From cytologic features to serum markers: advancing prognostication of uveal melanoma

Abstract

Uveal melanoma (UM) is the most common primary intraocular malignancy in adults. As surveys show, almost all patients diagnosed with UM want prognostic information. However, the tools available today for early prognostication in UM are limited; beyond what information can be obtained by tumor size and location, current methods typically rely on tumor tissue obtained from enucleated eyes or invasive biopsies, or on repeated sampling of peripheral plasma or serum. Therefore, improving the toolbox for prognostication of aggressive vs. less aggressive disease is the focus of this dissertation.

In paper I, we explored the scope of digital image analysis of cytologic features in uveal melanoma. We know that cytologic features such as shape and size of tumor cells can predict metastatic death in uveal melanoma. Currently, however, cytologic analysis is conducted manually by visual inspection of pathologists, which is time-consuming and impacts its reproducibility. In this study, we analyzed twelve morphometric variables in a large number of uveal melanoma cells. We then correlated the results with BRCA associated protein-1 (BAP-1) expression and BAP-1 gene mutation status, monosomy 3, gene expression classification as well as patient survival. We demonstrated that digital image analysis of variables describing the shape and size of tumor nuclei correlated to BAP-1 status, to monosomy 3, and to gene expression class. Mean time consumption per tumor was less than 2.5 min. Thus, digital image analysis is a fast and highly reproducible technique, that for the first time allows to objectively quantify cytologic tumor features in large tumors and gives prognostic information on survival in UM.

In paper II, we explored if biopsies from posterior UM hold the same prognostic information as tissue from enucleated specimens. We classified BAP-1 expression in transvitreal biopsies of posterior UM and correlated our results with BAP-1 expression in subsequent biopsies. We found that BAP-1 expression in transvitreal biopsies was concordant with BAP-1 in enucleated specimens. Moreover, BAP-1 expression in transvitreal biopsies identified patients with poor metastasis-free survival. Thus, transvitreal biopsies can be a good prognostic tool for patients that do not undergo enucleation. The results also indicate the need to further study BAP-1 expression in even smaller UMs, to infer the tumor size at which loss of BAP-1 starts to occur, and to correlate it with the early seeding of micrometastases.

In paper III, we explored prognostic testing on serum samples obtained from patients at diagnosis of uveal melanoma. We first screened for cancer-related proteins by protein profiling in order to find potential biomarkers. Second, we performed ELISA to evaluate the serum levels of the best candidates. Third, receiver operating characteristics were used to define thresholds for metastatic risk. This led to a prognostic test (serUM-Px) that stratifies patients into low, intermediate, and high-risk categories which we tested in a training cohort and validation cohort. We found that serUM-Px is a prognostic test, based on a single peripheral venous blood sample at the time of UM diagnosis, which can be used to stratify patients into low, intermediate and high metastatic risk categories. Thus, this test predicts metastases many years in advance without the need for biopsy or repeated sampling.

In paper IV, we wanted to assess the prognostic utility of Tenascin C (TNC) in uveal melanoma. Therefore, we collected peripheral blood samples from 82 patients with small posterior uveal melanomas between 1996 and 1999 as described above for paper III. TNC concentrations were examined 2021. RNA sequencing data of TNC from an additional 80 larger tumors were collected. These levels were subsequently linked with the cumulative incidence of metastasis-related death through competing risk data analysis. In our analysis we observed no significant disparities in tumor size, age at diagnosis, visual acuity, serum protein levels, or treatment modality between patients with above or below median serum or tumor TNC levels. However, above median TNC RNA was associated with BAP1 mutation, monosomy 3, and epitheloid cytomorphology. The competing risk analysis indicated increased metastatic mortality in patients with above median TNC and primary tumor TNC RNA compared to their below median counterparts. We conclude that TNC is a prognostic biomarker in uveal melanoma that exhibits elevated levels in peripheral blood and tumors at diagnosis in patients destined for metastatic death.