The approved therapies for patients with mCRPC that can significantly improve OS are next-generation anti-hormonal therapies (abiraterone and enzalutamide), first- and second-line chemotherapies with docetaxel and cabazitaxel, and alpha radionuclide therapy with radium-223 . Currently, several novel agents, such as immunotherapeutics or therapies targeting poly (adenosine diphosphate–ribose) polymerase (PARP) inhibitors, are under advanced clinical investigation . Radionuclide imaging and therapy using PSMA as a target have been investigated for more than two decades beginning with monoclonal antibodies that were later replaced by low molecular weight PSMA ligand inhibitors . Since 2015, there have been increasing numbers of published data showing promising efficacy and a low toxicity profile of these agents in mCRPC patients [6, 7, 11, 22].
Although this therapy is still experimental and is mainly given in compassionate-use scenarios only, different societies have published guidelines or recommendations trying to standardize this therapy for patients with mCRPC [23,24,25,26]. The most important limitations of the published data are the limited number of included patients and the heterogeneity of patients regarding prior therapies.
Ahmadzadehfar et al. reported a median OS of 60 weeks in 100 patients who received a median of 3 cycles of 177Lu-PSMA-617 (total 347 cycles). All of the patients had a history of therapy with either enzalutamide or abiraterone, or both. At least one line of chemotherapy had been performed in 70% of the patients, and 36% had a history of radionuclide therapy with radium-223. Here, a PSA decline after the first RLT, as well as a decline ≥ 50%, was significant prognosticators of longer OS. Rahbar et al.  reported an OS of 56 weeks in 104 patients treated with 351 cycles of 177Lu-PSMA-617. All of these patients had a history of therapy with at least one line of chemotherapy, as well as either abiraterone or enzalutamide. Both studies showed that patients who respond to PSMA therapy live longer than those who do not. In these studies, prior therapies, such as chemotherapy, had no impact on OS. Heck et al.  reported the results of therapies using [177Lu]Lu-PSMA-I&T in 100 patients with mCRPC who received a total number of 319 cycles (median 2 cycles). Here, the median OS was 12.9 months (95% CI 9.9–15.9). The included patients were comparable with those of our cohort regarding prior therapies.
Heck et al. also showed a significant correlation between PSA response under RLT and survival. They analyzed the PSA changes within 12 weeks of RLT. According to their analysis, a maximum PSA decline of ≥ 50% was associated with longer OS (median 16.7 (n = 32) vs 6.2 (n = 60) months, p = 0.007). In our current study, the median OS of patients with rising PSA, decline < 50%, and decline ≥50% were 7.2 months, 13.9 months, and 14.3 months, respectively (p < 0.0001); however, there was no significant difference between patients with more or less than 50% decline regarding OS (p = 0.6) by measuring PSA 8 weeks after the first cycle. This difference may be due to the measuring time point of PSA. The best response, within 12 weeks, means that the majority of patients got two cycles of therapy; on the other hand, it seems that Heck et al. did not differentiate between rising PSA and a decline of less than 50%.
Despite PCWG3’s suggestion to measure PSA decline after 12 weeks in clinical trials, prior studies as well as the present study found that a good response to the first cycle (a PSA decline, measured 2 months after the first cycle) is associated with a favorable response to further cycles in more than 90% of patients [13, 27].
Barber et al. reported the results of 167 mCRPC patients who were treated with 177Lu-PSMA-617 or 177Lu-PSMA-I&T . The patients were divided into two groups according to prior therapy with taxane-based chemotherapy. The median OS in 83 patients in the taxane-pretreated group was 10.7 months; it was 27.1 months in 84 patients in the taxane-naïve group . In the taxane-pretreated group, 76% and 14% of patients prior to RLT had received abiraterone or enzalutamide and Ra-223 treatments, respectively, while only 38% and 2% of patients in the taxane-naïve group had received abiraterone or enzalutamide and Ra-223 therapies, respectively . This likely means that the long OS of 27.1 months in the taxane-naïve group was caused by the fact that about 60% of the patients in this group had received an RLT as a first-line therapy, thus the natural course of their prostate cancer was associated with a significantly longer OS than that of the patients who had been treated with various mCRPC-approved compounds prior to 177Lu-PSMA-RLT. Thus, despite longer OS in the taxane-naïve group, in the multivariate analysis, prior chemotherapy was not a significant prognosticator of overall survival .
In our multicenter analysis, all of the patients had received abiraterone or enzalutamide, and 53.6% had been treated with both agents. A total of 75.5% had at least a history of first-line chemotherapy with docetaxel. A therapy with Ra-223 had been done in 20.4% of the patients. Altogether, the included patients in this retrospective multicenter analysis were heavily pretreated. Despite prior therapies and advanced disease (92.8% and 20.9% bone and liver involvement, respectively), the median OS was 11.1 months (95% CI 9.7–12.5 months) and was comparable with the taxane-pretreated group of Barber et al. and other prior studies [11,12,13,14].
In terms of treatment planning, having predictive parameters for a favorable response and prognosticators of longer OS is of importance for us as clinicians, first, to decide on the indication of a therapy, and second, to accurately inform patients about the treatment response rate and their prognosis for survival. In the current study, age, GS, prior therapies with abiraterone or enzalutamide as well as Ra-223, and the existence of lymph node as well as lung metastases were not significant prognosticators of OS, while prior chemotherapy, ECOG status and the existence of bone and liver metastases were prognosticators of OS in both univariate and multivariate analysis.
As mentioned above, a prior therapy with enzalutamide did not have a significant impact on OS; however, there was a significant difference between the OS of patients with a history of enzalutamide and the OS of patients who were under ongoing usage of enzalutamide (12.3 vs 10.8 months, respectively; p = 0.045). These findings should be further explored in future studies. One explanation could be the negative impact of enzalutamide on PSMA expression, reducing the tumor-absorbed dose, or an agonistic effect of enzalutamide on patients who do not respond any more to this agent. The potential impact of enzalutamide and abiraterone on the efficacy of PSMA-RLT when used concurrently is therefore very interesting, even more so since the current phase III registrational trial, VISION, tests PSMA-RLT in combination with enzalutamide or abiraterone only. Thus, in the future, depending on the results of VISION  and subgroup analyses, a PSMA monotherapy may have to be compared with a combination therapy including abiraterone or enzalutamide. For some patients, PSMA-RLT alone may be sufficient.
Patients who had received a prior therapy of Ra-223 showed a longer OS during the first 10 months as compared with patients without any Ra-223 (Fig. 3). Ra-223 is a bone-seeking alpha-emitting radionuclide acting in reactive bone forming cells adjacent to cancer cells. This therapy has demonstrated a median OS of 12.4 months alone in 150 patients . The early improvement of OS is thus possible due to existing long-acting synergistic Ra-223 effect; this effect lasts typically 4–5 months, and thus an average of 7–9 months benefit is expected. After 10 months, the 177Lu-PSMA-617 showed better OS in those patients who did not receive Ra-223.
The existence of visceral metastases is a negative prognostic factor, as other studies have shown [11, 12, 30]. Although in our multicenter study patients without prior chemotherapy showed a significantly longer OS, we should take into consideration that some of these patients who had initially avoided chemotherapy received it after having progressed on PSMA treatment. Future trials will have to elucidate the ideal position of PSMA-RLT within the ever-growing armamentarium of therapies for patients with mCRPC. This issue cannot be analyzed in a retrospective setting, and it should thus be evaluated in a prospective setting.
Since we included patients from different departments, we did not analyze baseline laboratory parameters such as blood count, alkaline phosphatase, LDH, and different tumor markers other than PSA, first, because different laboratories have different ranges of normal, and second, because apart from blood counts, the other parameters were not checked routinely in all clinics.
One of the most important limitations of this study is its retrospective design; however, we tried to exclude all patients with unclear follow-up or documentation from the analysis. The high proportion of patients who were excluded in this analysis is a major drawback of the analysis and is due to the retrospective design and recorded data mostly in clinical routine and different countries.
Additional limiting factor is the timepoint of calculation of survival rates, especially according to performed chemotherapy, this might represent a lag-time bias. This might be overcome with results of the prospective trials, which are currently running [28, 31].