18. Detecting, Localizing, and Treating the Multiparametric Magnetic Resonance Imaging Invisible Lesion: Utilizing Three-Dimensional Transperineal Mapping

Multiparametric magnetic resonance imaging (mpMRI) is quickly becoming the new “gold standard” to improve prostate cancer detection and to guide focal therapy. Combined with a targeted biopsy approach, it is beginning to replace transrectal ultrasound (TRUS)-guided prostate biopsy, which was introduced almost 30 years ago and had been the favored technique for the urologic community to diagnose prostate cancer [1]. Prostate-specific antigen (PSA) testing was also introduced almost simultaneously with the TRUS biopsy. PSA and TRUS biopsy combined with intense screening efforts tripled the detection rate of the disease and identified the majority of palpable and locally advanced prostate cancer lesions. In 1994 the Prostate Cancer Education Council (PCEC) reported the results of Prostate Cancer Awareness Week (PCAW) 1989–1992 [2]. A total of 14,900 men participated in the screening study in 1989, 150,000 in 1990, 400,000 in 1991, and >500,000 in 1992. The cancer detection rate by PSA rose from 4.98 % to 10 % in the first 2 years of the study and then fell to 6.48 % and 3.57 % for 1991 and 1992, respectively. Brawer reported a similar decrease from 2.6 % in year 1 to 2.0 % in year 2 and 1.8 % in year 3 [3]. The American Cancer Society (ACS) screening project also noted a decrease from 5.4 % to 1.0 % from the first to the third year of screening [4]. These efforts have resulted in a decrease of men presenting with an elevated PSA and a concomitant decrease in the volume of disease in the prostate. Today, most patients present with non-palpable disease (T1c).

When TRUS prostate biopsy was introduced, most prostate cancers were large and detectable by both digital rectal exam (DRE) and ultrasound. In addition, a positive diagnosis was readily available because these lesions were mostly peripheral and easily targeted by puncture through the anterior rectal wall. Over time, peripheral lesions have become smaller and more difficult to detect by TRUS and have in part been overshadowed by anterior lesions. Anteriorly located prostate cancer is not ideally suited for TRUS biopsy.

The decreasing volume of the lesions has also been associated with an increased interest in only treating the disease confined to the gland (pT2). A treatment methodology similar to that used for small-volume breast cancer, in essence the “male lumpectomy,” has been proposed by Onik [5]. However, unlike breast cancer, where the solitary lesion is easily imaged, prostate cancer presents with small, mostly ultrasound-invisible multifocal disease [6]. The inability to image and detect all prostate cancers within the gland has generated interest in finding different imaging technologies and limiting detection and treatment to only the potentially lethal prostate cancers, thus the interest in multiparametric magnetic resonance imaging (mpMRI), which can reliably detect the larger and higher-grade prostate cancer, often termed the “index lesions.” MRI-detected index lesions have also created interest in using focal ablative technologies to only treat these lesions as opposed to prostatectomy or whole-gland irradiation.

This chapter will make the argument that this approach may not be the best way to find and focally ablate isolated prostate cancer lesions. Multiparametric MRI, while an advance in prostate cancer detection compared to the “semi-blind” 12-core TRUS biopsy, is not ready to replace TRUS biopsy. We will review transperineal mapping biopsy (TPMB), which is a superior method to identify and select lesions for targeted focal therapy (TFT) because it improves intraprostatic staging. New technology, which corrects some of the deficiencies of the current TPMB technique, will also be discussed.

The enhanced detection rate of clinically significant prostate cancer by mpMRI-targeted biopsy compared to the systematic 12-core technique is an accepted fact. Unfortunately, it is also now recognized that MRI alone does not have the necessary sensitivity, requiring most centers to add 12-core systematic biopsy after performing the targeted biopsy. Mendhiratta et al. compared targeted mpMRI biopsy with the additional 12-core systematic biopsy in 452 men undergoing prostate cancer detection [7]. Systematic biopsy detected more prostate cancer than the targeted biopsy (49.2 % vs. 43.5 %, p = 0.006). However MRI did have the advantage in detecting more Gleason score (GS) 7 lesions than did the 12-core procedure (88.6 % vs. 77.3 %, p = 0.037). Nassiri and others recently reported the experience from the University of California, Los Angeles (UCLA), in more than 1200 targeted biopsies and noted that 15–30 % of “potentially important prostate cancers” are MRI invisible [8, 9].

Several review studies have been published evaluating the utility of mpMRI in detecting significant prostate cancer. Fütterer et al. evaluated 12 articles using prostatectomy data as the reference standard [10]. The negative predictive value for the exclusion of clinically significant disease ranged from 63 to 98 %. Part of the difficulty is analyzing data from different institutions is the lack of a common definition of significant disease. These included maximum cancer core length, grade at biopsy, number of positive cores, and PSA.

Wysock investigated the use of 12-core biopsy in men with negative 3 T mpMRI [11]. In the 75 men studied, the negative predictive value (NPV) for all cancers was 82 % and for GS >7 was 98 %. While this study might provide some reassurance that a negative mpMRI does not require biopsy confirmation of lack of significant disease, it should be recognized that the confirmation in the Wysock study relied on the same technology that is being replaced by MRI: the 12-core semi-blind procured from the 1980s. To overcome this limitation, several centers have compared the MRI-detected lesions to prostatectomy specimens. Le et al. investigated 122 men who had mpMRI with prostate cancer detected and compared them to whole-mount histopathology [12]. Overall mpMRI sensitivity for tumor detection was 47 % (132/283) with increased sensitivity for larger (102/141 [72 %] >1.0 cm) and higher-grade (96/134 [72 %] Gleason >7) and index tumors (98/122 [80 %]). One of the limitations in this study was that visual concordance was used by the uroradiologist and genitourinary pathologist to determine agreement between the MRI and prostatectomy specimens. The lack of digital co-registration adds additional error into their estimates of concordance.

Vargas et al. used the Prostate Imaging Reporting and Data System (PI-RADS) v2 for detection of clinically significant prostate cancer and compared the results in 150 prostatectomy specimens [13]. In this study the whole-mount specimens were digitized, but no information was provided about whether the images were co-registered to the MR images. PI-RADS correctly identified high-grade tumors >0.5 mL 94–95 % of the time. For lesions smaller than 0.5 mL, it was only successful in 20–26 %. It is important to recognize a lesion with a volume of 0.5 mL had a linear dimension of 8 mm in three planes (assuming a sphere). Missing high-grade lesions of this size in more than three-quarters of men who may be harboring them is worrisome.

It is interesting to note that both the ultrasound-guided TRUS biopsy and TPMB were described in 1987 [14]. TPMB has emerged as an alternative to TRUS biopsy, but due to lack of proper instrumentation, the need for anesthesia other than local has stymied its development. Nonetheless, many investigators have published their results and compared them to TRUS and targeted MRI biopsy results and to prostatectomy specimens. Symons et al. performed TPMB on 409 men and found prostate cancer in 208 (50.9 %) of which 75 % were GS >7 [15]. In a study of 431 radical prostatectomy (RP) specimens, of which prostate cancer was diagnosed by TRUS (283) or TPMB (184), those men who had the latter were more likely to be assigned the actual clinical risk category [16]. Serrao et al. performed transperineal-targeted MRI biopsy followed by 24–36 sectoral mapping biopsies [17]. MRI-positive scans (mean 1.57 lesions/patient, median 2) had positive pathology in 75 %. Of the 220 positive biopsies, 46 (20.9 %) were in areas determined falsely negative on MRI.

Sivaraman et al. investigated the use of TPMB in men with negative mpMRI [18]; 27/75 (36 %) had prostate cancer. The detection of clinically significant cancer (depending on the definition) ranged from 22.7 to 30.7 %. Toner performed a MEDLINE and PubMed database search comparing mpMRI with RP or TPMB histology [19]. The analysis found that compared with RP and TPMB specimens, the sensitivity of mpMRI for prostate cancer detection was 80–90 % and the specificity for suspicious lesions is between 50 % and 90 %.

While it is clear that TPMB is superior to TRUS and perhaps targeted mpMRI biopsy, it is not without challenges. There is no standardized protocol on how many biopsies to obtain. While most agree that the 5 mm external template should be used, how thoroughly to sample the gland remains undefined. Pham et al. performed an investigation using two TPMB approaches [20]. The biopsy technique was based on a 24-core template with 12 anterior and 12 posterior cores or a template based on gland volume with 1 core per cc (median 62 cores). No significant difference was noted in upgrading or complications between the two techniques. Valerio et al. utilized different mapping zones when performing a 20-core TPMB approach. Strategy 1 involved excluding the anterior areas of the prostate representing the transition area, but not the anterior horns, which were sampled within the lateral zones. Strategy 2 and strategy 3 involved a reduced sampling density from 5 to 10 mm by omitting intervening areas. Strategies 1, 2, and 3 had sensitivities of 78 % (95 % confidence interval [CI] 73–84 %), 85 % (95 % CI 80–90 %), and 84 % (95 % CI 79–89 %), respectively. The NPVs of the three strategies were 73 % (95 % CI 67–80 %), 80 % (95 % CI 74–86 %), and 79 % (95 % CI 72–84 %), respectively. The authors stated that altering the TPMB sampling strategy by preferential sampling of certain locations or reducing the sampling density led to significant reductions in the ability of the test to exclude clinically significant cancers.

The data presented herein has made it clear that the clinician cannot rely on MRI to exclude significant disease or to exclude disease in regions of the prostate not identified as regions of interest (ROI) for biopsy. However, even when an ROI is found to contain prostate cancer and targeted focal therapy is considered, how reliable is the MRI in demarcating the volume to be ablated? Cornud et al. evaluated 84 consecutive patients who underwent mpMRI before radical prostatectomy [21]. The volume of each suspicious area detected on magnetic resonance imaging and of all surgical histological foci was determined by planimetry. Histology revealed 99 significant tumors with a volume of greater than 0.2 cc and/or a Gleason score of greater than six. Of the tumors, 16 (16.2 %) were undetected by mpMRI. Linear regression analysis showed that tumor volume estimated by T2-weighted or diffusion-weighted imaging correlated significantly with pathological volume (r ² = 0.82 and 0.83, respectively). Nevertheless, diffusion-weighted imaging underestimated pathological volume in 43 of 87 cases (49 %) by a mean of 0.56 cc (range 0.005–2.84). Multiparametric and target volumes significantly overestimated pathological volume by a mean of 16 % and 44 %, with underestimation in 28 (32 %) and 15 cases (17 %), respectively. Volume underestimation was significantly higher for tumor foci less than 0.5 cc.

Failure to identify all significant lesions, incorrect estimation of tumor size, and inability of creating a sufficiently accurate treatment map are some of the reasons focal prostate ablation remains an investigational exercise. Other issues also compromise the successful introduction of a focal ablation program. While mpMRI can identify larger high-grade disease with high sensitivity, its ability to find GS 6 cancer is very limited. The elimination from consideration of identifying and potentially treating these lesions could compromise any focal therapy program. Haffner et al. used whole-genome sequencing and molecular pathological analyses to characterize the lethal cell clone in a patient who died of prostate cancer [22]. The lethal clone arose from a small, relatively low-grade cancer focus in the primary tumor, and not from the bulk, higher-grade primary cancer or from a lymph node metastasis resected at prostatectomy. In an opposing view editorial, Gonzalgo summarized many of the concerns surrounding focal therapy for prostate cancer. Treating only the index lesion may compromise cancer control because of inadequate treatment of potentially lethal disease [23]. There is significant interfocal tumor heterogeneity in prostate cancer and, from a molecular standpoint, metastasis may arise from secondary tumor foci. To date, there remains no good scientific evidence to support the fundamental basis of the index lesion theory. In fact, recent data suggest that more aggressive and potentially lethal disease can be found outside of the index lesion [24]. These and other concerns prompted the US Food and Drug Administration (FDA) to hold a workshop at the 2015 American Urological Association (AUA), which culminated in a publication in 2016 stating: “The general consensus was that currently available technologies are capable of selective ablation with reasonable accuracy, but that criteria for patient selection remain debatable, and long term cancer control remains to be established in properly designed and well-performed prospective clinical trials. Concerns include the potential for excessive, unnecessary use in patients with low risk cancer and, conversely, that current diagnostic techniques may underestimate the extent and aggressiveness of some cancers, leading to inadequate treatment” [25].

After nearly three decades of performing TRUS-guided biopsy, the diagnosis and treatment of prostate cancer has become very challenging. TRUS-guided biopsies identify many cancers that do not need to be treated while missing potentially lethal disease. Men who go on active surveillance are not appropriate candidates because they were incorrectly identified as low risk based on the TRUS biopsy results. In order to improve the thorough sampling, Crawford described the use of computer simulation to map prostate transperineal biopsies [26]. This innovative approach could record lesion location but was limited because biopsy sites were annotated off-line. Stone, working with interactive software, developed a real-time computer-simulated prostate brachytherapy program [27]. 3DBiopsy, Inc., was formed in 2012 with the intention of refining the TPMB technique so its biopsy plan accurately finds all the disease within the prostate. The clinician would then decide what therapy is most appropriate and, in the case of focal therapy, which lesions (some or all) would need to be treated. In men with truly very-low-risk prostate cancer, active surveillance would become accurate surveillance as future biopsies, and frequent PSA testing could be eliminated.

In order to accomplish this goal, several components needed to be changed. First is a requirement that a biopsy plan needs to be created that had a high probability of sampling all lesions within the gland, regardless of Gleason score. Kepner discussed an approach to distributing transperineal prostate biopsy cores that yields data on the volume of a tumor that might be present when the biopsy is negative [28]. If the biopsy sites are parallel and spaced 5 mm apart and a 15-gauge biopsy needle is used, then the theoretical probability of encountering a lesion of 2.5 mm radius is 98 % (Fig. 18.1).

In order to put this approach into practice, three devices needed to be developed: First, a software program that generates a biopsy plan based on a reconstructed three-dimensional (3D) model of the prostate generated from the transrectal ultrasound images. Second, a variable biopsy needle apparatus that samples the prostate from apex to base as one core. And third, a pathology carrier system where the physician places the specimen after biopsy that preserves the integrity of the core and allows the pathologist to render a diagnosis that included location and length of cancer on the specimen. The results of these three are then integrated into a precise 3D model of the prostate and the areas of cancer.

The biopsy needle and actuator (gun) being used for TRUS biopsy (targeted or systematic) is the same one used for TPMB (including the aforementioned study). This needle, developed 30 years ago, is no longer appropriate for prostate biopsy, regardless of the approach. A properly designed biopsy needle would have no minimal deflection (entry and end points are in the same plane), and the amount of tissue collected is consistent with the amount required (a 4 cm specimen length should be 4 cm). The biopsy needles used today suffer from significant deflection and marked core inconsistency. Nobody really cared previously because the focus was to make a diagnosis of cancer. Today with lesions substantially smaller (and in most cases microscopic) and when trying to reach an anterior ROI, needle deflection and core integrity are more critical. Brede and Jones examined needle defection using a 5 mm grid and found only 22 % shallow and 7 % deep precision [29]. The Bard MaxCore biopsy needle (C.R. Bard, Inc., Covington, Georgia, USA) was used for this study, which has a bevel or lancet tip design.

Satasivam et al. examined fragmentation of specimen cores taken by TRUS biopsy [30]. Although the biopsy needle type was not specified, substantial fragmentation was noted with the standard swipe technique, and core length and specimen length varied from 12.4 to 13.4 mm (73–79 % of full length). Öbek et al. performed a similar study and noted mean core length in patients with prostate cancer was 12.3 mm (72 %) vs. 11.4 mm (66 %) in those without (p = 0.015). Patients with a core length greater than 11.9 mm were 2.57 times more likely to be diagnosed with cancer [31].

In order to overcome these limitations, a new needle design and actuator have been under development. The needle tip was changed from lancet to trocar with various designs tested in gelatin matrix simulating prostate density (Fig. 18.6).

The Bard needle deflected a median of 0.9 mm (range 0.0–1.3), while the trocar tip needles deflected a median of 0 mm (range 0–1.7 mm, p < 0.001). Needle size (15-G vs. 18-G) did not affect deflection. Obtaining full intended length core length is also a necessary component of the needle design. When trying to obtain the entire length of the prostate in one sample, if the software designates 4 cm and only 2.8 cm is delivered, the remaining 1.2 cm is questionable. Testing of different needle designs demonstrated the 15-G trocar tip needle depicted in the top of Fig. 18.6 performed no better than the Bard needle. However, the addition of ridges to the core bed, which secures the specimen as the cutting portion of the cannula passes over it, resulted in a 92 % core consistency rate (over various lengths between 2 and 6 cm).

The new variable-length needle would not be useful without a companion actuator to fire it a variable distance. This device has also been developed and tested. The physician can dial the desired length of the specimen as dictated by the software. Each specimen length will vary depending on the length of the prostate when the probe is obliquely rotated away from the midline of the prostate. This actuator combined with the new needle would also be ideal for targeted MRI biopsies. When anterior lesions are present, the user can puncture the posterior capsule of the gland, dial the distance to the anterior capsule, and fire the gun to take a full-length specimen. The specimen will not only contain the entire cancer but also the noncancerous tissue on either side of it. Planning focal therapy with this additional information would be invaluable.

The process to handle tissue specimens has not substantially changed in more than 100 years. Take a specimen, drop it in formalin, and send it to the lab is the routine. In the past, there was not much concern because clinicians were only interested in knowing whether cancer was on the specimen and what type it was. The new biopsy system necessitates a substantial change in the carrier mechanism. The longer (up to 6 cm) cores need to be preserved intact so the pathologist can render an accurate diagnosis on cancer location and length on every positive core. In addition, staff have to manage an increased number of specimens (16–20 for MRI-targeted biopsy and 30 for the new 3DBiopsy). Specimen providence errors are not uncommon, and handling each core, in the OR and pathology lab, increases the risk of fragmentation [30]. Wojno et al. estimated a 2.5 % specimen providence error rate for prostate biopsies costing $879,900,000 per year in the USA [32]. Most of these cases were 12-core systematic TRUS biopsies. Increasing the number of specimens will likely increase the costs of managing these errors.

The solution to these problems was to develop a carrier system that secures the specimen, delivers it to the technicians intact, and eliminates the need to remove the specimen from the carrier. The specimen remains in the carrier through formalin fixation, processing, paraffin embedding, microtome sectioning, staining, and reading. The device is also under development (Fig. 18.7).

The diagnosis and management of prostate cancer is evolving rapidly. The current focus is to biopsy men at risk of harboring significant cancers. In most other cancers that we treat, whether it be lung, breast, or colorectal cancers, we stage them. We employ positron emission tomography (PET) scans, computed tomography (CT) scans, bone scans, and others to determine the extent of disease. In localized prostate cancer, these tests do not help. We want to know what is in the gland and where. Currently TRUS and even saturation biopsies do not help. Multiparametric MRI is much more helpful; however, as already reviewed, it can miss up to 20–30 % of potentially progressive and lethal cancers.

Having precise knowledge of the 3D location of the cancers is clearly going to change how we manage these patients and thereby reduce the current dominant roles of both radical prostatectomy and radiation therapy. Precise intraprostatic staging will afford some patients the opportunity to be followed without treatment because the fear of missing a lethal cancer will be greatly diminished. A large group of newly diagnosed men, perhaps as many as 1 in 3, will be candidates for targeted focal therapy because clinicians will have precise knowledge of the location of all lesions and be able to deliver the ablative treatment to the exact sites. Whether MRI or an improved biopsy strategy, as depicted here, will be the best means to achieve these results will need to be determined by prospective clinical trials.