3.1 Patient characteristics
The flowchart for patient screening was displayed in Fig 1. In total, 568 patients fit the inclusion criteria. No chemotherapy or radiotherapy was administered to any of the patients prior to surgery, and there was no perioperative mortality included. This study included 442 men and 126 women. The average age of the population group was 60.29 ± 9.79 (25–87). The average BMI ratio prior surgery for all patients was 22.20 ± 3.37. 31.7% (n = 180) of the patients were operated with open,40.0% (n = 227) with laparoscopic approach,and 28.3% (n = 161) with robot-assisted. On the basis of the AJCC staging standards, 119 (21.0%) patients were categorized as stage I, stage II patients accounted for 178 (31.3%), and stage III patients made up 271 (47.7%). No patient was disregarded in the follow-up process. A median follow-up time of 45 months was established for all patients, ranging from 12 to 61 months.
The clinical characterization of the study population was shown in Table 1, along with a comparison of the characteristics and clinical aspects of the two group of patients who had no complications (no) and/or experienced minor complications and those who had major complications. The description of the features and clinicopathological comparison between the group of who did not experience recurrence or metastasis and the group of patients who did recurrence or metastasis were displayed in Table 5.
3.2 Postoperative complications
A total of 89 (15.7%). patients in our statistics suffered serious complications. The occurrence of early postoperative complications in individuals experiencing radical gastrectomy was showed in Table2. Complications included duration of enteral nutrition was longer than 2 weeks in 26 patients, infection-related complications (incision infection, abdominal infection, pulmonary infection) in 234 patients, anastomotic fistula in 6 patients, pyloric or intestinal obstruction in 14 patient, thrombosis or embolism in 15 patients, and seven patients developed postoperative shock, they were all rescued after treatment.
In accordance with Table 1, age (p = 0.046), BMI (p = 0.003), tumour size (<3/≥3cm/) (p = 0.014), resection range (p = 0.019), perioperative transfusion (p < 0.001), and hospital stay (p < 0.001) were statistical significance between the two groups. For laboratory parameters, lymphocytes (p < 0.001), neutrophils (p < 0.001), platelets (p = 0.013), monocytes (p = 0.032), albumin (p < 0.001), fibrinogen (p < 0.001), CEA (p = 0.011), SIRI (p < 0.001) and AFR (p < 0.001) also differed significantly between groups.
3.3 Correlations between SIRI, AFR and the clinicopathological characteristics of gastric cancer.
In accordance with the results in Table 3, Preoperative SIRI level was related to the sex (p = 0.002), resection range (p = 0.008) among patients of gastric cancer. AFR had an association with the degree of tumor differentiation (p = 0.002) and duration of enteral nutrition (p = 0.01). Meanwhile, both preoperative conditions were related to age, tumour size (<3/≥3cm), TNM stage, perioperative transfusion, CA199, CEA, amount of bleeding, relapse or metastasis (p < 0.05).Upon further analysis, among patients under 60 years of age, SIRI levels were lower and AFR levels were higher (SIRI, p = 0.038; AFR, p < 0.001), and SIRI levels were higher and AFR levels were lower in individuals with a maximum tumor diameter >3 cm (SIRI, p < 0.001; AFR, p < 0.001). Furthermore, the level of SIRI in stage III was the highest of the clinical stages, the level of AFR in stage III was the lowest of the clinical stages (SIRI, p < 0.001; AFR, p < 0.001). For the perioperative blood transfusion patients, the level of SIRI was higher and the level of AFR was lower (SIRI, p < 0.001; AFR, p < 0.001). Group of CA199 and CEA positive patients, SIRI levels were higher and AFR levels were lower (SIRI, p = 0.023, p < 0.001; AFR, p = 0.001 p < 0.001). The highest SIRI levels and lowest AFR levels were observed in the group with intraoperative blood loss >400 ml (SIRI, p < 0.001; AFR, p < 0.001). The SIRI level of patients with gastric cancer with relapse or metastasis was noticeably raised (p < 0.001), and the AFR level was prominently reduced (p < 0.001).
3.4 Significance of preoperative SIRI and AFR levels for early serious postoperative complications in respectable gastric cancer
Table 4 listed the outcomes of the Univariate regression analysis that was executed to establish the OR values for the complication estimation. The result suggested that high preoperative SIRI was substantially related with early serious postoperative complications (P < 0.001; HR 1.429; 95% CI 1.175-1.738). Meanwhile, elevated preoperative AFR levels was a protective factor against postoperative complications (P < 0.001; HR 0.729; 95% CI 0.665-0.799;). Additionally, age, BMI, tumor size (<3/≥3 cm), resection range, perioperative transfusion and CEA (<5/≥5 ng/mL) were other noteworthy variables revealed by univariate analysis (P < 0.05). Regards to multivariable analyses, preoperative SIRI and AFR remained an independent influencing indicator for postoperative complications. (SIRI: P = 0.02; HR 1.222; 95% CI 1.031-1.446; AFR: P < 0.001; HR 0.771; 95% CI 0.701-0.848). Furthermore, resection range (P=0.044; HR 1.682; 95% CI 1.015-2.787) and perioperative transfusion (P = 0.008; HR 2.028; 95% CI 1.202-3.422) were other contributing factors.
3.5 Evaluation of predictive abilities for SIRI and AFR.
Since the previous statistical findings concluded that high levels of AFR are a protective parameter for postoperative complications, in order to facilitate the calculation of the predictive power of SIRI combined with AFR, we take fibrinogen to albumin ratio (the reciprocal of AFR) as the calculation amount. ROC curve generation and AUC calculation were used to determine the prediction capabilities of SIRI and AFR. The AUC values of SIRI, AFR, and SIRI combine AFR levels were summarized in Fig 2. The AUC values computed for the SIRI: AUC 0:765; 95% CI 0.714‐0.815), the AFR: AUC 0:743; 95%CI 0.689‐0.797, the SIRI-AFR: AUC 0:779; 95%CI 0.737‐0.820.
3.6 Establishment of the SIRI-AFR Score.
Based on the appropriate cut-off values for each determinant which were established using receiver operating characteristic curves with Youden's index, patients were grouped accordingly. Corresponding to the SIRI-AFR system, 219(38.6%), 224 (39.4%), and 125 (22.0%) patients had scores of 0, 1, and 2, accordingly.
3.7 The correlation between clinicopathological and relapse or metastasis.
The relationship between relapse or metastasis and clinicopathological factors was exhibited in Table 5. Recurrence or metastasis within 3 years in patients receiving radical resection of gastric cancer was associated with these factors: age (p = 0.019), tumour location (p = 0.012), tumour size (<3/≥3cm/) (p < 0.001), differentiated degree (p = 0.033) ,Her-2 (p = 0.042), TNM stage (p < 0.001), perioperative transfusion (p < 0.001), operation time (p = 0.001) , lymph node metastasis rate (p < 0.001), lymphocytes (p < 0.001), neutrophils (p < 0.001), platelets (p = 0.002), monocytes (p < 0.001), albumin (p < 0.001), fibrinogen (p < 0.001), SIRI (p < 0.001), AFR (p < 0.001), SIRI-AFR score (p < 0.001), CA199 (p =0.001), CEA (p < 0.001), postoperative complication (p < 0.001) and postoperative adjuvant chemotherapy (p =0.012).
3.8 Univariate and multivariate Cox regression analysis for relapse or metastasis.
Among patients of stomach carcinoma, univariate analyses identified that the greater risk of relapse or metastasis was profoundly associated with correlated with older age (p = 0.009), lower tumour location (p = 0.006), large tumor size (p < 0.001, later clinical stage (p < 0.001),longer operating time (p = 0.002), perioperative transfusion (p < 0.001), positive CA199 (p < 0.001), positive CEA (p < 0.001), major postoperative complication (p < 0.001),no postoperative adjuvant chemotherapy was performed (p = 0.006), high SIRI-AFR score (p < 0.001). Multivariate analysis revealed that TNM stage (p = 0.002; HR 5.100, 95% CI 1.847-14.086), operation time (p = 0.029; HR 1.003, 95.0% CI 1.000-1.005), perioperative transfusion (p = 0.009; HR 1.660, 95.0% CI 1.135-2.428), positive CEA (p = 0.025; HR 1.528; 95% CI 1.054-2.213), postoperative adjuvant chemotherapy (p = 0.008; HR 0.475, 95% CI 0.273-0.826), SIRI-AFR score (p < 0.001; HR 4.363, 95% CI 2.170-9.037) were the independently determined prognostic variables for relapse or metastasis (Table 6). Further, as presented in Fig 3, we observed that the SIRI-AFR score could effectively differentiate patients into three distinguishing risk groups for recurrence or metastasis.