Integrated Multi-Omics Analysis Constructs an Intratumoral Heterogeneity-Corrected Prognostic Signature for Cervical Cancer.

Introduction: Tumor heterogeneity challenges the accuracy of existing prognostic models for cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC). Prognostic models that correct for tumor heterogeneity remain unexplored. Methods: Based on the GSE5787 dataset comprising 33 multiregional samples from 11 patients with CESC, we calculated both interpatient heterogeneity (IPH) and intratumoral heterogeneity (ITH) scores. Genes with low heterogeneity were selected to construct a prognostic model using the CESC cohort from The Cancer Genome Atlas (TCGA). Key gene expression was validated in HeLa and SiHa cell lines. The functional consequences of gene knockout were assessed through Cell Counting Kit-8 (CCK-8) proliferation assays, wound healing migration assays, and Transwell assays. Results: Genes exhibiting low heterogeneity were predominantly associated with energy metabolism pathways, whereas those displaying high heterogeneity were enriched in immune-related signaling pathways. The risk score based on Armadillo Repeat Containing X-Linked 1 (ARMCX1) and Signal Transducing Adaptor Family Member 2 (STAP2) achieved area under the curve (AUC) values of 0.74, 0.68, and 0.74 for predicting 1-year, 3-year, and 5-year overall survival, respectively. Importantly, this signature demonstrated robust performance with a risk classification inconsistency rate of merely 18%, outperforming 10 previously published CESC prognostic models. The integrative nomogram incorporating both the risk score and clinical staging demonstrated enhanced prognostic accuracy. Although patients in the high-risk group exhibited diminished immunotherapy efficacy, drug sensitivity screening identified seven potentially effective therapeutic agents as alternative treatment options. In vitro experiments demonstrated that STAP2 knockdown suppressed HeLa cell proliferation, migration, and invasive capacity. Conclusion: This study establishes a robust gene signature through ITH correction, which may mitigate single-region sampling bias and offers a promising tool for prognostic stratification in CESC.