The construction of a prognostic nomogram model for colorectal cancer and the prediction of immune characteristics and immune treatment responses based on the bioinformatics analysis of soluble mediator-related genes

Accurate prognosis prediction in colorectal cancer (CRC) is essential for personalized treatment. Soluble mediators are promising predictive biomarkers for evaluating outcomes. We sourced transcriptome data of CRC (COAD+READ) from TCGA and GEO. Soluble mediator-related genes (SMRGs) were identified via GeneCards. Through univariate Cox and Lasso regression analyses, prognosis-related feature genes were determined. A prognostic model was created using multivariate Cox regression, categorizing patients into high-risk (HR) and low-risk (LR) groups based on the median riskscore. KEGG pathway enrichment analysis and GSEA were undertaken on groups. ssGSEA assessed immune cell scores, while ESTIMATE analysis evaluated stromal and immune cell scores along with tumor purity. The CellMiner database identified potential drugs for HR patients. Pearson correlation analysis revealed the relationship between mismatch repair (MMR) genes and model genes. We identified 10 SMRGs. Pearson correlation analysis indicated positive correlations among these genes. GO analysis showed that most feature genes were linked to binding functions. KEGG analysis revealed that the HR group was enriched in pathways like Basal cell carcinoma and Glycosaminoglycan biosynthesis. The ssGSEA indicated higher immune cell scores in the LR group, alongside lower stromal scores. LR group also exhibited a lower TIDE score and higher immunophenoscore. Drug sensitivity analysis suggested PF-4708671, PI-103, and XAV939 as potential treatments for HR patients. There was significant correlation between model gene and MMR genes. The CRC prognostic model based on SMRGs effectively predicts patient prognosis and guides treatment strategies.

结直肠癌（colorectal cancer, CRC）的精准预后预测对个体化治疗至关重要。可溶性介质是评估患者临床结局的极具潜力的预测性生物标志物。我们从TCGA与GEO数据库中获取了结直肠癌（COAD+READ）的转录组数据。通过GeneCards数据库鉴定得到可溶性介质相关基因（soluble mediator-related genes, SMRGs）。经单变量Cox回归与Lasso回归分析，筛选出与预后相关的特征基因。采用多变量Cox回归构建预后模型，以中位风险评分为界将患者划分为高风险（high-risk, HR）与低风险（low-risk, LR）组。对两组开展KEGG通路富集分析与基因集富集分析（Gene Set Enrichment Analysis, GSEA）。通过单样本基因集富集分析（single-sample Gene Set Enrichment Analysis, ssGSEA）评估免疫细胞浸润评分，同时利用ESTIMATE分析评估基质细胞、免疫细胞评分及肿瘤纯度。借助CellMiner数据库筛选出针对高风险患者的潜在治疗药物。通过Pearson相关分析揭示错配修复（mismatch repair, MMR）基因与模型基因之间的关联。本研究共鉴定出10个SMRGs，Pearson相关分析显示这些基因间均呈正相关关系。GO富集分析结果表明，多数特征基因与结合功能密切相关。KEGG富集分析显示，高风险组显著富集于基底细胞癌、糖胺聚糖生物合成等通路。ssGSEA分析结果显示，低风险组的免疫细胞评分更高，而基质评分更低。低风险组同时表现出更低的TIDE评分与更高的免疫评分（immunophenoscore）。药物敏感性分析提示PF-4708671、PI-103及XAV939可作为高风险患者的潜在治疗药物。模型基因与MMR基因之间存在显著相关性。本研究构建的基于SMRGs的结直肠癌预后模型可有效预测患者预后并指导临床治疗策略。