Journal list menu

Volume 6, Issue 2 p. 242-250
Review
Open Access

Cancer immunotherapy

Thomas F. Gajewski

Corresponding Author

Thomas F. Gajewski

University of Chicago, 5841 S. Maryland Ave., MC2115, Chicago, IL 60637, USA

Tel.: +1 773 702 4601; fax: +1 773 702 3163.Search for more papers by this author
First published: 08 January 2012
Citations: 67

Abstract

The remarkable specificity of the immune system through antigen recognition has long attracted investigators to the possibility of immune-based therapy for cancer. Previous cancer immunotherapeutics had been restricted to non-specific immunomodulatory agents, such as the cytokines IL-2 or IFN-α. However, the molecular definition of cancer-associated antigens introduced the possibility of specific vaccines and adoptive T cell approaches aiming to target the tumor cells more specifically. The recent introduction of total exome sequencing has enabled the identification of patient tumor-specific epitopes generated through somatic point mutations, raising the possibility of targeting tumor antigens in individual patients which are even more tumor-specific. Transcriptional profiling and immunohistochemistry analyses have revealed a subset of patients with a pre-existing T cell-inflamed tumor microenvironment. This phenotype may be predictive of clinical outcome to immunotherapies and offers the possibility of a predictive biomarker. Further analysis of these tumors has identified a set of defined immune suppressive factors which themselves are being targeted with new immunotherapeutics, already with interesting early phase clinical trial results. Understanding not only the expression of tumor antigens but also the dynamic between a growing tumor and the host immune response is thus generating a rich set of opportunities for the specific immunotherapy of cancer.

1 Dynamic between growing tumor and host immune response

How to develop and optimize immunotherapeutic interventions might best be facilitated through a greater understanding of the dynamic interplay between a tumor and the host response and the mechanisms of immune escape that allow the cancer to develop in the face of immune competence (Figure 1). The fact that tumors arise from transformation of normal cells had generated the notion that lack of immune-mediated elimination of non-virus-associated cancers might have been due to immune tolerance against self antigens. However, the molecular characterization of cancer-associated antigens has suggested that the vast majority of cancers do indeed express antigens that are capable of being recognized by the host immune system. The molecular mechanisms mediating neoantigen expression are multiple and have been reviewed (Van Pel et al., 1995; Kawakami and Rosenberg, 1996; Rosenberg, 1996), but a lack of antigens does not appear to be the reason for failed spontaneous immune rejection of cancers.

Details are in the caption following the image
Model for dynamic interaction between a cancer and the host immune response. Some patients show a spontaneous immune response against their tumor. Recent evidence suggests that a pre-existing T cell-inflamed tumor microenvironment may be a predictive biomarker for response to immunotherapies. Preclinical data indicates that a spontaneous anti-tumor T cell response is initiated when still undefined tumor-derived factors induced type I IFN production by host DCs. This facilitates cross-presentation of tumor antigens to CD8+ T cells, which promotes differentiation of naïve cells (nCD8) into effector cells (eCD8). Under the influence of chemokines and other inflammatory signals, effector T cells can migrate into the tumor microenvironment and gain access to tumor cells. The tumor microenvironment also can be regulated by immune suppressive mechanisms, including PD-L1, IDO, Tregs, and anergy. Knowledge of these processes has facilitated the rational development of new immunotherapeutic approaches for cancer.

If antigens are indeed present, then a next-level barrier to effective spontaneous immunity might be immunologic ignorance, as without the involvement of a pathogen for most cancers, innate immune activation might be minimal and therefore an adaptive immune response might not be primed. However, this assumption has recently been called into question based on gene expression profiling and immunohistochemical analysis of tumor specimens from a variety of cancers. In melanoma, transcriptional profiling and confirmatory studies have suggested that around 30–40% of patients show spontaneous inflammation of the tumor that includes the presence of CD8+ T cells (Gajewski et al., 2007, 2009, 2010; Harlin et al., 2009; Louahed et al., 2008; Vansteenkiste et al., 2008). In HLA-A2+ patients, analysis using peptide/HLA multimers has confirmed the presence of tumor antigen-specific T cells among this population (Harlin et al., 2006; Mortarini et al., 2003; Appay et al., 2006), arguing that priming and differentiation of anti-tumor T cells has occurred spontaneously. In ovarian cancer, patients with T cells infiltrating tumor have been shown to have improved outcome, arguing for prognostic significance of a natural immune response (Zhang et al., 2003). A major subset of colorectal patients also has been found to have spontaneous infiltration of tumor with activated CD8+ T cells (Galon et al., 2006). Interestingly, in stage I–III patients, the presence of an effector memory CD8+ T cell population infiltrating the tumor has been found to be more prognostic than stage by TNM classification (Mlecnik et al., 2011), suggesting that integration of host immune response information might be developed to improve the accuracy of cancer staging. A T cell-inflamed tumor microenvironment also has been observed in a subset of patients with non-small cell lung cancer (Vansteenkiste et al., 2008). Together, these results clearly show that immunologic ignorance is not the dominant limiting factor in at least a major subset of cancer patients. However, it may explain the failure of tumor elimination in the subset of patients with tumors that lack a T cell infiltrate. Categorizing a patient's tumor based on the presence or absence of an adaptive immune-inflamed phenotype may allow better selection of immunotherapies for individual cases based on the mechanism of failed spontaneous immune elimination in each instance.

Interestingly, the T cell-inflamed tumor microenvironment may have predictive value for response to immunotherapeutics. In melanoma, clinical responses to several tumor antigen vaccine platforms have been associated with a transcriptional signature indicative of an inflamed tumor microenvironment (Gajewski et al., 2010; Louahed et al., 2008). This signature includes chemokines that appear to mediate trafficking of effector T cells into tumor sites (Harlin et al., 2009). Recent presentations in abstract form have suggested that a similar gene expression pattern associated with clinical benefit from the anti-CTLA-4 mAb ipilimumab and also from IL-2 (Sullivan et al., 2009). Thus, this transcriptional signature may ultimately be useful for patient selection for the currently available immunotherapeutics, and is being tested prospectively in the multi-center vaccine trials being directed by GSK-Bio. Biologically, these data imply that non-inflamed tumors may be resistant to immune therapies, likely because of a lack of chemokines and other signals required for T cell homing into tumor sites.

The underlying mechanisms by which some patients appear to generate a spontaneous T cell response against their cancer are beginning to be elucidated. Interrogation of the metastatic melanoma gene expression data for innate immune signals that might correlate with the presence of a T cell infiltrate revealed the presence of a type I IFN transcriptional signature (Fuertes et al., 2011). Preclinical mechanistic experiments revealed that host type I IFN signaling was necessary upstream for a spontaneous CD8+ T cell response against tumor antigens in mice (Fuertes et al., 2011; Diamond et al., 2011). The mechanism of this effect was predominantly through the action of host type I IFNs on CD8α+ dendritic cells (DCs), the subset involved in cross-presentation of antigen to CD8+ T cells. Therefore, one potential reason for inter-patient heterogeneity in the spontaneous generation of an anti-tumor T cell response might be through variation at the level of the type I IFN pathway, either upstream at the level of type I IFN production, or downstream at the level of type I IFN sensitivity. The tumor-derived factors that trigger DCs to produce type I IFNs and thus bridge to an adaptive immune response are being elucidated.

2 Strategies to increase frequencies of tumor antigen-specific T cells

Whether a natural immune response has occurred against the tumor or not, a major goal therapeutically has been to increase the frequency of tumor antigen-specific T cells in order to tip the balance in favor of immune-mediated tumor control. The two major strategies pursued have been the administration of antigen-specific vaccines, or adoptive transfer of anti-tumor T cells.

2.1 Cancer vaccines

Numerous platforms have been developed and evaluated clinically to induce immune responses against tumor-associated antigens. While whole cell-based vaccines have represented the first major strategy pursued, such approaches do not take advantage of the opportunity for more focused antigen specificity, and it is not clear if the use of whole cells is desirable. The Canvaxin platform based on pooled allogeneic melanoma cell lines has suggested evidence of worse patient outcome in a randomized phase III trial. One hypothesis to explain a negative outcome is that induction of antibody responses against unknown surface receptors may paradoxically support tumor growth. Focus on defined antigens, therefore, has received the greatest level of investigation, which also allows for precise immunologic monitoring. Platforms based on single or multiple defined tumor antigens include recombinant viral or bacterial vectors, peptides or proteins prepared in a variety of immunologic adjuvants, and specific antigens or coding sequences of RNA or DNA loaded onto autologous antigen-presenting cells (e.g. DCs) which are re-administered to the patient (Zarour and Kirkwood, 2003). While a majority of such early phase clinical trials has been carried out in patients with melanoma, a broad array of patients with other cancer types has also been studied. Perhaps surprisingly, the first FDA-approved therapeutic cancer vaccine is Provenge, which has been developed for patients with castrate-resistant prostate cancer (Kantoff et al., 2010). The technology involves loading a fusion protein between GM-CSF and the antigen prostatic acid phosphatase (PAP) onto autologous peripheral blood cells obtained by leukapheresis. It is thought that the GM-CSF moiety favors delivery to DCs, and also contributes to DC activation. The product is then administered intravenously, whereby it is thought to induce a PAP-specific T cell response that in turn can attack prostate cancer tumor cells. The phase III clinical trial showed an improvement in median overall survival, from 21.7 months to 25.8 months (Kantoff et al., 2010), which is a modest but clinically meaningful benefit for this patient population.

Several other vaccination approaches are worth describing for illustrative purposes. As the major goal of most vaccines is to induce tumor antigen-specific CD8+ cytolytic T cells, a major strategy has been to utilize specific class I MHC-binding tumor antigen peptides. Because different peptides are presented by distinct polymorphic class I molecules, patients are usually HLA typed so that the appropriate peptide(s) can be utilized. The most commonly expressed class I MHC allele in Western populations is HLA-A2, and thus most clinical trials have focused on this patient subset. In parallel, most trials have analyzed the patient tumor for expression of the tumor antigens of interest. Therefore, eligibility for such studies requires both expression of the defined antigen(s) by the tumor, as well as the presence of the specific HLA allele in the host. Another key variable that has been addressed in these trials is the selection of a vaccine adjuvant to support optimal DC activation which in turn should induce a more potent effector T cell response. Some of the more potent vaccine adjuvant components explored have been combinations of toll-like receptor (TLR) agonists such as CpG 7909 (Speiser et al., 2005; Fourcade et al., 2008) or the cytokine IL-12 (Gajewski, 2002; Peterson et al., 2003; Lee et al., 2001). These studies have demonstrated induction of high frequencies of peptide-specific CD8+ T cells using direct ex vivo monitoring assays, and also have shown clinical activity in a subset of patients with melanoma.

In addition to peptides, whole protein-based approaches have also begun to be evaluated. These platforms are more costly and challenging, as clinical-grade full-length proteins must be prepared and quality controlled. However, an advantage is that they do not need to be restricted to patients having a restricted HLA type. The protein-based strategy furthest in development is the MAGE-3 vaccine being evaluated by GSK-Bio in both melanoma and in NSCLC. The vaccine adjuvant utilizes a combination of the TLR4 agonist MPL and the TLR9 agonist CpG 7909. Evidence for clinical benefit has been observed in a subset of patients with MAGE-3-expressing tumors in both disease contexts, and this vaccine is currently undergoing phase III clinical trial testing (Louahed et al., 2008; Vansteenkiste et al., 2008; Brichard and Lejeune, 2007).

A new level of complexity in vaccine development is emerging as the point mutations in individual tumors are becoming utilized for preparation of a truly patient-specific vaccine formulation. The expanding availability and diminishing cost of total exome sequencing has made it possible to identify mutations that would give rise to new antigenic epitopes in a given patient's tumor. A set of synthetic peptides can then be manufactured and used for immunization on a case-by-case basis. In fact, this approach may grow in importance with cancers such as renal cell carcinoma, in which dozens of genetic alterations can be identified in each patient's tumor, but without any gain of function mutations that might enable targeting of an active signaling enzyme with small molecule inhibitors (Dalgliesh et al., 2010). As such, capitalizing on immune recognition of such neoantigens may offer a particularly advantageous therapeutic strategy in these settings.

It is important to emphasize that therapeutic cancer vaccines typically deliver clinical benefit in only a subset of patients. This observation has led to the consideration of predictive biomarkers that may enable better patient selection. The magnitude of the immune response induced as measured in the peripheral blood has not reproducibly been associated with clinical response. Based on the presumption that raising a T cell response may be necessary but not sufficient for tumor control, attention has turned to probing characteristics of the tumor microenvironment, which is the tissue site where the immune system must act in order to mediate its therapeutic effect. In fact, several small studies have indicated that clinical benefit to cancer vaccines may preferentially occur in those patients that already have a T cell-inflamed tumor microenvironment (Gajewski et al., 2010). This notion makes sense, as local tissue inflammation including chemokine production may be required to support directed effector T cell homing across the vascular endothelium and into the tumor microenvironment. An implication of this model is that tumors that lack the appropriate inflammation may fail to respond to immunotherapies because of inadequate T cell recruitment, thereby representing an important mechanism of tumor resistance to anti-tumor immunity. The possibility that an inflamed tumor microenvironment may predict clinical benefit to active immunization is being tested prospectively in the MAGE-3 protein vaccine being evaluated by GSK-Bio (Brichard and Lejeune, 2007).

2.2 Adoptive T cell therapy

As an alternative approach to raise the frequency of tumor-specific T cell populations, adoptive T cell transfer has been pursued. This general strategy bypasses the need for the endogenous host immune system to respond to an exogenous vaccine, and can involve delivery of enormous numbers of cells, offering a quantitative advantage. The approach also allows for direct manipulation of the T cell population being administered, and also conditioning of the host to support optimal T cell persistence and functional maintenance.

Initial efforts at adoptive T cell therapy began with melanoma and met with limited success (Rosenberg et al., 1993). Antigen-specific CD8+ T cell clones infused intravenously failed to promote meaningful clinical responses, even when combined with high-dose IL-2 (Rosenberg and Dudley, 2009). Similarly, autologous tumor-infiltrating lymphocytes (TIL) that were harvested, expanded in vitro, and re-infused along with IL-2 did not show clinical activity that was meaningfully greater than what would be expected with IL-2 alone (Rosenberg et al., 1994). Analysis of the presence of the transferred T cells over time suggested a short half-life, arguing that techniques to improve expansion and persistence in vivo might be necessary. In fact, pre-conditioning of the patient with lympho-ablative doses of Cytoxan and fludarabine prior to infusion of TIL and administration of IL-2 has given response rates of greater than 50% of selected melanoma patients (Rosenberg and Dudley, 2004; Dudley et al., 2008). Preclinical models have indicated that the lymphopenic condition supports homeostatic proliferation of transferred T cells through release of endogenous IL-7 and IL-15 (Gattinoni et al., 2005). These cytokines not only support persistence but also have been shown to prevent and reverse T cell anergy (Brown et al., 2006), which likely helps maintain the function of transferred T cells. The Cytoxan and fludarabine combination also likely depletes suppressive cell populations, particularly CD4+CD25+FoxP3+ regulatory T cells (Tregs), which have been shown to inhibit the efficacy of anti-tumor effector T cells in vitro and in preclinical models in vivo (Viguier et al., 2004; Kline et al., 2008).

It is neither yet clear that massive lymphodepletion is actually required for optimal efficacy of adoptively transferred T cells, nor that high-dose IL-2 is absolutely required. Cytoxan alone has been shown to be efficacious in one model by Yee et al., explored based on the argument that fludarabine is immunosuppressive beyond lymphodepletion and might interfere with the function of APCs (i.e. DCs) of the host (Frank et al., 1999). In addition, the apparent requirement of high-dose IL-2 administration to the patient may be related to the fact that the T cell product itself is expanded using high concentrations of IL-2 in vitro. Expansion of T cells in vitro with other homeostatic cytokines such as IL-15 or IL-21 may eliminate the need to administer high-dose IL-2 following adoptive transfer. Indeed, low doses of IL-2 appear sufficient in one model in which the T cell culture was itself expanded using lower concentrations of IL-2 (Yee et al., 2002).

With the advent of lymphopenic conditioning, exploration of different culture parameters, and the consideration of alternative homeostatic cytokines, the possibility of using T cell clones having precise antigen specificity has begun to be re-explored. Yee et al. have observed clinical activity with infusion of either CD8+ or CD4+ T cell clones expanded from the peripheral blood (Yee et al., 2002; Hunder et al., 2008). Further work in this area is warranted using the next generation homeostatic cytokines, as well as a variety of host conditioning regimens. Direct administration of IL-15 or IL-21 to patients may obviate the need to administer lymphopenia-inducing chemotherapy.

Besides autologous T cell clones or TIL-derived products for therapy, additional strategies are being explored that involve genetic engineering of polyclonal autologous T cell populations to re-direct antigen specificity through introduction of a recombinant receptor. Preliminary clinical trial experience is being gained using recombinant T cell receptors (TCRs) specific for Melan-A/MART-1, gp100, and NY-ESO-1 (Rosenberg and Dudley, 2009). An expansion of this notion has been the design of chimeric antigen receptors (CARs) built as a fusion receptor between an antibody molecule and the signaling domains of the TCR and coreceptor molecules. Exciting preliminary data using a CAR specific for CD19 in patients with chronic lymphocytic leukemia have been reported (Porter et al., 2011).

3 How to promote local tumor inflammation and T cell trafficking into the tumor microenvironment

Inasmuch as clinical benefit to immunotherapies might be restricted to the subset of patients having pre-existing inflammation in the tumor microenvironment, consideration must be given to developing strategies to induce the desired inflammatory phenotype in patients with tumors that fail to generate it spontaneously. This principle should apply whether one considers the endogenous immune response, one induced by vaccination, or adoptively transferred T cell populations. Because a major functional component of this inflammation involves expression of chemokines that in turn support T cell recruitment, one strategy has involved direct introduction of selected chemokines. As proof of concept, these experiments are usually done by transfection or using viral vectors. While some preclinical models have shown anti-tumor effects of expression of specific chemokine cDNAs in tumor cells (Braun et al., 2000), it is unlikely that chemokines alone will have maximal therapeutic efficacy in patients. A broader inflammatory milieu may be needed that includes activation of the vascular endothelium, recruitment and activation of the appropriate DC subsets for T cell stimulation, as well as chemokines for attraction of the desired populations of CD8+ effector/memory T cells. Several considerations include the local application of TLR agonists, the introduction of innate immune cytokines such as type I IFNs, or the expression of ligands of the LTβR (lymphotoxin, LIGHT) which have been shown to induce a broad tissue restructuring effect that mimics creation of a secondary lymphoid structure (Wang et al., 2001; Yu et al., 2004). Intratumoral administration of an adenoviral vector encoding LIGHT in preclinical models has been shown to result in substantially augmented T cell recruitment and potent anti-tumor activity in vivo (Yu et al., 2007). A major technical hurdle for development of these strategies for clinical application is how to direct the expression of such factors selectively in the tumor sites following systemic administration. Topical application of the TLR7 agonist Imiquimod has shown high clinical response rates in cutaneous basal cell carcinoma (Geisse et al., 2004). How to guide delivery of this or other moieties to a range of metastatic tumor sites remains a challenge, but one that should receive a high priority for development.

4 Overcoming specific negative regulatory pathways in the tumor microenvironment

The fact that a natural immune response can be demonstrated in a substantial proportion of patients, through to the stage of effector T cell trafficking into tumors, but at the same time those tumors exist, argues that inhibitory mechanisms must be operational that suppress the function of those T cells within the tumor microenvironment. In fact, blockade of immune regulatory pathways as a strategy to augment the function of endogenous anti-tumor T cells has emerged as a major advance in cancer immunotherapy.

The first T cell negative regulatory pathway to be targeted in the clinic is the inhibitory receptor CTLA-4 which is expressed on activated T cells. Antagonistic anti-CTLA-4 mAbs were shown to have potent anti-cancer effects in a variety of preclinical models (van Elsas et al., 1999; Leach et al., 1996), and clinical-grade reagents targeting human CTLA-4 were developed. Based on strong phase II data indicating clinical activity in patients with metastatic melanoma, as well as induction of very prolonged disease stabilization in another major subset of patients (O'Day et al., 2010; Camacho et al., 2009), phase III studies were performed and have recently been reported. In fact, an improvement in overall survival was observed when compared to a first generation melanoma peptide vaccine in previously-treated patients (Hodi et al., 2010), and also in combination with dacarbazine versus dacarbazine alone in treatment-naïve patients (Robert et al., 2011). The median overall survival was 10.1 months with anti-CTLA-4 mAb compared to 6.4 months in control patients, with 23.5% of patients still being alive at two years. It is noteworthy that anti-CTLA-4 mAb is the first therapy ever to pass the hurdle of improved survival in a phase III clinical study in patients with metastatic melanoma (Gajewski, 2010).

Based on the success of targeting this first immune inhibitory checkpoint, excitement has grown over identifying and targeting other related pathways therapeutically. The metastatic melanoma gene expression profiling studies have revealed that the T cell-inflamed subset of tumors expresses high levels of several candidate immune inhibitory pathways (Gajewski et al., 2006; Gajewski, 2007). Indoleamine-2,3-dioxygenase (IDO) is a tryptophan-metabolizing enzyme that has been shown to regulate peripheral tolerance (Munn and Mellor, 2004). PD-L1 is a ligand for the inhibitory receptor PD-1, which much like CTLA-4 is expressed on activated T cells (Dong and Chen, 2003). Inflamed tumors also contain higher numbers of Tregs. Finally, indirect evidence has suggested that T cell-intrinsic dysfunctional state called anergy also is operational in the costimulatory ligand-poor tumor microenvironment (Harlin et al., 2006). In preclinical models, blockade of, or interference with, each of these four inhibitory mechanisms has been shown to improve immune-mediated tumor control in various model systems (Kline et al., 2008; Blank et al., 2004; Uyttenhove et al., 2003).

Clinical approaches to block or interfere with these additional immune suppressive mechanisms have been initiated and have already shown promise. Phase I/II studies of mAbs against PD-1 have been performed, with clinical responses seen in around 30% of the melanoma patients enrolled (Sznol et al., 2010). Responders were also seen among patients with renal cell carcinoma and NSCLC. Analogous studies with anti-PD-L1 mAbs also are ongoing, and multi-center phase II trials are being pursued. Strategies to deplete Tregs have focused upon agents that target the CD25 surface receptor. Denileukin diftitox, an IL-2-diptheria toxin fusion protein, was developed as a therapeutic for cutaneous T cell lymphoma. Early phase studies have shown evidence of partial Treg depletion with this agent (Dannull et al., 2005), and clinical responses have been seen in melanoma (Rasku et al., 2008), although not all studies have been positive (Attia et al., 2005). Multi-center phase II trials with this agent in melanoma are ongoing with careful correlative assays being integrated. mAbs against CD25 also have begun to be explored in the clinic, mimicking the most effective strategy that has been used in mouse models (Sutmuller et al., 2001). Early results have indicated durable reduction in Treg numbers after a single dose of one such agent, daclizumab (Rech and Vonderheide, 2009). Small molecule inhibitors that block the enzymatic activity of IDO have been developed. 1-methyltryptophan has shown efficacy in preclinical models (Uyttenhove et al., 2003) and is in phase I/II clinical trial testing. A more potent inhibitor developed by Incyte is also in phase I studies and has been reported to have biological activity on tryptophan metabolites in vivo (Liu et al., 2010). T cell anergy has proven more challenging to uncouple, but preclinical evidence has indicated that the process of homeostatic proliferation in response to cytokines that utilize the common γ chain can reverse anergy both in vitro and in vivo (Gattinoni et al., 2005; Brown et al., 2006; Boussiotis et al., 1994). One way to liberate such cytokines, particularly IL-7 and IL-15, is by rendering a host lymphopenic. Adoptive transfer of T cells into lymphopenic recipients can reverse anergy and maintain the function of anti-tumor T cells (Brown et al., 2006). This may be one mechanism explaining the success of TIL-based therapy in the setting of lymphopenic host conditioning (Rosenberg and Dudley, 2004). The use of exogenously delivered IL-7 or IL-15 has also been considered (Pellegrini et al., 2009, 2011), and clinical-grade material is entering clinical testing in cancer patients.

Inasmuch as multiple immune inhibitory mechanisms appear to operational concurrently in the tumor microenvironment, it is not unreasonable to consider that blockade of two or more pathways may be necessary for maximal therapeutic efficacy and to overcome compensatory effects. For example, preclinical data have indicated that Treg depletion combined with homeostatic proliferation as an anergy-reversal strategy can be potently synergistic in vivo (Kline et al., 2008). Combined blockade of CTLA-4 and PD-1 is synergistic in some models, as is anti-CTLA-4 mAb plus Treg depletion (Sutmuller et al., 2001). Once each of these agents has undergone phase II testing individually, it is hoped that combination studies can proceed to test for synergy in human cancer patients.

5 Sources of inter-patient heterogeneity—a potential role for host genetics, tumor somatic changes, and environmental factors

The observation that a subset of patients has a spontaneous T cell-inflamed tumor microenvironment has offered a potential predictive biomarker for patient selection for immunotherapeutics that is currently being evaluated prospectively. However, it also has generated a next-level question regarding the underlying molecular mechanism explaining these distinct phenotypes. Gaining a molecular understanding of these observations may further refine the opportunities for proper patient selection for specific immunotherapies, and also aid in focusing in on the key pathways that need to be modulated for optimizing therapeutic efficacy.

Three potential sources of inter-patient heterogeneity can be envisioned—genetic polymorphisms in the host, somatic differences between the tumors, and environmental differences that may have a global impact on immune function. Polymorphisms in immune regulatory genes may set thresholds for innate immune activation and peripheral tolerance, and also influence response to therapy. The first example reported is the case of CCR5 polymorphisms and clinical response to IL-2 (Ugurel et al., 2008). Distinct patterns of oncogene activation also could contribute to whether a spontaneous immune response is initiated against a cancer in an individual patient. One key example is Stat3, the silencing of which has been shown to induce expression of immunoregulatory genes, including chemokines and cytokines (Burdelya et al., 2005). Heterogeneity in activation of accessory oncogene pathways such as Stat3 could explain inter-patient variability in response to immunotherapies. In terms of environmental factors that could influence global immune function, recent work has drawn attention to the intestinal microbiome. The composition of intestinal flora has been shown in preclinical models to have a profound impact on the manifestation of autoimmune diseases, such as inflammatory bowel disease and rheumatoid arthritis (Ivanov et al., 2009; Wu et al., 2010; Ivanov and Littman, 2011). It is therefore not difficult to envision a role for selected species of commensal bacteria influencing the ability of the host to mount a spontaneous immune response against cancer. Each of these three categories of inter-patient heterogeneity deserves more detailed evaluation in patients, to further refine the ability to personalize immunotherapy application.

6 Conclusions and future directions

The rational development of immunotherapies is making major advances based on a more detailed understanding of the biology of the tumor/host interaction. Depending on the mechanism by which the tumor has escaped immune-mediated destruction in individual cases, the interventions required to restore tumor regression in specific subsets of patients may be distinct. In patients with T cell-inflamed tumors, blockade of the major immune suppressive mechanisms operational within the tumor microenvironment might be sufficient, either individually or in combination. In patients with non-inflamed tumors, strategies to promote local productive inflammation in the tumor microenvironment may be the most critical step to ensure. In both cases, strategies to increase the frequency of tumor antigen-specific T cells through vaccination or adoptive transfer may have additional utility. Ultimately, it is not difficult to envision that analysis of tumor material from an individual patient will be routinely employed, to explore the critical predictive biomarkers that will facilitate selection of the appropriate immune-based therapies. This would be analogous to analysis of Her2 expression, estrogen/progesterone receptor presence, B-Raf or c-kit mutations, etc. to restrict application of specific therapies to the individual patients having tumors possessing the relevant target. The future of this general strategy of scientifically developed, patient-specific application of immunotherapeutics is thus very bright indeed.