Endoscopic Ultrasound Features of Gastrointestinal Stromal Cell Tumors Can Be Predictive of Our Ability to Obtain Diagnostic Cytology Via EUS-Guided Fine Needle Aspiration

Abstract

Background: Endoscopic ultrasound (EUS) is frequently used to evaluate suspected subepithelial tumors seen on endoscopy. EUS guided fine needle aspiration (FNA) has been well documented to aid in diagnosis by providing cytologic material. One limitation of EUS-FNA is the frequency of non-diagnostic samples as a result of low cellularity of specimens. The purpose of this study was to determine if EUS features of GISTs are associated with ability to obtain diagnostic cytology via EUS-FNA. We hypothesized that the biologic behavior of a GIST was more predictive of positive cytologic specimens as compared to accessibility of the GIST by EUS. Methods: In a retrospective study, all patients at Massachusetts General Hospital with histologically confirmed c-kit positive GIST who underwent EUS-FNA between 1997-2006 were included. Diagnostic cytology via FNA was defined as a spindled or epithelioid population of cells on smear. EUS images were reviewed by a blinded endosonographer, and evaluated for mass size, shape, location, wall layer, heterogeneity, echogenicity, cystic spaces, lobulation, ulceration, and central umbilication. Needle gauge, # of passes, and presence of a cytologist during the EUS-FNA were also recorded. Results: 27 patients met inclusion criteria; 20 with diagnostic cytology (Group A), and 7 with non-diagnostic cytology (Group B). Table 1 summarizes the main findings. The size, shape, and location of the GIST were predictive of a diagnostic cytology specimen. A trend toward non-diagnostic specimens was seen with higher risk histology, absence of a cytologist, and ulceration, but these were not statistically significant. There was no significant difference between the two groups for the presence of heterogeneity, cystic spaces, echogenicity, lobulation, central umbilication, needle gauge, or # needle passes. Findings: 1) Gastric location other than antrum, smaller size, and round shape are associated with an EUS directed diagnostic cytology specimen. 2) Location in the antrum or duodenum, larger size, and irregular shape are associated with non-diagnostic specimens. 3) The histologic grade of GIST is not predictive of a diagnostic cytology specimen. Conclusion: Accessibility of GISTs is a more important determinant of diagnostic cytology than biologic behavior. Table 1Results Group A (Diagnostic cytology) n Group B (Non-diagnostic cytology n p-value Location in duodenum/antrum 2 (10%) 20 5 (71%) 7 0.005 Size (mm) 36.4 ± 18 20 75.7 ± 49 7 0.005 Irregular shape 1 (5%) 19 3 (60%) 5 0.016 Histology findings of high/intermediate risk 8 (28%) 20 4 (57%) 7 0.66 Presence of cytologist 5 (45%) 11 1 (25%) 4 0.662 Ulceration 5 (26%) 19 3 (50%) 6 0.344 Open table in a new tab