Plasma Methylated SEPT9 as a Novel Biomarker for Predicting Liver Metastasis in Colorectal Cancer

This study aimed to explore the role of plasma methylated SEPT9 (mSEPT9) in predicting liver metastasis (LM) in colorectal cancer (CRC) patients. The clinicopathological information of 115 consecutive CRC patients were collected. The differences of clinical characteristics and several biomarkers between CRC patients with LM and those with non-liver metastasis (NM) were analyzed. Multivariate logistic regression analysis was used to identify the risk factors for predicting LM in CRC patients. Receiver operating characteristic curve (ROC) analysis was applied to investigate the sensitivity and specificity of potential biomarkers in indicating the presence of LM in CRC. Compared with the CRC without LM, the levels of plasma mSEPT9 and carcinoembryonic antigen (CEA) were significantly increased in CRC with LM. Multivariate logistic regression analysis showed that plasma mSEPT9 was an independent risk factor for predicting LM in CRC. ROC curves showed that mSEPT9 and CEA could efficiently distinguish LM from NM in CRC. The area under the curve (AUC) of mSEPT9 was 0.850, which was slightly higher than that of CEA (0.842). The optimal cut-off value of mSEPT9 was 35.09 with a sensitivity of 81.82% and a specificity of 73.33%, both similar with that of CEA (sensitivity 87.27% and specificity 75.00%). In addition, the combination of mSEPT9 and CEA had a higher specificity than CEA alone (81.70% Vs 75.00%). Our findings suggest, for the first time, that plasma mSEPT9 might serve as a potential biomarker to predict LM in CRC, which deserves further in-depth study.

Automated summary

This study aimed to investigate the role of plasma mSEPT9 in predicting LM in CRC patients. The levels of plasma mSEPT9 and CEA were significantly increased in CRC patients with LM. Multivariate logistic regression analysis showed that plasma mSEPT9 was an independent risk factor for predicting LM in CRC. ROC curves showed that mSEPT9 and CEA could efficiently distinguish LM from NM in CRC.