Various studies were able to show that Gallium-68-labelled prostate-specific membrane antigen (PSMA) positron emission tomography (PET) can improve nodal and distant staging of prostate cancer patients (1, 2). An additional benefit of PET imaging is that imaging parameters can be quantified, e.g., by the calculation of standardized uptake values (SUV), PSMA expressing tumor volume (PSMA-TV) and its derivatives. The maximum SUV (SUV_max) of tumor lesions has been shown to be prognostic for a plethora of diseases and tumor stages and various PET tracers, including the most commonly used tracer [¹⁸F]fluorodeoxyglucose (FDG) but also less frequently used tracers (3, 4). Recent studies reported that (semi-)quantitative PSMA parameters appear to be a promising prognostic parameter. These investigations were mainly performed in advanced metastatic disease with patients prior to PSMA radioligand treatment (5, 6). In these cohorts of patients, high PSMA uptake seems to be associated with adverse outcome. So far, no data is available for locally confined disease and primary staging of prostate cancer, probably due to the relatively short follow-up time with this novel radiotracer.

Regarding focal radiotherapy treatment escalation in non-metastatic primary prostate cancer patients, an important issue is the potential correlation between quantitative PSMA ligand uptake measures and tumor aggressiveness, e.g. its correlation with the histopathological defined Gleason score. Additional PET parameters could help in the decision for more personalized treatment options like focal radiation boost to tumors, which has shown promising results in magnetic resonance imaging (MRI) guided boost delineation and is currently investigated in PSMA based focal dose escalation trials (7–9). Only weak to moderate correlation has been observed between PSMA PET SUV_max during initial staging of prostate cancer and Gleason scores obtained by biopsy. Similar modest correlations were reported for serum PSA values and SUV_max (2, 10, 11). Most studies only investigated SUV_max and did not analyse further quantitative PET metrics. A novel quantitative PET parameter is tumor asphericity (ASP). ASP is a measure of tumor shape irregularity and has shown a strong association with patient outcome in various diseases and for different PET tracers (12–15). In a recent study with a relatively small number of patients, ASP from [⁶⁸Ga]Ga-PSMA-11 PET was strongly associated with Gleason scores in patients with primary prostate cancer (16).

The aim of our study was to investigate the correlation between different quantitative PSMA parameters, including PSMA derived tumor volume (PSMA-TV) and ASP, with Gleason scores and PSA values and examine if one of these parameters outperforms SUV_max, especially regarding personalized treatment options of the primary tumour in patients without evidence of loco-regional or distant tumor lesions.

Most patients had high-risk prostate cancer. Table 1 summarizes clinical characteristics and quantitative imaging findings of the study cohort.

The investigated quantitative PSMA PET parameters were significantly inter-correlated with correlation coefficients between 0.20 and 0.95. The only exception was SUV_mean and ASP, which were not significantly correlated (p = 0.79). Details are shown in Supplementary Table 1 . Regarding correlation between quantitative parameters of the primary tumor and clinical parameters, a significant, however low to moderate correlation with initial serum PSA values (Spearman rho between 0.19 and 0.45, all p < 0.05; Table 2 ; Figure 1 ) was observed. Correlation with Gleason scores obtained by previous biopsy was slightly lower (Spearman rho between 0.17 and 0.31, all p < 0.05 except for ASP; Figure 2 ).

Figure 3 shows the distribution of quantitative PET parameters for each Gleason score.

AUC analysis regarding the differentiation of high-risk from low- or intermediate-risk prostate cancer patients revealed poor to fair sensitivity and specificity for all investigated imaging parameters. AUC plots are depicted in Figure 4 and the respective values are shown in Table 3 . Comparison between AUC characteristics for different PET parameters showed that SUV_max is significantly better suited than SUV_mean to predict high-risk prostate cancer (p = 0.035), no significant differences between other quantitative metrics could be observed as shown in Table 4 . Additionally, SUVpeak was investigated in the whole cohort, SUVpeak showed a very high correlation with SUVmax (r = 0.99, p < 0.001) and similar results regarding all investigated endpoints as shown in Supplementary Figure 1 .

Since Gleason scores obtained from biopsy might over- or underestimate surgically obtained Gleason scores of whole prostate specimens, a sub-group of 38 surgically treated patients was further evaluated. Similar correlation coefficients as in the main analysis (but with each p>0.05) were obtained between quantitative imaging parameters and surgical Gleason scores ( Supplementary Table 2 , Supplementary Figure 2 ).

PSMA PET has shown great potential for focal treatment strategies. Bettermann and colleagues were able to show that PSMA PET-based tumor delineation is superior to MRI regarding the sensitivity to detect prostate cancer foci on whole mount histopathology specimens (21). Several studies are currently investigating focal treatment escalation by the implementation of PET imaging. Identification of the optimal imaging parameter as a surrogate for tumor aggressiveness is therefore an important need.

In this study, we examined the correlation between PET parameters and clinical risk factors in non-metastatic primary prostate cancer patients. We were able to validate prior publications that reported a moderate correlation between clinical risk parameters like Gleason score, PSA levels or D’Amico risk category and SUV_max of primary prostate tumors. Further analysis of additional quantitative PET parameters like ASP or PSMA-TV did not show superiority compared to SUV_max in this monocenter investigation. Only a moderate correlation of any investigated parameter with Gleason scores could be observed.

The reported correlation coefficients in our study are comparable with published data on correlations between SUV_max and Gleason scores that ranged between 0.096 and 0.5 and correlation coefficients between SUV_max and PSA values that ranged between 0.071 and 0.57 (2, 10, 11, 22–27). All but one of these studies reported lower numbers of patients, Supplementary Table 3 gives an overview of the published data on correlation coefficients.

Gleason scores of needle biopsies show discrepancies with surgical Gleason scores in up to 50% of cases, especially upgrading to higher Gleason scores is a frequent observation (28, 29). This can influence the observed correlations with quantitative PSMA metrics, probably underestimating the real Gleason score. Analysis of the patient sub-group that underwent surgery did not show any significant correlation between the investigated quantitative PET metrics and surgical Gleason grades. However, this is most likely due to the comparatively low number of patients in this sub-group, because correlation coefficients were similar to the correlation coefficients for biopsy-based Gleason scores.

Data on the correlation between quantitative PSMA PET metrics other than SUV_max and clinical risk factors are sparse. Meißner and colleagues reported a strong correlation between ASP and Gleason scores (rho 0.88) and a moderate correlation between tumor volume and Gleason scores (rho 0.51) in a small cohort of 37 patients (16). However, patients with lymphatic or distant metastases were not excluded in their analysis, the exact number of patients with extraprostatic lesions was unfortunately not reported. Hoberück et al. evaluated various quantitative PSMA PET metrics including SUV_max, SUV_mean and PSMA-TV. In a small cohort of 21 patients with consecutive PSMA scans before and during androgen deprivation therapy, they observed a strong correlation between the investigated PET parameters and no superiority of a specific parameter (30). The same quantitative parameters were investigated by Schmidkonz et al. in patients with bone metastases. They reported that all quantitative metrics were higher for Gleason scores > 7, but did not provide further comparative details (31).

Our study has several limitations. First, the retrospective nature of the investigation with its known limitations. Second, no spatial correlation analyses with whole-mount histology was performed in surgically resected patients. Current analyses in this regard showed an excellent correlation of PET parameters with intraprostatic tumor foci (32, 33). Third, the used radiotracer might not be the best modality for local tumor assessment. The high urinary clearance of [⁶⁸Ga]Ga-PSMA-11 hampers automatic delineation in close vicinity to the bladder. The necessary manual modifications are observer-dependent and might complicate independent reproducibility. Furthermore, high bladder uptake can potentially affect quantitative PET metrics of the prostate, e.g. by halo artifacts (34). The F-18-labeled PSMA-1007 radiotracer might be superior for evaluation of primary prostate cancer due to its favorable biodistribution, in particular lower bladder activity (35). Furthermore, SUV_max in primary prostate cancer lesions are systematically higher with [18F]F-PSMA-1007 compared to [68Ga]Ga-PSMA-11 (36). Nonetheless, a current meta-analysis was not able to show clear superiority of one of the specific PSMA radioligands in the recurrent situation (37). If prolonged uptake times are encountered in routine clinical care, [18F]F-PSMA-1007 could be advantageous over [68Ga]Ga-PSMA-11 by providing beneficial count statistics due to its longer physical half-life. Additionally, the higher positron range of Gallium-68 compared to Fluor-18 results in decreased spatial resolution, although Soderlund et al. observed only marginal differences using clinical PET scanners (38). The range of uptake times in the current analysis was relatively high, which might hamper inter-patient comparability of SUV. Lesion uptake of [68Ga]Ga-PSMA-11 increases over time after injection and has been described as approximately irreversible (39). However, the average increase in lesion SUV between 1h and 3h post injection has been reported to be moderate (25%) (40). The same PET scanner was used in all patients, which benefits comparability of PET parameters between patients. However, strictly speaking, applicability to other scanner models with different image properties and reconstruction methods would require dedicated analyses.

An important strength of our analysis is the restriction to patients without evidence of metastases by imaging including PSMA PET. Inclusion of metastatic patients might partly explain the high heterogeneity between previous publications, especially regarding correlation coefficients with PSA values (which is highly correlated with the total tumor volume). Additionally image evaluation was performed in a standardized fashion and with the observer being blinded to clinical risk parameters.

Overall, the observed association of the investigated quantitative imaging parameters with clinical risk factors is only fair. Novel methods like radiomics might be more suitable to detect high-risk sub-volumes within the prostate (41, 42).

In summary, this comprehensive analysis of quantitative PSMA PET metrics confirms prior studies that showed a moderate correlation with clinical risk factors. All investigated quantitative PET metrics intercorrelated and showed similar association with Gleason score, PSA values or D’Amico risk groups. The widely used reporting of SUV_max only seems therefore reasonable for personalized treatment options like focal boost in primary prostate cancer. Further prospective studies in a large cohort are needed to confirm our results, especially regarding the outcome after PET-guided personalized treatment.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethikkomission der Charité Universitätsmedizin Berlin, Germany. The ethics committee waived the requirement of written informed consent for participation.

Study conception and design: SZ and KH. Drafting of manuscript: SZ and SA. Image processing and analysis: SZ, FH, and JR. Study Investigators: SZ, SA, HA, CF, JR, MB, FH, and KH. Interpretation of data: all authors. All authors contributed to the article and approved the submitted version.

HA declares research grants, travel grants, and lecture fees from Sirtex Medical Europe; HA confirms that none of the above funding sources were involved in the preparation of this paper.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.