New targets for the immunotherapy of colon cancer—does reactive disease hold the answer?

Abstract

Colorectal cancer (CRC) is one of the most commonly diagnosed cancers in both men and women, posing a serious demographic and economic burden worldwide. In the United Kingdom, CRC affects 1 in every 20 people and it is often detected once well established and after it has spread beyond the bowel (Stage IIA–C and Stage IIIA–C). A diagnosis at such advanced stages is associated with poor treatment response and survival. However, studies have identified two sub-groups of post-treatment CRC patients—those with good outcome (reactive disease) and those with poor outcome (non-reactive disease). We aim to review the state-of-the-art for CRC with respect to the expression of cancer-testis antigens (CTAs) and their identification, evaluation and correlation with disease progression, treatment response and survival. We will also discuss the relationship between CTA expression and regulatory T-cell (Treg) activity to tumorigenesis and tumor immune evasion in CRC and how this could account for the clinical presentation of CRC. Understanding the molecular basis of reactive CRC may help us identify more potent novel immunotherapeutic targets to aid the effective treatment of this disease. In this review, based on our presentation at the 2012 International Society for the Cell and Gene Therapy of Cancer annual meeting, we will summarize some of the most current advances in CTA and CRC research and their influence on the development of novel immunotherapeutic approaches for this common and at times difficult to treat disease.

The global challenge

‘You have cancer’—the diagnosis that millions of people around the world hear every year and also the second leading cause of death worldwide. The number of new cancer cases each year is gradually increasing and the mortality rate is expected to rise from 7.6 million in 2008 to over 17 million in 2030.^{1, 2} On average, one of every three people is expected to experience some type of cancer in the course of their lifetime and this frequency is expected to increase³ as the population ages. The World Health Organization reports that an increasing percentage of all newly diagnosed cancer cases annually occurs in low- and middle-income countries and this is attributed to a wide range of behavioral, genetic and environmental risk factors.⁴ Nevertheless, the increase in the average life expectancy and the adoption of an unhealthy lifestyle (such as smoking, physical inactivity and poor diet) worldwide have also contributed to the rise of a new trend in the cancer demography. Newly diagnosed cancers from the low- and middle-income countries today account for >51% of the total number and their share in the global burden is expected to continue to increase with the growth and aging of the population.⁴ In addition, the lack of resources for early cancer detection and effective treatments in the developing world contributes to an increase in cancer-related deaths.

Today, some of the most commonly diagnosed cancers both in economically developing and developed countries include prostate, breast, lung and colorectal cancer (CRC)—‘the big four’—accounting for nearly 50% of the total cases diagnosed.⁵ Cancer is a global challenge, opposing a serious demographic and economic burden with worldwide economic costs estimated to be as high as £572 billion per year. These alarming statistics resulted in the development of national strategies and action plans for cancer control including prevention, early detection and effective treatment.⁶ However, the demand for the development of new approaches to cancer screening and therapy is recognized globally and brings the research attention on these aims sharply into focus.

Colorectal cancer

CRC is one of the most commonly diagnosed cancers worldwide. It affects the bowel and the rectum and is rare in people under 40, with almost 85% of cases being diagnosed in persons over 65 years of age.⁶ Statistics show that men and women are affected equally, while it is the third most common type of cancer in men (after prostate and lung cancer) and the second most common cancer in women (after breast cancer).⁶ One in every 20 people in the UK develops CRC with only half of them surviving beyond 5 years, mainly because it is often detected once well established and after it has spread beyond the bowel. The disease stage at the time of diagnosis governs both the choice of treatment and the prognosis. CRC is staged to reflect how far the cancer has spread and whether or not it has reached nearby structures such as lymph nodes or distant organs. The most commonly used staging system for CRC is that of the American Joint Committee on Cancer, also known as tumor nodes metastases system.⁷ It describes three key pieces of information: ‘T’—how far has the primary tumor grown; ‘N’—the extent of spread to nearby lymph nodes; and ‘M’—describes whether the cancer has metastasized. The information from the T, N and M is combined to determine the cancer stage grouping from Stage I (the least advanced) to Stage IV (the most advanced). Two of the older staging systems include Duke’s⁸ and Astler–Coller⁹ but these are very rarely used today.

Stage I (A–C) (Duke’s A–C) CRC is reported to be an asymptomatic malignancy, developing slowly by the progressive accumulation of genetic mutations within precancerous bowel lesions and polyps. Diagnosis at this stage reduces the risk of death from CRC, giving 90% chance of survival beyond 5 years^{10, 11} and significantly low levels of disease recurrence. However, most cases of CRC are detected once the cancerous cells have moved beyond the middle layers of the colon (Figure 1). This is classified as Stage IIB (Duke’s B) and is one of the most commonly diagnosed forms of CRC. Currently, the course of treatment for CRC patients is fairly similar regardless of the significant differences in the biological features of each CRC case. Usually, the most effective approach is tumor resection, followed by chemo- or radiotherapy for Stage III and sometimes for Stage II CRC patients. However, recent studies suggest that not all Stage III patients benefit from these therapies and that 25% of Stage II cases are under treated.^{12, 13} Furthermore, a number of non-aggressive tumors are frequently overtreated, leading to the patient experience of unnecessary and severe side effects. Methods such as, fecal occult blood test, sigmoidoscopy, colonoscopy, virtual colonoscopy and double contrast barium enema offer improvements in the detection rates of CRCs.¹⁴ However, their diagnostic value is limited with regards to costs, risks, lack of sensitivity especially in early stages and inconvenience to the patient.¹⁵ Therefore, the focus remains on developing efficient methods for the early detection of CRC such as the identification of early disease biomarkers which could be used in non-invasive (urine and blood serum) tests. Such molecular biosensors for CRC would enable widespread screening alongside general health examinations and may further reduce the mortality rate associated with late stage detection of CRC.¹⁶

Figure 1

Shows a malignant primary tumor that has moved beyond the middle layers of the colon and has also metastasized to a nearby lymph node. These are stages IIB (Duke’s B) onwards.

Murphy et al¹⁷ suggested that there are two sub-groups of patients with Stage IIB CRC—those with good treatment outcome (reactive disease) and those with poor treatment outcomes (non-reactive disease). Whether the difference between these two groups of patients can be determined by differences in humoral responses warrants further investigation and provides the basis of our own current studies. Tumor biomarkers offer an opportunity to translate unique CRC biological features into diagnostically pertinent information and would enable personalized treatments, which could inform conventional and immunotherapeutic interventions. This would enable discerning treatment strategies for aggressive and non-aggressive cancers and the clear ‘up front’ distinction of reactive from non-reactive disease.

In part, to address this need, researchers are investigating immune responses in cancer patients to identify new immunotherapy targets and biomarkers. They are hoping to identify and evaluate tumor-associated antigens (TAAs) and cancer-testis antigen (CTAs) that could prove to be efficient diagnostic, prognostic or immunotherapeutic targets in CRC.¹⁸ Recent developments in the fields of genomics and proteomics have greatly contributed to these studies, enabling the identification of multiple potential antigens within a single experiment.¹⁹ Techniques, such as DNA microarray analysis, protein microarrays, peptide-major histocompatibility complex (pMHC) tetramers, serological identification of antigens by recombinant expression cloning (SEREX), serological proteome analysis (SERPA) and peptide elution from MHC for mass spectrometry analysis are now commonly used to evaluate the expression profiles of genes and proteins as well as antigen recognition within different types, sub-types and stages of cancer. Their application in a number of studies conducted on acute myeloid leukemia (AML),^{20, 21} diffuse large B-cell lymphoma,²² lung cancer²³ and osteosarcoma²⁴ have already proven successful and have led to the discovery of a large panel of antigens with prognostic and immunotherapeutic relevance. It has also provided an insight into the biological complexity and individuality of each cancer case, demonstrating considerable heterogenicity among patients and within tumor types. Therefore, genomics, proteomics or a combination of multiple methods could aid the discovery of novel biomarkers specific for CRC and contribute to the development of personalized therapies, which would maximize efficiency and minimize side effects for the patient.

The search for tumor antigens specific for CRC

The desire to identify TAAs has inspired and attracted researchers for more than four decades. Most efforts were driven by the aim to uncover specific epitopes on cancer cells, which could elicit immune responses in the autologous host.^{25, 26} In the 1970s, a method known as ‘autologous typing’ was used successfully to identify a number of antigens including alpha fetoprotein (in hepatoma and germ cell tumors), carcinoembryonic antigen (CEA) (in gastrointestinal cancers), prostate-specific antigen (in prostate cancer), CA125 (in ovarian cancer) and AU (in melanomas).^{27, 28} This novel serological technique offered a substantial improvement over existing methods as it enabled the analysis of the cellular (T-cell defined) anti-tumor human immune responses in an autologous manner where only tumor-specific antigens could be recognized.²⁷ However, autologous typing did not completely fulfill the original hopes for its success because of several limitations. The major disadvantage arose from its reliance on cultured tumor cell lines and not all tumor types could be propagated ex vivo to allow autologous typing to be performed. Furthermore, only a small fraction of the total number of patients was found to have demonstrable levels of autologous antibody with specificity for cell surface antigens on their tumor.^{25, 26} In the cases when the technique was successful and a tumor antigen was identified, the low titer of antibodies that were often detected made any further biochemical or molecular characterization of the antigens almost impossible. However, the method of autologous typing contributed to the identification of a number of human tumor antigens, which were categorized into four classes: differentiation antigens—for example, gp-100; mutational antigens—for example, abnormal forms of p53; retroviral antigens—Epstein–Barr virus and human papilloma virus; and CT antigens.²⁶

CT antigens

CTAs were classified as TAAs with restricted expression in testis, placenta and various types of cancer.²⁶ In the field of cancer serology they were quickly identified as the ideal targets for tumor-specific immunotherapeutic approaches.^{29, 30, 31, 32, 33}

The selective expression pattern of CTAs is considered to be a consequence of gene activation by DNA demethylation^{34, 35} and histone post-translational modifications³⁶ both occurring constitutively in the testis, but also in tumor cells.³⁵ Some CTAs have been found to be expressed in healthy tissues such as liver, pancreas and spleen but at a significantly lower level (<1%) when compared with their expression in testis.³¹ Around 70 families of CTAs have been identified to date and further classified as X-CTAs and non-X-CTAs depending on the chromosomal location to which the genes are mapped.³⁷ The genes for distinct X-CTAs have been previously reported to encode for different antigenic peptides that are presented with HLA class I or HLA class II allo-specificities, eliciting both cell-mediated and humoral immune responses.³⁸ The blood–testis barrier and the lack of HLA class I expression on the surface of germ cells prevents the cells of the immune system from interacting with the CTAs expressed there.³⁹ It has been suggested that as the B and T cells have not been previously challenged by CTAs the immune response should be able to recognize CTAs as non-self structures when expressed on cancer cells³⁹ circumventing the need to break tolerance. These findings suggest that CTAs can be viewed as promising molecular targets for the development of immunotherapeutic interventions for specific cancer types without the possibility of triggering autoimmune responses. The identification of CTAs specific for a particular type of cancer is a promising approach for the development of peptide or recombinant full-length protein anti-cancer vaccines and for antigen-specific adoptive T-cell transfer as part of cancer therapy.⁴⁰ In that respect, NY-ESO-1 (a CTA with strong immunogenicity still detectable at sera dilutions of 1 in 40 000⁴¹) has been of interest in relation to the development of cancer vaccine trials. Several ongoing trials have been based on this antigen alone or its use in combination with an immune enhancer are currently ongoing.^{42, 43} However, the expression of other CTAs within various types of cancer and at different disease stages appears to be heterogeneous and further investigation is required for the identification of effective immunotherapeutic targets.

Although, CTA expression is generally correlated with tumor progression and immunogenicity in various types of cancer, little is known about this in relation to CRC. In addition, CTAs were often found to be poorly expressed (low level, heterogeneous expression) in CRC⁴⁴ highlighting the need for the identification of further immunogenic CTAs involved in CRC development. Such proteins would be candidates for cancer-specific therapy trials and may provide effective biomarkers for diagnosis, prognosis and monitoring of CRC disease progression.

Serological identification of antigens by recombinant expression cloning

The limited success of autologous typing demonstrated the need for a more comprehensive experimental approach for the identification of TAAs. It was not until the mid-1990s when a new autologous immunoscreening technique was shown to circumvent some of the limitations of autologous typing. This method was called SEREX.⁴⁵ It allowed the identification of numerous TAAs based on their recognition by antibodies in diluted pre-cleared autologous patient sera and in a small number of experiments. The advantages of SEREX were quickly recognized and a substantial pool of data encompassing serologically relevant antigens within cancer began to collect. SEREX was capable of identifying immunoglobulin G antibody responses to both highly (NY-ESO-1) and weakly immunogenic TAAs.^{46, 47} Approximately one-third of these antigens were novel and referred to as SEREX-defined antigens. In addition, sera was shown to be able to detect CD8⁺ T-cell recognized TAAs, including NY-ESO-1,⁴¹ MAGE-1 and tyrosinase^{41, 45, 48} as well as TAAs recognized by immunoglobulin G antibodies, which are also known to require CD4⁺ T-cell help (NY-ESO-1).^{45, 46, 48, 49, 50, 51} These findings demonstrate that the cellular and humoral immune system work in concert and are both stimulated by TAAs in the case of cancer. Furthermore, several antigens (NY-ESO-1, MAGE-1 and hMena)⁵² that elicit humoral as well as cell-mediated immune responses have also been successfully identified through SEREX. Studies have investigated CD8+ and CD4+ T-cell recognition of SEREX-defined antigens and the data showed that co-immunization with such proteins in combination with a cytotoxic T lymphocyte (CTL) epitope enhanced CD8⁺ induction in a CD4⁺ T-cell-dependent manner.^{53, 54}

The continuously increasing number of SEREX antigens led to the creation of a Cancer Immunome Database, a free repository for the antigens identified by this serological approach.⁵⁵ The success of SEREX lies in its reliance on the construction of a cDNA library from human tissues (cell lines, primary tumor or normal donor testes) and their expression in a prokaryotic system.⁴⁵ Immunoscreening with SEREX permits quick and effective extraction, processing, sequencing and subsequent molecular analysis of the protein of interest as both the TAA and its coding cDNA are present in the same plaque. However, the use of SEREX has several limitations. It has been previously reported that SEREX-defined antigens are predominantly nuclear proteins which are often transcription factors⁴¹ and/or ubiquitously expressed.^{55, 56} As such, they rarely show evidence of mutations or other structural abnormalities and are often weakly immunogenic, incapable of eliciting and sustaining strong humoral immune responses.²⁵ In addition, phage can only express proteins in their primary structure, which may cause the failure to detect antigenic sequences consequent to eukaryotic post-translational modifications.⁵⁷ Regardless of the limitations, SEREX-defined antigens are already being investigated as diagnostic markers (testis-specific protease (TSP50) in ovarian and CRCs)⁵⁸ and as immunotherapeutic targets (OVA66 in ovarian cancer,⁵⁹ KIF20A in pancreatic cancer⁶⁰).

Serological proteome analysis

SERPA is a powerful tool used for the identification and validation of immunogenic TAAs. Similar to SEREX, it uses the antibody repertoire contained within the sera of a cancer patient to successfully identify TAAs.⁶¹ In comparison to SEREX, SERPA does not require the use of a cDNA library, making this method less time consuming and less labor intensive. Furthermore, SERPA is better suited for the detection of possible post-translational modification and protein isoforms as it relies on the separation of complex mixtures of proteins extracted from cell cultures or tumors. The separation is performed via the use of a two dimensional gel electrophoresis (2-DE).⁶² A number of TAAs have been identified to date using SERPA in various types of cancer including lymphoma,⁶³ renal cell carcinoma,⁶⁴ ovarian cancer,⁶⁵ and CRC.⁶⁶ These studies have established the specificity of SERPA and its great potential in uncovering immunogenic TAAs and identifying tumor markers. However, similar to SEREX, most of the antigens identified by SERPA are predominantly weakly immunogenic intracellular proteins and very rarely membrane-associated ones.^{67, 68} This and the finding that a number of antigens found by T-cell cloning have also been found by SEREX suggests that most antigens induce B and T-cell responses,⁶⁹ although not necessarily to the same epitope(s). In addition, SERPA has demonstrated a drawback associated with the use of 2-DE: low abundance (such as regulatory and signal transduction proteins and receptors⁷⁰), hydrophobic or insoluble proteins (such as membrane proteins⁷⁰) are inherently difficult to detect.⁶⁷

SEREX and SERPA are two methodologies that seem complementary to each other as they identify two different sets of antigens (SEREX for the identification of antigens with altered expression; SERPA for the identification of antigenic proteins resulting from post-translational modifications). Therefore, the identification and validation of CRC-related CTAs using SERPA and SEREX independently or as a combined approach is worthy of further consideration.

Recombinant antigen expression on yeast surface (RAYS)

RAYS is another serological strategy applied to the process of discovering immunogenic CTAs. RAYS permits the expression of immunogenic proteins on the surface of yeast allowing for a more natural folding of the protein and partial glycosylation (by virtue of it being a eukaryotic system).⁵⁷ This permits the analysis of proteins in their natural conformation when compared to the prokaryotic expression utilized by SEREX. This method has demonstrated specificity and sensitivity for the detection of an antibody response to a conformation-dependent epitope—the CRC antigen A33. Therefore, the CTAs that have escaped detection because of the fact that they elicit immune responses only after undergoing the appropriate post-translational modifications could potentially be identified via RAYS. To date, RAYS has allowed the confirmation of antigen immunogenicity through screening of eukaryotic cDNA expression libraries derived from pancreatic cancer⁷¹ and prostate cancer.⁷² RAYS offers a less time-consuming analysis of the serological autoreactivity in cancer patients and it has provided an effective anti-cancer vaccine platform (recognizing NY-ESO-1) in prostate cancer patients.⁷² However, its further development is required to allow the detection of novel target antigens (such as is achieved with SEREX or SERPA) and its application in CRC requires further analysis.

Serum antibody detection array

To evaluate the seroreactivity of a number (or a panel) of SEREX-defined antigens in a particular type of cancer, a spot immunoassay, known as serum antibody detection array, has been successfully developed and utilized.⁷³ Although, several antigens have been evaluated in cases of colon cancer (MAGE-A3, SSX2 and NY-ESO-1⁷³), additional improvements in the sensitivity of serum antibody detection array are still required.

Protein microarrays

Protein microarrays allow the rapid and easy detection of tumor antigens using patient sera.^{74, 75} Recent studies have demonstrated the efficiency of the technique in the rapid identification of immunogenic membrane-based TAAs with high reproducibility of the experimental analyses of lung and brain⁷⁴ and ovarian⁷⁶ cancers. In this context, protein microarrays has a great advantage as this methodology allows the construction and simultaneous analysis of a large panel of candidate tumor biomarkers (∼9,000). However, it should be noted that although protein microarrays have vast screening potential they are limited to defined proteins from an albeit not insubstantive pool.

Understanding the molecular interactions between T-cell receptors on CTL and peptide/MHC class I complexes on tumor cells is an essential tool for the development of immunogenic vaccines.⁷⁷ Several such vaccines have already been designed based on TAA epitopes and have been implemented in Phase I and II clinical trials for different types of cancer (human papilloma virus,⁷⁸ WT1,⁷⁹ human telomerase reverse transcriptase⁸⁰ and HLA-A24⁺HRPC⁸¹ peptide vaccinations). The strength of vaccine-mediated immunological responses generally need to be enhanced and this will be much more feasible in patients in subsequent (Phase III) clinical trials who are likely to have less advanced disease. However, the results obtained from clinical trials to date warrants further investigation.

pMHC (tetramer) microarrays

The development of pMHC microarrays has allowed the rapid identification of antigen-specific populations of T cells in the peripheral blood of patients.⁸² This approach has proven useful for epitope prioritization, and for the detection of multiple T-cell populations in cancer patients undergoing conventional treatments or tumor-associated peptide vaccine trials.^{82, 83} We are using the pMHC array to identify which epitopes are recognized by peripheral T cells from colon cancer patients during conventional treatment (Bonney et al., in preparation).

Tumor profiling in patients with CRC

A number of studies have been conducted over the last decade aiming to identify novel biomarkers that would prove to be efficient in CRC profiling as prognostic, predictive or therapeutic biomarkers. The recent improvements in proteomics and genomics methods has greatly aided this aim and has led to the identification of a number of CTAs and TAAs relevant to CRC. However, the clinical significance of only a small fraction of these potential markers in CRC has been evaluated to date.

CTAs in CRC

NY-ESO-1 has been recently studied in relation to CRC. It was demonstrated that some CTAs are capable of eliciting strong humoral and cell-mediated immune responses in some patients with CRC.⁸⁴ However, its expression in CRC is often highly heterogeneous, when present, which poses an obstacle in the development of a generalized immunotherapy. As discussed earlier in this review, NY-ESO-1 has also been targeted in several vaccine clinical trials worldwide involving CRC patients. Recently, a new Phase I trial of a fusion protein vaccine is being organized targeting solid tumors expressing NY-ESO-1 and this investigation includes Stage I–IV CRC.⁸⁵ In addition, NY-ESO-1 expression was shown to correlate with CRC stages and local lymph node metastasis⁸⁶ making it a potential prognostic biomarker for CRC.

TSP50 was originally identified as abnormally expressed in breast cancer cells⁸⁷ and has recently been studied for the first time in CRC patient samples. The expression of TSP50 was found to correlate with the clinicopathological characteristics and disease-specific survival for CRC patients.⁸⁸ Furthermore, the study demonstrated that TSP50 expression is highly specific for CRC as compared with colorectal adenomas and normal tissues, allowing the easy differentiation between them. TSP50 is an attractive predictive biomarker for poor survival in patients with early stages CRC (Stage I and II), but not in patients with advanced stage disease. As such, it is the only effective predictive biomarker reported to date for patients with early stage CRCs.⁸⁸

CABYR is a calcium-binding tyrosine phosphorylation-regulated fibrous sheath protein and its expression was first identified in human spermatozoa (⁸⁹; reviewed in Chiriva-Internati et al.).⁹⁰ Subsequent detection of CABYR in lung carcinoma and its absence in healthy tissues, led it to be considered as a novel CTA, which has been shown to have some immunogenic properties that could serve as a basis for the development of immunotherapy for cancer patients.⁹¹ Additional studies on CABYR expression in brain,⁹² hepatocellular⁹³ and other carcinomas⁹⁴ have shown that there are at least five different isoforms of this protein, which could have unique roles in the process of carcinogenesis. Following these discoveries a recent study has reported a frequent overexpression of CABYR a/b and c isoforms in CRC tumors when compared with adjacent normal tissues.⁹⁵ However, a more comprehensive investigation is required to determine whether CABYR expression correlates with tumor stages and is a suitable therapeutic vaccine candidate in CRC.

SPAG9 is another antigen found to be exclusively expressed in testis⁹⁶ that is a particularly attractive target for immunotherapy in epithelial ovarian cancer,⁹⁷ thyroid cancer⁹⁸ and in CML.⁹⁹ A recent study has investigated the expression of SPAG9 in CRC patients aiming to explore its possible role in colon cancer tumorigenesis and its effectiveness in eliciting a humoral immune response.¹⁰⁰ Interestingly, the study has reported a close relationship between SPAG9 expression and early stages of CRC development suggesting that it could serve as an early diagnostic biomarker for CRC patients. The investigation had also demonstrated that SPAG9 could have a key role in the tumor development and could also serve as a target for the development of immunotherapeutic methods.

Other CTAs, found to exhibit a strong correlation with CRC presentation are listed in Table 1.

Table 1 Antigens determined to have potential as diagnostic, prognostic or immunotherapeutic targets in CRC

TAAs in CRC

CD133 is a cell surface protein marker found on undifferentiated cancer cells that exhibit stem-like properties. These cells account for the propagation, growth and recurrence of AML^{110, 111} and CRC^{112, 113} and for the resistance of these cancers to current therapies. The functional importance of CD133 expression has been investigated in several studies in relation to the initiation and behavior of CRC.^{113, 114, 115} The CD133⁺ CRC stem cells are reported to have exhibited the ability to transfer cancer to a secondary recipient maintaining the same immunophenotype and the global gene expression profile of the primary tumor when compared with CD133⁻ cells.^{113, 114} Furthermore, several studies have clearly identified the correlation between CD133 expression alone^{114, 115} or in combination with other protein markers¹¹⁶ with CRC patient survival and have revealed it to be a reliable prognostic marker. In combination with CD44 and CD166 (cell surface protein markers), CD133 expression has also been linked to the presentation of low-, intermediate- or high-risk CRC cases with the ability to distinguish between them at an early stage of the disease (stage II).^{114, 116} CD133 is a promising predictive and prognostic marker in the diagnosis of CRC and particularly of interest as it is applicable to the early stages of the disease. However, the presence of such cell surface markers on the CRC stem cells has not been investigated in relation to the underlying cause of the non-reactive type of CRC.

CEA is a TAA whose expression levels are often monitored pre- and post-treatment in CRC patients as they have been found to be indicative of cancer recurrence¹¹⁷ and poor disease prognosis.^{118, 119} Patients with elevated post-treatment CEA expression levels are often monitored more carefully for relapse of CRC and for local or distant recurrence.^{117, 119} To date, several studies have investigated the potential of CEA as an immunotherapeutic target in cases of CRC. Different research strategies have incorporated CEA peptides or CEA mRNAs in dendritic cell vaccines^{120, 121} and in plasmid DNA vaccines¹²² demonstrating that these vaccines are well-tolerated and have immune-stimulatory capacity in patients with CRC. However, the overall outcome of these studies indicated that additional vaccine modulation is necessary to attain significant clinical impact. More recently, a study to investigate whether the vaccination of toll-like receptor activated dendritic cells can induce more potent CTL responses and antitumour activity in CEA transgenic mouse tumor models was published.¹²³ It has demonstrated that the combined activation of TLRs can lead to better maturation status of dendritic cells and can also induce more effective antitumour immune responses against CRC. However, additional investigation is necessary to evaluate the effectiveness of this approach in human models.

Clinical significance of CTAs in CRC

The CTAs that have already been identified within different types of cancer could serve as biomarkers for discerning aggressive and non-aggressive cancers and for predicting treatment outcome and relapse. Their expression patterns and clinical significance is still under investigation, but there are very promising early results. Some of their clinical applications are described in the followings sections, and could expand the list of potential CTAs in CRC.

Potential biomarkers for discerning reactive from non-reactive disease

The expression profile of CTAs in relation to treatment outcome in particular types of solid cancers has been previously studied on several occasions^{24, 52, 58, 59, 60} but further investigation is necessary to compare these findings with CRC cases. In this context, a recent study has demonstrated that γ-irradiation de novo upregulates the expression of various CTAs and MHC-I in a randomized fashion. Therefore, irradiation could be accounted responsible for the increased immunological response to certain tumors owing to the inflammation and cell damage it causes. This would be anticipated to cause the immune system to attend the site of damage, mop up cellular debris and present proteins including CTAs to the immune system.¹²⁴ These findings fit with demonstrations in leukemia that elevated tumor antigen expression at disease presentation is associated with improved survival.^{125, 126} Identifying and evaluating such TAA and CTAs would be beneficial for profiling individual tumors and for combining radiotherapy (or other cancer therapy approach) with immunization to maximize the effect of treatment in CRC. However, in order to design a combined treatment a thorough understanding about the mechanisms of initiating CTAs expression and the likely order of their expression is necessary.

The exact role of the numerous CTAs in relation to tumor response to various treatments still remains poorly defined. Particularly, in cases where adjuvant therapy is in order, it would be beneficial to have a panel of biomarkers to predict the likely success of the therapy. Several CTAs have been evaluated in gastrointestinal stromal tumor—with regards to recurrence, while levels of MAGE-A1, MAGE-A3, MAGE-A4, MAGE-C1 and NY-ESO-1 expression were investigated in response to imatinib adjuvant therapy.¹²⁷ This study demonstrated that CTA⁺ gastrointestinal stromal tumors had a significantly shorter recurrence free survival compared with negative cases. Furthermore, the expression of NY-ESO-1 and MAGE-A3 was associated with elevated resistance to imatinib and therefore, with continuous tumor progression. Luetkens et al.¹²⁸ have also showed that PRAME expression remains stable under imatinib treatment and correlates with decreased overall survival in patients with CML.

Similar findings have been reported in several studies of prostate and lung cancer, multiple myeloma, AML, liposarcoma and others.^{129, 130, 131, 132, 133, 134} These suggest an important role of CTAs in the pathophysiological behavior of different tumors in response to treatments. Further investigation of a panel of antigens associated with particular types of cancer and their relationship to either reactive or non-reactive disease is yet to be attempted. In this relation, TSP50 is the only known CTA to have been characterized as a biomarker for disease prognosis in CRC,⁸⁸ but its association with treatment outcome is still to be analyzed. Furthermore, investigation of the interrelationships between groups of CTAs in CRC and their expression profiles could lead to significant discoveries about the underlying cause of a particular treatment response. This could benefit the design of a multivalent cancer vaccine targeting several antigens rather than just a single one.

Potential biomarkers for survival prognosis

Several studies have investigated the role of the CTAs as biomarkers of prognostic value regarding patient survival. Elevated levels of expression of particular CTAs have been shown to correlate with poor survival prognosis, particularly in solid tumors. In contrast to expectations some antigens have been shown to have above average levels and have better survival rates in patients with hematological malignancies. For example, expression of SSX2IP in the presentation of AML has been shown to predict good survival in patients with no detectable cytogenetic rearrangements.¹²⁵ The elevated levels of antigens provide targets for improved immune responses in patients post-conventional (chemotherapy, radiotherapy) treatment, when there is cancer cell damage and inflammation (danger) signals stimulating an immune response to clear up dead and dying cells. Furthermore, the CTAs expressed on non-solid tumors (such as AML) are more accessible and ‘easier to see’ by the immune system as they are not hidden within heterogeneous layers of cancer cells (as seen in solid tumors).

The prognostic value of CTAs has also been evaluated in the presentation of osteosarcoma by gene microarray where the high expression of MAGE-A could predict distant metastasis and poor survival. For patients with and without MAGE-A expressing tumors, the 5-year survival rates were found to be 39.6% and 80% respectively.²⁴ Similarly, increased levels of WT1 (another CTA) expression have been shown to correlate to poor prognosis and relapse in pediatric AML after induction therapy.¹³⁵ A number of products of translocations have been used in routine labs, which detect minimal residual disease and can indicate impending relapse with high accuracy.¹³⁶ The CTA expression in relation to patient survival in CRC cases has been the focus of few studies and further investigation is required to establish the clinical significance of this relationship.⁸⁸

Potential biomarkers for discerning aggressive from non-aggressive disease

A recent study on prostate cancer has demonstrated that several CTAs are preferentially expressed in either aggressive or non-aggressive disease.¹³⁷ Such biomarkers could be particularly useful in preventing the overdetection and overtreatment of potentially indolent CRC or the undertreatment of a more aggressive type of disease.

Potential immunotherapeutic targets

In 2005, cancer patients expressing NY-ESO-1 and LAGE-1 antigens entered a Phase I clinical trial on a plasmid DNA (pPJV7611) cancer vaccine,¹³⁸ demonstrating the effectiveness of CTAs as immunotherapeutic targets. Recently, the first clinical trial of cancer vaccine therapy with artificially synthesized helper/killer-hybrid epitope long peptide of MAGE-A4 cancer antigen was initiated.¹⁰⁴ A patient with pulmonary metastasis of CRC was vaccinated and had shown a significant reduction in the tumor growth.

Non-reactive disease—overcoming the underlying issues

CTAs are key molecules in the field of cancer serology. Their progressively expanding family is continuously providing potential novel targets for cancer immunotherapy or diagnostic/prognostic examinations. However, CTAs have also been evaluated for their role in oncogenesis, particularly their contribution to the immortality, invasiveness, immune evasion, hypomethylation and metastatic capacity of the neoplasms.¹³⁹ Investigating the correlation between CTAs and the clinical presentation of CRC (reactive or non-reactive type) requires knowledge of the molecular mechanisms that govern their expression and physiological functions—mechanisms that still remain poorly understood. According to recent studies, genome wide hypomethylation has accounted for the aberrant expression of CTAs within CRC cells.^{140, 141, 142} However, the epigenetic factors that dictate which panel of silenced genes to be reactivated and the physiological properties of these particular CTAs are still unknown. This is an area of actively ongoing research as it is believed that these factors are responsible for the heterogenicity in the CTA expression profiles of the individual CRC cases. A better understanding of the structural and serological properties of these antigens, their modes of expression and functions, should establish whether they are the ones governing the reactive and non-reactive CRC phenotypes.

CTAs in CRC have been associated with low immunogenicity under normal conditions, and only a few patients actually recognize the peptide epitopes and exhibit strong CTL or humoral immune responses to CTAs. This could be partially accounted to the structural stability of the proteins, encoded by the CT genes. It has been previously demonstrated that proteins must be cleaved to small peptides by intracellular proteinases prior to presentation to the immune system.¹⁴³ However, high stability proteins (α-helical secondary protein conformation known to provide the most optimal structural stability¹⁴¹) are less likely to undergo denaturation and polypeptide chain unfolding. Therefore, CTAs containing large proportions of α-secondary structure should be more resistant to protein unfolding and subsequent cleavage by the proteinases, thus resulting in an unsuccessful presentation to CTLs (low immunogenicity). For example, SCP-1 (a CTA of particularly low immunogenicity) has been shown to contain 76.6% α-helical structure from its total secondary conformation when compared with the highly immunogenic NY-ESO-1 CTA, containing only 20.7%.¹⁴¹ The identification of such antigens that are predominantly expressed in non-reactive CRC patients could potentially be accounted for the therapy-resistant presentation of the disease. In such cases epigenetic modulation to induce expression of other CTAs may highly favor the immunotherapeutic approach to non-reactive CRC disease. In this respect, a study has recently reported a successful induction of NY-ESO-1 expression in CRC cells (but not in normal nontransformed cells) both in vitro and in vivo subject to hypomethylating agent 5-aza-2′-deoxycytidine (DAC) treatment.¹⁴⁴ The study reports that DAC-treated CRC cells are susceptible to MHC-restricted recognition by CD8⁺ NY-ESO-1-specific T cells. A NY-ESO-1_157–165-specific T-cell receptor was successfully used to generate both CD8⁺ and CD4⁺ NY-ESO-1_157–165-specific T cells that selectively recognized DAC-treated CRC cells but not non-treated cells. These data reveal the great potential of combining epigenetic modulation and adoptive transfer of genetically engineered T lymphocytes targeting NY-ESO-1 or other CTAs for the development of a specific immunotherapy for CRC.

Another approach could also include the introduction of sequences capable of disrupting long α-helical stretches in the regions outside the potential epitopes.¹⁴¹ Such approaches have the potential to improve the treatment response in patients with non-reactive type CRC. However, several challenges remain to be overcome, including the insufficient antitumour responses due to immunosuppression driven by T lymphocytes known as regulatory T cell (Tregs).

Infiltrating T lymphocytes

During the last decade, the search to understand the causes underlying presentation CRC has focused on Tregs. They have been shown to have a major role in cancer immunoevasion by directly inhibiting or even eradicating both CTLs and T-helper lymphocytes.¹⁴⁵ Tregs suppress autoreactive T cells without killing them through incompletely understood, contact-dependent mechanisms.¹⁴⁶ In healthy individuals, they represent 5–10% of the population of CD4⁺ lymphocytes. However, in cancer patients, Tregs may increase up to 30% and are predominantly found among the tumor-infiltrating lymphocytes present.¹⁴⁵ Therefore, the failure of the T and B cells to recognize and eradicate immunogenic cancer cells could be accounted for by the ability of certain tumors to secrete chemokine CCL22, which recruits Tregs and immobilizes the function of the anti-cancer immune response. Indeed patients with ulcerative colitis and irritable bowel syndrome have been shown to have higher numbers of infiltrating T cells than healthy controls¹⁴⁷ which may have a role in controlling cancer-driven inflammation. Tregs in CRC have been shown to exhibit both pro- and anti-tumor activities governed by the level of inflammatory stimuli received and dependent of the phase of tumourigenesis (early or late).¹⁴⁸ Following treatment in a Phase II clinical trial for CRC patients, the level of circulating Tregs was reported to have almost reached normal levels accompanied by 70% increase in CTLs responses against CEA epitopes.¹⁴⁹ Therefore, it still remains to be determined whether presence of Tregs in CRC has a pro- or anti- tumor role, how this correlation changes with the stage of the disease and the clinical significance of these changes.

In addition, Tregs appeared to be highly specific for a distinct set of TAAs in CRC patients, suggesting that Tregs exert T-cell suppression in an antigen-selective manner.¹⁵⁰ Several key hallmarks of Tregs in CRC have been identified, highlighting their complex role in the progression of the disease, the survival prognosis and their potential as therapeutic targets. Overall, it was established that CRC patients develop multivalent and individual T-cell responses against a broad variety of different CRC-associated TAAs. Therefore, selecting a panel of antigens according to pre-existing T-cell responses as an intermediary could improve the efficacy of future immunotherapies for CRC and should be further investigated. Better understanding of the role and behavior of Tregs within CRC would improve tumor profiling on an individual basis and could aid the choice of the most adequate therapy for each individual case. It could also benefit the development of cancer vaccines and other immune-based therapies targeting particular CTAs in non-reactive type CRC.

Tumor-infiltrating CD45RO(+)-cell density is a prognostic biomarker associated with longer survival in CRC patients, independent of clinical, pathological and molecular features.¹⁵¹ Similar findings have also been reported from another study, linking the density of CD45RO(+) memory T cells in the different regions of CRC with dissemination to lymphovascular and perineural structures and to regional lymph nodes in patients—low densities were associated with a very poor prognosis.¹⁵² However, not only was the type and density of the infiltrating immune cells within human CRC predictive of clinical outcome but also their location within the tumor, that is, whether they are located in the center of the tumor or the invasive margin. A strong in situ immune reaction in both regions was shown to correlate with a favorable prognosis regardless of the stage (I, II and III CRC).¹⁵³ Conversely, a weak in situ immune reaction predicted a poor clinical outcome even in patients with stage I CRC. These data were further supported by Pagès et al.¹⁵⁴ who examined the relationship between the presence of CD8(+) and CD45RO(+) in two regions of the tumor in patients with early stage CRC with regards to tumor recurrence and overall patient survival. High densities of both CD8(+) and CD45RO(+) correlated significantly with lower rates of CRC recurrence and increased overall survival when compared with patients with low densities of both immune cell types within the primary tumors. In contrast, extramural vascular invasion and high FOXP3⁺ cell density in lymphoid follicles were independent factors for worse survival, whereas a high frequency of lymphoid follicles in histologically normal colonic mucosa was associated with better survival.¹⁵⁵ Therefore, the collective analysis of the type, density and the location of immune cells within the CRC could be used to predict patient survival and to identify the high-risk patients who would benefit most from adjuvant therapy.

Intrinsically disordered proteins (IDPs)

Following a bioinformatics approach that implements the application of two algorithms (FoldIndex and RONN) predicting the level of disorder within a sequence, the experimental outcome revealed that >90% of the examined 228 CTAs were IDPs.¹⁵⁶ The latter are proteins that lack the typical hydrophobic cores and therefore do not appear as having rigid 3D structures (along their entire length or in localized regions) under physiological conditions and instead, exist as dynamic ensembles.^{156, 157} However, IDPs can evade being detected as ‘misfolded’ and degraded by the cell’s surveillance system through their ability to undergo ‘disorder-to-order’ transitions upon binding to biological targets—a paradox known as the ‘order/disorder paradox’.¹⁵⁸ This is achieved as a segment of an IDP initially binds with the target, followed then by coalescing of the other protein segments facilitating the IDPs’ folding.¹⁵⁹ However, recent studies have challenged this general view by revealing a phenomenon—uncoupled binding and folding of IDPs.¹⁶⁰ The complexity of the binding mechanisms of IDPs has been investigated by others^{161, 162} and a summary of possible pathways have been outlined in Figure 2.

Figure 2

Diagrammatic representation of the different binding mechanisms and the disorder-to-order transition that the IDPs undergo before and upon binding to their targets (based on reference¹⁶⁰). Some IDPs (particularly immune signalling-related IDPs) do not fold upon binding to their targets (for example, Intermediate 1). Other IDPs undergo partial or complete folding upon binding (Intermediate 2 and Intermediate 3) or fold before binding to their targets (Intermediate 4). The images shown are representative of the disorder-to-order transition that a hypothetical IDP would undergo.

The above mentioned properties associated with IDPs give an interesting angle of perception towards the expression, behavior and function of CTAs identified as IDPs. The lack of rigid 3D structures is believed to be responsible for the exposure of primary contact sites, which enables the faster, more effective and promiscuous binding at high concentrations to target molecules.¹⁵⁷ Together with the fact that intrinsic disorder has been identified as a determinant of genes that are harmful when overexpressed,¹⁶¹ this could account for the correlation between CTA overexpression and disease prognosis. This is further supported by the fact that CTAs appear to occupy ‘hub’ positions (highly connected protein nodes) within the complex protein–protein interaction (PPI) network.^{156, 163} What was interesting in these findings was the fact that these networks are dynamic and grow incrementally by establishing new nodes. However, a desirable protein for recruitment to a hub position is a protein that is likely to participate in a large number of promiscuous interactions when overexpressed such as a CTA that is an IDP.¹⁵⁶ Eventually, the PPI network becomes dominated by such hubs leading to the overexpression of CTAs and to the creation of nodes with novel functions. This accounts for the poorer disease prognosis in later CRC stages and the phenotypic presentation of non-reactive CRC disease. This idea is further supported by the fact that a high abundance of IDPs is believed to result in undesirable interactions and potentially harmful effects of such interactions.¹⁶¹ Perhaps the failure to respond to treatment in non-reactive type CRC could be due to targeting common nodes in a protein network rather than CTAs that occupy hub positions. Therefore, the search for such antigens could potentially take a turn towards more thorough investigation of the structure of the PPI networks established in cases of reactive disease.

Conclusions

Research over the last four decades has proven that CTAs hold a prominent role within the field of cancer serology and could be excellent diagnostic, prognostic and immunotherapeutic biomarkers in CRC. Furthermore, examination of their structural properties and selective expression patterns may enable the understanding of the underlying molecular mechanisms that govern the reactive and non-reactive presentation of CRC. Nevertheless, the panel of CTAs with potential clinicopathological significance in relation to CRC still remains to be conclusively clarified. As this review pointed out, a large number of CTAs to date have been proven to be excellent prognostic biomarkers or targets for the development of peptide or protein vaccines in different solid tumors. However, the attempts to characterize such potent targets and to evaluate their significance in relation to CRC are still at their starting point. Up to date, the comprehensive analysis of the CTAs in relation to CRC have demonstrated that their secondary structural conformation and the presence of a large number of α-helical formations may be accounted for the reduced immunogenicity and treatment response of the CTAs associated with non-reactive type CRC. A solution to this potential problem has also been recently identified and evaluated—epigenetic modulation, which could induce the expression of highly immunogenic CTAs, that could subsequently be targeted by chemo- or immunotherapy. Particular attention should also be paid to the infiltration of Tregs within the tumor mass at different stages of the disease. Tregs have been associated with a rather complex and ambiguous role in the process of tumourigenesis in CRC patients as they have been demonstrated to exhibit both pro- and anti-tumor activities. Further study would be require to gain a better understanding on whether they play different roles in the presentation of reactive versus non-reactive CRC. Nevertheless, research on CTAs expression in CRC should follow the findings that these proteins are also exhibiting particular properties characteristic for IDPs.

In summary, the numerous studies and the significant data gathered regarding the expression pattern of CTAs and their role in carcinogenesis have not yet provided an explanation for the differences in the clinical presentation of CRC. However, they have demonstrated the complex variety of CTA expression mechanisms and their implications in hematological and solid cancers. The fact that certain CTA expressions could be indicative of poor prognosis in solid cancers but suggest good overall outcome in hematological malignancies brings to the fore a number of new challenges on the journey towards advancing clinical tumor immunotherapy. Still the main question remains unanswered: Can we transfer what triggers a good response in reactive disease to a patient with a poor anti-tumor response and improve the overall outcome? Identifying highly immunogenic peptides is a basic requirement for the development of immunotherapy vaccines. However, solid tumors, such as CRC pose additional obstacles on the road to mounting an immune response due to the fact that the antigens expressed within CRC cells are not readily accessible/easily recognized by the immune system. The heterogenicity in the nature of the tumor cells comprising CRCs and their multi-layered structure are the main reasons for evading immune recognition. However, understanding the underlying molecular basis of reactive disease could lead researchers to overcome these issues. Despite the amount of data gathered to date, the search for tumor-specific immunotherapy targets has remained elusive but as our understanding of how to induce effective immune responses grows so does our likelihood of improving patient outcomes.

Future perspectives

The identification of novel CTAs for CRC and evaluating the clinical significance of existing ones should be one of the main priorities in this field. As previous research has indicated, a number of CTAs could be effective prognostic biomarkers in CRC when evaluated as a panel of antigens rather than individually. The same approach could also be promising for the development of multivalent cancer vaccines that will target a panel of antigens in a particular tumor type.

It would also be beneficial to investigate the implications of high expression of novel CTAs on patient survival and treatment outcome. Furthermore, understanding the mechanisms of CTAs expression and epitope presentation to CTLs might provide new targets for epigenetic modulation or adjuvant therapies. Nevertheless, investigating the role of the known CRC-associated CTAs in relation to particular drug treatments or therapies should not be overlooked. As discussed earlier, a combined approach for the treatment of CRC might hold the key to the development of more successful therapies for CRC patients. Treatment with hypomethylating drugs on cancers with low CTA expression followed by immunotherapy has already shown promising results in both in vitro and in vivo studies. These findings should be further investigated in cases with CRC to provide a more conclusive idea of their effectiveness.

As suggested already, a more thorough investigation should enable the discovery of CTAs, which are also IDPs. Understanding the properties and the function of IDPs could assist a more comprehensive understanding of the physical and serological properties of CTAs and how they relate to CRC reactivity.