Staging 68 Ga-PSMA PET/CT in 963 consecutive patients with newly diagnosed prostate cancer: incidence and characterization of skeletal involvement

The aim of the study was to elaborate the incidence and type of skeletal involvement in a large cohort of patients with newly diagnosed prostate cancer (PCa) referred for Ga-68 PSMA-11 PET/CT staging in a single center. Study cohort included 963 consecutive patients with newly diagnosed PCa referred for Ga-68 PSMA-11 PET/CT study for staging. The incidence of bone involvement, type of bone metastases, and extent of disease were determined and correlated with the ISUP Grade Group (GG) criteria and PSA levels. Bone metastases were found in 188 (19.5%) of 963 patients. Bone metastases were found in 10.7% of patients with PSA < 10 ng/dL and in 27.4% of patients with PSA > 10 ng/dL and in 6.1% of patients with GG ≤ 2/3 and in 8.9% of patients with GG 4/5. In 7.6% of the patients, skeletal involvement was extensive, while 11.9% of patients had oligometastatic disease. Osteoblastic type metastases were the most common type of bone metastases presented in 133 of the patients with malignant bone involvement (70.7%). More than half of them had only osteoblastic lesions (72 patients (38.3%)), while the other (61 patients (32.5%)) had also intramedullary and/or osteolytic type lesions. Intramedullary metastases were found in 97 patients (51.6%), while 41 (21.8%) of them were only intramedullary lesions. Osteolytic metastases were detected in 36 patients (19.2%), of which 8 were only osteolytic lesions. Although traditionally bone metastases of PCa are considered osteoblastic, osteolytic and intramedullary metastases are common, as identified on PET with labeled PSMA. Skeletal spread may be present also in patients with GG ≤ 2/3 and PSA < 10 ng/dL.


Introduction
Advanced prostate cancer (PCa) metastasizes primarily to bone [1]. Autopsy series revealed that more than 90% of patients diagnosed with metastatic PCa show evidence of bone metastases [2]. Treatment algorithms for patients with newly diagnosed PCa reflect the need to identify metastatic involvement of the skeleton prior to definitive therapy, particularly in patients at high risk for advanced disease. Identification of early skeletal spread and accurate determination of disease extent has been an ongoing challenge of imaging modalities [1,3,4].
For decades, bone scintigraphy (BS) with Tc-99 m methylene diphosphonate (Tc-99 m MDP) has been the most commonly used modality for detection of bone metastases [5]. Bone metastases of PCa are considered traditionally as "osteoblastic" (OB) by conventional radiographs and CT. Uptake of Tc-99 m MDP is closely associated with osteoblastic activity and increased blood flow and therefore is highly sensitive for detection of OB type metastases. Clinical and imaging guidelines recommend BS for initial staging of patients with intermediate-risk and high-risk disease [6][7][8][9]. However, data accumulated over the years explored the limitations of BS in detecting intramedullary (IM) and osteolytic (OL) type metastases also present in PCa patients. The incidence of the latter types of metastases remains elusive [1,5].
The use of positron emitting radiotracers accumulating directly in tumor tissue gave rise to a completely new concept of assessment of malignant skeletal involvement. F-18 fluorodeoxyglucose (F-18 FDG), the most commonly used tracer in oncologic PET/CT, is unfortunately not always suitable for imaging of PCa [1]. The role of C-11 and F-18 choline PET/CT for lymph nodes or bone involvement in PCa was investigated. Labeled choline was found highly specific as well as allowing for early identification of IM and OL type metastases [10,11].
Labeled prostate-specific membrane antigen (PSMA) is a relatively novel PET recently introduced in routine staging of prostate cancer patients with unequivocal superiority over routine BS in detecting bone metastases [12][13][14][15][16][17][18]. In the meta-analysis of 24 articles by Zhou J et al. on the role PSMA PET/CT, choline PET/CT, F-18 fluoride PET/CT, MRI, and BS in the diagnosis of bone metastasis, PSMA PET/CT was found with highest per patient sensitivity and specificity [17).
The purpose of the current study was to elaborate the incidence and type of skeletal metastases in a large cohort of patients with newly diagnosed prostate cancer referred for PSMA PET/CT for staging prior to definitive therapy. Labeled PSMA PET/CT is nationally reimbursed for staging of patients with newly diagnosed prostate cancer, which allowed us to perform the study on consecutive patients.

Patients
Staging of intermediate-and high-grade newly diagnosed prostate cancer using PET/CT with labeled PSMA is nationally reimbursed since 2016; thus, Ga-68 PSMA PET/CT has been introduced in the staging algorithm of the latter patient population. Having the institutional review board approval, we queried our database to retrieve 963 consecutive patients with newly diagnosed PCa, who had a staging Ga-68 PSMA-11 at our department between November 2015 and September 2020 prior to treatment. Characteristics of the patients are summarized in Table 1. Gleason score was available in 839 of 963 patients. Patients were categorized based on the International Society of Urological Pathology (ISUP) Grade Group (GG) criteria [19].

PET/CT imaging
An amount activity 148-166 MBq Ga-68 PSMA-11 was injected intravenously 50-100 min before acquisition. The patients were instructed to void immediately prior to acquisition. PET/CT studies were performed from the tip of the skull to mid-thigh using Discovery 690 or MI PET/CT systems (GE Healthcare).
CT acquisition of the PET/CT was performed using automatic mA-modulation and 120 kV. CT scans were reconstructed to a slice thickness of 2.5 mm. PET acquisition was performed with acquisition time of 3 min per bed position in 3D mode. PET images were reconstructed in a matrix size of 128 × 128 with a pixel size of 5.5 mm and slice thickness of 3.3 mm for the 690 system and a matrix size of 256 × 256 with a pixel size of 2.7 mm and slice thickness of 2.8 mm for the Discovery MI system. Reconstruction method was VUE Point FX by GE Healthcare that uses time of flight information and includes a fully 3D OSEM algorithm with 3 iterations. For the 690 systems, reconstruction method used 24 subsets and filter cutoff of 8 mm, and for the Discovery MI, we used 8 subsets and filter cutoff of 6.0 mm. VUE Point FX algorithm also includes normalization and image corrections for attenuation, scatter, randoms, and dead time. A heavy Z-filter was applied to smooth between transaxial slices.

Image analysis
All scans were reviewed by nuclear medicine physicians (MK, JK, EES), body radiologist (CL), and musculoskeletal radiologist (ID). Any skeletal lesion showing above normal uptake not associated with physiological uptake was recorded [20]. Equivocal bone lesions were evaluated according to the PSMA-RADS, version 1.0 system [21], which has been previously validated [22]. Lesions were classified based on their morphologic appearance on CT as osteoblastic type (OB), focal increased uptake corresponding to a lesion showing higher HU compared to surrounding bone tissue; osteolytic type (OL), focal increased uptake corresponding to a lesion showing bone destruction; and intramedullary type (IM), focal increased uptake with intact bone on CT [1,18]. The three types of lesions are illustrated in Fig. 1.
Based on the number of metastatic bone lesions, patients were categorized as having a single metastasis and oligometastatic disease (2-5 metastatic skeletal lesions) and patients with extensive metastatic spread (> 5 metastases) [23]. The incidence of bone metastasis and extent of disease and metastasis type were correlated with ISUP Grade Group (GG) criteria and with PSA levels.
In addition to skeletal lesions, increased uptake in nonosseous metastatic sites was also recorded.

Statistical analysis
Categorical variables were described as frequency and percentage. Continuous variables were evaluated for normal distribution using histogram. Normally distributed variables were reported as mean and standard deviation (SD) and skewed variables as median and interquartile range (IQR). Independent sample t-test and Mann-Whitney test were applied to compare continuous variables between metastatic categories. The area under the receiver operating characteristic (ROC) curve was used to evaluate the discrimination ability of Gleason score, ISUP Grade Group, and PSA. All statistical tests were two sided, and p < 0.05 was considered statistically significant. Statistical analysis was performed using SPSS statistical software (IBM SPSS Statistics for Windows, version 27, IBM Corp., Armonk, NY, USA, 2020).

Results
Bone metastases were detected in 186 patients. In 45 additional patients, bone lesions detected on PET were interpreted as benign or equivocal: 16 were single sites of tracer accumulation in the ribs with no corresponding morphological changes on CT with no evidence of metastatic skeletal spread on follow-up [24]. Twenty-seven scintigraphic lesions corresponded to benign findings on CT including fibrous dysplasia (n = 16), hemangioma (n = 5), fracture (n = 3), enchondroma (n = 2), and Paget's disease (n = 1). Two equivocal lesions found in two patients were metastases based on the findings on follow-up PET. The latter patients were included in the results analysis to the group of patients with metastatic skeletal spread.
In 839 of the study cohort, ISUP data was available for correlation. Bone metastases were found in 11.1% (14/126) patients with GG 1 representing 1.5% of the patients with low-risk disease; in 10.5% (44/418) patients with GG 2/3 representing 4.6% of the patients with intermediate-risk disease; and in 21.6% (36/167) patients with GG 4 and 39.1% (50/128) patients with GG 5 representing 8.9% of the patients with high-risk disease. A total of 10.7% (58/544) of patients with Gleason score ≤ 7 were found to have metastatic skeletal spread based on PET/CT findings (Fig. 2).
Mean PSA ± SD in patients with no evidence of metastatic skeletal involvement was significantly lower than in patients with bone metastases: 16 (Fig. 3). Incidence of the various types of metastases is presented in Table 2. OB type metastases were the most common type of bone metastases identified in 133 of the patients with malignant bone involvement (70.7%). Of these, 72 patients (38.3%) were found to have only osteoblastic lesions, while the other 61 patients (32.5%) had also IM and/or OL type lesions. IM metastases were found in 97 patients (51.6%). In 41 patients (21.8%), IM metastases were the only type of lesion detected. OL metastases were detected in 36 patients (19.2%). In 8 (4.3%) patients, OL lesions were the only type of bone metastasis identified. Histogram 1, 2, and 3 illustrate the type of bone metastases in correlation with GG score, PSA level, and the extent of the metastatic skeletal spread. Table 3 summarizes the non-osseous sites of metastases found in 188 patients having skeletal involvement.
Follow-up imaging data was available for assessment in 109 of 188 patients (58%) with bone metastases-102 patients (54.3%) had a follow-up PSMA PET/CT studies, and additional 7 patients (3.7%) had CT. Follow-up period was 3-78 months. Of these 109 patients, 97 (89%) received anti-androgen hormonal therapy based on the baseline PET-CT findings, 5 (4.6%) received hormonal therapy in combination with chemotherapy, and four patients (3.7%) with lytic metastases received radiotherapy. Seven patients (6.4%) were not treated.
On follow-up imaging, bone involvement improved in 56 patients (51.4%) with bone lesions showing sclerosis on CT and no increased PSMA uptake or reduction in intensity of uptake. Nine patients (8.3%) had unchanged bone findings. In 29 patients (26.6%), skeletal involvement progressed having new bone lesions on repeat PET/ CT study. Mixed response to treatment with improvement of some lesions and progression of other lesions and/or new skeletal lesions was identified in 15 patients (13.7%).

Discussion
Malignant skeletal involvement of PCa develops as a multistep process initiating as intramedullary lesions in the distribution of the red active marrow [25]. In the marrow, tumor cells follow a lag phase of undetermined duration comprising a dormant phase followed by a more aggressive active phase with tumor-associated osteolysis and tumor-induced bone formation that dictates the final appearance of the lesion as OB or OL. Thus, metastatic processes are present in the bone marrow long before they give rise to derangement of the normal bone tissue architecture [26]. The incidence of bone metastases at staging of patients with newly diagnosed high-risk PCa varies between 6.6 and 36.1% in different publications depending on the size of cohort, and either SPECT or PET technology was used [4,[27][28][29][30][31][32]. Clinical guidelines [6][7][8][9] recommend performance of BS at staging when PSA is ≥ 20, clinical stage is T2 and PSA is ≥ 10 ng/dL, clinical stage is T3 or T4, Gleason score is ≥ 8, or any symptoms are suggestive of bone metastases. In the current study, we found metastatic involvement in 19.5% of patients with newly diagnosed prostate cancer. As expected, patients with skeletal metastatic involvement were those with higher risk based on ISUP score and PSA levels. It should be noted however that metastatic skeletal spread was found also in 1.5% of patients with low risk (GG 1) and 4.6% of patients with intermediate risk (GG 2/3) (a total of 10.7% of patients with Gleason score ≤ 7) and in 10.7% of patients with PSA ≤ 10 ng/dL.
Metastatic skeletal spread was found in almost 9% of high-risk patients with GG 4 or 5 and in 27% of patients with PSA > 10 ng/dL (including as high as 39% of patients with PSA level of 20 ng/dL and higher).
In the clinical setting, the extent of skeletal involvement influences the choice of the treatment. Patients with 5 or fewer metastatic bone lesions were reported to have a longer overall survival than patients with more than 5 metastases (73% vs 45% at 5 years and 36% vs 18% at 10 years), as well as longer disease-free survival. A specific cohort of patients with "oligometastases" is now recognized at staging evaluation [22,33]. Treatment options for oligometastatic and extensive disease vary greatly [34]. In the current study, while a total of 19.5% of the study cohort were found to have malignant skeletal involvement, only 7.6% had extensive disease, and the others had a single lesion or oligometastatic disease.
Bone metastases of PCa are considered OB based on their appearance on conventional radiographs. However, studies have shown a high heterogeneity of lesions, with synchronous osteolysis, even when the blastic appearance predominates. Histomorphometric studies have shown that OB lesions are mixed in nature with increased activity of both osteoblasts and osteoclasts contributing to the histological frailty observed in the skeleton in PCa patients, even in the presence of dense metastatic lesions [1].
Among patients with bone metastases, we found OB type metastases in 70.7%. Of these, 72 patients (38.3%) were found to have only OB metastases, while the remaining 61 patients (32.5%) metastases were not only OB but also OL and/or IM. Patients with extensive disease, with GG 4/5 and with high PSA levels, had more often combined IM + OB metastases, probably reflecting the aggressiveness of the disease, when mature metastases ones and new marrow deposits co-exist.
In almost one-third of patients with bone metastases, only OL type and/or IM metastases were present. We did not perform a head-to-head comparison between the PET findings and bone scintigraphy, but these results raise a question regarding the ability to relay on BS for staging of newly diagnosed prostate cancer and for assessment of disease extent in view of the limited sensitivity of BS in detection of lytic type and marrow-based metastases.
Two recent manuscripts, one evaluating 30 patients and the other evaluating 35 PCa patients, reported a high sensitivity of PSMA PET/CT in detection of the various types of bone metastases [18,29]. In 2015, Ceci et al. [33] published a study on the incidence of various metastases type in 140 newly diagnosed prostate cancer patients with bone involvement as identified on C-11 choline PET/CT. There are some differences in the findings presented by the above publication and ours probably. We found a higher incidence of IM metastases (21.8% of the patients with metastatic spread), compared to 6.5% in the C-11 choline PET/CT report. We found only OB type metastases in 38.3% of the patients and only OL type metastases in 4.3% compared to 62.3% and 26.1%, respectively, on the C-11 choline PET/CT study. Possible causes for the differences may be the different performance of labeled choline and labeled PSMA, difference in cohort size, and the fact that our cohort is composed of consecutive patients. The national reimbursement of PSMA PET/CT in newly diagnosed patients with intermediate and high-risk prostate cancer decreases the potential bias in referral of patients for PET/CT staging. As previously described by others, increased uptake of labeled PSMA in the skeleton may be seen also in nonmalignant lesions [21,23,35]. A list of benign bone lesions were reported to be associated with labeled PSMA accumulation including fibrous dysplasia, Paget's disease, enchondroma, and fracture. These lesions can be accurately diagnosed on the data of contemporaneous CT and should not be reported as metastases. Increased uptake in the rib with no morphological abnormality or other skeletal lesions is a common finding that is likely to be a false positive site of uptake and should be mentioned with caution. Only two lesions reported as equivocal were found in the current study to be metastases.
In summary, PSMA PET/CT staging of 963 consecutive patients with newly diagnosed intermediate-and high-risk PCa revealed malignant skeletal involvement in 19.5% of the patients, but only 7.6% had extensive disease, while the others had a solitary lesion or oligometastatic disease. Extensive disease was closely related with high ISUP Grade and high PSA level but was found also in patients with lower ISUP Grade and/or PSA level. Although bone metastases of PCa are considered OB, OL type and IM metastases are commonly identified on PET/CT. Author contribution MK, OY, and EES contributed to the conception of the study, data collection, study design, writing, and revising the manuscript; KK, ID, JK, CL, DS, and DK contributed to the data collection, writing, and revising the manuscript; all authors read and approved the final manuscript.

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations
Ethics approval and consent to participate All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This article does not contain any studies with human participants or animals performed by any of the authors. The study protocol was approved by the local institutional ethics committee (reference ID TLV-0327-20). All patients gave written informed consent.

Competing interests
The authors declare no competing interests.