A transformer-based prognostic signature integrating tumor and body composition CT images predicts postoperative recurrence in gastric cancer

Sep 30, 2026

Accurate preoperative prognosis prediction is vital for gastric cancer treatment planning, yet existing models largely ignore body composition. This study integrates multimodal data—skeletal muscle (SM), adipose tissue (AT), and primary tumor CT images—to improve prognostic stratification across 1862 patients. A Vision Transformer-based approach produces the SM-AT-Tumor-Clinical (SMAT-TC) integrated score for predicting recurrence-free survival (RFS). The SMAT-TC score shows strong C-index performance and superior discrimination versus baseline models. Improved reclassification supports incremental value of body composition, and SMAT-TC serves as an independent recurrence risk factor enabling risk-group stratification with distinct 3- and 5-year RFS rates.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Clinical need for improved preoperative prognosis

Limitations of current TNM-based stratification

Role of malnutrition and body composition

Promise of CT-based multimodal imaging

Deep learning approach

Vision Transformer-based integrated scoring

npj | digital medicine Article | Published in partnership with Seoul National University Bundang Hospital | | | | --- | --- | --- | | [https://doi.org/10.1038/s41746-025-02183-z](https://doi.org/10.1038/s41746-025-02183-z) | | | | Atransformer-based prognostic signature integrating tumor and body composition CT images predicts postoperative recurrence in gastric cancer
Check for updates | | | | Qiuying Chen1,5, Hua Xiao2,5, Yueyue Li1,5, Lian Jian1,3,5, Lu Zhang1,5, Bo Lai1, Xuewei Wu1,4, Jingjing You1, Zhe Jin1, Hui Shen1, Jie Sun1, Wenle He1, Shuixing Zhang1 & Bin Zhang1 | | | | Accurate preoperative prognosis prediction is crucial for gastric cancer (GC) treatment planning, yet existing models overlook body composition integration. This study demonstrates the potential of integrating multimodal data, including skeletal muscle (SM), adipose tissue (AT), and primary tumor computed tomography images, to improve prognostic stratiﬁcation in GC patients using an entire cohort of 1862 patients. By leveraging a Vision Transformer-based deep learning approach, we developed and validated a SM-AT-Tumor-Clinical (SMAT-TC) integrated score to predict recurrencefree survival (RFS) in GC patients. The SMAT-TC score achieved a C-index of 0 .966 (95% CI: 0.937–0.990), 0.890 (95% CI: 0.866–0.915), and 0.855 (95% CI: 0.829–0.881) in the training, internal validation, and external validation cohorts, respectively, outperforming the Clinical, SM, AT, Tumor, Tumor-Clinical (TC), and SM-Tumor-Clinical (SM-TC) models. The net reclassiﬁcation improvement and integrated discrimination improvement conﬁrmed the incremental value of body composition. The SMAT-TC score was an independent risk factor for recurrence. The SMAT-TC model could stratify patients into high-, medium-, and low-risk groups with distinct 3- (99.6% vs. 67.0% vs. 10.9%) and 5-year RFS rates (98.8% vs. 61.7% vs. 2.4%) . Collectively, the SMAT-TC score may serve as a novel imaging biomarker for GC patients, enhancing risk stratiﬁcation and guiding individualized treatment strategies. | | | | Gastric cancer (GC) ranks as the ﬁfth most common cancer and the ﬁfth leading cause of cancer-related mortality worldwide1. Although the prognosis of GC patients is gradually improving, more than half of patients remain suffering recurrence, even after curative resection2,3. Reliable preoperative assessment is crucial for guiding treatment strategies and improving patient outcomes. Current prognostic stratiﬁcation relies primarily on the established TNM staging system4. However, signiﬁcant survival disparities remain among patients with identical tumor stages undergoing similar therapeutic regimens5. Therefore, there is an urgent need to identify more powerful biomarkers that can provide prognostic information, thereby optimizing therapeutic planning and monitoring. | Malnutrition represents a prevalent and serious concern among GC patients. The progression of GC correlates with a drastic decline in nutritional and functional status, leading to poor treatment response and unfavorable prognosis6. Numerous studies have identiﬁed malnutrition as a crucial risk factor for postoperative complications and survival in GC patients7,8. Currently, body composition analysis is gaining prominence in nutritional research. Skeletal muscle (SM) and adipose tissue (AT), as key body composition parameters, serve as indicators of nutritional status8,9. The gold standard for evaluating body composition quality is computed tomography (CT)10. As CT is routinely used for GC diagnosis and staging, preoperative scans contain not only tumor images that reﬂect biological | | | 1Department of Radiology, the First Afﬁliated Hospital, Jinan University, Guangzhou, Guangdong, China. 2Department of Hepatobiliary and Intestinal Surgery, Hunan Cancer Hospital and the Afﬁliated Cancer Hospital of Xiangya School of Medicine, Central South University, Changsha, Hunan, China. 3Department of Radiology, Hunan Cancer H

FAQs

What problem does the study address in gastric cancer prognosis modeling?

It addresses the limitation of existing prognostic models that overlook body composition integration when predicting outcomes for gastric cancer patients.

Which data modalities are combined in the proposed model?

The model integrates skeletal muscle (SM), adipose tissue (AT), and primary tumor CT images, alongside clinical information in the SMAT-TC framework.

How does the SMAT-TC score perform compared with other models?

The SMAT-TC score outperforms the Clinical, SM, AT, Tumor, Tumor-Clinical (TC), and SM-Tumor-Clinical (SM-TC) models in predicting recurrence-free survival, supported by C-index results and improved discrimination.

Sep 30, 2026

Upload by Valentina