Cancer Therapy Vol 1, 353-362, 2003

1Division of Immunology, Infection & Inflammation and ²Division of Cancer Sciences & Molecular Pathology, University of Glasgow, Western Infirmary, Glasgow, G11 6NT, Scotland, U.K.

__________________________________________________________________________________

Key words: B lymphocyte, breast cancer, Medullary carcinoma, Her 2^neu, MUC-1, p53, Carcinoembryonic antigen, Antibody-mediated tumour cell killing

Abbreviations: Ductal carcinoma in situ, (DCIS); Infiltrating ductal carcinoma, (IDC); Medullary carcinoma, (MC); tumour-infiltrating lymphocytes, (TIL); complementarity determining regions, (CDR); framework regions, (FR); epidermal growth factor receptor, (EGFR); carcinoembryonic antigen, (CEA); antibody-dependent cell-mediated cytotoxicity, (ADCC); follicular dendritic cell, (FDC)

Breast cancer is the most common malignancy of women in many parts of the world, with a lifetime incidence around 1/12. Despite advances in screening, diagnosis and treatment, many people, nearly all women, still die from the disease each year. A variety of distinct histopathological subtypes are recognised. Ductal carcinoma in situ (DCIS), confined to the mammary ducts and lobules, has a cure rate approaching 100%. Infiltrating ductal carcinoma (IDC) accounts for 80% of breast cancer. Malignant growth starts in the ducts and lobules prior to invasion and ultimately metastasis to distal sites. Lobular carcinoma in situ and invasive lobular carcinoma, which represent 10-15% of breast cancer, behave analogously. Medullary carcinoma of the breast (MC) is relatively uncommon, comprising a mere 1-5% of breast cancers. Despite this, it has received the most attention with respect to the role of the immune response in breast carcinomas.

Prior to the 1940s, virtually no attention was paid to tumour-infiltrating lymphocytes (TIL) in breast cancer. As interest in lymphocyte infiltration grew, more reports were published and attention turned to the types of cells involved in the infiltrates. Most breast cancers (60-80 %) have detectable lymphocyte infiltrates . These infiltrates consist largely of T cells, with variable numbers of macrophages, natural killer cells and B-cells . Presently, there is disagreement on the relative abundance of T-lymphocyte sub-populations. Some investigators report CD4⁺ T cells as predominant, especially in large tumours while others show a higher proportion of CD8⁺ T-lymphocytes . Owing to the abundance of T lymphocytes in tumour tissue and the belief that cytotoxic T lymphocytes constitute the major contribution to immune responses against tumours, most in-depth studies on TIL have concentrated on T cells . In contrast, tumour-infiltrating B cells are less well characterised despite the presence of TIL B-cells in at least 20% of breast cancers .

B lymphocytes control humoral immunity and represent the effector cell type involved in the production of antibodies. The generation of antibody-secreting plasmacytes and memory B cells occurs in secondary lymphoid organs. An immune response is initiated when specialised antigen-presenting cells such as dendritic cells acquire antigen in the periphery and migrate to the draining lymph node where the antigen comes into contact with naďve or memory lymphocytes. Antigen-specific B cells presented with their complementary antigen become activated and undergo developmental events in the germinal centre, including somatic mutation of immunoglobulin V genes and affinity selection, which lead to a large clonal expansion of antigen-specific plasma cells and memory B cells . High-affinity antigen-specific B cells leave the lymph node and migrate to the periphery where they re-circulate. Plasma cells entering the circulation migrate principally to the bone marrow.

While the appreciation of lymphocyte infiltration in breast cancer has grown, controversy surrounds the prognostic significance of lymphocytic infiltrates and the relative importance of various types of lymphocytes. The controversy around the precise role of tumour infiltrates has produced two competing hypotheses: one that lymphocytic infiltrates merely reflect non-specific inflammatory reactions resulting from tumour-derived chemokines and cytokines, and the other, that they represent defensive reactions, overridden by metastatic disease.

For most cancers including breast cancers, the immune response seems to have a limited effect on tumour behaviour. However, detailed analysis of the host response has led to the identification of a number of tumour-associated and tumour-specific antigens that are over-expressed, mutated or structurally modified. The identification of several self-antigens such as the HER-2^neu protein, p53 , CEA, c-Hras, c-myc and MUC-1 indicates that at least some cancers are immunogenic, and elicit humoral and cellular immune responses, but a tumour-specific immune response does not necessarily translate into tumour rejection. On the other hand, it is possible that many early tumours rejected by the immune system do not progress and only those that the immune response fails to eliminate do progress to a clinically detectable stage. This could account for the high incidence of malignancies in immuno-suppressed individuals.

MC is characterised by a syncytial growth pattern, well circumscribed tumour borders, and dense lymphoplasmacytic cell infiltrates in the tumour stroma . Brisk mitotic rates, large pleomorphic nucleoli sparse necrosis and a lack of acinar differentiation are also characteristic. This type of breast carcinoma has also been described as bulky carcinoma, circumscribed breast carcinoma, solid circumscribed as well as medullary carcinoma with lymphoid stroma.

MC is relatively infrequent, representing 1-5 % of breast cancer but has received considerable attention since the late 1940s when it was reported that patients with MC had a five-year survival of 82.7% compared to 50% for patients with the usual infiltrating forms of carcinoma of the breast . In addition, nodal metastases at mastectomy were 43% in MC compared to 60% in IDC. Moore and Foote suggested that the extensive plasma cell infiltrate could be partly responsible for the unusual survival they observed. Gorski and co-workers confirmed these findings . In their study, axillary metastases were found in 16% of patients with MC and 49% of patients with other types of breast carcinoma and axillary metastases in patients with MC did not diminish survival. Ten-year survival for patients with MC was 68% compared with 52% for other breast cancer patients. Maier et al (1977) also reported a better prognosis for MC patients. This group found that lymph node involvement was less common in MC than IDC, but considered the size of the tumour more important than lymph node involvement.

A review of 192 patients confirmed the favourable prognosis of MC compared to IDC diagnosed and treated over the same time span in the same institute . An attempt was made to re-define MC reproducibly using 11 histological criteria to include precisely defined patients in the study. In contrast to other reports, Ridolfi and co-workers did not report a significantly lower frequency of axillary metastases in MC than other types of breast cancers. However, survival of node-positive patients with MC was significantly better than node-positive patients with IDC.

Rapin and co-workers (1988), using the same diagnostic criteria as Ridolfi et al, reported better 5 and 10-year survival, which they attributed to less frequent involvement of axillary nodes in MC. Whether this was because of a lower tendency to metastasize or a specific immunologic response was not determined. Richardson, working on 117 patients, also reported a better prognosis for MC patients . The five-year survival of patients with MC and axillary metastases was better than of patients with non-MC and axillary metastases, 72% and 50%, respectively. According to Richardson however, the better prognosis was not related to the intensity of lymphoid infiltration. He attributed the better prognosis of MC to its poor stroma formation and reduced ability to metastasize.

Several papers do not support the conclusion that MC has an especially favourable prognosis. When identical cancer stages were compared, the 5 and 10-year survival rates for patients with MC paralleled those of patients with ductal carcinomas . These results corroborated earlier work on 140 patients not matched for disease stage, which found that patients with "circumscribed carcinoma" had the same survival rate as those used for comparison . On the other hand, a study of 104 cases of MC in a series of 1411 with similar stage reported that after twenty years, 74 % of cases with operable MC were alive compared with 14% for non-MC breast cancers . Although 68% of the MC cases in this study were reported to be high grade, axillary metastases were present in only 39% of cases. Compared with other types of breast cancer, unsuccessfully treated MC patients died fastest of all groups, with deaths due to the carcinoma being rare after 5 years. The authors concluded that MC are essentially aggressive malignancies, as indicated by their tumour grade, but their biological potential is countered to a considerable extent by the host's immune response.

Two decades after the findings of Moore and Foote, Schwart reported that 10-year survival was, in fact, worse for patients with "circumscribed carcinoma" among both those with axillary metastasis (28% v 37%) and without (60% v 76%), compared to other types of breast cancers . Flores et al (1974) concluded that lymphocytic infiltration in MC was associated with an over-all poorer response, as it was positively associated with a higher rate of metastasis.

Between the 1940s and early 1990s controversies surrounding MC included its definition, its prognosis and the role of the infiltrating lymphocytes. Distinction of MC from several other forms of invasive breast carcinoma can be difficult. Much of the controversy concerning the prognosis of MC may be attributable to the use of varied or uncertain classification criteria. An attempt was made by Ridolfi et al to group several histological criteria and thus define MC more precisely with a reproducible classification. Pedersen and co workers (1991), further simplified these criteria by eliminating those with poor inter-/intra-observer agreement and those deemed to imply no or little impact on prognosis. Another reason for the differing conclusions concerning the prognosis of MC could be that survival of patients is estimated across different treatments. During this period, there was general agreement on one aspect of MC, namely, that it has less nodal metastasis and when it does occur, fewer nodes are positive than with infiltrating ductal carcinoma .

Towards the end of the 1990s, data were emerging on the role of tumour antigens and the humoral immune response in MC. Kotlan and co-workers (1999) identified somatically mutated immunoglobulin variable region genes expressed by tumour tissue B lymphocytes infiltrating MC. In addition, certain V-genes were preferentially represented, suggesting selection of sub-populations of tumour infiltrating B cells. Coronella et al (2001) demonstrated clonal expansion of tumour-infiltrating B cells in MC. It is well established that B-cells undergoing a germinal centre response, i.e. clonal proliferation accompanied by somatic hypermutation and affinity selection, are driven by antigen, which is required as both the initial trigger and at the later stage of affinity selection . The observation of clonally expanding, mutating B cells infiltrating the tumour therefore implies a local, antigen-driven response.

The plasma cells associated with breast carcinomas are predominantly of IgG subtype, with some IgA and few IgM . In contrast, the plasma cells associated with normal breast tissue and other secretory epithelia are predominantly IgA . The prominence of IgG in MC versus the usual IgA plasma cells, and the association of infiltrates with the tumour stroma, suggest a specific immune response to tumour-derived antigens but the putative driving antigens have largely not yet been identified.

Hansen et al (2002) showed that actin becomes exposed on the cell surface of a large proportion of apoptotic MC cells as an early apoptotic event and that cloned anti-actin antibodies bind specifically and with high affinity. b -actin is prevalent in most cells and is normally non-immunogenic, although low affinity antibodies can be found in healthy individuals. Hansen and co-workers proposed that owing to the increased rate of apoptosis of MC cells, the anti-actin immune response was generated when b -actin was exposed on the surface of apoptotic cells. The clinical significance of this antigen remains to be elucidated.

Non-medullary carcinomas account for most breast cancer but unlike MC, they have more variable numbers of tumour-infiltrating lymphocytes and the role of these cells in tumour immunity has not received much attention. The most comprehensive study on the role of lymphocytes in breast cancers was carried out recently by Ben-Hur et al (2002) who investigated different types of breast cancer; fibrocytic disease, fibroadenoma, carcinoma in situ, invasive lobular carcinoma and infiltrating ductal carcinoma. Modest lymphoid infiltration was observed in fibrocytic disease, fibroadenoma and some infiltrating ductal and lobular carcinomas. In contrast, carcinoma in situ and some infiltrating ductal and lobular carcinomas were associated with extensive lymphoid infiltration. DCIS but not lobular carcinoma in situ is often associated with a dense lymphocytic infiltrates. Ramachandra et al (1990) could not interpret the role of the lymphocytes but Ben-Hur et al considered the infiltrating lymphocytes to indicate a host reaction to the tumour.

An investigation of tumour-infiltrating B lymphocytes in IDC demonstrated the presence of T cells, follicular dendritic cells, B cells and plasma cells in the tumour-infiltrating lymphoid clusters (Figure 1) . The presence of FDCs is particularly significant as they are normally localised to germinal centres in peripheral lymphoid tissues where they play a crucial role in affinity maturation by antigen selection of B cells with high-affinity mutated antigen receptors .

Secondly, highly mutated B cell clones were identified in the B cell clusters in the carcinoma and draining axillary lymph nodes (Figure 2) . As already noted in the previous section, B cells undergoing a germinal centre reaction as indicated by clonal proliferation and somatic hypermutation are driven by antigen which is required as both the initial trigger and during affinity selection. B cell clones in the tumour originate from the first tumour-draining lymph node, (sentinel lymph node), and possibly other axillary nodes (Nzula et al, manuscript submitted for publication). A few B cells undergoing antigen-driven clonal proliferation and somatic hypermutation in germinal centres of the tumour-draining lymph nodes migrate into the breast tumour where they undergo further rounds of clonal proliferation, somatic hypermutation and affinity selection (Figure 3).

Germline, unstimulated V_H gene family usage in normal peripheral blood lymphocytes is well understood . Significant deviations from this baseline state are indicative of antigen-driven selection. Highly biased representation of the uncommon V_H5 gene family and V_H1 in tumour-infiltrating B cells is indicative of antigen-driven selection within the tumour (Figure 4). This is corroborated by the high proportion of replacement mutations in complementarity determining regions (CDR) relative to the framework regions (FR) . The six CDRs (three in the heavy chain and three in the light chain) form the antigen binding sites of an antibody whereas the FRs form the structural domains of the antibody. Although CDRs are more susceptible to mutations leadingto amino acid replacements, a high replacement to silent ratio in the CDR is believed to result from selection of mutations providing the best 'fit' for the antigen, whereas the lower ratio in the FR preserves the domain structure supporting the antigen-binding site. Replacement mutations that improve antibody affinity in the CDR are therefore selected for, whereas replacement mutations in the FRs, which may disrupt domain structure, are selected against.

Fabs cloned from lymphocytes infiltrating IDC recognised a breast cancer cell line and autologous tumour tissue lysate , suggesting that some of the lymphocytes are responding to tumour antigen rather than non-specific inflammatory or cytokine signals. However, the nature of the antigens recognised by tumour infiltrating plasmacytes in IDC is still to be determined.

Figure. 1. Immunohistochemical staining of ductal breast carcinoma. B cells were identified with anti-CD20 (x200), T cells with anti-CD3 (x200), FDCs with anti-FDC (x200), plasma cells with anti- plasma cell (x200). (reproduced with permission, from Nzula et al, Cancer Research 63:3275-80, 2003)

Figure. 2. A B cell clone showing clonal proliferation and somatic hypermutation within a cluster of lymphocytes in breast carcinoma tissue, based on analysis of clonally related Ig V_H-gene sequences. The clone is shown as proliferating from the best matching germline V_H-gene segment. Each circle symbolises a B cell and letters within depict individual V_H-gene sequences. Deduced intermediates are shown as dotted circles. The numbers alongside the arrows represent the minimum number of mutations between the different sequences. Reproduced with permission, from Nzula et al, Cancer Research 63:3275-80, 2003)

Characterisation of antigens driving humoral and cellular immune responses has been intensely investigated over the last few decades, and detailed molecular analysis of the pathways that control tumour growth has identified tumour-associated growth factor receptors as potential targets for therapy. The epidermal growth factor receptor (EGFR) belongs to the immunoglobulin super-family and bind to different growth factor ligands, resulting in increased DNA synthesis . The ErbB family, a subset of the EGFR family, consist of four members designated ErbB1-ErbB4. ErbB interaction with growth factors results in the initiation of a signalling cascade with the activation of tyrosine kinase as a crucial step.

ErbB2, also called (human epidermal growth factor receptor 2 (Her2), c-erbB-2, Her2^neu), is over-expressed in about 20% of breast cancers. Her2^neu over-expression is associated with a poor relapse-free survival following treatment in early stage, node negative patients with breast carcinoma . A humanised anti-Her2^neu monoclonal antibody herceptin/trastuzumab, was developed by Genentech and tested in clinical trials of patients with metastatic disease. Phase I and II trials demonstrated objective clinical and radiographic responses, particularly when used in combination with cytotoxic chemotherapy . The results from extensive clinical trials led to herceptin being FDA-approved for the treatment of breast cancer in 1998, followed by approval in the UK. Herceptin is indicated for the treatment of metastatic breast cancer where the tumour cells over-express Her 2^neu protein and the patients are receiving chemotherapy . Therapy directed against Her-2^neu is now well established in this clinical setting, a clear indication of the beneficial effects of passive immunotherapy with a humanised monoclonal antibody for treating human cancers.

The mucins are another group of important tumour antigens. These glycoproteins are major secretory products of various epithelial cells including breast. The human epithelial mucin MUC-1 is a large, complex and heavily glycosylated transmembrane glycoprotein expressed at the luminal surface of most glandular epithelial tissues. Expression of MUC-1 is increased in many epithelial malignancies, such as breast, gastric, pancreatic and ovarian cancers, and a proportion of colonic and lung cancers . Over-expression of MUC-1 occurs in 90% of breast cancers and is believed to be an indicator of poor prognosis . In malignancy, under-glycosylation occurs so that shorter carbohydrate chains are produced, allowing the exposure of potentially antigenic cryptic carbohydrate and peptide segments .

Figure 3. B cells clones identified in breast tumour tissue originate in the sentinel lymph node (SLN). The clone in the SLN is shown as proliferating from a founder memory B cell. Each circle symbolises a B cell. B cells migrate from the clones in the germinal centre of the SLN to the lymphocytic cell cluster in the tumour, where they undergo further rounds of somatic hypermutation and clonal proliferation

High levels of anti-mucin antibodies have been detected in pre-treatment breast carcinoma patients . B cells and monoclonal antibodies specific for MUC-1 have been isolated from the tumour-draining lymph nodes of breast as well as other carcinomas . Production of IgG and IgM anti-MUC-1 antibodies is associated with better disease-specific survival . The good prognosis of patients with anti MUC-1 antibodies indicates that treatment of MUC-1 expressing carcinomas with monoclonal antibodies could be beneficial. Moreover, vaccination with MUC-1 derived glycopeptides might also favourably influence the outcome of disease.

A synthetic vaccine, Theratope, which corresponds to the glycan epitope of MUC-1, is currently undergoing a multinational phase III trial to test its effectiveness in metastatic breast cancer. Theratope stimulates both the production of antibodies and CD4⁺ T cell responses . If successful, the vaccine may be applicable to other types of cancers such as ovarian and pancreatic where MUC-1 is over-expressed on the tumour cells.

The p53 gene is one of the best-known tumour suppressor genes in humans. Wild-type p53 is a critical regulator of cell cycle control pathways whereas the mutated p53 gene acts as an oncogene.

Figure 4. VH gene family usage of combined patient data differs significantly from the expected frequencies and from normal PBL (p=0.000015 & <0.0001 respectively). (reproduced with permission, from Nzula et al, Cancer Research 63:3275-80, 2003)

Accumulation of p53 protein, owing to mutations in the p53 gene, is a common event in breast, lung, cervical, colon and gastric carcinomas . Accumulation of mutated p53 can trigger the humoral immune response and p53 antibodies have been detected in the sera of patients with different types of tumours . These antibodies can also recognise the native form of p53 . Anti-p53 antibodies can indicate breast cancer relapse . In a retrospective study of 24 breast cancer patients, Regele et al reported a decrease in p53 antibodies that paralleled therapy in 64% of the patients. In 18% of patients, relapse was preceded by an increase of the antibody titre. Thus, monitoring serum levels of p53 antibodies could provide essential information about the clinical course of the disease.

Expression of p53 is correlated with resistance to paclitaxel, an important agent in the pharmacological treatment of metastatic breast cancer . A study of 33 patients with metastatic breast cancer found that none of the tumours with p53 expression responded to paclitaxel. In contrast, all of the patients without p53 expression responded to treatment. Thus, p53 expression could be used to identify paclitaxel-resistant patients prior to using ineffective therapy with major side effects.

Human carcinoembryonic antigen (CEA) is a glycoprotein normally expressed in colon epithelium and some fetal tissues. CEA is over-expressed on approximately 50% of breast cancers and carcinomas of the colon, pancreas, lung and gastrointestinal tract . Despite the poor immunogenicity of endogenous CEA, recent recombinant vaccine strategies have demonstrated cellular and humoral immune responses that recognise CEA and kill tumour cells . In addition, Phase I clinical trials of a recombinant vaccinia virus-CEA vaccine in metastatic colorectal, breast and lung cancer patients showed enhancement of immune responses with no toxicity . Consequently, CEA represents another potential target for recombinant vaccines against different types of cancers including breast.

There are several mechanisms through which antibodies bound to tumour antigens can induce tumour cell death. The main mechanism is called antibody-dependent cell-mediated cytotoxicity (ADCC), in which antibodies bound to tumour cells activate effector cells of the immune system. Receptors expressed on natural killer cells and other leucocytes bind to the antibody-tumour cell complex through the antibody Fc region, release cytotoxic granules containg perforin and granzymes and destroy the tumour cell . Antibody binding to tumour antigens can also trigger the classical complement cascade, leading to complement-mediated cell lysis . When the first component of complement, C1q, binds to the Fc portion of the antibody-tumour cell complex, this stimulates binding and activation of the remaining components of the complement system (C1-C9) at the cell surface. The activation of complement results in the release of anaphylatoxic and chemotactic factors (C3a, C5a) and the formation of the membrane attack complex (C5b-C9), which initiates lysis of the tumour cell membrane.

The function and prognostic significance of tumour-infiltrating lymphocytes has been controversial for over 50 years. It is generally believed that they indicate protective involvement of the host's immune system. Significant effort has been devoted to the identification of antigens recognised by T cells, with antibodies assumed to have little impact on the growth of breast tumours. In reality, both cellular and antibody-mediated mechanisms are likely to be important in controlling and eliminating tumours. Passive immunotherapy with a humanised monoclonal antibody against HER-2^neu is now well established for patients with HER-2^neu over-expressing breast cancers. The effectiveness of herceptin demonstrates that antibodies against breast cancer antigens can be potent agents of breast cancer control, if not cure. Since only about 20% of patients have HER-2^neu positive cancers, this immunotherapeutic strategy currently excludes a sizeable fraction of women with breast cancers. Antibodies against more widely expressed tumour antigens show potential in the treatment of breast and other carcinomas.