Patients
We performed retrospective analysis for one hundred and one patients with primary PCa confirmed by biopsy or radical prostatectomy between March 2019 and June 2020. All participants included in the data analysis were evaluated by 18F-PSMA-1007 PET/CT and had total PSA value measured within 4 weeks prior to the 18F-PSMA-1007 PET/CT imaging. Diagnosis of PCa proven through histological examination served as reference for the PET imaging analyses. Patients were excluded from analysis if they 1) had received local or systemic treatment, 2) lacked histological examination proven diagnosis or tPSA value, 3) had incomplete imaging data. This study was approved by local ethics committee (No. 2019LSYZD-J1-H) and was conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from each patient.
18F-PSMA-1007 and image acquisition
All 18F-PSMA-1007 PET/CT data was acquired on a PET/CT scanner (Gemini 64TF, Philips, Netherlands) at a single location (The First Affiliated Hospital of Xi’an Jiaotong University). Radiolabeling was performed using a fully automated radiopharmaceutical synthesis device based on a modular concept (MINItrace, GE Healthcare, USA). Over 99% radiochemical purification yield 18F-PSMA was obtained and examined by both radio-thin layer chromatography (TLC) and high performance liquid chromatography (HPLC) analysis.
Patients received intravenous injection of 18F-PSMA-1007 (3.7 MBq/kg body weight), and completed PET and CT scans 90 minutes after the injection. Low-dose CT scans from head to the proximal thighs (pitch 0.8 mm, 60 mA, 140 kV [peak], tube single turn rotation time 1.0 s and 5-mm slice thickness) for PET attenuation were acquired (pitch 0.8 mm, automatic mA, 140 kV [peak] and 512 × 512 matrix). Whole-body PET scans were performed in three-dimensional mode (emission time: 90 s per bed position, scanned at a total of 7-10 beds).
Imaging analysis
All 18F-PSMA-1007 PET/CT images were analyzed using Fusion Viewer software in the Extended Brilliance Workstation (EBW, Philips, Netherlands). Two experienced nuclear medicine specialists jointly interpreted all 18F-PSMA-1007 PET/CT scans, and performed comprehensive analysis of available clinical data. Consensuses were achieved through discussion when conclusions between the two specialists were discordant. The SUVmax of the primary PCa was calculated automatically with a manually adapted isocontour threshold centered on lesions with focally increased uptake corresponding to the tumor site verified by TRUS biopsy or radical prostatectomy [18]. The SUVmax values were also calculated for metastases, defined by an uptake higher than the local background and not associated with physiologic uptake per the guideline of the Society of Nuclear Medicine and Molecular Imaging and the European Association of Nuclear Medicine. The identified metastases were also consistent with PCa lesions pathologic tracer accumulation [22, 23, 5]. PET/CT scan findings were classified as (a) primary tumor, and (b) distance metastasis (abdominal lymph nodes, bone and internal organs) (Fig 1).
Statistical analysis
The Shapiro‐Wilk W test was used to test for normal distribution of all continuous variables. Based on data normality, independent two‐sample t test and the Mann‐Whitney U test were used for group comparison accordingly. Based on PET/CT findings, the participants were divided into non-metastasis and metastasis groups. The diagnostic performance of a prediction tPSA value was evaluated using the ROC curve and the area under the curve (AUC). Additionally, a logistic regression model was created with the metastasis results entered as a binary outcome variable. Four independent variables were included: a binary variable of tPSA value greater or lesser than the prediction tPSA value, a binary variable of tPSA value greater or lesser than the reference value of tPSA per the European Association of Nuclear Medicine guideline, and continuous variables of age and SUVmax. Based on the optimal prediction tPSA cutoff value, we divided all patients into two groups (i.e., high-tPSA and low-tPSA). We then divided each group into three subgroups based on their metastasis status (i.e., primary prostate tissue without metastasis, primary prostate tissue with metastasis, and primary PCa metastasis foci). We performed ANOVA analysis to compare the SUVmax among the three subgroups of high-tPSA and low-tPSA participants, respectively. Finally, we performed path analyses to examine the potential mediating role of tPSA level on SUVmax across different metastatic statuses. Specifically, the mediation model included the SUVmax of primary PCa lesions as the predictor, tPSA level as the mediator, and the SUVmax of metastasis foci as the outcome. A significance level of α = 0.05 (two-tailed) was applied. Statistical analyses were performed using IBM SPSS Statistics version 22 (IBM Corp., Armonk, NY, USA), GraphPad Prism software, version 8.4 (GraphPad Software, Inc., La Jolla, CA, USA) and MedCalc version 19.0 (MedCalc Software Ltd, Belgium).