Prognostic scoring system for osteosarcoma using network-regularized high-dimensional Cox-regression analysis and potential therapeutic targets

With the recent emphasis on the importance of personalized genomic medicine, studies have performed prognostic stratification using gene signatures in cancers. However, these studies have not considered gene networks with clinical data. Therefore, this study aimed to develop a novel prognostic score using grouped variable selection for patients with osteosarcoma. We assessed messenger RNA (mRNA) expression and clinical data from Gene Expression Omnibus to develop a novel prognostic scoring system for patients with osteosarcoma. Variable selection using Network-Regularized high-dimensional Cox-regression analysis with information regarding gene networks obtained from six large pathway databases was performed. We determined the risk score on the linear combination of regression coefficients and mRNA expression values. Log-rank test, UNO's c-index, and area under the curve (AUC) values were determined to evaluate the discriminatory power between the low- and high-risk groups. A recently reported next-generation Connectivity Map was used to identify future therapeutic targets for osteosarcoma. Our novel model had significantly high discriminatory power in predicting overall survival. An optimal c-index of 0.967 was obtained and time-dependent receiver operating characteristic analysis revealed an acceptable predictive value of AUC between 0.953 and 1.000. Knockdown of BACE2 or ING2 and linifanib treatment may improve the prognosis of patients with osteosarcoma. Herein, this novel prognostic scoring system would not only facilitate a more accurate prediction of patient prognosis, but also contribute to the selection of suitable therapeutic alternatives for osteosarcoma patients.