The clinicopathological characteristics of EOGC and LOGC differ significantly.
The SEER database included 27,948 patients, with 2,698 diagnosed with EOGC and the remaining 25,250 with LOGC (Table 1). We compared the clinicopathological characteristics between the two groups. EOGC patients were characterized by a higher proportion of females, fewer tumors in the upper third of the stomach, a higher incidence of poorly differentiated carcinoma and signet ring cell carcinoma (SRCC), and increased rates of lymph node and distant metastasis, as well as more frequent use of chemotherapy and radiotherapy. Table 1 presents the detailed baseline characteristics of the SEER cohort.
Table 1
Clinicopathological characteristics between EOGC and LOGC patients in the SEER cohort.
Variable | | EOGC(<=45) | LOGC(>45) | p-value |
---|
N (%) | n = 2698 | n = 25250 |
---|
Race: | | | 0.083 |
Black | 295 (10.9%) | 2943 (11.7%) | |
Other | 501 (18.6%) | 4597 (18.2%) | |
Unknown | 23 (0.85%) | 128 (0.51%) | |
White | 1876 (69.6%) | 17585 (69.6%) | |
Sex: | | | < 0.001 |
Female | 1245 (46.2%) | 8759 (34.7%) | |
Male | 1450 (53.8%) | 16494 (65.3%) | |
Tumor Location: | | | < 0.001 |
Lower | 537 (19.9%) | 5602 (22.2%) | |
Middle | 693 (25.7%) | 5129 (20.3%) | |
Others | 279 (10.4%) | 1930 (7.64%) | |
Unknown | 434 (16.1%) | 2697 (10.7%) | |
Upper | 752 (27.9%) | 9895 (39.2%) | |
Tumor size: | | | < 0.001 |
>4cm | 591 (21.9%) | 7178 (28.4%) | |
<=4cm | 826 (30.6%) | 8616 (34.1%) | |
Unknown | 1278 (47.4%) | 9459 (37.5%) | |
Differentiation: | | | < 0.001 |
Moderately | 286 (10.6%) | 6188 (24.5%) | |
Poorly | 1910 (70.9%) | 14123 (55.9%) | |
Undifferentiated | 37 (1.37%) | 280 (1.11%) | |
Unknown | 409 (15.2%) | 3584 (14.2%) | |
Well | 53 (1.97%) | 1078 (4.27%) | |
Pathological type: | | | < 0.001 |
Adenocarcinoma | 1644 (61.0%) | 20574 (81.5%) | |
SRCC | 1051 (39.0%) | 4679 (18.5%) | |
TNM stage: | | | < 0.001 |
I | 309 (11.5%) | 5836 (23.1%) | |
II | 405 (15.0%) | 5232 (20.7%) | |
III | 479 (17.8%) | 4675 (18.5%) | |
IV | 1502 (55.7%) | 9510 (37.7%) | |
T stage: | | | < 0.001 |
T1 | 235 (19.6%) | 3487 (27.4%) | |
T2 | 130 (10.9%) | 1550 (12.2%) | |
T3 | 432 (36.1%) | 4714 (37.0%) | |
T4 | 381 (31.8%) | 2892 (22.7%) | |
TX | 19 (1.6%) | 92 (0.7%) | |
N stage: | | | < 0.001 |
N0 | 974 (36.1%) | 10871 (43.0%) | |
N1 | 700 (26.0%) | 6758 (26.8%) | |
N2 | 285 (10.6%) | 2655 (10.5%) | |
N3 | 342 (12.7%) | 2714 (10.7%) | |
NX | 394 (14.6%) | 2255 (8.93%) | |
M stage: | | | < 0.001 |
M0 | 1193 (44.3%) | 15743 (62.3%) | |
M1 | 1502 (55.7%) | 9510 (37.7%) | |
Gastrectomy: | | | < 0.001 |
No | 1497 (55.5%) | 12468 (49.4%) | |
Unknown | 1 (0.04%) | 50 (0.20%) | |
Yes | 1197 (44.4%) | 12735 (50.4%) | |
NACT: | | | 0.002 |
No/unknown | 2158 (80.1%) | 20824 (82.5%) | |
Yes | 537 (19.9%) | 4429 (17.5%) | |
ELNs: | | | 0.057 |
<16 | 1870 (69.4%) | 17986 (71.2%) | |
>=16 | 730 (27.1%) | 6534 (25.9%) | |
Unknown | 95 (3.53%) | 733 (2.90%) | |
Chemotherapy: | | | < 0.001 |
No/Unknown | 432 (16.0%) | 8752 (34.7%) | |
Yes | 2263 (84.0%) | 16501 (65.3%) | |
Radiotherapy: | | | < 0.001 |
No/Unknown | 2001 (74.2%) | 17788 (70.4%) | |
Yes | 694 (25.8%) | 7465 (29.6%) | |
SRCC: Signet-ring cell carcinoma; NACT: Neoadjuvant chemotherapy; ELNs: examined lymph nodes; |
In the Chinese cohort of 5,727 patients, 510 were diagnosed with EOGC, while 5,217 were diagnosed with LOGC (Table 2). The findings from the SEER cohort were validated in the Chinese cohort. Additionally, we found a higher incidence of vascular tumor thrombus in EOGC patients. In terms of serum tumor markers, EOGC patients had lower positivity rates for CEA and CA242, but a higher positivity rate for CA125. Table 2 presents detailed baseline characteristics of the Chinese cohort.
Table 2
Clinicopathological characteristics between EOGC and LOGC patients in the Chinese cohort.
Variable | | EOGC(<=45) | LOGC(>45) | p-value |
---|
N (%) | | n = 510 | n = 5217 |
---|
Sex: | | | < 0.001 |
Female | 263 (51.6%) | 1374 (26.3%) | |
Male | 247 (48.4%) | 3843 (73.7%) | |
BMI: | | | 0.141 |
<18.5 | 61 (12.0%) | 503 (9.64%) | |
18.5–23.9 | 308 (60.4%) | 3346 (64.1%) | |
>=24 | 141 (27.6%) | 1368 (26.2%) | |
Family history: | | | 0.409 |
No | 407 (79.8%) | 4247 (81.4%) | |
Yes | 103 (20.2%) | 970 (18.6%) | |
Tumor location: | | | < 0.001 |
Lower | 320 (62.7%) | 2929 (56.1%) | |
Middle | 122 (23.9%) | 855 (16.4%) | |
Total | 17 (3.4%) | 133 (2.6%) | |
Upper | 51 (10.0%) | 1300 (24.9%) | |
Tumor size: | | | < 0.001 |
>4cm | 254 (49.8%) | 2254 (43.2%) | |
<=4cm | 242 (47.5%) | 2892 (55.4%) | |
Unknown | 14 (2.7%) | 71 (1.4%) | |
Differentiation: | | | < 0.001 |
Moderate | 87 (17.1%) | 2610 (50.0%) | |
Poor | 417 (81.8%) | 2506 (48.0%) | |
Well | 6 (1.1%) | 101 (2.0%) | |
Pathological type: | | | 0.018 |
Adenocarcinoma | 301 (59.0%) | 2787 (53.4%) | |
SRCC | 209 (41.0%) | 2430 (46.6%) | |
Vascular tumor thrombus: | | | 0.018 |
No | 301 (59.0%) | 2787 (53.4%) | |
Yes | 209 (41.0%) | 2430 (46.6%) | |
Nerve invasion: | | | 0.351 |
No | 242 (47.5%) | 2594 (49.7%) | |
Yes | 268 (52.5%) | 2623 (50.3%) | |
TNM stage: | | | < 0.001 |
I | 156 (30.6%) | 1330 (25.5%) | |
II | 82 (16.1%) | 962 (18.4%) | |
III | 227 (44.5%) | 2700 (51.8%) | |
IV | 45 (8.8%) | 225 (4.31%) | |
T stage: | | | 0.001 |
T1 | 153 (30.0%) | 1140 (21.9%) | |
T2 | 55 (10.8%) | 622 (11.9%) | |
T3 | 32 (6.3%) | 351 (6.7%) | |
T4 | 270 (52.9%) | 3104 (59.5%) | |
N stage: | | | 0.068 |
N0 | 173 (33.9%) | 1799 (34.5%) | |
N1 | 77 (15.1%) | 894 (17.1%) | |
N2 | 89 (17.5%) | 1046 (20.0%) | |
N3 | 171 (33.5%) | 1478 (28.3%) | |
M stage: | | | < 0.001 |
M0 | 465 (91.2%) | 4992 (95.7%) | |
M1 | 45 (8.8%) | 225 (4.3%) | |
ELNs: | | | 0.560 |
<16 | 49 (9.61%) | 550 (10.5%) | |
>=16 | 461 (90.4%) | 4667 (89.5%) | |
NACT: | | | 0.159 |
No/unknown | 446 (87.5%) | 4673 (89.6%) | |
Yes | 64 (12.5%) | 544 (10.4%) | |
Chemotherapy: | | | < 0.001 |
No/Unknown | 173 (33.9%) | 2584 (49.5) | |
Yes | 337 (66.1%) | 2633 (50.5%) | |
Complication: | | | 0.181 |
No | 484 (94.9%) | 4865 (93.3%) | |
Yes | 26 (5.1%) | 352 (6.7%) | |
Postoperative recurrence: | | | 0.107 |
No | 507 (99.4%) | 5133 (98.4%) | |
Yes | 3 (0.6%) | 84 (1.6%) | |
Distant metastasis (Pre-and Postoperative): | | | < 0.001 |
No | 405 (79.4%) | 4548 (87.2%) | |
Yes | 105 (20.6%) | 669 (12.8%) | |
Postoperative recurrence and metastasis: | | | 0.138 |
No | 462 (90.6%) | 4827 (92.5%) | |
Yes | 48 (9.4%) | 390 (7.5%) | |
CEA: | | | < 0.001 |
Negative | 469 (92.0%) | 4155 (79.6%) | |
Positive | 41 (8.0%) | 1062 (20.4%) | |
CA199: | | | 0.080 |
Negative | 426 (83.5%) | 4184 (80.2%) | |
Positive | 84 (16.5%) | 1033 (19.8%) | |
CA125: | | | < 0.001 |
Negative | 459 (90.0%) | 4909 (94.1%) | |
Positive | 51 (10.0%) | 308 (5.9%) | |
AFP: | | | 0.429 |
Negative | 484 (94.9%) | 4900 (93.9%) | |
Positive | 26 (5.1%) | 317 (6.1%) | |
CA242: | | | 0.003 |
Negative | 470 (92.2%) | 4566 (87.5%) | |
Positive | 40 (7.8%) | 651 (12.5%) | |
CA724: | | | 0.689 |
Negative | 417 (81.8%) | 4222 (80.9%) | |
Positive | 93 (18.2%) | 995 (19.1%) | |
SRCC: Signet-ring cell carcinoma; NACT: Neoadjuvant chemotherapy; ELNs: examined lymph nodes; |
The prognosis of EOGC patients is better than that of LOGC patients.
Since approximately 50% of patients in the SEER cohort either did not undergo surgery or had unknown surgical status, while all patients in the Chinese cohort had undergone surgery (either curative or palliative), we included only 13,932 post-surgical patients from the SEER cohort in the prognostic analysis (Table S1). RCS analyses revealed a significant nonlinear relationship between age and HR in both the SEER and Chinese cohorts (P-non-linear < 0.001). In the SEER cohort, 66 years was identified as the cut-off point, after which HR significantly increased (Fig. 2A); in the Chinese cohort, 61 years served as the cut-off point, showing a similar upward trend in HR (Fig. 2B). This suggests that as the age at gastric cancer diagnosis increases, mortality risk gradually rises and then significantly accelerates after a specific age threshold. These two cut-off points reveal the staged impact of diagnosis age on gastric cancer mortality risk: early changes in mortality risk are relatively slow, but increase rapidly after the cutoff, underscoring the importance of diagnosis age in assessing mortality risk for gastric cancer.
Prognostic analyses of the SEER and Chinese cohorts demonstrated that EOGC patients had significantly better OS compared to LOGC patients (p < 0.001). In the SEER cohort (Fig. 2C), the 5-year OS rate for EOGC patients was 48.7%, significantly higher than the 45.4% for LOGC patients (HR = 0.82, p < 0.001). In the Chinese cohort (Fig. 2D), the 5-year OS rate for EOGC patients was 72.2%, again significantly higher than the 62.8% for LOGC patients (HR = 0.65, p < 0.001).
Given the significant baseline differences between EOGC and LOGC, we performed 1:2 PSM to adjust for potential confounding factors affecting prognosis. In the SEER cohort, PSM resulted in a total of 3,434 matched patients, comprising 1,158 EOGC and 2,276 LOGC patients (Table S2). In the Chinese cohort, PSM yielded 1,353 matched patients, with 469 EOGC and 884 LOGC patients (Table S3). In the PSM-adjusted SEER cohort (Fig. 2E), the 5-year OS rate for EOGC patients was 48.3%, compared to 43.1% for LOGC patients, with EOGC continuing to show a significant survival advantage (HR = 0.83, p < 0.001). Similarly, in the PSM-adjusted Chinese cohort (Fig. 2F), the 5-year OS rate for EOGC patients was 72.6%, significantly higher than the 64.3% for LOGC patients (HR = 0.71, p < 0.001). These results suggest that in both the SEER cohort of Western populations and the Chinese cohort, EOGC patients maintain a survival advantage even after adjusting for baseline differences, further validating the relatively favorable prognosis of EOGC across different populations.
We subsequently performed univariate and multivariate Cox regression analyses on EOGC patients to identify independent prognostic factors. In the SEER cohort, multivariate analysis identified independent prognostic factors for EOGC patients, including tumor size, differentiation, T stage, N stage, distant metastasis, radiotherapy, and the number of ELNs (Table S3). In the Chinese cohort, independent risk factors for prognosis included tumor size, nerve invasion, T stage, N stage, distant metastasis, as well as the tumor markers CA199, CA125, and CA724 (Table S4).
Subgroup analyses of clinicopathological features in EOGC and LOGC patients and the impact of chemotherapy on prognosis
We performed a subgroup analysis of the PSM-adjusted SEER cohort to investigate EOGC and LOGC patients (Fig. 3). The results revealed significant subgroup effects (p for interaction < 0.05) in four variables: tumor size, pathology type, TNM stage, and chemotherapy. Specifically, EOGC patients with smaller tumor size (HR = 0.63, p < 0.001), SRCC pathology (HR = 0.72, p < 0.001), and stage I-III (HR = 0.76, p < 0.001) demonstrated a survival advantage over LOGC patients. In evaluating the impact of chemotherapy on prognosis, although EOGC patients receiving chemotherapy still had a survival advantage (HR = 0.59, p < 0.001), this advantage was significantly weaker compared to those who did not undergo chemotherapy (HR = 0.88, p = 0.022). This suggests that although chemotherapy benefits EOGC patients, the relative survival benefit they gain is less than that of LOGC patients.
In the subgroup analysis of the PSM-adjusted Chinese cohort (Fig.S1), we similarly identified significant subgroup effects in sex, tumor size, TNM stage, T stage, and chemotherapy (P for interaction < 0.05). EOGC patients with smaller tumors and stage I-III had significantly better prognosis compared to LOGC patients, consistent with the findings in the SEER cohort. Regarding the impact of chemotherapy, the Chinese cohort once again validated the SEER cohort findings: EOGC patients who did not undergo chemotherapy had a significant survival advantage over LOGC patients (HR = 0.34, p < 0.001), while those who received chemotherapy still showed a survival advantage, though reduced (HR = 0.79, p = 0.032). This further supports the conclusion that EOGC patients derive relatively less survival benefit from chemotherapy.
Analyses of differences in distant metastasis sites and prognosis in EOGC and LOGC patients
The incidence of distant metastasis in EOGC patients was significantly higher than in LOGC patients, a finding validated across both cohorts. To further investigate the characteristics of distant metastasis in EOGC and LOGC patients and its impact on prognosis, we included 11,012 patients with distant metastasis from the SEER cohort and 774 from the Chinese cohort for analysis. We first compared the differences in metastasis sites between EOGC and LOGC patients. In the SEER cohort, bone and other metastasis sites were more common in EOGC patients, while liver and lung metastasis occurred more frequently in LOGC patients (Table 3). Similar results were observed in the Chinese cohort (Table 4), where liver and lung metastasis were more common in LOGC patients, while ovarian metastasis was more frequent in EOGC patients.
Table 3
Distribution of distant metastasis sites between EOGC and LOGC patients in the SEER cohort
Variable | | EOGC(<=45) | LOGC(>45) | p-value |
---|
N (%) | n = 1502 | n = 9510 |
---|
D.M-distant.LN: | | | 0.194 |
No | 439 (29.2%) | 2666 (28.0%) | |
Unknown | 623 (41.5%) | 3838 (40.4%) | |
Yes | 440 (29.3%) | 3006 (31.6%) | |
D.M-liver: | | | < 0.001 |
No | 1111 (74.0%) | 5327 (56.0%) | |
Unknown | 49 (3.2%) | 286 (3.0%) | |
Yes | 342 (22.8%) | 3897 (41.0%) | |
D.M-bone: | | | 0.009 |
No | 1226 (81.6%) | 8038 (84.5%) | |
Unknown | 56 (3.8%) | 342 (3.6%) | |
Yes | 220 (14.6%) | 1130 (11.9%) | |
D.M-brain: | | | 0.935 |
No | 1416 (94.3%) | 8975 (94.4%) | |
Unknown | 61 (4.1%) | 370 (3.9%) | |
Yes | 25 (1.6%) | 165 (1.7%) | |
D.M-lung: | | | < 0.001 |
No | 1303 (86.8%) | 7851 (82.6%) | |
Unknown | 65 (4.3%) | 409 (4.3%) | |
Yes | 134 (8.9%) | 1250 (13.1%) | |
D.M-others: | | | < 0.001 |
No | 559 (37.2%) | 5679 (59.7%) | |
Unknown | 14 (0.9%) | 86 (0.9%) | |
Yes | 929 (61.9%) | 3745 (39.4%) | |
D.M-distant.LN: Distant.metastasis-distant lymph nodes; D.M-others: Distant.metastasis-other sites. |
D.M-multiple: Distant.metastasis-multiple organs. |
Table 4
Distribution of distant metastasis sites between EOGC and LOGC patients in the Chinese cohort
Variable | | EOGC(<=45) | LOGC(>45) | p-value |
---|
N (%) | n = 105 | n = 669 |
---|
DX-distant.LN: | | | 0.669 |
No | 87 (82.9%) | 540 (80.7%) | |
Yes | 18 (17.1%) | 129 (19.3%) | |
DX-bone: | | | 0.107 |
No | 92 (87.6%) | 621 (92.8%) | |
Yes | 13 (12.4%) | 48 (7.2%) | |
DX-brain: | | | 0.607 |
No | 105 (100%) | 661 (98.8%) | |
Yes | 0 (0.00%) | 8 (1.2%) | |
DX-lung: | | | 0.018 |
No | 103 (98.1%) | 607 (90.7%) | |
Yes | 2 (1.9%) | 62 (9.3%) | |
DX-ovary: | | | < 0.001 |
No | 77 (73.3%) | 638 (95.4%) | |
Yes | 28 (26.7%) | 31 (4.63%) | |
DX-liver: | | | 0.006 |
No | 90 (85.7%) | 486 (72.6%) | |
Yes | 15 (14.3%) | 183 (27.4%) | |
DX-peritoneum | | | 0.160 |
No | 87 (82.9%) | 586 (87.6%) | |
Yes | 18 (17.1%) | 83 (12.4%) | |
DX-others: | | | 0.076 |
No | 101 (96.2%) | 605 (90.4%) | |
Yes | 4 (3.8%) | 64 (9.6%) | |
D.M-distant.LN: Distant.metastasis-distant lymph nodes; D.M-other: Distant.metastasis-other sites. |
D.M-multiple: Distant.metastasis-multiple organs. |
We subsequently performed subgroup analyses on the prognosis of EOGC and LOGC patients with different metastasis sites. In the SEER cohort, aside from other metastasis sites, no statistically significant difference in prognosis was observed between EOGC and LOGC patients for each metastasis site (Fig. 4A). Notably, EOGC patients with metastasis to other sites exhibited a better prognosis than LOGC patients (HR = 0.91, p = 0.031). In the Chinese cohort, the prognosis of EOGC patients differed (Fig. 4B): those with peritoneal metastasis had significantly worse prognosis than LOGC patients (HR = 2.23, P = 0.031), whereas those with multiple organ metastasis had significantly better prognosis than LOGC patients (HR = 0.70, p = 0.044). This suggests that the impact of different metastasis sites on prognosis varies between EOGC and LOGC patients across the two populations.
Additionally, we performed univariate COX regression analyses on EOGC patients with distant metastasis to examine prognostic differences by metastasis site. In the SEER cohort, the prognosis for distant lymph node, liver, and other single metastasis sites was significantly better than for multiple organ metastasis, with distant lymph node metastasis demonstrating the greatest survival advantage (Fig. 4C). In the Chinese cohort, peritoneal metastasis had the worst prognosis, while single organ metastasis, bone metastasis, liver metastasis, distant lymph node metastasis, and ovarian metastasis all had better prognoses than peritoneal metastasis (Fig. 4D). These findings highlight significant prognostic differences for EOGC patients based on metastasis site, suggesting that the metastasis pattern plays a crucial role in survival outcomes.