Intra-individual dynamic comparison of 18F-PSMA-11 and 68Ga-PSMA-11 in LNCaP xenograft bearing mice

Recently, a 18F-labeled derivative of the widely used 68Ga-PSMA-11 was developed for PET imaging of prostate cancer. Although 18F-PSMA-11 has already been evaluated in a Phase I and Phase II clinical trial, preclinical evaluation of this radiotracer is important for further understanding its dynamic behavior. Saturation binding experiments were conducted by incubation of LNCaP cells with 18F-PSMA-11 or 68Ga-PSMA-11 for 1 h, followed by determination of the specific and aspecific binding. Mice bearing LNCaP or PC-3 xenografts each received ± 3.7 MBq 18F-PSMA-11 and 68Ga-PSMA-11 followed by dynamic acquisition of 2.5 h as well as ± 15 MBq 18F-FDG followed by static acquisition at 1 h post injection (p.i.). Uptake was evaluated by comparison of uptake parameters (SUVmean, SUVmax, TBRmean and TBRmax). Mice underwent ex vivo biodistribution where 18F-PSMA-11 activity was measures in excretory organs (kidneys, bladder and liver) as well as bone fragments (femur, humerus, sternum and skull) to evaluate bone uptake. The dissociation constant (Kd) of 18F-PSMA-11 and 68Ga-PSMA-11 was 2.95 ± 0.87 nM and 0.49 ± 0.20 nM, respectively. Uptake parameters were significantly higher in LNCaP compared to PC-3 xenografts for both 18F-PSMA-11 and 68Ga-PSMA-11, while no difference was found for 18F-FDG uptake (except for SUVmax). Tumor uptake of 18F-PSMA-11 showed a similar trend over time as 68Ga-PSMA-11, although all uptake parameter curves of the latter were considerably lower. When comparing early (60 min p.i.) to delayed (150 min p.i.) imaging for both radiotracers individually, TBRmean and TBRmax were significantly higher at the later timepoint, as well as the SUVmax of 68Ga-PSMA-11. The highest %ID/g was determined in the kidneys (94.0 ± 13.6%ID/g 1 h p.i.) and the bladder (6.48 ± 2.18%ID/g 1 h p.i.). No significant increase in bone uptake was seen between 1 and 2 h p.i. Both radiotracers showed high affinity for the PSMA receptor. Over time, all uptake parameters were higher for 18F-PSMA-11 compared to 68Ga-PSMA-11. Delayed imaging with the latter may improve tumor visualization, while no additional benefits could be found for late 18F-PSMA-11 imaging. Ex vivo biodistribution demonstrated fast renal clearance of 18F-PSMA-11 as well as no significant increase in bone uptake.


Scientific Reports
| (2020) 10:21068 | https://doi.org/10.1038/s41598-020-78273-7 www.nature.com/scientificreports/ PSA values ≥ 2.0 ng/mL and 93.0%, 72.7% and 57.9% for PSA values of 1 to < 2 ng/mL, 0.5 to < 1 ng/mL and 0.2 to < 0.5 ng/mL, respectively. Despite the high affinity for the PSMA receptor and the excellent results with regard to currently used PET probes [4][5][6][7] , the use of 68 Ga as radionuclide is associated with some unfavorable physical properties. In comparison to 18 F, 68 Ga has a shorter half-life (68 min vs 110 min), as well as a lower positron emission (89% vs 97%) and a higher maximum positron energy (1.90 meV vs 0.63 meV), resulting in a longer positron range and lower spatial resolution 8 . Furthermore, the cyclotron-based production of 18 F makes large batch production possible as opposed to the limited capacity of 2-3 patient doses for the generator-produced 68 Ga 9 . Amongst others, the well-established use of 68 Ga-PSMA-11 has led to the development of the fluorine-18 derivative 18 F-PSMA-11 by Malik et al. 10 and Boschi et al. 11 and was further optimized by Kersemans et al. 12 to enable semi-automated production. The Phase I clinical trial conducted in our hospital evaluated safety, dosimetry and biodistribution 13 . The recently published Phase II study reported on an optimized scan protocol where dosage, scan time and administration of a diuretic were studied 14 .
Although the use of 18 F-PSMA-11 has already been investigated in 107 patients, preclinical evaluation of this radiotracer is warranted in order to gain a deeper understanding of its dynamic character, biological behavior and excretion kinetics. Therefore, imaging characteristics of 18 F-PSMA-11 and 68 Ga-PSMA-11 were compared in a preclinical setting. To our knowledge, no dynamicly acquired intra-individual comparison of these two radiotracers as well as extensive in vivo and ex vivo evaluation of bone uptake of 18 F-PSMA-11 tracer has been published before.

Materials and methods
Synthesis of PET radiotracers. Synthesis of 18 F-PSMA-11 was performed as described by Kersemans et al. 12 on a modified SynthraFCHOL synthesis module (Synthra GmbH, Hamburg, Germany). 68 Ga-PSMA-11 was prepared using a lyophilized sterile cold kit (ANMI, Liege, Belgium) by reconstitution of 25 µg PSMA-11 precursor in acetate buffer (pH 4. 1-4.4). 68 Ga was eluted from a 68 Ge/ 68 Ga generator (50 mCi; IRE-Elit, Fleurus, Belgium) in an evacuated sterile vial using 1.1 mL of 0.1 M HCl and added to the precursor solution. Labeling was performed at room temperature for 5 min.

Affinity.
Saturation binding experiments were conducted as described by Verhoeven et al. 15 to determine the K d of 18 F-PSMA-11 and 68 Ga-PSMA-11. Wells were seeded with 2 × 10 5 LNCaP cells 48 h prior to the experiments using poly-lysine coated 24-well-plates (VWR, USA). After removal of the culture medium, wells were washed twice with 1 mL HEPES buffer (pH 7.4, 37 °C). Six dosing solutions between 2.5 and 50 nM of both radiotracers were prepared in HEPES buffer and evaluated in triplicate. Non-specific binding was determined by co-incubation with 100 µM 2-(phosphonomethyl)-pentanedioic acid (2-PMPA, Sigma Aldrich, Belgium). After an incubation period of 1 h at 37 °C, plates were cooled on ice and 1 mL ice-cold 1% BSA/PBS was added to stop radiotracer uptake. Cells were washed twice with 2 mL ice-cold PBS and subsequently lysed with 0.1 M NaOH (VWR, USA). 18 F-PSMA-11 and 68 Ga-PSMA-11 uptake in the cells was measured with an automated gamma counter (Cobra-inspector 5003, Canberra Packard, Meriden, CT, USA) and corrected for amount of protein by a Bicinchonic Acid (BCA) assay (ThermoFisher Scientific, Belgium). The K d value was calculated by non-linear regression using Graphpad Prism 5.0 (GraphPad Software, San Diego, CA, USA, http://www.graph pad.com).
Inoculation of mice. The study was approved by the Ghent University Ethical Committee on animal experiments (ECD 17/14). All animals (n = 10) were kept and handled according to the European guidelines (Directive 2010/63/EU) and housed under environmentally controlled conditions (12 h normal light/dark cycles, 20-24 °C and 40-70% relative humidity) with food and water ad libitum. On the day of the inoculation, LNCaP and PC-3 cells were washed twice with FBS-free RPMI 1640 medium and two cell suspensions of 5 × 10 6 cells/100 µL were prepared and kept on ice until inoculation. Four-week-old male athymic nude mice (swiss nu/nu, Charles River Laboratory, France) were subcutaneously injected with 200 µL 1:1 cell:Matrigel suspension using precooled insulin syringes on either side of each mouse (LNCaP, n = 6; PC-3, n = 4) at shoulder height. Tumor growth was monitored weekly for 5-6 weeks until tumors reached a diameter between 5 and 10 mm.
Biodistribution. Eight male athymic nude mice (swiss nu/nu, Charles River Laboratory, France) were subjected to ex vivo biodistribution. One additional mouse bearing LNCaP xenograft was added to evaluate tumor uptake. All mice received 1.95 ± 0.10 MBq 18 F-PSMA-11 and were sacrificed at 1 h (n = 4 + 1) or 2 h (n = 4) post injection (p.i.). Excretory organs (kidneys, bladder and liver) and bone fragments (femur, humerus, sternum and skull) were removed, weighted and measured using a gamma counter. Image analysis. Images were analyzed using the Amide software 16 . After co-registration of PET and CT images, volumes of interest (VOIs) were drawn manually for delineation of the tumor, kidneys, bladder and bone fragments (spine, femur, sternum and humerus). A background region was drawn in the same transversal slice as tumor VOIs. The tracer uptake in each tumor VOI was calculated as mean and maximum standardized uptake value (SUV mean and SUV max ) according to Formula 1.
Besides SUV mean and SUV max , tumor-to-background ratios (TBR mean and TBR max ) were determined. For nontumor tissues, only SUV mean was determined. Semi-quantitative analysis of tumor uptake was performed for every 5 min time frame and plotted at 5, 10, 15, 20, 25, 30, 60, 90, 120 and 150 min.
Immunohistochemical evaluation. After the last scan, mice were sacrificed and tumors were collected for immunohistochemical (IHC) evaluation as described by Braeckman et al. 17 . Sections were either stained using Hematoxylin and Eosin or incubated with a primary PSMA antibody (1:400, 2 h, Abcam, ab133579) and counterstained using hematoxylin (Mayer). Sections were digitally scanned with a virtual scanning microscope (Olympus BX51, Olympus Belgium SA/NV, Berchem, Belgium) at high resolution (40 × magnification).

Statistical analysis.
All uptake parameters (SUV mean , SUV max , TBR mean and TBR max ) were expressed as mean ± SEM. Curves were constructed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA, http://www.graph pad.com). The statistical analysis was performed in R 18 using the Wilcoxon-signed Rank test for the cross-over intra-individual comparison of radiotracer uptake and the Mann-Whitney U test for comparison of uptake between PSMA positive and negative tumors. The significance level was set on p ≤ 0.05.

Results
Synthesis. 18 Fig. 1. Colormaps were adapted in order to optimally visualize the tumor, images comparing radiotracers at identical thresholds can be found in the Supplementary Data ( Figure S1). PSMA-targeting radiotracers showed less background activity in adjacent tissues compared to 18 F-FDG. LNCaP tumors could be clearly identified with all three radiotracers, while PC-3 tumors were only visible with 18 F-FDG. The specificity of 18 F-PSMA-11 and 68 Ga-PSMA-11 was visualized and semi-quantified by comparing radiotracer uptake in PSMA positive (LNCaP) and PSMA negative (PC-3) tumors. SUV mean , SUV max , TBR mean and TBR max were significantly higher in LNCaP compared to PC-3 xenografts for both 18 F-PSMA-11 and 68 Ga-PSMA-11, while no difference was found for these parameters with regard to 18 F-FDG uptake, except for SUV max ( Table 1). The presence and absence of PSMA expression in respectively LNCaP and PC-3 cells was demonstrated with IHC analysis (Fig. 2).
Tumor uptake of 18 F-PSMA-11 in LNCaP tumors increased rapidly within the first 30 min post radiotracer administration for all uptake parameters (Fig. 3 www.nature.com/scientificreports/  www.nature.com/scientificreports/ increase could be seen after 20 min. When comparing early (60 min p.i.) to delayed (150 min p.i.) imaging for both radiotracers individually, TBR mean and TBR max were significantly higher at the later timepoint, whereas for 68 Ga-PSMA-11 also an increased SUV max was observed (Fig. 4). When comparing both radiotracers at 60 min and 150 min p.i., all uptake parameter values were higher for 18 F-PSMA-11 compared to 68 Ga-PSMA-11. These differences were significant, except for TBR max and SUV max 150 min p.i. (Fig. 5).
Time activity curves of the excretory organs (kidneys, bladder and liver) demonstrated higher 18 F-PSMA-11 radioactivity in the kidneys (SUV mean 30 min p.i. of 12.98 ± 0.82 vs 7.20 ± 1.09) while 68 Ga-PSMA-11 was more prominent in the bladder (SUV mean 60 min p.i. of 49.71 ± 4.93 vs 16.82 ± 3.87) (Fig. 6), which was also visible on maximum intensity projection (MIP) images at 1 h p.i. (Fig. 7). Liver uptake decreased rapidly for both radiotracers indicating limited hepatobiliary clearance. Bone uptake was assessed using VOIs drawn in the spine, femur, sternum and humerus. The resulting SUV mean of both 18 F-PSMA-11 and 68 Ga-PSMA-11 in these VOIs decreased during the first 60 min p.i. Between 60 and 150 min p.i., presence of 68 Ga-PSMA-11 in the bone continued to decrease while the uptake of 18 F-PSMA-11 slightly increased (SUV mean from 0.71 ± 0.07 to 0.75 ± 0.07 in the     Biodistribution. Blood levels of 18 F-PSMA-11 decreased between 1 and 2 h p.i. from 0.75 ± 0.31%ID/g to 0.47 ± 0.03%ID/g. The highest %ID/g was determined in the kidneys (94.0 ± 13.6%ID/g 1 h p.i. and 82.5 ± 10.9%ID/g 2 h p.i.) and the bladder (6.48 ± 2.18%ID/g 1 h p.i. and 11.7 ± 2.51%ID/g 2 h p.i.) (Fig. 8). No significant increase in bone uptake was observed between 1 and 2 h p.i. ( Table 2). The LNCaP tumor showed radiotracer uptake of 9.11%ID/g.  www.nature.com/scientificreports/ 18 F-PSMA-11 has already been evaluated in a Phase I and Phase II clinical trial in our hospital. The Phase II study was set up in order to determine an optimized scan protocol. Although several parameters such as dosage, scan duration and time of imaging post radiotracer administration were investigated, the latter was limited to two timepoints (early (1 h p.i.) and delayed (3 h p.i.) imaging) due to practical considerations inherent to a clinical trial involving human participants 14 . In vitro and in vivo evaluation of 18 F-PSMA-11 involving dynamic imaging in mice may provide more insight into the affinity, scan time window and biological behavior of the radiotracer.

Discussion
In vitro characterization of 18  A significantly higher uptake in LNCaP compared to PC-3 xenografts indicated high specificity of PSMAtargeting radiotracers for PSMA-positive tumors. Due to the poor-differentiated and highly aggressive character of PC-3 cells, 18 F-FDG uptake was expected to be higher compared to LNCaP cells [22][23][24] . However, a significant difference could only be observed in SUV max .
All mice underwent dynamic imaging for 2.5 h to evaluate the optimal scan window and to assess the feasibility of delayed imaging with either 68 Ga or 18 F as radio-isotope. The SUV mean and SUV max values of 18 F-PSMA-11 suggest a wide scan time window as no significant difference was found between early (60 min p.i.) and delayed (150 min p.i.) imaging. TBR mean and TBR max values continued to rise up to 150 min p.i. This can be attributed to decreasing background activity due to fast radiotracer clearance. These preclinical results suggest an optimal  www.nature.com/scientificreports/ scan time window between 1 and 2 h p.i. to obtain the highest SUV mean and SUV max values. Rising TBR mean and TBR max values at later timepoints could potentially be beneficial for suspicious lesions that were unclear on early images. This corresponds with the results obtained in the Phase II study, which suggested early imaging at 1 h p.i. 14 . Based on the preclinical data, the scan time could potentially be extended to up to 2 h p.i. A preclinical study by Cardinale et al. 25 evaluating one LNCaP xenograft bearing mouse after administration of 25 MBq 18 F-PSMA-1007 revealed an SUV mean of approximately 1.1 in the tumor 10 min p.i., which remained constant up to 1 h and showed limited bone uptake (SUV mean of approximately 1) which was reduced by half over time.
The tumor was visible 20-40 min p.i. and displayed increasing image contrast over time. Comparable results were reported for 18 F-DCFPyL where five mice were injected with 2-10 MBq and underwent dynamic PET imaging for 60 min. SUV mean values reached a maximum 10 min p.i. and remained constant over time (1.1 ± 0.1 at 60 min p.i.) 26 . Similar trends regarding tumor uptake in function of time were found for 68 Ga-PSMA-11, although the curves for all uptake parameter were considerably lower in comparison with those for 18 F-PSMA-11. SUV mean reached its maximum value at 20 min p.i. and remained constant over time (1.05 ± 0.07 20 min p.i. to 0.97 ± 0.12 150 min p.i.).
Both for early (60 min p.i.) and delayed (150 min p.i.) acquisition, uptake parameters were significantly higher for 18 F-PSMA-11 compared to 68 Ga-PSMA-11 (except for TBR max and SUV max 150 min p.i.). Results on early vs delayed imaging (Fig. 4) suggest improved imaging with 68 Ga-PSMA-11 at later timepoints as SUV max , TBR mean and TBR max were significantly higher at 150 min p.i. Delayed imaging using 68 Ga-PSMA-11 seems to be favorable and may provide improved tumor visualization compared to early imaging, while limited additional benefits could be found for 18 F-PSMA-11 imaging at later timepoints. A comparable conclusion was reached in the Phase II clinical study where no additional lesions were found between 1 and 3 h p.i. for 18 F-PSMA-11 14 . Several clinical trials have evaluated delayed imaging with 68 Ga-labeled PSMA-targeting radiotracers such as 68 Ga-PSMA-11 and 68 Ga-PSMA-I&T. A study by Afshar-Oromieh et al. 27 reported higher lesion uptake and contrast at 3 h p.i. which lead to an increased detection rate. Schmuck et al. 28 confirmed improved lesion contrast, but only found a limited impact on detection rates due to higher image noise and low residual activity 3 h p.i. Rahbar et al. 29 and Derlin et al. 30 found no additional benefit to delayed imaging with 68 Ga-PSMA-11 because of high and variable urinary activity. However, combined with the administration of a diuretic, it could be beneficial for unclear lesions on early images and for improved assessment of the prostate gland/bed and pelvic lymph nodes. Since these studies do not report an unambiguous result, there is a need for further clinical research regarding the benefits of delayed imaging.
Even though increasing TBR values seem to be in favor of delayed acquisition, early imaging as soon as 20 min p.i. was shown to be feasible by Behesti et al. 31 , which would be beneficial in clinical practice due to the short half-life of 68 Ga.
Qualitative comparison of PET images revealed improved tumor visualization and delineation with 18 F-PSMA-11. This can be attributed to the lower positron energy of 18 F (0.65 vs 1.90 meV) resulting in a shorter positron range (R max 2.4 mm vs 9.2 mm), as well as the higher positron yield (97% vs 89%), which both contribute to a better image spatial resolution 9,32 . These observed differences will likely be less significant in clinical practice due to the difference in spatial resolution between preclinical (1.3 mm) and clinical PET cameras (4.5 mm), as the resolution is the limiting factor instead of isotope ranges 33 . This will be further investigated in a Phase 3 clinical trial (ClinicalTrials.gov identifier NCT03911310).
Ex vivo biodistribution of 18 F-PSMA-11 in healthy mice demonstrated a high %ID/g in the kidneys and bladder, which can be attributed to both renal clearance of the radiotracer as well as specific binding due to PSMA expression in mouse kidneys 34 . Lütje et al. reported lower 18 F-PSMA-11 uptake in the kidneys (36.7 ± 9.3%ID/g vs 94.2 ± 13.6%ID/g 1 h p.i. and 43.5 ± 5.7%ID/g vs 82.5 ± 10.8%ID/g 2 h p.i.) 35 . They also reported higher renal accumulation of 68 Ga-PSMA-11 which was in agreement with high SUV mean values in the bladder (Fig. 6). 18 F-PSMA-11 could therefore be more suitable for the detection of lesions in the proximity of the bladder although administration of sufficient fluids, co-administration of a diuretic and voiding prior to imaging may be sufficient to decrease activity in the urinary system. Low and constant liver values of 0.40%ID/g both at 1 h and 2 h p.i. as well as rapidly decreasing SUV mean values confirmed limited hepatobiliary clearance, which is advantageous for the detection of prostate cancer lesions in the pelvic region and/or abdominal cavity and potential liver metastasis 1 .
Potential defluorination of 18 F-labeled PSMA tracers is of great concern because free 18 F could lead to aspecific bone uptake, causing the detection of false positive lesions. Therefore, bone uptake was evaluated by in vivo PET imaging and ex vivo biodistribution. SUV mean values of the spine, femur, sternum and humerus showed decreasing time activity curves up to 30 min p.i., corresponding to tracer distribution in the blood, followed by a limited rise of SUV mean between 60 and 150 min p.i., although this increase was not significant. Ex vivo biodistribution showed similar results, no significant increase in bone uptake was found between 1 and 2 h p.i. The highest uptake in bone was seen 2 h p.i. in the humerus (1.96%ID/g) and skull (1.94%ID/g), which is considerably lower than tumor uptake (9.11%ID/g). Bone uptake was also lower compared to previously published results. Lütje et al. 35 reported bone uptake of 3.3 ± 0.6 and 5.0 ± 0.6%ID/g at 1 h and 2 h p.i. They administered additionally 10% free 18 F-AlF together with 18 F-PSMA-11, which evidently led to increased bone activity (7.1 ± 1.3%ID/g and 7.0 ± 0.8%ID/g at 1 h and 2 h p.i.) but did not cause interference on the visualization of subcutaneous xenograft tumors. A comparative study between 68 Ga-PSMA-11, 18 F-PSMA-1007 and 18 F-PSMA-11 by Ioppolo et al. 36 reported bone uptake of 1.5 ± 0.3%ID/g and 0.9 ± 0.1%ID/g 4 h p.i. for 68 Ga-PSMA-11 (n = 3) and 18 F-PSMA-1007 (n = 3) compared to 4.0 and 10.2%ID/g 1 h and 4 h p.i. for 18 F-PSMA-11 (n = 2) 36 , which was explained by rapid degradation due to instability of the HBED-CC and 18 F-AlF complex. Although there are contradicting results regarding stability of 18 F-PSMA-11 in serum 10,11,37,38 , ex vivo biodistribution results and PET images in this study as well as in the clinical trials did not suggest extensive tracer degradation, as the Phase 1 study showed only limited amounts of free fluoride in blood over time (increase of 1.4% and 2.5% at 50 versus 20 min p.i. and Scientific Reports | (2020) 10:21068 | https://doi.org/10.1038/s41598-020-78273-7 www.nature.com/scientificreports/ 90 versus 50 min p.i., respectively) 13 . Evaluation of possible interference of free 18 F on bone lesion visualization should be further investigated in a preclinical bone metastasis model. A limitation of this study was the difference in specific activity between the two PSMA-11 radiotracers. The specific activity of 18 F-PSMA-11 was set on 20 MBq/µg as this was practically achievable due to the longer halflife and the semi-automated production method, while the short half-life of 68 Ga and limited yield of a 68 Ge/ 68 Ga generator, especially at the end of its life cycle, only allowed lower specific activities of 1.5 MBq/µg. However, the difference in SA reflects a major advantage of 18 F-labeled radiotracers in clinical practice. Two mice were scanned per day and equal specific activities per radiotracer were aimed for.

Conclusion
This paper evaluated the intra-individual comparison of 18 F-PSMA-11 and 68 Ga-PSMA-11 for imaging of PSMA positive tumors. Both radiotracers showed high affinity for the PSMA receptor. All uptake parameters (except for SUV max and TBR max at 150 min p.i.) were significantly higher for 18 F-PSMA-11 compared to 68 Ga-PSMA-11. Delayed acquisition imaging with the latter may improve lesion detection compared to early imaging, while no additional benefits could be found for late 18 F-PSMA-11 imaging. No significant increase in bone uptake could be found. In the preclinical setting, 18 F-PSMA-11 demonstrated excellent imaging characteristics. Whether these can be translated to a clinical setting, will be further investigated in a Phase 3 clinical trial.