Theranostics of Primary Prostate Cancer: Beyond PSMA and GRP-R

Simple Summary The accurate assessment of the aggressiveness and localization of primary prostate cancer lesions are essential for treatment decision making. Around 15% of lesions are missed by PSMA Positron-Emission tomography/computed Tomography (PET/CT). The aim of our study was to investigate the potential of novel surface markers to detect PSMA-negative lesions using immunohistochemistry and autoradiography techniques. Our work demonstrates that targeting both PSMA and neurotensin receptors might detect all intra-prostatic lesions. This new finding has implications for the future theranostics of primary prostate cancer. Abstract The imaging of Prostate-Specific Membrane Antigen (PSMA) is now widely used at the initial staging of prostate cancers in patients with a high metastatic risk. However, its ability to detect low-grade tumor lesions is not optimal. Methods: First, we prospectively performed neurotensin receptor-1 (NTS1) IHC in a series of patients receiving both [68Ga]Ga-PSMA-617 and [68Ga]Ga-RM2 before prostatectomy. In this series, PSMA and GRP-R IHC were also available (n = 16). Next, we aimed at confirming the PSMA/GRP-R/NTS1 expression profile by retrospective autoradiography (n = 46) using a specific radiopharmaceuticals study and also aimed to decipher the expression of less-investigated targets such as NTS2, SST2 and CXCR4. Results: In the IHC study, all samples with negative PSMA staining (two patients with ISUP 2 and one with ISUP 3) were strongly positive for NTS1 staining. No samples were negative for all three stains—for PSMA, GRP-R or NTS1. In the autoradiography study, binding of [111In]In-PSMA-617 was high in all ISUP groups. However, some samples did not bind or bound weakly to [111In]In-PSMA-617 (9%). In these cases, binding of [111n]In-JMV 6659 and [111In]In-JMV 7488 towards NTS1 and NTS2 was high. Conclusions: Targeting PSMA and NTS1/NTS2 could allow for the detection of all intraprostatic lesions.


Introduction
Prostate cancer is the most common cancer in men and the third leading cause of cancer death [1]. Prostate tumors are typically multifocal, composed of a combination of cells at different stages of differentiation; the histo-prognostic grade (ISUP score) obtained from biopsy samples then guides the management. However, prostate biopsies only provide a limited representation of the intraprostatic tumoral process. Indeed, the ISUP score is frequently modified after analysis of prostatectomy specimens. In addition, it is not uncommon for biopsies to be negative, despite a strong suspicion of prostate cancer. Several studies have shown that performing multiparametric magnetic resonance imaging (mpMRI) before a series of biopsies increases the detection of lesions [2][3][4], but no imaging method is currently able to accurately estimate the histo-prognostic grade and their sensitivity is not optimal.
Focal therapies using focused ultrasound (HIFU-High Intensity Focused Ultrasound) or stereotactic radiotherapy are becoming increasingly important in the management of low-grade localized prostate cancers, mainly because of their low rates of complications. The accurate localization and characterization of the tumor lesion is therefore essential. Indeed, in many cases, no target is identified on mpMRI-thus preventing the use of these treatments. High-performance molecular imaging would guide these focal therapies.
The development of novel radiopharmaceuticals supports innovations in molecular imaging by improving sensitivity and specificity in the diagnosis and characterization of primary prostate tumors. For example, [ 68 Ga]Ga-PSMA (Prostate Specific Membrane Antigen) PET/CT (Positron Emission Tomography/Computed Tomography) is now widely used at the initial staging of prostate cancers in patients with high metastatic risk and in the context of biochemical recurrence [5,6]. However, its ability to detect low-grade tumor lesions is not optimal. Novel radiopharmaceuticals with a role in this setting would be helpful.
Tissue micro-imaging is a technique that allows for the pre-clinical evaluation of radiopharmaceuticals [7,8]. We recently compared the targeting of PSMA and GRP-R (Gastrin Releasing Peptide Receptor), by means of [ 111 In]In-PSMA-617 and [ 111 In]In-RM2, respectively. We showed good detection of low-grade tumor lesions by [ 111 In]In-RM2, superior to that of [ 111 In]In-PSMA [8]. Next, we translated these results into a Phase II study using [ 68 Ga]Ga-PSMA-617 PET/CT and [ 68 Ga]Ga-RM2 PET/CT. Again, we demonstrated a better detection of low-grade lesions by targeting GRP-R using [ 68 Ga]Ga-RM2 [9]. However, 15.6% of the lesions remained undetectable by both modalities.
Thus, the main objective of this study was to evaluate alternative targets for the better identification of intraprostatic lesions. Our strategy was based on a sequential approach: First, we prospectively performed NTS 1 IHC in a series of patients receiving both [ 68 Ga]Ga-PSMA-617 and [ 68 Ga]Ga-RM2 before prostatectomy. In this series, PSMA and GRP-R IHC were also available [9]. Next, we aimed at confirming the PSMA/GRP-R/NTS 1 expression profile by a retrospective autoradiography study and also aimed to decipher the expression of less-investigated targets such as NTS 2 , SST 2 and CXCR4.

Patient Characteristics
Study 1: Formalin-fixed paraffin-embedded samples were prospectively available from patients enrolled in the NCT03604757 study, comparing [ 68 Ga]Ga-PSMA-617 PET/CT to [ 68 Ga]Ga-RM2 PET/CT in patients with localized prostate cancer that were candidates for radical prostatectomy. PSMA and GRP-R staining were performed during this study [9]. For the current study, 16 samples were available for additional NTS 1 staining and comparison with GRP-R and PSMA staining (six samples were considered as non-contributors). National Informatique et Libertés). None of these patients received hormone therapy or other systemic therapy prior to surgery. For each case, thirteen adjacent sections were used: one for hematoxylin-eosin-saffron (HES) staining and twelve for high-resolution microimaging (one section per radiopharmaceutical for total binding and another adjacent section for non-specific binding). An experienced pathologist manually drew tumoral areas on the HES-stained section. All patients were characterized according to their clinical and biochemical criteria including age, tumoral size (clinical and pathological sizes), PSA value and ISUP score.

NTS 1 -Immunohistochemistry
Immunohistochemical assessments were performed as previously described [10]. Immunohistochemistry results were expressed as an immunoreactive score (IRS) that considered staining intensity and the percentage of stained tumor cells [13]. The final IRS score (the product of the staining intensity score and the percentage of positive cells score) thus ranged from 0 to 12: IRS 0-1 means no clear expression; IRS 2-3 indicates weak expression; IRS 4-8 indicates moderate expression; IRS 9-12 indicates strong expression. In order to study associations with other parameters, IHC results were dichotomized into two groups: low expression (regrouping absent/weak expression) and high expression (regrouping moderate/strong expression).

Binding Assay
The protocol and recommendations edited by Reubi and co-workers for binding assays were strictly adhered to [20]. Frozen samples were kept at −80 • C. Three days before handling, samples were placed at −20 • C. The day of the experiment, samples were pre-incubated for 10 min at 37 • C in Tris-HCl buffer at pH 7.4. Then, a binding solution containing 10 nM of the radiopharmaceuticals (except [ 111 In]In-JMV 7488 and [ 67 Ga]Ga-pentixafor, which were used at 75 nM and 50 nM, respectively) in Tris-HCl buffer at pH 8.2, 1% of BSA (Sigma A2153), 40 µg/mL of bacitracin (Sigma ® 11,702), and 10 nM of MgCl 2 (Sigma M8266) was applied. In order to determine the amount of non-specific binding, a large excess of cold ligand was added-more precisely, 1µM of [ nat Ga]Ga-RM2 (Life Molecular Imaging), [ nat Ga]Ga-PSMA-617 (ABX), neurotensin (Bachem), or [ nat Lu]Lu-DOTATATE (ABX), 7.5 µM of levocabastine or 10 µM pentixafor were used. Samples were incubated at 37 • C for 2 h. Afterward, samples were rinsed five times for 8 min in cold Tris-HCl buffer at pH 8.2 with 0.25% of BSA, two times for 8 min in cold Tris-HCl buffer at pH 8.2 without BSA and finally, two times for 5 min in distilled water.

Tissue Microimaging
A Beta Imager-2000 (Biospace Lab) device was used to image and quantify radioactivity in the samples. Acquisition duration was about 10 h (4 × 10 6 counts).

Data Analysis
Imaging analysis was performed as previously described [7]. Briefly, HES and autoradiographic slides were fused and regions of interests (ROIs) were used to calculate the amount of specific binding. A first ROI-drawn by the pathologist to delineate tumor areas-was applied to estimate total binding, and a second ROI-corresponding to background noise-was placed around the tissue. The same ROIs were then transferred to the adjacent slice to determine non-specific binding. After subtracting background noise, specific binding (total binding-non-specific binding) was expressed as a percentage of total binding as follows: Specific binding(%) = (Total binding background) − (non specific binding background) Total binding background × 100

Statistical Analysis
Data, presented as the mean ± standard deviation (SD), were compared using a nonparametric ANOVA. Statistical analyses were performed using GraphPad software (v 6.01, San Diego, CA, USA). p values < 0.05 were considered statistically significant.
Interestingly, all samples with negative PSMA staining (two patients with ISUP 2 and one with ISUP 3) were strongly positive for NTS 1 staining (IRS 0 versus 12; 1 versus 12; 2 versus 12). One lesion was negative for both PSMA and GRP-R staining and strongly positive for NTS 1 staining. On the other hand, all samples with negative NTS 1 staining (n = 6) were positive for PSMA and five of them were positive for GRP-R. Figure 1 shows an example of a prostatic ISUP-2 sample with positive staining for NTS 1 immunochemistry  Immunochemistry was conducted on samples from prostatectomies of patients included in our previous in vivo study [9]. Sixteen samples were available for GRP-R, PSMA and NTS1 staining. Staining was cytoplasmic for PSMA and GRP-R and nuclear for NTS1 ( Figure 1). GRP-R staining was considered positive (IRS ≥ 4) in 11 (68.8%) of 16 lesions. The median GRP-R IRS score was 4 (3-6). PSMA IRS was considered positive (IRS ≥ 4) in 15 (83.3%) of 18 lesions. The median PSMA IRS score was 11 (6)(7)(8)(9)(10)(11)(12). NTS1 IRS was considered positive (IRS ≥ 4) in 10 (62.5%) of 16 lesions. The median NTS1 IRS score was 5 (1-12). Interestingly, all samples with negative PSMA staining (two patients with ISUP 2 and one with ISUP 3) were strongly positive for NTS1 staining (IRS 0 versus 12; 1 versus 12; 2 versus 12). One lesion was negative for both PSMA and GRP-R staining and strongly positive for NTS1 staining. On the other hand, all samples with negative NTS1 staining (n = 6) were positive for PSMA and five of them were positive for GRP-R. Figure 1 shows an example of a prostatic ISUP-2 sample with positive staining for NTS1 immunochemistry but negative staining for PSMA and GRP-R. No prostatic lesion showed negativity with all three stains for PSMA, GRP-R and NTS1.
Finally, when correlating the current NTS1 staining results with clinical PET imaging data from patients included in the trial, four lesions were positive for NTS1 staining with a low [ 68 Ga]Ga-PSMA-617 uptake (SUVmax < 4). One lesion was positive for NTS1 staining with a low [ 68 Ga]Ga-RM2 uptake (Table 1). Finally, when correlating the current NTS 1 staining results with clinical PET imaging data from patients included in the trial, four lesions were positive for NTS 1 staining with a low [ 68 Ga]Ga-PSMA-617 uptake (SUVmax < 4). One lesion was positive for NTS 1
For each radiopharmaceutical, there was no significant difference in binding intensity between various ISUP scores. Overall, binding of [ 111 In]In-PSMA-617 was high in all ISUP groups. However, it was interesting to see that some samples did not bind or bound weakly [ 111 In]In-PSMA-617 (9%). Therefore, a search for novel targets is needed. Below, we report the number of samples for which the binding intensity of the radiopharmaceutical was at least equal to that of [ 111 In]In-PSMA-617 (Table 4) NTS1  SST2  NTS2  CXCR4  1  1  1  2  1  2  2  0  0  0  1  0  3  0  2  1  1  0  4  0  3  0  0  0  5  1  0  0  1  1  Total  2  6  3  4  3 The number of samples for which the specific binding of a radiopharmaceutical was equal or higher than [ 111 In]In-RM2 is reported in Overall, binding of [ 111 In]In-PSMA-617 was high in all ISUP groups. However, it was interesting to see that some samples did not bind or bound weakly [ 111 In]In-PSMA-617 (9%). Therefore, a search for novel targets is needed. Below, we report the number of samples for which the binding intensity of the radiopharmaceutical was at least equal to that of [ 111 In]In-PSMA-617 (Table 4) NTS1  SST2  NTS2  CXCR4  1  1  1  2  1  2  2  0  0  0  1  0  3  0  2  1  1  0  4  0  3  0  0  0  5  1  0  0  1  1  Total  2  6  3  4  3 The number of samples for which the specific binding of a radiopharmaceutical was equal or higher than [ 111 In]In-RM2 is reported in For each radiopharmaceutical, there was no significant difference in binding intensity between various ISUP scores.

Discussion
Several radiopharmaceuticals have been developed to help in the staging of prostate cancer. The radiolabeled analog of the essential amino acid leucine 18 F-FACBC ( 18 F-Flucicovine) does not demonstrate high specificity for imaging in primary prostate cancer [21]. Furthermore, 11 C-Acetate-marking lipid metabolism-cannot reliably distinguish benign prostatic hyperplasia from prostate tumors. Finally, 11 C/ 18 F-Choline-another marker of lipid metabolism-shows lower sensitivity than mpMRI for the detection of . Comparison between PSMA, GRP-R, NTS 1 , NTS 2 and CXCR4-specific radiopharmaceuticals on an ISUP-5 sample. The red line drawing on HES slice corresponds to the tumor area. PSMA, GRP-R and NTS 1 samples showed no uptake on the tumoral area. Contrarily, NTS 2 , and CXCR4 showed a strong tumor uptake. Specific binding = 0% for PSMA, 0% for GRP-R, 0% for NTS 1 , 71% for NTS 2 and 100% for CXCR4. Color scale refers to cps/mm 2 .

Discussion
Several radiopharmaceuticals have been developed to help in the staging of prostate cancer. The radiolabeled analog of the essential amino acid leucine 18 F-FACBC ( 18 F-Flucicovine) does not demonstrate high specificity for imaging in primary prostate can-cer [21]. Furthermore, 11 C-Acetate-marking lipid metabolism-cannot reliably distinguish benign prostatic hyperplasia from prostate tumors. Finally, 11 C/ 18 F-Choline-another marker of lipid metabolism-shows lower sensitivity than mpMRI for the detection of primary prostate cancer [22]. Thus, the search for novel targets appears necessary for the initial assessment of the aggressiveness of primary prostate tumors.
PSMA and GRP-R have been investigated for the initial staging of prostate cancer. In a prospective study enrolling 56 intermediate grade prostate cancer patients before prostatectomy, PSMA PET was found to be accurate in detecting intraprostatic lesions of ISUP ≥ 2. Contrarily, the detection rate of PSMA PET was low for ISUP-1 lesions. Touijer et al. prospectively investigated [ 68 Ga]Ga-RM2 PET/CT in 16 patients before radical prostatectomy; the performance of [ 68 Ga]Ga-RM2 PET/CT imaging did not significantly differ compared to mpMRI in terms of sensitivity, specificity or accuracy [23].
Our previous study showed similar findings, as [ 68 Ga]Ga-PSMA-617 PET/CT was useful for depicting higher ISUP score lesions and [ 68 Ga]Ga-RM2 PET/CT had a higher detection rate for low-ISUP tumors [9]. In the lesion-based analysis (including lesions < 0.1 cc), [ 68 Ga]Ga-PSMA-617 PET/CT detected 74.3% of all tumor lesions and [ 68 Ga]Ga-RM2 PET/CT detected 78.1%. However, paired examinations showed negative uptake in 15.6% of lesions by both modalities. Therefore, the objective of this work was to explore new targets to detect these unseen lesions.
The prospective immunochemistry study performed in this work confirms the interest in NTS 1 , as all PSMA negative lesions were strongly positive for NTS 1 . Moreover, all negative NTS 1 staining lesions (37.5%) were positive for PSMA staining and positive for GRP-R staining in five patients (31%). Our results consolidate a previous study demonstrating that PSMA-negative samples from Gleason scores of 5, 6 or 7 were all NTS 1 -positive [24]. Thus, the interest in NTS 1 might be greater than for GRP-R in low histological grade tumors, but comparison with GRP-R is obviously needed. Unfortunately, no NTS 1 imaging radiopharmaceutical has yet shown interesting results when applied to humans [25]. Work is ongoing to find stabilized NTS 1 analogues suitable for imaging [16,26]. These new data should also be considered with caution as IHC results do not necessarily translate into similar findings in vivo.
With this in mind, we performed a retrospective micro-imaging study comparing PSMA, GRP-R, NTS 1 as well as NTS 2 , SST 2 and CXCR4 expression using specific radiopharmaceuticals that would be more representative of in vivo behavior. Overexpression of the NTS 2 receptor in prostate cancer has been reported; an in vitro study has assessed the potential use of the NTS 2 receptor as a target by analyzing its expression patterns in human prostate cell lines and primary prostate cell cultures-NTS 2 was found in cells with luminal phenotype [27]. Other studies are needed to confirm these results. SST 2 is also overexpressed in prostate cancer-especially in cases of neuroendocrine differentiation [11,28]. CXCR4 overexpression has also been reported in prostate cancer; studies have shown that CXCR4 is a key regulator of tumor dissemination [12]. An in vitro study comparing adjacent normal endothelial cells to prostate tumor vasculature highlighted CXCR4 as a potential novel target to interfere with prostate cancer angiogenesis [29].
While our work shows the superiority of PSMA for the detection of intraprostatic lesions, with a significant higher binding of In-JMV 7488 for all ISUP-score groups (no significant difference was found for CXCR4-mostly due to a lack of power), and PSMA PET has now entered into guidelines [30], alternative targets are necessary in the event of PSMA negativity. In a previous study enrolling fifty newly diagnosed patient with prostate cancer, the [ 68 Ga]Ga-PSMA-617 PET/CT was negative in 12.5% [31]. Targeting the GRP-R is expected to cover the limitations of PSMA [9]. In our work, in ISUP scores 1 Overall, the most interesting targets to cover PSMA-negative lesions appear to be NTS 1 and NTS 2 -with, respectively, four and six cases with superior or equivalent detection than PSMA, covering all ISUP scores. It is interesting to note that combining PSMA and NTS 1 /NTS 2 could allow for the detection of all intraprostatic lesions. The new findings in this work also highlight the potential of multireceptor-targeting radioprobes that can still bind one target (NTS 1 or NTS 2 or GRP-R) when the other is lost (PSMA). Works are ongoing to optimize radiolabeled PSMA/GRP-R heterobivalent probes [32], while the development of PSMA/NTS 1 heterodimers has only been described once [33]. Overall, this work sheds light on the abundance of different neuropeptide receptors (mainly neurotensin receptors) in different physiopathological states of prostate cancer.
The improved detection of lesions allows for better mapping of prostate tumor pathology, which is necessary for biopsy guiding to decrease the discordance rate of staging of biopsies and final staging of prostatectomy samples. Finally, the possibility of a more precise detection and characterization of intra-prostatic lesions opens new avenues for radiotherapy planning and/or focal treatments.
The reader should be aware that it was not our aim to compare radiopharmaceuticals, but rather to use them to quantify receptor density in primary prostate cancer samples. Moreover, in this work, we were not able to provide the uptake (as a percentage of the applied dose) of each radiopharmaceutical.

Conclusions
In this work, we have compared GRP-R, PSMA, NTS 1 , NTS 2 , SST 2 and CXCR4 expression in vitro in primary prostate cancer samples. Our results confirm that PSMA remains the best target in tumor detection at initial staging-especially for high grade lesions. Interestingly, targeting NTS 1 and NTS 2 allowed us to detect all PSMA-negative lesions more precisely than GRP-R in vitro. Future in vivo prospective studies must confirm these data.