New guidelines for use of endoscopic ultrasound for evaluation and risk stratification of pancreatic cystic lesions may be too conservative

• Published in: Surgical endoscopy Vol. 32; no. 5; pp. 2420 - 2426
• Main Authors: Sahar, Nadav, Razzak, Anthony, Kanji, Zaheer S., Coy, David L., Kozarek, Richard, Ross, Andrew S., Gluck, Michael, Larsen, Michael, Irani, Shayan, Gan, S. Ian
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.05.2018; Springer Nature B.V
• Subjects: Abdominal Surgery; Cysts; Gastroenterology; Gynecology; Hepatology; Medicine; Medicine & Public Health; Proctology; Surgery; EUS; Pancreatic cysts; Guidelines
• ISSN: 0930-2794; 1432-2218
• DOI: 10.1007/s00464-017-5941-y

Background The role of EUS in managing asymptomatic pancreatic cystic lesions (PCLs) remains unresolved. We retrospectively evaluated EUS in risk stratification of PCLs when adhering to the most recent AGA guidelines. Methods Asymptomatic PCLs that were evaluated by EUS from January 2014 to December 2014 were retrospectively reviewed including associated cytology, fluid analysis, and relevant surgical pathology. Cross-sectional imaging reports were reviewed blindly by an expert radiologist using AGA risk stratification terminology. Accepted imaging high-risk features (HRF) included cyst diameter > 3 cm, dilated upstream pancreatic ducts, and a solid component in the cyst. Results We reviewed 125 patients who underwent EUS. Expert review of cross-sectional imaging resulted in a different interpretation 25% of the time including 1 malignant cyst. Ninety-three patients (75%) had no HRFs on cross-sectional imaging; 28 patients (22%) were diagnosed with 1 HRF and 4 patients (3%) had 2 HRFs. Adhering to AGA guidelines using 2 HRF as threshold for use of EUS, the diagnosis of malignant and high-risk premalignant lesions (including pancreatic adenocarcinoma, mucinous cystadenoma, neuroendocrine tumors, and IPMN with dysplasia) had a 40% sensitivity and 100% specificity. Had EUS been utilized based on a threshold of 1 HRF on imaging, malignant and high-risk premalignant lesions would have been identified with 80% sensitivity and 95% specificity. By adding EUS to radiographic imaging, the specificity for detecting carcinomas ( p  = 0.0009) and detection of all premalignant lesions ( p  = 0.003) statistically improved. Furthermore, EUS allowed 14 patients (11%) to avoid further surveillance by lowering their risk stratification. Conclusion EUS remains an essential risk stratification modality for incidental PCLs. Current guideline suggestions of its utility may be too stringent. Our study justifies expert radiology review when managing PCLs. Further studies are required to identify the optimal approach to PCL management.