The SmartTarget Biopsy Trial: A Prospective, Within-person Randomised, Blinded Trial Comparing the Accuracy of Visual-registration and Magnetic Resonance Imaging/Ultrasound Image-fusion Targeted Biopsies for Prostate Cancer Risk Stratification

Background Multiparametric magnetic resonance imaging (mpMRI)-targeted prostate biopsies can improve detection of clinically significant prostate cancer and decrease the overdetection of insignificant cancers. It is unknown whether visual-registration targeting is sufficient or augmentation with image-fusion software is needed. Objective To assess concordance between the two methods. Design, setting, and participants We conducted a blinded, within-person randomised, paired validating clinical trial. From 2014 to 2016, 141 men who had undergone a prior (positive or negative) transrectal ultrasound biopsy and had a discrete lesion on mpMRI (score 3–5) requiring targeted transperineal biopsy were enrolled at a UK academic hospital; 129 underwent both biopsy strategies and completed the study. Intervention The order of performing biopsies using visual registration and a computer-assisted MRI/ultrasound image-fusion system (SmartTarget) on each patient was randomised. The equipment was reset between biopsy strategies to mitigate incorporation bias. Outcome measurements and statistical analysis The proportion of clinically significant prostate cancer (primary outcome: Gleason pattern ≥3 + 4 = 7, maximum cancer core length ≥4 mm; secondary outcome: Gleason pattern ≥4 + 3 = 7, maximum cancer core length ≥6 mm) detected by each method was compared using McNemar's test of paired proportions. Results and limitations The two strategies combined detected 93 clinically significant prostate cancers (72% of the cohort). Each strategy detected 80/93 (86%) of these cancers; each strategy identified 13 cases missed by the other. Three patients experienced adverse events related to biopsy (urinary retention, urinary tract infection, nausea, and vomiting). No difference in urinary symptoms, erectile function, or quality of life between baseline and follow-up (median 10.5 wk) was observed. The key limitations were lack of parallel-group randomisation and a limit on the number of targeted cores. Conclusions Visual-registration and image-fusion targeting strategies combined had the highest detection rate for clinically significant cancers. Targeted prostate biopsy should be performed using both strategies together. Patient summary We compared two prostate cancer biopsy strategies: visual registration and image fusion. A combination of the two strategies found the most clinically important cancers and should be used together whenever targeted biopsy is being performed.


1.
Introduction Current management strategies for prostate cancer (PC) depend heavily on appropriate risk stratification, for which adequate tumour sampling and localisation are pivotal. The standard 12-core systematic transrectal ultrasound (TRUS)guided prostate biopsy has, however, led to overdiagnosis of indolent cancers in some patients and undersampling in others. Development of multiparametric magnetic resonance imaging (mpMRI) has improved diagnostic sensitivity for clinically significant disease while reducing overdetection of clinically insignificant cancer [1][2][3].
Several mpMRI-directed or targeted biopsy methods have been established to improve risk stratification: in-bore targeted biopsies, visual registration (also called cognitive registration, mentally translating mpMRI targets onto realtime ultrasound images), and software-based MRI/ultrasound image-fusion systems overlaying MRI targets onto real-time ultrasound images [3]. None has yet established superiority in a clinical setting. Whether visual-registration targeting is sufficient or whether it needs augmentation with image-fusion software has been debated [4]. Smart-Target Biopsy trial (ClinicalTrials.gov NCT02341677) was conducted to compare visual registration with image fusion using a validated [5] MRI/ultrasound fusion system developed in our institution (SmartTarget; technical details included in the Supplementary material).

2.
Patients and methods

Study design and participants
SmartTarget Biopsy was a prospective, blinded, within-person randomised, paired, validating clinical trial designed in accordance with the IDEAL Collaboration recommendations for medical device evaluation [6]. The

Intervention and follow-up
Before enrolment in the study, all men underwent mpMRI as standard of care in accordance with the British Society of Urogenital Radiology and European Society of Urogenital Radiology standards [7][8][9] in sequences as described in our previous studies [1]. These procedures were reported in both written and pictorial form by an experienced uroradiologist with access to clinical information. Each lesion was scored using a five-point   Table 1) [12]. Each patient completed established, validated instruments for the detection of change in genitourinary function, the International Prostate Symptom Score (IPSS) and International Index of Erectile Function-15 questions (IIEF-15) [13,14], and the EuroQol-5 Domains-5 Levels (EQ-5D-5L) questionnaire before the procedure, and at 1 and 6 wk after the procedure. Adverse events (AEs) were recorded until 6 wk after the procedure, and serious AEs (SAEs) were reported for 90 d after the procedure.

Outcomes
The prespecified primary outcome was the proportion of men with UCL definition 2 clinically significant disease [15] based on image-fusion biopsy versus that based on visual registration. The primary outcome was therefore Gleason pattern of !3 + 4 = 7 or cancer core length of !4 mm in any core. The prespecified secondary outcome was the proportion of men with UCL definition 1 clinically significant disease detected by image-fusion biopsy versus visual-registration biopsy, that is, Gleason pattern of !4 + 3 = 7 or cancer core length of !6 mm in any core. Questionnaires and AEs were used to evaluate the quality of life and safety related to performance of both targeting strategies.

Statistical analysis
Our null hypothesis was that the accuracy of these two biopsy strategies

Results
Between 14 November 2014 and 23 September 2016, 341 patients were screened for the study, 141 were enrolled, and 129 underwent both visual-registration and image-fusion biopsies and were analysed for the primary and secondary endpoints. Baseline and demographic characteristics are shown in Table 1. The reasons for randomisation without completion of both biopsies for the other 12 men are provided in Figure 1. UCL definition 2 PC was detected using both biopsy strategies combined in 93/129 (72%) men (Table 2). Each strategy detected 80/93 (86%; p = 1) of these significant cancers with an overall detection rate of 80/129 (62%). Each method identified 13 cancers that the other missed. The combination of the two methods resulted in a 14% (13/93 cases) improvement in the detection of clinically significant PC. Post hoc analysis of this difference showed it to be statistically significant (95% confidence interval: 7.6-22.5). UCL definition 1 PC was detected in 66/129 (51%) men. Of these 66 men, 52 (79%) were identified by visual-registra-[ ( F i g . _ 1 ) T D $ F I G ]  Table 3. No differences in patient age, prostate-specific antigen level, total cancer core length, or lesion volume between the concordant and discordant cases was apparent (Supplementary Table 2 and Supplementary Fig. 1-4). An increase in cumulative cancer detection was seen for each additional core taken (Table 4).
Safety findings after the two biopsy strategies were consistent with the safety profile associated with either strategy performed alone [12,16]. Three patients experienced AEs related to the biopsy procedure: two patients with events (urinary retention with catheterisation for 1 wk and urinary tract infection) that were mild in severity, and one patient with moderate nausea and vomiting. No SAEs were reported. No statistically significant difference in patient-reported outcome scores (IPSS, IIEF-15, and EQ-5D-5L) was seen between baseline and follow-up (median [interquartile range] of follow-up of 74 [49-105] d; Supplementary Table 3 and Supplementary Fig. 5-7). Individual variation was seen in IIEF-15 scores (overall and domain specific) and in health-related quality-of-life scores measured by EQ-5D-5L ( Supplementary Fig. 5-7).

Discussion
This study, which directly compared transperineal imagefusion and visual-registration biopsy strategies, found no statistically significant difference in overall detection rates of clinically significant PC. Both strategies missed clinically significant cancers detected by the other strategy and so should be used in combination to optimise cancer detection. The recent publication of the PROMIS trial [1] will increase demand from patients and policymakers to implement an mpMRI-based pathway given the degree of diagnostic superiority that was shown for this method compared with the standard of care. Moreover, the number of studies that have demonstrated increased detection of  clinically significant PC using mpMRI targeting (of whatever kind) compared with TRUS-guided biopsy continues to grow [3,4,17,18]. The recent PRECISION randomised controlled trial demonstrated the superiority of MRI-targeted biopsies over systematic biopsies [19]. Omitting the systematic template biopsy and performing only targeted biopsies may have maintained high diagnostic yield for clinically significant cancer in this trial, and reduced patient and healthcare resource burden. However, the optimal method for targeting biopsies has yet to be defined. The SmartTarget Biopsy study design had a number of important strengths. First, the paired cohort design (both strategies conducted in each patient) allowed both a comparison of detection rates between them and an evaluation of the benefits and risks of combining the two strategies. To minimise potential incorporation bias sources, we randomised the order of the two biopsy strategies for each patient and reset the equipment to a default setting before each biopsy strategy. A double-blind, parallel-group clinical trial would provide confirmation of detection rate similarity afforded for the two strategies, but a design that also assessed the additive value might be more challenging. Second, biopsy conducted by 14 urologists who were considered experienced by our senior assessors in visualregistration biopsy performance meant that the visualregistration strategy was as optimised as possible, providing a robust comparator to image fusion. However, this factor might have provided a comparison level that may, in fact, not represent performance of MRI-targeted biopsies elsewhere by urologists with less experience. Last, the careful prespecified sample size calculation and subsequent increase assured a sample size that minimised type II error, that is, with sufficient power to detect a true difference in PC detection rates between the two biopsy strategies.
Key limitation of this study included capping the biopsy sample number to three per strategy, which may have reduced detection rates for both visual registration and image fusion. This limit may also potentially confound an effect of increasing sample number with the apparent additive effect of the two strategies-the increase in detection rate for the combination of the two strategies. A parallel-group trial, which could maximise the number of needle deployments in an ethically acceptable fashion, could further distinguish the role played by each factor. Furthermore, evaluation of only transperineal biopsy may have limited applicability to transrectal biopsy.
Although our study was conducted in a different population and using different methods, our results are consistent with those of others. Wegelin and colleagues [3] published a systematic review comparing MRI in-bore targeting with a targeted biopsy utilising both visual registration and image fusion. They showed that each method had similar overall cancer detection rates. However, both MRI in bore and image fusion proved to be superior to visual registration for clinically significant cancer detection, although the confidence that we can attribute to the data was limited by the wide variability in detection rates. Among comparative studies performed in expert centres where skill-based biopsy strategies (visual registration) will be an optimal control, most reports suggest that a biopsy using some form of image registration will approximate expert performance [20,21]. Wysock et al. [21] demonstrated some benefit associated with an imagefusion system for anterior tumours, but that study was conducted using a transrectal approach, which might have made the sampling of these tumours that were furthest away from the needle deployment subject to some systematic error. In contrast, work by Lee and colleagues [22] found that the sampling of transition zone lesions (with the exception of basal lesions) yielded higher cancer detection rates when image-fusion software was used. This difference was attributed to limited registration contouring to the base or difficulty targeting the base on axial views. We found no baseline or imaging parameter that might explain the discordant cases in our study.
Cost is an important consideration but may vary widely depending on both the capital cost of the system and patient volume. A cost-benefit analysis is a complex question beyond this study's scope. However, our results suggest potential benefits of a faster learning curve and higher repeatability that may enable less experienced centres to increase throughput and achieve cancer detection rates equivalent to those of highly experienced centres.

Conclusions
Visual-registration and image-fusion targeting strategies combined had the highest detection rate for clinically collection, data analysis, data interpretation, or writing of the report.