Evaluating the Learning Curve for In-Office Freehand Cognitive Fusion Transperineal Prostate Biopsy

Abstract

Objective:

To define the learning curve of the in-office, freehand MRI-ultrasound cognitive fusion transperineal prostate biopsy (CTPB) by assessing cancer detection, biopsy core quantity and quality, procedure times, and complications over the initial experience.

Methods:

We reviewed 110 consecutive CTPB performed March 2021-September 2022 by a urologist inexperienced with the PrecisionPoint^® platform. The study period was divided into quarters to assess for temporal variation in outcomes. Univariable and multivariable analysis modeled the learning curve.

Results:

Across quarters, there were no differences in the detection of clinically significant prostate cancer (Q1:50%, Q2:52%, Q3:50%, Q4:48%, p>0.9) or Gleason grade group upgrading by targeted versus systematic biopsy (p=0.6). Median procedure times improved with experience (Q1:17min, Q2:14min, Q3:12min, Q4:13min, p=0.018). On multivariable analysis, procedure times decreased by one minute per 20 cases (p<0.001). On linear regression, CTPB procedure times approximated transrectal biopsy times after 90 cases (p<0.001). The histopathologic core quality did not differ, as evidenced by consistent core length (p=0.13) and presence of minimal fibromuscular tissue (p>0.9).The most common complications, hematuria and hematospermia, were similar across quarters (p=0.7, p=0.3, respectively). There was a single episode of urinary retention and no reported infections.

Conclusions:

There is no evidence of a learning curve for CTPB as shown by consistent clinically significant prostate cancer detection, high quality biopsy cores, and low complications. However, CTPB procedural times begin to approximate cognitive targeted transrectal biopsy times after 90 cases.

Introduction

The transrectal ultrasound-guided prostate biopsy is the predominant approach for the evaluation of prostate cancer (PC) in the United States.¹ However, transrectal biopsy needles traverse the rectal mucosa, potentially seeding infectious bacteria into the urinary tract. Despite antibiotic prophylaxis, the rate of infectious complications after transrectal biopsy has risen to 7%, with up to 3% requiring hospitalization.^2,3

As an alternative to the transrectal approach, the transperineal biopsy needle samples the prostate through the skin, avoiding the rectal flora and eliminating the need for antibiotic prophylaxis.^4,5 Transperineal biopsy also improves PC detection within the anterior zone of the prostate, which is challenging to sample from a transrectal approach in men with benign prostatic hyperplasia.⁶ The transition to transperineal biopsy may address unmet needs in racial disparities. African American men are twice as likely to harbor anterior tumors and are more than twice as likely to die from PC than Caucasian men.^7,8 Transperineal biopsy with a brachytherapy grid was historically performed under general anesthesia due to pain of multiple skin punctures, but the procedure is well tolerated in the office setting with local anesthesia, facilitated by freehand techniques such as the needle-stabilizing PrecisionPoint^® Transperineal Access System (Perineologic, Cumberland, MD).⁹ This tool couples the ultrasound probe and needle cannula in place, allowing the surgeon to complete the biopsy with one skin puncture per prostate lobe.

Over the last decade, prebiopsy multiparametric Magnetic Resonance Imaging (MRI) has increased the accuracy of both transrectal and transperineal biopsy by providing clearer anatomical guidance and/or visualization of target lesions. Currently, the American Urological Association (AUA) and the European Association of Urology (EAU) recommend MRI prior to prostate biopsy, but no U.S. consensus exists regarding the competing approaches of MRI-ultrasound guided fusion biopsy.^10,11

With the integration of MRI into routine prostate cancer evaluation, in-office freehand MRI-ultrasound cognitive fusion transperineal biopsy (CTPB) may serve as the highest value-based biopsy method. However, the learning curve for CTPB remains unknown. A learning curve, or the number of cases required to reach comfort and proficiency with a procedure, estimates the experience needed to safety and reliably adopt new skills. While each procedure has its own set of relevant measures, surgical learning curves typically trace patient outcomes and procedural efficiency.¹² For example, the learning curve of MRI-ultrasound software transrectal biopsy demonstrates that a user’s accuracy may improve up to the 98^th case.¹³ These outcomes may contextualize negative biopsy findings in the setting of high-risk imaging results (Prostate Imaging-Reporting and Data System version 2.1 [PI-RADS v2.1] score ≥3) and inform the decision for re-biopsy or earlier follow-up. Although proponents of the transrectal approach cite higher rates of urinary retention and longer procedural times with transperineal biopsy,¹⁴ there is an evidence gap to demonstrate whether these differences may become clinically insignificant with experience. Finally, published CTPB aggregate outcomes may not portray new user outcomes during adoption.¹³ Therefore, our objective was to evaluate the learning curve of CTPB by assessing cancer detection, procedural times, complications, and pathologic biopsy core characteristics over time.

Methods

Patient Selection & Biopsy

We performed a retrospective study of the first 110 consecutive men who underwent CTPB using the PrecisionPoint^® device between March 2021 and September 2022. At the time of this study, the urologist (JCH) performing all CTPB had no prior experience with this method. However, they previously performed hundreds of transrectal biopsies, which they continued to offer during the course of this study. No antibiotic prophylaxis or bowel preparation (e.g.rectal enema, polyethylene glycol solution) was prescribed prior to CTPB. Demographic and radiographic characteristics were collected, including age, body mass index (BMI), pre-biopsy prostate-specific antigen (PSA), race/ethnicity, history of previous prostate biopsy, prostate volume, PI-RADS score, and primary region of interest (ROI) size. Clinically significant prostate cancer (csPC) was defined as Gleason Grade Group (GG) ≥2.

Inclusion criteria for biopsy included first time, repeat after previous negative, and active surveillance of PI-RADS ≥3 lesions or PSA density ≥ 0.15. Biopsies performed after focal therapy were excluded as the treatment effects alter the anatomy of the prostate. CTPB comprised 12 systematic cores and targeted cores (2–3 per target), except in 8 men who underwent systematic biopsy alone for PSA density ≥ 0.15 in the absence of suspicious lesions on MRI. There were no formal exclusion criteria as all patients were offered both biopsy options, of which some men preferred one approach over the other. The only factor affecting biopsy allocation for the provider was schedule availability in the clinic for CTPB based on resource availability. Total time spent by the patient in the procedure room (room time) and total time the physician was in the room, including time for anesthetic block placement, (procedure time) were recorded during the study period for 87 of the 102 systematic and targeted CTPB. Fifteen patients were excluded due to missing or incomplete time records. For comparison, room and procedure times were also captured for 43 transrectal biopsies with cognitive MRI guidance conducted by the same surgeon during the second half of the study period. CTPB pathology reports were evaluated to capture the number of fibromuscular cores per biopsy, presence and/or GG of prostate cancer, and GG upgrading by targeted biopsy in comparison to systematic biopsy.

Patients were queried about post-biopsy complications at their follow-up appointment, and responses were stratified by the Clavien-Dindo Classification.¹⁵ The study period was divided into quarters to assess changes over time.

Pathology

The first and last twenty available patients of the study period were selected for evaluation of biopsy quality. Two experienced genitourinary pathologists assessed a total of 221 cores for total core length and lengths of peripheral zone, transition zone, and fibromuscular tissue. The number of fibromuscular cores was assessed to evaluate the proficiency of prostate sampling.

Statistical Analysis

Descriptive statistics were generated to evaluate outcomes per quarter. For continuous variables of patient, biopsy and pathological characteristics, the difference in medians between quarters was calculated using the Kruskall-Wallis test. Categorical variables were analyzed using Chi-Square or Fisher’s exact tests of independence. Linear regression compared the procedure time of transperineal biopsy to the transrectal control. A multivariable regression model including age, BMI, initial PSA, prostate volume, and primary ROI size assessed the variation of room and procedure time with respect to sequential case number. Piecewise linear regression models were fitted to delineate the effect of surgeon experience on procedural efficiency and the total number of cores collected per biopsy. A piecewise linear regression models two connected lines, each with an independent slope and intercept, that quantify the learning curve for their respective time segments. The two lines meet at a breakpoint, determined by the minimized mean-squared error, which indicates a locus of change in learning curve. All p<0.05 were considered statistically significant, and statistical analysis was performed with SAS v9.4 (Cary, North Carolina).

Results

Biopsy

There were no significant differences in patient demographic and MRI characteristics across the study population (Table 1). Detection of overall PC and csPC in the first quarter was 61% and 50%, respectively, and remained comparable throughout the four quarters (Table 2). The GG upgrading by targeted versus systematic biopsy was also similiar between quarters (Table 2, Supplementary Table 1).

Table 1:

Baseline clinical characteristics across the four quarters of the study period.

Median Characteristics [IQR]	1st Quarter (No. = 28)	2nd Quarter (No. = 27)	3rd Quarter (No. = 28)	4th Quarter (No. = 27)	p-value
Age	68 [59–72]	61 [57–67]	66 [60–72]	67 [56–72]	0.12
BMI, kg/m2	25.6 [23.7–28.6]	27 [25.3–28.6]	26.7 [23.6–31.8]	26.3 [24–31.5]	0.5
PSA, ng/mL*	6.5 [4.3–9.5]	5.8 [4.4–7.4]	5.1 [3.6–7.3]	6.8 [4.6–8.2]	0.3
Prostate volume, mL	53.0 [33.9–82.0]	39.0 [25.9–52.0]	40.8 [31.9–54.8]	39.0 [36.0–55.8]	0.3
Primary ROI size, mm**	11.0 [6.6–15.6]	10.4 [8.4–12.8]	11.0 [7.8–16.8]	11.6 [7.8–15.7]	0.8
No. (%)
Race/Ethnicity
White	15 (54)	14 (52)	20 (71)	14 (52)	0.19
Black	2 (7.1)	3 (11)	3 (11)	3 (11)	--
Asian	2 (7.1)	3 (11)	4 (14)	1 (3.7)	--
Declined or Other	9 (32)	7 (26)	1 (3.6)	9 (33)	--
Underwent previous prostate biopsy	8 (29)	6 (22)	9 (32)	7 (26)	0.9
PI-RADS score**
3	9 (33)	7 (30)	7 (25)	7 (29)	0.8
4	13 (48)	12 (52)	12 (43)	13 (54)	--
5	5 (19)	4 (17)	9 (32)	4 (17)	--
Biopsy type
Targeted & systematic	27 (96)	23 (85)	28 (100)	24 (89)	0.092
Systematic only	1 (4)	4 (15)	0 (0)	3 (11)	--

BMI = body mass index; PSA = prostate-specific antigen; PI-RADS = Prostate Imaging Reporting & Data System; ROI = region of interest.

^*Two patients missing value.

^**Only 102 targeted & systematic biopsy patients had MRI imaging findings that were included in these variables.

Table 2:

Biopsy characteristics and post-biopsy complications across the four quarters of the study period.

Median Biopsy Characteristics [IQR]	1st Quarter (No. = 28)	2nd Quarter (No. = 27)	3rd Quarter (No. = 28)	4th Quarter (No. = 27)	p-value
Room time, min*	48 [41–54]	45 [39–50]	42 [36–50]	47 [38–50]	0.075
Procedure time, min*	17 [11–19]	14 [12–15]	12 [10–16]	13 [10–15]	0.018
No. of cores per biopsy	18 [16–20]	16 [14–18]	16 [15–20]	15 [14–17]	0.009
Biopsy Results, No. (%)
All prostate cancer	17 (61)	19 (70)	20 (71)	18 (67)	0.8
Clinically significant prostate cancer	14 (50)	14 (52)	14 (50)	13 (48)	>0.9
Upgrading by targeted biopsy**	3 (11)	2 (8.7)	3 (11)	5 (21)	0.6
> 1 Fibromuscular core per biopsy	8 (29)	6 (22)	6 (21)	4 (15)	0.7
> 2 Fibromuscular cores per biopsy	5 (18)	1 (3.7)	3 (11)	4 (15)	0. 4
Complications, No. (%)
Hematuria	12 (43)	12 (44)	11 (39)	15 (56)	0.7
Hematospermia	6 (21)	12 (44)	9 (32)	7 (26)	0.3
Urinary retention	0	0	1 (3.6)	0	>0.9
Lower urinary tract symptoms	0	0	1 (3.6)	0	>0.9
Unplanned visit to the emergency room or clinic	2 (7.1)	0	3 (11)	2 (7.4)	0.5

^*Times captured for 87 systematic & targeted biopsies across the four quarters.

^**Variable and percentiles only counted for 102 systematic & targeted biopsies.

Procedure times significantly declined throughout the duration of the study (Table 2). The median procedure time of targeted and systematic CTPB decreased from 17 min (Interquartile Range [IQR]: 11–19) in the first quarter to 13 min (IQR: 10–15) in the fourth quarter (p=0.018). However, the median room time did not differ across the study period. In comparison, cognitive targeted transrectal biopsies were found to have median procedure and room times of 11 min (IQR: 8–13) and 38 min (IQR: 32–42), respectively. On univariable linear regression, targeted and systematic CTPB procedure times gradually approached the median transrectal procedure time towards the end of the study period (correlation: −0.38, 95% Confidence Interval [95% CI]: −0.55- −0.18, p<0.001; Figure 1). The median number of cores collected per biopsy decreased significantly across the quarters (Q1: 18, Q2: 16, Q3: 16, Q4: 15, p=0.009). However, the presence of more than one or two fibromuscular cores per biopsy did not vary over the study period (Table 2).

Figure 1:

Linear regression line (blue) of CTPB procedure time (black circles; correlation: −0.38, 95% Confidence Interval [95% CI]: −0.55- −0.18, p<0.001) compared to the median transrectal biopsy procedure time (black line) of 11 min.

On multivariable linear regression, procedure time decreased by approximately one minute per 20 consecutive cases (coefficient: −1.08, 95% CI: −1.72- −0.45, p<0.001; Supplementary Table 2). However, room time did not significantly change with increasing case number in this multivariable model (Supplementary Table 2). Moreover, piecewise linear regression did not demonstrate a significant breakpoint in total core number, room time, nor procedure time with respect to increasing case number (Supplementary Table 3).

Complications

The incidence of reported seven-day complications after CTPB did not vary (Table 2). The most common complications in each quarter were hematuria and hematospermia, and all cases resolved without intervention. One case of urinary retention was reported in the third quarter and resolved after catheterization. One patient reported lower urinary tract symptoms with negative urine cultures in the third quarter that resolved with empiric antibiotic treatment. All complications were Clavien-Dindo ≤ Grade I. No culture-proven urinary tract infection, septic events, or other serious complications, including hospitalizations, were identified during the study period. Seven men reported an unplanned visit to the clinic or emergency room, with no difference in visits across quarters. These visits were due to the previously stated episodes of urinary retention and lower urinary tract symptoms, as well as subjective fever, testicular swelling, hematuria and hematospermia.

Pathology

Of the twenty biopsies from the first and fourth quarters, there were no differences in pathologic characteristics of biopsies from both groups (Table 3). The median lengths of the total core, peripheral zone, transition zone, and fibromuscular tissue did not differ. Likewise, biopsy cores with ≤5 mm of peripheral zone tissue or with >5 mm of fibromuscular tissue remained consistent across the quarters.

Table 3:

Pathological characteristics of 221 biopsy cores of 10 biopsies selected from each the first and fourth quarters.

Median Pathologic Characteristics [IQR]	1st Quarter (Core No. = 104)	4th Quarter (Core No. = 117)	p-value
Total length of each core, mm	10 [6–12]	10 [8–12]	0.13
Length of peripheral zone, per core, mm	10 [6–12]	9 [4–12]	0.15
Length of transition zone, mm	0 [0–0]	0 [0–0]	0.6
Length of fibromuscular tissue, mm	0 [0–0]	0 [0–0]	0.5
No. (%)
Cores with ≤5mm of peripheral zone tissue	25 (24)	35 (30)	0.3
Cores with >5mm of fibromuscular tissue	1 (1.0)	2 (1.7)	>0.9

Discussion

This is the first study to evaluate the learning curve of CTPB, which is of great interest to urologists adopting this technique across the United States and Europe as our findings gauge potential improvement in patient outcomes and procedural efficiency with user experience. Of significance, we demonstrate the absence of a significant learning curve for csPC detection or upgrading by targeted biopsy. The first quarter detection of csPC was 50%, which is comparable to the 43% CTPB detection by others using the PrecisionPoint^® device.⁹

We demonstrate significant improvements in the technical efficiency of CTPB as procedure times decreased by 24% during the study and approached the median cognitive targeted transrectal biopsy time beginning at 90 cases. In comparison to a brachytherapy stepper grid, the PrecisionPoint^® procedure time is almost 50% shorter.¹⁶ Moreover, as no plateau was reached on the initial linear regression of 110 CTPB procedure times, there is potential for this procedure to become as, if not more efficient than a transrectal biopsy. Given that procedure time is defined by provider presence in the room, procedural efficiency frees up time for the provider to see additional patients or perform other duties, enhancing overall clinic throughput and revenue. In regard to room time, the fourth quarter times were 9 minutes longer for CTPB vs transrectal biopsy, suggesting longer setup time due to leg stirrups or technician unfamiliarity with the setup/tools required. It is likely that this time difference in setup could eventually be decreased with experienced staff and optimized rooms.

The total number of collected biopsy cores decreased from the first to the second quarter, reflecting a gradual shift in sampling style or increased confidence per core collected. This carries limited clinical relevance as templates vary widely in their number of systematic and targeted cores despite similar cancer detection rates.¹⁷

There was no evidence of a significant learning curve associated with the quality of cores across time, including the number of cores that missed the prostate (fibromuscular cores). Moreover, pathologic review confirmed that cores were comprised of predominantly peripheral zone tissue. This is noteworthy as the CTPB needle must pass through a greater distance of fascia, subcutaneous fat, and pelvic floor muscle prior to reaching the prostate as compared with transrectal biopsy. The greater distance may contribute to increased needle deflection,^18,19 yet our cores were consistently collected from the intended prostate zone without involvement of extraprostatic tissue Thus, a urologist without prior CTPB experience should not experience a prolonged learning curve in targeting the correct regions of the prostate.

Complications reported after CTPB were similar to other prior series.^9,20 All complications were minor (Clavien-Dindo Class I) and did not change in incidence during the initial experience with CTPB. Specifically, urinary retention remained uncommon throughout the study (0.9%) and was similar to the transrectal biopsy rate at our institution (1.6%). There were no infectious complications in the absence of antibiotic prophylaxis, which contrasts with the reported 3.1% readmission rate for infections after transrectal biopsy.³ Given that each episode of sepsis admission costs $8,672 to $19,100 USD, the transition from transrectal to transperineal biopsy may save $623 million USD annually in the United States alone.^3,21 The publicly-funded National Healthcare Service of the United Kingdom already performs majority transperineal biopsies, where decreased hospitalization costs with the transperineal approach are estimated to save taxpayers £556 GBP per admission, or £2.1 million GBP annually.^22,23 These readmissions savings alone will likely offset the costs of prolonged room time and initial staff training required to adopt CTPB, although further cost-effectiveness research is warranted. Moreover, our elimination of routine antibiotic use with transperineal biopsy aligns with accountable antibiotic stewardship, and others have found no difference in infectious complications with or without antibiotics for this clean, percutaneous approach.⁵ Eliminating antibiotic prophylaxis use also saves medication costs and limits antimicrobial resistance potential side-effects.

Although both the AUA and the EAU recommend the routine use of MRI prior to biopsy, provider unfamiliarity and cost concerns are two key barriers to incorporating this novel clean, percutaneous CTPB approach. Spiraling United States healthcare costs have focused interest on value-based care, which aims to maximize health outcomes while minimizing the associated cost burden. The transperineal approach raises the value of biopsy care as it virtually eliminates the risks and costs of infectious complications, such as hospitalizations. Nevertheless, some providers may be hesitant to adopt targeted transperineal biopsy because software-fusion platforms require an investment of >$120,000 USD without a corresponding increase in reimbursement. The effectiveness of software-fusion platforms may be further hindered by large prostate size and pubic arch interference, whereas CTPB remains unconstrained by such technological limitations.²⁴ Additionally, CTPB may be more cost-efficient as each procedure only requires the purchase of the single-use PrecisionPoint^® device (~$250 USD) that may be reimbursed by the payer.²⁵ Even in the public healthcare model of the United Kingdom, each in-office transperineal biopsy using the PrecisionPoint^® device is comparable in cost to the transrectal biopsy (£375 vs £332 GBP). Further reduction in CTPB device costs is expected with the emergence of cost-conscious competitors of the PrecisionPoint^® device (e.g. CamPROBE^®, JEB Technologies, Suffolk, UK; Surefire^®, Advance Medical Designs, Marietta, GA).^22,26,27

Our findings must be interpreted in the context of the study design. First, we did not directly measure the precision of the cognitive targeted biopsy to target the ROI, as can be tracked with software fusion.¹³ Despite this limitation, our analysis of CTPB used measures of cancer detection and clinicopathologic outcomes to estimate user accuracy over time. Second, this study analyzed only the learning curve of a single urologist using PrecisionPoint^® who has over a decade of experience with various methods of prostate biopsy. Our results may have limited generalizability to newer providers or other freehand or grid-based transperineal biopsy platforms. Third, the absence of a breakpoint on piecewise linear regression indicates that the CTPB procedural times may continue to decrease past the duration of the study period, but further improvement may be slower and carry less clinical relevance. Fourth, the outcomes of the biopsy are also dependent on the expertise of the radiologist and the pathologist, both of whom have been shown to undergo a learning curve in PC tumor mapping, detection and staging.^28,29 Our center has fellowship trained expert genitourinary radiologists with extensive experience interpreting prostate MRIs, but a recent study demonstrated high variability in the positive predictive value of PI-RADS scores across expert centers, such as a 27%–48% range for scores ≥3.³⁰ Finally, the superiority of CTPB over transrectal biopsy cannot be confirmed without high-quality randomized clinical trials to evaluate both clinical and patient-reported outcomes. Cost-effectiveness analyses are also warranted to review the wider financial sequelae of CTPB adoption.

Conclusions

Our novel findings demonstrate a lack of a measurable learning curve with CTPB regarding detection of clinically significant prostate cancer, collection of high quality cores, and complications. However, there is a learning curve of of approximately 90 cases to approach transrectal biopsy times, without the need for antibiotic prophylaxis. We demonstrate that CTPB may be incorporated into practice without compromising outcomes during the learning curve.