State of the Art in Prostate-specific Membrane Antigen–targeted Surgery—A Systematic Review

Take Home Message We found good evidence to suggest that prostate-specific membrane antigen targeting helps identify prostate cancer during surgery. The oncological benefits have yet to be investigated further.


Introduction
During both primary and salvage prostate cancer (PCa) surgery, identifying the target PCa tissue among the surrounding healthy tissue provides a key challenge [1]. Patients are significantly more likely to have biochemical recurrence (BCR) and undergo adjuvant or early salvage cancer treatment when tumor-containing tissue remains in situ following PCa surgery. In general, surgeons rely on experience, anatomical knowledge, and the ability to correctly interpret preoperative imaging to resect PCa tissue [2]. The use of intraoperative imaging and radioguidance helps better distinguish between cancerous and healthy tissue during surgery [3,4].
Prostate-specific membrane antigen (PSMA) is a transmembrane glycoprotein that is highly overexpressed in PCa cells and is used as the target for positron emission tomography (PET) imaging [5]. Owing to its high specificity, it is more accurate for nodal staging than magnetic resonance imaging (MRI), abdominal contrast-enhanced computed tomography (CT), or choline PET/CT, making its use increasingly common in staging of primary and recurrent PCa [5,6]. However, the technique is less reliable for identifying small lymph node metastases (micro metastases <3 mm), and the PSMA-PET tracers are typically excreted by the kidneys, making it difficult to locate the primary cancer site [5,7].
PSMA targeting has been proposed to extend beyond cancer diagnosis and into surgical guidance [7]. Multiple groups have explored a variety of tracer designs to realize this application, leading to the development of, for example, 99m Tc-PSMA-targeted radiotracers [8,9]-tracers that support noninvasive single photon emission computed tomography (SPECT)/CT, providing a surgical roadmap, as well as allow for intraoperative image guidance ( Fig. 1) [5].
As of today, different PSMA-targeting surgical approaches have been used in PCa patients. In order to provide a comprehensive overview of the techniques and initial outcome data, a systematic review of the available clinical literature was conducted.

Protocol registration and search strategy
The protocol was registered on PROSPERO (CRD42022304195) in January 2022. A systematic web search was conducted using MEDLINE (OvidSP), Embase.com, and Cochrane Library (Fig. 2). The search was last updated in August 2022. The search was executed with the help of an expert information specialist (S.v.d.M.) and checked by a second information specialist. Search terms can be found in the Supplementary material. Conference abstracts from Embase.com were removed based on their indexed publication type. Citation chasing was done by one person. No other methods to acquire additional reports and no other limits were used. The results were deduplicated in EndNote 20 using the method of Bramer et al. [10]. After removal of duplicates, two authors (A.C.B. and S.K.) screened all abstracts and reviewed the full-text reports for eligibility using Rayyan software [11]. Discrepancies were resolved through consensus or by consultation with a third author (G.M.). Data collection focused on demographics and surgical and oncological outcomes, and was collected by two authors in a prespecified form in Excel (Microsoft Corporation, Redmond, WA, USA). The review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [12].

Inclusion and exclusion criteria
As the first reports on the subject of PSMA targeting in surgery became available in 2015, the studies included in this review date from 2015 to 2022. Our review incorporated research that assessed the effectiveness of intraoperative PSMA-targeted surgical guidance in directing the surgeon toward the targeted tissue in vivo or that could confirm the target through an ex vivo analysis at the back table. Surgical procedures assisted by only preoperative PSMA PET for decision-making without intraoperative or back-table guidance were excluded. Only studies that were registered with a study protocol were considered prospective. Given the dynamic growth of PSMA-targeted surgery activities, case reports were included as well. Only original Englishlanguage literature full-text reports were considered. Pure preclinical work was excluded.

Assessment of risks of bias
The studies were defined according to the Idea, Development, Exploration, Assessment, Long-term (IDEAL) framework (Table 1) [13]. The risk of bias was assessed through the Risk Of Bias In Non-randomized Studies-of Interventions(ROBINS-I) tool ( Fig. 3) [14]. As the bias of case reports is not contributory, these were excluded from the assessment of the risk of bias.

Data analysis and objectives
Owing to the high heterogeneity found among the included studies in terms of different procedures, different reporting, and different definitions of outcomes, a meta-analysis was not possible. A comprehensive narrative synthesis of the included studies was performed. Descriptive statistics were used to summarize baseline characteristic data.

Study quality and baseline features
We included 29 reports on PSMA-targeted surgery (Fig. 2), of which eight were prospective studies [15][16][17][18][19][20][21][22], 12 retrospective analyses [4,[23][24][25][26][27][28][29][30][31][32][33], and nine case reports [34][35][36][37][38][39][40][41][42] Generally, a small number of patients were included: in eight of 29 (28%) reports 20 patients were included, of which two (retrospective) studies included >100 patients (7%). The remaining 72% of studies can be considered small-scale pilots or first-in-man reports (Fig. 4). In the primary setting (41% of reports), patients included were at an intermediate or a high risk with or without nodal involvement. In the salvage setting (45% of reports), the included patients had recurrent disease on PSMA PET/CT (in local recurrence a maximum of one lesion and in nodal recur-rence a maximum of five) and were eligible for surgery. The maximum standard uptake values of the PSMA-PET/ CT scans were not taken into account. A detailed description of patients' demographics can be found in Supplementary  Table 1. Seventeen studies (59%; including ten case reports) were considered stage 1 according to the IDEAL framework: proof of concept or first in man. These studies describe the clinical use of novel tracers or an adapted version of a previously studied tracer. Nine were considered exploring stage (stage 2a; 31%) and three were in developmental stage (stage 2b; 10%). None of the studies could be considered to be at stage 3 (assessment and long term; see Table 1) [13].

Preoperative features
Surgical resections were always guided by preoperative imaging roadmaps acquired >24 h prior to surgery (PSMA PET/CT 97% and [additional] PET/MRI 7% [28,40]). In general, the guidance provided by PSMA PET/CT was considered leading even in cases where a SPECT/CT scan was done (see below). Fifteen (52%) of the studies reported on the outcomes of PSMA PET/CT at lesion level. Out of 473 tumor-positive lesions at histopathology, PSMA PET/CT identified 382 (81%). The smallest lesion identified on PSMA PET was 3 mm in size [4]. For the most commonly used tracers ( 99m Tc-PSMA-I&S and 111 In-PSMA-I&T), the mean injection time to surgery was around 24 h (range 16-28 h), with a median injected activity varying between 541 and 638 MBq and between 140 and 150 MBq, respectively. Studies used 68 Ga-PSMA-11 injected between 76 and 127 MBq [16,20]. A detailed description of the timing and injected activity of all tracers can be found in Supplementary Table 2.
When 111 In-or 99m Tc-labeled PSMA tracers were used, an additional preoperative SPECT/CT scan was performed within 24 h before surgery (in 43% of the procedures with 111 In and in 93% with 99m Tc). Of the studies that included SPECT/CT, only seven reported their findings, indicating that 60 out of 133 histopathologically confirmed tumor-positive lesions (45%) could be identified by SPECT/CT. The only studies that described all lesions being identified on SPECT/CT were case reports with a maximum of two lesions on the PSMA PET/CT [37,38,42].

Radioguidance
Clinically, PSMA guidance is employed in two main approaches: radioguidance and optical guidance. The former is most frequently described (22/29 reports [76%]; ranging from one to 364 patients) and makes use of gamma/beta-emitting radioisotopes. Clinical use of gamma-emitting radioisotopes started with the use of 111 In-PSMA-I&T, followed not shortly by studies using 99m Tc-PSMA-I&S and 111 In-PSMA-617 [29,31,41]. In all these reports, guidance was facilitated by real-time tracing using a gamma probe. The design of the gamma probe varied depending on the surgical approach. A hand-held gamma probe for open surgery was used in 17/29 studies      (59%; ranging from one to 364 patients). One study reported the combined use with a probe-based freehand SPECT scan [29]. The development of a DROP-IN gamma probe design enabled the performance of the first robotic PSMAtargeted surgery, a technique that was used in 5/29 studies (17%; ranging from one to 20 patients). In one study, a robotic DROP-IN beta probe has been employed ex vivo to confirm the presence of 68 Ga-PSMA-11 in the prostate and nodal tissue [15].  Table 3) [40]. Figure 4 shows that in 2020 111 In-PSMA-617 ceases to be reported on and that new 99m Tc-based tracers such as 99m Tc-MIP-1404 are still being introduced.

Optical guidance
A different approach of PSMA guidance relies on converting the beta-emission of tracers such as 68 Ga-PSMA-11 into a secondary optical signal (Cerenkov Luminescence; k em max < 450 nm) that has a very limited degree of tissue penetration (<3 mm) but can be recorded in a back- An alternative approach to combining beta emissions and optical guidance makes the use of the so-called hybrid (radioactive and fluorescently labeled) PSMA tracers [34,36]-a strategy that was used in 2/29 (7%) studies (three patients). Eder et al. [36] reported the use of a PSMA-11derived hybrid molecule tracer, PSMA-914 ( 68 Ga-PSMA-914), using fluorescence imaging [44], while Aras et al. [34] used 18 F-BF3-Cy3-ACUPA in an ex vivo setting. Unfortunately, the dyes 800 CW (k em max = 789 nm [45]) and Cy3 (k em max = 565 nm) are not optimally compatible with the da Vinci Firefly endoscope (Intuitive Surgical Inc., Sunnyvale, CA, USA), which is designed for the detection of indocyanine green (k em max = 820 nm; Supplementary Table 2 and Supplementary Fig. 1). As a result, the fluorescence of 18 F-BF3-Cy3-ACUPA was imaged only ex vivo, for neither compound was investigated if radioguidance and optical guidance could complement each other.

3.5.
Tumor-to-background values A comparison between tracers or modalities was difficult as the in vivo as well as ex vivo reporting of tumor-tobackground ratio (TBR) was not consistent, and different cutoffs to consider a lesion positive were used [18,28].
Mostly this was defined as at least twice the background, with different definitions of the background (fatty tissue, lymph nodes, and psoas muscle). TBR ranged from 2.   [19] reported the only discrepancy with in vivo specificity of 69% (sensitivity of 76%; 99m Tc-PSMA-I&S; robotic procedure). A notable outlier was sensitivity of 50%, reported by Gandaglia et al. [18], in primary LND using 99m Tc-PSMA-I&S. The influence of the tracers and the approach (open/ robotic) on the accuracy could not be assessed given the different preoperative patient characteristics, differences in timing of injection, differences in location of the lesions, and differences in reporting.
In prostate-focused surgery, the accuracy of identifying positive surgical margins in vivo by radioguided surgery (RGS) was described only in one patient by Gondoputro et al. [19]. They successfully removed residual cancerous tissue but advised caution due to potential urinary contamination. The ex vivo evaluation of 68 Ga-PSMA-11 with the         DROP-IN beta probe correctly identified positive surgical margins in 40% (seven patients) [15]. Ex vivo Cerenkov imaging of 68 Ga-PSMA-11 yielded 60-83% (37 patients) agreement with histopathology [16,[20][21][22]. Eder et al. [36] and Aras et al. [34] describe the visibility of the fluorescent component of the tracer (in and ex vivo, respectively) but did not quantify their results.

(Oncological) outcomes
Between the studies, the evidence varied as shown in Table 1; no study exceeded IDEAL framework stage 2b ''development.'' Follow-up data after PSMA-targeted salvage surgery were analyzed retrospectively in ten studies (34%) reporting a median follow-up of 1-25.7 mo and were provided only for 99m Tc-PSMA-I&S and 111 In-PSMA-I&T. Nothing has yet been reported on the difference in oncological outcomes between these tracers.
A postoperative prostate-specific antigen (PSA) decline was reported in six of the 22 reports addressing salvage lymph node surgery [4,23,27,30]. The PSA decline ranged between 67% and 100% of the patients, and a decline of >90% was reported in 22-100% (Table 2), indicating variations in treatment effectiveness. Knipper et al. [4] were the first to look at the difference in PSA decline after conventional salvage LND versus the addition of PSMAtargeted radioguidance ( 99m Tc-PSMA-I&S, 42 patients) and found a significantly better outcome for the latter. However, long term follow-up data are missing.
BCR was reported in five of 22 studies regarding salvage surgery, ranging from 50% to 61.8% of the patients with follow-ups ranging from 10.3 to 25.7 mo. One study looked at BCR in the primary setting and found a BCR of 16.7% within a median time of 13 mo (IQR 4-22) [19]. Horn et al. [24] concluded from their BCR-free survival (BCRFS) data that patients with low preoperative PSA and a single lesion on preoperative PSMA PET benefitted most from PSMA-targeted surgery. A range of 0-100% of the patient population was in need of additional treatment after PSMA-targeted surgery within a median time ranging from 1 to 25.7 mo.
Complications were classified according to Clavien-Dindo [46]. Seven of the ten studies that reported on complications report on patients with a grade I/II complication, with a percentage ranging from 8.3% to 38.7%. Five studies observed complications of grade 3 (percentage ranging from 3% to 25%), and only three patients from the total population of all studies experienced a complication of grade IV [17,24,27]. None of the complications were ascribed to the tracer or image guidance procedure. The overall survival was again mentioned only in studies using 99m Tc-PSMA-I&S and 111 In-PSMA-I&T, and was 94.7-100% after a median follow-up ranging from 1 to 23.5 mo. Details concerning follow-up and outcomes can be found in Table 2.

Discussion
This systematic review summarized the existing literature on PSMA-targeted surgery in PCa patients. PSMA targeting provides a promising strategy to identify PCa both pre-and intraoperatively, and it seems that we have only just begun to find out what this technique can offer. Currently, the most widely implemented approach is PSMA-RGS using 99m Tc-PSMA-I&S (50% of the studies on this topic and the study with the largest number of patients [n = 364]). Studies that used a conventional open surgical approach were the majority (69%) in comparison with those using robotassisted surgery (31%).
Most studies present data on the value of PSMA-RGS in men with nodal recurrence (salvage surgery). One study suggested a benefit of PSMA-RGS versus conventional salvage LND [4]. Sensitivity for detecting nodal metastases during salvage surgery was dependent on the size and location of the lesion and ranged widely from 50% to 100% (eight trials). One of the most critical factors is selection of patients. Men with lower preoperative PSA and one lesion on imaging were most likely to benefit from RGS with a complete biochemical response rate of 45-66% in the largest series.
PSMA-PET imaging is known to be less dependable in identifying lesions under 3 mm [6,7] This corresponds to our findings for intraoperative detection, where the median size of metastases found in this review ranged from 2 to 9 mm. Smaller nodes (<3 mm) were most frequently missed [17,19]. The level of reporting of the correlation between RGS findings and histology in studies varied, with some studies reporting at the nodal level and others at the patient level. Moreover, whether an ex vivo analysis corresponded to intraoperative findings was not reported in all studies. This makes it difficult to compare detection rates among studies.
Tracer choice may also impact detection accuracy, a choice usually based on pharmacokinetics and pharmacodynamics as well as the sensitivity and specificity of the tracer. The studies with the largest patient groups (ranging from one to 364) included in these trials received RGS facilitated by 99m Tc-PSMA-I&S. The properties of 99m Tc-PSMA-I&S are well known, but this cannot be said for many of the other tracers, and comparative studies are lacking [47]. As 99m Tc-PSMA-I&S has successfully been used in both the primary and the salvage setting, it is currently the favored tracer.
The oncological benefit of PSMA-targeted surgery was studied mainly using a biochemical response as an endpoint. Looking at all the data combined, there was a large variation in PSA decline. Knipper et al. [26] showed that addition of PSMA guidance to conventional salvage LND improved PSA decline, suggesting that PSMA-RGS may improve outcome in men with recurrent nodal disease when compared with conventional surgery. Around 50% of salvage RGS patients were BCR free at a median follow-up of 13.2 mo. Hereby, a low preoperative PSA level and a single lesion on preoperative PSMA PET yielded better BCRFS after salvage RGS [24].
In six studies, PSMA-RGS was used in primary LND. Sensitivity and specificity for the detection of nodal metastases was comparable with the salvage setting, but considering the short reported follow-up, no other oncological outcome data are available. Optical cancer detection by Cerenkov or fluorescence imaging has been studied only in the primary setting. No study has proved the oncological value of PSMA-targeted surgical margin imaging. PSMA tracers that solely rely on fluorescence are in development [48,49]. However, preclinical or first-in-man data may not immediately translate to patient care. In fact, our search worryingly suggests that thus far only two optical (hybrid) tracers mentioned in preclinical reviews ( 68 Ga-PSMA-914 and 18 F-BF3-Cy3-ACUPA) were tested clinically in two case series [8,9,50]. This review is limited by the retrospective nature of the majority of the included studies with an unclear overlap in numbers of patients. The included studies had a high or an unclear risk of bias, were noncomparative, lacked a standardized way of reporting outcomes, and had short follow-ups. Furthermore, the sample sizes were generally small. A direct comparison of studies is also hampered by overlapping patient populations in several studies. Only one prospective first-in-man study that presented the proof-of-concept data contained 20 patients so far. Based on the results from this review, it is essential that further clinical trials are conducted based on standardized methodology and proper study endpoints. In addition, consensus on what PSMA-targeted surgery should provide for wider clinical implementation is desirable.

Conclusions
Consolidation of the existing literature on PSMA-targeted guidance during surgery in PCa indicates that the most common technique used is radioactive gamma-tracing in the open salvage setting. Techniques for use in robotic surgery and the addition of optical detection possibilities are in the pipeline. Intraoperative PSMA targeting has been proved to be technically sound, but no clear oncological benefit has yet been objectified. Lacking solid outcome data, currently PSMA-targeted surgical guidance should be considered an experimental treatment. Randomized controlled studies may be considered after consensus on the optimal surgical approaches and most valid clinical endpoints.
Author contributions: Henk G. van der Poel 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. Other: None.
Financial disclosures: Henk G. van der Poel certifies that all conflicts of interest, including specific financial interests and relationships and affili-ations 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: None.