the trigger(s) for radical intervention or the role of PSA kinetics in monitoring. 16 Two more recent systematic reviews10,17 again found no consensus on eligibility criteria for AM/AS or the AM/AS regimens used.

Prostate-specific antigen kinetics

In order to use PSA in AM/AS, a model for normal PSA levels is needed as a comparator for observed PSA levels in men on AM/AS. PSADT, PSAv and absolute level of PSA are commonly used measures for monitoring a man’s PSA level. 18 There is little consensus on which of these to use or what threshold should be employed for each measure. There remains an absence of clinical or statistical evidence for their use in AM, and retrospective analyses have found very little association with clinical outcomes, such as metastases or PCSM. 18,25–28 Furthermore, there are concerns about the various methods of calculation of PSADT29,30 and PSAv,31,32 as well as a great deal of variation and uncertainty about how many PSA values should be used for calculation. 33 PSA levels increase naturally with age, so a method is needed to indicate when increases in PSA are beyond normal age-related change to avoid reviews being triggered when they are not necessary.

Modelling prostate-specific antigen change with age

Modelling PSA in AM/AS has received little attention in the literature,33–36 despite the fact that repeated measures of PSA from a cohort of men represent a wealth of information about the behaviour of PSA. An accurate model for PSA change would allow men to observe where they lie in comparison to the general trend of PSA in similar men and whether or not any increase in their PSA is consistent with normal age-related change. Multilevel models37,38 allow such data to be fully exploited by modelling the variation in PSA between men in an AM cohort, as well as the variation in PSA within each individual during follow-up. We have previously used multilevel models to model PSA change with age in men without cancer and compared this with men with screen-detected and clinically presenting cancer. 35 Limitations of this previous work include the use of only two cohorts of men with cancer and the derivation of the model on cancer-free men which meant that potential explanatory variables such as Gleason score, or age at diagnosis, were not included in the model. Initial examination of the comparison of observed PSA to that predicted for a cancer-free man of the same age indicated that this model might result in fewer alerts for clinical review. 35 However, clinical outcomes were not available for comparison. In order to study the effect of PSA change on clinical outcomes, it may be necessary to combine AM/AS cohorts, because very few deaths or cases of metastases have been reported at this stage. 10,16,17

Qualitative research in active monitoring/active surveillance

Qualitative methodology has not been widely applied in AM/AS research. Few studies have considered patients’ and clinicians’ views and experiences of AM/AS in depth, and the published literature largely refers to protocols that more closely follow AS than AM.

A considerable proportion of the qualitative literature relating to AM/AS considers men’s treatment decision-making. Qualitative evidence largely places clinician recommendations and communication as principal drivers for patients’ decisions to opt for or against AM/AS. This has tended to be framed in the literature as patients deferring the decision-making process to the clinician,39–42 or justifying their decision to opt for AM/AS in light of clinicians’ recommendations. 39,41,43 Similar decision-making tendencies are seen in earlier studies of localised PCa management that do not make specific reference to AM/AS. 44,45 Of these studies, one suggests that passing on the decision-making power to the clinician can be an active choice on the man’s part. 45 Qualitative interviews also suggest that men come to the decision to choose AM/AS by reading cues from their clinicians (e.g. descriptive terminology of the low-risk status of the cancer). 39,41

The qualitative findings provide insight into how clinicians may shape the decision-making process, but do not represent the prevalence of treatment decision-making preferences. A review of the literature found that men prefer to be actively involved in the treatment decision-making process. 46 Despite this, there is a substantial body of evidence that suggests that men’s values and preferences are not reflected in the final treatment choices made. 47 A systematic review of treatment decision-making concluded that men’s treatment choices tend to be reflective of information they receive, rather than their underlying values and preferences. 48

Only one qualitative study has considered AM/AS men’s information needs, focusing broadly on psychosocial and educational needs. The single focus group study found that men had poor knowledge of their disease and tended to turn to the internet as their main source of information. 49 A number of studies have designed decision-support interventions to increase men’s involvement in, and satisfaction with, decision-making in relation to treatment of PCa through increasing knowledge and education. Successful interventions have included providing audio-recordings of consultations, presenting individualised information to men and their partners,50,51 and providing counselling via nurses. 52

A handful of qualitative studies have sought to explore men’s experiences of being on AM/AS. These have tended to focus on psychosocial issues, particularly issues of uncertainty. Two qualitative interview-based studies in Sweden and Canada report uncertainty as a major theme to emerge from interviews with men on AM/AS. 53,54 Men talked about uncertainty across a number of domains, including disease progression, prognosis, symptom interpretation and morbidity associated with potential imminent treatment. However, there was a suggestion that uncertainty was temporal, peaking at times of PSA testing or rebiopsy. Receipt of reassuring test results enabled men to ‘continue with their normal lives’. The study by Oliffe et al. 53 also reported that men’s strategies for coping with uncertainty included attempting to ‘live a normal life’ and to do all they can to assert control over their bodies (i.e. through lifestyle and diet changes, and use of complementary/alternative therapies).

Some qualitative studies have intentionally focused on the issue of uncertainty in AS, rather than identified this as a key theme through grounded theory approaches. 55 A study by Bailey et al. analysed men’s accounts of being on AM/AS through the lens of a pre-existing theoretical model of uncertainty. 55 The authors identified uncertainty surrounding cancer progression and the questioning of whether or not one had made the ‘right choice’ as major themes to emerge from interviews. Linked with the uncertainty related to cancer progression was men’s perceptions of how PSA monitoring was not a fail-safe indicator of disease status. 55

The issue of uncertainty has also been identified as a key issue through survey-based research. A survey asked men to provide open responses to a question that asked about the key advantages and disadvantages of being on AM/AS (n = 103). 56 One-third of men reported the risk of disease progression being a key disadvantage, one-quarter made non-specific reference to ‘uncertainty’ and ‘distress’ and 10% talked about the bothersome nature of regular follow-up.

Although research focusing on men’s experience of AM/AS is limited, qualitative studies on decision-making have made some reference to men’s AM/AS experiences. Davison et al. reported that men felt able to ‘live a normal life’, although the authors also emphasise patients’ reported relaxed mentality. 39 A systematic review of psychological effects of AM/AS concluded that men could experience anxiety in response to the prospect of ‘no intervention’, uncertainty and lack of patient education/support. The authors suggest that these concerns are likely to be particularly relevant to men on AM/AS. 57 Overall, the literature on measures of anxiety/psychological distress and/or health-related quality of life suggests there are no significant negative outcomes associated with following an AM/AS programme,58–63 although this can depend on other factors (e.g. personality type, how men conceptualise the role of the clinician, marital status). 59,64 However, it has also been reported that PSA testing and rebiopsies can raise anxiety levels in men on AM/AS, usually temporarily. 65 Furthermore, comparisons of psychological distress/health-related quality of life in men on AM/AS compared with men who undergo radical treatments is somewhat mixed. 59,60,65–67

To our knowledge, there has been no qualitative research that seeks to understand patients’ and/or clinicians’ experiences of AM/AS protocols. More generally, men’s understanding of and views surrounding AM/AS protocols, including thresholds for moving off AM/AS, have not yet been reported. Furthermore, although guidance on AS protocols has been produced by authoritative bodies [e.g. the National Institute for Health and Care Excellence (NICE), the British Association of Urological Surgeons, the European Association of Urology], AM/AS protocols adopted by clinicians have not yet been explored in practice.

NHS policy and practice

The recently published 2014 NICE guidelines68 suggest that men with low-risk localised PCa should be offered AS, which acknowledges that many men would not benefit from radical treatment. NICE recommends a protocol for the management of men on AS, including frequency of PSA testing, DRE and rebiopsy. NICE also suggests that AS could be offered to men with intermediate-risk localised PCa if these men do not wish to undergo immediate radical treatment. There is only a little information about treatments, including AS, as part of the NHS Prostate Cancer Risk Management Programme (see www.cancerscreening.nhs.uk/prostate/faq17.html).

Inclusion criteria

Studies involving men with T1–T2 localised PCa, which was initially managed conservatively, were included if pre-defined clinical, pathological or biochemical criteria for clinical review were outlined. Studies involving recurrence after radical prostatectomy or radiotherapy (i.e. not initially managed conservatively) or non-clinical studies were excluded. We excluded any conference abstracts, non-systematic reviews, editorial comments, background papers or studies involving different treatments or diseases of the prostate. Where several papers reporting on the same cohort of men were found, the latest full analysis of the study was included. Title and abstract screening was undertaken independently by AJS, CM, KT and RMM. Each full paper was screened by AJS and independently by one of CM, KT and RMM.

Data extraction

Eligibility criteria, monitoring protocols, sample size, age, PSA, follow-up times, management change triggers, management change rates, metastases, PCSM and reasons for changing treatment were extracted manually from each paper by AJS and checked by one of CM, KT or RMM. Authors of publications found in our search were contacted to provide information where specific values were not given in a publication and to check that data extraction was correct. Management change rates are considered here as key short-term outcome measures for AS, whereas occurrences of metastases or PCa death are considered to be longer-term outcomes.

Data synthesis

Person-years per study were estimated as the median follow-up time multiplied by the sample size for each study; total person-years were the sum of these across included studies. A metaregression was conducted to investigate whether or not elements of study design were associated with the rate of management change per person-years. This rate was calculated for each study as:

Heterogeneity between studies was measured using the I² statistic;69 the higher the I² value, the more between-study heterogeneity is present. The covariates for the metaregression were (1) year of first recruitment; (2) whether or not a Gleason score of 3 + 3 or less was used for eligibility; (3) whether or not a PSA level of 10 ng/ml or less was used for eligibility; (4) the number of PSA tests in the first 3 years; (5) whether or not mandatory rebiopsy was specified in the protocol; and (6) whether or not PSA or PSA kinetic measures, such as PSADT or PSAv, were used to recommend clinical review or radical treatment. Together with univariable analysis, two multivariable metaregressions were performed grouping eligibility variables (1)–(3) and monitoring procedure variables (4)–(6). Average variable effects on management rate are presented alongside 95% confidence intervals (CIs).

Meta-analysis was performed in Stata 1270 (StataCorp LP, College Station, TX, USA) to combine the effects of prognostic factors for management change. Factors were combined if they appeared in more than two studies that reported estimates of the association with change to radical treatment. Where applicable, a sensitivity analysis was used to check if the pooled combination of odds ratios and hazard ratios (HRs) led to different results from just pooled HRs or pooled odds ratios. We also used meta-analysis to study the reasons for management changes. Reasons for changing to radical treatment were given within studies and the proportions of each reason were combined to get an overall average estimate.

Bias in reporting

Publication bias could be an issue in our review, as we examined only abstracts published in English. However, we would expect that studies showing that AM/AS was not a safe option would be more likely to be published in English-language journals, and, therefore, if anything, this systematic review will overestimate the number of adverse events.

Eligibility criteria

Table 1 contains a summary of eligibility criteria, monitoring protocol and baseline characteristics. A combined Gleason score of 6 (3 + 3) or lower was required for entry into 12 of the 22 studies. 71,72,74,77–79,82–84,87–89 A further five cohorts broadened this criterion to include a Gleason score of 7 (3 + 4),75,76,81,85,86 and two allowed up to a Gleason sum of 7,22,80 leaving three cohorts without an explicitly specified Gleason entry requirement. 21,23,73 Additional pathology findings were used with some degree of consistency across the studies. Five studies allowed only those men with < 50% cancer in any single core into the study,74,77,82–84 with one study allowing only 20% involvement in any single core. 78 The number of positive cores allowed varied between an absolute number or a proportion; two studies required no more than half of the biopsy cores to be positive,85,88 whereas seven allowed two or fewer positive cores,74,78,82–84,86,89 and one study raised this to allow those men with three positive cores to enter. 77

Clinical tumour stage (T-stage) was used in 17 cohorts with thresholds of T2 (six cohorts75,76,80,81,83,89), T2a (five cohorts74,77,85,87,88), T1c (three cohorts23,82,84), T1b73 and T1a (two cohorts21,71). There was no consensus among the 15 cohorts that used a PSA threshold for entry, with seven cohorts permitting men with PSA ≤ 10 ng/ml,74,77–79,86,87,89 three cohorts allowing PSA ≤ 15 ng/ml75,85,88 and three cohorts allowing PSA < 20 ng/ml. 76,80,84 Two further studies used PSA density (PSAD) as an eligibility criterion, with limits of PSAD < 0.15 ng/ml/cc82 and PSAD < 0.2 ng/ml/cc. 89 Seven studies did not explicitly use PSA for eligibility,21–23,71–73,83 although six of these involved men diagnosed between 1983 and 199121–23,71–73 (i.e. in the pre- and early-PSA-screening era). Age restrictions were used in just four studies. 74,79,84,88

Surveillance or monitoring protocol

Serial PSA measurement as a measure for surveillance or monitoring during follow-up was common to all studies, but the frequency of measurement was not consistent. Eleven studies maintained PSA testing every 3–6 months throughout;21,71,72,75,77,80–83,86,88 eight reduced the frequency of PSA testing after men had been on surveillance beyond 1 or 2 years. 22,23,73,76,78,84,85,89 Frequent testing occurred in the Kagawa and modern Royal Marsden Hospital (RMH) studies, with men scheduled for 2284 and 1885 PSA tests, respectively, in the first 3 years.

Digital rectal examination was regularly performed in all but five73,79,84,86,88 of the studies, with inconsistency of frequency between the study protocols. In the 17 studies with both DRE and PSA testing, 14 assessed both at the same frequency, whereas the other three had DRE less often than PSA tests. 74,85,89

Repeat biopsy testing was used by 19 studies, with 16 having a scheduled biopsy within 2 years of beginning surveillance. 22,74,77–79,81–89 Biopsies were repeated at least every 2 years in nine studies. 72,78,81–85,87,88 Three studies undertook repeat biopsy only where worsening PSA or DRE results were evident or at the request of the patient or clinician. 21,73,80 Three studies did not include routine rebiopsy or rebiopsy as a result of worsening DRE or PSA. 23,75,76

Triggers for clinical review or radical treatment

Triggers for recommending clinical review and radical treatment are summarised in Table 2. Almost all studies used histological findings such as an increase in Gleason score (18 cohorts22,71–74,77–89), number of positive cores or percentage of cancer on a single core (15 cohorts22,71–74,77–79,82–86,88,89) to recommend radical treatment. Less commonly, T-stage (six cohorts77,79,80,82,84,89) and PSA (13 cohorts21–23,76,77,79–82,84–86,89) were used as triggers to recommend rebiopsy or radical treatment. In four studies, a threshold of PSADT was used, with this being 2 years in two of those studies81,84 and 3 years in the other two studies. 79,86 In one study, radical treatment was offered to men with a PSA velocity > 1ng/ml/year,85 while in another such men were rebiopsied. 71 In two studies, a combination of PSA kinetics with other factors such as DRE and rebiopsy findings was used to recommend radical intervention. 22,87

Radical treatment rates

In 17 studies, up to 33% of men received radical treatment after a median of 3.6 years. 21,23,71–76,78,79,82,83,85–89 In one of these studies, all 16 men remained on monitoring after 7.5 years, having started with very low-risk T1a disease. 71 The University of California, San Francisco intermediate-risk group radically treated 35% of men after 3.9 years;81 the early Memorial Sloan Kettering study radically treated 35% of men after 3.7 years;22 and the later Memorial Sloan Kettering cohort changed the treatment of 36% of men. 77 In Gothenburg, 37% of men received radical treatment after a median follow-up of 6 years. 80 The highest proportion of men who changed from AS to radical treatment was in Kagawa, with 47% leaving surveillance for radical treatment by 4.5 years of follow-up. 84

Study factors associated with the rate of management change

Total person-years were estimated to be 22,545 in the 22 cohorts. The average rate of change of management (found through metaregression) was 84 (95% CI 61 to 106) per 1000 person-years. In the context of the follow-up found here (i.e. roughly 4 years), we estimate that 84 out of 250 men on monitoring would receive radical treatment during four years of follow-up (Figure 2). There was evidence (I² = 96%; p < 0.0005) that the rate of change differed substantially between cohorts. The minimum rate of change was 11 per 1000 person-years21 and the maximum was 218 per 1000 person-years. 88

FIGURE 2.

Forest plot of the proportion of men changing to radical treatment per person-years, sorted by year of beginning recruitment. ERSPC, European Randomized Study of Screening for Prostate Cancer; PRIAS, Prostate Cancer Research International: Active Surveillance; UNC, University of North Carolina.

The rate of management change was greater in studies with recruitment periods that started in more recent years; for each calendar year increase in start date, an extra four men changed treatment per 1000 person-years (95% CI 1 to 7 treatment changes per 1000 person-years; p = 0.014). Admitting men with higher Gleason scores was associated with an average decrease of 50 treatment changes per 1000 person-years (95% CI 9 to 88 treatment changes per 1000 person-years; p = 0.025), compared with studies with eligibility criteria including Gleason score of ≤ 6. Studies that had a mandatory scheduled rebiopsy had, on average, 50 more men change treatment per 1000 person-years (95% CI 7 to 94 treatment changes per 1000 person-years; p = 0.025) compared with those that rebiopsied on worsening results from PSA tests or DRE only. Each of these associations was found in univariable analyses. Two multivariable metaregressions were performed to account for mutual confounding. In these models, the results were slightly attenuated, although there still remains some evidence for associations with Gleason score of > 3 + 3 (46 fewer changes, 95% CI –5 to 96 fewer changes; p = 0.07) and regular biopsy (49 more changes, 95% CI –5 to 98 more changes; p = 0.07).

Reasons for changing management

Eighteen studies reported reasons men had for changing treatment. 21–23,72,76–89 Up to an average of 37% of changes to radical treatments were a result of a reclassified Gleason score (95% CI 23% to 51%),72,77–89 whereas an average of 38% were caused by men meeting PSA, PSADT or PSAv triggers (95% CI 25% to 51%). 76,77,79–81,84,85,89 Other tumour-related reclassifications, such as the number of positive cores, caused an average of 23% of the management changes (95% CI 8% to 37%). 72,77–79,81–83,89 Among those studies detailing reasons, an average of 21% (95% CI 13% to 30%) of treatment changes were a result of patient choice or anxiety. 22,72,76–80,82,84–89 Not all of these four types of reasons (Gleason score, PSA, other tumour and patient choice) were possible for all studies and, hence, the sum of the given averages is over 100%. For example, if study 1 has 100 men who change management from AS to radical treatment, with 50 changing as a result of Gleason score reclassification and 50 changing as a result of a rising PSA, then each reason contributes 50%. In study 2, 100 men receive radical treatment; however, in this study, men can only receive radical treatment as a result of Gleason score. Thus, averaging the two studies, we can either say that Gleason score contributes 75% [= (50% + 100%/2)] and PSA 25% [= (50%+ 0%)/2] of the reason why men receive radical treatment or we can ignore the PSA contribution from study 2 (i.e. 0%) and calculate that Gleason score contributes 75% [= (50% + 100%/2)] and PSA contributes 50% (= 50%/1) of the reasons for a change in management. We present the second method of reporting, which results in a total value over 100%.

Prostate-cancer-specific mortality and metastases

Overall, eight PCa deaths were reported among the 7111 men and 22,545 person-years of follow-up. In Gothenburg, one man changed to hormone therapy after 8.6 years on surveillance; he developed distant metastases and died 12.7 years after diagnosis. 80 Five PCa-specific deaths occurred in the Toronto cohort, with all having a PSADT < 2 years, such that triggers for intervention were met and treatment recommended; two men received radiation therapy, one underwent radical prostatectomy and two refused treatment. The deaths occurred 3.7, 5.2, 5.3, 8.7 and 9.6 years after diagnosis. 79 In the RMH cohort, two men died from PCa 4 years and 8 years after diagnosis, respectively. 85 Both had a PSAv of > 1 ng/ml/year, subsequent Gleason score upgrade on rebiopsy and received androgen-deprivation therapy. Six further cases of metastases have been reported among the remaining men. 21,72,80,89

Study population

The ProtecT study is a large UK-based multicentre clinical trial comparing surgery, radiotherapy and AM in the treatment of screen-detected (PSA ≥ 3 ng/ml) and clinically localised PCa. 14,15 Data were available from 512 men who declined random allocation of their treatment but chose to be managed according to the AM protocol. They had been followed over an average period of 4.8 years [standard deviation (SD) 2.4 years] with a median of 14 visits per individual (7438 total PSA tests). Covariates collected include Gleason score (or grade),94 a histological grading system for PCa, which allows for the two main patterns of tumour-cell structure evident upon biopsy to be graded and combined into a single score. All participants in this cohort have a Gleason score of 6 or 7; however, a Gleason score of 7 can refer to either a 3 + 4 (pattern 3 predominant 4, also present) or a 4 + 3.

We have centred age at 50 years (i.e. considering age 50 years to be the baseline age), given that the inclusion criterion for ProtecT is men aged between 50 and 69 years old. Follow-up of men is irregular, as both the duration and timing of blood tests vary across men and, because of this, classical longitudinal approaches for balanced designs are inappropriate. Furthermore, age, rather than time in the study, is used as the covariate over which change in PSA is estimated. Using age makes the model more portable to populations where age at, and method of, diagnosis for PCa may vary. Furthermore, as there is a great deal of variation in when a diagnosis is made, and, consequently, in when treatment is started in relation to the natural history of the disease, and as no PSA altering treatment is used in this cohort of men, changes in PSA level will be aligned with age rather than the date of diagnosis. Here, given that we are interested in model comparison, we treat Gleason score as a binary variable (score of 6 vs. 7) for simplicity.

Prostate-specific antigen has a positively skewed distribution and, for this reason, a log transformation of the response is the usual initial step in its analysis. 29,35,95,96 Figure 3 displays simulated PSA trajectories for five men on AM, the high variation in PSA level and steady upwards trend being typical of a real data series from men on ProtecT. These data are simulated, because presentation of post-randomisation results from the trial is restricted until primary analysis in 2015.

FIGURE 3.

Plot of PSA trajectories for five simulated men.

Methods for modelling repeated measures of prostate-specific antigen

Four approaches were used to fit the PSA data from ProtecT, namely linear mixed models (LMMs), fractional polynomial mixed models (FPMMs), regression spline mixed models (RSMMs) and principal components analysis through conditional expectation (PACE). These methods are described in detail in Appendix 4.

Prostate-specific antigen reference ranges

A suitable model for PSARRs must satisfy two criteria. First, it must offer an accurate description of the trend of PSA on average and in individuals. Second, it must be able to make accurate predictions about new PSA observations for an individual and about the entire PSA trajectory for a new individual. Once the model is chosen, 95% of observations will be below:

We are only interested in an upper bound for PSA, given that the method would be used to alert men with high observations of PSA.

Conditional predictions for multilevel models

Using methods developed for LMMs by Pan and Goldstein97 and Tilling et al. ,98 we can enhance predictions for future observations by conditioning on already observed PSA. This allows us to make better use of models when moving to a different population of interest or for new individuals from the same population. To predict a future PSA for an individual, conditional on a previously observed PSA, we regress the deviation from the fixed part of the model at the second measurement on the observed deviation from the model at the previous measurement. Mathematically, to condition the prediction of PSA₂ on PSA₁:

Linear mixed model

A linear mixed model was used to estimate the linear effect on log(PSA) of age, Gleason score and their interaction. The fixed parameter estimates are given in Table 3. Those with a higher Gleason score have an almost 30% increased exponential growth rate per year in their PSA (i.e. from 0.073 to 0.106). Given that the model must follow a log-linear trend, the mean LMM fit is lower to begin with in order to fit the data across later ages where PSA is generally higher.

The interpretation of this model is straightforward for use in a clinical setting. Its simplicity also allows for easy transfer to different populations, where more complex models may not fit well as a result of, say, knot placement or polynomial degree being too sample specific. This model could be used to predict PSA trajectory for a newly diagnosed man aged between 50 and 70 years at diagnosis, given his Gleason score. Once an initial PSA is observed, the model can be updated by conditioning on this new observation, thus making a second prediction more accurate.

Fractional polynomial mixed model

The fractional polynomial model of degree 1 (FP1) with the lowest deviance was that with a cubic age term as a covariate, the FP2 selected involved linear age and cubic age. Table 4 shows how the FP2 model was chosen in favour of the FP1 and linear model. We include random coefficients for both polynomial terms to allow for each individual to have their own linear and cubic effect of age on log(PSA).

The interaction between the cubic effect of age and Gleason score is omitted because it was not found to improve the model fit. The inclusion of the cubic age term indicates the inadequacy of the linear model for age. It is important to realise that the cubic term-only model was found to be the best FP1, and the addition of the linear term to it (in the FP2 model) was found to improve the fit through a hypothesis test (i.e. Table 4, Step 3). The interaction between age and Gleason score more than trebles the effect of age on log(PSA) for those in the higher Gleason category (from 0.011 to 0.036). If a Gleason score by cubic age interaction were included, the interpretation of this interaction coefficient would not be useful or straightforward.

The intercept [estimate of log(PSA) at age 50 years] is much larger here than in the LMM. In Figure 4 we see that the FPMMs allow a more versatile range of fits to the simulated PSA profiles compared with LMMs. The cubic term allows for a steeper gradient in PSA for older people. Overall, there is a positive effect of age on log(PSA), which concurs with the LMM.

FIGURE 4.

Fits to the five simulated PSA profiles. (a) LMM; (b) FPMM; (c) RSMM; and (d) PACE.

Regression spline mixed model

Here, K = 2 knots were selected, based on AIC, to gain modelling flexibility while avoiding potential overfitting of these PSA profiles. Given that no evidence of natural ‘jumps’ of PSA occur at any particular age for all men, these two knots were globally placed at the 33.3 and 66.7 centiles of the distribution of age. All interactions between spline terms and Gleason score were added initially but then removed owing to no model improvement.

Table 3 shows that the linear relationship between age and PSA becomes stronger as a man gets older. Indeed, the strength is doubled at 63 years and trebled at 68 years, compared with the trend for a man aged < 63 years. As can be seen in Figure 4, the fits are allowed to break from a linear trend to capture some features missed by the LMM.

Functional principal components analysis

The first two principal components of the PSA data accounted for 92% of the variation in PSA. In Figure 4 the serial PSAs for each man are fitted with smooth curves, which are deviations from the overall mean function μ(t) displayed in Figure 5. These individual curves are not restricted by a parametric shape. The mean curve μ(t) suggests a much less extreme trend than each of the other approaches. PSA is estimated to rise slowly up to the age of 70 years, at which point a sudden change of slope occurs with PSA rising at a faster rate. This fit appears similar to a linear regression spline fit with one knot at 70. The underlying model gives this fit far more flexibility, and, of all methods, it estimates the slowest trend for PSA.

FIGURE 5.

Mean PSA over age (solid line) with fits using LMM (dashed), FPMM (dotted), RSMM (dash dotted) and PACE (long dashed).

Describing the mean pattern of prostate-specific antigen change

Tables 5 and 6 summarise the difference in fitted and observed PSA and the RMSE, respectively, across age categories. In each age category, PACE is superior for describing these data. Apart from in the right tail (at ages > 71 years), the difference between the observed PSA and that fitted by the model is smaller for PACE than for the other methods. If the sole interest here was in representing PSA trend for this group of men on AM, then PACE would be the best choice. Among the other methods, from ages 50–60 years, the LMM performs quite badly. As can be seen in Table 5, the fitted PSA values are obviously not high enough on average. Over half of the total observations of PSA were taken from men aged 61–70 years, and this range sees comparable performance of the parametric methods. Figure 5 serves to illustrate the benefit of using RSMM instead of FPMM in this example. Up to the age of roughly 70 years, both have a similar mean estimate. The FPMM then continues on this path because it uses a global polynomial basis for model fitting, which leads to severe overestimation of mean PSA in the right tail of age. However, the regression spline fit has the flexibility to change at the second knot (i.e. at age 68 years) and this change then becomes apparent with a flatter fit of the regression spline in the right tail. Outlying PSA measurements in older age affect only the spline-based estimate of the age effect in that region, whereas with FPMM they have a global effect on the estimate.

From Table 7, the FPMM and RSMM have lower AIC compared with the LMM, but only marginally more variation is explained by the two more complex models. The RMSE is lowest using PACE followed by the RSMM approach, which corroborates the results from Table 6. Based on these diagnostic elements, the RSMM is the best fitting of the three parametric models for PSA in men on AM.

Table 3 compares the estimates of the associations between Gleason score and PSA trajectory from the three parametric models. There are some similarities in the effect of Gleason score and its interaction with age; however, the intercept is much lower in the LMM than in either of the other two methods (which have very similar estimates). In the LMM, with a constant trend over time, the intercept must be low to allow for the model to fit the higher PSA values in the right tail of age. We can see in Table 5 that this leads to an underestimate of PSA when age is between 50 and 60 years. For the same reason (i.e. to allow those with higher Gleason score to have higher PSA in later age) the LMM underestimates the association between Gleason score and PSA at age 50 years and overestimates the association between Gleason score and PSA slope, compared with the FPMM and RSMM. The RSMM or FPMM methods are preferred here for estimating the effect of Gleason score on PSA change over time.

Predicting prostate-specific antigen for a future patient

Here a parametric model is preferred, given that, by using methods that condition on new observations of the response,98,100 we can allow the parametric models to become individualised to newly diagnosed men on monitoring. Without a ‘factor loading’ ξi1, a new individual i would simply be predicted to lie on the mean PACE trajectory given his age.

Table 7 presents the RMSE of fitted PSA for each of the 512 men conditioned on their initial PSA. Here we can see that the conditioning improves on the parametric methods such that they are better at fitting these data than PACE. Between the three approaches, the RSMM is still preferred. Surprisingly, the LMM performs slightly better than the FPMM when fitted values are conditioned on the first observation of PSA.

Regression splines offer the optimum performance among parametric methods overall (with the lowest AIC and RMSE for both unconditional and conditional predictions) and also perform well in all segments of age, as shown in Table 5 and Table 6. Thus, using regression splines for prediction of PSA in men in the future seems appropriate. This reflects the changing trajectory of PSA over age, given that the RSMM allows for three separate effects of age on PSA.

Study populations

The RMH AS cohort is an ongoing study into the impact of initial conservative management of clinically localised PCa. 101 Data from 499 men, comprising 9472 PSA tests along with Gleason score and several other clinical covariates, were available. PSA test results were obtained between 1999 and 2012. The study eligibility criteria were baseline PSA < 15 ng/ml, Gleason score of ≤ 3 + 4 and percentage of positive biopsy cores ≤ 50%. In most men, diagnosis was based on a raised PSA and subsequent positive biopsy, so they represent a modern AS cohort. Men on AS were followed up with PSA tests every 3–4 months in the first 2 years and every 6 months thereafter. Biochemical progression was defined as PSA velocity > 1 ng/ml/year, whereas histological progression on rebiopsy was defined as having a Gleason score of 4 + 3 or > 50% positive biopsy cores. Clinical outcomes of metastases and PCSM were collected.

The Johns Hopkins (JH) AS programme began recruitment of men with clinically localised PCa in 1995. 82 Men were eligible if they had a Gleason score of ≤ 3 + 3, T1c, PSAD < 0.15 ng/ml/cm³, two or fewer positive biopsy cores and maximum involvement of 50% per core. Data from 961 men, comprising 9993 PSA test results performed between 1993 and 2012, along with diagnostic Gleason score and several other clinical covariates, were available. Radical treatment was recommended once men no longer met the eligibility criteria described above. Clinical outcomes of all-cause mortality and PCSM were included in the data.

The SPCG4 data contained 290 men with 2987 PSA tests. 12 These men were randomised to WW as part of the SPCG4 RCT comparing WW and radical prostatectomy. They were diagnosed between 1989 and 1999, for the most part through clinical presentation with symptoms. To be suitable for randomisation the men were required to be < 75 years of age, with a life expectancy of > 10 years. Data were available on whether or not the men developed metastases and/or died from PCa.

Data received from the University of Connecticut Health Center (UCHC) cohort consist of 114 men with 884 PSA test results102 followed for an average of 4.7 years (SD 3.9 years). Men were diagnosed between 1989 and 1993 (before the advent of widespread PSA testing), for the most part by clinical presentation with symptoms. Fewer baseline clinical exposures were available than from other cohorts, and information was not present on whether or not men developed metastases or died from PCa.

These four cohorts come from two eras, the PSA detection era (RMH and JH) and the clinical detection era (UCHC and SPCG4). As described above, the advent of PSA screening resulted in many more men being diagnosed with PCa at an earlier stage than before. The modern cohorts also come from populations with different screening prevalence. In the USA there is widespread PSA screening, whereby men are likely to have several PSA tests through their lifetime. Thus, men diagnosed with PCa in the USA are likely to have had mostly ‘normal’ PSA results before the raised value, which resulted in a biopsy and subsequent diagnosis. In the UK there is no such screening programme, and the men diagnosed with PCa in RMH may have had any level of PSA before the single high PSA that led to their diagnosis. Hence, the US men are likely to be a lower-risk group with lower PSA on average than their UK counterparts. These issues, and their impact on the poorly understood natural history of PCa, need to be considered when interpreting findings from the cohorts.

The coefficients found in the ProtecT trial model were applied to data from the RMH and JH AS cohorts as well as the UCHC and SPCG4 WW cohorts. However, all data were restricted to PSA test results ≤ 50 ng/ml and to men with an initial PSA ≤ 20 ng/ml. This is to eliminate any atypical values, in terms of a modern AM/AS study. Two cohorts of men without PCa were also included to examine differences in PSA change between men with and without cancer. Data from the Baltimore Longitudinal Study of Aging (BLSA)103 contained repeated 5012 PSA measurements of 1032 men. A model for PSA change in 1432 men without cancer from the Krimpen study,104 a large prospective community-based study in the Netherlands, has appeared elsewhere. 35

Measuring performance

First, to investigate the behaviour of PSA between cohorts, we fit a simple random intercept and slope multilevel model to log(PSA) in each of these four cohorts:

where u0i is the random intercept, which allows each man to have their own adjustment to the intercept β₀ [i.e. the average value of log(PSA) at age 50 years] and u1i is the random slope, which allows each man to have their own adjustment to the slope β₁ (i.e. the average change in log(PSA) per year increase in age). This is carried out to compare the age-related change in PSA for men with and without PCa.

The coefficients for our model of PSA change have been previously estimated using a cohort of 512 men with clinically localised PCa participating in the ProtecT study. 93 In the present analysis, the model is used to predict PSA in each of four external cohorts. We measure the accuracy of the predictions using the average difference between observed and predicted PSA value per PSA test:

where PSA_ij is the observed PSA test result for person i = 1, . . . , n measured at time j = 1, . . . , n_i and PSAiJ^ is the predicted PSA test result for person i at time j. The number of PSA tests can be different from person to person, and the total number of PSA tests in the cohort is N. The average absolute difference between observed and predicted PSA will always be positive, with a value of zero if the model predicted PSA perfectly.

For each of the four cohorts, predictions are made using (a) a model with coefficients derived from the external data themselves; (b) the ProtecT model coefficients; and (c) the ProtecT model updated using the first three PSA values for each man in the external cohort. 97,98 Prediction (a) gives the hypothetical upper limit of performance but is not clinically useful, because the model coefficients cannot be estimated until all the PSA measurements have been taken over the duration of AM. Predictions (b) and (c) indicate what could be achieved in clinical practice, as they apply coefficients estimated using the ProtecT cohort to the other data sets, and so can be applied each time a new measure becomes available for a man on monitoring.

To calculate the coverage of the model in predicting PSA, we check whether or not a 95% prediction interval (calculated using unconditional standard errors) from the ProtecT model contains the corresponding observed value of PSA. We measure performance of the models further by tabulating the model failures, which we define as absolute difference between predicted and observed PSA > 5 ng/ml and model successes, defined as predicted PSA within 2 ng/ml of observed PSA. These are tabulated to obtain the proportion of test level failures/successes (i.e. for how many PSA test results does the model fail/succeed) and subject level failures/successes (i.e. for how many men does the model fail/succeed on average across all their PSA test results). These cut-offs were chosen to reflect what we believe to be clinically significant ranges.

Age-related prostate-specific antigen change

Men on AM/AS and men without PCa have similar age-related PSA change (Table 8). For example, the PSA change per year is very similar in the Krimpen, BLSA, RMH and JH cohorts. Figure 6 shows the predicted average pattern of change if each cohort had an average PSA of 2 ng/ml at age 50 years. This hypothetical graph shows the similarities of the four modern cohorts involving men with or without PCa. However, the results from the multilevel models suggest that men without cancer have much lower average PSA values at age 50 years. In Figure 7 we see that the estimated average PSA at age 50 years is much lower in the Krimpen and BLSA cohorts.

FIGURE 6.

Hypothetical PSA change in the cohorts if average PSA at 50 years was 2 ng/ml in each.

FIGURE 7.

Estimated PSA change in the cohorts.

Screen-detected cohorts

The model updated with the initial three PSA results is, on average, 2 ng/ml away from the true PSA for RMH and 1.8 ng/ml away from the true PSA for JH (Table 9). Models fitted using all the PSA data from these AS cohorts lead to an average difference between observed and predicted PSA of 1.1 ng/ml for RMH and 0.82 ng/ml for JH. This indicates that the best prediction that can be achieved after collecting all PSA data from this cohort. Furthermore, the coverage of the model (i.e. whether the observed value lies within the 95% prediction interval) is close to the nominal value of 95%, at 98% for RMH and 97% for JH. Updating the model using the initial three PSA values has improved the accuracy of the ProtecT model. In both RMH and JH the difference between the observed and predicted PSA values reduces by 1.3 ng/ml per PSA test when using the updated model instead of the standard one.

Clinically detected cohorts

The ProtecT model achieves less accurate predictions in the older cohorts of SPCG4 and UCHC (see Table 9). Although an optimal model fitted using all the SPCG4 data gives values of PSA within an average of 1.7 ng/ml from the observed PSA, the ProtecT predictions differ by 4.6 ng/ml on average. Similarly, in the UCHC cohort, the ProtecT model leads to predictions that differ by 3.7 ng/ml from the observed PSA on average, whereas, for an optimal model fitted using all the UCHC data, the average difference between observed and predicted PSA is 1.3 ng/ml. However, coverage was 95% for SPCG4 and 96% for UCHC. The results from both these cohorts further demonstrate the improvements gained using the model updated with the initial three PSAs, with the average difference per test reducing by 2.8 ng/ml in SPCG4 and 1.9 ng/ml in UCHC.

Comparison across cohorts

A clear dichotomy between the cohorts is found in Table 10. Roughly 3–4% of men have an average absolute difference between predicted and observed PSA of > 5 ng/ml using the external model in the modern AS cohorts. However, this rises to 14% and 30% for the UCHC and SPCG4 cohorts, respectively. A similar divide is apparent in the percentage of PSA tests where the model fails. This is 7–8% for the modern AS cohorts but 20% and 25% in the UCHC and SPCG4 cohorts, respectively. The reverse of these results is also true, in that the amount of model successes shows up a dichotomy between the cohorts (Table 11). Between 67% and 79% of men in the modern AS cohorts have an average difference between predicted and observed PSA of < 2 ng/ml across all their PSA test results. The two older cohorts have a lower proportion of men, between 39% and 51%, whose average absolute difference between predicted and observed PSA is within 2 ng/ml. The percentage of individual predicted PSAs within 2 ng/ml of the actual test results is 70–73% in the modern cohorts but 55% and 47% in UCHC and SPCG4 respectively.

Prostate-specific antigen markers

Prostate-specific antigen doubling time is the most popular PSA kinetic measure used to identify those patients on AM who need clinical review. PSADT is calculated from a model of log(PSA) against time:

The doubling time is then calculated as log(2)/β and is interpreted as the estimated time for a man’s PSA to double. Thus, low values of doubling time coincide with rapidly rising PSA and are a cause for concern. Conversely, high values correspond to a steady trend of PSA, and those patients above a threshold of PSADT are advised to remain on monitoring. For a patient’s doubling time to be calculated using the regression method, at least three PSA test results are required. We consider a PSADT of < 3 years to be a marker for rapidly rising PSA. 79,84,86,89

An alternative to PSADT is PSAv. This explicitly measures the rate of change in PSA over time calculated as the slope term (β) from a model of PSA against time on monitoring:

Given that PSAv is the rate of increase of PSA, low positive values refer to steady PSA change and high PSAv equates to sharply increasing PSA. Several studies have used a threshold of PSAv > 1 ng/ml/year22,71,100 as a trigger for clinical review during monitoring. For this paper, we have calculated each man’s PSAv at each visit beyond their third (calculation of PSAv requires at least three observations of PSA). We have used PSAv > 1 ng/ml/year as a marker for rapidly rising PSA.

Prostate-specific antigen reference ranges have been established as alternatives to these two PSA kinetic measures. 33 A model for such reference ranges has been previously developed93 on a cohort of 512 men enrolled in the AM arm of the ProtecT trial. 15 For example, a 95% reference range, above which the fastest increasing 5% of men with clinically localised PCa lie, is constructed as:

where PSA^ and var(PSA^) are found using the ProtecT trial model described in Chapter 3. If subsequent PSA test results fall above this reference range, this implies that the patient’s PSA growth is above what would be expected as a result of normal ageing, such that a recommendation for clinical review could be warranted (e.g. Figure 8). Using the reference range model, we calculate whether or not a man’s PSA result is inside the reference range at each visit; if it falls outside the reference range, this suggests an increasing PSA beyond expected normal changes and that further clinical review may be necessary.

FIGURE 8.

An example of a PSARR. The PSA result represented by a triangle would lead to a recommendation for clinical review.

Study populations

The PSA markers were tested on the RMH AS cohort and the SPCG4 and UCHC WW cohorts described in Chapter 4, Methods, Study populations (see also Table 24). The data were again restricted to remove any single PSA test result > 50 ng/ml and any man with a PSA at diagnosis of > 20 ng/ml. Hence, the sample sizes differ (Table 12 and see Table 25).

Outcomes

All-cause mortality data from UCHC are not used in this analysis to validate reference ranges, because they probably contain many non-PCa-related deaths. Although the data from RMH include PCSM, no such deaths are reported. To exploit these cohorts fully, we introduce outcomes pertaining to a PSA milestone, namely whether or not a man reaches a PSA of 20 ng/ml during follow-up. The outcomes used for validation, along with the proportion of men experiencing these outcomes, are given in Table 12. Given that it is not clinically useful to have an alert (through PSADT, PSAv or PSARRs) close to, say, PCSM, we have only included those alerts occurring at least 5 years prior to PCSM or metastases and 2 years prior to reaching a PSA of 20 ng/ml.

Statistical analysis

Statistical analyses were performed in Stata 12. 70 We report results solely for men with Gleason score of 3 + 3, because this low-risk group represents the vast majority of modern AM cohorts. 17 We first compared 90%, 95% and 99% PSARRs to choose the optimal cut-off for rapidly rising PSA using the area under the receiver operator characteristic (ROC) curve statistic for onset of metastases and PCSM in SPCG4. Cox proportional hazards (Cox PH) models105 were used to estimate the association between markers and time-to-event outcomes. Time zero was set to be the date of beginning of monitoring, whereas date of an event or end of monitoring was the time at risk. For each cohort we fit a base model controlling for T-stage (where applicable), baseline PSA and baseline age. We report the HRs for each PSA marker and outcome, together with their 95% CIs. Finally, we compare PSADT, PSAv and PSARR using the difference of deviances from the Cox models using each of these in turn. A better model will have a lower deviance (see Table 16).

The ability of each PSA marker to discriminate between higher- and lower-risk men is calculated using Harrell’s concordance statistic (or c-statistic) from each Cox PH model. 105,106 This is equivalent to an area under the ROC curve for time-to-event responses, such as those of interest here. The c-statistic estimates the probability that, of two randomly chosen men, the man with the higher-predicted probability of an outcome will reach an outcome before the patient with the lower-risk score. By adding each PSA marker in turn to the baseline models, we can compare the change caused by the marker, by looking at the difference in c-statistic from base model to marker model.

Metastases

Using an alert based on PSARR alone to recommend clinical review would miss 15 out of 30 future metastatic cancers (sensitivity 50%, 95% CI 31% to 69%) while causing 38 unnecessary rebiopsies to be performed in the 107 men with non-life-threatening disease (specificity 65%,95% CI 55% to 74%). The PSADT < 3-year marker would lead to 13 missed metastatic cancers out of 30 (sensitivity 57%, 95% CI 37% to 75%; Table 13) and would unnecessarily biopsy 42 of the 107 men with disease that would not progress to metastases during the follow-up of the cohort (specificity 61%, 95% CI 51% to 70%). Finally, the PSAv > 1 ng/ml/year marker alone would miss just eight of the 30 future metastatic cancers (sensitivity 73%, 95% CI 54% to 88%) but would cause 70 unnecessary biopsies (specificity 35%, 95% CI 26% to 45%).

Initial age, PSA and T-stage were not useful to identify those men developing metastases (c-statistic = 0.58, see Table 15). There was weak evidence of positive associations between the PSADT marker (HR 2.1, 95% CI 0.97 to 4.4) and PSARR (HR 2.4, 95% CI 0.9 to 6.7) and development of metastases (see Table 15), and no evidence that men alerted by the PSAv > 1 ng/ml/year marker were more likely to develop metastases in the future (HR 1.1, 95% CI 0.5 to 2.7; see Table 15). However, the PSA markers slightly improved the discriminative ability of the baseline model, with increases in c-statistic of 9% (PSADT), 7% (PSARRs) and 2% (PSAv). Comparing the PSA marker models, there was some weak evidence of a better predictive ability using doubling time over velocity (p = 0.06; see Table 16).

Prostate cancer-specific mortality

The PSARR marker led to 56% sensitivity (95% CI 31% to 79%; Table 14) and 61% specificity (95% CI 52% to 70%) when used alone to determine which men would die from PCa. This would mean that 8 out of 18 men with lethal cancer would be missed by the PSARR, and 48 unnecessary biopsies would be performed. Using the PSADT marker alone would miss 6 out of 18 lethal cancers (sensitivity 67%, 95% CI 41% to 87%) and cause 52 unnecessary rebiopsies to be performed in the 124 men without lethal cancer (specificity 58%, 95% CI 49% to 67%). If the PSAv marker were used as a standalone recommendation for clinical review, just three of the 18 lethal cancers would be missed (sensitivity 83%, 95% CI 59% to 96%), but 83 unnecessary biopsies would be performed in the 124 men with non-life-threatening disease (specificity 33%, 95% CI 25 to 42%).

A model using baseline characteristics of PSA, T-stage and age was not useful to discriminate between those who died as a result of PCa and those who did not (c-statistic = 0.53; Table 15). Those men with a PSADT < 3 years were more likely to die from PCa (HR 3.2, 95% CI 1.1 to 9.2) and this improved the c-statistic from the baseline model by 0.13 (25%). There was some weak evidence for an association between PSARR (HR 3.3, 95% CI 0.9 to 12.1) and little evidence for an association between PSAv (HR 2.1, 95% CI 0.6 to 7.7) and death from PCa. There was some weak evidence that the PSADT model was a better predictor of outcome than the PSAv (p = 0.05; Table 16) model.

Reaching a prostate-specific antigen of 20 ng/ml during surveillance

In the RMH cohort, baseline age, PSA and T-stage had a moderate discriminative ability in separating men who would reach a PSA of 20 ng/ml (c-statistic = 0.74; see Table 15). There was no evidence that those men with a PSA outside their reference ranges were more likely to reach a PSA of 20 ng/ml (HR 1.5, 95% CI 0.5 to 4.1), and including PSARR only trivially improved the c-statistic. There was some evidence that men with a PSADT < 3 years were more likely to reach a PSA of 20 ng/ml (HR 2.2, 95% CI 1.0 to 4.6); including PSADT improved the c-statistic by 0.03 (5%). Those who would have been alerted by a PSAv > 1 ng/ml/year were more likely to reach a PSA of 20 ng/ml (HR 10.8, 95% CI 1.5 to 80.8). The PSAv marker led to an increase of 0.05 (7%) in the c-statistic from the baseline model. Comparing these models, the PSAv marker model was preferred to those including PSARR (p = 0.001; see Table 16) and PSADT (p = 0.006), with some weak evidence that the PSADT model improved on the PSARR model (p = 0.06).

For the UCHC data, baseline age and PSA were highly discriminative of reaching PSA 20 ng/ml in at least 2 years, with a c-statistic for the baseline model of 0.82 (see Table 15). Men with a PSA result over their PSARR were more likely to reach 20 ng/ml (HR 12, 95% CI 5.4 to 26.9; see Table 15). Given that there are very few men in the Gleason score of 3 + 3 category who reach a PSA of 20 ng/ml, the CI is very imprecise. Including PSARR led to an increase in c-statistic of 0.03 (4%) over the baseline model. Associations were found between both PSADT (HR 6.0, 95% CI 2.8 to 12.8) and PSAv (HR 4.3, 95% CI 2.5 to 7.5) and reaching a PSA of 20 ng/ml in the UCHC cohort. The inclusion of each of these markers led to increases of 0.02 over the baseline model, or 3%. There was strong evidence that the PSARR model had a higher predictive ability than both the PSAv and PSADT models (both p < 0.0001; see Table 16).

Design

Semi-structured, face-to-face interviews were conducted with clinicians, nurses and patients from four hospitals, each in a separate acute trust. Acute trusts (study sites) were purposively selected on the basis of having an established, fully running AS programme. One of the four sites was involved in the ProtecT study, but clinicians interviewed at this site had no direct involvement in the study.

Sampling and recruitment of clinicians/nurses

Clinicians were purposefully selected based on their roles and non-involvement in the ProtecT study. Clinicians who led routine AS appointments were prioritised, but staff with more peripheral involvement in AS patient care were also interviewed (e.g. nurse specialists).

Principal investigators at each site suggested appropriate potential participants and helped to organise introductory meetings in two of the four sites. These meetings allowed the research team to gain preliminary insights into the structure and running of the clinics and to recruit clinicians. A degree of snowball sampling, where interview participants suggested potentially eligible colleagues, occurred at three of the four sites. Clinicians/nurses were invited to take part in the study through a letter from the research team and a study information sheet. These information sheets were either sent to the hospitals or distributed at the introductory meetings. There was potential to continue recruitment across some of the sites, but sampling ceased once data saturation had been achieved; this was defined as the point at which four consecutive interviews generated no new themes.

Sampling and recruitment of men on active surveillance

Recruitment of men on AS took place across three of the four sites. The site involved in the ProtecT trial was excluded from this phase of the study owing to concerns that men’s care could have been influenced by the ProtecT study protocol for AM if patients had received care from clinicians involved in the ProtecT study. Clinical teams from each site identified eligible men in line with inclusion and exclusion criteria stated in Box 1. Recruitment packs were distributed to eligible men via post or within clinics. Each pack contained a letter from the researcher (LR), a reply slip, a study information sheet and a pre-paid return envelope. Men interested in participating in the study were asked to contact the researcher (LR) directly.

Data-collection methods

Data collection took place from April to November 2013 in private rooms within hospital trust premises or men’s homes. Each participant was interviewed once, and LR conducted all bar one interview (which was conducted by JW). Men’s partners were present in four interviews (see Table 17 for details). Where present, partners were considered as participants and subjected to the same consent processes as the men. Interviews were recorded following receipt of written consent from all present and transcribed in full using standard notation. Interview times ranged from 20 minutes to 1 hour and 5 minutes for clinicians, and 18 minutes to 1 hour and 36 minutes for men. Interview schedules were used to ensure topics were consistently covered (see Appendix 3). These documents evolved as data collection proceeded, with inclusion of new topics and omission of prompts that were known to be unsuitable/irrelevant based on initial interviews conducted. Clinician/nurse interview schedules explored the criteria informing decisions to offer AS, current AS protocols and perspectives on the AMS. Interview schedules for men on AS covered pathways leading to joining an AS programme, experiences of being monitored and perspectives on the AMS. Brief field notes were taken by the researcher during and after interviews, which recorded pertinent details to pursue within interviews and initial analytical thoughts (if time permitted).

Vignettes displaying three hypothetical men’s PSA values over time were used throughout the interviews with clinicians/nurses and men on AS. Each hypothetical man’s data were presented in two formats: as a table on paper showing PSA values over time (‘paper vignettes’), and as they would appear as a graph within the AMS, presented on a laptop using Microsoft Excel^® 2013 software [(Microsoft Corporation, Redmond, WA, USA); Appendix 3, ‘Active monitoring system vignettes’]. Each of the three hypothetical vignettes presented distinct PSA trends: vignette 1, where the values were steadily increasing (‘Mr Smith’); vignette 2, where the PSA values were erratic (‘Mr Jones’); and vignette 3, where there was a gradual rise in PSA with a steeper gradient for the last few readings (‘Mr Evans’). This final vignette was the only scenario in which a man’s PSA eventually exceeded a value of 10 ug/ml.

Presentation of the AMS using Excel allowed for addition and removal of various reference lines, each of which were individually superimposed on PSA graphs by the researcher (with a brief explanation of what each reference line represented). Addition and removal of these reference lines allowed the researcher to explore different presentations of PSA values within the AMS.

A core component of the qualitative work was to investigate current AS provision from the perspectives of men and clinicians; this was a broad aim, with scope for exploration of new lines of enquiry. Given this, data saturation was assessed primarily through consideration of the principal aim of the qualitative work in the context of the wider project: to evaluate the AMS tool.

Analysis

Audio-recordings of interviews were transcribed in full using standard notation. Transcripts were analysed thematically using the constant comparison method derived from grounded theory methodology. 107 This involved line-by-line coding of transcripts, categorising codes into themes and developing codes and themes as transcripts were reread in light of newly collected data. Analysis was primarily conducted by LR using NVIVO (V.9) (QSR International, Warrington, UK). A sample (10%) of transcripts from clinician and patient interviews were independently analysed by JW midway through data collection. Any differences in coding and thematic interpretations were discussed and additional areas were suggested for addition to the topic guides. Descriptive accounts of observation and interview findings were written and discussed with the research team. Matrices for major themes from interviews were drawn up. Participants were arranged according to centre and – in the case of clinical professionals – according to role to enable patterns of meaning within each group of participants’ accounts to emerge, facilitating identification of ‘negative’ cases that conflicted with emerging theories. These exceptions were revisited by rereading transcripts and were described accordingly in reported findings. Final matrices of the main themes described in this report were constructed prior to writing to ensure that the full range of perspectives was reported.

Participant characteristics

A total of 38 interviews were conducted (13 consultant urologists, two specialist registrars in urology, three urology nurse specialists, 20 men). All clinicians who were invited to take part in the research agreed, and three of the 23 invited men declined (without giving a reason). None of the participants withdrew from the research. Characteristics of men who participated can be found in Table 17. The sample of men interviewed was spread across the age range of 55–70 years, with most aged 65 years and above (as is typical in routine practice).

Presentation of data

Quotations from interviews have been selected on the basis of how clearly and succinctly they illustrate the dominant themes to emerge from the research. Tensions and inconsistencies have been discussed, and quotations from divergent cases are presented where relevant.

Clinicians are referred to as ‘he’ throughout, for purposes of anonymity.

The following identifiers have been used for quotations:

U(number) = consultant urologist or specialist registrar in urology; SPN(number) = specialist urology nurse; P(number) = man on AS; P(number)partner = partner of man on AS; INT = the interviewer (LR/JW)

Overview

The chapter covers two broad areas explored within this research:

1. deciding whether or not to follow an AS programme

2. current practice within AS programmes: follow-up procedures and triggers for change of management.

For clarity, data from interviews with clinicians/nurses and men on AS have been presented separately within each section. The term ‘AS’ rather than ‘AM’ was used by all sites, although clinicians did not make any distinctions between the two terms. They did however distinguish ‘WW’ from ‘AS’, with the main difference being that WW did not have a curative intent. A more detailed description of how clinicians distinguished AS from WW can be seen in Appendix 3.

Deciding whether or not to follow an active surveillance programme

Current practice within active surveillance programmes: follow-up procedures and triggers for radical treatment

This section will consider AS protocols from the perspective of clinicians and men. It should be noted that sites varied in how protocol-driven their practices were. At one site (site 1), clinicians had been following a site-specific protocol for a number of years, whereas clinicians from another site (site 2) were in the process of developing a protocol at the time of data collection. As such, there was more variation across clinicians’ accounts from site 2, although it became apparent that individual variation between clinicians existed irrespective of the presence of a protocol.

Clinicians’ views on data presented in the active monitoring system

Men’s perspectives on data presented in the active monitoring system

Clinicians’ and men’s views on presentation of data within the active monitoring system

Some clinicians expressed concern about the potential complexity of information presented in the AMS and the potential difficulties in trying to explain this within time-constrained consultations:

And you also have to be careful with patients because patients often don’t understand the nuances of this, they don’t understand an average is an average so you’ve got to be very careful how you explain these sorts of figures to patients, it can actually confuse them.

U2

In contrast, one clinician felt that the potential to add or remove reference lines would enable him to pitch the AMS at the right level, depending on his interpretation of patients’ comprehension of graph/statistics:

To the clinicians, brilliant, so I think the ability to have, as you just did, add them and take them away, that would be a good thing. Because you might be just showing his line first and then showing the average man with cancer as an add on, a build on – that might be the way to do it.

U1

Men who provided comments on the potential benefits of the AMS offered little input into how the presentation could be improved. A few insights emerged from the researcher’s observations of how men interacted with the AMS. For instance, two men found it difficult to see the green line, and one man viewed the orange line as red, a colour that can have ‘danger-’ or ‘risk-’related connotations. If the AMS were to be developed in the future, these issues will need careful consideration.

Although eliciting views on presentation was generally challenging, a number of men were able to suggest additional information that they would have liked to have been displayed. These included CIs around the ‘Predicted PSA for a man with PCa’ (green) line, and the option to add in a line of best fit for patients’ actual PSA values:

And if I, again, not very good on graphs, but if I were a statistician again or a graphologist [sic], whatever he is, I would want to take mine – that black line – entirely separately and I’d want to work out the average and reconfigure that, on that grid. My average. ’Cos this is my exact.

Oh I see, yes. Oh, OK.

Now because there’s a dip there, and these are two serious sort of dips. My mean line, when I work that out, may go somewhere like this [demonstrates].

So you would want to do your – kind of – line of best fit?

Yeah. I’d like to take my mean average [sic].

A handful of men expressed an interest in seeing reference lines that specifically indicated when they needed to take action:

I would have thought [. . .] rather than just the red line being sort of, y’know, saying that that’s the highest of people without prostate cancer, maybe the red line could be, you know, a red line, if you cross over this red line, our experience would show that, you know, there is a [. . .] serious problem.

Mm. There is another point, and that is, what would . . . without all this, at what the consultant says you’d better have an operation now. I wondered if you could put a line from past data, at the point when a decision was made. So I don’t know how it would work exactly [. . .]. So basically he has the treatment when he’s that much above deviation for that period. I don’t know how you’d express that on the graph, but it. . . it [pause].

So almost, pooling together of the tipping points. . . .

The tipping point. Exactly.

Recruitment and design

The ProtecT trial successfully gained consent from over 1600 men to be randomised to AM, radical prostatectomy or radiotherapy. However, as the failure of the START (Phase III Study of Active Surveillance Therapy against Radical Treatment in Patients Diagnosed with Favourable-Risk Prostate Cancer) trial108 demonstrates, recruiting enough men who are willing to be randomised to such different approaches remains an extremely difficult task. Lessons learned from the ProtecT trial would need to be implemented. 109 Recruitment of men to a trial of AS/AM would need to take place following PSA testing in a healthy population, or following ad hoc testing in clinical practice. A key issue in recruiting to a PSA-tested population is the huge numbers required – over 100,000 tested in the ProtecT trial, for example. 14,15 The alternative is to recruit from the pool of all new cases of clinically localised PCa who according to the latest NICE guidance,68 should be offered this option if they have low-risk disease or higher-risk disease but are unwilling to have radical treatment. Men would need to accept an initial conservative approach to their newly diagnosed cancer and to give consent to be randomised to one of several AM/AS strategies.

A decision about whether or not the study would involve randomisation at the cluster (clinic) or individual level would need to be made based on the AM/AS strategies under investigation. For example, a trial comparing monitoring with the AMS and an alternative monitoring strategy (e.g. current practice) would need to be cluster randomised, because there would likely be contamination between interventions if members of the same clinic were randomised to different monitoring strategies. Blinding of clinicians and patients would not be possible in a trial of different monitoring strategies, given the obvious distinctions between potential interventions, but blinding of outcome assessment would be considered carefully. Pre-trial and ongoing training of doctors or nurses responsible for AM/AS check-ups would be vital in any such trial. The interpretation of PSADT, PSAv and the PSARR graphs would need to be clear and consistent.

Eligibility

An issue for any trial will be to decide on the eligibility criteria for enrolment. The eligibility criteria for the ProtecT trial mean that it will be difficult to reach conclusions in the higher-risk groups (i.e. men with Gleason scores of 3 + 4 or 4 + 3). However, as worldwide AS studies accrue men and time in follow-up, the safety of allowing such men to enrol in AM/AS may be established.

Several studies have published eligibility criteria along with validation, the most popular of these being the D’Amico criteria,110,111 where low risk is: PSA < 10 ng/ml, Gleason score at most 3 + 3 and stage up to T2a; intermediate risk is PSA between 10 ng/ml and 20 ng/ml, Gleason score up to 4 + 3 and stage T2b; high risk is PSA > 20 ng/ml, Gleason score of at least 4 + 4 and stage T2c or T3a. Recently, there has been a proposal to use a single early PSA test to stratify men as low, intermediate or high risk. 112 If the ProtecT trial demonstrates that AM is an effective option for low-risk cancer, then a trial might be needed to establish whether or not it is also effective for those with intermediate-risk localised cancer.

Recent developments in MRI113 and potential genetic variants associated with lethal PCa114 may lead to better stratification of those men with PCa who will not experience progression to life-threatening disease. It might be that, in future, more precise stratification might be employed to determine eligibility to enter a trial comparing different management strategies.

Outcomes

The results of this project, including the systematic review, suggest that PCSM is rare in the short term for patients on AM/AS. However, PCSM or metastases are the most important primary outcomes by which to compare any monitoring strategies and should form the primary clinical outcome for future trials. Either a large sample size or long-term follow-up (or both) would be required to produce enough events to allow for any risk-group comparisons. However, if the ProtecT trial demonstrates that AM is effective, then other primary outcomes could be considered when comparing AM/AS strategies, such as the number of men remaining on AM/AS over time or to a specific age, adherence to monitoring protocols, costs and ease of implementation in the NHS, patient satisfaction and quality of life. The chosen primary outcome would have a crucial impact on the design issues of sample size and length of study. Repeated biopsy results are another potential clinical outcome that could be considered, but the sampling procedures (e.g. number of cores used) introduce a large amount of measurement error with either under- or oversampling of the prostate. For example, in the PRIAS [(Prostate Cancer Research International: Active Surveillance) see www.prias-project.org] study 2494 men were diagnosed with PCa and followed AS; of the 1858 repeat biopsies taken in the study, 687 found no cancer. 89

Active monitoring system

A trial comparing the use of the AMS with use of current PSA kinetic thresholds and methods might be of interest to test the acceptability of the AMS, but the issues around what form of AS or AM would be included remain. Our qualitative research has revealed that a minority of patients and clinicians express concerns about the complexity of using a system or graph for PSA monitoring, although most could see some potential for reassurance through using the AMS. However, consistent interpretation of the PSARRs presents a potential problem for nurses, doctors and patients. A short training session for nurses and/or doctors would be needed before commencing any trial to ensure accurate delivery of the system and to minimise interobserver bias in interpretation and informing men. It would be helpful to have objective criteria that are strictly adhered to, such that once a single PSA value falls above the 95% reference range, the nurse/doctor recommends re-evaluation of the cancer, although whether or not such clear criteria will ever be discovered remains uncertain. Furthermore, in the interest of time, nurses/doctors would need adequate training to explain clearly and concisely what the graph is telling the patient, such that consultation time is not wasted. Interviews with patients have made clear that these graphs should be shown to the men in a clinical setting and not sent out to men to interpret without guidance.

Qualitative research has suggested that PSADT and PSAv are sometimes calculated crudely and in a variety of ways in a clinical setting, leading to variable results. There are also many different methods of calculation, as have been discussed previously29,31 and, thus, a computer is needed at every visit to compute PSA kinetics consistently. It would be necessary to identify the optimum method and to ensure that it is applied consistently by suitably training those doctors/nurses in this arm of the trial.

This trial would require a cluster-randomised design, with clinics or GP surgeries as the unit of randomisation. The primary outcome could be the number of men remaining on monitoring at a median follow-up of, for example, 5 years, with a potential further follow-up depending on the rate of management change in the trial. Adherence to the monitoring system, quality of life and satisfaction with monitoring could be secondary outcomes. Reasons for leaving AM would be recorded but not limited to a single reason, given that several distinct factors could lead to a decision to seek radical intervention.

Active monitoring compared with active surveillance

Although ongoing studies with AM and AS may produce useful comparisons, they will not be able to establish as strong a conclusion as a RCT comparing scheduled repeat biopsy with using biopsy only when radical treatment is being considered. A trial comparing AM with AS (i.e. men randomised to have a rebiopsy only when triggered or at scheduled intervals) would determine whether or not this extra procedure needs to be regularly performed to increase the effectiveness of monitoring. Chapter 2 found several frequencies of rebiopsy even among the 16 AS studies, ranging from annual biopsies to biopsy at 1, 4 or 7 years.

The patients would need to be selected from participating clinics and randomised to AM or AS. A cluster randomised design may be necessary in this trial to avoid anxiety due to perceived reduced testing in the AM arm and possible anxiety due to overtesting in the AS arm, as well as contamination with men in the other arm of the trial.

The primary outcome could be the number of men remaining on AM or AS at a median follow-up of, for example, 5 years, with potential extension depending on the rate of management change. Secondary outcomes would be adherence to AM or AS, patient satisfaction, number of complications from rebiopsy, quality-of-life score and reasons for seeking radical intervention.

Active monitoring for higher-risk or locally advanced prostate cancer

If the results of the ProtecT trial show that AM is an effective option for men with clinically localised T1–T2 PCa, then a future trial could investigate whether or not AM/AS is effective in men with higher-risk or more advanced T3a disease. This option for a RCT would be difficult to launch currently because of concerns about the levels of progression likely to be experienced, as well as the NICE guidance advising men in these groups to have radical treatment. 68

Nurse-led active monitoring

Qualitative research within the ProtecT trial has shown that both doctors and patients find nurse-led monitoring acceptable. As such, a potential trial that could be undertaken now could compare nurse-led clinics with usual care involving consultant urologists. Some patients in this study have indicated that appointments with urologists were often so short that they wondered if there was a need for them to be involved, and travel and parking considerations were also raised. If this trial were conducted, men randomised to monitoring by nurses would need the support of consultants either at fixed time periods (perhaps annually) or when any change in disease status or anxiety required a clinical review. A nurse-led AM/AS programme, such as the one employed in the ProtecT trial, could lead to NHS savings from freeing up consultant time. A full economic evaluation would need to be undertaken, and cost could be a key outcome. The primary outcome in such a trial could include satisfaction with and adherence to monitoring, with quality-of-life score and number remaining on monitoring as secondary outcomes. The randomisation could be performed at the individual level, with men visiting either nurses or consultants at that clinic, or it may need to be a cluster design.

Modelling the relationship between prostate-specific antigen and prostate cancer outcomes

We recommend that studies in this area share data in order to provide the best evidence for future management of men with clinically localised PCa. The PSA era began in the early 1990s, and the earliest ongoing study of AS began in 1995. In our systematic review, just eight of 7111 men followed for roughly 4 years had died from PCa. It will take some time before the worldwide cohorts of men on AM/AS mature, and, even then, the number of deaths from PCa may be small because men in these cohorts die from other causes. Thus, the best examination of the effect of PSA change on PCSM or metastases will be through combining data from various research cohorts, using meta-analysis and advanced modelling techniques, including, for example, the US NIH-funded PROMISS study now under way.

Reporting in active monitoring/active surveillance studies

The systematic review in Chapter 2 found little consensus on study design between 22 cohorts of men followed by AM/AS. Furthermore, there was little consensus on what to report, despite the similar aims of these studies. For instance, although the reasons men changed AM/AS to radical treatment were not provided by each study, this information is likely to be crucial in future comparisons (through meta-analysis) between studies. Given the low short-term PCSM rate found in the systematic review, now would be the ideal time for ongoing studies to agree on a core outcomes set (as in clinical trials), whereby each study records instances of metastases and PCSM, for instance. As discussed in Chapter 2, it is likely that future comparisons to decide on the optimal strategy for AM/AS will be carried out through meta-analysis. Thus, it is crucial that a wide range of studies is undertaken and that they report design, triggers and results in as similar a way as possible. Future studies should also collect data on those who choose to leave AM/AS and the reasons for this decision, and should continue to follow these patients. Data on quality of life will be important for future AM/AS studies, along with an increasing focus on imaging, biomarkers and genetic information.

Changing clinical environment

The developments in functional magnetic resonance imaging (fMRI) and other imaging technologies may change the diagnosis and risk-stratification of men with PCa. These developments, and others in relation to clearer understandings of genetic risk, will need to be pursued in future research. We would recommend that cohorts of men with PCa build up a biorepository allowing retrospective measurement and evaluation of any new biomarkers. This will increase power to detect if new biomarkers are useful for monitoring men with clinically localised disease.

Future qualitative research in active monitoring/active surveillance

In terms of qualitative research, future work may consider supportive interventions to help clinicians and men cope with the various uncertainties in AS. Clinical centres may want to consider the use of formal protocols, with the opportunity to reduce practice variation and increase clinician confidence and patient reassurance. These ideas were not pursued in the current project, but our observations of subtle differences between sites with and without protocols suggest that this may be an important avenue to consider in future work.

The research in Chapter 6 suggested that there may be greater emphasis on patient choice and preference where clinical uncertainty is heightened, although this was generated as a hypothesis rather than a finding. Future work may investigate these notions more fully and could consider the use of observational techniques to capture clinician–patient interactions, particularly at key decision-making points in patient pathways. We uncovered evidence to suggest that there were differences in how central the patient’s role in AS decision-making may be, although these ideas will require more thorough investigation. Future research could explore the impact of centre protocols on levels of patient involvement and how optimum patient involvement can be achieved for individual patients regardless of centre protocols.

Qualitative interviews with men on AS suggested that the value of outpatient clinical appointments would benefit from further investigation. In particular, future work may consider the content and duration of appointments. Although only raised by a small number of individuals, there was a suggestion in this study that routine face-to-face appointments with urologists may not be deemed essential by men and their partners, although this appeared to depend on the content and duration of appointments. Where outpatient appointments are unlikely to call for clinical examination or discussion of change in management, there may be options to promote non-face-to-face forms of communication [i.e. telephone, Skype™ (Microsoft Corporation, Redmond, WA, USA)], or appointments led by specialist nurses. A further option may be to conduct outpatient appointments in a targeted, rather than routine, manner. These different options could have considerable implications for NHS resources and patients alike, especially given the increasing prevalence of men on AS.

Finally, future qualitative work could reconsider the application of the AMS once more is known about the effectiveness of PSA monitoring and the clinical implications of PSA values/behaviour. These gaps in evidence were the primary reason for clinicians’ reluctance to view the AMS as being capable of influencing clinical decision-making at present. Ultimately, the development of AS and AM programmes, and systems such as the AMS, await the publication of ongoing research including the ProtecT trial.

Mr Smith active monitoring system vignette

FIGURE 11.

Mr Smith vignette presented on AMS.

Mr Jones active monitoring system vignette

FIGURE 12.

Mr Jones vignette presented on AMS.

Mr Evans active monitoring system vignette

FIGURE 13.

Mr Evans vignette presented on AMS.

Clinicians’ views on active surveillance eligibility

Prostate-specific antigen testing

Clinicians from three sites reported that patients received 3-monthly blood tests via their GPs. Some clinicians suggested that there was flexibility surrounding the 3-monthly PSA testing if dealing with stable patients who had been on AS for a prolonged period (i.e. 18 months to 2 years):

After 18 months or so you can go to 6-monthly.

U15

The consistency of PSA testing could be considered as one of the more fixed components of the AS programme for most sites. There was more variation observed across clinicians’ accounts at the site where the protocol was under development, with discrepancies in how regularly PSA was reportedly monitored, and how and when these patterns could change:

Well they um – 3- to 4-monthly PSAs first year, then variable.

SPN3

So following an outpatient discussion on the available evidence that that’s what we’re going to do, then the patients will have another PSA three months after that. And then depending on the results, there will be a review of the PSA 3 to 6 months after that.

U8

Overall, all clinicians suggested that there would be an initial PSA 3 months into starting an AS programme. Beyond this, reports tended to align with 3–4-monthly PSAs within the first year, followed by a flexible approach:

Um we generally do it for at least a year, I would say probably 2 years, and then we step it down a bit after that. But again, you know, if you’ve got somebody who is 69 on surveillance you’d probably drop it after a year.

U10

Digital rectal examinations

Clinicians’ reported practices of conducting DREs were similar across three of the sites. Nonetheless, clinicians’ views on the value of DREs varied by individual, irrespective of which site they were from.

Clinicians from three sites generally reported conducting DREs every 6 months, corresponding to whenever a patient had an outpatient appointment. Two urologists were distinct in conducting annual DREs. The fact that clinicians from one site produced wide-ranging responses was likely to be connected to the absence of a protocol:

How often do you do DREs then?

Ah that varies on whether – I mean if these are – on the basis that usually these are all T1C tumours, so impalpable, um then they will have a DRE when they come back obviously for the first repeat biopsy. And then I will do it afterwards in the outpatients [. . .] but there isn’t a set protocol, the same with the follow-up intervals, they are not prescribed, but I think it is important that one doesn’t rely just on the PSA.

In line with the comment above, clinicians across all sites often framed the DREs as a way to compensate for the limitations of PSA measures:

We do occasionally get patients whose PSAs remain stable and we do a rectal, in fact we had one just recently and did a rectal examination and it’s a significantly advanced disease and the dynamics have changed dramatically.

U2

Because PSA has been well shown not to be a perfect marker for progression and the DRE sometimes picks things up.

U1

Some clinicians commented that DREs were highly subjective – especially if performed by different clinicians over time:

I think if you’re having the same person carrying out the DRE every time, that that’s probably more useful. I know that that isn’t practical. But you’re getting different people’s interpretations of, you know, moderately enlarged to one person, you know, what degree is moderately enlarged? And um feeling a nodule had – you know, it’s personal interpretation, isn’t it? And I think there isn’t enough consistency probably.

SPN3

Some clinicians implied that DREs were not sensitive enough to detect change, yet still conducted these regularly for reasons stated earlier (i.e. to avoid relying on PSA alone). In the case of the clinician below, DREs were thought to have psychological benefits for patients:

I think that it’s valid psychologically to the patients. I think men find it very reassuring to be examined for something that’s not being treated. How much that really is a sensitive test of the clinical progression in somebody with low risk disease in the first year when realistically things tend not to change, I question.

U3

The greatest range of views on DREs came from the site at which the protocol was under development. Quotations demonstrating the disparate views at this site are shown in the main body of the report (see Chapter 6). Views ranged from one urologist believing that there was no value in conducting DREs in an AS programme that included rebiopsy, to another urologist implying that examining the prostate was one of the most fundamental procedures that should be carried out when trying to monitor prostate cancer.

Rebiopsy

Rebiopsy provoked the greatest variation of views across clinician interviews, and, for some, the greatest expression of uncertainty. The site protocol was consistently reported by most clinicians from site 1: rebiopsies were offered one year after starting on AS, followed by 2-yearly intervals. As before, clinicians viewed this as an outline, making it clear that there was flexibility within this depending on individual patient cases.

The greatest range of rebiopsy practices emerged from the site at which the protocol was under development. Variation was found in relation to ‘initial’ rebiopsy times (ranging from 6-monthly to yearly) and subsequent rebiopsies (ranging from yearly to ‘based on need’). Clinicians from this site were distinct from others in that they were explicit about the uncertainties over when to rebiopsy (see Chapter 6, Current practice within active surveillance programmes: follow up procedures and triggers for radical treatment for quotations that succinctly demonstrate this).

Patients’ perceptions of clinicians’ triggers for considering radical treatment

Men were asked to describe any changes in PSA that would concern them, but all responses were framed around explanations they recalled hearing from their clinicians. The most common response involved rises in PSA, described as a series of consecutive rises, or a steeply rising gradient of PSA figures:

And they’ve told me that one bad, one high figure, doesn’t mean anything. They wouldn’t react to [.] one high PSA figure. It’s only a, you know, an increasing pattern that they’ll respond to.

P8

Well I got the impression that if it started . . . I think that if it was rising fast, then I should be considering treatment.

P13

Three of 20 men talked about the PSA value of 10 ng/ml being a significant threshold, although each of these qualified this by saying that the trend in PSA values was more important than absolute figures. All three felt that clinicians had indicated there was some significance to a PSA value of 10 ng/ml. One of these men felt that this might have been an outdated idea:

I can remember from years ago that 10 used to be a point where they were started to get a bit panicky about it and think that perhaps they ought to do something about it. It sort of stuck in my mind. Now, I don’t think they’re quite concerned so much about that now. I think they’re more interested in a trend for you as a person, rather than 10 as a nasty point.

P2

Other individual examples of personal triggers for concern included a doubling of the most recent PSA figure (P1), and deviations from age-adjusted reference lines for PSA (P4). The men who mentioned these triggers had based this on discussion with their clinicians. P1 recalled asking his consultant what a ‘significant change’ in PSA value would be. P4 was the only patient to report being given age-adjusted reference lines that denoted average PSA values in men without PCa. Hovering slightly above this line was deemed acceptable to this man, in light of his PCa diagnosis:

In other words, I’m behaving like a mid-70-year-old in my late sixties. So, that bit of information has been very helpful to me, to put it into perspective.

P4

None of the men reported that he would look into radical treatment in the face of concerning PSA values. All participants responded that the next step would be to initiate a discussion with their consultant (and in one case, their GP), or to wait for the results of a rebiopsy.

Quality assurance checklist for reporting of qualitative research

The reporting of the qualitative methods were checked against the Consolidated criteria for Reporting Qualitative research (COREQ), a 32-item checklist for interviews and focus groups. 122 The checklist is shown below; the final column details where evidence can be found that the relevant criterion is satisfied.