Comparison of PET imaging with a ⁶⁸Ga-labelled PSMA ligand and ¹⁸F-choline-based PET/CT for the diagnosis of recurrent prostate cancer

For this study we originally selected 38 male patients who underwent both choline-based PET/CT and ⁶⁸Ga-PSMA PET/CT within a time window of 30 days. One of the patients was deleted from the study as the histological evaluation revealed multiple metastatic and converging lymph node packages. This precludes radiological differentiation of individual lymph nodes and a meaningful quantitative comparison of lymph node detection rate between choline and PSMA PET/CT. Therefore, the total number of patients analysed in this study was 37. In all cases there was suspected progressive disease following prior conventional treatment of PC (e.g. hormone therapy, chemotherapy, radiation therapy and/or surgery).

All patients signed a written informed consent form for the purpose of anonymized evaluation and publication of their data. The quantity of patients presented in this study reflects our data collected during the past 3 years. Follow-up could be conducted in 33 of 37 patients up until the time this manuscript was submitted.

With the exception of patients 26 and 29 (organizational reasons), all patients were first investigated by ¹⁸F-fluoromethylcholine PET/CT. This is currently the method of choice in routine clinical practice. However, due to unsatisfying imaging results of choline PET/CT and to evaluate possible treatment with ¹³¹I-labelled PSMA ligands (therapy data not yet published), further investigation using ⁶⁸Ga-PSMA PET/CT was suggested. This additional PET scan was conducted only in case both patients and their referring physicians gave their consent.

All reported investigations were conducted in accordance with the Helsinki Declaration and with our national regulations. This study was approved by the Ethics Committee of the University of Heidelberg (permit S-321/2012).

Patient characteristics are summarized in Table 1. Twenty-eight patients had previously undergone prostatectomy (Table 1, patients 1, 3–4, 7–8, 10–16, 19, 22–23 and 25–37), whereas nine were treated with prior radiation therapy and androgen deprivation without surgical removal of the prostate. The average age was 69.3 ± 7.1 years (range 57–85, median 70.0) with a mean Gleason score (GSC) of 7.4 ± 1.1 (range 5–9, median 7.0) and a mean PSA level of 11.1 ± 24.1 ng/ml (range 0.01–116, median 4.0 ng/ml). PSA was measured in blood samples taken at the time of the first PET/CT scan. The average time between both investigations was 12.1 ± 8.4 days (range 1–30, median 11.0).

Fifteen patients (3, 5–14, 16–17 and 19–20) of the present study have coincidentally been analysed in a different study evaluating the biodistribution of the PSMA tracer in humans [26]. This concordance was not intentional. The authors are convinced that these patients had to be integrated into the present manuscript instead of being excluded from the novel study. Their inclusion improves data quality and confirms the findings of this study.

⁶⁸Ga-PSMA PET/CT was obtained with the ⁶⁸Ga-labelled HBED-CC conjugate of the PSMA-specific pharmacophore Glu-NH-CO-NH-Lys that was synthesized as described previously [16]. ⁶⁸Ga³⁺ was obtained from a ⁶⁸Ge/⁶⁸Ga radionuclide generator and complexed with the HBED-CC conjugate as previously published [16, 17]. The final product was formulated in isotonic phosphate-buffered saline (PBS) with subsequent sterile filtration. The radiolabelling and purification of the PSMA ligand was done using an automated module. As a consequence, the radiochemical yield was not determined regularly. Typically, the labelling efficiency (radiochemical yield) is >98 % as determined by module validation. The ⁶⁸Ga-PSMA complex solution was applied to patients via an intravenous bolus (mean 139.6 ± 46.3 MBq, range 59–263 MBq, median 132 MBq). Targeted ⁶⁸Ga-PSMA was 2 MBq/kg. Variation of injected radiotracer activity was caused by the short half-life of ⁶⁸Ga and variable elution efficiencies obtained during the lifetime of the ⁶⁸Ge/⁶⁸Ga radionuclide generator. However, in our experience with PSMA PET/CT during the last 3 years, all injected activities were sufficient in detecting PC (see also in the “Discussion”). All injections contained 2 nmol PSMA ligand resulting in a median specific radioactivity of 66 GBq/μmol.

Choline-based PET/CT was performed with ¹⁸F-labelled fluoromethylcholine (IASOcholine®, Iason, Graz, Austria). The production was according to common radiopharmaceutical standards and regulations. The choline solution was applied to patients via an intravenous bolus (mean 241.6 ± 59.4 MBq, range 114–374 MBq, median 237 MBq). Targeted choline activity was 3 MBq/kg.

A non-contrast-enhanced CT scan was performed 1 h post tracer injection using the following parameters: slice thickness of 5 mm, increment of 0.8 mm, soft tissue reconstruction kernel, 130 keV and 80 mAs. Immediately after CT scanning, a whole-body PET was acquired in 3D (matrix 164 × 164). For each bed position (16.2 cm, overlapping scale 4.2 cm) we used 4-min acquisition time with a 15.5-cm field of view (FOV). The emission data were corrected for randoms, scatter and decay. Reconstruction was conducted with an ordered subset expectation maximization (OSEM) algorithm with 2 iterations/8 subsets and Gauss-filtered to a transaxial resolution of 5 mm at full-width at half-maximum (FWHM). Attenuation correction was performed using the low-dose non-enhanced CT data. PET and CT were performed using the same protocol for every patient on a Biograph 6 PET/CT scanner (Siemens, Erlangen, Germany).

Image analysis was performed using an appropriate workstation and software (Syngo TrueD, Siemens, Erlangen, Germany). Two board-certified specialists in nuclear medicine with 8 and 9 years of clinical experiences read all data sets independently and resolved any disagreements by consensus. Choline and PSMA PET/CT were analysed in a randomized fashion. Lesions that were visually considered as suggestive of PC were counted and analysed with respect to their localization (local relapses, lymph node, bone and soft tissue metastases) and to their maximum standardized uptake values (SUV_max) as it is common practice in daily routine. SUV_max was chosen due to its higher reproducibility between different investigators when compared to SUV_mean. The latter one is always dependent on the volume of interest (VOI) drawn by the investigator, whereas SUV_max is independent [26].

For calculation of the SUV, circular regions of interest were drawn around areas with focally increased uptake in transaxial slices and automatically adapted to a three-dimensional volume of interest at a 70 % isocontour. In the case of suspicious focal tracer uptake in one investigation and simultaneous virtually inconspicuous finding in the same area using the corresponding different PET/CT technique, SUV was measured in the inconspicuous areas as well.

SUV_max of the same lesions in both PSMA ligand- and fluoromethylcholine-based PET/CT were defined as clearly less, equal or clearly more with intensity differences of ≤10 %, between −10 and +10 % or >10 %, respectively. SUV values of the same lesions in both PSMA ligand- and fluoromethylcholine-based PET/CT as well as their ratio to background signal (in SUV_max) were statistically analysed using a Wilcoxon signed rank test.

When analysing the contrast of lesions with pathological tracer uptake and therefore visually highly suggestive of PC (characteristic for PC), several background tissues corresponding to the localization of the lesions were selected. This method is more accurate and better reflects the contrasting ability of the imaging modality in the region of interest when compared to the selection of one general background tissue that might show differing background uptake than the region of interest. However, particularly in cases of a variety of metastases the selection of multiple backgrounds is not always feasible. For our study we chose the background according to the following algorithm:

In cases of bone metastases in the vertebral column, of local relapses, of lung metastases and of liver metastases, we selected adjacent normal tissue as background. In cases of non-vertebral bone metastases, we selected the contralateral normal bone tissue. In cases of soft tissue metastases and of lymph node metastases, the gluteal musculature was selected as background.

The selection of two different backgrounds in cases of bone metastases (vertebral and non-vertebral) as mentioned above is due to the well-known fact that the background signal in choline-based PET imaging is higher in the vertebral column when compared to other skeletal structures (e.g. Fig. 4d). The statistical differences between the background signal of the vertebral column and other parts of the skeletal system in choline PET/CT are described in the “Results”.

For statistical analysis, Excel 2010 (Microsoft, Redmond, WA, USA) and SigmaPlot version 11 software (Systat Software, Inc., Chicago, IL, USA) were used. Significance of differences was evaluated by:

In all cases a p value of <0.05 was considered statistically significant. Furthermore, regression analysis between PSA and SUV_max was done for both investigations.