Release of an inhibitor of angiogenesis upon induction of wild type p53 expression in glioblastoma cells.

The earliest genetic alteration in human astrocytoma progression is mutation of the p53 tumour suppressor gene, while one of the earliest phenotypic changes is the stimulation of neovascularization. Here, we tested the role of p53 in the angiogenic process by introducing a tetracycline-regulated wild type p53 gene into null glioblastoma cells. The parental cells expressed strong angiogenic activity while upon induction of wild type, but not mutant, p53 expression, the cells secreted a factor able to neutralize the angiogenicity of the factors produced by the parental cells as well as of basic fibroblast growth factor.

Upregulation of interleukin 8 by oxygen-deprived cells in glioblastoma suggests a role in leukocyte activation, chemotaxis, and angiogenesis.

Leukocyte infiltration and necrosis are two biological phenomena associated with the development of neovascularization during the malignant progression of human astrocytoma. Here, we demonstrate expression of interleukin (IL)-8, a cytokine with chemotactic and angiogenic properties, and of IL-8-binding receptors in astrocytoma. IL-8 expression is first observed in low grade astrocytoma in perivascular tumor areas expressing inflammatory cytokines. In glioblastoma, it further localizes to oxygen-deprived cells surrounding necrosis. Hypoxic/anoxic insults on glioblastoma cells in vitro using anaerobic chamber systems or within spheroids developing central necrosis induced an increase in IL-8 messenger RNA (mRNA) and protein expression. mRNA for IL-8-binding chemokine receptors CXCR1, CXCR2, and the Duffy antigen receptor for chemokines (DARC) were found in all astrocytoma grades by reverse transcription/PCR analysis. In situ hybridization and immunohistochemistry localized DARC expression on normal brain and tumor microvascular cells and CXCR1 and CXCR2 expression to infiltrating leukocytes. These results support a model where IL-8 expression is initiated early in astrocytoma development through induction by inflammatory stimuli and later in tumor progression increases due to reduced microenvironmental oxygen pressure. Augmented IL-8 would directly and/or indirectly promote angiogenesis by binding to DARC and by inducing leukocyte infiltration and activation by binding to CXCR1 and CXCR2.

Thrombospondin-1 is downregulated by anoxia and suppresses tumorigenicity of human glioblastoma cells.

Angiogenesis, the sprouting of new capillaries from preexisting blood vessels, results from a disruption of the balance between stimulatory and inhibitory factors. Here, we show that anoxia reduces expression of thrombospondin-1 (TSP-1), a natural inhibitor of angiogenesis, in glioblastoma cells. This suggests that reduced oxygen tension can promote angiogenesis not only by stimulating the production of inducers, such as vascular endothelial growth factor, but also by reducing the production of inhibitors. This downregulation may significantly contribute to glioblastoma development, since we show that an increase in TSP-1 expression is sufficient to strongly suppress glioblastoma cell tumorigenicity in vivo.

Proteomic identification of the wt-p53-regulated tumor cell secretome.

Tumor-stroma interactions play a major role in tumor development, maintenance and progression. Yet little is known on how the genetic alterations that underlie cell transformation elicit cell extrinsic changes modulating heterotypic cell interactions. We hypothesized that these events involve a modification in the complement of secreted proteins by the cell, acting as mediators of intercellular communication. To test this hypothesis, we examined the role of wt-p53, a major tumor suppressor, on the tumor microenvironment through its regulation of secreted factors. Using a combination of 2-DE and cICAT proteomic techniques, we found a total of 111 secreted proteins, 39 of which showed enhanced and 21 inhibited secretion in response to wt-p53 expression. The majority of these were not direct targets of p53 transcription factor activity, suggesting a novel role for wt-p53 in the control of intracellular protein trafficking and/or secreted protein stability. Evidence for p53-controlled post-translational modifications on nine secreted proteins was also found. These findings will enhance our understanding of wt-p53 modulated interactions of the tumor with its environment.

Fas ligand expression by astrocytoma in vivo: maintaining immune privilege in the brain?

Astrocytomas are among the most common brain tumors that are usually fatal in their malignant form. They appear to progress without significant impedance from the immune system, despite the presence of intratumoral T cell infiltration. To date, this has been thought to be the result of T cell immunosuppression induced by astrocytoma-derived cytokines. Here, we propose that cell contact-mediated events also play a role, since we demonstrate the in vivo expression of Fas ligand (FasL/CD95L) by human astrocytoma and the efficient killing of Fas-bearing cells by astrocytoma lines in vitro and by tumor cells ex vivo. Functional FasL is expressed by human, mouse, and rat astrocytoma and hence may be a general feature of this nonlymphoid tumor. In the brain, astrocytoma cells can potentially deliver a death signal to Fas+ cells which include infiltrating leukocytes and, paradoxically, astrocytoma cells themselves. The expression of FasL by astrocytoma cells may extend the processes that are postulated to occur in normal brain to maintain immune privilege, since we also show FasL expression by neurons. Overall, our findings suggest that FasL-induced apoptosis by astrocytoma cells may play a significant role in both immunosuppression and the regulation of tumor growth within the central nervous system.

Identification of the putative brain tumor antigen BF7/GE2 as the (de)toxifying enzyme microsomal epoxide hydrolase.

Malignant gliomas are the main cause of death from primary brain tumors. Despite surgery, radiation, and chemotherapy, patients have a median survival of less than a few years; therefore, it is clearly imperative to investigate new ways of treatment. The development of new therapeutic strategies for brain tumors is dependent on a better understanding of the differences between normal and tumoral brain cells. Our group had described previously a Mr 48,000 antigen defined by reactivity with two monoclonal antibodies (GE2 and BF7) obtained by immunization of mice with human glioblastoma cells. Here, we describe the identification of the GE2/BF7 antigen as microsomal epoxide hydrolase (mEH), a drug-metabolizing enzyme that is involved both in toxification and detoxification of carcinogens. We initially used immunoaffinity purification using GE2 and BF7 and analyzed the purified proteins by microsequencing. Edman degradation identified 15 amino acids of the NH2-terminal sequence that were 100% identical to mEH. To further confirm the identity of the BF7/GE2 antigen as mEH, we showed that the protein immunopurified with GE2 and BF7 was recognized by an anti-mEH antibody and that in vitro and in vivo synthesized human mEH is recognized by BF7 and GE2 antibodies. Furthermore, anti-mEH antibody recognizes an antigen expressed both in gliomas and reactive astrocytes, as do BF7 and GE2. Finally, we demonstrate that in contrast to what has been reported in rat embryo fibroblasts, p53 does not regulate mEH mRNA expression in glioma cells.

The PTEN lipid phosphatase domain is not required to inhibit invasion of glioma cells.

The tumor suppressor PTEN negatively controls the phosphoinositide 3-kinase pathway for cell survival by dephosphorylating the phospholipid substrates phosphatidylinositol 3,4-bisphosphate and phosphatidylinositol 3,4,5-trisphosphate. PTEN has been proposed to dephosphorylate focal adhesion kinase and is implicated in the regulation of cell spreading and motility. We analyzed the role of PTEN in invasion using the two highly infiltrative glioma cell lines U87MG (which lacks functional PTEN) and LN229 (wild-type PTEN). After constitutive overexpression of wild-type and phosphatase-deficient (C124S) PTEN, we found significant inhibition of invasion (50-70%) independent of the PTEN status of the cell and of the catalytic core domain of PTEN. Although wild-type but not mutant (C124S) PTEN decreased PKB/Akt phosphorylation and induced a stellate morphology in U87MG cells, an accompanying reduction of focal adhesion kinase phosphorylation was not seen. We conclude that phosphatase-independent domains of PTEN markedly reduced the invasive potential of glioma cells, defining a structural role for PTEN that regulates cell motility distinct of the PKB/Akt pathway.

Variant CD44 adhesion molecules are expressed in human brain metastases but not in glioblastomas.

Although human glioblastomas are highly invasive tumors intracerebrally, only rarely do they metastasize outside the central nervous system. In contrast, the brain is a major target for metastatic spread of many systemic tumors. Recently, it was demonstrated that expression of splice variants of CD44 (CD44v), but not standard CD44 (CD44s), was sufficient to confer metastatic potential to low- or nonmetastatic rat tumor cells. Because CD44 is expressed in brain tumors, we examined whether differential expression of CD44 isoforms was correlated with the metastatic behavior of these tumors. We compared CD44s and CD44v expression in 17 human glioblastomas, 18 glioma cell lines, and metastases of 15 other tumors to the brain by reverse transcription/polymerase chain reaction, Northern blotting, and immunocytochemistry. These experiments showed that 0 of 17 glioblastomas and 0 of 18 glioma cell lines expressed CD44v as compared to 12 of 15 brain metastases. These data show a correlation between CD44v expression and the metastatic ability of the tumors analyzed (P < 0.01). This suggests (a) that the biological significance of the lack of CD44v expression in human glioblastomas warrants further examination with regard to their inability to metastasize extraneurally and (b) that CD44v expression may play a role in the intracerebral spread of about 80% [corrected] of the brain metastases. Therefore, CD44v expression should be further considered as a potential marker for differential diagnosis and prognosis of patients with brain metastases.

Induction and intracellular regulation of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) mediated apotosis in human malignant glioma cells.

Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) preferentially triggers apoptosis in tumor cells versus normal cells, thus providing a therapeutic potential. In this study, we examined a large panel of human malignant glioma cell lines and primary cultures of normal human astrocytes for their sensitivity to TRAIL. Of 13 glioma cell lines, 3 were sensitive (80-100% death), 4 were partially resistant (30-79% death), and 6 were resistant (< 30% death). Normal astrocytes were also resistant. TRAIL-induced cell death was characterized by activation of caspase-8 and -3, poly(ADP-ribose) polymerase cleavage, and DNA fragmentation. Decoy receptor (DcR1 and DcR2) expression was limited in the glioma cell lines and did not correlate with TRAIL sensitivity. Both sensitive and resistant cell lines expressed TRAIL death receptor (DR5), adapter protein Fas-associated death domain (FADD), and caspase-8; but resistant cell lines expressed 2-fold higher levels of the apoptosis inhibitor phosphoprotein enriched in diabetes/phosphoprotein enriched in astrocytes-15 kDa (PED/PEA-15). In contrast, cellular FADD-like IL-1beta-converting enzyme-like inhibitory protein (cFLIP) expression was similar in sensitive and resistant cells. Transfection of sense PED/PEA-15 cDNA in sensitive cells resulted in cell resistance, whereas transfection of antisense in resistant cells rendered them sensitive. Inhibition of protein kinase C (PKC) activity restored TRAIL sensitivity in resistant cells, suggesting that PED/ PEA-15 function might be dependent on PKC-mediated phosphorylation. In summary, TRAIL induces apoptosis in > 50% of glioma cell lines, and this killing occurs through activation of the DR pathway. This caspase-8-induced apoptotic cascade is regulated by intracellular PED/PEA-15, but not by cFLIP or decoy receptors. This pathway may be exploitable for glioma and possibly for other cancer therapies.

Analysis of the p53 gene and its expression in human glioblastoma cells.

Chromosome 17p has been shown to be an early and frequent target for loss of heterozygosity through mitotic recombination in astrocytomas. These losses are frequently accompanied by point mutations in the p53 gene of the remaining allele, resulting in loss of wild type p53 function. However, a fraction of astrocytomas retain constitutional heterozygosity and do not have p53 mutations; some of these lose wild type p53 activity through binding to the protein product of amplified mdm2 genes. To test whether loss of wild type p53 biological function is a necessary step in astrocytoma progression we analyzed p53 expression and biological function in 13 glioma cell lines. All the cell lines expressed a 2.8-kilobase p53 transcript and showed various amounts of p53 protein by immunoprecipitation, except for cell line LN-Z308 which had only a small truncated p53 mRNA and no protein expression. To test whether the p53 expressed in these cell lines was functionally wild type or mutant we transfected them with a plasmid construct harboring a chloramphenicol acetyltransferase (CAT) reporter gene under the control of transcriptional elements that are induced by wild type but not mutant p53. Four lines were shown to retain wild type p53 function. Sequencing of the p53 gene in two of these cell lines confirmed the wild type genotype. These results show that inactivation of the p53 gene is not an obligatory step in glioblastoma genesis. This suggests either that two pathways (p53 inactivation dependent or independent) may lead to a tumor group classified histologically as glioblastoma or that in some cases p53 mutations are bypassed due to the presence of mutations in downstream effector genes.

Turcot's syndrome of glioma and polyposis occurs in the absence of germ line mutations of exons 5 to 9 of the p53 gene.

The term "Turcot's syndrome" has been used to describe approximatively 55 patients with an association of colonic polyposis and primary neuroepithelial tumors of the central nervous system. The p53 tumor suppressor gene is a possible candidate underlying the syndrome because (a) mutations in the p53 gene are ubiquitous in human cancer, including colon carcinoma and gliomas, and (b) somatic or germ line mutations of the p53 tumor suppressor gene cause the Li-Fraumeni syndrome, which is characterized by the association of breast and soft tissue tumors. We determined the DNA sequence of the conserved regions of the p53 gene (exons 5 to 9) in the tumor tissues and lymphocytes of two patients with glioma-polyposis and found that mutations did occur as independent tumor-specific alterations but did not involve the germ line of these patients, suggesting that p53 may play a role in progression but not initiation of the disease.

Genetic instability leads to loss of both p53 alleles in a human glioblastoma.

Little is known about the relationship between genetic recombination mechanisms and loss of tumour suppressor genes in solid tumours. Here, we demonstrate deletion and truncation of both p53 alleles in a primary human glioblastoma and a derived cell line as the combined result of a t(17;20) reciprocal translocation and a 1.1 Mbp genomic deletion on chromosome 17p, starting in intron 4 of the p53 gene and ending at the telomeric CA-repeat marker D17S960. These results (i) suggest that genetic instability can lead to loss of tumour suppressor gene function in solid cancers, (ii) provide mapping of one such recombination event at the nucleotide level, and (iii) establish the orientation of the p53 gene on chromosome 17 as: centromere 5'-3'-telomere.

Regulation of interleukin-8 expression by reduced oxygen pressure in human glioblastoma.

Oxygen deprivation is an important biological feature of tumor growth. We previously showed that in glioma, anoxia increases expression of IL-8, a chemokine and angiogenic factor. Here, we analysed for the first time the biochemical mechanisms inducing the IL-8 gene upon anoxia in glioma cells, and showed that they differ from those inducing the VEGF gene. Both genes are induced in biologically and genetically heterogenous glioblastoma cell lines (LN-229, LN-Z308, U87MG, T98G), whereas, in gliosarcoma cells (D247MG), only the VEGF gene is induced. The kinetics of IL-8 and VEGF mRNA inductions differ in these cells and reoxygenation experiments showed that the induction is due to the anoxic stress per se. Furthermore, in LN-229 and LN-Z308 cell lines actinomycin D, DRB and nuclear run-on experiments showed that anoxia stimulates increased transcription of both genes. Electromobility shift assays show increased protein binding to the AP-1 site on the IL-8 promoter following anoxia treatment. Finally, in situ hybridization on glioblastoma sections shows that the in vivo expression patterns of IL-8 and VEGF genes overlap, but are not identical. Since intratumoral augmentation of IL-8 and VEGF secretion, following microenvironmental decreases in oxygen pressure, may promote angiogenesis, further definition of these pathways is essential to appropriately target them for antitumoral therapy.

Quantitative real-time PCR does not show selective targeting of p14(ARF) but concomitant inactivation of both p16(INK4A) and p14(ARF) in 105 human primary gliomas.

In many human cancers, the INK4A locus is frequently mutated by homozygous deletions. By alternative splicing this locus encodes two non-related tumor suppressor genes, p16(INK4A) and p14(ARF) (p19(ARF) in mice), which regulate cell cycle and cell survival in the retinoblastoma protein (pRb) and p53 pathways, respectively. In mice, the role of p16(INK4A) as the critical tumor suppressor gene at the INK4A locus was challenged when it was found that p19(ARF) only knock-out mice developed tumors, including gliomas. We have analysed the genetic status of the INK4A locus in 105 primary gliomas using both microsatellite mapping (MSM) and quantitative real-time PCR (QRT-PCR). Comparison of the results of the two methods revealed agreement in 67% of the tumors examined. In discordant cases, fluorescence in situ hybridization (FISH) analysis was always found to support QRT-PCR classification. Direct assessment of p14(ARF) exon 1beta, p16(INK4A) exon 1alpha and exon 2 by QRT-PCR revealed 43 (41%) homozygous and eight (7%) hemizygous deletions at the INK4A locus. In 49 (47%) gliomas, both alleles were retained. In addition, QRT-PCR, but not MSM, detected hyperploidy in five (5%) tumors. Deletion of p14(ARF) was always associated with co-deletion of p16(INK4A) and increased in frequency upon progression from low to high grade gliomas. Shorter survival was associated with homozygous deletions of INK4A in the subgroup of glioblastoma patients older than 50 years of age (P=0.025, Anova test single factor, alpha=0.05).

New deletion in low-grade oligodendroglioma at the glioblastoma suppressor locus on chromosome 10q25-26.

Loss of heterozygosity on chromosome 10 is considered to be associated with the progression of glioblastomas. Two closely related regions have recently been proposed to contain the glioblastoma suppressor locus on chromosome 10q25-26; a 1 cM region between the polymorphic (CA)n-repeat markers D10S587 and D10S216, and an area of 5 cM between the markers D10S221 and D10S209. To confirm and further delineate this region, we analyzed 51 glioblastomas and 11 intermediate and low-grade gliomas for loss of heterozygosity on chromosome 10. 47/62 mostly malignant gliomas displayed complete loss of chromosome 10 and nine tumors were unaltered, whereas four glioblastomas and two low-grade oligodendrogliomas had partial loss on distal 10q. With these six tumors, we constructed a deletion map with increased marker density at 10q25-26 which shows two centromeric breakpoints at the markers D10S587 and D10S216, thus only confirming the distal, but not the proximal candidate glioblastoma suppressor locus. Two out of four low-grade oligodendrogliomas displayed partial deletions on 10q25-26. This suggests that deletion on chromosome 10 is not merely a late event in the progression of glioblastomas, but could play a role earlier in the development of gliomas.

p53 gene mutation and ink4a-arf deletion appear to be two mutually exclusive events in human glioblastoma.

P16 and P14ARF are two tumor suppressors encoded by the locus ink4a-arf which is frequently deleted in human tumors. Recent experiments performed with mouse embryonic fibroblasts have shown that P14ARF is an upstream regulator of the P53 pathway. This raises the question as to whether in human tumors the loss of p14arf and mutation of p53 are mutually exclusive events which segregate with genetic alterations at other loci. To examine this question we performed a multigenic analysis on 29 gliomas. We analysed p53 and p14arf in relation with five other genetic loci encoding the most frequently mutated genes in human gliomas: cdkn2a, mdm2, egfr, pten and the chromosomal regions 10q23.3 and 10q25-26. Our study shows for the first time that p53 mutations and p14arf deletions appear mutually exclusive in human glioblastoma, suggesting that they may be functionally redundant in glioma tumorigenesis. The P53 pathway is, therefore, disrupted in 81.8% of malignant gliomas (WHO grades III and IV), either by mutation of the p53 gene (31.8%) or by p14arf deletion (54.5%). These tumors further showed MDM2 overexpression (9.1%), egfr oncogene amplification/egfr overexpression (50%), pten mutations (27.3%) and loss of heterozygosity (LOH) at the chromosomal regions 10q23.3 (86.4%) and 10q25-26 (100%). These alterations did not segregate with p53 mutations or p14arf deletions, while p14arf and cdkn2a were always deleted.

Somatic deletion mapping on chromosome 10 and sequence analysis of PTEN/MMAC1 point to the 10q25-26 region as the primary target in low-grade and high-grade gliomas.

The 10q25-26 region between the dinucleotide markers D10S587 and D10S216 is deleted in glioblastomas and, as we have recently shown, in low-grade oligodendrogliomas. We further refined somatic mapping on 10q23-tel and simultaneously assessed the role of the candidate tumor suppressor gene PTEN/MMAC1 in glial neoplasms by sequence analysis of eight low-grade and 24 high-grade gliomas. These tumors were selected for partial or complete loss of chromosome 10 based on deletion mapping with increased microsatellite marker density at 10q23-tel. Three out of eight (38%) low-grade and 3/24 (13%) high-grade gliomas exclusively target 10q25-26. We did not find a tumor only targeting 10q23.3, and most tumors (23/32, 72%) showed large deletions on 10q including both regions. The sequence analysis of PTEN/MMAC1 revealed nucleotide alterations in 1/8 (12.5%) low-grade gliomas in a tumor with LOH at l0q21-qtel and in 5/21 (24%) high-grade gliomas displaying LOH that always included 10q23-26. Our refined mapping data point to the 10q25-26 region as the primary target on 10q, an area that also harbors the DMBT1 candidate tumor suppressor gene. The fact that we find hemizygous deletions at 10q25-qtel in low-grade astrocytomas and oligodendrogliomas - two histologically distinct entities of gliomas - suggests the existence of a putative suppressor gene involved early in glial tumorigenesis.

Cells with TP53 mutations in low grade astrocytic tumors evolve clonally to malignancy and are an unfavorable prognostic factor.

It is important to understand how low grade tumors recur and progress to malignant lesions since this dramatically shortens patient survival. Here, we evaluated the concept that malignant progression and poor prognosis of low grade astrocytic tumors are TP53 dependent through clonal expansion of mutated cells. TP53 status was established in primary and recurrent tumors from 36 patients with WHO grade II astrocytic tumors and two tumor types were found. Tumors from 14 patients (39%; type 1) had TP53 mutated cells, and 92% of these recurred with 57% progressing to malignancy. The evolution of TP53 mutated cells before and after progression was examined using a clonal analysis procedure in yeast. Malignant progression was accompanied by an increased percentage of mutant TP53 (red) yeast colonies resulting from monoclonal expansion of cells with mutated TP53. The presence of TP53 mutations in WHO grade II astrocytic tumors was associated with malignant progression (P=0. 034, chi2 test) and shorter progression-free survival (PFS; 47.6+/-9. 6 months for TP53-mutated tumors vs 67.8+/-8.2 months for TP53-wild type tumors, P<0.05, log-rank test). Tumors from 22 patients (61%; type 2) were without TP53 mutations, and 64% of these recurred without a change in TP53 status, although 41% progressed to malignancy. This suggests that TP53 mutation is not an initiating or progression event in the majority of low grade astrocytic tumors. Our study also indicates that irradiation for WHO grade II astrocytic tumors might be associated with poor outcome (P<0.0001) and this was independent of TP53 status. These findings have important implications in the clinical management of patients with low grade astocytoma and provide new support to the clonal evolution model for tumor progression.

Brain tumor-polyposis syndrome: two genetic diseases?

PURPOSE AND DESIGN: This report presents a comprehensive and statistical analysis of the brain tumor-polyposis (BTP) cases referred to as Turcot's syndrome in the literature. RESULTS: BTP patients encompass a heterogeneous group that can be classified into two statistically distinct clinical entities based on phenotype of the polyps (P = .0001), presence of colorectal cancer (P = .0001), type of brain neoplasm, ie, glioma or medulloblastoma (P = .0001), presence of skin lesions (P = .0004) and cafe-au-lait spots (P = .0008), as well as consanguinity (P = .0135). CONCLUSION: The first entity (BTP syndrome type 1) consists of patients who have glioma and colorectal adenomas without polyposis (non-FAP cases), and their siblings with glioma and/or colorectal adenomas. For these patients, we show that the patient's age at malignant glioma occurrence is less than 20 years (50 to 80 years in the general population), which strongly supports the existence of an underlying genetic cause. The neoplasms of these patients show DNA replication errors, which suggests a relationship with hereditary nonpolyposis colorectal cancer (HNPCC), a disease characterized by germline alterations in DNA mismatch repair genes. The second entity (BTP syndrome type 2) consists of patients with a CNS tumor that occurs in a familial adenomatous polyposis kindred (FAP cases). These patients carry germline mutations in the APC gene, which suggests that mutations in this gene might predispose to brain tumors. Risk analysis shows increased incidence of medulloblastoma in FAP patients, but APC mutations are not found in sporadic glioma or medulloblastoma. Therefore, further investigations should establish whether the occurrence of medulloblastoma in an FAP family represents a variant of FAP.

Thrombospondins and tumor angiogenesis.

The thrombospondins (TSPs) are a family of five secreted proteins that are widely distributed in the extracellular matrix of numerous tissues. TSPs are multimodular and each domain specifies a distinct biological function through interaction with a specific receptor. TSP1 and TSP2 have anti-angiogenic activity, which, at least for TSP1, involves interaction with the microvascular endothelial cell receptor CD36. Expression of TSP1 and TSP2 is modulated by hypoxia and by oncogenes. In several tumors (thyroid, colon, bladder carcinomas), TSP1 expression is inversely correlated with tumor grade and survival rate, whereas in others (e.g. breast carcinomas), it is correlated with the stromal response and is of little prognostic value. Recent studies suggest that TSPs or TSP-derived peptides retaining biological activity could be developed into promising new therapeutic strategies for the anti-angiogenic treatment of solid tumors.