Performance of multiparametric prostate magnetic resonance imaging validated by targeted and systematic transperineal biopsies

Abstract Objective To measure the performance of multiparametric (mp) magnetic resonance imaging (MRI) to identify intraprostatic tumour deposits using a systematic and targeted MR‐guided transperineal prostate biopsy technique. Materials and Methods Patients underwent a combined systematic and targeted MR‐guided transperineal biopsy procedure in the dorsal lithotomy position under general anaesthesia. Systematic biopsies were spaced 10 mm or less apart and additional biopsies targeted any Prostate Imaging–Reporting and Data System (PI‐RADS) 3, 4 or 5 lesions identified on mpMRI. Cancer detection rates were calculated on a per patient and per lesion basis. Results A total of 125 patients underwent the biopsy procedure. The positive predictive value (PPV) of mpMRI per patient was 59% for any cancer and 49% for Gleason score (GS) ≥ 7 cancer. The negative predictive value (NPV) of mpMRI per patient was 67% for any cancer and 88% for GS ≥ 7 cancer. On a per lesion basis, the PPV of PI‐RADS 3 lesions for any and GS ≥ 7 cancer was 24% and 10%. For PI‐RADS 4 lesions it was 42% and 32%. For PI‐RADS 5 lesions, it was 76% and 70%. MpMRI failed to identify GS ≥ 7 cancer found on systematic biopsy in 22% of patients. Conclusion Based on a combination of systematic and targeted transperineal prostate biopsies, mpMRI showed a high NPV and low PPV for GS ≥ 7 cancer on a per patient basis. The PPV of mpMRI on a per lesion basis increased with increasing PI‐RADS score. However, there were a significant number of both false positive as well as false negative (mpMRI invisible) areas within the prostate that contained GS ≥ 7 cancer. Therefore, pathologic confirmation using both targeted and systematic mapping biopsy is necessary to accurately identify all intraprostatic tumour deposits.


| INTRODUCTION
The introduction of multiparametric magnetic resonance imaging (mpMRI) scans of the prostate has dramatically changed the diagnostic pathway for prostate cancer over the last decade. Current NCCN (National Comprehensive Cancer Network) guidelines include consideration of mpMRI and targeted MR fusion prostate biopsy as acceptable options for the prostate cancer screening process (NCCN Early Detection Guidelines). In response to the widespread adoption of mpMRI into clinical practice and excessive variation in interpretation and reporting, a standardized reporting system was introduced in 2012 known as PI-RADS (Prostate Imaging-Reporting and Data System) version 1 and updated to PI-RADS version 2 in 2015. 1,2 Studies utilizing this system showed the use of mpMRI can, in general, lead to a reduction in overdiagnosis of clinically indolent prostate cancer and improved detection of clinically significant cancers. 3,4 Most of the reports on the performance of mpMRI, however, have focused on a per patient rather than per lesion basis in order to measure its diagnostic reliability. Extensive analyses of the sensitivity, specificity, positive and negative predictive values (PPV and NPV) to establish the presence or absence of prostate cancer in a given patient have been reported. 5 If the assumption is that patients diagnosed with prostate cancer who require treatment (i.e., those with clinically significant disease) will undergo whole gland therapy, the delineation of all tumours within a prostate is not critical. However, radical therapy such as surgical removal or radiation therapy can result in significant urinary, bowel and sexual dysfunction. As a result, there has been increased interest in focal prostate therapy in an effort to control tumour progression while avoiding or minimizing the toxicities associated with whole gland therapy. Over the past 15 years, reports have been published describing the use of focal therapy either as primary treatment, as an intraprostatic boost for whole gland therapy and as salvage treatment for locally recurrent prostate cancer. [6][7][8] Many of these studies utilized mpMRI to delineate treatment targets within the prostate. As current and future clinical trials continue to incorporate mpMRI into their treatment strategies, it will be important to take into consideration the abilities and limitations of these scans in identification of tumour within the prostate.
The purpose of this study was to evaluate the performance of prostate mpMRI on both a per patient and per lesion basis validated by targeted and systematic transperineal prostate biopsies.

| Patient population
In this study, the data from patients in a community practice setting undergoing mpMRI followed by MR fusion transperineal prostate biopsy from 2015 through 2021 were collected. Subjects undergoing both initial and repeat prostate biopsy after initial transrectal ultrasound guided (TRUS) biopsy were included. IRB approval was obtained to retrospectively analyse the records of each patient to obtain prebiopsy characteristics, mpMRI results, biopsy procedure details and pathology reports of the biopsy specimens. Patients were excluded if they could not receive gadolinium contrast or had undergone prior MRI fusion biopsy or saturation biopsy. Patients with hip replacement were also excluded due to extensive susceptibility artefact and image distortion on MRI, which could preclude reliable identification and delineation of PI-RADS lesions. In total, a final cohort of 125 patients met the study criteria.
All studies were interpreted by a single radiologist (MH) at the diagnostic radiology facility housing the MRI scanner. This physician had 25 years of clinical radiology experience prior to initiation of this procedure. MRI scans were read as part of routine clinical care using PI-RADS v2. No standardized training or formal performance feedback was given prior to or during the study period.

| Transperineal MRI-fusion prostate biopsy protocol
All biopsies were performed by a single physician (RAH). This physician is a radiation oncologist who is a brachytherapy specialist with 18 years of clinical experience prior to initiation of this procedure. All PI-RADS 3, 4 and 5 lesions were targeted based on results of the imaging from the above mpMRI protocol. Targets were confirmed by demarcation by the radiologist on captured key images available in the radiology facility PACS system or by personal review by RAH with the radiologist. Biopsy planning was performed by importing T2-weighted images into a commercially available transperineal biopsy planning system (Variseed/Varipath, Varian Medical Systems, Palo Alto, CA, USA). Images were then reoriented from supine (position mpMRI images were obtained) to dorsal lithotomy (position in which the transperineal biopsies were performed). The prostate was divided into an apical and base half and systematic biopsies were then planned 10 mm or less apart (from each other, the edge of the prostate or the urethra) for each half. Additional biopsies were also planned for each PI-RADS 3, 4 or 5 lesion. When no lesions were identified on mpMRI, only systematic biopsies were performed. Rigid fusion was used and confirmed using anatomic landmarks referenced to a transperineal template grid (see Figure 1). All biopsies were car- All patients received oral antibiotic prophylaxis for 3 days (the day prior, of and after the biopsy procedure) and a bowel prep consisting of clear liquids the day prior and Fleet ® enemas the night before and morning of the procedure. All patients had a urinary catheter placed during the procedure to identify the urethra.

| Statistical analysis
Descriptive statistics were used to present patient characteristics and cancer detection rates. We reviewed the cancer detections rates on a per patient and individual lesion basis. We determined the PPV and NPV of the PI-RADS v2 system in this series. The 95% confidence intervals (95% CI) were calculated for all proportional outcomes.  Table 1 The

| DISCUSSION
The utilization of mpMRI has expanded from diagnosis and screening to surgical and radiation therapy treatment planning and new treatment approaches such as focal and focal boost therapy. Despite widespread adoption of mpMRI, its statistical performance on a per patient and more so on a per lesion basis, remains somewhat variable.
This study, in summary, evaluated the performance of mpMRI on a per patient and per lesion basis in patients undergoing systematic and targeted transperineal prostate biopsy using the PI-RADS v2 system.

| Per patient analysis
We evaluated the positive and NPV of mpMRI in this study on a per patient basis to study its performance as a diagnostic tool. Large multi-institutional trials such as the PROMIS trial 9  Thus, the relatively high NPV and low PPV of mpMRI found in the current study seems to be consistent with other published series and confirms the conclusion that mpMRI is a relatively good screening tool to avoid biopsy, but improvements must still be made to more reliably and consistently predict the presence of clinically significant prostate cancer.

| Per lesion analysis
We evaluated the PPV of mpMRI in this study on a per lesion basis to study its performance in correct identification of intraprostatic tumour deposits. The PPV increased with increasing PI-RADS score

| mpMRI invisible prostate cancer
The effect of unidentified (mpMRI invisible) tumour has differing relevance depending upon whether the imaging is utilized for diagnosis or guidance for treatment decisions. In this current study, 22% of patients were found to have mpMRI invisible GS ≥ 7 disease. Other reports have found similar rates of mpMRI invisible tumour ranging from 10% to 28%. [18][19][20] However, in our series, only 6% of patients had mpMRI invisible GS ≥ 7 prostate cancer without other same or higher grade tumours identified elsewhere within the same prostate. Thus, from purely diagnostic standpoint, 94% of patients with clinically significant cancer somewhere within the prostate were identified by mpMRI. This high rate of detection would seem acceptable if mpMRI is used as a tool to help screen and perhaps direct prostate biopsy when trying to establish the diagnosis of GS ≥ 7 prostate cancer.
However, the presence of mpMRI invisible prostate cancer may have more significance when using mpMRI to guide focal therapy both as primary as well as intraprostatic boost treatment. The rationale for using focal therapy in an attempt to reduce toxicity of whole gland prostate treatment is based on the hypothesis that an index lesion (the largest tumour focus within the prostate) drives the natural history of an individual's prostate cancer and that anatomically distinct metastatic foci within a patient may share a monoclonal origin from a single progenitor found in that index lesion. 21 However, this hypothesis is far from proven and other data would suggest that non index lesions can be the origin of both local invasion with extracapsular spread and as well as metastatic disease. 22 higher in the focal boost cohort with no difference in prostate specific or overall survival. 7 Again, follow up is short, but it could be postulated that with histologic confirmation of the boost target, the outcomes might have been even better in favour of the boost arm.

| Strengths and limitations
The limitations of this study are first and foremost in its retrospective nature. Obvious referral bias could affect outcomes. This report consisted of patient populations that had either no prior prostate biopsy or prior negative biopsies. Although the PPV and NPV rates are similar to those reported in the literature, the differences between groups are notable. However, many reports as seen in Table 3 also include the same mixed group of patients. This variety of patients, however, also represents the reality of clinical practice. Finally, the definition of clinically significant prostate cancer (GS ≥ 7) is somewhat arbitrary.
Volume of the identified lesions was also not specifically taken into account, although the PI-RADS system does partially account for this issue as the main difference between PI-RADS 4 and 5 lesions is lesion size >1.5 cm. Also, the present analysis did not take into account the presence of high volume Gleason score 6 disease or patients with Gleason 6 disease with a very high PSA (e.g., >20). These situations may also represent 'clinically significant' prostate cancer.
The strengths of this study include the fact that all MRI scans were reviewed by a single radiologist (MH) and all procedures were carried out by a single radiation oncologist (RAH) in a community setting with results comparable to other published series both in terms of radiographic and pathologic correlation. However, each physician did have >18 years of clinical experience prior to initiation of this programme and the radiation oncologist is a specialist in prostate brachytherapy. Therefore, this clinical experience and expertise could also be considered a limitation as it may not be generalizable to other programs. Another strength of this series is that all patients underwent the same MRI protocol and all interpretations used the same PI-RADS v2 reporting system. In addition, the same biopsy planning software and equipment was used for all biopsy procedures. The reference standard was a systematic transperineal technique, which has been shown to be an effective sampling system, particularly of the anterior prostate when compared to a transrectal approach. 25 Although no whole mount prostatectomy specimens were available for correlation, it could be argued that using a reference with previously established cancer also introduces bias. Finally, the template guided system used in this series has the potential to be translated into a focal therapy plan. Although the mpMRI images may underestimate tumour volume, if systematic and target biopsies are utilized, a treatment target volume could be generated from the negative cores around the pathologically positive cores thus creating a pathologically confirmed margin around the target. 18

| CONCLUSION
Based on a combination of systematic and targeted transperineal prostate biopsies, mpMRI showed a high NPV but did miss a small (about 10%) number of patients with GS ≥ 7 prostate cancer. The PPV for GS ≥ 7 prostate cancer was relatively low on a per patient basis.
The PPV of mpMRI on a per lesion basis increased with increasing PI-RADS score and was consistent with other reports in the literature.
However, there were a significant number of both false positive as well as false negative (mpMRI invisible) areas within the prostate that contained GS ≥ 7 prostate cancer. Therefore, it would not be advisable to rely solely on mpMRI to target areas within the prostate and pathologic confirmation using both targeted and systematic mapping biopsy prior to focal therapy procedures should be performed.