Comparative transcriptomic analysis of early- and late-onset colorectal cancer across TCGA, GEO, and CPTAC.

234 Background: Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of cancer-related mortality worldwide. Notably, the incidence of early-onset CRC (EO-CRC; < 50 years) has been on the rise. While clinical differences from late-onset CRC (LO-CRC; ≥50 years) have been described, transcriptomic distinctions between the two require further elucidation. This study evaluated whether EO and LO tumors display reproducible gene- and pathway-level programs across independent cohorts, with external RNA-seq dataset validation. Methods: Bulk tumor transcriptomes from TCGA-COAD+READ (RNA-seq) and two microarray cohorts (GSE39582, GSE41258) were analyzed. LO–EO contrasts were modeled with limma/voom, and preranked enrichment was performed using GO Biological Process and MSigDB Hallmark pathways. Cross-cohort concordance was assessed via symbol mapping, directional agreement, and rank-decile overlaps. Independent validation was performed in CPTAC Colon (RNA-seq). Overall survival was evaluated in TCGA using Kaplan–Meier and Cox regression adjusted for age, sex, and stage. Results: In TCGA, seven genes reached FDR significance (LO-higher: PITX2, HOXB9, HOXB8, LINC01224; EO-higher: PRND, RPL39L, DNALI1). GSE39582 identified one EO-higher gene (TAF5L), and two in GSE41258 (PNMT, DZIP1). Overall, gene-level differences were modest (most |log2FC| ≤1). Pathway analysis revealed reproducible EO enrichment for RNA splicing/processing, ribosome biogenesis, and DNA replication/repair, with consistent down-ranking of MYC/E2F/G2M programs. LO tumors emphasized chromatin and transcriptional regulation in GSE41258 and mitotic spindle/checkpoint pathways (KIF11, PLK1, CDC20, UBE2C, MYBL2) in CPTAC. Genome-wide directional concordance was modest but above random (~55–57%), improving to (~75–77%) among jointly nominal genes. CPTAC validation (EO n = 5, LO n = 98) confirmed EO enrichment for immunomodulatory signatures, including negative regulation of NK-cell cytotoxicity and lymphocyte-mediated immunity (all q ≤0.05). In TCGA, EO patients showed a non-significant trend toward improved survival (HR LO vs EO 1.85, 95% CI 0.95–3.61; p = 0.069). Conclusions: Across three discovery cohorts and independent validation, EO-CRC consistently demonstrated enrichment for RNA-processing, biosynthetic, and immune-modulatory programs, whereas LO-CRC emphasized chromatin and mitotic spindle biology. These reproducible, age-linked transcriptomic signatures suggest distinct tumor biology in EO-CRC that warrants further mechanistic and therapeutic exploration.