Introduction
For patients with metastatic castration-resistant prostate cancer (mCRPC), prostate-specific membrane antigen (PSMA)-targeted radioligand therapy (RLT) has shown survival benefits in recent prospective clinical trials [
1,
2], leading to approval by the Food and Drug Administration [
3]. Despite those beneficial results, a substantial portion of patients does not respond to therapy, and biomarkers to identify patients prone to treatment failure include regular assessment of prostate-specific antigen (PSA) levels. In this regard, early biochemical response 8 weeks after the first cycle has been advocated to be linked to prolonged overall survival (OS) [
4‐
6]. However, failure to achieve substantial decline of PSA levels does not necessarily imply treatment failure. In this regard, a PSA flare phenomenon has been described for various prostate-specific therapies, e.g., docetaxel, abiraterone, or other therapeutic radionuclides, such as radium-223 dichloride [
7‐
9]. Defined as an initial increase in PSA levels in the first weeks after therapy, followed by a delayed decrease in PSA levels, the outcome in patients with such a PSA flare was similar to those in men exhibiting early biochemical response [
7‐
9]. Flare phenomenon has also been discussed in RLT [
10], but appears to be very rare [
5]. Nevertheless, previous studies have reported on up to 30% of initial non-PSA responders having a delayed decrease in PSA levels [
10,
11].
Enrolling the largest cohort to date for this purpose, we aimed to assess OS in patients experiencing initial and sustained PSA increase, continuous decrease, or fluctuations during therapy (defined as change in PSA levels after initial decrease or increase). We then compared survival among subgroups, thereby allowing to determine whether survival benefits of patients with PSA fluctuations under RLT are comparable to those under sustained PSA decline.
Discussion
In the present analysis including 176 patients under PSMA-directed RLT, we demonstrated that patients with signs of response or disease stabilization have longer OS compared to patients with PD, when current PCWG3 criteria are applied [
16]. Independent of those criteria, we demonstrated that patients with an initial decrease in PSA levels are highly likely to sustain this decrease over four cycles of therapy and that this trend was linked to prolonged OS. Men with initial PSA increase, however, exhibited a continued rise of their tumor marker levels during follow-up, ultimately experiencing shorter survival. PSA fluctuations, defined as change after initial decrease or increase, occurred relatively frequently and survival rates were comparable to the beneficial outcome seen in the group of patients with sustained PSA decline. Thus, such tumor marker oscillations should not lead to discontinuation of PSMA-RLT, in particular in patients with initial decline followed by a rise in PSA levels, as this scenario may be commonly mistaken as disease progression, especially in the absence of radiological progression.
Recommendations of response assessment for anti-tumor treatments in men with PC have been established by PCWG3 [
16]. Using those commonly applied response criteria, we demonstrated that patients with PD (category 6) had shorter survival when compared to men with signs of response or disease stabilization (categories 1–5; Fig.
1). As the latter categories also incorporate patients with PSA fluctuations, PCWG3-based assessment revealed then identical findings similar to our analyses using a simplified definition of PSA response (Fig.
4a). Of note, there were also no significant differences between patients within the categories 1–5 for OS, and thus, individuals with PSA fluctuations had also comparable survival benefits to those subjects with PCWG3-defined sustained response (category 1). Taken together, independent of defining PSA response, patients with tumor marker oscillations exhibit comparable survival to those individuals which always show biochemical response under RLT.
Our analysis indicates that patients exhibiting an initial PSA response sustain this trend over multiple cycles, while in patients with initial increase of PSA, there is a high likelihood of a continued rise of tumor marker levels. These findings corroborate previous results, in particular for [
177Lu]Lu-PSMA-617 [
4,
10]. In a matched-pair analysis, we have also demonstrated no relevant survival differences in patients treated with either [
177Lu]Lu-PSMA-617 or -I&T [
13]. Thus, given the results of this previous and the present study providing comparable percentages of patients maintaining an initial PSA decrease or increase, the treating physician can be confident that these trends are to be expected for both PSMA-targeted agents [
13]. These considerations are further fueled by a sub-analysis (Supplementary Fig.
1), where we also observed comparable results for patients treated with [177Lu]Lu-PSMA-617 or -I&T, as seen for the entire cohort.
Among the entire cohort, we observed a late response (i.e., PSA increase followed by a decrease) in only 7/176 (4%). As such, enrolling a substantially larger cohort, we corroborated previous findings by Gafita et al., which also observed this phenomenon in approximately 1% of their patients [
5]. Previous work, however, reported on approximately 8% [
4], which were comparable to the ones reported by Soydal et al. with 10% and up to 17% by Rahbar and coworkers (% derived by extrapolation taking the entire cohorts into account) [
10,
11]. As a possible explanation, the number of included individuals was lower relative to the present work and unifying definitions along with prospective studies further evaluating this aspect in mCRPC patients under RLT are warranted.
Within the subgroup showing PSA fluctuations, the majority of men experienced an increase of PSA levels after initial decrease (> 75%). In clinical routine, this finding is normally explained by disease progression. However, these patients still showed a longer OS, which was comparable to those with initial and sustained decrease of PSA levels and significantly longer when compared to patients with continuous increase (Fig.
4a). These results suggest that patients initially showing a decrease in their PSA levels may rather not discontinue treatment, in particular if they maintain this trend or even when they exhibit a PSA rise during follow-up, especially in the absence of radiological progression. Moreover, results of the present study are also in line with other reports on prostate-specific therapies and PSA flare under therapy, including chemotherapy, anti-hormonal treatment, or the use of alpha emitters primarily targeting osseous disease manifestations. For those therapies, early discontinuation due to transient PSA rises was also not recommended [
7‐
9]. In addition, an initial (or later) PSA response might also emerge as a powerful prognosticator for OS, further supporting previous analyses on a positive predictive value of an initial PSA response after the first treatment cycle [
4‐
6]. Nonetheless, PSA monitoring can only partially address the complexity of responding to PSMA-directed RLT, and thus, other markers of response may be also of relevance, e.g., circulating tumor cells [
17]. Those established or currently emerging biomarkers may be less prone to oscillations as observed for PSA in the present investigation.
Our results are limited due to the retrospective design. Nonetheless, the PSA values were consistently assessed on a regular basis and time-points of blood samples were in accordance to current procedure guidelines [
18]. Relative to previous reports, the number of patients is high and consisted of two cohorts treated with the two most commonly used radiopharmaceuticals, which have been reported to be comparable in their therapeutic efficacy [
13]. In addition, we did also not observe any differences in the patient’s characteristics between both cohorts in our study (Table
1). In our cohort, only a few patients had documented post-RLT treatments (four patients had radiotherapy, eight had chemotherapy, two had abiraterone, and one subject had received olaparib). However, given the fact that PSMA-RLT is currently considered the last-line treatment option [
18], we assume that the possible impact of follow-up therapies on survival may have been rather limited. In addition, future studies may also perform sub-analyses for patients which had received the identical treatment algorithm prior to RLT, preferably in a larger number of subjects. Moreover, radiological assessment is also used for identifying patients that do not respond to RLT and combined approaches by PSA measurements and PSMA-PET/CT have already shown to be prognostic for OS [
19]. Thus, the herein observed PSA fluctuations should also be evaluated together with such PET/CT-based response evaluations. Further investigations may then also focus on patients which showed PSA increase after initial decrease and whether those individuals also experience radiological progression.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.