Risk factors and nomogram for early mortality in young patients with non-metastatic gastric cancer

Introduction

Gastric cancer remains a major health burden both globally and in China. It most commonly occurs in individuals aged 50–70 years, with a relatively lower incidence in younger populations (1). Although the overall incidence of gastric cancer has been gradually declining in recent years, the incidence among young adults has shown an upward trend (2). Compared with elderly patients, young patients with gastric cancer (YGC) exhibit distinct clinicopathological characteristics. YGC tends to be more prevalent in females, often arises in the antrum, and presents with diffuse-type histology. A subset of patients also has a genetic predisposition (3,4). On the one hand, younger patients generally have better overall health status and can better tolerate comprehensive treatment, with higher expectations for treatment outcomes. On the other hand, timely diagnosis and treatment of YGC can lead to improved prognoses and help reduce the socioeconomic burden.

Currently, there is no universally accepted definition of YGC, and whether its prognosis differs from that of elderly gastric cancer remains controversial. However, in clinical practice, some young patients are observed to have poor survival outcomes. Tumor-related early mortality events significantly impact quality of life and increase the societal burden (5). For patients without distant metastasis, favorable prognoses can often be achieved through comprehensive treatments, including surgery, chemotherapy, radiotherapy, and immunotherapy. The median survival time for patients with non-metastatic gastric cancer is approximately 4 to 8 years. However, among young patients with non-metastatic gastric cancer, we still observe a subset who die within one year, with a survival time comparable to that of patients with metastatic advanced gastric cancer receiving chemotherapy. This imposes a substantial socioeconomic burden. Early identification of this subgroup and timely modification of the initial treatment strategy may substantially improve both overall survival and progression-free survival in this population. The Tumor-Node-Metastasis (TNM) staging system is the most commonly used staging system for predicting prognosis, yet it does not take into account factors such as pathological subtype and chemotherapy response, limiting its precision in identifying high-risk patients. Therefore, early identification of young gastric cancer patients with poor prognosis remains an urgent clinical need.

Nomograms are visualized clinical models that incorporate outcome-related risk factors and offer unique advantages in clinical application. Several nomogram models for gastric cancer have already been developed and published, playing important roles in clinical decision-making (6-8). However, there is a lack of studies analyzing risk factors and developing predictive models for early death in young, non-metastatic gastric cancer patients. To address this gap, we utilized data from the Surveillance, Epidemiology, and End Results (SEER) database as well as clinical data from our hospital to identify risk factors associated with early death in young gastric cancer patients without distant metastasis. Based on real-world data, we constructed and validated a nomogram model to predict early mortality in this population, aiming to facilitate early identification of high-risk patients and guide timely therapeutic interventions to improve prognosis. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1657/rc).

Methods

Data sources and study population

Based on existing literature, we defined YGC as gastric cancer diagnosed at age ≤45 years (9). The inclusion and exclusion criteria for this study were as follows: (I) patients with a pathologically confirmed diagnosis of gastric cancer according to International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) codes; (II) age at diagnosis ≤45 years; (III) complete follow-up records; (IV) a clearly documented cause of death; (V) no missing key clinical data (e.g., sex, age); and (VI) absence of distant metastasis at the time of diagnosis.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by Sun Yat-sen University Cancer Center Gansu Hospital (Approval No. A2025011100001, approved on January 10, 2025) and individual consent for this analysis was waived due to the retrospective nature. We extracted clinical data of gastric cancer patients diagnosed between 2012 and 2021 from the SEER database using SEER*Stat software version 8.4.4 in January 2025. Additionally, we collected data on patients diagnosed with gastric cancer at Sun Yat-sen University Cancer Center Gansu Hospital between 2011 and 2025. Based on the inclusion and exclusion criteria, a total of 2,087 eligible cases from the SEER database and 78 cases from Sun Yat-sen University Cancer Center Gansu Hospital were included in the study.

The included variables were: sex, age, T stage, N stage, surgical treatment, tumor grade, tumor location, tumor size, marital status, presence of multiple primary tumors, chemotherapy, and radiotherapy. Tumor location was categorized as cardia/fundus versus non-cardia/fundus. Tumor size was grouped into <2 cm, 2–5 cm, and >5 cm. Marital status was classified as single versus non-single, and tumor grade was divided into differentiated and undifferentiated types. The primary endpoint of this study was early death. In accordance with previous studies, early death was defined as death due to gastric cancer occurring within 12 months of diagnosis (10,11).

Model development, validation and performance evaluation

In this study, the 2,087 eligible patients from the SEER database were randomly divided into a training cohort (n=1,461) and a validation cohort (n=626) at a 7:3 ratio. Randomization and all statistical analyses were performed using R software. Prognostic factors were identified through univariate and multivariate logistic regression analyses. Based on the significant variables identified in the training cohort, a nomogram was developed to predict early death in young gastric cancer patients.

The nomogram was internally validated using the Hosmer-Lemeshow goodness-of-fit test, and calibration curves were generated with 1,000 bootstrap resamples to compare the predicted probabilities with actual outcomes. The predictive performance of the nomogram was further assessed using Harrell’s concordance index (C-index) and receiver operating characteristic (ROC) curves.

For the finalized model, additional validation was conducted using both the internal validation cohort and the external validation cohort from Sun Yat-sen University Cancer Center Gansu Hospital. The Hosmer-Lemeshow test and calibration curves (based on 1,000 bootstrap resamples) were again used to evaluate model calibration, while the C-index and ROC curves were employed to assess the model’s discrimination ability.

Statistical analysis

Statistical analyses were conducted using R software version 4.0.2 and Python version 3.8. Data preprocessing, filtering, and visualization were performed using various R and Python packages, including Tableone, rms, pROC, dplyr, pandas, and ggplot2. Visualizations such as nomograms, calibration plots, ROC curves, and decision curve analysis (DCA) curves were generated. No formal sample size calculation was performed. All eligible cases that met the inclusion criteria during the study period were included to maximize the statistical power and representativeness of the dataset. Prognostic factors included in the nomogram were selected based on the results of univariate and multivariate regression analyses. Variables with P<0.05 in univariate analysis were further entered into a multivariate model, and independent predictors were incorporated into the final nomogram. Categorical variables were expressed as frequencies and percentages. To compare clinical characteristics between the training and validation cohorts, the chi-square test (χ² test) or Fisher’s exact test was applied for categorical variables, while the t-test or one-way analysis of variance (ANOVA) was used for continuous variables. A two-sided P value of less than 0.05 was considered statistically significant.

Results

Baseline clinical characteristics

The clinical characteristics of the 2,087 eligible cases from the SEER database are summarized in Table 1. A total of 450 patients (21.6%) experienced early death events. The mean age of patients was 38.7±5.8 years. Among them, 907 patients (43.5%) were female, and 1,180 (56.5%) were male. Regarding T stage, 428 patients (20.5%) were classified as T1, 761 (36.5%) as T2, 471 (22.6%) as T3, and 270 (12.9%) as T4. For N stage, 810 patients (38.8%) were N0, 711 (34.1%) N1, 342 (16.4%) N2, and 108 (5.2%) N3. In terms of tumor location, 575 patients (27.6%) had tumors in the cardia/fundus, while 1,512 (72.4%) had tumors in the antrum. Undifferentiated tumors were observed in 1,539 patients (73.7%), and differentiated tumors in 309 patients (14.8%), with tumor grade unknown in 239 cases (11.5%). Tumors smaller than 2 cm were found in 296 patients (14.2%). Surgical treatment was received by 1,616 patients (77.4%). Chemotherapy was administered to 70.5% of patients, while 29.5% did not receive chemotherapy. Radiotherapy was performed in 752 patients (36.0%), and 192 patients were diagnosed with multiple primary malignancies.

In the external validation cohort from Sun Yat-sen University Cancer Center Gansu Hospital, consisting of 78 cases, 30 patients (38.5%) experienced early death events. The mean age was 32.8±7.2 years. Of these patients, 46 (59.0%) were female, and 32 (41.0%) were male. T stage distribution included 22 patients (28.2%) at T1, 24 (30.8%) at T2, 12 (15.4%) at T3, and 20 (25.6%) at T4. For N stage, 47 patients (60.3%) were N0, 16 (20.5%) were N1, 8 (10.3%) were N2, and 7 (9.0%) were N3. Regarding tumor location, 30 patients (38.5%) had tumors in the cardia/fundus and 48 (61.5%) in the antrum. A total of 51 patients (65.4%) had undifferentiated tumors, while 27 (34.6%) had differentiated tumors. Tumor size was <2 cm in 15 cases (19.2%). Surgical treatment was administered to 50 patients (64.1%). Chemotherapy was received by 49 patients (62.8%), and radiotherapy was administered to 17 patients (21.8%). Among these, 58 patients (74.4%) were married or non-single. Most cases (76 patients, 97.7%) were single primary tumors.

Compared to the training cohort, notable differences were observed in several baseline characteristics, indicating that the external validation cohort represented a distinct and independent dataset.

Analysis of risk factors for early mortality

We first conducted univariate analyses. As shown in Table 2, univariate analysis identified several factors significantly associated with early death, including T stage, N stage, radiotherapy, marital status, surgical treatment, tumor grade, and tumor size. In contrast, age, chemotherapy, and tumor location were not significantly associated with early death.

Subsequently, variables found to be significant in the univariate analysis were included in a multivariate logistic regression analysis. The results indicated that all these factors remained statistically significant in the multivariate model (Table 2). Therefore, we confirmed that T stage, N stage, radiotherapy, marital status, surgical treatment, tumor grade, and tumor size were independently associated with early death in young gastric cancer patients.

Construction and validation of the nomogram

To facilitate clinical application of our findings, we incorporated the independent risk factors identified in the multivariate analysis to construct a nomogram predicting the risk of early death (Figure 1). The nomogram visually highlights that surgical treatment status, T stage, and N stage carry the greatest weight in predicting early death among young gastric cancer patients, indicating their critical role in prognosis. Further evaluation demonstrated that the predictive model achieved a C-index of 0.748 and an area under the curve (AUC) of 0.81. Compared with using TNM staging alone, our model showed superior discrimination, as reflected by a larger AUC (0.81 vs. 0.64) on the ROC curve (Figure 2A), suggesting improved accuracy in predicting early death in young gastric cancer patients. Calibration curves generated via bootstrap resampling showed good agreement between predicted probabilities and observed outcomes (Figure 2B). We also validated the model using the internal validation cohort, where the C-index was 0.792 and the AUC was 0.791. Bootstrap validation on this cohort showed an absolute error of 0.02 between predicted and observed outcomes (Figure 3A,3B). Moreover, external validation was performed using data from Sun Yat-sen University Cancer Center Gansu Hospital. As shown in Figure 3C,3D, the model yielded a C-index of 0.758 and an AUC of 0.757 in the external cohort, with an absolute error of 0.013 between predicted and actual outcomes based on bootstrap sampling. These results indicate good calibration and consistent predictive performance across both internal and external datasets, demonstrating the model’s strong clinical generalizability.

Figure 1 Nomogram for predicting early mortality in young patients with non-metastatic gastric cancer.

Figure 2 Receiver operating characteristic curves (A) and calibration plot (B) of the prediction model. AUC, area under the curve; TNM, Tumor-Node-Metastasis.

Figure 3 ROC curves and calibration plots for the validation sets. (A) ROC curve for the internal validation set; (B) calibration plot for the internal validation set; (C) ROC curve for the external validation set; (D) calibration plot for the external validation set. AUC, area under the curve; ROC, receiver operating characteristic.

Finally, we assessed the clinical utility of the model using DCA. As illustrated in Figure 4, the decision curves for the training cohort, internal validation cohort, and external validation cohort all demonstrate a net clinical benefit from applying the nomogram for early death risk prediction.

Figure 4 Decision curve analysis of the nomogram. (A) Training set; (B) internal validation set; (C) external validation set.

Discussion

In this study, analysis of real-world data on young gastric cancer patients without distant metastasis confirmed that T stage, N stage, radiotherapy, marital status, surgical treatment, tumor grade, and tumor size were significantly associated with early death. Based on these findings, we constructed a nomogram and validated its performance using both internal and external datasets, demonstrating its potential clinical applicability.

Although the incidence of gastric cancer has been declining in recent years, it remains a major cause of cancer morbidity and mortality worldwide, with an increasing incidence observed in younger populations (5,12). YGC exhibits distinct clinicopathological features compared to elderly patients, including a higher proportion of female patients, diagnosis often at advanced stages, and a higher prevalence of diffuse-type histology. Whether these characteristics translate into different prognoses remains controversial, with some studies reporting comparable outcomes between young and elderly patients, while others indicate poorer prognosis in YGC (13-15). This inconsistency may be partly due to the lack of a universally accepted definition of young gastric cancer. Notably, a subset of YGC cases is attributable to hereditary factors, such as hereditary diffuse gastric cancer, which typically presents before age 50 and is characterized by diffuse, poorly differentiated histology including signet-ring cells (16).

Our dataset shows a roughly equal male-to-female ratio in YGC patients, predominance of undifferentiated histology, and mostly non-cardia tumor location, with the majority receiving surgery and chemotherapy, and a smaller proportion undergoing radiotherapy. The gender distribution in our cohort differs somewhat from previous reports, while other features are consistent, underscoring the representativeness of our sample. Alarmingly, up to 20% of patients experienced early death, imposing a significant societal burden. Despite over 70% of young patients receiving surgery and chemotherapy, the high early mortality rate warrants careful consideration. Based on existing literature and our findings, it is plausible that the etiological factors underlying YGC differ from those in elderly patients, resulting in a heterogeneous subgroup within gastric cancer (17). The poor prognosis observed highlights the urgent need to identify high-risk individuals accurately and to explore more effective treatment strategies and therapeutic targets.

Nomograms are widely used tools for visualizing risk and aiding clinical decision-making. Numerous prognostic nomograms for gastric cancer have been developed, demonstrating good predictive performance (18-21). However, due to the heterogeneity of gastric cancer, no single model is universally applicable across all patient groups, emphasizing the importance of developing nomograms tailored to specific subpopulations to facilitate precision medicine. In this study, we identified and confirmed the association of T stage, N stage, radiotherapy, marital status, surgical treatment, tumor grade, and tumor size with early death in young non-metastatic gastric cancer patients. T and N stages, as core components of the TNM staging system, have long been recognized as strong prognostic indicators (22). Radiotherapy remains an important modality in cancer treatment, though its role in gastric cancer is still under debate. Some studies suggest radiotherapy can improve tumor staging in the short term but does not significantly prolong overall survival (23-25). Our analysis revealed radiotherapy as a significant factor influencing early death in YGC, whereas chemotherapy did not reach statistical significance. This may indicate a potential survival benefit from radiotherapy in young patients, warranting further investigation in larger, well-designed clinical trials. The lack of significance for chemotherapy might reflect the generally good performance status and strong treatment tolerance of young patients, leading to a high rate of combined modality treatment, with relatively few untreated patients. Additionally, the high surgical rate could mask chemotherapy’s contribution, as chemotherapy is primarily given as an adjunct to surgery. Unexpectedly, marital status emerged as an important predictor of early death. This suggests that personal and social lifestyle factors may influence gastric cancer outcomes, possibly paralleling urban-rural incidence disparities (26). We speculate that marital status may reflect differences in daily routines, lifestyle habits, and potentially socioeconomic status, all of which could impact prognosis (27).

There are several limitations in this study. First, the inherently low incidence of YGC restricted the sample size and may limit the generalizability of our findings. Larger external validation cohorts are needed to further verify model robustness. Second, clinical factors often interact intrinsically—for example, T and N stages influence surgical candidacy, while chemotherapy and radiotherapy may complement surgery or serve as primary treatments in unresectable cases (28). These interactions, although not evident in statistical collinearity tests, may affect model performance. Third, YGC remains a highly heterogeneous entity, with variability in staging, genetic background, and pathology, underscoring the need for inclusion of more comprehensive features in future analyses. Lastly, patient prognosis is multifactorial; incorporating additional variables and their interactions could improve predictive accuracy, but this conflicts with the limited sample size inherent to YGC populations. Advanced statistical methods may be required to address these challenges.

Conclusions

In summary, we analyzed the clinical characteristics of young gastric cancer patients, identified independent risk factors for early death, and developed a clinically applicable nomogram to aid prognosis prediction and personalized treatment strategies.

Acknowledgments

The authors are thankful to Sun Yat-sen University Cancer Center Gansu Hospital for their management of our patient database.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1657/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1657/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1657/prf

Funding: This work was supported by Gansu Provincial Natural Science Foundation (Grant No. 24JRRA1148) and Healthcare Industry Project of Gansu Province (Grant No. GS-62000000003-2024-034).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1657/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. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by Sun Yat-sen University Cancer Center Gansu Hospital (Approval No. A2025011100001, approved on January 10, 2025) and individual consent for this analysis was waived due to the retrospective nature.

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. López MJ, Carbajal J, Alfaro AL, et al. Characteristics of gastric cancer around the world. Crit Rev Oncol Hematol 2023;181:103841. [Crossref] [PubMed]
2. Zhang H, Cheng X, Guo W, et al. Metastasis patterns and prognosis in young gastric cancer patients: A propensity score matched SEER database analysis. PLoS One 2024;19:e0301834. [Crossref] [PubMed]
3. Decourtye-Espiard L, Guilford P. Hereditary Diffuse Gastric Cancer. Gastroenterology 2023;164:719-35. [Crossref] [PubMed]
4. Li Y, Hahn AI, Laszkowska M, et al. Clinicopathological Characteristics and Risk Factors of Young-Onset Gastric Carcinoma: A Systematic Review and Meta-analysis. Clin Transl Gastroenterol 2024;15:e1. [Crossref] [PubMed]
5. Gamble LA, Heller T, Davis JL. Hereditary Diffuse Gastric Cancer Syndrome and the Role of CDH1: A Review. JAMA Surg 2021;156:387-92. [Crossref] [PubMed]
6. Miao X, Guo Y, Ding L, et al. A dynamic online nomogram for predicting the heterogeneity trajectories of frailty among elderly gastric cancer survivors. Int J Nurs Stud 2024;153:104716. [Crossref] [PubMed]
7. Dong D, Fang MJ, Tang L, et al. Deep learning radiomic nomogram can predict the number of lymph node metastasis in locally advanced gastric cancer: an international multicenter study. Ann Oncol 2020;31:912-20. [Crossref] [PubMed]
8. Ding P, Yang J, Guo H, et al. Multimodal Artificial Intelligence-Based Virtual Biopsy for Diagnosing Abdominal Lavage Cytology-Positive Gastric Cancer. Adv Sci (Weinh) 2025;12:e2411490. [Crossref] [PubMed]
9. Ramos MFKP, Pereira MA, Sagae VMT, et al. Gastric cancer in young adults: a worse prognosis group? Rev Col Bras Cir 2019;46:e20192256. [Crossref] [PubMed]
10. Downing A, Aravani A, Macleod U, et al. Early mortality from colorectal cancer in England: a retrospective observational study of the factors associated with death in the first year after diagnosis. Br J Cancer 2013;108:681-5. [Crossref] [PubMed]
11. Qiu ZC, Cai HZ, Wu YW, et al. Nomogram for predicting early cancer-related death due to recurrence after liver resection in hepatocellular carcinoma patients with Barcelona Clinic Liver Cancer (BCLC) stage B/C: a multicenter study. BMC Gastroenterol 2025;25:14. [Crossref] [PubMed]
12. Smyth EC, Nilsson M, Grabsch HI, et al. Gastric cancer. Lancet 2020;396:635-48. [Crossref] [PubMed]
13. Kono Y, Kanzaki H, Iwamuro M, et al. Reality of Gastric Cancer in Young Patients: The Importance and Difficulty of the Early Diagnosis, Prevention and Treatment. Acta Med Okayama 2020;74:461-6. [PubMed]
14. Li J, Kuang XH, Zhang Y, et al. Global burden of gastric cancer in adolescents and young adults: estimates from GLOBOCAN 2020. Public Health 2022;210:58-64. [Crossref] [PubMed]
15. Kono Y, Kanzaki H, Tsuzuki T, et al. A multicenter observational study on the clinicopathological features of gastric cancer in young patients. J Gastroenterol 2019;54:419-26. [Crossref] [PubMed]
16. Pilonis ND, Tischkowitz M, Fitzgerald RC, et al. Hereditary Diffuse Gastric Cancer: Approaches to Screening, Surveillance, and Treatment. Annu Rev Med 2021;72:263-80. [Crossref] [PubMed]
17. Yan HHN, Siu HC, Law S, et al. A Comprehensive Human Gastric Cancer Organoid Biobank Captures Tumor Subtype Heterogeneity and Enables Therapeutic Screening. Cell Stem Cell 2018;23:882-897.e11. [Crossref] [PubMed]
18. Xia F, Zhang Q, Ndhlovu E, et al. A Novel Nomogram to Predict Resectable Gastric Cancer Based on Preoperative Circulating Tumor Cell. Clin Transl Gastroenterol 2024;15:e00561. [Crossref] [PubMed]
19. Cao M, Hu C, Pan S, et al. Development and validation of nomogram for predicting early recurrence after radical gastrectomy of gastric cancer. World J Surg Oncol 2024;22:21. [Crossref] [PubMed]
20. Gou M, Qian N, Zhang Y, et al. Construction of a nomogram to predict the survival of metastatic gastric cancer patients that received immunotherapy. Front Immunol 2022;13:950868. [Crossref] [PubMed]
21. Nakauchi M, Court CM, Tang LH, et al. Validation of the Memorial Sloan Kettering Gastric Cancer Post-Resection Survival Nomogram: Does It Stand the Test of Time? J Am Coll Surg 2022;235:294-304. [Crossref] [PubMed]
22. López Sala P, Leturia Etxeberria M, Inchausti Iguíñiz E, et al. Gastric adenocarcinoma: A review of the TNM classification system and ways of spreading. Radiologia (Engl Ed) 2023;65:66-80. [Crossref] [PubMed]
23. Chong IY, Chau I. Is there still a place for radiotherapy in gastric cancer? Curr Opin Pharmacol 2023;68:102325. [Crossref] [PubMed]
24. Lordick F, Nilsson M, Leong T. Adjuvant radiotherapy for gastric cancer-end of the road? Ann Oncol 2021;32:287-9. [Crossref] [PubMed]
25. de Steur WO, van Amelsfoort RM, Hartgrink HH, et al. Adjuvant chemotherapy is superior to chemoradiation after D2 surgery for gastric cancer in the per-protocol analysis of the randomized CRITICS trial. Ann Oncol 2021;32:360-7. [Crossref] [PubMed]
26. Li B, Hu X. Time-Varying Effects of Marital Status on Gastric Cancer: A Population-Based Study. Cancer Manag Res 2019;11:10949-55. [Crossref] [PubMed]
27. Zhang L, Zhou B, Luo P, et al. A model established using marital status and other factors from the Surveillance, Epidemiology, and End Results database for early stage gastric cancer. J Investig Med 2022;70:1373-80. [Crossref] [PubMed]
28. Ajani JA, D'Amico TA, Bentrem DJ, et al. Gastric Cancer, Version 2.2025, NCCN Clinical Practice Guidelines In Oncology. J Natl Compr Canc Netw 2025;23:169-91. [Crossref] [PubMed]