gms | German Medical Science

To determine the distinct classes of the patients with uveal melanoma (UM) by several clinical and pathomorphological data, and the survival of the patients in these classes.

Clinical (age, sex, tumor height and the largest diameter of the tumor) and pathomorphological (cellular type, nucleoli area, DNA ploidy index) data of 146 patients were subjected to cluster analysis. Patient examination results were acquired in 1983-1985, end point observation in December of 2001.

Three clusters of patients were identified, differing by seven criteria under investigation. Reliability of paired distinctions of patient survival in clusters was assessed with Cox's F-test (F1-2 =1,87 (ð=0,038), F1-3 =2,0 (ð=0,019) è F2-3 =1,08 (ð=0,41). Survival median in the I cluster was 60 months, 106 months in the II cluster and 170 months in the III cluster. The worst prognosis is obvious for the I cluster of patients, in which appeared 64% of tumors with epithelioid and mixed cellular type (spindle A cells type were completely absent), which had the highest values of DNA ploidy index (3.6, SD=0.68) and nucleolus area (105,8; SD=21). Most tumors of the II and III cluster had spindle cells type (68,8% and 72,7%, respectively) with lower values of DNA polidy index (respectively, 2,7; SD=0,39 and 2,97; SD=0,52) and nucleolus area (respectively, 78,1; SD=17,2 and 87,96; SD=19,1). Thus, the approximate ratio of melanomas with high and low malignancy is about 1:3, hence, 25% of explored tumors had high malignancy.

Using the multivariate sampling methods we defined three classes of patients with UM, differing by seven clinical and pathomorphological criteria. One of those classes was found to develop tumors with highly malignant phenotype. In disease prognosis usage of cellular type only often results in discrepancies. Employment of such criteria as DNA ploidy index and nucleoli area allowed us to improve the differentiation of UM cases in classes. Usage of these additional criteria improves the reliability of multivriative sampling in contrast to usage of cellular type criterion only. Additional data about UM features are needed to raise the quality of classification, which can shed light on UM's biological background.