Osteosarcoma Multi-omics Landscape and Subtypes

Abstract

Abstract Background Osteosarcoma is the most common primary bone malignancy exhibiting remarkable histologic diversity and genetic heterogeneity. The complex nature of osteosarcoma has confounded precise molecular categorization, prognosis and prediction for this disease. Despite intensive studies aimed at identifying genes or biomarkers involved in pathogeneses, linking clinical outcomes with omics profiles in osteosarcoma has far remained elusive. Results 86 osteosarcoma tumors with matched profiles of somatic copy-number alteration, gene expression and methylation were categorized into three subgroups by similarity network fusion. The subgrouping criteria was validated on another cohort osteosarcoma tumors. Then the differences among these three subgroups were then investigated based on single-platform profiles. Conclusions The multiplatform analysis yields three molecularly distinct and clinically relevant subtypes for osteosarcoma. Previously unappreciated osteosarcoma-type-specific changes at genomic, transcriptomic and epigenetic level were revealed. Several novel factors, such as copy number in 17p13.1-17q11.2, expression of CDK6 or EGFR, and methylation status of Hippo signaling pathway, were found to be closely related to the diverse clinical outcomes in osteosarcoma patients. These findings provide a comprehensive genomic architecture for osteosarcoma and emphasize the need for data integration from different platforms.