Clinical Features
Based on the above inclusion and exclusion criteria, a total of 1966 patients, including 1300 males (66.1%) and 666 females (33.9%), were enrolled in this study. The clinical characteristics of all patients and each cohort are summarized in Table 1. Patients older than 65 years accounted for 22.2% of all patients. The differentiation degree was mainly low (66.5%), and high differentiated tumors appeared only in a few patients (0.9%). More than 70% of patients received adjuvant chemotherapy after surgery, while the remaining did not undergo any postoperative treatment due to the early staging (22%) or personal requirements. The pre- and postoperative cohorts had similar population distributions in age, gender, BMI, etc. There were differences in the clinicopathological characteristics and TNM stage between each cohort. Patients with more elevated pre- or postoperative tumor markers had relatively more advanced TNM stages and larger tumor size.
Table 1
General characteristics of 1966 patients with gastric cancer.
Variables | n (%) /x ± S |
All patients | Pre-C0 | Pre-C1 | Pre-C2 | Pre-C3 | Post-C0 | Post-C1 | Post-C2 | Post-C3 |
Age (years) | 55.6 ± 11.6 | 54.2 ± 11.9 | 57.0 ± 10.8 | 59.1 ± 10.7 | 58.6 ± 11.3 | 54.6 ± 11.7 | 57.0 ± 11.6 | 57.4 ± 10.4 | 56.2 ± 10.9 |
BMI (kg/m2) | 22.1 ± 3.1 | 22.1 ± 3.2 | 22.3 ± 3.1 | 21.8 ± 3.0 | 21.7 ± 3.3 | 22.1 ± 3.2 | 22.2 ± 3.0 | 21.8 ± 3.2 | 22.2 ± 3.6 |
Tumor size (mm) | 4.7 ± 2.9 | 4.2 ± 2.8 | 5.0 ± 2.7 | 5.9 ± 2.9 | 6.9 ± 4.6 | 4.5 ± 2.9 | 4.7 ± 2.7 | 5.9 ± 3.2 | 6.3 ± 4.0 |
Gender | | | | | | | | | |
Male | 1300(66.1) | 715(62.3) | 418(70.5) | 147(75.4) | 20(64.5) | 718(63.0) | 448(69.8) | 114(73.1) | 20(71.4) |
Female | 666(33.9) | 432(37.7) | 175(29.5) | 48(24.6) | 11(35.5) | 422(37.0) | 194(30.2) | 42(26.9) | 8(28.6) |
Differentiated degree | | | | | | | | | |
High | 17(0.9) | 12(1.0) | 4(0.7) | 1(0.5) | 0(0.0) | 10(0.9) | 7(1.1) | 0(0.0) | 0(0.0) |
Middle | 641(32.6) | 352(30.7) | 193(32.5) | 85(43.6) | 11(35.5) | 369(32.4) | 208(32.4) | 56(35.9) | 8(28.6) |
Low | 1308(66.5) | 783(68.3) | 396(66.8) | 109(55.9) | 20(64.5) | 761(66.8) | 427(66.5) | 100(64.1) | 20(71.4) |
Signet-ring cell component | | | | | | | | | |
Yes | 1488(75.7) | 875(76.3) | 445(75.0) | 145(74.4) | 23(74.2) | 858(75.3) | 494(76.9) | 116(74.4) | 20(71.4) |
No | 478(24.3) | 272(23.7) | 148(25.0) | 50(25.6) | 8(25.8) | 282(24.7) | 148(23.1) | 40(25.6) | 8(28.6) |
T stage | | | | | | | | | |
pT1 | 467(23.8) | 356(31.2) | 92(15.5) | 13(6.7) | 4(12.9) | 321(28.2) | 129(20.1) | 14(9.0) | 3(10.7) |
pT2 | 214(10.9) | 141(12.3) | 62(10.5) | 8(4.1) | 3(9.7) | 125(11.0) | 76(11.8) | 12(7.7) | 1(3.6) |
pT3 | 692(35.2) | 368(32.1) | 223(37.6) | 88(45.1) | 13(41.9) | 399(35.0) | 220(34.3) | 65(41.7) | 8(28.6) |
pT4a | 575(29.2) | 272(23.7) | 210(35.4) | 82(42.1) | 11(35.5) | 287(25.2) | 212(33.0) | 60(38.5) | 16(57.1) |
pT4b | 18(0.9) | 8(0.7) | 6(1.0) | 4(2.1) | 0(0.0) | 8(0.7) | 5(0.8) | 5(3.2) | 0(0.0) |
N stage | | | | | | | | | |
pN0 | 648(33.0) | 423(36.9) | 179(30.2) | 39(20) | 7(22.6) | 420(36.8) | 188(29.3) | 36(23.1) | 4(14.3) |
pN1 | 332(16.9) | 190(16.6) | 104(17.5) | 36(18.5) | 2(6.5) | 179(15.7) | 116(18.1) | 35(22.4) | 2(7.1) |
pN2 | 332(16.9) | 184(16.0) | 98(16.5) | 42(21.5) | 8(25.8) | 166(14.6) | 127(19.8) | 34(21.8) | 5(17.9) |
pN3a | 389(19.8) | 203(17.7) | 126(21.2) | 52(26.7) | 8(25.8) | 230(20.2) | 124(19.3) | 27(17.3) | 8(28.6) |
pN3b | 265(13.5) | 147(12.8) | 86(14.5) | 26(13.3) | 6(19.4) | 145(12.7) | 87(13.6) | 24(15.4) | 9(32.1) |
TNM stage | | | | | | | | | |
I | 432(22.0) | 323(28.2) | 91(15.3) | 14(7.2) | 4(12.9) | 292(25.6) | 121(18.8) | 16(10.3) | 3(10.7) |
II | 589(30.0) | 345(30.1) | 186(31.4) | 52(26.7) | 6(19.4) | 342(30.0) | 198(30.8) | 45(28.8) | 4(14.3) |
III | 945(48.1) | 479(41.8) | 316(53.3) | 129(66.2) | 21(67.7) | 506(44.4) | 323(50.3) | 95(60.9) | 21(75.0) |
Borrmann type | | | | | | | | | |
I | 254(12.9) | 189(16.5) | 49(8.3) | 12(6.2) | 4(12.9) | 174(15.3) | 63(9.8) | 12(7.7) | 5(17.9) |
II | 458(23.3) | 323(28.2) | 111(18.7) | 21(10.8) | 3(9.7) | 277(24.3) | 167(26.0) | 11(7.1) | 3(10.7) |
III | 1168(59.4) | 584(50.9) | 408(68.8) | 155(79.5) | 21(67.7) | 652(57.2) | 381(59.3) | 117(75) | 18(64.3) |
IV | 86(4.4) | 51(4.4) | 25(4.2) | 7(3.6) | 3(9.7) | 37(3.2) | 31(4.8) | 16(10.3) | 2(7.1) |
Adjuvant chemotherapy | | | | | | | | | |
No | 478(24.3) | 407(35.5) | 132(22.3) | 27(13.8) | 3(9.7) | 390(34.2) | 157(24.5) | 17(10.9) | 5(17.9) |
Yes | 1397(71.1) | 740(64.5) | 461(77.7) | 168(86.2) | 28(90.3) | 750(65.8) | 485(75.5) | 139(89.1) | 23(82.1) |
Abbreviations: BMI: Body mass index TNM: Tumor, Lymph node, ,Metastasis |
Survival Analysis of Patients According to Single Pre- and Postoperative Tumor Markers
The 5-year OS rate for the entire cohort was 77.4% (95% CI: 75.1%-79.7%), and the 5-year OS rates for the patients with stage I-III disease were 96.4% (95% CI: 94.5%-98.4%), 87.5% (95% CI: 84.4%-90.1%), and 60.8% (95% CI: 56.9%-65.0%), respectively. The 5-year RFS rates for the entire cohort were 57.0% (95% CI: 54.1%-60.1%), and the 5-year OS rates for patients with stage I-III disease were 89.3% (95% CI: 85.7%-93.2%), 67.2% (95% CI: 62.0%-72.8%), and 36.4% (95% CI: 32.3%-41.0%), respectively. The elevated rates of preoperative CEA, CA72-4 and CA19-9 were 15.4%, 28.9% and 11.5%, and for postoperative CEA, CA72-4 and CA19-9, they were 22.4%, 23.8% and 14.8%, respectively.
The OS and RFS by Kaplan-Meier curves and the log-rank test for each preoperative tumor marker are shown in Supplementary Fig. 1. Patients with normal preoperative CEA were associated with longer OS (HR:1.919, 95% CI:1.506–2.447, P < 0.001) and RFS (HR: 1.552, 95% CI: 1.269–1.898, P < 0.001) compared with those with elevated preoperative CEA, with 5-year OS of 79.7% (95% CI: 77.4%-82.1%) versus 63.7% (95% CI: 56.9%-71.3%), 5-year RFS of 60.2% (95% CI: 57.0%-63.5%) versus 39.6% (95% CI: 32.4%-48.4%) (Figure S1a-b). Parallel results were observed for those with preoperative CA72-4 (Figure S1c-d) and CA19-9 levels (Figure S1e-f).
Next, the prognostic values of postoperative tumor markers were explored. Patients with normal postoperative CEA were associated with longer OS (HR: 1.480, 95% CI: 1.179–1.858, P < 0.001) and RFS (HR: 1.659, 95% CI: 1.397–1.970, P < 0.001) compared with those with elevated postoperative CEA, with 5-year OS of 79.3% (95% CI:77.0%-81.9%) versus 70.1% (95% CI:64.8%-75.9%), 5-year RFS of 60.6% (95% CI:57.3%-64.0%)versus 44.6% (95% CI: 38.3%-51.9%) (Figure S2a-b). Parallel results were observed in preoperative CA72-4 (Figure S2c-d) and CA19-9 levels (Figure S2e-f).
Survival Analysis of Patients According to Combined Preoperative and Postoperative Tumor Markers
As previously described, patients were divided into four groups (C0-C3) according to the total number of elevated tumor markers. The survival outcomes of the preoperative cohort are shown in Fig. 2a-b. The 5-year OS rates for the patients in preoperative groups C0-C3 were 83.8% (95% CI: 81.3%-86.4%), 72.8% (95% CI: 68.5%-77.4%), 58.9% (95% CI: 50.4%-68.9%), and 18.5% (95% CI:4.0%-33.0%), respectively. The 5-year RFS rates for the patients in preoperative groups C0-C3 were 66.1% (95% CI: 62.5%-69.9%), 49.7% (95% CI: 44.3%-55.7%), 33.1% (95% CI: 24.3%-44.9%), and 18.7% (95% CI: 10.1%-27.2%), respectively.
The Kaplan–Meier and log-rank tests showed significant differences in OS and RFS between each group. Pairwise comparisons were performed for adjacent subgroups (C0-C1, C1-C2, C2-C3), and there were significant differences between each of these pairs (P < 0.05 by the log-rank test). Similarly, the survival outcomes of the postoperative cohort are shown in Fig. 3a-b. The 5-year OS rates for the patients in postoperative groups C0-C3 were 82.1% (95% CI: 79.4%-84.8%), 76.1% (95% CI: 72.2%-80.3%), 57.6% (95% CI: 48.4%-68.5%), and 16.8% (95% CI: 5.1%-28.5%), respectively. The 5-year RFS rates for the patients in postoperative groups C0-C3 were 63.2% (95% CI: 59.4%-67.2%), 56.1% (95% CI: 51.0%-61.7%), 31.0% (95% CI:22.2%-43.4%), and 10.1% (95% CI: 3.5%-16.7%), respectively. There were also significant differences between each of these paired pre-cohort.
Next, we further performed the subgroup analysis stratified by different TNM stages. We found that in early GC (stage I-II) (Fig. 2c-f), the difference among the various cohorts was not significant (P > 0.05 by the log-rank test). However, in stage III GC (Fig. 2g-h), the difference between the overall cohort and most pairwise subgroups was significant (P < 0.05 by the log-rank test). The 5-year OS rates for preoperative groups C0-C2 in stage III patients were 66.6% (95% CI: 61.3%-86.4%), 60.6% (95% CI: 54.2%-67.7%), 48.4% (95% CI: 38.4%-60.9%), respectively ,and 3-year OS rates for stage III patients in preoperative group C3 was 42.1% (95% CI: 29.7%-54.5%). Similar results in post-cohort are shown in Fig. 3c-h; the difference among the various cohorts was significant in stage III but not in stages I-II. The 5-year OS rates for postoperative groups C0-C3 in stage III patients were 66.0% (95% CI: 60.9%-71.7%), 61.8% (95% CI: 55.6%-68.7%), 42.8% (95% CI: 31.5%-58.1%), and 11.7% (95% CI: 1.3%-22.1%), respectively.
Prognostic Factor Analysis and Efficacy Test
Cox regression analysis was conducted to further adjust other covariates. The univariable analysis showed that age, tumor size, TNM stage, adjuvant chemotherapy, Borrmann type, and tumor markers were associated with OS (Table 2). Multivariate analysis for OS further showed that elderly age (HR: 1.100, 95% CI: 1.392–1.763, P = 0.006), larger tumor size (HR: 2.411, 95% CI: 1.923–3.024,P < 0.001), advanced TNM stage (HR: 12.242, 95% CI: 7.282–20.570, P < 0.001), preoperative (HR: 6.001, 95% CI: 3.523–10.221, P < 0.001) and postoperative (HR: 8.149, 95% CI: 4.962–13.528, P < 0.001) elevated tumor markers and high grade Borrmann type (HR: 19.227, 95% CI: 9.908–37.311, P < 0.001) were independent poor prognostic factors. Multivariate analysis for RFS also indicated that preoperative (HR: 4.095, 95% CI: 2.623–6.392, P < 0.001) and postoperative (HR: 5.415, 95% CI: 3.538–8.288, P < 0.001) elevated tumor markers were the independent poor prognostic factors (Table 3). In addition, we conducted model optimization and prognostic Cox regression analysis of the preoperative and postoperative cohorts (Table 4). Adjusting other critical factors, the number of elevated tumor markers (both preoperative and postoperative) was an independent risk factor for the prognosis of GC patients. The survival rates decreased as the number of elevated tumor markers increased.
Table 2
Univariable and multivariable analyses of clinicopathological and tumor markers in all 1966 patients with gastric cancer for OS.
Characteristic | n(%) | Univariable analysis | Multivariable analysis |
P | HR(95%CI) | P | HR(95%CI) |
Age | > 65 | 437(22.2) | 0.006** | 1.100(1.392–1.763) | 0.093 | 1.229(0.966–1.564) |
BMI | < 18.5 | 248(12.6) | | | | |
| 18.5–24 | 1189(60.5) | 0.737 | 1.056(0.769–1.449) | | |
| > 24 | 529(26.9) | 0.542 | 0.895(0.628–1.277) | | |
Tumor size(mm) | < 5 | 1098(55.8) | | | | |
| 5–10 | 731(37.2) | < 0.001** | 2.411(1.923–3.024) | 0.134 | 1.204(0.944–1.535) |
| > 10 | 132(6.7) | < 0.001** | 5.954(4.363–8.123) | < 0.001* | 2.189(1.504–3.187) |
Gender | Female | 666(33.9) | 0.179 | 1.163(0.933–1.450) | | |
Differentiated degree | High | 17(0.9) | | | | |
| Middle | 641(32.6) | 0.411 | 1.793(0.446–7.209) | | |
| Low | 1308(66.5) | 0.734 | 1.275(0.314–5.169) | | |
Signet-ring cell component | Yes | 478(24.3) | 0.909 | 0.986(0.774–1.257) | | |
T stage | pT1 | 467(23.8) | | | | |
| pT2 | 214(10.9) | < 0.001** | 3.244(1.649–6.379) | | |
| pT3 | 692(35.2) | < 0.001** | 7.227(4.159–12.558) | | |
| pT4a | 575(29.2) | < 0.001** | 13.837(8.048–23.791) | | |
| pT4b | 18(0.9) | < 0.001** | 15.684(5.645–43.579) | | |
N stage | pN0 | 648(33.0) | | | | |
| pN1 | 332(16.9) | < 0.001** | 2.089(1.403–3.111) | | |
| pN2 | 332(16.9) | < 0.001** | 3.336(2.312–4.812) | | |
| pN3a | 389(19.8) | < 0.001** | 4.554(3.229–6.423) | | |
| pN3b | 265(13.5) | < 0.001** | 5.487(7.763–10.984) | | |
TNM stage | Stage I | 432(22.0) | | | | |
| Stage II | 589(30.0) | < 0.001** | 3.412(1.941–5.999) | 0.002* | 2.577(1.409–4.714) |
| Stage III | 945(48.1) | < 0.001** | 12.242(7.282–20.579) | < 0.001* | 7.818(4.378–13.958) |
Borrmann type | I | 254(12.9) | | | | |
| II | 458(23.3) | 0.015* | 2.253(1.171–4.333) | 0.046* | 1.974(1.012–3.851) |
| III | 1168(59.4) | < 0.001** | 6.096(3.335–11.143) | 0.002** | 2.729(1.455–5.121) |
| IV | 86(4.4) | < 0.001** | 19.227(9.908–37.311) | < 0.001* | 4.803(2.348–9.825) |
Tumor markers | preCEA(+) | 302(15.4) | < 0.001** | 1.919(1.506–2.447) | | |
| postCEA(+) | 440(22.4) | 0.001** | 1.480(1.179–1.858) | | |
| preCA724(+) | 568(28.9) | < 0.001** | 1.829(1.485–2.253) | | |
| postCA724(+) | 468(23.8) | < 0.001** | 1.922(1.551–2.380) | | |
| preCA199(+) | 206(10.5) | < 0.001** | 2.430(1.866–3.165) | | |
| postCA199(+) | 130(6.6) | < 0.001** | 3.250(2.439–4.331) | | |
Pre cohort | C0 | 1147(58.3) | | | | |
| C1 | 593(30.2) | < 0.001** | 1.789(1.419–2.254) | 0.094 | 1.227(0.966–1.560) |
| C2 | 195(9.9) | < 0.001** | 2.821(2.101–3.788) | 0.029* | 1.413(1.036–1.929) |
| C3 | 31(1.6) | < 0.001** | 6.001(3.523–10.221) | < 0.001* | 3.028(1.742–5.261) |
Post cohort | C0 | 1140(58.0) | | | | |
| C1 | 642(32.7) | 0.001** | 1.486(1.183–1.868) | 0.106 | 1.215(0.960–1.538) |
| C2 | 156(7.9) | < 0.001** | 2.860(2.093–3.910) | < 0.001* | 1.820(1.309–2.531) |
| C3 | 28(1.4) | < 0.001** | 8.149(4.962–13.528) | < 0.001* | 4.861(2.879–8.208) |
* indicates p-value of Cox regression less than 0.05, ** indicates p-value of Cox regression less than 0.01. |
Abbreviations: BMI: body mass index; OS: overall survival |
Table 3
Univariable and multivariable analyses of clinicopathological and tumor markers in all 1966 patients with gastric cancer for RFS.
Characteristic | n(%) | Univariable analysis | Multivariable analysis |
P | HR(95%CI) | P | HR(95%CI) |
Age | > 65 | 437(22.2) | 0.017* | 1.254(1.041–1.51) | 0.243 | 1.118(0.927–1.348) |
BMI | < 18.5 | 248(12.6) | | | | |
| 18.5–24 | 1189(60.5) | 0.153 | 0.845(0.671–1.065) | | |
| > 24 | 529(26.9) | 0.156 | 0.831(0.643–1.073) | | |
Tumor size(mm) | < 5 | 1098(55.8) | | | | |
| 5–10 | 731(37.2) | < 0.001** | 2.637(2.218–3.134) | < 0.001** | 1.438(1.193–1.734) |
| > 10 | 132(6.7) | < 0.001** | 5.379(4.168–6.943) | < 0.001** | 2.437(1.799–3.301) |
Gender | Female | 666(33.9) | 0.501 | 1.059(0.896–1.251) | | |
Differentiated degree | High | 17(0.9) | | | | |
| Middle | 641(32.6) | 0.142 | 2.384(0.706–11.366) | | |
| Low | 1308(66.5) | 0.242 | 2.297(0.57–9.257) | | |
Signet-ring cell component | Yes | 478(24.3) | 0.144 | 0.868(0.717–1.05) | | |
T stage | pT1 | 467(23.8) | | | | |
| pT2 | 214(10.9) | < 0.001** | 2.639(1.686–4.129) | | |
| pT3 | 692(35.2) | < 0.001** | 5.555(3.908–7.898) | | |
| pT4a | 575(29.2) | < 0.001** | 8.727(6.174–12.336) | | |
| pT4b | 18(0.9) | < 0.001** | 14.133(7.186–27.796) | | |
N stage | pN0 | 648(33.0) | | | | |
| pN1 | 332(16.9) | < 0.001** | 1.857(1.412–2.441) | | |
| pN2 | 332(16.9) | < 0.001** | 2.185(1.675–2.849) | | |
| pN3a | 389(19.8) | < 0.001** | 3.105(2.433–3.962) | | |
| pN3b | 265(13.5) | < 0.001** | 5.011(3.89–6.454) | | |
TNM stage | Stage I | 432(22.0) | | | | |
| Stage II | 589(30.0) | < 0.001** | 3.05(2.136–4.356) | 0.001** | 1.798(1.233–2.621) |
| Stage III | 945(48.1) | < 0.001** | 7.299(5.256–10.133) | < 0.001** | 3.526(2.459–5.057) |
Borrmann type | I | 254(12.9) | | | | |
| II | 458(23.3) | 0.069 | 1.5(0.969–2.322) | 0.23 | 1.319(0.839–2.073) |
| III | 1168(59.4) | < 0.001** | 4.718(3.199–6.959) | < 0.001** | 2.206(1.449–3.357) |
| IV | 86(4.4) | < 0.001** | 10.096(6.39-15.952) | < 0.001** | 2.719(1.62–4.562) |
Tumor markers | preCEA(+) | 302(15.4) | < 0.001** | 1.552(1.269–1.898) | | |
| postCEA(+) | 440(22.4) | < 0.001** | 1.659(1.397–1.97) | | |
| preCA724(+) | 568(28.9) | < 0.001** | 1.632(1.387–1.92) | | |
| postCA724(+) | 468(23.8) | < 0.001** | 1.646(1.39–1.949) | | |
| preCA199(+) | 206(10.5) | 0.010** | 2.166(1.749–2.684) | | |
| postCA199(+) | 130(6.6) | < 0.001** | 2.554(2.017–3.235) | | |
Pre cohort | C0 | 1147(58.3) | | | 0.156 | 1.142(0.951–1.372) |
| C1 | 593(30.2) | < 0.001** | 1.636(1.371–1.951) | 0.034* | 1.308(1.021–1.676) |
| C2 | 195(9.9) | < 0.001** | 2.544(2.013–3.214) | 0.005** | 1.932(1.218–3.063) |
| C3 | 31(1.6) | < 0.001** | 4.095(2.623–6.392) | | |
Post cohort | C0 | 1140(58.0) | | | | |
| C1 | 642(32.7) | 0.001** | 1.341(1.123–1.601) | 0.271 | 1.108(0.923–1.332) |
| C2 | 156(7.9) | < 0.001** | 2.845(2.249–3.598) | < 0.001** | 1.915(1.494–2.453) |
| C3 | 28(1.4) | < 0.001** | 5.415(3.538–8.288) | < 0.001** | 3.132(2.012–4.878) |
* indicates p-value of Cox regression less than 0.05, ** indicates p-value of Cox regression less than 0.01. |
Abbreviations: BMI: body mass index; RFS: recurrence-free survival |
Table 4
Survival analysis after stepwise model optimization.
Cohort | Model 1 | Model 2 | Model 3 |
P | HR(95%CI) | P | HR(95%CI) | P | HR(95%CI) |
Pre-C0 | Reference | Reference | Reference |
Pre-C1 | < 0.001** | 1.768(1.401–2.232) | 0.003** | 1.431(1.129–1.815) | 0.029* | 1.301(1.027–1.648) |
Pre-C2 | < 0.001** | 2.748(2.041-3.700) | < 0.001** | 2.072(1.531–2.806) | 0.001** | 1.692(1.253–2.286) |
Pre-C3 | < 0.001** | 5.694(3.332–9.733) | < 0.001** | 4.716(2.756–8.069) | < 0.001** | 4.287(2.507–7.329) |
Post-C0 | Reference | Reference | Reference |
Post-C1 | 0.001** | 1.462(1.162–1.840) | 0.007** | 1.372(1.089–1.729) | 0.026* | 1.301(1.032–1.639) |
Post-C2 | < 0.001** | 2.791(2.039–3.821) | < 0.001** | 2.374(1.731–3.257) | < 0.001** | 2.245(1.639–3.077) |
Post-C3 | < 0.001** | 7.850(4.748–12.979) | < 0.001** | 6.390(3.858–10.583) | < 0.001** | 5.328(3.214–8.834) |
Model 1 = Adjusted by age + BMI + sex; Model 2 = Model 1 + tumor size + differentiated degree + Borrmann type + adjuvant chemotherapy; Model 3 = Model 2 + TNM stage. |
* indicates p-value of Cox regression less than 0.05, ** indicates p-value of Cox regression less than 0.01. |
By integrating several important clinical information and the number of elevated tumor markers in patients with gastric cancer, we established a nomogram prognostic model((Fig. 4a). All patients were divided into points-low group and points-high group by nomogram total points. The overall survival time of points-low group was significantly longer than that of points-high group (HR: 6.321, 95% CI: 4.823–8.282,P < 0.001;Figure 4b).With calibration curve, it is found that the survival prediction model with elevated number of tumor markers has a good fit. The C-index was used to evaluate the prediction accuracy of the prognostic model of GC patients by calculating the discrimination between the predicted value and the actual accuracy of the Cox regression model in the survival analysis. The C-indexes of different predictive indicators are shown in Fig. 5. The C-index for a single tumor marker ranged from 0.53 to 0.56. Among the three tumor markers, the C-index of pre- and post-CA72-4 (both 0.56) was the highest. The combined use of tumor markers had a higher C-index (0.65–0.66) for prognostic analysis than a single tumor marker. Moreover, the combination of TNM staging and preoperative (0.76) or postoperative (0.74) cohort could further improve the C-index for the survival analysis. There was no significant difference between preoperative and postoperative tumor markers in predicting patient survival (P = 0.12).
Patient survival and adjuvant chemotherapy in stage II/III patients
In this study, we also analyzed the relationship between receiving adjuvant chemotherapy and patient survival. The 5-year OS rates for patients receiving adjuvant chemotherapy in stage II/III patients were 71.9% (95% CI: 68.9%-75.0%), while the 5-year OS rates for patients without adjuvant chemotherapy in stage II/III patients were 66.8% (95% CI: 58.9%-75.8%). We further analyzed the survival difference between C0-C3 group with or without adjuvant chemotherapy. (Supplementary Fig. 3). In patients without adjuvant chemotherapy, there was no significant difference in survival between the cohort subgroups (P > 0.05 for the log-rank test), and in patients with adjuvant chemotherapy, the difference was significant (P < 0.05 for the log-rank test). This was consistent in the pre- and postoperative cohorts.