MicroRNA-mRNA networks define translatable molecular outcome phenotypes in osteosarcoma [article]

Christopher E. Lietz, Cassandra Garbutt, William T. Barry, Vikram Deshpande, Yen-Lin Chen, Santiago A. Lozano-Calderon, Yaoyu Wang, Brian Lawney, David Ebb, Gregory M. Cote, Zhenfeng Duan, Francis J. Hornicek (+5 others)
2019 medRxiv   pre-print
There is a lack of well validated biomarkers in osteosarcoma, a rare, recalcitrant disease with variable outcome and poorly understood biologic behavior, for which treatment standards have stalled for decades. The only standard prognostic factor in osteosarcoma remains the amount of pathologic necrosis following pre-operative chemotherapy, which does not adequately capture the biologic complexity of the tumor and has not resulted in optimized patient therapeutic stratification. New, robust
more » ... rkers are needed to understand prognosis and better reflect the underlying biologic and molecular complexity of this disease. Methods: We performed microRNA sequencing in 74 frozen osteosarcoma biopsy samples, the largest single center translationally analyzed cohort to date, and separately analyzed a multi-omic dataset from a large (n = 95) NCI supported national cooperative group cohort. Molecular patterns were tested for association with outcome and used to identify novel therapeutics for further study by integrative pharmacogenomic analysis. Results: MicroRNA profiles were found predict Recurrence Free Survival (5-microRNA profile, Median RFS 59 vs 202 months, log rank p=0.06, HR 1.87, 95% CI 0.96-3.66). The profiles were independently prognostic of RFS when controlled for metastatic disease at diagnosis and pathologic necrosis following chemotherapy in multivariate Cox proportional hazards regression (5-microRNA profile, HR 3.31, 95% CI 1.31-8.36, p=0.01). Strong trends for survival discrimination were observed in the independent NCI dataset, and transcriptomic analysis revealed the downstream microRNA regulatory targets are also predictive of survival (median RFS 17 vs 105 months, log rank p=0.007). Additionally, DNA methylation patterns held prognostic significance. Through machine learning based integrative pharmacogenomic analysis, the microRNA biomarkers identify novel therapeutics for further study and stratified application in osteosarcoma. Conclusions: Our results support the existence of molecularly defined phenotypes in osteosarcoma associated with distinct outcome independent of clinicopathologic features. We validated candidate microRNA profiles and their associated molecular networks for prognostic value in multiple independent datasets. These networks may define previously unrecognized osteosarcoma subtypes with distinct molecular context and clinical course potentially appropriate for future application of tailored treatment strategies in different patient subgroups.
doi:10.1101/19007740 fatcat:powk6534uvc7tekdmdulhnk3va