Objective      To summarize the survival outcomes of colorectal cancer with liver metastasis (CRCLM) after curative resection

and analyze the risk factors associated with intrahepatic recurrence.       Methods      From January 1993 to December 2013

a total of 164 patients with liver metastasis from colorectal cancer

who accepted radical resection

were enrolled into the study. Survival curve and disease-free survival curve were constructed. Univariate analysis and multivariate analysis were performed to study the risk factors associated with intrahepatic recurrence after surgery.       Results      The median disease-free survival (DFS) rate and overall survival (OS) rate after surgery were 12 months and 32 months (1~162 months)

respectively. The 1-year

2-year

3-year and 5-year DFS rates were 49.4%

38.4%

25.0% and 14.0%

respectively

and the corresponding OS rates were 90.9%

72.1%

57.1% and 26.0%

respectively. Postoperative mortality was 16.5%

and 30-day morbidity was 1.2%. Intrahepatic recurrence was found in 46.3% of patients postoperatively. The frequencies of intrahepatic recurrence were as follows: marginal 6.6%

same segment 9.2%

adjacent segment 9.2%

distal segment 17.1%

and multiple segment 57.9%. There were no significant differences in the sites of intrahepatic recurrence. Based on univariate analysis

the primary tumor cell differentiation (P=0.035)

carcinoembryonic antigen (CEA) (P=0.013)

N-status (P=0.043) and resection margin width (P=0.012) were risk factors of postoperative intrahepatic recurrence. Based on multivariate analysis

the primary tumor cell differentiation (P=0.021) and CEA (P=0.015) were independent risk factors of postoperative intrahepatic recurrence.       Conclusion      Radical resection for CLCLM results in long-term survival

and it offers potentially curative treatment for hepatic metastasis from colorectal cancer. Primary tumor cell differentiation and preoperative plasma CEA level are independent risk factors for predicting postoperative intrahepatic recurrence.