DNA alteration-based classification of uveal melanoma gives better prognostic stratification than immune infiltration, which has a neutral effect in high-risk group

Abstract

Background: In uveal melanomas, immune infiltration is a marker of poor prognosis. This work intended to decipher the biological characteristics of intra-tumor immune population, compare it to other established biomarkers and to patients' outcome.

Methods: Primary, untreated, and mainly large uveal melanomas with retinal detachment were analyzed using: transcriptomic profiling (n = 15), RT-qPCR (n = 36), immunohistochemistry (n = 89), Multiplex Ligation-dependent Probe Amplification (MLPA) for copy number alterations (CNA) analysis (n = 89), array-CGH (n = 17), and survival statistics (n = 86).

Results: Gene expression analysis divided uveal melanomas into two groups, according to the IFNγ/STAT1-IRF1 pathway activation. Tumors with IFNγ-signature had poorer prognosis and showed increased infiltration of CD8⁺ T lymphocytes and macrophages. Cox multivariate analyses of immune cell infiltration with MLPA data delineated better prognostic value for three prognostic groups (three-tier stratification) than two (two-tier stratification). CNA-based model comprising monosomy 3, 8q amplification, and LZTS1and NBL1 deletions emerged as the best predictor for disease-free survival. It outperformed immune cell infiltration in receiver operating characteristic curves. The model that combined CNA and immune infiltration defined risk-groups according to the number of DNA alterations. Immune cell infiltration was increased in the high-risk group (73.7%), where it did not correlate with patient survival, while it was associated with poorer outcome in the intermediate risk-group.

Conclusions: High degree of immune cell infiltration occurs in a subset of uveal melanomas, is interferon-gamma-related, and associated with poor survival. It allows for two-tier stratification, which is prognostically less efficient than a three-tier one. The best prognostic stratification is by CNA model with three risk-groups where immune cell infiltration impacts only some subgroups.