Introduction
Up to 50% of patients with localized prostate cancer (PCa) undergoing radical prostatectomy (RPE) experience biochemical relapse within the first 5 years after treatment [
1‐
3]. Early salvage radiation therapy (sRT) with or without androgen deprivation therapy (ADT) is recommended as the only curative treatment option [
4]. However, recurrence rate after sRT is represented by heterogeneous patterns and influenced by clinico-pathological features such as pre-treatment prostate-specific antigen values (PSA), International Society of Urological Pathology Grade (ISUP), extracapsular extension (ECE), and seminal vesicle infiltration (SVI) and surgical margins with progression-free survival rates vary between 20 and 70% after 5 years [
2,
5]. Therefore, additional prognostic markers are needed to improve risk stratification and guide personalized treatment approaches such as dose escalation, adaption of RT fields, or intensification of systemic treatments. Implementation of positron emission tomography (PET) targeting the prostate-specific membrane antigen (PSMA), a cell-surface transmembrane protein over-expressed in PCa cells [
6], improves detection rates of recurrent PCa lesions even at low PSA levels and outside the prostatic fossa [
7,
8]. These findings have high impact on the management of salvage treatments with alteration in approximately half of patients [
9‐
11]. While retrospective evidence supports putative improvements of biochemical recurrence free survival rates due to PSMA-PET-guided sRT [
12,
13], results of prospective trials are pending ([
14], NCT04794777, PATRON–NCT04557501). Emerging data suggest that the maximum standardized uptake value (SUVmax) can be used as a biomarker to prognosticate clinically significant PCa [
15], Gleason-Score (GS) [
16,
17], and distant metastases [
18] in primary PCa, but no data exist on the applicability of this PSMA-PET feature in patients with relapse undergoing sRT. In search of new, non-invasive biomarkers for personalized risk stratification, this retrospective multicenter study aims to evaluate the impact of the SUVmax on biochemical recurrence free survival (BRFS) in patients with recurrent or persistent PCa cancer after RPE and
68 Ga-PSMA11-guided salvage radiotherapy.
Discussion
SRT is the last curative option for patients with recurrent or persistent PCa after surgery, but heterogeneous responses demonstrate the need for improved and differentiated risk stratification and subsequently appropriate adaptions in disease management. While implementation of PSMA-PET led to relevant improvements in disease localization, with results of prospective trials pending, it is not yet clear whether the sole spatial information of tumor burden and the accompanying changes in treatment management have a relevant impact on progression rates. Considering that PSMA-PET-positive findings might only represent the “tip of the iceberg,” the additional biological information provided by this molecular imaging bears great potential. To our knowledge, this is the first large multicenter retrospective study to evaluate the potential of SUVmax values in PSMA-PET as a new biomarker in patients with PCa persistence/recurrence. The findings from this study suggest that SUVmax values may significantly contribute to identify patients who are at higher risk for progression after sRT and therefore might benefit from treatment intensification, guiding personalized treatment approaches.
In contrary to prospective trials investigating sRT, most of the patients included in our study are likely to be at advanced recurrent disease stages with 70% having a PSA ≥ 0.5 ng/ml prior to sRT and 59% having nodal recurrence. Therefore, the comparison with data of recently published randomized controlled trials [
26‐
28] is limited. Nevertheless, sRT in fossa-confined patients yielded BRFS in 80% after 2 years. Considering slightly different definitions of endpoints, these results are comparable to freedom from biochemical failure in approximately 50–70% for patients with a pre-sRT PSA between 0.2 and 2.0 ng/ml in a large retrospective study of conventionally staged patients [
5,
29]. BRFS rates for patients with node-positive PSMA-PET findings dropped dramatically to 65% after 2 years, which is in line with findings from a prospective trial, reporting 3-year BRFS rates of 45% with PSMA-PET-positive disease outside the prostatic fossa [
30]. Our results thereby confirm that PSMA-PET findings are highly prognostic for BRFS.
In our retrospective analysis, presence of PSMA-PET-positive local and nodal recurrence positive on PSMA-PET were prognosticators for BRFS after sRT, whereas classical pathological and clinical parameters were not. Since Emmett et al. prospectively demonstrated the prognostic value of PET-positive findings, it is likely that they dominate established parameters in our cohort, since we included only patients with PSMA-PET-positive lesions. The improved BRFS associated with the presence of LR (HR 0.5,
p = 0.016) is explainable by the dose escalation of PET-positive LRs yielding sufficient RT dose coverage, which was previously reported to be beneficial [
10].
Despite favorable outcomes for patients with LR only, still nearly one-third of patients suffered from progression after sRT. In multivariate analysis, our results demonstrate a significant association of SUVmax ≥ 75% quartile (HR 2.3,
p = 0.022) with unfavorable BRFS in this subgroup. Sensitivity analysis confirmed these results with a HR of 3.9 in patients with LR only and FU of > 12 months. These findings are in line with the biological understanding of PSMA, with high PSMA-expression being associated with more aggressive disease [
31] and SUVmax correlating with PSMA-expression [
32]. Therefore, patients with LR and high SUVmax values might represent a subgroup with more aggressive PCa, potentially suffering from micro metastases outside the prostatic bed at the time of imaging and therefore benefiting from intensified treatments. Cox-regression and ROC analysis furthermore suggest that SUVmax in LR might be a valuable prognosticator in patients with both LRs and NR. However, these results need to be interpreted carefully, since ADT is administered more often and for a longer period of time in patients with NR (44% in our cohort).
Whether extraction of additional radiomic features from PSMA-PET images enables identification of additional prognosticators [
33] needs to be evaluated in future studies. However, implementation of SUVmax offers great potential in this scenario, since it is easily and non-invasively determinable with minimal resources and without additional costs and is not affected by interobserver variability [
34]. AUC values of the time-dependent ROC analysis showed the best discrimination in patients with > 12 months FU and LR only. In an exploratory analysis, we calculated SUVmax cut-off value for optimal discrimination. In all patients, the optimized cut-off value was slightly lower than the 75% quartile (11.8 vs 12.8). In the subgroup with FU > 12 months, the cut-off value was, however, more similar (13.0). Thus, a SUVmax threshold of approximately 13.0 should be validated in future studies in PSMA-PET imaging. This being said, the role of SUVmax in these patient subgroups needs to be evaluated in future studies including new tracers to validate putative cut-off values and design studies, which evaluate treatment intensification such as extension of RT fields to elective nodes or intensified systemic treatments. To define optimal RT fields, patterns of metastases need to be vigorously analyzed. Intensification of systemic treatments in sRT is currently investigated by the FORMULA-059 RCT (NCT03141671). Keeping in mind that sRT should be initiated at low PSA levels, implementation of SUVmax into risk stratification might even be relevant in this scenario, with approximately 50% and 65% of patients having PET-positive findings at PSA values < 0.2 ng/ml and between 0.2 and 0.5 ng/ml [
30]. Furthermore, administration of ADT was associated with favorable BRFS in the entire cohort in univariate analysis, but not in multivariate analysis. Additionally, ADT was not associated with BRFS in patients with LR only. Despite RCTs demonstrated a benefit of adding short-term ADT to sRT [
35], our results show that these findings cannot directly be transferred into the PSMA-PET era. Our findings suggest that patients with recurrence confined to the prostatic fossa in PSMA-PET might not benefit from systemic, but rather from local treatment intensification. Considering recent results of the multicenter retrospective SPIDER 01 (Abstract OC-0607), we therefore evaluated the effect of doses ≥ 72 Gy to the PSMA-PET-defined local recurrence. We could not identify a statistical significant difference but a trend for favorable BRFS in patients with LR only, who received dose escalation ≥ 72 Gy (
p = 0.086). However, this analysis is hampered by the relatively small number of patients receiving dose escalation and short follow-up. Future studies are needed to evaluate this aspect in depth.
Interestingly presence of NR was associated with significantly unfavorable BRFS (HR 2.1) but not SUVmax values in NR, suggesting that the additional biologic information provided by SUVmax values does not contribute to this patient subgroup, who already suffer from relevantly poorer prognosis. These patients might benefit from systemic treatment intensification, since despite dose escalation to PET-positive nodes, it is likely that sRT might not cover non-visible tumor spread.
Our study has some limitations. First, due to its retrospective design, protocols for PSMA-PET scans, sRT, and follow-up varied between centers and are prone to selection bias. Since the tracer kinetics depends on the time between injection and image acquisition, we want to point out that all scans were acquired in line with recent guidelines in all centers, but use of different PET scanners might affect comparability. Second, it is known that different reconstruction methods such as time-of-flight and point-spread-function have implication to SUV values. Most importantly, SUV values might be underestimated when taken from small lesions, i.e., sub-centimeter. In order to quantify this aspect, we analyzed SUVmax values with respect to lesion volume. We found a tendency of smaller SUVmax values for lesions < 1 ml, which could only be found in a minority of patients (
n = 14). Thus, we are convinced that this limitation is very limited towards the overall results. Furthermore, there are many PSMA-ligands on the market and to date both
68 Ga-PSMA-11 and
18F-DCFPyL received FDA approval, with different pharmacokinetics and pharmacodynamics as well as isotopes, raising questions about the reproducibility of SUV values. In our study, we employed only
68 Ga-PSMA-11, in order to generate more standardized data [
36]. Therefore, the different reconstruction algorithms between the different institutions pose the risk for deviation. In order to limit the inter-institution variability, we only employed PET/CT systems with EARL1 accreditations. However, this heterogeneity can also be regarded as a strength of our retrospective analysis as it makes the data more applicable in clinical routine when different scanners and reconstructions are used between different centers. Third, no central review of PET images was performed with potential differences in interpretation between centers. Fourth, the median FU is relatively short with a median FU of 24 months. Lastly pathological data from RPE was missing in up to 27% of patients, likely contributing to inferior Cox-regression results.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.