基于RNA编辑的风险模型预测结直肠癌肝转移中的肝外转移

Background: The management of colorectal cancer liver metastases (CRLM) remains an urgent challenge in metastatic colorectal cancer. The efficacy of liver transplantation for unresectable CRLM has recently introduced a new treatment option. However, extrahepatic metastasis (EHM), which is one of the key exclusion criteria for liver transplantation, is a poor prognostic factor in CRLM. Therefore, establishing a method to accurately detect EHM prior to treatment is needed. Existing imaging diagnostics have limitations in detecting small lesions, necessitating complementary biomarkers. Adenosine-to-inosine RNA editing, which is a post-transcriptional modification driven by adenosine deaminase acting on RNA (ADAR), promotes tumor malignancy and the acquisition of metastatic potential. We have previously reported that ADAR1 expression in liver metastases is a predictor of residual liver recurrence, but the significance of RNA editing in extrahepatic metastasis remains unclear. This study aimed to analyze RNA editing profiles in primary tumors and liver metastases to clarify the clinical significance of RNA editing in EHM. Methods: We analyzed gene profiling datasets from normal tissue, primary tumor and liver metastasis in silico discovery from GEO (Gene Expression Omnibus) for both gene expression analysis and RNA editing analysis. For clinical validation, we analyzed 70 CRLM cases (39 EHM and 32 non-EHM). Results: Public data analysis revealed significantly higher ADAR1 expression in primary tumors and liver metastases compared to normal colon tissue (P = 0.03, P < 0.001, respectively). RNA editing analysis identified numerous RNA editing events specific to both primary tumors and liver metastases, in addition to common editing events shared between the two sites. These findings suggest that the RNA editing program regulated by ADAR1 changes according to the metastatic site, implying involvement in organ specificity and multiorgan metastasis. In clinical validation, primary tumor ADAR1 expression was an independent predictor of EHM alongside sex (female) and poorly differentiated histology. Furthermore, a risk model that incorporated CEA and the presence of multiple CRLM in addition to the three independent predictors effectively predicted EHM (AUC = 0.78), suggesting clinical utility. Further characterization of site-specific RNA editing events may improve predictive performance. Conclusion: ADAR1 expression is a promising biomarker for predicting EHM in patients with CRLM. Identification of EHM-specific RNA editing events may enable the development of non-invasive and highly accurate metastasis prediction models with clinical applicability.