Department of Molecular Neuropathology, Beijing Neurosurgical Institute, Capital Medical University, Beijing, China.
Chinese Glioma Genome Atlas, Beijing, China.
JCI Insight. 2019 Aug 13;5(17):130591. doi: 10.1172/jci.insight.130591.
Aberrant expression of RNA processing genes may drive the alterative RNA profile in lower-grade gliomas (LGGs). Thus, we aimed to further stratify LGGs based on the expression of RNA processing genes.
This study included 446 LGGs from The Cancer Genome Atlas (TCGA, training set) and 171 LGGs from the Chinese Glioma Genome Atlas (CGGA, validation set). The least absolute shrinkage and selection operator (LASSO) Cox regression algorithm was conducted to develop a risk-signature. The receiver operating characteristic (ROC) curves and Kaplan-Meier curves were used to study the prognosis value of the risk-signature.
Among the tested 784 RNA processing genes, 276 were significantly correlated with the OS of LGGs. Further LASSO Cox regression identified a 19-gene risk-signature, whose risk score was also an independently prognosis factor (P<0.0001, multiplex Cox regression) in the validation dataset. The signature had better prognostic value than the traditional factors "age", "grade" and "WHO 2016 classification" for 3- and 5-year survival both two datasets (AUCs > 85%). Importantly, the risk-signature could further stratify the survival of LGGs in specific subgroups of WHO 2016 classification. Furthermore, alternative splicing events for genes such as EGFR and FGFR were found to be associated with the risk score. mRNA expression levels for genes, which participated in cell proliferation and other processes, were significantly correlated to the risk score.
Our results highlight the role of RNA processing genes for further stratifying the survival of patients with LGGs and provide insight into the alternative splicing events underlying this role.
RNA 处理基因的异常表达可能导致低级别胶质瘤（LGG）中 RNA 谱的改变。因此，我们旨在根据 RNA 处理基因的表达进一步对 LGG 进行分层。
本研究纳入了来自癌症基因组图谱（TCGA，训练集）的 446 例 LGG 和来自中国胶质瘤基因组图谱（CGGA，验证集）的 171 例 LGG。采用最小绝对收缩和选择算子（LASSO）Cox 回归算法构建风险特征。采用接收者操作特征（ROC）曲线和 Kaplan-Meier 曲线研究风险特征的预后价值。
在测试的 784 个 RNA 处理基因中，有 276 个与 LGG 的 OS 显著相关。进一步的 LASSO Cox 回归确定了一个 19 基因风险特征，其风险评分也是验证数据集的独立预后因素（P<0.0001，多因素 Cox 回归）。该特征在两个数据集的 3 年和 5 年生存率方面均具有比传统因素“年龄”、“分级”和“2016 年 WHO 分类”更好的预后价值（AUCs > 85%）。重要的是，该风险特征可进一步分层特定 2016 年 WHO 分类亚组的 LGG 患者的生存情况。此外，发现 EGFR 和 FGFR 等基因的剪接事件与风险评分相关。参与细胞增殖和其他过程的基因的 mRNA 表达水平与风险评分显著相关。
我们的研究结果突出了 RNA 处理基因在进一步分层 LGG 患者生存方面的作用，并深入了解了这一作用背后的剪接事件。
