Tumour antigen-targeted immunotherapy for chronic myeloid leukaemia: is it still viable?

Abstract

In haematological cancers, malignant cells circulate in the blood and lymphatic system. This may make leukaemic cells easier to target by immunotherapy than in other types of cancer. Various immunotherapy strategies have been trialled in several leukaemias including chronic myeloid leukaemia (CML) and in general, these have been aimed at targeting tumour-associated antigens (TAA). There are numerous TAA expressed by CML patients including WT1, proteinase 3, BCR-ABL and HAGE amongst others. The immunogenicity of the CML-specific tumour antigen, BCR-ABL, has been the subject of much debate and its role in the development of the disease and its unique sequence spanning the breakpoint region make it an ideal target for immunotherapy. However, there are a limited number of immunogenic epitopes across the junctional region, which are restricted to only a few HLA types, namely A2, A3 and B7 (Clark et al. in Blood 98:2887–2893, 2001). The second CML-associated antigen is the helicase antigen HAGE, a cancer-testis antigen found to be over-expressed in more than 50% of myeloid leukaemias (Adams et al. in Leukaemia 16:2238–2242, 2002). Very little is known about the function of this antigen and its significance to CML. However, its membership of the DEAD-box family of ATP-dependent RNA helicases and the involvement of other members of this family in tumour cell proliferation (Eberle et al. in Br J Cancer 86:1957–1962, 2002; Yang et al. in Cell Signal 17:1495–504, 2005) suggest a crucial role in the RNA metabolism of tumour cells. For these reasons, HAGE also seems to be a good target for immunotherapy as it would be applicable for the majority of patients with CML. This review aims to discuss the potential of immunotherapy for the treatment of leukaemia, in particular CML, and the prospect of targeting three CML associated antigens: BCR, ABL and HAGE. During his career, Prof. Tony Dodi made a significant contribution in this area of leukaemia research, confirming the identity of immunogenic HLA-A3 and B7-restricted peptides as targets for CTL. Published, as a highlighted paper in Clark et al. (Blood 98:2887–2893, 2001), this study demonstrated the expression of MHC-peptide complexes on the surface of CML cells and the presence of tetramer-positive CTL activity in CML patients positive for these two HLA alleles. His drive and dedication for research excellence will be remembered by all who knew and worked with him.

Chronic myeloid leukaemia: why focus on immunotherapy?

Chronic myeloid leukaemia is a malignant myeloproliferative disorder of pluripotent haemopoietic stem cells that is characterised by the presence of a Philadelphia chromosome, formed as a result of a translocation t(9:22) [22, 31]. The outlook for CML has been transformed by the advent of the tyrosine kinase inhibitor (TKI) imatinib (Gleevec) and more recently, second generation TKIs. However, it appears unlikely that TKIs alone are curative and whilst allogeneic stem cell transplant (alloSCT) still remain an effective longer lasting therapy, they are only feasible in 30% of patients [25]. Interferon α, hydroxycarbamide and TKIs are also not capable of eradicating leukaemic stem cells [21, 24]. There is evidence that immune responses play a part in the development of CML [26]. In particular, T cells have been implicated in the control of CML; this has been shown by the fact that donor lymphocyte infusions (DLI) can successfully complete remission in up to 90% of CML patients, especially if used in early molecular relapse [4, 25, 29]. Furthermore, T cell depletion of alloSCT grafts results in a higher incidence of relapse [32].

A number of CML-specific and tumour-associated antigens (TAA) have been identified and studied, including Wilm’s tumour protein (WT1) [5], proteinase 3 [27], helicase antigen (HAGE) [33, 34], and BCR-ABL [12]. Various peptides from these antigens have been used in trials, and have been shown to activate CML-specific T cell responses [36]. These responses depended on the epitopes used, and were either or both of a specific CD4 + T cell proliferative response or peptide-specific CD8 + T cell interferon-γ production [9]. In this review, the potential of BCR, ABL and HAGE as immunotherapeutic targets will be examined including some recent results.

Leukaemia and vaccine strategies

Leukaemia is an ideal candidate for studying the effectiveness of cancer vaccines because leukaemic cells circulate in blood and lymphatic systems making them accessible to cells of the immune system. Key to development of these immunotherapies is the discovery of tumour-specific antigens, which provide a way to distinguish between tumours and normal cells [1]. Following this, specific immunogenic epitopes can be identified from these tumour antigens to either provide a way to measure the presence of specific T cells in the circulation, using tetramer technology and flow cytometry [35], or for use as vaccines to activate an immune response against tumour cells [7]. Cancer vaccines developed in relation to the treatment of leukaemia include DNA-, dendritic cell (DC)- and peptide-based vaccines as well as cellular adoptive therapy.

DNA encoding for tumour antigens can be given directly as a vaccine. These are simple, safe to use and easily manufactured in bulk [3, 16, 53]. DNA vaccines can induce both humoral and cell-mediated responses [54, 56] and may also help avoid immunodominance and pre-existing immunity [58]. However, unlike viral vaccines, the capacity of these types of vaccines to break tolerance in vivo and provide long-lasting immunity is limited [52]. In a clinical setting, this limited potency means that the vaccine can fail to provide effective immunity [53]. DNA vaccines have been used in both mouse and human trials for the treatment of leukaemia. In a mouse model of acute promyelocytic leukaemia, a DNA vaccine was used with positive results. This vaccine was made by the fusion of the human promyelocytic leukaemia-retinoic acid receptor-α (PML-RARA) oncogene to tetanus fragment C (FrC) sequences [42]. It was shown to be particularly successful when used in combination with all-trans retinoic acid (ATRA). In a clinical setting, a DNA vaccine, which encoded for a chimeric Ig molecule derived from the patient’s tumour and linked to mouse Ig heavy-chain and light-chain sequences, was used to treat follicular B cell lymphoma. Seven of the 12 patients treated showed a humoral and/or a cellular proliferative response to the mouse component of the vaccine [36].

Another type of vaccine involves the use of DC, which are ideal “natural” adjuvants as they are potent mediators of specific immune responses [41, 45]. One of the main reasons for this potency is that they cross-present to activate both CD4 + and CD8 + T cells, something other vaccines such as peptide vaccines fail to do. New technologies have allowed the generation of high quantities of autologous DC from monocytes or CD34 + cells of patients by culturing them in vitro in the presence of GM-CSF and IL-4 [50]. These can then be loaded with tumour antigens either in the form of peptides, protein, lysate, DNA or RNA from tumour extracts or from a defined TAA [6, 8]. Autologous DC loaded ex vivo with peptides or tumour cell lysates were capable of efficiently presenting TAA to T cells [37, 38]. Following loading of tumour antigens, DC are matured with specific signals through toll-like receptors, TNF-α receptors and CD40-ligand receptors before being infused back into the patient. This maturation state is important as immature DC or partly mature DC can induce tolerance [36], a problem that all vaccines need to overcome. DC vaccines are an attractive therapy for leukaemia as progenitor cells express all the TAA including BCR-ABL in the case of CML and can be used to generate antigen-presenting cells (APC) [36, 39]. DC vaccines have been used for patients with CML where the autologous DC were pulsed with BCR-ABL peptides from the b3a2 type. No major adverse effects were observed and although a peptide-specific cellular response was observed, there were no clinical effects (Table 1). This type of vaccine can produce a response but more work needs to be carried out to increase the clinical effect [51].

Table 1.

Table showing clinical trials in leukaemic patients

Cancer	Vaccine	Clinical setting/study design	Response	Immunologial responses	References
CML	Peptide: BCR-ABL1 linked to PADRE with GM-CSF given ID	CP, MD Phase I/II, n = 19	CR 13/14	14/19 T cell response to BCR-ABL	[46]
CML	Peptide: BCR-ABL1 wih QS-21 given subcutaneously	CP, MD Phase II, n = 14, HLA restricted class I/II	OR 5/14 = MR	14/14 CD4 + T cell proliferation and DTH 11/14 Th cytokines 4/14 CTL cytokines	[13]
CML	Peptide: BCR-ABL1 with QS-21 given subcutaneously	Partial/CR Phase I DE, n = 12, HLA restricted A3, A11, B8 and DR	OR not reported	2/12 DTH 3/12 proliferation 0/12 cytotoxicity	[44]
AML	Peptide: WT-1 with KLH and GM-CSF given ID and subcutaneously	MD Pilot study, n = 1	CR 1/1	1/1 T cell response frequency and CK release	[30]
CML	Autologous CML cells and DCs with KLH given ID	IFN Peptide: Bcr-abl1 with QS-21 given subcutaneously resistant Pilot study, n = 3, no HLA matched donors	OR = 0/3	2/3 DTH and Cytokine 0/3 cytotoxicity	[40]
B-CLL	Autologous tumours with CD40L	Progressive, intermediate or high-risk disease, Phase I-II DE, n = 11	OR n = 11 Decreased BL count	3/3 cytokine release and proliferation	[51]
CML	Peptide: CMLVAX 100 With QS-21 and GM-CSF and MM given subcutaneously	Stable RD following ST1-571/IFN-α treatment Phase-II, n = 16, HLA-restricted A3, A11, B8 and DR	OR = 15/16	11/16 DTH 13/14 CD4 + T cell proliferation 5/5 IFN-γ	[9]

Various clinical trials involving vaccines have been used this include peptide and DC vaccines. Some vaccines have generated an overall response with T cells responses, although none have sent patients into complete remission with no residual disease

CR complete remission, OR overall remission, DE dose escalation, ID intradermal injection, KLH keyhole limpet haemocyanin, RD residual disease, CP chronic phase, MD measurable disease

A further type of vaccine that has been used in the treatment of leukaemia is peptide vaccines. They rely on the in vivo loading of synthetic peptides onto MHC molecules [7], except when they are used to pulse DC in vitro and then re-infused back into the patient. Various strategies for this type of vaccine have been used involving single epitopes, multiple epitopes, class I and class II epitopes. Peptides can also be modified to enhance binding and interaction with the TCR [58]. The advantages of these types of vaccines are that they are cost effective and measuring immunological responses is easy [17]. However, the problem with peptides is that they are HLA-restricted, often target single antigens and can lead to tolerance. In the case of leukaemia, in particular CML, different peptide vaccines have been trialled (Table 1) and these may be either single or multi-epitope vaccines. Some of these vaccines have induced T cell responses and overall remission in a majority of patients, although no peptide-specific cytotoxicity was seen [36].

The design of novel vaccine strategies relies often on pre-clinical models gathering valuable information regarding optimal conditions for tumour clearance and follow-up, such as mode of vaccination, use of adjuvants, systems of delivery, administration schedule and immune response monitoring. It is critical that these vaccines are also suitable for clinical use. Therefore, the ideal vaccine should combine the following characteristics in order to overcome immunological barriers: low toxicity, ability to stimulate all the anti-tumour arms of the immune response and ability to generate immunomodulatory signals such as cytokines or co-stimulatory molecules for DC and T cell stimulation and eventually, a strategy to deplete immunosuppressive cells such as regulatory T cells (Tregs).

BCR-ABL and CML

BCR-ABL is an unique protein formed as a result of a chromosomal translocation [38], which generates the Philadelphia chromosome. This antigen is central to the pathogenesis of CML (Fig. 1), and is classed as both a tumour-specific antigen on account of its unique junctional region and an over-expressed antigen as it is over-expressed in comparison to normal levels of BCR and c-ABL. The presence of BCR-ABL-specific T cells has been detected in CML patients using tetramer technology [12, 15], i.e. CD8 + T cells specific for the HLA-A3-restricted epitope (KQSSKALQR) were observed and the majority of patients were found to have low levels of circulating BCR-ABL specific CD8 + T cells [12]. Interestingly, it was also recently seen that ex vivo specific CD8 + T cell responses against extra-junctional BCR-ABL epitopes were also detected in CML patients [19]. We have also shown that selected extra-junctional epitopes are capable of generating specific CD8 + T cells response in vivo in transgenic HLA-A2.1 (HHDII) mice as well as in vitro using PBMC from CML patients (Figs. 2, 3).

Fig. 1.

Schematic diagram showing the effects of increased tyrosine kinase activity of BCR-ABL. BCR-ABL subverts normal cellular processes in which normal BCR and c-ABL play a role in control of cellular growth and apoptosis. It can cause production of reactive oxygen species (ROS), which are known to be involved in causing mutations leading to a transformed phenotype

Fig. 2.

Representative results generated from HHDII mice immunised with FLN, MLT and LYG epitopes. At the top ratio of 100:1 effectors to targets, there is approximately 80% cytotoxicity in response to these peptides. These peptides can be classed as immunogenic as a significant level of cytotoxicity was detected. The graphs c and d are using CTL generated from HHDII mice immunised with the LYG peptide and then cultured with T2 cells pulsed with the LYG peptide c or the FLN peptide d. As can be seen a strong cytotoxic response could be generated against both these epitopes. This could be due to the fact that the 12-mer epitope could be acting as both a MHC class I and class II epitope stimulating CD8 + and CD4 + T cells. ***P < 0.001. See Table 1 for total number of mice tested

Fig. 3.

Results from ELISPOT assay carried out on PBMC stimulated with either a LYG, b MLT or c FLN epitopes. T cells were generated from PBMC by re-stimulating in vitro with LYG, MLT and FLN peptides and following this, the level of IFN-γ releasing cells was measured by ELISPOT assay. A significant response was detected in patient sample 6 T2 (LYG 12-mer a), patient samples 17 and 6 T2 in response to the MLT peptide b, patient samples 4, 10, 19 T1, 19 T 2 and 19 T3 after one stimulation with FLN peptide c. *P < 0.05, **P < 0.01 and ***P < 0.001

The SYFPEITHI computer algorithm, which calculates peptide binding affinity to various HLA haplotypes according to length, sequence and key anchor residues, was utilised in this experiment [11]. Three peptides derived from the BCR-ABL extra-junctional region and potentially capable of binding HLA-A2.1 molecules were selected, synthetically made and used to immunise HHDII mice (Fig. 4). Cytotoxicity assays were performed with antigen-specific cytotoxic T lymphocytes (CTL) generated from these mice; these results are shown in Fig. 2. Two of these peptides (FLN peptide: FLNVIVHSA and MLT peptide: MLTNSCVKL) stimulated CTL capable of killing target T2 cells pulsed with the peptide and up to 80% cytotoxicity was observed at an effector:target ratio of 100:1 (Fig. 2). The third peptide was a 12-mer epitope that was selected and although the optimal length for a peptide is 8–10 amino acids, it has been shown that longer peptides of 12–14 amino acids are capable of binding and stabilising MHC molecules [11]. The results of cytotoxicity assays carried out with splenocytes generated from mice immunised with this peptide are shown in Fig. 2. As can be seen, the 12-mer generated specific CTL capable of killing target cells pulsed with both the 12-mer (LYG peptide: LYGFLNVIVHSA) and a 9-mer peptide derived from the 12-mer (FLN peptide: FLNVIVHSA). The overall results of these experiments are summarised in Table 2. These results suggest that these three peptides are immunogenic, capable of binding to HLA-A2.1 molecules.

Fig. 4.

Partial sequence of BCR-ABL sequence across the junctional region showing the underlined selected peptides. The amino acid sequence above refers to a portion of the 210 kDa BCR-ABL protein. The sequence in capitals is from the BCR sequence and following this, is the ABL sequence, apart from the unique lysine (K) residue highlighted in bold. The two 9-mers and the 12-mer (see Table 2) used in this study are, respectively, underlined and framed above and were from across the BCR part of the sequence

Table 2.

Summary of peptide immunisations

Protein	Name	Sequence	SYFPEITHI score	*Number of mice responding
BCR	FLN	FLNVIVHSA	23	6/6
BCR	MLT	MLTNSCVKL	23	4/4
BCR	LYG	LYGFLNVIVHSA	ND	3/4 (4/4 FLNVIVHSA)

As can be seen, all mice tested with the MLT and FLN peptides generated a significant response. The LYG peptide generated a significant response in 75% of mice tested (ND not determined)

These epitopes were then used to generate T cells in PBMC from both healthy donors and CML patient samples. It is essential to compare the results from these different models to get a true representation of the specific T cell responses to peptides [33] as strong T cell responses to peptides in a mouse model does not always convert into a strong response in humans. This is due to the fact that whilst transgenic mouse models give an indication of what might occur in vivo with human HLA molecules, they express murine co-stimulatory molecules and TCR repertoire differences may occur [14]. These variations could potentially lead to differences in peptide responses across species. Therefore, human PBMC models can indicate whether peptide-specific T cells can be generated and can also provide information about whether specific T cells are present in cancer patients. Following a single stimulation, responses were seen in the HLA-A2.1 positive patients in response to the MLT (1/3) and FLN (2/3) epitope. No responses were seen in these patients in response to the LYG epitope after a single stimulation. Following further stimulation, these responses changed and the HLA-A2.1 positive healthy donor also responded (results not shown) suggesting a new T cell response as opposed to a “memory” or activation of precursor T cells. Another important observation of these preliminary results is that HLA-A2.1 negative patients also responded to stimulation with these epitopes, i.e. patient 6 (time point 2) to LYG and MLT (Fig. 3) and patient 19 (time point 1, 2 and 3) to FLN epitope; the overall results of these experiments are summarised in Table 3.

Table 3.

Summary of patients who generated specific IFNγ-secreting cells in responses to selected peptides following single stimulation (a) and a summary of their specific HLA-class I types (b)

(a)
Patient	LYG	MLT	FLN
4			Yes
6 (Time point 1)			ND
6 (Time point 2)	Yes	Yes
10	ND		Yes
15
17		Yes
19 (Time point 1)	ND		Yes
19 (Time point 2)	ND		Yes
19 (Time point 3)			Yes

(b)
Patient	HLA-A	HLA-B	HLA-Cw
4	2, 3	7, 38	7w
6	11, –	35, 62	4, 9
10	2, 23	56, 65	1, 8
15	3, 31	44, 51	15, 16
17	2, 3	7, 18	7, –
19	1, 23	44, 62	4, 9

For the full HLA details and treatment regime of these patients, see Rojas et al. [46]. The HLA-A2 + ve samples are highlighted in bold

It is possible that peptides bound to the T2 cells, which express HLA-A2 molecules and used to stimulate PBMC, have been produced and released in the in vitro culture. These peptides could be taken up by the patient’s own cells, thereby cross-presenting them to activate T cells. If this was true, these peptides could be important for therapy to treat a wider range of patients with different haplotypes. However, another possibility is that the presence of IL-2 in the media may have activated other IFN-γ-secreting effector cells within the cultures, such as NK cells. This could explain the lack of specificity of IFN-γ release in HLA-A2.1 negative donors as NK cells lack target specificity of T cells and do not require sensitisation for the expression of their activity as T or B cells do. The problem with this theory is that the T cell alone cultures were carried out in the presence of IL-2 and yet no specific release of IFN-γ was seen suggesting that the peptide was the stimulatory factor. The final possibility is that the peptides could have been cross presented by APCs present in the culture to activate CD4 + T cells, another group of effector cells capable of secreting IFN-γ. Again this would be relying on the fact that the peptides bound to the T2 cells were released into the in vitro culture. These results are, however, only preliminary and were performed on a small cohort of patients. Further work needs to be carried out to investigate responses to these epitopes in a larger cohort of CML patients with varying HLA types. This work would also need to include the assessment of the immune cells responsible for the release of IFN-γ, i.e. by FACS analysis. Nonetheless, these results when combined with those from the mouse study carried out in our laboratory suggest BCR-ABL extra-junctional epitopes can provide immunogenic targets.

HAGE and CML

Another antigen associated with CML is HAGE, a cancer-testis antigen. The HAGE gene was mapped on chromosome 6 (6q12-q13) by radiation hybrid analysis and encodes a putative 73 kDa protein. Analysis of the protein sequence of HAGE revealed that it has a DEAD box characteristic of the family of ATP-dependent RNA helicases and has 55% homology with DDX5 (p68), another member of this family. Although it was initially identified in a human sarcoma by Martelange et al. [33] using representational difference analysis, several studies have reported the over-expression of the cancer-testis antigen HAGE in more than 50% of myeloid leukaemias [2, 33], as well as being frequently expressed in about 30% of brain, colon and lung cancers amongst others [48]. Alongside this, we have assessed the expression of HAGE by real time PCR on 20 CML samples. Real time PCR detected HAGE levels from five to almost 180-fold higher in 70% of samples tested than in normal peripheral blood cells (PBC); the proportion of HAGE-positive patients is superior to that described by Adams et al. [2]. Interestingly, these CML samples were composed of two groups of ten having either low or high BCR-ABL transcript levels respectively, samples with low transcript levels corresponding to patients successfully responding to treatment with imatinib. Real time PCR showed that the transcript levels correlated with HAGE expression with the latter being 5 to 20-fold higher than in PBC in 50% of samples with a low transcript level, and 20 to 180-fold higher than in PBC in 90% of samples with a high transcript level (Fig. 5). These results suggest that HAGE is not only associated with advanced CML disease and poor prognosis as shown in a recent study by Roman-Gomez et al. [47] but may also be involved in the pathogenesis of CML. This is of major interest if one wants to use this cancer-testis antigen in strong immunotherapeutic interventions, as it is very unlikely that even under pressure leukaemic cells undergoing chemotherapy or TKI therapy will stop expressing it.

Fig. 5.

Expression of HAGE and BCR-ABL in chronic myeloid leukaemia. Real time PCR analysis was carried out on 20 CML samples and normal total blood cells (TBC). In a 10 CML samples from patients with high BCR-ABL transcript levels and in b 10 CML samples with low BCR-ABL transcript levels. The experiment was carried out once and data are expressed relative to the mRNA level of TBC, arbitrarily set as 1

Like DDX1 [20], DDX2 [18] and DDX5 [57], evidence suggests that the DEAD-box RNA helicase HAGE is involved in tumour cell proliferation as well as having a crucial role in the RNA metabolism of tumour cells making HAGE an ideal immunotherapeutic target for the treatment of CML patients in the later stages of the disease. However, when targeting this antigen in in vitro experiments using CML cell lines, it is important to consider that the methylation status of the HAGE promoter is critical to gene transcription. This is shown by the fact that CML cell lines have very high hypermethylation status of the HAGE promoter, hence preventing transcription of the HAGE gene [47]. Like other cancer-testis antigens, low levels of HAGE expression have been shown to be easily reversed in CML cell lines by the addition of a demethylating agent, which has shown synergy with a histone deacetylase inhibitor confirming that the methylation status of the HAGE promoter was critical to the HAGE gene transcription (results not shown).

Limited work has been undertaken to determine the immunogenicity of HAGE or the presence of HAGE-specific T cells in CML patients. Previously, we identified several immunogenic HAGE-derived HLA-A2 and HLA-DR1/-DR4 peptides using reverse immunology and transgenic mouse models amongst which one class I and two class II peptides were also found to be naturally processed. Moreover, responses detected in a murine model with one of the class II peptides have also been observed in healthy donors, indicating that this peptide was processed identically in both species [34]. Further work is still required to test the ability of the two remaining peptides to generate in vitro human T cells capable of specifically responding to target cells expressing the appropriate HLA molecules as well as HAGE protein/peptides. It also remains to be established whether the administration of class I and class II peptides derived from the same protein will result in an increased CTL activity, long-lasting immunity and whether tumour protection and/or clearance can be achieved with the use of helper and cytotoxic T cells.

A tumour model was developed in HHDII-DR1 double transgenic mice (transgenic for HLA-A2 and -DR1, and knockout for H-2 and Ia/Ie). This tumour model relied on the creation of stable HLA-transfectants expressing HAGE, the transplantation of these genetically modified tumour cells in mice and the assessment of several DNA-based vaccine strategies circumventing the need to identify specific epitopes and allowing the generation of an immune response against multiple epitopes regardless of HLA haplotype. Using a standard analysis of tumour regression, tumour protection and/or clearance were achieved in a majority of mice immunised with HAGE DNA vaccine administrated intra-dermally by gene gun or intra-muscular injection of plasmid DNA, indicating that HAGE as a whole antigen is strongly immunogenic, at least in mice (Fig. 6). The ability of HAGE DNA vaccine to generate an immune response was not surprising as several studies targeting cancer-testis antigens using plasmid vaccines in mice have demonstrated similar promising results [28, 43]. Despite being a transplantable solid tumour model, these preliminary findings suggest that HAGE could prove to be an interesting immunotherapeutic target for the treatment of CML. However, more work is required to identify suitable peptide targets that can be confirmed by in vitro experiments with healthy donors or patients PBMC. Confirmation of clinical utility using in vivo models with HAGE-expressing (target) tumour cells should be performed in order to closely mimic human cancer development, to avoid failure of future HAGE-based vaccine strategies.

Fig. 6.

Therapy studies using gene gun immunisation in DR1/HHDII mice. Three days after injecting 6 × 10⁵ ALC/HAGE cells to two groups of 10 mice, mice were immunised three times with gold particles coated with either pBudCE4.1/(−) or pBudCE4.1/HAGE at 7-day intervals and monitored for tumour growth a and survival b. Results shown are representative of two independent experiments. *P < 0.05 are the statistical differences between HAGE DNA immunisation and control DNA immunisation determined by unpaired Student t test. Down arrow indicates timepoints of immunisation, dagger indicates termination

Discussion

There has been a large amount of research and clinical trials focussed on the immunotherapy of leukaemia. The tumour specific BCR-ABL antigen provides the ideal immunotherapeutic target. However, there has been a considerable debate about the immunogenicity of this antigen [23]. This is mainly because research has focussed on the junctional region of this fusion antigen, where there are limited potential HLA-class I epitopes and because high avidity T cells are often not detected in CML patients with or without immunotherapy [48]. Recent data has also shown that a number of genes are upregulated by BCR-ABL including PRAME [55], survivin, adipophilin, WT-1 and BCL-2 [10]. Following treatment imatinib, the expression of these genes was decreased and this affected the immunogenicity of these cells, leading to impaired CTL responses [10]. Further to this, a CML cell line resistant to imatinib did not demonstrate these reductions in immunogenicity [49], which led to the suggestion that the immunogenicity of CML is linked to these BCR-ABL inducible genes as opposed to BCR-ABL itself [49]. With this in mind, some consideration would be to target both BCR-ABL and other antigens such as HAGE, at the same time to ensure a complete response and to avoid immune escape. Although little is known about the function and characteristics of the HAGE protein, its expression does appear to be correlated with BCR-ABL expression and it is immunogenic. Its potentially crucial role in RNA metabolism in tumours provides another way to inhibit the development of cancer cells. A greater understanding of this antigen may point the way to the development of more effective immunotherapy in CML, that can be superimposed onto the dramatic improvement in outlook that have been achieved with TKI therapy.