Skip to main content

Research Repository

Advanced Search

Prediction of Distant Metastasis of Lymph-Node-Negative Primary Breast Cancer From Gene Expression Profiling Using Cox-Boost Regression Model

Hamidi, Omid; Amini, Payam; Tapak, Leili; Zohrab Beigi, Yasaman; Afshar, Saeid; Dinu, Irina

Authors

Omid Hamidi

Leili Tapak

Yasaman Zohrab Beigi

Saeid Afshar

Irina Dinu



Abstract

Backgrounds:: Distant metastasis in breast cancer patients contributes to increased breast cancer mortality, highlighting the urgent need for effective predictive strategies. Understanding metastasis mechanisms and identifying relevant biomarkers are crucial for improving patient outcomes and informing targeted therapies. This study employed a high-dimensional regression model to identify biomarkers linked to distant metastasis-free survival in breast cancer patients, with the goal of enhancing prognostic accuracy and guiding clinical decisions. Methods:: We utilized the publicly available breast cancer dataset (GSE2034), which includes gene expression profiles for 22 283 genes across 286 samples. To identify relevant genes, we applied Cox-Boost regression and a random forest (RF) model. We then explored the association between the selected genes and metastasis-free survival outcomes using quantile regression, chosen for its ability to assess the impact of these genes across different survival quantiles (P < .05). This approach complements the Cox-Boost model by providing a more detailed understanding of gene-survival relationships at various points in the survival distribution, thereby strengthening the robustness of our findings. Results:: We identified 222 significant transcripts using univariate Cox regression models. By applying Cox-Boost, both with and without adjustment for ER+/− status, we identified 7 genes associated with time-to-relapse/metastasis in breast cancer patients: SNU13, CLINT1, ACBD3, NEK2, COL2A1, WFDC1, and RACGAP1. A similar approach was used for ER-positive patients. Patients were classified as high or low risk for metastasis based on the median prognostic index calculated from the identified genes (P < .001). The top-ranked genes associated with high/low risk groups using RF were RACGAP1, NEK2, CCNA2, DTL, ACBD3, ARL6IP5, WFDC1, and PDCD4. Conclusions:: We identified eleven key genes, including SNU13, CLINT1, ACBD3, NEK2, COL2A1, WFDC1, and RACGAP1, as well as CCNA2, DTL, ARL6IP5, and PDCD4, that are related to the risk of distant metastasis and may be used as biomarkers to predict distant metastasis of breast cancer.

Citation

Hamidi, O., Amini, P., Tapak, L., Zohrab Beigi, Y., Afshar, S., & Dinu, I. (in press). Prediction of Distant Metastasis of Lymph-Node-Negative Primary Breast Cancer From Gene Expression Profiling Using Cox-Boost Regression Model. Cancer Informatics, 23, https://doi.org/10.1177/11769351241297493

Journal Article Type Article
Acceptance Date Oct 7, 2024
Online Publication Date Nov 30, 2024
Deposit Date Dec 9, 2024
Publicly Available Date Dec 9, 2024
Journal Cancer Informatics
Publisher Libertas Academica
Peer Reviewed Peer Reviewed
Volume 23
DOI https://doi.org/10.1177/11769351241297493
Keywords Bioinformatics, machine learning, biomarker, prognosis, Gene expression
Public URL https://keele-repository.worktribe.com/output/1013125
Publisher URL https://journals.sagepub.com/doi/10.1177/11769351241297493

Files

Prediction of Distant Metastasis of Lymph-Node-Negative Primary Breast Cancer From Gene Expression Profiling Using Cox-Boost Regression Model (478 Kb)
Archive

Licence
https://creativecommons.org/licenses/by/4.0/

Publisher Licence URL
https://creativecommons.org/licenses/by/4.0/

Copyright Statement
This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).






You might also like



Downloadable Citations