Skip to main content
main-content
Top

09-07-2018 | Prostate cancer | Article

Targets missed: predictors of MRI-targeted biopsy failing to accurately localize prostate cancer found on systematic biopsy

Journal:
Prostate Cancer and Prostatic Diseases

Authors: Michael Austin Coker, Zachary A. Glaser, Jennifer B. Gordetsky, John V. Thomas, Soroush Rais-Bahrami

Publisher: Nature Publishing Group UK

Abstract

Background

Magnetic resonance imaging (MRI)/ultrasound (US) fusion-guided biopsy has improved the ability to localize and detect prostate cancer (PCa) with efficiency surpassing systematic biopsy. Nevertheless, some patients have PCa missed using the MRI-targeted biopsy sampling alone. We aim to identify clinical and imaging parameters associated with cases where targeted biopsy did not detect PCa compared to systematic biopsy.

Methods

We conducted a retrospective review of patients who underwent MRI/US fusion-guided biopsy in addition to concurrent systematic, extended-sextant biopsy between 2014 and 2017. For patients with PCa detected on systematic biopsy not properly localized by MRI/US fusion-guided biopsy, the sextant distance from MRI-targeted lesion to the cancer-positive sextant was calculated and parameters potentially predicting this targeting miss were evaluated.

Results

In all, 35/127 (27.6%) patients with single-session MRI/US fusion-guided biopsy plus standard biopsy finding PCa had lesions incorrectly localized. Of these, 15/35 (42.9%) were identified as possible fusion-software misregistrations. The remainder, 12/35 (34.3%), represented targeted biopsies one sextant away from the cancer focus and 8/35 (22.9%) targeted biopsies two sextants away from the cancer focus. Only 7/35 (20.0%) patients were determined to have clinically significant PCa, which represents 7/127 (5.5%) of the overall population. Lower MRI lesion volumes (p = 0.022), lesion density (p < 0.001), and PI-RADS scores (p < 0.001) were significantly associated with targeted biopsy missing PCa detected on systematic biopsy.

Conclusion

Clinically significant PCa is rarely missed utilizing MRI/US fusion-guided biopsy. With the majority of missed tumors representing targeting misregistrations or cases of low-grade cancer in sextants immediately adjacent to MRI suspicious lesions. Lower MRI lesion volumes, lesion density, and PI-RADS are predictors of cases with targeted biopsies missing cancer, for which systematic sampling of the sextants containing MRI targets and adjacent sextants would most optimize PCa detection.

Introduction

Prostate cancer (PCa) has the highest incidence among solid-organ malignancies in American men, accounting for nearly one in five new cancer diagnoses [1]. The current diagnostic method for patients who are suspected to have PCa is a systematic, extended-sextant biopsy. The majority of these men present with elevated serum prostate-specific antigen (PSA) or abnormal findings on digital rectal examination. PCa has traditionally been diagnosed by systematically sampling cores of prostatic tissue from the different sextant regions of the gland rather than sampling any particular areas based on imaging suspicion for harboring cancer. The conventional approach of systematic transrectal ultrasound (TRUS)-guided biopsy results in a high degree of false-negative biopsies from missing occult cancer foci as well as undergrading PCa.
The advent of advanced imaging with multiparametric magnetic resonance imaging (MP-MRI) has allowed for improved localization of PCa foci and permits the targeting of biopsies from areas suspicious for malignant tissue [2, 3]. Current improvements in imaging modalities have made it possible to electronically superimpose MRI images with real-time TRUS images, permitting the utilization of MRI/ultrasound (US) fusion-guided prostate biopsy [24]. Prior trials suggest that targeted biopsy is superior to systematic biopsy alone for the detection of clinically significant PCa [5, 6]. Despite this finding, there are still a small subset of patients who undergo concurrent standard biopsy with targeted biopsy and are determined to have clinically significant PCa on the standard biopsy alone, with benign prostatic tissue reported in the targeted biopsy. The purpose of this study was to identify potential explanations for this phenomenon, as well as ascertain possible mechanisms by which targeted biopsy fails to detect clinically significant PCa in efforts to improve the process and optimize cancer detection in the future.

Methods

Patient subjects

An institutional review board-approved, HIPPA-compliant retrospective review of prostate MRI/US fusion-guided biopsy records from January 2014 to April 2017 was performed. During this time period, 413 men with no prior PCa treatment or other prostate surgeries underwent MP-MRI and subsequent MRI/US fusion-guided targeted biopsies of one or more MRI regions of interest suspected to harbor PCa. Of these, 262 patients included in the study underwent concurrent standard extended-sextant core biopsy, as well as targeted biopsy sampling of MRI suspicious lesions. Additionally, for patients with multiple biopsy sessions, only the initial biopsy session results were included in this study. Exclusion of patients with serial fusion biopsy resulted in a patient cohort of 198 patients. After exclusion of patients with benign pathology on both MRI/US fusion-guided biopsy and concurrent systematic biopsy cores, a final study cohort of 127 patients was established for the analyses performed. Of this cohort, 35 patients were determined to have had fusion biopsy cores composed of only benign prostatic tissue while systematic biopsy found PCa. Figure 1 summarizes the inclusion and exclusion criteria as applied to select the MRI/US fusion-guided biopsy patient records for inclusion into this study analysis.

Imaging and biopsy protocols

Our institutional MP-MRI prostate protocol included a triplanar T2-weighted, axial diffusion-weighted, and axial dynamic contrast-enhanced pelvic MRI, along with axial post-contrast abdominopelvic T1-weighted MRI. All scans were performed on 3 T MRI systems unless the patient had a history of hip prosthesis, pacemaker, or other indication to undergo imaging on a lower field strength 1.5 T magnet at our center. Each MRI was prospectively reviewed and a consensus read was defined in the setting of a multidisciplinary prostate imaging conference attended by fellowship-trained abdominal radiologists and urologic oncologists with experience in prostate MRI and targeted biopsy, respectively, as previously described [7]. In this setting, MRI lesions with cancer suspicion were segmented by urologic oncologists in three dimensions and assigned a Prostate Imaging Reporting and Data System (PI-RADS) v2 score by the group using DynaCad, a post-image processing software (Philips/InVivo, Gainesville, FL, USA). All MRI studies read prior to implementation of the second version of PI-RADS were retrospectively reread for assignment of a PI-RADSv2 score to suspicious lesions with readers blinded to biopsy pathology results.
Those patients with suspicious lesions on MP-MRI underwent MRI/US fusion-guided targeted biopsy using the UroNav software fusion biopsy system (Philips/InVivo) sampling at least two cores from each biopsy-targeted lesion as previously recommended [8]. Target biopsies were subsequently followed by the systematic 12-core, extended-sextant biopsy sampling during the same biopsy procedure.

Data analyses

The designation of “target miss” was given to those cases in which targeted biopsy cores failed to detect malignancy, while PCa was found on one or more of the systematic biopsy cores. Furthermore, the sextant distance from the targeted biopsy sextant to nearest focus of malignant tissue detected by standard biopsy was calculated using the mapping system depicted in Fig. 2. The nomenclature of “misregistration” was provided to those lesions in which the targeted biopsy failed to detect cancer and the systematic core biopsy positive for cancer were located in the same sextant of the prostate gland. In these cases, it is likely that the MRI appropriately detected the tumor but the imaging modalities were not appropriately matched based on anatomic location during the MRI/US fusion biopsy session.
Subsequently, a retrospective review of patient records was performed for patients with concurrent targeted biopsy and systematic core biopsy documenting patient demographics, number of targeted biopsy lesions, serum PSA level at time of biopsy (ng/mL), PSA density (ng/mL2), prostate volume measured by US (mL), prostate volume measured by MRI (mL), total MRI-based lesion volume (mL), lesion density (defined as total MRI-based lesion volume divided by prostate gland volume measured on MRI), and PI-RADS score for those patient populations with and without malignant prostatic tissue present in the targeted biopsy.
Pathologic analysis of the biopsy tissues for this study were all performed by a single fellowship-trained genitourinary pathologist following the recommendations of the most recent World Health Organization regarding PCa reporting, as previously described [9]. For this study, clinically significant PCa is defined as any case with PCa risk greater than the National Comprehensive Cancer Network (low-risk category, which is defined as: clinical staging T1–T2a; PSA <10 ng/mL; PSA density <0.15 ng/mL/g; and Gleason score ≤3 + 3 = 6 (grade group 1)) PCas.
The reported statistical significance levels were all two-sided, with the predetermined threshold of statistical significance set to a p-value of <0.05.

Results

A total of 413 patients were identified to have undergone MRI/US fusion-guided biopsy without history of prior prostate surgery or PCa treatments. Furthermore, 262 (63.4%) of these patients underwent concurrent systematic, extended-sextant biopsy. With the exclusion of patients who underwent serial fusion biopsy sessions, 198 (47.9%) patients were determined to fit the inclusion. Further exclusion of patients with benign pathology on both MRI/US fusion-guided biopsy as well as systematic biopsy resulted in a study cohort of 127 (30.8%) patients available for analysis for this study. The mean age of patients with PCa found in targeted biopsy was 65 years compared to 64.1 years for the “target miss” patient cohort (p = 0.592). Pre-biopsy PSA for the cohort of patients with PCa found on targeted biopsy cores was 8.22 ng/mL, compared to 8.99 ng/mL for the “target miss” patient cohort (p = 0.618). Table 1 demonstrates the patient demographic and associated clinical data for the 127 patients with targeted and concurrent standard biopsy with findings of PCa on biopsy pathology.
Table 1
Demographic, clinical, and pathologic features of patients undergoing single-session MRI/US fusion-guided biopsy and standard-of-care 12-core extended-sextant TRUS-guided biopsy found to have PCa
 
Total patients (n = 127)
Patients with PCa found in target (n = 92)
Patients with “target miss” (n = 35)
p-Value
Race
AA = 25 (19.7%)White = 91 (71.7%)Other = 11 (8.7%)
AA = 15 (16.3%)White = 70 (76.1%)Other = 7 (7.6%)
AA = 10 (28.6%)White = 21 (60.0%)Other = 4 (11.4%)
0.194
Age (y)
64 ± 7.45
65 ± 8.05
64.1 ± 6.17
0.592
Number of targeted biopsies
2.00 ± 0.96
2.10 ± 0.95
2.26 ± 1.29
0.448
PSA at time of biopsy (ng/mL)
8.43 ± 7.80
8.22 ± 5.37
8.99 ± 12.16
0.618
PSAD (ng/mL2)
0.22 ± 0.22
0.23 ± 0.21
0.20 ± 0.26
0.497
TRUS volume (mL)
46.57 ± 23.05
45.04 ± 24.54
50.57 ± 18.29
0.229
MRI volume (mL)
52.19 ± 27.40
50.38 ± 27.32
56.95 ± 27.43
0.329
Total lesion volume (mL)
1.73 ± 1.22
1.88 ± 1.29
1.33 ± 0.90
0.022
Lesion density
0.037 ± 0.024
0.041 ± 0.025
0.025 ± 0.015
<0.001
PI-RADS
4.00 ± 0.79
4.23 ± 0.70
3.40 ± 0.69
<0.001
AA African American, MRI magnetic resonance imaging, PCa prostate cancer, PI-RADS Prostate Imaging Reporting and Data System, PSA prostate-specific antigen, PSAD prostate-specific antigen density, TRUS transrectal ultrasound
The total MRI lesion volume, lesion density, and PI-RADS score were prognosticators of a missed targeted biopsy on univariate analysis. The average total MRI-based lesion volume was 1.88 mL for the PCa-positive target biopsy cohort, and 1.33 mL for the “target miss” cohort (p = 0.022). Lesion density for the PCa-positive targeted biopsy cohort was 0.041 vs. 0.025 for the “target miss” cohort (p < 0.001). A mean PI-RADS score of 4.23 was determined for the PCa-positive target biopsy cohort vs. 3.40 for the “target miss” cohort (p < 0.001). Total prostate volume as measured by TRUS and MRI was not statistically significantly different when comparing the PCa-positive targeted biopsy cohort vs. the “target miss” cohort (Table 1).
The mapping of sextant distance from MRI-targeted lesion to the nearest cancer-positive systematically sampled sextant was analyzed for the 35 patients where the target biopsy sampling failed to detect malignant prostatic tissue. Figure 2 depicts the sextant prostate map used to determine the distance systematically sampled sextants with PCa were separated from the targeted biopsy lesions. The majority of these cases, 15/35 (42.9%), had targeted lesion biopsies located in the same sextant where the systematic biopsy core(s) localized malignant prostate tissue. These cases were thereby defined as “misregistration.” Twelve of 35 (39.3%) targeted areas were measured one sextant adjacent to the area with malignancy on standard biopsy. A smaller subset, 8/35 (22.9%), had benign targeted biopsies two sextants from the positive systematic biopsy. No patients included in the study had a negative targeted biopsy found to be three sextants from a positive systematic biopsy (Table 2).
Table 2
Sextant distance between MRI target lesion and systematic biopsy sextant found to contain PCa
 
n (%)
Total
35
Same sextant/fusion misregistration
15 (42.9)
1
12 (34.3)
2
8 (22.9)
3
0 (0)
The 35 patients with targeted biopsy that missed cancer were further categorized by the modern prognostic grade group system for reporting PCa pathology [10]. The majority of these patients, 28/35 (77.7%), had grade group 1 tumors. However, there were a small number of patients with missed cancer in targeted biopsy cores that were found to have clinically significant PCa on systematic biopsy: 6/35 (16.6%) patients had grade group 2 tumors and 1/35 (2.8%) patient had a grade group 3 PCa. No patients with a benign result on targeted biopsy were determined to have high-grade PCa (grade group ≥ 4) discovered on systematic biopsy (Table 3).
Table 3
Prostate cancer pathology in patients with “target miss” results
 
n (%)
Total
35
GG 1
28 (77.7)
GG 2
6 (16.6)
GG 3
1 (2.8)
GG 4
0 (0)
GG grade group
The seven patients with PCa missed via targeted biopsy who had clinically significant disease (grade group > 1) were further analyzed by prostate sextant mapping. Four of the seven (57.1%) patients had lesions one sextant away from the MRI-targeted lesion detected via systematic biopsy approach. One of the seven (14.3%) patients had PCa detected two sextants away. Two of the seven (28.6%) patients had possible software misregistrations with cancer found in the same sextant as the MRI-targeted lesion.

Discussion

In modern clinical practice, MP-MRI and subsequent MRI/US fusion-guided prostate biopsy are methods used for the optimized detection, staging, and active surveillance of PCa [1113]. MRI/US fusion technology affords improved visualization of areas suspicious for cancer and aids in detecting PCa foci occult on systematic biopsy [11, 14]. Furthermore, multiple investigations have demonstrated that performing MRI-targeted biopsy superiorly detects clinically significant PCa, and less often samples insignificant cancers compared to a systematic extended-sextant biopsy. Thus, MRI-targeted biopsy would theoretically spare both clinicians and patients from detection of indolent, clinically insignificant cancers and the associated downstream psychological effects and public health costs [9, 11, 15, 16]. With this improved diagnostic accuracy, evidence suggests that selected men on active surveillance meeting certain clinical criteria may safely forgo repeat biopsy for at least 2-year intervals when incorporating MP-MRI in the surveillance algorithm [12, 17]. This would thereby decrease the incidence of iatrogenic complications from repeat biopsies, lower cost of PCa care, and reduce the psychological burden on a patient who would otherwise be undergoing more frequent biopsy procedures.
However, using MRI-targeted biopsy alone for the detection of new clinically significant PCas remains controversial. For men on active surveillance (AS) who are undergoing MRI-targeted biopsy, several studies have reported an optimized detection rate of clinically significant cancer detection using MRI-targeted biopsy [12, 15]. However, Marliere et al. [18] observed improved cancer detection when a concurrent standard template biopsy is additionally performed. A recent multi-institutional, international randomized trial supported MRI-targeted biopsy without systematic sampling performed better than a systematic 10- to 12-core TRUS-guided biopsy schema in detecting clinically significant PCa [19].
In the present study, we observed the majority (77.7%) of the “target miss” biopsies identified were histologically determined to be grade group 1 PCa. Only seven patients had “target miss” biopsies with a higher grade group 2 or grade group 3 PCa. These cases of clinically significant, intermediate-risk PCa are essential to detect early in the disease course to allow for a discussion regarding definitive treatment with curative intent given the recognized heightened risk for stage progression. The goal of PCa screening has appropriately shifted toward early detection of clinically significant cancers with limiting the detection and potential overtreatment in cases of more indolent disease. Hence, optimizing the detection of higher grades of PCa is critical for which MRI has augmented screening efficacy [20].
The PI-RADS suspicion score is generated using a 5-point scale, with a score of 5 being the most concerning for a malignant tumor, and has become an increasingly utilized tool to describe suspicious lesions detected on MRI of the prostate [21, 22]. We observed the PI-RADS score demonstrated statistical significance for the comparative determination of fusion-guided biopsy that resulted in PCa detection vs. cases of “target miss” disease where systematic biopsies detected cancer (p < 0.001). Furthermore, total lesion volume (p = 0.022) and lesion density (p < 0.001) met criteria for statistical significance in our study. Due to limited statistical power, multivariable analyses were not performed, but we acknowledge that lesion volume, and in turn, lesion density may have some overlap with the PI-RADS score as greatest axial dimension of targetable lesions is incorporated into the PI-RADSv2 scoring system. It is acknowledged that not all MRI lesions with PI-RADS suspicion scores assigned would represent PCa foci. Hence, we enriched our population for those with PCa proven to be present on systematic or targeted biopsy core sampling for internal comparison for each patient included in the analysis.
Muthigi et al. [23] described a retrospective review of MRI/US fusion-guided biopsy in relation to the upgrading of high-risk prostatic disease uncovered by standard biopsy to lower-risk disease discovered through targeted biopsy. Their multivariable analysis found upgrading of disease to be significant relative to PSA (p < 0.001), prostate gland volume (p < 0.001), and a lower number of target cores (p = 0.001). Based on these findings, predictors of missing targets may be used in the future to guide those cases best served by added systematic biopsy sampling to optimize detection of all clinically significant cases of cancer while obviating the need for always conducting the full systematic extended-sextant schema of needle sampling [24].
To determine the distance of a “target miss” lesion from a positive cancer sextant sampled with the systematic biopsy approach, a sextant mapping guide was used for spatial comparison (Fig. 2). The majority of the “target miss” biopsies fell either in the same sextant or within one sextant of the malignant lesion: 77.2%. This suggests that the targeted and systematic template cores may have sampled the same relative area but yielded different pathological results. The underlying reason for this finding is not fully elucidated in the present study and is most likely multifactorial. Tumor heterogeneity, with varying grades of cancer in a single lesion and even detection of atypical small acinar proliferation, which is a benign finding often found adjacent to cancer foci, is a well-substantiated finding in PCa. In addition, MRI/US fusion-guided biopsy is subject to potential error in the multiple steps required to achieve a precise biopsy of an MRI target, including MRI segmentation, TRUS segmentation, fusion misregistration, and patient anatomic differences (bladder filling, rectal contents, and prostate gland deformation with TRUS). Furthermore, it is well recognized that deployment of the TRUS-guided needle biopsy is not always achieving biopsy in the sextant pursued in the systematic biopsy approach. This may further complicate the baseline comparison, though the MRI-US software fusion devices allow for three-dimensional mapping of systematic biopsy core samples for retrospective evaluation of spatial deployment within the three-dimensional prostate gland volume. Additionally, patient mobility during targeted sampling was not specifically collected as a parameter in our dataset, but has been reported as a known risk of altering lesion location and configuration during fusion alignment between real-time TRUS and the previously acquired diagnostic MP-MRI [23, 25]. Based on the findings from the current study, perhaps systematic sampling may be of higher yield sampling within the sextant MRI-targeted lesions are found or in immediately adjacent sextants within the prostate rather than a full extended-sextant sampling throughout the gland.
Strengths of this study include a well-corroborated and standardized procedure for the diagnostic MP-MR imaging, MRI/US fusion-guidance for targeted biopsy, and extended-sextant mapping for the special localization of PCa foci. The most current MRI suspicion scoring system, PI-RADSv2, and the newly adopted PCa grading system using prognostic grade groups were utilized to characterize malignant lesions in this study. Despite these advances in the established workflow used, 35 out of 127 patients (27.6%) who underwent targeted biopsy were determined to have negative targeted lesions with cancer found via the systematic approach. MRI/US fusion-guided biopsy shows promise to continue to be an effective method for the localization of suspicious prostatic tissue lesions, but has room for improvement. As the majority of “target miss” cases had PCas found in the same sextant or immediately adjacent sextant to the targeted lesion(s), potential elimination of systematic sampling in other, more distant sextants may be considered in an effort to decrease the burden of biopsy sampling.
This study is limited in being a single-institution patient cohort with overall patient numbers limiting statistical power. While all biopsies were performed by two experienced, fellowship-trained urologic oncologists at a tertiary care center, this may not reflect urologic practice in academic centers newly adopting these techniques or the community practice setting at-large. Furthermore, this study made the assumption that each biopsy collected an accurate representation of the prostatic tissue histology while it is recognized that systematic biopsy does not always represent the highest-grade disease when radical prostatectomy specimens have been evaluated. Though this is a limitation, evaluation of targeted biopsy and systematic biopsy on the subset of patients who have gone on to radical prostatectomy would significantly limit the number of patients evaluated and would also incorporate a selection bias that all patients would have PCa that prompted early definitive treatment. Future directions of validation via a larger, multi-institution patient cohort may build upon these findings and better define the systematic biopsy schema needed to supplement targeted biopsy sampling in order to optimize detection of clinically significant cancers without oversampling.

Conclusion

Clinically significant PCa is rarely missed utilizing MRI/US fusion-guided biopsy. With the majority of missed tumors representing targeting misregistrations or cases of low-grade cancer in sextants immediately adjacent to MRI suspicious lesions. Lower MRI lesion volumes, lesion density, and PI-RADS are predictors of cases with targeted biopsies missing cancer, for which systematic sampling of the sextants containing MRI targets and adjacent sextants would most optimize PCa detection.

Compliance with ethical standards

Conflict of interest

SR-B serves as a consultant for Philips/InVivo Corp. The remaining authors declare that they have no conflict of interest.
Literature

Related topics