Immunotherapy for Melanoma: Current Status and Perspectives

Interleukin-2 (IL-2)

The biological effects of IL-2 are complex. Relevant for cancer therapy is the enhancement of CTL and NK-cell lysis. In response to IL-2 stimulation, a mixture of NK and CD8⁺ cells acquire cytolytic properties, which lead to tumor cell killing in vitro, even in the absence of HLA-class I restriction. Complete responses (CR) were produced in 6% and partial responses (PR) in 16% to 20% of patients with metastatic disease that failed standard therapy that were treated with a high-dose (HD) regimen of 600,000 to 720,000 IU/kg i.v., repeated every 8 hours for 8 to 14 doses per cycle with two cycles spread by a week considered a treatment course.^5,17 However, many patients are not able to tolerate more than 8 doses per cycle of treatment, with the second cycle typically consisting of fewer doses than the first. Although initial murine studies supported improved results with the addition of LAK cells to HD IL-2 therapy, subsequent clinical data in patients with metastatic disease showed an equivalent efficiency with the HD IL-2 alone.^18,19 With a median response duration of 8.9 months and 44% of responders being alive at 6 years, HD IL-2 offered the possibility of cure for a small fraction of patients.²⁰ An analysis of 374 metastatic melanoma patients receiving high-dose IL-2 suggested that patients having only cutaneous metastasis had a higher response as compared to however patients with diseases at other sites. Additionally, in the same analysis, it was found that development of vitiligo and thyroid dysfunction was associated with response.²¹ An elegant study by Sabatino et al²² used a multiplex antibody-targeted protein array platform to identify pretreatment elevation in VEGF and fibronectin as being indicators of poor response to IL-2 therapy.

The toxicity of IL-2 therapy prompted studies assessing feasibility of lowering the doses of IL-2. While this results in reduced adverse effects, it is associated with an inferior RR (frequently under 5%), duration and quality of responses.²³ A frequently employed continuous infusion of IL-2 of 18 MU/m²/d for 5 days is characterized by a similar toxicity and response rate as HD IL-2, but a shorter duration of response.⁶ The current consensus is that although the high dose regimens are currently approved by the FDA, the issue of optimum dose appears to still be unresolved, primarily due to lack of randomized data controlled for prognostic factors across institutions.

Interferon-α (IFN-α), IL-2⁺IFN-α Combinations

IFN-α exerts anticancer activities through numerous mechanisms, including direct antiproliferative/apoptotic effects,²⁴ increasing immunogenicity of tumors,²⁵ suppression of angiogenesis,^26,27 as well as modulating the innate and adaptive immune response.²⁸ Response rates of 14% with rhIFN-α2a and up to 23% with IFN-α2b were obtained in Phase II trials in metastatic melanoma using various schedules and dosages.²⁹ Accumulating clinical experience showed that responses to IFN-α can be delayed for a couple of months, and continuous treatment is more effective. One potential interpretation of these findings is that IFN-α stimulates clonal expansion of cytotoxic T cells that naturally are subjected to tumor-mediated immune suppression.³⁰

The rationale for combining IL-2 with IFN-α resides in the up-regulation by the latter of HLA and tumor-associated antigen (TAA) expression on tumor cells, which may enhance the T-cell lysis induced by IL-2.³¹ However, clinical trials have not confirmed the anticipated benefits of combination therapy over HD IL-2 alone.^7,32 Similarly, trials combining IL-2 with granulocyte-macrophage colony stimulating factor (GM-CSF), tumor necrosis factor (TNF), IFN or IL-4 did not result in any meaningful clinical improvement as reviewed in ^ref.23

Peptide Vaccines

Poor efficacy of complex tumor cell-derived polyvalent vaccines has prompted interest in more defined antigens. Although highly selective approaches have the advantage that: (a) immunogenic epitopes can be selective for; and (b) epitopes can be modulated (eg, altered peptide ligands) to artificially increase immunogenicity. The first defined human cancer-testes antigen, MAGE-1, was characterized in 1991.³ Since then a long and growing list of potential targets for cancer immunotherapy has emerged. In this context, metastatic melanoma served as the first model to test immune responses to peptide vaccines.

Spontaneous T-cell responses to the native gp100 antigen have been noted in patients with metastatic melanoma who experienced tumor regression following adoptive therapy with TIL and IL-2. Rosenberg et al⁶⁸ utilized a cancer vaccine consisting of immunodominant peptides derived from the gp100 melanoma-associated antigen, administered with a synthetic peptide designed to increase binding to HLA-A2 molecules. Of patients receiving the vaccine plus IL-2, 42% had an objective response. However, three different schedules of high-dose IL-2, administered with the 209-2M peptide did not produce better activity than expected with IL-2 alone (OR 14.5%, CR 8%).⁶⁹ Vaccination with the synthetic modified gp100: 209 to 217 (210 M) peptide, created by altering the anchoring amino acid (methionine in place of threonine at position 2), produced an increase in specific CD8⁺ T cells after vaccination in 28 of 29 patients, of whom 77% remained disease-free at 23 months.⁷⁰ Another approach, consisting of addition of tetanus peptide as a nonspecific helper epitope resulted in Th-cell responses in 79% of patients and an OS of 75% at 4.7 years of follow-up, which compared favorably with the expected survival.⁷¹ However, clinical responses are not always correlating with the immune activation, as exemplified by a study involving vaccination with the HLA-A*0201-restricted Mart-1/Melan-A(27 to 35) and incomplete Freund's adjuvant (IFA), where no clinical activity was observed, despite a 94% rate of CTL-specific responses.⁷² Conversely, no evidence of enhancement of the systemic immune response could be documented in nine patients immunized with the peptide MAGE-A12: 170 to 178 administered in IFA, although one patient had an ongoing PR.⁷³

A paradoxical dichotomy between elicitation of immune responses and clinical activity has also been seen with the modified gp100 epitope g209-2M. While 91% of patients showed successful immunization, no clinical responses were observed. Moreover, only 16% of patients treated with IL2⁺g209-2M vaccination developed immune reactivity, whereas 13 of 31 (42%) demonstrated clinical responses.⁷⁴ An explanation for this phenomenon can reside in trafficking of the vaccine-specific T cells to the tumor site, although it is possible that the discrepancy originated in a differential susceptibility of CTLs to in vitro sensitization.⁷⁵

A recent breakthrough was seen in a multicenter, randomized, prospective trial compared HD IL-2 alone with IL-2 and gp100 209 to 217 peptide in 185 stage III/IV patients expressing HLA0201. The response rate was significantly higher in the combination treated (22.1%) as compared to the IL-2 alone treated patients (9.7%). A significant improvement was seen also for the progression free survival 2.9 months (1.7 to 4.5) compared to 1.6 months (1.5 to 1.8), respectively.⁷⁶ This appears to be the first description of clinical benefit in metastatic melanoma using vaccination.

In the E1696 trial assessing multiepitope (MART-1, gp-100, and tyrosinase) vaccine alone, in combination with either GM-CSF, IFN-α2b, or in combination with both, of 115 patients who were analyzed, median survival of patients exhibiting immune response to at least one of the antigens was prolonged as compared to those lacking induction of immunity (21.3 mo vs. 13.4 mo; P=0.046). There was no difference in terms of immunity between patients receiving the vaccine alone or in combination with cytokines.⁷⁷

IL-12⁺ Peptide Vaccines

Murine studies have shown that IL-12 promotes potent antitumor immunization when co-administered with peptides loaded onto other class I MHC⁺ cells, thus potentially bypassing the need to use DCs. In a Phase I clinical trial, patients with metastatic melanoma received autologous PBMCs pulsed with a MAGE-3 or Melan-A peptides and co-administered with various doses of rhIL-12. Antigen-specific CD8⁺ T-cell responses were demonstrated in six out of eight patients having sustained clinical responses.⁷⁸ Another study found no clear dose-dependent effect of rhIL-12 on the responses to Melan-A and influenza matrix peptides administered intradermally. However, rhIL-12 was well tolerated at doses of 10 to 100 ng/kg, and 3/24 melanoma patients demonstrated clinical activity.⁷⁹

GM-CSF⁺ Peptide Vaccines

Use of GM-CSF as a vaccine adjuvant is appealing because of its role as one of the primary growth and maturation factors for DCs. A randomized trial compared three different adjuvants in HLA*A0201+ patients with stage III or IV melanoma immunized with tyrosinase and gp100 peptides. 44% and 50% of patients immunized using QS-21 (a purified saponin) and GM-CSF, respectively, developed increased frequencies of CD8⁺ T cells against tyrosinase 370D peptide, compared with 0 of 9 patients immunized using IFA (P=0.045).⁸⁰ Immunization against 3 HLA-A2-binding peptides of the cancer-testis antigen NY-ESO-1, a strongly immunogenic tumor antigen, followed by administration of GM-CSF resulted in 4 of 7 NY-ESO-1 antibody-negative patients developing a specific CD8⁺ T-cell response, and was associated with stabilization or regression of metastases in 5 of 7 cases.⁸¹ There may be some rationale for restricting dose and site of GM-CSF administration. Tumor-secreted GM-CSF has been speculated to actually play an immune suppressive role through increasing numbers of CD34 expressing myeloid suppressor cells.^82,83 In addition, some tumor vaccination studies have demonstrated enhanced tumor growth and immune suppression depending on dose of GM-CSF administered.⁸⁴

Recombinant (Gene-modified), Viral, and Plasmid Vaccines

Gene modified tumor vaccines are commonly designed using autologous melanoma cells that have been transfected with an immunostimulatory gene, such as the expression of an immune enhancing cytokine. Immune stimulation that translates into antitumor activity was demonstrated in a B16 melanoma model, in which irradiated tumor cells expressing murine GM-CSF (and to a lesser extend cells expressing IL-4 and 6) stimulated long-lasting, and specific anti-tumor immunity, requiring both CD4⁺ and CD8⁺ cells.⁸⁵ Stimulation of immunity may have been the result of enhanced antigen presentation by local dendritic cells whose maturation was induced by tumor secreted GMCSF, although this was not formally demonstrated. In a Phase I trial investigating autologous melanoma cells engineered to secrete human GM-CSF, induction of tumor-specific responses were observed. Metastatic lesions were densely infiltrated with T lymphocytes and showed extensive destruction in 11 of 16 patients, which was associated with anti-melanoma CTL and antibody responses. One PR, one mixed response, and three minor responses were achieved, and three patients remained disease-free at 36, 36, and 20 months.⁸⁶ Another approach used mixtures of autologous and allogeneic irradiated melanoma cells secreting IL-6 and sIL6-R. This evoked immune activation and promising clinical results (22% CR⁺PR, 32% SD).⁸⁷ Several factors need to be contemplated when deciding appropriate choice of genes for tumor transfection, for example, ability to stimulate DC maturation, ability to increase immunogenicity of tumor directly, and possibility of synergizing with existing immune responses. Interestingly, in some situations, transfection of tumors with agents that are considered to be immune suppressive, such as interleukin-10, actually evoke anti-tumor immunity (reviewed in⁸⁸).

An alternative approach to immunizing with gene-modified tumors is to locally transfect muscle tissue using recombinant adenovirus vector encoding tumor antigens. For example, intramuscular delivery of gp100 or MART-1 using this approach, either alone or followed by IL-2, resulted in one CR in 16 patients pertaining to the group receiving the recombinant adenovirus MART-1 alone.⁸⁹ Detection of high titers of neutralizing antibodies to the adenoviral vector may explain the relatively low efficacy of this approach. In any case, the low number of patients in this study does not rule out the possibility of expanding on this approach given similar low rates of responses to other immunotherapeutic approaches.

Given that the tumor already expresses a wealth of tumor antigens, studies have been conducted to enhance tumor immunity by inducing a potent local inflammatory response in the tumor site itself. One example of this is Allovectin-7, a plasmid DNA encoding HLA-B7 and beta-2 microglobulin.⁹⁰ When injected into melanoma lesions, it resulted in up to 15% rate of PRs, which occurred even at remote sites. An OS of 21.3 months was achieved with high doses of the bicistronic vaccine, and corresponded to a 12.7 months median duration of responses.⁹¹

Dendritic Cell (DCs) Vaccines

The CD8⁺ cytotoxic T-cell compartment, which is believed to possess major anti-tumor effector function, usually is poorly activated in cancer patients (Fig. 1). One of the postulated mechanisms is the inadequate tumor antigen presentation by DCs. This may be explained by a variety of factors including tumor secretion of factors inhibiting DC function,^92,93 as well as indirect inhibition of DC function by Treg cells,⁹⁴ which are increased in melanoma patients.⁹⁵ Therefore, use of de novo generated dendritic cells has been part of the new strategies to enhance CTL responses (Table 2). The goal of dendritic cell vaccines is to induce a Th1 immune response and to activate CTLs, in order to facilitate tumor elimination.

DCs are capable of processing and presenting peptides derived from tumor protein antigens to CD4⁺ and CD8⁺ T cells, and of regulating the activity of natural killer (NK) cells. Matured from CD14⁺ precursors, ex-vivo DCs are loaded with antigens in the form of whole proteins, tumor lysates, peptides, necrotic and apoptotic bodies, or messenger RNA. Specific epitopes can be pulsed in the form of synthetic HLA-binding peptides, or DNA and RNA sequences carried by viral vectors. Immature DCs are able to process antigens, but it is the mature DCs that are able to fully stimulate T cells by upregulating cytokine secretion, adhesion and costimulatory molecules (Fig. 1).^113–115 Animal experiments with primarily mature DCs consistently showed inducement of tumor-specific CTL responses, and occasional regression of metastases.

In a landmark study using mature DCs pulsed with Mage-3A1 peptide and a recall antigen (tetanus toxoid or tuberculin), Thurner et al¹¹⁶ demonstrated significant expansion of CTL in 8/11 patients with advanced stage IV disease, and objective responses in 6/11 patients. In another trial, 5 clinical responses occurred in 16 metastatic melanoma patients treated with tyrosinase 370D and gp100 210M peptides restricted to HLA class I A*0201. The peptides were modified to increase immunogenicity by altering one aminoacid from the wild type, and were pulsed into DC derived by incubation of plastic-adherent peripheral blood mononuclear cells (PBMC) with IL-4 and GMCSF. Among five patients having a CTL response to gp100 or tyrosinase, four were clinically stable or had tumor regression. Immune responses to gp100 or tyrosinase were demonstrated by gamma intereferon ELISA assay in 31% of patients.⁹⁸ Vaccination with mature DCs loaded with three tumor lysates (M44, SK-MEL 28 and COLO 829), tested in a phase II trial evaluating patients with low volume or in-transit melanoma, produced two clinical responses and 4 stable disease in 33 metastatic melanoma patients. Ten patients showed evidence of enhanced anti-TAA tumor specific CD8⁺ T cells.¹⁰⁹ A correlation between the enhanced immune response and survival was seen with DCs pulsed with MART1, tyrosinase, MAGE3, gp100 and influenza matrix peptides, along with keyhole limpet hemocyan (KLH). Six out of seven patients with immunity to two or fewer antigens progressed, in contrast to only 1/10 patients with immunity to more than 2 antigens.¹¹⁷

NK T-cell activation has also been demonstrated in phase I/II testing of vaccination of patients with metastatic melanoma with intermediate-maturity DCs engineered to express MART-1 through an adenoviral vector. CD8⁺ and/or CD4⁺ MART-1-specific T-cell responses were observed in 6/11 and 2/4, respectively, of the patients being evaluated.¹¹⁸ Objective responses were evident in 5 of 16 patients treated with immature DCs pulsed with tumor lysate or a cocktail of peptides in the presence of GM-CSF and IL-4. In this study, elicitation of delayed type hypersensitivity (DTH) reactions was seen in 11 patients, and recruitment of peptide-specific CTLs was demonstrated.⁹⁶

In a review of 32 clinical trials on dendritic cell vaccines, Engell-Noerregaard et al¹¹⁹ found that clinical response (defined as CR, PR, or SD) was significantly correlated with the use of peptide antigens, use of helper antigen or adjuvant, and induction of tumor antigen-specific T cells. The majority of studies, however, show that despite repeated T-cell activation with antigen-loaded mature DCs, expansion of tumor-antigen-specific immune responses is often transient, and only rarely produces stable disease or regression of tumor metastases.^97,120,121 Although significant expansions of specific CD8⁺ cytotoxic T lymphocytes have been demonstrated for Mage 3A1 in over 70% of patients,¹¹⁶ clinical results with peptide-pulsed DCs remain disappointing, indicating a failure of immunological data to translate into clinical success. Schadendorf et al¹²² actually reported no benefit of peptide-pulsed DC in comparison to dacarbazine (DTIC) treatment. Regulatory T cells (Tregs) may produce a potent down-regulation of anti-tumor responses that counteract the protection conferred by dendritic cell vaccines (Fig. 1). Tumors can also mediate DC suppression, and potential molecular targets are now being identified, including p44/42 MAPK, which is hyperactivated in melanoma, and its upstream activator MEK1/2. Blockade of the MEK1/2-p44/42 axis has been shown to increase IL-12 production and enhance Th1 immune responses.¹²³

Immunomodulatory Monoclonal Antibodies

As the understanding of the dynamic and complex interaction between the immune system and tumors improves, new immunotherapies targeting critical regulatory elements of the immune system can be developed in order to provide treatments with greater specificity and better safety profiles. This includes the recent development of anti-CTLA-4 monoclonal antibodies, Toll-like receptor (TLR) agonists, CD40 agonists, and anti-ganglioside monoclonal antibodies. A family of receptors on T cells serves as a natural braking mechanism for T-cell activation, functioning to reestablish homeostasis following an immune response, and to maintain peripheral self-tolerance. This includes CTLA4 and PD1.¹³⁸

CTLA-4 expression on T cells outcompetes CD28 for binding to CD80 (B7–1) and CD 86 (B7–2), resulting in suppression of T-cell activation and depending on the model assessed, modulation of cytokine production.¹³⁹ In addition, the high affinity binding of CTLA-4 to CD80/86 stimulates the antigen presenting cell to generate high concentrations of the immune suppressive enzyme indola-mine 2,3 deoxygenase.¹⁴⁰ This understanding provided the rationale for development of anti-CTLA4 monoclonal antibodies that could inhibit interaction between B7 and CTLA4, thus releasing the “brakes” against T-cell activation that was hypothesized to enhance antitumor immune response. This was confirmed in animal studies showing that CTLA4 blockade enhanced antitumor T-cell function and inhibited tumor recurrence in murine prostate cancer and melanoma models.^141–143

Ipilimumab

Anti-CTLA4 monotherapy with ipilimumab has produced clinical response rates in 7% to 15% of patients with metastatic melanoma.¹⁴⁴ In one study, the CTLA-4 antibody ipilimumab (formerly MDX-010), produced objective responses in 17% of patients, including 3 patients who demonstrated complete responses (ongoing at 23+, 52+, and 53+ mo).¹⁴⁵ To enhance antitumor response, CTLA4 blockade has been investigated in various dosing regimens as well as in combination with cancer vaccines, standard therapies such as chemotherapy, and IL-2 administration (Table 3). Ipilimumab administered in conjunction with a gp-100 peptide was able to induce a 21% response rate (14% CR).¹⁴⁶ The same antibody alone or in conjunction with dacarbazine produced durable responses in patients with melanoma, some lasting over 1 year.¹⁴⁷ Combination therapy with ipilimumab and IL-2 showed an objective response rate of 22% in patients with metastatic melanoma. An additive, but not synergistic, effect between IL-2 and ipilimumab was observed.¹⁴⁸

Recent studies have also noted cases of late onset objective response among patients who previously experienced stable disease or disease progression.^150,154 This development prompted a suggestion to continue observation or treatment in patients with initial progressive disease or stable disease. Objective antitumor response is often associated with immune related adverse events, most commonly involving the skin and gastrointestinal tract.^145,155 For example, in the study of Sanderson et al¹⁵⁵ involving a multiple peptide vaccine in conjunction with CTLA-4 blockage, only 37.5% of patients with autoimmune effects experienced a relapse, compared to 81.8% of those without autoimmunity.

Other studies suggest that prior cytokine therapy may pose a negative prognostic factor for survival in patients later receiving treatment with anti-CTLA-4 monoclonal antibodies, although differing results have been reported between IFN and IL-2.^145,156

Tremelimumab

Tremelimumab (CP-675206), a fully human immunoglobulin G-2 anti-CTLA-4 mAb, was shown in phase I and II studies to produce durable objective responses in patients with melanoma, ranging from 7% to 14% at dose levels between 0.01 to 15 mg/kg.^151,152 Subsequently, a phase III randomized clinical trial was initiated to compare tremelimumab to standard chemotherapy (dacarbazine or temozolomide) in patients with advanced relapsed or refractory melanoma. Among 655 previously untreated patients, no statistically significant difference in overall survival was observed between the two treatment arms (median OS 11.8 and 10.7 for tremelimumab and chemotherapy, respectively, HR 1.04), however differences in long-term survival are still unknown.¹⁵² In the setting of advanced Stage IV melanoma, CTLA-4 blockade with tremilimumab demonstrated restoration of effector and memory CD4⁺ and CD8⁺ T cells, and induction of transient T-cell resistance to Treg-mediated suppression, which was correlated with clinical outcome.¹⁵⁷

Increased recognition that common genetic variation in drug targets could affect clinical response to CTLA-4 blockade therapy has led to the recent incorporation of pharmacogenetic analysis to evaluate common polymorphisms in the CTLA4 gene and their influence on the response to tremelimumab. Correlation of patient genotype with clinical response has so far demonstrated inconsistent trends^158,159 and further studies are necessary.

Anti-PD-1 mAbs

PD-1 is another inhibitory molecule found on the surface of T cells that is associated with tolerance induction upon binding to its ligands PD-L1 and PD-L2. PD-1 is also expressed by several tumors, including melanoma, where it inhibits antitumor responses and mediates tumor evasion.^160,161 Preclinical studies demonstrate that monoclonal antibodies against PD-1 improve immune functions of tumor-specific T cells, enhance cytokine production, and increase tumor lysis.¹⁶² Phase I trials of MDX-1106/ONO-4538, a fully human anti-PD1 blocking antibody, are ongoing.

CD40 Agonist mAbs

The CD40 cell-surface costimulatory molecule is naturally expressed on dendritic cells, B lymphocytes, monocytes, even solid tumors, and has a broad range of functions. CD40 is upregulated on activated DCs. The engagement with its natural ligand CD154 (CD40L), primarily expressed on activated CD4+ T cells, triggers cytokine secretion and enhanced expression of costimulatory molecules required for efficient T-cell activation.¹⁶³ It has been demonstrated in murine models that in vivo delivery of CD40-activating antibodies overcomes the immunosuppressive mechanism of tumors, increases T-cell activation, and enhances antitumor immunity, leading to regression of established tumors.¹⁶⁴ Other preclinical studies testing CP-870,893, a recombinant human agonist monoclonal antibody against CD40, have also demonstrated enhanced in vitro anti-tumour T-cell responses, evidenced by a significant expansion of IL-2 and IFN-γ-producing cells, as assessed by ELISpot assay.¹⁶⁵ In early phase I trials, CP-870,893 was well tolerated and was able to produce objective tumor response in 27% of melanoma patients.¹⁶⁶

Activation of Innate Immunity (TLR Agonists)

Newer strategies to overcome tumor-induced immune suppression involve attempts to improve presentation of tumor-associated antigens through enhanced expression of co-stimulatory molecules on the surface of DCs.

Toll receptors (TLR) are signaling molecules that recognize conserved molecular patterns on common pathogens, which are able to influence the activity of DCs and induce T-cell responses. TLR-stimulating ligands are being evaluated for their potential to enhance DC activation and heighten antitumor immune responses. Attention has largely focused on TLR-9 agonists, although TLR7/8 ligands also show promise (Table 4).

The use of PF-3512676 (formerly CpG 7909), a synthetic deoxycytidyl-deoxyguanosine oligonucleotide which activates TLR-9, has been associated with a 10% PR and with one response lasting for 13+ months.¹⁶⁸ Clinical response has been associated with the stimulation of NK-cell cytotoxicity, as well as with direct effects on increased activation of DCs, which subsequently induce potent innate immune responses via proinflammatory cytokine secretion, activation of other immune effectors (eg, NK-cells), and increased antigen presentation.¹⁷⁰

TLR7/8 agonists demonstrated in preclinical studies an ability to markedly enhance the antitumor responses though diverse mechanisms involving maturation, activation, and/or migration of critical effector cells, including dendritic cells, B cells, T cells, NK cells, and mast cells. Dendritic cells have been shown to respond to TLR 7/8 agonists by increased secretion of IFN-α, IL-12, TNF-α, as well as upregulation of costimulatory molecules such as CD80 and CD86, increased polarization towards Th1-type responses, and enhanced tumor lysis.^172–174 B-cells, upon treatment with TLR7/8 agonists, are stimulated to increase production of cytokines and antibodies, as well as to upregulate CD80/86 which is essential for T-cell activation. These agonists also act directly on T cells by increased production of Th1-polarizing cytokine.¹⁷⁵ The effects described above may render tumor cells more immunogenic and more susceptible to chemotherapy-induced tumor lysis. Inhibition of angiogenesis and promotion of apoptosis have also been associated with TLR7/8 agonists.¹⁷⁶

Imiquimod, a synthetic TLR7 agonist, when utilized to activate DCs in a group of patients vaccinated with influenza, Melan-A, tyrosinase and NY-ESO peptides, elicited CD8⁺ T-cell responses in 5/8 patients, and one patient in 12 achieved a PR.¹⁷⁷ Other cases have been reported of complete clinical clearance of locally metastatic melanoma when treated with topical 5% imiquimod¹⁷⁸ alone, or in combination with tazarotene cream.¹⁷⁹ These responses are likely mediated by several effects on dendritic cells, ranging from increased recruitment to the skin, enhanced migration to lymph nodes upon antigen uptake,¹⁸⁰ and functional maturation,¹⁸¹ as demonstrated in preclinical studies. Alternatively, imiquimod has been demonstrated to directly increase immunogenic molecule expression of melanoma cells, as well as to induce apoptosis, thus providing an ample supply of local immunogenic proteins.¹⁸²

TLR agonists are able to produce key alterations in the tumor microenvironment, but despite successful induction of Th1 antibody responses and of tumor antigen-specific CD8⁺ T cells, the development of clinically meaningful responses and improved survival have yet to be seen in patients with melanoma being treated with stand-alone TLR agonists. The potential use of TLR agonists as adjuvants in cancer vaccine or adoptive immunotherapy approaches are under ongoing investigation.

Interferons

Immunomodulatory effects of interferons (IFNs), along with the proven activity in metastatic disease, triggered considerable interest for testing this class of agents in the adjuvant setting. The E1684 trial randomized 287 patients between 52 weeks of high-dose interferon (HDI) versus observation. The first results of this trial reported in 1996 showed a significant improvement in RFS (1.72 y vs. 0.98, P=0.002) and OS in the HDI arm compared with observation (3.82 vs. 2.78 y P=0.02), leading to FDA approval of this treatment for stage IIB, IIC and III melanoma. A later update of data at 12.6 years of follow-up still shows a significant RFS advantage for HDI (HR=1.38, P=0.02), but a decrease in the OS benefit (HR=1.22, P=0.18), attributed partially to death from competing causes.²⁰⁸ Despite considerable toxicity, use of HDI in this setting was associated with an improvement in quality of life compared to observation only. The larger E1690 trial, designed using a cure-rate model derived from the results of E1684, employed a randomized comparison of HDI and low-dose interferon (LDI) against observation alone. Again, a significant RFS benefit was recorded for HDI compared with observation (HR=1.28, P=0.025), but no impact was seen with LDI, possibly influenced by the very high salvage rate of relapsed patients in the observation group. Furthermore, neither IFN dose led to any improvement in OS. Similar to E1684, a later follow-up analysis at 6.6 years still revealed a RFS advantage (HR=1.38, P=0.02).²⁰⁸ In a randomized comparison of HDI versus Gm2-klh/Qs-21 (GMK) vaccine (E1694), an initial significant advantage for both RFS and OS with HDI (HR=1.47, P=0.001 and HR=1.52, P=0.009, respectively), was maintained after 2.1 years (HR=1.33, P=0.006, and HR=1.32, P=0.04, respectively). A direct comparison of GMK alone versus GMK with either concurrent or sequential HDI (E2696) demonstrated the superiority of both HDI combinations over the vaccine alone (HR=1.75 and 1.96), reaching significance after adjusting for gender, performance status, time to resection, nodal status and age (P=0.016 and 0.03, respectively). However a survival update at 2.8 years did not confirm the advantage held initially by the two HDI groups.²⁰⁸

Composite data resulting from these four trials support the clinical benefit of HDI, with three trials demonstrating a significant improvement in disease-free survival (DFS), and 2 of them indicating an OS benefit.²⁰⁸ A pooled analysis of E1684 and E1690 data at a median follow-up of 7.2 years shows a significant superiority of HDI over observation in regard to RFS (HR=1.30, P<0.06), but not OS (HR=1.08, P=0.42), which is consistent with the survival data pooled by a large meta-analysis of eight trials comprising 3,178 patients.²⁰⁹ Further studies are necessary to better delineate melanoma prognostic groups and likelihood of response to adjuvant therapy before HDI can be accepted as a universal standard of care.

As discussed above in the sections regarding IL-2 and vaccine therapy, stimulation of anti-melanoma responses is associated in some cases with appearance of autoimmunity. A study by Gogas et al²¹⁰ reported a correlation between the response to HDI therapy and various manifestations of autoimmunity, including antithyroid, antinuclear, anti-DNA, and anticardiolipin autoantibodies, and vitiligo. Development of autoimmunity represented an independent prognostic factor for relapse-free and overall survival (P<0.001). Although lead time bias can conduct to this correlation, similar results with CTLA-4 antibodies indicate an association between autoimmune phenomena (thyroiditis, hypophysitis, enteritis, hepatitis, and dermatitis) and prolonged survival in metastatic disease.^146,155

Alternative interferon schedules have been designed in order to avoid toxicities associated with high-doses or prolonged courses of therapy. No survival benefit was achieved using an induction phase of 10 MU s.c. daily 5 days per week followed by 2 years of LDI.²¹¹ Likewise, two studies employing two and three years of LDI (3 MU s.c. three times a week), had no DFS or OS benefit compared to observation.^212,213 Reduction of the duration of adjuvant HDI to 3 months (20 MU/m² trice weekly) did not produce a median OS benefit (6.6 y for IFN-α 2a and 5.0 y for observation, P=0.40), but a possible improvement was suggested for selected high-risk node-positive patients (OS 4.1 vs. 2.7 y P=0.44). A comparison between one month induction therapy versus one full year of high-dose adjuvant interferon was reported in 364 patients with stage IIB, IIC, and III melanoma treated no later than 56 days of curative surgery with no significant differences in OS and RFS between the regimens of 1 month and 1 year of treatment.²¹⁴

Based on data showing synergistic antitumor efficacy in vitro and in vivo,²¹⁵ a combination of LDI-α2b and low-dose IL-2 was tested in 225 node-negative, pT3 and T4 patients, and demonstrated no superiority in regard to DFS or OS compared to LDI-α2b alone.²¹⁶ Adjuvant treatment for high-risk stages IIA to IIIB with DTIC and low-dose natural interferon-a resulted in a significantly higher 7-year calculated OS rate of 51 versus 30% (P=0.007), with a greater benefit on late mortality, especially in high-risk patients.²¹⁷ However, this data needs to be further confirmed.

Pegylated IFN-alpha-2b (PEG-IFN-alpha-2b) has raised interest for the chronic adjuvant treatment of melanoma given its convenient administration and positive outcomes recorded in the European trials. In a study of 1256 patients with stage III melanoma, PEG-IFN-alpha-2b (n=627) induction with 6 micrograms/kg/wk for 8 weeks followed by maintenance with 3 micrograms/kg/wk was administered against placebo for an intended total duration of 5 years. After 3.8 years of follow-up, the risk for recurrence-free survival (RFS) was reduced by 18% (hazard rate=0.82; P=0.01), along with an expected negative effect on global quality of life score.²¹⁸ The exposure to Peg-IFN alpha-2b appears to be sustained during the long-term adjuvant treatment in melanoma, which is consistent with the European Organization for Research and Treatment of Cancer 18991 data indicating a significant, sustained, relapse-free survival benefit.²¹⁹

High-dose Interferon (HDI) administered neoadjuvantly in doses of 10 MU/m² three times weekly resulted in clinical responses in 55% of patients, and 50% of patients had no recurrence of their disease at over 1.5 years of follow-up. In this patient population of stage IIIB melanoma, clinical responds were associated with larger amounts of tumor infiltrating CD3⁺ and CD11c⁺ lymphocytes.²²⁰ Neoadjuvant biochemotherapy with cisplatin, vinblastine, dacarbazine, interleukin-2 and interferon-alpha 2a produced a 50% response rate and a 65% DFS at 31 months,²²¹ and a phase III intergroup trial comparing a short intensive course of biochemotherapy (IL-2, IFN-α, cisplatin, DTIC and vinblastine) with standard HDI is currently addressing the high-risk stage III patients. Equivalent clinical results with less toxicity than HDI were obtained with a combination of intermediate-dose IFN and Melacine vaccine in stage III melanoma (RFS 31 vs. 25 mo, P=0.85).²²²

Microarray analysis of PBMCs gene induction in vitro may be a useful predictor of the in vivo response of T cells, NK cells or monocytes to treatment with IFN-α. Various genes involved in immune responses, nucleic acid binding and metabolism, protein catabolism, or cyclin-dependent protein kinase activity (OAS2, OASL, HERC5, ISG20, IFI44, LIR7, LGP2, MT1H, MT2A, N4BP1, PLSCR1, USP18, TREX1, ZCCHC2) are activated in both settings, therefore being potential markers of patient response to interferon.²²³

Vaccines

Several melanoma-specific vaccines have been prepared from whole cells,^224–226 or from antigens shed from allogeneic cell lines.²²⁷ A prolongation of survival in comparison to historical controls or to patients not developing an immune response to the vaccine was reported in the initial studies. However, randomized phase III studies did not confirm a significant benefit.

In a 689 patients trial comparing an allogeneic melanoma cell lysate vaccine, Melacine, to observation in patients with stage II disease, a similar 5-year disease-free survival was reported (65% for the vaccine and 63% for observation arm).²²⁸ An OS analysis could only be performed in a follow-up study, which showed a survival advantage for the HLA-A2⁺ and/or C3⁺ group, but no significant benefit for the overall cohort of patients.²²⁹

Canvaxin is an allogeneic whole cell vaccine that was tested in double blind trials of post resection patients with high probability of relapse. Initial nonrandomized studies demonstrated promising results.²³⁰ 2602 AJCC stage III melanoma patients who underwent lymphadenectomy were enrolled in a study where 935 received Canvaxin vaccine between 1984 and 1998, while 1667 historical controls did not. A significantly higher median OS and 5-year OS was observed in patients receiving the vaccine (56.4 vs. 31.9 mo and 49% vs. 37%, respectively; P=0.0001). The difference was maintained when patients treated with Canvaxin were matched with non-PV patients by six covariates forming 739 pairs.²³¹ Another study involving 150 stage IV melanoma patients treated adjuvantly with Canvaxin after surgical resection demonstrated induction of delayed-type hypersensitivity (DTH), which correlated with the OS (39% for vaccinated and 19% for non-vaccinated patients) (P=0.0001). In a multivariate analysis model, vaccine therapy was the most significant prognostic variable (P=0.0001).²³² However, subsequent prospective randomized trials showed less favorable results, with a trend towards decreased survival in treated patients. Among 1656 patients with resected stage III and IV melanoma, Canvaxin failed to demonstrate improvements in overall survival (OS), as compared to BCG.²³³

Similarly, a polyvalent, shed-antigen Bystryn vaccine did not produce a significant statistical improvement in median OS, even after adjusting for risk factors. Thirty-eight patients with stage III melanoma with a particularly poor prognosis were immunized intradermally in a 2:1 vaccine/placebo ratio every 3 weeks×4, monthly×3, every 3 months×2, and then every 6 months for 5 years or until disease progression. At 2.5 years median length of observation, the median time to disease progression was 2.5 times longer in the active arm (P=0.03). Median overall survival was 40% longer in the active treatment group (3.8 vs. 2.7 y, P=nonsignificant).²³⁴

Interest in targeting more specific surface antigens has resulted in vaccines derived from GM2 gangliosides, a well-defined melanoma associated antigen. Anti-GM2 antibodies have been detected in approximately 5% of melanoma patients, and their presence was associated with an increase in the relapse-free survival (RFS).²³⁵ An IgM antibody response was obtained in 85% of patients with resected stage III melanoma who were immunized with purified GM2 adherent to BCG. An improved DFS was observed for patients with anti-GM-2 titers >1:40, but such a benefit could not be confirmed in a subsequent randomized phase III trial comparing GM2+BCG vaccine with BCG alone.²³⁶ However, when the six patients who produced GM2 antibodies before randomization were excluded, an increased DFS of 23% (P=0.02) and a trend toward longer OS were observed for the GM2+BCG. Another phase III randomized trial comparing the efficacy of high-dose interferon alfa-2b therapy (HDI) versus vaccination with GM2 (GMK) demonstrated an overall benefit for HDI in terms of RFS and OS in melanoma patients. Antibody responses to GM2, however, were associated with a trend toward improved RFS and OS.²³⁷

However, no survival benefit was noted in two large phase III trials which used adjuvant GM2-KLH with QS21 adjuvant. A negative effect on survival might have been observed in the E1694 trial, where 880 patients with resected stage IIB and III melanoma have been randomized between GM2-KLH vaccine and HD IFNa-2b. Lower RFS (relapse-free survival) and OS (overall survival) were seen in the GM2-KLH arm.²³⁷ A negative effect on the rate of development of distant metastases and a lower OS (HR 1.57, P=0.03) were observed in an European study European Organization for Research and Treatment of Cancer 18961 involving 1314 patients with resected stage II (T3-4N0M0) disease who were randomized between the GM2-KLH vaccine and placebo, although the disease-free survival was similar in the two groups.²³⁸ Both studies were closed prematurely because of the apparent detrimental effect on survival of the GM2-KLH vaccine.

The most immunogenic peptides among 12 cancertestis and melanocyte differentiation proteins tested are tyrosinase, gp100, MAGE-A1, and MAGE-A10, and it was determined that administration of multiple peptides is safe and immunogenic.²³⁹ An ongoing Intergroup trial (E4697) is currently testing vaccination with HLA-A2-restricted peptides (tyrosinase, gp210M, and MART-1), with GM-CSF used either alone (in HLA-A2- patients), or added to the peptide vaccine for potential synergy.

Anti-ganglioside Monoclonal Antibodies

A humoral response can be generated against the monoclonal antibody active sites. Anti-GD3 antibodies were induced in 3 of 14 patients immunized with BEC2, an anti-idiotypic monoclonal antibody that mimics GD3 and BCG, but the immunogenicity of anti-idiotypic monoclonal antibodies has generally remained low. A 71% survival and a 64% disease-free survival at 2.4 years of follow-up were considered to be encouraging results.²⁴⁰

TLR Agonists

In a study involving 24 patients with stage I to III melanoma, local administration of PF-3512676 (formerly CpG 7909), a synthetic deoxycytidyl-deoxyguanosine oligonucleotide which activates TLR-9, has induced melanoma-specific CD8⁺ T-cell responses against at least one melanoma-associated antigen in the draining lymph node in 50% of the patients receiving the compound, vs. none of the control patients injected with saline (P=0.01). Clinical response has been associated with enhanced local and systemic melanoma-specific CD8⁺ T-cell reactivity and NK cell mobilization.²⁴¹ Intradermally injected PF-3512676 appears promising as an adjuvant therapy for early-stage melanoma, being associated with increased dendritic cell activation, enhanced type I cytokine secretion, and a significant reduction in CD4⁺CD25⁺ Tregs in sentinal lymph nodes.²⁴²

The authors like to thank Famela Ramos for assistance in literature collection and critical review of the manuscript.