Predictive signature of malignant recurrence in G-CIMP tumors
Souza C, and Noushmehr H. Predictive signature of malignant recurrence in G-CIMP tumors. Neuro-Oncology 2017; 19(suppl 6):vi98.
Histomorphology and grading schemes are unable to predict glioma relapse and malignant tumor progression. We reported IDH-mutant associated Glioma-CpG Island Methylator Phenotype (G-CIMP) can be further divided into two clinically distinct subtypes independent of neuropathological grading (G-CIMP-high and -low) with evidence of tumor progression. Glioma biomarkers that can predict the risk of progression of G-CIMP-high tumors (favorable survival) to G-CIMP-low tumors (unfavorable survival) are lacking. To find molecular markers of G-CIMP progression, we first defined G-CIMP by analyzing genome-wide DNA methylation data of 181 longitudinal glioma fragments derived from 74 (21 from our own cohort) patient biopsies using the HM450 DNA methylation platform, the Random Forest computational method, and supervised analysis of G-CIMP tumors using either Wilcoxon test or Fisher's exact test (P < 0.05) and a beta-value difference cutoff of 0.2. We found 134 G-CIMPs out of 181 glioma fragments and then we defined a set of candidate biomarker signature composed by seven hypomethylated CpG sites in primary G-CIMP-high gliomas that undergo malignant progression to WHO grade IV recurrent G-CIMP-low gliomas (n=7) when compared to G-CIMP-high primary gliomas that retain the G-CIMP-high epigenetic profiling through tumor recurrence (n=32) (Fisher-test P < 0.05). Our molecular signature was able to predict the risk to progression to G-CIMP-low tumors in an independent set of TCGA and non-TCGA primary CIMP gliomas (n=79 out of 271; 29%) which associates significantly with patient overall survival (log-rank P value = 0.02, hazard ratio = 2.19). Our study provides insights into the tumorigenic events that may drive G-CIMP malignant recurrence and opportunities for further targeted therapy exploitation and clinical trials design.