Development of a novel chemokine signaling-based multigene signature to predict prognosis and therapeutic response in colorectal cancer

Development of a novel chemokine signaling-based multigene signature to predict prognosis and therapeutic response in colorectal cancer 基于趋化因子信号通路开发新型多基因模型用于预测结直肠癌的预后和治疗反应 corresp first-author Xin QI, qixin@usts.edu.cn Xin QI, qixin@usts.edu.cn qixin@usts.edu.cn https://orcid.org/0000-0001-9038-4287 School of Chemistry and Life Sciences, Suzhou University of Science and Technology, Suzhou215011, China School of Chemistry and Life Sciences, Suzhou University of Science and Technology, Suzhou215011, China 目的 2 开发一种基于趋化因子信号通路的多基因模型，可以有效预测结直肠癌患者的总生存期和治疗反应。 创新点 2 结直肠癌具有高度异质性，迫切需要开发可靠的标志物用于预测患者的预后和治疗效果。研究显示趋化因子信号通路关键基因表达变化与结直肠癌转移和预后密切相关，但尚缺乏基于该信号通路的基因模型用来预测患者的临床预后。本研究基于趋化因子信号通路构建了一种新的多基因模型（CSbMgSig），用于结直肠癌患者的预后风险分层和总体生存时间预测，为结直肠癌患者的治疗提供指导。 方法 2 首先，考虑到转移对结直肠癌预后的重要影响，我们在原发和转移性结直肠癌样本中鉴定了差异表达的趋化因子信号通路相关基因。然后，结合单因素Cox回归分析和LASSO Cox回归分析建立基于趋化因子信号通路的预后模型CSbMgSig，并在另一个独立数据集通过Kaplan-Meier生存分析和时间依赖性受试者工作特征（ROC）分析进一步验证其预后性能。此外，我们还进行了基因本体（GO）富集分析、基因集富集分析（GSEA）、单样本基因集富集分析（ssGSEA）和化疗反应分析，以探索CSbMgSig在结直肠癌致病机制中的功能及其对免疫浸润和化疗反应的影响。 结论 2 构建的由8个趋化因子信号通路相关基因组成的预后模型CSbMgSig，可以有效地在训练集和验证集中区分高危结直肠癌患者，并被证明是总生存期的独立预测因子。功能分析结果表明，该特征在结直肠癌患者的免疫浸润和对药物的反应中起关键作用。因此，基于趋化因子信号通路的多基因模型CSbMgSig是一种很有前景的工具，可以进行风险分层、生存预测和治疗评估，从而有利于结直肠癌患者的个性化管理。 趋化因子信号 多基因模型 结直肠癌 预后 治疗反应 Colorectal cancer (CRC) is the most lethal gastrointestinal cancer in both males and females worldwide ( Sung et al., 2021 Sung et al., 2021 ). Because of the high heterogeneity of tumors, robust prognostic biomarkers are urgently needed in CRC management ( Koncina et al., 2020 Koncina et al., 2020 ). Chemokine signaling is a well-known pivotal player in immunity, inflammation, and cancer metastasis ( Lacalle et al., 2017 Lacalle et al., 2017 ; Poeta et al., 2019 Poeta et al., 2019 ; Do et al., 2020 Do et al., 2020 ), and multiple genes involved in chemokine signaling have been demonstrated as potential prognostic biomarkers for CRC (Cabrero-De Las Heras and Martínez-Balibrea, 2018; Ottaiano et al., 2020 Ottaiano et al., 2020 ; Yu et al., 2020 Yu et al., 2020 ). Therefore, the aim of our study was to develop a chemokine signaling-based multigene signature (CSbMgSig) that could effectively predict overall survival (OS) and therapeutic response for patients with CRC. Gene expression data from human CRC tissues and corresponding clinical data were retrieved from the public Gene Expression Omnibus (GEO) database ( https://www.ncbi.nlm.nih.gov/geo https://www.ncbi.nlm.nih.gov/geo ) with the accession numbers GSE131418, GSE39582, and GSE17536. Given the critical role of distal metastasis in influencing patient survival, we first downloaded the GSE131418 dataset (with 332 primary and 184 metastatic human CRC samples) from the Moffitt Cancer Center ( Kamal et al., 2019 Kamal et al., 2019 ) in order to identify differentially expressed genes (DEGs) involved in CRC metastasis. The GSE39582 ( n =582) and GSE17536 ( n =145) cohorts were then respectively used as the training set and validation set for CSbMgSig construction. We also collected 192 genes in the chemokine signaling pathway from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database ( https://www.genome.jp/kegg https://www.genome.jp/kegg ) with the entry number hsa04062 to explore the prognostic roles of chemokine signaling-related genes (CSRGs) in CRC. The present research used the strategy (Fig. S1) modified from our previous study ( Qi et al., 2021 Qi et al., 2021 ). We found that 59 CSRGs were significantly differentially expressed between primary and metastatic human CRC samples in the GSE131418 dataset. Furthermore, univariate Cox regression analysis showed that the abnormal expression patterns of nine CSRGs, namely adenylate cyclase 2 ( ADCY2 ), protein kinase B γ ( AKT3 ), chemokine (C-C motif) ligand 7 ( CCL7 ), CCL8 , CXC motif chemokine ligand 2 ( CXCL2 ), CXC chemokine receptor 3 ( CXCR3 ), nuclear factor-‍κB (NF-‍κB) inhibitor β ( NFKBIB ), phosphatidylinositol 3-kinase catalytic subunit α ( PIK3CA ), and phospholipase C β4 ( PLCB4 ), were significantly correlated with OS in CRC patients in the GSE39582 dataset (Fig. S2). Least absolute shrinkage and selection operator (LASSO) Cox regression analysis was also performed to select the most robust prognostic genes from the above nine CSRGs. Based on the optimal λ value, we were able to establish the CSbMgSig ( Figs. 1 Figs. 1 a–‍1c). Notably, most genes of the signature have been shown to be involved in the pathogenesis of CRC. For example, overexpression of CCL7 is closely associated with liver metastasis and shorter OS of CRC patients ( Kurzejamska et al., 2019 Kurzejamska et al., 2019 ). The CRC patients were divided into high- and low-risk groups based on the median value of risk scores, which were calculated based on the regression coefficient and expression level of each gene in the signature ( Fig. 1 Fig. 1 d). The expression level of each signature-associated CSRG was obviously changed between the high- and low-risk patients ( Fig. 1 Fig. 1 f), and patients with high risk scores had significantly shorter OS (Figs. 1e 1e and 1g 1g ). Furthermore, we found that based on the time-dependent receiver operating characteristic (ROC) curve analysis results generated by the CSbMgSig, the area under the curve (AUC) value reached 0.700 at one year, 0.624 at two years, and 0.619 at three years ( Fig. 1 Fig. 1 h). Therefore, the results indicated that the established CSbMgSig possessed a powerful predictive ability for the OS of CRC patients. To investigate the relationship between the CSbMgSig and clinicopathological characteristics, we determined whether there were significant differences in risk scores among distinct subgroups stratified by TNM stage, T stage, N stage, M stage, tumor location, age, and sex. The results revealed that CRC patients with advanced TNM stage, T stage, N stage, M stage, or proximal tumor location exhibited higher risk scores ( Fig. 2 Fig. 2 a). To further assess the prognostic power of the CSbMgSig, we performed a survival stratification analysis in different subgroups of CRC patients. As shown in Fig. 2 Fig. 2 b, high-risk patients had worse outcomes than low-risk patients in the following subgroups: TNM stage III/IV, T3/4 stage, N stage (N0 and N1/2/3), M stage (M0 and M1), tumor location (primary and distal), and sex (female and male). These results indicated that the established CSbMgSig had a strong predictive ability for the prognosis of CRC patients in these clinical subgroups. To confirm the prognostic performance of the constructed CSbMgSig, we divided the CRC patients from the GSE17536 cohort into high- and low-risk groups based on the median risk score calculated by the established signature formula ( Fig. 3 Fig. 3 a). As expected, the survival status distribution plot and Kaplan-Meier survival curve indicated that CRC patients with higher risk scores had lower survival probability and shorter OS time (Figs. 3b 3b and 3d 3d ). The time-dependent ROC curve analysis also indicated that the CSbMgSig had a strong prognostic ability for CRC patients in the GSE17536 dataset ( Fig. 3 Fig. 3 e). Moreover, the expression levels of the eight genes involved in the signature were obviously different between high- and low-risk patients ( Fig. 3 Fig. 3 c), and the expression patterns of the representative signature-related genes (PLCB4 and CXCR3) were validated in CRC and normal control tissues utilizing the Human Protein Atlas (HPA) database ( https://www.proteinatlas.org https://www.proteinatlas.org ) (Figs. 3f 3f and 3g 3g ). These results implied that the CSbMgSig had a robust ability to predict clinical outcomes of CRC patients. In addition, to further examine whether the prognostic power of the CSbMgSig was independent of other clinicopathological factors, we conducted multivariate Cox regression analysis for the established signature and clinical factors including age, gender, and TNM stage. As shown in Fig. S3, the CSbMgSig-based risk score was strongly associated with OS of CRC patients in both the GSE39582 and GSE17536 datasets, indicating that the CSbMgSig was an independent prognostic factor for CRC patients. To investigate the potential biological processes and pathways related to the CSbMgSig, we firstly identified 311 DEGs between high-and low-risk groups in the training cohort and performed KEGG pathway and gene ontology (GO) functional enrichment analyses on the identified DEGs. As shown in Fig. S4a, the results of GO analysis showed that these DEGs were significantly enriched in GO terms related to cell migration and chemotaxis. In addition, KEGG pathway analysis indicated that these DEGs were primarily enriched in multiple hallmark biological pathways involved in CRC pathogenesis, such as extracellular matrix (ECM)‍-receptor interaction and the NF-‍κB and chemokine signaling pathways (Fig. S4b). Moreover, gene set enrichment analysis (GSEA) indicated that multiple tumor hallmarks, e.g., angiogenesis, epithelial-mesenchymal transition, inflammatory response, interferon-γ (IFN-γ) response, and kirsten ratsarcoma viral oncogene homolog (KRAS) signaling, were closely associated with high-risk CRC patients (Fig. S4c), while cancer-related pathways, e.g., DNA repair, E2F targets, G2/M checkpoint, myelocytomatosis viral oncogene homolog (MYC) target, oxidative phosphorylation, and peroxisome, were significantly enriched in the low-risk group (Fig. S4d). Considering the critical role of immunity in cancer development and progression, we first determined the correlation between the CSbMgSig-derived risk score and immune status in the GSE39582 cohort through single-sample GSEA (ssGSEA) enrichment analysis. As shown in Fig. 4 Fig. 4 a, the scores of multiple immune cell types, such as antigen-loaded dendritic cells (aDCs), CD8 + T cells, B cells, helper T cell 2 (Th2) cells, macrophages, tumor infiltrating lymphocytes (TILs), and regulatory T cells (Tregs), were dramatically increased in the high-risk group (all adjusted P < 0.05). Moreover, the scores of several pivotal immune-related functions, e.g., antigen presenting cell (APC) co-stimulation, check-point, inflammation-promoting, T cell co-inhibition, and IFN response, were also significantly elevated in the high-risk group (all adjusted P < 0.05; Fig. 4 Fig. 4 b). The link between risk score and molecular subtype was determined in the GSE39582 dataset to investigate its association with immune status. Molecular subtypes C1 and C3 were associated with down-regulated immune pathways, while C2 was correlated with up-regulated immune pathways ( Marisa et al., 2013 Marisa et al., 2013 ). As shown in Fig. 4 Fig. 4 c, the risk score of the C2 subtype was significantly higher than those of the C1 and C3 subtypes, indicating the close relationship between risk score and immune status. Also, the results of correlation analyses showed that the risk score was significantly correlated with the immune score and stromal score (Figs. 4d 4d and 4e 4e ), which were tightly linked to immune infiltration in the tumor microenvironment. In addition, as immune checkpoints have been demonstrated as an important target for CRC immunotherapy, we further examined the expression levels of well-known immune checkpoints between high- and low-risk groups in the GSE39582 dataset. As shown in Fig. 4 Fig. 4 f, cytotoxic T lymphocyte-associate protein-4 ( CTLA-4 ) and programmed cell death-1 ( PD-1 ) were dramatically increased in patients with low risk scores (both P < 0.05), and programmed cell death-ligand 1 ( PD-L1 ) and PD-L2 were significantly decreased in patients in the low-risk group (both P < 0.05). The results were further confirmed through correlation analyses ( Fig. 4 Fig. 4 g), supporting the important role of the CSbMgSig in mediating immune response. Considering the important role of chemotherapy in CRC treatment ( Ichikawa et al., 2020 Ichikawa et al., 2020 ; Glimelius et al., 2021 Glimelius et al., 2021 ), we assessed the chemotherapeutic responses of high- and low-risk patients based on the GDSC database, which contains sensitivities and genomic profiles of large-scale anti-tumor compounds ( Yang et al., 2013 Yang et al., 2013 ). Notably, 53 chemotherapeutic drugs had significant differences in the half-maximal inhibitory concentration (IC 50 ) between high- and low-risk patients in the GSE39582 and GSE17536 cohorts. Of these, 16 anti-tumor compounds have been proven to be chemotherapeutic drugs for CRC patients (Fig. S5). Therefore, these data indicated that the established CSRG signature provides potential candidates for personalized treatment of CRC patients. In summary, we established a novel CSbMgSig as a promising tool for prognostic risk assessment in CRC patients. The prognostic signature was confirmed to be independently associated with OS of CRC patients in both training and validation cohorts. In particular, functional analysis demonstrated that the signature plays a key role in immune infiltration and drug response of CRC patients. Therefore, evaluating the CSbMgSig-based risk score of individual CRC patients will not only contribute to risk stratification and OS prediction, but also provide valuable insights into therapeutic efficacy. Materials and methods 1 Detailed methods are provided in the electronic supplementary materials of this paper. References Cabrero-de Las Heras S , Martínez-Balibrea E , 2018 . CXC family of chemokines as prognostic or predictive biomarkers and possible drug targets in colorectal cancer . World J Gastroenterol , 24 ( 42 ): 4738 - 4749 . https://doi.org/10.3748/wjg.v24.i42.4738 https://doi.org/10.3748/wjg.v24.i42.4738 CXC family of chemokines as prognostic or predictive biomarkers and possible drug targets in colorectal cancer Do HTT , Lee CH , Cho J , 2020 . Chemokines and their receptors: multifaceted roles in cancer progression and potential value as cancer prognostic markers . Cancers , 12 ( 2 ): 287 . https://doi.org/10.3390/cancers12020287 https://doi.org/10.3390/cancers12020287 Chemokines and their receptors: multifaceted roles in cancer progression and potential value as cancer prognostic markers Glimelius B , Stintzing S , Marshall J , et al. , 2021 . Metastatic colorectal cancer: advances in the folate-fluoropyrimidine chemotherapy backbone . Cancer Treat Rev , 98 : 102218 . https://doi.org/10.1016/j.ctrv.2021.102218 https://doi.org/10.1016/j.ctrv.2021.102218 Metastatic colorectal cancer: advances in the folate-fluoropyrimidine chemotherapy backbone Ichikawa N , Kamiyama T , Yokoo H , et al. , 2020 . Preoperative chemotherapy in colorectal cancer patients with synchronous liver metastasis . Mol Clin Oncol , 12 ( 4 ): 374 - 383 . https://doi.org/10.3892/mco.2020.1992 https://doi.org/10.3892/mco.2020.1992 Preoperative chemotherapy in colorectal cancer patients with synchronous liver metastasis Kamal Y , Schmit SL , Hoehn HJ , et al. , 2019 . Transcriptomic differences between primary colorectal adenocarcinomas and distant metastases reveal metastatic colorectal cancer subtypes . Cancer Res , 79 ( 16 ): 4227 - 4241 . https://doi.org/10.1158/0008-5472.CAN-18-3945 https://doi.org/10.1158/0008-5472.CAN-18-3945 Transcriptomic differences between primary colorectal adenocarcinomas and distant metastases reveal metastatic colorectal cancer subtypes Koncina E , Haan S , Rauh S , et al. , 2020 . Prognostic and predictive molecular biomarkers for colorectal cancer: updates and challenges . Cancers , 12 ( 2 ): 319 . https://doi.org/10.3390/cancers12020319 https://doi.org/10.3390/cancers12020319 Prognostic and predictive molecular biomarkers for colorectal cancer: updates and challenges Kurzejamska E , Sacharczuk M , Landázuri N , et al. , 2019 . Effect of chemokine (C-C motif) ligand 7 (CCL7) and its receptor (CCR2) expression on colorectal cancer behaviors . Int J Mol Sci , 20 ( 3 ): 686 . https://doi.org/10.3390/ijms20030686 https://doi.org/10.3390/ijms20030686 Effect of chemokine (C-C motif) ligand 7 (CCL7) and its receptor (CCR2) expression on colorectal cancer behaviors Lacalle RA , Blanco R , Carmona-Rodríguez L , et al. , 2017 . Chapter five‍‒‍chemokine receptor signaling and the hallmarks of cancer . Int Rev Cell Mol Biol , 331 : 181 - 244 . https://doi.org/10.1016/bs.ircmb.2016.09.011 https://doi.org/10.1016/bs.ircmb.2016.09.011 Chapter five‍‒‍chemokine receptor signaling and the hallmarks of cancer Marisa L , de Reyniès A , Duval A , et al. , 2013 . Gene expression classification of colon cancer into molecular subtypes: characterization, validation, and prognostic value . PLoS Med , 10 ( 5 ): e1001453 . https://doi.org/10.1371/journal.pmed.1001453 https://doi.org/10.1371/journal.pmed.1001453 Gene expression classification of colon cancer into molecular subtypes: characterization, validation, and prognostic value Ottaiano A , Scala S , Normanno N , et al. , 2020 . Prognostic and predictive role of CXC chemokine receptor 4 in metastatic colorectal cancer patients . Appl Immunohistochem Mol Morphol , 28 ( 10 ): 755 - 760 . https://doi.org/10.1097/PAI.0000000000000828 https://doi.org/10.1097/PAI.0000000000000828 Prognostic and predictive role of CXC chemokine receptor 4 in metastatic colorectal cancer patients Poeta VM , Massara M , Capucetti A , et al. , 2019 . Chemokines and chemokine receptors: new targets for cancer immunotherapy . Front Immunol , 10 : 379 . https://doi.org/10.3389/fimmu.2019.00379 https://doi.org/10.3389/fimmu.2019.00379 Chemokines and chemokine receptors: new targets for cancer immunotherapy Qi X , Wang R , Lin YX , et al. , 2021 . A ferroptosis-related gene signature identified as a novel prognostic biomarker for colon cancer . Front Genet , 12 : 692426 . https://doi.org/10.3389/fgene.2021.692426 https://doi.org/10.3389/fgene.2021.692426 A ferroptosis-related gene signature identified as a novel prognostic biomarker for colon cancer Sung H , Ferlay J , Siegel RL , et al. , 2021 . Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries . CA Cancer J Clin , 71 ( 3 ): 209 - 249 . https://doi.org/10.3322/caac.21660 https://doi.org/10.3322/caac.21660 Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries Yang W , Soares J , Greninger P , et al. , 2013 . Genomics of drug sensitivity in cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells . Nucleic Acids Res , 41 ( D1 ): D955 - D961 . https://doi.org/10.1093/nar/gks1111 https://doi.org/10.1093/nar/gks1111 Genomics of drug sensitivity in cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells Yu L , Yang XM , Xu CC , et al. , 2020 . Comprehensive analysis of the expression and prognostic value of CXC chemokines in colorectal cancer . Int Immunopharmacol , 89 : 107077 . https://doi.org/10.1016/j.intimp.2020.107077 https://doi.org/10.1016/j.intimp.2020.107077 Comprehensive analysis of the expression and prognostic value of CXC chemokines in colorectal cancer Xin QI designed the research and drafted the manuscript. Xin QI, Donghui YAN, Jiachen ZUO, and Rui WANG collected the data and performed the computational analyses. Xin QI, Donghui YAN, Jiachen ZUO, Rui WANG, and Jiajia CHEN revised the manuscript. All authors have read and approved the final manuscript, and therefore, have full access to all the data in the study and take responsibility for the integrity and security of the data. This work was supported by the National Natural Science Foundation of China (Nos. 31900490 and 31770903) and the Natural Science Foundation of the Jiangsu Higher Education Institutions of China (No. 19KJB180027). 10.1631/jzus.B2100412 1673-1581（2021）12-1053-07 This work was supported by the National Natural Science Foundation of China (Nos. 31900490 and 31770903) and the Natural Science Foundation of the Jiangsu Higher Education Institutions of China (No. 19KJB180027). 2021-05-09 2021-08-16 2021-12-15 2022-05-09 2021-11-22 Copyright © Zhejiang University and Springer-Verlag GmbH Germany, part of Springer Nature 2021 2021 浙江大学出版社 Xin QI, Donghui YAN, Jiachen ZUO, Rui WANG, and Jiajia CHEN declare that they have no conflict of interest. This article does not contain any studies with human or animal subjects performed by any of the authors. Compliance with ethics guidelines 浙江大学学报（英文版）（B辑：生物医学和生物技术） 12 22