By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Due to essential maintenance the subscribe/renew pages will be unavailable on Wednesday 26 October between 02:00- 08:00 BST/ 09:00 – 15:00 SGT/ 21:00- 03:00 EDT. Apologies for the inconvenience.
Dr G. Cook, MBChB, PhD, FRCP(Glas), FRCPath, Blood and Marrow Transplantation, BMTU and Transplant Immunology Group, Leeds Teaching Hospitals, St James’ University Hospital, Leeds, UK. E-mail: firstname.lastname@example.org
In a small number of patients with multiple myeloma (MM), long-term disease-free survival has been achieved by harnessing the immune phenomenon, ‘graft-versus-tumour’ effect, induced by allogeneic haemopoietic stem cell transplantation. This has prompted many investigators to examine ways in which a patient's own immune system can be more effectively directed against their disease, with the ultimate aim of tumour eradication. In this review we assess the current understanding of immunobiology in MM, and how the different components of the immune system, such as dendritic cells, T cells and natural killer cells, may be harnessed using in-vitro and in-vivo priming techniques. We look at the clinical immunotherapy trials reported to date and whether, in light of the current information, immunotherapy for MM is an achievable goal.
The immune system is capable of eradicating malignancies. The first evidence that cancer cells were susceptible to immune attack emerged in the early 1960s from the animal models used for haemopoietic stem cell transplantation (HSCT) (Mathe et al, 1965). This was first demonstrated in human HSCT in the late 1980s with the observation that patients who survived T-cell-replete allogeneic (allo) HSCT had a lower relapse rate compared with autologous or T-cell-depleted allo HSCT (Gale et al, 1989; Hughes et al, 1989). This ‘graft-versus-tumour’ effect has also been demonstrated using steady-state, unprimed, donor lymphocyte infusions (DLI) to treat relapse following allo HSCT and in the setting of minimal residual disease (Kolb et al, 1990).
Multiple myeloma (MM) is a clonal B-cell malignancy characterised by an excess of mature plasma cells in the bone marrow (BM), in association with monoclonal protein in serum and/or urine, decreased normal immunoglobulin (Ig) levels and lytic bone disease. MM remains essentially incurable by conventional anti-tumour therapy. The use of allo HSCT in the treatment of MM has resulted in a higher rate of molecular remission, with lower rates of relapse and disease progression as compared with patients treated with autologous HSCT (Bensinger et al, 1996; Corradini et al, 1999).Those patients who survive 1 year have significantly improved disease-free survival (Bjorkstrand et al, 1996). This is partly related to the intensity of the chemo-radiotherapy conditioning regimen, and it is also a result of the graft-versus-myeloma (GVM) effect (Tricot et al, 1996; Verdonck et al, 1998; Perez-Simon et al, 2003). However, these improvements in disease control are achieved at the expense of higher treatment-related morbidity/mortality in the first year and have prompted the experimental use of potentially less toxic non-myeloablative HSCT (Singhal et al, 2000; Garban et al, 2001; Perez-Simon et al, 2003; Crawley et al, 2005). The close relationship between graft-versus-host disease (GVHD) and GVM in published studies suggests that donor allo-reactive T cells, directed against minor histocompatibility antigens (Ags) present on both normal and myeloma cells, mediate the latter effect. The successful use of DLI in some cases of MM that relapse following allo HSCT has led to increased interest in the possibility that other forms of immune therapy might be effective in this disease (Lokhorst et al, 1997, 2000). However, the doses of T cells required to induce these remissions are higher in MM than in other DLI responsive diseases, such as chronic myeloid leukaemia (CML) (Verdonck et al, 1998), and are associated with an increased incidence of GVHD (Salama et al, 2000; Huff & Jones, 2002). This may be partly explained by the influence of the malignant clone on the function of the immune effector cells resulting from both passive and active suppression, although it is possible that MM tumour Ags are less immunogenic compared with other those on other malignant cells. MM is associated with several defects in the host's immune system (Cook & Campbell, 1999).
Multiple myeloma tumour cells produce a number of immunologically active agents that can modulate the immune response, such as transforming growth factor-beta (TGF-β), interleukin-10 (IL-10) (Brown et al, 2004), Fas ligand (Villunger et al, 1997), vascular endothelial growth factor (VEGF) (Oyama et al, 1998) and mucin 1 (MUC-1) (Gimmi et al, 1996; Agrawal et al, 1998; Treon et al, 1998). It is postulated that, by producing these agents, the tumour cell modifies both the microenvironment to support growth and differentiation of the clone, and the host immune response to prevent tumour rejection. This duality of function is important in understanding the possible interactions of the malignant clone with the tumour-bearing host, especially if we are to design immunotherapy strategies that will achieve their true potential and result in improved survival in MM.
In an attempt to boost the GVM effect, Kwak et al (1995) and Yiwen et al (2000) have demonstrated the successful transfer of myeloma idiotype-specific immunity from an actively immunised bone marrow donor to a recipient with MM, demonstrating major histocompatibility complex (MHC) class I-restricted CD8+ T-cell recognition of freshly isolated, recipient myeloma tumour cells. It still remains to be seen whether the GVM effect can be separated from GVHD. However, there is evidence that a patient's own immune system may play an important role in the control of their disease. It has been shown that T cells specific for MM-associated Ags, such as Mucin-1 (MUC-1 or CD227) (Beckhove et al, 2003), idiotype protein (Yi et al, 1995) and NY-ESO-1 (van Rhee et al, 2005), are present in the peripheral blood of MM patients, while other studies have shown that the presence of expanded T-cell clones at any time during the disease course is associated with prolonged overall survival (Brown et al, 1997; Raitakari et al, 2003).
Immunotherapeutic strategies attempt to utilise the immune system for disease control and they have mainly been tested in the setting of relapsed or resistant disease. The use of such strategies in the setting of minimal residual disease following conventional chemotherapy or HSCT offers an increased potential for tumour cell control by adoptively transferring immune effectors at a time when they are likely to have most impact (Hsu et al, 1997; Stevenson et al, 2004).
The aim of this review is to assess, with the current level of understanding of immunobiology in MM, whether immunotherapy for MM is an achievable goal.
Discovered in the 1970s (Steinman & Cohn, 1973, 1974; Steinman et al, 1974, 1975), dendritic cells (DC) have been identified as the sentinels of the immune system. As a result it has become theoretically possible to direct the immune response against a specific chosen (tumour) Ag towards immunity or tolerance (Matzinger & Guerder, 1989). It is important to consider how the different components of the immune system [DC, T cells and natural killer (NK) cells] develop and interact in vivo and how this may be altered in patients with MM. Subsequently, we will consider how cells of the immune system may be manipulated ex vivo to overcome any MM tumour cell suppressive effects and eradicate the malignant clone.
In vivo DC development
Dendritic cells precursors are derived from the BM haematopoietic stem cell (HSC), differentiating into two phenotypically distinct populations. Myeloid DC (mDC) classically arise from the common myeloid progenitor and myelomonocytic precursors are characterised by the surface phenotype of CD11c++/CD123−/CD1c+, and tend to induce T-helper cell type 1 (Th1) responses. Plasmacytoid DC (pDC) arise from the common lymphoid progenitor, are CD11c−/CD123++, and induce Th2 responses. However, other factors, such as the strength of the T-cell receptor (TCR)/MHC class II interaction, Ag density and the microenvironment, are also important in the balance between mounting a Th1 or Th2 T-cell response. Recent studies also suggest that the system for DC generation may exhibit more plasticity than previously thought, in that both myeloid and lymphoid DC may arise from either myeloid or lymphoid progenitors (Shigematsu et al, 2003). These cells migrate via the blood to tissues, such as skin, portal triads, mucosa and lung, under the influence of chemokines such as macrophage inflammatory protein-1α (MIP-1α), MIP-3α and regulated on activation, normal T-cell expressed and secreted (RANTES), where they exhibit an immature DC (iDC) phenotype [Fig 1] (Austyn & Larsen, 1990). The iDC actively sample the environment by phagocytosis and process Ag from bacteria, viruses and apoptotic bodies (Hengel et al, 1987; Svensson et al, 1997; Albert et al, 1998), presenting them in the context of MHC class I and II molecules. To mature, iDC require a second ‘danger’ signal, such as interferon (IFN) α or IL-1β, or microbial proteins, such as bacterial lipopolysaccharide (LPS) (Matzinger, 1998). This maturation signal causes the DC to down-regulate their phagocytic and Ag processing functions, up-regulate expression of MHC class I and II, co-stimulatory, adhesion molecules and chemokine receptors such as CCR-7 and 8 (Qu et al, 2004). The mature DC (matDC) then rapidly migrate to the secondary lymphoid tissues via the afferent lymphatic system, attracted by chemokines such as MIP-3β and secondary lymphoid-tissue chemokine (Dieu et al, 1998; Chan et al, 1999), where they present Ag to T cells.
Dendritic cell, T-cell interactions
Dendritic cells have the unique ability to present Ag to naïve T cells. Techniques, such as 2-photon imaging of lymph nodes, have shown that this process follows an orderly progression. Within 2 h of DC migration, T-cell behaviour changes from a random stochastic nature, to making short-lived contacts with multiple DC dendrites. During the next 14 h, T cells form stable clusters around DC before breaking down into dynamic swarms, and by 24 h, T-cell proliferate and migrate out of the node (Miller et al, 2004). Elegant studies such as these provide further evidence that the ‘immune synapse’ is a dynamic, orderly process.
Effective priming of naïve T- cell induces clonal expansion and differentiation into effector and memory cells. This is achieved via the engagement of the TCR/human leucocyte antigen (HLA)-I/II/Ag, CD40/CD40 Ligand (CD40L) and CD28/CD80 complexes and the secretion of IL-7, IL-12 and perhaps IL-2 (Granucci et al, 2001). CD4+ T-cell help is required at the time of priming to generate CD8+ effector cells, and is mediated by CD40/CD40L engagement, which induces DC production of IL-12. This CD40 licensing of DC may also be induced by CD40L on some apoptotic bodies (Propato et al, 2001). The CD40/CD40L interaction seems particularly important in that CD40 ligation alone is sufficient to drive the maturation of iDC (Caux et al, 1994), and allows matDC to prime CD8+ cytotoxic T cells (CTL) without further CD4+ T-cell help. Of the many other factors important in the T-cell response, the maturity of the DC is one of the most important.
Immature DC induce ‘abortive proliferation’ of T cells, i.e. initial proliferation but short-term survival and perhaps clonal deletion. Mature DC induce T-cell survival and differentiation by displaying Ag in the context of the appropriate co-stimulatory molecules (Dhodapkar et al, 2001; Jonuleit et al, 2001a; Liu et al, 2001). Thus, it is crucial that DC achieve a fully mature state if they are to be used as an anti-cancer therapy. This may also partly explain how the tumour is able to evade detection by the immune system at an early stage, when novel tumour Ags may be encountered by antigen-presenting cells (APC) in the absence of an inflammatory, pro-maturation environment (danger signal). Thus APC remain in an immature state and a toleragenic response may be generated (Matzinger, 1998).
How can DC be produced/obtained for clinical use?
Circulating blood mDC and pDC can be collected using apheresis machines and enriched with magnetic cell separation protocols in a similar manner to CD34+ stem cells (Fearnley et al, 1997; Lopez et al, 2003). These cells are present at very low concentrations in peripheral blood (Dzionek et al, 2000; MacDonald et al, 2002) but the number of circulating DC can be dramatically increased using FLT-3 (Maraskovsky et al, 1996; Morse et al, 2000). This is a relatively inexpensive and straight forward process, however, the yield of cells is relatively modest and there is conflicting evidence relating to whether circulating DC are functionally normal in patients with MM (Brown et al, 2001; Ratta et al, 2002). Brown et al (2001) also reported that circulating DC from MM patients fail to up-regulate CD80 and 86 in response to CD40L, which may be corrected by co-culture with IL-12, IFN- or anti-TGF-β antibodies, suggesting that TGF-β1 and IL-10 are implicated in causing this defect.
Alternatively, DC may be generated from either CD34+ HSC (34DC) (Caux et al, 1992) or from monocytes (MoDC) (Zhu et al, 2000). 34DC were first produced using a cytokine cocktail containing granulocyte-macrophage colony-stimulating factor (GM-CSF) and tumour necrosis factor alpha (TNFα) (Reid et al, 1992; Santiago-Schwarz et al, 1992). Many groups have subsequently refined this by adding cytokines, such as FLT-3, stem cell factor and IL-4 (Strunk et al, 1996; Evans et al, 2000). The DC generated by these methods have a mature myeloid phenotype, low phagocytic activity, and express high levels of co-stimulatory molecules, such as CD40, CD80, CD83 and CD86 (Ferlazzo et al, 2000). They are also able to induce an Ag-specific cytotoxic response from naïve T cells (Caux et al, 1995). Unfortunately, patients must first undergo HSC mobilisation with either granulocyte colony-stimulating factor alone, or in combination with chemotherapy. In addition, the methods by which 34DC can be loaded with Ag, such as transfection or fusion with tumour cells, are not straightforward and are further discussed below.
Dendritic cells may be generated from MoDC isolated from peripheral blood mononuclear cells separated by either plastic adherence or positive selection of CD14+ cells using immunomagnetic beads. In vitro culture of MoDC occurs classically in three stages. Immature DC are produced by culture with IL-4 and GM-CSF for 5–10 d. These cells are CD1a+/CD1c+/CD14−/CD86+ and HLA-DR+, and they have a high phagocytic capacity, which facilitates Ag uptake. A number of groups have substituted IL-13 for IL-4 at this stage to produce functional iDC which may have a more stable phenotype compared with those generated with IL-4 (Boyer et al, 1999; Morse et al, 1999).
The next stage in the process is to load iDC with Ag. Normal iDC are phagocytic, so they can be simply ‘fed’ Ag as a purified naked form, coated with liposome, crude tumour lysate, tumour-derived heat shock protein, whole apoptotic tumour cells, antibody-coated tumour cells or genetically modified Ag (Mylovenge, idiotype-GM-CSF fusion; Dendreon) (Galea-Lauri et al, 2004), summarised in Table I. However, there is some evidence that DC from patients with malignancy (CML and MM) have reduced phagocytic capacity because of cytoskeletal abnormalities, reduced processing capacity for exogenous Ag and reduced chemokine-induced migration (Ratta et al, 2002; Dong et al, 2003). There are at least three approaches that may circumvent this problem. The first is the generation of DC from a malignant precursor cell. This is well characterised in myeloid malignancies, such as acute myeloid leukaemia and CML, and there have been recent reports of the generation of DC-like cells from patients with acute lymphoblastic leukaemia (Blair et al, 2001). These cells appear to have the ability to stimulate naïve T cells, although whether they function as efficiently as normal DC is still open to debate. At present there is no evidence that MM plasma cells can be converted into more efficient APC.
Table I. Sources of antigen and methods of DC loading.
Type of antigen
Method of Ag loading
Tumour-associated proteins, e.g. MUC-1 and idiotype protein
Easy to monitor immune response with tetramers. Ubiquitous Ag
Relies on a single antigen for immune response. Some tumour antigens are not processed efficiently, e.g. MUC-1
Tumour lysate, e.g. freeze/thaw heat or sonication
All tumour Ags available. Simple preparation
May expose DC to immunologically active tumour molecules. Difficult to monitor. Need to collect tumour cells from individual patients at diagnosis
Tumour-derived heat shock proteins, e.g. HSP70 GP96
Ag preloaded. Specific mechanism for uptake and processing
Complex Ag preparation for individual patients. Difficult to monitor
All tumour Ags available. Specific mechanism for uptake
Need to collect tumour cells from individual patients at diagnosis. Difficult to monitor
Live tumour cells
All tumour Ags available
Can MM plasma cells be converted to DC? Are tumour DC efficient APC? Tumour cell contamination. Difficult to monitor
Tumour cell DC fusion
All tumour Ags available
Complex Ag preparation for individual patients. Fusion process toxic to DC. Tumour cell contamination. Difficult to monitor
Ubiquitous Ags. Easy to monitor immune response with tetramers
Relies on a single antigen for immune response poor efficiency of transfection. Vectors and tumour DNA may be toxic
The next approach is to generate normal matDC (maturation is discussed below) and then fuse them with tumour cells by either electroporation or pegylation (Gong et al, 2002; Hao et al, 2004; Raje et al, 2004). These approaches make the assumption that the resulting fusion or malignant DC retain normal function. Such cells must also express all of the relevant tumour Ags, in a state that the immune system is able to recognise as abnormal, so that a cytotoxic response is mounted rather than inducing tolerance. The final approach is to again generate normal matDC but then transfect them with tumour-associated Ags (TAA), such as MM idiotype protein (Id) or NY-ESO-1 (van Rhee et al, 2005) using viral vectors (i.e. adenovirus and lentivirus). It is possible to achieve high rates of transfection but these approaches have been hampered by viral effects on DC function and recent reports of leukaemic transformation in two of nine undergoing gene therapy for X-linked severe combined immunodeficiency due to the activation of the oncogene LMO2 (Hacein-Bey-Abina et al, 2003a,b; McCormack & Rabbitts, 2004).
Usually, iDC are converted into matDC after Ag loading, using various combinations of cytokines and inflammatory stimulants. Classically, TNFα (Thurnher et al, 1997) or monocyte conditioned medium (Romani et al, 1996) are used, but many other compounds and combinations are described including poly (I:C), CD40 Ligand, prostaglandin E2, IL-1β and IL-6, to name but a few. Mature DC up-regulate surface expression of the co-stimulatory molecules CD80, CD83, CD40, HLA class I and II, lose their phagocytic ability, increase their migratory capacity to lymph nodes via increased CCR-7 and become much more efficient at inducing responses from naïve T cells (De Vries et al, 2003a,b; Tarte et al, 2000).
The production of MoDC (and 34DC) is costly, labour intensive and requires autologous cells to be generated for each individual patient. Recently groups have been looking at speeding up the process of producing matDC from 1–2 weeks down to 2 days (Fast DC) (Dauer et al, 2003; Xu et al, 2003), which may more closely represent what is happening physiologically during the inflammatory response and has the added benefits of being quicker and less labour intensive.
Unfortunately, the MM tumour cells actively produce cytokines (IL-6, IL-10, TGF-β and VEGF) that interfere with this process at many points, resulting in DC being produced in fewer numbers, that are functionally abnormal and fail to mature with appropriate stimulation (Brown et al, 2001) [Fig 2]. When functionally normal matDC are generated, it will be critical to choose the correct Ag(s) and adjuvants with which to prime the immune response for clinical use. A number of groups have started clinical trials looking at the use of DC as treatment for MM.
There is a growing body of evidence that, as well as providing immunogenic ‘help’ to the immune system, CD4+ T cells are also important negative regulators of the immune response. Many of these effects are mediated by a subset of cells known as naturally occurring CD4+/CD25+ regulatory T cells (TReg cells), which compose 5–10% of all circulating CD4+ T cells in adults (Sakaguchi et al, 1995; Asano et al, 1996). Unfortunately, this picture is further complicated by the interaction of iDC and naïve T cells, which may cause T-cell depletion, anergy and the induction of other subsets of CD4+ and CD8+ TReg cell which secrete IL-10 and TGF-β (Dhodapkar et al, 2001; Jonuleit et al, 2001b).
TReg cells are produce in the thymus by positive selection of CD4+cells with TCR that have intermediate affinity for MHC class II–self Ag complexes, and exert their immunomodulatory effects in vitro in a cell contact-dependent manner, although there is some in vivo evidence that immuosuppressive cytokines may also be involved. TReg cells play a role in maternal tolerance to the fetus (Aluvihare et al, 2004), tolerance following transplantation (Wood & Sakaguchi, 2003) and they have been implicated in impeding natural anti-tumour immunity and immunotherapy (Sakaguchi et al, 2001). One of the proposed mechanisms for these observations is the interaction between TReg cell CTLA-4 and iDC B7-1 and B7-2 which induces the up-regulation of indolemine 2,3-dioxygenase (IDO) on the DC (Mellor et al, 2003). These IDO-DC inhibit T-cell proliferation, induce T-cell apoptosis (Munn et al, 2002) and large numbers of these cells have been found in the tumour-draining lymph nodes of patients with malignancies (Munn et al, 2002). The Dana Faber group recently reported a significant increase in CD4+/CD25+ TReg in MM patient samples compared with normal donors (23 ± 4% vs. 6 ± 3%) (Prabhala et al, 2004). They found that the proliferation of T cells depleted of TReg cells was significantly lower in monoclonal gammopathy of undetermined significance (MGUS; n = 9, SI = 12 ± 2) and MM (n = 9, SI = 28 ± 8) compared with normal donors (n = 9, SI = 74 ± 9, P < 0·01), and LPS was unable to overcome suppression of T-cell proliferation by TReg cells in MGUS (49%) and MM (24%) compared with normal controls (110%). These observations suggest that TReg cells may play a role in the immune dysfunction seen in MM patients.
Natural killer cells
In contrast to T cells, NK cells do not require preactivation or immunisation in order to recognise and kill targets, such as tumour cells or virally infected cells. NK cells arise from the HSC and develop under the influence of IL-2 and/or IL-15, acquiring effector functions, such as lytic ability and cytokine production (Williams et al, 1997). NK cells recognise abnormal cells via a number of mechanisms, of which the ‘missing self hypothesis’ is the best understood (Ljunggren & Karre, 1990). Tumour and virus-infected cells may lose or down-regulate MHC class I expression in an attempt to evade recognition by CD8+ CTL. NK cells detect this via surface receptors (CD94:NKG2A heterodimer or killer Ig-like receptors), which, under normal circumstances, engage the MHC class I molecules and deliver an inhibitory signal to prevent NK cells attacking normal healthy cells (Vilches & Parham, 2002). Myeloma cells are susceptible to NK cell lysis (Frohn et al, 2002) and there is evidence to suggest that the number and state of activation of NK cells are increased in MM. These cells have inherent anti-MM cytotoxic activity and drugs, such as thalidomide, may further augment this effect (Gonzalez et al, 1992; Frohn et al, 2002; Zheng et al, 2002; EL-Sherbiny et al, 2003).
There are reports of natural NK cytotoxicity to circulating (Ferlazzo et al, 2002) and ex vivo generated, autologous Ag-pulsed DC in patients with MM (Zheng et al, 2002). Immature DC are efficiently targeted by NK cells while matDC are protected by the up-regulation HLA class I. The interaction between NK cells and DC at sites of inflammation is complex (Ferlazzo & Munz, 2004). DC are able to activate NK cells via IL-12 and perhaps IL-18 (Andrews et al, 2003), while DC maturation is stimulated by IFN-α released by the NK cell. Why should NK cells kill iDC while at the same time inducing DC maturation and homing to lymph nodes? It has been suggested that this paradox is a control mechanism used during infective episodes. Both results of DC/NK cell interaction result in the depletion of DC at sites of inflammation. While this may deprive DC trophic pathogens of their host, it may also be an important feedback mechanism to prevent excessive production of pro-inflammatory cytokines by limiting recruitment of iDC to the matDC pool (Moretta et al, 2003). These observations provide evidence that there is an active interface between the innate and adaptive arms of the immune response that could be harnessed in future anti-MM NK cell and DC based therapies.
As our understanding of how the immune system regulates itself improves, we should be better able to manipulate it to produce more effective immune therapies. We shall now review the use of immunotherapeutic strategies in the treatment of MM.
Anti-tumour vaccines in MM
Vaccination is the most effective intervention modern medicine has developed, and has almost eradicated diseases such as small pox and polio. An ideal anti-tumour vaccine would be produced from a TAA that is only expressed on tumour cells but is shared between different patients and tumour types. It should be highly immunogenic, be able to produce both humoral and cellular immune responses and it should be essential for tumour cell survival, thus not susceptible to mutation or deletion. As yet, such ‘ideal’ TAAs have yet to be identified in any malignant disease. In B-cell malignancies, the only compounds that come close to this ideal are the Id proteins produced by the clone of tumour cells. Unfortunately, these molecules are specific to each individual patient, requiring the vaccine to be tailor-made for each patient and are thus labour intensive and very expensive to produce. Furthermore, Id are weakly immunogenic when administered in vivo and, in the case of MM, expressed at low levels on the surface of the tumour cells. However, it has been shown that anti-Id antibody and T cells are present in the blood of MM patients (Yi et al, 1995), and experimental data shows that the anti-Id immune response is able to kill MM tumour cells in vitro and in animal models (Li et al, 2000; Wen et al, 2001). A number of different strategies have been employed to produce an effective Id vaccine. These have employed the use of DNA, purified Id protein or light and heavy chain variable regions (VH and VL), which can then be linked to an adjuvant molecule, such as keyhole limpet haemocyanin (KLH) or cytokines like IL-2, IL-12 or GM-CSF, in order to render them more immunogenic (King et al, 1998; Osterborg et al, 1998; Rasmussen et al, 2003; Stritzke et al, 2003). DNA acts as a natural vaccine adjuvant. Bacterial DNA is even more potent, because of the 20-fold increase in CpG motifs, recognised by molecules such as the Toll-like receptor 9 (TLR-9) present on cells of both the innate and adaptive immune system (Hemmi et al, 2000).
A number of studies in both MM and B-cell lymphoma have demonstrated that it is possible to use these vaccines to boost the immune system but the clinical results so far have been largely disappointing, although most series have been too small to produce statistically significant results. Coscia et al (2004) reported that, despite inducing anti-Id antibodies and skin-prick sensitivity in MM patients in first CR following high dose chemotherapy, the residual tumour burden was not eliminated. This is in contrast to the study reported by Bendandi et al (1999), in which Id protein vaccination was able to clear circulating tumour cells in eight of 11 patients who were otherwise in remission following chemotherapy for follicular lymphoma (FL). The group in Southampton has developed a DNA fusion vaccine, which contains the VH and VL genes of the Id protein assembled as a single-chain variable fragment (scFv) sequence. To enhance T-cell help, this has been fused to either the fragment C (FrC) sequence, a non-toxic part of the Clostridium tetani toxin (Spellerberg et al, 1997), or a plant viral protein coat sequence (Savelyeva et al, 2001). This scFv–FrC fusion gene vaccine has been used in clinical trials in patients with FL and MM (King et al, 1998). Antibody responses to FrC were seen in eight of 10 FL patients treated, with five patients having detectable T-cell responses to Id (Stevenson et al, 2004). The trial in MM patients is in its very early stages, but some immune and clinical responses have been reported (Stevenson et al, 2004). This platform continues to be modified in order to induce greater CD8+-mediated immunity by increasing Ag presentation via MHC class I. In a similar approach to Kim et al (2003), Stevenson (2003) also conducted a clinical trial of vaccinating normal donors prior to collection of DLI for use in MM patients who relapse following allo BMT.
A novel approach has been taken by Cell Genesys with the Gvax® myeloma vaccine. Irradiated, autologous MM cells are administered with K562 cells, genetically engineered to produce GM-CSF. Twenty-two patients have been enrolled in the phase I/II trial, and 17 patients have received at least one vaccination. Interim data has demonstrated that chemotherapy followed by auto HSCT and vaccination resulted in six complete responses, five partial responses, three patients with stable disease and two patients with progression (Borrello et al, 2004). Three patients with early progression after transplantation then demonstrated potential antitumour activity following initiation of vaccination, as measured by reductions in the myeloma-associated circulating protein (M-spike) of 92%, 37% and 25%. Treatment with GVAX® myeloma vaccine to date has been well tolerated, with only self-limiting skin rashes (two patients) and colitis (one patient) reported.
Dendritic cell vaccination in MM
Immunotherapy in MM may be more effective if it is delivered via professional APC, such as DC. A number of studies have examined the use of DC that have been pulsed with tumour-derived Id protein or peptides (Tarte et al, 1997; Dabadghao et al, 1998; Zeis et al, 1998; Reichardt et al, 1999; Titzer et al, 2000; Ridgway, 2003). It has been shown that moDC pulsed with purified patient-specific Id can serve as cellular vaccine for MM patients after high dose therapy and autologous peripheral blood stem cell transplantation (Lim & Bailey-Wood, 1999; Reichardt et al, 1999; Liso et al, 2000; Yi et al, 2003). In the Stanford study, 26 patients were immunised with Id-pulsed DC derived from autologous monocytes under serum-free conditions and vaccines consisting of matDCs (HLA-DR+/CD83+/CD80+/CD54+/CD86+, median number of 5 × 106 DC/injection) were administered without serious adverse events. Four patients have demonstrated Id-specific T-cell proliferative responses and two patients demonstrated the induction of Id-specific T-cell cytotoxicity (Reichardt et al, 1999; Liso et al, 2000). As all patients received Id/KLH boosters post-vaccination, it is not surprising that 24 of 26 patients developed KLH-specific T-cell responses after two to three Id/KLH booster injections. This protocol has been further developed in Tubingen, by the addition of GM-CSF, 250 μg/m2 s.c., as an adjuvant to vaccination. Two of 12 patients developed T-cell proliferative response and one patient formed a T-cell cytotoxic response (Reichardt et al, 2003). Investigators from Arkansas have also modified this approach by giving Id/KLH loaded DC by weekly intranodal injection for 4 weeks, each followed by low dose IL-2 (5 × 105 IU/injection, s.c.) (Yi et al, 2003). Seven of eight patients developed T-cell responses and all enhanced the anti-idiotypic B cell response. The serum paraprotein fell by 30% in one patient and five others continue to have stable disease with longer follow-up required. This group is also examining the use of MM tumour cells lysates as a source of Ag for DC priming (Wen et al, 2002; Szmania et al, 2003) whilst Kim et al (2003) are investigating the use of in vitro priming of allo donor T cells with tumour Id-pulsed DC.
Dendritic cell based Id vaccination of MM patients is feasible and can induce Id-specific immune responses in MM patients. However, the clinical effectiveness of such vaccinations in MM still needs to be proven. These therapies may need to be ‘boosted’ with adjuvants or other tumour-specific peptides in conjunction with Id-pulsed DC. Such an approach has been taken by Dendreon with the Mylovenge DC vaccine, which uses Id protein fused with GM-CSF, and is currently in phase II trials (Rice & Hart, 2002). On-going studies in this area will shed further light on the use of DCs as cellular vaccines but clearly this is an area worthy of further investigation.
T cells and reversal of tumour-induced immune suppression
The effects of the tumour cell population and associated microenvironment in MM are well described (Cook & Campbell, 1999). It has been known for many years that patients with MM are immunosuppressed, being more prone to infections (Zinneman & Hall, 1954; Glenchur et al, 1959; Perri et al, 1981), and MM is less responsive to normal T cells when given as DLI to treat relapse post-allo HSCT compared with other diseases, such as CML (Verdonck et al, 1998). This is at least partly because of MM tumour cells secreting immunologically active compounds such as IL-6, IL-10, VGEF and TGFβ among many others. It has been shown that MM cell lines and freshly isolated myeloma cells from patients produce excess TGFβ, and that this agent is responsible, at least in part, for suppressing T-cell responses against tumour cells (Cook et al, 1999; Campbell et al, 2001). Using the natural inhibitor to TGFβ, latency-associated peptide (LAP), the TGFβ suppressive effect against T cells has been shown to be specific and mediated through the inhibition of IL-2 autocrine pathways in the T cells (Cook et al, 1999; Campbell et al, 2001). Ex vivo activation, using anti-CD3 monoclonal antibody (MoAb) in the presence of exogenous IL-15, is able to overcome this inhibition and, crucially, the IL-2 autocrine pathways are reinstated in T cells, rendering them resistant to further TGF-β suppression (Campbell et al, 2001). Such a strategy may be employed to reinstate T-cell effector function in patients with MM, however we have demonstrated that these cells may be rendered cytokine-dependent and fail to respond to Ag's presented by DC in the absence of IL-2/IL-15 (unpublished data).
Other groups have identified that the T-cell repertoire in MM patients is severely skewed after auto BMT (Mariani et al, 2001) and that there is a significant association between survival and lymphocyte recovery post-auto BMT. In the Xcellerate T-cell trial (Vij et al, 2003), T cells are activated with anti-CD3 and anti-CD28 MoAb-coated beads and infused i.v. 3 d following stem cell infusion. Initial results suggest that this process partially corrects the skewing of the Vβ TCR repertoire and also rapidly corrects the lymphocyte count following auto BMT. It has been previously shown that patients with MM who have less restriction of the TCR Vβ repertoire have a better prognosis (Brown et al, 1997). At present there is little evidence that the cells manipulated by the Xcellerate process have any specific anti-tumour activity. However, the presence of expanded T-cell clones is associated with prolonged overall survival (Brown et al, 1997; Raitakari et al, 2003; Sze et al, 2003), and MM Ag-specific T cells circulating in the peripheral blood of MM patients have been identified (Yi et al, 1995; Beckhove et al, 2003; van Rhee et al, 2005). These cells could now be isolated from the peripheral blood of myeloma patients using immunomagnetic or flowcytometric methods, expanded ex vivo using some of the methods described above and re-infused into the patient.
It has been known for some time that immunosuppression with cyclophosphamide may increase the effectiveness of adoptively transferred anti-tumour T cells (Rosenberg et al, 1994), and recently it has been shown that the specific depletion of CD4+ CD25+ TR cells by anti-CD25 antibodies increases the efficiency of the anti-tumour immune response of tumour-bearing animals, although the tumours are not completely rejected (Jones et al, 2002). It is possible to enhance this effect by using CD25 depletion along with matDC vaccination (Sutmuller et al, 2001).
Tumour-directed MoAb therapy has been the ‘holy grail’ of many haemato-oncologists since the 1970s. With the development of compounds, such as Rituximab (anti-CD20), Campath-1H (anti-CD52) and Myelotarg (anti-CD33), the potential has developed into a viable treatment for lymphoma and leukaemia. This form of immunotherapy is now beginning to be investigated in the context of MM. One major hurdle in this area has been the selection of a suitable surface Ag that would permit the generation of a MoAb with satisfactory specificity and sensitivity for the targeting of the malignant cell, resulting in its destruction. Potential candidate molecules include CD38, CD138 (syndecan-1), CD54 (ICAM), CD40, VEGF (Yang et al, 2003a) and the unclustered surface type II transmembrane glycoprotein, HM1.24. The use of anti-CD20 MoAb is limited by the fact that CD20 is expressed on <20% of fresh myeloma cells. Despite the success of anti-CD20 MoAb in FL (McLaughlin et al, 1998; Zinzani et al, 2004), diffuse high-grade lymphoma (Coiffier et al, 1998) and Waldenstrom's macroglobulinaemia (Dimopoulos et al, 2002; Gertz et al, 2003), to date the experience in myeloma is limited (Treon et al, 2001; Musto et al, 2003; Lim et al, 2004), although it may have a role against the MM ‘stem cell’ as discussed later.
The use of anti-CD38 MoAb in clinical trials to date has been limited although some clinical efficacy has been demonstrated (Ellis et al, 1995). Initial difficulties associated with human-anti-mouse antibodies have largely been temporised with the use of a humanised variant.
One advance with respect to improving efficacy is the engineering of MoAb as carriers of toxin genes. Workers in the Mayo Clinic have generated scFv that represent the linkage of the carboxyl terminus of one variable region (IgH) with the amino terminus of the other (IgL), using nucleotides that encode a series of hydrophilic peptides that retain the original antibody specificity (Chen et al, 1995). The investigators have attached diphtheria toxin-A plasmid DNA to a scFv directed against CD38, which is internalised into the Ag-expressing cell by receptor-mediated endocytosis. Using this technology, in vitro and animal studies have demonstrated cell suicide by expression of the diphtheria toxin-A and the investigators intend to directly target plasma cells using tissue-specific transcriptional regulatory elements (e.g. Ig heavy chain enhancer) to limit non-specific cell killing. The translation into clinical trials is eagerly awaited using this novel technology.
Immunologically active drugs
Thalidomide and its analogues
Thalidomide has been successfully used to treat patients with relapsed/resistant myeloma (Singhal et al, 1999). While thalidomide may act as an anti-angiogenic factor in myeloma, several other potential mechanisms of action have been proposed. First, thalidomide may directly inhibit tumour cell growth and differentiation, mediated by reduced cell adhesion interactions and inhibition of cytokine secretion (e.g. TNF-α) or through free-radical generation and apoptosis, either directly or in response to drugs, such as dexamethasone. Secondly, thalidomide may inhibit the activity of VEGF and basic fibroblast growth factor (bFGF-2), which act as growth and survival factors for myeloma cells. Thirdly, thalidomide may act by promoting a Th1 T-cell response resulting in the secretion of interferon-γ and IL-2. Whether, one or all of these mechanisms mediate the clinical effects of thalidomide in myeloma remains to be clarified. Several thalidomide analogues are in the early phases of clinical development aiming to either inhibit cytokines (phosphodiesterase 4 inhibitors) or induce immunomodulation, with fewer side-effects (Corral et al, 1999). There is tremendous future potential in the use of thalidomide or its analogues and studies, such as Medical Research Council Myeloma IX, aim to assess the efficacy of thalidomide in combination with induction chemotherapy and as a maintenance strategy following auto BMT.
Interleukin-2 is a central regulator of immune responses mainly produced by activated CD4+ T cells. However, transformed T cells, B cells, leukaemia cells, lymphokine-activated killer cells, NK cells and DC may also secrete IL-2. IL-2 is a growth factor for all T-cell subpopulations, inducing Ag-unspecific proliferation of T cells by inducing cell cycle progression in resting cells, and clonal expansion of activated T-lymphocytes. IL-2 also promotes the proliferation of activated NK cells and B cells (but this requires the presence of additional factors, for example, IL-4).
The observation that leucocytes cultured in IL-2 were able to lyse tumour cells (Lotze et al, 1981; Grimm et al, 1982) ultimately led to IL-2 being administered to patients with melanoma, renal cell carcinoma and B-cell lymphoma (Rosenberg et al, 1985, 1987; Lotze et al, 1986, ) with a number of complete and partial remissions being reported (Yang et al, 2003b). IL-2 has no direct effect on tumours, as high concentrations do not inhibit tumour cell growth in vitro and the anti-tumour properties are mediated by the cellular immune system, i.e. NK and T cells. Investigators have combined the use of immune effector cells (tumour infiltrating lymphocytes) and IL-2 in the search for more potent therapies (Rosenberg & Lotze, 1986; Dudley et al, 2002).
Intravenous IL-2 is a toxic therapy. The major side-effects include hypotension, capillary leak syndrome, organ dysfunction and a treatment-related mortality of 1·5%. This has prompted investigators to reduce the dose and change to the subcutaneous route of administration. Although there is debate as to whether this reduces the efficacy of IL-2 therapy (Ravaud et al, 1994; Yang et al, 2003b), Kiss et al (2003) have reported that low dose s.c. IL-2 (1 MU b.i.d. for 1 week, 2 MU b.i.d. for 3 weeks) induced a complete remission of relapsed Hodgkin's disease following sibling allo HSCT complicated by acute skin GVHD. Slavin et al (1996) have reported long-term remissions using in vitro activation of DLI and SC IL-2 for the treatment of relapsed leukaemia following allo HSCT that was unresponsive to DLI alone.
These observations suggest that IL-2 may be a useful adjuvant to cellular immunotherapy of MM by providing continuing stimulation of the immune system following the administration of immune effector cells. IL-2 could also be given following allo HSCT in place of, or in addition to DLI to boost any immune mediated effect.
Efficacy measurements in immunotherapy trials
Progress in the understanding of the basic mechanisms involved in generating immune responses to tumour Ags has never been greater. However, there is a lack of accurate, reproducible and readily transferable measurements of efficacy to assess the immune responses generated in the context of tumour immunotherapy. The role of efficacy measurements in association with immunotherapy trials is fundamental to the full assessment of such novel strategies. It has been assumed that current treatment modalities are themselves curative in patients who achieve long-term disease-free survival. The role of an autologous anti-tumour immune response remains to be clarified. It has been shown that it is an autologous immune response to the residual disease that prevents relapse in acute leukaemia (Lowdell et al, 1997; Lowdell et al, 1999; Lowdell & Koh, 2000) and the data suggests that the immune system has some impact on the disease course of MM patients. Knowing whether there is an anti-tumour immune response, albeit inadequate, allows one to construct testable hypotheses about how it may be enhanced or induced when absent. Such hypotheses inevitably lead on to the design of appropriate assays of immune function during and after administration of the immunological intervention under trial.
Measurements of immune function have been notoriously unreliable and poorly reflect the true status of the patient. However, recent advances in measurements of cell activation, cytokine production and cell-mediated cytotoxicity have radically changed the reproducibility of such experiments. The identification of relevant peptides from TAA is enabling the construction of peptide/HLA tetramers, which can be used to enumerate and analyse the function of Ag-specific T cells (Howard et al, 1999). In all such assays it remains important that the true measure of a relevant immune response is not that the patient or animal model can be demonstrated to have responded to the vaccine but rather that the response is measurable against the primary tumour. This is plainly easier said than done. Nonetheless it should be an aim at the outset of research so that the primary tumour is collected and stored in an appropriate manner at presentation for use in later experiments. Clearly, robust, reproducible and relevant outcome measurements of immune function are essential to determine the efficacy of any immunotherapy intervention.
In view of the disappointing clinical response seen thus far in immunotherapy trials, a number of groups have looked at whether immune therapies may be more effective if used in combination. Shimizu et al (1999) showed that immune responses to a tumour lysate-pulsed DC vaccine could be increased by the addition of IL-2 in a murine sarcoma model. Yi et al (2002) have reported the use of SC low dose IL-2, 5 d following administration of idiotype-pulsed DC. Anti-idiotype T-cell responses developed in four and B-cell responses in all five patients, including a 50% reduction in M-protein in one patient (Shimizu et al, 1999; Yi et al, 2002). In a murine model, 81% of animals treated with Id vaccine, plus FLT-3 ligand (FLT-3L) and IL-2 survived more than 180 d compared with none given the individual therapies, and 27% and 41% given Id plus IL-2 or FLT-3L respectively (Zeis et al, 2002). Another approach has been taken by the group in South Carolina, who have combined Rituximab and low dose IL-2 s.c. in non-Hodgkin lymphoma. Responses correlated with NK cell numbers and the addition of IL-2 resulted in an increase in Ag-dependent T cell cytotoxicity in responders (Gluck et al, 2004). However, some of the most promising results to date have come from the National Institutes of Health, where the combination of vaccination, adoptive transfer of T cells and IL-2 therapy resulted in some long-term cures in a mouse melanoma model (Overwijk et al, 2003).
The field of cancer immunotherapy stands at a threshold. There have been great advances in the understanding of the immune system and how it can be manipulated, and in a number of diseases, immune-based therapies are beginning to realise their potential. The use of antibody therapy, such as Rituximab, in combination with conventional chemotherapy for the treatment of lymphoma has resulted in improved disease responses and is becoming standard practise. Even more impressive results have been achieved using cellular immunotherapy for the treatment of post-transplant lymphoproliferative disorder (PTLD) (Bollard et al, 2003). PTLD is a complication of HSCT or solid organ transplantation, associated with regimens that contain antibodies, such as anti-lymphocyte or anti-thymocyte globulin, that generate more profound cellular immunodeficiency than those that do not contain such antibodies. This may allow uncontrolled expansion of Epstein–Barr virus (EBV)-infected B cells. These infected B cells express the full spectrum EBV-latent Ags, which are well characterised and highly immunogenic, and there appears to be no additional immune suppression associated with the malignant clone. Adoptively transferred autologous or HLA class I-matched EBV-specific T cells are able to restore cellular immunity, eradicating the disease without the need for conventional chemotherapy. However, whilst we can consistently generate autologous cellular and humoral anti-MM immune responses in vitro and in vivo, they have yet to be translated into effective clinical tools that are capable of inducing sustained disease responses. We have also yet to show that the anti-MM immune effects seen following allo HSCT can be separated from those because of major and minor histo-incompatibility. If immunotherapy of MM is to develop into an effective therapy, a number of crucial questions still need to be answered.
What are the most effective anti-MM Ag, can they be used alone, or are they more effective in combination?
A number of new and old target Ags are currently under investigation. Of these molecules, Id protein is the most widely studied. We await with interest the results of studies, such as those from Southampton, which look at combining Id with adjuvant compounds to increase the potency of the anti-MM immune response. Newer molecules, such as HM1.24 and NY-ESO-1 are in the early phases of development for the treatment of MM (Chiriva-Internati et al, 2003; Rew et al, 2003). Most MM immunotherapeutic strategies are directed at mature MM cell Ags and will thus be effective against the bulk of the tumour cells. However, as with other targeted therapies, such as Glivec in CML, this approach has the potential to leave these tumour stem cells unscathed, resulting in disease relapse (Elrick et al, 2005). MM tumour cells are classically thought of as mature plasma cells (CD45+ve/CD19-ve/CD56+ve/CD138+ve/CD38+ve), with a low proliferative index. However, as with other leukaemia, there may be a small population of ‘tumour stem cells’ (CD45+ve/CD19+ve/CD22+ve/CD138-ve) that retain the capacity to self-renew and proliferate (Matsui et al, 2004). These cells are able to repopulate non-obese diabetic severe combined immunodeficient mice and if removed by purging with Rituximab, the MM fails to engraft (Jones, 2003; Matsui et al, 2004). It may be a mistake to select MM Ags that may only be present on the non-self-renewing cells in a bid to get good clinical responses, when we need to direct our efforts at the cells that contribute to relapse, thus improving long-term survival. Further studies of the putative MM stem cell are warranted. Examining their phenotypic and genotypic characteristics may identify novel immunotherapeutic targets. It remains to be seen whether a single Ag can effectively eradicate MM in-vitro. Approaches that combine multiple Ags, such as tumour lysate or apoptotic tumour cells, have the theoretical advantage of being less likely to induce the MM clone to delete the target Ag, a phenomenon known as Ag escape.
Can simple vaccination induce an effective anti-MM response or do we need to deliver the Ag(s) via professional APC generated ex vivo?
At present, immunotherapy of MM is developing along two main themes, the more traditional vaccination-based approach, and ex vivo generated cellular therapies. In the future it may be possible to combine these two approaches by preconditioning the site of vaccination with growth factors and chemokines in order to recruit APC. Efficient trafficking of these Ag-loaded APC to lymph nodes could then be induced by vaccines containing adjuvants designed to promote maturation and migration of APC. This type of approach would avoid all the inherent costs and risks of having to collect cells and then ex vivo generate Ag-loaded, mature APC, for each individual patient. In addition, humoral as well as cellular immune responses may be induced.
How does the MM tumour cell evade immune surveillance?
The evidence to date indicates that the MM tumour cell is able to disable the immune system at multiple crucial points [Fig 2]. We are beginning to understand this interplay between the malignant cell and the immune system of the tumour-bearing host. This is prompting researchers to devise strategies that overcome the immunosuppressive effects of the MM tumour cell, such as T-cell preactivation, in order to increase the effectiveness of immune therapy against the malignant clone.
Can immunotherapies be effective when used alone, or are they best used in combination in the treatment of MM?
In view of the hostile in vivo environment induced by the MM tumour cell, and the disappointing clinical response seen thus far in immunotherapy trials in MM, it is our view that immune therapies may be more effective if used in combination. A number of groups are examining the effectiveness of combinations of immunotherapy, such as Id-pulsed DC and IL-2, CTL and IL-2, and antibody therapy with IL-2. Another approach would be to combine a DC-based vaccine to prime the immune system, prestimulation of T cells to correct tumour-induced anergy, followed up with low dose s.c. IL-2 to maintain the activation of the immune system. As previously discussed, a number of groups are examining the use of combinations, such as the adoptive transfer of Ag-specific T cells and IL-2, and Ag-loaded DC and IL-2.
Until we can answer these questions, effective immunotherapy in myeloma will remain a possibility not a probability. The on-going basic and translational research is encouraging, but the proof of principal in the clinical arena remains to be demonstrated. The design of good clinical trials will be crucial, as is optimal immunological monitoring to measure what, if any, effect such strategies will have on the host's immune system. As with other biological therapies in haemato-oncology, it is likely that these immune-based therapies will achieve their maximum effect when the tumour burden is reduced to a minimum. Thus, immune strategies against MM are likely to be useful adjuncts to conventional chemo-radiotherapy and autologous HSCT.
SJH is a clinical research fellow funded by the Scottish National Blood Transfusion Service. Thanks to Prof. I.M. Franklin and Prof. T.L. Holyoake for their assistance in the preparation of this manuscript.