Clinicopathological characteristics of the patients
Of all 57 cases, 45 patients were male and 12 patients were female. The average age was 62.1 years old, ranging from 30 to 79 years. There were 32 cases in the <65 year-old group, while 25 cases in ≥65 year-old group. In terms of surgical methods, 11 patients underwent Billroth I anastomosis, 9 patients received distal gastrectomy and Roux en-Y anastomosis, while 37 patients experienced total gastrectomy plus Roux en-Y anastomosis. Only 5 cases received neoadjuvant chemotherapy. Most of the patients (52 cases) were with two malignant lesions, while 3 cases and 2 cases were with three and four malignant lesions respectively. Only 4 patients were with tumors <2 cm, and 53 patients with tumors ≥2 cm. Histologically, 11 patients with well-differentiated type, and 46 patients with poorly-differentiated type. Consistency of histology was positive in 33 cases. According to pTNM staging criteria, pT1 was regarded in 20 patients, 10 patients were defined as pT2, and 27 patients were regarded as pT3-T4. The consistency of tumor pT staging was positive in 27 cases. Ulcer presented in most of the patients (53 cases). Lymphovascular cancer plug and nerve invasion were detected positively in 13 and 14 cases respectively. Distant metastasis appeared in 2 cases. Postoperative pathology showed that LNM occurred in 24 patients. There were 3 patients with postoperative complications, including anastomotic fistula, intraabdominal hemorrhage and cardiac insufficiency.
As summarized in Table 3, due to 15 patients (26.3%) were loss to follow-up, there were 42 patients who had the data of follow-up in this study. Tumor recurrence occurred in 8 patients (19.0%), among which 7 cases (16.7%) died from recurrence. The median of follow-up time was 27.5 months (ranging from 3 to 33 months) in patients with tumor recurrence, and it was 28.5 months (ranging from 1 to 91 months) in patients without recurrence.
Univariate analysis of influence factors of LNM
According to the presence of LNM, all 57 patients with SMGC were divided into positive group (n=24) and negative group (n=33). Univariate analysis was performed to evaluate the influence factors of LNM of SMGC. Histologically, only one case (9.1%) of 11 patients with well-differentiated type was with positive LNM, and 23 cases (50.0%) of 46 patients with poorly-differentiated type were with positive LNM. The incidence of LNM in patients with poorly-differentiated type was significantly higher than that in patients with well-differentiated type (P=0.017). In patients with pT1 and pT2, 3 cases (15.0%) of 20 patients and 3 cases (30.0%) of 10 patients were with positive LNM, respectively. While LNM was detected as positive in 18 cases (66.7%) of 27 patients with pT3-T4. Compared to patients with pT1 and pT2, the rate of LNM was obviously higher in patients with pT3-T4 (P=0.001). The incidence of LNM in patients without lymphovascular cancer plug (27.3%, 12 of 44 patients) was significantly lower than that (92.3%, 12 of 13 patients) in patients with lymphovascular cancer plug (P=0.000). Similarly, the rate of LNM was remarkably lower in patients without nerve invasion (27.9%, 12 of 43 patients) compared to the patients with nerve invasion (85.7%, 12 of 14 patients) (P=0.000). In positive group, the median of preoperative CA125 level was 13.355 U/ml (range from 4.46 to 39.09 U/ml), which was significantly higher than that of 10.05 U/ml (range from 3.28 to 18.87 U/ml) in negative group (P=0.001) (Figure 1). It showed that histological type, tumor pT staging, lymphovascular cancer plug, nerve invasion and preoperative CA125 level were the risk factors of LNM in patients with SMGC. (Table 1)
Multivariate analysis of the independent risk factors of LNM
Based on the outcomes of univariate analysis, histological type, tumor pT staging, lymphovascular cancer plug, nerve invasion and preoperative CA125 were defined as the independent variables, and dummy variable was set in tumor pT staging. LNM was regarded as the dependent variable. Binary logistic regression was performed to validate the independent predictive risk factors of LNM. Compared with the patients without lymphovascular cancer plug, the risk of LNM increased significantly in patients with lymphovascular cancer plug (P=0.004; 95%CI, 6.445~24782.173). The increase of preoperative CA125 level was significantly positively associated with the risk of LNM of SMGC (P=0.007; 95%CI, 1.131~2.192). Multivariate analysis indicated that lymphovascular cancer plug and preoperative CA125 were the independent predictive risk factors of LNM in patients with SMGC. (Table 2)
Univariate analysis of influence factors of tumor recurrence
Because of the cases of loss to follow-up, only 42 patients were ultimately included in survival analysis. According to the presence of tumor recurrence, all patients were separated into positive group (n=8) and negative group (n=34). Univariate analysis was used to investigate the influence factors of recurrence in SMGC patients. No significant difference of follow-up time existed between the positive and negative groups. There were 6 cases (35.3%) with tumor recurrence in 17 patients with LNM, and 2 cases (8.0%) with tumor recurrence in 25 patients without LNM. The incidence of recurrence was significantly higher in patients with LNM compared to those without LNM (P=0.045). 4 of 10 patients (40.0%) with nerve invasion had tumor recurrence, and 4 cases (12.5%) had tumor recurrence in 32 patients without nerve invasion. There was a trend that incidence of recurrence in patients with nerve invasion was obviously higher than that of patients without nerve invasion, but with no statistically difference (P=0.075). The median of preoperative AFP level was 3.37 ng/ml (range from 1.18 to 8.25 ng/ml) in patients with tumor recurrence, tendentiously higher than 2.72 ng/ml (range from 0.947 to 5.92 ng/ml) in patients without recurrence, but no significant difference existed (P=0.0791). Log-rank test showed that the difference of recurrence-free survival (RFS) was statistically significant between the patients with and without LNM (P=0.0498) (Figure 2). It revealed that LNM was the risk factor of tumor recurrence in patients with SMGC. However, nerve invasion and preoperative AFP might be the risk factors of recurrence, but without sufficient evidence. (Table 3)
Cox regression analysis of the independent risk factors of tumor recurrence
According to results of univariate analysis, LNM, nerve invasion and preoperative AFP were regarded as independent variables, and tumor recurrence was deemed as the dependent variable. Survival analysis of Cox regression was adopted to verify the independent predictive risk factors of recurrence in patients with SMGC. The increase of preoperative AFP level was tendentiously positively associated with the risk of tumor recurrence of SMGC patients, but with no significant difference (P=0.081; 95%CI, 0.957~2.128). There was no significant difference of relationships between LNM or nerve invasion and risk of tumor recurrence. We found that preoperative AFP might be the independent risk factor of recurrence of SMGC patients, but need further validation. (Table 4)
Comparison of clinicopathological features between early and advanced SMGC
In order to assess the difference of clinicopathological features between early and advanced SMGC, all patients were divided into early SMGC group (n=20) and advanced SMGC group (n=37). 15 patients had follow-up outcomes in early SMGC group, among whom one patient appeared tumor recurrence and then died. In advanced SMGC group, there were 27 patients with follow-up outcomes, among those 7 patients appeared tumor recurrence and 6 cases died from it. The RFS and overall survival (OS) curves of early and advanced SMGC patients were showed in Figure 3. There were 11 and 9 patients underwent distal and total gastrectomies respectively in early SMGC group, while most of the patients (28 of 37 patients) received total gastrectomy in advanced SMGC group (P=0.018). In early SMGC group, 4 patients (20.0%) were <2 cm of tumor size, but no patient was <2 cm in advanced SMGC group (P=0.012). Compared to 12 patients (60.0%) with poorly-differentiated type in early SMGC group, most of patients (91.9%) in advanced SMGC group were the poorly-differentiated type (P=0.011). The occurrence rate of LNM was 15% (3/20) in early SMGC patients, significantly lower than that of 56.8% (21/37) in advanced SMGC patients (P=0.002). Ulcer existed in most of patients with SMGC. 16 cases (80.0%) appeared ulcer in early SMGC patients, and it occurred in all the patients with advanced SMGC (P=0.012). No nerve invasion appeared in early SMGC group, while there were 14 patients (37.8%) with nerve invasion in advanced SMGC group (P=0.001). The median of preoperative CEA level in early SMGC patients was 2.08 ng/ml (range from 0.848 to 4.7 ng/ml), which was remarkably lower than 2.75 ng/ml (range from 0.2 to 23.5 ng/ml) in advanced SMGC patients (P=0.0384) (Figure 4). (Table 5)