The files of patients who underwent Ga-68 PSMA PET/CT for staging of PCa in the Nuclear Medicine Department of our hospital between September 2017 and December 2020 were browsed retrospectively. Out of a total of 862 patient files, 42 patients with no known secondary malignancy, diagnoses confirmed by biopsy, and high-volume denovo metastatic castration-susceptible PCa followed by at least 6 cycles of DTX (75 mg/m2 + prednisone 5 mg) treatment  were included in the study (Figure 1). These patients had also received routine androgen deprivation therapy (ADT). Criteria defined in the ChemoHormonal Therapy Versus Androgen Ablation Randomized Trial for Extensive Disease in Prostate Cancer (CHAARTED) study was used to distinguish between HV and LV while including the patients into the study. Accordingly, the presence of 4 or more bone metastases, at least one of which is outside the vertebral column and pelvis, and/or visceral organ metastases were considered sufficient for the patients to be evaluated as having a high-volume disease.
In this retrospective study, the results of routine diagnostic procedures such as PSA tests, pathology results (ISUP scores), and demographic data of eligible patients who were screened in our clinic were obtained from the hospital’s information management system. In addition, the information regarding whether the patients were alive or not and the date of death in case they were dead, was obtained from the Death Notification System (DNS).
As determined by the criteria of the Prostate Cancer Study Group 2 (PCWG-2) for the definition of progression, an increase in three consecutive PSA measurements and a PSA value above 2.0 ng/ml were adopted.  Since there was no progression detected by anatomical imaging without PSA progression in our study group, additional radiological progression criteria were not needed. This study was a graduation/specialization thesis and was carried out with the approval of the Ethics Committee of Prof. Dr. Cemil Taşçıoğlu City Hospital, (protocol number 391).
Ga-68 PSMA PET CT preparation, imaging, and evaluation
Gallium68 radionuclide was obtained by milking from a Germanium/Gallium (Ge68/Ga68) generator (iThemba LABS©, South Africa) containing 30 mCi (milicurie) of Germanium68. In a fully automated synthesis device (Scintomics©, Germany), 25 µg of PSMA I&T compound was labeled with Ga68 according to the procedure. After the labeling process, the radiochemical purity was evaluated in the liquid chromatography test and the radiopharmaceutical with a result of 95% and above was considered suitable for application to the patient.
Patients are instructed to drink contrast medium (Omnipaque© 350 mg/50ml) diluted with 1.5 liters of water 6 hours before the examination. No additional preparation was required. Ga-68 labeled PSMA compound was administered intravenously at a dose of 1.8-2.2 MBq/kg (0.049-0.060 mCi/kg).
Patients voided just prior to imaging and they were scanned at the 60th minute with low-dose CT scan for attenuation correction and anatomical correlation, with arms up in the supine position and covering the vertex-upper thigh proximal (including lower extremity if necessary). Immediately afterwards, PET imaging was performed for 3 minutes per bed. All imaging was performed on a Siemens Biograph 6 LSO HI-RES integrated PET/CT device (Siemens Medical Solutions, Chicago, IL, USA). All PET images were acquired in 3D mode and reconstructed with the attenuation-weighted OSEM algorithm (four iterations, eight subgroups). Gaussian smoothing filter (5 mm full width at half-width) was applied to the post-reconstruction images.
Gallium-68 PSMA PET CT images were analyzed as MIP (maximum intensity projection) and cross-sectional images. All areas which visually exceed the normal background activity uptake in these images were evaluated as metastasis when they are outside of the prostate and as primary tumors when in the prostate bed if they do not correspond to the known physiological involvement areas and if the described area does not correspond to any known non-malignant activity involvement (such as inflammation, urinary activity, etc.).
SUVmax, SUVmean, SUVpeak, and PSMA tumor volume (PSMA-TV) (cm3) for each lesion were automatically generated from user-specified VOIs by the Workstation. The borders of the target lesion within the VOI were automatically drawn with isocontour curves and voxels greater than the 45% threshold of SUVmax in the VOI were defined as the lesion area to measure PSMA-TV and SUVmean[9-10]. In cases where the lesion area observed on CT and the VOI drawn were inconsistent, the threshold value was readjusted to cover the entire lesion and exclude the extralesion tissue. In cases where the borders of PSMA-positive lesions overlapped with the areas of physiological uptake of neighboring organs, the borders of the areas drawn on the VOI were determined manually by using CT images.
The whole body PSMA-TV(WB) value of each patient was defined as the sum of the PSMA-TV of all lesions. Whole-body Total Lesion PSMA, namely TL-PSMA(WB), was obtained by multiplying the SUVmean determined in the volume selected with the isocontour and PSMA-TV(WB). For the primary lesion, the TL-PSMA (primary) value was obtained by multiplying the SUVmean (primary) obtained in the determined VOI area with the PSMA-TV (primary) value. For each patient, SUVmax(primary), SUVmean(primary), SUVpeak(primary), PSMA-TV(primary), TL-PSMA(primary) and SUVmax(WB), SUVmean(WB), SUVpeak(WB), PSMA-TV (WB), and TL-PSMA(WB) data were calculated on ADW 4.7 (GE Healthcare, Chicago, USA) consoles. In addition, bone metastases, visceral metastases, and intra-pelvic-extra-pelvic lymphatic metastases of the patients were also noted while the images were being evaluated.
SPSS 24.0 for Windows (IBM, New York, USA) package program was used to evaluate the data obtained in the study and the statistical significance limit was determined as p<0.05. While descriptive statistics were presented as mean ± standard deviation for normally distributed numerical variables, categorical variables, ordinal variables, and non-normally distributed numerical variables were presented as median (minimum, maximum).
ROC curves were plotted to determine the appropriate threshold value for continuous variables. This procedure was done separately for overall survival (OS) and progression-free survival (PFS) data for each variable. Threshold values were determined using the Youden index in the drawn ROC curves. After the threshold values of the continuous variables were determined, the survival graphs for the OS and PFS data separately for the variables and the appropriate area under the curve (AUC) in the ROC curves were drawn using the Kaplan-Meier method. At this stage, univariant analyzes were performed using the log-rank method and statistical significance was evaluated. Spearman's correlation analysis was used to evaluate the probability of multicollinearity for the variables with univariate analysis p<0.25. The Cox hazard-ratio model was then applied as a multivariate analysis to evaluate the potential independent effects of these variables on OS and PFS.