We report here about the first phase I trial (Fig. 1) which administers 4 exosome vaccinations intradermally and subcutaneously at 1 week intervals (Fig. 2). The study was approved by the Ethics Committee, local IRBs and regulatory authorities, and informed written consent was given by all patients. The authors were in compliance with the Helsinki Declaration. Fifteen patients bearing melanoma fullfilling the inclusion criteria were enrolled in the study (see Additional file 1 and Table 1). Inclusion criteria were: biopsy-proven American Joint Committee on Cancer stage IIIB and IV metastatic melanoma, HLA-A1⁺, or -B35⁺ and HLA-DPO4⁺ phenotype (as defined by serological and molecular typing), tumor expressing MAGE3 antigen (assessed by RT-PCR, as described elsewhere 2728), inclusion at least 4 weeks after cytotoxic chemotherapy, surgery, or radiation therapy, intradermal skin test positivity to one or more recall antigens using Pasteur Mérieux multi DTH test, age > 18 years, Karnofsky performans status >80%, lymphocyte counts >1000 /mm³. Exclusion criteria were: prior chemotherapy or biotherapies <4 weeks before trial entry, brain metastasis, pregnancy, concurrent steroids or immunosuppressive therapy, history of asthma, congestive heart failure, autoimmune disease, active infections. Three patients presenting with skin or LN involved sites declined conventional therapies and received exosomes upfront (see Additional file 1 and Table 1). Patients received a 4 week outpatient vaccination course with antigen loaded DC derived-exosomes given intradermally (1/10^th) and subcutaneously (9/10^th) every week for a total of 4 vaccinations (Fig. 2). The injection sites were rotating between both thighs and forearms. The study was scheduled in two steps i.e 1) MHC class I peptide loading was «indirectly» performed on the DC culture at 10 μg/ml, and 2) MHC class I peptide loading was «directly» performed onto purified and acid eluted exosomes (Fig. 1). Indeed, our preclinical studies aimed at comparing the exosome immunogenicity after indirect versus direct peptide loading showed a superiority of the latter process 2022. Exosomes (quantified as concentrations of MHC class II molecules, see below) or peptides that were pulsed onto exosomes were administered in a dose escalation design. In the first step, 0.13 or 0.40 × 10¹⁴ MHC class II molecules were inoculated in a cohort of three patients each. In the second step, the concentration of peptide loading onto exosomes varied from 10 μg/ml to 100 μg/ml in a cohort of 3 and 6 patients respectively (see Additional file 1 and Table 1).

Insurance quality criteria allowing exosome release in patients were qualitative (expression of CD81 tetraspanins), quantitative (at least 1 × 10¹⁴ MHC class II molecules in immunocapture assays) and included a functional assay (bioactivity in the superantigen SEE test of potency).

Adverse events were graded according to the WHO Toxicity criteria. All patients underwent assessment of tumor status at baseline and 2 weeks after the fourth exosome vaccination. Tumor evaluation was performed using the RECIST criteria. Disease progression was defined as >20% increase in target lesions and/or the appearance of new lesions, partial response as a > 30% decrease in the sum of the longest diameters of target lesions.

Pre-, per- and post-immunization PBMCs were collected before vaccination (week 0, 5% of total leukocytes collected from the starting leukapheresis), after 2 (week 4, 50 ml heparinized blood sample) and 4 vaccines (week 7, by leukapheresis), isolated by Ficoll density gradient and conditioned either for immediat testing or kept frozen for further analyses. In patient #12, single cell suspensions from dissociated cervical lymph nodes resected at the end of the induction and retreatment phases were also collected.

Immunostainings were performed on sections obtained from formalin-fixed and paraffin-embedded lymph node samples. Sections were deparaffinized, placed in 10 mmol/L Na-citrate buffer (pH.7), and heated in a microwave during 20 min. Endogenous peroxidase was blocked with 1% hydrogen peroxide in methanol for 30 min. Slides were stained using anti-CD3 and anti-CD57 mAb (Pharmingen, France) and secondary antibodies with appropriate controls.

Leukaphereses of 1.5 blood mass performed in the 15 metastatic melanoma patients enrolled in the study allowed the recovery of 9.7 × 10⁹ ± 0.8 PBMC (range: 4.4–15) containing 20.7% ± 1.8 CD14⁺ cells (range: 12.4–33.0). These monocytes differentiated into immature MD-DC in rhu GM-CSF and rhu IL-4 (means: 313 × 10⁶ ± 100, range: 50–1200) as assessed at day 7 in flow cytometry highlighting loss of CD14 molecules, acquisition of CD1a, poor cell surface expression of CD83. In 3 patients out of 15, the first leukapheresis did not allow CD14⁺ cells to adhere and consequently, a second leukapheresis was required to harvest exosomes. The GMP process allowed to purify a mean of about 5.22 × 10¹⁴ ± 0.9 (range: 1.2–15.0) exosomal MHC class II molecules from supernatants of day 6 to day 7 MD-DC required for up to 41 ± 6.7 (9–115) id/sc vaccinations/patient at the lowest dosage (i.e 0.13 × 10¹⁴ exosomal MHC class II molecules /vaccine). As for the patients who benefited from continuation treatment requiring a second leukapheresis (pt #3, #12 and #14), the second exosome recovery was not significantly different from the first one. The quality control parameters required for GMP exosome batch release are reminded in M&M.

GMP exosomes were successfully produced from DC cultures derived from all melanoma patients enrolled in this phase I study, allowing at least 4 exosome inoculations.

No major (>grade II) toxicity was observed. A grade 1 fever was recorded in 5 patients. We noted slight inflammatory reactions at vaccine sites without outward delayed type hypersensitivity reactions. Some patients (#4, #8, #11, #12 and #15) reported a transient swelling and sensitivity of cutaneous and lymph nodes metastases 48 hrs after each exosome inoculation.

Among the six patients vaccinated in the first part of the study, only one patient (#3) presenting with a stage III disease exhibited a minor response (disappearance of one sc lesion out of 3). This patient benefited from a continuation therapy with exosomes every other 3 weeks for 21 months and remained stable for up to24 months. It is noteworthy that this patient was progressing despite vaccination with MAGE3 protein prior to enrollment in the Phase I exosome trial. In the second part of the study («direct loading»), responses were only observed in the second group of patients receiving the highest dosages of peptides (100 μg/ml) pulsed onto exosomes. One female patient (#12) presenting with five supraclavicular invaded lymph nodes after conventional chemotherapy (DTIC, 2 cycles) exhibited a partial response (PR) after the induction therapy. Interestingly, disappearance of arterial neovasculature concomittant with tumor shrinkage and necrosis as assessed by doppler pulsed ultrasonography could be demonstrated (Fig. 3). Halo of depigmentation around neavi appeared 10 months after the initiation of vaccination (Fig. 3). A continuation therapy with exosomes was administered for 4 months every other 3 weeks allowing stabilisation without toxicity supporting the indication of surgery that confirmed the partial response (Jul. 2002, Fig. 3) followed by a second leukapheresis allowing to pursue vaccination for 10 months. This patient relapsed in contralateral nodes six months after exosomes discontinuation but exhibited a slow pace of tumor growth. Two additionnal patients (#11, #14) presenting with only skin and /or LN lesions exhibited transient stabilisation and started a continuation therapy. Interestingly, patient # 14 who displayed a SD with exosomes had previously progressed despite biotherapy with IFNα and vaccination with poxviruses recombinant for the MAGE 3A1 cDNA. It is of note that patient #9 in a M1b stage exhibited a regression on a subcutaneous nodule but did progress in the pulmonary sites (mixed response).

Exosome therapy promoted 2 stable diseases, 1 minor response, 1 partial response and 1 mixed response in skin or LN sites even in patients progressing with biotherapies or alternate vaccines. Patient #3 and #12 with respectively minor and partial responses are still alive. Neither of these patients exhibited a delayed type hypersensitivity response to the immunizing epitopes after the completion of the 4 vaccines.

The phenotypic analyses of lymphocyte subsets did not reveal any significant changes in the percentages nor absolute counts following exosome therapy (not shown). Interestingly, the CD122 molecule i.e IL-2Rβ chain was upregulated in both CD4 and CD8 T cell subsets after exosome therapy (p < 0.05 using Student t'test, week 7 versus week 1 for CD4+ T cells). Serial ex vivo ELISPOT analyses of PBMC before (W1, Fig. 4A) and after (W7, Fig. 4B) exosome vaccination did not reveal any significant Th1 (using TT or DP04 MAGE 3 peptides) or Tc1 (using MAGE 3. A1/B35 peptides) type immune responses induced by exosomes. Recall responses (proliferative responses to TT protein using autologous DC or restimulation with autologous DC pulsed with MAGE 3.DP04) did not allow to conclude that exosomes boosted MHC class II specific responses at these time points (not shown). In accordance with these findings, the skin reactivities to iterative injections of id/sc exosomes did not correspond to clinically significant Delayed Type Hypersensitivity responses. Extensive immunomonitoring by screening of 15 day-microcultures of MAGE3 peptide pulsed PBMC performed in the presence of IL-2, IL-4, IL-7 30 using the MAGE3.A1/B35 tetramers allowed to detect and clone specific CTL precursors only in 3 patients (#3, 9 and 11) but estimated at a low frequency (10^-6 - 10^-7 of the CD8⁺ T cells). These studies did not allow to conclude for significant anti-MAGE3A1/B35 CTL responses induced by exosome therapy nor to suggest the clonal expansion of discrete T cell specificities (not shown) in patients' peripheral blood.

Patient#12 presenting with five supraclavicular lymph nodes exhibited a partial regression after 4 exosome vaccines in November 2001 and underwent a continuation therapy from Jan. 2002 to April 2002 (Fig. 3). The HLA class I haplotype of the patient was A2.A26.B35.B44 and the initial tumor sites expressed both MAGE1, MAGE3 and other melanoma antigens such as Melan-A/MART1 and NA-17 (as assessed using semiquantitative RT-PCR). The antitumor effects did not correlate with an enhanced frequency of MAGE3 specific CTL precursors as demonstrated using microculture detection assays after the first 4 vaccines or after the continuation treatment (not shown). However, tumor shrinkage after the continuation therapy prompted a partial surgical resection in July 2002 and comprehensive analyses of tumor infiltrating lymphocytes (TILN) and tumor cells in 4 avalaible sites (#1, 3, 5, 6 in Fig. 3). Site 5 was scored tumor free by the pathologists. Site 1 and 3 maintained the transcription of Mage3 mRNA. In these TILN sites, detection of HLA-A2/MART1_26–35, -/gp100_209M-217 -/Tyrosinase_239–251 and HLA-B3501/MAGE3_168–176 -specific CD8+ lymphocytes using soluble fluorescent tetramers were performed directly ex_vivo and after microculture restimulation assays using MART1_26–35 or MAGE3 peptides.Only high frequencies of MART1 specific CD8+ T cells were detected in TILN from site 1 and 6 (1.22 and 0.85 % respectively) and in peripheral blood (0.53% in May 2002 versus 0.15% in Oct. 2001) that were confirmed by expansion in all microculture assays (Fig. 5A and 5B). In parallel, halo of depigmentation around naevi appeared (Fig. 3). Moreover, a cell line established from site 1 (CUR1) did not express the HLA-A2 allele in flow cytometry using the MA2.1mAb but maintained the expression of HLA-BC molecules (Fig. 5C). In November 2003 while disease was slowly progressive, another LN was removed allowing to re-establish another cell line (CUR2) that no longer express HLA-BC molecules (as assessed in flow cytometry using W6.32 or anti-HLA-BC mAb, Fig. 5C).

Immunohistochemistry performed on the primary (April 2001) and secondary (July 2002) tumor specimen revealed quantitative modifications of tumor infiltrating CD3+CD57+ T cells. While activated T cells were mostly surrounding the tumor bed initially, a two fold expansion or recruitment of activated T cells invade tumor areas after exosome vaccines (Fig. 6).