Introduction
Prostate-specific membrane antigen (PSMA)-targeted approaches have gained increasing acceptance for the diagnosis and treatment of patients afflicted with metastatic castration-resistant prostate cancer (mCRPC) [
1‐
3]. As a type 2 transmembrane protein, PSMA is heavily overexpressed on the surface of prostate cancer cells [
4], providing a target for radiolabeled imaging and therapy [
1,
5‐
8]. For instance, [
177Lu]Lu-PSMA-RLT demonstrated high efficacy and safety in a first prospective phase 2 study (LuPSMA) with over half of the investigated patients demonstrating a biochemical response [prostate-specific antigen (PSA) decline of over 50%] [
1]. Further building on these promising results, Hofman et al. also recently revealed a favorable outcome for patients treated with [
177Lu]Lu-PSMA-RLT when compared to the standard chemotherapeutic regimen using cabazitaxel [
9]. Finally, the recently published prospective phase 3 VISION trial reported on longer progression-free (PFS) and overall survival for patients treated with [
177Lu]Lu-PSMA-RLT compared to standard care [
3,
10].
However, although patients had to demonstrate significant tumor PSMA expression to be enrolled into the VISION trial, 54% of the RLT-treated patients did not reach biochemical response [
3]. In previous studies, high PSMA expression on pretherapeutic
68Ga-labeled PSMA-targeted PET/CT emerged as a strong predictor for treatment response [
7,
11,
12]. Yet, all studies reported patients who did not respond although demonstrating increased PSMA expression [
11‐
13]. In light of these results, more reliable predictors are needed. Gafita et al. recently reported on the predictive performance of nomograms containing both PET-based and clinical parameters [
14], emphasizing the complexity of predicting treatment response in mCRPC patients receiving [
177Lu]Lu-PSMA RLT. Thus, investigating the underlying pathophysiological response caused by RLT-based radiation might help to better understand the biological mechanisms of response and non-response.
Used for PSMA-targeted treatment,
177Lu as a primarily beta-emitting isotope induces a plurality of DNA damages, including DNA double-strand breaks (DSBs), ultimately leading to subsequent cell death [
15,
16]. Following DSBs, a cascade of cellular DNA damage response (DDR) mechanisms is activated, involving the histone-variant γ-H2AX and the p53-binding protein 1 (53BP1) [
16‐
18]. However, beta-radiation from [
177Lu]Lu-PSMA does induce DSBs not only in tumor cells but also in benign cells such as peripheral blood lymphocytes (PBLs) which become exposed during circulation [
19,
20]. The time course of DSB formation in PBLs and the potential to obtain data on the individual dose–response relationships in vivo have recently been reported in [
177Lu]Lu-PSMA RLT [
19,
20]. Marked interindividual heterogeneity on both amount and kinetics of DNA damage response (DDR) foci during [
177Lu]Lu-PSMA RLT was also reported [
19,
20]. Given the broad range of measured γ-H2AX and 53BP1 levels potentially reflecting individual radiosensitivity, these parameters may also hold potential for risk stratification, serving as “liquid biopsy” to identify patients that most likely benefit from RLT [
21,
22].
In the present prospective study, we investigated the predictive performance of DSB markers relative to other established parameters, for the prediction of both early progression and progression-free survival in [177Lu]Lu-PSMA RLT.
Discussion
In the present prospective study, we evaluated the predictive performance of γ-H2AX and 53BP1 foci in PSMA-targeted RLT. There was a consistent trend emphasizing the predictive value of reduced DDR-markers at baseline for subsequent poor outcome under [177Lu]Lu-PSMA RLT, but likely due to the small patient cohort significance could not be reached. In this regard, low baseline DDR-marker levels prior to treatment on-set probably reflected low individual radiosensitivity and thus, were linked to early PSA-PD and shorter PSA-PFS. Thus, we herein provide for the first time data on the role of individual radiosensitivity in the context of PSMA-targeted RLT along with predictive capability, which requires further studies with larger patient populations to provide definite evidence. Nonetheless, the herein presented consistent results may also enable future studies investigating the potential of pretherapeutic administration of radiosensitizing agents, e.g., in patients having low baseline levels of γ-H2AX and 53BP1 foci.
Schumann et al. were the first to investigate changes in DDR-marker levels during [
177Lu]Lu-PSMA RLT [
27]. Enrolling 16 mCRPC patients, a distinctive time-dependent kinetic pattern of both γ-H2AX and 53BP1 foci was observed, with an early increase of both DDR-marker levels within + 4 h after administration and a subsequent decrease thereafter [
27]. Ritt et al. corroborated these findings by reporting on similar DDR-marker kinetic patterns enrolling a larger cohort of 48 patients under
177Lu-DOTATOC (n = 26) or [
177Lu]Lu-PSMA RLT (
n = 22) [
20]. Of note, in both studies, DDR-marker remained elevated compared to baseline even + 24 h after initiation of RLT [
20,
27]. This is in line with our results also showing a time-dependent kinetic pattern of DDR-marker levels in a prospective cohort of 20 mCRPC patients. Of note, a recent study reported on the predictive value of DDR-markers for therapy response under
177Lu-DOTATATE [
22], while the present investigation is the first to evaluate the predictive performance of γ-H2AX and 53BP1 foci for men treated with [
177Lu]Lu-PSMA RLT. Consistent with findings of that study investigating DDR-markers in patients with gastroenteropancreatic neuroendocrine neoplasms treated with somatostatin receptor-targeted peptide receptor radionuclide therapy (PRRT), low levels of 53BP1 foci were linked to shorter PFS [
22], supporting the notion of an incremental value of those DDR-markers for patients scheduled for endoradiotherapy.
Despite favorable results of PSMA-targeted RLT compared to other therapies in advanced disease [
3,
9], a substantial portion of patients will experience early PD, and thus, reliable predictors for subsequent treatment response would be crucial for optimizing patient care [
13]. In this regard, the averaged SUV
max derived from tumor lesions on pretherapeutic [
68Ga]Ga-PSMA PET/CTs has shown promise for predicting early response to treatment and overall survival [
11,
12]. However, due to various advantages such as improved image quality and longer half-life allowing for a higher throughput in a busy PET practice, recent years have witnessed an increased use of
18F-labeled PSMA radiotracers, e.g., [
18F]F-PSMA-1007 [
6]. Given this shift from
68Ga to
18F radiotracers in clinical routine, the predictive potential of
18F-PSMA PET in patients scheduled for RLT needs further exploration. We demonstrated that a low SUV
max derived from a pretherapeutic [
18F]F-PSMA-1007 PET/CT was significantly associated with early PSA-PD after two cycles of RLT (
P = 0.04). Nonetheless, SUV
max did not reach significance for the prediction of PSA-PFS in Kaplan–Meier analysis and univariate Cox-regression which might be clinically even more important than early PD for the long-term outcome of mCRPC patients. Although the study population was rather small, this may also be a hint at restricted predictive capability of
18F-PSMA PET-based quantification. While a strong relation between uptake of PSMA-ligands and histologic PSMA-expression has already been proven for [
68Ga]Ga-PSMA-11 in a dedicated ex vivo animal model [
28] and immunohistochemical validation of clinical primary prostate cancer samples [
29], such evidence has not been provided for [
18F]F-PSMA-1007 to date. However, there is an increasing body of evidence of potential pitfalls on scan interpretation for this
18F-labeled PSMA radiotracer [
30]. Along with the high rate of indeterminate findings on
18F-PSMA especially in the skeleton [
31] and moderate interobserver agreement rates [
32], one may question the reliability of
18F-PSMA PET-based predictors. One way to maximize the predictive performance of PET-based parameters could be a dual tracer approach for patient selection targeting both PSMA and glucose metabolism [
9]. Excluding patients with PSMA-negative and simultaneously FDG-positive metastases, Hofman et al. were able to achieve biochemical response in 65% of the RLT-treated patients compared to 46% in the VISION trial, in which only PSMA PET/CTs were performed for patient selection [
3,
9].
In addition, blood-based markers such as AP have also shown promise for the prediction of treatment response [
33,
34] and we also confirmed the value of AP levels in the present investigation (
P = 0.04), but inconsistent results question the suitability of this parameter to identify patients prone to early therapeutic failure [
13]. For instance, Gafita et al. reported on no predictive value of AP for overall survival under [
177Lu]Lu-PSMA RLT [
35]. As such, liquid biopsies assessing DNA damage due to ionizing radiation, either emitted through external beam radiation or internal RLT, may provide a more causal-system approach to identify patients prone to treatment failure [
17].
Following RLT-induced DSBs, a cascade of biological response mechanisms is activated including the phosphorylation of the histone H2AX to γ-H2AX and the assembling of 53BP1, subsequently leading to either repair, arrest, or apoptosis of the referring cell [
15‐
17]. Thus, one might speculate that low levels of the DDR-markers 53BP1 and γ-H2AX are associated with reduced responsiveness to radiation subsequently causing ineffective cell repair and less apoptosis [
36,
37]. In this regard Xiao et al. were able to demonstrate, that low levels of 53BP1 were associated with reduced expression of proteins fundamental for p53-dependent apoptosis such as the ataxia-telangiectasia mutated kinase and the checkpoint kinase-2 in a dedicated colorectal cancer cell line after radiation [
38]. Decreased expression of 53BP1 led to radiotolerance with reduced apoptosis and elevated proliferation rates after treatment [
38]. However, beta-irradiation due to
177Lu not only induces DDR mechanisms in malignant but also in benign cells, e.g., PBLs [
20,
22]. Of note, the evolution of γ-H2AX and 53BP1 foci in PBLs have previously been investigated under RLT [
20,
27].
In the present study, we expanded the concept of blood-based assessment of DNA damage and demonstrated clinical value of these parameters beyond dosimetry. We were able to demonstrate a clear and consistent trend emphasizing the predictive value of reduced DDR-markers assessed prior to treatment onset for subsequent poor outcome under [
177Lu]Lu-PSMA RLT. Future studies, however, may also measure DDR-foci in circulating tumor cells (CTCs) possibly enabling an even more accurate determination of tumor radiosensitivity [
21]. However, relative to PBLs, previous single-cell studies assessing mutations in CTCs of breast cancer patients revealed remarkable heterogeneity within investigated cells from the same patient [
39], and thus, DDR-marker in CTCs derived from men affected with prostate cancer may also be prone to such intra-patient heterogeneity, limiting their usefulness for prediction.
The herein presented results also support the hypothesis that low DDR-marker levels reflect low individual radiosensitivity explaining their predictive value for treatment outcome. Although other clinical parameters demonstrated similar or even superior predictive performance, e.g., AP, this is of importance as these parameters can hardly be influenced. By contrast, simultaneous chemoradiotherapy to boost radiosensitivity upon external beam radiation is widely practiced in numerous other tumor entities e.g. non-small-cell lung cancer [
40]. Thus, our study provides a rationale for future studies exploring the use of radiosensitizing agents under [
177Lu]Lu-PSMA RLT, especially in high-risk individuals exhibiting low baseline DDR-marker levels. Among others, a combination of [
177Lu]Lu-PSMA RLT and docetaxel has already demonstrated high response rates with only limited side effects [
41]. An even more sophisticated approach was presented by Guo et al. [
42]. Conjugating docetaxel-loaded nano-particles with a PSMA-ligand achieved high tumor doses while the whole-body dose was reduced [
42]. Given the herein reported findings, patients with low γ-H2AX and 53BP1 foci in PBLs at baseline may benefit from such combination approaches, e.g., by conducting a docetaxel-based radiosensitizing prior to RLT.
This study has several limitations. First, although we were able to demonstrate consistent and robust results, the baseline DDR-marker closely failed to reach significance in predicting early PSA-PD and PSA-PFS due to the small number of enrolled patients in this prospective study. Therefore, future studies need to verify these results in a substantially larger cohort with a longer follow-up interval, which would then also allow to investigate overall survival. In addition, factors influencing the individual expression of DDR-markers in PBLs before RLT remain unclear and should also be investigated. Moreover, additional characterization of PBLs, e.g., CD4 + or CD8 + , might also help in better understanding radiosensitivity of specific immune cell types and their influence on treatment outcome after radiation therapy [
43]. Future studies may also investigate the additional predictive role of DDR markers when combined with established outcome predictors, e.g., recently suggested nomograms [
14].
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.