| 20 | 0 | 40 |
| 下载次数 | 被引频次 | 阅读次数 |
目的：利用肉瘤血清样本内微小RNA(microRNA,miRNA)间的相对表达秩序关系识别肉瘤特异miRNA诊断标志物并构建诊断模型。方法：分析了GEO数据库中6个数据集的8 492例血清miRNA样本（包括肉瘤、非癌对照以及12种其他癌型），以癌型覆盖最广、样本分布相对均衡的GSE113486为训练集，其余数据集为测试集。基于样本内miRNA间相对表达秩序关系筛选肉瘤特异miRNA标志物对（miRNA pair,miRPair），结合最小绝对收缩和选择算子（Least absolute shrinkage and selection operator,LASSO）、随机森林（Random forest,RF）、极限梯度提升（Extreme gradient boosting,XGBoost）等5种机器学习算法构建肉瘤诊断模型，并采用受试者工作特征曲线（Receiver operating characteristic curve,ROC）的曲线下面积（Area under curve,AUC）、准确率、敏感度与特异度等指标进行评估。结果：以GSE113486的非癌和肉瘤样本作为训练集样本，根据miRPair相对表达秩序变化，筛选了9个miRPair作为肉瘤诊断模型。该模型在训练集与2个独立测试集（GSE112264和GSE106817）中区分非癌和肉瘤样本的AUC均超过0.978，诊断准确率、敏感度与特异度均高于95%。然而，该模型难以有效鉴别其他癌型，将12种其他癌症中超过99%的样本误判为肉瘤，缺乏肉瘤特异性。基于数据集GSE113486中12种其他癌型与肉瘤样本构建肉瘤特异诊断模型。通过比较2组间血清miRNA相对表达秩序关系，筛选出399个肉瘤特异性miRPair，基于这些miRPair，采用5种机器学习算法构建肉瘤特异诊断模型，除RF模型中有1个测试集的诊断准确率为78.0%，其余模型在5套独立测试集的诊断准确率均达到85.3%以上，具有良好的肉瘤诊断能力。通路富集分析发现，肉瘤特异的miRNA主要富集在精氨酸生物合成等相关生物通路中。结论：血清中存在肉瘤特异的miRNA相对表达秩序关系，基于此开发的肉瘤特异诊断模型具有良好的诊断效能和跨数据集稳健性。该研究为进一步开发临床可用的肉瘤特异诊断生物标志物提供了理论基础。
Abstract：Objective : To identify sarcoma-specific miRNA diagnostic biomarkers and construct a diagnostic model by using the relative expression orderings between microRNAs(miRNAs) in sarcoma serum samples. Methods : A total of 8 492 serum miRNA samples(including sarcoma, non-cancer controls, and 12 other cancer types) from six datasets in the gene expression omnibus(GEO) database were analyzed. The GSE113486 with the most extensive coverage and relatively balanced sample distribution was used as the training set, and the remaining datasets were used as the test set. Sarcoma-specific miRNA pairs(miRPair) were screened based on the with-sample relative expression orderings between miRNAs. Five machine learning algorithms, including least absolute shrinkage and selection operator(LASSO), random forest(RF), and extreme gradient boosting(XGBoost), were combined to construct a sarcoma diagnosis model. Model performance was evaluated using the area under the curve(AUC), accuracy, sensitivity, and specificity. Results : Using non-cancer and sarcoma samples of GSE113486 as training samples, 9 miRPairs were identified as a diagnostic model according to the changes in the relative expression orderings of miRPairs. The AUC values of the model for distinguishing non-cancerous samples and sarcoma samples in both the training set and the two independent test sets, GSE112264 and GSE106817, were all more than 0. 978, and the diagnostic accuracy, sensitivity, and specificity were above 95%. However, this model is difficult to effectively identify other cancer types, and erroneously classified over 99% of samples from 12 other cancer types as sarcoma. A sarcoma-specific diagnostic model was constructed based on 12 other cancer types and sarcoma samples in the data set GSE113486. By comparing the relative expression orderings of serum miRNAs between the two groups, 399 sarcoma-specific miRPair were identified. Based on these miRPairs, five machine learning algorithms were used to construct sarcoma-specific diagnostic models. Apart from RF with an accuracy of 78. 0% in one dataset, all other models achieved diagnostic accuracy over 85. 3% on five independent test sets, demonstrating favorable predictive performance, demonstrating favorable predictive performance. Pathway enrichment analysis indicated that these sarcoma-specific miRNA were mainly enriched in pathways related to arginine biosynthesis. Conclusion : Sarcoma-specific relative expression orderings among serum miRNAs were identified. Diagnostic models developed based on these orderings exhibited robust diagnostic performance and strong cross-dataset performance. This study provides a theoretical foundation for the further development of clinically applicable sarcoma-specific diagnostic biomarkers.
[1]Siegel R L,Giaquinto A N,Jemal A. Cancer statistics,2024[J]. CA Cancer J Clin,2024,74(1):12-49.
[2]Grünewald T G,Alonso M,Avnet S,et al. Sarcoma treatment in the era of molecular medicine[J]. EMBO Mol Med,2020,12(11):EMMM201911131.
[3]Gao W,Chen Y,Yang J,et al. Clinical perspectives on liquid biopsy in metastatic colorectal cancer[J]. Front Genet,2021,12:634642.
[4]Li X,Seebacher N A,Hornicek F J,et al. Application of liquid biopsy in bone and soft tissue sarcomas:present and future[J]. Cancer Lett,2018,439:66-77.
[5]Stiller C A ,Trama A ,Serraino D ,et al. Descriptive epidemiology of sarcomas in Europe:report from the RARECARE project[J]. Eur J Cancer,2013,49(3):684-695.
[6]Eid J E,Garcia C B. Reprogramming of mesenchymal stem cells by oncogenes[J]. Semin Cancer Biol,2015,32:18-31.
[7]刘梦莹，蔡浩，何明，等.液体活检技术在乳腺癌领域中的研究及应用现状[J].赣南医科大学学报，2025,45(1):11-17.
[8]Chen X,Ba Y,Ma L,et al. Characterization of microRNAs in serum:a novel class of biomarkers for diagnosis of cancer and other diseases[J]. Cell Res,2008,18(10):997-1006.
[9]Nakamura K,Hernández G,Sharma G G,et al. A liquid biopsy signature for the detection of patients with earlyonset colorectal cancer[J]. Gastroenterology,2022,163(5):1242-1251. e2.
[10]Wang X,Jia Z,Shi H,et al. Identification and evaluation of 2 circulating microRNAs for non-small cell lung cancer diagnosis[J]. Clin Exp Pharmacol Physiol,2018,45(10):1083-1086.
[11]Kachakova D,Mitkova A,Popov E,et al. Combinations of serum prostate-specific antigen and plasma expression levels of let-7c,miR-30c,miR-141,and miR-375 as potential better diagnostic biomarkers for prostate cancer[J]. DNA Cell Biol,2015,34(3):189-200.
[12]Cong C,Wang W,Tian J,et al. Identification of serum miR-124 as a biomarker for diagnosis and prognosis in osteosarcoma[J]. Cancer Biomark,2018,21(2):449-454.
[13]Liu W,Zhao X,Zhang Y J,et al. MicroRNA-375 as a potential serum biomarker for the diagnosis,prognosis,and chemosensitivity prediction of osteosarcoma[J]. J Int Med Res,2018,46(3):975-983.
[14]Li M,Li H,Hong G,et al. Identifying primary site of lung-limited cancer of unknown primary based on relative gene expression orderings[J]. BMC Cancer,2019,19(1):67.
[15]Wu P,Li D,Zhang C,et al. A unique circulating microRNA pairs signature serves as a superior tool for early diagnosis of pan-cancer[J]. Cancer Lett,2024,588:216655.
[16]Li H,Zheng T,Chen B,et al. Similar blood-borne DNA methylation alterations in cancer and inflammatory diseases determined by subpopulation shifts in peripheral leukocytes[J]. Br J Cancer,2014,111(3):525-531.
[17]Hong G,Chen B,Li H,et al. Similar source of differential blood mRNAs in lung cancer and pulmonary inflammatory diseases:calls for improved strategy for identifying cancer-specific biomarkers[J]. PLoS One,2014,9(9):e108104.
[18]Hong G ,Li H ,Zhang J ,et al. Identifying diseaseassociated pathways in one-phenotype data based on reversal gene expression orderings[J]. Sci Rep,2017,7:1348.
[19]Li H,Ma L,Luo F,et al. Construct of qualitative diagnostic biomarkers specific for glioma by pairing serum microRNAs[J]. BMC Genomics,2023,24(1):96.
[20]Breiman L. Random forests[J]. Mach Learn,2001,45(1):5-32.
[21]Tibshirani R. Regression shrinkage and selection via the Lasso[J]. J R Stat Soc Ser B Stat Methodol,1996,58(1):267-288.
[22]Chen T,Guestrin C. XGBoost:a scalable tree boosting system[PP/OL]. V3. arVix(2016-06-10)[2026-02-04].https：//arxiv. org/pdf/1603. 02754.
[23]Charbuty B ,Abdulazeez A. Classification based on decision tree algorithm for machine learning[J]. J Appl Sci Technol Trends,2021,2(1):20-28.
[24]Engebretsen S,Bohlin J. Statistical predictions with glmnet[J]. Clin Epigenetics,2019,11(1):123.
[25]Alderden J,Pepper G A,Wilson A,et al. Predicting pressure injury in critical care patients:a machinelearning model[J]. Am J Crit Care,2018,27(6):461-468.
[26]Loh W Y. Classification and regression trees[J]. WIREs Data Min&Knowl,2011,1(1):14-23.
[27]Ritchie M E,Phipson B,Wu D,et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies[J]. Nucleic Acids Res,2015,43(7):e47.
[28]Aparicio-Puerta E,Hirsch P,Schmartz G P,et al.miEAA 2023:updates,new functional microRNA sets and improved enrichment visualizations[J]. Nucleic Acids Res,2023,51(W1):W319-W325.
[29]Zhang A ,Hu H. A novel blood-based microRNA diagnostic model with high accuracy for multi-cancer early detection[J]. Cancers,2022,14(6):1450.
[30]Jin Z,Liu S,Zhu P,et al. A novel serum miRNA-pair classifier for diagnosis of sarcoma[J]. PLoS One,2020,15(7):e0236097.
[31]Bean G R,Kremer J C,Prudner B C,et al. A metabolic synthetic lethal strategy with arginine deprivation and chloroquine leads to cell death in ASS1-deficient sarcomas[J]. Cell Death Dis,2016,7(10):e2406.
[32]Lyu G,Li D. Unveiling the role of DPYS:a new prognostic biomarker in sarcoma[J]. Curr Protein Pept Sci,2025,26(9):761-772.
[33]Shi F, Li L, Cheng Y. FAT10 stimulates the development of osteosarcoma by regulating the JAK/STAT signaling pathway[J]. J BUON,2021,26(5):2090-2096.
[34]Hong G ,Li H ,Li M ,et al. A simple way to detect disease-associated cellular molecular alterations from mixed-cell blood samples[J]. Brief Bioinform,2018,19(4):613-621.
基本信息:
中图分类号:R738.7
引用信息:
[1]王玲琍,阮佳怡,何云龙,等.基于血清样本内miRNA表达秩序特征识别肉瘤特异miRNA标志物[J].赣南医科大学学报().
基金信息:
江西省高层次人才创新创业“千人计划”(Jxsq2020101096)
2026-06-25
2026-06-25
2026-06-25