基于多序列MRI的多类放射组学预测儿童低级别胶质瘤BRAF突变状态
这项回顾性双机构研究纳入511例儿童低级别胶质瘤患者,利用多序列MRI提取放射组学特征,训练随机森林分类器进行三分类(BRAF融合 vs BRAF V600E vs 非BRAF),发现FLAIR是单序列中预测性能最佳者(AUC 0.82),多序列整合可改善并平衡分类性能。
要点
- FLAIR是预测BRAF突变状态的最佳单序列(AUC 0.82),其次为T2WI(0.80)、ADC(0.77)、CE-T1WI(0.75)。
- 联合临床-放射组学模型持续优于单一来源模型。
- 在180例多序列队列中,放射组学特征拼接与集成建模(macro-AUC均为0.79)均优于单序列方法(P < .001)。
核心结论
- 基于MRI的机器学习模型可能支持无创预测儿童低级别胶质瘤的BRAF突变状态;FLAIR是预测最佳的单序列,但多序列整合与改善且更平衡的性能相关。
研究设计与样本
- 回顾性双机构研究,纳入2009年1月至2023年1月期间接受术前MRI且BRAF突变状态已知的儿童低级别胶质瘤患者511例(平均年龄8.5 ± 5.1岁;45%为女孩),手动分割肿瘤并用PyRadiomics提取放射组学特征,训练随机森林分类器进行三分类(BRAF融合 vs BRAF V600E vs 非BRAF),采用留一交叉验证评估性能,遵循TRIPOD+AI指南。
关键影像表现/征象
- FLAIR序列可用495例,T2WI可用454例,增强T1WI可用285例,ADC图可用252例,全部4个序列可用180例。
- 特征分析显示FLAIR衍生特征占主导,但加入T2、ADC和CE-T1WI可改善各亚型间的平衡分类。
诊断标准与关键数值
- 分子亚型分布:BRAF融合223/511(44.6%),BRAF V600E 105/511(21.0%),非BRAF肿瘤172/511(34.4%)。
- 单序列最佳为FLAIR(AUC 0.82),其次T2WI(0.80)、ADC(0.77)、CE-T1WI(0.75);AUC值为三分类分子类别的一对多宏平均性能。
- 在180例多序列队列中,放射组学特征拼接(macro-AUC 0.79)与集成建模(macro-AUC 0.79)均优于单序列方法(P < .001)。
临床应用与阅片要点
- 联合临床-放射组学模型持续优于单一来源模型。
- 多序列放射组学可能作为指导儿童低级别胶质瘤精准治疗的有前景工具,尤其在组织采样不可行时。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
BACKGROUND AND PURPOSE: Pediatric low-grade gliomas (pLGGs) are the most common brain tumors in children and frequently harbor BRAF alterations, most commonly KIAA1549-BRAF fusions and BRAF V600E mutations, which have distinct therapeutic implications. The aim of our study was to assess multiclass radiomics-based prediction of BRAF mutation status in children with pLGGs using multisequence MRI. This study follows TRIPOD+AI guidelines.
MATERIALS AND METHODS: This retrospective bi-institutional study included pediatric patients with pLGGs and known BRAF mutation status who underwent presurgical MRI between January 2009 and January 2023. Tumors were manually segmented, and radiomics features were extracted using PyRadiomics. Random forest classifiers were trained for 3-class classification (BRAF fusion vs BRAF V600E vs non-BRAF) using clinical-only, radiomics-only, and combined models. Performance was evaluated with leave-one-out cross-validation, and results were compared across single-sequence and multisequence approaches. Single-sequence models were trained using all available patients for each MRI sequence, whereas multisequence models were restricted to the subset of 180 patients with all 4 sequences available.
RESULTS: Five hundred eleven children were included (mean age 8.5 ± 5.1 years; 45% girls). Molecular subtypes included BRAF fusion (223/511; 44.6%), BRAF V600E (105/511; 21.0%), and non-BRAF tumors (172/511; 34.4%). FLAIR sequences were available for 495, T2WI for 454, contrast-enhanced T1WI (CE-T1WI) for 285, and ADC maps for 252 children. All sequences were available for 180 children. FLAIR was the best-performing single sequence (area under the curve [AUC] 0.82), followed by T2WI (0.80), ADC (0.77), and CE-T1WI (0.75). Reported AUC values represent macro-average one-vs-rest performance across the 3 molecular classes. Combined clinical-radiomics models consistently outperformed single-source models. In the 180-patient multisequence cohort, radiomics feature concatenation (macro-AUC 0.79) and ensemble modeling (macro-AUC 0.79) both outperformed single-sequence approaches (P < .001). Feature analysis showed FLAIR-derived features dominated, but adding T2, ADC, and CE-T1WI improved balanced classification across subtypes.
CONCLUSIONS: MRI-based machine learning models may support noninvasive prediction of BRAF mutation status in pLGGs. FLAIR is the best-predicting single sequence, but multisequence integration was associated with improved and more balanced performance. These findings support multisequence radiomics as a promising tool to guide precision treatment in pLGGs, particularly when tissue sampling is not feasible.
引用
- Yahav-Dovrat A, Namdar K, Wagner MW, et al. Multiclass Radiomics-Based Prediction of BRAF Mutation Status in Pediatric Low-grade Gliomas Using Multisequence MRI. AJNR Am J Neuroradiol. 2026. PMID: 41819797. doi:10.3174/ajnr.A9296
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。