| 3 | 0 | 25 |
| 下载次数 | 被引频次 | 阅读次数 |
目的 探讨11C-蛋氨酸(11C-MET)正电子发射断层显像(PET)代谢参数及影像组学模型对脑干胶质瘤H3 K27M基因突变状态的预测价值。方法 选取78例脑干胶质瘤患者，将患者按照7∶3随机分为训练集55例、验证集23例。其中训练集H3 K27M基因野生型18例(野生组)、H3 K27M基因突变型37例(突变组)；验证集H3 K27M基因野生型6例(野生组)、H3 K27M基因突变型17例(突变组)。术前均接受11C-MET PET扫描。在11C-MET PET图像上半自动勾画肿瘤ROI，测量最大肿瘤靶本底比值(TBRmax)、平均肿瘤靶本底比值(TBRmean)、肿瘤靶本底比峰值(TBRpeak)、肿瘤代谢体积(MTV)及病灶总蛋氨酸摄取量(TLMU)。提取ROI影像组学特征，采用最大相关最小冗余、最小绝对收缩和选择算法筛选影像组学特征，通过逻辑回归(LR)、随机森林、支持向量机、朴素贝叶斯和决策树五种机器学习方法构建预测脑干胶质瘤H3 K27M基因突变状态的影像组学模型。采用ROC曲线评估11C-MET PET代谢参数与各影像组学模型的预测性能。结果 在训练集中，突变组TBRmax、TBRmean及TBRpeak均显著低于野生组(P均<0.05)。在验证集中，突变组TBRmax显著低于野生组(P<0.05)。在训练集和验证集中，突变组和野生组的MTV和TMLU比较差异均无统计学意义(P均>0.05)。在训练集中，TBRmax、TBRmean、TBRpeak及TBRmax+TBRmean+TBRpeak多参数联合模型预测脑干胶质瘤H3 K27M基因突变状态的AUC分别为0.642、0.709、0.651、0.709，在验证集中的AUC分别为0.725、0.706、0.696、0.735。五种影像组学模型中，LR影像组学模型表现最优，在测试集和验证集中的AUC分别为0.875、0.794。训练集中LR影像组学模型的AUC高于TBRmax、TBRmean、TBRpeak及联合预测模型(P均<0.05)。结论 11C-MET PET代谢参数与影像组学模型均可有效预测脑干胶质瘤H3 K27M基因突变状态，其中LR影像组学模型具有最佳的预测性能。
Abstract：Objective To evaluate the value of semi-quantitative metabolic parameters and radiomics models derived from 11C-methionine(11C-MET) positron emission tomography(PET) in predicting H3 K27M gene mutation in brainstem glioma(BSG). Methods A total of 78 patients with BSG were enrolled and randomly divided into a training cohort(n=55) and a validation cohort(n=23) at a ratio of 7∶3. In the training cohort, 37 patients had H3 K27M mutation, and 18 were H3 K27M wildtype. In the validation cohort, 17 patients had H3 K27M mutation, and 6 were H3 K27M wildtype. Their preoperative 11C-MET PET images were analyzed. Regions of interest(ROIs) were semi-automatically delineated on 11C-MET PET. Maximum tumor-tobackground ratio(TBRmax), mean TBR(TBRmean), peak TBR(TBRpeak), metabolic tumor volume(MTV), and total lesion methionine uptake(TLMU) were measured. Radiomics features were extracted from the ROIs and selected using the minimum redundancy and maximum relevance(mRMR) and least absolute shrinkage and selection operator(LASSO). Five machine learning algorithms, i.e.,Logistic regression(LR), Random Forest, Support Vector Machine, Naive Bayes, and Decision Tree, were employed to develop radiomics models. The operating curve was used to evaluate the model performance. Results In the training cohort, TBRmax, TBRmean, and TBRpeak were significantly lower in H3 K27M mutant BSGs compared to wildtype tumor(all P<0. 05). In the validation cohort, TBRmax was significantly lower than in the wildtype group(P<0. 05). There was no statistically significant difference in MTV and TMLU between the mutant and wildtype groups in the training and validation cohorts(P>0. 05). In the training cohort, the multivariate radiomics model with TBRmax, TBRmean, TBRpeak, and TBRmax+TBRmean+TBRpeak achieved an AUC of 0. 642, 0. 709, 0. 651, and 0. 709, respectively, for the training set, and 0. 725, 0. 706, 0. 696, and 0. 735, respectively, for the validation set. Among the five radiomics models, the LR radiomics model performed best, with AUCs of 0. 875 and 0. 794 in the test set and validation set, respectively. The AUC of the LR radiomics model in the training set was higher than that of TBRmax, TBRmean, TBRpeak, and the combined prediction model(P<0. 05). Conclusion 11C-MET PET metabolic parameters and radiomics models can effectively predict the gene mutation status of brainstem glioma H3 K27M, with the LR radiomics model exhibiting the best predictive performance.
[1]UPPAR A M, SUGUR H, PRABHURAJ A R, et al. H3 K27M,IDH1, and ATRX expression in pediatric GBM and their clinical and prognostic significance[J]. Childs Nerv Syst, 2019, 35(9):1537-1545.
[2]LO GRECO M C, MARANO G, LA ROCCA M, et al. Latest advancements in the management of H3 K27M-mutant diffuse intrinsic pontine glioma:a narrative review[J]. Cancers(Basel),2025, 17(3):420.
[3]ZHAO X B, LI D L, QIAO Z, et al. 11C-methionine PET imaging characteristics in children with diffuse intrinsic pontine gliomas and relationship to survival and H3 K27M mutation status[J]. Eur J Nucl Med Mol Imaging, 2023, 50(6):1709-1719.
[4]赵晓斌，武玉亮，李德岭，等. 11C-蛋氨酸PET/CT在脑干胶质瘤诊断及治疗中的价值[J].中华神经外科杂志，2018, 34(4):338-343.
[5]WU C Q, ZHENG H, LI J N, et al. MRI-based radiomics signature and clinical factor for predicting H3 K27M mutation in pediatric high-grade gliomas located in the midline of the brain[J].Eur Radiol, 2022, 32(3):1813-1822.
[6]LOUIS D N, PERRY A, REIFENBERGER G, et al. The 2016world health organization classification of tumors of the central nervous system:a summary[J]. Acta Neuropathol, 2016, 131(6):803-820.
[7]LAW I, ALBERT N L, ARBIZU J, et al. Joint EANM/EANO/RANO practice guidelines/SNMMI procedure standards for imaging of gliomas using PET with radiolabelled amino acids and[18F]FDG:version 1.0[J]. Eur J Nucl Med Mol Imaging,2019, 46(3):540-557.
[8]GIELEN G H, GESSI M, HAMMES J, et al. H3F3A K27M mutation in pediatric CNS tumors:a marker for diffuse high-grade astrocytomas[J]. Am J Clin Pathol, 2013, 139(3):345-349.
[9]WANG Y, PAN C C, XIE M G, et al. Adult diffuse intrinsic pontine glioma:clinical, radiological, pathological, molecular features, and treatments of 96 patients[J]. J Neurosurg, 2022,137(6):1628-1638.
[10]VENNETI S, KAWAKIBI A R, JI S, et al. Clinical efficacy of ONC201 in H3 K27M-mutant diffuse midline gliomas is driven by disruption of integrated metabolic and epigenetic pathways[J]. Cancer Discov, 2023, 13(11):2370-2393.
[11]KIM D, CHUN J H, KIM S H, et al. Re-evaluation of the diagnostic performance of 11C-methionine PET/CT according to the2016 WHO classification of cerebral gliomas[J]. Eur J Nucl Med Mol Imaging, 2019, 46(8):1678-1684.
[12]CHENG Y, SONG S S, WEI Y K, et al. Glioma imaging by O-(2-18F-Fluoroethyl)-L-Tyrosine PET and diffusion-weighted MRI and correlation with molecular phenotypes, validated by PET/MR-guided biopsies[J]. Front Oncol, 2021, 11:743655.
[13]PICCARDO A, TORTORA D, MASCELLI S, et al. Advanced MR imaging and 18F-DOPA PET characteristics of H3 K27Mmutant and wild-type pediatric diffuse midline gliomas[J]. Eur J Nucl Med Mol Imaging, 2019, 46(8):1685-1694.
[14]OKUBO S, ZHEN H N, KAWAI N, et al. Correlation of Lmethyl-11C-methionine(MET)uptake with L-type amino acid transporter 1 in human gliomas[J]. J Neurooncol, 2010, 99(2):217-225.
[15]BEURIAT P A, FLAUS A, PORTEFAIX A, et al. Preoperative11C-methionine PET-MRI in pediatric infratentorial tumors[J].Clin Nucl Med, 2024, 49(5):381-386.
[16]RHEIMS S, RUBI S, BOUVARD S, et al. Accuracy of distinguishing between dysembryoplastic neuroepithelial tumors and other epileptogenic brain neoplasms with[11C]methionine PET[J]. Neuro Oncol, 2014, 16(10):1417-1426.
[17]FALK DELGADO A, FALK DELGADO A. Discrimination between primary low-grade and high-grade glioma with 11Cmethionine PET:a bivariate diagnostic test accuracy metaanalysis[J]. Br J Radiol, 2018, 91(1082):20170426.
[18]ZHUO Z Z, QU L Y, ZHANG P, et al. Prediction of H3 K27Mmutant brainstem glioma by amide proton transfer-weighted imaging and its derived radiomics[J]. Eur J Nucl Med Mol Imaging, 2021, 48(13):4426-4436.
[19]PAN C C, LIU J, TANG J, et al. A machine learning-based prediction model of H3 K27M mutations in brainstem gliomas using conventional MRI and clinical features[J]. Radiother Oncol,2019, 130:172-179.
[20]ZHOU W Y, WEN J B, HUANG Q, et al. Development and validation of clinical-radiomics analysis for preoperative prediction of IDH mutation status and WHO grade in diffuse gliomas:a consecutive L-[methyl-11C] methionine cohort study with two PET scanners[J]. Eur J Nucl Med Mol Imaging, 2024, 51(5):1423-1435.
[21]BAI J, CUI B X, LI F Q, et al. Multiparametric radiomics signature for predicting molecular genotypes in adult-type diffuse gliomas utilizing 18F-FET PET/MRI[J]. BMC Med Imaging, 2025,25(1):187.
[22]LEIBETSEDER A, LEITNER J, MAIR M J, et al. Prognostic factors in adult brainstem glioma:a tertiary care center analysis and review of the literature[J]. J Neurol, 2022, 269(3):1574-1590.
基本信息:
DOI：10.20258/j.cnki.1006-9011.2026.05.001
中图分类号:R739.41;R730.44
引用信息:
[1]张姝,赵晓斌,李晓桐,等.基于~(11)C-蛋氨酸PET代谢参数及影像组学模型对脑干胶质瘤H3 K27M基因突变状态的预测价值[J].医学影像学杂志,2026,36(05):1-6+12.DOI:10.20258/j.cnki.1006-9011.2026.05.001.
基金信息:
国家自然科学基金资助项目(编号：81771143)
2026-05-30
2026-05-30