Integrative machine learning identifies lactylation-related gene signature prognostic for chemotherapeutic efficacy in colorectal carcinoma

整合机器学习识别与乳酸化相关的基因特征,该特征可预测结直肠癌化疗疗效

阅读:1

Abstract

BACKGROUND: Lactylation, a recently discovered post-translational modification, has emerged as a critical regulator in cancer biology. Although chemotherapy remains the first-line treatment for metastatic colorectal cancer (CRC), only a subset of patients responds to it. This study aimed to identify key lactylation-related genes in CRC and evaluate their potential as predictive biomarkers for chemotherapy response. METHOD: Gene expression profiles and corresponding clinical data from CRC patients were obtained from The Cancer Genome Atlas (TCGA), Genotype-Tissue Expression (GTEx), and Gene Expression Omnibus (GEO) databases. Differentially expressed genes (DEGs) were identified using the limma R package, and key modules were selected through weighted gene co-expression network analysis (WGCNA). Intersecting genes were determined by aligning DEGs with WGCNA module genes. A predictive model was developed utilizing 11 machine learning algorithms and 92 algorithm combinations. Furthermore, the correlation between lactylation-related gene score and immune infiltration as well as drug sensitivity in CRC were also investigated with "CIBERSORT" and "oncoPredict" package. RESULTS: Eight lactylation-related genes in CRC were identified and used to construct a predictive model employing Random Forest (RF) and Gradient Boosting Machine (GBM) algorithms. The model demonstrated strong predictive efficacy for chemotherapy response in CRC patients. Using lactylation gene scores, we effectively stratified patients into high- and low-score groups, which showed distinct patterns in immune cell infiltration, tumor mutational profile, and response to conventional antitumor drugs. Notably, the high-lactylation score group exhibited reduced Treg immune characteristics and increased sensitivity to 5-Fluorouracil. CONCLUSIONS: In summary, our findings demonstrate that machine learning-driven analysis of lactylation biomarkers represents a promising approach for advancing personalized therapy and optimizing clinical management in CRC.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。