Background: Submucosal tumors (SMTs) comprise both benign and malignant lesions, and most of the gastric lesions tend to be malignant. The addition of EUS-guided FNA (EUS-FNA) has the potential to improve this distinction, but published series are limited. Objective: To evaluate the yield of EUS-FNA in gastric SMTs with referral to a criterion standard final diagnosis. Design: Retrospective study. Setting: Tertiary-care referral center. Patients: This study involved 141 consecutive patients with gastric SMTs, who underwent EUS-FNA from January 2000 to December 2008. Immunohistochemical staining with c-kit, CD34, actin, and S-100 antibodies was done if a spindle cell tumor was found. Based on FNA sample adequacy, and whether a specific diagnosis could be established, EUS-FNA results were categorized as diagnostic, suggestive, or nondiagnostic. The criterion standards for final diagnosis were the surgical histopathological results or the follow-up course for malignant, inoperable cases. Intervention: EUS-FNA. Main Outcome Measurements: Diagnostic yield of EUS-FNA and factors related to sampling adequacy for cytological and immunohistochemical evaluation. Results: A total of 141 patients (52% female, mean age 56.7 years) underwent EUS-FNA (range 1-5 passes). The overall results of EUS-FNA were diagnostic, suggestive, and nondiagnostic in 43.3%, 39%, and 17.7% of cases, respectively. Adequate specimens were obtained in 83% of cases, and 69 cases (48.9%) had a definitive final diagnosis. The most common gastric SMT was GI stromal tumor (59.5%). EUS-FNA results were 95.6% accurate (95% confidence interval [CI], 87.5%-99%) for the final diagnosis and 94.2% (95% CI, 85.6%-98.1%) accurate for differentiating potentially malignant lesions. A heterogeneous echo pattern was the only independent predictor for sampling adequacy (adjusted odds ratio 6.15; P = .002). There were no procedure-related complications. Limitations: Possibility of selection bias. Conclusion: EUS-FNA is an accurate method for diagnosis of gastric SMTs and for differentiating malignant lesions.
All Science Journal Classification (ASJC) codes
- Radiology Nuclear Medicine and imaging