Prognostic value of postoperative CEA/CA19-9 positivity and elevation relative to preoperative levels in gastric cancer

Abstract

BACKGROUND

Gastric cancer (GC) is a common malignant tumor. Elevated preoperative levels of serum markers are significantly associated with poor long-term survival. However, few studies have systematically analyzed the prognostic significance of postoperative elevation of carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) levels compared with preoperative levels in GC.

AIM

To evaluate whether postoperative positivity and elevation of CEA and CA19-9 relative to preoperative levels were associated with overall survival (OS) and disease-free survival (DFS) in patients with GC after curative gastrectomy.

METHODS

Patients who received radical gastrectomy for GC at the First Affiliated Hospital of Anhui Medical University and Anhui Provincial Hospital between January 2013 and December 2017 were included. OS and DFS were evaluated using the Kaplan-Meier method and compared using the log-rank test. Univariate and multivariate Cox proportional hazards regression analyses were used to assess prognostic factors for OS, and DFS was additionally analyzed to evaluate recurrence-related prognostic value. Logistic regression methods were used to explore the potential risk factors for postoperative elevation of CEA and CA19-9.

RESULTS

Patients with elevated postoperative CEA/CA19-9 had poorer prognosis, while patients with normal postoperative CEA/CA19-9 without elevation had the best prognosis. Postoperative CEA or CA19-9 had a negative impact on prognosis of GC. Postoperative elevation of tumor markers was more informative of prognosis of GC compared to preoperative tumor marker levels. Multivariable analyses demonstrated that postoperative elevation of CEA and CA19-9 were independent prognostic risk factors for OS and DFS (P < 0.05). Postoperative CEA elevation was significantly associated with clinical stage, T stage, N stage, tumor differentiation, vascular invasion, recurrence pattern, postoperative CEA and CA19-9 status, PNI, and PLR (P < 0.05). Postoperative CA19-9 elevation was significantly associated with clinical stage, N stage, tumor differentiation, vascular invasion, recurrence pattern, postoperative CEA and CA19-9 status and LMR (P < 0.05).

CONCLUSION

Postoperative CEA and CA19-9 positivity or elevation independently predicted worse OS and DFS after curative gastrectomy for GC; dynamic changes enhanced risk stratification.

Key Words: Gastric cancer; Carcinoembryonic antigen; Carbohydrate antigen 19-9; Prognosis; Postoperative tumor markers

Core Tip: Carcinoembryonic antigen and carbohydrate antigen 19-9 are representative tumor markers of gastric cancer and are considered useful for detecting recurrence after curative gastrectomy. Postoperative elevation of tumor markers was more informative of prognosis of gastric cancer compared to preoperative tumor marker levels. Postoperative elevation of carcinoembryonic antigen and carbohydrate antigen 19-9 was an independent prognostic risk factor for overall and disease-free survival.

INTRODUCTION

Gastric cancer (GC) is the fifth most common cancer worldwide[1]. Despite the worldwide decline in incidence and mortality over the past five decades, GC remains the third leading cause of cancer-related death[2]. According to 2022 estimates from the International Agency for Research on Cancer GLOBOCAN project, there were 968350 new cases of GC (representing 4.9% of all cancer cases diagnosed) and 659853 deaths related to GC[3]. In China, the incidence and mortality rates of GC have decreased gradually in recent years[4].

Because elevated serum markers are associated with tumor progression, most previous reports have concluded that elevated preoperative serum markers are significantly associated with poor long-term survival. Carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), alpha-fetoprotein, CA72-4, and CA125 are representative tumor markers of GC and are considered useful for detecting recurrence after curative gastrectomy[5]. CEA and CA19-9 are the most established tumor markers and are routinely measured after gastrectomy in clinical practice[6]. CEA and CA19-9 play important roles in predicting the occurrence of GC. Previous studies have indicated that patients with elevated CEA levels are more likely to have liver metastasis than those with normal CEA levels. It appears that tumors with elevated CEA levels are more invasive and prone to hematogenous metastasis[7]. Some studies have indicated that CA19-9 rather than CEA is likely to be more useful in detecting recurrence after curative gastrectomy for GC[8]. Takahashi et al[9] reported that CEA and CA19-9 levels increased at the time of recurrence in 54.7% and 40.0% of patients, respectively. The sensitivity of CEA, CA19-9, and their combination for detecting recurrence was 65.8%, 55.0%, and 85.0%, respectively. More than 90% of patients with elevated preoperative CEA levels had a subsequent increase at recurrence. Similarly, CA19-9 levels increased again at recurrence in > 90% of patients with high preoperative levels. Choi et al[10] reported 90% of patients with liver recurrence had elevated CEA levels, whereas postoperative elevation of CA19-9 was more predictive of peritoneal recurrence (78.9%). CA19-9 may be a particularly useful marker of peritoneal recurrence, whereas CEA may be a useful marker of liver recurrence[10]. Moreover, the combined use of tumor markers significantly improved the predictive value compared with using a single marker[11].

The prognostic value of CEA and CA19-9 in GC is well established. However, few studies have systematically analyzed the significance of postoperative elevation of CEA and CA19-9 in the prognosis of GC after radical resection. Therefore, this retrospective study analyzed the importance of postoperative elevation of CEA and CA19-9 levels compared with preoperative levels in the survival of patients with GC.

MATERIALS AND METHODS

Patient cohort

This retrospective study included two comprehensive regional hospitals. Patients who received radical gastrectomy for GC at the First Affiliated Hospital of Anhui Medical University and Anhui Provincial Hospital between January 2013 and December 2017 were included. The inclusion criteria were: (1) Histologically confirmed primary GC; (2) No preoperative neoadjuvant therapy; and (3) Undergoing radical gastrectomy (R0). The exclusion criteria were: (1) Distant metastasis; (2) Age > 80 years; (3) Other critical illnesses; and (4) Incomplete follow-up information.

Clinical characteristics

The clinical characteristics of the patients, including age, sex, TNM stage, degree of differentiation, vascular invasion, and postoperative and preoperative CEA and CA19-9 values, were obtained from electronic medical records. Clinical stage was defined according to the 8^th American Joint Committee on Cancer criteria for GC. Preoperative CEA and CA19-9 levels were tested within 1 week prior to surgery. Postoperative CEA and CA19-9 values were the last test results obtained within 2 months after surgery. Adjuvant chemotherapy may be indicated for patients with advanced GC and for selected early-stage patients with high-risk features. Therefore, postoperative CEA and CA19-9 values were obtained before initiation of adjuvant chemotherapy, to minimize potential chemotherapy-related effects on tumor marker levels. The reference upper limits for CEA and CA19-9 levels were 5 ng/mL and 37 U/mL, respectively. An elevation in tumor markers was defined as a postoperative CEA or CA19-9 value higher than the preoperative value. In addition, based on previous studies, we included the prognostic nutritional index (PNI) and inflammation-related indicators, including the lymphocyte to monocyte ratio (LMR) and platelet to lymphocyte ratio (PLR)[12,13]. Cutoff values for continuous variables were obtained using receiver operating characteristic curves. The optimal cutoff was selected according to the maximum Youden index. Patients were divided into four groups based on postoperative CEA or CA19-9 positivity and whether these markers were elevated relative to preoperative levels. To further explore the prognostic significance of elevated postoperative CEA and CA19-9 levels, patients were grouped according to preoperative CEA and CA19-9 levels and postoperative elevations. There were 522 patients who received postoperative adjuvant chemotherapy, which mainly comprised the SOX (S1 and oxaliplatin) or XELOX (capecitabine and oxaliplatin) regimen, once every 3 weeks for six cycles.

All patients were followed up via telephone, text message, and outpatient visits. The follow-up endpoint was December 2020. Overall survival (OS) was defined as the time from surgery to death from any cause. Disease-free survival (DFS) was defined as the time from surgery to the first documented recurrence, death from any cause, or last recurrence-free follow-up, whichever occurred first. Recurrence pattern was classified as no recurrence, peritoneal recurrence, and other recurrence patterns included liver, lung, bone, and multiple-site recurrence.

Statistical analysis

All data were analyzed using IBM SPSS Statistics software (version 26.0) and GraphPad Prism 9. Pearson’s χ² test was used to compare different categorical variables. The Kaplan-Meier method was used to evaluate survival, and the log-rank test was used to compare survival curves between different groups. DFS curves were generated for postoperative elevation of CEA and CA19-9, and univariate Cox regression was used to estimate the association between marker elevation and DFS. Univariate and multivariate Cox proportional hazards regression analyses were used to test the effect of each variable on GC prognosis. Univariate and multivariate logistic regression methods were used to explore the potential risk factors for postoperative elevation of CEA and CA19-9. A two-tailed P < 0.05 was considered statistically significant. Because multiple exploratory comparisons were performed, the Benjamini-Hochberg false discovery rate method was used as a sensitivity correction for exploratory subgroup and supplementary analyses. The primary Cox regression analyses were interpreted based on effect estimates, 95% confidence intervals (CIs), and clinical relevance, rather than P values alone.

RESULTS

Clinicopathological characteristics

A total of 562 postoperative patients with GC were included in this study, which comprised 385 (68.5%) men and 177 (31.5%) women. A total of 314 patients (55.9%) were aged ≥ 60 years, and 248 patients (44.1%) were aged < 60 years. There were 61 (10.8%), 169 (30.1%), and 332 (59.1%) patients diagnosed with clinical stage I, stage II, and stage III GC, respectively. In detail, 337 (60%) patients had T4 stage, 117 (20.8%) T3 stage, and 63 (11.2%) T2 stage, and 45 (8%) T1 stage. Moreover, 151 (26.9%) patients had no lymph node metastasis, while 108 (19.2%), 146 (26%), and 157 (27.9%) patients had N1, N2, and N3 stage disease, respectively. There were 303 (53.9%) patients with poorly differentiated GC, 238 (42.4%) with moderately differentiated GC, and only 21 (3.7%) with well-differentiated GC. Vascular invasion was observed in 153 (27.2%) patients. Postoperative CEA elevation was observed in 184 (32.7%) patients, and 235 (41.8%) had postoperative CA19-9 elevation (Table 1). Positive preoperative CEA was observed in 129 patients (23.0%), and positive preoperative CA19-9 was observed in 62 patients (11.0%). Regarding recurrence patterns, 369 patients (65.7%) had no recurrence, whereas 79 patients (14.1%) developed peritoneal recurrence and 114 patients (20.3%) had recurrence at other sites (Table 1).

Table 1 Demographic and clinicopathologic characteristics of patients, n (%).

Variables	Patients (n = 562)
Sex
Male	385 (68.5)
Female	177 (31.5)
Age, years
< 60	248 (44.1)
≥ 60	314 (55.9)
Clinical stage (pTNM)
I	61 (10.8)
II	169 (30.1)
III	332 (59.1)
T stage
T1	45 (8)
T2	63 (11.2)
T3	117 (20.8)
T4	337 (60)
N stage
N0	151 (26.9)
N1	108 (19.2)
N2	146 (26)
N3	157 (27.9)
Differentiation
Good	21 (3.7)
Moderate	238 (42.4)
Poor	303 (53.9)
Vascular invasion
Positive	153 (27.2)
Negative	409 (72.8)
Preoperative CEA
Positive	129 (23.0)
Negative	433 (77.0)
Preoperative CA19-9
Positive	62 (11.0)
Negative	500 (89.0)
Post-preoperative CEA elevation
Yes	184 (32.7)
No	378 (67.3)
Post-preoperative CA19-9 elevation
Yes	235 (41.8)
No	327 (58.2)
Adjuvant chemotherapy
Yes	522 (92.9)
No	40 (7.1)
Recurrence status
Yes	193 (34.3)
No	369 (65.7)
Survival status
Alive	338 (60.1)
Dead	224 (39.9)
Recurrence pattern
No recurrence	369 (65.7)
Peritoneal recurrence	79 (14.1)
Other recurrence sites	114 (20.3)

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9.

Prognostic significance of postoperative elevation of CEA and CA19-9

The Kaplan-Meier method and log-rank test were used to assess the effects of postoperative elevation of tumor markers on survival. Patients with elevated postoperative CEA and CA19-9 levels had a poorer prognosis (P < 0.05; Figure 1). Patients with positive postoperative CEA and postoperative elevation had the worst prognosis while patients with normal postoperative CEA without postoperative elevation had the best prognosis (P < 0.05; Figure 2A). The group analysis for CA19-9 was similar (Figure 2B). The survival curves also showed that, regardless of the postoperative levels of CEA and CA19-9, an elevation in CEA or CA19-9 after surgery indicated a poor prognosis (Figure 2).

Figure 1 Kaplan-Meier curves of overall survival for patients with gastric cancer. A: Kaplan-Meier curves of overall survival based on postoperative carcinoembryonic antigen elevation; B: Kaplan-Meier curves of overall survival based on postoperative carbohydrate antigen 19-9 elevation. CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9.

Figure 2 Kaplan-Meier curves of overall survival for patients with gastric cancer. A: Kaplan-Meier curves of overall survival based on postoperative carcinoembryonic antigen level and postoperative carcinoembryonic antigen elevation; B: Kaplan-Meier curves of overall survival based on postoperative carbohydrate antigen 19-9 level and postoperative carbohydrate antigen 19-9 elevation.

Univariate and multivariate analysis verified that postoperative CEA and CA19-9 elevations were independent prognostic risk factors for GC

Univariate Cox regression analysis showed that age, clinical stage, T stage, N stage, differentiation, vascular invasion, CEA elevation, and CA19-9 elevation were significantly associated with OS. Multivariable analysis showed that the following factors remained independently associated with poorer OS: T3 stage [hazard ratio (HR): 2.10, 95%CI: 1.02-4.33, P = 0.044]; T4 stage (HR: 3.98, 95%CI: 2.06-7.70, P < 0.001); N2 stage (HR: 1.62, 95%CI: 1.12-2.34, P = 0.010); N3 stage (HR: 1.56, 95%CI: 1.08-2.25, P = 0.017); CEA elevation (HR: 2.71, 95%CI: 2.01-3.66, P < 0.001); and CA19-9 elevation (HR: 2.84, 95%CI: 2.15-3.76, P < 0.001). Poor differentiation showed a borderline association with OS (HR: 1.33, 95%CI: 0.98-1.80, P = 0.065), whereas vascular invasion was no longer significant after adjustment (HR: 1.18, 95%CI: 0.88-1.58, P = 0.259) (Table 2). Although vascular invasion was significantly associated with OS in univariate analysis, its effect was attenuated after adjustment for T stage, N stage, and tumor marker elevation. In contrast, N3 stage remained independently associated with poorer OS.

Table 2 Univariate and multivariate Cox regression analyses for overall survival of patients with gastric cancer.

Variables	Univariate analysis		Multivariable analysis
	HR (95%CI)	P value	HR (95%CI)	P value
Sex
Male	Reference		Reference
Female	1.17 (0.89-1.54)	0.265	1.03 (0.77-1.38)	0.822
Age
< 60	Reference		Reference
≥ 60	1.34 (1.02-1.75)	0.034	1.24 (0.94-1.64)	0.124
Clinical stage
I, II	Reference
III	4.00 (2.85-5.59)	< 0.001
T stage
T1, T2	Reference		Reference
T3	2.53 (1.36-4.71)	0.003	2.10 (1.02-4.33)	0.044
T4	5.44 (3.16-9.39)	< 0.001	3.98 (2.06-7.70)	< 0.001
N stage
N0, N1	Reference		Reference
N2	2.49 (1.75-3.54)	< 0.001	1.62 (1.12-2.34)	0.010
N3	4.33 (3.12-6.01)	< 0.001	1.56 (1.08-2.25)	0.017
Differentiation
Moderate, good	Reference		Reference
Poor	1.96 (1.49-2.59)	< 0.001	1.33 (0.98-1.80)	0.065
Vascular invasion
Negative	Reference		Reference
Positive	2.11 (1.61-2.77)	< 0.001	1.18 (0.88-1.58)	0.259
CEA elevation
No	Reference		Reference
Yes	4.21 (3.23-5.49)	< 0.001	2.71 (2.01-3.66)	< 0.001
CA19-9 elevation
No	Reference		Reference
Yes	3.28 (2.50-4.31)	< 0.001	2.84 (2.15-3.76)	< 0.001

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; HR: Hazard ratio; CI: Confidence interval.

DFS

In univariate Cox analysis, advanced clinical stage, higher T stage, higher N stage, poor differentiation, vascular invasion, CEA elevation and CA19-9 elevation were significantly associated with worse DFS. In the multivariable model, T stage, N stage, vascular invasion, CEA elevation, and CA19-9 elevation remained independently associated with DFS (Table 3).

Table 3 Univariate and multivariable Cox regression analyses for disease-free survival of patients with gastric cancer.

Variables	Univariate analysis		Multivariable analysis
	HR (95%CI)	P value	HR (95%CI)	P value
Sex
Male	Reference		Reference
Female	1.12 (0.86-1.44)	0.406	1.03 (0.78-1.35)	0.842
Age, years
< 60	Reference		Reference
≥ 60	1.28 (1.00-1.64)	0.053	1.18 (0.92-1.53)	0.190
Clinical stage
I, II	Reference
III	3.48 (2.59-4.67)	< 0.001
T stage
T1, T2	Reference		Reference
T3	2.78 (1.62-4.79)	< 0.001	2.30 (1.22-4.31)	0.010
T4	4.94 (3.05-8.01)	< 0.001	3.59 (2.00-6.43)	< 0.001
N stage
N0, N1	Reference		Reference
N2	2.41 (1.76-3.32)	< 0.001	1.74 (1.25-2.42)	< 0.001
N3	3.86 (2.87-5.21)	< 0.001	1.59 (1.14-2.22)	0.006
Differentiation
Moderate, good	Reference		Reference
Poor	1.74 (1.36-2.24)	< 0.001	1.18 (0.90-1.56)	0.222
Vascular invasion
Negative	Reference		Reference
Positive	2.20 (1.71-2.83)	< 0.001	1.39 (1.07-1.82)	0.015
CEA elevation
No	Reference		Reference
Yes	3.58 (2.80-4.57)	< 0.001	2.37 (1.80-3.13)	< 0.001
CA19-9 elevation
No	Reference		Reference
Yes	2.71 (2.11-3.46)	< 0.001	2.33 (1.81-3.00)	< 0.001

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; HR: Hazard ratio; CI: Confidence interval.

Compared with T1/T2 disease, T3 and T4 tumors were associated with increased risk of DFS (T3: HR = 2.30, 95%CI: 1.22-4.31, P = 0.010; T4: HR = 3.59, 95%CI: 2.00-6.43, P < 0.001). N2 and N3 disease also remained significant predictors compared with N0/N1 disease (N2: HR = 1.74, 95%CI: 1.25-2.42, P < 0.001; N3: HR = 1.59, 95%CI: 1.14-2.22, P = 0.006). Vascular invasion was independently associated with poorer DFS (HR = 1.39, 95%CI: 1.07-1.82, P = 0.015). CEA and CA19-9 elevation remained strong independent predictors of worse DFS (CEA elevation: HR = 2.37, 95%CI: 1.80-3.13, P < 0.001; CA19-9 elevation: HR = 2.33, 95%CI: 1.81-3.0, P < 0.001; Table 3).

We further evaluated DFS to determine whether postoperative CEA/CA19-9 dynamics were associated with recurrence-related outcomes. Patients with postoperative CEA elevation relative to preoperative levels had significantly worse DFS than those without CEA elevation, with 3-year DFS rates of 29.6% and 74.6%, respectively (log-rank P < 0.001; univariate HR 3.58, 95%CI: 2.80-4.57; P < 0.001; Figure 3A). Similarly, patients with postoperative CA19-9 elevation had poorer DFS than those without CA19-9 elevation, with 3-year DFS rates of 42.3% and 72.5%, respectively (log-rank P < 0.001; univariate HR 2.71, 95%CI: 2.11-3.46; P < 0.001; Figure 3B).

Figure 3 Kaplan-Meier curves of disease-free survival according to postoperative carcinoembryonic antigen elevation and postoperative carbohydrate antigen 19-9 elevation. A: Disease-free survival according to postoperative carcinoembryonic antigen elevation relative to preoperative carcinoembryonic antigen levels; B: Disease-free survival according to postoperative carbohydrate antigen 19-9 elevation relative to preoperative carbohydrate antigen 19-9 levels.

Risk factors associated with postoperative elevation of CEA and CA19-9

To verify the correlation between CEA and CA19-9 elevation and clinicopathological characteristics, χ² tests were performed. Subsequently, univariate and multivariate logistic regression analyses were performed to screen for risk factors for postoperative CEA and CA19-9 elevation. χ² tests showed that postoperative elevation of CEA was related to clinical stage, T stage, N stage, degree of tumor differentiation, vascular invasion, postoperative CEA and CA19-9 levels, PNI, recurrence pattern, and PLR (all P < 0.05) (Table 4). Additionally, postoperative elevation of CA19-9 differed in terms of clinical stage, N stage, degree of tumor differentiation, vascular invasion, recurrence pattern, postoperative CEA and CA19-9 levels, and LMR (all P < 0.05; Table 4). According to the logistic regression methods, the following were independent risk factors for elevated postoperative CEA: N stage [N2, OR: 2.726 (1.590-4.675), P < 0.001; N3, OR: 6.603 (3.815-11.426), P < 0.001]; degree of tumor differentiation [OR: 2.015 (1.284-3.163), P = 0.002]; vascular invasion [OR: 3.953 (2.506-6.235), P < 0.001]; postoperative CEA level [OR: 2.296 (1.152-4.575), P = 0.018]; and PNI [< 49.98, OR: 1.824 (1.149-2.896), P = 0.011] (Table 5). N stage [N3, OR: 2.344 (1.490-3.686), P < 0.001], postoperative CEA level [OR: 2.559 (1.380-4.746), P = 0.003], and LMR [< 4.32, OR: 1.442 (1.009-2.062), P = 0.044] were independent risk factors for postoperative CA19-9 elevation (Table 6).

Table 4 The relationship between clinicopathological characteristics and postoperative carcinoembryonic antigen and carbohydrate antigen 19-9 elevation.

Variables	Post-preoperative CEA elevation			Post-preoperative CA19-9 elevation
	Yes (n = 184)	No (n = 378)	P value	Yes (n = 235)	No (n = 327)	P value
Sex			0.172			0.856
Male	119	266		160	225
Female	65	112		75	102
Age, years			0.612			0.570
< 60	84	164		107	141
≥ 60	100	214		128	186
Clinical stage			< 0.001			0.002
I, II	38	192		78	152
III	146	186		157	175
T stage			< 0.001			0.605
T1, T2	23	85		41	67
T3	29	88		48	69
T4	132	205		146	191
N stage			< 0.001			< 0.001
N0, N1	37	222		80	179
N2	47	99		62	84
N3	100	57		93	64
Differentiation			< 0.001			0.014
Moderate, good	53	206		94	165
Poor	131	172		141	162
Vascular invasion			< 0.001			0.001
Positive	92	61		81	72
Negative	92	317		154	255
Adjuvant chemotherapy			0.104			0.247
Yes	175	345		221	299
No	9	33		14	28
Recurrence pattern			< 0.001			< 0.001
No recurrence	83	286		118	251
Peritoneal recurrence	38	41		49	30
Other recurrence sites	63	51		68	46
Post-CEA			< 0.001			< 0.001
Positive	37	22		41	18
Negative	147	356		194	309
Post-CA19-9			< 0.001			0.006
Positive	24	14		24	14
Negative	160	364		211	313
PNI			< 0.001			0.072
< 49.98	128	189		143	174
≥ 49.98	56	189		92	153
PLR			< 0.001			0.436
< 138.41	76	229		123	182
≥ 138.41	108	149		112	145
LMR			0.650			0.014
< 4.32	86	169		121	134
≥ 4.32	98	209		114	193

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; LMR: Lymphocyte to monocyte ratio; PLR: Platelet to lymphocyte ratio; PNI: Prognostic nutritional index.

Table 5 Univariate and multivariate logistic regression analyses for postoperative carcinoembryonic antigen elevation.

Variables	Univariate analysis		Multivariable analysis
	OR (95%CI)	P value	OR (95%CI)	P value
T stage
T1, T2, T3	Reference		Reference
T4	2.142 (1.466-3.131)	< 0.001	0.942 (0.590-1.505)	0.802
N stage
N0, N1	Reference		Reference
N2	2.848 (1.742-4.657)	< 0.001	2.726 (1.590-4.675)	< 0.001
N3	10.526 (6.537-16.950)	< 0.001	6.603 (3.815-11.426)	< 0.001
Differentiation
Moderate, good	Reference		Reference
Poor	2.960 (2.029-4.320)	< 0.001	2.015 (1.284-3.163)	0.002
Vascular invasion
Negative	Reference		Reference
Positive	5.197 (3.490-7.739)	< 0.001	3.953 (2.506-6.235)	< 0.001
Post-CEA
Negative	Reference		Reference
Positive	4.073 (2.323-7.142)	< 0.001	2.296 (1.152-4.575)	0.018
Post-CA19-9
Negative	Reference		Reference
Positive	3.900 (1.966-7.736)	< 0.001	1.349 (0.584-3.114)	0.483
PNI
≥ 49.98	Reference		Reference
< 49.98	2.286 (1.574-3.320)	< 0.001	1.824 (1.149-2.896)	0.011
PLR
< 138.41	Reference		Reference
≥ 138.41	2.184 (1.526-3.126)	< 0.001	1.418 (0.904-2.223)	0.128

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; PLR: Platelet to lymphocyte ratio; PNI: Prognostic nutritional index; OR: Odds ratio; CI: Confidence interval.

Table 6 Univariate and multivariate logistic regression analyses for postoperative carbohydrate antigen 19-9 elevation.

Variables	Univariate analysis		Multivariable analysis
	OR (95%CI)	P value	OR (95%CI)	P value
N stage
N0, N1	Reference		Reference
N2	1.651 (1.084-2.516)	0.019	1.489 (0.968-2.291)	0.07
N3	3.251 (2.151-4.915)	< 0.001	2.344 (1.490-3.686)	< 0.001
Differentiation
Moderate, good	Reference		Reference
Poor	1.528 (1.088-2.145)	0.014	1.245 (0.857-1.808)	0.250
Vascular invasion
Negative	Reference		Reference
Positive	1.863 (1.280-2.711)	0.001	1.414 (0.945-2.116)	0.092
Post-CEA
Negative	Reference		Reference
Positive	3.628 (2.026-6.496)	< 0.001	2.559 (1.380-4.746)	0.003
Post-CA19-9
Negative	Reference		Reference
Positive	2.543 (1.286-5.029)	0.007	1.301 (0.613-2.760)	0.493
LMR
≥ 4.32	Reference		Reference
< 4.32	1.529 (1.091-2.143)	0.014	1.442 (1.009-2.062)	0.044

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; OR: Odds ratio; CI: Confidence interval; LMR: Lymphocyte to monocyte ratio.

Impact of postoperative CEA and CA19-9 elevations on GC prognosis across different preoperative biomarker levels

To explore the impact of postoperative elevation of CEA and CA19-9 on GC prognosis, we conducted subgroup analyses. The subgroups were divided based on preoperative and postoperative tumor marker levels. Survival curves were further stratified according to preoperative marker status and postoperative elevation relative to preoperative levels. For CEA, patients with negative preoperative CEA and no postoperative CEA elevation had the most favorable OS, whereas patients with positive preoperative CEA and postoperative CEA elevation had the poorest OS (Figure 4A). Patients with positive preoperative CEA but no postoperative elevation and those with negative preoperative CEA but postoperative elevation showed intermediate outcomes. Similar findings were observed for CA19-9 (Figure 4B). The results of subgroup analysis of CA19-9 levels were similar (Figure 4B). Exploratory Kaplan-Meier analyses further suggested a graded association between the magnitude of postoperative marker increase and OS. Patients with greater increases in CEA or CA19-9 had worse OS than those with mild or no increase (both log-rank P < 0.001; Figure 5). These delta thresholds were based on median positive delta values and should be interpreted as exploratory rather than validated clinical cutoffs. We further evaluated whether the combined elevation pattern of CEA and CA19-9 could improve risk stratification. Patients were divided into four groups: Neither marker elevated, CEA only elevated, CA19-9 only elevated, and both markers elevated. The four groups showed significantly different OS and DFS (both log-rank P < 0.001; Figure 6), with the both-elevated group showing the poorest prognosis. After Benjamini-Hochberg false discovery rate correction, the exploratory subgroup comparisons for Figures 5 and 6 remained statistically significant. These results demonstrated that the presence of postoperative tumor marker elevation was more informative for the prognosis of GC than preoperative tumor marker levels.

Figure 4 Kaplan-Meier curves of overall survival for patients with gastric cancer. A: Kaplan-Meier curves of overall survival based on preoperative carcinoembryonic antigen level and postoperative carcinoembryonic antigen elevation; B: Kaplan-Meier curves of overall survival based on preoperative carbohydrate antigen 19-9 level and postoperative carbohydrate antigen 19-9 elevation.

Figure 5 Exploratory Kaplan-Meier curves of overall survival according to the magnitude of postoperative tumor marker changes. A: Overall survival according to size of change in carcinoembryonic antigen; B: Overall survival according to size of change in carbohydrate antigen 19-9. For carcinoembryonic antigen, mild and greater increases were split at the median positive change of 0.65 ng/mL. For carbohydrate antigen 19-9, mild and greater increases were split at the median positive change of 3.12 U/mL. These change thresholds were data-derived for exploratory analysis and should not be interpreted as validated clinical cutoffs.

Figure 6 Combined carcinoembryonic antigen and carbohydrate antigen 19-9 elevation stratifies overall survival and disease-free survival. Kaplan-Meier curves showing survival according to combined post-preoperative carcinoembryonic antigen and carbohydrate antigen 19-9 elevation. Patients were classified into four groups: Neither marker elevated, carcinoembryonic antigen only elevated, carbohydrate antigen 19-9 only elevated, and both markers elevated. A: Overall survival; B: Disease-free survival.

DISCUSSION

Surgical resection is recommended as the first-line treatment for GC. However, postoperative recurrence of GC remains a major clinical challenge[5]. Clinically, common risk factors for GC recurrence include tumor size and TNM stage. Recent studies have extensively investigated prognostic serum markers, which are crucial to predict high-risk GC patients, and to choose better treatments after recurrence to improve OS. CEA, CA72-4, CA19-9, and CA125 are representative tumor markers for GC and play important roles in detecting recurrence and predicting the prognosis of patients with GC[14]. CEA and CA19-9 are the best established tumor markers that need to be measured after surgery[15], including in our hospital. Many studies have evaluated preoperative, but not postoperative, serum levels of tumor markers including CEA and CA19-9, but few have explored the relationship between the postoperative tumor marker positivity and prognosis.

Our results suggested that postoperative elevation of CEA was related to T stage, N stage, degree of tumor differentiation, vascular invasion, postoperative CEA level, postoperative CA19-9 level, PNI, and PLR (all P < 0.05). In addition, postoperative CA19-9 elevation was differentiated according to N stage, degree of tumor differentiation, vascular invasion, postoperative CEA and CA19-9 levels, and LMR. After screening though multivariate logistic regression analyses, N stage, degree of tumor differentiation, vascular invasion, postoperative CEA level, and PNI were identified as independent risk factors for postoperative CEA elevation, whereas N stage, postoperative CEA level, and LMR were identified as independent risk factors for postoperative elevation of CA19-9. Multivariable Cox analysis demonstrated that postoperative elevations of CEA and CA19-9 were independent prognostic risk factor for GC. Patients with elevated postoperative CEA and CA19-9 had a poorer prognosis, while patients with normal postoperative CEA/CA19-9 and without postoperative elevation had the best prognosis, and positive postoperative CEA or CA-199 had a negative impact on the prognosis of GC. These results confirmed the significant role of CEA and CA19-9 in predicting the prognosis of patients with GC.

Similar to our study, Yu et al[16] concluded that CEA, CA19-9, and CA72-4 were superior to other markers in GC. Another study also reached the same conclusion[8], that preoperative serum CA19-9 level was better than CEA level as a prognostic factor, and several analyses identified preoperative positivity for CEA (but not CA19-9) as a significant prognostic factor for GC[17,18]. Our results demonstrated that elevation of postoperative tumor markers was more informative for prognosis of GC than preoperative tumor markers. Postoperative elevations of CEA and CA19-9 were independent prognostic risk factors for GC, which indicated a significant increase in postoperative CEA and CA19-9 levels compared with preoperative levels. Therefore, in addition to preoperative levels, postoperative levels of CEA and CA19-9 should be measured during regular examinations. Our study also identified high-risk factors for postoperative elevation of CEA and CA19-9 levels. For high-risk individuals, the frequency of postoperative tumor marker detection must be strictly controlled. The DFS analysis supports the relevance of postoperative CEA/CA19-9 dynamics to recurrence-related outcomes, because postoperative elevation of CEA and CA19-9 was associated with markedly poorer DFS. These findings suggest that postoperative marker elevation may help identify patients who require closer recurrence surveillance after curative gastrectomy. In the future, we will analyze more tumor markers and compare their prognostic values in patients with GC in more centers.

In our recurrence-pattern analysis, postoperative CA19-9 elevation was associated with a higher proportion of peritoneal recurrence. In the present study, peritoneal recurrence accounted for 41% of recurrences among patients with postoperative CA19-9 elevation, which is consistent with the findings of Wu et al[19], but our results are more convincing in that they were based on two centers. This suggests that postoperative CA19-9 elevation is associated with an increased risk of peritoneal recurrence. Biologically, CA19-9 may be linked to peritoneal recurrence through aberrant tumor glycosylation[20]. As a sialyl Lewis(a) carbohydrate epitope carried by mucin-type glycoproteins, CA19-9 may reflect a highly sialylated, adhesive, and invasive tumor phenotype[21]. Because peritoneal dissemination depends on tumor-cell detachment, survival in the peritoneal cavity, and adhesion to mesothelial cells, CA19-9 elevation may indicate occult residual disease or a glycosylation phenotype that favors peritoneal implantation. However, this potential mechanism requires validation in prospective studies with tissue and imaging-based analyses.

Our study had several limitations. First, we only analyzed 3-year OS and DFS in patients with GC. Future studies involving more centers and longer follow-up periods are needed to validate our findings. Second, the present study only evaluated the combined prognostic significance of CEA and CA19-9, and we should include more tumor markers to study their value in the diagnosis and treatment of GC. Third, because detailed imaging data were unavailable in this retrospective cohort, the association between CA19-9 elevation and imaging-defined peritoneal metastasis could not be directly assessed. Future studies with standardized imaging review are warranted to clarify this relationship.

CONCLUSION

This study showed that postoperative positivity and elevation of CEA and CA19-9 relative to preoperative levels were independently associated with poorer prognosis in patients with GC after curative gastrectomy. Postoperative CEA and CA19-9 levels may play significant roles in predicting the prognosis of patients with GC.

ACKNOWLEDGEMENTS

Thanks to all the participants included in this study.