Epigenetic studies in rhabdomyosarcoma

Division Of Basic Sciences - Nci

Investigators

Linked publications, trials & patents

Abstract

In previous studies, we established a cohort of frozen samples derived from 223 RMS tumors to investigate the biological and clinical signi?cance of global DNA methylation patterns in pediatric RMS. Unsupervised analysis of the genome-wide DNA methylation patterns in this large cohort con?rmed that these RMS cases can be divided into two major clusters, one cluster consisting of fusion-positive (FP) cases (usually PAX3-FOXO1 [P3F] or PAX7-FOXO1 [P7F]) and a second cluster consisting of fusion-negative (FN) cases. Furthermore, we found that the FP cluster can be divided into P3F-enriched and P7F-enriched subsets, and the FN cluster can be divided into two subsets (designated as FN1 and FN2). The large FN2 category is composed of tumors with classical embryonal histology and frequent mutations of the RAS signaling pathway, whereas the FN1 subset is composed of multiple groups, including a group with embryonal histology and TP53 or RAS pathway mutations (FN1-E), a group of sclerosing/spindle cell tumors with recurrent MYOD1 mutations (MYOD1), a group of fusion-negative tumors with alveolar histology (AFN), and a heterogeneous group of other tumors (FN1-O). Based on these findings, we proposed that DNA methylation analysis could identify at least seven discrete categories: P3F-enriched, P7F-enriched, FN2, FN1-E, MYOD1, AFN and FN1-O. To validate these findings, we analyzed a second independent cohort of recent cases from the Children's Oncology Group (DOG) for which exome sequencing, fusion detection and DNA methylation profiling were performed as part of the Molecular Characterization Initiative (MCI) of the Childhood Cancer Data Initiative (CCDI). Starting with a cohort of 172 RMS MCI cases, DNA methylation data was used to cluster the cases according to similarities and differences in overall DNA methylation pattern with hierarchical clustering and t-distributed stochastic neighbor embedding (t-SNE) algorithms. These methods identified two major subsets, a fusion positive-cohort of 44 cases and a fusion-negative cohort of 128 cases. Using the published sarcoma methylation classifier, all cases in the fusion-positive cohort were classified as alveolar RMS. Within the fusion-negative cohort, there were two major subsets, an FN1 cohort containing 37 cases and an FN2 cohort containing 91 cases. All cases within the FN2 cohort were classified by the sarcoma methylation classifier as embryonal RMS and contained frequent mutations of the RAS signaling pathway (51/91 cases = 56%). The FN1 cohort could then be divided into four additional subsets corresponding to our previously determined FN1-E, MYOD1, AFN and FN1-O categories. In particular, the FN1-E category consisted of 6 cases, which were classified as embryonal RMS and contained frequent mutations of RAS pathway genes (3/6 cases = 50%) and TP53 (5/6 cases = 83%, including 3 cases (50%) with germline TP53 mutations. The MYOD1 category consisted of 3 cases classified as MYOD1-mutant RMS, all with MYOD1 Leu122Arg mutations. The AFN category consisted of 5 cases, which were classified as alveolar RMS but did not contain fusions in the three cases for which fusion analysis was performed. Finally, the FN1-O category consisted of 23 cases, which were classified into a variety of non-RMS diagnostic categories, often with a low confidence score. For the FN1-O cases classified with high confidence score, three were classified as inflammatory myofibroblastic tumor, two were classified as RMS-like sarcoma and one each was classified as Langerhans cell histiocytosis, blood, osteosarcoma and small blue round cell tumor. Five of these FN1-O cases contained DICER1 mutations (22%) and five contained RAS pathway mutations (22%). Overall, these DNA methylation studies in the MCI cohort showed marked similarities to our previous studies of our initial discovery cohort, thus validating our overall results and conclusions.