Gastric cancer (GC) is a significant global public health issue, especially in East Asian countries (1). In 2020, GC was ranked the fifth most frequently diagnosed cancer, and the third leading cause of cancer-related deaths in the word, with more than 1 million cases and over 768,000 deaths (2). GC has the highest incidence and mortality rate in China. Upon diagnosis, approximately 75% of cases were already in the advanced stage of GC, with the majority accompanied by lymph node metastasis (3). Lymph node metastasis is an important influencing factor in the prognosis and the selection of optimal treatment approach among patients with GC. Therefore, it is necessary to evaluate whether lymph node metastasis occurs in patients with GC before surgery (4). At present, the clinical diagnosis of preoperative GC with lymph node metastasis mostly relies on auxiliary examinations, such as upper endoscopy, 18-fluorodeoxyglucose (FDG) positron emission tomography integrated with computed tomography (PET-CT), and abdominal ultrasonography(AUS), which cannot be reliably utilized to confirm or exclude lymph node metastasis (5, 6). Furthermore, tissue biopsy cannot be routinely used in patients with GC due to its invasiveness and high cost. Thus, it is crucial to explore inexpensive and powerful predictors for lymph node metastasis in patients with GC (7).

Previous studies have focused on tumor markers, which are quantifiable circulating substances associated with malignant diseases to predict GC (8). Among several tumor markers, carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) are common tumor associated antigens (9, 10). However, previous studies regarding the association of CEA and CA19-9 with GC yield inconsistent findings (11, 12). Furthermore, only a few investigations have examined the diagnosis significance of these markers for lymph node metastasis in patients with GC.

Accumulating evidence have shown a critical role of the inflammatory microenvironment and nutritional status in the tumorigenesis and progression of GC (13). It has been found that the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), which are indicators of systemic inflammation, can affect tumor progression by inhibiting apoptosis, promoting angiogenesis, and damaging DNA (14). Moreover, the systemic immune-inflammation index (SII), a novel inflammatory marker derived from the counts of peripheral blood neutrophils, platelets, and lymphocytes, can serve as a promising indicator of various inflammatory and immune pathways within the body and exhibit greater stability in diagnosis compared to PLR and NLR (15). Several studies showed that SII can predict the prognosis of malignant tumors, including squamous cell carcinoma (16), lung cancer, and breast cancer (17). Furthermore, it has been found that malnourished patients affect immune cell metabolism, resulting in immune dysregulation and the development of cancer. Evidence have also shown that the prognostic nutrition index (PNI), a novel index of an individual’s nutritional status, plays an important role in predicting the metastasis of various malignant tumors (18, 19). PNI can be used as a prognostic factor for survival in GC patients receiving adjuvant chemotherapy (20). Besides, PNI variability was associated with survival outcomes in gastrointestinal cancers (21). Although PLR, NLR, and CEA have been reported to be associated with GC prognosis (9, 22), their clinical correlations with GC remain controversial. Furthermore, the diagnostic values for the combined effects of these indicators in GC are unclear.

Therefore, we conducted a retrospective study to evaluate the associations of CEA, CA125, CA19-9, CA724, NLR, PLR, SII, and PNI in patients with GC, and further examined their combination on the diagnostic value of lymph node metastasis in patients with GC.

Table 1 displays the characteristics of patients. Of the 395 patients aged 68.5 ± 9.1 years, 192(48.6%) were diagnosed with lymphatic metastasis, and 282 (71.4%) were males. The average tumor diameter was 4.2 ± 2.2 cm. The results show that lymph node metastasis was significantly correlated with age (P=0.017), tumor stage, TNM, degree of differentiation and tumor diameter (P<0.001).

The spearman correlation analysis showed that the correlation coefficient between SII, PNI, CEA and other clinical variables (e.g., age, tumor diameter, differentiation, etc.) was less than 0.3, and the variables were independent. There was significant collinearity between TNM stage and lymphatic metastasis (0.817)( Figure 1 ). Table 2 shows the univariate and multivariate logistic regression analysis of diagnostic factors in patients with lymph node metastasis. After adjusting for confounding factors, CEA (Odd ratio (OR):2.11; 95% confidence interval (CI): 1.17-3.81), SII (OR:1.02; 95%CI: 1.01-1.04), tumor diameter (OR:2.43; 95%CI: 1.23-4.79)were identified as significant risk factors that affect lymph node metastasis, while PNI (OR:0.90; 95%CI: 0.85-0.96) was associated with a lower risk for lymph node metastasis in GC.

The ROC analysis of CEA, PNI, and SII is shown in Figure 2 . The AUC of the PNI, SII, and CEA for lymph node metastasis was 0.679, 0.673, and 0.574, respectively. After applying the Bonferroni correction, the PNI, and SII demonstrated higher diagnostic accuracy compared to CEA in detecting lymph node metastasis (p<0.001). The combined diagnosis of lymph node metastasis in patients with GC using CEA, SII, PNI, differentiation, and tumor diameter yielded an AUC of 0.788, which was significantly higher than single diagnostic indicator. In male patients, the AUC for the joint diagnosis of CEA, SII, PNI, differentiation, and tumor diameter was 80.0%, while in female patients, the AUC was 78.5% ( Figure 3 ). The accuracy and comprehensive performance of comprehensive model in predicting lymphatic metastasis were better than that of single index ( Table 3 ). The ROC analysis of PLR, NLR, CA19-9, CA125, and CA724 is shown in Supplementary Figure S1 . The AUC of the PLR, NLR, CA19-9, CA125, and CA724 for lymph node metastasis was 0.644, 0.649, 0.549, 0.527, and 0.560, respectively.

Based on the results of multivariate logistic regression, CEA, SII, PNI, differentiation, and tumor diameter were included in the nomogram model ( Figure 4A ). DCA showed greater net benefits ( Figure 4B ). In addition, calibration curves also show good agreement between predictions and observations ( Figure 4C ).

Given the asymptomatic presentation in most patients with GC prior to advanced disease stages, early GC detection assumes paramount importance (27). The main manifestation of GC metastasis is the involvement of lymph node metastasis, which is closely related to the prognosis of patients with GC. In addition, accurate assessment of preoperative lymph node status is important for the choice of GC treatment options. Therefore, it is of great significance to explore accurate preoperative diagnosis of lymph node metastasis for the formulation of treatment plans and prognosis evaluation of patients with GC (28). After a series of analyses, we used CEA, SII and PNI to develop a clinical diagnostic model for lymph node metastasis in patients with GC.

The relationship between inflammatory makers and various tumors has become the current research hotspot. Systemic inflammatory responses have been significantly associated with poor prognosis in patients with cancers (16, 29, 30). Persistent inflammatory responses can stimulate tumor growth, invasion, and metastasis (31). Of the three inflammatory factors (SII, PLR, and NLR) in the present study, SII was the only independent prognostic factor for lymph node metastasis in patients with GC in the full-adjusted statistical models. Our results were consistent with the findings by Cao et al (32). Matsubara et al. also showed that SII was an independent prognostic factor in patients with endometrial cancer and could more precisely predict survival than PLR and NLR (33). In contrast to PLR and NLR, SII consists of three peripheral blood parameters (neutrophils, platelets, and lymphocytes) (34). It can comprehensively reflect the balance of host immunity and inflammation, and dynamically obtain information about host inflammation, immune response and clinical treatment (35). Platelets deliver adenosine triphosphate into the circulation and promote tumorigenesis (36). Lymphocytes control tumor growth by secreting cytokines such as interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α) which interact with each other (37). Neutrophil and platelets can stimulate cancer progression by producing vascular endothelial growth factor (VEGF) and interleukin-6 to enhance tumor angiogenesis and tumor cell epithelial-mesenchymal transition (38, 39).

Patients with cancers are at a high risk of malnutrition due to the proliferation of tumor cells and adverse effects of treatment (34, 40). Previous studies have found that PNI was a prognostic marker for various malignancies. In this study, PNI had a significant negative association with lymph node metastasis in patients with GC. The reason that PNI could be used as a prognostic factor for lymph node metastasis in GC lies in the fact that PNI was calculated from serum albumin concentration and peripheral blood lymphocyte count. Serum albumin may reflect the overall nutritional status of the body, and low serum albumin levels indicate poor nutrition in the host, which can negatively affect overall health, such as damage to the body’s cellular immunity, humoral immunity and phagocytic function and other defense mechanisms, increase inflammatory response, and promote tumor cell invasion function, thereby promoting the occurrence and progression of tumors (41, 42).

Previous research suggest that CEA, CA19-9, CA724, and CA125 are classical tumor markers for GC (43). In the present study, the multivariate analysis showed that CEA with a cut-off of 5ng/ml was an independent risk factor for lymph node metastasis in patients with GC, which could be used for early screening. Nevertheless, CA19-9, CA724, and CA125 were not significantly associated with lymph node metastasis in GC. CEA, a carcino-embryonic antigen, has been shown to play a role in programmed cell death and cell adhesion. Our results are similar with a previous study by Feng et al., who found that among several tumor markers including CEA, CA19-9, and CA125, elevation of CEA level was an independent prognostic factor for the poor prognosis of early GC (9). In addition, the combination of these tumor markers had a relatively low diagnostic value. Therefore, a more sensitive prediction model is warranted to be constructed in further studies.

In the present study, we found that the combination of CEA, SII, PNI, differentiation, and tumor diameter increased the AUC under the ROC to 0.788, which means that the combined index could improve the diagnosis of lymph node metastasis in patients with GC than single marker. To the best of our knowledge, this is the first study integrating tumor markers, inflammatory and nutritional indicators to diagnose lymphatic metastasis in GC. In subgroup analyses, the diagnostic sensitivity in male (AUC=0.800) was significantly higher than females (AUC=0.785). This result is consistent with the findings of a previous study showing higher diagnostic significance of systematic markers of inflammation in male than female GC patients (14). This might be explained by a more intense response immune system to lymphatic metastasis of GC cells in males than females, resulting in an increase in the proportion of neutrophils and platelets, and CEA in patients with GC is closely related to lymphatic metastasis (5). Therefore, it is warranted to include the gender factors in the early diagnosis of GC with lymph node metastasis to improve the application value of combined diagnostic indicators.

There are some limitations in this study. Firstly, blood indicators might have been affected by infection and blood circulation capacity. Secondly, we used a single-center retrospective data to build the predictive model, which may be subject to selection bias, Thus, it is necessary to conduct a large-scale prospective study with long-term follow-up from multi-centers to validate our findings. Lastly, the blood neutrophil, lymphocyte, platelet and monocyte was collected at a single time point when admission. Thus, future research should assess the dynamic changes of blood samples in multiple time points.

Our study demonstrated that CEA, SII, PNI, differentiation, and tumor diameter were significantly associated with lymph node metastasis in patients with GC, indicating these indices can be used for monitoring of GC patients. In addition, the established nomogram by incorporating CEA, SII, PNI, differentiation, and tumor diameter can effectively predict lymph node metastasis in GC. Therefore, the combined screening of tumor markers and blood routine indicators are recommended in clinical practice to assist in the early diagnosis of lymph node metastasis in patients with GC. Our results may provide a new biological target for the clinical diagnosis and facilitate the process of treatment of GC.