Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature
OBJECTIVES: MicroRNAs (miRNAs) play a key role in governing posttranscriptional regulation through binding to the mRNAs of target genes. This study is to assess miRNAs expression profiles for identifying brain metastasis-related miRNAs to develop the predictive model by microarray in tumor tissues....
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2018-02-01
|
Series: | Translational Oncology |
Online Access: | http://www.sciencedirect.com/science/article/pii/S1936523317303625 |
_version_ | 1811308611641540608 |
---|---|
author | Shuangtao Zhao Jiangyong Yu Luhua Wang |
author_facet | Shuangtao Zhao Jiangyong Yu Luhua Wang |
author_sort | Shuangtao Zhao |
collection | DOAJ |
description | OBJECTIVES: MicroRNAs (miRNAs) play a key role in governing posttranscriptional regulation through binding to the mRNAs of target genes. This study is to assess miRNAs expression profiles for identifying brain metastasis-related miRNAs to develop the predictive model by microarray in tumor tissues. METHODS: For this study, we screened the significant brain metastasis-related miRNAs from 77 lung adenocarcinoma (LUAD) patients with brain metastasis (BM+) or non-brain metastasis (BM−). A predictive model was developed from the training set (n = 42) using a random Forest supervised classification algorithm and a Class Centered Method, and then validated in a test set (n = 35) and further analysis in GSE62182 (n = 73). The independence of this signature in BM prediction was measured by multivariate logistic regression analysis. RESULTS: From the training set, the predictive model (including hsa-miR-210, hsa-miR-214 and hsa-miR-15a) stratified the patients into two groups with significantly different BM subtypes (90.4% of accuracy). The similar predictive power (91.4% of accuracy) was obtained in the test cohort. As an independent predictive factor, it was closely associated with BM and had high sensitivity and specificity in predicting BM in clinical practice. Moreover, functional enrichment analysis demonstrated that this signature involved in the signaling pathways positively correlated with cancer metastasis. CONCLUSION: These results suggested that the three-miRNA signature could develop a new random Forest model to predict the BM of LUAD patients. These findings emphasized the importance of miRNAs in diagnosing BM, and provided evidence for selecting treatment decisions and designing clinical trials. |
first_indexed | 2024-04-13T09:26:17Z |
format | Article |
id | doaj.art-207c8cd917714a71871908f5f38268ca |
institution | Directory Open Access Journal |
issn | 1936-5233 1944-7124 |
language | English |
last_indexed | 2024-04-13T09:26:17Z |
publishDate | 2018-02-01 |
publisher | Elsevier |
record_format | Article |
series | Translational Oncology |
spelling | doaj.art-207c8cd917714a71871908f5f38268ca2022-12-22T02:52:25ZengElsevierTranslational Oncology1936-52331944-71242018-02-0111115716710.1016/j.tranon.2017.12.002Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA SignatureShuangtao ZhaoJiangyong YuLuhua WangOBJECTIVES: MicroRNAs (miRNAs) play a key role in governing posttranscriptional regulation through binding to the mRNAs of target genes. This study is to assess miRNAs expression profiles for identifying brain metastasis-related miRNAs to develop the predictive model by microarray in tumor tissues. METHODS: For this study, we screened the significant brain metastasis-related miRNAs from 77 lung adenocarcinoma (LUAD) patients with brain metastasis (BM+) or non-brain metastasis (BM−). A predictive model was developed from the training set (n = 42) using a random Forest supervised classification algorithm and a Class Centered Method, and then validated in a test set (n = 35) and further analysis in GSE62182 (n = 73). The independence of this signature in BM prediction was measured by multivariate logistic regression analysis. RESULTS: From the training set, the predictive model (including hsa-miR-210, hsa-miR-214 and hsa-miR-15a) stratified the patients into two groups with significantly different BM subtypes (90.4% of accuracy). The similar predictive power (91.4% of accuracy) was obtained in the test cohort. As an independent predictive factor, it was closely associated with BM and had high sensitivity and specificity in predicting BM in clinical practice. Moreover, functional enrichment analysis demonstrated that this signature involved in the signaling pathways positively correlated with cancer metastasis. CONCLUSION: These results suggested that the three-miRNA signature could develop a new random Forest model to predict the BM of LUAD patients. These findings emphasized the importance of miRNAs in diagnosing BM, and provided evidence for selecting treatment decisions and designing clinical trials.http://www.sciencedirect.com/science/article/pii/S1936523317303625 |
spellingShingle | Shuangtao Zhao Jiangyong Yu Luhua Wang Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature Translational Oncology |
title | Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature |
title_full | Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature |
title_fullStr | Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature |
title_full_unstemmed | Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature |
title_short | Machine Learning Based Prediction of Brain Metastasis of Patients with IIIA-N2 Lung Adenocarcinoma by a Three-miRNA Signature |
title_sort | machine learning based prediction of brain metastasis of patients with iiia n2 lung adenocarcinoma by a three mirna signature |
url | http://www.sciencedirect.com/science/article/pii/S1936523317303625 |
work_keys_str_mv | AT shuangtaozhao machinelearningbasedpredictionofbrainmetastasisofpatientswithiiian2lungadenocarcinomabyathreemirnasignature AT jiangyongyu machinelearningbasedpredictionofbrainmetastasisofpatientswithiiian2lungadenocarcinomabyathreemirnasignature AT luhuawang machinelearningbasedpredictionofbrainmetastasisofpatientswithiiian2lungadenocarcinomabyathreemirnasignature |