A phase II trial of recombinant MAGE-A3 protein with immunostimulant AS15 in combination with high-dose Interleukin-2 (HDIL2) induction therapy in metastatic melanoma

The eligibility criteria included age at least 18 years; histologically confirmed, stage III or IV unresectable melanoma; an Eastern Cooperative Oncology (ECOG) performance status of 0 or 1; and adequate organ function. This was a single-institution study with all patients recruited from UT MD Anderson Cancer Center. Patients were required to have available formalin-fixed paraffin-embedded (FFPE) tumor tissue for MAGE-A3 expression screening [16]. The RNA from these samples was extracted to assess the MAGE-A3 expression by quantitative PCR. The patients were considered eligible if at least one tested block was MAGE-A3-positive. Patients were additionally required to have measurable disease, defined by Response Evaluation Criteria in Solid Tumor (RECIST) v1.1 [17] and at least 1 accessible tumor for biopsy obtained within 30 days from starting treatment. Key exclusion criteria included any prior systemic therapy with the exception of interferon in the adjuvant setting, history of brain metastases, autoimmune disease and/or ANA titer > 1:80, Hepatitis B/C, or HIV, and chronic steroid use.

Study schematic is provided as Additional file 1. In the induction phase, 300 μg of recMAGE-A3 protein (MAGE-A3) + adjuvant system 15 immunostimulant (AS-15) (see Additional file 2 for composition) were given via intramuscular injection every 2 weeks for 6 cycles and then every 3 weeks for 6 cycles. HDIL-2 was initiated on the day following MAGE-A3 CI immunization or up to eight cycles on weeks 1, 3, 9, 11, 18, 21, 27 and 30 at 720,000 IU/kg every 8 h for up to 14 doses each cycle. Following completion of the 30-week induction HDIL-2 + MAGE-A3 CI, patients who remained on study were continued on the maintenance MAGE-A3 CI alone, given every 6 weeks for 4 cycles, then every 12 weeks for 4 cycles, and then every 24 weeks for 4 cycles. Treatment was continued until disease progression, intolerable toxicity, patient withdrawal of consent, or investigator decision. Patients who permanently discontinued HDIL-2 before completing 8 cycles because of an adverse event (AE) and no signs of disease progression were allowed to proceed to maintenance phase and resume MAGE-A3 CI after resolution of the AE to grade 0 or 1. Tumor images were obtained at baseline and before every other cycle or every 8 weeks during the induction phase and then 1 week prior to each dose of maintenance MAGE-A3 CI. Assessment of antitumor activity was based on RECIST v1.1 criteria [17]. AEs were collected throughout treatment and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0. Attribution of AEs to study treatment was determined by the treating physician based on their assessment of exposure, time course, likely cause, and consistency with known treatment profile.

The primary endpoints were overall response rate (ORR), with decision on proceeding to 2nd stage of trial based on response rate and safety profile. Tumor responses were assessed every 8 weeks during induction phase and 1 week prior to MAGE-A3 CI immunization during maintenance phase. Secondary endpoints included progression-free survival (PFS), overall survival (OS), duration of response, and biomarker correlatives.

FFPE tissue was required for MAGE-A3 expression screening, either from a fresh tumor biopsy at the time of screening or archival tissue obtained within 1 year. A baseline biopsy for gene expression profiling was performed prior to therapy initiation and optional biopsies were performed at week eight. Frozen specimens were stored in RNAlater and mRNA was extracted by standard methods [16]. Blood samples were collected at baseline and at regular intervals thereafter and peripheral blood mononuclear cells (PBMCs) isolated. MAGE-A3 gene expression with quantitative PCR (qPCR) was performed on FFPE tissue using TaqMan (Applied Biosystems, Foster City, CA) low-density arrays as previously described (see Additional file 2) [18]. Immunohistochemistry (IHC) on FFPE tumor tissue sections was performed using antibodies against PD-1, PD-L1, CD8, Granzyme B, and CD45RO (see Additional file 2). Percentage of positivity for each IHC marker was evaluated by an experienced dermatopathologist (CA Torres-Cabala). RNA from frozen baseline specimens was amplified and assayed on Affymetrix HG-U133.Plus 2.0 (Affymetrix, Santa Clara, CA) microarray gene chips by standard methods to assess gene expression of an 84-gene set previously found to be predictive of benefit in melanoma patients treated with MAGE-A3 CI [16].

Blood samples for immune biomarker studies were collected pre-MAGE-A3 dose on day 1 and at completion of HDIL-2 of each cycle in induction phase and pre-MAGE-A3 dose on day 1 of each cycle in maintenance phase. Flow cytometry was used to detect changes in immune cell populations over time and quantify MAGE-A3 specific CD8+ T cells in HLA-A 0201+ patients using an HLA-A 0201-restricted MAGE-A3 epitope.

The study used a Two-Stage Fleming’s design, [19] with response measured by overall response (OR: complete response [CR] or partial response [PR] as defined above) in 15 patients treated in the 1st stage. The null hypothesis was that the ORR was 5%, matching the historical control rate, and the alternative hypothesis was that this could be improved to 20% with the study regimen. If none had responded, the study would have been terminated for futility. If 4 or more had responded, the study would have been terminated early for efficacy. Neither of these criteria were met, and the study was scheduled to enroll an additional 15 patients in the 2nd stage. The study was terminated for slow patient accrual after only 2 patients were enrolled in 2nd stage. Response duration was measured from the time of response until evidence of disease progression. PFS was measured from the time of treatment initiation to evidence of disease progression or death, and OS was measured from the time of treatment initiation to the time of death or last follow-up. Kaplan–Meier estimates were used to plot survival curves. Logistic regression models and Fisher’s exact test were used to assess associations of response with covariates of interest. P-values of < 0.05 were considered significant. R (v 3.2.2) was used for data analysis.

Forty-four patients were screened for eligibility. Eighteen patients were eligible and enrolled between March 2011 and March 2014. Follow-up was continued until October 2017 when the last patient completed maintenance therapy. One patient withdrew consent prior to initiation of therapy. Seventeen patients were evaluable for safety and sixteen for response because one patient received one cycle of HDIL-2 without ever receiving the MAGE-A3 CI. As shown in Table 1, most patients had Stage IV disease, including 35% with M1c disease.

The adverse events were similar to those commonly seen with HDIL-2 monotherapy, including, hypotension edema (capillary leak and fluid retention), hypoalbuminemia, fatigue, fever, chills, nausea, vomiting, rash and pruritus, myalgia, arthralgia, and elevated liver enzymes (Table 2). One patient discontinued HDIL-2 after 2 cycles secondary to significant AEs (atrial fibrillation, excess fluid retention, and hypotension) but continued with maintenance MAGE-A3 CI. Few toxicities were attributed to the MAGE-A3 CI by the investigators: five patients had pain or swelling at the injection site (all Grade 1), and one patient had acute renal dysfunction with elevated creatinine (Grade 3). The patient with acute renal dysfunction discontinued MAGE-A3 CI after 2 cycles and continued HDIL-2 off-study without recurrent Grade 3 renal dysfunction.

Patients completed a mean of four cycles of induction therapy cycles of combination HDIL-2 and MAGE-A3 CI. Four patients continued to maintenance MAGE-A3 monotherapy. The overall response rate was 25% (3 patients with a CR and 1 patient with PR) (Fig. 1). The median time to response was 5.7 months (range 2.4 to 12.2 months). One patient (6%) achieved a PR after 8 weeks induction therapy. Three patients achieved a CR while receiving maintenance MAGE-A3 monotherapy at 11.7, 23.3, and 33.9 months respectively. Two of these patients completed all 8 cycles of HD-IL2, while the third patient only completed 2 cycles of HD-IL2 secondary to toxicity prior to proceed to maintenance MAGE-A3 CI monotherapy. The patient with a PR during the induction phase was converted to a CR with surgical resection of residual disease. With a median follow-up of 36.8 months, the median duration of response was not reached with an ongoing response up to 57.1 months. An additional six patients had stable disease (SD); thus, the overall disease control rate was 63%. The duration of SD was 2.5, 3.2, 3.3, 4.5, 5.5, and 27.5 months respectively. Median PFS was 3.6 months (95% CI 1.8 - NR), and median OS was not reached.

Ten patients had baseline biopsies with sufficient samples for gene expression profiling. An 84-gene signature was positive in 8 of 10 specimens (80%). Two of two patients (100%) with clinical response had a positive gene signature as did 6 of 8 patients (75%) who did not respond. Eleven baseline biopsies were analyzed by IHC to quantify immune infiltrates. PD-L1 expression was not predictive of response. Baseline tumor specimens from responders tended to have increased CD8+, CD45RO+, Granzyme B+ and PD-1+ tumor infiltrating lymphocytes though there were no statistically significant differences between responders (CR + PR) and non-responders (progressive disease, PD) (Fig. 2).

Flow cytometry was performed at baseline and week 3 on PBMCs of patients who had a response (CRs + PR, n = 4) and compared to those with primary progressive disease (PDs, n = 6). There were no differences in baseline cell populations. There were significant differences in the change of expression of multiple immune checkpoint proteins on the T regulatory cells from baseline to week 3 in the responding vs. non-responding patients with checkpoint protein expression increasing in the non-responding patients and decreasing in the responding patients from baseline to week 3 [41BB (p = 0.04), CTLA4 (p = 0.01), and LAG3 (p = 0.02)] (Fig. 3). Expression of 41BB on CD8+ T cells also exhibited an increase over time in non-responders compared to responders (p = 0.04). Flow cytometry for MAGE-A3 specific CD8+ T cells could be performed in 9 patients with appropriate HLA type. As only one patient with a clinical response had the appropriate HLA type (HLA-A0201+), statistical comparison of responders and non-responders was not feasible. However, the patient that achieved a CR had a high level of baseline MAGE-A3 specific CD8+ T cells and then had a further robust increase in MAGE-A3 specific CD8+ T cells at week 3 with much higher levels than any of the non-responding patients (Fig. 4).