A multicenter, open‑label phase II clinical trial (NCT01804465) was undertaken that enrolled and followed patients from April 2014 to November 2020. Eligible patients had asymptomatic or minimally symptomatic mCRPC, defined as progressive prostate cancer by PCWG2 criteria in the face of castrate levels of testosterone.8 Patients with liver metastases were excluded, as were patients with ECOG performance status of 3 or worse. Adequate end organ function (liver, kidney, hematologic) was required. Androgen deprivation was continued in all patients, and prior chemotherapy for mCRPC was an exclusion criterion. All patients received sipuleucel-T administered in standard fashion (intravenous infusion once every 2 weeks for a total of 3 doses). Patients were subsequently randomized to receive their first dose of ipilimumab either immediately (within minutes) following their last sipuleucel-T infusion (immediate arm), or 3 weeks after their last sipuleucel-T infusion (delayed arm) (Figure S1). After initiating ipilimumab therapy, all patients received an additional 3 doses of ipilimumab, 3mg/kg IV every 3weeks, for a total of 4 ipilimumab doses given 3 weeks apart. Patients remained enrolled on the clinical trial until disease progression. The study allowed for patients who experienced an initial response,either by radiographic assessment or by at least a 30% reduction in pre-treatment PSA levels in blood, and then subsequently progressed could be reinduced with another 4 cycles of ipilimumab every 3 weeks. The primary endpoint was to determine the impact of timing of ipilimumab treatment on the induction of antibodies to PAP and PA2024 following treatment. Secondary endpoints included efficacy (as measured by the proportion of patients experiencing a >30% decline in their serum PSA level, duration of PSA decline, radiologic progression-free survival (rPFS) and overall survival, safety as measured by the severity and distribution of adverse events, andother immunomodulatory effects of the treatment. Patients were accrued at The University of California San Francisco (UCSF) and The University of Texas MD Anderson Cancer Center (MDACC). Each patient provided signed informed consent; and relevant institutional review boards approved the study protocols, including the collection of biospecimens.

Immunoglobulin G (IgG) levels against PAP and PA2024, the PAP-GM-CSF fusion protein used to manufacture sipuleucel-T, were evaluated by Life Technologies Corporation using Luminex xMAP technology, which uses multiplexed antigen-coated spectrally distinguishable fluorescence dyed beads. Serum samples (30 µL) were assessed at 1:200 dilution, and normalized signal intensities were log2-transformed for analysis. Non-specific background signal was measured using bovine serum albumin, glutathione s-transferase (GST) and anti-GST-conjugated beads, and data were normalized based on a linear model to reduce technical variability.9 Antibody titers ≥1:400 were considered to be positive.

Cryopreserved PBMCs were thawed and rested overnight at 37°C for batch analysis. The cells were then plated in triplicate of 3.0×10⁵ cells/well and incubated with recombinant PAP or PA2024 at 25 µg/mL, leucoagglutinin PHA-L at 10 µg/mL (Sigma, Cat # L2769) or without antigen for 48 hours at 37°C in MultiScreen Filter Plates (Millipore, Cat # S2EM004M99). Cells secreting interferon-γ (IFN-γ) were visualized by anti-human-IFN-γ enzyme-linked immunospot assay (ELISpot) (MABTECH, Cat # 3420-2A). Plates were scanned with an automated ELISpot plate reader (CTL-ImmunoSpot Analyzer). Spots were counted using CTL Immunospot V.5.0 analyzer software. Final counts of antigen-specific IFN-γ secreting cells were obtained by subtracting the number of spots counted in no-antigen control wells from test wells. Samples were accepted for inclusion in final analysis if positive control PHA wells had an average >100 spots/well, and negative control (no antigen) wells had <100 spots/well.

For the proliferation assays, the cells were plated in triplicate of 1.0×10⁵ cells/well and incubated with recombinant PAP or PA2024 at 25 µg/mL, leucoagglutinin PHA-L at 3 µg/mL (Sigma, Cat# L2769) or without antigen for 120 hours at 37°C. ³H-thymidine 1 mCi was then added, and the cells were incubated for another 8 hours at 37°C. Cells were harvests and assessed on a Perkin Elmer MicroBeta Trilux.

Available cryopreserved PBMC samples were thawed, barcoded, and stained with heavy metal-labeled antibodies for analysis by mass cytometry by time-of-flight (CyTOF). These samples were then washed and acquired on a mass cytometer (Helios, Fluidigm). Contour plot data were generated for CD3+ T cells with Cytobank.10 We used flowCore V.2.0.1 (https://rdrr.io/bioc/flowCore/) to load the data into R V.4.0.2, uwot (https://arxiv.org/abs/1802.03426) to make the UMAP projection, Rphenograph11 for clustering, and ggplot2 (https://ggplot2.tidyverse.org) for plotting. The calculation of both the UMAP projection and the clustering was based on the relative staining levels of CD3, CD4, CD8a, CD11b, CD11c, CD14, CD16, CD19, CD25, CD31, CD33, CD45RA, CD56, CD66, CD117, CD123, CD127, CD235ab/CD61, BDCA3, CCR7, FceRIa, FoxP3, γδTCR, HLA-DR, T-bet, TCR Va24-Ja18, and VISTA. Rphenograph clustering with a k value of 200 yielded eight clusters, which were annotated based on the relative staining levels of CD4, CD8, FoxP3, CD25, CD127, CCR7, CD45RA, T-bet, and HLA-DR. For clarity of visualization, the expression values of Ki-67 were capped at the 99th quantile. Cytobank was used to perform manual gating and to create landmark nodes for statistical scaffold analysis.12 Live T cells (Singlets, Intercalator+, Cisplatin-, CD45+, CD61-, CD235ab-, CD19-, CD3+) were extracted from the fcs files and divided into 30 unsupervised clusters using statistical scaffold. Clusters were assigned vectors associated with the average median value of markers and edges, which are defined as similarity between vectors to produce graphs, which show the relationships between different clusters. Cluster frequencies and Boolean expression for functional markers for each cluster were passed through the Significance Across Microarrays algorithm and results were formulated into the scaffold maps for visualization (github.com/nolanlab/scaffold). For scaffold heatmaps, fold change, significance, cell count, and nearest landmark node, data were extracted from the outputs of the scaffold analysis using a custom script. The heatmap was created in R using pheatmap (https://CRAN.R-project.org/package=pheatmap), with log₂ fold change capped at 2.

This is a multicenter non-comparative randomized phase II trial with the primary objective of immunologic efficacy in addition to safety. Immune response was defined as an IgG antibody response to either PAP or PA2024 by study week 20. The IgG antibody response was determined by ELISA with a positive response defined as titer ≥1:400. To achieve a 40% immune response rate indicating immune activity versus a null hypothesis of 15% indicating lack of immune activity assuming a two-sided type I error of 5% and 81% power would require 27 total patients per arm, based on the exact binomial test. Patients were accrued at UCSF and MDACC with the randomization to ipilimumab timing stratified by institution.

Baseline characteristics were summarized as frequencies and percentages or medians and ranges. Comparison of the categorical and continuous variables between two arms was performed by χ² test and Wilcoxon rank-sum test, respectively. rPFS was analyzed using the Kaplan-Meier method and compared between arms using the unstratified log-rank test and Cox proportional hazards regression model. Time to radiographic progression was defined as the time from randomization to the date of documented radiographic progression or last available follow-up date. OS was defined as the time from randomization to the date of death or last available follow-up date. Changes in PSA and immune response parameters within each arm were evaluated using a non-parametric Wilcoxon signed-rank test. A clinical response was defined as at least 30% reduction in serum PSA level at any time during treatment compared with pretreatment value for that patient. Analogously, changes in PSA and immune response parameters between arms were evaluated using a non-parametric Wilcoxon rank-sum test. All reported p values are two-sided, and p<0.05 was used to define statistical significance.

Fifty patients were randomized, with 26 allocated to the delayed ipilimumab arm and 24 to the immediate ipilimumab arm as in the study design (online supplemental figure 1). Patient disposition is summarized in online supplemental figure 2. The baseline clinical characteristics of patients were similar between the arms (table 1). Overall, 5 of 50 patients (10%) had >50% decline in their serum PSA level, and 1 additional patient had a PSA decline >30% but <50%. The duration of PSA decline of >50% to PSA progression in the five responding patients was 55, 84, 140, 435 and 689 days (figure 1A). These responses were distributed between the treatment arms (figure 1B). The rPFS for the entire group was 5.72 months (95% CI 3.95 to 6.58) (figure 1C), and rPFS was similar between the two arms (figure 1D). The median OS was 31.9 months (95% CI 27.2 to 43.7) (figure 1E), with no significant difference between the arms (figure 1F). In univariate analyses, baseline PSA, baseline hemoglobin and prior radiation therapy (RT) were associated with rPFS (online supplemental table 1). Multivariate analyses did not reveal any significant findings.

The treatment was well tolerated, and immune-related adverse events (irAEs) were consistent with prior reports of irAEs with ipilimumab. There was a single grade 4 event consisting of colitis with colonic perforation requiring surgery and a total of nine grade 3 events in seven patients (table 2). There was no difference in frequency or distribution of irAEs between the two treatment arms. Patients with an irAE were more likely to have a PSA response (any grade, p=0.001, grade 3/4, p=0.037).

Antigen-specific IgG antibody responses and ELISpot responses to PA2024 were associated with improved survival with sipuleucel-T.1 13 In this trial, the primary endpoint was to determine the proportion of patients in each study arm who achieved an antibody titer of ≥1:400 to PA2024 and/or PAP following treatment. Overall, the majority of the patients (78.6%) had induced antibodies post-treatment above this titer. There was no significant difference between the two arms (71.4% for the immediate arm and 81% for the delayed arm, p=1). T cell immune responses were evaluated by assessing proliferative responses of PBMC to antigen. Therapy with the combination of sipuleucel-T and ipilimumab induced significant proliferation to PA2024 and PAP, which were detectable at multiple timepoints (figure 2A,B), with no significant difference between arms (online supplemental figure 3). Immune responses assessed by IFN-γ ELISpot also identified the induction of T cell responses to PA2024 and PAP (figure 2C,D), although these responses were less durable than the proliferative responses. There were again no differences between arms (online supplemental figure 3). RT prior to immunotherapy did not significantly affect any of the antigen-specific immune responses either before or at any timepoint after immunotherapy treatment.

High-dimensional mass cytometry (CyTOF) was used to more broadly determine the effects of immunotherapy in the circulating T cell compartments (figure 3A, left panel). Consistent with our prior observations,14 we found that ipilimumab treatment induced Ki-67 expression in both CD4 and CD8 T cells (figure 3A, right panel). This induction was most pronounced at timepoint (TP) 3 with immediate arm and at TP4 with the delayed arm (figure 3A, right panel). To assess the changes in the activation state of the different T cell subtypes, we applied statistical scaffold analysis for the percentage of T cells that were positive for various markers that characterize the functional state of the T cells. When compared with the pretreatment baseline (TP1), T cells across various subtypes showed Ki-67 induction at both TP3 and TP4 in the immediate arm, but only at TP4 in the delayed arm (figure 3B). In the immediate arm, intense proliferation was induced across all T cell subtype clusters, except in CD4-naïve cells, while in the delayed arm, proliferation was observed primarily in the memory CD4 and CD8 T cell compartments at TP4. Consistent with previous studies demonstrating increased expression of inducible costimulator (ICOS) on circulating T cells as a pharmacodynamic biomarker of anti-CTLA-4 therapy,15 16 induction of ICOS was also seen on T cells in our current study, primarily on CD4 and CD8 memory T cell subtype clusters (figure 3C). Similar to Ki-67, ICOS induction was observed at both TP3 and TP4 in the immediate arm but was more pronounced at TP4 in the delayed arm (figure 3C). These ICOS findings were confirmed via manual cellular gating using Cytobank (online supplemental figure 4).