MRI影像组学预测较低级别胶质瘤免疫亚型的临床研究

Clinical investigation of MRI radiomics predicting immune subtype of lower grade glioma

目的:探讨影像组学特征预测较低级别胶质瘤(LGG)与不同免疫亚型的相关性。方法:对133例LGG患者的MRI影像特征和免疫相关基因数据进行影像基因组学分析。利用3D-slicer软件提取胶质瘤患者的影像组学特征,通过最小绝对收缩和选择算子(LASSO)回归进一步筛选与预后显著相关的特征构建影像组学评分。利用聚类算法确定最佳的免疫分型,为了探索不同影像分型预后差异的遗传背景,差异性分析不同免疫分型与影像组学评分的关系。结果:共纳入133例LGG患者,异柠檬酸脱氢酶(IDH)野生型28例,IDH突变型105例,其中野生型1p19q共缺失68例。最终由11个影像组学特征构建影像组学风险评分,影像组学风险评分中位值将患者分为高风险组(n=66)和低风险组(n=67),高评分组患者生存期预后明显较差。受试者操作特征(ROC)曲线表明影像组学预测LGG患者1年、5年生存风险的ROC曲线下面积(AUC)分别为0.79和0.86。聚类分析将LGG患者分为免疫炎症型(n=13)和免疫沙漠型两种亚型(n=113)。免疫炎症型患者影像组学评分较高,提示肿瘤纯度较低,异质性较高,预后较差。结论:影像组学评分可以有效预测LGG预后,影像组学评分与免疫分型存在一定平行关系,大部分LGG患者属于免疫沙漠型。少部分免疫炎症型患者影像组学评分较高,异质性明显高于免疫沙漠型,免疫炎症型患者可能获益于免疫治疗。

Objective:Explore the association between radiomics features and different immune subtypes in patients with lower-grade glioma(LGG).Methods:Performing radiogenomics analysis on the MRI imaging features and immune-related gene data of 133 LGG patients.The 3D-slicer software was used to extract MR imaging features.The Least Absolute Shrinkage and Selection Operator(LASSO)regression was employed to further screen features significantly associated with prognosis to construct radiomics scores.The optimal immune classification was determined using clustering algorithms.To explore the genetic background of the different prognosis among different radiomics scores,differential analysis was conducted to investigate the relationship between different immune classifications and radiomics scores.Results:A total of 133 cases of LGG patients were included in this study,including 28 cases with wild-type Isocitrate dehydrogenase(IDH)and 105 cases with mutated IDH,among which 68 cases had a combined loss of 1p19q in wild-type IDH.Eventually,radiomics risk scores was constructed using 11 radiomics features,and the median value of the radiomics risk score divided patients into high-risk(n=66)and low-risk(n=67)groups.Patients in the high-risk group had a significantly poorer prognosis.The ROC curve showed that the radiomics prediction accuracy of 1-year and 5-year survival rates for LGG patients was over 80%.Cluster analysis divided LGG patients into two subtypes:immune-inflammatory type(n=13)and immune-desert type(n=113).Patients with the immune-inflammatory subtype had a higher radiomics scores,indicating lower tumor purity,higher heterogeneity,and poorer prognosis.Conclusion:Radiomic risk scores can effectively predict the prognosis of LGG.There is a certain parallel relationship between radiomic scores and immune subtype,most LGG patients belong to immune-desert type.Few patients with immune-inflammatory type with higher imaging score,and the heterogeneity is significantly higher than that of immune desert type.Patients with immune inflammatory type may benefit from immunotherapy.