Independent validation of a 16-protein test to assess malignant potential of small uveal melanocytic tumors.

9590 Background: Uveal melanocytic tumors of indeterminate malignant potential (UMTIMP) are usually managed by a “watch and wait” approach before the decision to treat. Although the aggressiveness of these lesions can be assessed by tumor-biopsy based molecular testing, serial tumor biopsies may be impractical and not always feasible for routine clinical management of UMTIMPs. In contrast, aqueous humor (AH) sampling can be safely performed as an outpatient procedure and is repeatable. AH protein biomarkers strongly associated with aggressive UM could be used as a sensitive and objective biological marker of malignant transformation, facilitating earlier tumor biopsy and treatment when clinically indicated. The purpose of this study was to validate a previously developed 16-protein algorithmic test for identification of high-risk tumor biology in AH sample. Methods: All study participants were clinically diagnosed with UM and had the 15-GEP, PRAME and 7-gene UM panel next-generation sequencing (NGS) test results available. AH samples (N=71) were prospectively collected at 3 independent sites under IRB approved protocols. The samples were analyzed with the Olink Target 96 Oncology II panel. The low-risk group included Class 1 and BAP1 wild type samples, and the high-risk group included Class 1 BAP1-mutant samples and all Class 2. Algorithmic analyses were performed in R and demographics analysis was performed in GraphPad Prism (version 10). Results: The sample distribution was representative of a typical UM patient cohort: average age was 62.4±15.1 years, tumor diameter 11.16±3.56 mm, and tumor thickness 4.85± 2.94 mm. The 15-GEP identified 48/71 of tumors as Class 1, and 23/71 tumors as Class 2. There was a significant difference in tumor diameter (P=0.003) and tumor thickness (P=0.001) between Class 1 and Class 2 patients. The proteins primarily belonged to Signal Transduction, Disease, and Cytokine Signaling pathways, based on Olink’s Pathway Browser. The 16-protein test had a sensitivity of 92%, specificity 52%, NPV 92%, and PPV 51%. Conclusions: A novel 16-protein algorithmic test for predicting high-risk tumor biology of the uveal melanocytic lesions was independently validated in a multi-center study. This high sensitivity test would help to accurately identify high-risk melanocytic lesions based on AH sample and provide a clinically useful ancillary approach for guiding decisions for definitive tumor biopsy and treatment.