Background and study aims: Endoscopic ultrasonography (EUS) has been shown to be the most accurate test for locoregional staging of upper gastrointestinal tumors; however, recent studies have questioned its accuracy level in daily clinical application. The present retrospective study analyzes the accuracy of EUS in guiding interdisciplinary treatment decisions.
Patients and methods: 123 primarily operated patients (63 % men, mean age 61.4 years) were included; only cases with tumor-free resection margins and without evidence of distant metastases were selected. EUS and histopathological findings were compared. Main outcome parameter was the distinction between tumors to be primarily operated (T1 /2N0) and those to be treated by neoadjuvant or perioperative chemotherapy (T3/4, or any N + ), based on an assumed algorithm for treatment stratification.
Results: Overall staging accuracy of EUS was 44.7 % for T and 71.5 % for N status irrespective of tumor location. Overstaging was the main problem (44.9 % for T, 42.9 % for N staging). The overall EUS classification was correct in 79.7 % (accuracy), with a sensitivity 91.9 % and specificity 51.4 %; only 19 out of 37 cases with histopathological T1/2N0 were correctly classified by EUS. Positive and negative predictive values of EUS in diagnosing advanced tumor stage for assignment to neoadjuvant therapy were 81.4 % and 73.1 %, respectively.
Conclusions: Whereas EUS has a high sensitivity in the diagnosis of locally advanced gastric cancer, endosonographic overstaging of T2 cancers appears to be a frequent problem. EUS stratification between local (T1 /2N0) and advanced (T3/4 or any N + ) tumors would thus result in incorrect assignment to neoadjuvant treatment in half of cases.
