Abstract
Purpose :
Over the past 10 years, gene expression profiling (GEP) has been used to estimate the risk for metastatic death in UM. However, inaccurate genomic results can be obtained if normal tissue is inadvertently analyzed or due to tumor heterogeneity. Our group reported on the role of on-site rapid cytologic evaluation to minimize these errors. In this study, we examine the significance of cytologic identification of epithelioid cells (predictor of poor prognosis) related to the GEP prognostic classes (1A=good, 1B=intermediate, 2=poor) and patient outcomes.
Methods :
This was a retrospective review of 82 consecutive patients diagnosed with UM between 2013-16 who underwent FNA biopsy at the time of brachytherapy. Review of the clinical variables, GEP results, and detailed analysis of the morphology of the cells in cytological samples was performed. The association between morphologic features, GEP classification, and vital status was assessed.
Results :
Sixty six of 82 cases of UM in this series had complete cytology and GEP results available for analysis (mean follow up: 1.3 years; 0.1-3.8). The remainder were excluded due to inadequate clinical, morphologic, or GEP data, or inadequate follow up. Chi-square analysis demonstrated overall consistency between the cytology and GEP (p=0.027). Analysis of morphology along with GEP class demonstrated disparity in 27/65 cases (41%) (see Table).
Conclusions :
In our study, 27 patients (41%) had some disparity between morphologic features and GEP, two of whom died of metastatic disease. Our preliminary data suggests that in a subset of cases, identification of epithelioid cells during FNAB adds valuable prognostic data to GEP by providing independent information especially in the rare cases with low discriminant value or unavailable GEP and support previous reports of tumor heterogeneity. Longer follow up of this cohort will establish survival outcomes in the remainder of discordant cases and prospective trials should be initiated to validate these findings.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.