BACKGROUND: Endoscopic ultrasound is widely used following endoscopy for evaluation of suspected submucosal lesions and may guide further management of patients. PATIENTS AND METHOD: A total of 181 consecutive patients with suspected submucosal lesion in the upper gastrointestinal tract were diagnosed by endoscopic ultrasound between 1990-97. We evaluated: 1) the potential of endoscopic ultrasound criteria to predict histological type of submucosal lesions in 69 patients with available histology, 2) the ability of endoscopic ultrasound alone or with clinical presentation, to predict malignancy in 86 patients with available histology or follow-up of >12 months. RESULTS: Sensitivity and specificity for diagnosing 44 gastrointestinal stromal tumours were 95 and 72%, respectively, while 25 miscellaneous lesions were diagnosed correctly in only 56% by endoscopic ultrasound. Diagnosis of malignancy, using any two of three endoscopic ultrasound criteria (heterogeneous echotexture, size >3 cm, irregular margins) showed a sensitivity of 80% and specificity of 77%, giving accurate endoscopic ultrasound diagnosis in 16/20 malignant and 51/66 benign submucosal lesion. Heterogeneous echotexture, size >3 cm, and irregular margins showed a relative risk of 7.2, 5.4 and 4.6, respectively, for presence of malignancy. The presence of symptoms, potentially suggesting malignancy (dysphagia, gastrointestinal bleeding, pain and weight loss), had a relative risk of 4.2, however this did not increase the accuracy of diagnosing malignancy based on endoscopic ultrasound criteria alone. CONCLUSION: The accuracy of endoultrasound is high in diagnosing gastrointestinal stromal tumours, which show a significant potential of malignancy. Endoscopic ultrasound morphology appears to be helpful in selection of patients for surgical or conservative treatment. The accuracy of endoscopic ultrasound in differential diagnosis of non-gastrointestinal stromal tumour lesions is limited.
BACKGROUND: Endoscopic ultrasound is widely used following endoscopy for evaluation of suspected submucosal lesions and may guide further management of patients. PATIENTS AND METHOD: A total of 181 consecutive patients with suspected submucosal lesion in the upper gastrointestinal tract were diagnosed by endoscopic ultrasound between 1990-97. We evaluated: 1) the potential of endoscopic ultrasound criteria to predict histological type of submucosal lesions in 69 patients with available histology, 2) the ability of endoscopic ultrasound alone or with clinical presentation, to predict malignancy in 86 patients with available histology or follow-up of >12 months. RESULTS: Sensitivity and specificity for diagnosing 44 gastrointestinal stromal tumours were 95 and 72%, respectively, while 25 miscellaneous lesions were diagnosed correctly in only 56% by endoscopic ultrasound. Diagnosis of malignancy, using any two of three endoscopic ultrasound criteria (heterogeneous echotexture, size >3 cm, irregular margins) showed a sensitivity of 80% and specificity of 77%, giving accurate endoscopic ultrasound diagnosis in 16/20 malignant and 51/66 benign submucosal lesion. Heterogeneous echotexture, size >3 cm, and irregular margins showed a relative risk of 7.2, 5.4 and 4.6, respectively, for presence of malignancy. The presence of symptoms, potentially suggesting malignancy (dysphagia, gastrointestinal bleeding, pain and weight loss), had a relative risk of 4.2, however this did not increase the accuracy of diagnosing malignancy based on endoscopic ultrasound criteria alone. CONCLUSION: The accuracy of endoultrasound is high in diagnosing gastrointestinal stromal tumours, which show a significant potential of malignancy. Endoscopic ultrasound morphology appears to be helpful in selection of patients for surgical or conservative treatment. The accuracy of endoscopic ultrasound in differential diagnosis of non-gastrointestinal stromal tumour lesions is limited.