Gastric cancer is among the most common malignancies worldwide, ranking fifth in both incidence and mortality rate (1). Gastric signet ring cell carcinoma (GSRCC) is a distinct histological subtype, accounting for approximately 30% of all gastric cancer cases. Predominantly affecting younger individuals and females, GSRCC is characterized by unique biological features, including aggressive progression, frequent metastasis, and resistance to conventional therapies. As a result, GSRCC is associated with a significantly worse prognosis, with reported 5-year survival rates ranging from 18.4% to 44.1% (2, 3).

Inflammation has been recognized as the 7th hallmark of cancer and plays significant roles in tumor progression and metastasis. Tumor cells can trigger systemic inflammatory responses, which in turn promote cancer cell proliferation, invasion, and dissemination, while also facilitating angiogenesis (4–7). In recent years, peripheral blood inflammatory biomarkers have been widely utilized to predict prognosis across diverse cancers. Among these, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), and lymphocyte-to-C-reactive protein ratio (LCR) have been validated as potential prognostic indicators in gastric cancer (8–11). However, relying on a single inflammatory marker may be inadequate to fully capture the complex and multifactorial prognostic landscape of patients with GSRCC.

In addition to inflammation, nutritional status plays a critical role in determining cancer prognosis. Malnutrition not only contributes to weight loss but also increases risk of infection, mortality and systemic inflammation, thereby accelerating tumor progression (12–14). Previous studies have demonstrated that nutritional prognostic scores possess strong predictive value in a variety of solid tumors, including gastric, esophageal, pancreatic, lung, and renal cancers (15–19). Given that single factor often fail to capture the dynamic interplay between inflammation and nutrition, integrated scoring systems combining both parameters have been developed. For example, the modified Glasgow Prognostic Score (mGPS) incorporates serum albumin and C-reactive protein (CRP) levels, while Naples Prognostic Score (NPS) is derived from serum albumin, total cholesterol, LMR, and NLR (20, 21). These combined indices enhance prognostic accuracy and offer more reliable tools for guiding clinical decision-making.

However, unlike conventional gastric adenocarcinoma, GSRCC predominantly affects younger patients, whose nutritional status is often underestimated or overlooked in clinical practice. Despite the well-recognized interplay among inflammation, nutrition and prognosis, no validated scoring system currently exists that specifically integrates these factors for patients with GSRCC. In this study, we aimed to develop and validate a novel integrated inflammatory-nutritional prognostic scoring system to improve the accuracy of survival prediction in GSRCC.

GSRCC is a distinct histological subtype of gastric cancer, characterized by increasing global incidence, aggressive biological behavior, and poor clinical outcomes (27). Growing evidence underscores the crucial role of systemic inflammation and nutritional status in cancer initiation, progression, and prognosis (4, 7). In this study, we conducted a retrospective study to develop and validate a prognostic model for GSRCC that integrates both inflammatory and nutritional biomarkers. Our findings revealed that the inflammatory marker LCR and the nutritional marker GNRI were both significantly associated with OS in GSRCC patients. Multivariate Cox regression analysis further identified advanced age (≥60 years), upper tumor location, T4 staging, N2–N3 lymph node involvement, low LCR (< 8,785.7) and low GNRI (< 112.9) as independent predictors of poor OS. Based on these variables, we constructed a nomogram that demonstrated robust prognostic performance, with a C-index of 0.831. Calibration curves indicated strong agreement between predicted and actual 1-, 3-, and 5-year OS, and DCA confirmed a substantial net clinical benefit. Notably, for the variable selection process using LASSO regression, we compared two commonly used penalty parameters: λ.min, which minimizes the cross-validation error, and λ.1se, which selects the model within one standard error of the minimum cross-validation error. We compared the models with λ.min and λ.1se, and found that the λ.min model demonstrated superior predictive accuracy, as evidenced by a higher concordance index and improved AUC (Supplementary Figure S6). Although λ.1se could have been used to select a simpler model with fewer variables, we found that it resulted in a higher cross-validation error and reduced predictive performance. Therefore, we believe that λ.min was the most appropriate choice for this study to ensure a robust and accurate prognostic model. Taken together, these results support the clinical utility of our model as a statistically sound and practically meaningful prognostic tool. It may facilitate individualized risk stratification and assist clinicians in formulating more tailored management strategies for patients with GSRCC.

Our study first integrates inflammatory and nutritional markers for prognostic prediction in GSRCC patients. Conventional prognostic models predominantly rely on clinicopathological parameters such as TNM stage and tumor size (28, 29). In contrast, our study incorporates LCR and GNRI, providing a more comprehensive and biologically informed approach to risk stratification. LCR, as a novel inflammatory marker, reflects the balance between host immune competence and systemic inflammation by combining lymphocyte count and CRP levels (8). Lymphocytes are essential mediators of antitumor immunity, limiting tumor progression through cytokine production and cytotoxic killing (30). Accordingly, tumor-infiltrating lymphocytes reflect the strength of the local cytotoxic immune response (31). In contrast, lymphopenia indicates impaired host immune competence and is associated with worse oncological outcomes (32). CRP, an acute-phase reactant, is frequently elevated in the presence of tumor-promoting inflammation (33). Cells activated by monomeric CRP can stimulate intracellular signaling pathways, including the activation of the NF-kB transcription factor (34). CRP directly binds fibronectin stimulating the retention of monocytes in the TIME (tumor immune microenvironment) (35). Moreover, CRP can enhance cytotoxic T lymphocyte-mediated cell lysis and, on the other hand, can promote a sustained pro-inflammatory and pro-tumor immune microenvironment (34). Tumor progression and metastasis result from interactions between tumor cells and TIME. Therefore, LCR may reflect the antitumor cytotoxicity of the intertumoral immune response in TIME and affect tumor. Regarding gastric cancer, a particular focus has been placed on the LCR. Several studies have shown that LCR outperforms other inflammation-based scores, such as CAR, NLR, LMR, and PLR, in predicting overall survival in gastric cancer (36, 37). Notably, a multicenter study comparing the predictive performance of 16 systemic inflammatory indices identified LCR as the strongest marker of systemic inflammatory burden (38). In our study, patients with LCR < 8,785.7 had significantly worse overall survival, likely due to the combined impact of systemic inflammation and impaired immune surveillance.

GNRI, derived from serum albumin and BMI, reflects nutritional reserve and chronic catabolic status. Malnutrition can impair immune function, reduce tolerance to anticancer treatment, and potentially shape an unfavorable tumor microenvironment (39). Recent studies have consistently demonstrated its prognostic value in various malignancies (15–19). Malnutrition not only reduces treatment tolerance but may also facilitate tumor progression by weakening host immune defenses (40). In line with previous findings (16), we observed that a GNRI < 112.9 was significantly associated with poorer survival in GSRCC patients. Notably, GNRI was also associated with several clinicopathological features in our cohort, including age, tumor size, pT stage, and tumor location, suggesting that nutritional compromise may partially reflect baseline disease burden and anatomic factors (Supplementary Table S1), consistent with previous reports (41, 42). Importantly, GNRI showed no significant association with LCR, indicating that GNRI captures a nutritional dimension that is not redundant with systemic inflammation and therefore provides complementary prognostic information. The integration of LCR and GNRI highlights their potential to reflect distinct yet complementary biological processes influencing cancer progression. Inflammation may promote tumor aggressiveness through mechanisms such as NF-κB pathway activation, while malnutrition may compromise antitumor immunity (16). The combined use of inflammation- and nutrition-based markers strengthens the biological rationale of our model and may more accurately reflect the overall health status and outcome risk in patients with GSRCC. These findings provide a foundation for future research and hold promise for broader clinical application.

Using X-tile analysis, we identified optimal cutoffs of 8,785.7 for LCR and 112.9 for GNRI. Published studies in gastrointestinal cancers generally report an LCR cutoff between 6,300 and 12,000 (43–45), and our threshold lies within this interval, supporting the robustness of our findings. Okugawa et al. (43) evaluated 551 patients with gastric cancer undergoing curative surgery and applied an LCR cutoff of 8,350, which is close to our value. The variability in reported LCR cutoffs likely reflects differences in patient case mix, perioperative management, treatment strategies, and sample size. For GNRI, our optimal cutoff of 112.9 is higher than the thresholds of 104 and 98 reported by Tanabe et al. (47) and Ide et al. (46), respectively. This difference may relate to the distinctive characteristics of our gastric signet-ring cell carcinoma cohort, abbreviated as GSRCC. Nearly 60% of our patients were younger than 65 years, and the cohort also had relatively high serum albumin levels, with a mean of 44 g/L, which may reflect the benefits of nutritional management and contribute to higher GNRI values. These cutoffs may therefore provide more sensitive prognostic stratification in GSRCC. Nevertheless, they should be validated in larger independent cohorts, and future studies should evaluate their generalizability across gastric cancer subtypes and diverse populations.

Age was reaffirmed in this study as a critical prognostic factor in GSRCC. Our results demonstrated that patients aged ≥60 years had significantly worse survival outcomes, consistent with findings from previous studies (48). Older individuals often present with comorbidities such as cardiovascular disease or diabetes, which may exacerbate disease burden, reduce tolerance to anti-cancer therapies, and adversely affect overall prognosis (49). Moreover, age-related immune senescence may impair antitumor immune surveillance, thereby facilitating tumor progression (50). Although GSRCC tends to occur more frequently in younger individuals, our data underscore that advanced age remains a non-negligible determinant of poor prognosis in this patient population. Another noteworthy finding is the prognostic significance of tumor location. Patients with tumors located in the upper stomach exhibited significantly worse survival compared to those with tumors in other gastric regions. This disparity may be attributed to distinct anatomical and biological characteristics associated with upper gastric tumors (51). Tumors in this region often involve the gastroesophageal junction (GEJ) and lie in close proximity to the diaphragm, making them more prone to mediastinal lymph node metastasis (42). Notably, we also observed an association between nutritional status and tumor location: GNRI categories differed significantly across tumor locations (Supplementary Table S1, p = 0.026), with a higher proportion of upper and entire-stomach tumors in the low-GNRI group, supporting the possibility of greater nutritional compromise with proximal/GEJ involvement. These findings suggest that patients with upper gastric or GEJ involvement may benefit from more intensive preoperative evaluation and tailored multimodal treatment strategies to improve survival outcomes. Although “entire stomach” tumor location showed borderline significance in multivariable analysis, our subgroup analysis suggested potential heterogeneity by sex (Supplementary Figure S3). However, the number of patients with “entire stomach” involvement was very limited, leading to imprecise estimates and occasional non-estimable effects in some strata. Therefore, we did not include this variable in the final nomogram to reduce the risk of overfitting and to maintain model stability. Further validation in larger, multicenter cohorts is warranted to clarify its prognostic value.

Our nomogram demonstrated superior discriminative performance, with a C-index of 0.831, substantially higher than those reported in previous prognostic models for GSRCC. For example, Shao et al. (52) reported a C-index of 0.76 in a SEER-based model, while Wang et al. (25) achieved 0.78 in a multicenter cohort. Additionally, DCA confirmed a meaningful net clinical benefit across multiple time points, with particularly strong utility in predicting 5-year OS. The combination of robust discriminatory ability and practical utility positions our model as a promising tool for prognostic evaluation in GSRCC (53). Despite these strengths, several limitations should be acknowledged. First, the retrospective nature of this study may introduce selection bias. Although strict inclusion and exclusion criteria were employed to mitigate this risk, the inherent constraints of retrospective data remain. The training and validation cohorts were separated by surgical date, using January 2016 as a temporal cutoff, to perform temporal validation. This design evaluates whether a model developed from earlier patients remains applicable to patients treated in a later period. However, advances in treatment over time may introduce a cohort effect, as improvements in surgical techniques, adjuvant therapies, and postoperative care could contribute to better outcomes in the later cohort. Second, the study was conducted at a single center, which may affect the generalizability of the findings to broader populations. Although our cohort shares similar characteristics, such as age, gender, and tumor location, with those reported in multicenter or population-based studies, these results should be interpreted with caution (54–56). To enhance the external validity of our model, we propose that future studies be conducted across multiple centers, incorporating diverse patient populations from different geographic locations. Third, although receipt of adjuvant chemotherapy was recorded, detailed treatment exposure (e.g., cycles, dose intensity, and treatment modifications) was not uniformly available, and adjuvant chemotherapy was not retained in the final prognostic model after LASSO selection. Previous studies have suggested that reductions in inflammation during therapy may be associated with improved survival outcomes (57). Meanwhile, as neoadjuvant chemotherapy–associated lymphopenia may alter the LCR_GNRI score, thereby identifying a high-risk subgroup, it seems prudent to include neoadjuvant therapy. Therefore, future studies should incorporate treatment-related factors to provide a more comprehensive assessment of prognosis in patients with GSRCC, as these variables may modify the associations between inflammation, nutrition and survival. Fourth, this study was not designed to explore underlying biological mechanisms. Although our findings support the prognostic value of inflammation–nutrition markers, the mechanistic links between systemic inflammation, nutritional status, and tumor behavior in GSRCC remain to be clarified. Future basic research should investigate immune-cell modulation and cytokine signaling in the tumor microenvironment, as well as how nutritional deficits influence tumor progression, metastasis, and treatment response. Additionally, this study did not evaluate the prognostic significance of dynamic changes in inflammatory and nutritional markers over time. Serial monitoring of these markers may offer incremental prognostic value and should be investigated in future prospective studies to clarify how they vary with disease progression and treatment response. Future prospective, multicenter studies are warranted to validate the accuracy and applicability of this model across diverse populations.

Refinement of the model may offer new avenues for individualized treatment strategies in patients with GSRCC. Based on our findings, we propose that patients with LCR_GNRI score of 0 or 1 should be considered for more aggressive treatment approaches, including neoadjuvant chemotherapy and closer postoperative monitoring. These patients have a higher risk of poor survival outcomes, and personalized management strategies could improve their prognosis. On the other hand, patients classified as low-risk (LCR_GNRI scores of 2) could be managed with standard clinical protocols and less intensive monitoring, in line with current clinical guidelines. This risk-based approach would allow for more tailored treatment, improving the overall management of GSRCC patients.

Our research integrated inflammation and nutritional indicators to develop and validate a GSRCC prognostic nomogram, outperforming existing models in discrimination and clinical utility. It offers a novel tool for precise prognostic evaluation and personalized treatment, particularly for identifying and managing high-risk GSRCC patients.