Immmunotherapy of cancer: from vision to standard clinical practice

Abstract

Until the end of the 19th century the possibility that a tumor could be rejected merely by the body’s immune defense was no more than a vision. After more than 100 years of preclinical and clinical research in the field, the vision of cancer immunotherapy became real and has, with multiple tools, successfully entered clinical standard practice. Non-specific mediators of immune defense, such as BCG for treatment of superficial bladder cancer or interferon-α for treatment of chronic myelogenous leukemia and hairy cell leukemia, can induce durable remissions. In particular, antigen-specific mediators of immune defense represent promising agents for targeted cancer therapies. Antibody-based treatment of B-cell lymphomas and breast cancer has dramatically improved disease response without major toxicity. A large number of new antibodies targeting different epitopes on a variety of malignant cells are now approaching clinical approval. Allogeneic stem cell transplantation for treatment of leukemia and certain cancers represents the first T cell-based adoptive immunotherapy with large-scale clinical application. Experimental T cell-based immunotherapies with promising clinical perspectives include tumor vaccines, adoptive transfer of autologous tumor-specific effector cells and the genetic transfer of tumor-specific T cell receptors into the patient’s lymphocytes. These facts demonstrate that immunotherapy has now been established as the fourth column of cancer therapy besides surgery, radiotherapy, and chemotherapy. On the basis of its already proven efficacy, the usually favorable toxicity profiles and the development perspectives of the experimental approaches a further and more rapid expansion can be expected

Introduction

The immune response is the body’s way of defending itself against disease. Immune defense against pathogens and tumors is mediated through antigen-specific and nonspecific immune mechanisms (Fig. 1). Non-specific immune responses are procured by cells of the macrophage and NK cell lineage and/or by soluble factors such as inflammatory cytokines. Nonspecific immunostimulation with bacteria or cytokines have both been applied for treatment of cancer. The functioning of the antigen-specific immune system is based on a division of labor between T cells and antibody-producing B cells [1]. Adoptive transfer of antibodies or T cells as well as cancer vaccines represent tools for cancer immunotherapy in the clinics. This article aims to review basic aspects of tumor immunology and its translation into the clinics.

Fig. 1.

Mediators of immune defense and immunotherapeutic strategies

Tumor therapy with non-specific mediators of immune defense

Non-specific immunostimulation with bacteria or bacterial components

Non-specific stimulation by live bacteria was the first immunotherapeutic approach to demonstrate antineoplastic efficacy in the clinics. Already in the early 1890s, the American surgeon Dr. Coley noticed that certain cancer patients experienced regression of their tumors when they contracted acute bacterial infections. Based upon this observation Coley than injected live bacteria into a patient with advanced cancer and subsequently went on to develop a safe and more effective mixture of bacteria for treatment of cancer patients [2]. Starting from this observation, the field has evolved during the subsequent 100 years into a better understanding of the mechanisms of tumor rejection induced by non-specific immunostimulants [3] and the development of the first standard clinical practice. Intravesical instillation of live Bacillus Calmette-Guerin (BCG) now represents the standard treatment for certain stages of superficial bladder cancer [4]. The limited efficacy and significant toxicity of this approach has led to the exploration of the immunostimulatory potential of bacterial components or genetically modified bacteria. Among the first non-specific immunostimulation with CpG oligodeoxynucleotides has attracted particular attention [5].

Recombinant cytokines

With the discovery of interferons in the mid 1950s, the vast research field of inflammatory cytokines has been opened [6]. Cytokines not only control resistance against infectious pathogens but, depending on the cytokine and tumor type, can either inhibit or enhance tumor growth. Interferons (IFN) display antineoplastic efficacy in vitro and in vivo [7] and interferon-receptor knockout mice demonstrate enhanced tumor development [8]. IFNs mediate their antineoplastic activity by direct action and indirectly by modulating a variety of host immune as well as host stroma cell functions. Antitumor effects of IFNs in vivo seem to involve both direct and indirect mechanisms including upregulation of HLA antigens, enhancement of cellular cytotoxicity, and antiangiogenic activity. On the other hand, immunosuppressive cytokines such as TGF-β and IL-10 are constitutively expressed in many tumors and can mediate their escape from immunosurveillance [9].

In the early 1980s, after cloning of multiple cytokine genes and the introduction of production of proteins by recombinant DNA technology [10], large-scale clinical testing of cytokines became feasible. This, together with promising preclinical results of cytokine treatment in various experimental animal tumors, has subsequently precipitated numerous and, in part, highly redundant clinical cancer trials. In the following two decades, IFN-α, -β, -γ, tumor necrosis factor (TNF)-α, the interleukins IL-2, IL-3, IL-4, IL-6, IL-12, granulocyte-monocyte stimulating factor (GM-CSF), and other cytokines were studied with great expectations. In spite of enormous efforts beneficial responses were rare, limited to a few diseases, and associated with significant side effects [11]. Interferon-alpha can induce complete remissions in patients with hairy cell leukemia and chronic myelogenous leukemia. With the exception of a small subset of patients with advanced renal cell carcinoma and malignant melanoma responding to IFN-α or interleukin-2 treatment, most solid tumors, high-risk leukemias and lymphomas are largely resistant to administration of cytokines.

Tumor therapy with antigen-specific mediators of immune defense

T cell-based immunotherapy of cancer

Basic aspects

In the 1970s, it was thought that most spontaneous tumors in animal and man, in contrast to chemically or virus-induced tumors, are non-immunogenic. It was also believed that antibody responses dominate the host’s immunity against immunogenic tumors. Subsequent studies disproved both dogmata. They demonstrated that the lack of immunogenicity of most tumors was not caused by the absence of tumor antigens (TA) but rather by the deficiency of tumor cells to activate the immune system [12]. They also revealed that T cell-mediated cellular and not humoral immune responses are crucial for immunosurveillance against cancer. This view was based upon the observation of an increased frequency of tumors in congenital and acquired cellular immunodeficiency syndromes [13, 14], the identification of the tumor-antigens (TA) recognized by cytotoxic T cells (CTL) [15, 16], and a positive correlation of the extent of intratumoral T cell infiltration with disease prognosis [17].

A very complex network of interacting cells and molecules mediates cellular immune responses against infectious pathogens and cancer. Specificity of the system is determined by the interaction of antigen binding receptors (TCR) on tumor-specific T cells. The latter are presented to the TCR as peptides assembled into the antigen-presenting groove of HLA-I or HLA-II antigens on the surface of tumor cell and/or antigen presenting dendritic cells (DC). T cells represent the only immune mechanism which can see inside a cancer cell (Fig. 2). This is important because most cancer specific proteins, including those determining the malignant phenotype, are not expressed on the cell surface and not accessible to antibodies. Subsequent to fragmentation of tumor proteins in the proteasome the resulting peptides associate with HLA-I and ß2-microglobulin in the endoplasmatic reticulum. This peptide-HLA-ß2-microglobulin complex is than stable enough to integrate into the cell surface where it can be recognized by TA-specific CTLs.

Fig. 2.

T cell can see inside a cancer cell

Since the early 1980s, both ourselves and others have identified a rapidly increasing number of tumor antigens (TA) by genetic or biochemical approaches. For further details the interested reader is referred to: http://www.cancerimmunity.org/peptidedatabase/Tcellepitopes.htm or http://www.licr.org/SEREX.html.

The genetic approach transduces eukaryotic cells or bacteria with tumor cDNA libraries. Transfectants are subsequently screened with autologous tumor-specific T cells [15, 16] or patients’ IgG antibodies, respectively [18]. The biochemical approach uses mass spectrography to characterize T cell stimulatory peptides eluted from HLA molecules on the surface of tumor cells [19]. More recently, both ourselves and others have used subtractive libraries and virtual cloning procedures [20] to identify selectively expressed cancer genes, which might code for attractive targets for immune intervention. According to their expression patterns and the similarity to self-proteins TAs can be grouped into individual and shared antigens [21] (Fig. 3). Individual TAs are tumor-specific and derived from mutated genes frequently related to the process of oncogenesis. Mut-CDK4 represents the prototype of such a TA [16]. Individual TAs represent neoantigens presumably not subjected to self-tolerance. This advantage is counterbalanced by the large heterogeneity of individual TAs, which limits their utility as universal targets for cancer immunotherapy. Shared TAs are either tumor-specific or tissue-restricted. Shared tumor-specific TAs are either derived from frequently mutated genes such as mutant p53 or from transforming viruses such as HPV-16. Tissue-restricted TAs are either coded for by so-called cancer-testis genes (CT), which due to promotor hypomethylation are selectively expressed in germ and cancer cells. The other group of tissue-restricted TAs are differentiation antigens. Melanocyte specific proteins such as tyrosinase or CEA and PSA represent examples of this group. Tissue-restricted TAs are structurally intact self-proteins and therefore controlled by tolerance mechanisms. This and the danger of induction of autoimmune processes make these TAs less attractive candidates for cancer immunotherapy. Although multiple TAs are expressed on most tumor cells, their hierarchy and relevance for prognosis, spontaneous tumor rejection, and immunotherapy of cancer still remain to be determined in clinical studies.

Fig. 3.

Human tumor antigens

The different stages and response patterns of a vaccination-induced tumor rejection are now increasingly well-understood [22]. Injection of TAs plus adjuvants into the skin leads to local inflammatory responses with uptake of antigen and maturation of DC (Fig. 4.). Ag-loaded DCs migrate to the regional lymph nodes and present TAs to naive and central memory T cells. Ag-experienced peripheral effectors or memory T cells than enter the circulation and eventually establish a peripheral response at the tumor site. This response can either lead to tumor rejection or persistence. Reasons for a failure of a host immune system to reject a tumor after vaccination are multifold. They may involve the use of a wrong TA, immune escape due to loss of TA or its processing and presentation, release of immunosuppressive cytokines by tumor cells or indifference of the stroma. Responses set in motion by vaccination with tumor peptides are highly individual and hit a heterogeneous and highly movable target (Fig. 5). Most tumors are heterogeneous with respect to the expression patterns of different TAs. Some sub-clones may express TA₁, others TA₂, a third TA₁+TA₂, and others none. It is likely that only individuals who can mount T cell responses which are adequate to set in motion a multitude of secondary responses, including amplification of pre-existing tumor responses directed against other TAs and modulation of stroma cell functions, can lead to partial or complete tumor rejection. Needless to say, selection pressure by TA-specific effector cells on a genetically unstable tumor target will favor the survival and expansion of TA-negative subclones.

Fig. 4.

Stages of T cell-mediated tumor rejection (modified from [22])

Fig. 5.

Response patterns of T cell mediated tumor rejection (modified from [22])

In parallel with our improved understanding of the molecular basis of T cell responses, the makings of TAs, and modes of tumor rejection, three T cell-based cancer immunotherapies have successfully entered the clinics. They are: (i) allogeneic stem cell transplantation and donor lymphocyte infusion; (ii) adoptive transfer of specific T cells or TCRs; and (iii) tumor vaccination. The three treatment modalities will be discussed in the following paragraphs.

Allogeneic stem cell transplantation and donor lymphocyte infusion as adoptive immunotherapy.

The original concept of allogeneic stem cell transplantation (allo-SCT) in the 1960s and 1970s was to restore normal hemopoiesis after myeloablation with high-dose chemotherapy and total body irradiation. In the subsequent two decades it was learned that immune-mediated graft-versus-leukemia (GvL) effects induced by histocompatibility differences between donor and recipient are crucial for the antileukemic efficacy of allo-SCT. GvL effects are eliminated by the use of syngeneic grafts from identical twins [23] and by elimination of mature T cells from allogeneic stem cell transplant [24]. More recently it was shown that a simple infusion of donor lymphocytes (DLI) in patients with CML and to a lesser extent with other hemopoetic malignancies could induce durable remissions in CML patients relapsing after allo-SCT [25]. Allo-SCT does not only mediate GvL effects but at least in renal cell carcinoma also displays a graft-versus-tumor (GvT) effect [26].

The recognition by donor T cells of minor histocompatibility antigens with expression restricted to hemopoetic tissues and leukemic cells represents one mechanism of antileukemic activity of allo-SCT [27]. It also appears that donor NK cells activated through “intelligent” HLA-CI mismatches contribute to the GvL reactivity [28]. Clinical manifestations and modes of action of GvL effects frequently overlap graft-versus-host disease (GvH-D) [23, 26, 27]. To reduce the unfavorable side effects of allo-SCT, we, and others, attempted to selectively eliminate GvH reactive lymphocytes while leaving leukemia-and pathogen-reactive immune cells intact. Targeting activation-induced surface structures of alloantigen-stimulated donor T cells represent promising steps in this direction [29, 30]. Allo-SCT and DLI, with the increasingly well-understood mechanisms underlying the GvL/GvT phenomena, can now be considered as the first form of adoptive immunotherapy with proven clinical efficacy.

Adoptive transfer of tumor-reactive T cell clones or T cell receptors (TCR)

Adoptive immunotherapy with large numbers of autologous TA-specific effector cells is a straightforward concept, which has been investigated both in infectious (e.g., cytomegalovirus, Epstein-Barr virus, HIV) and malignant disease [31]. Although evidence of migration of the transferred tumor-peptide specific CTLs to tumor sites and regression of individual metastases was observed in melanoma patients [31] the basic limitations to generate sufficient numbers of effector cells in each patient has still not been solved [32]. Perhaps making room for T cells by a non-myeloablative conditioning regimen prior to the adoptive transfer of tumor-specific CTLs increases the probability of success for this new form of cellular therapy [33].

In an attempt to circumvent the problems of generating large numbers of tumor effector cells on an individual basis, we have started to develop TCRs specific for universal TAs for genetic transfer into T cells from cancer patients [34]. The approach taken to generate TCRs with specificity for universal TAs was to bypass tolerance to self-peptides by immunizing HLA-A*0201 transgenic mice. The human oncoproteins p53, mdm-2, and others were used as immunogens. Peptide-specific CTL clones recognizing human peptides in context with HLA*0201 were than isolated. The α-and β-chains of TCRs were cloned and subsequently chimerized and grafted to human T cells by retroviral gene transfer. Human T cells grafted with these TCRs were capable to specifically kill human HLA-A*0201 positive cells, which at the same time also expressed the relevant peptide. On this basis, a cellular therapy concept to redirect ineffective peptide responses in cancer patients by genetic transfer of specific TCRs was formulated. Preclinical development of this concept is advanced and clinical testing is planned in the near future.

Tumor vaccination

Over many decades’, physicians only vaguely understanding the mechanisms involved in specific defense have applied undefined TAs such as crude cellular vaccines for prevention of relapse or treatment of malignant disease. The assessment of efficacy in the early trials was solely based upon clinical end points. From these approaches very little about the clinical utility of tumor vaccines and their immunologic consequences was learned in the past. The availability of well-defined TAs in the 1990s opened the doors for a more rational development. Using defined TAs, the assessment of correlations between clinical endpoints and TA-specific immunologic surrogate parameters became possible. To measure the frequency of tumor-peptide-specific T cells prior and during immune interventions we, and others, have developed immune monitoring techniques such as the computer-assisted video image ELISPOT assay and tetramer-technologies, which are now routinely applied in tumor vaccination trials [35, 36].

Preventive vaccination against infectious disease or experimental animal tumors is much more effective than the therapeutic vaccination of established and advanced disease. Similar rules seem to control vaccination responses against human tumors. Vaccination with human papilloma virus 16 (HPV-16) virus-like-particles in a recent double blind study effectively prevented persistent HPV-16 infections as well as the establishment of HPV-16-related cervical intraepithelial neoplasia [37]. As with all new cancer therapeutics almost all studies with defined TAs were performed as therapeutic vaccines in advanced disease stages. This must be kept in mind when discussing the efficacy of the early vaccination studies. For further details the interested reader is referred to http://clinicaltrials.gov/ct/search?term=cancer+vaccines&submit=search or http://c-imt.org. The Brussels Branch of the Ludwig Institute of Cancer Research (LIC) under T. Boon pioneered the clinical testing of the first defined TA vaccines [38]. Most of the LIC and other institutions vaccination studies used melanoma peptides of the cancer-antigen or the differentiation antigen families. For vaccination all formats of TAs from peptides to peptide-loaded DCs, mRNAs or DNAs were applied in parallel. In spite of other statements, it is still not clear which of these formats is ideal, and comparative clinical studies are urgently required. The early vaccine trials, at least in the melanoma setting, have shown clinical responses in a relatively small minority of patients with advanced disease and displayed mild side effects [39]. When peptides derived from melanocyte differentiation antigens were applied in melanoma vaccination studies, induction of autoimmune phenomena such as vitiligo was only rarely observed. So far, immunomonitoring has failed to disclose a strong correlation between induction of clinical response surrogate immune parameters. This is unfortunate, because in view of the multiple variables involved, a surrogate parameter-based development strategy is urgently required. Prominent and still open questions in the field of therapeutic tumor vaccination relate to the choice of optimum TAs and their ability to induce tumor rejection, to their efficacy in non-melanoma indications, and optimum formats, doses, and schedules.

Antibody based immunotherapy of cancer

Although frequently present on human cancer cells, most of the widely expressed TAs are self-proteins and therefore weakly immunogenic. Crossing the species barrier represents the most effective means to break self-tolerance. Mice immunized with human tumor proteins recognize multiple non-homologous human epitopes as foreign and generate strong and qualitatively unique B-and T-cell responses which, due to self-tolerance mechanisms, cannot be achieved within the homologous species.

Using this approach, the first murine monoclonal antibodies (MCAB) were generated and were subsequently chimerized or humanized to reduce their immunogenicity. Approximately 25 years after Köhler and Milstein’s famous description of the hybridoma technique [40], the first MCABs have successfully entered the clinics for the treatment of breast cancer or B-cell lymphomas, respectively. Rituximab, recognizing the B cell antigen CD20, and herceptin, an anti-HER-2/neu antibody, were the first to demonstrate single agent activity in CD20⁺ B cell lymphomas or HER-2/neu⁺ breast cancer. They can also potentiate the effect of chemotherapy [41, 42]. For the first time, it was shown that antibodies could break chemotherapy-refractory states in most advanced lymphoma or cancer and are capable of inducing durable remissions in patient groups with otherwise very poor prognoses. On the basis of their efficacy, it is not surprising that antibodies have become the most rapidly expanding class of pharmaceuticals for treating neoplastic disease. Four years after approval in the US, the world wide annual sales of these two drugs have exceeded 4 billion Euros. Naked antibodies were shown to be effective cancer agents by still not completely understood effector mechanisms involving Fc region-dependent processes [43] as well as interference with intracellular signaling of targeted surface structures [44] . Radiolabeling or conjugation with toxins further improved efficacy of monoclonal antibodies. Mylotarg, an anti-CD33-immunotoxin, was the first agent of this type to successfully enter clinical practice. Mylotarg has been approved for treatment of CD-33⁺ AML of the elderly on the basis of comparable antileukemic efficacy to chemotherapy and is associated with fewer side effects [45]. More than 100 new therapeutic antibodies targeting different epitopes, such as the EGF-receptor [46], and multiple modifications of the antibody structure [47], promise to improve antineoplastic efficacy and are now in different phases of their preclinical or clinical testing.

Abbreviations

CT

Cancer-testis genes

CTL

Cytotoxic T lymphocytes

DC

Dendritic cells

DLI

Donor lymphocyte infusion

GvL

Graft versus leukemia

GvT

Graft versus tumor

CTL

Cytotoxic T lymphocytes

MCAB

Monoclonal antibodies

MHC-C I

Major histocompatibility complex class I antigens

MHC-C II

Major histocompatibility complex class II antigens

TA

Tumor antigen

TCR

T cell receptor