Background

Prostate-specific membrane antigen (PSMA) is overexpressed in nearly all primary prostate cancer [1], and PSMA expression further increases in de-differentiated, metastatic, or hormone-refractory disease [2, 3]. In recent years, PSMA targeting ligands have been investigated for molecular imaging and radioligand therapy of prostate cancer. Recent developments increasingly focus on 18F-based, rather than 68 Ga-based PSMA ligands for imaging given the advantages of 18F-based agents in terms of availability, ease of production, and image resolution.

Several 18F-based PSMA ligands have been developed with different kinetic profiles. The kidneys and the urinary tracts are critical organs in the development of novel PSMA ligands as intense urinary retention may interfere with accurate evaluation of the prostate and adjacent area. 18F-labelled, urea-based inhibitors of PSMA such as 18F-DCFBC and 18F-DCFPyL demonstrated diagnostic capabilities in clinical trials [4, 5], but they still have shortcomings due to urinary excretion as their limitation to a diagnostic use. 18F-PSMA-1007 was developed aiming for similar structure, biodistribution and tumour uptake compared to PSMA-617 which is currently used for radioligand therapy. 18F-PSMA-1007 is predominantly excreted via the hepatobiliary tract and has very low urinary retention but its optimal imaging timepoint (120 min) is suboptimal for routine clinical practice [6].

Radiohybrid PSMA (rhPSMA) ligands are a new class of fully theranostic agents that allow fast 18F synthesis and labelling with radiometals. They are silicon-fluoride-acceptor (SiFA)–conjugated radiopharmaceuticals that are labeled by isotopic exchange. One example, 18F-rhPSMA-7, has been shown to have a biodistribution typical of other established PSMA ligands [7] and its clinical performance for imaging patients with prostate cancer has been reported previously [8, 9]. 18F-rhPSMA-7 is comprised of four diastereoisomers (rhPSMA-7.1, rhPSMA-7.2, rhPSMA-7.3 and rhPSMA-7.4), of which rhPSMA-7.3 exhibits the most promising targeting characteristics for clinical translation and has been selected for further development based in preclinical results [10].

Here, we analyzed the preclinical biodistribution and dosimetry of 18F-rhPSMA-7 and 18F-rhPSMA-7.3 at different timepoints after a single intravenous administration in mice. We further assessed biodistribution and image quality of clinical 18F-rhPSMA-7 and 18F-rhPSMA-7.3 PET scans in patients with histopathologically proven prostate cancer.

Methods

Synthesis of 18F-rhPSMA-7 and 18F-rhPSMA-7.3

The rhPSMA-7 and rhPSMA-7.3 peptide precursors were labelled with 18F and dissolved in phosphate buffered saline solution as previously described [11]. The 18F-labeling of rhPSMA-7 and rhPSMA-7.3 was performed in a fully automated, Good Manufacturing Practice-compliant procedure using a GRP™ synthesis module (Scintomics, Fürstenfeldbruck, Germany).

Preclinical biodistribution and dosimetry

Severe combined immunodeficiency (SCID) mice were supplied by Charles River Laboratories. (Freiburg, Germany). The preclinical evaluation study was performed in accordance with the German Animal Welfare Act (Deutsches Tierschutzgesetz, approval #55.2-1-54-2532-216-2015). Male mice (≥ 6 weeks old) were used for the study after reaching sexual maturation.

The biodistribution study was performed at multiple timepoints; 10, 20, 40, 60, 120 and 180 min post-injection of 18F-rhPSMA-7, and at 10, 60, 120, 180 and 300 min post-injection of 18F-rhPSMA-7.3. Based on initial experiments exhibiting prolonged renal uptake for 18F-rhPSMA-7.3, a late timepoint (300 min) was applied for the final experiments. At each timepoint, 3–6 mice were injected intravenously in the tail vein with a mean 25.6 ± 3.6 MBq of 18F-rhPSMA-7 and 28.5 ± 4.8 MBq of 18F-rhPSMA-7.3, respectively. For the biodistribution study, the mice were dissected and samples collected from urine, blood, heart, lung, spleen, pancreas, liver, stomach (emptied), small intestine (emptied), large intestine (emptied), kidneys, bladder, testis, fat, muscle (partial, femoral), femur, tail and brain. An automatic gamma counter (PerkinElmer-Wallac, Waltham, USA) was used to measure count rate and the percentages of the injected dose (%ID and %ID/g) were calculated.

Of note, to achieve a similar number of timepoints for both calculations, the 18F-rhPSMA-7 10 min and 20 min timepoints were interpolated to a 15 min timepoint. The time-integral of activity for the accumulation in the investigated source organs (AUCs) were generated both with numerical integration and physical decay [12].

To extrapolate from preclinical data to human dosimetry, linear scaling of %ID from the mice by the ratio of the organ weights and total body weights of phantoms compared to humans was necessary [13]. Normal-organ radiation doses were estimated for the 70 kg standard adult anatomic model using time-dependent organ activity concentrations in %ID/g and total-body activities measured in the biodistribution studies in mice. Tissue activity concentrations in mice were converted to tissue fractional activities in the 70 kg standard adult using the relative fractional organ masses in the standard adult and the standard 25 g mouse. Time-dependent total-body activity was fit to an exponential function and the difference between the injected activity and the total-body activity was assumed to be excreted via the urine because activity concentrations in the liver and gastrointestinal tract were low at all timepoints studied.

Organ residence time was calculated by numerical integration using the trapezoidal rule and the rest-of-body. 18F residence times were calculated as the difference between the total-body residence time and the sum of the organ and urine residence times. The bladder contents residence time was estimated using the dynamic voiding model in the OLINDA/EXM 1.0 dosimetry software (Vanderbilt University, Nashville, TN, USA). Finally, the standard adult mean absorbed dose to organs (in µGy/MBq) and total effective dose (in µSv/MBq) were calculated using OLINDA/EXM 1.0 [14].

Human biodistribution

Patients

Data from patients with histopathologically proven prostate cancer who underwent a clinically indicated PET/CT with 18F-rhPSMA-7 or 18F-rhPSMA-7.3 between October 2017 and November 2018 were retrospectively analyzed. 18F-rhPSMA-7 PET/CT was performed in 47 patients (mean [range] age: 69.8 [52–80] years) with a mean injected activity of 324 (range, 236–424) MBq and mean acquisition time of 84 (range, 42–166) min. 18F-rhPSMA-7.3 PET/CT was performed in 33 patients (mean [range] age: 70.8 [57–85] years) with a mean injected activity of 345 (range, 235–420) MBq and mean acquisition time of 76 (range, 59–122) min. The mean prostate-specific antigen level at the time of imaging was 42.9 ng/mL (median range, 0–1459 ng/mL) for 18F-rhPSMA-7 and 20.0 ng/mL (range, 0–202 ng/mL) for 18F-rhPSMA-7.3, respectively. Patients received an injection of 20 mg of furosemide at the time of tracer application. A comparison of the disease stages and primary Gleason Scores of both cohorts are presented in Additional file 1: Table S5.

All patients gave written informed consent for the original procedure. All reported investigations were conducted in accordance with the Helsinki Declaration and with national regulations. This retrospective analysis was approved by the Local Ethics Committee (permits 290/18S and 99/19S) and the need for patient consent was waived. The administration of 18F-rhPSMA-7 and 18F-rhPSMA-7.3 was in accordance with The German Medicinal Products Act (AMG §13 2b) and the responsible regulatory body (Government of Oberbayern).

PET/CT imaging

All the PET/CT scans were obtained from the skull base to mid-thigh using a Biograph mCT Flow scanner (Siemens Medical Solutions, Erlangen, Germany). PET scanning was operated in 3D mode with an acquisition time of 1.1 mm/s in continuous table movement, and a diagnostic CT scan (240 mAs, 120 kV, 5 mm slice thickness) was acquired in the portal venous phase 80 s after the intravenous injection of an iodinated contrast agent (Imeron 400, Bracco Imaging Deutschland GmbH, Konstanz, Germany). Reconstruction of the PET images was performed based on iterative algorithms with an ordered-subsets expectation maximization (4 iterations, 8 subsets) followed by a post-reconstruction smoothing Gaussian filter (5 mm full width at one-half maximum).

Image analysis

For both quantitative and qualitative analyses PET datasets (non-Time-of-Flight/non-True X) were used. The maximum standardized uptake value (SUVmax) and the mean standardized uptake value (SUVmean) with an isocontour of 50% of the SUVmax were determined applying circular volumes of interest (VOIs) using OsiriX MD® 11.0.2 (Pixmeo SARL, Geneva, Switzerland).

The circular VOIs were placed over normal organs; parotid gland, submandibular gland, mediastinal aortic arch (blood pool), lungs, liver, spleen, pancreas, duodenum, kidneys, bladder, sacral promontory, and background (gluteus maximus muscle). Up to 3 lesions per patient were analyzed in decreasing order of SUVmax. Organ/tumour to background ratios (ratio-SUVmean, ratio-SUVmax) were calculated.

To evaluate overall image quality, non-specific blood pool activity and background uptake in bone/marrow was analysed using 3- or 4-point scales as previously described [7]. All the analyses were performed by a board-certified nuclear medicine physician.

Statistical analysis

Prior to analysis, the Kolmogorov–Smirnov test was used to assess the normality of the data distribution. The independent Student t-test was performed to compare means between groups for the normal parameters. The Mann–Whitney U test was conducted for the non-normal parameters. The Chi-square test or the Fisher's exact test was adopted to compare differences among groups for the analyses of ordinal variables. Data were expressed as mean ± standard deviation (SD) for continuous variables and frequencies (percentages, %) for categorical variables, respectively. All statistical analyses were performed using the IBM SPSS Statistics version 25 (IBM Inc. Armonk, NY, USA) and R version 3.5.2 (http://www.r-project.org). P values less than 0.05 were considered statistically significant.

Results

Preclinical biodistribution and dosimetry

Significant accumulation of radioactivity was observed (i.e. source organs) in the kidney, spleen, lung, liver and heart for both 18F-rhPSMA-7 and 18F-rhPSMA-7.3. The activity distributed to individual organs is presented for the source organs and whole organs in Additional file 1: Table S1 and S2, respectively. The AUCs in the source organs are summarized in Table 1, and only results calculated from the AUCs with extrapolation and physical decay were used in the following analyses. Radioactivity distribution of both 18F-rhPSMA-7 and 18F-rhPSMA-7.3 was found to be the highest in the kidneys at most timepoints. With respect to activity accumulation and clearance, a rapid clearance from blood and clearance to urine but relatively slow build-up in the kidneys was present for both 18F-rhPSMA-7 and 18F-rhPSMA-7.3.

Table 1 Time-integral of activity for accumulation in significant source organs (AUCs)
Full size table

Extrapolating the preclinical dosimetry data to humans revealed that at 1 h and 3.5 h bladder voiding intervals, the total effective doses for humans are 12.2 and 26.6 µSv/MBq for 18F-rhPSMA-7, and 12.8 and 21.7 µSv/MBq for 18F-rhPSMA-7.3, respectively (Table 2). The absorbed doses for the individual organs at 1 h and 3.5 h bladder voiding intervals are provided in Additional file 1: Table S3. On the individual organ level, the kidneys were among the organs receiving the highest absorbed doses for both 18F-rhPSMA-7 and 18F-rhPSMA-7.3. For both bladder voiding intervals, the absorbed dose in the kidneys was slightly higher with 18F-rhPSMA-7.3 (71.60 and 71.80 µGy/MBq at 1 h and 3.5 h, respectively) than with 18F-rhPSMA-7 (64.70 and 65.10 µGy/MBq at 1 h and 3.5 h, respectively; Additional file 1: Table S3).

Table 2 Dosimetry results using 1 h and 3.5 h voiding interval
Full size table

Human biodistribution in normal organs and tumours

Quantitative normal organ biodistribution was very similar for 18F-rhPSMA-7 and 18F-rhPSMA-7.3 (Fig. 1, Table 3). SUVmean were 16.9 versus 16.2 (parotid gland), 19.6 versus 19.9 (submandibular gland), 2.0 versus 1.9 (blood pool, p < 0.005), 0.7 versus 0.7 (lungs), 7.0 versus 7.3 (liver), 9.1 versus 8.4 (spleen), 32.4 versus 35.7 (kidney), 2.5 versus 2.8 (pancreas), 10.9 versus 11.0 (duodenum), 1.1 versus 1.3 (non-diseased bone, p = 0.005) and 4.6 versus 2.0 (bladder; p < 0.001) for 18F-rhPSMA-7 versus 18F-rhPSMA-7.3, respectively. Relative to the other organs, the kidneys exhibited particularly high uptake values of both rhPSMA ligands, but radiotracer retention in the urinary bladder was comparatively low. Moreover, the mean bladder SUVmean and ratio-SUVmean were significantly lower for 18F-rhPSMA-7.3 than for 18F-rhPSMA-7 (p < 0.001; Table 3). Qualitative analyses of image quality revealed no significant differences between 18F-rhPSMA-7 and 18F-rhPSMA-7.3 PET (Additional file 1: Table S4).

Fig. 1
figure 1

Human biodistribution in normal organs and tumours of 18F-rhPMSA-7 (A) and 18F-rhPSMA-7.3 using SUVmean. SMG: submandibular gland

Full size image
Table 3 Comparisons of the individual organ uptake and tumour uptake in human biodistribution
Full size table

Tumour uptake was analyzed in 89 lesions (26 primary tumours/local recurrences, 23 bone, 38 lymph node and 2 visceral metastases) and 63 lesions (14 primary tumours/local recurrences, 30 bone, 18 lymph node and 1 visceral metastases) for 18F-rhPSMA-7 and 18F-rhPSMA-7.3, respectively. Mean tumour SUVmean (20.0 ± 20.2 for 18F-rhPSMA-7 vs. 32.5 ± 42.7 for 18F-rhPSMA-7.3) and mean ratio-SUVmean (36.0 ± 45.5 for 18F-rhPSMA-7 vs. 50.9 ± 68.7 for 18F-rhPSMA-7.3) showed a trend towards higher tumour uptake values for 18F-rhPSMA-7.3 but without statistical significance (p > 0.05, Fig. 1, Table 3).

Figure 2 demonstrates the maximum intensity projection images of patients with normal biodistribution of 18F-rhPSMA-7 and 18F-rhPSMA-7.3. Additional file 1: Figure S1 demonstrates the maximum intensity projection images and PET/CT fused images of two patients with early biochemical recurrence who have undergone 18F-rhPSMA-7 and 18F-rhPSMA-7.3 PET/CT, respectively.

Fig. 2
figure 2

Maximum intensity projection images of patients with normal biodistribution of 18F-rhPSMA-7 (A) and 18F-rhPSMA-7.3 (B), respectively

Full size image

Discussion

Here, we conducted a series of preclinical and human biodistribution and dosimetry studies to compare the profiles of two lead rhPSMA ligands for prostate cancer imaging. The results of the preclinical elements of this study indicate that the biodistribution and dosimetry profiles for both 18F-rhPSMA-7 and its single diastereoisomer form, 18F-rhPSMA-7.3, are favourable for prostate cancer imaging. The human clinical biodistribution results confirm and extend the preclinical data to indicate that while the profiles of 18F-rhPSMA-7 and 18F-rhPSMA-7.3 are similar, the significantly lower bladder uptake, and the trend towards higher uptake in tumours with 18F-rhPSMA-7.3 suggest that it is the more promising agent for further clinical translation.

The AUCs in the mouse dosimetry studies showed that 18F-rhPSMA-7.3 exhibited appropriate profiles for human study. Compared with 18F-rhPSMA-7, kidney uptakes of 18F-rhPSMA-7.3 were higher, but its accumulation speed was relatively slow. Thus, dosimetry studies of 18F-rhPSMA-7.3 were planned to be performed at a late timepoint (300 min) to acknowledge this slow build-up in the kidneys. This kinetic profile could be helpful for PSMA ligand imaging. Since urinary excretion of the radiotracer is slow, excretion may not contribute to the kidney uptakes at early imaging timepoints, which in turn leads to lower bladder radioactivity levels and lower levels of background.

While direct conversion of preclinical dosimetry studies using mice into human data will not truly model the human profile, it is reasonable to assume that kinetic profiles are similar in humans. In our earlier work with 18F-rhPSMA-7, we recommended an early imaging timepoint (50–70 min) for 18F-rhPSMA-7 as lower tracer retention in the urinary bladder is an important feature for PSMA ligand PET imaging [7] and it is expected that PET/CT with 18F-rhPSMA-7.3 could also achieve good image quality at early imaging timepoints, especially given the significantly lower bladder SUV that we observe here.

The extrapolated effective doses for humans calculated from the preclinical dosimetry show estimated effective doses of both rhPSMA ligands that are comparable with those of other 18F-based PSMA ligands. The estimated effective doses using 1 h voiding interval were 12.2 µSv/MBq for 18F-rhPSMA-7 and 12.8 µSv/MBq for 18F-rhPSMA-7.3, respectively, leading to an effective dose of less than 5 mSv after injection of 370 MBq of either rhPSMA ligand. Radiation dosimetry studies revealed a mean effective dose of 19.9 µSv/MBq for 18F-DCFBC [4], 13.9 µSv/MBq for 18F-DCFPyL [5], and 22.0 µSv/MBq for 18F-PSMA-1007 [6]. 18F-rhPSMA-7 and 18F-PSMA-7.3 also provide favourable effective doses in comparison with 18F-FDG, which has a reported effective dose of 19.0 µSv/MBq [15]. In the present study the kidneys received among the highest doses with both rhPSMA ligands. This is in line with data previously reported for 18F-rhPSMA-7.3 in healthy volunteers [16]. Tolvanen et al. reported the kidneys to be one of the organs with the highest mean absorbed doses (172 µGy/MBq at a 3.5 h voiding window) [16]. All PSMA ligands share similar physiological distribution; organ absorbed doses are known to differ minimally among various PSMA ligands [6] and the data for 18F-rhPSMA-7.3 appear comparable with other agents.

Our clinical data demonstrate that, in humans, the normal biodistribution of 18F-rhPSMA-7 and 18F-rhPSMA-7.3 are similar, but with some slight, but notable differences that are relevant to prostate cancer imaging. Defining the extent of urinary excretion of novel PSMA ligands is imperative given that PSMA ligands that are mainly excreted via the urinary tract often show high radiotracer uptake in the bladder that can interfere with image interpretation in the prostate region [17, 18]. Our data show low relative bladder uptake with both rhPSMA ligands, but, of the two ligands, bladder uptake is significantly lower with 18F-rhPSMA-7.3 than 18F-rhPSMA-7. Moreover, while the tumour uptake of both rhPSMA ligands is higher than the kidney uptakes, facilitating easy localization of tumours with both rhPSMA ligands via PET/CT, there was a trend towards higher tumour SUV with 18F-rhPSMA-7.3. The qualitative PET/CT image interpretation analysis confirm the suitability of both agents for prostate imaging.

Some limitations to the present study should be acknowledged. First, direct conversion of preclinical dosimetry studies to humans may not truly reflect human dosimetry, and thus, interpretation of the estimated effective doses reported here should be performed with caution. Second, our observations of low bladder retention of both ligands may be confounded by the application of furosemide at the time of tracer injection which is routinely done at our institution. It is known from 68Ga-PSMA-11 that this can substantially lower bladder retention due to forced bladder emptying prior to the PET scan. Third, we acknowledge a trend towards higher tumour SUV using 18F-rhPSMA-7.3 compared with 18F-rhPSMA-7. However, an inter- rather than an intra-patient comparison can be affected by several clinical variables (e.g. tumour stage, differences in PSMA-expression, prior treatments). Fourth, definitive conclusions about the uptake characteristics in tumors comparing both cohorts are limited as they base on different patient cohorts. Despite the patient cohorts are weighted for disease stage differences in the distribution of primary Gleason Scores and PSA at imaging have to be acknowledged.

Conclusion

Radiation dosimetry is favourable both for 18F-rhPSMA-7 and 18F-rhPSMA-7.3 with an estimated effective dose of less than 5 mSv, assuming a 1 h voiding interval after injection of 370 MBq of the radiotracer. Human biodistribution of 18F-rhPSMA-7 and 18F-rhPSMA-7.3 were similar to each other, but significantly lower bladder uptake and a trend towards higher uptake in tumour lesions was seen with 18F-rhPSMA-7.3.