Consensus-Derived Risk Stratification to Guide Surveillance in Primary Uveal Melanoma: The PRIME Delphi Study

Background Primary uveal melanoma (PUM) carries a substantial risk of metastatic progression yet, approaches to risk stratification and surveillance vary widely across Brazilian centers. The PRIME Delphi study ( Individual Risk-Based Prognosis in Primary Uveal Melanoma study ) aimed to define expert consensus on prognostic determinants and surveillance strategies to support practical, harmonized clinical guidance in Brazil. Methods A multidisciplinary Delphi process was conducted in two phases. In Phase 1, ocular oncologists, clinical oncologists, and radiotherapists completed qualitative questionnaires exploring prognostic interpretation, treatment considerations, and surveillance practices. Thematic analysis informed the structured items used in Phase 2, which comprised two quantitative Delphi rounds. Consensus required ≥ 70% agreement. Key outputs included a consensus-derived risk classification table and a simplified clinical pathway. Results Fourteen ocular oncologists, seventeen clinical oncologists, and two radiotherapists completed at least one quantitative round. Consensus identified three factors that independently classify patients as high metastatic risk: monosomy 3, BAP1 mutation, and tumor thickness ≥ 10 mm. Confirmed extrascleral extension and chromosome 8q gain were recognized as prognostically relevant but not independently determinative of high-risk status. Magnetic resonance imaging (MRI) was endorsed as the preferred modality for hepatic surveillance, with computed tomography (CT) considered an acceptable alternative when MRI is unavailable High-risk patients should undergo imaging every three months during the first two years after diagnosis and every six months thereafter. For intermediate- and low-risk patients, surveillance intervals did not reach consensus in the initial quantitative round but were defined following re-iteration, resulting in consensus-based surveillance schedules across all risk categories. A consensus-driven framework was constructed to operationalize risk classification and guide surveillance strategies.