Multiparameter magnetic resonance imaging-based radiomics models for the preoperative prediction of microsatellite instability status in endometrioid adenocarcinoma
Zhao Jintao1,2, Cui Yanfen2, Jia Yaju2, Ren Jialiang3, Hou Lina2
1Department of Medical Imaging, Shanxi Medical University, Taiyuan 030000, China; 2Department of Radiology, Shanxi Province Cancer Hospital/ Shanxi Hospital Affiliated to Cancer Hospital, Chinese Academy of Medical Sciences/ Cancer Hospital Affiliated to Shanxi Medical University, Taiyuan 030000, China; 3GE Healthcare China, Beijing 100176, China
摘要目的 探讨基于多参数MRI影像组学模型术前预测子宫内膜样腺癌患者微卫星不稳定(MSI)状态的应用价值。方法 回顾性队列研究。纳入2017年1月—2021年12月山西省肿瘤医院经病理证实为子宫内膜样腺癌的患者225例。患者年龄29~78(55.1±7.5)岁,均行全子宫+双附件切除术,并于术前行盆腔多参数MRI检查。将225例患者按7∶3的比例随机分为训练集(158例)和验证集(67例),根据手术标本病理免疫组织化学检查结果分为高度微卫星不稳定(MSI-H)组58例和低度微卫星不稳定/微卫星稳定(MSI-L/MSS)组167例。对每例患者的快速自旋回波压脂序列T2加权像、增强T1加权像、表观弥散系数(ADC)3个序列进行图像分割和特征提取,在训练集中采用组内相关系数(ICC)、Mann-Whitney U检验、Spearman相关及Boruta算法进行图像特征筛选,并使用朴素贝叶斯(NB)、随机森林(RF)及支持向量机(SVM)3种分类器构建影像组学模型。在训练集、验证集中分别采用受试者工作特征(ROC)曲线、Delong检验、决策分析曲线(DCA)评估并验证3种模型对MSI状态的诊断性能、预测性能、净收益。将筛选出的影像组学特征纳入诊断效能较高的最佳模型中,对每例患者进行影像组学评分,分别在训练集、验证集中评估MSI-H组和MSI-L/MSS组影像组学评分的分布情况。结果 训练集与验证集间比较,患者肿瘤分化程度差异有统计学意义(χ2=8.40,P=0.015),患者年龄、体质量指数、国际妇产科联盟分期,以及肌层浸润深度等临床病理资料差异均无统计学意义(P值均>0.05)。在训练集中,MSI-H组与MSI-L/MSS组比较,患者是否绝经2组间差异有统计学意义(χ2=4.56,P=0.033),其他临床病理资料2组间差异均无统计学意义(P值均>0.05);在验证集中,2组间各临床病理差异均无统计学意义(P值均>0.05)。多参数MRI图像经特征提取后,每例患者获得4 245个影像组学特征,在训练集中筛选出6个关键特征构建模型。在训练集中NB模型、RF模型及SVM模型预测子宫内膜样腺癌患者MSI状态的ROC曲线下面积(AUC)分别为0.764[95%可信区间(CI)0.682~0.846]、0.821(95% CI 0.751~0.892)和0.905(95% CI 0.848~0.961),验证集中AUC分别为0.712(95% CI 0.568~0.856)、0.812(95% CI 0.710~0.915)、0.875(95% CI 0.762~0.988),结果表明SVM模型的诊断性能最佳。Delong检验结果显示,训练集中SVM模型与NB模型、RF模型AUC间差异均有统计学意义(Z=-3.45、-2.33,P值均<0.05),而在验证集中AUC差异均无统计学意义(P值均>0.05)。 DCA曲线显示,在训练集和验证集中,SVM模型预测MSI状态的净收益均优于NB模型、RF模型,也优于将所有患者MSI状态都看作MSI-H或MSI-L/MSS。 采用最佳模型SVM模型对每例患者进行影像组学评分,训练集及验证集中MSI-H组与MSI-L/MSS组影像组学评分分布有明显差异。结论 基于多参数MRI的SVM模型对于子宫内膜样腺癌患者MSI状态具有一定的术前预测价值。
Abstract:Objective This study aimed to evaluate the application value of multiparameter magnetic resonance imaging (MRI)-based radiomics models for the preoperative prediction of microsatellite instability (MSI) status in patients with endometrioid adenocarcinoma. Methods This retrospective cohort study included 225 patients, with an average age of 29-78 (55.1±7.5) years, with pathologically proven endometrioid adenocarcinoma in Shanxi Cancer Hospital between January 2017 and December 2021. All patients underwent pelvic MRI scans before complete hysterectomy with oophorotomy and were randomly divided into training set (158 cases) and validation set (67 cases) at a ratio of 7∶3. Patients were divided into MSI-H group (58 cases) and MSI-L/MSS group (167 cases) according to pathological IHC results. Image segmentation and features were extracted on T2-weighted imaging fat suppression, contrast-enhanced T1WI images, and apparent diffusion coefficient (ADC) maps for each patient. In the training set, intraclass correlation coefficient (ICC), Mann-Whitney U test, Spearman correlation analysis, and Boruta algorithm were used for feature selection. Three classification algorithms including naive Bayesian algorithm (NB), random forest (RF), and support vector machine (SVM) were applied to build radiomics models. The diagnostic performance, predictive performance, net benifit, and reliability of the three radiomics models were tested by ROC curve, Delong test, and decision analysis curve (DCA) and verified in the validation set. The selected imaging features were included in the best model, and radscore measurement was performed for each patient. The distribution of radscore in the MSI-H and MSI-L/MSS groups was evaluated in the training and validation sets, respectively. Results Between the training and validation sets, the difference between patients' tumor grade was significant (χ2=8.40, P=0.015) but not between other indicators such as age, body mass index, Federation of Gynecology and Obstetrics, and depth of myometrial invasion (all P values > 0.05). In the training set, patients before or after menopause had statistically significant difference between the MSI-H and MSI-L/MSS groups (χ2=4.56, P=0.033). The other indicators were not statistically different between the two groups (all P values > 0.05). In the validation set, clinical and pathological data were not different between the two groups (all P values > 0.05). After feature extraction of multiparameter MRI images, 4 245 imaging features were obtained for each patient. Six key radiomics features were selected for model building. The area under curve (AUC)of the NB, RF, and SVM models were 0.764 (95% credibility interval [CI] 0.682-0.846), 0.821 (95% CI 0.751-0.892), 0.905 (95% CI 0.848-0.961) and 0.712 (95% CI 0.568-0.856) as well as 0.812 (95% CI 0.710-0.915), 0.875 (95% CI 0.762-0.988) in the training and validation sets, respectively. The SVM model had the best performance in the training and validation sets. Delong test results showed that in the training set, the AUC difference in the SVM, NB, and RF models was statistically significant (Z=-3.45, -2.33, all P values < 0.05), and that in the validation set was not statistically significant (all P values > 0.05). The DCA results showed that the SVM model achieved higher net benefits than the other models and was better than treating all patients as MSI-H or MSI-L/MSS in the training and validation sets. The best model SVM was used for radscore measurement for each patient, and no significant differences in the distribution of radscore were found between the MSI-H and MSI-L/MSS groups in the training and validation sets. Conclusion The SVM model based on multi-parameter MRI can be used for the preoperative evaluation of MSI status in patients with endometrioid adenocarcinoma.
赵锦瑫, 崔艳芬, 贾亚菊, 任嘉梁, 侯丽娜. 基于多参数MRI的影像组学模型对子宫内膜样腺癌微卫星不稳定状态的术前预测价值[J]. 中华解剖与临床杂志, 2024, 29(2): 88-96.
Zhao Jintao, Cui Yanfen, Jia Yaju, Ren Jialiang, Hou Lina. Multiparameter magnetic resonance imaging-based radiomics models for the preoperative prediction of microsatellite instability status in endometrioid adenocarcinoma. Chinese Journal of Anatomy and Clinics, 2024, 29(2): 88-96.
Guo YE, Liu Y, Zhang W, et al.The clinicopathological characteristics, prognosis and immune microenvironment mapping in MSI-H/MMR-D endometrial carcinomas[J]. Discov Oncol, 2022,13(1):12. DOI: 10.1007/s12672-022-00466-5.
[3]
齐月,许天敏,赵丽苹,等.不同分子学筛查方法在Lynch综合征相关子宫内膜癌患者中的对比分析[J].中国妇产科临床杂志, 2022, 23(2): 116-120. DOI: 10.13390/j.issn.1672-1861.2022.02.002.Qi Y, Xu TM, Zhao LP, et al.Comparison of different molecular screening methods for Lynch syndrome in patients with endometrial cancer[J]. Chin J Clin Obstet Gynecol, 2022, 23(2): 116-120. DOI: 10.13390/j.issn.1672-1861.2022.02.002.
[4]
Oaknin A, Bosse TJ, Creutzberg CL, et al.Endometrial cancer: ESMO clinical practice guideline for diagnosis, treatment and follow-up[J]. Ann Oncol, 2022,33(9):860-877. DOI: 10.1016/j.annonc.2022.05.009.
Concin N, Matias-Guiu X, Vergote I, et al.ESGO/ESTRO/ESP guidelines for the management of patients with endometrial carcinoma[J]. Radiother Oncol, 2021,154:327-353. DOI: 10.1016/j.radonc.2020.11.018.
[7]
Olave MC, Graham RP.Mismatch repair deficiency: the what, how and why it is important[J]. Genes Chromosomes Cancer, 2022,61(6):314-321. DOI: 10.1002/gcc.23015.
[8]
Deshpande M, Romanski PA, Rosenwaks Z, et al.Gynecological cancers caused by deficient mismatch repair and microsatellite instability[J]. Cancers (Basel), 2020,12(11):3319. DOI: 10.3390/cancers12113319.
[9]
Karpel H, Slomovitz B, Coleman RL, et al.Biomarker-driven therapy in endometrial cancer[J]. Int J Gynecol Cancer, 2023,33(3):343-350. DOI: 10.1136/ijgc-2022-003676.
[10]
Rizzo A.Immune checkpoint inhibitors and mismatch repair status in advanced endometrial cancer: elective affinities[J]. J Clin Med, 2022,11(13): 3912. DOI: 10.3390/jcm11133912.
[11]
O'Malley DM, Bariani GM, Cassier PA, et al. Pembrolizumab in patients with microsatellite instability-high advanced endometrial cancer: results from the KEYNOTE-158 study[J]. J Clin Oncol, 2022, 40(7): 752-761. DOI: 10.1200/JCO.21.01874.
[12]
王芳,夏雨薇,柴象飞,等.影像组学分析流程及临床应用的研究进展[J].中华解剖与临床杂志,2021,26(2):236-241. DOI: 10.3760/cma.j.cn101202-20200701-00200.Wang F, Xia YW, Chai XF, et al.Analysis process and clinical application of radiomics[J]. Chin J Anat Clin, 2021,26(2):236-241. DOI: 10.3760/cma.j.cn101202-20200701-00200.
[13]
Yan BC, Li Y, Ma FH, et al.Radiologists with MRI-based radiomics aids to predict the pelvic lymph node metastasis in endometrial cancer: a multicenter study[J]. Eur Radiol, 2021,31(1):411-422. DOI: 10.1007/s00330-020-07099-8.
[14]
贾亚菊,杨晓棠,崔艳芬,等. 基于多参数MR的影像组学融合模型术前预测宫颈鳞癌脉管浸润的应用价值[J].中华解剖与临床杂志, 2022,27(11) : 737-744. DOI: 10.3760/cma.j.cn101202-20220307-00070.Jia YJ, Yang XT, Cui YF, et al.Multiparameter magnetic resonance imaging-based radiomics model for the preoperative prediction of lymph-vascular space invasion in cervical squamous carcinoma[J]. Chin J Anat Clin, 2022,27(11) : 737-744. DOI: 10.3760/cma.j.cn101202-20220307-00070.
[15]
Long L, Sun J, Jiang L, et al.MRI-based traditional radiomics and computer-vision nomogram for predicting lymphovascular space invasion in endometrial carcinoma[J]. Diagn Interv Imaging, 2021,102(7-8):455-462. DOI: 10.1016/j.diii.2021.02.008.
[16]
Zheng T, Yang L, Du J, et al.Combination analysis of a radiomics-based predictive model with clinical indicators for the preoperative assessment of histological grade in endometrial carcinoma[J]. Front Oncol, 2021,11:582495. DOI: 10.3389/fonc.2021.582495.
[17]
Zhang K, Zhang Y, Fang X, et al.Nomograms of combining apparent diffusion coefficient value and radiomics for preoperative risk evaluation in endometrial carcinoma[J]. Front Oncol, 2021, 11: 705456. DOI: 10.3389/fonc.2021.705456.
[18]
Blüthgen C, Patella M, Euler A, et al.Computed tomography radiomics for the prediction of thymic epithelial tumor histology, TNM stage and myasthenia gravis[J]. PLoS One, 2021, 16(12): e0261401. DOI: 10.1371/journal.pone.0261401.
[19]
Abu-Rustum N, Yashar C, Arend R, et al.Uterine neoplasms, version 1.2023, NCCN clinical practice guidelines in oncology[J]. J Natl Compr Canc Netw, 2023, 21(2): 181-209. DOI: 10.6004/jnccn.2023.0006.
[20]
Li Y, Liu X, Wang X, et al.Using amide proton transfer-weighted MRI to non-invasively differentiate mismatch repair deficient and proficient tumors in endometrioid endometrial adenocarcinoma[J]. Insights Imaging, 2021, 12(1): 182. DOI: 10.1186/s13244-021-01126-y.
[21]
田士峰,刘爱连,陈丽华,等.酰胺质子转移成像和扩散峰度成像评估子宫内膜癌微卫星不稳定状态[J].中国临床医学影像杂志, 2022, 33(5): 345-349. DOI:10.12117/jccmi.2022.05.009.Tian SF, Liu AL, Chen LH, et al.Evaluation of microsatellite instability in endometrial carcinoma by amide proton transfer imaging and diffusion kurtosis imaging[J]. Chin Clin Med Imaging, 2022, 33(5): 345-349. DOI:10.12117/jccmi.2022.05.009.
[22]
马长军, 田士峰, 陈丽华, 等. T2 mapping联合mDixon-Quant多参数成像定量评估子宫内膜癌微卫星不稳定状态的研究[J].磁共振成像, 2022, 13(8): 48-54. DOI: 10.12015/issn.1674-8034.2022.08.009.Ma CJ, Tian SF, Chen LH, et al.Quantitative assessment of microsatellite instability in endometrial cancer by T2 mapping combined with mDixon-Quant multiparameter imaging[J]. Chin J Magn Reson Imaging, 2022, 13(8): 48-54. DOI: 10.12015/issn.1674-8034.2022.08.009.
[23]
Bhosale P, Ramalingam P, Ma J, et al.Can reduced field-of-view diffusion sequence help assess microsatellite instability in FIGO stage 1 endometrial cancer?[J]. J Magn Reson Imaging, 2017, 45(4):1216-1224. DOI: 10.1002/jmri.25427.
[24]
Wang H, Xu Z, Zhang H, et al.The value of magnetic resonance imaging-based tumor shape features for assessing microsatellite instability status in endometrial cancer[J]. Quant Imaging Med Surg, 2022,12(9):4402-4413. DOI: 10.21037/qims-22-77.
[25]
李丹丹, 崔艳芬, 杨晓棠. 基于MR T2加权成像的影像组学标签预测直肠癌KRAS基因突变的价值[J].中华解剖与临床杂志, 2021,26(1): 7-14.DOI: 10.3760/cma.j.cn101202-20200407-00116.Li DD, Cui YF, Yang XT.Radiomic signature based on MR T2-weighted images for predicting KRAS mutation statue in rectal cancer[J]. Chin J Anat Clin, 2021,26(1): 7-14.DOI: 10.3760/cma.j.cn101202-20200407-00116.
[26]
Zhang W, Yin H, Huang Z, et al.Development and validation of MRI-based deep learning models for prediction of microsatellite instability in rectal cancer[J]. Cancer Med, 2021,10(12):4164-4173. DOI: 10.1002/cam4.3957.
[27]
Pei Q, Yi X, Chen C, et al.Pre-treatment CT-based radiomics nomogram for predicting microsatellite instability status in colorectal cancer[J]. Eur Radiol, 2022,32(1):714-724. DOI: 10.1007/s00330-021-08167-3.
[28]
Li Z, Zhong Q, Zhang L, et al.Computed tomography-based radiomics model to preoperatively predict microsatellite instability status in colorectal cancer: a multicenter study[J]. Front Oncol, 2021,11:666786. DOI: 10.3389/fonc.2021.666786.
[29]
Lin Z, Wang T, Li H, et al.Magnetic resonance-based radiomics nomogram for predicting microsatellite instability status in endometrial cancer[J]. Quant Imaging Med Surg, 2023,13(1):108-120. DOI: 10.21037/qims-22-255.
[30]
潘美宇,蔡望洲,陈亮,等.基于临床-影像组学建立子宫内膜癌微卫星不稳定性的评估模型[J].分子影像学杂志,2023,46(1):48-52. DOI:10.12122/j.issn.1674-4500.2023.01.09.Pan MY, Cai WZ, Chen L, et al.Establishment of microsatellite instability assessment model for endometrial carcinoma by clinic-imaging omics[J]. Journal of Molecular Imaging, 2023,46(1):48-52. DOI:10.12122/j.issn.1674-4500.2023.01.09.
[31]
彭涛,肖建明,张仕慧,等.基于多参数MRI及影像组学建立机器学习模型诊断临床显著性前列腺癌[J].中国医学影像技术,2019,35(10):1526-1530.DOI:10.13929/j.1003-3289.201902129.Peng T,Xiao JM,Zhang SH,et al.Establishment of machine learning models for diagnosis of clinically significant prostate cancer based on multi-parameter MRI and radiomics[J]. Chin J Med Imaging Technol, 2019,35(10):1526-1530. DOI:10.13929/j.1003-3289.201902129.
[32]
Zheng Y, Zhou D, Liu H, et al.CT-based radiomics analysis of different machine learning models for differentiating benign and malignant parotid tumors[J]. Eur Radiol, 2022,32(10):6953-6964. DOI: 10.1007/s00330-022-08830-3.
[33]
Mainenti PP, Stanzione A, Cuocolo R, et al.MRI radiomics: a machine learning approach for the risk stratification of endometrial cancer patients[J]. Eur J Radiol, 2022,149:110226. DOI: 10.1016/j.ejrad.2022.110226.
[34]
Song XL, Luo HJ, Ren JL, et al.Multisequence magnetic resonance imaging-based radiomics models for the prediction of microsatellite instability in endometrial cancer[J]. Radiol Med, 2023,128(2):242-251. DOI: 10.1007/s11547-023-01590-0.
[35]
Lafata KJ, Wang Y, Konkel B, et al.Radiomics: a primer on high-throughput image phenotyping[J]. Abdom Radiol (NY), 2022,47(9):2986-3002. DOI: 10.1007/s00261-021-03254-x.
[36]
Cao Y, Zhang G, Zhang J, et al.Predicting microsatellite instability status in colorectal cancer based on triphasic enhanced computed tomography radiomics signatures: a multicenter study[J]. Front Oncol, 2021,11: 687771. DOI: 10.3389/fonc.2021.687771.