Search

Osteopontin (OPN) is overexpressed in metastatic uveal melanoma (UM). S-100beta and melanoma-inhibitory activity (MIA) serum levels are elevated in metastatic cutaneous melanoma. The ability of OPN, S-100beta and MIA serum levels to be used as non-invasive markers for detecting metastatic UM was tested.OPN, S-100beta and MIA levels were measured by ELISA assays in 18 patients with metastatic UM and in 38 patients who were disease-free (DF) for at least 10 years after treatment of the primary tumor. Paired serum samples from 8 patients before and after development of metastasis were analyzed. Forty-four healthy controls (C) were compared to the other two groups.Serum OPN, MIA, and S-100beta levels were significantly higher in patients with metastatic UM as compared to patients who were DF for at least 10 years after treatment (p = 0.0001) or with age-matched controls. Serum OPN, MIA and S-100beta levels were significantly higher (p0.005) after metastasis formation than before the clinical detection of metastasis in the 8 patients. Receiver operator characteristic analysis was performed for metastatic patients vs. DF and vs. C, and the area under the curve was calculated for each marker and for the combination of the 3 markers, which was 91%.Elevated serum OPN, MIA and S-100beta levels correlate with metastatic UM to the liver. When used in combination, these markers provide a highly sensitive and specific method to detect hepatic metastases and therefore provide for earlier therapeutic intervention that can prolong survival.

Human uveal melanoma disseminates initially and preferentially to the liver. This study describes the relationship between the expression of the c-met proto-oncogene (receptor for hepatocyte growth factor/scatter factor (HGF/SF)) in interconverted uveal melanoma cells (co-expressing vimentin and keratin intermediate filaments) and the regulation of their motogenic response to HGF/SF, a key step in local invasion and targeted dissemination to the liver. Expression of c-met in uveal melanoma cell lines correlates with both the appearance of an interconverted phenotype and invasive ability (measured in vitro). Using chemotactic checkerboard analysis, the greatest motogenic response to HGF/SF was achieved by invasive, interconverted, c-met-positive uveal melanoma cells. C-met was observed histologically in a uveal melanoma containing interconverted cells but was absent in a tumor composed of non-interconverted cells (vimentin positive/keratin negative). The c-met ligand, HGF/SF, although not expressed by uveal melanoma cell lines, was localized in tissue sections of primary uveal melanomas and metastatic melanoma to the liver. In the primary tumor, staining for HGF/SF was most intense at the level of the choriocapillaris, a finding that is significant because 1) highly remodeled neovascular loops and networks, which appear in tumors likely to disseminate, can be traced to the choriocapillaris and the draining vortex veins and 2) HGF/SF plays a role in tumor angiogenesis. Foci of metastatic melanoma to the liver stain diffusely for HGF/SF. Regulation of the uveal melanoma interconverted phenotype by HGF/SF may play an important role in the dissemination of this tumor.

Choroidal and ciliary body melanomas disseminate exclusively by a hematogenous route because there are no lymphatics inside the eye. Although angiogenesis is an absolute precondition for metastasis in this tumor system, not all morphologic expressions of tumor angiogenesis are associated with metastasis from choroidal and ciliary body melanomas. Specifically, the remodeling of the microcirculation to form vascular networks is very strongly associated with metastasis. Type VI collagen is upregulated in tissue remodeling and the generation of tissue patterns and is either not present in the normal choroid or present at very low levels. This study was designed to investigate the possible expression of type VI collagen in the stroma of choroidal and ciliary body melanomas. Type VI collagen was detected in tissue sections from five primary choroidal melanomas and three melanomas involving the choroid and ciliary body in the subendothelial compartment of the microcirculation and in avascular areas by immunohistochemistry. Melanoma cell lines were established from each of these tumors. Cultured melanoma cells invaded into type I collagen gels and expressed type VI collagen by immunohistochemistry. Using specific primers for human type VI collagen, the expected band size (413 base pairs) was isolated from one of the cell lines by reverse transcriptase PCR. The presence of type VI collagen in the melanoma tumor stroma reflects active remodeling of the uveal extracellular matrix microenvironment by the melanoma cells themselves. Before the formation of the microvasculature, the expression of type VI collagen and of the other matrix components, such as hyaluronan, to which it binds, may erect a scaffold permitting the formation of higher order stromal patterns such as vascular networks. These stromal patterns, which are markers of tumor progression, may be detectable clinically by a specialized form of ultrasonography that detects backscatterers of the same dimension as tissue compartments encircled by vascular loops in networks.

The long-range goal of our research is to develop intervention strategies based on newly discovered biologic mechanisms responsible for the invasive dissemination of metastatic uveal melanoma. To accomplish this goal, we have focused on the biologic relevance of novel marker proteins contributing to the uveal melanoma metastatic phenotype. The expression of vimentin intermediate filaments (IFs), a mesenchymal marker, is typical of melanomas, whereas carcinomas typically express keratin IFs, which are markers for epithelia. Thus, cells that coexpress both IFs are regarded as "interconverted" in that they display both mesenchymal and epithelial phenotypes. Although the biologic functions of IFs have remained enigmatic, there is substantial support to suggest that the significance of vimentin/keratin coexpression is linked with poor patient outcome in cutaneous melanoma. Our data demonstrate that human uveal melanoma cell lines (isolated from primary choroidal or ciliary body melanomas and from foci of metastatic uveal melanoma to the liver), which contain predominant populations of cells that coexpress vimentin/keratins 8 and 18 (keratins 8,18) IFs, were 6-fold more invasive through collagenous extracellular matrices in vitro, compared with uveal melanoma cells expressing vimentin only, and were 8- to 13-fold more invasive than normal uveal melanocytes. Colocalization of vimentin/keratins 8,18 in cell cultures was corroborated by immunohistochemistry in histologic sections of tumors from which the cell lines were derived. Minor populations of these cells also coexpressed keratins 13 and 17. Experimental down-regulation of the predominant keratins 8,18 in the interconverted cells, using 16-mer antisense oligonucleotides, resulted in a significant decrease in the migratory ability of the cells-similar to levels achieved by cells positive only for vimentin. These findings provide justification for additional studies of the association between coexpression of IFs vimentin/keratins 8,18 and uveal melanoma metastasis.

Pathology observational reports and experimental data suggest that keratin and vimentin intermediate filament (IF) coexpression in breast cancer confers a more aggressive "interconverted" phenotype, expressing both epithelial and mesenchymal markers. In this study, we extended previous observations by measuring the expression of keratin and vimentin, in relation to other selected biomarkers of disease progression, in postmenopausal women with breast cancer. Using immunohistochemical analysis of 54 archival, formalin-fixed, paraffin-embedded invasive breast cancers from a well-defined cohort, we examined relative IF (keratin and vimentin) expression in a semiquantitative fashion and compared these results with other biological markers and survival. By univariate analysis, we found that vimentin expression was inversely associated with keratin expression alone (P = 0.0089) and directly related to histological grade (P = 0.017), nuclear grade (P = 0.027), Ki67 growth fraction (P = 0.024), and epidermal growth factor receptor immunostaining (P = 0.019). The relative expression of keratin and vimentin in approximately similar amounts characterized tumors with the poorest prognosis, as compared with keratin-high/vimentin-negative or keratin-low/vimentin-positive tumors. These latter two groups demonstrated similar Kaplan-Meier survival curves; the former group (keratin and vimentin in approximately similar amounts) demonstrated a poorer survival, with a hazard ratio of 2.1 (95% confidence interval, 0.5-9.6). These data suggest that relative keratin and vimentin IF expression is more indicative of prognosis and tumor phenotype than either IF marker detected independently.