DIAGNOSTICS ASSESSMENT PROGRAMME

DIAGNOSTICS ASSESSMENT PROGRAMME Evidence overview Diagnosis and monitoring of prostate cancer: PROGENSA PCA3 assay and the Prostate Health Index (PHI) This overview summarises the key issues for the Diagnostics Advisory Committee s consideration. It includes a brief description of the topic, a description of the analytical structure and model, a discussion of the analytical difficulties, and a brief summary of the results. It is not a complete summary of the diagnostics assessment report, and it is assumed that the reader is familiar with that document. This overview contains sections from the original scope and the diagnostics assessment report, as well as referring to specific sections of these documents. 1 Background 1.1 Introduction The PROGENSA PCA3 assay was selected by the Medical Technologies Advisory Committee (MTAC) for the Diagnostics Assessment Programme to develop recommendations on its use in the NHS. One other technology, the was identified during the scoping phase and is included in the assessment as an alternative technology. The purpose of this assessment is to consider the clinical and cost effectiveness of using the PROGENSA PCA3 assay or the Prostate Health Index as an aid in diagnosis in people undergoing investigations for suspected Issue date: November 2014 Page 1 of 78

prostate cancer. Provisional recommendations on the use of these technologies will be formulated by the Diagnostics Advisory Committee at the Committee meeting on 21/01/2015. 1.2 The Conditions 1.2.1 Prostate cancer Prostate cancer is the most common cancer in men in the UK, and represented 26% of all male cancers in England and Wales in 2010 (NICE CG175). It is estimated that there are around 40,000 new cases of prostate cancer diagnosed in the UK every year, and in 2010 there were 34,892 new cases of prostate cancer registered in the England. Prostate cancer is more likely to affect older men, and most cases are diagnosed in men over 50 years of age. In the UK, between 2008 and 2010, it is estimated that 75% of prostate cancer diagnoses were in men aged 65 years and older, with less than 1% of diagnoses being in men age under 50 years. Prostate cancer is also more common among men of African- Caribbean and African descent, whilst it is relatively rare amongst men of Asian and South and Central American descent. Men with a close male relative who has developed prostate cancer are also thought to be at an increased risk of the disease (around 120-150%). Prostate cancer also has a strong inverse relationship with deprivation, with incidence rates lower in men from lower socio-economic groups. Transgender women can also be affected by prostate cancer. Prostate cancer is the second most common cause of cancer death among men in the UK and in 2010 there were 10,721 deaths from prostate cancer in the UK (Cancer Research UK 2012). Around 90% of men who are diagnosed with stage 1or 2 prostate cancers will be alive 5 years after their diagnosis and it is estimated that between 68-90% of men will be alive 10 years after their diagnosis. Men with early stage prostate cancer may have disease that is slow to develop, meaning that they are more likely to die from other causes. Issue date: November 2014 Page 2 of 78

Men with locally advanced (stage 3) prostate cancer have a survival rate of 70-80% 5 years after diagnosis. Around 20-30% of men already have metastatic disease (stage 4) when they are diagnosed with prostate cancer; it is estimated that 30% of men with advanced prostate cancer will be alive 5 years after they are diagnosed (Cancer Research UK 2012). Diagnosis of prostate cancer often involves a biopsy of the prostate gland. According to Hospital Episode Statistics (HES) Online (Hospital Outpatient Activity), in 2012-13, there were 17,284 outpatient attendances in England associated with a rectal needle biopsy of the prostate and 1353 with perineal biopsy of the prostate. It should be noted that these figures are likely to substantially underestimate the number of biopsies performed because HES data only record inpatient episodes and day cases, and prostate biopsies are commonly performed as an outpatient procedure. Based on the number of men diagnosed with prostate cancer every year, anecdotal evidence suggests that the number of biopsies performed every year is more likely to be in the region of 80,000.Prostate biopsies are associated with side effects; relatively common minor complications include haematospermia, haematuria and rectal bleeding which subsides after intervention, whilst major complications, which are comparatively rare, include sepsis, prostatitis, fever, urinary retention, epididymitis, and rectal bleeding for longer than two days (NHS Prostate Cancer Risk Management Programme 2010). An important, and also under recognised side effect of prostate biopsies is the detection of clinically insignificant prostate cancer which may lead to unnecessary invasive treatments. Patient issues and preferences 1.2.2 Patient issues and preferences A biopsy is an invasive procedure and is associated with complications such as discomfort, bleeding and urinary retention. Biopsies can be a significant source of anxiety for patients, particularly when associated with uncertain diagnostic results and a need to repeat the biopsy procedure. The levels of anxiety associated with the biopsy procedure can be higher for a second Issue date: November 2014 Page 3 of 78

biopsy than the initial biopsy because of the patient s previous experience of the procedure. This experience leads to many patients preferring to have a general anaesthetic for a second biopsy procedure. Using PROGENSA PCA3 assay or the PHI assay may reduce the number of invasive diagnostic tests for a patient and shorten the length of time that a patient needs to experience exploratory diagnostic tests. It is reported that early treatment decisions for prostate cancer can be difficult, with patients often feeling that they have to choose between quality of life and longevity, especially given the uncertainty of treatment outcomes (Roth et al. 2008). Many patients may consider multiple second opinions regarding primary therapy; often the amount of information they are faced with leads to significant anxiety while trying to make a reasonable treatment decision. Once treatment has commenced, the side effects of the treatments, such as hormonal therapy, and pain medications, can also cause distress. The major quality of life concerns are around sexual function, incontinence, pain, and fatigue. 1.3 Diagnostic and care pathways The process for diagnosing prostate cancer is described in NICE Guideline 175 (CG175) Prostate cancer: Diagnosis and treatment. The guideline recommends the following before performing a biopsy: To help men decide whether to have a prostate biopsy, discuss with them their prostate-specific antigen (PSA) level, digital rectal examination (DRE) findings (including an estimate of prostate size) and comorbidities, together with their risk factors (including increasing age and black African-Caribbean family origin) and any history of a previous negative prostate biopsy. Do not automatically offer a prostate biopsy on the basis of serum PSA level alone. Issue date: November 2014 Page 4 of 78

Give men and their partners or carers information, support and adequate time to decide whether or not they wish to undergo prostate biopsy. Include an explanation of the risks (including the increased chance of having to live with the diagnosis of clinically insignificant prostate cancer) and benefits of prostate biopsy. If the clinical suspicion of prostate cancer is high (usually indicated by an abnormal DRE and a PSA value >100), patients will be referred immediately on entering secondary care for investigation of bone metastases using an isotopic bone scan or plain radiograph. If there is evidence of bone metastases, a patient begins treatment and does not undergo a prostate biopsy for histological confirmation of the cancer diagnosis. It is also recommended in CG175 that prostate biopsies should be carried out following the procedure recommended by the Prostate Cancer Risk Management Programme (2006), Undertaking a transrectal ultrasound guided biopsy of the prostate. This recommends that the prostate should be sampled through the rectum unless there is a specific condition that prevents this and also that the scheme used at first biopsy should be a 10 to 12 core pattern that samples the midlobe peripheral zone and the lateral peripheral zone of the prostate only. Transrectal ultrasound (TRUS) biopsy is usually carried out with the patient under local anaesthetic and involves using thin needles to take around 10 to 12 small pieces of tissue from the prostate. For people who have a negative first prostate biopsy, CG175 states that a core member of the urological cancer MDT should review the risk factors of all men who have had a negative first prostate biopsy, and discuss with the man that: there is still a risk that prostate cancer is present and the risk is slightly higher if any of the following risk factors are present: Issue date: November 2014 Page 5 of 78

the biopsy showed high-grade prostatic intra-epithelial neoplasia (HGPIN) the biopsy showed atypical small acinar proliferation (ASAP) abnormal digital rectal examination. The guideline (CG175) also recommends that multiparametric MRI (using T2 and diffusion-weighted imaging) be considered for men with a negative transrectal ultrasound 10 12 core biopsy to determine whether another biopsy is needed. A repeat biopsy should not be offered if the multiparametric MRI (using T2- and diffusion-weighted imaging) is negative, unless any of the risk factors (listed above) are present. In current NHS practice, a multiparametric MRI may not be carried out until 6 to 12 weeks after the TRUS biopsy because any haemorrhage after the biopsy can cause artifacts in the images and this may reduce the diagnostic accuracy of the prostate multiparametric MRI. A second biopsy may be undertaken using a template biopsy. A template biopsy is undertaken using a template grid either with a cross sectional MRI (where available) or using transrectal ultrasound imaging with transperineal sampling of the prostate under general anaesthetic. Usually, around 25 to 40 samples of the prostate are taken during a template biopsy. In NICE Interventional Procedure Guidance 364 (2010), it states that Current evidence on the efficacy of transperineal template biopsy shows an increase in diagnostic yield in patients with suspected prostate cancer who have had negative or equivocal results from other biopsy methods. There are no major safety concerns. Therefore, this procedure may be used for this indication provided that normal arrangements are in place for clinical governance, consent and audit. However, the guidance also states that Evidence was not found to support the use of transperineal template biopsy of the prostate as a mapping technique to determine the exact location and extent of prostate cancer in order to guide focal therapy, nor as part of an active surveillance Issue date: November 2014 Page 6 of 78

regime. Therefore the procedure should be used with these intentions only with special arrangements for clinical governance, consent and audit or research. Another type of biopsy that is used for a second biopsy is the saturation biopsy, which involves more than 20 cores being taken from the prostate. A saturation biopsy may be performed transrectally or using a transperineal approach. The transperineal approach is generally performed as a stereotactic template-guided procedure under general anaesthesia. The saturation approach leads to improved sampling of the anterior zones of the gland, which may be under-sampled in a TRUS biopsy and which may lead to cancer cells being missed. A third option that can be used for a second biopsy is the targeted approach. This approach is similar to the saturation approach in that it also aims to improve sampling of the anterior zones of the gland. There are three methods of MRI guidance for the performance of targeted prostate biopsy: cognitive fusion, in which the ultrasound operator directs the biopsy needle to the prostate area where a lesion was identified on the MRI (reviewed prior to biopsy); direct MRI-guided biopsy, a biopsy performed within an MRI tube; using a fusion device, in this method information from the MRI of the prostate, which is performed before the biopsy and stored in the device, is fused with real-time ultrasound using a digital overlay. The fusion creates a three-dimensional reconstruction of the prostate, and allows the aiming and tracking of biopsy sites. The targeted approach allows urologists to progress from blind, systematic biopsies, to biopsies which are mapped, targeted and tracked. The use of this approach relies on the availability of radiologists with relevant expertise and experience, and access to multiparametric MRI. Issue date: November 2014 Page 7 of 78

Nomograms or risk calculators can also be used to help predict repeat biopsy results. Specific risk calculators are not cited in NICE CG58. An example of a European risk calculator for assessing the likelihood of prostate cancer after a negative biopsy is risk calculator number four available from the Prostate Cancer Research Foundation, Rotterdam. This tool requires the DRE findings (normal or abnormal), transrectal ultrasound findings (normal or abnormal), prostate volume (ml), PSA level (ng/ml) and history of previous negative biopsy to be input and calculates both the chance of having a subsequent positive biopsy and the chance of having a high grade or advanced prostate cancer. Nomograms are not commonly used in the NHS as aids to shared decision making, but often the various parameters that are included in nomograms are used when discussing the risk of cancer with patients following an initial biopsy. Management and treatment Following the introduction of PSA testing, most prostate cancers are now either localised or locally advanced at diagnosis, with no evidence of spread beyond the pelvis. The treatment of prostate cancer is also described in NICE CG175, on page 163 the guideline acknowledges that with the uncertain, and often indolent, natural history of prostate cancer, and the wide range of management options, treatment decision-making in localised prostate cancer is difficult. For patients that have an initial negative TRUS biopsy, with no indication of risk from other risk factors, the usual care is PSA surveillance, with repeat PSA testing taking place every 3 months. A number of management approaches and treatments are available for localised disease, including: active surveillance, radical prostatectomy, external beam radiotherapy and brachytherapy. Hormone therapy (androgen deprivation or anti-androgens) is the usual primary treatment for metastatic Issue date: November 2014 Page 8 of 78

prostate cancer, but it is also increasingly being used for men with locally advanced, non-metastatic disease. The function of hormone therapy on prostate cancer is to stop testosterone feeding prostate cancer and encouraging growth. Androgen deprivation therapy (ADT) blocks the production of androgens including testosterone, with the aim of slowing the growth of prostate cancer cells. Most men who receive ADT for prostate cancer will receive the treatment for anything between a few months up to a few years. Around 9000 newly diagnosed men (26% of all new diagnoses) in the UK will receive ADT each year. Active surveillance is a way of monitoring prostate cancer that aims to avoid or delay unnecessary treatment in men with less aggressive cancer. It is suitable for men with low risk early stage prostate cancer that is contained within the prostate gland. It may also be suitable for some men with intermediate risk cancer. In active surveillance prostate cancer is monitored with regular tests that involve PSA testing every three to six months and DRE every six months for the first two years, then annually for subsequent years. Repeat biopsies are also likely to form part of an active surveillance programme. Watchful waiting is a way of monitoring prostate cancer that isn t causing any symptoms or problems. The aim is to monitor the cancer over the long term, and avoid treatment unless symptoms or disease progression occur. This approach involves PSA testing in primary or secondary care every 3 to 12 months. Additional DRE and other blood and urine tests may also be carried out depending on clinical judgment. Additional biopsies are not carried out. Radical prostatectomy involves removal of the entire prostate gland and seminal vesicles. Surgery has been traditionally performed by Issue date: November 2014 Page 9 of 78

an open retropubic or perineal approach. The risks associated with surgery include incontinence, erectile dysfunction and the chance of involved surgical margins. Recently, laparoscopic or robotically assisted techniques have shortened inpatient stays and reduced blood loss. Radical prostatectomy is a major operation that is typically only offered to fitter men without co-morbidities. External beam radiotherapy is the most common treatment in the UK for men diagnosed with localised prostate cancer. It is usually preceded by a period of hormonal therapy, and is given in daily fractions over 4 8 weeks as an outpatient procedure. The side effects of this treatment can include alteration in urinary and bowel function and erectile dysfunction. Brachytherapy (internal radiotherapy) may be used to try to cure prostate cancer that is completely contained within the prostate gland. 1.4 The population The population covered in this assessment is people who are suspected of having clinically significant prostate cancer, in whom the results of an initial prostate biopsy were negative for malignant disease, who are being considered for a second biopsy. 2 The technologies The PROGENSA PCA3 assay and the are in vitro diagnostic tests that are intended for use in people who are undergoing investigations for suspected prostate cancer, who have had a negative transrectal ultrasound prostate biopsy. The tests detect specific biomarkers that, when present at high levels, can suggest the presence of cancer. Both tests are intended to be used in conjunction with a review of risk factors, such as raised prostate specific antigen (PSA) levels and digital rectal examination findings, to help determine the need for a second biopsy to rule out the presence of prostate cancer. Detection rates of prostate cancer are around Issue date: November 2014 Page 10 of 78

14-25% for first biopsy (NICE CG175), and it is estimated that a significant proportion of people may receive a false negative result and require further investigations, including a second biopsy, to rule out prostate cancer. Biopsy procedures are invasive and are commonly associated with minor complications. In rare cases, biopsies can lead to major complications such as sepsis. The use of the PROGENSA PCA3 assay or the Prostate Health Index may avoid second biopsies and associated complications by indicating which patients have a decreased likelihood of a positive biopsy result and therefore, are unlikely to have prostate cancer and do not need a second biopsy. 2.1 The PROGENSA PCA3 assay The PROGENSA PCA3 test is an in vitro nucleic acid amplification test and is intended for the quantitative determination of PCA3 ribonucleic acid (RNA) in urine. The PROGENSA PCA3 assay requires a digital rectal examination, designed to release prostate cells and RNA into the urinary tract followed by a first catch urine sample of at least 2.5ml. Once collected, 2.5ml of the sample is added to a transport tube containing Urine Transport Medium that triggers the lysis of any remaining prostate cells and stabilises the RNA. The PROGENSA PCA3 assay incorporates two nucleic acid amplification tests; one test for detecting PCA3 RNA and one test for detecting prostate specific antigen (PSA) RNA. PCA3 RNA is highly overexpressed in prostate cancer tissue cells compared to adjacent benign tissue, whereas PSA gene expression is similar in both cancerous and benign prostate cells. By combining the detection of both genes in one assay, the PSA RNA levels can be used to normalise the amount of PCA3 RNA in molecular test samples so that higher levels of PCA3 RNA can be detected. The PROGENSA PCA3 assay quantifies the amount of both PCA3 and PSA RNA present in the sample and generates a PCA3 score based on the ratio of PCA3 RNA: PSA RNA. The PCA3 score can then be used to aid the risk stratification of people Issue date: November 2014 Page 11 of 78

for repeat biopsies; higher PCA3 scores are associated with a higher probability of a positive biopsy. For the purposes of this assessment the threshold PCA3 score used to indicate the likelihood of a positive biopsy will be 25. It should be noted that this assay analyses the levels of PSA RNA found in the urine following a digital rectal examination. The PSA values reported by this assay are not the same as those that would be reported using the standard PSA test for prostate cancer, since the standard test detects the levels of PSA protein in the serum of a blood sample. PROGENSA PCA3 Assay can be used with the Hologic Gen-Probe Direct Tube Sampling (DTS) 400, 800 and 1600 molecular laboratory systems. The PCA3 Assay is not compatible with other analysers. Each PROGENSA PCA3 assay kit is suitable for 2 x100 reactions and includes reagents, controls and calibrators for both the PCA3 and PSA reactions. The instructions for use document states that the PROGENSA PCA3 Assay should not be used for patients who are taking medications known to affect serum PSA levels such as finasteride (Proscar, Propecia), dutasteride (Avodart), and anti-androgen therapy (Lupron). The effect of these medications on PCA3 gene expression has not yet been evaluated. Certain therapeutic and diagnostic procedures such as prostatectomy, radiation, prostate biopsy, and others may affect the viability of prostatic tissue and subsequently impact the PCA3 Score. The effect of these procedures on assay performance has not yet been evaluated. Samples for PCA3 testing should be collected when the clinician believes prostate tissue has recovered. 2.2 The is manufactured by Beckman Coulter. It is an In Vitro Diagnostic Multivariate Index Assay which uses a calculation that combines the results of three quantitative blood serum immunoassays (Access Hybritech PSA, fpsa and p2psa) for different types of PSA (total PSA, free PSA and p2psa) into a single numerical result, PHI. To calculate Issue date: November 2014 Page 12 of 78

PHI the laboratory must run the 3 separate assays total PSA; free PSA and p2psa. They can be run from the same patient blood sample with no special handling or preparation of the sample and therefore can be conducted in a routine blood sciences laboratory using Beckman Coulter analysers that have the PHI algorithm incorporated in the software. Beckman Coulter validated the PHI to perform equivalently with the traditional Beckman Coulter Hybritech calibration and the Beckman Coulter WHO calibration for both the Access Hybritech PSA and free PSA assays. Separate, but clinically equivalent, reporting criteria are available for both assay calibrations. The Beckman Coulter PHI is not intended to be calculated using PSA, or free PSA results, from any other company's assay and the PHI assay is only compatible with Beckman Coulter Access instruments (Access2, DxI600, DxI800, DxC600i, DxC680i, DxC800i, DxC880i). The Prostate Health Index is calculated using the equation, ([ 2]proPSA/free prostate specific antigen) total PSA PHI is designed for prostate cancer detection in men aged 50 years and older with total PSA between 2 to 10 ng/ml and with digital rectal examination findings that are not suspicious for cancer. PHI can be used to categorize patients into low, moderate and high probabilities of prostate cancer being found on biopsy. A score of 0 20.9 indicates low risk of cancer (8.4% risk of cancer), 21 39.9 indicates moderate risk (21%) and 40 + indicates high risk of cancer (44%). It should be noted that these figures are from tests using Hybritech calibration. For PHI using WHO calibration, the equivalent cancer risk figures are a score of 0 20.9 (8.7% risk), 21 39.9 (20.6%) and 40 + (43.8%) The company claims that PHI demonstrates more than three times the specificity at 90% clinical sensitivity than PSA alone. In the NHS, it is likely that a second biopsy would be carried out for people with a PHI score of 21 Issue date: November 2014 Page 13 of 78

and above, that is, those classified as at moderate and high risk of cancer. For people with a PHI score of less than 21, indicating a low risk of cancer, it is likely they would be monitored over time by PSA testing rather than having a second biopsy, although this decision is dependent on other risk factors. Information provided by the company states that the effect of medication for benign prostate hyperplasia (BPH), and specifically the 5 alpha reductase inhibitors (5ARI) on the level of [-2]proPSA is not known. As consequence, the PHI results cannot be interpreted in patients under 5ARI medication. PHI should not be offered to men under 5ARI treatment. In addition they have stated that stability studies have demonstrated that the [- 2]proPSA molecule is not stable on coagulated blood. When left on clotted sample at room temperature the [-2]proPSA concentration increases significantly after 3 hours, probably because of the degradation of other propsa molecules. However, the analyte is stable in serum at room temperature. Therefore it is important that the serum sample must be prepared (separated from the clot by centrifugation) within 3 hours of taking a blood sample. 2.3 The comparator The comparator used in this assessment is information used to decide if a repeat prostate biopsy should be conducted based on clinical judgement and previous findings such as age, prostate size, PSA levels, close male relative (brother or father) with prostate cancer, presence of ASAP or HGPIN or abnormal digital rectal examination. Information based on the above plus the results of a multi-parametric MRI. Issue date: November 2014 Page 14 of 78

3 The evidence This section summarises data from the diagnostics assessment report compiled by the External Assessment Group (EAG). 3.1 Effectiveness The External Assessment Group conducted three systematic reviews to assess the clinical effectiveness of the PCA3 assay and PHI in the diagnosis of prostate cancer: A review of the analytic validity (how well laboratory tests measure the substances they are intended to measure) of the intervention tests to assess how accurately the tests measure PCA3 score/p2psa level present in a sample. A review of the clinical validity (accuracy of the diagnostic tests) of comparator and intervention pathways to assess how the addition of the PCA3 score or PHI might contribute to the diagnosis of prostate cancer. A review of the clinical utility of the intervention test pathways to evaluate how the addition of the intervention tests might affect patient outcomes, including long-term outcomes such as mortality and morbidity from prostate cancer, and intermediate outcomes such as side effects from tests. Details of the systematic reviews can be found starting on page 39 of the diagnostics assessment report (DAR). Studies considered for inclusion were assessed against criteria developed for each systematic review. In total, 37 studies met the inclusion criteria and were included in this assessment: 6 studies reported the analytical validity and 31 studies reported the clinical validity of the tests. No studies reporting clinical utility of the tests Issue date: November 2014 Page 15 of 78

were identified. Critical appraisal of the studies was undertaken using the Tuetsch checklist or the QUADAS-2 tool. 3.1.1. Analytical validity Studies reporting analytical validity were reviewed if they met the following inclusion criteria (Table 3.1): Table 3.1: Inclusion criteria (Analytical validity) Patient population Intervention test All adult men PCA3 assay or p2psa or phi score Outcomes Measures of consistency and accuracy between, and within, laboratories such as coefficient of variation Sensitivity and specificity against external standard Assay robustness Test failure rate Study design All study designs including collaborative studies, external proficiency testing, peerreviewed repeatability studies, internal reports and manufacturer data In total, 6 studies reporting analytical validity were identified: 3 of these used the PCA3 assay and the remaining 3 studies used the p2psa assay. All 3 studies using the PCA3 assay and 1 study using the p2psa assay were carried out in the USA. The remaining 2 studies using the p2psa assay were conducted in Germany. Company information and FDA data (Summary of Safety and Effectiveness Data (SSED) report) on the analytical validity of each test were also reviewed by the External Assessment Group. Full details of the quality of the studies can be found starting on page 43 of the diagnostics assessment report. PCA3 Assay: summary of results Full details of the results on the analytical validity review of the PCA3 assay can be found on page 44 of the diagnostics assessment report. One study, Sokoll (2008) showed that in 179 patients, 74.4% of urine samples taken before a DRE were informative compared to 95.5% of urine samples Issue date: November 2014 Page 16 of 78

that were taken after DRE. The collection method (number of strokes per lobe - 3 or 8) did not increase the information rate of tests. One study, the SSED report and company information reported the analytic sensitivity of the test and these are shown in table 3.2 below. The analytic specificity was also reported in the company information. Table 3.2: Analytic sensitivity of the PCA3 assay Sokoll 2008 SSED 2012 Methods LoB: 95 percentile of zero Calibrator LoD: lowest measureable concentration of controls LoQ: < 130% recovery and CV <35% 4 blank female urine and 4 female urine spiked to Calibrator 2 concentrations LoD=LoB + 1.65 SD LoB Copies/ml LoD Copies/ml LoQ Copies/ml PCA3 176 259 259 PSA 831 2338 2338 PCA3 90 239 239 PSA 254 3338 3338 Pack insert Diluted in-vitro transcripts. LoQ assessed according to CLSI EP17- A PCA3 NR 80 c/ml Calibrator 2~ 750 PSA NR 1438 c/ml Calibrator 2 ~ 7500 Accuracy The accuracy of the PCA3 assay was reported in 2 studies, the SSED report and company information. As no gold standard is available, accuracy was calculated by the percentage recovery of measured PCA3 or PSA RNA copies/ml compared to UV-determined copies/ml of female urine samples spiked with varying concentration of in-vitro transcripts or compared to control samples. Accuracy varied from 90% to 118% for PCA3 and from 85% to 121% for PSA (Table 3.3 ). Issue date: November 2014 Page 17 of 78

Table 3.3 Percent recovery: PCA3 assay Study Methods Minimum Maximum Sokoll 2008 3 controls. Tested in two sites PCA3 104.1% 110.8% SSED 2012 Pack insert Groskopf 2006 Precision 8 member panel of female urine spiked with in-vitro transcript 8 member panel of female urine spiked with in-vitro transcript PSA 93.2% 108.8% PCA3 90% 118% PSA 85% 121% PCA3 94% 108% PSA 111% 120% 3 controls PCA3 102% 109% PSA 94% 111% Two studies, the SSED report and company information reported withinlaboratory variation. One of these studies and the SSED report included data on within and between-laboratory variation. The within-laboratory total coefficient of variation (CV%) ranged from 4% to 27% for PCA3 and from 7% to 18.7% for PSA. In the SSED report and company information, PCA3 score varied from 12% to 28% (full details on page 48 of the diagnostics assessment report). The within and between-laboratory total CV% ranged from 5.9% to 17.2% for PCA3 and 10.1% and 19.3% for PSA (full details on page 48 of the diagnostics assessment report).only one study 48 reported within and between-laboratory total CV for PCA3 score; this ranged from 12.3% to 25.0%. Most variation appeared to occur within-laboratory; betweenlaboratory results contributed little additional variation. The External Assessment Group highlight that with a CV of 25%, a sample with a true score of 25 would give a result between 19 and 31 in two thirds of tests, with a third of tests lying outside this range. The interpretation of scores near the cut-off point should therefore be interpreted with caution. Issue date: November 2014 Page 18 of 78

p2psa / PHI assay: summary of results Full details of the results on the analytical validity review of the PCA3 assay can be found on page 49 of the diagnostics assessment report. Two studies and the SSED report reported the analytic sensitivity of the test and these are shown in table 3.4 below. The analytic specificity was also reported in the company information and the SSED report. Table 3.4 : Analytic sensitivity of the p2psa / PHI assay Methods SSED LoB: 95 percentile of zero analyte 2012 56 LoD: LOB+ 1.65 SD [SD from patient serum LoQ- dilutions of calibrators from LoD to 7x LoD] LoQ: concentration with CV20% from quadratic model Sokoll Methods as for SSED. Appears to be 2012 72 same study results as in SSED Stephan LoD: repeat measurement of zero 2009 73 calibrator + 2SD p2psa LoB LoD LoQ pg/ml pg/ml pg/ml 0.50 0.69 3.23 0.50 0.70 3.23 Not reported 2.27 Not reported Accuracy The accuracy of the p2psa / PHI assay was reported in 2 studies and the SSED report. As no gold standard is available, accuracy was calculated by the percentage recovery of measured p2psa pg/ml in male serum samples spiked with varying concentration of purified p2psa (Table 3.5). The reference material used was based on purified p2psa. Issue date: November 2014 Page 19 of 78

Table 3.5: Percent recovery: p2psa assay 6 spiked patient serum p2psa Methods Minimum Maximum Other Stephan 93% 103% Regression 2009 73 samples r=0.959-1.055 SSED 6 spiked patient serum 2012 56 samples Sokoll 6 spiked patient serum 2012 72 samples 90% 96% Mean=93% 90% 96% Precision Two studies and the SSED report reported within-laboratory variation. The SSED report included data on within and between-laboratory variation. Both studies and the SSED report reported CV% for the p2psa assay but only the SSED report included CV% for the PHI assay. The within-laboratory total coefficient of variation (CV%) ranged from 3% to 12% for the p2psa assay and from 8.5% to 12% for the PHI assay. The within- and between-laboratory total CV% ranged from 5.4% to 9.4% for the p2psa assay and from 4.9% and 7.3% for the PHI assay. (Full details can be found on page 52 of the diagnostics assessment report). 3.1.2. validity Full details of the clinical validity review can be found starting on page 55 of the diagnostics assessment report. Studies reporting clinical validity were reviewed if they met the following inclusion criteria (Table 3.6): Issue date: November 2014 Page 20 of 78

Table 3.6: Inclusion criteria (clinical validity) Patient population Intervention Comparator Reference standard Outcomes Study design Men suspected of having prostate cancer who had had at least one negative or equivocal biopsy. The review was restricted to studies where at least six cores were taken in initial biopsy. Studies of men taking medications known to affect serum PSA levels such as finasteride (Proscar, Propecia), dutasteride (Avodart), and anti-androgen therapy (Lupron) were excluded Diagnostic test or test pathway including PCA3 and/or phi Diagnostic test or test pathway without PCA3 or phi and including one or more of following comparator tests: individual clinical risk factors such as age or DRE standard clinical care/nomograms PSA levels MRI results: T2-MRI/ DW-MRI. Studies that directly compared the performance of PCA3 with that of phi, with or without other comparators, were also included Eligible studies compared the performance of comparator or intervention pathways to a histological analysis of prostatic tissue. This could have been obtained from a second prostatic biopsy or from a prostatectomy specimen. Biopsy must have taken place within 1 year of the intervention test Studies with all types of second biopsy were included: repeat standard TRUS biopsy saturation template MRI targeted biopsies use of prostatectomy specimens. Studies reporting any of the following were included: estimates of the intervention or comparator test (means and standard deviation (SD), proportion positive) in men with positive and negative results on second biopsy specificity and sensitivity for different threshold points of PCA3, phi or PSA comparison of area under the curve (AUC) for different tests or test combinations gain in sensitivity and specificity estimates by adding intervention test as derived from ROC curves results of logistic regression analyses test failure rate adverse effects of test or subsequent biopsies risk group and stage of cancers detected. Studies reporting within-study comparison of interventions/comparators: paired design. Cross-sectional or longitudinal studies in which intervention test(s), comparator test(s) and reference standard test were performed in the same group of people unpaired design. Trials in which people were randomised to either the intervention or comparator test(s) and then all received the reference standard test. Studies for inclusion in between-study comparisons of interventions/ comparators: systematic reviews with meta-analyses of the clinical validity of the intervention or any of the comparator tests Issue date: November 2014 Page 21 of 78

The use of the PCA3 assay and PHI was considered in 3 possible diagnostic pathways: PCA3 score/phi alongside established risk factors (including histopathology results, PSA level and DRE) to inform the decision to perform a second biopsy PCA3 score/phi alongside established risk factors (including histopathology results, PSA level and DRE) to inform the decision to perform a mpmri scan before second biopsy. If the mpmri is positive a second biopsy would be performed PCA3 score/ alongside established risk factors (including histopathology results, PSA level and DRE) to inform the decision to perform a second biopsy in men who have had a negative mpmri scan. Comparisons between the performance of the intervention tests (PCA3 assay and PHI) and the comparators (clinical assessment and MRI) were made using either data from studies carried out in the same study population (withinstudy or direct comparisons) or from data from studies where intervention and comparator tests are carried out in different populations (between-study or indirect comparisons). Given the uncertainty in the diagnostic pathways used in NHS practice and because MRI is not always available, the External Assessment Group included all studies with a direct comparison of the PCA3 assay or PHI with one or more of the comparators listed in the inclusion criteria. 25 publications met the inclusion criteria for the within-study comparisons. Twenty one of these publications reported within-study comparisons between clinical assessment and PCA3, and/or clinical assessment and PHI versus a comparator. The remaining 4 publications reported univariate assessments of Issue date: November 2014 Page 22 of 78

the PCA3 assay or PHI compared to univariate PSA which only provided limited data (see appendix 6 of the diagnostics assessment report). Of the 21 publications which reported within-study comparisons (clinical assessment and intervention versus a comparator), 17 studies, which reported data from15 different study populations, were included in the review; results from two study populations (European Cohort, and REDUCE) were published in two publications each. The remaining 4 publications included 3 reporting data from the European cohort study and one study by Pepe et al. (2013). These 4 publications did not present additional study results but are included in the number of eligible studies (n=21) for information. Fourteen studies were observational cohort studies and one was a randomised controlled trial. Eleven of these studies were prospective cohorts, 3 were retrospective cohorts, and 1 study was of mixed design. Six papers reporting five systematic reviews and meta-analyses were identified which considered between-study comparisons and met the inclusion criteria. None of these reviews consider clinically relevant comparisons. Critical appraisal of the identified studies was undertaken using the QUADAS- 2 tool. Full details of the critical appraisal can be found starting on page 70 of the diagnostics assessment report. In summary, the main potential sources of bias in the included studies related to patient selection and a lack of reported details on the intervention tests, comparators and biopsies. None of the studies were based in the UK. No meta-analyses were carried out because of the heterogeneity in the included studies. Data available from the included studies could only address the first diagnostic pathway because the results of the tests were most often presented as outputs from logistic regression models. It is therefore not possible to determine from the data available whether diagnostic accuracy is improved if the PCA3 assay or PHI test is carried out before or after an MRI scan. Issue date: November 2014 Page 23 of 78

Results were most frequently reported using multivariate logistic regression models using AUC statistics, ROC curves, multivariate odds ratios and derived sensitivity and specificity values. Only one study presented independent sensitivity and specificity estimates. Six studies reported results using decision curve analysis. Decision curve analysis calculates the net benefit of a diagnostic model by subtracting the harm of unnecessary biopsies from the benefit of diagnosed cases of prostate cancer. Unlike the conventional trade-off between sensitivity and specificity, in decision curve analysis there is an attempt to weight the relative harms and benefits using the threshold probability of cancer at which the patient or clinician will opt for a biopsy. Further details describing this analysis method can be found in appendix 1 of the diagnostics assessment report. Seven studies reported diagnostic accuracy results for the PCA3 assay for the detection of more aggressive cancers usually based on a Gleason score of 7 or greater. In 6 studies the authors employed univariate analyses and showed the ability of PCA3 score to predict a Gleason score of 7 or greater. One study reported how the use of PCA3 score in combination with clinical assessment contributed to the prediction of more aggressive cancers. One other study considered the relationship between PHI and the Gleason score. The External Assessment Group considered the 4 most clinically relevant comparisons for health professionals in the NHS to be: clinical assessment vs clinical PCA3 (Comparison 1) clinical assessment vs clinical PHI (Comparison 2) clinical MRI vs clinical MRI + PCA3 (Comparison 3) clinical MRI vs clinical MRI + PHI. (Comparison 4) Issue date: November 2014 Page 24 of 78

These are discussed in greater detail below (Tables 3.7 to 3.10). Comparison 1: assessment vs clinical PCA3 Table 3.7 assessment vs clinical PCA3 Area under the curve Study assessment PCA3 Variables included Result Threshold Result Difference & p value if given European cohort (Haese 2008)44# European cohort (Ankerst 2008) 111# REDUCE placebo (Aubin 2010)84 Age, prostate volume, DRE, tpsa, %fpsa PCPT nomogram: family history, number of previous biopsies, DRE, PSA Age, family history, prostate volume, PSA, %fpsa Scattoni Age, DRE, prostate volume, 2013 100 tpsa, fpsa 0.67 Continuous 0.71 +0.04 P<0.001 0.65 (0.59-0.71) 0.72 (0.68-0.76) 0.75 (0.64-0.87) Busetto Age, DRE, PSA 0.55 2013 88 (0.46-0.64) Porpiglia Age, DRE 0.62 2014 97 (0.53-0.72) Gittelman Age, family history, race, 2013 43 number of previous biopsies, DRE REDUCE placebo (Aubin 2010)84 Age, family history, prostate volume, PSA, %fpsa Continuous 0.70 (0.64-0.75) Continuous 0.75 (0.71-0.79) Continuous 0.76 (0.64-0.88) Continuous 0.74 (0.66-0.82) Continuous 0.69 (0.60-0.78) +0.04 P<0.05 +0.04 p=0.0009 + 0.01 p=0.719 +0.19 p=0.0002 + 0.06 0.65 25 0.74 +0.09 (0.04-0.14) p=0.0007 0.72 (0.68-0.76) 35 0.74 (0.70-0.78) Goode Unclear 0.61 2013 89 Age, DRE, prostate volume, race, family history Perdona Continuous 0.74 2011 95 PCPT nomogram: age, race, PSA, family history, DRE, previous biopsies (0.63-0.83) Perdona Continuous 0.74 2011 95 Chun nomogram: age, PSA, DRE, previous biopsies, (0.64-0.83) prostate volume Multivariate odds ratio for PCA3 Study + PCA3 Variables included Threshold Result REDUCE placebo (Aubin 2010)84 Age, family history, prostate volume, tpsa, %fpsa Continuou s 1.02 (1.01-1.02) Wu 2012 105 DRE, TRUS, PSA, PSA Continuou s 1.02 (1.00-1.03) Gittelman 2013 43 Age, family history, race, number of previous biopsies, DRE, tpsa 25 4.56 (2.65-7.84) Porpiglia 2014 97 Age, DRE Unclear 3.88 (1.27-12.95) REDUCE placebo (Aubin 2010)84 Age, family history, prostate volume, tpsa, %fpsa 35 2.65 (1.86-3.79) Bollito 2012 87 Age, PSA, %fpsa 39 9.44 +0.02 p=0.0558 Issue date: November 2014 Page 25 of 78

(5.15-17.31) Bollito 2012 87 Age, PSA, %fpsa 50 9.29 (5.11-16.89) Sensitivity and specificity + PCA3 Variables Sensitivity Specificity Threshold Sensitivity Specificity REDUCE placebo (Tombal 2013)104 REDUCE placebo (Tombal 2013)104 Pepe 2013 94 Pepe 2013 94 included Best clinical judgement (life expectancy, DRE, prior biopsy, prostate volume and PSA): all cancers Best clinical judgement: Gleason score >=7 75% (68-81) 75% (61-85) 26% (23-30) 26% (23-29) Grouped: <20,20-34,35-50,>50 Grouped: <20,20-34,35-50,>50 66% (58-72) 85% (73-93) 71% (67-74) 67% (64-70) Derived sensitivity and specificity at various risk thresholds + PCA3 Sensitivity Specificity Threshold Sensitivity Specificity PCPT nomogram: 100% 1% PCPT + 100% 8% age, race, family continuous history, PSA, PCA3 DRE, prior biopsy 25% risk 25% risk threshold threshold PCPT nomogram: age, race, family history, PSA, DRE, prior biopsy 40% risk threshold 75% 26% PCPT + continuous PCA3 40% risk threshold 85.8% 25% Derived sensitivity for various set specificity levels Study assessment PCA3 Variables included Result Threshold Result Difference & p value if given 80% specificity European Cohort (Ankerst 2008)83 PCPT nomogram: age, family history, number of previous biopsies, DRE, PSA 43.9% Continuous 46.3% +2.4% Porpiglia Age, DRE 48.0% Continuous 38.5% -9.5% 2014 97 90% specificity European Cohort (Ankerst 2008)83 PCPT nomogram: age, family history, number of previous biopsies, DRE, PSA 24.4% Continuous 28.5% +4.1% Porpiglia Age, DRE 23.0% Continuous 26.9% +3.9% 2014 97 95 % specificity European Cohort (Ankerst 2008)83 Age, family history, number of previous biopsies, DRE, PSA 11.4% Continuous 17.1% +5.7% Porpiglia Age, DRE 17.3% 32.5 19.2% +1.9% 2014 97 Issue date: November 2014 Page 26 of 78

Derived specificity for various set sensitivity levels Study assessment PCA3 Variables included Result Threshold Result Difference & p value if given 80% sensitivity Scattoni 49% Continuous 47% -2% 2013 100 Age, DRE, tpsa, f/tpsa, prostate volume Porpiglia Age, DRE 27.1% Continuous 37.3% +10.2% 2014 97 90% sensitivity Scattoni Age, DRE, tpsa, f/tpsa, 35% Continuous 25% -10% 2013 100 prostate volume Gittelman 2013 43 Age, family history, race, number of previous biopsies, DRE 18.9% (10.3-36.9) 25 41.5% (32.5-49.9) +22.6% (90% CI 9.0-33.1) Porpiglia Age, DRE 12.7% Continuous 11.0% -1.7% 2014 97 95% sensitivity Porpiglia 2014 97 Age, DRE 0.8% Continuous 8.5% +7.7% Area under the Curve (AUC) Eight AUC results were reported from six study populations for the comparison of clinical assessment versus clinical PCA3; one study reported the results from two models, 1 using PCA3 score as a continuous variable and 1 employing a threshold value of 35. Results from the same study population were reported in two separate papers. The studies showed an increase in discrimination of between 1% and 19% when the PCA3 score was added to the clinical assessment model, either as continuous or binary variable. In addition, 2 studies reported AUC results only for models of clinical PCA3 and these results were similar to the AUC results reported in other studies: Goode et al. (2013) reported an AUC of 0.61 for a multivariate logistic regression model; and Perdona et al. (2011) reported an AUC of 0.74 for the Chun nomogram and an AUC of 0.74 for the PCPT nomogram. Issue date: November 2014 Page 27 of 78

Multivariate odds ratio for PCA3 Five studies reported seven multivariate ORs for clinical PCA3. Four studies presented statistically significant results (ORs were above 1 and confidence intervals did not include 1). One study had an OR above 1 with a confidence interval that included 1. Haese et al. (2008) reported that the multivariate odds ratio for PCA3 score was significant (p=0.006) in the model but did not report the effect size. These results are consistent with the AUC results and indicate that the addition of the PCA3 score to the clinical assessment model increases discrimination. Two studies reported ORs for PCA3 using the PCA3 score as a continuous variable; in the remaining studies various different thresholds were used to divide PCA3 scores into a dichotomous variable. Sensitivity and specificity One study presented independent sensitivity and specificity estimates. In this study the addition of PCA3 scores to best clinical judgment reduced sensitivity from 75% to 66% and increased specificity from 26% to 71%. In this population (prevalence of all cancers=17.9%) adding PCA3 score to clinical assessment meant that 18 cancers would have been missed and 371 biopsies would have been avoided compared with clinical assessment alone. However, when the analyses were repeated for cancers with a Gleason score of 7 or above (prevalence=5.4%), the addition of PCA3 score increased sensitivity from 75% to 85% and specificity from 26% to 67% meaning that six more cancers would have been detected and 395 biopsies would have been avoided compared with clinical assessment alone. Derived sensitivity and specificity Pepe et al. (2013) reported sensitivity and specificity for various risk thresholds in the logistic regression model. At a 25% risk threshold, models of the PCPT nomogram alone and of the PCPT + PCA3 had 100% sensitivity Issue date: November 2014 Page 28 of 78