Introduction: The disease course of patients with diffuse low-grade gliomas (LGGs) is notoriously unpredictable. Detailed analysis of the genetic make-up is therefore key to understand this wide variety in overall survival of patients diagnosed with LGGs. Reports on the prognostic value of CNAs other than 1p/19q co-deletion are conflicting, which may be explained by spatial copy number heterogeneity.

Aim: Prognostic implications of spatial and temporal copy number aberrations in LGGs by whole genome sequencing.

Methods: Approximately 25% of LGG patients have a life expectation of more than 20 years following diagnosis, which necessitates collection of samples with long clinical follow up information. Therefore archival material was selected for this project and WGS was developed to access these samples without the requirement for a normal reference. We collected 156 formalin-fixed and paraffin-embedded (FFPE) samples in a discovery cohort of 98 LGG patients with extensive clinical follow-up data, which included recurrent tumours and spatially distinct regions. An independent cohort of 126 samples was obtained from a recently published study (Alentorn et al, Neuro-oncology 2014). Technical challenges for genome-wide inference of copy number aberrations include repetitive and common DNA sequences in the genome and sequence variation across the general population, as well as the compromised and variable quality of DNA obtained from archived tissues. We developed a robust and cost effective method that infers copy number aberrations from WGS data of approximately 0.1x coverage, without the need for a reference signal. This method implements (1) a combined LOESS correction for mappability and GC content, and improves on previous methods by (2) comprehensive filtering based on public genome project data, 1000-Genomes project and ENCODE blacklists.

Results: Both prognostic value, temporal evolution and spatial heterogeneity of CNA were assessed by WGS. We confirmed prognostic favourable value of 1p/19q co-deletion, and demonstrated loss of 10q to be an unfavourable marker. In paired recurrences 10q loss was invariably maintained and surfaced in 4 additional recurrences of the discovery cohort. In spatial regions of LGGs we recognized extensive copy number heterogeneity; 15 of 17 LGGs show spatial variability of CNAs. 1p/19q co-deletion is homogeneous, while loss of 10q is heterogeneously present.

Conclusions: we present clinically relevant CNAs, but also demonstrate extensive spatial copy number heterogeneity in diffuse LGGs that might complicate unequivocal biomarker discovery.

Citation Format: Ilari Scheinin, Hinke F. van Thuijl, Daoud Sie, Hendrik F. van Essen, Paul P. Eijk, Francois Rustenburg, Ahmed Idbaih, Agusti Alentorn, Gerrit A. Meijer, Mark A. van der Wiel, Henrik Bengtsson, Adam Olshen, Eleonora Aronica, Jan J. Heimans, Jaap C. Reijneveld, Pieter Wesseling, Donna G. Albertson, Dan Pinkel, Bauke Ylstra. A novel approach to copy number assessment by whole genome sequencing reveals extensive spatial heterogeneity in diffuse low-grade glioma. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 3426. doi:10.1158/1538-7445.AM2014-3426