Future therapeutic strategies for metastatic prostate cancer

In 1994 it was discovered that dendritic cells (DCs) could ex vivo be generated from monocytes by culturing them in the presence of IL‑4 and GM-CSF [4]. However, the DC function was negatively affected probably by the long culture period and the addition of compounds required to differentiate them into DCs. During the last decade it has been discovered that natural DCs (plasmacytoid and myeloid DCs) can easily be obtained by leukapheresis without the need of culturing and several clinical trials are demonstrating its safety and effectiveness [5, 6].

Several small studies have been conducted to investigate whether monocyte-derived DCs can be used to evoke or enhance tumor-specific immune responses [7‐9]. Although immune response was often observed, these did not always translate into clinical meaningful responses in patients with prostate cancer. Presumably, the presence of high levels of immune suppressive cells in the tumor microenvironment, such as myeloid-derived suppressor cells (MDSCs), prevented full activation of the induced T cells. To overcome these potential suppressive factors co-administration of docetaxel chemotherapy with DCs was investigated in mCRPC, since taxanes were shown to decrease MDSC levels in murine models [10]. This randomized phase II trial compared docetaxel in combination with monocyte-derived DCs to monotherapy docetaxel. Although no difference was seen in progression free survival (PFS) and disease specific survival (DSS), a significant decline of MDSCs was observed in patients who received docetaxel and DCs, which appeared to be an independent prognostic factor for DSS [11]. A phase III trial (NCT02111577) is ongoing investigating overall survival (OS) in docetaxel with DCs versus monotherapy docetaxel. The disappointing results in earlier trials could possibly also be explained by the use of monocyte-derived DCs instead of natural DCs. The immunologic efficacy and clinical response of natural DCs tested in a Dutch randomized phase IIa trial (NCT02692976) in first-line patients with mCRPC will be presented this year.

Prostatic acid phosphatase (PAP) is expressed in the epithelial cells of the prostate, and was discovered in 1936 by Gutman before the introduction of prostate specific antigen (PSA). Its expression is increased in prostate adenocarcinoma, making it an interesting antigen for vaccination strategies. Sipuleucel‑T is a therapeutic vaccine consisting of CD54-positive monocytes directed in vitro into activated antigen-presenting cells by short-term culture with PA2024, a fusion protein of PAP and granulocyte macrophage colony-stimulating factor (GM-CSF) [12]. The clinical effects of sipuleucel‑T were investigated in several phase III studies, and while PFS endpoints were not met, the pivotal phase III registration study powered for OS benefit was a positive study [13‐15]. Proliferation of T cells directed against PAP and PA2024 was more often seen in patients receiving sipuleucel‑T, with a positive relationship between antibody titers and the duration of OS [15]. Based on these studies sipuleucel‑T was FDA-approved for asymptomatic or minimally symptomatic mCRPC in 2010 and EMA-approved in 2013. It was withdrawn by the EMA in 2015 requested by the marketing authorization holder for commercial reasons. Many trials are testing Sipuleucel‑T together with regimens that may boost neo-antigenicity (radium-223, taxanes, PARPi) and immune checkpoint-inhibitors, and are discussed further in the future perspectives section.

PROSTVAC-VF is a DNA-based vaccine that utilizes PSA as the tumor-specific antigen, and contains two live pox virus-based vectors enclosing transgenes for PSA with 3 co-stimulatory molecules. Due to the positive outcomes of OS (8.5 months longer OS after 3 years follow up) in mCRPC patients receiving PROSTVAC-VF in a phase II trial in 2010 [16], which was again confirmed in a revision analysis in 2017 [17], a phase III trial was conducted and recently results were published [18]. A total of 1286 patients were included and randomized to be treated with PROSTVAC-VF + GM-CSF or PROSTVAC-VF + placebo or placebo + placebo. Surprisingly, no OS benefit was seen between these arms. There were also no differences in secondary endpoints (time to event, HLA-A2 status and tumor response results). These unexpected results are possibly explained by an underpowered OS analysis in the phase II trial (in total 122 patients) [17], biased data collection (OS data collection after data lock up and unblinding), a control arm with worse clinical outcomes than expected (16.6 months vs 20.2 months according to Halabi criteria) and the introduction of new therapies since start of the phase III trial [18]. Based on translational data within this phase III trial, the immunological responses do not translate to the clinical benefit that would be expected, with suggestion that expression of inhibitory checkpoints and other factors dampen the clinical benefit.

Treatment of patients with a DNA-vaccine encoding PAP (pTVG-HP) induced a PAP-specific Th1-biased T cell response [19, 20]. However, this didn’t lead to meaningful PSA decline in cohorts with HSPC or non-metastatic CRPC patients. No randomized studies with monotherapy pTVG-HP have yet been conducted in mCRPC patients, however a combination with chemotherapy or immunotherapy will be investigated in currently recruiting trials (Tab. 1).

Whereas PD-L1 expression is predictive for the response to anti-PD1 therapy in many types of cancer [26, 27], its expression did not associate with objective response or clinical benefit in the largest study to date in prostate cancer [28]. Utilizing different PD-L1 antibodies and cut-off values by different research groups, PD-L1 positivity in primary prostate cancer is estimated between 15–60% [29‐31], and in CRPC between 14–32% [32, 33].

In the phase Ib KEYNOTE-028 trial 23 patients with PD-L1 positive mCRPC were treated with pembrolizumab. In 4 patients a partial response (PR) occurred, and 8 patients experienced stable disease (SD). The median objective response was 13.5 months, and median OS was 7.9 months [33].

Bishop et al. showed for the first time that timing of checkpoint inhibition may be crucial in prostate cancer, as levels of PD-L1-positive circulating immune cells were altered following resistance to enzalutamide [34]. This led to a small phase II trial by Graff et al. [35], who included 10 chemotherapy-naive patients with mCRPC following enzalutamide resistance, to continue on enzalutamide with addition of pembrolizumab. The remarkable objective responses (60% non-PD) [35] led to an extended trial and follow-up data that were presented in 2018 showing a biochemical response in 18% and ORR of 25% with a median radiographic progression free survival (rPFS) of 10.8 months and median OS of 22.2 months [36].

Outcomes of the previous studies have led to the larger phase II KEYNOTE-199 study, comprising 5 different cohorts. The results of the first three cohorts were presented in 2018 with a recent presentation of longer follow up data at the ASCO GU in 2019 [28]. Although responses to pembrolizumab were durable in most of the patients with a partial or complete response (CR), only a specific subset in the entire cohort, comprising 13% of unselected patients with mCRPC, seems to take advantage of PD‑1 blockage.

The fourth and fifth cohort contain patients who progressed on enzalutamide and continued enzalutamide with the addition of pembrolizumab, to investigate and corroborate the results from Graff et al. [35] whether the combination yields increased responsiveness [28]. The combination of enzalutamide and pembrolizumab will also be investigated in a currently recruiting phase III trial, and will demonstrate the differences in response when pembrolizumab was directly added at the start of enzalutamide compared to enzalutamide alone (NCT03834493). Additionally, two phase III trials are starting to recruit that are testing the combination of pembrolizumab with docetaxel (NCT03834506) and olaparib (NCT02861573), and are discussed further in this review.

Only one study has investigated monotherapy of PD-L1 inhibition in prostate cancer. This phase I study with avelumab included 18 patients with mCRPC who progressed on previous treatment. No responses were observed [37]. In currently active phase 2/3 trials PD-L1 inhibitors are investigated with or without second generation anti-androgen therapy (ADT), or in combination with poly ADP ribose polymerase (PARP) inhibitors (Tab. 1).

Ipilimumab was the first checkpoint inhibitor showing an OS benefit in metastatic melanoma [38]. It is an inhibitor of the CTLA4-B7 pathway, which negatively regulates priming of tumor antigen specific T‑cells [1] in lymphoid organs. In earlier phase I/II trials responses were observed in mCRPC patients receiving ipilimumab [39, 40]. Despite the expectations based on these trials, ipilimumab did improve PFS, but no improvement was observed for OS in two phase III trials [41, 42]. Here, ipilimumab was compared to placebo in chemo-naive [42] and chemotherapy-treated patients [41]. Based on a post hoc multivariate analysis in the study by Kwon et al. [41] OS benefit of ipilimumab was seen only in the patient group with favorable prognostic characteristics. Particularly the presence of visceral metastases was associated with a lack of response to ipilimumab [41].

Both phase III trials [41, 42] suggest that ipilimumab can induce clinical meaningful antitumor responses, however only present in a small subset of patients. Due to a lack of predictive biomarkers associating with ipilimumab responses, monotherapy has limited future in mCRPC, and is mainly of interest in the neo-adjuvant or oligometastatic setting in combination with radiotherapy.

A retrospective study by Gao et al. demonstrated an upregulation in PD-L1 immune checkpoint expression on both tumor infiltrating T‑cells and tumor cells when treated with ipilimumab [43], suggesting co-administration of PD1/PD-L1 checkpoint inhibitors might be able to generate a more effective immune response. This has led to the phase II Checkmate-650 study of which the first results were recently presented [44]. Of the 78 patients who had a minimum follow up of 6 months the objective response rate (ORR) was 26% vs 10% in chemo-naive and chemotherapy pretreated patients respectively. In both cohorts ORR was higher in patients with PD-L1 expression >1%, DNA repair deficiencies or above median tumor mutational burden (TMB) [44]. The toxicity profile of the combo (grade 3–4 treatment-related AE’s 39 and 51% resp) does not warrant further evaluation in phase III, and additional less toxic regimens are to be studied.

Based on data suggesting that AR-V7 positive prostate cancer harbors more defects in DNA repair genes and has a higher TMB [45] the phase II (STARVE-PC) trial [46] was conducted including 15 patients with AR-V7 positive circulating tumor cells who were treated with the combination of nivolumab and ipilimumab. Preliminary data indicate a PSA response in 13% and 2 out of 8 patients with measurable disease had an objective response. Interestingly, PFS was significantly longer in DNA repair deficient patients but did not (yet) translate to differences in OS [46].

Other new combinations with ipilimumab and nivolumab are mentioned in Tab. 1.