A Cancer Associated Fibroblasts-Related Six-Gene Panel for Anti-PD-1 Therapy in Melanoma Driven by Weighted Correlation Network Analysis and Supervised Machine Learning

作者全名："Tian, Luyao; Long, Fei; Hao, Youjin; Li, Bo; Li, Yinghong; Tang, Ying; Li, Jing; Zhao, Qi; Chen, Juan; Liu, Mingwei"

作者地址："[Tian, Luyao; Long, Fei; Tang, Ying; Li, Jing; Zhao, Qi; Chen, Juan; Liu, Mingwei] Chongqing Med Univ, Coll Lab Med, Key Lab Clin Lab Diagnost, Chongqing, Peoples R China; [Hao, Youjin; Li, Bo] Chongqing Normal Univ, Coll Life Sci, Cell Biol & Bioinformat, Chongqing, Peoples R China; [Li, Yinghong] Chongqing Univ Posts & Telecommun, Key Lab Big Data Bio Intelligence, Chongqing, Peoples R China"

通信作者："Chen, J; Liu, MW (通讯作者)，Chongqing Med Univ, Coll Lab Med, Key Lab Clin Lab Diagnost, Chongqing, Peoples R China."

来源：FRONTIERS IN MEDICINE

ESI学科分类：CLINICAL MEDICINE

WOS号：WOS:000791577200001

JCR分区：Q1

影响因子：3.9

年份：2022

卷号：9

期号：　

开始页：　

结束页：　

文献类型：Article

关键词：melanoma; anti-PD-1 therapy; CAFs-related biomarker panel; WGCNA; supervised machine learning

摘要："BackgroundMelanoma is a highly aggressive skin cancer with a poor prognosis and mortality. Immune checkpoint blockade (ICB) therapy (e.g., anti-PD-1 therapy) has opened a new horizon in melanoma treatment, but some patients present a non-responsive state. Cancer-associated fibroblasts (CAFs) make up the majority of stromal cells in the tumor microenvironment (TME) and have an important impact on the response to immunotherapy. There is still a lack of identification of CAFs-related predictors for anti-PD-1 therapy, although the establishment of immunotherapy biomarkers is well underway. This study aims to explore the potential CAFs-related gene panel for predicting the response to anti-PD-1 therapy in melanoma patients and elucidating their potential effect on TME. MethodsThree gene expression datasets from melanoma patients without anti-PD-1 treatment, in a total of 87 samples, were downloaded from Gene Expression Omnibus (GEO) as the discovery sets (GSE91061) and validation sets (GSE78220 and GSE122220). The CAFs-related module genes were identified from the discovery sets by weighted gene co-expression network analysis (WGCNA). Concurrently, we utilized differential gene analysis on the discovery set to obtain differentially expressed genes (DEGs). Then, CAFs-related key genes were screened with the intersection of CAFs-related module genes and DEGs, succeeded by supervised machine learning-based identification. As a consequence of expression analysis, gene set enrichment analysis, survival analysis, staging analysis, TME analysis, and correlation analysis, the multidimensional systematic characterizations of the key genes were uncovered. The diagnostic performance of the CAFs-related gene panel was assessed by receiver operating characteristic (ROC) curves in the validation sets. Eventually, the CAFs-related gene panel was verified by the expression from the single-cell analysis. ResultsThe six-gene panel associated with CAFs were finally identified for predicting the response to anti-PD-1 therapy, including CDK14, SYNPO2, TCF4, GJA1, CPXM1, and TFPI. The multigene panel demonstrated excellent combined diagnostic performance with the area under the curve of ROC reaching 90.5 and 75.4% ~100% in the discovery and validation sets, respectively. ConclusionConfirmed by clinical treatment outcomes, the identified CAFs-related genes can be used as a promising biomarker panel for prediction to anti-PD-1 therapy response, which may serve as new immunotherapeutic targets to improve survival outcomes of melanoma patients."

基金机构：　

基金资助正文：