Comparison of the binding of the gastrin-releasing peptide receptor (GRP-R) antagonist 68Ga-RM2 and 18F-FDG in breast cancer samples

The Gastrin-Releasing Peptide Receptor (GRPR) is over-expressed in estrogen receptor (ER) positive breast tumors and related metastatic lymph nodes offering the opportunity of imaging and therapy of luminal tumors. 68Ga-RM2 binding and 18F-FDG binding in tumoral zones were measured and compared using tissue micro-imaging with a beta imager on 14 breast cancer samples (10 primaries and 4 associated metastatic lymph nodes). Results were then assessed against ER expression, progesterone receptor (PR) expression, HER2 over-expression or not and Ki-67 expression. GRPR immunohistochemistry (IHC) was also performed on all samples. We also retrospectively compared 68Ga-RM2 and 18F-FDG bindings to 18F-FDG SUVmax on the pre-therapeutic PET/CT examination, if available. 68Ga-RM2 binding was significantly higher in tumors expressing GRPR on IHC than in GRPR-negative tumors (P = 0.022). In ER+ tumors, binding of 68Ga-RM2 was significantly higher than 18F-FDG (P = 0.015). In tumors with low Ki-67, 68Ga-RM2 binding was also significantly increased compared to 18F-FDG (P = 0.029). Overall, the binding of 68Ga-RM2 and 18F-FDG displayed an opposite pattern in tumor samples and 68Ga-RM2 binding was significantly higher in tumors that had low 18F-FDG binding (P = 0.021). This inverse correlation was also documented in the few patients in whom a 18F-FDG PET/CT examination before surgery was available. Findings from this in vitro study suggest that GRPR targeting can be an alternative to 18F-FDG imaging in ER+ breast tumors. Moreover, because GRPR antagonists can also be labeled with lutetium-177 this opens new avenues for targeted radionuclide therapy in the subset of patients with progressive metastatic disease following conventional treatments.


Introduction
The Gastrin-Releasing Peptide Receptor (GRPR, also named BB2) is a G-protein coupled receptor of the bombesin family. Its over-expression on the membrane of tumor cells offers the opportunity of a selective targeting, using suitable radiolabelled bioconjugates, for positron emission tomography (PET) imaging and targeted radionuclide therapy (TRT). Tumors that can be targeted with GRPR-based radiotracers are notably, prostate cancer, breast cancer, lung cancer and colorectal cancer among others [1]. We have recently studied, using immunohistochemistry, the expression of GRPR ina large series of primary breast cancers and found that GRPR was overexpressed in 83.2% of ER-positive tumors but only in 12% of ER-negative tumors (p < 0.00001) [2]. When examined in molecular subtypes, GRPR is over-expressed in 86.2% of luminal-A and 82.8% of luminal-B HER2 negative tumors while triple negative breast cancers and HER2-enriched phenotypes exhibit GRPR over-expression in only 7.8% and 21.3% of cases. Importantly, lymph nodes metastases of GRPR-positive tumors also showed GRPR overexpression [2]. The association between GRPR and ER has also been documented at mRNA level by Dalm and colleagues [3]. Recently, GRP-R antagonists radiolabelled for PET imaging, demonstrated promising results in breast cancer patients. For example, in a small pilot study that used 68 Ga-SB3, metastases were successfully visualized in 4 out of 6 patients [4]. In another study, 68 Ga-RM2 could image with high contrast 13/18 primary breast tumors and detected metastatic lesions [5]. In a more recent study conducted in 34 women with suspected breast cancer, a novel GRPR antagonist, 68 Ga-NOTA-RM26, was able to delineate primary breast tumors in 29/34 patients and lymph nodes metastases in 15/18 patients with node-positive disease [6]. Comparison of breast cancer imaging using GRP-R based radioantagonists and 18 F-FDG is now needed to elucidate the place of GRP-R in the complex landscape of breast cancer imaging. This in vitro study aimed to assess the binding of 18 F-FDG and that of the GRPR antagonist 68 Ga-RM2 on representative breast cancer samples.

Breast cancer samples
This study was approved by our institutional review board "Institut Bergonié Groupe Sein". The project and data collection were performed according to the national French commission on informatics and liberty (CNIL). Prior to surgery, patients had given written consent to the use of part of the tumor material for research, after diagnostic procedures had been performed. Fourteen samples of formalin-fixed, paraffin-embedded breast cancer tissues (10 primary tumors and 4 associated metastatic lymph nodes) were retrospectively selected at Institut Bergonié. Sample characteristics' are presented in Table 1. No patients had received neoadjuvant hormone therapy or chemotherapy. For each case, 6 successive slices were used: 1 for HES staining, 1 for GRP-R immunohistochemistry and 4 for micro-imaging of tissue radioactivity (one slice per radiopharmaceutical for total binding and one slice per radiopharmaceutical for non-specific binding). GRP-R immunohistochemistry was carried-out as previously described [2].
Immunohistochemistry. IHC analyses were performed on 3μm tumor sections using specific antibodies directed against ER, PR, HER2/neu, Ki-67 and GRPR. All immunohistochemical techniques were performed on a Roche Ventana Benchmark ultra-automat. Details of antibody clones, manufacturers, dilutions used, incubation times, pretreatment buffers and staining kits are summarized in Table 2.
Results for GRP-R immunohistochemistry were expressed as previously described [2]. An experimented pathologist (GMG) quantified GRP-R expression and manually drew tumoral regions on the HES slice for quantification.

Sample ER (%) PR (%) HER2 over-expression Ki-67 (%) Molecular phenotype GRPR status
Primary tumors Belgium) and 5mg of ascorbic acid. The raw solution was then purified on a C 18 cartridge (WAT023501) preconditioned with 1mL ethanol (Merck) and 5 mL water (GE Healthcare). The final product was then eluted with 1 mL ethanol and formulated in PBS. 68 Ga-RM2 was checked for radiochemical purity using HPLC (Phenomenex Luna C 18 ; 250mm x 4.6mm x 5μm; 2.5 mL/min, λ = 220nm). The analytical HPLC system used was a JASCO system with ChromNAV software, a PU-2089 Plus quaternary gradient pump, a MD-2018 Plus photodiode array detector and Raytest Gabi Star detector. Amount of 68 Ga-RM2 was determined by UV-HPLC by linear regression of the calibration curve established using the reference compound nat Ga-RM2 (Life Molecular Imaging). Tracer incubation and tissular micro-imaging. After dewaxing, rehydratation and unmasking, samples were pre-incubated during 10min at 37˚C in Tris-HCl buffer at pH 7.4. Then, binding solution containing 5nM of 68 Ga-RM2 or 1MBq (amount of 18 F-FDG is not determined by the supplier) of 18 F-FDG in Tris-HCl buffer pH 8.2, containing 1% of BSA (Sigma-Aldrich), 40μg/mL of bacitracin (Sigma-Aldrich) and 10nM of MgCl 2 (Sigma-Aldrich) was applied. To assess non-specific binding, 1μM of reference compounds nat Ga-RM2 or nat F-FDG was added in adjacent slices. Samples were then incubated at 37˚C for 2 hours. Afterwards, samples were rinsed 5 times during 8min in cold Tris-HCl buffer at pH 8.2 with 0.25% of BSA, 2 times during 8 minutes in cold Tris-HCl buffer at pH 8.2 without BSA and finally 2 times during 5 minutes in distilled water. Finally, samples were dried using air stream and were imaged using a beta imager 2000 (Biospace Lab).
Signal quantification. The M3Vision software was used for signal quantification. Total binding and non-specific binding were determined using the region of interest (ROI) method. First, a manual fusion by affine transformation of homologous structures was performed using the HES slice to match the radioactivity distribution to histology. Afterwards, on the total binding image ( 68 Ga-RM2 or 18 F-FDG alone) a first ROI (tumoral ROI) was placed on the tumoral zone and a second ROI (noise ROI), corresponding to noise, was placed around the tissue. Then, the same ROIs were applied on quantitative images from adjacent slices representing non-specific binding ( 68 Ga-RM2 or 18 F-FDG plus excess of reference compound) to define non-specific binding. Finally, data were exported on Excel software for processing. Parameters "Signal to Noise Ratio (SNR)" and "Delta" were then calculated. SNR was defined as the signal in tumoral ROI minus signal in noise ROI. Delta was calculated as follow: Delta % ð Þ ¼ SNRtotal binding À SNRnon specific binding SNRtotal binding x 100 Statistical analysis. Differences between mean values were assessed using non parametric t-test. A P-value of less than 0.05 was considered statistically significant. Statistical analysis was performed using GraphPad Prism software v 6.01.
Retrospective analysis of 18 F-FDG PET/CT. We retrospectively analyzed pre-therapeutic 18 F-FDG PET/CT performed at the Nuclear Medicine Department of Institut Bergonié. PET/ CT had been performed before surgery in only 2 patients, corresponding to tissue samples 1 and 5 in Table 1. 18 F-FDG uptake was measured as SUV max in a VOI drawn on the breast tumor.

68
Ga-RM2 radiosynthesis 68 Ga-RM2 was obtained at a mean specific activity of 47.3 ± 16.7 GBq/μmol and a mean radiochemical purity of 99.52 ± 0.18% suitable for in vitro experiments.

Comparison of 68 Ga-RM2 binding and GRP-R immunohistochemistry
As a validation step we assessed whether tissular micro-imaging may accurately reflect IHC results. We stratified samples according to their GRP-R status determined by IHC and our results showed that the mean 68 Ga-RM2 delta was significantly higher in GRP-R expressing tumors than in GRP-R-negative tumors (33.93 ± 17.55% vs 0.0 ± 0.0%; P = 0.022).

Ga-RM2 and 18 F-FDG bindings
Qualitative analysis. Qualitative analysis showed a good signal-to-noise ratio, and a binding in agreement with GRPR IHC with clear differences between total and non-specific bindings (Fig 1).
There were no significant differences in the HER2 + and HER2groups for 68 Ga-RM2 or for 18 F-FDG binding (Table 3).
Quantitative analysis: Comparison of 68 Ga-RM2 and 18 F-FDG bindings. In ER + tumors, binding of 68 Ga-RM2 was largely higher than 18 F-FDG (45.31 ± 13.23% vs 16.51 ± 28.45%; P = 0.015), while in ERtumors binding of 18 F-FDG was comparable to that of 68 Ga-RM2 (P = 0.483). Therefore, the ratio of mean 68 Ga-RM2 binding to 18 F-FDG was 3.42 in ER + tumors vs 0.71 in ERtumors. There was also a strong trend for higher 68 Ga-RM2 binding than 18 F-FDG in PR + tumors (P = 0.089) while no differences were observed in the PRgroup (P = 0.626). In these subgroups, the ratio of mean 68 Ga-RM2 binding to 18 F-FDG was 1.99 in PR + tumors vs 0.86 in PRtumors. In tumors with low Ki-67, 68 Ga-RM2 binding was also significantly increased compared to 18 F-FDG (49.24 ± 9.15% vs 10.40 ± 12.35%; P = 0.029). There was no differences in the bindings of 68 Ga-RM2 and 18 F-FDG in tumors with high Ki-67 (P = 0.783). These differences translate in a higher ratio of mean 68 Ga-RM2 binding to 18 F-FDG in low Ki-67 tumors (4.73 vs 0.80). In HER2-tumors, the ratio of mean 68 Ga-RM2 binding to 18 F-FDG was 1.70 while in HER2+ this ratio reaches only 0.53.

F-FDG PET/CT
Among patients studied using tissular micro-imaging, two had undergone 18 F-FDG PET/CT imaging for staging before surgery ( Table 4). The first patient had a low 18 F-FDG uptake in vivo (SUV max = 2.5), a negative 18 F-FDG delta ex vivo, a high 68 Ga-RM2 delta of 37.46% and a positive GRP-R IHC. The second patient had a high 18 F-FDG uptake (SUV max = 9.2), a high 18 F-FDG delta of 42.97%, no 68 Ga-RM2 binding and a negative GRP-R IHC.

Discussion
The correlation between GRP-R overexpression in breast cancer and estrogen receptor positivity at protein level or mRNA level has been recently highlighted [2,3]. Moreover, it has been documented that when the breast primary is GRPR-positive, lymph node metastases also show GRPR overexpression [2,3]. Several clinical pilot studies have illustrated, in vivo, the potential of GRP-R for breast cancer imaging using radiolabelled GRP-R antagonists such as 68 Ga-SB3, 68 Ga-RM2 or 68 Ga-NOTA-RM26 [4,5,6]. In some of these studies it was shown that ER-positive tumors can be visualized with high contrast [5,6]. 18 F-FDG PET/CT is also a valuable tool for staging of invasive breast cancer [10]. Highly 18 F-FDG-avid tumors are generally Elston and Ellis grade 3, have a high proliferation index and negative hormone receptor status, while  somewhat lower uptake can be encountered in low grade ER-positive tumors and in lobular carcinoma [10]. Indeed, imaging ER-positive breast tumors, especially the luminal-A phenotype, might be challenging using 18 F-FDG PET/CT in some patients [11]. Therefore, how GRP-R imaging would perform compared to 18 F-FDG in ER-positive breast cancer deserves investigation. We aimed to compare on breast cancer samples the binding of a radiolabelled GRP-R antagonist, 68 Ga-RM2, to that of 18 F-FDG in order to better understand the potential of GRP-R imaging as a first step before a clinical study comparing the two tracers was launched.
Results of the present study on breast cancer samples showed that GRP-R targeting would be highly relevant in breast cancer, specifically in ER-positive tumors. Mean specific binding of 68 Ga-RM2 was 45.31 ± 13.23% in ER-positive tumors and only 14.32 ± 9.20% in ER-negative tumors (P = 0.030). The opposite pattern was noted as regards 18 F-FDG bindings. As a result, the ratio of mean 68 Ga-RM2 binding to that of 18 F-FDG binding in ER + tumors was 3.42 vs 0.71 in ERtumors. Another important finding is the high 68 Ga-RM2 binding in tumors with low Ki-67 (49.24 ± 9.15%) while tumors with high Ki-67 exhibited lower 68 Ga-RM2 binding (20.62 ± 17.88%)(P = 0.023). Also, the ratio of mean 68 Ga-RM2 to 18 F-FDG binding in tumors with low Ki-67 was significantly higher than in tumors with high Ki-67 (4.73 vs 0.80). Overall, these results suggest a role for GRP-R PET imaging that could be complementary or superior to 18 F-FDG imaging in ER-positive tumors with a low proliferation index.
Thus, 18 F-FDG PET/CT and GRP-R imaging may be complimentary for imaging breast cancer and more specifically so the ER-positive subtypes. A study comparing a GRPR targeting radiotracer and 18 F-FDG for primary staging or for restaging recurrent breast cancer would be appreciated. Another approach that could enhance tumor detection, is the possibility of a multiple targeting as demonstrated by 68 Ga-BBN RGD that targets both GRP-R and integrin α v β 3 . In a pilot study, this heterodimeric radiopharmaceutical performed better than 68 Ga-BBN (that targets only the GRP-R) in the detection of primary tumor and bone lesions in 11 patients [12]. Comparison with 18 F-FDG would also be helpful for clinicians.
Finally, GRP-R targeting opens also attractive perspectives for radiopharmaceutical therapy of this subgroup of metastatic luminal patients with antagonists labelled with beta-emitters such as the lanthanides 177 Lu [13] or 161 Tb [14,15] or with alpha emitters.
Limitations of this study, apart the number of samples and its retrospective nature, is the 18 F-FDG tissular micro-imaging which may appear questionable. Cristallographic studies at the human glucose transporter 1 (GLUT-1) revealed that glucose uptake is a 2-step mechanism involving glucose binding before active transport [16]. Moreover, enhanced 18 F-FDG uptake in tumors is not only related to overexpression of glucose transporters but also to enhanced hexokinase activity which was not assessed here. Therefore, our 18 F-FDG tissular micro-imaging is relevant (clear displacement of 18 F-FDG with reference compound) and revealed at least the 18 F-FDG binding site but may over-estimate or underestimate 18 F-FDG uptake. In total, our data point that GRPR targeting should be helpful for imaging breast cancer and more specifically so the ER-positive subtypes. A study comparing a GRPR targeting radiotracer and 18 F-FDG for primary staging and for restaging recurrent breast cancer is clearly needed.