Effect of information on prostate biopsy history on biopsy outcomes in the era of MRI-targeted biopsies

Purpose To describe the predictive value of information on previous benign biopsy for the outcome of MRI-targeted biopsies. Methods An exploratory analysis was conducted using data from a prospective, multicenter, paired diagnostic study of 532 men undergoing diagnostics for prostate cancer during 2016–2017. All men underwent 1.5 T MRI; systematic prostate biopsies; and MRI-targeted biopsies to MRI lesions with Prostate Imaging Reporting and Data System version 2, PI-RADS ≥ 3. The main outcome was numbers of detected prostate cancer characterized by grade group (GG) where GG ≥ 2 defined clinically significant cancer (csPCa). Results Men with previous biopsies had significantly more often negative MRI (26% vs. 17%, p < 0.05) compared to men without previous biopsies. Men with previous biopsies showed higher rates of benign biopsies (41% vs. 26%, p < 0.05) and lower rates of GG2 (17% vs. 30%, p < 0.05) and GG ≥ 3 (5% vs. 10%, p < 0.05) cancer. Biopsy-naïve men had higher proportions of highly suspicious MRI lesions (PIRADS 5; p < 0.05) and a higher proportion of significant cancer in those lesions (p = 0.05). In multivariate regression analysis, a previous benign prostate biopsy was associated with less than half the odds of csPCa (OR 0.38; 95% CI 0.20–0.71). Conclusion In this large prospective multicenter trial, we showed that men with a previous prostate biopsy had higher proportions of MRIs without lesions and lower proportion of highly suspicious lesions than biopsy-naïve men. Further, biopsy-naïve men showed higher detection of clinically significant cancer when using MRI-targeted biopsies. Also, in the era of MRI-targeted biopsy strategies, biopsy history should be carefully considered in biopsy decisions. Trial registration NCT02788825 (ClinicalTrials.gov). Date of registration June 2, 2016. Electronic supplementary material The online version of this article (10.1007/s00345-020-03277-x) contains supplementary material, which is available to authorized users.


Introduction
Several studies have shown that MRI-targeted biopsy is superior to systematic transrectal ultrasonography-guided biopsy in men at clinical risk for prostate cancer [1][2][3]. For example, the PRECISION trial reported that prostate biopsies targeted to MRI-visible lesions detected more clinically Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s0034 5-020-03277 -x) contains supplementary material, which is available to authorized users. 1 3 significant cancer, less non-significant cancer and would safely spare some men with MRI not suggestive of prostate cancer from biopsy as compared with standard biopsies [1].
Recently, it was further demonstrated that a combined procedure combining targeted and systematic biopsies in men with visible MRI lesions enhances cancer detection and decreases risk of disease misclassification [4]. This supports the statement in EAU guidelines that this combined approach might be a suggested clinical standard [5].
Published studies however lack evidence whether this novel strategy for detection holds for both biopsy-naïve men and men with a previous prostate biopsy with a benign result. We recently published results from the paired design STHLM3MRI study where men without prior diagnosis of prostate cancer scheduled for prostate biopsy underwent both magnetic resonance imaging, systematic and two to three targeted fusion biopsies to any significant lesion defined as PIRADSv2 ≥ 3 at any of the three Scandinavian sites [6]. We found that a combination of a blood-based risk prediction tool and MRI-targeted biopsies can be used to inform biopsy decision making.
The aim of this analysis was to perform an exploratory analysis in the above cohort to evaluate how information on previous prostate biopsies affects MRI-targeted prostate biopsy outcomes to provide support for clinical decision making.

Methods
This study describes an exploratory analysis using data from the previously published STHLM3 MRI Phase 1 study [6].

Design
The STHLM3 MRI Phase study is a prospective, multicenter, paired diagnostic study registered as NCT02788825 (ClinicalTrials.gov). As previously described, men without prostate cancer, aged 45-75 years and referred for prostate biopsies or prebiopsy MRI, were recruited and underwent MRI and a combined prostate biopsy procedure. Any previous biopsy (yes/no) was reported by patients at study inclusion.

MRI procedure
MRI was performed using a standardized, short, detection protocol compliant with European Society of Radiology Guidelines PI-RADS v2, except that dynamic contrast enhancement was omitted to decrease protocol complexity and acquisition time. We used 1.5 T magnetic field strength scanner at all study sites, without an endorectal coil. The protocol included T1-and T2-weighted images; diffusion-weighted images; b-values of 100, 450, and 800 with a calculated apparent diffusion coefficient map (ADC); and a separately acquired b1500. The time of acquisition was 16 min for the MRI sequences per participant.
MRI scans were reported according to the modified Prostate Imaging Reporting and Data System version 2 (following the PI-RADS v2 guidelines, with the modification that a PI-RADS 3 lesion in the peripheral zone may not be upgraded to PI-RADS 4 based on dynamic contrast) with up to three lesions with PI-RADS grade ≥ 3 marked for targeted biopsies. Any of six highly experienced uroradiologists reviewed the MRI series.

Biopsy procedure
All participants underwent 10-12 systematic biopsies and patients with PI-RADS ≥ 3 underwent a combined biopsy procedure with two to three targeted biopsies to any marked lesion, performed by the same urologist. The systematic biopsies were sampled from the base, midpart and apex of the peripheral zone according to the Swedish National Guidelines for Prostate Cancer. The targeted biopsy procedure was performed before the systematic biopsies and by the same biopsying urologist. Targeted biopsies were undertaken using the Koelis system (Koelis Inc.), Artemis system (Eigen Inc.), or BioJet system (D&K Technologies GmbH). The systematic biopsies were performed without considering the location of the MRI lesions. The pathological specimen was locally reviewed by experienced uropathologists (SiV Tønsberg, Unilabs Stockholm, Oslo University Hospital). The biopsy cores were formalin fixed using separate containers and graded according to the International Society of Urological Pathology (ISUP) 2014 modification.

Statistical methods
A PI-RADS score of ≥ 3 defined significant MRI lesions. The highest recorded ISUP Grade Group (GG) from MRItargeted and systematic biopsies in men with significant MRI lesions defined the primary outcome. Secondary outcome was the GG of MRI-targeted biopsies alone. Clinically significant cancer was defined as ISUP grade group ≥ 2 cancer. We report the biopsy outcomes stratified by PI-RADS score (≤ 2, 3, 4, 5). Proportions were compared using the McNemar test where p < 0.05 was considered statistically significant. We modelled the primary outcome in a univariate and a multivariate logistic regression including biopsy status (any previous prostate biopsy; yes/no), age (years), PSA (ng/ml), PSA density (ng/ml 2 ), PI-RADS score and pathological finding on digital rectal examination (DRE; yes/no) as predictors. We evaluated the receiver operating characteristics of the fitted logistic models calculating the area under the curve (AUC), compared using the DeLong method. All analyses were performed using Stata/MP 13.1 (Stata Corp., College Station, TX).

Cancer detection
There was a higher proportion of men with cancer findings among biopsy-naïve men than in men with a previous biopsy (57% vs. 33%, p < 0.05). Specifically, the proportion of men with clinically significant cancer was higher in biopsy-naïve men (33% vs. 22%; p < 0.05; Table 2). Using systematic biopsies as end point, biopsy-naïve men showed a higher proportion GG1, GG2-3 and GG ≥ 4 findings, but lower proportion benign biopsy results. The median length of any cancer findings in systematic biopsies were longer in biopsy-naïve men than in previously biopsied men (10 mm vs. 5 mm; p < 0.05). The median length for any cancer finding on targeted biopsies was 15 mm (IQR 8-22) in biopsynaïve men and 12 mm (IQR 4-22) in previously biopsied men (p = 0.2). Both when using systematic biopsies only and when omitting systematic biopsy results from the analysis, resembling the strategy of performing MRI-targeted biopsies only, we found a similar pattern (Table 2).
We then performed a multivariate regression model including biopsy status (yes/no) to assess any independent association of biopsy status with the risk of clinically significant cancer. Illustrated in Table 4, men with a previous biopsy show significantly lower odds of significant prostate cancer (OR 0.38; 95% CI 0.20-0.71) independent of the other clinical parameters. When using systematic biopsies, a previous biopsy was independently associated with a lower risk of significant cancer (OR 0.25; 95% CI 0.13-0.49).
Comparing AUC values for discrimination of csPCa, previous biopsy status added value when only PSA was considered. However, this was not conveyed in an increasing AUC when adding biopsy status to a full clinical model using age, digital rectal examination, PSA and PSA density ( Table 5). The ROC curves are displayed in Supplementary Fig. 1.
Finally, we assessed the added value of systematic biopsy in men with a significant lesion on MRI based on their biopsy status. Notably, omitting systematic biopsies in patients with a prior negative biopsy would have missed only two men (1.9%) with csPCa, but 18 (5.6%) in the biopsynaïve cohort (p = 0.07).

Discussion
In this exploratory analysis using data from a large prospective study of men undergoing MRI-targeted biopsies with or without the addition of systematic biopsies for prostate cancer detection, we demonstrate the added value of previous biopsy status for contemporary biopsy decisions. Our results show that men with a previous biopsy had significantly higher odds of MRI without significant lesions and lower odds of clinically significant prostate cancer, but no statistically significant difference in detection of non-significant prostate cancer. Correspondingly, biopsy-naïve men show higher odds of highly suspicious MRI lesions and also a higher frequency of significant cancer in such lesions. Adjusting for other clinical factors, we show that the odds of finding significant cancer in a man is more than halved if he has undergone a previous prostate biopsy.
Our results show that a history of a previous negative biopsy is strongly associated with lower risk of finding significant prostate cancer. Furthermore, our results show that systematic biopsies miss 20% of csPCa and that the added value of systematic biopsies in men with a previous biopsy are limited when performing MRI-targeted biopsies. In an  Multiple studies have shown that MRI-targeted biopsies more accurately detect clinically significant PC while also detecting less indolent PC, compared to systematic transrectal ultrasonography-guided biopsies in men at clinical risk for prostate cancer [1,2]. MRI and targeted biopsies have been extensively analyzed in the literature in distinction to different scenarios, as addition to or replacement for systematic biopsies, in the setting of a previous negative biopsy, as initial biopsy or during active surveillance. Regarding previous biopsy status, some studies have shown that MRI-targeted biopsies improve the detection rate in men with a previous biopsy compared to systematic biopsies, however, not in biopsy-naïve men [9,10]. Furthermore, existing randomized controlled trials, performed in biopsy-naïve men, have reported conflicting evidence in the detection rate of clinically significant prostate cancer evaluating MRI with MRI-targeted biopsies vs. systematic biopsies [1,[11][12][13]. Thus, EAU guidelines recommend the use of MRI and targeted biopsies, in the setting of persistent clinical suspicion of prostate cancer in men with a previous negative biopsy. If the MRI is negative and clinical suspicion of prostate cancer remains, guidelines suggest performing systematic biopsy based on shared decision making with the patient.
Is there any evidence to support whether men with normal MRI need systematic biopsies if they had a prior benign biopsy? In a prospective multicenter randomized trial, including 665 men with a prior negative systematic biopsy and a persistent suspicion of prostate cancer (by either suspicious DRE or PSA > 4.0 ng/ml), results show that only 1.3% of clinically significant prostate cancers would have been missed if systematic biopsies had been omitted [14]. The authors conclude that the value of repeat systematic biopsies is limited in this setting.
Our results show when comparing AUC values for discrimination of clinically significant prostate cancer, In a stratified analysis based on previous biopsy status, using MRI-targeted biopsies, men with a previous negative biopsy were 44% more likely to make the correct diagnosis of clinically significant prostate, buts 5% in biopsy-naïve men, compared to systematic biopsies [3]. Thus, current evidence indicates that knowledge of previous biopsy history should be included in biopsy decisions and prediction models for clinically significant prostate cancer. Our study has several strengths. First, this multi-center, prospective study using a paired design was specifically designed to study real-life performance of targeted biopsies. We use a structured and high-qualitative, short radiology protocol developed for early detection of prostate cancer and highly experienced uroradiologists to ensure high radiological quality. However, there are also obvious limitations. We present a pragmatic, exploratory analysis that was not pre-specified in the study protocol and conclusions are therefore primarily hypothesis generating. Second, we lack information on type and time span from the previous biopsy procedure which could impact the results. Third, the systematic biopsy was performed unblinded from the MRI results, possibly affecting the results of the systematic biopsies. Finally, although it has previously been shown that targeted biopsies decrease disease misclassification as compared with traditional biopsies [4], the true disease prevalence is unknown.

Conclusion
In this large prospective multicenter study, we showed that men with a previous prostate biopsy had higher proportions of MRIs without lesions and lower proportion of highly suspicious lesions than biopsy-naïve men. Further, biopsy-naïve men showed higher detection of clinically significant cancer when using MRI-targeted biopsies and the added value of systematic biopsies in these men was limited. Also, in the era of MRI-targeted biopsy strategies, biopsy history should be carefully considered in biopsy decisions. the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.