An ABCC1-based risk model is effective in the diagnosis of synchronous peritoneal metastasis in advanced colorectal cancer.

Abstract

Background: The presence of peritoneal metastasis (PM) in colorectal cancer (CRC) patients indicates an advanced CRC stage. Prompt diagnosis and early PM detection are difficult, and the underlying mechanisms are unclear, resulting in limited treatment options and unsatisfactory therapeutic effects. We aimed to identify applicable biomarkers for accurately diagnosing synchronous PM in CRC patients.

Methods: Differentially expressed genes between synchronous and non-synchronous PM groups were identified via label-free proteomic analysis of primary tumors from 31 CRC patients. Quantitative real-time PCR, multiplex and conventional immunohistochemistry were used to validate gene expression. We attempted to construct a logistic regression risk model for the diagnosis of PM, which was tested in a training cohort and validated in an independent cohort.

Results: Utilizing the results from multi-omics, we established an ABCC1-based risk model. In CRC patients with imaging-negative diagnoses, the model identified patients with metastases including PM (AUC = 0.892, 95% CI: 0.840-0.944) or those with PM only (AUC = 0.859, 95% CI: 0.794-0.924); these results were validated in an independent cohort of patients with metastases including PM (AUC = 0.831, 95% CI: 0.729-0.933) or PM only (AUC = 0.819, 95% CI: 0.702-0.936). In CRC patients with CEA-negative, this model more effectively distinguishes those with exclusive peritoneal involvement, with consistent results across both training (AUC = 0.913, 95% CI: 0.854-0.972) and validation (AUC = 0.869, 95% CI: 0.795-0.943) cohorts. Additionally, in CRC patients with PM, low ABCC1 may serve as a predictive marker for chemotherapy efficacy.

Conclusions: The ABCC1-based risk model effectively predicts PM in CRC, complementing traditional diagnostics. ABCC1 may serve as a predictive marker for chemotherapy efficacy in PM.