The CSF-470 vaccine consists of 1.6 × 10⁷ lethally irradiated cells derived from four CM cell lines established in-house (30), MEL-XY1, MEL-XY2, MEL-XY3, and MEL-XX4, which are deposited at the German Collection of Microorganisms and Cell Cultures (DSMZ), Braunschweig, Germany. All cell lines were established from metastatic CM tumors. Cells were cultured as previously described (30). The cell lines were grown in a GMP core facility at the Centro de Investigaciones Oncologicas-FUCA. For vaccine preparation, the four cell lines were thawed, washed, mixed, and subsequently irradiated at 70 Gy (Siemens Lineal Accelerator). The cells were frozen in vaccine doses (1.6 × 10⁷ total cells) in liquid nitrogen until use. For vaccine administration each dose was thawed, washed, and resuspended in Dulbecco’s Modified Eagle Medium.

RNASeq was performed for each cell line with the sequencing platform Illumina Hiseq 4000, with more than 20 M high-quality single-end reads per sample (BGI Americas). Quality control of reads was performed with FASTX-Toolkit (31) and FastQC (32). Reads were aligned to the latest human Hg38 reference genome using the STAR spliced read aligner (33). Fragment counts were derived using HTSeq package (34). Differentially expressed genes were identified by a ranking based on adjusted p-values ≤ 0.005 and a false discovery rate ≤ 0.1 using the R/Bioconductor package edgeR (35). In Figure 1A, we show the expression profile of several melanoma-associated Ags, selected from RNASeq data that have been previously shown to be immunogenic in CM pts (36). Also, vaccine cells express GD2 and GD3 gangliosides as detected by flow cytometry as previously described (37) (Figure 1B).

CASVAC-0401 is a single institution, randomized, open label Phase II/III study, to investigate in CM pts stages IIB, IIC, and III postsurgery (adjuvancy), the efficacy and safety (primary objectives) as well as the quality of life (QOL) and immune response (secondary objectives) of the CSF-470 vaccine vs medium-dose IFN-α2b treatment (ClinicalTrials.gov: NCT01729663). A total of 108 pts will be enrolled in the Phase II–III study. As of March 2016, 47 pts were evaluated, of which 31 were eligible and randomized in a 2:1 ratio (vaccine:IFN-α2b). The randomization scheme was generated by using the Web site Randomization.com (http://www.randomization.com); Nobeltri S.R.L. (C.R.O.) was responsible for pts randomization. The study was approved by the Ethics Committee of the Instituto Médico Especializado Alexander Fleming (Buenos Aires, Argentina) and by the Argentine Regulatory Agency (ANMAT), and it is being conducted according to the Declaration of Helsinki Principles. All pts signed informed consent.

Vaccine was injected intradermally in the arms or thighs without lymphadenectomy. On day 1, pts received CSF-470 admixed with 1 × 10⁶ colony forming units BCG (Pasteur strain, Instituto Malbrán, Buenos Aires, Argentina) and 100 µg rhGM-CSF (Laboratorio Pablo Cassará, Argentina). On days 2–4, 100 µg rhGM-CSF/day was i.d. injected at the edematous vaccination site. During the 2-year study, pts in the vaccine arm should receive 13 vaccinations: 4 vaccinations 3 weeks apart; 5 vaccinations 2 months apart during the first year and 4 vaccinations 3 months apart during the second year. The pts assigned to the IFN-α2b arm received 10 million units (MU) IFN-α2b (Laboratorio Pablo Cassará, Argentina)/day/5 days a week for 4 weeks; then 5 MU thrice weekly for 23 months (Figure 2).

Enrollment conditions were as follows: (a) age between 18 and 65 years; (b) CM stages IIB, IIC, or III (AJCC); (c) Performance status Eastern Cooperative Oncology Group (ECOG) 0–1; (d) life expectancy >6 months; (e) no previous chemotherapy or immunotherapy; (f) less than 4 months after surgery; (g) non-evident disease as asserted by CT scans; (h) adequate hematological, hepatic, and renal function; (i) absence of pregnancy, with serum β-HCG determined before each vaccination in premenopausal women; (j) negative serology for HIV, hepatitis B, and hepatitis C. Exclusion criteria were as follows: (a) Type I or II diabetes; (b) pregnancy; (c) morbid obesity; (d) autoimmune diseases or hepatitis; (e) chronic medication with corticosteroids or NSAIDs; and (f) psychiatric disorders. More information about the CASVAC-0401 study is presented in File S1 in Supplementary Material.

During the study, pts should complete 15 visits; visit 1 was for pt selection and visit 15 for closure participation. Visits 2–14 lasted 4 days for CSF-470 pts and 1 day for IFN-α2b pts. At each visit, after evaluation of laboratory results, pts received a complete clinical examination. Imaging studies were performed as indicated in Figure 2. At baseline (PRE) and after 6 (POST1), 12 (POST2), and 24 months (POST3) of treatment, 200 ml of blood were obtained to monitor immunological parameters (Figure 2). AEs were graded and recorded according to the NCI Common Terminology Criteria for Adverse Events, version 3.0. Clinical evaluation: the end points of the study were DMFS, DFS, OS, toxicity, QOL, and immune response.

On the vaccination day, DTH was performed in the forearm with 1/10th of the CSF-470 dose. The reaction was measured at 1, 24, 48, and 72 h and recorded as follows: 0: macular erythema < 0.5 cm diameter; 1: macular erythema 0.5–1.0 cm; 2: macular erythema 1.1–2.0 cm; 3: macular erythema > 2.0 cm; and 4: papular erythema > 2.0 cm. A DTH score corresponding to the sum of the four values was calculated for each vaccination.

Peripheral blood mononuclear cells obtained by density gradient were incubated with anti-human MAbs: PerCP-CD3 (clone SK7), FITC-CD4 (clone RPA-T4), FITC-CD8 (clone RPA-T8), FITC-CD3 (clone UCHT1), and APC-CD56 (clone B159) (BD Biosciences, San Jose, CA, USA). For detection of Tregs, 1 × 10⁶ PBMC were stained with mAbs [FITC-CD4, AF647-FoxP3 (clone 259D/C7) and PE-CD25 (clone M-A251) (BD Biosciences)] following the manufacturer’s protocol. Lymphocytes were gated in FSC/SSC dot plot (≥30,000 events), and NK (CD3⁻CD56⁺), CD4⁺, CD8⁺, and Tregs (CD4⁺ CD25⁺ FoxP3⁺) cells were determined. Isotype-matched irrelevant MAbs were used as negative controls.

NK cells were tested for degranulating activity (CD107a membrane expression in the CD3⁻CD56⁺ population) by flow cytometry, using canonical K562 cells (ATCC, CCL-243) (38) or MelC (non-irradiated CM cells composing CSF-470 vaccine: MEL-XY1, MEL-XY2, MEL-XY3, and MEL-XX4 cell lines) as targets in a 10:1 ratio as previously described (39). All samples were acquired on a BD FACSCalibur using Cellquest Pro software (BD Biosciences, Franklin Lakes, NJ, USA) and analyzed with FlowJo 7.6.2 software (FlowJo, Ashland, OR, USA).

Frequency of myeloid-derived suppressor cells (CD14⁺CD3⁻CD20⁻CD56⁻HLA-DR⁻) was analyzed by flow cytometry. The following mAbs (BD Biosciences) were used: PE-CD14, FITC-CD3, FITC-CD20, APC-CD56, and PerCPCy5.5-HLA-DR (Mouse IgG2a isotype). Data acquisition (≥10,000 events) was performed using a FACSCanto II cytometer, using FACSDiva software (BD Biosciences). Data analysis was performed with FlowJo software.

PRE and POST1 pts’s PBMC and healthy donor (HD) PBMC were thawed, allowed to rest for 2 h at 37°C and then seeded (1 × 10⁶) in 1 ml of Complete Medium consisting of RPMI 1640 (Invitrogen) supplemented with 10% heat-inactivated human AB sera, 2 mM glutamine, 100 U/mL penicillin, 100 µg/ml streptomycin, 2.5 mM HEPES, and 50 U/mL of IL-2 (Laboratorio Pablo Cassará SRL), in 24-well plates (Costar, Corning, NY, USA). PBMC were stimulated with CSF-470 lysate in a 3:1 ratio (presenting cells from PBMC population: lysed CSF-470 cells), and cultured at 37°C, in 5% CO₂ for 12 days. Every 3 days, fresh complete medium with IL-2 was added. CSF-470 lysate was obtained after five cycles of freezing (liquid nitrogen, 3 min) and thawing (7 min, 37°C) and then centrifuged at 1,000 × g for 10 min to separate nuclei. Lysate aliquots were frozen at −80°C until use.

MultiScreen-IP opaque 96-well plates (High Protein Binding Immobilon-P membrane, Millipore, Bedford, MA, USA) were coated overnight at room temperature with 100 μl/well of 5 µg/ml mouse anti-human IFN-γ mAb (BD Biosciences) in 1X PBS (Gibco, ThermoFisher, USA). Then, the plates were washed and blocked for 2 h at room temperature with 200 µl/well of culture medium consisting of RPMI 1640 (Invitrogen) supplemented with 10% heat-inactivated fetal bovine sera (FBS, Gibco), 2 mM glutamine, 100 U/mL penicillin, 100 µg/ml streptomycin, and 2.5 mM HEPES.

Harvested PBMC from 12-day culture from pts (PRE and POST1 samples) and HD were added to the plates in duplicate at 2.5 × 10⁵ cells/well and cultured with either culture medium alone or with addition of OKT3 (30 ng/ml, BD Pharmingen, San Diego, CA, USA) plus PHA (1/1,000, M form, Gibco) for 24 h at 37°C, 5% CO₂. The plate was washed twice with deionized water and then thrice with PBS–0.05% Tween-20. A total of 1,000 μl/well of 2 µg/ml anti-human IFN-γ-biotinylated mAb (BD Pharmingen) in 1X PBS/10% FBS were added, and the plates were incubated for 2 h at room temperature. After three washes with PBS–0.05% Tween-20, the plate was incubated with 50 μl/well of streptavidin HRP (BD Pharmingen) diluted 1:100 in PBS 1X/10% FBS for 1 h at room temperature. The plates were then washed four times with PBS–0.05% Tween-20 and then twice with PBS. Spots were visualized by adding 50 μl/well of AEC Substrate (BD Biosciences) for 2 min. Substrate reaction was stopped by washing the plate with deionized water. Plates were scanned using an AID iSPOT ELR088IFL analyzer to quantify the number of spots per well.

Humoral responses were measured against MelC. For pt #006, her own CM cells were obtained from an excised subcutaneous metastasis and grown as a xenograft in NIH nude mice, from which a cell line was established. CM cells were seeded in 96-well culture plates (10⁴ cells/well) and cultured for 48 h. Reactivity was detected by ELISA in PRE, POST1, POST2, and POST3 serum samples (1/10 to 1/1,000 dilutions) using peroxidase-conjugated goat anti-human IgG (Abcam) and o-phenylenediamine as substrate (450 nm). A HD serum sample was used to set the background value. In the case of pt #006 cells, a serum sample obtained 2 years after finishing the clinical trial [2 years follow-up (F-UP)] was also analyzed. Pt #006 cells were also incubated with 1/10 diluted sera and bound antibodies (Abs) were detected using FITC-rabbit anti-IgA, G, M (DAKO) and observed by fluorescence microscopy (Olympus BX40 microscope, DP2-BSW software).

Formalin-fixed, paraffin-embedded CM tissue biopsies were studied. Histopathological features were determined according to AJCC-UICC staging (40) (Table S1 in Supplementary Material). Proliferative index (PI) was determined by Ki-67⁺ staining (41) (%): Ki-67⁺ tumor cells/(Ki-67⁺ tumor cells + Ki-67⁻ tumor cells) × 100. PI was determined in 1 mm² tumor hot spot zone (clone MIB-1, Dako). HLA-I expression (%) was determined in tumor tissues with clone EMR8-5 (Abcam). The avidin–biotin–peroxidase (ABC) system (Vectastain, Vector Labs) was afterward used. Sections were examined by optical microscopy (Olympus BX40 microscope, DP2-BSW software), and digitalized pictures were analyzed with ImageJ software (NIH).

It was determined after DNA extraction from tumor biopsies, PCR amplification, and Sanger sequencing (42).

Quality of life was evaluated by the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (“EORTC QLQ-C30, version 3.0”). Mean scores were calculated for each visit, only considering values from pts who were actually receiving treatment. ECOG performance status was also registered at each visit.

GraphPad Prism statistics software was used. Distribution of pts in both arms of treatment according to different characteristics of pts and tumors was assessed by χ² or t-tests. Time to event distribution was estimated by the Kaplan–Meier method. The significance was evaluated by the log-rank test (Gehan–Breslow–Wilcoxon test). For immune monitoring comparisons, paired t-tests were used. For DTH and tumor-associated immune populations analysis, the Mann–Whitney test was used. For QOL, unpaired t-test with Welch’s correction was used. The significance level was set as p < 0.05.

The Phase II stage of the CASVAC-0401 study was performed on 31 CM pts who were recruited at the Instituto Alexander Fleming between March 2009 and April 2014. The characteristics and evolution of the participating pts are listed in Table 1 and Table S1 in Supplementary Material, in which it can be seen an equitable distribution of pts in both arms of treatment with regard to age, sex, and AJCC-UICC stage stratification. However, primary tumor ulceration was more frequent in the vaccine arm. A Consort Flow chart is shown in Figure 3 describing pts demography and treatment assignment. Twenty-one pts (70%) had the BRAF^V600E mutation. Pt #030 was withdrawn from protocol after receiving four vaccines due to confirmation that millimetric lung lesions detected by CT-scan at screening, and informed as non-specific, later developed as overt metastases; the pt was, therefore, at stage IV and understaged at screening. This pt was included in the AEs record but excluded from efficacy calculation.

As of March 2016, pts from CSF-470 vaccine arm received a total of 219 vaccinations (Table 1). A total of 6/19 pts from the vaccine arm and 8/11 pts from the IFN-α2b arm were withdrawn from protocol due to appearance of distant metastases. With a mean and maximum F-UP of 39.4 and 83 months, respectively, DMFS was significantly higher for CSF-470 vaccinated pts than for IFN-α2b pts (p = 0.022). DMFS proportion at 25 months was 72.8% for the vaccine arm and 27.2% for IFN-α2b arm (Figure 4A and Table S1 in Supplementary Material). For vaccinated pts, the absolute risk reduction was 51.7% (95%, CI: 19.6–83.7%) and necessary number to treat was 2 (95%, CI: 1–5) at 24 months. However, when both regional and distant progressions were considered (DFS), the difference was not statistically significant as estimated by Kaplan–Meier (p = 0.12) (data not shown). Median OS has not been reached for pts in the vaccine arm and it was 39 months for IFN-α2b pts (data not shown).

The AEs are listed in Table 2 (CSF-470-vaccinated pts) and Table 3 (IFN-α2b pts). The most common toxicity in CSF-470 pts (20/20) was an inflammatory reaction at the vaccination site, which were grades 1–2 for most pts. Ninety percentage of the pts presented mild to moderate pain and 40% had pruritus at the vaccination site. In 6/20 pts, the BCG dose had to be reduced by 75% and pts continued vaccination. Systemic AEs were mild (grade 1); one pt with gastritis required treatment with omeprazole and ranitidine and symptoms resolved within 24 h. A total of 4/20 pts presented allergic reactions, mainly cutaneous rash, pruritus, and dyspnea (grades 2–3). In these pts, GM-CSF dose was reduced to 75 μg/day; in only one pt GM-CSF had to be suspended. Fifty milligrams of diphenhydramine were administered p.o. before each vaccination to pts with allergic reactions to prevent symptoms. In the IFN-α2b arm, 3/11 pts interrupted treatment due to excessive toxicity, 5/11 had transitory interruptions and dose reduction and one pt refused to continue treatment within the first month. Only 2/11 pts could complete the 2-year treatment. QOL was significantly better for CSF-470 pts than for IFN-α2b pts as evidenced by both scores obtained from the QOL questionnaire (Figure 4B).

Delayed-type hypersensitivity skin test increased in every vaccinated pt after receiving CSF-470 (Figure 5A). After seven vaccinations (visit 8), DTH reactivity was higher in pts with no evidence of disease as compared to those with progressive disease (PRO) (Figure 5B). As for PBMC, population frequencies of PRE and POST1 samples were evaluated. In most CSF-470 pts, only CD3⁻CD56⁺ NK cells significantly increased after vaccination. NK cells from both PRE and POST1 samples from pts receiving vaccine or IFN-α2b were functionally competent, as detected by in vitro degranulation assays after co-culturing pts’ PBMCs with K562 or MelC as target cells (data not shown). CSF-470 pts had a modest reduction in the CD4⁺CD25⁺FoxP3⁺ Treg lymphocyte population, but no changes in CD3⁺, CD4⁺ and CD8⁺ T cell proportions were observed throughout the vaccination. IFN-α2b treated pts showed a significant decrease in CD3⁺ cell frequency in POST1 samples, particularly in CD4⁺ and CD8⁺ T cells (Figure 5C).

Myeloid suppressor (CD14⁺HLA-DR⁻) cells were less than 1% of CD14⁺ cells in PRE and POST1 samples from all pts (not shown).

In order to determine peripheral blood T cells reactivity to melanoma Ags in both arms of treatment, IFN-γ ELISPOT analysis was performed after stimulation with a vaccine lysate. PRE and POST1 PBMC samples were analyzed (Figure 6). In the vaccine arm, 15/19 pts could be analyzed. Pt #003 was not analyzed due to early recurrence before obtaining the POST1 sample; PBMC from pt #006 could not be expanded in vitro after thawing, pt #011 samples were scarce and in pt #025 blood samples could not be obtained due to bad venous access. In the IFN-α2b arm, 7/11 pts were analyzed. Pt #014, pt #019, pt #027, and pt #028 were not analyzed since their participation in the protocol was discontinued before obtaining the POST1 blood sample. In the vaccine arm, the IFN-γ secreting T cells increased in 14/15 pts and in case of pt #013 they were already high in the PRE sample and remained unchanged in the POST1 sample The difference between IFN-γ secreting T cells in PRE and POST1 samples was statistically highly significant (p = 0.0001). Thus, CSF-470 vaccination induced T cell reactivity against melanoma Ags contained in the vaccine cells. On the contrary, in the IFN-α2b arm, in 4/7 pts the number of IFN-γ secreting T cells was low and did not increase with treatment, while in 1/7 they slightly increased. It is remarkable that in pt #002 the number of spots increased with treatment and in pt #024 it was high before treatment and it remained at the same levels at the POST1 sample; both pts remain NED after long F-UP. In the IFN-α2b arm, the difference between the PRE and POST1 samples was not statistically significant (Figure 6).

Thus, there are pts in both arms who recognize Ags contained in CSF-470 vaccine even before any treatment and vaccination induces a significant increase in T cell response.

All vaccinated pts significantly increased their serum Abs reactivity against MelC (PRE vs POST1 samples) (Figure 7A). In the case of pt #006, increased serum reactivity against autologous tumor cells derived from a s.c. metastasis was detected (Figure 7B); the Abs title returned to the PRE level 2-year after ending her participation in the study (2-year F-UP sample). Particularly, tumor cell surface reactivity was evident in the POST1 sample (Figure 7C).