Abstract
Abstract: :
Purpose:It has been shown that microvascular density (MVD) is associated with death from metastatic melanoma. It has also been shown that melanoma cells label with CD34 and CD31. This study was designed to further explore the relationship between the labeling of melanoma cells by putative endothelial cell markers, calculation of MVD, and the influence of endothelial cell marker labeling by tumor cells on outcome. Methods:Tissue sections were stained with CD34 to calculate MVD as described previously, and a Cox model was developed to test the contribution of MVD to outcome in this series. Tissue sections were labeled with either melanoma marker S100 protein or Melan-A and then stained with either CD31, CD34, or factor 8 related antigen to study co-labeling of melanoma cells for putative endothelial cell markers. Co-localization of CD34 and S100 protein was determined in each of the 200 cases of uveal melanoma in this series and the effect of labeling of tumor cells by CD34 was tested in a Cox model. Results:Without consideration for whether tumor cells labeled with CD34, MVD entered into a Cox model along with the presence of loops/networks, largest tumor dimension, mitotic figures, tumor infiltrating lymphocytes, epithelioid cells, age, and gender. However, tumor cell labeling by CD34 was associated with MVD (p = 0.0007), and when permitted entry into the Cox model, the attribute of CD34 labeling by tumor cells remained in the model as a prognostic factor. Conclusion:Counting discrete points of CD34 labeling may detect not only blood vessels but also tumor cells that label for this marker. Thus, the tumor attribute conventionally known as «MVD» may not accurately reflect the degree of tumor vascularization, and the effect of «MVD» on outcome may be attributed, in part, to the inappropriate expression of tumor cells for endothelial cell markers such as CD34.
Keywords: 464 melanoma • 507 pathology: human • 474 microscopy: light/fluorescence/immunohistochemistry