Comparison of urinary bladder activity from [⁶⁸Ga]Ga-PSMA-11 and [¹⁸F]PSMA-1007 in patients with prostate cancer undergoing PSMA-ligand PET/CT

Abstract

Aim

PET/CT with radiolabelled PSMA-ligands has become the mainstay of diagnostic imaging for prostate cancer. The several available PSMA tracers differ modestly with respect to their avidity for lesions, but show distinct patterns of off-target uptake and routes of excretion. We sought to compare urinary bladder activity from [⁶⁸Ga]Ga-PSMA-11 and [¹⁸F]PSMA-1007 with regard to their differing renal elimination, which can interfere in the detection of perivesical lesions.

Materials and methods

We retrospectively compared urine activity in patients undergoing PET/CT after injection of either [⁶⁸Ga]Ga-PSMA-11 patients in combination with hydration and diuresis (n = 132) or [¹⁸F]PSMA-1007 patients without hydration and diuresis (n = 100). Furthermore, we compared urinary bladder uptake from [¹⁸F]PSMA-1007 with lesional uptake in locally recurrent prostate cancer and locoregional lymph node metastasis, and performed an exploratory analysis of urine activity from [¹⁸F]PSMA-1007 relative to patient characteristics.

Results

Urine activity from [⁶⁸Ga]Ga-PSMA-11 was normally distributed, being lowest in acquisitions 90 min p.i. with oral hydration and a single dose of furosemide injected 60 min after the tracer. In contrast, urine activity from [¹⁸F]PSMA-1007 followed an exponential distribution in the population, with approximately one quarter of cases exceeding an SUV_mean threshold of 7. Urine activity of [¹⁸F]PSMA-1007 exceeded the uptake in prostate cancer lesions in more than 20% of patients.

Conclusions

Urinary uptake from [¹⁸F]PSMA-1007 was in the range of that seen in [⁶⁸Ga]Ga-PSMA-11 patients with hydration and diuretics. Some 25% of the [¹⁸F]PSMA-1007 cases showed problematic urinary uptake, which would potentially eclipse the uptake in prostate cancer lesions.

Introduction

Spillover of progressively increasing urinary signal from radiopharmaceuticals into the surrounding structures limits the assessment of perivesical structures by PET/CT, for instance in patients scanned for prostate cancer (PC) with PSMA-targeted radioligands (Freitag et al. 2017). This is an issue of particular importance in the context those PSMA-targeted diagnostic radioligands that exhibit high renal excretion. This issue has led to the adoption of hydration and forced diuresis protocols prior to image acquisition at some centres (Piron et al. 2020; Donswijk et al. 2022; Derlin et al. 2016; Wondergem et al. 2019; Uprimny et al. 2021; Bayerschmidt et al. 2021). In the case of [⁶⁸Ga]Ga-PSMA-11 PET, a protocol with administration of furosemide in conjunction with late imaging improved lesion-to-bladder ratios, ultimately motivating its adoption for routine clinical imaging at our center (Alberts et al. 2021a).

A search for a diagnostic ligand synthesizable at larger scales led to the radiofluorinated ligand [¹⁸F]PSMA-1007, which, in contrast to PSMA-11, fortuitously (and somewhat idiosyncratically) undergoes predominantly hepatobiliary excretion and less urinary excretion in humans (Cardinale et al. 2017; Giesel et al. 2017). This property facilitated the adoption of [¹⁸F]PSMA-1007 as theoretically allowing for improved assessment of intraprostatic tumor localisation and detection of perivesical tumor masses due to absent spillover (Privé et al. 2021). However, in clinical practice, tumor detection rates are broadly comparable amongst available PSMA radioligands (Alberts et al. 2021b). Moreover, in a recent retrospective study, [¹⁸F]PSMA-1007 showed considerable variability in urinary bladder signal between individuals (Allach et al. 2022). This potentially compromises the general sensitivity and specificity of [¹⁸F]PSMA-1007 PET/CT for the detection of perivesical lesions. Additionally, the considerable expansion in the indications of PSMA-targeted PET/CT (Weiner et al. 2024) has substantially modified the functional requirements of PSMA-targeting radioligands according to the clinical context. In particular, unspecific bone uptake of [¹⁸F]PSMA-1007 (Vollnberg et al. 2022; Rizzo et al. 2024; Popescu et al. 2024) and the relatively lower signal from [⁶⁸Ga]Ga-PSMA-11 in healthy liver parenchyma (Popescu et al. 2024) could influence the discernment of metastases. In primary staging, where a prior biopsy has confirmed local disease and pelvic MRI is available to adjudicate invasion into periprostatic structures, the primary focus of PSMA-PET/CT is to exclude the presence of metastases inaccessible to surgery or primary radiotherapy. The renal excretion of [⁶⁸Ga]Ga-PSMA-11, is thus of little consequence, while its lower physiological bone uptake may enable a more confident rule-out of osseous metastases. This trade-off, however, does not hold for patients examined for biochemical recurrence, as the pre-test probability for local disease is high and the localization of recurrent disease to the vesical neck may enable targeted salvage radiotherapy (Dragonetti et al. 2025). Although the reduced radioactivity signal within the urinary bladder is a key rationale for the application of [¹⁸F]PSMA-1007 in such patients, we have observed many cases with intravesical activity comparable to that typical of [⁶⁸Ga]Ga-PSMA-11. Given this background, we undertook a systematic comparison of the urinary bladder activity of the both mentioned tracers.

Materials and methods

The Canton of Berne research ethics committee granted IRB approval for this retrospective study (KEK file number 2020 − 01162). Patients examined within the approved time frame were screened chronologically. Starting in October 2018 until May 2020, we included all available patients who underwent PET/CT for imaging prostate cancer after injection of [⁶⁸Ga]Ga-PSMA-11 (n = 132) or [¹⁸F]PSMA-1007 (n = 100). We extracted data on demographics and disease characteristics from the available electronic patient records. Data processing and storage were compliant with applicable Swiss privacy and data protection laws.

Imaging protocol

For all patients, [⁶⁸Ga]Ga-PSMA-11 and [¹⁸F]PSMA-1007 had been obtained commercially from SWAN Isotopen AG (Berne, Switzerland) and administered by intravenous bolus injection. In each case, we acquired low-dose attenuation CTs (120 keV, adaptive tube current) extending from the vertex to the mid-thighs on a Siemens Biograph Vision 600 scanner (Siemens Healthineers, Erlangen, Germany), followed by emission tomography with continuous bed motion. The cohort of patients examined with [⁶⁸Ga]Ga-PSMA-11 were orally hydrated with one liter of water 30 min post injection (p.i.), followed by injection of 20 mg furosemide (FRS) 60 min p.i. Thereafter, we obtained single frame PET/CT recordings at 90 and 150 min p.i. in (n = 76) cases. In a second [⁶⁸Ga]Ga-PSMA-11 patient cohort (n = 56), an additional dose of 10 mg FRS was injected at 120 min p.i. between acquisitions at 90 and 150 min p.i. The group examined with [¹⁸F]PSMA-1007 (n = 100) did not receive diuretics or hydration. Images were reconstructed iteratively with the manufacturer’s implementation of an ordered subset expectation maximization algorithm (TrueX TOF, 5 subsets, 4 iterations) into a 440 × 440 voxel matrix with 4 mm slice thickness and post-reconstruction smoothing by a 2 mm Gaussian filter.

Image evaluation

Images as reconstructed for routine clinical reporting were downloaded into Syngo.Via (Siemens Healthineers, Munich, Germany) for the present analysis. We determined for each case the urinary SUV_mean and lesion SUV_max and SUV_peak measurements in locally recurrent lesions and lymphatic node metastases using the isocontour VOI tool (40% intensity threshold). Where multiple lesions were present, the most intense lesion was chosen for analysis. In some cases, bladder segmentation proved impossible due to surrounding PSMA-avid structures (e.g., ileum) or low urinary bladder activity. In these cases, we manually placed a spherical VOI within the urinary bladder.

Statistical analysis

Statistical analysis was performed using SPSS Version 25 (IBM Corporation, Armonk, USA). Distributions were assessed by the Kolmogorov-Smirnov test. Spearman rank correlations served to assess relationships of bladder uptake and lesion to bladder ratio (LBR). Between-group comparisons were made using the Kruskal-Wallis test. Statistical significance was defined as p ≤ 0.05 with Bonferroni correction for multiple testing.

We calculated the population distributions of intravesical urine activities for the two tracers, and then tested the hypothesis that [⁶⁸Ga]Ga-PSMA-11 administration with FRS pretreatment would give lower intravesical signal and higher tumor-to-urine ratio as compared to [¹⁸F]PSMA-1007 without FRS. In addition, we made an exploratory analysis of the population distributions of [¹⁸F]PSMA-1007 ratios of urinary activity to locally recurrent and distal lymphatic metastases and [¹⁸F]PSMA-1007 urine uptake as a function of tumor stage, clinical characteristics, and demographics.

Results

Patient characteristics

Demographic data on patients examined with [⁶⁸Ga]Ga-PSMA-11 and [¹⁸F]PSMA-1007 are presented in Table 1.

Table 1.

Patient characteristics (median ± standard deviation / range in parentheses)

Radiotracer	[68Ga]Ga-PSMA-11		[18F]PSMA-1007
Scanning protocol	90 min p.i. + 150 min p.i. FRS 20 mg 60 min p.i.	90 min p.i. + 150 min p.i. FRS 20 mg 60 min + 10 mg 120 min p.i.	120 min p.i. no FRS
Age	68 ± 7	69 ± 6	71 ± 7
Gleason score	7 (6–10)	7 (6–9)	7 (6–10)
Serum PSA	1.1 ± 33	0.5 ± 46	2.6 ± 49

Urinary bladder activity of PSMA-targeting tracers

As demonstrated by Fig. 1, for [⁶⁸Ga]Ga-PSMA-11, urinary SUV_mean at 90 min p.i. ranged from 1.2 to 79.9 (median 5.2, IQR 3.7–6.7), while urinary SUV_mean at 120 min p.i. ranged from 2.2 to 65.4 (median 10.9, IQR 8.0–16.5) for patients receiving 20 mg FRS 90 min p.i., versus 3.3–25.8 (median 6.5, IQR 5.5–8.8) for those receiving 20 mg FRS at 90 and 10 mg FRS at 120 min p.i., as described above. The SUV_mean values followed normal distributions, albeit with substantial deviations (p < 0.0002, Kolmogorov-Smirnov test) arising from right-hand side tails of the distributions. For [¹⁸F]PSMA-1007 urinary SUV_mean ranged widely from 0.42 to 25.8 (median 3.6, IQR 1.7–7.1), best described by an exponential distribution (Kolmogorov-Smirnov test, p = 0.166), indicating a preponderance of cases with low bladder uptake. Histograms of urinary SUV_mean for [⁶⁸Ga]Ga-PSMA according to FRS regimen and for [¹⁸F]PSMA-1007, along with normal and exponential fittings of the respective distributions are shown in Fig. 2.

Fig. 1.

SUV_mean of urinary bladder content for different imaging agents and protocols (A). Histograms of urinary SUV_mean for [⁶⁸Ga]Ga-PSMA-11 according to FRS regimen and for [¹⁸F]PSMA-1007, along with the respective distributions (B, [⁶⁸Ga]Ga-PSMA-11 fitted with normal distributions, [¹⁸F]PSMA-1007 fitted with exponential distribution)

Fig. 2.

[¹⁸F]PSMA-1007 bladder-to-lesion ratios for locoregional recurrence (A) and pelvic lymph node metastasis (B)

Comparison of distributions by one-factorial Kruskal-Wallis analysis of variance with post-hoc pairwise comparisons revealed Bonferroni-corrected, statistically significant differences in urinary activity between the [⁶⁸Ga]Ga-PSMA-11-based protocols with single FRS administrations at 90 and 150 min p.i. (p < 0.001). The urine activities did not differ significantly between [⁶⁸Ga]Ga-PSMA-11 recorded 90 min p.i. versus [¹⁸F]PSMA-1007 recorded 120 min p.i. (p = 0.16) and [⁶⁸Ga]Ga-PSMA-11 recorded 90 min p.i. compared to [⁶⁸Ga]Ga-PSMA-11 recorded 150 min p.i. after two successive doses of FRS (p = 0.075).

Lesional uptake of [¹⁸F]PSMA-1007 in relation to urinary bladder activity

In patients scanned with [¹⁸F]PSMA-1007, pathologic lesions suggestive of locoregional PC lesions were reported in 74/100 patients, with SUV_max ranging from 1.3 to 52 (median 10.2, IQR 5.6–17.2) and SUV_peak from 0.7 to 33.7 (median 4.8, IQR 2.8–9.6). Tumor-to-urine ratios ranged from 0.29 to 56.6 (median 3.0, IQR 1.14–5.4) for SUV_max measurements and 0.15 to 175 (median 12.4, IQR 7.0 to 25.2) for SUV_peak measurements (Fig. 3A). In 17/100 patients, urine SUV_mean exceeded tumor SUV_max, while in 28/100 patients, the urine SUV_mean exceeded tumor SUV_peak values.

Fig. 3.

[¹⁸F]PSMA-1007 urinary activity by initial T/N stage (A, B), resection status (C), previous therapy (D) and initial Gleason score (E). Correlations of [¹⁸F]PSMA-1007 urinary activity with patient age (F), injected activity (G) and PSA (H), illustratively showing locally estimated smoothed lines, RPV: radical prostatovesiculectomy, RT: radiotherapy

PSMA-avid locoregional lymph nodes were reported in 52/100 [¹⁸F]PSMA-1007 patients, with tumor-to-urine ratios ranging from 0.47 to 90.8 for SUV_max (median 4.4, IQR 1.6–8.0) and 0.14 to 28.8 (median 1.7, IQR 1.0–3.1) for SUV_peak measurements (Fig. 3B). In 7/100 patients, bladder SUV_mean surpassed lymph node SUV_max, and in 13/100 patients, the urine SUV_mean exceeded the lymph node SUV_peak.

Exploratory analysis of urinary bladder activity from [¹⁸F]PSMA-1007

Urinary SUV_mean did not differ between T groups (p = 0.47, Kruskal-Wallis test), N groups (p = 0.21, Kruskal-Wallis test), or according to R status (p = 0.09, Kruskal-Wallis test, Fig. 3A-C). There were no statistically significant differences in urinary SUV_mean between the different treatment options (radical prostatovesiculectomy, radiotherapy or both, Kruskal-Wallis test, p = 0.18, Fig. 3D), nor were there any significant differences in bladder SUV_mean according to Gleason scores (p = 0.51, Kruskal-Wallis test, Fig. 3E).

Urinary bladder activity did not correlate with patient age at time of examination, injected activity, or PSA level (Fig. 2A-C, Spearman rank correlation, p = 0.778 for age, p = 0.931 for injected activity and p = 0.172 for serum PSA). Similarly, there were no statistically significant associations for lesion-to-urine ratios (Spearman rank correlation, local recurrence: p = 0.32 for age, p = 0.56 for activity, and p = 0.14 for serum PSA – locoregional lymph nodes: p = 0.25 for age, p = 0.61 for activity, and p = 0.97 for serum PSA).

Lesion-to-bladder ratios for local recurrence did not differ between T groups (p = 0.43, Kruskal-Wallis test), N groups (p = 0.13, Kruskal-Wallis test), or according to R status (p = 0.89, Kruskal-Wallis test). There were no statistically significant differences in LBRs according to treatment option (radical prostatovesiculectomy, radiotherapy, or both, Kruskal-Wallis test,p = 0.21), nor were there any significant differences in LBRs by Gleason scores (p = 0.41, Kruskal-Wallis test). Similarly, there were no statistically significant differences for LBRs calculated for locoregional lymph nodes (data not shown).

We present an example of difficulties in reading arising from elevated urinary activity in Fig. 4. An 81-year-old Caucasian male with previously resected PC had undergone [¹⁸F]PSMA-1007-PET/CT (Fig. 4A, MIP, SUV 0–10) for staging of biochemically recurrent PC. Under Syngo.Via’s default viewing conditions (hot metal scale, SUV 0–5), there was no gross abnormality present within the urinary bladder (Fig. 4B). Upon expanding the hot metal scale (SUV 0–20), the same section reveals intramural metastasis in the bladder wall (white arrow, Fig. 4C).

Fig. 4.

Intense urinary bladder accumulation of [¹⁸F]PSMA-1007 in an 81-year-old man with a biochemical recurrence after radical prostatovesiculectomy. Maximum intensity projection, greyscale SUV 0–10 (A). Transaxial fused slice through the urinary bladder, hot metal scale SUV 0–5 (B). Transaxial fused slice though the urinary bladder, hot metal scale SUV 0–20 (C). Arrow in 4 C points to intramural metastasis in the bladder wall, with the potential of being obscured by high urinary bladder activity

Discussion

Spillover of tracer signal from the urinary bladder can interfere in the detection of pervesical lesions, leading to false negative readings. In this study, we sought to evaluate the magnitude of urine activity arising in PSMA-targeted images in relationship to perivesical PC lesions, depending on the choice of PSMA ligand or imaging/diuresis protocol. We compared the proclivity of [¹⁸F]PSMA-1007 imaging to impart a urinary signal to our previous findings with [⁶⁸Ga]Ga-PSMA-11 imaging protocols incorporating diuresis with FRS. We observed a wide range of urinary activity in [¹⁸F]PSMA-1007 images, and likewise with [⁶⁸Ga]Ga-PSMA-11 imaging, with considerable overlap in the distributions of SUV_mean values. Notably, [¹⁸F]PSMA-1007 and the 90-minute [⁶⁸Ga]Ga-PSMA-11 protocol resulted in similar urinary bladder activity (p = 0.16, Kruskal-Wallis test). Examining the distributions of bladder SUV_mean values in Fig. 1B, [⁶⁸Ga]Ga-PSMA-11 results relatively consistently gave an SUV_mean of 4–6, and only a few outliers with SUV_mean > 7 when using hydration and FRS and acquisition 90 min p.i. On the other hand, [¹⁸F]PSMA-1007 yielded urinary values below 4 in approximately half of patients, albeit with a broad tail in the distribution, with approximately 25% of patients exhibiting urinary SUV_mean of > 7. Since previous work proposed lesional SUV_max > 7 as an upper threshold between unspecific/reactive tissue uptake of [¹⁸F]PSMA-1007 and frank malignancy (Orevi et al. 2022), patients with urinary activity exceeding this threshold may be at risk for false negative perivesical results due to signal spillover from the bladder. Notably, the late imaging acquired 150 min p.i. for [⁶⁸Ga]Ga-PSMA-11 resulted in higher urine activity than both the 90-minute [⁶⁸Ga]Ga-PSMA-11 protocol and the 120 min [¹⁸F]PSMA-1007 recordings (p < 0.001, Kruskal-Wallis tests), which was incompletely resolved by injection of a second dose of 10 mg FRS 120 min p.i. (p = 0.075 relative to the 90 min frame). We therefore assume that 10 mg of FRS is not sufficient to suppress completely the radioactivity signal within the urinary bladder. A higher dose, e.g. 20 mg, could prove to be better suited.

To assess better the possible relationship between high urinary activity and eclipsing of perivesical lesions, we calculated bladder-to-lesion ratios in well-defined locally recurrent disease and in local lymph node metastases. In 17% of [¹⁸F]PSMA-1007 patients, urinary activity exceeded the uptake measured in lesions indicative of recurrent prostate cancer. Less frequently (7%), this also applied to locoregional lymph node metastases. We take this as an indication that urinary activity could therefore mask lesions close to the bladder, particularly at the bladder neck in the prostate resection bed. Indeed, comparing a protocol with and without diuresis in [⁶⁸Ga]Ga-PSMA-11 showed additional lesions in 5.2% of patients (Alberts et al. 2021a).

Despite the predominantly hepatobiliary elimination of [¹⁸F]PSMA-1007, we frequently found high urinary activity in similar concentration ranges to that of [⁶⁸Ga]Ga-PSMA-11 with FRS and hydration. We have earlier proposed a serial FRS protocol for reducing [⁶⁸Ga]Ga-PSMA-11 urinary activity at 90 min and 150 min p.i. (Alberts et al. 2021a). However, even using the single-time point hydration and diuresis protocol outlined above, and with acquisition after 90 min, [⁶⁸Ga]Ga-PSMA-11 afforded scans with low urinary activity. One such representative case is depicted in Fig. 5. On the other hand, we have no records of how many patients had to interrupt their PET/CT due to the urge to urinate, because this data was not collected. Such cases do occur, although in our experience they are not very common.

Fig. 5.

75-year old patient examined for recurrent prostate cancer by PET/CT 90 min after injection of [⁶⁸Ga]Ga-PSMA-11 with 20 mg furosemide injected 60 min after tracer loading. (A: maximum intensity projection, greyscale SUV 0–5. B: Fused sagittal slice through urinary bladder and infundibular recurrence (hot metal scale, SUV 0–5). C: Fused transaxial slice through urinary bladder and infundibular recurrence (hot metal scale, SUV 0–5)

We note some limitations of our analyses. Despite our best efforts to obtain homogeneity, the sample size of 232 patients may lack sufficient power for the detection of small differences and weak associations. Furthermore, we did not restrict our dataset to lesions with histological confirmation, but rely on the demonstrated high positive predictive value of PSMA-targeted PET/CT for detection of prostate cancer metastasis (Light et al. 2024). Lastly, as our data derive from routine clinical practice, a group of patients undergoing PET/CT after the injection of [¹⁸F]PSMA-1007 with diuresis was unavailable, as this approach was not common practice or recommended by applicable guidelines at the time of scanning. The findings above would, thus, need confirmation in a study providing diuresis to patients undergoing [¹⁸F]PSMA-1007 to reveal the true number of lesions otherwise obscured by urinary activity. While a recent randomised controlled trial demonstrated lower urinary bladder uptake from [¹⁸F]PSMA-1007 after i.v. FRS and oral or i.v. hydration, neither a comparison to [⁶⁸Ga]Ga-PSMA-11 nor data on clinical impact were made available (Khalimon et al. 2025). Further prospective, randomised trials are needed to explore the clinical impact of these findings and compare the positive and negative predictive value of both radiotracers under similar conditions. Similarly, further exploratory work is necessary to optimize scanning protocols. We note that present FRS dosages and timing of injections were based on physiological considerations and clinical feasibility, but have not been subject to prospective trials. Two-dose FRS regimens employing higher doses than currently used may further reduce urinary activity.

Conclusion

We demonstrate substantial overlap in the distributions of urinary SUV_mean values between [¹⁸F]PSMA-1007 alone and [⁶⁸Ga]Ga-PSMA-11 along with hydration and FRS. Results did not indicate important group differences in the end-point of urinary uptake. However, a clinically relevant proportion of the patients examined with [¹⁸F]PSMA-1007 PET/CT exhibited intravesical activity corresponding to an SUV_mean >7, which risks eclipsing perivesical lesions. Indeed, in agreement with previous reports, approximately one in four patients scanned with [¹⁸F]PSMA-1007 demonstrated urinary bladder SUV_mean exceeding the perivesical lesional SUV_max and SUV_peak. This likely imparts a risk of false-negative reports of PC lesions adjacent to the urinary bladder, thereby diminishing the advantage of [¹⁸F]PSMA-1007 over the predominantly renally eliminated tracer, [⁶⁸Ga]Ga-PSMA-11. Although [¹⁸F]PSMA-1007 had lower renal excretion in some patients, this was not the case for all patients. In contrast, the urinary activity in [⁶⁸Ga]Ga-PSMA-11 could be adequately and reliably mitigated with diuresis, resulting in lower overall bladder activities.

Author Contribution

KK: data collection, analysis, visualisation, interpretation; writing and editing. PKC: data interpretation; writing and editing. ILA: data interpretation; writing and editing. AOR: writing and editing. AAO: data interpretation, writing and editing.

Funding

KK was supported by an ETH MedLab Fellowship and a research grant by the International Centers for Precision Oncology Foundation. PC, ILA, AOR and AAO declare no relevant funding.

Data availability

Data may be made available upon reasonable request, subject to and limited by applicable Swiss data protection and privacy law.

Declarations

Ethical approval

All patients included in this manuscript signed a written informed consent form allowing anonymized evaluation and publication of their data. This evaluation was approved by the ethics committee of the University of Bern (KEK-Nr.2020 − 01162).

Conflict of interest

KK: Speaking fees & travel stipend, Novartis Pharma Schweiz AG, Rotkreuz, Switzerland; PKC: no conflicts of interest, ILA: speaking fees from Novartis, Canada, AAO: honoria as a member of an advisory board of Novartis, AOR: no relevant conflict of interest.