Methylation of the TERT promoter and risk stratification of childhood brain tumours: an integrative genomic and molecular study

Background: Identification of robust biomarkers of malignancy and methods to establish disease progression is a major goal in paediatric neuro-oncology. We investigated whether methylation of the TERT promoter can be a biomarker for malignancy and patient outcome in paediatric brain tumours.

Methods: For the discovery cohort, we used samples obtained from patients with paediatric brain tumours and individuals with normal brain tissues stored at the German Cancer Research Center (Heidelberg, Germany). We used methylation arrays for genome-wide assessment of DNA. For the validation cohort, we used samples obtained from several tissues for which full clinical and follow-up data were available from two hospitals in Toronto (ON, Canada). We did methylation analysis using quantitative Sequenom and pyrosequencing of an identified region of the TERT promoter. We assessed TERT expression by real-time PCR. To establish whether the biomarker could be used to assess and predict progression, we analysed methylation in paired samples of tumours that transformed from low to high grade and from localised to metastatic, and in choroid plexus tumours of different grades. Finally, we investigated overall survival in patients with posterior fossa ependymomas in which the identified region was hypermethylated or not. All individuals responsible for assays were masked to the outcome of the patients.

Findings: Analysis of 280 samples in the discovery cohort identified one CpG site (cg11625005) in which 78 (99%) of 79 samples from normal brain tissues and low-grade tumours were not hypermethylated, but 145 (72%) of 201 samples from malignant tumours were hypermethylated (>15% methylated; p<0.0001). Analysis of 68 samples in the validation cohort identified a subset of five CpG sites (henceforth, upstream of the transcription start site [UTSS]) that was hypermethylated in all malignant paediatric brain tumours that expressed TERT but not in normal tissues that did not express TERT (p<0.0001). UTSS had a positive predictive value of 1.00 (95% CI 0.95-1.00) and a negative predictive value of 0.95 (0.87-0.99). In two paired samples of paediatric gliomas, UTSS methylation increased during transformation from low to high grade; it also increased in two paired samples that progressed from localised to metastatic disease. Two of eight atypical papillomas that had high UTSS methylation progressed to carcinomas, while the other six assessed did not progress or require additional treatment. 5-year overall survival was 51% (95% CI 31-71) for 25 patients with hypermethylated UTSS posterior fossa ependymomas and 95% (86-100) for 20 with non-hypermethylated tumours (p=0.0008). 5-year progression-free survival was 86% (68-100) for the 25 patients with non-hypermethylated UTSS tumours and 30% (10-50) for those with hypermethylated tumours (p=0.0008).

Interpretation: Hypermethylation of the UTSS region in the TERT promoter is associated with TERT expression in cancers. In paediatric brain tumours, UTSS hypermethylation is associated with tumour progression and poor prognosis. This region is easy to amplify, and the assay to establish hypermethylation can be done on most tissues in most clinical laboratories. Therefore the UTSS region is a potentially accessible biomarker for various cancers.

Funding: The Canadian Institute of Health Research and the Terry Fox Foundation.