Reliability Of Eus In Exclusion Of Pancreatic Cancer-Results Of The Hamburg-Eppendorf Study

Introduction: Endoscopic ultrasound (EUS) has greatly increased the sensitivity and specificity of pancreatic cancer detection. However, not many studies have dealt with the reliability of EUS in excluding pancreatic cancer. Aim: To study the reliability of EUS in excluding pancreatic cancer. Method: 593 patients with a suspicion of pancreatic cancer referred to our center for EUS from 2002 and 2003 were analyzed. All these patients underwent EUS examination with a linear EUS system (GF UC140P-AL5; Olympus Optical Co., Ltd., Tokyo, Japan and SSD-5000, Aloka Co. Ltd., Tokyo, Japan) by 4 experienced examiners (SS, OS, IH, SU). The pancreas was examined systematically for signs of any lesions. If a suspicious lesion was imaged, EUS-guided fine needle aspiration (EUS-FNA) was performed with a 22G Echotip Ultra needle (Wilson-Cook, Winston-Salem, NC, USA). All patients were followed-up by a single physician by telephone calls to the primary physicians or the patients. Results: EUS-FNA diagnosed 181 malignancies; 412 patients with normal EUS findings or who were negative for malignancy by FNA were retrospectively analyzed (median age 60.9 years; 224M:188F). The median follow-up was 14 months (range 6-42). Patients without visible EUS lesions [group A] (n = 253) were further subdivided into those with “normal” pancreas [N] (n = 122) and those with signs of chronic pancreatitis [CP] (n = 131). Patients with EUS-FNA negative “benign” lesions [group B] (n = 159) were further divided into those with cystic [CL] (n = 50) and solid lesions [SL] (n = 109). The solid lesions were further defined as either “circumscribed” [CSL] (n = 49), “non-circumscribed” [NSL] (n = 25) or “lobulated” [LSL] (n = 35). Two patients in Group A and 17 patients in Group B were finally found to have pancreatic adenocarcinoma (p < 0.001). Subanalysis of group A revealed pancreatic adenocarcinoma in 1.5% of CP patients. In Group B, there were no patients with pancreatic adenocarcinoma in the CL group. However 18.4%, 20% and 8.6% patients developed pancreatic adenocarcinoma in the CSL, NSL, LSL groups respectively (CL vs SL p < 0.01; CSL vs NSL vs LSL p = NS). The sensitivity, specificity, positive predictive value and negative predictive value of EUS in excluding pancreatic adenocarcinoma is 90.5%, 100%, 100% and 95.4% respectively. Conclusion: Pancreatic adenocarcinoma can be reliably excluded in patients with a normal EUS appearance. In the subset of patients with abnormal findings such as chronic pancreatitis and “benign” solid lesions, there is still a risk of malignancy and hence these patients should be closely followed-up.