Abstract
Colorectal cancer (CRC) is a leading cause of cancer-related mortality globally, necessitating biomarkers for early detection and targeted therapies. We integrated RNA-seq (GSE180440) weighted gene co-expression analysis with experimental validation to identify coding and long non-coding RNA biomarkers associated with tumor biology and treatment response. Bioinformatics analysis of GSE180440 dataset (145 tumor, 45 normal samples) identified differentially expressed genes (DEGs) and lncRNAs (DELs) using DESeq2. Weighted gene co-expression network analysis (WGCNA) identified modules linked to CRC traits. Functional enrichment, protein-protein interaction (PPI), and miRNA-gene networks were constructed. Validation used TCGA-COAD and RT-qPCR on 61 paired CRC samples. Diagnostic performance was assessed by ROC/AUC (pROC). Drug-response and survival analyses were performed with GEPIA3. From 672 DEGs and WGCNA, LINC02577, LINC00294 and CKAP2 emerged as hub candidates together with miR-548k and miR-941. In TCGA-COAD CKAP2 (log2FC = 1.501, P < 0.001) and LINC02577 (log2FC = 6.676, P < 0.001) were significantly upregulated, while LINC00294 was downregulated (log2FC = - 1.104, P < 0.001). Diagnostic AUCs (tumor vs. normal) were: LINC02577 0.982, CKAP2 0.797 and CEA 0.663; LINC00294 showed poor discrimination (AUC = 0.055). In our cohort CKAP2 associated with tumor size > 5 cm (P = 0.025, AUC = 0.667) and LINC02577 associated with nodal involvement in females (P = 0.028, AUC = 0.736). miR-548k overexpression correlated with early invasion (P = 0.044, AUC = 0.667). LINC02577 expression was higher in patients with progressive disease after leucovorin-containing regimens (P = 0.009). None of the selected genes were prognostic for overall survival. We identified CKAP2, LINC02577, miR-941, and miR-548k as key biomarkers with distinct expression patterns in early-stage CRC. Their altered expression in small, early-stage tumors reveales their potential for early CRC detection. LINC02577 and CKAP2 show strong diagnostic utility (AUCs 0.982 and 0.797) outperforming CEA, and LINC02577 may predict poor response to leucovorin-containing therapy (P = 0.009). These candidates warrant functional studies to define mechanisms and evaluate clinical applicability.