Comprehensive analysis of intratumoural heterogeneity of somatic copy number alterations in diffuse glioma reveals clonality-dependent prognostic patterns.

AIMS:Intratumoural heterogeneity (ITH) has been implicated in tumour growth and progression as well as therapy resistance. However, the extent of ITH of somatic copy number alterations (ITH-SCNAs) as a result of tumour evolution and its influence on clinical outcomes in diffuse glioma (DG) remain poorly understood. METHODS:We used an integrated computational method to infer clonal and subclonal SCNAs in 760 untreated primary DGs from The Cancer Genome Atlas. ITH-SCNAs at the genome-wide, peak (region with recurrent SCNAs) and gene level were calculated. We used the Kaplan-Meier estimators and Cox proportional hazards models to examine the associations of ITH-SCNA with patient outcomes. An independent cohort of 243 patients with paired initial and recurrent tumours from the Glioma Longitudinal Analysis Consortium was used for validation. RESULTS:DGs showed widespread ITH-SCNA, with a median of 25.5% of SCNAs identified as subclonal. We found that clonal SCNA burden had stronger prognostic power than total SCNAs in IDH-mutant astrocytoma. Coamplifications of receptor tyrosine kinases (RTKs) tended to be subclonal, and subclonal RTK amplification was significantly associated with high tumour proliferative potential and unfavourable clinical outcomes in IDH-wild-type glioblastoma. In addition, we found that the prognostic values of the peak-level SCNAs were related to their mutated clonal architecture, from which three clonality-dependent prognostic patterns of SCNAs were proposed, including clonal-dominant, subclonal-dominant and clonality-independent schemas. CONCLUSIONS:The systematic analysis of ITH-SCNAs in large cohorts of DGs highlighted the importance of considering the clonality of SCNA in discovery of tumour prognostic markers.