Use of brain protein I3 (BRI3) to predict disease fate in glioblastoma.

Abstract

2047 Background: Genome-wide characterization has illuminated the molecular complexity of human gliomas. Genetic alterations help predict the clinical behavior of gliomas, but variability persists. Accordingly, there is a need to expanding molecular signatures that refine prognostication. The Brain Protein I3 (BRI3) gene, localized on chromosome 7, is associated with high-grade glioma, yielding the highest hazard ratio among all high-risk genes in an in-silico glioma model. Given GBM's known chromosome 7 gain and BRI3's chromosomal location, we hypothesized that gene dosage gains and overexpression serve as prognostic biomarkers. Methods: We used the Rembrandt (n = 461 patients) and TCGA low-grade glioma (LGG) and GBM (n = 1,148 patients) databases for multi-omic analyses. RNA sequencing (Illumina HiSeq) and DNA methylation profiles (Illumina 450K) were analyzed in a combined LGG and GBM cohort, with high/low expression and hyper/hypomethylation defined by median values. CNV analysis (GISTIC 2.0) classified 2 copies as gene-copy neutral and > 2 as gene dosage gain. Somatic mutation data (SNPs/INDELs) were derived from whole-exome sequencing to decipher IDH-wt and IDH-mutant gliomas. Univariate and adjusted Cox models, Kaplan-Meier estimates, and receiver operating characteristic (ROC) curve analysis were performed. Results: Of 461 Rembrandt patients, 47.29% were GBM, 31.89% astrocytoma, 14.53% oligodendroglioma, and 6.29% normal brain. Among 1,148 TCGA LGG/GBM patients, 551 had complete molecular and clinical data, with 28% IDH-wt status, 49% MGMT hypermethylation, and 36.66% BRI3 gene dosage gains. BRI3 expression was significantly higher in GBM, grade IV tumors, IDH-wt, and mesenchymal subtypes. Among the IDH-wt cases, 75% showed BRI3 gene dosage gains with significantly elevated mRNA expression. EGFR, co-amplified in 80% of IDH-wt cases, did not affect survival (25.27 vs. 17.90 months, p = 0.346). Conversely, BRI3 gene dosage gain correlated with worse survival (79.3 vs. 17.93 mo, p = 0.001), as did BRI3 high vs. low expression (17.90 vs. 25.27 months, p = 0.015) and MGMT hypermethylation (25.27 vs. 18.63 mo, p = 0.021). Univariate analysis linked patient age (HR: 2.858 [1.781–4.586], p < 0.001), MGMT hypermethylation (HR: 2.632 [1.307–5.301], p = 0.007), BRI3 high expression (HR: 1.938 [1.131–3.320], p = 0.016) and BRI3 gene dosage gain (HR: 2.504 [1.425–4.398], p = 0.001) to worse OS. Adjusted multivariate Cox regression confirmed BRI3 gene dosage gain, age, and MGMT methylation as independent OS predictors in IDH-wt cases. ROC analysis revealed stronger prognostic performance for BRI3 gene dosage gain than MGMT hypermethylation (AUC: 0.737 vs. 0.616, p < 0.001) in IDH-wt cases. Conclusions: Elevated BRI3 gene dosage and expression portend poor prognosis and could be incorporated into models predicting disease fate in GBM.