This investigation has been conducted in accordance with the ethical standards, the Declaration of Helsinki, and national and international guidelines. The present study was approved by the Institutional Review Board of the Severance Hospital (IRB no. 4-2017-0157).

A total of 52 patients who received curative resection between 2004 and 2005 at Yonsei University Severance Hospital, Seoul, Korea for pathologically confirmed adenocarcinoma originating in the colon were enrolled. Patients who had metastasis, palliative resection, previous cancer treatment, or no available paired blood samples from both the tumor DV and the PV were excluded.

Blood samples (5 mL) were drawn from the PV and tumor DV and placed in plain tubes. PV blood was obtained 1 hour prior to surgical incision. DV blood was obtained prior to the ligation of any branch of the SMV in patients with proximal colon cancer and prior to ligation of the IMV in patients with distal colon cancers. Samples were immediately centrifuged, and plasma and serum were separately stored at −70°C until analysis. enzyme-linked immunosorbent assay (ELISA) was performed using commercially available kits (IBL-Hamburg GmbH, Hamburg, Germany) according to manufacturer’s instructions. Diluted serum was transferred to wells of plates pre-coated with primary antibody. Following the recommended incubation period, plates were washed with buffer solution, and substrate solution was added and incubated as instructed. The wells were developed with a color-reagent. Stop solution was added to each well after incubation and the optical density was measured at a wavelength of 450 nm using automated optical densitometry. Each sample was run in triplicate, and the mean value was used for analysis. If the R² values of the standard solutions was < 0.98, data from the plate were excluded.

Statistical analysis was performed by using SPSS ver. 18.0 (SPSS Inc., Chicago, IL, USA).

Analyzed clinical parameters included age, gender, tumor node metastasis (TNM) stage, histological grade, and lymphovascular invasion. Data on molecular factor levels are shown as the mean± standard deviation. The Mann-Whitney U test, Fisher’s exact test, and McNemar test were used to compare differences in molecular factor levels between the clinical and pathological features. A Cox proportional hazards analysis was used to evaluate the relationship between pathological features, plasma molecular factor levels, and overall survival. A P-value < 0.05 was considered significant.

A total of 52 patients were enrolled in this study. The mean age was 60 years; approximately half of patients had stage 3 disease (Table 1). The mean pCEA and dCEA levels were 14.4± 32.0 ng/mL and 65.2± 144.7 ng/mL (P= 0.001), respectively. There was a moderate correlation between pCEA and dCEA levels (Pearson correlation coefficient: 0.6736).

Patients were then divided into 2 groups according to whether their peripheral CEA exceeded the reference value of 5.9425 ng/mL (Fig. 1A). Clinicopathologic variables were compared; none was significantly different (Table 1). The dCEA levels in each group were 31.8± 31.9 ng/mL and 93.8± 192.0 ng/mL, respectively, but were not statistically different (P= 0.103).

To determine cut-off values for pCEA and dCEA levels, receiver operating characteristic (ROC) curves were analyzed. Through ROC curve analysis, a point most correlated to recurrence with the highest sensitivity and specificity was calculated for both pCEA and dCEA levels. For pCEA, it was determined to be 5.9425 ng/mL, the same as the reference value, for which sensitivity and specificity were 0.667 and 0.588, respectively (Fig. 1A). For dCEA, the point was 20.192 ng/mL; sensitivity and specificity were 0.833 and 0.529, respectively (Fig. 1B).

For given cut-off values of pCEA and dCEA, Kaplan-Meier survival analyses with log rank tests were performed. When the analysis was performed according to a pCEA level below 5.9425 ng/mL, both disease-free and overall survival curves were not significantly different (Fig. 2). However, patients with dCEA below 20.192 ng/mL showed better disease-free (P = 0.009) and overall survival (P= 0.033) curves than those with dCEA = 20.192 ng/mL (Fig. 3).

The Cox proportional hazard model analysis revealed the dCEA level to be a significant prognostic factor for disease-free (hazard ratio [HR]=399; 95% confidence interval [CI], 16.4–9,747; P<0.001) and overall survival (HR= 9.39; 95% CI, 1.29–68.006; P= 0.026) (Table 2). For disease-free survival, significant poor prognostic factors other than high dCEA level included a poorly differentiated tumor, complications, a Clavien-Dindo grade = 3, patients > 60 years of age, TNM stage 3 or 4, and intraoperative blood transfusion. For overall survival, factors included poorly differentiated tumors, complications, a Clavien-Dindo grade = 3, or a lymph node harvest below 12.

A subgroup analysis was performed to determine whether the dCEA level has prognostic significance in patients with a normal pCEA level. Of 51 patients in this study, 24 (47.06%) had a pCEA level in the normal range of < 5.9425 ng/mL; these patients were divided into 2 groups according to whether their dCEA level was below the calculated cut-off value (20.192 ng/mL). Kaplan-Meier survival analysis revealed that patients with a dCEA level = 20.192 ng/mL showed significantly poorer disease-free survival than those with dCEA < 20.192 ng/mL (P= 0.009) (Fig. 4). A Cox proportional hazard model revealed that none of the analyzed variables, including the dCEA level, were significant prognostic factors for both disease-free and overall survival.