A Modular Trial of Androgen Signaling Inhibitor Combinations Testing a Risk-Adapted Strategy in Patients with Metastatic Castration-Resistant Prostate Cancer

Abstract

Purpose: To determine the efficacy and safety of risk-adapted combinations of androgen signaling inhibitors and inform disease classifiers for metastatic castration-resistant prostate cancers.

Patients and methods: In a modular, randomized phase II trial, 192 men were treated with 8 weeks of abiraterone acetate, prednisone, and apalutamide (AAPA; module 1) and then allocated to modules 2 or 3 based on satisfactory (≥50% PSA decline from baseline and <5 circulating tumor cell/7.5 mL) versus unsatisfactory status. Men in the former were randomly assigned to continue AAPA alone (module 2A) or with ipilimumab (module 2B). Men in the latter group had carboplatin + cabazitaxel added to AAPA (module 3). Optional baseline biopsies were subjected to correlative studies.

Results: Median overall survival (from allocation) was 46.4 [95% confidence interval (CI), 39.2-68.2], 41.4 (95% CI, 33.3-49.9), and 18.7 (95% CI, 14.3-26.3) months in modules 2A (n = 64), 2B (n = 64), and 3 (n = 59), respectively. Toxicities were within expectations. Of 192 eligible patients, 154 (80.2%) underwent pretreatment metastatic biopsies. The aggressive-variant prostate cancer molecular profile (defects in ≥2 of p53, RB1, and PTEN) was associated with unsatisfactory status. Exploratory analyses suggested that secreted phosphoprotein 1-positive and insulin-like growth factor-binding protein 2-positive macrophages, druggable myeloid cell markers, and germline pathogenic mutations were enriched in the unsatisfactory group.

Conclusions: Adding ipilimumab to AAPA did not improve outcomes in men with androgen-responsive metastatic castration-resistant prostate cancer. Despite the addition of carboplatin + cabazitaxel, men in the unsatisfactory group had shortened survivals. Adaptive designs can enrich for biologically and clinically relevant disease subgroups to contribute to the development of marker-informed, risk-adapted therapy strategies in men with prostate cancer.

Figure 1.. Trial schema and profile.

Shown in panel (A) is the trial schema. Men with mCRPC began treatment with abiraterone acetate plus prednisone and apalutamide (AAPA) within 3 days of registration. Each cycle lasted 3 weeks. After 8 weeks of treatment, patients were allocated to Satisfactory or Unsatisfactory modules based on whether they achieved a ≥50% decline in PSA levels from baseline and had ≤5 CTC/7.5 mL. Patients allocated to Module 2 were randomized to the addition of up to 4 cycles of ipilimumab (IPI) and patients allocated to Module 3 had up to 10 cycles of cabazitaxel and carboplatin (CC) added to their treatment. At the completion of IPI or CC, patients continued AAPA on Module 4 until criteria for treatment discontinuation were met. Panel (B) details the trial profile. 192 patients were included in the Satisfactory vs Unsatisfactory group comparisons (shown in the dotted boxes, ∙∙∙∙∙). Overall and failure-free survival analyses were performed for patients that received treatment on Modules 2 and 3 (shown in the black boxes with white font). Of the 64 patients treated on Module 2B, 39 (61%) completed 4 cycles of IPI, 8 (13%) completed 3, 6 (9%) completed 2, and 11 (17%) only 1. Patients treated on Module 3 received a median of 6 cycles of CC (range 4–10). ¹Stopped study treatment for reasons other than PD (1 for toxicity, 1 for financial reasons, 1 for unknown reasons). ²Stopped study treatment for reasons other than PD (1 for a second malignancy requiring other systemic therapy, 1 declined chemotherapy, 1 for unknown reasons).

Figure 2.. Change from baseline PSA after 8 weeks of ApA treatment, and overall survival in Modules 2 and 3.

Baseline changes are shown in panel (A). Red bars correspond to 128 patients with a Satisfactory circulating marker status (i.e., ≥50% decline in PSA from baseline and ≤5 CTC/7.5 mL) after 8 weeks of treatment with ApA. Blue bars correspond to 64 patients with Unsatisfactory marker status. Sixteen patients had ≥50% decline in PSA from baseline but also had either RECIST progression prior to week 8 (n = 4, denoted by *) or >5 CTC/7.5 mL (n = 12). Shown in panel (B) is overall survival in Module 2 with 95% CI, Arms 2A (AAPA alone) and 2B (AAPA plus IPI). Shown in panel (C) overall survival in Module 3 with 95% CI, AAPA plus CC. Patients alive at last contact were censored on that date.

Figure 3.. Allocation of baseline tumor biopsies, immunohistochemistry results, and multivariate logistic regression of selected baseline clinical and tumor characteristics.

(A) Of the 192 participants, 154 (80%) underwent tumor biopsies prior to start of study treatment from lymph nodes (n = 78, 51%), bone (n = 56, 36%), prostate or prostate bed (n = 9, 6%), liver (n = 6, 4%), or other (n = 5, 3%). Of the 154, 99 (64.3%) had FFPE blocks available and 84 (84.8%) of these contained sufficient tumor tissue for IHC analysis. Those allocated to the Satisfactory (S) group are represented by the red boxes and those allocated to the Unsatisfactory (US) group are represented by the blue boxes. Immunohistochemistry results are shown in panel (B). Samples were considered negative for RB1, PTEN, AR_N, and AR_C if Li was ≤10% and positive for p53 if Li was ≥10%. Li was calculated as the number of positively stained epithelial cells (with 2+3+ intensity for p53 and RB1, and 1+2+3+ for the remainder) divided by the total number of epithelial cells, at x200 magnification. Abnormal results are represented in solid boxes and normal results in dotted boxes. Comparisons in A and B used univariate chi-square tests, with Fisher’s exact test as needed for small sample sizes.

Figure 4.. Immune cell populations in baseline metastatic lymph node biopsies within Satisfactory and Unsatisfactory groups.

(A-B) Myeloid/macrophage subpopulations and targets based on deconvolution analyses of bulk RNA sequencing in baseline metastatic lymph node samples from patients in the Satisfactory (n = 27, red) and Unsatisfactory (n = 13, blue) groups. Shown in (A) are grouped scatter plots showing enrichment scores for SPP1+, IGFBP2+, C1Q+, and TREM2+ myeloid/macrophages and in (B) expression of targetable myeloid markers CXCR2, LILRB1, LILRB2, SIRPA, and VISTA. Shown in (C) are correlation plots evaluating co-expression of SPP1+ (top row) and IGFBP2+ (bottom row) with RNA expression of CXCR2, LILRB1, LILRB2, SIRPA, and VISTA. Because these are exploratory analyses, p-values are not shown.