CEH-EUS for Differentiating GISTs and Leiomyomas: A Multicenter Prospective Self-Controlled Study

This study adopts a prospective, multicenter, self-controlled (pre-post) design. Patients with upper gastrointestinal subepithelial lesions (SELs) detected under white-light endoscopy will be consecutively enrolled. Each patient will first undergo endoscopic ultrasound (EUS) examination, during which lesion characteristics-including tumor size, margins, internal echogenicity, and originating layer-will be recorded. The contrast-specific extended pure harmonic detection (ExPHD) mode will then be activated, the mechanical index (MI) will be set to 0.3, and 2.4 mL of contrast agent (SonoVue) will be injected intravenously, followed by a 5 mL saline flush. Real-time dynamic imaging will be recorded for 120 seconds, capturing the arterial, venous, and delayed phases.

Histopathological results from surgical or endoscopic resection will serve as the reference standard. The diagnostic performance of EUS and contrast-enhanced EUS (CE-EUS) for differentiating gastrointestinal stromal tumors (GISTs) from leiomyomas and for predicting GIST risk stratification will be compared.

Patients meeting the inclusion criteria will be consecutively recruited. The primary outcome indicators are the sensitivity of CEH-EUS versus EUS in the differential diagnosis between GIST and leiomyoma, as well as in risk stratification of GIST. A prospective, self-controlled design will be used, with a type I error (α) of 0.05 (two-sided) and type II error (β) of 0.20, yielding a power of 0.80. Based on prior studies, the sensitivity of CEH-EUS for differentiating GISTs from leiomyomas is 87%, while that of EUS is approximately 72%-73%. Using McNemar's test for paired binary data (e.g., the same subject evaluated by two methods for positive diagnosis), the minimum sample size required for GIST patients is 121. Given that GISTs account for 67%-68% of all SELs, the required combined sample size of GIST and leiomyoma cases is 178. Accounting for an estimated 20% dropout rate, the final minimum sample size is 222 cases. For the GIST risk stratification analysis, prior studies report a sensitivity of 93% for CEH-EUS and 80% for EUS. McNemar's test was again used for sample size estimation. To meet the statistical requirements, 110 low-risk GIST cases are needed. Considering that low-risk GISTs comprise about 70% of all GISTs, and factoring in a 20% dropout rate, the total number of prospectively enrolled GIST patients required is 196. Since the risk stratification task requires a larger sample size and GISTs are the target of both diagnostic and stratification objectives, the final planned total sample size is 288 patients with GISTs or leiomyomas, which satisfies the statistical requirements for all primary study endpoints.

The study team will screen patients based on inclusion and exclusion criteria, ensure that all necessary examinations are completed to confirm eligibility, and obtain written informed consent from all prospective participants before conducting any study-related procedures.

This is a purely observational study. No additional interventions will be performed on participants, nor will they incur any extra costs. Patient access to optimal diagnostic or therapeutic options will not be affected. The primary potential risk is the breach of patient privacy. A strict data security and monitoring plan will be implemented, and participants will be informed that their data will be used for clinical research purposes.

The diagnostic performance of EUS and CE-EUS in differentiating GISTs from leiomyomas and in risk stratification of GISTs will be compared, with histopathological diagnosis serving as the gold standard. Diagnostic performance will be evaluated using paired analysis. All statistical tests will be two-sided, with significance defined as P \< 0.05. Continuous variables will be presented as mean ± standard deviation, and categorical variables as counts and percentages.(1) Diagnostic performance: Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and area under the curve (AUC) will be calculated for both CEH-EUS and EUS as interpreted by expert endoscopists. To address class imbalance (e.g., GIST vs. other lesions), F1-score (harmonic mean) and balanced accuracy will also be computed. (2) Continuous variables: Comparisons with baseline will be performed using paired t-tests, analysis of variance (ANOVA), or rank-sum tests, as appropriate to data distribution. (3) Categorical variables: Group comparisons will use Chi-square tests (including Cochran-Mantel-Haenszel Chi-square) or Fisher's exact test. (4) Baseline comparability: Demographic and baseline characteristics will be compared using independent t-tests or Chi-square tests to assess balance between groups. (5) Effectiveness analysis: The primary effectiveness endpoint is the diagnostic accuracy for upper gastrointestinal subepithelial lesions. Differences in proportions and the Youden index will be compared using approximate Z-tests or Chi-square tests, with adjustment for center effects. (6) Statistical software: All statistical analyses will be performed using SPSS version 26.0.

• Comparison of CEH-EUS Imaging Characteristics Between GIST and Leiomyoma [Time frame: Within 1 month after histopathological diagnosis]
• Quantitative TIC Parameter Analysis for Differential Diagnosis and Risk Stratification [Time frame: Within 1 month after histopathological diagnosis]
• Correlation of CEH-EUS Perfusion Parameters with Tumor Type and Risk Classification [Time frame: Within 1 month after histopathological diagnosis]