Development and external validation of a prognostic nomogram for cancer-specific survival in elderly patients with early-stage gastric cancer

Introduction

According to the Cancer Statistics of 2020, gastric cancer (GC) is a disease with a high mortality rate worldwide, accounting for 768,793 new deaths (7.7% of all cancer-related deaths) in 2020 (1). Eastern Asia, especially Japan and Mongolia, has the highest incidence rates of GC by continent (1). Owing to the insidious onset of GC and its high risk of metastasis, early detection and early surgical resection are recommended as the standard approaches for improving clinical outcomes (2).

Elderly individuals, especially those over 60 years of age, are at increased risk of developing GC (3). With the widespread application of artificial intelligence (AI) technology, the sensitivity and accuracy of early-stage gastric cancer (EGC) detection have improved significantly, as demonstrated by the Integrative Multi-Regional Convolutional Neural Network (IMR-CNN) model proposed by Zhang et al. (4,5). However, despite the improved survival outcomes associated with EGC, elderly patients constitute a particularly heterogeneous population, making it clinically challenging to identify individuals at increased risk of adverse cancer-specific outcomes (6).

For postoperative management, the clinical prognosis of GC patients is widely recognized to be stratified according to the traditional tumor-node-metastasis (TNM) staging system (7), with N staging playing a decisive role in determining the subsequent adjuvant therapy for EGC (8,9). However, given the potential biases and limitations of conventional N staging, additional lymph node (LN) assessment approaches, such as the log odds of positive lymph nodes (LODDS), have been proposed to complement existing staging systems (10,11). Building on this concept, previous studies suggested that LODDS provides different prognostic information, particularly in patients with EGC, by partially addressing the inherent limitations of N staging (12-14).

In elderly patients, postoperative management of EGC is often complicated by heterogeneity in cancer-specific prognosis, even within the same TNM stage (15). Inaccurate risk estimation may lead to overtreatment or insufficient surveillance, potentially affecting clinical outcomes (16). Therefore, this study aimed to develop and externally validate an oncology-focused prognostic model to stratify the risk of cancer-specific survival (CSS) in elderly patients with EGC. We present this article in accordance with the TRIPOD reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-aw-815/rc).

Methods

Ethical approval

The data were obtained from the Surveillance, Epidemiology, and End Results (SEER) database (Mar. 2024 sub). This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Additionally, the data used for external validation were from the Department of Gastrointestinal Surgery, Zhongshan Hospital (The Affiliated Hospital of Fudan University). This study was approved by the Ethics Committee of Zhongshan Hospital (No. B2023-196), and all patients provided signed informed consent.

Study population

A total of 95,222 elderly patients over 60 years of age who were diagnosed with EGC between 2000 and 2020 in the SEER database were examined in this study. The exclusion criteria were as follows: (I) not the first primary maligancy; (II) the cancer-related surgery was not performed; (III) death for unknown reasons; (IV) non-T1 [American Joint Committee on Cancer (AJCC) 8^th] stage; (V) non-M0 (AJCC 8^th) stage; (VI) the follow-up data were not complete; and (VII) the number of detected LNs was unknown. Ultimately, 1,898 patients were eligible for the subsequent model construction. We enrolled 174 patients for external validation. The overall workflow is illustrated in Figure 1.

Figure 1 The flowchart of the patient filtering. SEER, Surveillance, Epidemiology, and End Results; T, tumor.

Study variables and outcomes

The 17 variables extracted from the SEER database were as follows: age, race, sex, marital status, primary site, differentiation, histologic type, tumor (T) stage, node (N) stage, radiotherapy, chemotherapy, tumor size, LODDS score, median household income, time of diagnosis to treatment, and the number of examined LNs and positive LNs. In the SEER database, chemotherapy was recorded as a binary variable, and information regarding treatment timing (neoadjuvant, adjuvant, or salvage) and specific regimens was not available. Geriatric-specific variables were unavailable in SEER and therefore not included. For the external validation cohort, additional surgical information, including type of gastrectomy, extent of lymphadenectomy, margin status, and postoperative complications were collected, as these data were not available in the SEER database. Median household income was not available in the external validation cohort due to institutional data availability constraints and therefore was not included in the analysis.

TNM classification followed the 8^th edition of the AJCC (17). Elderly patients were defined as those aged 60 years and older (3). EGC is a clearly defined gastric malignancy that is restricted to the mucosa or submucosa regardless of the degree of LN metastasis (18). The LODDS was defined as the log (pLN + 0.5/nLN + 0.5), where pLN represented the number of positive LNs and nLN indicated the number of negative LNs. To prevent singularity, 0.5 was incorporated into both the numerator and denominator (19).

CSS was defined as the time from diagnosis to cancer-specific death or last follow-up and was the primary outcome of this study. Data were extracted from the SEER database for the training and internal validation cohorts, while the CSS of the external validation cohort was followed up via telephone or outpatient visits.

Statistical analysis

R software (version 4.4.1), SPSS software (version 29.0) and X-tile software were the main tools used for the statistical analyses in this study. Categorical variables were summarized as frequencies and proportions and were evaluated via the Chi-squared test or Fisher’s exact test. Continuous variables were reported as medians and interquartile ranges (IQRs). For continuous variables following a normal distribution, an independent sample t-test was used, while those not conforming to normality were analyzed via the Mann-Whitney U test. Median follow-up time was calculated using the reverse Kaplan-Meier (K-M) method.

Variables with statistical significance in univariate analyses, as well as clinically relevant factors, were included in the multivariable Cox regression models. The hazard ratio (HR) and 95% confidence interval (CI) were calculated, and P<0.05 was considered statistically significant. The LODDS score was calculated from the number of regional nodes examined and the number of positive regional nodes. The optimal cutoff value for the LODDS was determined via X-tile software, which divides the LODDS into three groups: the low (≤−1.65) group, medium (−1.65< LODDS <−1.25) group, and high (≥−1.25) group.

Univariate and multivariate Cox regression analyses were performed via SPSS software and R (survival package) for nomogram development. The predictive performance of the nomogram was subsequently assessed in the internal validation and external validation cohorts. Receiver operating characteristic (ROC) curves, area under the curve (AUC) values, and the concordance index (C-index) were used to evaluate the model’s accuracy and reliability across the three cohorts. The calibration curves and decision curve analysis (DCA) were additionally used to validate and evaluate the model’s accuracy for predicting 3-, 5-, and 10-year CSS, as well as the model’s potential clinical applicability. Finally, the risk scores from the total cohort were categorized into low-risk and high-risk groups (the cutoff value was 1.57) via X-tile based on CSS. This cutoff value was consistently applied in subsequent overall survival (OS) analyses. K-M survival curves, along with log-rank tests, were used to compare CSS outcomes between the two groups.

Subgroup analyses were conducted to evaluate whether the prognostic effect of LODDS on CSS varied across clinically relevant subgroups. Patients were stratified by age, using two cutoffs (75 and 80 years). Multivariable Cox proportional hazards models were fitted within each subgroup, and interaction terms between LODDS and subgroup variables were included to formally assess effect modification. P values for interaction were reported.

Sensitivity analyses were performed to assess the robustness of the primary findings using alternative modeling strategies and patient restrictions. Consistent results across models were considered indicative of robustness.

Results

Demographic and clinical characteristics of the patients

A total of 1,898 individuals extracted from the SEER database were randomly assigned to the training cohort (n=1,138) or the internal validation cohort (n=760) at a ratio of 6:4. The detailed demographic and clinical characteristics of the elderly patients with EGC are summarized in Table 1. No significant differences were observed in the baseline characteristics between the two cohorts. With respect to the N stage, the proportion of tumors in the N0 stage was over 85% in the training cohort (86.8%) and in the internal validation cohort (86.4%). However, the proportion of patients with higher LODDS scores was greater in both cohorts, accounting for 61.0% and 63.8%, respectively.

In addition, 174 patients from Zhongshan Hospital were included in the external validation. The baseline characteristics are shown in Table 2. Similarly, patients with LODDS scores >−1.65 accounted for a larger proportion (60.4%), while those with N0 stages accounted for 85.6% of all patients.

The median follow-up times for OS/CSS were 92/80 months in the training cohort, 90/81 months in the internal validation cohort, and 79.77 months for CSS in the external validation cohort.

Cause of death distribution

During follow-up, a substantial proportion of deaths were attributable to causes other than GC. In the overall cohort, non-cancer-related deaths accounted for 23.97% of all patients, whereas cancer-specific deaths accounted for 17.02%. Similar distributions were observed in the training and internal validation cohort (Table S1). Given that cancer-specific mortality remained considerable despite the presence of competing non-cancer risks, CSS was selected as the primary endpoint for model development.

Screening the predictors correlated with CSS in the training cohort

To identify the prognostic factors associated with CSS, univariate and multivariate Cox regression analyses were performed in the training cohort. Collectively, age (P=0.003; 95% CI: 1.105–1.604), sex (P<0.001; 95% CI: 0.429–0.789), histologic type (P=0.002; 95% CI: 0.639–0.901), T stage (P=0.03; 95% CI: 1.017–1.473), N stage (P<0.001; 95% CI: 1.195–1.484), radiotherapy (P<0.001; 95% CI: 2.257–4.433), chemotherapy (P<0.001; 95% CI: 1.662–3.160), tumor size (P=0.003; 95% CI: 1.172–2.178), LODDS score (P<0.001; 95% CI: 1.229–1.953), median household income (P=0.002; 95% CI: 0.541–0.869), and median positive LNs (P<0.001) were identified as potential predictors for CSS by the univariate Cox regression analysis. Furthermore, multivariate Cox regression analysis revealed that age (P=0.004; 95% CI: 1.114–1.745), sex (P=0.002; 95% CI: 0.444–0.830), histologic type (P=0.003; 95% CI: 0.608–0.902), T stage (P=0.004; 95% CI: 1.114–1.745), radiotherapy (P=0.001; 95% CI: 1.477–4.711), tumor size (P=0.046; 95% CI: 1.006–1.911), and LODDS score (P=0.004; 95% CI: 1.116–1.812) were independent prognostic factors for CSS (Table 3).

Predictors associated with OS

To further explore the clinical relevance of prognostic factors in the context of patient management, multivariable Cox regression analysis for OS was additionally performed using the same candidate variables. Several factors remained independently associated with OS (Table S2). Specifically, LODDS score (P=0.04; 95% CI: 1.017–1.743) was an independent prognostic factor for OS. And the C-index was 0.689 (95% CI: 0.656–0.719) in the internal validation cohort.

Development and validation of the nomogram model

Based on the results of the variable screening in the training cohort, independent risk factors were utilized to develop a prognostic nomogram model (Figure 2). This model converts each risk factor into a numerical value and calculates the aggregate of all values along the top axis for each individual, which could then predict 3-, 5-, and 10-year CSS for elderly patients with EGC effectively.

Figure 2 The nomogram of CSS at 3-, 5-, and 10-year in the elderly patients with EGC. CSS, cancer-specific survival; EGC, early gastric cancer; LODDS, log odds of positive lymph nodes; NOS, not otherwise specified; T, tumor.

To evaluate the predictive accuracy of the nomogram, the C-index was calculated. The C-index was 0.697 (95% CI: 0.659–0.732) in the training cohort, 0.668 (95% CI: 0.619–0.712) in the internal validation cohort, and 0.697 (95% CI: 0.608–0.778) in the external validation cohort, indicating that the model demonstrated good discriminative ability. Additionally, ROC curve analysis was performed, and AUCs were calculated. In the training cohort, the AUCs were 0.716 for 3-year CSS, 0.737 for 5-year CSS, and 0.780 for 10-year CSS. In the internal validation cohort, the AUCs were 0.683, 0.693 and 0.692, respectively. The AUCs were 0.729, 0.655 and 0.636, respectively, in the external validation cohort (Figure 3). The calculated AUCs indicated that this model performed well in terms of discrimination ability.

Figure 3 The ROC curves for CSS of the training cohort (A), internal validation cohort (B), and external validation cohort (C) at 3, 5, and 10 years. AUC, area under the curve; CSS, cancer-specific survival; ROC, receiver operating characteristic.

Calibration curves were subsequently plotted to assess the agreement between the predicted and observed CSS. The predicted CSS probabilities are plotted on the x-axis, whereas the observed outcomes are displayed on the y-axis. As shown in Figure 4, the calibration curves for 3- and 5-year CSS demonstrated good concordance between the predicted and actual survival rates across the training, internal validation, and external validation cohorts. However, the calibration for the 10-year CSS showed some deviation, indicating less accurate long-term prediction. Overall, at all three time points, the nomogram exhibited satisfactory predictive accuracy for CSS.

Figure 4 The calibration curves for 3-, 5-, and 10-year CSS in the training cohort (A), internal validation cohort (B), and external validation cohort (C). CSS, cancer-specific survival.

Clinical utility of the nomogram model

DCA was conducted to assess the clinical utility of the nomogram. As shown in Figure 5, the model demonstrated a greater net clinical benefit for predicting 3-, 5-, and 10-year CSS. Furthermore, patients were stratified into high-risk and low-risk groups based on a cutoff value of 1.57 derived from the risk scores in the total cohort. K-M survival analysis revealed a statistically significant difference in CSS between the two groups, with P values of <0.001 in both the training and internal validation cohorts and <0.001 in the external validation cohort (Figure 6). These results indicated that the nomogram effectively distinguished high-risk patients with a high level of sensitivity. Overall, the nomogram demonstrated promising predictive performance and potential clinical utility in guiding individualized prognostic assessment. Additionally, for the entire SEER cohort (n=1,898), the 3-, 5-, and 10-year CSS rates were 87.4%, 84.1%, and 79.3%, respectively (Figure S1A). And for the entire external validation cohort (n=174), the 3-, 5-, and 10-year CSS rates were 94.0%, 82.8%, and 62.8%, respectively (Figure S1B).

Figure 5 DCA of the nomogram in the training cohort (A), internal validation cohort (B) and external validation cohort (C). DCA, decision curve analysis.

Figure 6 Kaplan-Meier curves of CSS for patients in the different risk groups in the training cohort (A), internal validation cohort (B), and external validation cohort (C). CSS, cancer-specific survival.

Subgroup analyses

Subgroup analyses suggested that the prognostic effect of LODDS on CSS was generally consistent across age groups. Using a cutoff of 75 years, no significant interaction between age and LODDS was observed. In contrast, when using a cutoff of 80 years, a significant interaction was observed, with a markedly stronger association in patients aged ≥80 years and a more modest association in those aged <80 years. However, the estimate in the ≥80-year group was imprecise, as reflected by the wide CI (Figure S2).

Sensitivity analyses

Radiotherapy is not routinely recommended for EGC, therefore two additional analyses were conducted: excluding patients who received radiotherapy and excluding radiotherapy from candidate variables. We further restricted the analysis to patients with more than 15 examined LNs to ensure adequate nodal assessment. The association between LODDS and CSS remained consistent across above sensitivity analyses. Notably, in the N0-only subgroup, LODDS was not significantly associated with survival outcomes (Table S3).

Discussion

With improvements in diagnostic techniques, GC is increasingly detected at an early stage. However, the benefit of postoperative management strategies remains heterogeneous, as not all patients derive sustained benefits from additional interventions (18). Previous studies have predominantly focused on younger patients with EGC, largely due to their longer life expectancy (20,21). With increasing longevity, long-term CSS has become an increasingly relevant outcome in elderly patients with EGC (3).

Although the overall prognosis of EGC is generally favorable, treatment decision-making in elderly patients should be guided by a comprehensive assessment of functional status, comorbidities, and competing risks of non-cancer mortality (22,23). Nevertheless, although geriatric-specific variables are limited in population-based registries such as SEER, oncology-focused models can still provide clinically useful information by characterizing tumor-related risk within TNM-defined early-stage disease.

In this study, CSS at multiple time points (3-, 5-, and 10-year) was evaluated to characterize tumor-related outcomes in elderly patients with EGC. We particularly emphasized 10-year CSS to better capture long-term oncologic risk, as short-term survival is generally favorable in EGC, whereas differences in cancer-related mortality may become more apparent with extended follow-up (24). Given the substantial competing mortality in this age group, OS was additionally analyzed to contextualize CSS. Although LODDS was independently associated with both endpoints, its association was stronger for CSS than for OS, possibly due to attenuation by non-oncologic deaths, while discriminative performance remained comparable across endpoints.

Using SEER data and external validation from Zhongshan Hospital in China, we developed a nomogram to support long-term cancer-specific risk stratification in elderly patients with EGC, with LODDS contributing substantially to risk separation. Importantly, it should be interpreted as a complementary tool to support individualized clinical decision-making alongside comprehensive geriatric assessment and overall health evaluation (25).

Within this framework of cancer-specific risk stratification, it is important to further interpret the role of LN-related parameters, particularly the complementary value of traditional N staging and LODDS. Traditional TNM staging remains the cornerstone of prognostic assessment in GC (26,27). However, conventional N staging relies primarily on the number of positive LNs and does not account for the total number of examined nodes, which may introduce bias, particularly in early-stage disease with limited LN retrieval (9,28). To address this limitation, alternative indicators such as LODDS have been proposed to provide complementary prognostic information (29,30).

Consistent with previous studies (30,31), LODDS integrates information from both positive and negative LNs and provides a more comprehensive assessment of nodal status. In the present study, different LN staging methods resulted in different risk group distributions within the same population. While conventional N staging classified a large proportion of patients as low risk, LODDS-based stratification identified a broader and more heterogeneous risk distribution. These findings indicate that LODDS and N staging capture distinct aspects of nodal involvement and should be viewed as complementary rather than competitive. In particular, LODDS may offer additional prognostic value in populations with a high prevalence of LN-negative disease, with potential implications for postoperative risk assessment and follow-up strategies (32). In further sensitivity analyses, LODDS was not significant in the N0-only subgroup, possibly because LODDS provided limited additional prognostic information when all examined nodes were negative.

In addition to LODDS, advanced age and higher T stage were independently associated with poorer CSS in elderly patients with EGC. Male patients were also identified as an unfavorable prognostic factor, consistent with previous studies (33,34), and might warrant closer postoperative risk assessment and follow-up. Radiotherapy was associated with reduced CSS. However, this finding should be interpreted with caution, as radiotherapy is not a standard treatment for EGC, and radiotherapy indications cannot be ascertained in the SEER database, raising concerns about selection bias. To address this, we performed two sensitivity analyses, excluding patients who received radiotherapy and excluding radiotherapy from candidate variables, and both of the results remained consistent. Chemotherapy was not retained in the final model, possibly reflecting heterogeneous treatment indications and incomplete staging information in SEER, and this finding should not be interpreted as evidence against the benefit of systemic therapy in appropriately selected patients (35).

To improve clinical applicability in geriatric practice, we explored age-specific subgroups. Overall, the prognostic value of LODDS appeared largely stable among older patients, while a potentially stronger association was suggested in the very elderly (≥80 years). This observation should be interpreted cautiously, as the limited precision likely reflects small sample sizes in this subgroup, highlighting the need for validation in larger cohorts of very elderly patients.

Taken together, these findings support the use of our model for risk stratification in the overall study population. Beyond that, the identification of high-risk patients in this study was intended to support cancer-specific risk stratification rather than treatment escalation. In elderly patients with EGC, the nomogram may assist postoperative risk assessment and follow-up planning when interpreted alongside clinical judgment.

There are several strengths in this study. This is the first study to explore a nomogram for predicting CSS in elderly EGC patients up to 10 years after surgery based on a large population-based database. External validation was conducted to evaluate the robustness of the model, thereby enhancing its credibility. Although the external validation cohort was relatively small, the nomogram demonstrated stable predictive performance, as indicated by the C-index, AUC, calibration curves, and DCA across the three cohorts. Taken together, these findings suggest that the model may offer additional prognostic value in elderly EGC patients, with potential implications for postoperative risk assessment and follow-up strategies.

There are several limitations in this study. First, the study population consisted of elderly patients aged 60 years and older, and the prediction of 10-year CSS might be inaccurate, as illustrated by the calibration curve analysis. Second, key surgical quality indicators are unavailable in SEER, limiting interpretation of LN-based indicators such as LODDS. Although these variables were collected in the external cohort, its relatively small sample size and modest C-index may reduce the robustness of validation. Third, geriatric-specific variables were unavailable in SEER database and were also lacking in the external cohort, limiting the applicability of the model to settings where comprehensive geriatric assessment is routinely performed. Additionally, socioeconomic variables were available in the SEER cohort, but were unavailable in the external validation cohort due to institutional data availability constraints, which may limit generalizability. Furthermore, although radiotherapy is generally not recommended for early gastric cancer, some patients in the SEER database received radiotherapy. Since SEER does not provide information on the indication for radiotherapy, this may introduce selection bias.

Conclusions

We developed and externally validated an oncology-focused prognostic nomogram for elderly patients with early gastric cancer. The nomogram complements TNM staging by enabling cancer-specific risk stratification and may provide additional prognostic information when interpreted alongside clinical assessment.

Acknowledgments

We would like to acknowledge the Surveillance, Epidemiology, and End Results (SEER) database for its support. We are grateful to all the contributors for making the SEER data available for research. And we acknowledge all the authors whose publications are referred in our article.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-aw-815/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-aw-815/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-aw-815/prf

Funding: This study was sponsored by Shanghai Clinical Nutrition Research Center, China (grant No. 2025XKPT30).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-aw-815/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the World Medical Association Declaration of Helsinki and its subsequent amendments, and was approved by the Ethics Committee of Zhongshan Hospital (The Affiliated Hospital of Fudan University, No. B2023-196), and all patients signed informed consent.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
2. Li P, Li Z, Linghu E, et al. Chinese national clinical practice guidelines on the prevention, diagnosis, and treatment of early gastric cancer. Chin Med J (Engl) 2024;137:887-908. [Crossref] [PubMed]
3. Nomura S, Kaminishi M. Surgical treatment of early gastric cancer. Dig Surg 2007;24:96-100. [Crossref] [PubMed]
4. Chen Y, Wang B, Zhao Y, et al. Metabolomic machine learning predictor for diagnosis and prognosis of gastric cancer. Nat Commun 2024;15:1657. [Crossref] [PubMed]
5. Zhang K, Wang H, Cheng Y, et al. Early gastric cancer detection and lesion segmentation based on deep learning and gastroscopic images. Sci Rep 2024;14:7847. [Crossref] [PubMed]
6. Zheng H, Huang Q, Lin J, et al. A machine learning-based predictive model for lymph node metastasis in elderly patients with early gastric cancer and prognosis. Updates Surg 2025; Epub ahead of print. [Crossref] [PubMed]
7. Chen D, Fu M, Chi L, et al. Prognostic and predictive value of a pathomics signature in gastric cancer. Nat Commun 2022;13:6903. [Crossref] [PubMed]
8. Japanese gastric cancer treatment guidelines 2018 (5th edition). Gastric Cancer 2021;24:1-21.
9. Yin K, Jin X, Pan Y, et al. Revolutionizing T3-4N0-2M0 gastric cancer staging with an innovative pathologic N classification system. J Gastrointest Surg 2024;28:1283-93. [Crossref] [PubMed]
10. Wang Q, Ge JT, Wu H, et al. Impacts of neoadjuvant therapy on the number of dissected lymph nodes in esophagogastric junction cancer patients. BMC Gastroenterol 2023;23:64. [Crossref] [PubMed]
11. Liu S, Wang Z, Ge Y, et al. Prognostic significance of an innovative staging system based on the logarithmic odds of positive lymph nodes for resectable gastroesophageal cancer after neoadjuvant chemoradiation: a population-based study with external validation of data. J Transl Med 2024;22:801. [Crossref] [PubMed]
12. Díaz Del Arco C, Estrada Muñoz L, Sánchez Pernaute A, et al. Prognostic role of the log odds of positive lymph nodes in Western patients with resected gastric cancer: A comparison with the 8th edition of the TNM staging system. Am J Clin Pathol 2024;161:186-96.
13. Gu P, Deng J, Sun Z, et al. Superiority of log odds of positive lymph nodes (LODDS) for prognostic prediction after gastric cancer surgery: a multi-institutional analysis of 7620 patients in China. Surg Today 2021;51:101-10. [Crossref] [PubMed]
14. Zhou R, Zhang J, Sun H, et al. Comparison of three lymph node classifications for survival prediction in distant metastatic gastric cancer. Int J Surg 2016;35:165-71. [Crossref] [PubMed]
15. Kamiya A, Hayashi T, Sakon R, et al. Prognostic Impact of Long-term Postoperative Pneumonia in Elderly Patients with Early Gastric Cancer. J Cancer 2022;13:2905-11. [Crossref] [PubMed]
16. Song P, Zhao Y, Tian W, et al. Prognostic factors for overall survival in patients with early gastric cancer: a retrospective cohort study based on SEER database. J Gastrointest Oncol 2025;16:1878-89. [Crossref] [PubMed]
17. Amin MB, Greene FL, Edge SB, et al. The Eighth Edition AJCC Cancer Staging Manual: Continuing to build a bridge from a population-based to a more "personalized" approach to cancer staging. CA Cancer J Clin 2017;67:93-9.
18. Yang K, Lu L, Liu H, et al. A comprehensive update on early gastric cancer: defining terms, etiology, and alarming risk factors. Expert Rev Gastroenterol Hepatol 2021;15:255-73. [Crossref] [PubMed]
19. Qiu MZ, Qiu HJ, Wang ZQ, et al. The tumor-log odds of positive lymph nodes-metastasis staging system, a promising new staging system for gastric cancer after D2 resection in China. PLoS One 2012;7:e31736. [Crossref] [PubMed]
20. Tan N, Wu H, Cao M, et al. Global, regional, and national burden of early-onset gastric cancer. Cancer Biol Med 2024;21:667-78. [Crossref] [PubMed]
21. Ben-Aharon I, van Laarhoven HWM, Fontana E, et al. Early-Onset Cancer in the Gastrointestinal Tract Is on the Rise-Evidence and Implications. Cancer Discov 2023;13:538-51. [Crossref] [PubMed]
22. Sekiguchi M, Oda I, Morita S, et al. Management of elderly patients with early gastric cancer in Japan. Jpn J Clin Oncol 2022;52:425-32. [Crossref] [PubMed]
23. Jun JK, Choi KS, Lee HY, et al. Effectiveness of the Korean National Cancer Screening Program in Reducing Gastric Cancer Mortality. Gastroenterology 2017;152:1319-1328.e7. [Crossref] [PubMed]
24. Zhao L, Niu P, Wang W, et al. Application of Survival Quilts for prognosis prediction of gastrectomy patients based on the Surveillance, Epidemiology, and End Results database and China National Cancer Center Gastric Cancer database. J Natl Cancer Cent 2024;4:142-52. [Crossref] [PubMed]
25. Guan X, Liu J, Xu L, et al. Development and validation of a machine learning-based prognostic model for gastric cancer: a multicenter retrospective study. Langenbecks Arch Surg 2025;411:15. [Crossref] [PubMed]
26. Yu J, Yao R, Han N, et al. Evaluating the prognostic significance of tumor deposits in gastric cancer and strategies for their integration into the TNM staging system: a single-center retrospective study. Cell Oncol (Dordr) 2025;48:761-73. [Crossref] [PubMed]
27. Di Gregorio F, Polici M, Pilozzi E, et al. Staging of Gastric Cancer: CT Patterns and Correlation with Pathologic Findings. Radiographics 2025;45:e240186. [Crossref] [PubMed]
28. Fu Y, Qin C, Li M, et al. Comparative Evaluation of 68Ga-FAPI-04 PET for Initial N and M Staging in Gastric Cancer: A Study Against Histopathology and Contrast-Enhanced CT. Clin Nucl Med 2025;50:394-403. [Crossref] [PubMed]
29. Wang C, Fang Y, Zhang Y, et al. A LODDS-based nomogram for overall and cancer-specific survival in stage III-IV gastric signet ring cell carcinoma. Front Mol Biosci 2025;12:1600626. [Crossref] [PubMed]
30. Deng MC, Chen K, Bao QM, et al. Evaluating log odds of positive lymph nodes as a prognostic tool in differentiated gastric cancer: A retrospective study. World J Gastroenterol 2025;31:107029. [Crossref] [PubMed]
31. Xia H, Qian Y, Pang Q, et al. Prognostic value of LODDS in late-onset gastric adenocarcinoma: analysis of SEER, TCGA, and external multicenter cohorts. Front Oncol 2025;15:1619504. [Crossref] [PubMed]
32. Que SJ, Zhong Q, Chen QY, et al. A Novel ypTLM Staging System Based on LODDS for Gastric Cancer After Neoadjuvant Therapy: Multicenter and Large-Sample Retrospective Study. World J Surg 2023;47:1762-71. [Crossref] [PubMed]
33. López MJ, Carbajal J, Alfaro AL, et al. Characteristics of gastric cancer around the world. Crit Rev Oncol Hematol 2023;181:103841. [Crossref] [PubMed]
34. Li H, Wei Z, Wang C, et al. Gender Differences in Gastric Cancer Survival: 99,922 Cases Based on the SEER Database. J Gastrointest Surg 2020;24:1747-57. [Crossref] [PubMed]
35. Zhang S, Zheng L, Zhang Y, et al. A web-based prediction model for long-term cancer-specific survival of middle-aged patients with early-stage gastric cancer: a multi-institutional retrospective study. J Cancer Res Clin Oncol 2023;149:16551-61. [Crossref] [PubMed]