Pre-operative to post-operative serum carcinoembryonic antigen ratio is a prognostic indicator in colorectal cancer

We explored the prognostic significance of the pre-operative-to-post-operative serum carcinoembryonic antigen (pre-post-CEA) ratio in colorectal cancer (CRC). We detected pre- and post-operative CEA levels in 2035 CRC patients surgically treated at First Affiliated Hospital of Zhengzhou University between June 2001 and June 2011. Univariate analysis revealed the pre-post-CEA ratio is associated with distant metastasis and degree of tumor differentiation (both P < 0.05). Multivariate analysis showed that the pre-post-CEA ratio is associated with lymphatic and distant metastasis, tumor-node-metastasis (TNM) stage and degree of tumor differentiation (all P < 0.01). The pre-CEA levels, pre-post-CEA ratios, distant metastasis, TNM stage and degree of tumor differentiation were all associated with 5-yr overall survival (all P < 0.05) based on multivariate analysis. Consequently, pre-CEA levels, pre-post-CEA ratios, distant metastasis and TNM stage are independent risk factors for CRC. We have thus demonstrated that the pre-post-CEA ratio is a prognostic indicator for CRC patients.


INTRODUCTION
Colorectal cancer (CRC) is the third most commonly diagnosed cancer and the fourth leading cause of cancer death worldwide with steadily increasing mortality rates [1,2]. In recent years, lifestyle changes including high-fat diet, lack of exercise and mental stress has resulted in a rise in CRC cases [3]. Radical surgery still remains the best therapeutic option for CRC patients [4]. However, high recurrence and metastasis rates have resulted in poor overall survival of CRC patients [5,6]. Therefore, identification of novel prognostic risk factors is necessary for improving survival rates.
In recent years, tumor biomarkers have been widely used in clinical diagnosis, post-operative monitoring of tumor recurrence, prognosis and curative therapy of CRC patients [7]. The serum carcinoembryonic antigen (CEA) is one of the tumor biomarkers used for predicting recurrence, prognosis and therapeutic efficacy in CRC patients [8,9]. The potential clinical use of changes in pre-operative and post-operative CEA levels (pre-post-CEA) has been recognized in few cancer studies [10,11]. Increased serum CEA levels were observed after radical surgery in some patients [12,13]. The significance of preand post-operative CEA levels is controversial. Hotta et al. reported that the pre-post-CEA ratio is prognostic predictor after surgery for stage III rectal cancer patients [10]. However, another study reported that post-operative serum CEA (post-CEA) levels were more definitive as prognostic prediction than pre-post-CEA ratios in nonsmall cell lung cancer patients. Therefore, in this study, we explored the prognostic relevance of pre-post-CEA ratios in predicting the survival time of CRC patients after radical surgery.

Univariate analysis of correlation between pre-CEA levels and pre-post-CEA ratios with clinicopathological parameters of CRC patients
The patients were divided into low and high expression groups based on pre-CEA levels and prepost-CEA ratios based on their mean values. Univariate analysis showed that pre-CEA levels were associated with smoking, tumor size, lymphatic and distant metastases and TNM stage (P < 0.05 for all parameters; Table 1). The pre-post-CEA ratios were associated with distant metastasis and degree of tumor differentiation (both P < 0.05; Table 1). Pre-CEA levels were not associated with gender, age, family history, tissue type, degree of tumor differentiation and tumor location. Meanwhile, pre-post-CEA ratios were not associated with gender, age, family history, smoking, tumor size, lymphatic metastasis, TNM stage, tissue type and tumor location (all P > 0.05).

Multivariate analysis of the correlation between pre-CEA levels and pre-post-CEA ratios with clinicopathological parameters of CRC patients
Multivariate analysis showed that pre-CEA levels were associated with family history, smoking, tumor size, distant metastasis and TNM stages (all P < 0.05; Table 2). However, they were not associated with other cancer in combination, lymphatic metastasis, tissue type, general tumor type, degree of tumor differentiation and tumor location (all P > 0.05). Meanwhile, pre-post-CEA ratios were associated with lymphatic and distant metastases, TNM stage and degree of tumor differentiation (all P < 0.01; Table 2). But, they were not associated with family history, smoking, other cancer in combination, tumor size, tissue type, general tumor type and tumor location (all P > 0.05).
Univariate analysis of the correlation between pre-CEA levels and pre-post-CEA ratios with 5-year OS As shown in Table 3, univariate analysis demonstrated that pre-CEA levels, pre-post-CEA ratios, tumor size, TNM stage, lymphatic metastasis, distant metastasis and degree of tumor differentiation were associated with 5-year overall survival (OS, all P < 0.05). Factors like age, gender, family history, smoking, tissue type and tumor location were not related with 5-year OS (all P > 0.05).
Multivariate analysis of the correlation between pre-CEA levels and pre-post-CEA ratios with

5-year OS
Multivariate survival analysis showed that preoperative CEA, pre-to post-operative CEA ratio, distant metastasis, TNM stage and degree of tumor differentiation were associated with 5-year OS (all P < 0.05, Table 4). Meanwhile, gender, age, family history, smoking, tumor size, lymphatic metastasis, tissue type, general tumor type and tumor location were not associated with 5-year OS (all P > 0.05, Table 4). Therefore, our analysis demonstrated that pre-CEA levels, pre-post-CEA ratios, distant metastasis, TNM stage and differentiation degree were independent risk factors for 5-year OS of CRC patients.

Kaplan-Meier analysis of association of overall survival with pre-CEA levels and pre-post-CEA ratios
Kaplan-Meier survival curve analysis showed that overall survival decreased post-operatively ( Figure 2). Interestingly, patients with low pre-CEA levels demonstrated survived longer than those with high pre-CEA levels (log-rank test, P < 0.001; Figure 3). Meanwhile, patients with a lower pre-post-CEA ratios survived longer compared to those with higher pre-post-CEA ratios (log-rank test, P < 0.001; Figure 4). Also, patients with low pre-CEA levels and pre-post-CEA ratios (group 1) showed longer survival than patients with high pre-CEA levels and pre-post-CEA ratios (group 2). However, patients with lower pre-CEA levels, but high pre-post-CEA ratios (group 3) and patients with higher pre-CEA levels and low pre-post-CEA ratios (group 4) showed no differences with overall survival (log-rank test, P > 0.05). Also, patients in groups 3 and 4 showed lower OS than group 1 patients and higher OS than group 2 patients (log-rank test, both P > 0.05; Figure 5). This suggested that the pre-CEA levels and pre-post-CEA ratios had a synergistic effect in prognostic prediction of overall survival of CRC patients.

DISCUSSION
Carcino-embryonic antigen (CEA) represents a group of macromolecular acid glycoproteins expressed highly during fetal development and reduced in the adults [14,15]. High CEA levels were first observed in colorectal cancer and then in other malignant tumors of Oncotarget 54674 www.impactjournals.com/oncotarget the esophagus, stomach, liver and pancreas [16,17]. CEA is secreted by cancer cells resulting in high serum CEA levels.
Many studies have suggested that CEA levels can be used for predicting metastasis although few contradictory reports exist [18][19][20]. High CEA levels are observed in cancers, whereas changes are also observed in smoking, inflammation, pregnancy, gynecological disease and hepatitis conditions [21,22]. Furthermore, prognostic potential of pre-post-CEA ratio has been reported for    patients with high (red) and low (blue) pre-operative CEA levels compared to total (green) CRC patients based on Kaplan-Meier curves and log-rank test. As shown, patients with low pre-CEA levels survived longer than those with high pre-CEA levels (log-rank test, P < 0.001). with low (blue) and high (red) pre-post-CEA ratios compared to total (gray) number of CRC patients based Kaplan-Meier curves and log rank test. As shown, patients with low pre-post-CEA ratios survived longer than those with high pre-post-CEA ratios (log-rank test, P < 0.001).
Oncotarget 54677 www.impactjournals.com/oncotarget CRC patients. In gastric cancer, elevated pre-operative CK19 and CEA mRNA levels were associated with lymph node metastasis and predicted poor prognosis [22]. Yang et al. also reported that elevated pre-and post-operative serum CEA levels were associated with recurrence and poor clinical outcomes in few CRC patients [23]. Duan et al showed that high serum CEA or CYFRA 21-1 levels before and after surgery were associated with poor overall survival (OS) and progress-free survival (PFS) in stage I non-small cell lung cancer [11]. In this study, we explored the prognostic significance of pre-post-CEA ratios for CRC patients. We observed that pre-post-CEA ratios were closely related with smoking, tumor size, lymph node and distant metastases and post-operative pathological TNM staging. Also, our analysis demonstrated that distant metastasis and post-operative TNM stage were independent risk factors along with pre-CEA levels and pre-post-CEA ratios, consistent with previous reports. Lymph node metastasis [24] and distant metastasis, particularly liver [25,26] are independent risk factors with pre-CEA levels for CRC. The cell recognition, adhesion and immunosuppressive properties of CEA, which are Ca2+ dependent cell adhesion molecules aids aggregation, adhesion, invasion and metastasis of tumor cells [27,28]. Therefore, excessive serum CEA levels are beneficial to CRC metastasis. Our study also demonstrated that lymphatic and distant metastases were independent risk factors in advanced stage III and IV CRC.
In previous studies, pre-CEA levels were identified as reliable indicators of prognosis and therapeutic efficacy in stage II and III CRC patients [18,29]. Many studies also reported that pre-CEA levels were an independent risk factor for CRC patients [18,[30][31]. Our study further confirmed that pre-CEA level was an independent prognostic factor. Therefore, pre-CEA levels can be used evaluate therapeutic efficacy, post-operative tumor recurrence and predict prognosis of CRC patients.
Apart from the prognostic value of pre-CEA levels, few studies have also reported the prognostic value of preoperative and post-operative CEA levels in combination [10]. Therefore, we analyzed the significance of pre-post-CEA ratios and found that it was associated with lymphatic and distant metastases and post-operative pathological TNM stage. These three indicators were also shown to be independent risk factors along with pre-post-CEA ratios.
Based on tumor source, the degree of malignancy determined by histology has been used to classify different pathological types of tumors [32,33]. Most CRC patients are associated with moderately differentiated tumors, whereas high CEA levels are associated with poorly differentiated or undifferentiated CRC that are highly malignant [34]. In our study, pre-operative serum CEA levels were higher in lowly differentiated and undifferentiated carcinomas. Further, the differentiation degree of tumors affected the pre-post-CEA ratios in CRC patients. CEA is an important risk factor affecting CRC prognosis [13]. We observed that patients with decreasing post-operative CEA demonstrated better prognosis than those with higher post-operative CEA. Therefore, prepost-CEA ratio is potentially an important prognostic marker.
During malignant transformation of CRC cells, overexpression of CEA lowers the function of other adhesion molecules and interferes in the interaction between normal cells, thereby weakening the singlelayer structure of glandular tube [27][28][29][30][31][32][33][34][35]36]. Therefore, excessive CEA in combination with genetic damage would promote malignancy of tumors. According to the NCCN guidelines, CRC of TNM stages 3 and 4 are accompanied by lymphatic and distant metastases, severely damaged CRC patients with low pre-CEA levels plus low pre-post-CEA ratio (blue) survived longer than those with high pre-CEA levels plus high pre-post-CEA ratios (red). CRC patients with low pre-CEA levels plus high pre-post-CEA ratios (green) and those with high pre-CEA levels plus low pre-post-CEA ratios (gray) show comparable survival rates (log-rank test, P > 0.05) and are intermediate in comparison to the previous two groups (log-rank test, all P > 0.05).
Oncotarget 54679 www.impactjournals.com/oncotarget colon tissues and CEA overexpression. Therefore, high pre-CEA levels and lymphatic or distant metastases were used as prognostic indicators and lymph node or microdistant metastasis was neglected during surgery [37]. After surgery, post-CEA levels would increase leading to high pre-post-CEA ratios. In our study, pre-post-CEA ratio was closely related with lymphatic and distant metastases, TNM stage and degree of tumor differentiation, consistent with previous reports.
There is consensus that lymphatic metastasis, distant metastasis and TNM stage dictate prognostic status of CRC patients. In our study, these three indicators were also independent risk factors in relation to for pre-post-CEA ratios. Therefore, pre-post-CEA ratios are an essential risk factor for prognosis of CRC patients.
In conclusion, pre-post-CEA ratios were associated with the malignant phenotype of lymphatic metastasis, distant metastasis, TNM stage and degree of tumor differentiation in CRC patients. It was also an independent risk factor for CRC patients. Therefore, we postulate that pre-post-CEA ratio is a potential tumor biomarker to evaluate prognosis and therapeutic efficacy in CRC patients.

Enzyme linked immunosorbent assay (ELISA) assay
Blood samples (4ml each) were obtained within the week before surgery for preoperative CEA estimation and 1-2 weeks after the surgery for post-operative CEA estimation. Serum CEA levels were estimated by ELISA kit (ab183365, Abcam, England) according to manufacturer's instructions.

Patient follow-up
After surgery, the patients were assessed once a month for the first 6 months, once every 3 months for the first 2 years, once every 6 months until 5 years followed by once a year. Follow-ups were either by outpatient or inpatient review or by telephone. 216 patients did not participate in the follow-up analyses because they did not communicate with the physicians after surgery. In addition, 13 patients developed dysthymia and were unable to cooperate with the study; 3 patients committed suicide and 19 patients did not participate in the follow-up for unknown reasons. Therefore, the total follow-up rate in the study was 87.67%.

Chemotherapy and radical surgery
FOLFOX6 was used as the first line adjuvant or neoadjuvant therapy scheme for CRC patients with high-risk IInd or IIIrd stage. CapeOX was used as the chemotherapy scheme as either 1st or 2nd line adjuvant or neoadjuvant therapy for CRC patients with high-risk IInd, IIIrd or IVth stage, drug resistance or postoperative recurrence. FOLFIRI was used as the chemotherapy scheme for CRC patients in advanced IVth stage, postoperative recurrence, metastasis or drug resistance.
Radical surgery was performed according to complete mesocolic excision (CME) for colon cancer patients and total mesorectal excision (TME) principle for rectal cancer patients.

Statistical analysis
All statistical analyses were carried out using SPSS for Windows version 20 (SPSS Inc., Chicago, IL, United States). Univariate analysis was performed by χ 2 test to analyze pre-CEA levels and pre-post-CEA ratios and clinicopathological parameters. Kaplan Meier survival curves and log-rank test were used to compare the low and high level groups of pre-CEA levels and pre-post-CEA ratios. Multivariate survival analysis was performed by the Cox regression model to determine relative risk (RR) and 95% confidence intervals (CI). Statistical significance was defined as P < 0.05.