FormalPara Key Points for Decision Makers

Subject to a number of assumptions and based on current information, we estimate that transperineal biopsies are cost effective compared with transrectal as long as the per-procedure acquisition cost is below £81.

Future research should focus on comparative diagnostic accuracy and infection risk, focusing particularly on men with intermediate-risk disease.

1 Introduction

Prostate cancer is the fourth most common cancer globally, with an estimated 1.8 m cases world-wide in 2018 [1] and 48,500 cases in the UK every year [2]. Current practice for the diagnosis of prostate cancer is for referred men to undergo a multiparametric MRI scan, followed by a transrectal ultrasound-guided biopsy (TRUSBx) [3]. However TRUSBx is not without risk, with an incidence of sepsis of around 1%, and other less serious complications occurring with greater frequency [3]. The cause of infection is most likely due to piercing of the bowel wall with the biopsy needle, before insertion into the prostate to retrieve the sample. To address this, there has been a move to perform prostate biopsies using the much more sterile transperineal route [4]. These ‘template biopsies’ (making use of a grid placed over the perineum to guide needle insertion points) usually necessitate general anaesthesia (GA) with significant attendant costs. To overcome this, devices have been developed to permit transperineal biopsies (TPUSBx) to be performed under local anaesthesia (LA) and hence be more suited to the outpatient clinic setting [5,6,7]. To date however, it is not known how cost effective these devices are, especially given the relatively low cost and wide availability of the transrectal biopsy method.

Here we investigated the cost effectiveness of TPUSBx devices compared with TRUSBx in the diagnosis of prostate cancer in a UK secondary care setting from the perspective of the UK National Health Service (NHS). As a case study, we used the novel Cambridge Prostate Biopsy (CamPROBE) device, which has been recently evaluated for clinical effectiveness and safety [7]. The CamPROBE is based on the concept of a co-axial cannula, but designed specifically for transperineal prostate biopsies under LA. The device and how it is used can be seen at https://www.youtube.com/watch?v=Q3XYLq5po8s&t=196s. Although we have used this device in this analysis, our findings should be broadly transferable to any TPUSBx device.

2 Method

We developed a decision model comprising a decision tree with Markov models at the terminal nodes. Data informing the decision model were taken from a prospective case series representing the first rigorous data on the safety and acceptability of the CamPROBE [7], and other data from the literature to inform the likely cost effectiveness of the device, at various price points. We also conducted a value of information analysis (VoI) to guide the direction of further research to reduce uncertainty as to the cost effectiveness of TPUSBx devices.

The design of the model mirrors the clinical diagnostic pathway (described below) to compare the expected lifetime costs and QALYs accrued with TPUSBx and TRUSBx.

2.1 Clinical Diagnostic Pathway: Current Management

Men presenting in primary care for suspected prostate cancer are typically offered a prostate-specific antigen (PSA) test. If sufficiently raised and concerning [8], patients are referred to secondary care. Current (2019) NICE guidance [3] recommends multiparametric MRI (mpMRI) as first-line investigation of men with suspected localised prostate cancer. Results are reported on a five-point Likert scale (PI-RADS score); those scoring 3 or more are recommended biopsy (TRUSBx) using the results of the MRI to inform placing of the needle. Those scoring 1 or 2 may omit a biopsy, or be offered systemic prostate biopsy (i.e. sampling a wide area of the prostate rather than focusing on anomalies identified in the MRI scan).

For those with MRI Likert of ≥ 3 but a negative biopsy, repeat biopsy might be considered. For those with a raised PSA but MRI Likert of 1 or 2 who have not had a biopsy, or for whom the biopsy was negative, the PSA test may be repeated at 3–6 months and biopsy offered if there is ongoing strong suspicion of cancer. Alternatively, the patient can be discharged to primary care with referral if pre-determined PSA thresholds are reached. Treatment options for localised cancer include active surveillance, radical prostatectomy or radical radiotherapy depending on prognosis.

Much of the current NICE guidance on the appropriate diagnostic pathway is based on a recent decision model comparing many permutations of testing strategies, comprising combinations of one or two TRUSBx or template-guided biopsies (under GA), with and without mpMRI (Faria et al. [9]). This model itself draws on data from a large trial of MRI imaging and biopsy in diagnosing prostate cancer (the diagnostic Prostate MR Imaging Study, PROMIS [10]), inter alia. The cost-effective pathway is mpMRI first line, followed by mpMRI-guided TRUSBx where clinically significant cancer is suspected, followed by a second TRUSBx if the first is negative (strategy ‘M7’ of Faria et al. [9]).

2.2 Modelled Clinical Diagnostic Pathway and Structure

In this analysis, we replicated strategy M7 as modelled by Faria et al. [9], but added TPUSBx as a comparator strategy in place of TRUSBx, and included the cost and consequences of adverse events such as fever and sepsis associated with TRUSBx (Fig. 1). Entry point into the model is referral to secondary care. In the TRUSBx arm (status quo), a man referred has a probability of having either clinically significant cancer (which is defined as either high or intermediate risk), clinically non-significant cancer or no cancer, according to the prevalence in the referred population. Firstline diagnostic is mpMRI, the results of which are either negative (i.e. no cancer), clinically non-significant or clinically significant.

Fig. 1
figure 1

Decision tree structure. CNS clinically non-significant, CS clinically significant, HR high risk, IR intermediate risk, mpMRI multi-parametric magnetic resonance imaging, NC no cancer, TPUSBx transperineal ultrasound-guided biopsy, TRUSBx transrectal ultrasound-guided biopsy, UTI urinary tract infection

A man with an mpMRI result of no cancer or clinically non-significant cancer is discharged to routine follow-up and exits the model. A man diagnosed with clinically significant cancer by mpMRI undergoes transrectal ultrasound-guided biopsy (TRUSBx), with an associated risk of fever, urinary tract infection (UTI), sepsis (with risk of death), or no infection. As per mpMRI, the result of the TRUSBx can be either no cancer, clinically non-significant or clinically significant. A man with a clinically significant TRUSBx result enters the treatment pathway, whilst a man with clinically non-significant or no cancer is given a repeat TRUSBx (with associated infection risks), for which the results can again be no cancer, clinically non-significant or clinically significant.

A man with a positive mpMRI followed by two negative biopsies (either no cancer or clinically non-significant) is discharged to routine follow-up and exits the model. A man with a positive mpMRI, a negative first biopsy (no cancer or clinically non-significant), and a positive second biopsy enters the treatment pathway.

The structure of the model is identical for a man presenting with high, intermediate, low risk or no cancer, but the conditional probabilities (e.g. diagnostic accuracy of TRUSBx) and long-term cost and consequences vary (long-term model described below). Likewise, the structure of the model for the TPUSBx arm is identical, except populated with different probabilities of infection and procedure cost (see model inputs below and Table 1).

Table 1 Model inputs

Long-term costs and outcomes were calculated from Markov models appended to the terminal nodes dividing disease into ‘progression free’, ‘metastatic disease’ and dead (Fig. 2). Transition probabilities, costs and health state utilities were assigned from the literature (Table 1), with a transition period of 1 year. Six possible scenarios were estimated, dependent on the true state of the disease and the subsequent treatment strategy. For men with no cancer, no further monitoring was assumed, and lifetime QALYs calculated based on UK lifetable statistics (scenario ‘no cancer’). For men diagnosed with clinically non-significant cancer, a strategy of active surveillance was assumed, comprising one urology follow-up appointment and three PSA tests per annum (scenario ‘clinically non-significant’). Transition probabilities reflected the possibility of this becoming metastatic disease in the future (intermediate cancer stages are not explicitly modelled). For men with intermediate-risk and high-risk disease, the treatment strategy could be either active surveillance or radical prostatectomy (scenarios ‘intermediate-risk active surveillance’, ‘intermediate-risk radical prostatectomy’, ‘high-risk active surveillance’ and ‘high-risk radical prostatectomy’).

Fig. 2
figure 2

Long-term model structure

The relevant scenario lifetime costs and outcomes were appended to the terminal nodes of the decision tree: patients with no cancer were discharged back to primary care (scenario ‘no cancer’). Patients with clinically non-significant cancer were assigned to the active surveillance strategy (scenario ‘clinically non-significant’). Patients with correctly identified intermediate- and high-risk cancers were assigned to the respective radical prostatectomy strategy (‘intermediate-risk radical prostatectomy’ or ‘high-risk radical prostatectomy’ scenarios); and those whose cancers were misdiagnosed as clinically non-significant or no cancer were assigned to the intermediate- and high-risk active surveillance strategy/scenarios, respectively.

2.3 Model Inputs

Chance node probabilities and associated hyperparameters as well as health state utilities were replicated from the same sources as Faria et al. [9] (see Table 1). Key inputs included the sensitivity and specificity of mpMRI and first and second TRUSBx, which were extracted from the PROMIS study [10] (published 2017) and a 2015 systematic review [11]. These data suggested the sensitivity of both mpMRI and TRUSBx is 1 in the presence of high-risk cancer; that is, a high-risk cancer will never be misdiagnosed, whilst diagnosis of clinically non-significant cancer and intermediate-risk cancer carries a risk of misdiagnosis as per Table 1. After Faria et al. [9], we assumed perfect specificity of TRUSBx (i.e. a patient with no cancer will always be correctly diagnosed), but imperfect sensitivity as per Table 1. Dirichlet, Connor-Mosimann or modified Connor-Mosimann distributions were fitted [12] to the summary statistics reported in Faria et al. [9], which were then inserted into the model (see Appendix 1 in the Electronic Supplementary Material [ESM]). Note, as there are three possible outcomes from the biopsy (no cancer, clinically non-significant or clinically significant), there is no single measure of ‘sensitivity’ or ‘specificity’ as such, but probabilities of one of the three results, conditional on the true disease state. From hereon we refer to these measures as ‘diagnostic accuracy’.

We assigned the same probability distributions to the TPUSBx diagnostic accuracy as TRUSBx. The implied assumption of this is that on average we expect there to be no difference between the two methods in these parameters, but as the distributions are modelled independently, the sensitivity and specificity can each vary according to current levels of uncertainty.

The risk of infection associated with TRUSBx was extracted from the treatment arm of a Cochrane systematic review of the effects of antibiotic prophylaxis for TRUSBx [13], and mortality from sepsis was estimated [14] from US Surveillance, Epidemiology and End Results (SEER) data [15]. Our base case assumed a zero risk of infection associated with TPUSBx. This was based on the prospective case series of 40 TPUSBxs from the recent CamPROBE published results [7].

To estimate long-term model transition probabilities, parametric distributions were fitted [12] to the summary statistics reported in Faria et al. [9] (see Appendix 1 in the ESM). Incidence of adverse events following radical prostatectomy and active surveillance were extracted from a randomised controlled trial (RCT) of the two modalities [16], converted to 1-year probabilities.

Unit costs were extracted from routine NHS unit cost sources [17,18,19] for the price year 2018–19. The unit cost of the TPUSBx and TRUSBx procedures was based on a microcosting exercise (Appendices 2 and 3, see ESM). As this is a small sample from one centre only and the price of the TPUSBx device itself is unknown, our base-case analysis assumes equal procedure time and a zero price for the TPUSBx device: We explore the various price points in sensitivity analysis. Other cost inputs were based on previously described treatment regimens [20]. Health state utilities were extracted from relevant previous reports and sources [9, 21] (Table 1).

2.4 Analysis

The model and inputs were reviewed for face validity and clinical plausibility by the clinical lead (VG). Internal validity checks were conducted to test for bugs and errors (details and model code available from corresponding author). Analysis was based on a 50-year-old male and run for 30 years, representing the expected life span of the individual. Future costs and QALYs were discounted at the UK recommended rate of 3.5% [22]. Analysis was conducted probabilistically via Monte Carlo simulation, repeatedly running the model with sets of inputs drawn from their respective distributions. Stability testing determined the appropriate number of simulations, with a coefficient of variation of estimates of (a) mean incremental net benefit and (b) standard error of mean incremental net benefit below 2% declared stable.

We reported mean cost and QALYs associated with TRUSBx and TPUSBx, increments and 95% credibility intervals, incremental cost-effectiveness ratios and incremental net benefit (INB) at a threshold of £20,000 per QALY. As no price is currently set for the model TPUSBx device (CamPROBE), our base case assumes a zero price, and we draw readers’ attention to our sensitivity analyses:

  1. 1.

    We present a one-way sensitivity analysis on the price of the TPUSBx device, identifying the price associated with an ICER of £20,000.

  2. 2.

    Our base case assumes the risk of infection with TPUSBx is zero, thus we present a one-way sensitivity analysis on risk of infection with TPUSBx, varying the risk between 0 and 100% of that of TRUSBx. The base-case price for CamPROBE is assumed in this analysis.

  3. 3.

    As the reduction in risk of infection is considered the primary benefit of TPUSBx, we present a two-way sensitivity analysis showing the maximum cost-effective per-procedure price of the TPUSBx device as a function of the infection risk.

Finally, we present a VoI, estimating the expected value of perfect information (EVPI) and expected value of perfect parameter information (EVPPI) to help guide future research efforts to where they will be of most value. VoI parameters are calculated for the relevant patient population of England over a time horizon of 10 years (337,516; Appendix 4, see ESM). As the value for further information is dependent on the point estimate cost effectiveness, which itself is dependent on the price of the TPUSBx device, we present two analyses: one at the base case zero price and one at the maximum cost-effective price. At this price, the expected ICER is £20,000/expected INB is £0 and so decision uncertainty is maximised. This thus represents an absolute upper bound to the value of further information.

3 Results

Stability testing suggested that 200,000 iterations were sufficient for purpose (Appendix 5, see ESM).

At a zero price of the TPUSBx device, there is a 59% probability that it is cost effective compared with TRUSBx (Table 2 and Appendix 6, Table A6.1 [see ESM]). The cost-neutral per-procedure price (i.e. the price yielding a zero incremental cost) is £40.82. The maximum cost-effective price (yielding an ICER of £20,000 per QALY) is £81.17 (Fig. 3a). At this price there is a 50% probability of cost effectiveness. Note, this is the per-procedure price; thus, for the example of CamPROBE where two devices are required, the maximum cost effective price is £40.59.

Table 2 Point estimate cost effectiveness at equal procedure cost, and maximum cost-effective per-procedure price for TPUSBx device, mean (95% credible interval [CrI])
Fig. 3
figure 3

Sensitivity analyses on a price and b price vs risk of infection. a INB vs per-procedure price. Price expressed as per procedure price of TPUSBx, not per unit (e.g. if two units required, then the unit price is half that stated). Data reported in Appendix 6, Table A6.1 (see ESM). b Maximum cost-effective price vs risk of infection. Shading shows INB; lighter shading indicates higher INB. Solid line shows locus of points yielding an INB of £0 (= ICER of £20,000), and thus shows the maximum cost-effective price as a function of the infection risk. Risk of infection is expressed as a proportion of the risk associated with TRUSBx (i.e. relative risk, so risk = 1 means same probability of infection, risk = 0.5 means 50% probability of infection). Price is expressed as per procedure price; thus, if TPUSBx requires 2 units, the max unit cost is half the stated price. Points to the south-west of the line yield an INB > £0 (ICER < £20,000), those to the north-east yield an INB < £0 (ICER > £20,000). One-way sensitivity analysis against risk of infection alone is shown in Appendix 6, Fig A6.2 (see ESM). INB incremental net benefit, ESM electronic supplementary material, ICER incremental cost-effectiveness ratio, TPUSBx transperineal ultrasound-guided biopsy, TRUSBx transrectal ultrasound-guided biopsy

The above prices assume a zero infection risk with TPUSBx. As the risk of infection approaches that of TRUSBx, the maximum per-procedure cost-effective price falls to approximately £14.50 (Fig. 3b).

Value of information analysis suggests the overall value of eliminating all decision uncertainty is worth between £56 m and £65 m to the population of England, depending on the price of TPUSBx (Fig. 4 and Appendix 6, Tables A6.2–A6.4, see ESM). This is particularly focused on the diagnostic accuracy of a second transperineal or transrectal biopsy following a first biopsy result of ‘clinically non-significant cancer’, when the true disease state is intermediate-risk cancer (group EVPPI £46.2 m to £54.7 m, depending on the price of TPUSBx). There may also be value in reducing uncertainty in the probability of infection with TRUSBx (up to £5.0 m), long-term prognosis (£4.4 m) and the diagnostic accuracy of biopsy in patients with a true disease state of anything other than no cancer (between £1.8 m and £3.2 m each). However, this is only true when the price for CamPROBE is towards the upper end of its maximum cost-effective price; at a lower price for CamPROBE, there is minimal value in reducing uncertainty in these parameters.

Fig. 4
figure 4

Expected value of perfect (parameter) information (EVPPI) to England, assuming maximum cost-effective price for CamPROBE (£81.17). Figure shows EVPI and EVPPI associated with various groups of parameters. CaP cancer of the prostate, CNS clinically non-significant, HR high risk, IR intermediate risk, NC no cancer, TP transperineal, TR transrectal

4 Discussion

4.1 Interpretation of Results

Given current information, a per-procedure price for a TPUSBx device of up to £81.17 is expected to yield an incremental cost-effectiveness ratio at or below £20,000 per QALY gained. This price is derived from both the differences in consumables required between the procedures and the assumed elimination of risk of infection associated with TRUSBx, which both averts loss of QALYs and reduces health system cost. We also assumed that on average, TRUSBx and TPUSBx had the same diagnostic accuracy.

At this maximum price, decision uncertainty is also maximised, with a 50% probability that TPUSBx is cost effective. At a zero price, there is a 59% probability of TPUSBx being cost effective. This decision uncertainty is reflected in an EVPI of £56 m or £65 m (at zero or £81.17 per-procedure price, respectively). This represents the opportunity loss due to uncertainty and can be equally expressed as 2818 or 3271 QALYs foregone to patients in England.

The interpretation of this result is that at or below the maximum price, TPUSBx is on average the cost-effective option, but this is uncertain, and if wrong there is a loss to society in terms of foregone health gain: either a TRUSBx would have been better for diagnosing prostate cancer or TPUSBx is overpriced leading to a net loss of health gain as excessive resources are diverted to TPUSBx. The probability of being wrong (incurring an opportunity loss) is around 41% (50% at maximum price). So, in 41 (50) of 100 possible realisations of the world, TPUSBx is not cost effective. The expected loss due to uncertainty (= 41% * loss or = 50% * loss) is around 2818 (3271) QALYs, or £56 m (£65 m) when each QALY is valued at £20,000. This represents the maximum amount that should be paid for research to eliminate uncertainty. At the cost-neutral price of £40.82, the EVPI is £61 m, or 3034 QALYs.

Exploring further, the parameters responsible for the greatest opportunity loss due to uncertainty are those relating to the diagnostic accuracy of a second biopsy (whether TP or TR), following a first biopsy result of clinically non-significant cancer, when the true health state is intermediate-risk cancer (EVPPI = £46m or £55m, accounting for 82% or 84% of the entire EVPI, Appendix 6, Table A6.2, see ESM). There is also some value in reducing uncertainty in the probability of infection with TRUSBx, long-term prognosis from different management strategies, and diagnostic accuracy of the entire biopsy pathway (both first and second biopsy) in patients with a true state of high-risk cancer or clinically non-significant cancer.

These recommendations for research are for very specific subpopulations of those men undergoing prostate biopsy and must be considered within the context of the overall assumptions of the model. In our base case we assumed (1) a zero risk of infection with TPUSBx, (2) on average equal sensitivity and specificity between TRUSBx and TPUSBx, (3) perfect specificity and (4) perfect sensitivity in high risk disease. Whilst recent systematic reviews do not contradict these assumptions [23, 24] (see Sect. 4.3), this may not always be the case in every-day practice. The EVPPI of £46 m to £55 m represents the upper limit for a trial budget that would eliminate uncertainty. A trial of finite sample size can only reduce, and not eliminate uncertainty, so the expected value of sample information (EVSI), which is a function of the sample size, will be lower than this. Calculation of EVSI is extremely computationally expensive. Several statistical approximation methods are available [25], but unfortunately due to the nature of our data we were unable to generate solutions to the EVSI parameters.

Taking the EVPPI results, sensitivity analyses and base-case assumptions into account, a suitably powered comparative study of the risk of infection and diagnostic accuracy associated with TRUSBx devices versus TPUSBx may be warranted. Diagnostic accuracy can be established with patients acting as their own controls and/or in those whose disease status is already known, if they are willing to undergo both procedures, whereas establishing risk of infection may require an RCT design. Consideration should be given to enriching the enrolled population in these studies with men with a diagnosis of, or high prior probability of intermediate-risk cancer.

4.2 Strengths and Weaknesses

Our analysis presents a synthesis of current evidence on the cost and consequences of transperineal in place of transrectal ultrasound-guided biopsy in the diagnosis of cancer of the prostate. The conduct and reporting of our analysis conforms to best practice in economic evaluation [22, 26]. However, there are a number of limitations.

Any decision model is only as reliable as its assumptions. Our estimate of the maximum price is based on differences in consumables cost and elimination of infection risk. The quality of the data (reflecting the stage of development of the model device) must be considered when addressing the generalisability of our results. For example, the observed difference in consumables cost is driven by replacing the transrectal biopsy needle and guide with the CamPROBE device and based on a small sample of biopsies conducted in one centre (a large research-intensive teaching hospital). Consideration must be given as to whether this is generalisable to other settings; we provide full microcosting data in Appendix 2 to assist this (see ESM).

The other major component of value is the reduction in biopsy-associated infection. Indeed, this is the major anticipated benefit. Whilst we feel the estimates of QALY loss and health service cost associated with infections are plausible, the biggest uncertainty is the probability of infection itself. Data and budgetary limitations prevented us from assigning a plausible probability distribution for this (e.g. by conducting a formal expert elicitation process), so we addressed this by conducting a two-way sensitivity analysis, showing the maximum price as a function of the risk of infection, but stress the need for good quality comparative data to establish the difference in infection rate.

We also assumed an on-average equal diagnostic accuracy for transrectal and transperineal biopsies. We conducted an additional one-way sensitivity analysis on the probability of a true positive in patients with intermediate-risk cancer, concluding that transperineal biopsies remained cost effective as long as the relative probability was at least 98.8–99.97% that of transrectal, depending on the price of the transperineal biopsy device, with a higher price giving less ‘room’ for a reduction in relative diagnostic accuracy (Appendix 6, Tables A6.6 and A6.7, see ESM). Note, the figures quoted above are for the relative true-positive rate, not the absolute.

Finally, we did not include any costs for training or the impact of any learning curve effect in the analysis. In the experience of our clinical lead (and inventor of CamPROBE, VG), transferring from transrectal to transperineal biopsies is relatively straightforward with minimal training requirements. So whilst our analysis could be regarded as a ‘steady state’, assuming surgeons are fully competent in the transperineal technique, any additional training costs should be minor.

4.3 Comparison with Other Studies

We are not aware of any other economic evaluations comparing the cost effectiveness of transrectal and transperineal ultrasound-guided devices for local anaesthetic biopsies. Several transperineal biopsy devices have recently entered the market; a 2020 systematic review [23] comparing transrectal and transperineal methods identified 14 studies enrolling approximately 2000 patients. However, all but one of these were single-arm cohort studies. The one comparative study compared both techniques in the same patients [27]. The review concluded that both techniques were of similar sensitivity and specificity, although the diagnostic odds ratio may favour the transrectal approach [23]. Another (2019) systematic review [24] identified four RCTs (and a number of observational studies), concluding there was no evidence of a difference between transrectal and transperineal biopsies, although this review did not distinguish between GA-based template and non-template LA transperineal biopsy and did not explore sensitivity and specificity, instead reporting the ratio of cancer detection rates. However, the authors noted that TPUSBx was associated with a lower risk of fever and rectal bleeding. High-quality comparative data specifically comparing infection rates with the two techniques are somewhat limited, although a number of smaller single-centre and retrospective database analyses suggest an effective zero rate of sepsis with transperineal biopsy [4, 28,29,30,31,32].

The most influential economic study on UK prostate biopsy policy is the Faria et al. [9] model, drawing on the PROMIS study [10], on which UK guidelines are largely based. Other economic studies support the use of mpMRI first line [33,34,35,36] and newer biomarker tests appear cost effective compared with a first-line biopsy [37, 38]. However, comparison with mpMRI first line is unknown.

5 Conclusion

Transperineal ultrasound-guided biopsies have the potential to be cost effective with a device priced at up to £81.17 per procedure. This price must be divided by the number of units required per procedure to obtain the maximum unit price: if two devices are used, this equates to £40.59 per unit. A cost-neutral price to the NHS is £40.82 per procedure. The greatest value of further research is in the diagnostic accuracy of TPUSBx versus traditional transrectal prostate biopsy, and in the risk of infection associated with the two biopsy modes. Consideration should be given to enriching the enrolled patient population with men with either known or a high prior probability of intermediate-risk disease. New biomarkers are showing potential to assist in the diagnosis of prostate cancer, which may change the prevalence of disease in those eventually referred for biopsy, and hence the cost effectiveness of different diagnostic pathways. Future economic modelling aimed at informing the next revision of guidelines and clinical pathways should consider both (non-template) transperineal biopsy procedures as well as biomarker tests.