Risk of metastatic disease using [18F]PSMA-1007 PET/CT for primary prostate cancer staging

Background Accurate prostate cancer imaging is critical for patient management. Multiple studies have demonstrated superior diagnostic accuracy of [68Ga]-PSMA-11 PET/CT over conventional imaging for disease detection, with validated clinical and biochemical predictors of disease detection. More recently [18F]PSMA-1007 offers theoretical imaging advantages, but there is limited evidence of clinical and biochemical predictors of scan findings in the staging population. This study investigates the association of clinical variables with imaging characteristics among patients who underwent [18F]PSMA-1007 PET/CT for primary staging of men with histopathologically confirmed prostate carcinoma. A retrospective review of 194 consecutive patients imaged between May 2019 to May 2020 was performed. Association between imaging variables (presence and distribution of metastatic disease, primary tumour SUVmax) and clinical variables (EAU risk criteria) were assessed using descriptive statistics, logistic regression model and ROC analysis. Results The median age, PSA level and ISUP grade were 70 years, 10 ng/mL and ISUP grade 3, respectively. There were 36.6% of patients with intermediate-risk and 60.8% of patients with high-risk disease. ISUP grade was associated with the presence of metastasis overall (p = 0.008) as well as regional nodal (p = 0.003), non-regional nodal (p = 0.041) and bone (p = 0.006) metastases. PSA level was associated with metastatic disease overall (p = 0.001), regional (p = 0.001) and non-regional nodal metastases (p = 0.004), but not with bone metastases (p = 0.087). There were too few visceral metastases for meaningful analysis. SUVmax of the primary prostatic tumour was associated with ISUP grade (p = 0.004), PSA level (p < 0.001) and AJCC stage (p = 0.034). PSA > 20 ng/mL and ISUP grade > 3 had a specificity of 85% (95% CI 78–91%) and 60% (95% CI 50–68%) and a sensitivity of 36% (95% CI 25–49%) and 62% (95% CI 49–74%), respectively, for detection of metastatic disease. Conclusion Metastatic disease according to [18F]PSMA-1007 PET/CT was associated with ISUP grade and PSA level. This is the largest study using [18F]PSMA-1007 PET/CT to confirm a positive correlation of PSA level, ISUP grade and stage with primary prostate tumour SUVmax. Supplementary Information The online version contains supplementary material available at 10.1186/s13550-021-00869-5.

staging techniques such as computed tomography (CT) and bone scintigraphy imaging have relatively low sensitivity for early and small volume disease detection [3].
PSMA can be coupled with either Gallium-68 ( 68 Ga) and Fluorine-18 ( 18 F) for diagnostic imaging. [ 68 Ga]Ga-PSMA-11 is currently the most widely used radiotracer for prostate cancer imaging [11]. However, 68 Ga-labelled compounds have both cost and logistical limitations for scanning a large volume of patients, leading to the development of 18 F-labelled PSMA radiotracers [12]. Advantages of 18 F-labelled radiotracers include: (1) longer half-life (110 min vs 68 min for 68 Ga) and cyclotron production, facilitating centralised production and distribution leading to reduced costs, and (2) lower positron energy of 18 F compared to 68 Ga resulting in better intrinsic spatial resolution. Minimal urinary radiotracer excretion is an additional advantage specific to [ 18 [13,14].
In clinical practice, risk stratification tools such as D' Amico Risk Classification for Prostate Cancer and Memorial Sloan Kettering Cancer Center (MSKCC) Pre-Radical Prostatectomy nomogram use Prostate-specific antigen (PSA) level, Gleason score and clinical stage to stratify patients as low, intermediate or high-risk for prognostication and management [15]. European Association of Urology (EAU) has adopted the risk stratification used by D' Amico [16,17]. Several studies have demonstrated the correlation between prognostic risk factors such as PSA level and Gleason score with metastatic disease extent [18][19][20][21][22][23] and SUVmax [24,25] of the primary tumour using Ga68-PSMA PET/CT. However, this correlation has not been confirmed with findings on [ 18

Study design and technique
A retrospective audit of 194 consecutive patients who underwent [ 18 F]PSMA-1007 PET/CT scan for initial staging of biopsy proven prostate carcinoma at Royal Brisbane and Women's Hospital (RBWH), a tertiary referral centre, over 12 months between May 2019 to May 2020 was performed. Analysis of the first [ 18 F]PSMA-1007 PET/CT was performed if multiple scans were undertaken during the study period. Referral criteria for scanning included PSA > 20 ng/mL or ISUP grade ≥ 3 or clinical stage ≥ T2c. Additionally, at least one PSA value within 12 months preceding the [ 18 F]PSMA-1007 PET/ CT scan was required for inclusion in the study (Fig. 1). However, 31 patients did not meet the referral criteria (i.e. ISUP ≤ 2 or PSA < 20 or clinical stage < T2c) but were scanned for other reasons (e.g. high-risk radiological features, perineural invasion on biopsy etc.).
Demographic, clinical, biochemical and radiological variables were collected using the electronic medical record systems.
A low-risk ethics exemption was obtained for this study from the Royal Brisbane and Women's Hospital Human Research Ethics Committee (HREC) with a waiver of written consent for this retrospective study.

[ 18 F]PSMA-1007 Radiosynthesis and Quality Control
On-site synthesis of [ 18 F]PSMA-1007 was performed using a kit-based approach on GE FASTlab or MX Tracerlab platforms, followed by comprehensive quality control testing to pharmacopoeia standards for other F-18 radiopharmaceuticals (i.e. radionuclide, radiochemical and chemical purity, endotoxin and sterility, pH). The [ 18 F]PSMA-1007 radiochemical purity was > 95% as measured using high-performance liquid chromatography (HPLC).

Imaging protocol
All [ 18 F]PSMA-1007 PET/CT scans were performed at the RBWH Nuclear Medicine and Specialised PET Service department on one of three Siemens Healthineers PET/CT (Biograph Vision, Biograph mCT or Biograph mCT Flow) scanners. [ 18 F]PSMA-1007 was administered intravenously (mean activity 250 MBq, range 133-284 MBq), with patients being scanned after a median imaging delay of 127 min (IQR 120-140 min). A low-dose non-contrast CT was also performed from the vertex to mid-thighs for attenuation correction and anatomical localisation. PET acquisition was performed in threedimensional mode (acquisition time of 4-min per bed position over the pelvis and 2.5 min bed position over rest of the imaged body) with emission data corrected for attenuation, scatter, randoms and dead time. The corrected emission data were iteratively reconstructed using ordered-subsets expectation maximisation (3 iterations, 21 subsets) with time of flight and point-spread function resolution recovery, followed by a post-reconstruction Gaussian filter.

Imaging analysis
All the [ 18 F]PSMA-1007 PET/CT scans were dualreported by two experienced nuclear medicine specialists. The images were reviewed for reporting in Syngio. via (VB50, Siemens Software) which allows CT, PET and fused images to be viewed in multiple planes. Focal [ 18 F]PSMA-1007 avidity above background with typical appearance of prostate cancer (with or without definitive anatomic correlate on CT) and distribution not correlating to physiologic uptake (such as cervical, celiac or sacral ganglia) or non-specific bone uptake (obvious indolent bone lesions, degenerative change or fractures) were defined as metastatic disease. Any disagreement was resolved by consensus with a third nuclear medicine specialist. Equivocal lesions (predominantly bone uptake) were not considered malignant.

Standard of reference
We use [ 18 F]PSMA-1007 PET/CT as a standard of reference instead of histopathological validation. This is based on strong evidence with multiple studies using both [ 18 F] PSMA-1007 PET/CT and [ 68 Ga]Ga-PSMA-11 PET/CT confirming an excellent diagnostic accuracy and high positive predictive value for the detection of metastatic disease with histopathological validation [4,13,14,[26][27][28][29]. Additionally, a recent prospective intra-individual comparative study has demonstrated excellent concordance between [ 18 F]PSMA-1007 and [ 68 Ga]Ga-PSMA-11 PET/CT [30,31]. Therefore, we consider [ 18 F]PSMA-1007 PET/CT as an acceptable substitute for histopathological validation based on this extensive literature.

Statistical analysis
Patient characteristics were summarised using count and percentage for categorical measures and mean and standard deviation (SD) or median and interquartile range (IQR) for continuous measurements. Logistic regression was performed to examine the association between preselected patient clinical characteristics including patient age, PSA level, ISUP grade and ADT exposure and scan outcomes including presence or absence of metastatic disease. For multivariable modelling, the variables with p < 0.20 in univariable analyses were included. The ISUP grades were grouped during analysis according to the EAU risk groups that use PSA level, ISUP grade/Gleason score (GS) and clinical stage to stratify patients. Lowrisk is defined as PSA < 10 ng/mL and ISUP grade 1/ GS < 7 and cT1-2a, intermediate-risk as PSA 10-20 ng/ mL or ISUP grades 2 and 3/GS 7 or cT2b and high-risk as PSA > 20 ng/mL or ISUP grades 4 and 5/GS > 7 or cT2c or higher. As the clinical stage data were unavailable, patients were grouped into stage IIIb or below, IIIc, IVa and IVb according to American Joint Committee on Cancer (AJCC) stage (8th edition, 2017).
SUVmax of the primary prostatic tumour was compared among groups based on PSA level, ISUP grade or AJCC stage using Kruskal-Wallis or Mann-Whitney U test, and the proportion of metastases were compared using Chi-squared test or Fisher's exact test when more than 20% of the expected counts were less than 5.
The sensitivity and specificity of EAU high-risk criteria was examined using [ 18 F]PSMA-1007 PET/CT scan outcome to detect metastatic disease. Median PSA doubling time (Dt) was obtained, including patients who had decreasing PSA (negative PSA doubling time) by using the slope on logged PSA (i.e. log(2)/Dt) [32]. The level of statistical significance was set at 0.05. STATA 15 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC) and R (R version 4.1.0) were used for statistical analysis and graphs.

Patient characteristics
A total of 194 patients were included in the final analysis ( Table 1). The median age at the time of PET/CT was 70 years (IQR 63-75). The median PSA level prior to the PET/CT was 10 ng/mL (IQR 6.5-18). Using the EAU risk stratification system, 71 (36.6%) patients were classified as intermediate-risk (25.8% of patients with PSA 10-20 ng/mL or 48.2% of patients with ISUP grades 2 and 3), and 118 (60.8%) patients were classified as high-risk (22.2% of patients with PSA > 20 ng/mL or 47.6% of patients with ISUP grades 4 and 5). Further, 7.2% of patients had exposure to ADT prior to the PET/ CT with a median of 2 months (IQR 2-5) duration of ADT exposure. Table 2 demonstrates the distribution of PSA levels within each ISUP grade in our imaged population.    (Table 3). Metastases were detected in 28.7% (29/101) with PSA ≤ 10 ng/mL, 26% (13/50) with PSA 10-20 ng/mL and 55.8% (24/43) with PSA > 20 ng/mL. Patients with PSA levels > 20 ng/mL were more likely to have metastasis, compared to those with PSA levels 10-20 ng/mL (OR 3.60; 95%CI 1.50-8.60) and PSA levels ≤ 10 ng/mL (OR 3.14; 95% CI 1.50-6.58; Table 3). Multivariable modelling including ISUP grade, PSA level and ADT exposure demonstrated association of ISUP grade and PSA level with the presence of metastatic disease (p = 0.008 and p = 0.001, respectively) such that patients with higher ISUP grade and PSA level were more likely to have metastatic disease (Table 3). Although a higher proportion of metastatic disease was detected in patients with previous or current use of ADT prior to the PET/CT scan (50% vs. 33%; aOR 2.16, 95%CI 0.67-6.96), this was not statistically significant. Figures 2 and 3 show examples of [ 18 F] PSMA-1007 PET/CT with localised prostate disease and extensive metastases, respectively.

ISUP grade, PSA level and AJCC stage with SUVmax of primary prostate tumour
There were seven negative [ 18 F]PSMA-1007 PET/CT scans in the staging group that were excluded from the primary tumour SUVmax analysis (Additional file 1: supplementary table). The median SUVmax of the primary prostate lesion was 17.2 (IQR 9.5-26.7). The SUVmax of the primary prostate tumour was higher in the ISUP grades 4 and 5 compared to ISUP grades 2 and 3 (median SUVmax 19.6 vs 14.9). Similarly, the primary tumour SUVmax was higher in the PSA > 20 ng/mL group than PSA 10-20 ng/mL group (median SUVmax 21.5 vs. 17.4). The primary tumour SUVmax also increased with increasing AJCC stage with stage IIIB and below (median SUVmax 14.2), stage IIIC (median SUVmax 17.5), stage IVA (median SUVmax 19.8) and stage IVB (median SUVmax 20). A statistically significant association between ISUP grade (p = 0.004), PSA level (p < 0.001) and AJCC stage (p = 0.034) with SUVmax of the primary prostatic tumour (Tables 4, 5, 6) was observed. Additionally, a higher primary tumour SUVmax was detected in the high-risk group compared to the intermediate-risk group (median SUVmax 20.2 vs 11.9; p < 0.001). SUVmax of the primary prostate tumour with increasing ISUP grade and PSA levels is illustrated in Fig. 5 using box plots.

Metastatic disease by EAU risk group
Overall, the high-risk group (PSA > 20 ng/mL or ISUP 4 and 5) had significantly higher proportion of patients with nodal (regional and non-regional) (34.7% vs 1.4%; p < 0.001) and bone (28% vs 8.5%; p = 0.001) metastases compared to the intermediate-risk group (PSA 10-20 ng/ mL or ISUP 2 and 3). Although the high-risk group had a higher proportion of patients with visceral metastases compared to the intermediate-risk group (4.2% vs 2.8%), there were too few patients for meaningful analysis (p = 0.71) ( Table 7).

PSMA scan referral criteria
Out of the 194 patients included in the study, 31 patients did not meet our referral criteria but underwent [ 18 F] PSMA-1007 PET/CT scan due to one of five reasons:

Discussion
To our knowledge, this is the first study demonstrating a positive correlation between ISUP grade and PSA level with an increased likelihood of metastatic disease using [ 18 F]PSMA-1007 PET/CT for primary staging of prostate cancer. Additionally, the distribution of metastasis was also analysed in our study. There was a strong positive correlation between PSA level and the likelihood of regional and non-regional nodal metastases and ISUP grade and regional, non-regional nodal and bone metastases. There was a trend toward increasing PSA level and bone/visceral metastases and ISUP grade and visceral metastases, without reaching statistical significance, potentially due to a small number of these cases. Furthermore, we demonstrate a higher proportion of patient with nodal (regional and non-regional) and bone metastases in the high-risk group compared to the intermediate-risk group. These results are concordant with the findings previously demonstrated using 68 Ga-PSMA PET/CT [19,21,22], and whilst expected, these findings are an important step to validate the role of [ 18 F]PSMA-1007 PET/CT in primary prostate cancer staging. The presence of an unexpectedly high proportion of metastases in the PSA ≤ 10 ng/mL (15% regional nodal and 18% bone metastases) and ISUP grade 1 (13% regional nodal metastases) likely represents selection bias due to our PSMA PET/CT scan referral criteria requiring the presence of high-risk features for imaging (Tables 2,  3 , 4, 5). For example, the one patient with regional nodal metastasis in ISUP grade 1 had a PSA level of 22 ng/ mL placing them in the high-risk group. Similarly, the PSA ≤ 10 ng/mL group had a high proportion of intermediate-risk ISUP grades 2 and 3 (49%) and high-risk ISUP grades 4 and 5 patients (48%) ( Table 2).
Ours is also the largest study to date using [ 18 F]PSMA-1007 PET/CT confirming a positive correlation between higher ISUP grade and PSA level with increasing SUVmax of the primary prostate tumour, consistent with increasing PSMA receptor expression (Fig. 5) (Table 8). However, ISUP grade > 3 only has a sensitivity of 62% (specificity of 60%), whereas our referral criteria of ISUP ≥ 3 has a much greater sensitivity of 94% (specificity 36%) for detecting metastatic disease using [ 18 F]PSMA-1007 PET/CT (Table 8). Therefore, using PSA > 20 ng/mL or ISUP ≥ 3 as selection criteria would be adequate for identifying most patients with metastatic disease using [ 18 F] PSMA-1007 PET/CT for prostate cancer staging. The current EAU guidelines use similar parameters to recommend imaging, albeit conventional imaging, for staging prostate cancer [17].
There are a few limitations to our study. Firstly, given the study's retrospective nature, data collection was incomplete for some patients (missing information on PSA levels, ISUP grade and treatment information) potentially introducing bias and loss of statistical power, possibly avoidable in future prospective studies. Secondly, similar to other published [ 18 F]PSMA-1007 PET/ CT studies, typical focal prostatic and extraprostatic avidity was considered disease without histopathological confirmation. It is not feasible to pathologically confirm every disease site due to patient factors, location or small lesion size. However, sufficient evidence now exists to use [ 18 F]PSMA-1007 PET/CT as a standard of reference and a substitute for histopathological validation, as explained in the methods section [4,13,14,[26][27][28][29][30]. Thirdly, a common issue with the [ 18 F]PSMA-1007 radiotracer is focal uptake in non-specific bone lesions (NSBLs) without an underlying CT correlate definite for prostate metastasis. Recent data suggest that these NSBLs rarely represent metastatic lesions, supporting our image interpretation that equivocal lesions are not considered malignant [36]. Lastly, a small number of patients were on ADT before the PET/CT scan during initial staging, presumably due to high-risk features, potentially influencing the PSMA targeted PET/CT results [37].

Conclusion
Our study confirms PSA level and ISUP grade as significant prognostic risk factors for the presence and distribution of metastatic disease using [ 18 F]PSMA-1007 PET/ CT for primary staging of prostate carcinoma. Furthermore, this is the largest study confirming a positive correlation of PSA level, ISUP grade and AJCC stage with primary prostate tumour SUVmax using [ 18 F]PSMA-1007 PET/CT. This study also demonstrates adequate performance of PSA > 20 ng/mL or ISUP grade ≥ 3 parameters for the detection of metastatic disease using [ 18 F]PSMA-1007 PET/CT. Validation of these findings further supports the use of [ 18 F]PSMA-1007 PET/CT for the primary staging of prostate carcinoma.