Background
The International Society for the Biological Therapy of Cancer (iSBTc) launched in collaboration with the USA Food and Drug Administration (FDA) a task force addressing the need to expeditiously identify and validate biomarkers relevant to the biotherapy of cancer [
1]. The task force includes two principal components: a) validation and application of currently used biomarkers; b) identification of new biomarkers and improvement of strategies for their discovery. Currently, biomarkers are either not available or have limited diagnostic, predictive or prognostic value. These limitations hamper, in turn, the effective conduct of biotherapy trials not permitting optimization of patient selection/stratification (lack of predictive biomarkers) or early assessment of product effectiveness (lack of surrogate biomarkers). These goals were summarized in a preamble to the iSBTc-FDA task force [
1]; the results are going to be reported on October 28
th at the
"iSBTc-FDA-NCI Workshop on Prognostic and Predictive Immunologic Biomarkers in Cancer", which will be held in Washington DC in association with the Annual Meeting [
2]; a document summarizing guidelines for biomarker discovery and validation will be generated. Several other agencies will participate in the workshop including the National Cancer Institute (NCI), the National Institutes of Health (NIH) Center for Human Immunology (CHI) and the National Institutes of Health Biomarker Consortium (BC).
With the generous support of the Office of International Affairs, NCI, the "US-Japan Workshop on Immunological Molecular Markers in Oncology" included, on the US side, significant participation of the iSBTc leadership, representatives from Academia and Government Agencies, the FDA, the NCI Cancer Diagnosis Program (CDP), the Cancer Therapy and Evaluation Program (CTEP), the Cell Therapy Section (CTS) of the Clinical Center, and the CHI, NIH. The participation of Japanese and US scientists provided the opportunity to identify shared or discordant themes across the distinct immunogenetic background and the diverse disease prevalence of the two Nations and compare scientific and clinical approaches in the development of cancer immunotherapy.
Primary goal of the workshop was to define the status of the science in biomarker discovery by identifying emerging concepts in human tumor immune biology that could predict responsiveness to immunotherapy and/or explain its mechanism(s). The workshop identified recurrent themes shared by distinct human tumor models, independent of therapeutic strategy or ethnic background. This manuscript is an interim appraisal of the state of the science and advances broad suggestions for the solutions of salient problems hampering discovery during clinical trials and summarizes emerging concepts in the context of the present literature (Table
1). We anticipate deficiencies in our attempt to fairly and comprehensively portray the subject. However, through Open Access, we hope that this interim document will attract attention. We encourage feed back from readers in preparation of an improved and comprehensive final document [
2]. Thus, we invite comments that can be posted directly in the
Journal of Translational Medicine website and/or interactive discussion through
Knol [
3].
Table 1
Emerging biomarkers potentially useful for the immunotherapy of cancer
Predictive biomarkers
|
Telomere length
| Adoptive therapy | Melanoma | |
VEGF
| IL-2 therapy | Melanoma | |
CCR5 polymorphism
| IL-2 therapy | Melanoma | |
Carbonic Anhydrase IX
| IL-2 therapy | Renal Cell Cancer | |
IFN-γ polymorphism
| Immuno (IL-2)-chemo | Melanoma | |
STAT-1, CXCL-9, -10, -11, ISGs
| IFN-α therapy | Several Cancers | |
IL-1α,-1β, IL-6, TNF-a, CCL3, CCL4
| IFN-α therapy | Melanoma | |
CCL5, CCL11, IFN-γ, ICOS, CD20
| GSK/MAGE3 vaccine | Melanoma | |
IL-6 polymorphism
| BCG vaccine | Bladder Cancer | |
MFG-E8
| GM-CSF/GVAX (pre-clin) | Prostate | |
T regulatory cells
| hTERT pulsed DCs | Solid Cancer | |
K-ras mutation
| Cetuximab | Colorectal Cancer | |
CCL2, -3, -4, -5 CXCL-9, -10
| Preclinical | Melanoma | |
T cell mulifunctionality
| Preclinical | - | |
SNAIL
| Preclinical | - | |
Prognostic Biomarkers (useful for patient stratification/data interpretation)
|
Oncotype DX, Mamma Print
| - | Breast Cancer | |
TGF-β
| - | Breast Cancer | |
Korn Score
| - | Prostate Cancer | |
IFN-γ, IRF-1, STAT-1, ISGs, IL-15, CXCL-9, -10, -11 and CCL5
| - | Prostate Cancer | |
IFN-γ, IRF-1, STAT-1
| - | Colorectal Cancer | |
VEGF
| - | Colorectal Cancer, Nasopharyngeal Ca | |
ARPC2, FN1, RGS1, WNT2
| | Melanoma | |
Mechanistic/End Point Biomarkers
|
IFN-γ, IRF-1, STAT-1, ISGs, IL-15, CXCL-9, -10, -11 and CCL5
| IL-2 therapy/TLR-7 therapy | Melanoma/Basal Cell Cancer | |
IRF-1, STAT-1, ISGs, IL-15, CXCL-9, -10, -11 and CCL5
| Vaccinia virus (Xenografts) | Solid tumors | |
CXCL-9, -10
| Herpes simplex virus (syngeneic model) | Ovarian CA | |
18F-FDG localization
| Anti-CTLA-4 therapy | Melanoma | |
Epitope Spreading
| DC-based therapy | Melanoma | |
Kinetic regression/growth model
| - | - | |
Basic insights about cancer immune biology
Much can be learned in human immunology by a comparative method that looks at immunological phenomena with an interdisciplinary approach [
124]. The relevance of IFN signatures in the context of various diseases represents a good example. He et al [
180] observed that decreased IFN signaling and decreased
ex vivo responsiveness of PBMCs to IFN-α stimulation were harbingers of non-responsiveness of HCV-infected patients to systemic administration of pegylated IFN-α and Ribavarin. These differences were interpreted as related to the genetic background of patients as it was observed that PBMCs from patients of African American (AA) origin were least likely to respond to IFN-α stimulation
ex vivo and to recover from hepatitis compared to patients of European American (EA) background. This observation raises the question of whether patients with melanoma or HCV that have better changes to respond to therapy are characterized by a different genetic background compared to those likely to do poorly. A recent analysis performed in our laboratories (Pos et al. in preparation) failed to demonstrated dramatic differences between the responses of the two ethnic groups to IFN-α (see later). Thus, alterations in IFN signaling are likely to represent a secondary effect due to the presence of cancer cells or viral particles that in turn may interfere with the innate immune response of the host. This being the case, it will be likely in the future that more insights about the mechanisms leading to altered IFN signaling in cancer patients will be gathered by a more in depth analysis of cancer biology and the products released by cancer cells that may affect immune cells activity locally and at the systemic level.
Indeed tumors, including melanoma, display strong differences in the expression of ISGs [
190,
191], which are coordinately associated with the expression of several chemokines, cytokines, growth and angiogenic factors [
190,
192]. Moreover, the presence of immune activation has been associated with the prognosis of melanoma [
193]. Thus, it is likely that melanoma and other cancers express an immune modulatory phenotype that may alter not only their own microenvironment but whose effects can reverberate at the systemic level. Whether these differences are due to distinct disease taxonomy [
194] or to disease progression [
126,
190] remains to be clarified.
Mohammed Kashani-Sabet proposed a model that may explain the dichotomy observed in the biological pattern of melanomas. Studying check points in the progression of melanoma, it was observed that BRAF mutations occur early in the development of the disease and do not account for the switch to an increasingly more aggressive phenotype. Transcriptional analysis was performed to compare radial to vertical growth, which identified predominantly loss of gene expression [
195,
196]. Two subtypes of melanoma were identified that could not be segregated only on account of BRAF mutations. Rather, modifiers associated with the vertical growth phase included immune regulatory genes such as IFI16, CCL2 and 3, CXCL-1, -9 and -10. These genes are up regulated in primary melanoma compared with nevi but become down-regulated in the metastatic phase in some but not all melanomas [
195], a phenomenon we had previously observed comparing the transcriptional profile of melanoma metastases to normal melanocytes [
190] and other cancers [
192]. A multi-marker diagnostic assay for melanoma was developed [
197]; a large training set of tissue microarrays with 534 samples including nevi and melanoma biopsies was validated on 4 independent test sets and found ARPC2, FN1, RGS1, SSP1 and WNT2 to be over-expressed in melanoma compared with nevi. Based on the 5 markers, a diagnostic algorithm was developed that could differentiate with high accuracy and specificity benign from malignant lesions [
197]. The markers were also evaluated on independent cohorts including the German Cancer Registry (Heidelberg/Kiel cohort). The multi-marker approach tested at several stages of disease could predict sentinel node status and disease specific survival (p < 0.001). The multi-marker score demonstrated higher accuracy than lesion depth or ulceration. A molecular map of melanoma progression is being built from melanocyte to various growth phases and metastatization and will be evaluated in the ECOG data set. Although this algorithm does not directly address the immune responsiveness of tumors, it will be important to include such information for patient stratification in future clinical trials to interpret immunotherapy results.
Constitutive activation of immune regulatory mechanism was also reported by Yutaka Kawakami, who discussed the molecular mechanisms of cancer cell induced immune-suppression and their potential as biomarkers of responsiveness to immunotherapy. In particular, regulatory mechanisms dependent on the MAPK, WNT and BRAF mutations were discussed. BRAF and NRAS mutations occur early in melanoma [
198]. Kawakami reported that inhibition of BRAF or STAT-3 depleted the expression of several cytokine including IL-6, CXCL8/IL-8 and IL-10 by cancer cells. Also a MEK inhibitor blocked the expression of IL-10. Finally, VEGF expression was inhibited by small interference RNA (siRNA) for ERK1/2.
In vivo studies, observed that inhibition of ERK induced the enhancement of T cell responses and protection of mice from cancer [
199]. Considering the recently described role of VEGF as a negative predictor of immune responsiveness of melanoma metastases to high dose IL-2 therapy [
9] and a poor prognostic marker of survival in colorectal cancer [
141], it is possible that this observation may provide an important target for a combination therapy for VEGF expressing melanomas. In particular, the melanoma cell line, 888-MEL previously extensively characterized [
200,
201] was found to be sensitive to MEK inhibition. Moreover, Kawakami reported that IL-10 production is strictly dependent (in this cell line) upon the expression of β-catenin a mutation inducing enhanced activation of the WNT pathway [
202]. Transfection of β-catenin induced production of IL-10; moreover, culture of DC with supernatant of melanoma cells with high β catenin induces IL-10-producing DC and it was decreased by siRNA blockade of β-catenin. Functionally, T cells produced less TNF-α when stimulated with DC cultured with supernatant from β-catenin positive melanomas and expressed higher levels of FOX P3. In a xenogenic model, the human melanoma cells 397-MEL that do not express constitutively high levels of activated β-catenin, were transfected to produce IL-10. Upon antigen exposure T cells were observed to produce less IFN-γ and display lowered lytic activity in animals implanted with the IL-10 expressing tumors. However, IL-10 blocking antibodies did not reverse the tolerogenic effect suggesting that a more complicated mechanism is responsible for the effect on T cells than the direct activity of IL-10. Of interest is the relationship between IL-10 expression and responsiveness. The high expression of IL-10 by 888-MEL contrasts with the observation that this cell line was derived from a patient who dramatically responded to immunotherapy and was a long-term survivor [
203]. However, the perceived immune suppressive role of IL-10 may be more complex than previously reported. We observed, that IL-10 expression by melanoma cells studied in pre-treatment biopsies is a positive predictor of tumor responsiveness to immunotherapy with high-dose IL-2 [
126,
204,
205]; moreover, the majority of pre-clinical models in which the effect of IL-10 was evaluated as a modulator of tumor responsiveness identified this cytokine as a factor favoring tumor regression suggesting a dual role of IL-10 promoting growth in natural conditions but favoring tumor rejection upon immune stimulation [
206]. Kawakami's work may shed light on this paradoxical observation; screening of siRNA against 800 kinases was done to identify which are involved in immune suppression; it was found that STKX kinase inhibits IL-10 and TGF-β production. Moreover, epithelial-mesenchymal transition is induced by SNAIL transfection, which also induces IL-10, VEGF and TGF-β and, in co-culture with human PBMCs, induces FOX-P3 expression. Co-culture of PBMCs with melanoma cells transfected with SNAIL increases the number of FOX-P3-expressing T cells and this is also reversed by SNAIL/TSP (downstream of SNAIL) blockade. Blocking SNAIL expression by tumors with siRNA induced increase in CD4 and CD8 T cells, thus
in vivo SNAIL may be involved in immune suppression. Similar results can be obtained by anti-TSP1 which can induce better T cell infiltrates. SNAIL transfected melanoma is resistant to immunotherapy in mouse models and may represent a new predictive biomarker of tumor responsiveness to immune therapy [
43].
Host's genetics vs cancer genetics; the riddle of tumor immunology
The relative contribution of the genetic background of the host, the genetic instability of cancer and the effects of the environment on the natural history of cancer is complex. A good example is nasopharyngeal carcinoma (NPC), which predominantly affects specific geographic areas and ethnicities, in particular the Asian Population [
207‐
210]. NPC etiology is clearly linked to Epstein-Barr virus (EBV) infection [
211] and the immune response to the EBV infection appears to bear a strong influence in both the natural history of the disease and response to therapy [
207,
212‐
218]. A recent observation linked elevated VEGF secretion by the tumor tissue to outcome; in that study, high VEGF secretion correlated with decreased survival. The reason for the prevalence of NPC in specific ethnic groups remains to be conclusively explained but there is evidence that the genetic background of the host plays an important role in familiar and sporadic cases [
209‐
211,
218‐
230]. However, as for most disease etiologies that are influenced by numerous genes, the genetic determinants of disease prevalence and clinical outcome are still not fully understood [
231‐
238]. In particular, cancer immune responsiveness can be influenced by either the genetic background of the host's or by disease heterogeneity [
1,
239]. Few lines of evidence suggest that the genetic make up of patients may affect the natural history of cancer or its responsiveness to therapy; a polymorphism of the IFN-γ gene was associated with responsiveness to combination therapy with IL-2 therapy and chemotherapy [
240]. Others found that variants of CCR5 are predictors of survival in patients with melanoma receiving immunotherapy [
161]. More recently, the responsiveness to IFN-α therapy in melanoma was found to be associated with autoimmune disease which in turn could be related to genetic predisposition [
241,
242]. Recently, Dudley et al [
8] reported that the adoptive transfer of tumor-infiltrating lymphocytes with shorter telomeres was associated with a strongly decreased chance of clinical response; although this effect has been explained by a senescent phenotype of lymphocytes, it is possible that genetic variations in the ability to conserve telomere length could be responsible for differences among patients as previously observed for other instances [
243‐
245].
In a broader sense, the heterogeneous response to IFN-α observed among patients with either cancer [
182,
183] or HCV [
180,
246,
247] can be plausibly explained by inherited genetic predispositions that determine the responsiveness to this cytokine. It has been proposed that single nucleotide polymorphisms in the IFN pathway are associated with the response to IFN-α therapy of HCV [
248]. Moreover, ISG polymorphisms have been associated with other immune pathologies and differences in the prevalence of IRF and STAT gene polymorphisms have been associated with the prevalence of systemic lupus erythematosus in AA [
249,
250]. Alternatively, racial differences in the responsiveness to a given treatment may come from effects that the disease exerts on the host's immune cells, and from differences to environmental exposures. Thus, AA may be genetically less protected against HCV infection for reasons unrelated to IFN-α activity; yet, the higher viral load or other factors associated with worse disease may, in turn, affect IFN-related pathways [
180,
246,
251,
252]. Whether the genetic background determines the responsiveness to IFN-α or whether acquired differences in the disease status are responsible for differences in the disease phenotype among populations, can only be answered by studying normal volunteers not bearing a disease, like cancer or HCV, that are known to affect the immune response [
118]. Based on the observation that AA patients with HCV infection are the least likely to respond to IFN-α stimulation, we tested whether immune cells from 48 AA and 48 EA normal volunteers matched for age and sex responded differently to IFN-α. We compared the levels of STAT-1 phosphorylation and global transcriptional profile of T cells between the two ethnic groups. The same subjects were genetically characterized by genome wide single nucleotide polymorphism analysis to determine the racial deviation of the two groups. This is an important task considering the genetic diversity of AA and their potential admixture with other ethnic groups [
253] Although there was clear separation among AA and EA at the genomic levels, no clear differences could be identified at the functional level (phospho-assays or transcriptional profiling, Pos et al. manuscript in preparation). Thus, it is likely that differences observed in IFN-α responsiveness among different individuals of distinct genetic background or within the same ethnic group affected by cancer or HCV may be secondary to a difference in the disease itself or a difference in the response of the host to the disease, which may affect secondarily the host's immune response. This observation may help interpret differences in tumor immune biology according to race/ethnicity reported by other groups.
Stefan Ambs (NCI, Bethesda, Maryland, USA) reported a comparison of transcriptional patterns between AA and EA in prostate and breast cancer [
254,
255]. It is noteworthy that AA have higher death rates from all cancer sites combined than other US populations [
256]. Ambs also presented an example for race/ethnic differences in the prevalence of a genetic susceptibility locus from published reports. Several genetic variants at the 8q24 cancer locus are most common among subjects with African ancestry and these differences can explain some of the excess risk of AA to develop prostate cancer. In their study, Ambs and coworkers compared 33 AA and 36 EA macro-dissected tumors by transcriptional analysis. Numerous genes were differently expressed between the two patient groups, but the biggest differences were found to be related to genes involved in the immune response and in particular associated with IFN signaling: IFN-γ, STAT1, CXCL9–11 CCL5 CCL4 CCR7, IL-15 and -16, USG15, Mx1, IRF-1, – 8, -2, OAS2, TAP1 and 2. These genes were over expressed in AA suggesting that in those tumors the cancer cells are in an anti-viral state. Interestingly, the expression of these genes in prostate and breast cancer was associated with resistance to chemotherapy and radiation and in general with a worse prognosis [
257] bearing the opposite significance than the expression of similar signatures in colorectal cancer [
134,
135,
141]. Their expression is associated with a poor prognostic connotation in the former and a good one in the latter. An explanation for this discordant and opposite observation is lacking. Similar differences in the tumor microenvironment were observed by Ambs studying breast tumors and comparing tumor stroma and micro-dissected tumor epithelium. Those data were further validated by immunohistochemistry in an extended set of tissues [
255]. In tumors from AA, an increased macrophage infiltration was observed, using CD68 as marker, and also a higher micro vessel density, as judged by CD31 expression, when compared with EA tumors
Xifeng Wu (MD Anderson Cancer Center, Houston, Texas, USA) emphasized the need for a systematic evaluation of genetic variants in inflammation-associated pathways as predictors of cancer risk and clinical outcome. The evolution of epidemiologic research from traditional to molecular and even more integrative epidemiology has rapidly changed the paradigm of cancer research. The integration of information at the pathway level is necessary because multiple inherited alterations in gene function can have additive effects as part of a pathway and different pathways can act synergistically or in antagonism. Additional assessment of the predicted or documented functional effects of genetic variants in the biology of disease should also be considered in these models. Wu's hypothesizes that the inflammatory response that plays a role in carcinogenesis is modulated by genetic variability. Fifty-nine SNPs in 36 genes were analyzed. SNPs were selected at promoter UTR or coding region segments according to the literature. Several cytokines were selected and were studied in 1,500 lung cancer cases and 1,700 matched controls. Comprehensive epidemiologic information was obtained and 7 SNPs were found to be relevant. Among them, IL-1α and IL-1β positively correlated with lung cancer prevalence in heavy smokers suggesting that deregulated inflammatory response to tobacco-induced lung damage promotes carcinogenesis [
258]. Five SNPs were associated with increased risk of developing bladder cancer including MCP1 and IFNAR2 and two variants of COX2 and IL4r (the COX-2 allele was observed to be associated with reduced mRNA expression) [
259]. Interestingly, an IL-6 polymorphism was associated with an increased risk of recurrence after treatment with BCG and with poor survival. In another study of about 400 cases of bladed cancer of whom half experienced recurrence after treatment, Wu and coworkers observed that the genes that were associated with risk of developing bladder cancer were also predictor of response; a survival analysis based on a combination of SNPs including those related to IFN genes could predict with a much higher accuracy risk of recurrence compared to clinical parameters and this observation is now under validation studying a 10,000 SNPs of which 400 belong to the already investigated inflammation-related pathways.
Predictors of responsiveness
Although the IFN pathways seem to be central to TSD, the large experience gained treating patients with adjuvant melanoma with IFN-α has shown limited success. John Kirkwood (University of Pittsburgh Cancer Center, Pittsburgh, Pennsylvania, USA) summarized the long term experience with this treatment emphasizing the importance of sufficiently large randomized studies to obtain conclusive information about usefulness of therapeutics and related biomarkers [
15,
242,
260]. An extensive meta analysis including all phase II trials suggested that while in various trials different outcome biomarkers are identified these are most likely to fail validation as larger patient cohorts are treated [
15]. A recent analysis looking for predictive biomarkers in melanoma and renal cell carcinoma [
261] suggested that the
ex vivo ability of IFN-α to revert STAT-1 phosphorylation signaling defects in melanoma patients may be useful [
182,
183]. In addition, development of autoimmunity during IFN-α therapy is a clear predictor of a 50-fold reduction in frequency of relapse [
241]. Finally, the concentration of various soluble factors in pretreatment sera of patients undergoing IFN-α therapy suggested that the pro-inflammatory cytokines IL-1β, IL-1α, IL-6, TNF-α and chemokines CCL2/MIP-1α and CCL3/MIP-1β are elevated in patients with longer relapse-free survival [
262]. Together with VEGF and fibronectin potentially predictive of immune responsiveness to high-dose IL-2 therapy [
9], these biomarker represent candidate parameters for validation in future trials. High VEGF, together with high IL-6 levels have also been reported as negative predictor of response to bio-chemotherapy [
263,
264].
This is advancement from previous analyses in which the majority of putative predictors of IL-2 response were related to post-treatment parameters [
265,
266]. In renal cell carcinoma an additional biomarker has been described, carbonic anhydrase IX, whose expression in pre-treatment lesions may be associated with higher likelihood of response [
267]; interestingly, carbonic anhydrase IX is not expressed by melanomas although they display a similar ranges of responsiveness to IL-2 therapy, suggesting, that this molecule may be a biomarker of a particular phenotype associated with responsive lesions but not the determinant of responsiveness [
268]. In any case, further validation, together with a better understanding of the biology of these tumors will hopefully enhance the usefulness of these candidate biomarkers.
It has recently been shown that treatment with anti CTLA-4 antibodies can induce clinical responses in few patients previously vaccinated with irradiated, autologous granulocyte-macrophage colony-stimulating factor (GM-CSF)-secreting cancer cells [
269]. However, a large phase III study on hormone refractory prostate cancer-bearing patients treated with the same vaccine (but not anti-CTLA-4 antibody) failed to demonstrate effectiveness leading to early termination of the clinical protocol [
270,
271].
Masahisa Jinushi (The University of Tokyo, Tokyo, Japan) reported the mechanisms hampering vaccine effectiveness and the potentials for combining anti-CTLA-4 therapy. It was observed that GM-CSF-deficient mice are defective in apoptotic cell phagocytosis and develop autoimmune manifestations including pulmonary alveolar proteinosis, SLE, insulitis and diabetes [
272]. GM-CSF transduction restores the production of cytokines that regulate T helper cell differentiation (TGF-β, IL-1b IL-4 IL-12p70 and IL-23p19) in response to apoptotic cells. GM-CSF regulates the phagocytosis of apoptotic cells by antigen presenting cells and modulates the function of the phagocyte receptors milk fat globule EGF 8 (MGF-E8), a protein secreted at high levels by melanomas during the vertical growth phase. MGF-E8 has pleiotropic functions in the tumor microenvironment including promoting cancer cell survival, invasion and immune suppression. While GM-CSF regulates T helper cell differentiation by MFG-E8, TLR stimulation suppresses MFG-E8 production by antigen presenting cells resulting in increased allo-mixed lymphocyte reaction in apoptotic cell loaded macrophages-driven splenocytes proliferation [
272]. Blockade of MFG-E8 in tumor cells potentiates GVAX therapeutic immunity in the B16 mouse melanoma model. GVAX/RGE (inhibitor of MFG-E8) vaccines decreases Tregs and decreases tumor specific CD8+ T cell effectors with decrease of FoxP3 and increase in CD69 expressing CD8 T cells [
273]. MFG-E8 expression in melanoma patients with advanced stage is high and not detected in non advanced stage melanoma and nevi [
274]. Thus, MFG-E8 might be considered a negative regulator of GVAX induced immunity by regulating Treg/Teff balance. It is a prognostic factor and may predict response to GVAX and possibly other types of immunotherapy as recently shown by Aloysius el al [
275] with various cancers vaccinated with hTERT peptide-pulsed DCs and by Tatsumi et al. [
276] in the context of renal cell carcinoma and melanoma.
Target Selection
The NCI has shown strong interest in developing a systematic approach to the prioritization of agents to be tested in immunotherapy trials including the type of immune response modifier ()()[
277,
278] or target cancer antigen [
279]. Criteria were developed for the selection of each agent with a non-parametric approach receiving feed back from several investigators; however, the ideal antigen and/or biologic modifier and their combination remain to be defined. An ideal candidate target could be considered a protein expressed consistently by cancer initiating cells. Sato et al. [
280] described their efforts in identifying such cells among which they describe sperm mitochondrial cystein rich protein and sex determining region Y box-2 protein as potential candidate targets of immunotherapy. They may be used against breast cancer as their expression correlates with poor prognosis and resistance to chemotherapy. Identification of epitopes is underway for HLA alleles common in the Asian population and this novel target could be considered a potential biomarker for patient selection. Another important target expressed by several tumors and potentially associated with the oncogenic process is NY-ESO, a prototype cancer/testis antigen, which induces strong antibody and T cell responses. Extensive work has been done in Japan on patients with esophageal and other solid cancers [
281]. NY-ESO was delivered as cholesterol-bearing hydrophobized pullulan nano-particles that absorb the protein and express it in the antigen presenting cells. Humoral and cellular immune responses were elicited in 9 of 13 treated patients and clinical responses were observed in 4 of 5 evaluable patients. Several examples of antigen spreading were observed and a restricted region of the NY-ESO protein was found to be most immunogenic; it is suggested that, for the future, only this region should used for immunization. This is an example of the relevance of careful immune monitoring related to a specific target antigen that provides insights for the design of future clinical trials.
For gastrointestinal tumors, EpCAM, a tumor associated antigen was proposed as a useful target in gastrointestinal cancers. Use of anti-EpCAM may affect tumor stage and progression. Recently a technique was developed to isolate circulating tumor cells using magnetic beads based on EpCAM expression. Cancer cells were isolated from 130 cancer patients and 40 normal controls. Highly significant differences in extractable cells were observed between cancer and normal patients and between patients with or without metastatic disease. The identification of ≥ 2 circulating cancer cells was associated with tumor stage, survival and pleural or peritoneal dissemination. In esophageal cancer cell lines a proliferation assay was performed showing that introduction of EpCAM increases the expression of cyclins suggesting that EpCAM expression accelerates cell cycle and may be an important novel target for the immunotherapy of gastrointestinal tumors. Indeed, anti-EpCAM antibodies decrease tumor growth in animal models and recent clinical trials have been initiated [
282,
283]. More recently, antibody-mediated targeting of adenoviral vectors modified to contain a synthetic immunoglobulin g-binding domain in the capsid was described that could be used to target tumor-specific antigens expressed on the surface of cancer cells [
284].
Furthermore, attention should be put to the status of methylation or acetylation patterns of various genes that may directly or indirectly affect immune function either by down-modulating the expression of putative tumor antigens, or by interfering with immune-regulatory pathways [
285‐
287].