Construction of a novel mRNAsi-related risk model for predicting prognosis and immunotherapy response in osteosarcoma

Zhe Li; Chi Jin; Xinchang Lu; Yan Zhang; Yi Zhang; Jia Wen; Yongkui Liu; Xiaoting Liu; Jiazhen Li

doi:10.21037/atm-22-6011

Construction of a novel mRNAsi-related risk model for predicting prognosis and immunotherapy response in osteosarcoma

Ann Transl Med. 2023 Jan 31;11(2):61. doi: 10.21037/atm-22-6011.

Authors

Zhe Li¹, Chi Jin¹, Xinchang Lu¹, Yan Zhang¹, Yi Zhang¹, Jia Wen¹, Yongkui Liu¹, Xiaoting Liu¹, Jiazhen Li¹

Affiliation

¹ Department of Orthopaedics, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China.

Abstract

Background: Targeting cancer stem cells (CSC) may represent a future therapeutic direction for osteosarcoma (OS), which mainly relies on the identification of CSC markers. This study aimed to classify OS based on messenger ribonucleic acid (mRNA) stemness indices (mRNAsi) and construct a mRNAsi-related risk model to predict the prognosis of OS.

Methods: The one-class logistic regression (OCLR) algorithm was applied to the RNA- sequencing (seq) data of human embryonic stem cells (hESC) and induced pluripotent stem cell (iPSC) lines to calculate mRNAsi. Weighted gene co-expression network analysis (WGCNA) was performed on data obtained from the TARGET database to screen the mRNAsi-related genes. Univariate Cox regression analysis was implemented to screen mRNAsi-related genes with prognostic significance for consensus clustering of OS. The least absolute shrinkage and selection operator (LASSO) and COX regression analysis were conducted to construct a risk model based on mRNAsi-related genes.

Results: Six gene modules were identified in the TARGET database. The yellow module showed the strongest negative correlation with mRNAsi and the strongest significant positive correlation with the immune score and stromal score. OS was divided into three molecular subtypes with significant survival differences based on 73 mRNAsi-related genes with prognostic value for OS. The survival rate was ranked as C3 < C1 < C2 from low to high. The levels of immune components in C2 was significantly higher than those in C1 and C3. HSD11B2, GBP1, RNF130, APBB1IP, and NPC2 in the yellow module were used as variables for building the mRNAsi-related risk model. The survival rate of the high-risk group (as defined by this model) was significantly higher than that of the low-risk group, and it had significant survival prediction ability in 28 types of cancer. In addition, the mRNAsi-related risk model was superior to the Tumor Immune Dysfunction and Exclusion (TIDE) model in predicting the prognosis and immunotherapy response in all three immunotherapy cohorts.

Conclusions: This study classified OS and constructed a mRNAsi-related risk model based on mRNAsi-related genes, which provides a potential tool for more accurate risk stratification of OS and prediction of immunotherapy response.

Keywords: immunotherapy response; mRNA stemness indices; osteosarcoma (OS); prognosis; tumor microenvironment.