GE-24 * CHARACTERIZING INTRATUMOR HETEROGENEITY AND TUMOR EXTENSION IN PEDIATRIC DIFFUSE INTRINSIC PONTINE GLIOMA

INTRODUCTION: Diffuse intrinsic pontine glioma (DIPG) is an understudied pediatric brain cancer. The validity of at-diagnosis tumor biopsy is currently under debate with no published studies investigating tumor heterogeneity within the same patient. METHODS: Two whole brains obtained at postmortem from patients with DIPG were studied. Frozen and formalin fixed paraffin embedded (FFPE) specimens were stained for H&E, GFAP and Ki67. Histologic analysis revealed a high-grade infiltrating astrocytoma with variable cellularity and mitotic activity. Core punches representing different tumor anatomical locations, as well as tumor grades were obtained, and used for DNA and RNA extractions. DNA was analyzed for methylation (Illumina 450), and mRNA for NanoString expression profiling. To investigate molecular disparities within each grade, specimens were reanalyzed based on geographic location and pathology. Each tumor grade was then designated as classic or non-classic (not fitting all criteria for a specific grade). RESULTS: We show that mRNA profiles (unsupervised clustering) correspond to tumor grade across the brainstem (n = 12 per case). The expression pattern of twelve genes is sufficient for accurate molecular grading (WHO II, III, IV) of tissue specimens when compared with neuropathological assessment. We show detectable differential mRNA expression patterns within each grade, helping to discriminate between classic and non-classic specimens. Methylation analysis (n = 24, 12/case) verified the heterogeneous nature of mRNA expression data (within tumor grades), where WHO grade III non-classic specimens clustered with WHO grade II classic specimens. The largest molecular changes were observed in tumors from distinct geographic locations (brainstem versus lateral ventricles). CONCLUSIONS: Significant intratumor heterogeneity may affect the accuracy of targeted diagnosis in small biopsy specimens. A better understanding of these differences will help in improving therapy and devising accurate molecular-based tumor grading.