3.1 Baseline Characteristics
There were 110 RGC patients included in this study. The data of these patients were compared to determine whether the sixth or the seventh version of the AJCC gastric cancer staging system is more suitable for RGC and analysis of prognostic factors.
Among them, 49 patients suffered from benign diseases, and 61 patients were diagnosed with malignant tumors. The average age of the patients was 59 years, and the male-to-female ratio was 4:1. The average interval time from initial gastric resection to RGC was 13.75 years, and the interval time for the occurrence of RGC in the two groups with different initial nature of the disease was dissimilar. The average time of RGC occurrence was 22 years in the benign disease group after initial gastric resection, and seven years in the malignant tumor group after initial gastric resection. The average size of tumors in RGC patients was 3 cm. According to pathological examinations, 23 patients were diagnosed with lymph node metastasis, 41 patients with tumor infiltration into the serosal layer, 43 patients with distant metastasis, 27 patients with peritoneal metastasis only, three patients with liver metastasis only, and 13 patients with peritoneal and liver metastasis. The peritoneal metastasis rate was relatively high and represented the most common metastatic site in the patients included in this study.
The time of RGC occurrence was significantly longer for patients who underwent subtotal gastrectomy for benign diseases than for those who underwent distal gastrectomy for malignant tumors (P = 0.01). Patients who underwent subtotal gastrectomy for benign diseases were significantly older than those in the malignant group (P = 0.01). Digestive reconstruction methods after initial gastric resection included Billroth-I and Billroth-II anastomosis in 33 and 70 cases, respectively, and seven cases underwent gastroesophageal anastomosis (GEA). We found 47 cases of RGC at the anastomotic site and 63 cases of RGC at non-anastomotic sites. We also found that 32 RGC patients had organ infiltration, and 15 patients among them underwent combined resection of the surrounding infiltrated organs during surgery. Following radical resection for RGC, some patients experienced postoperative complications, including intestinal obstruction in two cases, anastomotic bleeding in five cases, and postoperative abdominal abscess in five cases (Table 1).
3.2 Stage Changes
Due to significant differences between the 6th and 7th editions of tumor staging in T and N staging, patients classified as belonging to Stage IV in the 6th edition were reclassified in the 7th edition as belonging to Stages IIB, IIIA, IIIB, or IIIC. All other changes in staging resulted in an upward shift. Patients classified as T1N1M0 or T2bN0M0 Stage IB in the 6th edition were classified as Stage IIA in the 7th edition. Patients classified as T2bN1M0 Stage II in the 6th edition were classified as Stage IIIA in the 7th edition. Patients classified as T2bN2M0, T3N1M0, or T4N0M0 Stage IIIA in the 6th edition were classified as Stage IIIB in the 7th edition. Patients classified as T3N2M0 Stage IIIB in the 6th edition were classified as Stage IIIC in the 7th edition. Among the 110 patients included, 37 cases were downgraded in the 7th edition, relative to the 6th edition, and six cases were upgraded in the 7th edition. There were 67 cases whose staging remained unchanged; the shift rate was found to be 39% (Table 2).
3.3 Survival analysis
The three-year survival rates for AJCC sixth edition T staging of T1, T2a, T2b, T3, and T4 were 93%, 81%, 76%, 60%, and 20%, respectively, and the differences in the survival rates among groups were statistically significant (P = 0.01; Figure 1A). The three-year survival rates for AJCC 7th edition T staging of T1, T2, T3, T4a, and T4b were 93%, 81%, 76%, 60%, and 20%, respectively, and the differences in the survival rates among groups were statistically significant (P = 0.001; Figure 1B). We also found that the C-index of AJCC 6th and 7th edition T staging was 0.70. However, the C-index of the two editions was not significantly different (P = 0.12), as determined by the coxph function in the R package "survival" (Table 4).
Our results showed that the three-year survival rates for AJCC 6th edition N staging of N0, N1, N2, and N3 were 83%, 78%, 45%, and 8%, respectively, and the prognostic differences were statistically significant (P = 0.0001, Figure 1C). The three-year survival rates for AJCC 7th edition N staging of N0, N1, N2, N3a, and N3b were 83%, 70%, 59%, 55%, and 8%, respectively, and the prognostic differences were statistically significant (P = 0.012, Figure 1D). Moreover, the C-index of AJCC 6th edition N staging was 0.635, and the C-index of AJCC 7th edition N staging was 0.645, the difference was not significant between the two editions (P = 0.23) (Table 4).
The one-year survival rates for AJCC 6th edition stages I, II, III, and IV were 95%, 93%, 92%, and 83%, respectively, and the three-year survival rates were 91%, 88%, 60%, and 50%, respectively (Figure 1E). The one-year survival rates for AJCC 7th edition stages I, II, III, and IV were 92%, 90%, 88%, and 80%, respectively, and the three-year survival rates were 92%, 80%, 57%, and 50%, respectively (Figure 1F). The C-index of AJCC 6th edition overall staging was 0.76, and the C-index of AJCC 7th edition overall staging was 0.732; the difference between the C-indices was not significant (P = 0.09) (Table 4).
Among the 110 RGC patients, 91 patients died; all deaths occurred due to tumor recurrence or distant metastasis. The overall survival rates of the RGC patients at one, three, and five years were 71%, 45%, and 31%, respectively. Among them, 60 patients underwent radical resection, 35 underwent exploratory laparotomy, 11 underwent palliative resection, and four did not undergo surgery. Our results showed that the survival rate of patients with distant metastasis was significantly lower than the survival rate of patients without metastasis. Specifically, their one-, three-, and five-year survival rates were 60%, 39%, and 23%, respectively, which was lower than the survival rates (81%, 55%, and 40%, respectively) of patients without metastasis. Following radical resection (three-year survival rate of 60%), patients showed a significant survival advantage over those who underwent exploratory laparotomy (three-year survival rate of 40%) or palliative surgery (three-year survival rate of 51%).
3.4 Prognostic Analysis and Establishment of a Predictive Model
The univariate analysis of the survival data of the 110 patients in this study included the gender and age of the patients, the nature of the first disease, the interval between the first surgery and RGC, the method of digestive tract reconstruction after the first surgery, whether chemotherapy was administered after the first gastric cancer surgery, the location of RGC, direct infiltration of peripheral organs, the surgical approach to RGC, R0 resection for RGC, combined organ resection for RGC, Borrmann type of RGC, histological grading of RGC, tumor diameter of RGC, distant metastasis of RGC, whether chemotherapy was administered after RGC surgery, anastomotic leakage after RGC surgery, incision infection, intestinal obstruction, anastomotic bleeding, and abdominal abscess. Among these risk factors, the direct infiltration of peripheral organs by RGC, Borrmann type of RGC, peritoneal metastasis or liver metastasis of RGC, and anastomotic bleeding after surgery for RGC were correlated with the prognosis of RGC patients. The results of the multivariate analysis of these risk factors showed that they were all independent risk factors related to the prognosis of RGC patients (Table 3). We used these independent risk factors to construct a nomogram (Figure 2A), which had a total score of 300 and a C-index of 0.783. We used the Bootstrap method to evaluate the internal validity of the newly constructed column chart. We constructed a calibration plot by correcting 1,000 repeated Bootstrap samples in the dataset. In the plot, the horizontal axis represented the three-year survival probability predicted by 1,000 repeated Bootstrap samples, and the vertical axis represented the actual three-year survival rate of patients obtained through the Cox proportional hazard analysis. A dotted diagonal line marked the ideal reference line, indicating the site where the predicted probability exactly matched the observed probability. The two points represented the predicted and observed survival probabilities, respectively, for each of the two groups undergoing internal validation. Since the intersection line coincided with the dotted reference line, it indicated that the predicted overall survival rate was not significantly different from the actual observed survival rate of the patients. The survival prediction model constructed in this study for RGC patients performed well in predicting the three-year patient survival (Figure 2B).