Vaccination with LAG-3Ig (IMP321) and Peptides Induces Specific CD4 and CD8 T-Cell Responses in Metastatic Melanoma Patients—Report of a Phase I/IIa Clinical Trial

This article is featured in Highlights of This Issue, p. 1299

Melanoma is a highly aggressive cancer, with increasing incidence and mortality rates. Although surgery can cure melanoma detected at early stages, the prognosis of most patients with metastatic disease is unfavorable. Chemotherapy, irradiation and therapy with IFNα are of limited efficacy (1, 2). Recently, much more effective therapies have been developed, with the introduction of BRAF inhibitors active for patients whose tumor expresses V600 mutations in the BRAF gene (3). High-dose interleukin 2 (IL2) was showing marked successes but with considerable side effects and costs, and only for selected patients (4). More recently, a novel and more widely applicable form of immunotherapy has been introduced: a “checkpoint blockade,” consisting of a monoclonal antibody specific for cytotoxic T-lymphocyte–associated protein 4 (CTLA-4), named Ipilimumab. It was approved in 2011 for the treatment of metastatic melanoma (5). Subsequently, antibodies that block PD-1/PD-L1 (Programmed Death 1/Programmed Death-Ligand 1) pathway have been introduced and shown great progress in the treatment of metastatic melanoma (6) leading to the first market approvals in 2014. Of high interest, positive clinical studies are also reported in patients with carcinomas of lung, kidney, bladder, and head and neck, and large efforts with current studies are also ongoing in patients with other cancers. Besides CTLA-4 and PD-1/PD-L1, the targeting of further inhibitory lymphocyte receptors (“checkpoints”) is explored in an increasing number of preclinical studies (7, 8). One of these is lymphocyte activation gene-3 (LAG-3) or CD223, an inhibitory receptor that modulates T-cell homeostasis, proliferation, and activation (9). The LAG-3–specific antibody BMS986016 is already in clinical development, with two clinical trials recruiting patients with solid tumors or hematologic neoplasms, respectively (www.clinicaltrials.gov).

Therapeutic vaccination is a promising strategy against malignant diseases. Similar to other immunotherapies, a major challenge is to break immune tolerance, to induce a powerful and targeted immune response, and to avoid autoimmune side effects. Over the last decades major efforts were made to develop immunomodulatory molecules that act as vaccine adjuvants. An important class of innate immune receptors, the Toll-like receptors (TLR) are targeted and studied in animals and in humans, uncovering novel means of inducing strong immune responses (10). Beside TLRs, there are additional receptors for activating innate immune pathways, such as RIG-I–like receptors, NOD-like receptors, AIM2-like receptors, and STING (11). Although these receptors are interesting for future vaccine development, this field is less advanced than targeting TLRs.

Since several years, the non-TLR agonist, soluble LAG-3Ig (IMP321, Immutep S.A.) is in clinical development (12, 13). The rational is at least two-fold. First, this molecule can block inhibitory signals in T-cells similar to the “checkpoint blockade” principle. Second, it can stimulate antigen-presenting cells (APC), which makes LAG-3Ig an attractive vaccine adjuvant. LAG-3 is a CD4-like protein able to bind to major histocompatibility complex (MHC) class II molecules with a 100-fold higher affinity than CD4. IMP321 is one of a few clinical-grade non-TLR adjuvants for T-cell vaccination. LAG-3Ig stimulates the innate immune system by inducing activation and migration of APCs, particularly dendritic cells. Consequently, it enhances specific immune responses, by stimulation of antigen cross presentation to CD8 T-cells, and by decreasing regulatory T-cell frequency and function (9). Clinical trials with IMP321 showed already promising results when used as adjuvant for vaccination against hepatitis B and influenza viruses in healthy individuals (14, 15). Several trials have provided evidence for clinical activity in patients with advanced renal cell carcinoma, metastatic breast carcinoma, and metastatic melanoma (16–18).

Antigens used in cancer vaccines should be exclusively expressed by the tumor cells, or should be lineage specific in case of tumors arising from nonvital tissues/cells such as melanoma. The melanocyte differentiation antigen melanoma antigen recognized by T-cells-1 (MART-1)/Melan-A is one of the rare antigens that are expressed by a vast majority of patient's tumors (19, 20). Often, this antigen triggers T-cell responses upon tumor progression, which can be amplified by immunotherapy (21–24). Furthermore, cancer-testis antigens such as melanoma-associated antigen 3 (Mage-A3) and NY-ESO-1 are excellent antigens, due to their high degree of tumor specificity, the relatively high affinity of peptides derived from these antigens that bind to human leukocyte antigen (HLA)-A*0201 and presentation to specific CD8 T-cells. NA-17 represents also a promising target antigen for the development of melanoma immunotherapy as it triggers tumor-specific CD8 T-cells and showed promising clinical responses in a dendritic cell-based vaccine trial (25). Importantly, simultaneous targeting of multiple antigens likely reduces the risk of tumor outgrowth by escape variants.

Besides components that trigger innate immune cells and antigen specific ones, vaccines may also be enhanced by delivery systems with depot effect. One of the best adjuvants for T-cell vaccines is incomplete Freund's adjuvant (IFA), produced as Montanide ISA-51. Because of its nonresorbable mineral oil component, it persists at subcutaneous injection sites for several weeks up to multiple months and contributes to continuous or repetitive T-cell stimulation.

Based on the above outlined evidence and on our own previous experience (18), we designed a vaccine formulation composed of IMP321 (LAG-3Ig), Montanide ISA-51 (IFA), and tumor antigen–specific peptides. We combined four short HLA-A2–restricted peptides (Melan-A, NY-ESO-1, Mage-A3, and NA-17) to target CD8 T-cells, and one longer 15-mer peptide (Mage-A3), containing HLA class II epitopes to target CD4 T-cells. Sixteen metastatic melanoma patients were included in this study and vaccinated up to nine times.

For comprehensive immune monitoring, we developed a combinatorial tetramer staining allowing parallel quantification of 13 specific CD8 T-cell populations in T-cell cultures after one stimulation in vitro. In addition, Melan-A–specific CD8 T-cells were quantified directly ex vivo, without any culture steps. Finally, we quantified the frequency and evaluated the functional potential of Mage-A3–specific CD4 T-cells induced by vaccination.

The vaccinations were well tolerated, showing comparable (mostly local) adverse events like vaccination with Montanide alone. Remarkably, the vaccines induced specific CD4 T-cell responses in all 16 vaccinated melanoma patients (100%), and specific CD8 T-cells to at least one antigen formulated in the vaccine in 13 patients (81%).

More details are given in the Supplementary Materials and Methods section.

HLA-A2⁺ patients with histologically confirmed metastatic melanoma of stage III to IV, expressing Melan-A/MART-1 [determined by reverse transcription (RT)-PCR or immunohistochemistry] were included upon informed consent, in this phase I/IIa, single center, open, noncomparative study. The primary objective was measurement of antigen-specific immune responses besides safety and tolerability assessments. Local ethic review committees and responsible health authorities approved the study, which was carried out according to the Good Clinical Practice guidelines and the Declaration of Helsinki and was registered at www.clinicaltrials.gov (NCT01308294).

Eligible patients were immunized subcutaneously with the synthetic peptides and 250 μg IMP321/LAG-3Ig (Immutep S.A.). The first six vaccines were formulated with 1 mL Montanide ISA-51 (Seppic Inc.) and the last three vaccines without Montanide. Patients with an expected survival of at least 3 months (status at entry is described in Supplementary Table S1A) received nine vaccines scheduled in three cycles with three vaccinations (3 weeks intervals between vaccines, 6–14 weeks between cycles), based on observations that multiple boosters can not only induce but also maintain T-cell responses during many months or even years (26, 27). The clinical-grade peptides used in the vaccines were: Melan-A_26-35 native EAAGIGILTV or analogue (A27L) ELAGIGILTV, NY-ESO-1_157-165 SLLMWITQC, Mage-A3_112-120 KVAELVHFL, NA17 VLPDVFIRC, all representing known class I HLA-A2 peptides; and Mage-A3_243-258 KKLLTQHFVQENYLEY, containing a class II HLA-DP4 epitope. Blood samples were collected before the first vaccine (C0), 1 week after the third vaccination of each cycle (C1, C2, C3), and 6 months after the end of the third cycle (follow up: FU) for some patients. Peripheral blood mononuclear cells (PBMC) were isolated by density gradient using Lymphoprep (Axis-Shieldy) and immediately cryopreserved in RPMI1640 supplemented with 40% fetal calf serum (FCS) and 10% dimethyl sulfoxide (DMSO).

CD8⁺ T-cells or CD4⁺ T-cells were enriched using magnetic activated cell sorting (MACS) technology (Miltenyi Biotec). The negative fraction of cells was loaded with laboratory-grade peptides (listed in Supplementary Table S2), irradiated (30 Gray), and used as APCs to stimulate peptide-specific T-cells during 10 or 12 days in presence of IL2.

Freshly thawed cells or cultured cells were stained using similar protocols. As a first step, single tetramer staining or combinatorial tetramer staining (TCMetrix) was performed. Cells were incubated 45 minutes at room temperature with class I tetramers (combinatorial tetramers or tetramer Melan-A-PE) or 60 minutes at 37°C with class II tetramer (tetramer Mage-A3/DP4-PE). The combinatorial tetramers are listed in Supplementary Table S2. In addition, surface antibodies and dead cell exclusion marker were used to identify live CD4 and CD8 T-cells. Data were acquired on a Gallios Flow Cytometer (Beckman Coulter) and analyzed using FlowJo 7.6.5 software (TreeStar).

A blood sample from patient Lau1187 (collected after three vaccines; C1) was stimulated with Mage-A3_243-258 peptide for 16 days, as described above in IVS section. CD4⁺ tetramer-low and CD4⁺ tetramer-high cells were separately cloned. The specificity of each clone was checked by tetramer staining analyzed by flow cytometry (BD FACS Array). Twelve days after expansion, clones were phenotyped in separate tubes for their expression level of CD4, T-cell receptor (TCR) and for their tetramer brightness. In parallel, clones were stimulated with 0.5 and 5 μg/mL Mage-A3_243-258 peptide in presence of Brefeldin A 10 μg/mL for 6 hours. Intracellular cytokine staining (ICS) was performed as described in the Supplementary Materials and Methods section. Analysis of cytokine coexpression was done with SPICE software version 5.2. Finally, clones were clonotyped as previously described (28).

The Student t test (two-sample two-tailed comparison) or paired t test was used for statistical calculations. P < 0.05 was considered significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant).

In this study, 16 HLA-A2⁺ patients with metastatic melanoma received serial vaccines (up to nine vaccines; see Supplementary Table S1A). The vaccines contained IMP321, Montanide and four short peptides representing class I epitopes presented by HLA-A2, and one long 15-mer peptide containing a class II epitope presented by HLA-DP4.

The patient's baseline characteristics are shown in Supplementary Table S1A. There were eight women and eight men, with an age ranging from 21 to 84 years. Ten patients had stage III melanoma and six patients with stage IV melanoma at study entry. Fourteen patients finished the entire clinical trial protocol and two patients discontinued the study earlier due to disease progression.

The safety analysis of the treatment showed a comparable severity of side-effects commonly reported for cancer vaccines. The 16 patients experienced in total 588 adverse events. One event was of life-threatening severity and was qualified as serious adverse event as it involved hospitalization of the patient, but it was not related to the study treatment. Twelve events were of severe intensity, all caused by melanoma metastases and considered as not related to the study treatment. The majority (534 events, i.e. 91%) of all adverse events were of mild intensity. Supplementary Table S1B shows all adverse events with an incidence of more than 5, according to their severity and their relationship to the study treatment. A total of 135 injections were applied to 16 patients, reported adverse events were mainly expected/prelisted local reactions at vaccine injection sites (313 events, i.e. 53.2%), consisting of induration (100 events), erythema (83 events), pain (72 events), warmth (38 events), and injection site reactivation (20 events). They were mostly mild and were resolved rapidly except indurations that could last for several weeks. Indeed, Montanide ISA-51 is well known to form a depot allowing long-term local vaccine persistence and causing local inflammation. The systemic reactions reported during study corresponded to 113 events of all adverse events (19.2%). The ones that were considered as possibly or probably related to the study treatment were mainly of mild severity. The most frequent systemic events were headache, myalgia, chills, asthenia, and arthralgia. Tumor progression was the only type of grade 3 adverse event. No grade 4 adverse events related to the study treatment were observed.

All together, the combination of IMP321, Montanide, and five peptides was safe and well tolerated, with mild severity local adverse events, no cases of ulcerations at the vaccine injection sites, and no treatment related systemic adverse events higher than grade 2.

To obtain maximal information on specific CD8 T-cells with a minimum of blood volume, we chose to use combinatorial tetramer staining. This technique is based on the combination of multiple tetramers labeled with a combination of multiple dyes (29, 30), allowing the detection of multiple specific T-cell populations in the same sample (up to 15 specificities with four dyes). We set up a panel detecting 13 specificities allowing the quantification of 13 populations of virus and tumor-specific CD8 T-cells in 10 million PBMCs (see the list of specificities in Supplementary Table S2).

As a first step, we expanded the T-cells by IVS with peptides in two conditions, one with viral antigens [Epstein–Barr virus (EBV), influenza (Flu), cytomegalovirus (CMV), human immunodeficiency virus (HIV)] and the other one with tumor antigens; some of which corresponded to the antigens included in the vaccinations (Melan-A, NY-ESO-1, Mage-A3, NA17), and others not [“control” tumor antigens, namely melanoma-overexpressed antigen-1 (Meloe-1), glycoprotein 100-2 (Gp100-2), Meloe-2, Mage-A10, Tyrosinase]. In addition, to avoid competition of the well-growing specificities versus the less frequent (weaker-growing) ones, we split the cells in 2 to 12 independent cultures (96 wells) depending on the quantity of CD8 T cells we purified. After 12 days of culture, the frequencies of specific CD8 T-cells were quantified by combinatorial tetramer staining, determined using the gating strategy shown in Supplementary Fig. S1.

The majority of patients had antigen-specific T-cells for EBV, Flu, and/or Melan-A (13–15 patients of 16; detection limit at 0.1%; Fig. 1A; Supplementary Table S3). Furthermore, two thirds of the patients had detectable amounts of CMV and/or NY-ESO-1–specific CD8 T-cells; 1/3 Meloe-1, Gp100-2, and/or Meloe-2–specific cells. Mage-A10 or Tyrosinase-specific T-cells were only detected in one patient (Lau 616 and Lau 1366, respectively). None of them had HIV, Mage-A3, or NA17-specific CD8 T-cells above the detection limit, despite that the patients had been vaccinated with the two latter peptides. HIV was a reliable negative control, based on the negative HIV serology result available for each patient.

In addition, EBV, Flu, CMV, and Melan-A–specific CD8 T-cells were found in more than 80% of the cultures (Fig. 1A and B), and the frequencies of specific cells were mainly above 10% of the cells (Fig. 1B). NY-ESO-1-specific CD8 T-cells were found in 34% of culture wells, and Meloe-1, Gp100-2, and Meloe-2–specific CD8 T-cells maximally in 13% of cultures. For these specificities, the frequencies mainly ranged between 0.1% and 1%. Mage-A10–specific CD8 T-cells were at high frequency in one patient (Lau 616) who had previously been vaccinated with this peptide, with different adjuvants (31).

For further analysis of the impact of vaccination on antigen-specific CD8 T-cells, we focused on the specificities found in more than one patient, and calculated mean frequencies in positive culture wells, excluding wells with frequencies below the detection limit of 0.1%. Thus, we focused on EBV, Flu, and CMV for the control viral antigens; on Melan-A and NY-ESO-1 for the tumor antigens against which the patients were vaccinated; and on Meloe-1, Gp100-2, and Meloe-2 for the remaining tumor antigens.

We were able to detect Melan-A and/or NY-ESO-1–specific CD8 T-cells in 15 and 9 patients, respectively. Overall, we found no significant increase of the frequencies of Melan-A–specific CD8 T-cells after vaccination when analyzed after IVS, but a significant increase was found for NY-ESO-1–specific CD8 T-cells (Fig. 2A–C). However, it is well known that IVS with peptide efficiently amplifies previously primed and thus highly frequent Melan-A–specific T-cells, leading to the saturation of this kind of proliferation system. Therefore, we analyzed the frequency changes with vaccination on Melan-A–specific cells after excluding the three patients who showed high frequencies already at baseline, due to tumor priming or previous treatment (C0 > 15%). In the remaining 12 patients, we found a significant increase of Melan-A–specific CD8 T-cells following vaccination (Fig. 2D).

Concerning the tumor antigen specificities not included in the vaccine (Meloe-1, Meloe-2, and Gp100-2) and the viral antigens (EBV, Flu, and CMV), we did not observe any significant frequency changes (Supplementary Fig. S2).

Together, the IVS-combinatorial tetramer technique showed increased frequencies of circulating CD8 T-cells (corresponding to an increase of 2-fold or more as compared to baseline) to at least one of the four class I peptides administered, in 13 of 16 metastatic melanoma patients.

In the majority of vaccine studies in cancer patients, IVS is required to allow the detection of tumor antigen-specific T-cells, as their in vivo frequencies are below the detection limits (which is approximately 0.01% of CD8 T-cells for flow cytometry–based techniques). Detection and analysis of tumor-specific T-cells directly ex vivo is still a major challenge. Yet, we and others were able to detect, in blood samples, Melan-A–specific CD8 T-cells, directly ex vivo, without any culture step, as these cells are unusually frequent. Moreover, some vaccine components such as CpG-B are capable of inducing extraordinarily strong expansion of human CD8 T-cells (32). We monitored the frequencies of Melan-A–specific CD8 T-cells before and after each vaccination cycle, directly ex vivo (Fig. 3 and Supplementary Table S4). Six patients showed an increase of 2-fold or more in their frequencies of Melan-A–specific CD8 T-cells after vaccination (Supplementary Table S4). Overall, this approach allowed revealing that the frequencies of Melan-A–specific T-cells increased significantly after vaccination (Fig. 3A). Remarkably, data obtained from frequencies of Melan-A–specific CD8 T-cells as determined by the IVS-combinatorial tetramer technique strongly correlated (P < 0.0001) to those collected after direct ex vivo analysis with standard tetramers (Fig. 3B).

All together, our data demonstrate that vaccination with IMP321, Montanide, and short peptides induced significant expansion of tumor-specific CD8 T-cells in vivo.

In addition to the short peptides, one longer peptide (15-mer) derived of the Mage-A3 protein was used for vaccination, with the aim to activate tumor-specific CD4 T-cells. For immunologic monitoring, we stimulated PBMC in vitro with the peptide for 10 days, followed by challenging the cells with the Mage-A3_243-258 peptide for 4 hours, allowing the quantification of IFNγ- and TNFα-positive cells upon ICS (Fig. 4A and B). Remarkably, the frequencies of IFNγ and TNFα producing cells were always increased after vaccination, demonstrating that all 16 patients uniformly generated CD4 T-cell responses with at least 2-fold higher frequencies as compared to baseline (C0).

Patients had been included in the study irrespective of their HLA class II genes. Previous publications reported that the Mage-A3_243-258 peptide can be presented by HLA-DP4 and HLA-DQ6 (33, 34). Using blocking antibodies specific for HLA-DP, HLA-DQ, or HLA-DR, we found a dominant usage of HLA-DP (Fig. 4C). As 13 of the patients were HLA-DP*0401 and/or HLA-DP*0402 (Supplementary Table S1A), we synthesized a class II tetramer to quantify Mage-A3/DP4–specific CD4 T-cells (Fig. 4D). Indeed, the frequencies of Mage-A3/DP4–specific CD4 T-cells were increased in all HLA-DP4+ melanoma patients, with an expansion of 2-fold or higher, including patient Lau 616 who was not typed for HLA class II.

In contrast to the first six vaccines, Montanide ISA-51 was omitted for the last three vaccines, with the aim to reduce local toxicities, and to determine whether this strategy may nevertheless achieve the desired booster effects. Despite slight trends toward reduced frequencies of some of the T-cell populations analyzed after cycle 3 as opposed to the previous two cycles, there were no statistically significant differences, suggesting that booster vaccinations without Montanide are a valuable option.

While analyzing the Mage-A3–specific CD4 T-cells, we observed two populations of specific cells with different tetramer staining intensities in patient Lau 1187 (Fig. 5A), which we sorted and cloned separately. Each clone showed a tetramer-low (described thereafter as Low) or tetramer-high (High) staining, corresponding to the tetramer staining intensity before sorting (Fig. 5A).

First, we checked the tetramer staining characteristics after expansion of the two clonal families and confirmed that the tetramer-high and tetramer-low profiles (Fig. 5B) were stable over time (data not shown). Theoretically, higher intensity tetramer staining could be associated to a higher expression at the surface level of the TCR and/or the co-receptor (CD4), and/or to a higher TCR avidity (35, 36). We next measured the expression level of TCR and CD4 separately, in order to avoid competition of the staining antibodies. As shown in Fig. 5B, a higher tetramer staining was not associated with a higher TCR or CD4 staining. We further tested the functionality of the clones in a 6-hour peptide stimulation assay, and quantified cytokines by ICS (Fig. 5C). We found that the two families of clones differed in their cytokine profile (Fig. 5C), with tetramer-high clones producing more TNFα, IL2, and also the Th2 cytokine IL13, but the same amounts or less of the Th1 cytokine IFNγ than tetramer-low clones (Fig. 5C). Tetramer-high clones were also more polyfunctional than tetramer-low clones (Fig. 5D).

Sequencing of the complementarity determining region 3β (CDR3β) of the TCR revealed that all the tetramer-low clones were of the same clonotype (VB6, CDR3β SIGLAGGTDTQ, JB2.3), whereas all the tetramer-high clones came from a second clonotype (VB7, CDR3β SRGTLPPMNTEA, JB1.1). These data suggest that each of the two clonal populations originated from a single precursor.

Although it was not in the focus of the study, we documented the clinical results. Twelve of the 16 patients entered the study without detectable tumor, due to previous surgical resection. Ten of these 12 patients remained tumor-free throughout the entire study period, whereas 2 patients developed new metastases, causing 1 patient to stop after the second cycle of vaccination. The remaining 4 patients entered the study with detectable metastases and all had disease progression. Of those, 1 patient stopped study participation after the first vaccination cycle. At the end of the vaccination period, all patients were alive. Subsequently, two of the 16 patients died after 8 and 14 months, respectively. The median follow up time was 47.2 months (with a range from 35.9 to 57.3 months) at the time of analyses (July 8, 2015). After the study, 9 of the 16 patients received one or more additional antimelanoma treatments, which were surgery (7 patients), chemotherapy (3 patients), irradiation (3 patients), BRAF inhibitors (3 patients, 2 of which were the 2 deceased patients), and anti–CTLA-4 antibody followed by anti–PD-1 antibody (1 patient). Obviously, poststudy treatments may impact on the clinical outcome in an uncontrolled manner, a caveat inherent to most clinical trials.

In this phase I/IIa study, 16 metastatic melanoma patients were vaccinated with IMP321 (LAG-3Ig), Montanide ISA-51 (IFA), and five synthetic peptides, resulting in CD4 and CD8 T-cell responses that were antigen specific, as no effect was seen on T-cells specific for other antigens than those used for vaccination (summarized in Table 1).

The treatment was well tolerated; none of the patients discontinued the study due to treatment-related adverse events. Montanide ISA-51 has been used in a large number of cancer patients, and in cumulative doses up to 16 mL (37). Despite the overall favorable safety profile of Montanide ISA-51, the local side effects can be strong and may require surgical removal of the nonresorbed material when causing persistent local inflammation with ulceration leading to bacterial infections (38). This was, however, not encountered in this study; the maximal local toxicity was grade 2.

The safety and toxicity profile of IMP321 has been established in two randomized phase I trials of subcutaneous vaccination against Influenza virus (Flu, n = 60) and the Hepatitis virus (HBsAg, n = 48) in healthy volunteers (14, 15). Both studies revealed very good clinical tolerability with a low toxicity profile for the four dose levels of IMP321: 3, 10, 30, and 100 μg.

The vaccinations induced specific CD8 T-cell responses to at least one of the injected antigens in 13 of 16 melanoma patients (81%), and specific CD4 T-cell responses in all patients (16/16, 100%). Thus, the vaccine fulfilled the intended purpose to simultaneously induce CD8 and CD4 T-cell responses, according to the principle that CD8 cytotoxic T-cells are key players in the control and the killing of tumor cells, and CD4 T-cell help supports the CD8 T-cell responses (39, 40). We detected Melan-A– and NY-ESO-1–specific CD8 T-cells in the majority of patients (15/16 and 9/16 patients, respectively). Among them, 10 patients showed an increase in the frequency of Melan-A–specific cells and 8 patients in the frequency of NY-ESO-1–specific cells (Fig. 2 and Table 1). Unfortunately, we did not find CD8 T-cell responses to Mage-A3 and NA17. Although difficult to explain, it may be possible that the applied vaccine formulation has weakness for activating T-cells with low precursor frequencies, a problem that is also observed with other types of vaccines (26, 27, 41).

Remarkably, all 16 patients developed Mage-A3–specific CD4 T-cell responses, independently of their HLA class II genotypes. The Mage-A3_243-258 peptide is presented by HLA-DP4 and HLA-DQ6 (33, 34). Among the 16 patients, 13 were HLA-DP4 positive, in line with the reported high frequency of this allele (33). Therefore, it was not surprising that we could frequently detect HLA-DP–restricted cells (Supplementary Table S1A).

Besides studying cytokine production by the Mage-A3_243-258 peptide–specific CD4 T-cells, we used highly purified tetramers (42) produced with this peptide and recombinant HLA-DP*0401 protein, allowing to identify two CD4 T-cell populations with different tetramer staining intensities. Interestingly, they represented two dominant clonotypes with different cytokine profiles. It is worthwhile to note that the differences in tetramer staining intensity remained stable over time, and did not reflect different levels of TCR downregulation. On the contrary, the low tetramer staining cells expressed even higher TCR levels, which could be due to lower triggering and thus less TCR downregulation. This may possibly be associated with lower TCR affinity, and/or different fine specificity. Indeed, it has been shown that a given peptide can make different configurations on a particular MHC class II protein, resulting in different epitopes recognized by different TCR (fine) specificities (43–45). In this regard, our observations are based solely on a single patient. Generalization would require more extensive studies which are beyond the scope of this clinical trial.

In conclusion, the vaccinations of this study induced tumor-specific T-cells in the majority of patients. CD4 T-cell responses were very satisfactory. CD8 T-cell responses were less frequent, but still comparable to vaccines formulated with TLR2 ligands (46) or TLR4 ligands (24), and more frequent than with protein vaccines (20). The CD8 T-cell responses were however less frequent and less strong as compared to vaccines with short peptides, IFA and CpG-B, representing the currently most potent synthetic vaccine formulation for the induction of human CD8 T-cell responses (32). Future vaccine strategies may profit from combinations with multiple TLR/innate immune stimulators, potentially capable of mimicking immune responses to viruses that can generate more robust and long lasting T-cell responses (ref. 47 and manuscript in preparation). Thus, besides multiple antigens (possibly also including mutated antigens; ref. 48), future vaccines may also require multiple immune “adjuvants.” In such scenarios, the role of IMP321 and its activatory effects on APCs (9) remains to be determined.

Although the clinical results were relatively favorable, they cannot be firmly interpreted as this is a phase I study and has not been designed to determine clinical efficacy. Before designing larger phase III studies with clinical endpoints, it will be useful to further define the clinical role of IMP321, particularly with respect to vaccine component combinations as mentioned above. Carefully performed phase I/II studies may evaluate candidate treatment combinations, based on their capabilities of sound systemic T-cell activation and also overcoming immune suppression and T-cell exhaustion in the tumor microenvironment (49). Triggering HLA class II with IMP321 may have effects that are complementary or even synergistic to TLR stimulation (9). Furthermore, although we did not combine with checkpoint blockade (e.g., anti–CTLA-4 or anti–PD1/PD-L1 antibodies), there is a sound rationale for doing so, as the antigen-specific nature of vaccines enhances treatment specificity and thus may increase the efficacy/toxicity ratio of checkpoint blockade (8, 49).

O. Michielin is a consultant/advisory board member for Bristol-Myers Squibb, GlaxoSmithKline, Novartis, and Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Legat, P. Baumgaertner, D.E. Speiser

Development of methodology: A. Legat, F. Triebel, D.E. Speiser

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Legat, P. Baumgaertner, S.A. Maillard, E.M. Iancu, L. Lebon, D. Dojcinovic, O. Michielin, E. Romano, G. Berthod, D. Rimoldi, D.E. Speiser, L. Cagnon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Legat, P. Baumgaertner, L. Cagnon, E.M. Iancu, L. Lebon, E. Romano, D. Rimoldi, D.E. Speiser, H. Maby-El Hajjami

Writing, review, and/or revision of the manuscript: A. Legat, H. Maby-El Hajjami, P. Baumgaertner, L. Cagnon, O. Michielin, E. Romano, F. Triebel, I. Luescher, N. Rufer, D.E. Speiser

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Maby-El Hajjami, L. Cagnon, S.A. Maillard, C. Geldhof, P. Guillaume, D. Dojcinovic, I. Luescher, D.E. Speiser

Study supervision: O. Michielin, D.E. Speiser

Other (clinical trial coordination): H. Maby-El Hajjami

Other (clinical trial management and coordination): S.A. Maillard, C. Geldhof, L. Cagnon, H. Maby-El Hajjami

The authors thank the patients for their dedicated collaboration, and Immutep and Ludwig Cancer Research for providing IMP321 and clinical-grade peptides, respectively. The authors also thank B. Schuler-Thurner and G. Schuler for the Mage-A3/DP4 peptide; L.J. Old, J. O'Donnell-Tormey, L. Harmer, J. Skipper, R. Venhaus, L. Pan, M. Matter, C. Brignone, S. Leyvraz, C. Jandus, P.O. Gannon, P. Romero, J. Schmith, E. Devêvre, N. Montandon, L. Leyvraz, M. van Overloop, P. Marcos Mondéjar, A. Wilson, D. Labbes, S. Winkler, A. Digklia, K. Homicsko, S. Badel, H. Bouchaab, G. Buss, A. Christinat, F. Claude, N. Divorne, M. Figeri, M. Gavillet, A. Stravodimou, D. Taylor, E. Tzika, and J.-P. Zuercher for essential support, collaboration, and advice; and A. Erdmann-Voisin, L. Guihard, L. Valloton, and G. Wuerzner from the Clinical Research Center of Lausanne for their excellent monitoring support. The authors appreciate the support and assistance of the CHUV physicians, nurses, and staff of the Medical Oncology Service, Institute of Pathology, Clinical Investigation Units, and Blood Bank Donor Room.

This work was supported by the Cancer Research Institute (USA), Ludwig Cancer Research (USA), the Cancer Vaccine Collaborative (USA), Atlantic Philanthropies (USA), the Wilhelm Sander-Foundation (Germany), a Swiss Cancer Research grant (3507-08-2014), a Swiss National Science Foundation grant (Sinergia CRSII3_141879), and a SwissTransMed grant (KIP 18).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.