Avelumab in Men With Metastatic Castration-Resistant Prostate Cancer, Enriched for Patients Treated Previously With a Therapeutic Cancer Vaccine

Abstract

Therapeutic cancer vaccines including sipuleucel-T, a prostatic acid phosphatase (PAP) targeted vaccine that improves survival in metastatic castration-resistant prostate cancer (mCRPC), can produce immune responses that translate to clinical benefit. The effects of sequential checkpoint inhibitors after therapeutic vaccine on immune responses are unknown. Avelumab is an anti–programmed death ligand-1 monoclonal antibody evaluated in patients with mCRPC in the JAVELIN solid tumor phase 1 trial expansion cohort, enriched for patients with a previous therapeutic prostate cancer-targeted vaccine. mCRPC patients received intravenous avelumab 10 mg/kg every 2 weeks with imaging every 6 weeks. Peripheral blood T-cell responses to PAP and to PA2024, the peptide containing PAP utilized by the vaccine, were evaluated pre and posttreatment. Eighteen patients enrolled, and previous treatments included abiraterone or enzalutamide in 14 (78%), therapeutic cancer vaccine in 14 (78%), and chemotherapy in 4 (22%). Avelumab had a manageable safety profile. There were no sustained prostate specific antigen decreases. Of 17 patients evaluable for best overall response by RECISTv1.1, 12 had stable disease (SD) and 5 had progressive disease. Seven patients had SD for >24 weeks posttreatment. Fourteen patients had previously received therapeutic cancer vaccines. Eleven (79%) had SD as the best overall response. Of these 14 patients, 9 had previously received sipuleucelT. Analysis of antigen-specific T-cell responses pre and postavelumab treatment did not demonstrate changes in interferon-γ production or proliferation in response to PAP or PA2024. This unplanned analysis does not support the use of sequential therapeutic cancer vaccine therapy followed by programmed death ligand-1 inhibition in mCRPC.

Despite the clinical efficacy in other genitourinary malignancies like kidney and bladder cancer, immune checkpoint inhibitors (CPIs) have not demonstrated clinical benefit in unselected patients with prostate cancer. Two phase 3 studies evaluating CTLA-4 blockade (ipilimumab) did not meet overall survival primary endpoints.^1,2 Phase 1 and 2 trials assessing programmed death-1 (PD-1) or programmed death ligand (PD-L1) blockade reported minimal activity (objective or prostate specific antigen (PSA) response rates of ≤5 to 8.6%).^3–6 A combination of enzalutamide and atezolizumab (anti–PD-L1) did not improve overall survival, compared with enzalutamide alone.⁷ In addition, the phase 3 KEYNOTE-921 trial failed to provide an overall survival or radiographic progression-free-survival benefit with docetaxel plus pembrolizumab versus docetaxel alone in mCRPC patients. Patients with metastatic solid tumors [with microsatellite instability-high/mismatch repair deficient and/or high tumor mutational burden (≥10 mutations per megabase)] have been identified as subgroups, for which immune checkpoint inhibition is beneficial.^8,9 However, these make up a small proportion of prostate cancer patients (~5%).¹⁰ Some reports suggest that certain prostate cancer tumor characteristics (eg, tumoral PD-L1 expression ≥1%) may enrich for response to immune checkpoint inhibition;^11,12 however, these observations have not been validated in larger sample sizes.

Sipuleucel-T has demonstrated an ability to extend survival in patients with prostate cancer providing proof of concept that immunotherapy can be effective in this disease.¹³ Sipuleucel-T is a therapeutic cancer vaccine that generates immune responses that target the tumor antigen prostatic acid phosphatase (PAP) through PA2024, a fusion protein that encodes PAP and granulocyte-macrophage colony-stimulating factor. Most patients treated with sipuleucel-T develop PAP/PA2024-specific T-cell activation and/or PAP/PA2024-specific T-cell proliferation. These immune responses correlate with overall survival (P = 0.003).¹⁴ Neoadjuvant studies have demonstrated that sipuleucel-T is capable of enhancing immune infiltration of the prostate before surgical resection in newly diagnosed patients.¹⁵ Similarly, PROSTVAC is a poxviral-based therapeutic cancer vaccine that generates immune responses primarily against PSA. Although a phase 3 study of PROSTVAC did not improve survival in advanced prostate cancer, the immunotherapy has demonstrated the ability to generate systemic immune responses as well as increase immune cell infiltration of the prostate in newly diagnosed patients treated with surgery.^16,17

It has been hypothesized that the prostate cancer microenvironment is devoid of or contains immune effector cells incapable of anti-tumor activity, thereby resulting in a suboptimal response to immune CPIs. This perspective raises the possibility that by enhancing immune infiltration of the tumor microenvironment with vaccine-based immune strategies the treatment could prime the immune system to respond to immune CPIs at a high rate. With this in mind, this study enriched enrollment for patients with previous vaccine therapy in the mCRPC expansion cohort of the JAVELIN solid tumor phase 1 trial. This study evaluated avelumab, a fully human, IgG1 monoclonal antibody with FDA-approved indications for metastatic Merkel cell carcinoma,¹⁸ metastatic/advanced urothelial carcinoma,^19,20 and first-line treatment of renal cell carcinoma in combination with axitinib.²¹

PATIENTS AND METHODS

Eligibility

Patients were treated as part of an expansion cohort of the first human, phase 1 trial (JAVELIN solid tumor; EMR100070-001; NCT01772004). All patients signed Institutional Review Board-approved informed consents and the protocol was registered with clinicaltrials.gov (NCT01772004). The primary endpoint was the best overall response by RECISTv1.1. Patients were 18 years or older of age with an ECOG 0–1 and a life expectancy >3 months. Patients had adequate organ function and had histologically confirmed mCRPC with stable castrate levels of testosterone ≤50 ng/dL on androgen deprivation therapy (unless status postorchiectomy). Upon study entry, objective evidence of disease (nonmeasurable or measurable lesions by RECISTv1.1) and progressive disease (PD) on previous treatment were required. Patients were asymptomatic or minimally symptomatic (defined as not requiring consistent treatment with opiates during the month before enrollment for treatment of prostate cancer-related pain). Additional antiandrogen (eg, enzalutamide) therapy was allowed to be continued after enrollment. Key exclusion criteria were prior treatment with a monoclonal antibody therapy targeting T-cell regulatory molecules, anticancer treatment within 28 days of trial enrollment (excluding palliative radiotherapy and luteinizing hormone-releasing hormone agonists or antagonists), major surgery within 28 days of enrollment, and active or history of autoimmune disorders. Patients with diabetes type I, vitiligo, psoriasis, hypo or hyperthyroid disease not requiring immunosuppressive treatment were not excluded. Patients were not evaluated for PD-L1 status before enrollment as this was not a criterion of the JAVELIN Solid Tumor study.

Trial Design

Trial participants received avelumab 5 mg/kg intravenously infused over 1 hour every 2 weeks. Restaging imaging was performed every 6 weeks. Subjects received avelumab until disease progression or unacceptable toxicity. Prostate cancer working group 2 criteria²² were used to determine the PD.

Interferon-γ Enzyme-Linked Immunosorbent Spot

Interferon-γ (IFN-γ) enzyme-linked immunosorbent spot (ELISPOT) assay was performed using Millipore MultiScreen 96-Well Assay Plates for ELISPOT Assays, 0.45 µm pore size (Millipore Sigma) and Human IFN-γ ELISpot BASIC kit (Mabtech). After plates were coated with IFN-γ capture antibody overnight, they were plated with 300,000 PBMCs per well. For each tested antigen and control, samples were plated in triplicate and the median response value was used for analysis. IFN-γ ELISPOT data were adjusted for background by subtracting IFN-γ spots for PBMCs in no antigen control.

Proliferation

Proliferation responses to antigens were assayed in preculture PBMC using standard tritiated thymidine (3H-Thymidine; PerkinElmer) incorporation assay. The degree of proliferation was expressed as a stimulation index. This was defined as the value of 3H-Thymidine incorporation in the presence of antigens divided by the value of 3H-Thymidine incorporation in the presence of no antigen (media alone).

RESULTS

Patient Demographics

Between December 2013 and December 2015, 18 patients enrolled. Baseline demographic characteristics are summarized in Table 1. Four patients previously received chemotherapy and 14 patients previously received antiandrogen therapy including enzalutamide (n = 7) or abiraterone acetate (n = 7). Five patients continued enzalutamide while in the study. Most patients (78%) had previously received a therapeutic, prostate cancer-specific vaccine, either sipuleucel-T (n = 9) or PROSTVAC (n = 5).

TABLE 1 - Baseline Characteristics

	Number	Median (range)
Total patients	18	—
Age (y)	—	67 (47–85)
Baseline sites of disease	—	—
Bone only	9	—
Node only	2	—
Visceral disease only	1	—
Multiple sites	6	—
Gleason score	—	8 (7–10)
7	7	—
8	3	—
9	7	—
10	1	—
Prior therapeutic cancer vaccine	14	—
Sipuleucel-T	9	—
PROSTVAC	5	—
Median time since vaccine (mo)	—	6 (1–42)
Prior chemotherapy	4	—
Prior abiraterone acetate	7	—
Prior enzalutamide	7	—
Continued enzalutamide on tria1	5	—
Pretreatment PSA	—	16.9 (1.1–1274)
Hemoglobin (g/L)	—	12.75 (9.7–14.1)
Alkaline phosphatase (U/L)	—	93.5 (29–262)
Lactate dehydrogenase (U/L)	—	200 (142–381)

Table represents baseline characteristic/parameter (left column), number of patients with a given parameter (middle column), median and range for numerical parameters (right column).

Treatment-Related Adverse Events

Overall, avelumab demonstrated a tolerable safety profile. Grade≤2 treatment-related adverse events (TRAEs) occurred in 15 patients. The most common TRAEs were fatigue (28%) and infusion reaction (22%). The only grade 3 TRAE was asymptomatic amylase and/or lipase elevation observed in 2 patients. Otherwise, the grade 2 adverse events experienced in multiple patients included hypothyroidism (2 patients), elevated liver enzymes (2 patients), and infusion reactions (4 patients). There were no grade 4 or 5 TRAEs (Table 2).

TABLE 2 - Treatment-Related Adverse Events

Patients (n = 18)
	Number (%)
	Grade 1	Grade 2	Grade 3	Grade 4
Fatigue	5 (28)	0	0	0
Joint pain	1 (5)	0	0	0
Infusion reaction	0	4 (22)	0	0
Rash	2 (11)	0	0	0
Hypothyroidism	1 (5)	3 (17)	0	0
Elevated liver enzymes	0	2 (11)	0	0
Elevated amylase and/or lipase	0	1 (5)	2 (11)	0

Table represents the number of individual adverse events and the percent of patients experiencing that event in parentheses. There were no grade 5 events.

Clinical Response

Out of 17 evaluable patients, 12 had stable disease (SD) (67%) and 5 (28%) had PD. Seven patients had SD for >24 weeks after treatment. One patient was taken off trial after he developed a grade 2 aspartate transaminase and alanine aminotransferase elevation and was not evaluable for response. The percent changes in PSA from the pretreatment baseline for each patient are depicted in Figure 1. There were no sustained PSA decreases. One patient experienced a decrease in PSA but concurrently developed signs of clinical progression and was found to have disease progression on imaging.

FIGURE 1: Percentage change in PSA from baseline. Spider plot of PSA at baseline and subsequent change on treatment with avelumab. Y-axis represents the percentage change in PSA compared with baseline (% change = 0 at month = 0). X-axis represents time in weeks from the initiation of avelumab, up to 30 weeks. Each line represents the PSA curve for 1 patient post-avelumab. Dashed lines indicate a prior therapeutic vaccine; solid lines indicate no prior therapeutic vaccine. Closed circles indicate the BOR of SD (gray: SD <24 wk and black: SD ≥24 wk); open circles indicate BOR of PD. One patient experienced a decrease in PSA but concurrently developed clinical and radiographic progression. One patient not evaluable for response came off treatment after experiencing an avelumab infusion-related reaction and the PSA curve is not included here. BOR indicates best overall response; PD, progressive disease; SD, stable disease.

Among the 14 patients who had received therapeutic cancer vaccines, 11 (79%) had SD as the best overall response and 3 (21%) had PD. Of the 3 evaluable patients who had not previously received a therapeutic vaccine, 1 (33.3%) had SD and 2 (66.7%) had PD (Table 3).

TABLE 3 - BOR and Previous Therapeutic Vaccine

Prior therapeutic prostate cancer-targeted vaccine	Number	Number evaluable for response	Number with BOR SD; n (%)
Sipuleucel-T	9	8	8 (100)
PROSTVAC	5	5	3 (60)
No prior vaccine	4	4	2 (50)

Table depicts the number of patients evaluable for best overall response who previously received tumor-targeted vaccine.

BOR indicates best overall response; SD, stable disease.

All 5 patients who continued enzalutamide during avelumab treatment had previously received a therapeutic vaccine. Three (60%) had SD for >24 weeks; 2 received sipuleucel-T and 1 received PROSTVAC previously. One patient had SD for 13 weeks (prior sipuleucel-T) and 1 had PD (prior PROSTVAC) at the time of the first restaging scans.

Cellular Responses to Prostatic Acid Phosphatase and PA2024

We conducted an exploratory analysis to investigate the effect of anti-PD-L1 therapy with avelumab on T-cell responses to the target antigen of sipuleucel-T in patients who previously received sipuleucel-T. PBMC was analyzed for antigen-specific proliferation and IFN-γ production. PBMC samples (when available) were analyzed for responses to PAP and PA2024 at baseline (before receiving avelumab), day 15, day 43, and day 85. Changes in IFN-γ production and proliferation of tritiated thymidine are plotted in Figure 2. In general, patients who received prior sipuleucel-T had elevated immune responses against PAP and PA2024 relative to patients who were not treated with sipuleucel-T. Those patients, however, did not exhibit an increase in IFN-γ production or proliferation in response to PA2024 or PAP after initiating avelumab treatment. One patient’s proliferative response to PAP trended up (Fig. 2). However, this was not associated with the clinical response; PSA increased, and this participant had PD at 7 weeks on treatment after initially having SD at first restaging.

FIGURE 2: Cellular responses to PAP and PA2024 decrease. Line plots depict interferon-γ ELISPOT assay and proliferation responses by tritiated thymidine to designated antigens (PAP or PA2024) in PBMC over time. Y-axis represents mean spots per 3×10⁵ PBMC (ELISPOT) or stimulation index (proliferation assay). X-axis represents time in days from the initiation of avelumab. In patients who did not receive sipuleucel-T (right column), solid lines indicate prior PROSTVAC; dashed lines indicate no prior PROSTVAC. Closed circles indicate BOR of SD (gray: SD <24 wk, black: SD ≥24 wk); open circles indicate BOR of PD. BOR indicates best overall response; ELISPOT, enzyme-linked immunosorbent spot; PAP, prostatic acid phosphatase; PBMC, peripheral blood mononuclear cell; SD, stable disease.

DISCUSSION

Antigen-specific T-cell responses are an important component of a successful immune response in cancer patients responding to immune CPIs^23,24 and multiple studies demonstrated that sipuleucel-T and other prostate cancer-targeted vaccine platforms can provide this component of antitumor immune activity.^14,25–28 In addition, PD-1/PD-L1 interactions are relevant during antigen presentation to naïve T cells²⁹ and PD-1/PD-L1 blockade during antigen presentation may positively affect the anticancer capabilities of the resulting T cells. Furthermore, studies done with therapeutic cancer vaccines before surgery have demonstrated the ability to enhance T-cell trafficking to the tumor.¹⁵ With this understanding, this trial was enriched with patients who previously received a vaccine-based therapy as either standard of care (sipuleucel-T) or as part of a clinical trial (PROSTVAC).

These previous studies^15,16 supported the hypothesis that vaccines could increase the presence of immune cells in the prostate cancer microenvironment. In a study of 37 evaluable patients who had received sipuleucel-T before prostatectomy, increased T-cell proliferation and IFN-γ were noted in circulating blood.¹⁵ When the prostate was evaluated after prostatectomy, there was a >3-fold increase in CD3⁺, CD4⁺ FOXP3⁻, and CD8⁺T cells noted at the interface of benign tissue and prostate adenocarcinoma compared with pretreatment biopsies (P < 0.001). The majority of these T cells had an expression of PD-1 as well. A second neoadjuvant study involved PROSTVAC before surgery.¹⁶ In 26 patients there was a significant increase in T-cell infiltration when using multiplex immunofluorescence compared with baseline biopsies.

In consideration of these reports, we conducted an unplanned, exploratory analysis on available peripheral blood samples to interrogate antigen-specific T-cell responses to PAP and PA2024 pre and post-avelumab, in patients who had received sipuleucel-T. In the context of PD-L1 blockade after the previous vaccine, our analyses showed no evidence of synergy in most patients who had received sipuleucel-T. Furthermore, there were no objective responses or sustained PSA declines associated with disease control in the 18 mCRPC patients (14 of whom had either sipuleucel-T or PROSTVAC) treated with avelumab in this expansion cohort of the JAVELIN phase 1 trial. These results are in line with observations from other trials testing immune checkpoint inhibition in metastatic prostate cancer patients that showed minimal or no activity in unselected patients.

In addition, 5 patients were treated with avelumab and continued enzalutamide after evidence of progression on enzalutamide alone. Three of these patients had SD for >24 weeks (range, 25–34 wk). Although the sample size from this study is small, results from the phase 3 IMbassador 250 trial, which showed no survival benefit with atezolizumab plus enzalutamide, suggest that enzalutamide plus PD-L1 blockade does not provide meaningful clinical benefit for mCRPC patients.⁷

Although no responses were seen in the 14/18 nivolumab-treated patients who had received the vaccine, higher SD rates were seen in patients who received the tumor-targeted vaccine (79%) compared with patients who did not (33.3%) (Table 3). However, exploratory cellular immunity analyses (Fig. 2) do not demonstrate an increase in antigen-specific T-cell responses to PAP or PA2024. This suggests that the addition of avelumab did not enhance antigen-specific T-cell responses in peripheral blood after sipuleucel-T. It is possible that these vaccine-derived cells had already migrated to the tumor; however, tumor tissue is not available for interrogation of this question. It is also possible that other components of the immune system not evaluated in this study were activated by the subsequent therapy of avelumab but did not have a measurable antitumor effect. An additional variable and important limitation of this unplanned analysis is the time and dosing of avelumab to the last treatment with sipuleucel-T, which varied among patients.

A clinical study by McNeel et al²⁷ suggests that a sequential tumor-targeted vaccine followed sequentially by immune checkpoint inhibition may produce dysfunctional immune activity. In that trial, mCRPC patients were treated with a DNA vaccine encoding PAP plus pembrolizumab. Pembrolizumab was given concurrently with a vaccine for 12 weeks or for 12 weeks duration in the 12 weeks after vaccine treatment. PSA responses were observed in 8/13 (62%) and 1/12 (8%) patients treated concurrently and sequentially, respectively (P = 0.01).²⁷ Although they did not qualify as objective responses, 4/5 patients with measurable disease by RECISTv1.1 treated with vaccine plus pembrolizumab concurrently had measurable decreases in tumor volume at 12 weeks. This is in comparison to 1/3 of patients with measurable disease who had disease reduction with treatment in the sequential group. Although patients in both groups developed circulating PAP-specific T cells after vaccination, antitumor activity and CD8 T-cell infiltration of metastases were observed only in tumor samples from patients who received concurrent treatment. The authors proposed that vaccine-derived T cells present in the absence of PD-1 blockade quickly become dysfunctional. Therefore, at the time of postvaccine pembrolizumab treatment, this dysfunction is not reversible and blockade of PD-1/PD-L1 interaction is irrelevant. This theory offers a potential explanation for the absence of objective responses and scant evidence of enhanced T-cell activity to PAP and PA2024 in patients receiving avelumab who had received therapeutic vaccines. It may also explain the minimal immune activity seen when sequentially giving sipuleucel-T followed by atezolizumab.³⁰

This analysis of patients sequentially treated with vaccine and then PD-L1 blockade has many limitations that must be taken into consideration. First and foremost, this was not a preplanned analysis of the sequential use of vaccine and avelumab. The vaccine treatment was not an eligibility requirement for the protocol; furthermore, the timing of the vaccine before avelumab was not uniform (median = 6 mo). Given that the effects of vaccines can be delayed relative to pharmacologic interventions, this may have fewer implications but it is noteworthy nonetheless.^31,32 Even when vaccines were used, both sipuleucel-T and PROSTVAC were considered equivalent in their ability to work sequentially with avelumab despite different levels of proven efficacy and potential mechanisms of immune activation. As mentioned, the small sample size and lack of follow-up to include survival also are limitations of this analysis. A small minority of patients were cotreated with enzalutamide, which has been shown to perhaps increase systemic steroids, thereby affecting the potential for immune responses as well.³³ Despite these shortcomings, the lack of any evidence of confirmed radiographic or PSA responses is noteworthy and contributes to the body of evidence that does not support tumor-targeted vaccine sequentially followed by immune checkpoint blockade.^27,30

CONCLUSIONS

In conclusion, we report results from 18 mCRPC patients treated with avelumab from 2013 to 2015. As expected, considering reports published since, PD-L1 blockade with avelumab did not produce objective responses or confirmed and sustained PSA responses in unselected mCRPC patients. This trial enrolled 14 patients who received a prior tumor-targeted vaccine and these patients had higher rates of SD compared with those who did not receive a prior vaccine. However, peripheral blood T-cell responses to targeted antigens were not increased in patients who previously received sipuleucel-T. Data from this small sample suggest that although tumor-targeted vaccine and immune checkpoint inhibition may each be components of immune-mediated antitumor activity, they are not sufficient when a vaccine is given as monotherapy before immune checkpoint blockade monotherapy. This should be a consideration for future research studies, regardless of tumor type. Regimens that optimize timing and utilize other immune-enhancing treatments (eg, cytokines) may be required to produce clinically meaningful antitumor responses in mCRPC patients.