Xinyue Qiu1, Cheng Shen2, Wenjing Zhao1, Xunlei Zhang3, Dakun Zhao1, Yueyue Zhu4, Guoxing Li5, Lei Yang3. 1. Cancer Research Center, Affiliated Tumor Hospital of Nantong University, Nantong Jiangsu, China. 2. Department of Computer Science and Engineering, Tandon School of Engineering, New York University, Brooklyn, NY 11201, US. 3. Department of Oncology, Affiliated Tumor Hospital of Nantong University, Nantong Jiangsu, China. 4. Department of Oncology, Affiliated Hospital of Nantong University, Nantong Jiangsu, China. 5. Department of Surgery, Affiliated Tumor Hospital of Nantong University, Nantong Jiangsu, China.
Worldwide, gastric cancer ranks fifth among all cancers and is the third most common cause of cancer mortality 1, but ranks second among all cancers and is the third most common cause of cancer mortality in the People's Republic of China 2. As early-stage gastric cancer is often asymptomatic and difficult to detect, most patients have advanced-stage gastric cancer at the initial diagnosis. Radical gastrectomy is the mainstay of treatment for early-stage gastric cancer. However, different risk groups and additional potential prognostic biomarkers need to be identified for patients with gastric cancer who have different physical conditions and disease states. Therefore, the identification of accurate prognostic factors and the development of validated scoring systems to predict survival outcomes in patients with gastric cancer are essential for developing individualized treatment plans for patients with gastric cancer.Increasingly, several studies have shown that hematological biomarkers, determined via rapid and easy testing, can predict the prognosis of patients with gastric cancer. The commonest hematological biomarkers for gastric cancer include hemoglobin (HB), neutrophil-to-lymphocyte ratio (NLR) 3, platelet-to-lymphocyte ratio (PLR) 3, and tumor markers 4. Anemia is a common symptom in patients with cancer, of which the most intuitive reflection is a decrease in the hemoglobin level that often has a multifactorial causation 5. Cancer-related anemia occurs frequently in malignancies and is one of the commonest comorbidities. The prevalence of pretreatment anemia is high and ranges from 30% to 90% in various cancers 6. The causes of cancer-related anemia include tumor-related blood loss, bone marrow involvement, cytokine-mediated disorders, and nutritional deficiencies of iron or folic acid 7. The widespread pretreatment anemia adversely affects the treatment and prognosis of patients with cancer 8, 9. In the past decade, evidence has accumulated to indicate that anemia predicts poor patient survival in many cancers, including gastric cancer 10,11.Tumor markers can be used as indicators for determining the diagnosis, treatment, and prognosis of patients with gastric cancer. Serum levels of the tumor markers (e.g. carcinoembryonic antigen [CEA], carbohydrate antigen CA199 and CA 72-4, and alpha fetoprotein [AFP]) are elevated in some patients with gastric cancer 12. CEA is a common serum marker for malignant gastrointestinal tract tumors and facilitates a diagnosis of gastric cancer. Moreover, the expression of CEA is an independent risk factor for poor prognosis in gastric cancer 13. Elevated preoperative CEA has been proposed as a predictor of overall survival (OS) in early-stage gastric cancer 13. The combination of CEA with CA19-9 and CA72-4 has been reported to predict survival in gastric cancer and prove important for subsequent treatment-related decisions 14. Accordingly, we proposed the hypothesis that a combination of tumor markers and individual patient differences can be analyzed together to predict the survival of patients with gastric cancer.Therefore, the aim of this study was to assess the clinical value of the novel combination prognostic biomarker (HB-CEA) for gastric cancer. In addition, we assessed the relationship between HB-CEA and the clinicopathological features of gastric cancer.
Materials and Methods
This study involved a retrospective evaluation of patients with primary gastric cancer who underwent radical gastrectomy between 2007 and 2016 (training set) and between 2010 and 2012 (validation set) at the Affiliated Tumor Hospital of Nantong University and the Affiliated Hospital of Nantong University. The inclusion criteria: (1) availability of complete patient clinical data (including sex, age, tumor location, differentiation, diameter of lesion, cancer embolism, nerve invasion, pathological TNM stage (pTNM), depth of invasion, lymph node metastasis, and survival status); (2) without coexistence of other tumors and hematological diseases; (3) received radical gastrectomy. The exclusion criteria: (1) received neoadjuvant chemotherapy or radiotherapy before surgery; (2) with metastases to lung, liver and other organs in preoperative imaging examinations, and distant metastasis in intraoperative examinations; (3) hepatic or renal insufficiency; (4) acceptance of palliative surgery. Thus, 741 patients were enrolled in the present study.This retrospective study was approved by the Medical Institutional Ethics Committee of the Affiliated Tumor Hospital of Nantong University and the Affiliated Hospital of Nantong University, Jiangsu province. The data were anonymous and, therefore, the requirement for informed consent was waived.
Data Collection
Details of the clinicopathological information of the participants, including sex, age, tumor location, differentiation, diameter of lesion, cancer embolism, nerve invasion, pathological TNM stage (pTNM), depth of invasion, lymph node metastasis, and survival status, were obtained from the medical records. OS was defined as the interval from the date of the surgery to the date of death or the last follow-up. Pretreatment fasting peripheral blood samples were collected within 1 week preceding the surgery and were processed within 48 hours after collection to assess HB-CEA levels, liver and kidney function, coagulation function, and other predefined indicators. Clinical data of the patients with gastric cancer were collected from the hospital information system. Staging was performed according to the TNM classification of the American Joint Committee on Cancer (8th Edition).
Statistical Analysis
The optimal cutoff values for HB, CEA, and other variables were obtained via the receiver operating characteristic (ROC) curve analysis, and the participants were then divided into two study groups according to the cutoff. Additionally, the participants were divided into three groups according to the HB-CEA score: HB-CEA=0 (CEA <3.395 ng/mL and HB ≥125.5 g/L), HB-CEA =1 (CEA ≥3.395 ng/mL or HB <125.5 g/L), and HB-CEA=2 (CEA ≥3.395 ng/mL and HB <125.5 g/L) (Table 1). The chi-square or Fisher's exact test was used to analyze categorical variables. In the training set, propensity score matching (PSM) was performed to reduce selection bias and to quantify the possible association between HB-CEA and OS. The Kaplan-Meier method was used for survival analysis. Variables found to be significant in univariate analysis were included in multivariate analysis. Multivariate Cox model was used to analyze the factors that independently affect the survival and prognosis of gastric cancer patients. All statistical analyses were performed using SPSS 26.0 and R Package (http://www.r-project.org/). P<0.05 was considered indicative of statistical significance.
The training set comprised 353 patients with gastric cancer, including 117 (33.1%) women and 236 (66.9%) men; 179 (50.7%) and 174 (49.3%) participants were older than 65 years and aged 65 years or younger, respectively. The validation set comprised 388 participants with gastric cancer, including 122 (31.4%) women and 266 (68.6%) men; 165 (42.5%) and 223 (57.5%) participants were older than 65 years and aged 65 years or younger, respectively. The clinicopathological characteristics of the participants in the study cohorts are summarized in Table 2.
Table 2
The Clinicopathological Features of GC Patients
Characteristics
The training set
The validation set
Patients, n (%)
Patients, n (%)
Hemoglobin (g/L)
CEA (ng/mL)
Hemoglobin (g/L)
CEA (ng/mL)
≥125.5
<125.5
P value
<3.395
≥3.395
P value
≥125.5
<125.5
P value
<3.395
≥3.395
P value
Sex
<0.001*
0.010*
<0.001*
0.004*
Female
117(33.1%)
36
81
88
29
122(31.4%)
34
88
103
19
Male
236(66.9%)
137
99
145
91
266(68.6%)
154
112
188
78
Age (years)
0.428
0.108
0.002*
<0.001*
≤65
174(49.3%)
89
85
122
52
223(57.5%)
123
100
186
37
>65
179 (50.7%)
84
95
111
68
165(42.5%)
65
100
105
60
Tumor location
0.223
0.829
0.713
0.705
Upper third
80(22.7%)
44
36
52
28
71(18.3%)
33
38
52
19
Middle and lower third
273(77.3%)
129
144
181
92
317(81.7%)
155
162
239
78
Differentiation
0.041*
0.082
<0.001*
0.410
Moderate/Poor
330(93.5%)
157
173
214
116
343(88.4%)
152
191
255
88
Well
23(6.5%)
16
7
19
4
45(11.6%)
36
9
36
9
Diameter of lesion (cm)
0.003*
0.118
<0.001*
0.025*
<3
82(23.2%)
52
30
60
22
115(29.6%)
72
43
95
20
≥3
271(76.8%)
121
150
173
98
273(70.4%)
116
157
196
77
Cancer embolism
0.002*
0.583
0.484
0.401
None
242(68.6%)
132
110
162
80
345(88.9%)
165
180
261
84
Yes
111(31.4%)
41
70
71
40
43(11.1%)
23
20
30
13
Nerve invasion
0.697
0.784
0.494
0.275
None
227(64.3%)
113
114
151
76
333(85.8%)
159
174
253
80
Yes
126(35.7%)
60
66
82
44
55(14.2%)
29
26
38
17
pTNM
0.001*
0.006*
0.015*
0.003*
I-II
162(45.9%)
95
67
119
43
219(56.4%)
118
101
177
42
III-IV
191(54.1%)
78
113
114
77
169(43.6%)
70
99
114
55
Depth of invasion
<0.001*
0.002*
<0.001*
0.006*
T1-2
120(34.0%)
80
40
92
28
162(41.8%)
98
64
133
29
T3-4
233(66.0%)
93
140
141
92
226(58.2%)
90
136
158
68
LN metastasis
0.004*
0.011*
0.001*
0.002*
N0
123(34.8%)
73
50
92
31
168(43.3%)
97
71
139
29
N1/N2/N3
230(65.2%)
100
130
141
89
220(56.7%)
91
129
152
68
Survival status
<0.001*
<0.001*
<0.001*
<0.001*
Survival
221(62.6%)
130
91
163
58
241(62.1%)
136
105
200
41
Death
132(37.4%)
43
89
70
62
147(37.9%)
52
95
91
56
Note: HB-CEA: HB≥125.5 and CEA<3.395 represent 0; CEA≥3.395 or HB<125.5 represent 1; CEA≥3.395 and HB<125.5 represent 2.
In the training set, the optimal cutoff value of CEA for OS was 3.395 ng/mL (area under curve [AUC]=0.629, 95% confidence interval [CI]: 0.569-0.690, p<0.001; Figure 1A). The optimal cutoff value of hemoglobin for OS was 125.5 g/L (AUC=0.650, 95% CI: 0.592-0.709, p<0.001; Figure 1A). Based on these cutoff values, patients were divided into two groups: high CEA (≥3.395 ng/ mL, n=120) and low CEA (<3.395 ng/mL, n=233); high HB (≥125.5 g/L, n=173) and low HB (<125.5 g/L, n=180). The AUC of HB-CEA for OS was 0.677 (95% CI: 0.619-0.734, p<0.001; Figure 1A), which is slightly higher than that of hemoglobin (AUC=0.650, 95% CI: 0.592-0.709, p<0.001; Figure 1A) and CEA (AUC=0.629, 95% CI: 0.569-0.690, p<0.001; Figure 1A). These results demonstrate that HB-CEA may be superior to HB or CEA alone as a prognostic marker of gastric cancer. Similar results were obtained in the validation set. The AUC of HB-CEA for OS was 0.670 (95% CI: 0.615-0.725, p<0.001; Figure 1B), which is slightly higher than that of hemoglobin (AUC=0.605, 95% CI: 0.548-0.663, p=0.001; Figure 1B) and CEA (AUC=0.605, 95% CI: 0.546-0.665, p<0.001; Figure 1B).
Figure 1
Receiver operating characteristics curve analysis of HB, CEA and HB-CEA for OS in GC patients (the training set, A; the validation set, B). Note: HB-CEA: HB≥125.5 and CEA<3.395 represent 0; CEA≥3.395 or HB<125.5 represent 1; CEA≥3.395 and HB<125.5 represent 2. Abbreviations: GC: gastric cancer; CEA: carcinoembryonic antigen; HB: hemoglobin; OS: overall survival.
Relationship of Preoperative HB and CEA with Clinicopathological Features
The preoperative HB and CEA were associated with various clinicopathological features of patients with gastric cancer (Table 2). In the training set, high CEA significantly correlated with sex (p=0.010), pathological TNM stage (pTNM) (p=0.006), depth of invasion (p=0.002), lymph node metastasis (p=0.011), and survival status (p<0.001) but not with age, tumor location, diameter of lesion, cancer embolism, nerve invasion, and differentiation (all p>0.05). Similarly, preoperative hemoglobin levels significantly correlated with sex (p<0.001), differentiation (p=0.041), diameter of lesion (p=0.003), cancer embolism (p=0.002), pathological TNM stage (pTNM) (p=0.001), depth of invasion (p<0.001), lymph node metastasis (p=0.004), and survival status (p<0.001), but not with age, tumor location, and nerve invasion (all p>0.05). However, in the validation set, high CEA significantly correlated with sex (p=0.004), age (p<0.001), diameter of lesion (p=0.025), pathological TNM stage (pTNM) (p=0.003), depth of invasion (p=0.006), lymph node metastasis (p=0.002), and survival status (p<0.001) but not with the tumor location, differentiation, cancer embolism, and nerve invasion (all p>0.05). Similarly, preoperative hemoglobin levels significantly correlated with sex (p<0.001), age (p=0.002), differentiation (p<0.001), diameter of lesion (p<0.001), pathological TNM stage (pTNM) (p=0.015), depth of invasion (p<0.001), lymph node metastasis (p=0.001), and survival status (p<0.001), but not with tumor location, cancer embolism, and nerve invasion (all p>0.05).
Prognostic Significance of Preoperative HB and CEA Levels
The prognostic value of preoperative CEA and hemoglobin levels was demonstrated by Kaplan-Meier analysis and Log-rank test, and the overall survival rate and median overall survival time in the low CEA group were obviously higher than those in the high CEA group (p<0.001). Furthermore, the OS rate and median OS time in the high hemoglobin group were also obviously higher than those in the low hemoglobin group (p<0.001; Figure 2A). The correlation of CEA (p<0.001) and hemoglobin (p<0.001; Figure 2B) with the OS were similarly indicated in the validation set.
Figure 2
Kaplan-Meier survival curves for OS according to CEA and HB in GC patients (the training set, A; the validation set, B). Abbreviations: OS: overall survival; GC: gastric cancer; CEA: carcinoembryonic antigen; HB: hemoglobin.
Relationship of the HB-CEA Score with Clinicopathological Characteristics
We found significant intergroup differences among the three groups with regard to age (p=0.026), differentiation (p=0.014), diameter of lesion (p=0.005), cancer embolism (p=0.037), depth of invasion (p<0.001), lymph node metastasis (p=0.001), pathological TNM stage (pTNM) (p<0.001), and survival status (p<0.001) upon analyzing the relationship between HB-CEA and clinicopathological features in the training set (Table 3). Analysis of heterogeneity after PSM between the sub-cohorts stratified by the HB-CEA score cutoffs showed almost balanced distribution of the patient characteristics, including sex, age, tumor location, differentiation, diameter of lesion, cancer embolism, nerve invasion, pathological TNM stage (pTNM), depth of invasion, and lymph node metastasis, among the 64, 128, and 64 patients (HB-CEA=0, 1, and 2, respectively; p>0.05). Thus, the PSM-adjusted analysis could facilitate the investigation of whether the HB-CEA is associated with increased OS in a total of 256 patients. However, in the validation set, the HB-CEA score was significantly correlated with sex (p=0.037), age (p<0.001), differentiation (p<0.001), diameter of lesion (p<0.001), pathological TNM stage (pTNM) (p<0.001), depth of invasion (p<0.001), lymph node metastasis (p<0.001), and survival status (p<0.001) (Table 3).
Table 3
The Clinicopathological Characteristics Stratified by the HB-CEA Score
Characteristics
The training set
The validation set
Before PSM
After PSM
HB-CEA
HB-CEA
HB-CEA
P value
HB-CEA
HB-CEA
HB-CEA
P value
HB-CEA
HB-CEA
HB-CEA
P value
0
1
2
0
1
2
0
1
2
Sex
0.231
0.388
0.037*
Female
33
58
26
20
40
26
31
75
16
Male
84
114
38
44
88
38
103
136
27
Age (years)
0.026*
0.165
<0.001*
≤65
59
93
22
30
62
22
98
113
12
>65
58
79
42
34
66
42
36
98
31
Tumor location
0.401
0.832
0.669
Upper third
31
34
15
18
33
15
24
37
10
Middle and lower third
86
138
49
46
95
49
110
174
33
Differentiation
0.014*
0.711
<0.001*
Moderate/Poor
103
165
62
62
126
62
107
193
43
Well
14
7
2
2
2
2
27
18
0
Diameter of lesion (cm)
0.005*
0.830
<0.001*
<3
36
40
6
8
13
6
57
53
5
≥3
81
132
58
56
115
58
77
158
38
Cancer embolism
0.037*
0.336
0.989
None
89
116
37
45
81
37
119
188
38
Yes
28
56
27
19
47
27
15
23
5
Nerve invasion
0.898
0.663
0.947
None
77
110
40
35
76
40
116
180
37
Yes
40
62
24
29
52
24
18
31
6
pTNM
<0.001*
0.380
<0.001*
I-II
71
72
19
25
38
19
94
107
18
III-IV
46
100
45
39
90
45
40
104
25
Depth of invasion
<0.001*
0.449
<0.001*
T1-2
60
52
8
13
19
8
82
67
13
T3-4
57
120
56
51
109
56
52
144
30
LN metastasis
0.001*
0.556
<0.001*
N0
55
55
13
18
29
13
79
78
11
N1/N2/N3
62
117
51
46
99
51
55
133
32
Survival status
<0.001*
<0.001*
<0.001*
Survival
95
103
23
47
77
23
108
120
13
Death
22
69
41
17
51
41
26
91
30
Note: HB-CEA: HB≥125.5 and CEA<3.395 represent 0; CEA≥3.395 or HB<125.5 represent 1; CEA≥3.395 and HB<125.5 represent 2.
In order to evaluate the predictors of OS (overall survival) after radical gastrectomy, the clinicopathological characteristics were evaluated using univariate and multivariate analyses. The Kaplan-Meier analysis showed that a higher HB-CEA score was associated with worse OS (p<0.001; Figure 3). Before PSM, the univariate analysis revealed that differentiation (HR: 0.045, 95% CI: 0.002-0.916, p=0.044), diameter of lesion (HR: 1.707, 95% CI: 1.080-2.698, p=0.022), cancer embolism (HR: 2.069, 95% CI: 1.468-2.917, p<0.001), nerve invasion (HR: 1.607, 95% CI: 1.138-2.268, p=0.007), pathological TNM stage (pTNM) (HR: 3.243, 95% CI: 2.193-4.797, p<0.001), depth of invasion (HR: 2.530, 95% CI: 1.636-3.912, p<0.001), lymph node metastasis (HR: 3.806, 95% CI: 2.385-6.075, p<0.001), Hb levels (HR: 2.131, 95% CI: 1.481-3.067, p<0.001), CEA levels (HR: 2.061, 95% CI: 1.463-2.903, p<0.001), and HB-CEA (HB-CEA=1, HR: 2.362, 95% CI: 1.462-3.817, p<0.001; HB-CEA=2, HR: 4.339, 95% CI: 2.583-7.289, p<0.001) were significantly correlated with patient prognosis in the training set. Univariate analysis after PSM revealed that cancer embolism (HR: 1.863, 95% CI: 1.279-2.713, p=0.001), nerve invasion (HR: 1.490, 95% CI: 1.021-2.173, p=0.038), pathological TNM stage (pTNM) (HR: 3.219, 95% CI: 1.938-5.348, p<0.001), depth of invasion (HR: 1.978, 95% CI: 1.032-3.792, p=0.040), lymph node metastasis (HR: 4.829, 95% CI: 2.437-9.569, p<0.001), Hb levels (HR: 1.691, 95% CI: 1.124-2.545, p=0.012), CEA levels (HR: 1.814, 95% CI: 1.245-2.643, p=0.002), and HB-CEA (HB-CEA=1, HR: 1.615, 95% CI: 0.933-2.797, p=0.087; HB-CEA=2, HR: 2.915, 95% CI: 1.655-5.134, p<0.001) were related to the prognosis of patients (Table 4). In the validation set, univariate analysis showed age (HR: 1.430, 95% CI: 1.035-1.976, p=0.030), differentiation(HR: 0.358, 95% CI: 0.176-0.731, p=0.005),diameter of lesion (HR: 2.264, 95% CI: 1.491-3.437, p<0.001), cancer embolism (HR: 2.299, 95% CI: 1.492-3.542, p<0.001), nerve invasion (HR: 2.241, 95% CI: 1.511-3.323, p<0.001), pathological TNM stage (pTNM) (HR: 3.537, 95% CI: 2.511-4.982, p<0.001), depth of invasion (HR: 2.766, 95% CI: 1.905-4.018, p<0.001), lymph node metastasis (HR: 3.841, 95% CI: 2.591-5.694, p<0.001), Hb levels (HR: 1.968, 95% CI: 1.402-2.761, p<0.001), CEA levels (HR: 2.273, 95% CI: 1.628-3.173, p<0.001), and HB-CEA (HB-CEA=1, HR: 2.585, 95% CI: 1.671-3.999, p<0.001; HB-CEA=2, HR: 5.427, 95% CI: 3.203-9.195, p<0.001) were associated with survival (Table 4).
Figure 3
Kaplan-Maier survival curves for overall survival of patients in the training set before (A) and after (B) propensity score matching. Note: HB-CEA: HB≥125.5 and CEA<3.395 represent 0; CEA≥3.395 or HB<125.5 represent 1; CEA≥3.395 and HB<125.5 represent 2.
Table 4
Univariate and Multivariate Analyses of Clinicopathological Characteristics in GC Patients
Characteristics
The training set
The validation set
Before PSM
After PSM
Univariate Analysis
Multivariate Analysis
Univariate Analysis
Multivariate Analysis
Univariate Analysis
Multivariate Analysis
HR (95% CI)
P value
HR (95% CI)
P value
HR (95% CI)
P value
HR (95% CI)
P value
HR (95% CI)
P value
HR (95% CI)
P value
Sex
0.910
0.706
0.519
Female
Ref
Ref
Ref
Male
0.979 (0.683-1.404)
1.080 (0.724-1.611)
0.893 (0.632-1.260)
Age (years)
0.082
0.107
0.030*
≤65
Ref
Ref
Ref
>65
1.357 (0.962-1.914)
1.373 (0.934-2.019)
1.430 (1.035-1.976)
Tumor location
0.695
0.408
0.289
Upper third
Ref
Ref
Ref
Middle and lower third
1.084 (0.724-1.623)
1.200 (0.780-1.846)
0.805 (0.539-1.202)
Differentiation
0.044*
0.271
0.005*
Moderate/Poor
Ref
Ref
Ref
Well
0.045 (0.002-0.916)
0.048 (0.001-10.681)
0.358 (0.176-0.731)
Diameter of lesion(cm)
0.022*
0.895
<0.001*
<3
Ref
Ref
Ref
≥3
1.707 (1.080-2.698)
0.961 (0.527-1.751)
2.264 (1.491-3.437)
Cancer embolism
<0.001*
0.001*
<0.001*
None
Ref
Ref
Ref
Yes
2.069 (1.468-2.917)
1.863 (1.279-2.713)
2.299 (1.492-3.542)
Nerve invasion
0.007*
0.038*
<0.001*
0.043*
None
Ref
Ref
Ref
Ref
Yes
1.607 (1.138-2.268)
1.490 (1.021-2.173)
2.241 (1.511-3.323)
1.525 (1.013-2.295)
pTNM
<0.001*
0.014*
<0.001*
0.026*
<0.001*
0.004*
I-II
Ref
Ref
Ref
Ref
Ref
Ref
III-IV
3.243 (2.193-4.797)
1.800 (1.125-2.881)
3.219 (1.938-5.348)
1.925 (1.082-3.423)
3.537 (2.511-4.982)
1.894 (1.225-2.928)
Depth of invasion
<0.001*
0.040*
<0.001*
T1-2
Ref
Ref
Ref
T3-4
2.530 (1.636-3.912)
1.978 (1.032-3.792)
2.766 (1.905-4.018)
LN metastasis
<0.001*
0.008*
<0.001*
0.004*
<0.001*
0.010*
N0
Ref
Ref
Ref
Ref
Ref
Ref
N1/N2/N3
3.806 (2.385-6.075)
2.134 (1.217-3.741)
4.829 (2.437-9.569)
3.137 (1.444-6.813)
3.841 (2.591-5.694)
1.949 (1.175-3.233)
Hb levels (g/L)
<0.001*
0.012*
<0.001*
≥125.5
Ref
Ref
Ref
<125.5
2.131 (1.481-3.067)
1.691 (1.124-2.545)
1.968 (1.402-2.761)
CEA levels (ng/mL)
<0.001*
0.002*
<0.001*
<3.395
Ref
Ref
Ref
≥3.395
2.061 (1.463-2.903)
1.814 (1.245-2.643)
2.273 (1.628-3.173)
HB-CEA
0
Ref
Ref
Ref
Ref
Ref
Ref
1
2.362 (1.462-3.817)
<0.001*
1.840 (1.134-2.985)
0.013*
1.615 (0.933-2.797)
0.087
1.411 (0.813-2.447)
0.221
2.585 (1.671-3.999)
<0.001*
2.049 (1.313-3.198)
0.002*
2
4.339 (2.583-7.289)
<0.001*
3.255 (1.927-5.500)
<0.001*
2.915 (1.655-5.134)
<0.001*
2.722 (1.544-4.800)
0.001*
5.427 (3.203-9.195)
<0.001*
4.007 (2.335-6.877)
<0.001*
Note: HB-CEA: HB≥125.5 and CEA<3.395 represent 0; CEA≥3.395 or HB<125.5 represent 1; CEA≥3.395 and HB<125.5 represent 2.
In the training set, the multivariate analysis before PSM showed that the HB-CEA (HB-CEA=1, HR: 1.840, 95% CI: 1.134-2.985, p=0.013; HB-CEA=2, HR: 3.255, 95% CI: 1.927-5.500, p<0.001), pathological TNM stage (pTNM) (HR: 1.800, 95% CI: 1.125-2.881, p=0.014), and lymph node metastasis (HR: 2.134, 95% CI: 1.217-3.741, p=0.008) were predictors of the clinical outcome (Table 4). Similar results were obtained with multivariate analysis after PSM. In the validation set, HB-CEA (HB-CEA=1, HR: 2.049, 95% CI: 1.313-3.198, p=0.002; HB-CEA=2, HR: 4.007, 95% CI: 2.335-6.877, p<0.001), nerve invasion (HR: 1.525, 95% CI: 1.013-2.295, p=0.043), pathological TNM stage (pTNM) (HR: 1.894, 95% CI: 1.225-2.928, p=0.004), and lymph node metastasis (HR: 1.949, 95% CI: 1.175-3.233, p=0.010) were associated with the prognosis of patients (Table 4).
Nomogram and Calibration Curve for Predicting the OS of Patients with Gastric Cancer
Using the abovementioned covariates identified by the statistical analyses, nomograms before and after PSM were constructed for predicting the 3-, 5-, and 7-year OS according to the training set. The score details of each nomogram predictor are shown in Figure 4. Summing the scores of all selected variables can facilitate the accurate determination of the survival probability of individual patients. The analysis of the data from the validation set showed that C-index of OS can be accurately predicted with a value of 0.723. The calibration curve showed good concordance between the predicted and observed values of 3-, 5-, and 7-year OS in the validation set (Figure 5).
Figure 4
Nomograms for predicting 3-, 5-, and 7-year OS for patients in the training set (before PSM, A; after PSM, B) with the indicated prognosis factors. Summing up points from all predictors could obtain total points. The predicted probabilities of OS can be obtained by projecting the location of the total points to the bottom scales. Note: HB-CEA: HB≥125.5 and CEA<3.395 represent 0; CEA≥3.395 or HB<125.5 represent 1; CEA≥3.395 and HB<125.5 represent 2; pTNM (I-II) represent 0; pTNM (III-IV) represent 1; LN metastasis(N0) represent 0; LN metastasis(N1/N2/N3) represent 1. Abbreviations: CEA: carcinoembryonic antigen; HB: hemoglobin; pTNM: pathological TNM stage; LN metastasis: lymph node metastasis.
Figure 5
The calibration curves for predicting 3 years, 5 years and 7 years OS for GC patients in the validation set. The OS predicted by the nomogram model is plotted on the x-axis, and the actual OS is plotted on the y-axis. Abbreviations:OS: overall survival; GC: gastric cancer.
Discussion
Hematological biomarkers, such as CEA and hemoglobin, determined from peripheral blood samples are useful prognostic indicators of clinical outcomes in patients with cancer. In this study, we established the presence of a relationship between peripheral blood biomarkers and clinical outcomes in patients with resectable gastric cancer. A lower hemoglobin level or high CEA were associated with sex, larger tumor size, more advanced pathological TNM stage, deeper depth of tumor invasion, more severe lymphatic metastasis, and worse survival status. Notably, this study reveals for the first time that HB-CEA is a more effective prognostic marker than either HB or CEA alone in patients with gastric cancer scheduled to undergo radical gastrectomy. Patients with gastric cancer and an HB-CEA score of 2 had worse prognoses than those with an HB-CEA score of 0, indicating that a higher HB-CEA score is related to the tumor burden and tumor progression. Thus, HB-CEA is a promising prognostic marker in patients with gastric cancer as tumor progression is closely related to both patient and tumor characteristics.Gastric cancer is a wasting disease, and most patients are diagnosed at an advanced stage 15. Most patients with gastric cancer suffer from malnutrition, which can exacerbate perioperative stress and further reduce the body's immunity, leading to a poor prognosis 16. Hemoglobin can reflect the nutritional status of the patient as well as the clinical condition. Therefore, it has been proposed that early enteral nutrition support can significantly enhance the cellular immune function of patients undergoing radical gastric cancer surgery, improve their nutritional status, and reduce the risk of postoperative complications 17. The current primary treatment for severe anemia is blood transfusion, which is believed to reduce the incidence of anemia-associated postoperative complications. However, transfusions are associated with a number of adverse effects, including pulmonary complications, graft-versus-host disease, and transmission of infectious diseases 18-23. Tumor markers comprise antigens and bioactive substances that are produced by tumor cells following the abnormal expression of oncogenes, whereas under normal conditions, these markers are present in miniscule amounts or are absent. The high or low expression of tumor markers reflects changes in the expression of relevant genes during tumor progression 24. CEA is a conventional tumor biomarker that reflects tumor development, and CEA levels are significantly associated with lymphovascular invasion 25,26. The involvement of CEA in the adhesion of tumor cells to liver parenchyma is associated with liver metastasis 12. The combination of these two analyses can predict the prognosis of patients and is an important guide to the use of enteral nutrition and other methods to correct anemia in patients with advanced gastric cancer.To our knowledge, this report describes a pilot study of HB-CEA as a novel scoring system. Herein, we demonstrated that the combination of hemoglobin and CEA levels is effective in predicting survival in patients with gastric cancer and that the predictive power of HB-CEA is superior to that of CEA or HB levels alone. However, there are a number of limitations of this study that need to be mentioned. Firstly, there is a need to increase the patient sample size and select patients from different geographic locations to reduce potential bias in retrospective study designs. Secondly, the postoperative treatment of patients can have an impact on the OS and could thus affect the diagnostic accuracy of the HB-CEA score. Thirdly, we were unable to fully assess the prognosis of patients with gastric cancer due to the inadequacy of data on disease-free survival. Therefore, the results of this study need to be validated with a larger sample size and analysis of more adequate clinicopathological and survival data from prospective studies.
Conclusion
In this retrospective study, we demonstrated the high value of hemoglobin combined with CEA in the analysis of gastric cancer progression and patient prognosis. As a new scoring system, HB-CEA is of great significance for precise and individualized tumor treatment.
Authors: M Gierth; R Mayr; A Aziz; S Krieger; B Wullich; A Pycha; M Lodde; U Salvadori; J Bründl; H M Fritsche; F Hofstädter; M T Pawlik; W Otto; M May; M Burger; S Denzinger Journal: J Cancer Res Clin Oncol Date: 2015-04-02 Impact factor: 4.553
Authors: Koray Yuruk; Sebastiaan A Bartels; Dan M J Milstein; Rick Bezemer; Bart J Biemond; Can Ince Journal: Transfusion Date: 2011-08-29 Impact factor: 3.157
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5