Multicenter Phase 1 Trial of a DNA Vaccine Encoding the Androgen Receptor Ligand Binding Domain (pTVG-AR, MVI-118) in Patients with Metastatic Prostate Cancer

Abstract

Background:

Preclinical studies demonstrated that a DNA vaccine (pTVG-AR, MVI-118) encoding the androgen receptor ligand-binding domain (AR LBD) augmented antigen-specific CD8+ T cells, delayed prostate cancer progression and emergence of castration-resistant disease, and prolonged survival of tumor-bearing mice. This vaccine was evaluated in a multicenter phase 1 trial.

Patients and Methods:

Patients with metastatic castration sensitive prostate cancer (mCSPC) who had recently begun androgen deprivation therapy were randomly assigned to receive pTVG-AR on one of two treatment schedules over one year, and with or without GM-CSF as a vaccine adjuvant. Patients were followed for 18 months. Primary objectives were safety and immune response. Secondary objectives included median time to PSA progression, and 18-month PSA-progression-free survival (PPFS).

Results:

Forty patients were enrolled at three centers. Twenty-seven patients completed treatment and 18 months of follow-up. Eleven patients (28%) had a PSA progression event before the 18-month timepoint. No grade 3 or 4 adverse events were observed. Of 30 patients with samples available for immune analysis, 14 (47%) developed Th1-type immunity to the AR LBD, as determined by IFNγ and/or granzyme B ELISPOT. Persistent IFNγ immune responses were observed irrespective of GM-CSF adjuvant. Patients who developed T-cell immunity had a significantly prolonged PPFS compared to patients without immunity (HR=0.01, 95% CI: 0.0–0.21, p=0.003).

Conclusion:

pTVG-AR was safe and immunologically active in patients with mCSPC. Association between immunity and PPFS suggests that treatment may delay the time to castration resistance, consistent with preclinical findings, and will be prospectively evaluated in future trials.

INTRODUCTION:

Prostate cancer remains a leading cause of cancer-related death worldwide (1). While nearly 2/3 of patients with organ-confined disease will be cured with definitive surgery or radiation therapy, approximately one third will have recurrence. Prostate cancer, once metastatic, is not curable, and is generally initially treated with androgen ablation therapy. Median survival for patients with metastatic prostate cancer is approximately 5 years, however recent trials have demonstrated that the addition of docetaxel or androgen-targeted agents, such as abiraterone, enzalutamide, or apalutamide, can prolong survival in patients with metastatic prostate cancer beginning androgen deprivation therapy (2–5). Metastatic prostate cancer that develops resistance to surgical or medical castration represents the lethal form of the disease. While several agents have been shown to prolong survival of men with castration-resistant prostate cancer (CRPC), the median survival benefit of each drug is only on the order of a few months (6–10). Consequently, there is an urgent need to identify treatments able to prevent the establishment of metastatic disease and/or to prolong the period of response to androgen deprivation prior to the development of lethal CRPC.

Cancer vaccines aim to eliminate or slow tumor growth by expanding and activating tumor-specific T-cells. Sipuleucel-T, an antigen-presenting cell cancer vaccine, was approved by FDA as a treatment for metastatic CRPC following trials showing prolonged survival, providing proof of concept for the use of cancer vaccines in treating prostate cancer (11). In order to build on this approach, we have sought to identify other tumor-associated antigenic targets and optimize vaccine methods. In preclinical studies, we have focused on identifying immunologically recognizable proteins, shared among multiple patients, that might serve as vaccine antigens, and on cost-effective vaccine methods that can elicit immunity to these antigens. We hypothesized that the androgen receptor (AR) might be an ideal target antigen, as it is the core oncogenic driver of prostate cancer, and has been the primary treatment target of hormone sensitive disease for over 50 years (12). Moreover, we reasoned that T cells specific for the AR, augmented by vaccination, might better target castration-resistant tumors that overexpress AR as a mechanism of resistance to androgen deprivation. In support of our hypothesis, we have previously reported that some patients with prostate cancer, compared to men without prostate cancer, have antibody and CD8+ T cell immunity to the AR, suggesting that immune responses to the AR can arise with the development of prostate cancer (13, 14).

In preclinical models, we have evaluated a DNA vaccine encoding the ligand-binding domain (LBD) of the AR (pTVG-AR, MVI-118), given that the sequence of this domain is identical among multiple species, including mice, rats and humans. We found that vaccination of male, immunocompetent mice could elicit AR-specific CD8+ T cells with cytolytic function, which delayed the establishment of tumors in transgenic mice, and prolonged the survival of prostate tumor-bearing mice or rats (15). Treatment was not associated with adverse effects on normal tissues (16). Finally, we found that androgen deprivation of prostate tumor cells led to increased expression of the AR, making tumor cells more susceptible to lysis by AR-specific CD8+ T cells, and that AR-targeted vaccination combined with androgen deprivation led to a delay in the development of castration-resistant tumors in mice (17). Given these findings, we hypothesized that patients with newly metastatic prostate cancer beginning androgen deprivation would be the ideal population in which to evaluate this vaccine, with the ultimate clinical goal of targeting AR overexpression as a mechanism of resistance, and hence prolonging or avoiding the development of castration resistance. We report here the results from a multicenter first-in-human phase 1 clinical trial using this same DNA vaccine in patients with mCSPC.

MATERIALS AND METHODS:

Study Agent and Regulatory Information:

pTVG-AR (MVI-118) is a plasmid DNA encoding the ligand-binding domain of the human androgen receptor (AR LBD) cDNA downstream of a eukaryotic promoter (15). The study protocol was reviewed and approved by all local and federal entities (FDA, NIH Recombinant DNA Advisory Committee), as well as the institutional review board at each participating site. The study was conducted according to the provisions of the Declaration of Helsinki and Good Clinical Practice Guidelines of the International Council on Harmonization. All patients provided written informed consent before participating.

Patient Population:

Male patients with adenocarcinoma of the prostate and evidence of metastatic disease by CT, MRI and/or bone scintigraphy were eligible. Patients were required to have started standard ADT (bilateral orchiectomy or LHRH agonist, and with or without an androgen antagonist) within 6 months of study entry, with evidence of PSA decline on treatment, and have castrate levels of testosterone. Patients were not allowed to change the type of ADT used while on treatment, and were required to continue ADT throughout the study treatment period. Patients who were deemed candidates for docetaxel chemotherapy were not eligible, and this study was conducted before other agents (e.g. abiraterone or enzalutamide) were approved for use in this disease stage. Patients were further required to have an ECOG performance score ≤ 1, and normal bone marrow, liver and renal function.

Study Design:

This was an open-label, randomized, multi-institutional phase 1 trial designed to evaluate two different schedules of pTVG-AR administration with or without GM-CSF used as a vaccine adjuvant. The primary objectives were safety and immune response to pTVG-AR. Secondary objectives were to evaluate the impact of different schedules of pTVG-AR and GM-CSF on immune response, and to determine median time to PSA progression and 18-month PPFS while receiving ADT. The proposed sample size of 40 patients, 10 patients per study arm, was chosen to detect toxicity incidences given that the probability of observing at least one incidence of grade ≥ 3 toxicity in 10 patients was > 95%, assuming a true toxicity rate of 33%. This sample size was also adequate for identifying the arm with the highest immune response rate with a high probability (selection design with >80% power) so that the results of the trial could be used to prioritize a possible preferred schedule of administration for subsequent trials.

Study Procedures:

Patients were randomized to one of four treatment arms (Figure 1A). In arms 1 and 3, patients received six intradermal injections at 14-day intervals, and then quarterly for up to 12 months (10 total vaccinations), with 100 μg pTVG-AR plasmid DNA, and with (Arm 3) or without (Arm 1) 200 μg GM-CSF (Leukine®, sargramostim) admixed with the DNA (Figure 1B). In arms 2 and 4, patients received two intradermal injections 14 days apart with 100 μg pTVG-AR as a treatment cycle that was repeated every 12 weeks over one year (10 total vaccinations), and were again given with (Arm 4) or without (Arm 2) 200 μg GM-CSF admixed with the DNA (Figure 1B). All subjects were immunized with tetanus toxoid prior to receiving study treatment, as a positive immunological control, and immune responses were similarly evaluated to tetanus toxoid and PSA, as a tumor-specific non-targeted antigen. Safety labs were performed monthly for the first 3 months, and then quarterly. Serum PSA was obtained monthly throughout the 18-month study period. Patients underwent leukapheresis of up to 100 mL volume prior to treatment and at one year, and had additional quarterly blood collections, for immunological analyses. All toxicities were graded using NCI CTCAE, version 4. Radiographic imaging was not scheduled per protocol but permitted at any time per physician discretion. Patients were to come off trial at the time of PSA progression (defined as a serum PSA value at least 25% over the nadir, an absolute increase of 2 ng/mL over the nadir, and verified by repeat value at least 3 weeks later (18, 19)), radiographic disease progression, if undue toxicity, or at the discretion of the patient or physician that other therapies were warranted.

Figure 1:

Schema and patient allocation: Shown is the CONSORT diagram (panel A) and treatment schema (panel B).

Immunological Response Evaluation:

Antigen-specific T cells secreting IFNγ or granzyme B were detected by ELISPOT, using methods as previously reported (20, 21). This analysis was performed from 30 patients treated at one clinical site, with analysis performed for each sample in real time, without cryopreservation of samples. All antigens and sera used were from the same lots to control for possible variation over time. Test antigens included phytohemaglutinin (positive control), media alone, tetanus toxoid (EMD Millipore, Billerica, MA), and pools of 15-mer peptides spanning the amino acid sequence of either PSA (as a tumor-specific, non-targeted antigen) or the AR LBD, with adjacent peptides in each pool overlapping by 11 amino acids (Lifetein, LLC, Hillsborough, NJ). A response resulting from immunization was defined as an antigen-specific response detectable more than once post-treatment that was both significant (compared to media only control), at least 3-fold higher than the pre-treatment value, and with a frequency > ten spot-forming units (SFU) per million PBMC. Immune responses over time were evaluated as an area under the curve (AUC), using antigen-specific SFU at each time point following subtraction of the pre-treatment value (GraphPad Prism, version 5).

IgG antibody responses to tetanus, PSA, and the AR LBD were evaluated by Luminex. Beads coated with AR LBD protein (Invitrogen, Carlsbad, CA), PSA protein (Fitzgerald Industries International, Acton, MA), or tetanus toxoid (EMD Millipore, Billerica, MA) were prepared per manufacturer recommendations (xMAP antibody coupling kit and MagPlex microspheres, Luminex, Austin, TX). Serum samples were diluted 1:100 in PBS/1% bovine serum albumin (BSA), and incubated overnight at 4°C with the beads in wells of a 96-well plate. Plates were washed multiple times, IgG detected with biotinylated secondary antibody and avidin-fluorophore, and quantified by Luminex MAGPIX. Results are displayed as change in response from baseline over time.

Clinical Response Evaluation:

Time to treatment failure was evaluated from the start of study treatment and also from the start of ADT prior to study enrollment. Treatment failure was defined as the first evidence of castration-resistance (PSA at least 25% over the nadir, an absolute increase of 2 ng/mL over the nadir, and verified by repeat value at least 3 weeks later), at which point patients came off study. The time to the first rise in PSA at least 0.1 ng/mL over the nadir, verified by repeat value at least 3 weeks later, was also evaluated as a separate measure of early treatment failure, but not requiring study discontinuation. If there was no PSA progression event, then time to treatment failure was censored at the last available PSA assessment.

Statistical Analysis:

Baseline characteristics were summarized by frequencies and percentages or medians and ranges. Toxicity rates were compared between groups using Fisher’s exact test. Time to treatment failure/PSA progression was analyzed using the Kaplan-Meier method and compared between arms using the unstratified log-rank test and univariate Cox proportional hazard model. Analogously, changes in immune response parameters between arms were evaluated using a nonparametric Wilcoxon rank sum test. All reported P-values are two-sided and P<0.05 was used to define statistical significance. Statistical analyses were conducted using SAS software (SAS Institute Inc., Cary NC), version 9.4.

RESULTS:

Patient population and course of study:

Forty patients were enrolled between August 2015 and November 2017 at the University of Wisconsin, Rutgers Cancer Institute of New Jersey, and the University of Washington (Figure 1A, Table 1). The median age of participants was 69 years (range 51–84 years). Twenty-seven subjects (68%) completed treatment without progression at 18 months. Eleven patients (28%) came off study due to rising PSA, two of which did not meet study criteria for PSA or radiographic progression. Two additional patients ended treatment early, one due to a secondary cancer diagnosis, and the other due to experiencing grade 2 heart failure. The heart failure was believed to be at least possibly treatment related.

Table 1:

Demographics. Demographics for all patients enrolled. High volume or low volume metastatic burden is defined as per CHAARTED trial criteria, in which high volume disease included visceral metastases and/or ≥ 4 bone metastases with at least one outside of vertebrae and pelvis.

	Total (n=40)	Arm 1 (n=9)	Arm 2 (n=11)	Arm 3 (n=10)	Arm 4 (n=10)
Age (years)
Median:	69	67	67	71	67
Range:	(51–84)	(51–74)	(53–73)	(68–84)	(57–80)
Race / Ethnicity
Caucasian	38 (95%)	9	11	10	8
American Indian/Alaska Native	1 (2%)	0	0	0	1
Unknown	1 (2%)	0	0	0	1
Prior treatment
Prostatectomy	21 (52%)	3	8	4	6
Radiation therapy
Treatment	9 (22%)	4	0	2	3
Adjuvant / salvage	16 (40%)	3	5	2	6
De novo metastatic disease (no prior treatment)	10 (25%)	2	3	4	1
Gleason Score
< 7	3 (8%)	0	0	1	2
7	13 (33%)	4	3	2	4
8	5 (12%)	2	1	0	2
9	19 (48%)	3	7	7	2
Sites of metastases
Bones	22 (55%)	4	5	7	6
Lymph nodes	16 (40%)	4	6	2	4
Other	2 (5%)	1	0	1	0
High volume	12 (30%)	3	3	4	2
Low volume	28 (70%)	6	8	6	8
Type of androgen deprivation used
Leuprolide	40 (100%)	9	11	10	10
Bicalutamide	19 (48%)	5	5	6	3
PSA (ng/mL) at time of beginning
ADT - Median:	19.0	20.9	12.2	18.3	22.1
(Range):	(0.7–238)	(3.5 – 238)	(0.7–180)	(7.6–182)	(7.7 – 138)

Safety and adverse events:

A total of 79 treatment-related adverse events were observed over the period of study (10 for Arm 1, 11 for Arm 2, 23 for Arm 3, and 35 for Arm 4). No grade 3 or 4 events deemed at least possibly related to study treatment were observed. The most common grade 1 and 2 adverse events were fatigue, chills, arthralgias, and injection site reactions. Injection site reactions were more common in the groups receiving GM-CSF adjuvant (62% versus 16%, p=0.008). Otherwise, adverse events were not significantly different among the treatment arms (Table 2).

Table 2:

Adverse events: Shown are the number of patients experiencing any adverse events ≥ grade 1 that were determined to be at least possibly related to treatment, and with the highest grade reported per patient. No events greater than grade 2 were observed.

	Grade 1		Grade 2
	pTVG-AR (Arms 1 and 2) n=20	pTVG-AR + GM-CSF (Arms 3 and 4) n=20	pTVG-AR (Arms 1 and 2) n=20	pTVG-AR + GM-CSF (Arms 3 and 4) n=20
General / Constitutional
Fatigue	3	4		2
Chills		3
Flu-like symptoms	1
Weight gain			1
Insomnia	1
Injection site reactions	3	13
Cardiac
Heart failure			1
Respiratory
Dyspnea			1
Gastrointestinal
Diarrhea		1
Nausea		1
Musculoskeletal
Arthralgia	1	3
Myalgia		1
Back pain		1
Nervous system
Headache	1
Cognitive disturbance		1
Memory impairment		1
Skin disorders
Rash maculo-papular			1
Pruritis		1
Vascular
Hot flashes	1
Hypertension			2
Hypotension	1
Immune systems
Allergic reaction				1
Laboratory studies
CPK increased		1

Immunological response:

ELISPOT was used to identify AR LBD-specific T cells secreting IFNγ or granzyme B. As shown in Figure 2A, AR LBD-specific IFNγ-secreting T cells, in frequencies at least 3-fold higher than baseline, were detected at least twice in follow-up in 9/30 (30%) patients. AR LBD-specific granzyme B-secreting T cells, in frequency at least 3-fold higher than baseline, were detected at least twice in follow-up in 12/30 (40%) patients (Figure 2B). AR LBD-specific T cells secreting IFNγ or granzyme B were detected in 16/30 (53%) patients.

Figure 2:

T-cell immunological response: PBMC from 30 patients treated at one center were evaluated in real time at the indicated time points for antigen-specific IFNγ (panel A) or granzyme B (panel B) secretion. PBMC were stimulated in vitro with peptides spanning the amino acid sequence of the AR LBD or PSA, or tetanus toxoid, and evaluated by ELISPOT. A positive immune response (red) was defined as an antigen-specific response (statistically higher than the media-only control, p<0.05 by t-test) that was at least 3-fold higher than pre-treatment and with a frequency >1:100,000 cells. Black squares indicate a time point where an assessment was performed, but in which immune response criteria were not met. White squares indicate time points where no data were collected. Patients with at least two “positive” responses post-treatment were defined as immune “responders.”

As also shown in Figure 2, tetanus-specific T cells, secreting granzyme B or IFNγ, were detected in 13/30 (43%) patients. PSA-specific T cells, secreting granzyme B or IFNγ, were detected in 9/30 (30%) patients. Of note, PSA-specific T-cell responses were only detected in patients with immune response to the AR LBD, whereas 4/13 (31%) patients with response to tetanus did not have response to AR LBD, and 2/9 (22%) of patients with response to PSA did not have response to tetanus.

To evaluate differences in immune response due to vaccine schedule and/or the use of GM-CSF adjuvant, the magnitude of response to AR LBD, as detected by IFNγ ELISPOT, was evaluated over time compared with the baseline measurement. As shown in Figure 3A, immune responses to AR LBD were generally of greater magnitude in patients treated in Arms 1 and 3 (same schedule, with or without GM-CSF), and tended to increase over the first several immunizations. The durability of immunity was assessed by evaluating these data over time as an area under the curve. As shown in Figure 3B, immune responses to AR LBD were significantly more durable over time in these same treatment arms. No significant differences were observed in magnitude or durability of immune response in patients who received GM-CSF adjuvant compared with those that did not. IFNγ-secreting immune responses to tetanus toxoid were similar among all treatment arms (Supplemental Figure 1). Immune responses to PSA were generally lower in magnitude, and mostly detectable in patients treated with the schedule of Arms 1 and 3 (Supplemental Figure 2).

Figure 3:

T-cell immunological response over time: Panel A: IFNγ ELISPOT data as shown in Figure 2 were plotted over time, subtracting the pre-treatment value, to evaluate the magnitude and durability of immunity over time. Panel B: Immune response over time was plotted as “area under the curve,” and evaluated with respect to treatment arm. Lines show median and interquartile range. Comparison between arms was made by t test.

IgG serum antibody responses to AR LBD, PSA, and tetanus toxoid were evaluated over time for all 40 patients by Luminex assay. IgG responses to tetanus toxoid increased in the majority of patients. Antibody responses to AR LBD and PSA, however, did not significantly change over time (Supplemental Figure 3).

Clinical response:

Twenty-seven out of 40 (68%) patients were progression-free at 18 months. The median time to first PSA rise was 9.2 months from start of study treatment, and 11.7 months from starting ADT prior to study treatment. As shown in Figure 4A, the time to castration resistance or first PSA rise was not significantly different among study arms grouped with respect to use of GM-CSF adjuvant or not. However, as shown in Figure 4B, the time to first PSA rise was significantly longer (median time: 13.8 versus 4.6 months, HR=0.42, CI 0.19–0.92, p=0.02) in patients treated with the schedule of arms 2 and 4 compared to the other treatment schedule. As shown in Figure 4C, patients who developed IFNγ and/or granzyme B immunity to the AR LBD had a significantly prolonged time to castration resistance (median time: not reached versus 9.2 months, HR 0.01, 95% CI 0–0.021, p=0.003) or first PSA rise (median time: 12.5 versus 4.6 months, HR 0.18, 95% CI 0.05–0.64, p=0.008). There was no significant association observed between immunity to tetanus or PSA and time to castration resistance or time to first PSA rise (data not shown).

Figure 4:

Clinical outcomes: Time to progression (percentage free from castration resistance, right panels, and percentage free from first PSA rise, left panels) were evaluated from study start date in patients treated on arms 1 and 2 (no GM-CSF) versus arms 3 and 4 (with GM-CSF) (panel A), or in patients treated on arms 1 and 3 (schedule 1) versus arms 2 and 4 (schedule 2) (panel B). Time to castration-resistance (left panel) and time to first PSA rise (right panel) were also evaluated with respect to whether patients developed IFNγ and/or granzyme B T-cell immune response to the AR LBD target (panel C).

DISCUSSION:

In preclinical models, we previously demonstrated that androgen deprivation leads to an increase in expression of AR, which in turn can make prostate cancer cells more recognizable by CD8+ T cells specific for the AR (17). Vaccination targeting AR led to an increase in time to progression when combined with androgen deprivation in murine prostate cancer models (17). In this phase I study, we have evaluated the same DNA vaccine encoding the AR LBD in men with mCSPC. We found that: 1) vaccination with pTVG-AR (MVI-118) was safe; 2) vaccination elicited a T-cell response to the AR LBD in the majority of patients; 3) a preferred schedule of immunization was identified that led to greater magnitude and durability of Th1-type immunity; 4) the use of GM-CSF adjuvant did not substantially affect the development of immune response to the target antigen; 5) vaccination elicited immunity to a non-targeted tumor antigen (PSA) consistent with “antigen spread;” and 6) immune response to AR LBD augmented with immunization was associated with a longer time to castration resistance, and early PSA rise preceding the onset of castration resistance.

Vaccination with pTVG-AR was safe and well tolerated. We observed that injection site reactions were more common in patients receiving GM-CSF adjuvant, suggesting these were mostly due to GM-CSF. The absence of significant toxicity is notable, since the AR target antigen is also expressed in normal tissues (e.g. muscle, liver, and skin) (22). No toxicity was observed in these tissues. This was not unexpected as no toxicity was observed in preclinical models (16), and we have previously detected immunity to AR existing in patients with prostate cancer who otherwise had no evidence of autoimmunity (13). Most vaccine trials in prostate cancer to date have focused on tissue-restricted antigens, assuming that toxicity might be greater to antigens shared by normal tissues. Certainly that has not been the experience for other tumor types targeting antigens (e.g. CEA or HER-2/neu) with wider tissue expression profiles (23, 24). Hence, we expect that the increased expression of AR in prostate cancers, and notably after ADT, may favor this as an immunologically targetable antigen.

As described above, we found that immunization elicited T-cell responses to the AR LBD, but not antibody responses. This is consistent with preclinical findings in mice in which we also observed that DNA immunization favored the generation of CD4 and CD8 T cells rather than antibody responses (15). We have similarly found that a different DNA vaccine, encoding the prostate antigen prostatic acid phosphatase (PAP, ACP3), elicited Th1-type T-cell responses but not antibody responses in patients with prostate cancer (25). We assume that the immunity to AR elicited is Th1/Tc1-biased given that we observed IFNγ- and granzyme B-secreting responses. Further analyses are ongoing to identify whether antigen-specific Th2- or Th17-type responses may have been elicited in some individuals, and whether the responses observed were primarily CD4 or CD8 T cells. Of note, while all patients were immunized with tetanus toxoid, we did not detect T-cell responses in all individuals. We believe this is due to our definition of immune response, requiring detection at least 3-fold over baseline, which may be too stringent to detect responses in individuals with high pre-existing immunity. Antibody responses to tetanus, however, were common, and identifiably augmented in the majority of patients.

A previous retrospective review of tumor vaccine trials found that traditional dose-escalation phase I trial designs were not useful in identifying doses for subsequent trials because, as opposed to standard cytotoxic therapeutic approaches, immune response may not be strictly dose-dependent, and few toxicities were identified at higher doses (26). For this trial, we elected to evaluate the treatment schedule, rather than dose, in particular because the vector backbone used for the pTVG-AR plasmid is identical to that previously used in other trials using a different DNA vaccine targeting PAP (pTVG-HP). In those previous trials, we had fixed the dose at 100 μg of plasmid DNA as one that demonstrated biological activity in eliciting T cell immunity to the PAP target antigen and was observed to be safe and well tolerated (20, 21, 25). The trial schedules evaluated in the current trial were based on one previously identified in previous human trials using pTVG-HP using several immunizations at 2-week intervals followed by intermittent boosters (20), or one with a more intermitted schedule evaluated in animal studies using pTVG-AR (15). Using magnitude and durability of immune response to AR as markers of biological activity, we identified the schedule of arms 1 and 3 to be a preferred schedule of immunization. Specifically, we found that several repetitive immunizations appear necessary to elicit a consistent, detectable T-cell response. These findings are consistent with findings in a previous trial in which we evaluated the frequency of immunizations to elicit and maintain Th1-type immunity to the PAP target antigen (20). Optimal immunization schedules remain unknown for human vaccines, and likely differ in the contexts of preventative vaccines for infectious diseases and for treating cancer in which the target antigen persists. Nonetheless, it appears that multiple repetitive immunizations are necessary. While it is believed that breaks in schedule may help establish T-cell memory, episodic booster immunizations may be necessary to maintain immune responses elicited.

Curiously, we did not identify a greater immune response when GM-CSF was used as a vaccine adjuvant. This was somewhat unexpected, as GM-CSF has been used as an adjuvant for many other vaccines. This has included cellular vaccines modified to secrete GM-CSF (27), and the only current FDA-approved cancer vaccine, sipuleucel-T, includes GM-CSF as part of the antigen used for activating antigen-presenting cells ex vivo (28). GM-CSF has also been used as a protein adjuvant, co-delivered with the antigen. In particular, human trials using the hepatitis B vaccine have demonstrated improved immune responses when co-delivered with GM-CSF (29, 30). Notwithstanding, a large trial evaluating GM-CSF as an adjuvant with a peptide vaccine demonstrated inferior immunity when used with GM-CSF (31). While GM-CSF has demonstrated utility as an adjuvant for DNA vaccines in animal models (32), it has not specifically been evaluated in randomized human trials using DNA vaccines. It is conceivable that bacterial plasmid DNA itself has an adjuvant effect, and little is added by the addition of GM-CSF protein. Certainly, preclinical studies in mice and rats with the pTVG-AR vaccine have demonstrated efficacy without the use of an additional adjuvant (15–17). Future clinical trials using the pTVG-AR vaccine will be done without GM-CSF adjuvant.

We found that vaccination was associated with immune responses to PSA, a non-targeted prostate-specific antigen. These findings are consistent with AR-targeted vaccination eliciting antigen spread, notably because immune responses to PSA were only detected in individuals who developed immunity to the AR LBD antigen. All patients were immunized with tetanus toxoid, and several patients developed response to tetanus but not the AR. The absence of response to PSA in these individuals suggests that immune responses to PSA detected were indeed a result of antigen spread. We have previously observed responses to PSA after treatment with pTVG-HP, and others have reported immune responses elicited to PSA as evidence of off-target antigen spread (20, 33). Future studies will further explore the character of T cells elicited to PSA, and the repertoire of tumor-infiltrating T cells following vaccination.

The major limitation of this study was the small sample size, notably given the inclusion of patients with different prior treatments (surgery, radiation therapy, or no prior therapy) and volume of disease. We expect that patients with recurrent prostate cancer after extirpative therapy, compared with patients with de novo metastatic disease, may have a prolonged time to progression. Hence, clinical outcomes must be interpreted cautiously. This may, in fact, explain some differences in progression with respect to the different treatment arms, since more patients in arms 2 and 4 had undergone prior prostatectomy. Notwithstanding, our studies suggest that successful immunization to the AR, as demonstrated by the detection of Th1 immunity to the AR target antigen, was associated with delayed development of castration resistance. These are similar to our previous preclinical findings that immunization with this same vaccine led to prolonged survival of prostate tumor-bearing mice and rats, and prolonged time to castration resistance in mice treated with androgen deprivation (17). In addition, the time to castration resistance was similar for patients without evidence of immunity to the AR LBD to what has been observed in recent large clinical trials in this same population using standard androgen deprivation, and substantially longer in those patients with immunity (3, 5).

Taken together, these findings support the further evaluation of this vaccine in a phase 2 trial in patients with metastatic prostate cancer designed to rigorously evaluate whether treatment can prolong the time to castration resistance. When this trial began, docetaxel was the only agent demonstrated to prolong time to progression and overall survival when used concurrently with standard androgen deprivation for patients with mCSPC (2). Since this trial was conducted, several androgen-targeted therapies, including abiraterone, enzalutamide and apalutamide, have also been approved for use in combination with standard androgen deprivation, based on similar improvement in time to progression and overall survival (3–5). Because overexpression of AR is likely a mechanism of resistance to all of these AR-targeted therapies, we expect that targeting the AR by vaccination should similarly prolong the time to castration resistance with any of these approaches. Hence, phase 2 trials using pTVG-AR will be conducted using standard androgen deprivation with other AR-targeting agents. Of note, a trial evaluating this vaccine in combination with PD-1 blockade is also underway in patients with castration-resistant metastatic prostate cancer (NCT04090528).

STATEMENT OF TRANSLATIONAL RELEVANCE:

We report a first-in-human phase 1 trial using a DNA vaccine targeting the androgen receptor ligand-binding domain (AR LBD) in patients with newly metastatic prostate cancer. No significant adverse events were observed, and T-cell immunity to the AR LBD was augmented with immunization. Immunity was associated with prolonged time to castration resistance. These findings demonstrate, for the first time, that the androgen receptor is not only a pharmacological target, but a viable immunological target, for the treatment of human prostate cancer. Further evaluation of this vaccine in larger clinical trials is warranted.