Abstract
Ependymoma is the second most common malignant childhood brain tumour. Immunity has been implicated in recurrence but its role needs further definition to inform therapy. We profiled two independent ependymoma cohorts using RNA sequencing and DNA methylation profiling; one from fresh frozen (FF), and the other from paraffin embedded (FFPE), material. FF and FFPE cohorts contained 27 and 25 matched primary and recurrent pairs respectively. Following differential expression (DEA) and gene set enrichment analyses (GSEA), we used validated algorithms to calculate parameters associated with anti-tumour immune responses, including: immunophenoscore (IPS), cytolytic activity (CYT), and expression of immune checkpoint genes. Clustering using RNA-seq data, from primary and recurrent tumours, formed location based groups. Posterior fossa (PF) clusters formed two groups, PF1 and PF2, with expression patterns similar to EPN_PFA and EPN_PFB tumours, however, both contained predominantly EPN_PFA DNA methylation designations. GSEA comparing primary to first relapse was similar in both cohorts. PF1 demonstrated treatment independent downregulation of inflammatory responses, cytokine activity and chemotaxis. Radiotherapy treated PF1 relapses were associated with upregulation of immune related ontologies. PF2 and ST relapses showed upregulation of B- and T-cell ontologies. IPS did not change at recurrence, and was below levels associated with response to checkpoint inhibition. CYT and expression of checkpoint inhibition molecules were low. Whilst these independent datasets demonstrated shared immunological changes at recurrence; immune based algorithms indicated that activity may be inadequate to respond to immune checkpoint therapy. Other immunotherapeutic approaches may be effective, but better understanding of ependymoma’s immunological basis is essential.