- Email: [email protected]
Ultrasound Fusion-guided Prostate Biopsy
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
available at www.sciencedirect.com journal homepage: euoncology.europeanurology.com
The Learning Curve for Magnetic Resonance Imaging/Ultrasound Fusion-guided Prostate Biopsy Khushabu Kasabwala a, Neal Patel a, Eliza Cricco-Lizza a, Adrian A. Shimpi b, Stanley Weng a, Rose M. Buchmann b, Samaneh Motanagh c, Yiyuan Wu d, Samprit Banerjee d, Francesca Khani a,c, Daniel J. Margolis e, Brian D. Robinson a,c, Jim C. Hu a,* a
Department of Urology, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA;
Cornell University, Ithaca, NY, USA; d
c
b
Meinig School of Biomedical Engineering,
Department of Pathology, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA;
Department of Healthcare Policy and Research, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA;
e
Department of
Radiology, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA
Article info
Abstract
Article history: Accepted July 16, 2018
Background: Magnetic resonance imaging/ultrasound-guided fusion biopsy (FBx) is more accurate at detecting clinically significant prostate cancer than conventional transrectal ultrasound-guided systematic biopsy. However, learning curves for attaining accuracy may limit the generalizability of published outcomes. Objective: To delineate and quantify the learning curve for FBx by assessing the targeted biopsy accuracy and pathological quality of systematic biopsy over time. Design, setting, and participants: We carried out a retrospective analysis of 173 consecutive men who underwent Artemis FBx with computer-template systematic sampling between July 2015 and May 2017. Outcome measurements and statistical analysis: The accuracy of targeted biopsy was determined by calculating the distance between planned and actual core trajectories stored on Artemis. Systematic sampling proficiency was assessed via pathological analysis of fibromuscular tissue in all cores and then comparing pathology elements from individual cores from men in the first and last tertiles. Polynomial linear regression models, changepoint analysis, and piecewise linear regression were used to quantify the learning curve. Results and limitation: A significant improvement in targeted biopsy accuracy occurred up to 98 cases (p < 0.01). There was a significant decrease in fibromuscular tissue in the systematic biopsy cores up to 84 cases (p < 0.01) and an improvement in pathological quality when comparing systematic cores from the first and third tertiles. Use of a different fusion platform may limit the generalizability of our results. Conclusions: There is a significant learning curve for targeted and systemic biopsy using the Artemis platform. Improvements in accuracy of targeted biopsy and better sampling for systematic biopsy can be achieved with greater experience. Patient summary: We define the learning curve for magnetic resonance imaging/ultrasound-guided fusion biopsy (FBx) using targeted biopsy accuracy and systematic core sampling quality as measures. Our findings underscore the importance of overcoming learning curves inherent to FBx to minimize patient discomfort and biopsy risk and improve the quality of care for accurate risk stratification, active surveillance, and treatment selection. © 2018 The Authors. Published by Elsevier B.V. on behalf of European Association of Urology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Associate Editor: Gianluca Giannarini Keywords: Targeted prostate biopsy Magnetic resonance imaging Image fusion Artemis Fusion biopsy prostate biopsy Learning curve Prostate pathology
* Corresponding author. Department of Urology, New York Presbyterian Hospital-Weill Cornell Medical Center, 525 East 68th Street, New York, NY 10021, USA. E-mail address: [email protected] (J.C. Hu).
https://doi.org/10.1016/j.euo.2018.07.005 2588-9311/© 2018 The Authors. Published by Elsevier B.V. on behalf of European Association of Urology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
136
1.
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
Introduction
between July 2015 and May 2017. All procedures were performed by a single urologist (J.C.H.) who did not have prior FBx experience. The study
The 10–12-core transrectal ultrasound (TRUS)-guided systematic biopsy has been the standard diagnostic evaluation for men with elevated prostate specific antigen (PSA) and/or an abnormal digital rectal examination [1]. However, the limitations of conventional prostate biopsy include overdetection of indolent prostate cancer (PC), missing of clinically significant PC (CSPC) in up to 30% of cases [2,3], and Gleason upgrading on radical prostatectomy in up 48% of men [4–6]. This has led to increasing use of multiparametric magnetic resonance imaging (mpMRI) for PC risk stratification followed by MRI/US fusion targeted and systematic biopsies, which have relatively high accuracy in detecting CSPC, as recently demonstrated [7,8]. Currently, the European Association of Urology and American Urological Association both support the use of MRI/US-guided fusion biopsy (FBx) when suspicion for PC persists after a prior negative biopsy [9,10]. However, results from the recent PRECISION trial showed that targeted biopsy for elevated PSA was better in detecting CSPC compared to conventional TRUS biopsy [11]. Given this evidence, MRI/US fusion may be used as a preliminary diagnostic tool for initial biopsy, as noted in the 2018 National Comprehensive Cancer Network guideline for early PC detection [12]. Artemis (Eigen, Grass Valley, CA, USA), a widely used fusion platform, overlays a software-generated systematic template, stores the biopsy needle trajectory, and allows longitudinal tracking of targeted and systematic core trajectories via robotic arm guidance [13]. This platform has been validated for the detection of PC with up to 50% yield among men with previous negative biopsies and suspicious mpMRI [14]. Given the increasing evidence in support of FBx, utilization will probably continue to increase despite the higher cost compared to TRUS ($1072 vs $270), with MRI accounting for 62.5% of the cost ($670) [15]. As we move towards value-based care, it is critical to evaluate potential learning curves for this complex technology [16]. Moreover, quantifying the learning curve is critical for our understanding of why imaging results suspicious for CSPC (Prostate Imaging-Reporting and Data System version 2 [PI-RADS v2] category 3) may have negative FBx findings, which has bearing on the intensity and timing of subsequent monitoring or rebiopsy. Finally, FBx outcomes largely come from high-volume providers at referral centers; challenges with scaling beyond referral centers has significant implications for quality of care. Therefore, the objective of our study was to assess the learning curve for FBx via retrospective review of actual versus planned biopsy core trajectories and the pathologic yield over time. 2.
Patients and methods
2.1.
Patient selection
included only men with mpMRI region(s) of interest (ROI) categorized as a PI-RADS v2 score of 3 undergoing FBx [17]. At the time of starting MRI/US-guided fusion biopsies, the surgeon in our study had performed at least 285 TRUS biopsies during his career, with only 1.3% of biopsy cores identifying nonprostate tissue. Our study was approved by the Weill Cornell Medical Center institutional review board.
2.2.
MRI/US fusion biopsy
All men underwent contrast-enhanced mpMRI using a 3-T magnet without an endorectal coil before biopsy. Studies were performed with T1 and three-plane small field-of-view T2-weighted imaging, dynamic contrast-enhanced imaging, and diffusion-weighted imaging with apparent diffusion coefficient map and computed high b-value images. mpMRI scans were interpreted by fellowship-trained genitourinary (GU) radiologists with experience beyond 100 cases, using the recommendations in PI-RADS v2 [17]. Suspicious ROIs were subsequently mapped for MRI/US-targeted biopsy by a single experienced GU radiologist (D.J.M.). All targeted biopsies were performed using the Artemis platform. Artemis uses an elastic registration system, so mpMRI-identified ROIs are transferred and overlaid onto real-time US images during targeted biopsy. As the user moves the TRUS probe, the Artemis system detects the direction and probe trajectory, thereby implicitly tracking the prostate. Since the probe navigation is mechanical and follows a remote center of motion, the needle will be visible and is detected in all probe orientations during the biopsy, and recorded biopsy needle tracks are registered back to the reference three-dimensional prostate volume and stored. In the case of any unaccounted motion such as patient motion or prostate gland slippage, Artemis has a motion compensation mechanism (rigid registration) that realigns the prostate and the US probe. The targeted biopsy was performed first and the number of cores was varied according to the size of the ROI. After targeted biopsy of ROI(s), 12-core systematic biopsy was sampled using the Artemis system-generated template.
2.3.
Targeted biopsy accuracy
The trajectories of individual targeted biopsy cores were reviewed using data tracked on the Artemis platform in three dimensions. The distance between the ROI and the documented biopsied core trajectory was calculated using MatLAB 2013a (Mathworks, Natick, MA, USA). Specifically, the ROI border coordinates were imported from the US Digital Imaging and Communications in Medicine (DICOM) file and the tracked biopsy trajectories from the Extensible Markup Language (XML) file. Both files are generated and stored on the Artemis platform. ROI borders were encoded as a triangulation object consisting of both vertex coordinates (ie, Cartesian coordinates xn, yn, zn) and the corresponding face. Targeted biopsy core trajectories were recorded as three points: the distal, central, and proximal points. The minimum distance between the biopsy core trajectory and the ROI border was calculated using the following Euclidian distance formula, where (x1, y1, z1) and (x2, y2, z2) are coordinates along the border of the planned location and a midpoint of the biopsy core, respectively: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DðP1; P2Þ ¼ ðx2 x1 Þ2 þ ðy2 y1 Þ2 þ ðz2 z1 Þ2 :
2.4.
Pathology
The pathological quality of the MRI/US systematic template biopsy cores We performed a retrospective analysis of 173 consecutive men without
was quantified in two ways. First, the percentage of fibromuscular
prior PC treatment who underwent FBx with the Artemis platform
(nonprostate) tissue present was quantified. Second, two experienced
137
E U RO P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
Table 1 – Clinical characteristics of three sequential patient cohorts undergoing fusion biopsy performed by a single urologist Parameter Patients (n) Median age, yr (IQR) Median initial PSA, ng/ml (IQR) Medial prostate volume, ml, (IQR) Median BMI, kg/m2 (IQR) Median PI-RADS score (mean)
Tertile 1
Tertile 2
Tertile 3
Total
p value
58 67.7 (59.2–72.1) 7.5 (5.5–12.4) 44.3 (35.0–76.6) 26.7 (25.0–28.5) 4 (3.8)
57 68.2 (61.0–74.6) 6.4 (4.7–9.8) 50.0 (33.0–70.8) 25.7 (24.0–28.6) 4 (3.7)
58 66.5 (59.6–71.1) 6.6 (4.9–9.8) 48.6 (36.8–67.9) 26.5 (24.0–27.9) 4 (3.8)
173 66.2 (65.0–64.5) 13.1 (6.8–19.2) 57.3 (52.3–62.4) 26.6 (26.0–27.1) 4 (3.8)
0.34 0.41 0.96 0.41 0.86
IQR = interquartile range; PSA = prostate-specific antigen; BMI = body mass index; PI-RADS = Prostate Imaging-Reporting and Data System.
GU pathologists reviewed systematic biopsy cores from the first ten and last ten patients in the series and compared 120 cores for total length of tissue, length of the peripheral zone, length of the transition zone, and length of nonprostate tissue (e.g. fibromuscular, colonic wall/mucosa tissue).
2.5.
Data collection and statistical analysis
Patient characteristics were recorded, including age, body mass index, PSA, and prostate volume. To test whether patient and biopsy characteristics changed over time, the men were pooled into tertiles (first, second, and third according to the order of biopsy) and continuous variables were tested using a Kruskall-Wallis test. To investigate the effect of surgeon experience on accuracy, defined as the distance from ROI and MRI/US targeted biopsy core trajectory, a variety of trend models were fitted, and the model with lowest Akaike information criterion (AIC; a measure of goodness of fit) was chosen. The trend models included polynomial linear regression models (with linear, quadratic, and cubic terms for order of biopsy), change-point analysis, and piecewise linear regression. Change-point models allow the slope of the regression line to change at change-point(s), and we fitted models with at most one change-point. The piecewise linear regression model fits two distinct regression lines (with different intercept and slope parameters) and the break-point was determined as the point that
[CI] 4.8–8.6) to 5.4 mm (95% CI 3.6–7.2) and 0.06 mm (95% CI 0.02–0.09) over the three tertiles (p < 0.01). An improvement in specimen quality was also noted for the systematic biopsies. The percentage of fibromuscular tissue decreased significantly over the three tertiles (p < 0.01) from 10.5% (95% CI 7.2–13.8%) to 3.4% (95% CI 1.8–4.9%) and 2.7% (95% CI 0.7–4.7%). Piecewise linear regression demonstrated that the accuracy of MRI/US-targeted biopsy increased up to the 98th targeted biopsy. Estimates of intercept and slope for the piecewise linear model revealed a decrease in distance from ROI by 7.04 mm (95% CI 4.19–9.89; p < 0.01) after the break-point of 98. For systematic biopsy, trend analysis for the percentage fibromuscular tissue showed that a linear decline fitted the data the best (AIC = 361.8) and there was a significant decline of 0.07% (95% CI 0.05–0.1%) with every biopsy (p < 0.01). Piecewise linear regression demonstrated that the percentage fibromuscular tissue decreased significantly up to 84th systematic biopsy. We did not observe any change in CSPC detection over time for targeted (p = 0.43), systematic (p = 0.38), or combination biopsies (p = 0.42) over time.
minimized the mean-squared error. Statistical analysis was performed using STATA SE v13.1 (StataCorp, College Station, TX, USA).
3.2.
3.
Results
3.1.
Biopsy
The systematic biopsy core characteristics are shown in Table 3. The total length of each core did not differ between the first tertile (12.1 mm) and the last tertile (12.6 mm; p = 0.30). However, the amount of peripheral zone sampled increased by 1.5 mm (from 6.7 mm to 8.2 mm; p < 0.05) and amount of transition zone sampled increased by 0.5 mm (from 0.4 mm to 0.9 mm; p < 0.05). The amount of nonprostate tissue sampled decreased significantly from 3.4 mm for the first tertile to 1.9 mm for the third tertile(p < 0.05). Biopsy cores that contained < 5 mm of
Patient characteristics did not vary significantly between tertiles (Table 1). Table 2 demonstrates the variation in tumor and biopsy characteristics over time. The targeted biopsy accuracy improved significantly, with the mean distance between the ROI and the targeted biopsy core trajectory decreasing from 6.7 mm (95% confidence interval
Pathology
Table 2 – Biopsy characteristics for three sequential patient cohorts undergoing fusion biopsy performed by a single urologist Biopsy results
Mean distance from ROI, mm (95% CI) Mean fibromuscular tissue, % (95% CI) CSPC yield in men diagnosed with PC Patients (n) Targeted biopsy, n (%) Systematic (template) biopsy, n (%) Systematic and targeted biopsy, n (%)
Tertile 1 (n = 58)
Tertile 2 (n = 57)
Tertile 3 (n = 58)
Total (n = 173)
p value
6.7 (4.8–8.6) 10.5 (7.2–13.8)
5.4 (3.6–7.2) 3.4 (1.8–4.9)
0.06 (0.02–0.09) 2.7 (0.7–4.7)
4.0 (3.1–5.0) 5.5 (4.1–7.0)
<0.01 <0.01
30 18 (60.0) 15 (50.0) 21 (70.0)
33 14 (42.2) 20 (60.1) 22 (66.7)
26 12 (46.2) 14 (56.0) 16 (61.5)
ROI = region of interest; CI = confidence interval; PC = prostate cancer; CSPC = clinically significant PC.
89 44 (49.4) 49 (55.0) 59 (66.3)
0.43 0.38 0.42
138
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
Table 3 – Pathological characteristics of systematic biopsy cores Parameter
Mean Mean Mean Mean Mean Mean Mean
total length of each core (mm) length of peripheral zone (mm) length of transition zone (mm) length of nonprostate tissue (mm) cores containing <5 mm of peripheral-zone tissue (%) cores containing <5 mm of stromal tissue (%) cores that missed the prostate (%)
peripheral-zone tissue decreased from the first to the third tertile (35% vs 18% and 18% vs 6.7%, respectively; p < 0.05). Biopsy cores that missed the prostate entirely decreased from 18% in the first tertile to 3.3% in the last tertile (p < 0.05). 4.
Discussion
Originally developed for PC staging, prostate mpMRI has evolved into an important diagnostic tool that facilitates CSPC detection and risk stratification when used on MRI/US FBx platforms for direct sampling of suspicious ROIs [4,18]. The PRECISION trial recently showed that rates of CSPC detection were higher among men undergoing MRI/ US-targeted biopsy compared to those undergoing TRUS biopsy [11]. In addition, omitting the systematic biopsy decreased detection of clinically insignificant PC. Similar findings were seen in a prospective trial by Siddiqui et al. in 2015 [19]. On the basis of these recent studies, MRI/US FBx targeted biopsy has the potential to become the gold standard diagnostic tool for men with suspicion of PC [11,19]. However, barriers to widespread use of MRI-based biopsy technology include cost and the availability of prostate mpMRI radiologic expertise for interpretation [20]. Another limitation is the potential learning curve that may limit the accuracy of diagnosis by novice operators. Our study suggests that experience improves the accuracy of MRI/US-targeted biopsy, with a learning curve of 98 cases identified. Previous studies have alluded to the presence of a learning curve for cancer detection using FBx technology. Calio et al. [21] performed a study of targeted and systematic biopsies using the Artemis platform in 1528 patients and found that the PC detection rate with FBx significantly increased with surgeon experience, suggesting the presence of a learning curve, which they did not quantify. While we did not note any difference in PC detection, our study mathematically calculated the accuracy of MRI/US-targeted biopsy and tracking and found improvements in actual versus intended distance up to 98 cases. This finding suggests that the ROI may be missed during early operator experience with MRI/US-targeted biopsies. To overcome this, the computer-generated Artemis template, which geometrically spaces the 12-core systematic biopsy throughout to uniformly sample the prostate, is necessary to ensure thorough sampling, as two systematic biopsies overlap with the ROI on average
Tertile 1 (n = 10)
Tertile 3 (n = 10)
p value
12.1 6.7 0.4 3.4 35.0 18.0 18.0
12.6 8.2 0.9 1.9 18.0 6.7 3.3
0.30 <0.05 <0.05 <0.05 <0.05 <0.05 <0.05
[22]. Thus, it may be prudent to overcome the learning curve before adopting a target-only biopsy strategy as performed in the PRECISION trial. Given the growing discussion on morbidity associated with prostate biopsy, including sepsis and pain [23], and the risk of overtreatment of PC, obtaining fewer cores with target-only biopsies may have a role in more experienced centers. Spatial tracking of targets with MRI/US-targeted biopsies also has clinical value in active surveillance. Chang et al. [24] followed 352 men on active surveillance who underwent repeat MRI/US-targeted biopsies on the Artemis platform 1 yr after initial diagnosis. On subsequent biopsy, 91 men had PC upgrading and 48 (53%) of those were detected by tracking previous biopsy sites. Of note, this study fused the initial image marked with previous biopsy locations with the new model. Then repeat cores were taken from the previously biopsied areas. They did not compare the accuracy of actual versus intended biopsy sites. This fusion mapping technique was also used in a study by Palapattu et al. [25] with the Artemis platform. They were able to use MRI/US fusion mapping to obtain 96% concordance of PC clonal foci on repeat sampling of a specific biopsy site 1 yr later. Furthermore, accurate tracking of the biopsy trajectory can serve as a guide for future partial gland ablation when the biopsy-proven ROI is treated [26]. With regard to the pathological quality of systematic biopsies with Artemis, we also noted a significant learning curve, with a decrease in the percentage of cores with fibromuscular (nonprostate) tissue over time up to 84 cases. This suggests that novice users may be more prone to following a trajectory outside the borders of the prostatic capsule, which is not unusual given the complexity of the biopsy technique. Technically, the surgeon must follow the real-time US image rather than the Artemis-generated rendition of the spatial relationship of the systematic biopsy trajectories relative to the software-generated prostate border. Accurate sampling of the prostate is important given that histopathological biopsy quality, including tumor volume, affects clinical decision-making. In fact, Quintal et al. [27] found that the total percentage of carcinoma in all cores and the number and percentage of cores with cancer were significant predictors of biochemical recurrence. Our results demonstrate an improvement in biopsy core quality, including a reduction in nonprostate tissue on systematic biopsy over the time. Thus, it is important to consider the learning curve of the operator when using Artemisgenerated systematic templates given that experience
E U RO P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
portends an improvement in the diagnostic accuracy of systematic biopsy, as validated by histopathology. Our findings must be interpreted in the context of the study design. First, owing to the labor-intensive nature of capturing and analyzing the data in our study, including the Artemis trajectory calculations and in-depth pathological analysis, we were able to capture details of the initial experience a single surgeon. Given the resources needed to retrospectively review and quantify stored biopsy trajectories and assess pathologic sampling, we examined the learning curve for the initial 173 men undergoing FBx. However, the learning curve flattened at 98 FBx, and additional assessment would not have altered our results. Second, the study generalizability must be considered, as the characteristics of practitioners performing FBx are changing. Increasingly, non-urologist practitioners including physician extenders, radiologists and interventional radiologists are using this technology [28]. Third, the success of FBx depends on the combined performance and expertise of the radiologist, urologist, and pathologist, as well as the platform used. The presence of a learning curve for radiologists in tumor detection on prostate mpMRI [29] and for pathologists in accurate staging of prostate samples [5,30] has previously been defined. Although radiologists at our center are experienced in prostate mpMRI, a recent study demonstrated 40–80% cancer yield for PI-RADS 5 lesions at one academic medical center, depending on the radiologist [29]. Lastly, our findings for the Artemis FBx platform may not be generalizable to the technical nuances of other platforms. A significant determinant of accuracy for targeted biopsy with Artemis arises from optimal patient positioning to maximize the robotic arm articulation for access to all areas of the prostate. This step is done before three-dimensional US acquisition. The prostate should be centered in the US field of view, with the widest dimensions captured in the transverse orientation. In this probe orientation, the robotic arm must be attached to the probe. During attachment, it is important to ensure that the arm is in a neutral position (or optimal position of maximum arm articulation in all directions). Failure to initially position the US probe and robotic arm in a neutral position will compromise the subsequent movement needed to cover the prostate fully to conduct end-fire sampling. Given the complexity of this particular platform, early and frequent training protocols provided by the manufacturer may reduce the time to proficiency for this and alternative platforms. Despite this, in light of the PRECISION trial results, use of MRI/US-targeted biopsies will probably expand. Our findings demonstrate the importance of overcoming and recognizing learning curves inherent to FBx to minimize patient discomfort and biopsy risk and improve the quality of care.
139
experience must be considered when interpreting Artemis FBx outcomes and particularly in the setting of active surveillance and focal therapy. Author contributions: Khushabu Kasabwala had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kasabwala, Patel, Khani, Margolis, Robinson, Hu. Acquisition of data: Kasabwala, Patel, Cricco-Lizza, Shimpi, Weng, Motanagh, Khani, Margolis, Robinson, Hu. Analysis and interpretation of data: Kasabwala, Patel, Cricco-Lizza, Shimpi, Weng, Motanagh, Wu, Khani, Margolis, Banerjee, Robinson, Hu. Drafting of the manuscript: Kasabwala, Patel, Cricco-Lizza, Shimpi, Weng, Motanagh, Wu, Khani, Margolis, Banerjee, Robinson, Hu. Critical revision of the manuscript for important intellectual content: Kasabwala, Patel, Wu, Margolis, Banerjee, Robinson, Hu. Statistical analysis: Kasabwala, Motanagh, Wu, Banerjee, Robinson. Obtaining funding: Hu. Administrative, technical, or material support: None. Supervision: Robinson, Hu. Other: None. Financial disclosures: Khushabu Kasabwala certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: Jim C. Hu is on a speaker’s bureau for Genomic Health. Daniel J. Margolis is an ad hoc consultant for Blue Earth Diagnostics. The remaining authors have nothing to disclose. Funding/Support and role of the sponsor: This study was supported by The Frederick J. and Theresa Dow Wallace Fund of the New York Community Trust. The sponsor played no direct role in the study. Acknowledgments: The authors are grateful to Rajesh Venkataraman for helping to prepare the manuscript.
References [1] Bjurlin MA, Wysock JS, Taneja SS. Optimization of prostate biopsy: review of technique and complications. Urol Clin North Am 2014;41:299–313. [2] Lecornet E, Ahmed HU, Hu Y, et al. The accuracy of different biopsy strategies for the detection of clinically important prostate cancer: a computer simulation. J Urol 2012;188:974–80. [3] Serefoglu EC, Altinova S, Ugras NS, Akincioglu E, Asil E, Balbay MD. How reliable is 12-core prostate biopsy procedure in the detection of prostate cancer? Can Urol Assoc J 2013;7:E293–8. [4] Puech P, Potiron E, Lemaitre L, et al. Dynamic contrast-enhancedmagnetic resonance imaging evaluation of intraprostatic prostate cancer: correlation with radical prostatectomy specimens. Urology 2009;74:1094–9. [5] Kvale R, Moller B, Wahlqvist R, et al. Concordance between Gleason scores of needle biopsies and radical prostatectomy specimens: a population-based study. BJU Int 2009;103:1647–54.
5.
Conclusions
[6] Cooperberg MR, Broering JM, Kantoff PW, Carroll PR. Contemporary trends in low risk prostate cancer: risk assessment and treatment. J
We demonstrated that there is a significant learning curve for FBx, specifically for accurate ROI targeting and systematic biopsy core yield. This has significant implications for accurate risk stratification and diagnosis. Provider
Urol 2007;178:S14–9. [7] de Rooij M, Hamoen EH, Witjes JA, Barentsz JO, Rovers MM. Accuracy of magnetic resonance imaging for local staging of prostate cancer: a diagnostic meta-analysis. Eur Urol 2016;70:233–45.
140
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
[8] Futterer JJ, Briganti A, De Visschere P, et al. Can clinically significant
[20] Mager R, Brandt MP, Borgmann H, Gust KM, Haferkamp A, Kurosch
prostate cancer be detected with multiparametric magnetic reso-
M. From novice to expert: analyzing the learning curve for MRI-
nance imaging?. A systematic review of the literature. Eur Urol
transrectal ultrasonography fusion-guided transrectal prostate bi-
2015;68:1045–53.
opsy. Int Urol Nephrol 2017;49:1537–44.
[9] Rosenkrantz AB, Verma S, Choyke P, et al. Prostate magnetic reso-
[21] Calio B, Sidana A, Sugano D, et al. Changes in prostate cancer
nance imaging and magnetic resonance imaging targeted biopsy in
detection rate of MRI-TRUS fusion vs systematic biopsy over time:
patients with a prior negative biopsy: a consensus statement by
evidence of a learning curve. Prostate Cancer Prostat Dis
AUA and SAR. J Urol 2016;196:1613–8.
2017;20:436–41.
[10] Mottet N, Bellmunt J, Bolla M, et al. EAU-ESTRO-SIOG guidelines on
[22] Patel N, Cricco-Lizza E, Kasabwala K, et al. The role of systematic and
prostate cancer. Part 1: screening, diagnosis, and local treatment
targeted biopsies in light of overlap on magnetic-resonance image
with curative intent. Eur Urol 2017;71:618–29. [11] Kasivisvanathan V, Rannikko AS, Borghi M, et al. MRI-targeted or standard biopsy for prostate-cancer diagnosis. N Engl J Med [12] National Comprehensive Cancer Network. Prostate cancer early (version
plications of prostate biopsy. Eur Urol 2013;64:876–92. [24] Chang E, Jones TA, Natarajan S, et al. Value of tracking biopsy in men
2018;378:1767–77. detection
ultrasound fusion biopsy. Eur Urol Oncol 2018;1:263–7. [23] Loeb S, Vellekoop A, Ahmed HU, et al. Systematic review of com-
2.2017).
www.nccn.org/professionals/
physician_gls/pdf/prostate_detection.pdf. [13] Marks L, Young S, Natarajan S. MRI–ultrasound fusion for guidance of targeted prostate biopsy. Curr Opin Urol 2013;23:43–50. [14] Sonn GA, Natarajan S, Margolis DJ, et al. Targeted biopsy in the detection of prostate cancer using an office based magnetic resonance ultrasound fusion device. J Urol 2013;189:86–91.
undergoing
active
surveillance
of
prostate
cancer.
J
Urol
2018;199:98–105. [25] Palapattu GS, Salami SS, Cani AK, et al. Molecular profiling to determine clonality of serial magnetic resonance imaging/ultrasound fusion biopsies from men on active surveillance for low-risk prostate cancer. Clin Cancer Res 2017;23:985–91. [26] Natarajan S, Raman S, Priester AM, et al. Focal laser ablation of prostate cancer: phase I clinical trial. J Urol 2016;196:68–75.
[15] Laviana AA, Ilg AM, Veruttipong D, et al. Utilizing time-driven activity-
[27] Quintal MM, Meirelles LR, Freitas LL, Magna LA, Ferreira U, Billis A.
based costing to understand the short- and long-term costs of treating
Various morphometric measurements of cancer extent on needle
localized, low-risk prostate cancer. Cancer 2016;122:447–55.
prostatic biopsies: which is predictive of pathologic stage and
[16] Ahmed HU, El-Shater Bosaily A, Brown LC, et al. Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study. Lancet 2017;389:815–22. [17] Weinreb JC, Barentsz JO, Choyke PL, et al. PI-RADS Prostate Imaging-
biochemical recurrence following radical prostatectomy? Int Urol Nephrol 2011;43:697–705. [28] Hong CW, Amalou H, Xu S, et al. Prostate biopsy for the interventional radiologist. J Vasc Interv Radiol 2014;25:675–84.
Reporting and Data System: 2015, version 2. Eur Urol 2016;69:16–40.
[29] Sonn GA, Fan RE, Ghanouni P, et al. Prostate magnetic resonance
[18] Borkowetz A, Platzek I, Toma M, et al. Comparison of systematic
imaging interpretation varies substantially across radiologists. Eur
transrectal biopsy to transperineal magnetic resonance imaging/
Urol Focus. In press. https://doi.org/10.1016/j.euf.2017.11.010.
ultrasound-fusion biopsy for the diagnosis of prostate cancer. BJU
[30] Majoros A, Szasz AM, Nyirady P, et al. The influence of expertise of
Int 2015;116:873–9. [19] Siddiqui MM, Rais-Bahrami S, Turkbey B, et al. Comparison of MR/ ultrasound fusion-guided biopsy with ultrasound-guided biopsy for the diagnosis of prostate cancer. JAMA 2015;313:390–7.
the surgical pathologist to undergrading, upgrading, and understaging of prostate cancer in patients undergoing subsequent radical prostatectomy. Int Urol Nephrol 2014;46:371–7.
available at www.sciencedirect.com journal homepage: euoncology.europeanurology.com
The Learning Curve for Magnetic Resonance Imaging/Ultrasound Fusion-guided Prostate Biopsy Khushabu Kasabwala a, Neal Patel a, Eliza Cricco-Lizza a, Adrian A. Shimpi b, Stanley Weng a, Rose M. Buchmann b, Samaneh Motanagh c, Yiyuan Wu d, Samprit Banerjee d, Francesca Khani a,c, Daniel J. Margolis e, Brian D. Robinson a,c, Jim C. Hu a,* a
Department of Urology, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA;
Cornell University, Ithaca, NY, USA; d
c
b
Meinig School of Biomedical Engineering,
Department of Pathology, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA;
Department of Healthcare Policy and Research, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA;
e
Department of
Radiology, New York Presbyterian Hospital, Weill Cornell Medical College, New York, NY, USA
Article info
Abstract
Article history: Accepted July 16, 2018
Background: Magnetic resonance imaging/ultrasound-guided fusion biopsy (FBx) is more accurate at detecting clinically significant prostate cancer than conventional transrectal ultrasound-guided systematic biopsy. However, learning curves for attaining accuracy may limit the generalizability of published outcomes. Objective: To delineate and quantify the learning curve for FBx by assessing the targeted biopsy accuracy and pathological quality of systematic biopsy over time. Design, setting, and participants: We carried out a retrospective analysis of 173 consecutive men who underwent Artemis FBx with computer-template systematic sampling between July 2015 and May 2017. Outcome measurements and statistical analysis: The accuracy of targeted biopsy was determined by calculating the distance between planned and actual core trajectories stored on Artemis. Systematic sampling proficiency was assessed via pathological analysis of fibromuscular tissue in all cores and then comparing pathology elements from individual cores from men in the first and last tertiles. Polynomial linear regression models, changepoint analysis, and piecewise linear regression were used to quantify the learning curve. Results and limitation: A significant improvement in targeted biopsy accuracy occurred up to 98 cases (p < 0.01). There was a significant decrease in fibromuscular tissue in the systematic biopsy cores up to 84 cases (p < 0.01) and an improvement in pathological quality when comparing systematic cores from the first and third tertiles. Use of a different fusion platform may limit the generalizability of our results. Conclusions: There is a significant learning curve for targeted and systemic biopsy using the Artemis platform. Improvements in accuracy of targeted biopsy and better sampling for systematic biopsy can be achieved with greater experience. Patient summary: We define the learning curve for magnetic resonance imaging/ultrasound-guided fusion biopsy (FBx) using targeted biopsy accuracy and systematic core sampling quality as measures. Our findings underscore the importance of overcoming learning curves inherent to FBx to minimize patient discomfort and biopsy risk and improve the quality of care for accurate risk stratification, active surveillance, and treatment selection. © 2018 The Authors. Published by Elsevier B.V. on behalf of European Association of Urology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Associate Editor: Gianluca Giannarini Keywords: Targeted prostate biopsy Magnetic resonance imaging Image fusion Artemis Fusion biopsy prostate biopsy Learning curve Prostate pathology
* Corresponding author. Department of Urology, New York Presbyterian Hospital-Weill Cornell Medical Center, 525 East 68th Street, New York, NY 10021, USA. E-mail address: [email protected] (J.C. Hu).
https://doi.org/10.1016/j.euo.2018.07.005 2588-9311/© 2018 The Authors. Published by Elsevier B.V. on behalf of European Association of Urology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
136
1.
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
Introduction
between July 2015 and May 2017. All procedures were performed by a single urologist (J.C.H.) who did not have prior FBx experience. The study
The 10–12-core transrectal ultrasound (TRUS)-guided systematic biopsy has been the standard diagnostic evaluation for men with elevated prostate specific antigen (PSA) and/or an abnormal digital rectal examination [1]. However, the limitations of conventional prostate biopsy include overdetection of indolent prostate cancer (PC), missing of clinically significant PC (CSPC) in up to 30% of cases [2,3], and Gleason upgrading on radical prostatectomy in up 48% of men [4–6]. This has led to increasing use of multiparametric magnetic resonance imaging (mpMRI) for PC risk stratification followed by MRI/US fusion targeted and systematic biopsies, which have relatively high accuracy in detecting CSPC, as recently demonstrated [7,8]. Currently, the European Association of Urology and American Urological Association both support the use of MRI/US-guided fusion biopsy (FBx) when suspicion for PC persists after a prior negative biopsy [9,10]. However, results from the recent PRECISION trial showed that targeted biopsy for elevated PSA was better in detecting CSPC compared to conventional TRUS biopsy [11]. Given this evidence, MRI/US fusion may be used as a preliminary diagnostic tool for initial biopsy, as noted in the 2018 National Comprehensive Cancer Network guideline for early PC detection [12]. Artemis (Eigen, Grass Valley, CA, USA), a widely used fusion platform, overlays a software-generated systematic template, stores the biopsy needle trajectory, and allows longitudinal tracking of targeted and systematic core trajectories via robotic arm guidance [13]. This platform has been validated for the detection of PC with up to 50% yield among men with previous negative biopsies and suspicious mpMRI [14]. Given the increasing evidence in support of FBx, utilization will probably continue to increase despite the higher cost compared to TRUS ($1072 vs $270), with MRI accounting for 62.5% of the cost ($670) [15]. As we move towards value-based care, it is critical to evaluate potential learning curves for this complex technology [16]. Moreover, quantifying the learning curve is critical for our understanding of why imaging results suspicious for CSPC (Prostate Imaging-Reporting and Data System version 2 [PI-RADS v2] category 3) may have negative FBx findings, which has bearing on the intensity and timing of subsequent monitoring or rebiopsy. Finally, FBx outcomes largely come from high-volume providers at referral centers; challenges with scaling beyond referral centers has significant implications for quality of care. Therefore, the objective of our study was to assess the learning curve for FBx via retrospective review of actual versus planned biopsy core trajectories and the pathologic yield over time. 2.
Patients and methods
2.1.
Patient selection
included only men with mpMRI region(s) of interest (ROI) categorized as a PI-RADS v2 score of 3 undergoing FBx [17]. At the time of starting MRI/US-guided fusion biopsies, the surgeon in our study had performed at least 285 TRUS biopsies during his career, with only 1.3% of biopsy cores identifying nonprostate tissue. Our study was approved by the Weill Cornell Medical Center institutional review board.
2.2.
MRI/US fusion biopsy
All men underwent contrast-enhanced mpMRI using a 3-T magnet without an endorectal coil before biopsy. Studies were performed with T1 and three-plane small field-of-view T2-weighted imaging, dynamic contrast-enhanced imaging, and diffusion-weighted imaging with apparent diffusion coefficient map and computed high b-value images. mpMRI scans were interpreted by fellowship-trained genitourinary (GU) radiologists with experience beyond 100 cases, using the recommendations in PI-RADS v2 [17]. Suspicious ROIs were subsequently mapped for MRI/US-targeted biopsy by a single experienced GU radiologist (D.J.M.). All targeted biopsies were performed using the Artemis platform. Artemis uses an elastic registration system, so mpMRI-identified ROIs are transferred and overlaid onto real-time US images during targeted biopsy. As the user moves the TRUS probe, the Artemis system detects the direction and probe trajectory, thereby implicitly tracking the prostate. Since the probe navigation is mechanical and follows a remote center of motion, the needle will be visible and is detected in all probe orientations during the biopsy, and recorded biopsy needle tracks are registered back to the reference three-dimensional prostate volume and stored. In the case of any unaccounted motion such as patient motion or prostate gland slippage, Artemis has a motion compensation mechanism (rigid registration) that realigns the prostate and the US probe. The targeted biopsy was performed first and the number of cores was varied according to the size of the ROI. After targeted biopsy of ROI(s), 12-core systematic biopsy was sampled using the Artemis system-generated template.
2.3.
Targeted biopsy accuracy
The trajectories of individual targeted biopsy cores were reviewed using data tracked on the Artemis platform in three dimensions. The distance between the ROI and the documented biopsied core trajectory was calculated using MatLAB 2013a (Mathworks, Natick, MA, USA). Specifically, the ROI border coordinates were imported from the US Digital Imaging and Communications in Medicine (DICOM) file and the tracked biopsy trajectories from the Extensible Markup Language (XML) file. Both files are generated and stored on the Artemis platform. ROI borders were encoded as a triangulation object consisting of both vertex coordinates (ie, Cartesian coordinates xn, yn, zn) and the corresponding face. Targeted biopsy core trajectories were recorded as three points: the distal, central, and proximal points. The minimum distance between the biopsy core trajectory and the ROI border was calculated using the following Euclidian distance formula, where (x1, y1, z1) and (x2, y2, z2) are coordinates along the border of the planned location and a midpoint of the biopsy core, respectively: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DðP1; P2Þ ¼ ðx2 x1 Þ2 þ ðy2 y1 Þ2 þ ðz2 z1 Þ2 :
2.4.
Pathology
The pathological quality of the MRI/US systematic template biopsy cores We performed a retrospective analysis of 173 consecutive men without
was quantified in two ways. First, the percentage of fibromuscular
prior PC treatment who underwent FBx with the Artemis platform
(nonprostate) tissue present was quantified. Second, two experienced
137
E U RO P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
Table 1 – Clinical characteristics of three sequential patient cohorts undergoing fusion biopsy performed by a single urologist Parameter Patients (n) Median age, yr (IQR) Median initial PSA, ng/ml (IQR) Medial prostate volume, ml, (IQR) Median BMI, kg/m2 (IQR) Median PI-RADS score (mean)
Tertile 1
Tertile 2
Tertile 3
Total
p value
58 67.7 (59.2–72.1) 7.5 (5.5–12.4) 44.3 (35.0–76.6) 26.7 (25.0–28.5) 4 (3.8)
57 68.2 (61.0–74.6) 6.4 (4.7–9.8) 50.0 (33.0–70.8) 25.7 (24.0–28.6) 4 (3.7)
58 66.5 (59.6–71.1) 6.6 (4.9–9.8) 48.6 (36.8–67.9) 26.5 (24.0–27.9) 4 (3.8)
173 66.2 (65.0–64.5) 13.1 (6.8–19.2) 57.3 (52.3–62.4) 26.6 (26.0–27.1) 4 (3.8)
0.34 0.41 0.96 0.41 0.86
IQR = interquartile range; PSA = prostate-specific antigen; BMI = body mass index; PI-RADS = Prostate Imaging-Reporting and Data System.
GU pathologists reviewed systematic biopsy cores from the first ten and last ten patients in the series and compared 120 cores for total length of tissue, length of the peripheral zone, length of the transition zone, and length of nonprostate tissue (e.g. fibromuscular, colonic wall/mucosa tissue).
2.5.
Data collection and statistical analysis
Patient characteristics were recorded, including age, body mass index, PSA, and prostate volume. To test whether patient and biopsy characteristics changed over time, the men were pooled into tertiles (first, second, and third according to the order of biopsy) and continuous variables were tested using a Kruskall-Wallis test. To investigate the effect of surgeon experience on accuracy, defined as the distance from ROI and MRI/US targeted biopsy core trajectory, a variety of trend models were fitted, and the model with lowest Akaike information criterion (AIC; a measure of goodness of fit) was chosen. The trend models included polynomial linear regression models (with linear, quadratic, and cubic terms for order of biopsy), change-point analysis, and piecewise linear regression. Change-point models allow the slope of the regression line to change at change-point(s), and we fitted models with at most one change-point. The piecewise linear regression model fits two distinct regression lines (with different intercept and slope parameters) and the break-point was determined as the point that
[CI] 4.8–8.6) to 5.4 mm (95% CI 3.6–7.2) and 0.06 mm (95% CI 0.02–0.09) over the three tertiles (p < 0.01). An improvement in specimen quality was also noted for the systematic biopsies. The percentage of fibromuscular tissue decreased significantly over the three tertiles (p < 0.01) from 10.5% (95% CI 7.2–13.8%) to 3.4% (95% CI 1.8–4.9%) and 2.7% (95% CI 0.7–4.7%). Piecewise linear regression demonstrated that the accuracy of MRI/US-targeted biopsy increased up to the 98th targeted biopsy. Estimates of intercept and slope for the piecewise linear model revealed a decrease in distance from ROI by 7.04 mm (95% CI 4.19–9.89; p < 0.01) after the break-point of 98. For systematic biopsy, trend analysis for the percentage fibromuscular tissue showed that a linear decline fitted the data the best (AIC = 361.8) and there was a significant decline of 0.07% (95% CI 0.05–0.1%) with every biopsy (p < 0.01). Piecewise linear regression demonstrated that the percentage fibromuscular tissue decreased significantly up to 84th systematic biopsy. We did not observe any change in CSPC detection over time for targeted (p = 0.43), systematic (p = 0.38), or combination biopsies (p = 0.42) over time.
minimized the mean-squared error. Statistical analysis was performed using STATA SE v13.1 (StataCorp, College Station, TX, USA).
3.2.
3.
Results
3.1.
Biopsy
The systematic biopsy core characteristics are shown in Table 3. The total length of each core did not differ between the first tertile (12.1 mm) and the last tertile (12.6 mm; p = 0.30). However, the amount of peripheral zone sampled increased by 1.5 mm (from 6.7 mm to 8.2 mm; p < 0.05) and amount of transition zone sampled increased by 0.5 mm (from 0.4 mm to 0.9 mm; p < 0.05). The amount of nonprostate tissue sampled decreased significantly from 3.4 mm for the first tertile to 1.9 mm for the third tertile(p < 0.05). Biopsy cores that contained < 5 mm of
Patient characteristics did not vary significantly between tertiles (Table 1). Table 2 demonstrates the variation in tumor and biopsy characteristics over time. The targeted biopsy accuracy improved significantly, with the mean distance between the ROI and the targeted biopsy core trajectory decreasing from 6.7 mm (95% confidence interval
Pathology
Table 2 – Biopsy characteristics for three sequential patient cohorts undergoing fusion biopsy performed by a single urologist Biopsy results
Mean distance from ROI, mm (95% CI) Mean fibromuscular tissue, % (95% CI) CSPC yield in men diagnosed with PC Patients (n) Targeted biopsy, n (%) Systematic (template) biopsy, n (%) Systematic and targeted biopsy, n (%)
Tertile 1 (n = 58)
Tertile 2 (n = 57)
Tertile 3 (n = 58)
Total (n = 173)
p value
6.7 (4.8–8.6) 10.5 (7.2–13.8)
5.4 (3.6–7.2) 3.4 (1.8–4.9)
0.06 (0.02–0.09) 2.7 (0.7–4.7)
4.0 (3.1–5.0) 5.5 (4.1–7.0)
<0.01 <0.01
30 18 (60.0) 15 (50.0) 21 (70.0)
33 14 (42.2) 20 (60.1) 22 (66.7)
26 12 (46.2) 14 (56.0) 16 (61.5)
ROI = region of interest; CI = confidence interval; PC = prostate cancer; CSPC = clinically significant PC.
89 44 (49.4) 49 (55.0) 59 (66.3)
0.43 0.38 0.42
138
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
Table 3 – Pathological characteristics of systematic biopsy cores Parameter
Mean Mean Mean Mean Mean Mean Mean
total length of each core (mm) length of peripheral zone (mm) length of transition zone (mm) length of nonprostate tissue (mm) cores containing <5 mm of peripheral-zone tissue (%) cores containing <5 mm of stromal tissue (%) cores that missed the prostate (%)
peripheral-zone tissue decreased from the first to the third tertile (35% vs 18% and 18% vs 6.7%, respectively; p < 0.05). Biopsy cores that missed the prostate entirely decreased from 18% in the first tertile to 3.3% in the last tertile (p < 0.05). 4.
Discussion
Originally developed for PC staging, prostate mpMRI has evolved into an important diagnostic tool that facilitates CSPC detection and risk stratification when used on MRI/US FBx platforms for direct sampling of suspicious ROIs [4,18]. The PRECISION trial recently showed that rates of CSPC detection were higher among men undergoing MRI/ US-targeted biopsy compared to those undergoing TRUS biopsy [11]. In addition, omitting the systematic biopsy decreased detection of clinically insignificant PC. Similar findings were seen in a prospective trial by Siddiqui et al. in 2015 [19]. On the basis of these recent studies, MRI/US FBx targeted biopsy has the potential to become the gold standard diagnostic tool for men with suspicion of PC [11,19]. However, barriers to widespread use of MRI-based biopsy technology include cost and the availability of prostate mpMRI radiologic expertise for interpretation [20]. Another limitation is the potential learning curve that may limit the accuracy of diagnosis by novice operators. Our study suggests that experience improves the accuracy of MRI/US-targeted biopsy, with a learning curve of 98 cases identified. Previous studies have alluded to the presence of a learning curve for cancer detection using FBx technology. Calio et al. [21] performed a study of targeted and systematic biopsies using the Artemis platform in 1528 patients and found that the PC detection rate with FBx significantly increased with surgeon experience, suggesting the presence of a learning curve, which they did not quantify. While we did not note any difference in PC detection, our study mathematically calculated the accuracy of MRI/US-targeted biopsy and tracking and found improvements in actual versus intended distance up to 98 cases. This finding suggests that the ROI may be missed during early operator experience with MRI/US-targeted biopsies. To overcome this, the computer-generated Artemis template, which geometrically spaces the 12-core systematic biopsy throughout to uniformly sample the prostate, is necessary to ensure thorough sampling, as two systematic biopsies overlap with the ROI on average
Tertile 1 (n = 10)
Tertile 3 (n = 10)
p value
12.1 6.7 0.4 3.4 35.0 18.0 18.0
12.6 8.2 0.9 1.9 18.0 6.7 3.3
0.30 <0.05 <0.05 <0.05 <0.05 <0.05 <0.05
[22]. Thus, it may be prudent to overcome the learning curve before adopting a target-only biopsy strategy as performed in the PRECISION trial. Given the growing discussion on morbidity associated with prostate biopsy, including sepsis and pain [23], and the risk of overtreatment of PC, obtaining fewer cores with target-only biopsies may have a role in more experienced centers. Spatial tracking of targets with MRI/US-targeted biopsies also has clinical value in active surveillance. Chang et al. [24] followed 352 men on active surveillance who underwent repeat MRI/US-targeted biopsies on the Artemis platform 1 yr after initial diagnosis. On subsequent biopsy, 91 men had PC upgrading and 48 (53%) of those were detected by tracking previous biopsy sites. Of note, this study fused the initial image marked with previous biopsy locations with the new model. Then repeat cores were taken from the previously biopsied areas. They did not compare the accuracy of actual versus intended biopsy sites. This fusion mapping technique was also used in a study by Palapattu et al. [25] with the Artemis platform. They were able to use MRI/US fusion mapping to obtain 96% concordance of PC clonal foci on repeat sampling of a specific biopsy site 1 yr later. Furthermore, accurate tracking of the biopsy trajectory can serve as a guide for future partial gland ablation when the biopsy-proven ROI is treated [26]. With regard to the pathological quality of systematic biopsies with Artemis, we also noted a significant learning curve, with a decrease in the percentage of cores with fibromuscular (nonprostate) tissue over time up to 84 cases. This suggests that novice users may be more prone to following a trajectory outside the borders of the prostatic capsule, which is not unusual given the complexity of the biopsy technique. Technically, the surgeon must follow the real-time US image rather than the Artemis-generated rendition of the spatial relationship of the systematic biopsy trajectories relative to the software-generated prostate border. Accurate sampling of the prostate is important given that histopathological biopsy quality, including tumor volume, affects clinical decision-making. In fact, Quintal et al. [27] found that the total percentage of carcinoma in all cores and the number and percentage of cores with cancer were significant predictors of biochemical recurrence. Our results demonstrate an improvement in biopsy core quality, including a reduction in nonprostate tissue on systematic biopsy over the time. Thus, it is important to consider the learning curve of the operator when using Artemisgenerated systematic templates given that experience
E U RO P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
portends an improvement in the diagnostic accuracy of systematic biopsy, as validated by histopathology. Our findings must be interpreted in the context of the study design. First, owing to the labor-intensive nature of capturing and analyzing the data in our study, including the Artemis trajectory calculations and in-depth pathological analysis, we were able to capture details of the initial experience a single surgeon. Given the resources needed to retrospectively review and quantify stored biopsy trajectories and assess pathologic sampling, we examined the learning curve for the initial 173 men undergoing FBx. However, the learning curve flattened at 98 FBx, and additional assessment would not have altered our results. Second, the study generalizability must be considered, as the characteristics of practitioners performing FBx are changing. Increasingly, non-urologist practitioners including physician extenders, radiologists and interventional radiologists are using this technology [28]. Third, the success of FBx depends on the combined performance and expertise of the radiologist, urologist, and pathologist, as well as the platform used. The presence of a learning curve for radiologists in tumor detection on prostate mpMRI [29] and for pathologists in accurate staging of prostate samples [5,30] has previously been defined. Although radiologists at our center are experienced in prostate mpMRI, a recent study demonstrated 40–80% cancer yield for PI-RADS 5 lesions at one academic medical center, depending on the radiologist [29]. Lastly, our findings for the Artemis FBx platform may not be generalizable to the technical nuances of other platforms. A significant determinant of accuracy for targeted biopsy with Artemis arises from optimal patient positioning to maximize the robotic arm articulation for access to all areas of the prostate. This step is done before three-dimensional US acquisition. The prostate should be centered in the US field of view, with the widest dimensions captured in the transverse orientation. In this probe orientation, the robotic arm must be attached to the probe. During attachment, it is important to ensure that the arm is in a neutral position (or optimal position of maximum arm articulation in all directions). Failure to initially position the US probe and robotic arm in a neutral position will compromise the subsequent movement needed to cover the prostate fully to conduct end-fire sampling. Given the complexity of this particular platform, early and frequent training protocols provided by the manufacturer may reduce the time to proficiency for this and alternative platforms. Despite this, in light of the PRECISION trial results, use of MRI/US-targeted biopsies will probably expand. Our findings demonstrate the importance of overcoming and recognizing learning curves inherent to FBx to minimize patient discomfort and biopsy risk and improve the quality of care.
139
experience must be considered when interpreting Artemis FBx outcomes and particularly in the setting of active surveillance and focal therapy. Author contributions: Khushabu Kasabwala had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kasabwala, Patel, Khani, Margolis, Robinson, Hu. Acquisition of data: Kasabwala, Patel, Cricco-Lizza, Shimpi, Weng, Motanagh, Khani, Margolis, Robinson, Hu. Analysis and interpretation of data: Kasabwala, Patel, Cricco-Lizza, Shimpi, Weng, Motanagh, Wu, Khani, Margolis, Banerjee, Robinson, Hu. Drafting of the manuscript: Kasabwala, Patel, Cricco-Lizza, Shimpi, Weng, Motanagh, Wu, Khani, Margolis, Banerjee, Robinson, Hu. Critical revision of the manuscript for important intellectual content: Kasabwala, Patel, Wu, Margolis, Banerjee, Robinson, Hu. Statistical analysis: Kasabwala, Motanagh, Wu, Banerjee, Robinson. Obtaining funding: Hu. Administrative, technical, or material support: None. Supervision: Robinson, Hu. Other: None. Financial disclosures: Khushabu Kasabwala certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: Jim C. Hu is on a speaker’s bureau for Genomic Health. Daniel J. Margolis is an ad hoc consultant for Blue Earth Diagnostics. The remaining authors have nothing to disclose. Funding/Support and role of the sponsor: This study was supported by The Frederick J. and Theresa Dow Wallace Fund of the New York Community Trust. The sponsor played no direct role in the study. Acknowledgments: The authors are grateful to Rajesh Venkataraman for helping to prepare the manuscript.
References [1] Bjurlin MA, Wysock JS, Taneja SS. Optimization of prostate biopsy: review of technique and complications. Urol Clin North Am 2014;41:299–313. [2] Lecornet E, Ahmed HU, Hu Y, et al. The accuracy of different biopsy strategies for the detection of clinically important prostate cancer: a computer simulation. J Urol 2012;188:974–80. [3] Serefoglu EC, Altinova S, Ugras NS, Akincioglu E, Asil E, Balbay MD. How reliable is 12-core prostate biopsy procedure in the detection of prostate cancer? Can Urol Assoc J 2013;7:E293–8. [4] Puech P, Potiron E, Lemaitre L, et al. Dynamic contrast-enhancedmagnetic resonance imaging evaluation of intraprostatic prostate cancer: correlation with radical prostatectomy specimens. Urology 2009;74:1094–9. [5] Kvale R, Moller B, Wahlqvist R, et al. Concordance between Gleason scores of needle biopsies and radical prostatectomy specimens: a population-based study. BJU Int 2009;103:1647–54.
5.
Conclusions
[6] Cooperberg MR, Broering JM, Kantoff PW, Carroll PR. Contemporary trends in low risk prostate cancer: risk assessment and treatment. J
We demonstrated that there is a significant learning curve for FBx, specifically for accurate ROI targeting and systematic biopsy core yield. This has significant implications for accurate risk stratification and diagnosis. Provider
Urol 2007;178:S14–9. [7] de Rooij M, Hamoen EH, Witjes JA, Barentsz JO, Rovers MM. Accuracy of magnetic resonance imaging for local staging of prostate cancer: a diagnostic meta-analysis. Eur Urol 2016;70:233–45.
140
E U R O P E A N U R O L O GY O N C O L O GY 2 ( 2 019 ) 13 5 – 14 0
[8] Futterer JJ, Briganti A, De Visschere P, et al. Can clinically significant
[20] Mager R, Brandt MP, Borgmann H, Gust KM, Haferkamp A, Kurosch
prostate cancer be detected with multiparametric magnetic reso-
M. From novice to expert: analyzing the learning curve for MRI-
nance imaging?. A systematic review of the literature. Eur Urol
transrectal ultrasonography fusion-guided transrectal prostate bi-
2015;68:1045–53.
opsy. Int Urol Nephrol 2017;49:1537–44.
[9] Rosenkrantz AB, Verma S, Choyke P, et al. Prostate magnetic reso-
[21] Calio B, Sidana A, Sugano D, et al. Changes in prostate cancer
nance imaging and magnetic resonance imaging targeted biopsy in
detection rate of MRI-TRUS fusion vs systematic biopsy over time:
patients with a prior negative biopsy: a consensus statement by
evidence of a learning curve. Prostate Cancer Prostat Dis
AUA and SAR. J Urol 2016;196:1613–8.
2017;20:436–41.
[10] Mottet N, Bellmunt J, Bolla M, et al. EAU-ESTRO-SIOG guidelines on
[22] Patel N, Cricco-Lizza E, Kasabwala K, et al. The role of systematic and
prostate cancer. Part 1: screening, diagnosis, and local treatment
targeted biopsies in light of overlap on magnetic-resonance image
with curative intent. Eur Urol 2017;71:618–29. [11] Kasivisvanathan V, Rannikko AS, Borghi M, et al. MRI-targeted or standard biopsy for prostate-cancer diagnosis. N Engl J Med [12] National Comprehensive Cancer Network. Prostate cancer early (version
plications of prostate biopsy. Eur Urol 2013;64:876–92. [24] Chang E, Jones TA, Natarajan S, et al. Value of tracking biopsy in men
2018;378:1767–77. detection
ultrasound fusion biopsy. Eur Urol Oncol 2018;1:263–7. [23] Loeb S, Vellekoop A, Ahmed HU, et al. Systematic review of com-
2.2017).
www.nccn.org/professionals/
physician_gls/pdf/prostate_detection.pdf. [13] Marks L, Young S, Natarajan S. MRI–ultrasound fusion for guidance of targeted prostate biopsy. Curr Opin Urol 2013;23:43–50. [14] Sonn GA, Natarajan S, Margolis DJ, et al. Targeted biopsy in the detection of prostate cancer using an office based magnetic resonance ultrasound fusion device. J Urol 2013;189:86–91.
undergoing
active
surveillance
of
prostate
cancer.
J
Urol
2018;199:98–105. [25] Palapattu GS, Salami SS, Cani AK, et al. Molecular profiling to determine clonality of serial magnetic resonance imaging/ultrasound fusion biopsies from men on active surveillance for low-risk prostate cancer. Clin Cancer Res 2017;23:985–91. [26] Natarajan S, Raman S, Priester AM, et al. Focal laser ablation of prostate cancer: phase I clinical trial. J Urol 2016;196:68–75.
[15] Laviana AA, Ilg AM, Veruttipong D, et al. Utilizing time-driven activity-
[27] Quintal MM, Meirelles LR, Freitas LL, Magna LA, Ferreira U, Billis A.
based costing to understand the short- and long-term costs of treating
Various morphometric measurements of cancer extent on needle
localized, low-risk prostate cancer. Cancer 2016;122:447–55.
prostatic biopsies: which is predictive of pathologic stage and
[16] Ahmed HU, El-Shater Bosaily A, Brown LC, et al. Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study. Lancet 2017;389:815–22. [17] Weinreb JC, Barentsz JO, Choyke PL, et al. PI-RADS Prostate Imaging-
biochemical recurrence following radical prostatectomy? Int Urol Nephrol 2011;43:697–705. [28] Hong CW, Amalou H, Xu S, et al. Prostate biopsy for the interventional radiologist. J Vasc Interv Radiol 2014;25:675–84.
Reporting and Data System: 2015, version 2. Eur Urol 2016;69:16–40.
[29] Sonn GA, Fan RE, Ghanouni P, et al. Prostate magnetic resonance
[18] Borkowetz A, Platzek I, Toma M, et al. Comparison of systematic
imaging interpretation varies substantially across radiologists. Eur
transrectal biopsy to transperineal magnetic resonance imaging/
Urol Focus. In press. https://doi.org/10.1016/j.euf.2017.11.010.
ultrasound-fusion biopsy for the diagnosis of prostate cancer. BJU
[30] Majoros A, Szasz AM, Nyirady P, et al. The influence of expertise of
Int 2015;116:873–9. [19] Siddiqui MM, Rais-Bahrami S, Turkbey B, et al. Comparison of MR/ ultrasound fusion-guided biopsy with ultrasound-guided biopsy for the diagnosis of prostate cancer. JAMA 2015;313:390–7.
the surgical pathologist to undergrading, upgrading, and understaging of prostate cancer in patients undergoing subsequent radical prostatectomy. Int Urol Nephrol 2014;46:371–7.