Can EUS-FNA cytology reliably classify gastric submucosal spindle cell tumors?

Background: Cytological diagnosis using endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) for gastric submucosal spindle cell tumors, such as gastrointestinal stromal tumors (GISTs), leiomyomas, and schwannomas, is challenging because of their similar morphological characteristics. Objective and Materials: To clarify the cytological differential points, we reviewed the EUS-FNA cytology specimens of GISTs (37 cases), leiomyomas (11 cases), and schwannomas (4 cases). Method: Twelve cytomorphological features were evaluated: lymphocytes, crushed nuclei, naked spindle nuclei, mast cell, length of the streaming arrangement, cellularity, nuclei at the cluster margin (nuclei located at the periphery of the cell cluster), peripheral feathering (loosely aggregated cells at the margin of a cell cluster tended to taper like feathers), metachromasia, wavy nuclei, fishhook-type nuclei, and anisonucleosis. Results: Among these features, lymphocytes, naked spindle nuclei, length of the streaming arrangement, cellularity, nuclei at the cluster margins, peripheral feathering, and anisonucleosis were statistically significant for differentiation. Based on these findings, we developed an algorithm for cytodiagnosis. The algorithm was taught to four cytologists, and the interobserver agreement and correct diagnosis rates were compared before and after education, which showed a significant improvement. Discussion: The histological types of gastric submucosal spindle cell tumors can be estimated using this algorithm for EUS-FNA cytology. Furthermore, this algorithm can be applied for cytological diagnosis at bedside during rapid on-site evaluation.