Background
Gastric cancer was the fourth common cancer worldwide, and it was the second most common cause of death from cancer [
1]. Although the survival of gastric cancer had been improved dramatically in some countries, especially in Japan, possibly due to early diagnosis following a great number of endoscopic examinations performed for gastric disorders, survival was still relatively low in North America and Western Europe even in cases treated by radical surgery, which raised the question whether the molecular pathology of gastric cancer was similar worldwide [
2‐
4]. Therefore, a major impetus for the current study was the lack of previous studies on a large-scale Chinese population evaluating the expression of molecular prognostic markers in gastric cancer patients. Because of the variability of prognosis within a clinical or pathological stage of gastric cancer at presentation, which showed tumor stage could not provide complete information on the clinical behavior of gastric cancer, there has been a constant search for specific biological markers to identify subgroups of patients with more aggressive course of disease [
5‐
7]. Some serum tumor markers including alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), carbohydrate antigen (CA) 19-9, CA50, and CA72-4 have been reported to be elevated in some patients with gastric cancer [
8‐
10]. AFP, discovered about half a century ago by Abelev
et al., was a sensitive marker for diagnosis of hepatocellular cancer [
11]. The serum level of AFP also increased in AFP-producing gastric cancer [
12,
13]. CEA, originally described in 1965 by Gold and Freedman, was routinely used as a serum marker for colorectal cancer [
14]. CEA was a glycoprotein from the family of immunoglobulins whose function was to promote cellular binding. CA19-9, first described by Koprowski
et al. in 1979 as a marker for colorectal cancer, had become the most important tumor marker for pancreatic adenocarcinoma [
15]. CA19-9 was a high-molecular weight mucin that played a role in the adhesion of cancer cells to endothelial cells. CA50, initially screened out of colorectal cancer cell lines by Holmgren
et al. in 1984, was a kind of glycolipid antigen that played an important part in growth and differentiation of cell [
16]. CA72-4 was a complex glycoprotein that elevated in the serum of patients’ breast, pancreatic, ovarian, colon, and gastric cancers. CA72-4 was regarded as one of the most specific and sensitive markers for gastric cancers. However, it has not been routinely tested for gastric cancer patients in our hospital before 2005. At the present time, the value of these tumor markers in T4a stage gastric cancer was still elusive, which was responsible for more than 40% of gastric cancer and likely to present abnormal serum level of tumor markers. In this retrospective study, we evaluated the association between tumor markers and clinicopathological features and the prognostic value of tumor markers in T4a stage gastric cancer.
Discussion
The main findings of this study were: (1) tumor markers including CEA, CA19-9, and CA50 were independent prognostic factors for T4a stage gastric cancer; and (2) there were significant differences of overall 5-year survival rate between CEA (+) and CEA (-) according to stages II and III; between CA19-9 (+) and CA19-9 (-) according to stage III; between CA50 (+) and CA50 (-) according to stage III.
Various tumor markers have been identified since Gold and Freedman first reported the discovery of carcinoembryonic antigen (CEA) in 1965 [
14]. These tumor markers have been studied primarily for applicability and feasibility in terms of tumor early detection. Among these tumors, AFP, CEA, CA19-9, CA50, and CA72-4 were considered as relatively specific markers for gastric cancers. In particular, CA 72-4 was regarded as one of the most specific and sensitive markers for gastric cancers. However, as a retrospective study, we could only evaluate the prognostic value of AFP, CEA, CA19-9, and CA50 for gastric cancers, which were evaluated in our institute. A few of studies have investigated the value of various tumor markers, including CEA, CA19-9, and CA50 in gastric cancer [
21‐
23]. Some reports showed that one or more tumor markers could predict specific clinical outcomes such as prognosis, response to treatment, and recurrence, although results were not consistent. Furthermore, these studies have been counteracted by several methodologic flaws: (1) the inclusion of patients with various tumor stages raging from localized to metastatic; and (2) a heterogeneous treatment regimen. So it is necessary to clarify the value of tumor markers in homogeneous group of patients with gastric cancer.
We only evaluated patients with T4a N0-3 M0 gastric cancer, who received D2 gastrectomy. The reasons for including only T4a gastric cancers as follows: (1) There have been some studies on the prognostic impact of tumor markers in gastric cancer, but rarely have the previous studies evaluated the prognostic impact of tumor markers when specific depth of invasion was involved, especially in the T4a classification which was responsible for more than 40% of gastric cancers. Therefore, the precise determination of the prognostic value of tumor markers in T4a gastric cancer has substantial clinical importance. (2) Since some previous studies showed an association between the depth of tumor invasion and serum tumor markers levels [
24,
25], it was difficult to identify the important prognostic factors as a result of interrelated factors. In each case, 15 or more lymph nodes were dissected according the AJCC/TNM classification. In this study, the positive rate of patients with elevated serum AFP levels was 5.9%, which was similar to those reported by other investigators [
12,
24,
26]. The positive rate of CEA and CA19-9 were 16.1% and 32.6%, respectively, which are lower than that of other studies [
27,
28]. The corresponding proportion of patients with elevated serum CA50 levels was 29.7%. The positive rates of tumor markers are thought to be influenced by tumor progression at the time of presentation. When the serum positive rates of AFP, CEA, CA19-9, and CA50 were evaluated according to TNM stage grouping, it showed that the positive rates increased gradually with stage.
We compared other clinicopathological factors between patients with elevated tumor markers and those with normal levels of serum tumor markers. The proportion of patients with elevated serum AFP was significantly higher in those with Bormmann IV. The elevated serum CEA was associated with gender, tumor site, and pN stage. The elevated serum CA19-9 was associated with age, tumor site, lymphovascular invasion, and pN stage. CA50 positivity was significantly associated with age, tumor size, tumor site, and pN stage. We found that these tumor markers were associated with pN stage. This finding indicated that the positive rates of tumor markers increased as the tumor progressed. This was agreement with previous studies of AFP, CEA, and CA19-9 [
25,
26]. Additionally, we found that the positive rate of CEA was higher in males than in females, which was not consistent with previous some studies [
26,
29]. It was possible that the limited samples in these studies contributed to the negative correlation between CEA and gender. This result indicated that the serum CEA should be investigated in patients with gastric cancer, especially for male patients.
According to univariate analysis, our results showed that there was a significant difference in 5-year overall survival in terms of tumor markers and distinct clinicopathologic factors, which included age, tumor size, tumor site, Borrmann type, lymph node stage, nervous invasion, and lymphovascular invasion. We found that the serum levels of AFP, CEA, CA19-9, and CA50 were significantly correlated with survival rate in patients with T4a stage gastric cancer, which was in agreement with previous studies of AFP, CEA, and CA19-9 [
25,
26,
29]. These correlations indicated that patients with positive values of tumor markers have worse prognosis, which in turn may be due to the predominant proportion of advanced gastric cancer in this cohort of patients.
Multivariate analysis using Cox proportional hazards model showed that tumor markers including CEA, CA19-9, CA50, were independent prognostic factors, as tumor size, lymph node stage, and nervous invasion. The Cox proportional hazards regression analysis showed that patients with elevated levels of CEA, CA19-9, and CA50 had a higher risk of death than patients with low levels of these markers. Except CA50, the prognostic value of CEA and CA19-9 in gastric cancer had been widely studied. Tocchi
et al. [
30] found that CEA and CA19-9 provided independent predictive value in gastric cancer patients, but the other studies did not show consistent result [
26,
29]. This was likely due to the heterogeneity of patients included in these studies, including those with localized disease who undergo gastrectomy and those with locally advanced and disseminated disease who may not undergo resection.
To evaluate whether serum CEA, CA19-9, and CA50 could provide additional prognostic information on the basis of AJCC/TNM stage system, we compared cumulative survival curves according to CEA, CA19-9, and CA50. The results showed that there were significant differences between patients with elevated levels and those with normal serum CEA, CA19-9, and CA50 levels at stage III. These findings indicated that serum CEA, CA19-9, and CA50 levels could provide additional prognostic information in part of the patients with T4a stage gastric cancer. Similar results in patients with gastric cancer have been reported by others. A previous study reported that the survival rate of gastric cancer patients at stages I, II, and III with elevated serum CEA levels was significantly poorer than that of patients with normal levels [
31]. Another study reported that there were significant differences between patients with elevated CA19-9 levels and those with normal levels at stage I [
23]. The reason that preoperative levels of tumor markers could influence long-survival of T4a gastric cancer was still unclear; it was possible that a number of biological factors are involved. CEA and CA19-9 belonged to intercellular adhesion molecules, so cells expressing these glycoproteins may have a greater invasive potential [
8]. CA50 acted as a kind of glycolipid antigen that played a role in growth and differentiation of cell, suggesting that cells expressing this antigen would possess increased proliferating activity [
16].
Some limitations of this study should be acknowledged. Firstly, we only evaluated patients with T4a N0-3 M0 gastric cancer. Although this is a design of the study, only inclusion of T4a stage gastric cancers would limit the application of the results to the early stage or more unfavorable moderate stage gastric cancer. Secondly, CA72-4 was not routinely tested for gastric cancer before 2005 in our hospital, therefore its predicting value for gastric cancer was not known. Thirdly, as a result of uncompleted data about recurrence, it was unable to evaluate the correlation between tumor markers and recurrence. Fourthly, we could not determine the prognostic value of peritoneal cytology in our study, because we did not perform peritoneal cytology in the management of gastric cancer. Therefore, this may have influenced the survival data. In addition, some other molecular markers such as cell-free DNA (cfDNA) and miRNA should be investigated in future. Previous studies have demonstrated an increase of circulating cfDNA in different types of cancer [
32]. MicroRNA (miRNA) played an important role in regulating gene express. Chen
et al.[
33] found that expression pattern of serum miRNA was altered in reflection of various disease.
Competing interests
There is no any conflict of interest about the study.