The major hypothesis tested in our study was whether increased levels of PSMA as induced by ARB result in enhanced tumor targeting by 177Lu-PSMA617. As androgen receptor suppresses PSMA transcription , its inhibition with enzalutamide might lead to PSMA upregulation and, consequently, to a more effective treatment of tumors with 177Lu-PSMA617.
We therefore investigated in a mouse model of prostate cancer (i) the effect of ARB on PSMA expression over time and (ii) the effects of ARB on the efficacy of 177Lu-PSMA617 RLT. Our results indicate that a 21-day regimen of ENZ increases PSMA expression in C4-2 xenografts that peaks around days 23–29 and is maintained even after cessation of ENZ administration. However, ENZ-induced PSMA expression does not retard tumor growth more profoundly than 177Lu-PSMA617 alone. Furthermore, survival was not significantly improved.
First, we established ARB regimens with BIC or ENZ which effectively delayed tumor growth and induced a transient increase in PSMA expression in human prostate cancer xenografts in vivo. PSMA levels peaked 2–8 days (days 23–29 after ENZ start) after cessation of ARB treatment. Thus, when treated with ENZ for 21 days, PSMA levels remained above baseline for another 13–18 days (days 34–39 after ENZ start). This provided the rationale for administrating RLT on day 23 (see below).
The increases in 68Ga-PSMA11 tumor uptake in ARB-treated animals over controls were small (max. ~ 2-fold). Both the small increment and the PSMA kinetics corroborate and extend previous studies investigating the effect of androgen receptor inhibition on PSMA levels and PSMA-targeted imaging [4,5,6,7]. These studies reported enhanced PSMA expression in PC xenograft mouse models (i) upon androgen deprivation (6 days treatment) using 64Cu-J591 PET on day 7 (an antibody targeting the extracellular domain of PSMA)  and (ii) following orchiectomy or a 7-day regimen of apalutamide using 68Ga-PSMA11 PET on day 7 . The latter study also reports increased PSMA levels in a patient following a 30-day ARB therapy. Only one study investigated PSMA levels after ARB (ENZ) removal in vitro . Following a 21-day treatment with ENZ, PSMA levels peaked around day 21 and remained elevated (compared to untreated cells) 4–7 days after ENZ discontinuation. Importantly, this study demonstrated increased efficacy of PSMA antibody-drug conjugates with increasing PSMA expression.
Second, RLT and ENZ+RLT significantly delayed tumor growth and improved survival (compared to ENZ-only and vehicle). In addition, pre-treatment with ENZ led to more profound DNA damage (phospho-γH2A.X foci) compared to RLT alone. This is in line with in vitro studies reporting that treatment with ENZ combined with irradiation [15, 16] or the poly (ADP-Ribose) polymerase inhibitor olaparib  increased the number of phospho-γH2A.X foci resulting from either treatment (X-ray or olaparib) alone.
Although this would be expected to improve therapeutic efficacy, pre-treatment with ENZ did not result in additional tumor growth retardation when RLT was compared to ENZ+RLT. An accurate survival analysis was not possible as on day 81 all remaining mice (3/10 in the RLT group; 8/10 in ENZ+RLT) had to be sacrificed (compare last paragraph of result section).
While ENZ efficiently reduced tumor growth (Fig. 3a), tumors rapidly regrew in the ENZ-only group after termination of ENZ treatment (Fig. 5a shows tumor size change from the time after completion of ENZ and before RLT). This might suggest that ARB is only effective during treatment. However, as shown in Figs. 2 and 3b and c, increased PSMA expression persisted after cessation of ENZ up to day 34, i.e., almost 2 weeks after end of ENZ therapy. In fact, PSMA levels peaked between 23 and 29 days after start of enzalutamide treatment (Fig. 3b, c). 177Lu-PSMA617 radioligand therapy was performed on day 23 (i.e., 2 days after the last dose of ENZ), at a time when ENZ-induced PSMA expression peaked. 177Lu-PSMA617 binds to PSMA-expressing tumors within hours . It is then rapidly internalized . Unbound 177Lu-PSMA617 is excreted via the kidneys. Therefore, we ascertained high PSMA expression levels at the time of RLT administration. Maintaining high PSMA expression beyond the time of RLT is thus likely less (if at all) relevant for RLT efficacy. It remains unclear whether ARB-induced increases in PSMA expression might have been insufficient to enhance RLT efficacy. While we did not perform biodistribution studies to determine if the increased PSMA levels led to a higher tumor uptake of 177Lu-PSMA617, preclinical evaluation of 68Ga-PSMA11/177Lu-PSMA617 by Umbricht et al. demonstrated very similar uptake and internalization of the 68Ga- and 177Lu-ligand, respectively, by PC-3-PIP cells . Likewise, the %IA/g in PC3-PIP tumors was comparable for 68Ga-PSMA11 and 177Lu-PSMA617. In a similar study, Weineisen et al.  found that uptake and internalization of 68Ga/177Lu-PSMA I&T—a similarly used theranostic pair of PSMA ligands—into LNCaP cells was almost identical. Lastly, SUVmean on 68Ga-PSMA11 PET/CT images correlated with absorbed dose in men with mCRPC . Therefore, we propose that the uptake of 177Lu-PSMA617 into C4-2 tumors can be inferred from the uptake of 68Ga-PSMA11. Our data (Figs. 2 and 3) thus suggest that tumors conditioned with ENZ took up more 177Lu-PSMA617 than tumors without ENZ-induced PSMA upregulation. However, the actual absorbed dose of 177Lu-PSMA617 in tumor or organs cannot be calculated based on 68Ga-PSMA11 PET images, because the PET images only give information about the tracer activity. Future studies will need to define the relationship between the degree of target expression, tumor-absorbed dose, and RLT efficacy.
An alternative or additional explanation for the lack of synergy is based on the notion that AR signaling has been implicated in DNA damage repair [23,24,25]. In the setting of RLT, AR-triggered repair might protect tumor cells from cell death; its impairment by ARB could thus synergize with RLT. However, radiation causes the activation of several DNA damage repair pathways that act in parallel . As we did not observe an enhancement of RLT by ARB, either (i) impairment of AR-triggered repair might not be a critical determinant of RLT efficacy on its own or (ii) the ENZ-mediated AR blockade ceased with discontinuation of the ENZ administration. The latter might imply the need for conditioning plus continuation of ARB under RLT to first increase PSMA expression and, subsequently, block repair of DNA damage once it occurs.
Finally, it is possible that surgical castration or castration plus ENZ would have had a greater impact on AR signaling and, consequently, PSMA expression, tumor growth, and (AR-mediated) DNA damage repair. However, Evans et al. and Hope et al. directly compared the impact of surgical castration vs. treatment with anti-androgen (10 mg/kg ENZ and 10 mg/kg apalutamide, respectively) on PSMA levels and tumor growth [4, 5]. While both interventions decreased tumor growth to a similar extent, upregulation of PSMA was stronger following treatment with anti-androgens.
Combination therapies will be required to enhance the effectiveness of RLT. For instance, a current randomized phase III trial investigates the combination of abiraterone and the alpha emitter Radium 223 (NCT02043678) in mCRPC. Another prospective trial (e.g., ANZCTR12615000912583) combines second-line ARB in conjunction with 177Lu-PSMA617. Finally, a recently started trial in Australia combines RLT with DNA damage response inhibitors (NCT03307629).
While we failed to show improved tumor growth retardation with ENZ+RLT, a beneficial survival benefit cannot be ruled out with certainty. First, mice had to be sacrificed on day 81, mostly due to tumor ulceration. Thus, it is possible that survival benefits in the combination therapy group may have been missed. Secondly, responses to RLT combinations may differ between the current and other murine models. We selected the current model as C4-2 reflects the clinically relevant CRPC . As ENZ upregulated PSMA in C4-2 tumors more potently than BIC, it was chosen as the conditioning regimen. Furthermore, it remains to be determined if multiple (vs. single) injections of 177Lu-PSMA617 would enhance synergism with a PSMA-inducing drug. As PSMA levels started to decline around days 34–39, testing this hypothesis would have required either (i) to repeat the cycle of ENZ conditioning followed by 177Lu-PSMA617 a few weeks after the first cycle or (ii) to continuously administer ENZ until a further injection of 177Lu-PSMA617 could safely be given. In the current setting, the prerequisites for a potential positive impact of multiple 177Lu-PSMA617 administrations on treatment outcome would include that in scenario (i) subsequent cycles would lead to the same PSMA surge as observed after a single 21-day ENZ regimen and that in scenario (ii) continuous treatment with ENZ would maintain elevated PSMA levels.
Mice in the present study were treated with a single injection of 15 MBq 177Lu-PSMA617 (specific activity 84 GBq/μmol) which corresponds to 0.6 GBq/kg body wt in a 25 g mouse. This activity is sub-optimal for treating C4-2 tumors with RLT (our unpublished observations) and should allow the detection of potential synergistic effects of treatment combinations. Clinical protocols currently registered at clinicaltrials.gov use 4–6 cycles 177Lu-PSMA with cumulative activities ranging from 14 (4 × 3.7 GBq) to 51 GBq (6 × 8.5 GBq) . In the recently opened prospective phase 3 randomized VISION trial (NCT03511664), patients receive up to 6 cycles with 7.4 ± 10% GBq 177Lu-PSMA617 each; assuming an average weight of 80 kg, men thus receive 0.08–0.10 GBq/kg bodyweight 177Lu-PSMA617 per cycle. This is 6–7.5 times lower than the activity administered to the mice in our study. However, considering the cumulative activity patients receive over ≤ 6 cycles (maximum 39.96–48.84 GBq in the VISION trial), an 80-kg man would be treated with a total activity of 0.50–0.61 GBq/kg bodyweight. Therefore, the cumulative activity tested in clinical protocols and in our study is similar.
The differences in respective fold increase in PSMA expression measured by flow cytometry vs. 68Ga-PSMA11 PET-imaging in this study cannot be explained. However, a limited dynamic range of 68Ga-PSMA11 PET signal intensity (acquired with the Genisys 8 PET/CT) has previously been observed . In that study, PSMA expression (as quantified by flow cytometry from FNA tumor biopsies) and 68Ga-PSMA PET signal correlated at lower PSMA levels; this correlation was lost at higher PSMA expression levels.