Su1399 DIAGNOSTIC UTILITY OF LIQUID-BASED CYTOLOGY IN EUS-FNA FOR PANCREATIC LESION AND SUBEPITHELIAL TUMOR

Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is widely used in diagnosis of abdominal tumors, especially pancreatic tumors and subepithelial tumors (SET). Although rapid on-site cytopathological evaluation (ROSE) has been reported to improve diagnostic accuracy, it is difficult to routinely performed ROSE due to limitation of manpower in many institutions. Liquid based cytology (LBC) has been well established for gynecological specimen. However, the effect of LBC on EUS-FNA is still unclear. The aim of this study is to evaluate the diagnostic utility of LBC in EUS-FNA for pancreatic lesion and SET. A total of 130 patients who underwent EUS-FNA without ROSE for pancreatic tumors (n=94) and SET (n=36) from April 2009 to September 2017 were retrospectively investigated. LBC was carried out using residual specimens in a liquid medium after removal of core tissues for histological examination. We evaluated the adequacy rates of specimens obtained by EUS-FNA and diagnostic accuracy in LBC and histological examination. We also analyzed risk factors associated with diagnostic accuracy in LBC and histological examination (tumor location, tumor size, needle type, number of needle passes, and the experience with EUS-FNA). Final diagnosis was confirmed by surgical resection or clinical follow-up. The diagnostic adequacy of histological materials in pancreatic lesion was comparable to that in SET (82.9% vs 88.9%). However, the adequate specimen rates of LBC in SET were significantly lower than those in pancreatic lesion (69.4% vs 91.4%, P < 0.05). The diagnostic accuracy of LBC was significantly higher than that of histology in the 63 cases of pancreatic cancer (82.5% vs 65.1%, P < 0.05). On the other hand, in 36 cases of gastrointestinal stromal tumor (GIST), the diagnostic accuracy of LBC was significantly lower than that of histology (64.5% vs 87.1%, P < 0.05). The combination of LBC and histological examination increased the diagnostic accuracy in both pancreatic cancer (92.1%) and GIST (93.5%). Regarding risk factors for misdiagnosis, the less experience with EUS-FNA was associated with the low diagnostic accuracy of histological results (50% vs 77.1%, P < 0.05). Combination of LBC and histological examination without ROSE offers a high diagnostic efficacy for pancreatic cancer and GIST. However, the effect of LBC on EUS-FNA for SET was limited. It is noteworthy that LBC could improve the adequate specimen sampling of EUS-FNA for pancreatic tumors, especially in less-experienced endoscopist.