Role of magnetic resonance imaging in defining a biopsy strategy for detection of prostate cancer

Authors

  • Ashley J Ridout,

    Corresponding author
    1. Division of Surgical and Interventional Sciences, University College London, London, UK
    2. Department of Urology, University College London Hospitals NHS Foundation Trust, London, UK
    • Correspondence: Ashley J Ridout B.M.B.Ch., M.A.(Oxon), M.R.C.S., Division of Surgical and Interventional Sciences, University College London, 3rd Floor, Charles Bell House, 67-73 Riding House Street, London W1W 7EJ, UK. Email: ashley.ridout@doctors.org.uk

    Search for more papers by this author
  • Veeru Kasivisvanathan,

    1. Division of Surgical and Interventional Sciences, University College London, London, UK
    2. Department of Urology, University College London Hospitals NHS Foundation Trust, London, UK
    Search for more papers by this author
  • Mark Emberton,

    1. Division of Surgical and Interventional Sciences, University College London, London, UK
    2. Department of Urology, University College London Hospitals NHS Foundation Trust, London, UK
    Search for more papers by this author
  • Caroline M Moore

    1. Division of Surgical and Interventional Sciences, University College London, London, UK
    2. Department of Urology, University College London Hospitals NHS Foundation Trust, London, UK
    Search for more papers by this author

Abstract

Prostate cancer is the second most common male cancer worldwide. It has a broad spectrum, from low-risk, clinically indolent disease, to high-risk aggressive cancer. This variety conveys certain diagnostic and management challenges. The use of prostate-specific antigen as a screening test for prostate cancer is increasing the diagnosis of low-grade, low-volume disease. By targeting biopsies towards suspicious areas on multiparametric magnetic resonance imaging, we can accurately diagnose clinically significant prostate cancer, reducing identification of low-risk, clinically indolent disease. This could avoid the radical treatment of histopathological cancer that might never have become clinically apparent. In the present review, we consider the use of multiparametric magnetic resonance imaging to inform the biopsy strategy. By identification of suspicious lesions on multiparametric magnetic resonance imaging, biopsy targets can be identified, and the sampling bias associated with blind standard transrectal prostate biopsy can be reduced. We consider the reliability of these radiological lesions for detection of clinically significant prostate cancer, and the methods of targeting them to ensure the radiological lesion is accurately sampled. Evidence suggests that targeted biopsy is efficient and accurate for diagnosis of clinically significant prostate cancer. By rationalizing diagnosis, and subsequently preventing overtreatment of clinically insignificant disease, magnetic resonance imaging-informed prostate biopsy can provide a method for streamlining the diagnostic pathway in prostate cancer.

Abbreviations & Acronyms
ADC

apparent diffusion coefficient

DCE

dynamic contrast enhancement

DRE

digital rectal examination

DWI

diffusion weighted imaging

mpMRI

multiparametric magnetic resonance imaging

MRI

magnetic resonance imaging

MRS

magnetic resonance spectroscopy

PSA

prostate-specific antigen

RP

radical prostatectomy

TRUS

transrectal ultrasound

TTMB

transperineal template mapping biopsy

Introduction

Prostate cancer is the second most commonly diagnosed cancer in men worldwide,[1] and the sixth cause of cancer death.[2] There is a broad spectrum of disease, ranging from low-risk, clinically indolent, to high-risk aggressive cancer. The implementation of both formal and informal screening programs has increased early diagnosis of prostate cancer,[3, 4] improving the opportunity for curative radical treatments. However, there has been an associated increase in the diagnosis of low-grade, low-volume disease,[5] and it is expected that by 2030 there will be 1.7 million new diagnoses and 499 000 new prostate cancer-related deaths.[6] As well as the psychological impact of a cancer diagnosis, patients could be exposed to the potential side-effects of radical treatment for disease that might never have become clinically apparent. By using imaging techniques to identify suspicious lesions within the prostate, and directing biopsy towards these areas, there is the potential to ensure the accurate diagnosis of clinically important prostate cancer, without the identification of clinically insignificant disease and the problems associated with it.

Diagnosis of prostate cancer

Prostate cancer diagnosis has traditionally included serum PSA measurement, DRE and prostate biopsy. The role of rectal examination has been questioned, especially with the improved sensitivity of imaging techniques, such as mpMRI. A recent retrospective review of patients diagnosed with prostate cancer by prostate biopsy in the USA showed that just 14% of those diagnosed with prostate cancer had an abnormal DRE, and half of these individuals had a normal age-specific PSA.[7] This reminds us that, for localized disease, clinical examination and PSA measurement cannot always be relied on. Histopathological specimens are required for diagnosis and grading of prostate cancer, so reinforcing the need for a more accurate method of identifying and targeting suspicious areas for biopsy.

Standard prostate biopsy

Considerable advances have been made since the six-core transrectal prostate biopsy[8] and its modification to include sampling of the anterior horns of the peripheral zone.[9] The inclusion of greater numbers of tissue cores increased diagnosis of prostate cancer in both biopsy naïve men,[10] and those who had previous negative biopsy, but were still considered to have a clinical suspicion of prostate cancer.[11, 12] Although the average number of biopsy cores taken at TRUS-guided prostate biopsy has increased over time, this technique still has considerable limitations, including the blind nature of the sampling. Up to 25% of cancers might lie in the anterior prostate, outside of the standard TRUS-guided sampling zone, and so tumors in this area are more likely to be missed.[13] The potential benefits of diagnosing prostate cancer must be weighed against the risks of the biopsy itself, and the relevance of obtaining a detailed diagnosis and accurate risk stratification on patient management. The risk of post-biopsy infection (occasionally resulting in severe sepsis) is increasing,[14] likely in part because of antibiotic resistance, and other side-effects might be bothersome, including hematuria, hematospermia, rectal bleeding and prostatitis.[15]

Disparity between biopsy findings and corresponding radical prostatectomy specimens is well reported. Approximately one in three cases of low-volume, low-risk disease are upgraded or upstaged on radical prostatectomy after initial standard TRUS biopsy.[16] Duffield et al. showed that, in men from their active surveillance cohort who went on to radical prostatectomy due to disease progression, all tumors over 1 cm3 in size at prostatectomy were anteriorly located. They now incorporate anterior/transition zone biopsy in the annual repeat biopsy for men on active surveillance in their cohort.[17]

Enhanced prostate biopsy strategies

Enhanced biopsy strategies might include taking an increased number of tissue cores at a greater density, aiming to provide a more accurate representation of the prostate as a whole, and sampling areas of the prostate that are outside of the range of standard TRUS biopsy. Targeted strategies could also be used, whereby suspicious areas are identified, and biopsy targeted to these areas. An enhanced biopsy strategy could be used for diagnosis in men considered at risk for prostate cancer despite previous negative biopsy, to exclude more significant disease in men diagnosed with low-volume, low-risk disease, or to more accurately characterize those who would benefit from radical treatment. Confirmatory extended or targeted biopsy can also more accurately determine those men who could be managed with an observational strategy, such as active surveillance. The ultrasound-guided transperineal biopsy approach was adapted from the grid template used for brachytherapy.[18, 19] Initial studies described increased yield of prostate cancer diagnosis in men with previously negative standard prostate biopsies. Currently, the transperineal approach incorporates a range of sampling densities including high-density 5-mm sampling and a 20-zone approach. Merrick et al. reported their cohort of 102 men who had transperineal template-guided saturation prostate biopsy[20] – 101 of these had had prior negative TRUS biopsy. With template biopsy, 42.2% of these men were diagnosed with prostate cancer, with 65.1% having Gleason 7 disease or greater.

It is well recognized that increased numbers of biopsy cores provides a better risk-stratification for patients diagnosed with prostate cancer. Epstein et al. reported in 2005 that, in their group of 103 men who were thought to have low risk disease on needle biopsy (defined as <50% involvement of each core, Gleason score <7 and <3 cores involved), 29% were upgraded after radical prostatectomy.[21] Using a 5-mm interval saturation biopsy technique (average 44 cores) on the RP specimens resulted in increased diagnosis of significant cancer compared with standard biopsy, although not all significant cancers identified at RP were identified with either biopsy technique. Barzell et al. studied 124 men initially stratified with favorable-risk disease on standard TRUS biopsy, who then underwent repeat confirmatory biopsy with a combined transrectal and template prostate mapping approach. A total of 8–22% men were reclassified to clinically significant disease with repeat TRUS biopsy (range dependent on the stringency of the definition used), versus 41–85% men with template-guided prostate mapping. Of note, repeat TRUS biopsy did not detect up to 80% of clinically important prostate cancers detected by the template-guided biopsy, which was considered the reference standard.[22] Taira et al. carried out transperineal TTMB on 64 men who were diagnosed with clinically insignificant prostate cancer on TRUS biopsy (according to Epstein criteria – Gleason Score ≤6, no core with >50% involvement, PSA density <0.15 and <3 cores involved).[23] Although initially stratified as a low-risk cohort, after TTMB 46 patients (71.9%) were diagnosed with clinically significant cancer, with 39.1% having Gleason 7 disease or higher.

There are some limitations associated with radical prostatectomy studies, including reporting bias (the reporting radiologist is usually aware the patient is due to undergo prostatectomy, or has a diagnosis of prostate cancer), and the differential slicing of prostate after RP compared with MRI reporting. In an attempt to overcome this bias, LeCornet et al. used a computer simulation strategy of radical cystoprostatectomy samples from patients undergoing surgery for bladder cancer.[24] They modeled different biopsy approaches for detection of prostate cancer according to two definitions (definition 1: Gleason score ≥7 and/or lesion volume ≥0.5 mL; and definition 2: Gleason score ≥7 and/or lesion volume ≥0.2 mL). They repeatedly simulated different biopsy strategies, with varying degrees of error, and report the area under the receiver operating characteristic curve for ruling out definition 1 prostate cancer as 0.69 and 0.75 (for 12-core TRUS biopsy with random localization error of 15 mm and 10 mm, respectively), 0.82 (14-core TRUS biopsy) and 0.91 (5 mm interval template prostate sampling). This reinforces the poor detection of clinically significant prostate cancer with standard TRUS biopsy, the increased detection with the addition of anterior cores, and the best detection of significant cancer with template mapping biopsy. However, the clinical side-effects of such dense sampling strategies (hematuria, acute urinary retention) drive the need for a more targeted biopsy approach, in order to reduce the overall number of tissue cores required, while maintaining significant prostate cancer detection.

MRI and prostate cancer

During the MRI scan, the patient lies within a magnetic field – the magnetic field is created with an external coil, with either a pelvic phased-array coil situated around the pelvis, or an endorectal coil. Application of the magnetic field causes alignment of hydrogen ions within the body, and when a radiofrequency pulse is applied, the hydrogen atoms are raised to a higher energy state. The ions return to their original energy state (“relax”) when the radiofrequency pulse is stopped, and energy is emitted, producing a current used to create an image. T1- and T2-weighted sequences are created by the time taken for hydrogen ions to relax in the longitudinal and axial axes, respectively.[25]

Anatomical MRI sequences

T1- and T2-weighted imaging has traditionally been used for local staging, to evaluate extracapsular extension and lymph node disease. Fat has a high signal (bright) on T1-weighted imaging, and water has a low signal (dark). The opposite is true with T2-weighted imaging, where water is bright and fat is dark. Peripheral zone tumors usually appear as areas of low signal intensity on T2 sequences; however, this appearance might also be seen in benign prostatic hyperplasia, prostatitis, prostate atrophy, hemorrhage, or after biopsy or treatment.[26] Identification of central zone tumors on T2-weighted imaging is more difficult, and there is considerable similarity in appearance between cancer and benign hyperplasia.[27] The demonstration of prostate cancer on T2-weighted imaging is volume dependent; Roethke et al. reported a detection rate of 89% for tumors greater than 2 cm, with no tumors less than 0.3 cm detected, and just 3% of tumors 0.3–0.5 cm detected.[28]

The use of MRI after prostate biopsy is limited by artefact from swelling and bleeding. Artefact appears as low signal areas on T2-weighted imaging, with similar appearance to the low signal, suggestive of prostate cancer. In one in three men, this can last for up to 1 year, and artefact persists on T2-weighted imaging after obvious haemorrhage has resolved on T1-weighted imaging.[25] This affects both local staging and identification of lesions suspicious for cancer on T2-weighted sequences – White et al. showed accuracy of local staging within 21 days of biopsy was 46%, compared with 83% after 21 days.[29] The concept of MRI before prostate biopsy was, in part, proposed to overcome the inaccuracies associated with post-biopsy artefact, and also as a “triage” test for those with raised PSA. If required after prostate biopsy, staging MRI does not need to be delayed, so ensuring treatment can be expedited and reducing the risk of disease progression because of treatment delay. To refine the diagnostic pathway, the potential for MRI to identify men with radiological suspicion of prostate cancer, and/or to provide a biopsy target, was proposed.[30]

Functional MRI sequences

mpMRI incorporates functional parameters – DCE, DWI, ADC and MRS (Table 1).

Table 1. Details of MRI sequences
MR sequenceDetailsRelevance to prostate cancer
T1 and T2 weighted

T1: Produced by the time taken for proton relaxation in the longitudinal axis.

T2: Produced by the time taken for proton relaxation in the axial axis.

T1: Fat and blood have a high signal, therefore appears bright, and water has a low signal (dark).

T2: Water bright and fat dark. Prostate cancer appears as areas of low signal (volume-dependent and not specific).

DCEImages rapidly acquired after administration of intravenous paramagnetic contrast producing “wash-in/wash-out” curvesUptake and release of contrast is more rapid in prostate cancer, due to the increased vasculature compared with surrounding tissues.
DWIReflects the differential movement of water within tissues, according to their tissue architecture; cell density, cell membrane integrity and presence of necrosis. Multiple “b values” (to express diffusion characteristics including gradient duration and amplitude of a pulse) can be used to assess diffusion characteristics, with creation of apparent diffusion coefficient maps, showing signal differences.Prostate cancer has restricted diffusion, appearing bright on longer b value sequences and dark on ADC map.
MRSPortrays different metabolite concentrations within discrete prostate volumes, presenting a graphical representation of relative choline, creatinine and citrate concentrations.Ratio of choline and creatinine : citrate is increased in prostate cancer.

There is discussion amongst experts regarding the use of mpMRI in routine clinical practice. Two recent consensus meetings have aimed to formalize this – first, a European consensus meeting of MRI for the detection, localization and characterization of prostate cancer;[31] and second, the START meeting – STAndards for Reporting of studies of MRI-Targeted biopsies.[32] A European MRI consensus meeting recommended that all functional sequences, with the exception of spectroscopy, should be used as the minimum for mpMRI conduct, and that a five-point ordinal scoring scale should be used to report the radiological likelihood of prostate cancer. No consensus was reached on the use of endorectal coil in standard practice, although it was recommended for optimal practice. The START group aimed to define and recommend MRI reporting standards for MRI-targeted biopsy studies, in order to facilitate a comparison between targeted and standard biopsy approaches, and allow data synthesis and meta-analysis of MRI-targeted biopsy. There is considerable variation in reporting of biopsy procedures, biopsy results and the detection of clinically significant cancer. It was agreed that Gleason grade and maximum cancer core length were the most appropriate histological parameters in studies of MRI-targeted biopsy, but consensus as to what constitutes low, intermediate or high risk was not reached. A novel definition for clinically significant disease, based on the MRI-targeted sampling strategy, is required, but this was not defined during the meeting. This is important, because application of standard risk stratification criteria could result in higher risk attributed with targeted biopsy results. This is reinforced by computer simulation studies that modeled systematic 12-core TRUS biopsy and transperineal targeted biopsies on whole-mount radical prostatectomy specimens.[33] That study showed an increased maximum cancer core length and increased number of positive cores with transperineal targeted biopsies (because of the inherent oversampling associated with targeting).

MRI-informed prostate biopsy

Identifying the target

If MRI is to be used as a guide for prostate biopsy, the accuracy for identification of suspicious areas and the likelihood of MRI lesions being positive for clinically significant disease must be confirmed. Functional techniques have expanded the use of MRI to include identification of suspicious lesions within the prostate. As with T2-weighted imaging, detection is volume dependent. Villers et al. compared DCE and T2-weighted MRI with whole mount radical prostatectomy specimens. They showed sensitivity, specificity, positive and negative predictive values were 90%, 88%, 77%, and 95% for tumor volumes of 0.5 cm3 or greater. When the detection threshold was reduced to 0.2 cm3, these values reduced to 77%, 91%, 86% and 85%.[34] There is also the potential for risk-stratification of prostate cancer – Villeirs et al. showed increased sensitivity for disease of Gleason ≥4 + 3 disease (92.7% vs 67.6%) using combined MRI and MRS.[35] However, the clinical implications of this could be limited, as primary Gleason pattern 4 might not be considered stringent enough; for example, for active surveillance strategies. It has been shown that, compared with radical prostatectomy specimens, ADC mapping inversely correlates with Gleason grade in peripheral zone tumors, with good discrimination between low-, intermediate- and high-grade tumors.[36] ADC has also been used to inform targeted biopsy in patients suitable for active surveillance, in order to further risk stratify patients with previously undiagnosed higher-risk disease.[37] By correlating suspicious areas on ADC with targeted biopsy results, it was concluded that the median ADC value can identify prostate cancer and predict grade with significant accuracy. Other functional sequences have been implicated in MRI-based risk-stratification – Rastinehad et al. reported 101 patients who underwent 3T mpMRI with identification of suspicious lesions, with subsequent standard 12-core TRUS biopsy and real-time MRI/ultrasound fusion-targeted biopsies of suspicious areas.[38] Using their own scoring system, based on the number of positive MRI modalities, they concluded that there was a correlation between MRI-based suspicion of prostate cancer and D'Amico risk-stratification score. Figure 1 shows diffusion-weighted, ADC and T2-weighted images of the left peripheral zone lesion, corresponding to Gleason 3 + 4 disease, with maximum cancer length 7 mm.

Figure 1.

(a) Diffusion-weighted, (b) ADC map and (c) T2-weighted images showing left peripheral zone lesion (scored as 5/5 on a 5-point ordinal scale) between mid-gland level and apex. The remainder of the gland scored as low probability for significant disease (2/5). Corresponding histopathology from targeted biopsy to this lesion showed Gleason 3 + 4 disease, maximum cancer length 7 mm (60%).

Haffner et al. analyzed 555 patients with clinical suspicion of prostate cancer, who had had DCE-MRI before biopsy taking 10–12 TRUS-guided extended systematic biopsies and two targeted biopsies to suspicious areas on MRI. A total of 63% of men had suspicious MRI, and targeted biopsy identified 16% more Gleason grade 4 and 5 cancers than extended systematic biopsy. Significant prostate cancer was defined as total cancer core length ≥5 mm or Gleason ≥4 disease, and they report sensitivity, specificity and accuracy of targeted biopsies as 0.95, 1.0 and 0.98, respectively.[13] Lee et al. prospectively studied 87 men with increasing PSA despite negative prostate biopsies.[39] 94.2% of these patients had suspicious lesions identified on MRI, with 56% of this group having targeted biopsy positive for prostate cancer. MRI-targeted biopsy identified a large proportion of lesions in these men which were located outside of the standard TRUS biopsy sampling area (anterior/apical lesions), and detection of prostate cancer was more efficient using targeted cores than standard cores (28.8% vs 3.6%). Sciarra et al. report on their cohort of patients with persistently raised PSA and previous negative biopsy, who underwent MR spectroscopy and DCE MRI for identification of suspicious lesions.[40] For combined MRS and DCE they report sensitivity 92.6%, specificity 88.8%, positive predictive value 88.7%, negative predictive value 92.7% and accuracy 90.7% for prediction of prostate cancer detection. Hambrock et al. used 3T MRI with DCE and DWI-MRI (mixed endorectal coil and pelvic phased array coil) to identify target lesions in 71 men with at least 2 negative TRUS biopsies and PSA >4 ng/mL.[41] A single radiologist used a MR compatible biopsy device to obtain targeted biopsies from suspicious lesions. Their cancer detection rate was 59%, with a median average of 4 cores, and clinically significant cancer detection rate of 93% (defined as primary Gleason score ≥4, stage pT3 or tumor volume >0.5 mL in those who subsequently underwent radical prostatectomy, and, for those who did not go on to prostatectomy, clinical significance was defined as prebiopsy PSA >10 ng/mL, PSA density >0.15 ng/mL/cc or primary Gleason score ≥4 at biopsy).

Methods of targeting

Once identified, characterization of a lesion relies on accurate biopsy targeting to that lesion. When considering small lesions, and criteria for clinically significant prostate cancer that incorporate maximum cancer core length, any associated targeting error could be significant. Targeting might either be carried out in the MRI scanner itself, or with transrectal ultrasound guidance. Working within the MR scanner requires MR-compatible equipment and an open magnet, with serial MR scans used to target the lesion identified on the diagnostic scan, but this technology is not available widely in all centers. Reports suggest that this can be achieved with the patient supine, and with minimal modifications to the MRI table itself.[42] Ultrasound-guided targeting is either visually registered, where the biopsy operator visualizes the target after reviewing its position on MRI, or software registered, where the pre-procedural MRI images highlighting the location of the tumor are overlaid onto the real-time TRUS images. Kasivisvanathan et al. compared transperineal template prostate biopsy with cognitive MRI-targeted biopsy in 182 men with a suspicious lesion on mpMRI.[43] This was a mixed group, with 40% having had previous positive TRUS biopsy, 18% previous negative biopsy and 43% no previous biopsy. Clinically significant cancer (defined as maximum cancer core length ≥4 mm, and/or Gleason grade ≥3 + 4) was detected by MRI-targeted biopsy in 57%, and in 62% with 20-sector modified-Barzell technique template-guided prostate biopsy. Furthermore, detection of clinically insignificant disease was reduced in the targeted biopsy group (9.3% for MRI-targeted biopsy vs 17% for template-guided biopsy). That study reported detection of clinically significant and insignificant disease according to specified criteria, rather than overall cancer detection rate, and reported that 7% of clinically insignificant prostate cancer would have avoided diagnosis if only targeted biopsy was used.

A number of software systems are under development and in use, using both transrectal and transperineal approaches.[44-46] Software-assisted registration may be either rigid (the MR image is directly superimposed over the ultrasound image without accounting for gland deformation or changes resulting from different acquisition methods) or deformable, which is technically more challenging. Alternatively, the MRI scan can be carried out with a MR-compatible probe identical to that used at transrectal ultrasound,[47] which aims to reduce the effect of deformation during the biopsy procedure.

Accuracy of targeting

Labanaris et al. reported 70 men who had a MRI-targeted biopsy and subsequently underwent radical prostatectomy.[48] Of this group, 90.1% patients demonstrated exact Gleason score matching, with 8.5% showing significant upgrading and one patient showing significant downgrading with radical prostatectomy histology. There were no preoperative clinical variables that significantly predicted histopathological upgrading at radical prostatectomy.

Focal higher-grade tumor volume as suggested by ADC could be lower than the total tumor volume. Van de Ven et al. used a simulated ultrasound biopsy system to target such lesions, and reported that a technical registration accuracy of 1.9 mm is required to identify the high-grade components of suspicious lesions on mpMRI.[49]

In agreement with the data reported by Kasivisvanathan et al., Haffner et al. compared MR-targeted biopsy with extended systematic biopsy, and reported that diagnosis of 13% of clinically insignificant cancers would have been avoided with targeted biopsy alone.[13]

Recent studies using software assisted MR-targeting biopsy techniques have reported 59%[45] and 61%[44] overall cancer detection, but these did not differentiate between significant and insignificant disease. Both groups showed a higher proportion of positive cancer cores using MRI-targeted biopsy than systematic non-targeted biopsy. Pinto et al. reported their experience and cancer detection rates with a MRI/ultrasound fusion-guided prostate biopsy platform.[50] Patients underwent both 12-core TRUS biopsy and MRI/ultrasound fusion biopsy. The fusion system detected more cancer per core than standard TRUS biopsy for patients with low, moderate and high suspicion of cancer on MRI; detection rates were 20.6% (MRI/US fusion biopsy) and 11.7% (standard TRUS biopsy) when low-, moderate- and high-suspicion lesions were combined, but the results were most marked for high-suspicion lesions, with cancer detection 53.8% for MRI/ultrasound biopsy compared with 29.9% for standard TRUS biopsy. Delongchamps et al. prospectively compared visual targeting and software-assisted targeting with both rigid and deformable systems.[51] They reported significantly increased cancer diagnosis with software-assisted image registration compared with random biopsies. There was an associated decreased diagnosis of clinically insignificant cancers. If the targeted biopsy strategy alone had been used, 45% would have avoided unnecessary prostate biopsy, with a reduction in tissue cores sampled in the remainder who did require biopsy.

In order to overcome the inherent errors associated with human sampling, work into automated systems is ongoing. Song et al. have developed a motorized needle guide template for MRI-guided transperineal biopsy.[52] In this preliminary evaluation, they reported an average in-plane targeting error of 0.94 mm (standard deviation ± 0.34 mm), showing good targeting accuracy, with insignificant image artefact. MRI-compatible robotic techniques for needle placement are under development, with promising results for targeting accuracy using phantom studies.[53, 54] Long et al. have recently reported their 3-D ultrasound robotic system, which, combined with intraoperative prostate tracking, is being used for TRUS-guided needle deployment.[55]

In order to show which method of targeting is the most accurate, prospective, multicenter trials are required, comparing the available targeting methods with systematic biopsy.

Can we rely on targeting alone?

The option exists to only offer biopsy to men with a suspicious lesion on MRI. A recent systematic review[56] reported that in a cohort of men without previous biopsy, 38% did not show any suspicious areas on MRI. Of these men, 23% had prostate cancer on standard biopsy, but just 2.3% of this group had clinically significant cancer (defined as cancer core length >5 mm and/or Gleason pattern disease >3), which would have been missed if a targeted biopsy strategy alone was used. In the cohort of men with non-suspicious MRI who had previous negative biopsy, the overall cancer detection rate was 15%, but the proportion of disease identified as clinically significant was not reported separately in the studies. Yerram et al. analyzed 800 consecutive patients who underwent 3T mpMRI with endorectal coil, and identified 125 men with only prostatic lesions of low suspicion for prostate cancer.[57] These men underwent TRUS/MRI fusion biopsy – 88% had either no cancer (62%) or low-risk disease that would qualify for active surveillance according to the 2011 National Comprehensive Cancer Network Guidelines. Importantly, no cases of high-risk disease were identified, and in a subgroup that subsequently underwent radical prostatectomy (15/125), none showed histopathological upgrading to high-risk prostate cancer. A retrospective single center case series, due to be presented in 2013, reported that in men on surveillance, those without a suspicious lesion on mpMRI were less likely to show radiological progression during their surveillance.[58]

Conclusion

The MRI-guided approach to prostate biopsy is efficient, requiring fewer prostate cores and reducing the clinical load associated with systematic biopsy strategies, with studies suggesting improved accuracy and yield of cancer detection. Through streamlining and increasing efficiency of biopsy techniques, we are also aiming to identify only those men with clinically important cancer. Clarification of the optimal method of MR-registration and appropriate biopsy conduct (including the optimum number of tissue cores required to accurately characterise the biopsy target) should improve this further. The associated reduced diagnosis of clinically insignificant prostate cancer will have clinical, psychological and economic impacts.

Overtreatment of clinically indolent disease might be avoided, and we can appropriately counsel our patients and increase confidence in their treatment choices. We should aim to develop a method of early identification of clinically significant disease requiring treatment, with less importance placed on diagnosis of indolent disease – this approach could eventually obviate the need for observational strategies. There are some economic implications of an enhanced, targeted-biopsy approach, which must be considered. These include use of anesthesia or sedation, operating theater costs and additional costs associated with prebiopsy MRI. These might be weighed against the potential reduction in treatment of clinically indolent disease, and will be reduced as optimal sampling density and tissue core requirement are defined.

Advances in imaging and biopsy technologies, study of the natural history of low- and intermediate-risk disease, and developments in biomarker and genetic technologies are required to inform and rationalize the diagnosis and treatment of prostate cancer. Good quality prospective studies are required to further understand the potential uses of MRI in prostate cancer diagnosis, monitoring and treatment.

Conflict of interest

Ashley J Ridout receives research funding from the Bob Champion Trust. Veeru Kasivisvanathan received research grant support from the National Institute of Health Research UK. Caroline M Moore is a consultant for Steba Biotech, receives research funding from GSK, and has received speaker fees from Janssen and Lily. She receives research funding from the Wellcome Trust, the Department of Health and the Prostate Cancer Charity. Mark Emberton is a consultant to GSK, Sanofi Aventis, Jensen, STEBA Biotech, AMD and USHIFU. He receives research support from GSK, Sanofi Aventis, STEBA Biotech, AMD and USHIFU. He is Medical Director of Mediwatch Plc.

Ancillary