Clinical and histopathological features and immunoreactivity of human choroidal and ciliary melanomas as prognostic factors for metastasis and death

Purposes To determine the relationship between immunohistochemical reactivity to osteopontin, vimentin, keratin 8/18, LZTS1, and beta-catenin and clinical and histopathological prognostic factors for metastasis and death in archival specimens of primary uveal melanomas, and the prognostic value of the evaluated study variables for death from metastasis. Methods Retrospective analysis of clinical records and formalin-fixed, paraffin-embedded slides of primary uveal melanomas treated by enucleation during May 1 1999, through June 30 2009. Immunofluorescent staining of each tumor was assessed on newly prepared histologic slides after the application of antibodies directed against five biomarkers associated with unfavorable prognosis in uveal melanoma. Results After exclusions, our study group consisted of 82 cases. Immunofluorescence was observed in 40.2% of specimens evaluated for keratin, 50.0% evaluated for osteopontin, 26.8% evaluated for β-catenin, 65.9% evaluated for vimentin, and 70.7% evaluated for LZTS1. Through available follow-up, 27 patients (32.9%) were dead of confirmed or suspected metastatic uveal melanoma. None of the patients whose tumor exhibited strong immunoreactivity to β-catenin died of metastasis. In contrast, patients whose tumor exhibited immunoreactivity of any intensity to LZTS1 were more likely to develop metastasis. In multivariate Cox proportional hazards modeling, a composite variable that took into account the immunostaining for both β-catenin and LZTS1 had a statistically significant relationship with patient's survival time. Conclusions Our study suggests that conventional clinical and histopathological prognostic factors, and immunoexpression of β-catenin and LZTS1 combined may allow better prognostication of metastasis than clinical and histomorphological factors alone.