Abstract：
BACKGROUND: This study aimed to get a deeper insight into new osteosarcoma (OS) signature based on bone morphogenetic proteins (BMPs)-related genes and to confirm the prognostic pattern to speculate on the overall survival among OS patients.
METHODS: Firstly, pathway analyses using Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) were managed to search for possible prognostic mechanisms attached to the OS-specific differentially expressed BMPs-related genes (DEBRGs). Secondly, univariate and multivariate Cox analysis was executed to filter the prognostic DEBRGs and establish the polygenic model for risk prediction in OS patients with the least absolute shrinkage and selection operator (LASSO) regression analysis. The receiver operating characteristic (ROC) curve weighed the model\'s accuracy. Thirdly, the GEO database (GSE21257) was operated for independent validation. The nomogram was initiated using multivariable Cox regression. Immune infiltration of the OS sample was calculated. Finally, the three discovered hallmark genes\' mRNA and protein expressions were verified.
RESULTS: A total of 46 DEBRGs were found in the OS and control samples, and three prognostic DEBRGs (DLX2, TERT, and EVX1) were screened under the LASSO regression analyses. Multivariate and univariate Cox regression analysis were devised to forge the OS risk model. Both the TARGET training and validation sets indicated that the prognostic biomarker-based risk score model performed well based on ROC curves. In high- and low-risk groups, immune cells, including memory B, activated mast, resting mast, plasma, and activated memory CD4 + T cells, and the immune, stromal, and ESTIMATE scores showed significant differences. The nomogram that predicts survival was established with good performance according to clinical features of OS patients and risk scores. Finally, the expression of three crucial BMP-related genes in OS cell lines was investigated using quantitative real-time polymerase chain reaction (qRT-PCR) and western blotting (WB).
CONCLUSIONS: The new BMP-related prognostic signature linked to OS can be a new tool to identify biomarkers to detect the disease early and a potential candidate to better treat OS in the future.
METHODS: Firstly, pathway analyses using Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) were managed to search for possible prognostic mechanisms attached to the OS-specific differentially expressed BMPs-related genes (DEBRGs). Secondly, univariate and multivariate Cox analysis was executed to filter the prognostic DEBRGs and establish the polygenic model for risk prediction in OS patients with the least absolute shrinkage and selection operator (LASSO) regression analysis. The receiver operating characteristic (ROC) curve weighed the model\'s accuracy. Thirdly, the GEO database (GSE21257) was operated for independent validation. The nomogram was initiated using multivariable Cox regression. Immune infiltration of the OS sample was calculated. Finally, the three discovered hallmark genes\' mRNA and protein expressions were verified.
RESULTS: A total of 46 DEBRGs were found in the OS and control samples, and three prognostic DEBRGs (DLX2, TERT, and EVX1) were screened under the LASSO regression analyses. Multivariate and univariate Cox regression analysis were devised to forge the OS risk model. Both the TARGET training and validation sets indicated that the prognostic biomarker-based risk score model performed well based on ROC curves. In high- and low-risk groups, immune cells, including memory B, activated mast, resting mast, plasma, and activated memory CD4 + T cells, and the immune, stromal, and ESTIMATE scores showed significant differences. The nomogram that predicts survival was established with good performance according to clinical features of OS patients and risk scores. Finally, the expression of three crucial BMP-related genes in OS cell lines was investigated using quantitative real-time polymerase chain reaction (qRT-PCR) and western blotting (WB).
CONCLUSIONS: The new BMP-related prognostic signature linked to OS can be a new tool to identify biomarkers to detect the disease early and a potential candidate to better treat OS in the future.
摘要：
背景：本研究旨在基于骨形态发生蛋白（BMPs）相关基因更深入地了解新的骨肉瘤（OS）特征，并确认预后模式以推测OS患者的总生存期。
方法：首先，使用基因本体论(GO)和京都基因和基因组百科全书(KEGG)进行通路分析,以寻找OS特异性差异表达BMP相关基因(DEBRGs)可能的预后机制.其次,采用最小绝对收缩和选择算子(LASSO)回归分析,对OS患者进行单因素和多因素Cox分析,筛选出预后DEBRGs,并建立多基因模型进行风险预测.接收器工作特性（ROC）曲线权衡了模型的准确性。第三，对GEO数据库(GSE21257)进行独立验证.使用多变量Cox回归开始列线图。计算OS样品的免疫浸润。最后,这三个发现的标志性基因的mRNA和蛋白质表达得到了验证。
结果：在OS和对照样品中共发现46个DEBRG，和三个预后性DEBRG(DLX2，TERT，和EVX1)在LASSO回归分析下进行筛选。设计了多变量和单变量Cox回归分析来伪造OS风险模型。TARGET训练集和验证集均表明基于预后生物标志物的风险评分模型基于ROC曲线表现良好。在高风险和低风险人群中，免疫细胞，包括内存B，激活的桅杆,休息桅杆,等离子体,激活记忆CD4+T细胞，和免疫力，基质,和ESTIMATE评分存在显著差异。根据OS患者的临床特征和风险评分，建立了预测生存的列线图，表现良好。最后,使用定量实时聚合酶链反应（qRT-PCR）和蛋白质印迹（WB）研究了三个关键BMP相关基因在OS细胞系中的表达。
结论：与OS相关的新的BMP相关的预后特征可以成为识别早期检测疾病的生物标志物的新工具，也是未来更好地治疗OS的潜在候选者。
方法：首先，
结果：在OS和对照样品中共发现46个DEBRG，
结论：与OS相关的新的BMP相关的预后特征可以成为识别早期检测疾病的生物标志物的新工具，也是未来更好地治疗OS的潜在候选者。
