PDGF-BB regulates splitting angiogenesis in skeletal muscle by limiting VEGF-induced endothelial proliferation.

VEGF induces normal or aberrant angiogenesis depending on its dose in the microenvironment around each producing cell in vivo. This transition depends on the balance between VEGF-induced endothelial stimulation and PDGF-BB-mediated pericyte recruitment, and co-expression of PDGF-BB normalizes aberrant angiogenesis despite high VEGF doses. We recently found that VEGF over-expression induces angiogenesis in skeletal muscle through an initial circumferential vascular enlargement followed by longitudinal splitting, rather than sprouting. Here we investigated the cellular mechanism by which PDGF-BB co-expression normalizes VEGF-induced aberrant angiogenesis. Monoclonal populations of transduced myoblasts, expressing similarly high levels of VEGF alone or with PDGF-BB, were implanted in mouse skeletal muscles. PDGF-BB co-expression did not promote sprouting and angiogenesis that occurred through vascular enlargement and splitting. However, enlargements were significantly smaller in diameter, due to a significant reduction in endothelial proliferation, and retained pericytes, which were otherwise lost with high VEGF alone. A time-course of histological analyses and repetitive intravital imaging showed that PDGF-BB co-expression anticipated the initiation of vascular enlargement and markedly accelerated the splitting process. Interestingly, quantification during in vivo imaging suggested that a global reduction in shear stress favored the initiation of transluminal pillar formation during VEGF-induced splitting angiogenesis. Quantification of target gene expression showed that VEGF-R2 signaling output was significantly reduced by PDGF-BB co-expression compared to VEGF alone. In conclusion, PDGF-BB co-expression prevents VEGF-induced aberrant angiogenesis by modulating VEGF-R2 signaling and endothelial proliferation, thereby limiting the degree of circumferential enlargement and enabling efficient completion of vascular splitting into normal capillary networks despite high VEGF doses.

Governance of hospital alliances: lessons learnt from 6 hospital and non-hospital cases.

Hospital alliances can be a solution for the coordination that is required between providers as a result of the increasing fragmentation in health-care services. 3 hospital and 3 non-hospital alliances were studied with a qualitative case study methodology, to find common, or specific, practices with regards to the management of alliances. Striking similarities prevailed, hospital alliances are not unique and common principles of coordination were identified that appear pivotal for successful alliance management.

Perspective on the evolution of cell-based bone tissue engineering strategies.

Despite the compelling clinical needs in enhancing bone regeneration and the potential offered by the field of tissue engineering, the adoption of cell-based bone graft substitutes in clinical practice is limited to date. In fact, no study has yet convincingly demonstrated reproducible clinical performance of tissue-engineered implants and at least equivalent cost-effectiveness compared to the current treatment standards. Here, we propose and discuss how tissue engineering strategies could be evolved towards more efficient solutions, depicting three different experimental paradigms: (i) bioreactor-based production; (ii) intraoperative manufacturing, and (iii) developmental engineering. The described approaches reflect the need to streamline graft manufacturing processes while maintaining the potency of osteoprogenitors and recapitulating the sequence of biological steps occurring during bone development, including vascularization. The need to combine the assessment of efficacy of the different strategies with the understanding of their mechanisms of action in the target regenerative processes is highlighted. This will be crucial to identify the necessary and sufficient set of signals that need to be delivered at the injury or defect site and should thus form the basis to define release criteria for reproducibly effective engineered bone graft substitutes.

Intra-individual comparison of human ankle and knee chondrocytes in vitro: relevance for talar cartilage repair.

OBJECTIVE: As compared to knee chondrocytes (KC), talar chondrocytes (TC) have superior synthetic activity and increased resistance to catabolic stimuli. We investigated whether these properties are maintained after TC are isolated and expanded in vitro. METHODS: Human TC and KC from 10 cadavers were expanded in monolayer and then cultured in pellets for 3 and 14 days or in hyaluronan meshes (Hyaff-11) for 14 and 28 days. Resulting tissues were assessed biochemically, histologically, biomechanically and by real-time reverse transcriptase-polymerase chain reaction (RT-PCR). The proteoglycan and collagen synthesis rates in the pellets were also measured following exposure to Interleukin-1 beta (IL-1 beta). RESULTS: After 14 days of pellet culture, TC and KC expressed similar levels of type I collagen (CI) and type II collagen (CII) mRNA and the resulting tissues contained comparable amounts of glycosaminoglycans (GAG) and displayed similar staining intensities for CII. Also proteoglycan and collagen synthesis were similar in TC and KC pellets, and dropped to a comparable extent in response to IL-1 beta. Following 14 days of culture in Hyaff-11, TC and KC generated tissues with similar amounts of GAG and CI and CII. After 28 days, KC deposited significantly larger fractions of GAG and CII than TC, although the trend was not reflected in the measured biomechanical properties. CONCLUSION: After isolation from their original matrices and culture expansion, TC and KC displayed similar biosynthetic activities, even in the presence of catabolic stimuli. These in vitro data suggest a possible equivalence of TC and KC as autologous cell sources for the repair of talar cartilage lesions.

A novel implantation technique for engineered osteo-chondral grafts.

We present a novel method to support precise insertion of engineered osteochondral grafts by pulling from the bone layer, thereby minimizing iatrogenic damage associated with direct manipulation of the cartilage layer. Grafts were generated by culturing human expanded chondrocytes on Hyaff-11 meshes, sutured to Tutobone spongiosa cylinders. Through the bone layer, shaped to imitate the surface-contours of the talar dome, two sutures were applied: the first for anterograde implantation, to pull the graft into the defect, and the second for retrograde correction, in case of a too deep insertion. All grafts could be correctly positioned into osteochondral lesions created in cadaveric ankle joints with good fit to the surrounding cartilage. Implants withstood short-term dynamic stability tests applied to the ankle joint, without delamination or macroscopic damage. The developed technique, by allowing precise and stable positioning of osteochondral grafts without iatrogenic cartilage damage, is essential for the implantation of engineered tissues, where the cartilage layer is not fully mechanically developed, and could be considered also for conventional autologous osteochondral transplantation.

Cartilage tissue engineering using pre-aggregated human articular chondrocytes.

In this study, we first aimed at determining whether human articular chondrocytes (HAC) proliferate in aggregates in the presence of strong chondrocyte mitogens. We then investigated if the aggregated cells have an enhanced chondrogenic capacity as compared to cells cultured in monolayer. HAC from four donors were cultured in tissue culture dishes either untreated or coated with 1% agarose in the presence of TGFbeta-1, FGF-2 and PDGF-BB. Proliferation and stage of differentiation were assessed by measuring respectively DNA contents and type II collagen mRNA. Expanded cells were induced to differentiate in pellets or in Hyaff-11 meshes and the formed tissues were analysed biochemically for glycosaminoglycans (GAG) and DNA, and histologically by Safranin O staining. The amount of DNA in aggregate cultures increased significantly from day 2 to day 6 (by 3.2-fold), but did not further increase with additional culture time. Expression of type II collagen mRNA was about two orders of magnitude higher in aggregated HAC as compared to monolayer expanded cells. Pellets generated by aggregated HAC were generally more intensely stained for GAG than those generated by monolayer-expanded cells. Scaffolds seeded with aggregates accumulated more GAG (1.3-fold) than scaffolds seeded with monolayer expanded HAC. In conclusion, this study showed that HAC culture in aggregates does not support a relevant degree of expansion. However, aggregation of expanded HAC prior to loading into a porous scaffold enhances the quality of the resulting tissues and could thus be introduced as an intermediate culture phase in the manufacture of engineered cartilage grafts.

Cartilage tissue engineering by expanded goat articular chondrocytes.

In this study we investigated whether expanded goat chondrocytes have the capacity to generate cartilaginous tissues with biochemical and biomechanical properties improving with time in culture. Goat chondrocytes were expanded in monolayer with or without combinations of FGF-2, TGF-beta1, and PDGFbb, and the postexpansion chondrogenic capacity assessed in pellet cultures. Expanded chondrocytes were also cultured for up to 6 weeks in HYAFF-M nonwoven meshes or Polyactive foams, and the resulting cartilaginous tissues were assessed histologically, biochemically, and biomechanically. Supplementation of the expansion medium with FGF-2 increased the proliferation rate of goat chondrocytes and enhanced their postexpansion chondrogenic capacity. FGF-2-expanded chondrocytes seeded in HYAFF-M or Polyactive scaffolds formed cartilaginous tissues with wet weight, glycosaminoglycan, and collagen content, increasing from 2 days to 6 weeks culture (up to respectively 2-, 8-, and 41-fold). Equilibrium and dynamic stiffness measured in HYAFF M-based constructs also increased with time, up to, respectively, 1.3- and 16-fold. This study demonstrates the feasibility to engineer goat cartilaginous tissues at different stages of development by varying culture time, and thus opens the possibility to test the effect of maturation stage of engineered cartilage on the outcome of cartilage repair in orthotopic goat models.

Bi-zonal cartilaginous tissues engineered in a rotary cell culture system.

In this study, we aimed at validating a rotary cell culture system (RCCS) bioreactor with medium recirculation and external oxygenation, for cartilage tissue engineering. Primary bovine and human culture-expanded chondrocytes were seeded into non-woven meshes of esterified hyaluronan (HYAFF-11), and the resulting constructs were cultured statically or in the RCCS, in the presence of insulin and TGFbeta3, for up to 4 weeks. Culture in the RCCS did not induce significant differences in the contents of glycosaminoglycans (GAG) and collagen deposited, but markedly affected their distribution. In contrast to statically grown tissues, engineered cartilage cultured in the RCCS had a bi-zonal structure, consisting of an outgrowing fibrous capsule deficient in GAG and rich in collagen, and an inner region more positively stained for GAG. Structurally, trends were similar using primary bovine or expanded human chondrocytes, although the human cells deposited inferior amounts of matrix. The use of the presented RCCS, in conjunction with the described medium composition, has the potential to generate bi-zonal tissues with features qualitatively resembling the native meniscus.

Identification and intracellular location of MAGE-3 gene product.

The human MAGE-3 gene encodes a melanoma antigenic epitope recognized by specific cytotoxic T lymphocytes, but its gene product has not been identified thus far. We produced a recombinant MAGE-3 gene product by expression cloning of the entire reading frame in the context of a fusion protein characterized by a 10-histidine tail, allowing purification by metal chelation on a nickel Sepharose column. The semipurified product was used to generate MAGE-3-specific monoclonal antibodies. One reagent could identify by immunoblotting the native MAGE-3 gene product as a M(r) 48,000 protein in lysates of cell lines showing evidence of MAGE-3 gene expression. No apparent cross-reactivity with recombinant or native MAGE-1 gene product was observed. Immunohistochemistry shows that, closely resembling the MAGE-1 gene product, MAGE-3 is a cytoplasmic protein.

Enhanced generation of cytotoxic T lymphocytes using recombinant vaccinia virus expressing human tumor-associated antigens and B7 costimulatory molecules.

In this work, we addressed the possibility to enhance the "in vitro" generation of CTLs recognizing tumor-associated antigens (TAAs) by using an inactivated recombinant vaccinia virus encoding B7.1 and B7.2 costimulatory molecules (rVV-B7.1/2). Antigen presenting cells (APCs) infected by rVV-B7.1/2 and pulsed with MART-1/Melan-A27-35 HLA-A2.1-restricted peptide induced significantly higher specific cytotoxic activity than peptide-loaded APCs infected by wild-type VV, both in VV-sensitized and naive donors. When APCs were infected with a rVV encoding both MART-1/Melan-A27-35 and B7-1/2 (rVV-B7.1/2-M), a significantly more effective CTL generation was observed as compared with cultures stimulated by APCs infected with a rVV encoding the TAA epitope only (rVV-M). These enhancing effects were detectable irrespective of a previous VV-specific sensitization. Most importantly, fibroblasts, devoid of antigen-presenting capacity upon peptide pulsing or infection with rVV-M, could be turned into effective APCs after infection by rVV encoding TAA epitopes and costimulatory molecules. In these experiments, by using separate recombinant viral constructs, we observed a predominant role of B7-1 as compared with B7-2 in the induction of TAA-specific CTLs. Taken together, our data indicate that replication-incompetent rVV encoding TAA epitopes and costimulatory molecules are able to induce highly effective generation of tumor-specific CTLs. Therefore, these vectors could represent valuable clinical tools for immunotherapy of melanoma patients.

Structural characterization and reliable biomechanical assessment of integrative cartilage repair.

Structural and functional characterization of integrative cartilage repair in controlled model systems can play a key role in the development of innovative strategies to improve the long-term outcome of many cartilage repair procedures. In this work, we first developed a method to reproducibly generate geometrically defined disk/ring cartilage composites and to remove outgrown fibrous layers which can encapsulate cartilaginous tissues during culture. We then used the model system to test the hypothesis that such fibrous layers lead to an overestimation of biomechanical parameters of integration at the disk/ring interface. Transmission electron microscopy images of the composites after 6 weeks of culture indicated that collagen fibrils in the fibrous tissue layer were well integrated into the collagen network of the cartilage disk and ring, whereas molecular bridging between opposing disk/ring cartilage surfaces was less pronounced and restricted to regions with narrow interfacial regions (< 2 microm). Stress-strain profiles generated from mechanical push-out tests for composites with the layers removed displayed a single and distinct peak, whereas profiles for composites with the layers left intact consisted of multiple superimposed peaks. As compared to composites with removed layers, composites with intact layers had significantly higher adhesive strengths (161+/-9 vs. 71+/-11 kPa) and adhesion energies (15.0+/-0.7 vs. 2.7+/-0.4 mJ/mm2). By combining structural and functional analyses, we demonstrated that the outgrowing tissue formed during in vitro culture of cartilaginous specimens should be eliminated in order to reliably quantify biomechanical parameters related to integrative cartilage repair.

Phase I/II clinical trial of a nonreplicative vaccinia virus expressing multiple HLA-A0201-restricted tumor-associated epitopes and costimulatory molecules in metastatic melanoma patients.

We performed a phase I/II clinical trial in metastatic melanoma patients with an ultraviolet (UV)-inactivated nonreplicating recombinant vaccinia virus enabling the expression, from a single construct, of endoplasmic reticulum-targeted HLA-A0201-restricted Melan-A/MART-1(27-35), gp100(280-288), and tyrosinase(1-9) epitopes, together with CD80 and CD86 costimulatory proteins. Corresponding soluble peptides were used to boost responses and granulocyte-macrophage colony-stimulating factor was used as systemic adjuvant. Safety and immunogenicity, as monitored with in vitro-restimulated peripheral blood mononuclear cells by cytotoxic T lymphocyte precursor (CTLp) frequency analysis and tetramer staining, were specifically addressed. Of 20 patients entering the protocol, 2 had to withdraw because of rapidly progressing disease. Immune responses were evaluated in 18 patients (stage III, n = 5; stage IV, n = 13) and increases in specific CTLp frequencies were observed in 15. In 16 patients responsiveness against all 3 antigens could be analyzed: 7 (43%), including all stage III cases, showed evidence of induction of CTLs specific for the three epitopes, and 2 (12%) and 4 (25%), respectively, showed reactivity against two or one tumor-associated antigen. In three stage IV patients no specific CTL reactivity could be induced. Increases in CTLp frequency were detected mostly after viral vaccine injections. However, in a majority of patients final CTLp levels were comparable to initial levels. Tetramer characterization of Melan-A/MART-1(27-35)-specific CTLs during the protocol also suggested preferential expansion after recombinant virus administration. Vector-specific humoral responses, frequently undetectable in stage IV patients, did not appear to prevent tumor-associated antigen-specific CTL induction. Aside from a single occurrence of transient grade 3 leukopenia, no major clinical toxicity was reported. Seventeen of 18 patients completed the 3-month trial (one patient died before the last delayed-type hypersensitivity test). Three displayed regression of individual metastases, seven had stable disease, and progressive disease was observed in seven patients. This is the first report on the administration of a UV-inactivated recombinant vaccinia virus coexpressing five transgenes in cancer patients. The results described here, in terms of safety and immunogenicity, support the use of this reagent in active specific immunotherapy.

Immunogenicity of nonreplicating recombinant vaccinia expressing HLA-A201 targeted or complete MART-1/Melan-A antigen.

The effect on immunogenicity of different tumor T cell epitope formulations was evaluated in vitro using nonreplicating recombinant vaccinia vector expressing two forms of the melanoma-associated MART-1/Melan-A antigen. The first recombinant virus expressed a minigene encoding a fusion product between an endoplasmic reticulum (ER)-targeting signal and the HLA-A201 binding 27-35 peptide. The second viral construct encoded the complete MART-1/Melan-A protein. The capacity of HLA-A201 cells infected with either viral construct to generate and to stimulate MART-1/Melan-A 27-35 specific cytotoxic T-lymphocytes (CTL), was comparatively characterized. The results obtained here with a tumor antigen confirmed the capacity of vaccinia virus-encoded ER-minigene to generate a very strong antigenic signal. In cytotoxicity assays, recognition of target cells infected with high amounts of both recombinant viruses with activated specific CTL clones, resulted in similar lytic activity. With regard to calcium mobilization, TCR down-regulation, IFN-gamma release, and T cell proliferation assays, the targeted epitope elicited 10- to 1000-fold stronger responses. Remarkably, the immunogenic difference between the two formulations, in their respective capacity to generate CTL from naive HLA-A2 peripheral blood mononuclear cells in vitro as measured by tetramer detection, was lower (2- to 3-fold). Recombinant vectors expressing complete antigens have demonstrated their capacity to generate specific responses and such vaccines might take advantage of a broader potential of presentation. However, as demonstrated here for the HLA-A201-restricted MART-1/Melan-A immunodominant epitope, nonreplicative vaccinia virus expressing ER-targeted minigenes appear to represent a significantly more immunogenic epitope vaccine formulation.

Effects of different culture protocols on the expression of discrete T-cell receptor variable regions in human tumour infiltrating lymphocytes.

Therapeutic effects of tumour infiltrating lymphocytes (TIL) rely on T-cell receptor (TCR) engagement. In this work, the expression of five TCR alpha/beta variable (V) domains was quantitatively analysed by means of a panel of monoclonal antibodies (Mab) recognising gene products from TCR V alpha 2, V beta 5, V beta 6, V beta 8 and V beta 12 families in freshly isolated TIL and in autologous peripheral blood mononuclear cells (PBMC) from patients with neoplasms. In 3 out of 6 cases, differences in the expression of V beta 5, V beta 6, V beta 8 or V beta 12 could be detected. TIL populations were expanded by using recombinant human interleukin-2 (rhIL-2) alone or in addition to solid phase bound anti-CD3 Mab. Cultured TIL showed similar CD4/CD8 ratios and cytotoxic activity against autologous neoplastic target cells, regardless of the activation protocol. In 4 patients, the expression of TCR alpha/beta V gene products, as compared with TIL from freshly excised tumours, was found to be modified in cultured TIL, especially in cell populations activated with rhIL-2 only. These results indicate that TCR V gene usage in TIL may quantitatively differ from that in PBMC. TIL culture protocols using rhIL-2 alone or in combination with solid phase bound anti-CD3 may result in differential expression of discrete TCR V families.

Cytokine gene expression in primary brain tumours, metastases and meningiomas suggests specific transcription patterns.

To obtain an insight into the network of cytokine gene transcription in the brain tumour microenvironment, we investigated the expression of genes encoding for interleukin (IL)-1 alpha, IL-1 beta, IL-2, IL-4, IL-5, IL-6, IL-10, interferon (IFN)-gamma, granulocyte-macrophage colony-stimulating factor, tumour necrosis factor (TNF)-alpha and transforming growth factor (TGF)-beta 1, -beta 2 and -beta 3 in freshly excised brain tumour samples and autologous peripheral blood mononuclear cells. Tissue specimens from 15 primary brain tumours, three brain metastases, five meningiomas, autologous peripheral blood mononuclear cells (PBMC) and three brain tumour cell lines were tested by reverse polymerase chain reaction. Despite the presence of T-lymphocytes, cytokine gene transcripts typically detectable upon T cell receptor triggering could not be observed in central nervous system tumours of diverse histology. In primary brain neoplasms, transcription of genes encoding for the inhibitory cytokines TGF-beta and IL-10 was detectable in more than 50% of samples. IL-6 transcripts could only be detected in malignant gliomas. In brain metastases, virtually no cytokine gene transcripts could be observed. Surprisingly, TGF-beta transcripts were also detected in all meningiomas. Thus, transcription of genes encoding for inhibitory factors appears to prevail in primary brain neoplasms.

Functional and morphological outcome of knee joint transplantation in dogs depends on control of rejection.

BACKGROUND: The reconstruction of massive osteochondral defects extending to weight-bearing joints remains a surgical challenge. Total knee joint transplantation has been performed experimentally, but these studies lacked prospective evaluation of functional outcome, graft vascularization, and graft viability. METHODS: Replantation and transplantation of vascularized knee joints was performed in dogs (n=4 per group), comparing functional and morphological results during a 6-month follow-up. RESULTS: All replant recipients and three transplant recipients survived the 6-month follow-up period. At this time, duplex sonography and angiography revealed patent anastomoses in all animals. Increases in volumetric flow rates and vascular collateralization were observed in allografts, as compared with replanted joints (100+/-16 ml/min vs. 31+/-15 ml/min at 6 months after transplantation). Bone fusion at the graft-host interface was verified by fluorography in all animals at 3 months after transplantation. Six months after transplantation, microradiographies and computerized tomographies revealed spongialization of the cortical bone and filling of the medullary space by trabecular bone in transplanted joints. Such alterations were not detectable in replanted joints. Chondrocyte viability exceeded 80% in all but one transplanted joint. Lymphocyte infiltration of synovia and arterial walls was detected in all transplanted joints, suggesting the presence of chronic rejection. Weight-bearing capacity recovered in all replanted animals (weight-bearing index before transplantation: 0.499+/-0.080; 6 months after transplantation: 0.38+/-0.16) but only in two of four transplanted animals (weight-bearing index 6 months after transplantation: 0.37, 0.28, and 0.00). CONCLUSIONS: These data demonstrate the potential of joint grafting and the critical dependence of allotransplantation on the control of rejection.

Expression of HLA-DR in granulocytes of polytraumatized patients treated with recombinant human granulocyte macrophage-colony-stimulating factor.

MHC class II determinants are the restriction elements involved in antigen-specific activation of helper T lymphocytes and interaction with CD4 molecules. They are typically expressed on a limited number of cell types, mostly endowed with antigen-presenting capacity. Recently, expression of HLA-DR has been detected on granulocytes stimulated "in vitro" with GM-CSF. However, no evidence of "in vivo" expression in humans has been presented so far. We report here that class II determinant expression is detectable in vivo on peripheral blood granulocytes of polytraumatized patients upon intravenous administration of rhGM-CSF. Expression of these molecules appears to be an early effect of rhGM-CSF treatment, independent from endotoxemia or endogenous production of IL-6 or TNF-alpha, and rapidly declining upon discontinuation of therapy. Thus, this treatment might increase the number of cells potentially capable of presenting class-II-restricted antigens in these patients.

Modulation of dendritic cell phenotype and mobility by tumor cells in vitro.

To gain new insights into the functional interaction between DC and neoplastic cells, we have analyzed the effects of melanoma and colorectal cancer lines on the chemotaxis and the phenotype of monocyte-derived DC in vitro. Both types of tumor cells displayed effective chemoattractive capacity towards immature, but not mature DC. Furthermore, conditioned medium of discrete melanoma lines induced upregulation of CD80, CD86, MHC class I, and MHC class II molecules on immature DC. However, de novo expression of E-cadherin and strong upregulation of CD15 could also be detected in the absence of CD83 expression. Melanoma-conditioned DC exhibited an increased adhesion capacity to a melanoma cell line in vitro and did not migrate in response to SLC chemokine. Tumor-infiltrating CD15(+) cells displaying DC morphology could also be detected by immunohistochemistry in the original tumor specimens from which discrete melanoma cell lines under investigation were derived. Colorectal cancer cell lines, although able to chemoattract immature DC, were apparently unable to modulate their phenotype. Altogether our results suggest that tumor cells can attract immature DC in vitro and, eventually, modulate their phenotype. As a result, DC mobility could be severely impaired.

Expression of MAGE tumour-associated antigens is inversely correlated with tumour differentiation in invasive ductal breast cancers: an immunohistochemical study.

MAGE (Melanoma antigen E) family gene products encompass tumour-associated antigens (TAAs) recognised by human leukocyte antigen (HLA)-restricted specific T-cells. Agents inducing DNA demethylation, an event typically detectable in cellular de-differentiation processes, were shown to induce the expression of MAGE genes. By using a monoclonal antibody specific for MAGE family gene products, we have studied the expression of these TAAs in a group of 144 patients with invasive ductal breast cancers. Immunohistochemical data were correlated with tumour differentiation, lymphatic vessel invasion, oestrogen receptor expression, intratumoural necrosis, lymphocytic infiltration, perineural invasion, tumour microcalcifications and axillary lymph node metastases. MAGE immunoreactivity was undetectable in non-neoplastic cells. In poorly differentiated cancers positive staining was observed in 30/63 cases (47.6%) as compared with 13/51 (25.4%) and 5/30 (16.6%) in moderately and well-differentiated tumours, respectively (P<0.05). In addition, MAGE immunoreactivity was significantly correlated with lymphatic vessel invasion and intratumoural necrosis. Moreover, a significant inverse relationship with oestrogen receptor expression was also observed. However, no significant correlation could be established between MAGE immunoreactivity and defined phenotypic characteristics of tumour infiltrating lymphocytes, including expression of CD3, CD4, CD8, CD20 or granzyme B. Thus, expression of MAGE family gene products in invasive ductal breast cancers appears to be associated with poorly differentiated histological phenotypes. These data support the concept of specific immunotherapy in highly aggressive forms of breast neoplasms. Furthermore, they suggest that MAGE immunoreactivity could represent a tumour marker of potential prognostic relevance.

Immunohistochemical detection of MAGE tumor-associated antigens in esophageal squamous cell carcinoma.

Genes of the MAGE family encode tumor-specific antigens recognized by cytotoxic T-lymphocytes in a variety of neoplasms. We investigated the protein expression of these antigens as related to the gene expression, in esophageal squamous cell carcinoma by using monoclonal antibodies recognizing MAGE gene products. Esophageal squamous cell carcinomas were found to express both MAGE-1 (4 out of 15 samples) and MAGE-3 (7 out of 15 samples) genes, by RT-PCR. Immunoblotting revealed MAGE-1 and MAGE-3 gene products in 2 and 6 out of 15 samples, respectively. Immunohistochemistry performed on 12 samples showed MAGE-1 protein expression, limited to single tumor cells, in 2 cases. MAGE-3 gene product was detectable in 7 cases: in 5 of them over 50% of neoplastic cells were positive. Considering the high percentages of tumor cells expressing MAGE-3 antigen, the use of epitope-based vaccines could be envisaged in patients displaying appropriate HLA-class I phenotype.