Scholarlink

Background <p>Gastric gastrointestinal stromal tumours (gGISTs) are among the predominant subtypes of gastric submucosal tumors (SMTs) with malignant potential. Accurate differentiation between gGISTs and non-gastric gastrointestinal stromal tumours (non-gGISTs) using current imaging tools, especially for small-diameter lesions (&lt; 2.0&#xa0;cm), remains challenging. The aim of this study was to established a diagnostic nomogram model utilising endoscopic ultrasound (EUS) images to effectively distinguish small gGISTs from non-gGISTs.</p> Methods <p>We conducted a multicentre retrospective study of consecutive patients who underwent endoscopic resection (ER) for gastric SMTs at two centres from March 2020 to June 2025. Clinical data, EUS characteristics and pathological features were collected and analysed. A nomogram model for the diagnosis of small gGISTs was established, followed by internal and external validation.</p> Results <p>A total of 496 patients were included in this study. The independent predictors of gGIST diagnosis were age ≥ 60&#xa0;years (OR (95% CI) 2.30 (1.20–4.44), <i>P</i> = 0.013), gastric cardia-fundus/body location (OR (95% CI) 6.09 (1.55–23.98), <i>P</i> = 0.010), and muscularis propria/submucosa origin (OR (95% CI) 6.71 (2.24–20.04), <i>P</i> &lt; 0.001). The AUCs for the nomogram were 0.83 (95% CI 0.78–0.88), 0.81 (95% CI 0.73–0.89), and 0.87 (95% CI 0.81–0.92) in the training, internal validation, and external validation cohorts, respectively. Calibration curves showed excellent agreement between the predicted and actual probabilities for differentiating between small gGISTs and non-gGISTs. Decision curve analysis (DCA) demonstrated favourable clinical applications of the model. The external validation yielded an accuracy of 0.78, a sensitivity of 0.91, and a specificity of 0.71. A subgroup analysis between gGISTs and leiomyomas revealed that the AUC was 0.73 (95% CI 0.63–0.83), with an accuracy of 0.67, a sensitivity of 0.65, and a specificity of 0.74 in the external validation cohort.</p> Conclusion <p>Patient age, lesion location, and origin layer were independent diagnostic factors for small gGISTs. The proposed nomogram served as a valuable tool for differentiating between small gGISTs and non-gGISTs.</p> Graphical abstract

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Development and validation of an endoscopic ultrasound-based nomogram for differentiating small gastric stromal tumours from non-gastric stroma tumours

  • 期刊论文

  • 发表日期：

    2026年3月9日

查看全文
  • Weijia Dou,
  • Junjie Li,
  • Lei Shang,
  • Chun Song,
  • Haoying Wang,
  • Jing Ma,
  • Jun Wang,
  • Yan Wang,
  • Zhenxiong Liu

摘要

Background

Gastric gastrointestinal stromal tumours (gGISTs) are among the predominant subtypes of gastric submucosal tumors (SMTs) with malignant potential. Accurate differentiation between gGISTs and non-gastric gastrointestinal stromal tumours (non-gGISTs) using current imaging tools, especially for small-diameter lesions (< 2.0 cm), remains challenging. The aim of this study was to established a diagnostic nomogram model utilising endoscopic ultrasound (EUS) images to effectively distinguish small gGISTs from non-gGISTs.

Methods

We conducted a multicentre retrospective study of consecutive patients who underwent endoscopic resection (ER) for gastric SMTs at two centres from March 2020 to June 2025. Clinical data, EUS characteristics and pathological features were collected and analysed. A nomogram model for the diagnosis of small gGISTs was established, followed by internal and external validation.

Results

A total of 496 patients were included in this study. The independent predictors of gGIST diagnosis were age ≥ 60 years (OR (95% CI) 2.30 (1.20–4.44), P = 0.013), gastric cardia-fundus/body location (OR (95% CI) 6.09 (1.55–23.98), P = 0.010), and muscularis propria/submucosa origin (OR (95% CI) 6.71 (2.24–20.04), P < 0.001). The AUCs for the nomogram were 0.83 (95% CI 0.78–0.88), 0.81 (95% CI 0.73–0.89), and 0.87 (95% CI 0.81–0.92) in the training, internal validation, and external validation cohorts, respectively. Calibration curves showed excellent agreement between the predicted and actual probabilities for differentiating between small gGISTs and non-gGISTs. Decision curve analysis (DCA) demonstrated favourable clinical applications of the model. The external validation yielded an accuracy of 0.78, a sensitivity of 0.91, and a specificity of 0.71. A subgroup analysis between gGISTs and leiomyomas revealed that the AUC was 0.73 (95% CI 0.63–0.83), with an accuracy of 0.67, a sensitivity of 0.65, and a specificity of 0.74 in the external validation cohort.

Conclusion

Patient age, lesion location, and origin layer were independent diagnostic factors for small gGISTs. The proposed nomogram served as a valuable tool for differentiating between small gGISTs and non-gGISTs.

Graphical abstract