PD43-11: Methylation heterogeneity in ccRCC offers insights into tumorigenesis and biomarker development

Sabrina Rossi*, cambridge, United Kingdom, Victoria Dombrowe, Berlin, Germany, Sara Pita, Christopher Smith, Grant Stewart, Cambridge, United Kingdom, Roland Schwarz, Berlin, Germany, Charlie Massie, cambridge, United Kingdom

Introduction: Approximately 30% of patients with clear cell renal cell carcinoma (ccRCC) will develop a recurrence, however there are currently no validated prognostic biomarkers. This has been identified as a key research priority. Biomarker development has been hampered by significant genetic intra-tumoral heterogeneity (ITH), yet methylation ITH is largely uncharacterised. We therefore aim to evaluate methylation heterogeneity between patients, within patients and within individual tumour samples.

Methods: Multi-region samples were collected from fresh frozen nephrectomy specimens from individuals with ccRCC (Ethical approval: REC ID 03/018). Multiple tumour and normal samples were obtained from each patient (at least 5 samples per patient). Methylation analysis was performed using the Illumina EPIC-seq to obtain high-resolution sequence-level data. Matched samples were processed for copy-number alterations (whole exome sequencing) and gene expression (RNA-seq). An independent cohort of ccRCC tissue samples was sequenced for external validation.

Results: We sequenced 136 samples from 18 ccRCC patients, which consists of the largest multi-region methylation cohort to date. We observed extensive heterogeneity between patients, which dominated over within-patient heterogeneity. Although disordered methylation is believed to be a stochastic process, there were significant changes in methylation epipolymorphism between ccRCC and normal kidney at the promoter region of 47 known kidney cancer genes (including SLC16A3, MYC, WT1, KRT7 and KRT18). This finding was confirmed in an independent cohort of 71 ccRCC tumour and normal samples. Genes with significantly disordered methylation in tumour are more often associated with significant hypermethylation in ccRCC (87% genes hypermethylated) rather than hypomethylation, and are negatively correlated with gene expression, implying that methylation disorder may have a functional consequence. Several genes which have been implicated in ccRCC prognosis (including DKK2, SFRP1, CCND1, SERPINF1, SMAD3) demonstrated heterogeneous methylation patterns.

Conclusions: Heterogeneous methylation between patients and within a sample may be a key determinant of the noted failure to validate prognostic genes in ccRCC. Disordered methylation and ITH may lead to inconsistent results amongst samples, which is exacerbated by different methylation platforms covering disparate CpGs within the promoter of genes. This key finding may help identify prognostic methylation markers in future.

Source of Funding: Cancer Research UK