Su1438 DIAGNOSTIC ABILITY OF EUS-FNA FOR GASTROINTESTINAL SUBMUCOSAL TUMOR

Abstract

With gastrointestinal submucosal tumor (SMT), differentiating gastrointestinal stromal tumor (GIST) and malignant lymphoma from other benign tumors is clinically important. Distinguishing between pathologies based on imaging alone is difficult, and endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is useful. However, adequate specimen rate and the diagnostic ability with EUS-FNA are slightly lower for SMT than for pancreatic mass lesions and lymph nodes. To improve the adequate specimen rate and correct diagnosis rate of EUS-FNA for SMT, we examined correlations with lesion site, tumor diameter and puncture needle. We retrospectively analyzed 78 consecutive cases (79 lesions: esophagus, n=7; stomach, n=69; duodenum, n=1; rectal, n=2) with SMTs who had undergone EUS-FNA at our hospital between July 2013 and October 2017. The overall adequate specimen rate by EUS-FNA was 82.3%. Adequate samples were obtained for 100% of esophageal lesions, 81.2% of stomach lesions, 0% of duodenal lesions, and 100% of rectal lesions. In terms of tumor diameter, adequate samples were obtained from 73.1% (19 of 26) of lesions <20 mm in diameter, and 86.8% (46 of 53) of lesions ≥20 mm in diameter. Among the 79 lesions, a definitive final diagnosis was obtained for 36 (45.6%; 30 cases were surgically resected, and 6 cases were confirmed on follow-up to be malignant lymphoma or GIST). Of the 36 cases with definitive final diagnosis, 27 were GISTs, 3 were schwannomas, 1 was malignant lymphoma, 1 was ectopic pancreas, 1 was lymph node metastasis, 1 was metastatic gastric cancer, and 2 were others. The accuracy of EUS-FNA was 83.3%, and was relatively high at 96.3% for GIST. For GIST <20 mm in diameter, 3 of 5 cases showed a homogenous hypoechoic pattern on EUS, and differentiation from imaging alone was difficult. When examined by puncture needle separately, the adequate specimen rate with a conventional-type 22-G needle was 72.7% (40 of 55) in total and 60% (12 of 20) in lesions <20 mm in diameter, whereas use of a 22-G Franseen needle yielded adequate specimen rates of 88.9% (16 of 18) overall and 83.3% (5 of 6) in lesions <20 mm in diameter, representing relatively high levels (p = 0.21, 0.38). Usually, as tumor size increases, the postoperative metastatic rate increases in GIST. Heterogeneous internal echogenicity, ulceration, cystic changes, and presence of highly echoic spots are known as useful for diagnosing GIST, but distinguish between GIST and other diseases based on imaging alone was difficult in this study. Immunohistochemical diagnosis by EUS-FNA is necessary even for small lesions, and improvements in the adequate specimen rate are required. In EUS-FNA for SMT, a 22-G Franseen needle may be the first choice, but further accumulation of cases is necessary in the future.