BACKGROUND: Risk assessment of gastrointestinal stromal tumor (GIST) is challenging on cytology specimens. This stud y aims to determine whether Ki-67 index evaluated on fine-needle aspiration (FNA) specimens can correlate with the mitotic rate of GIST in surgical specimens and provide further risk assessment. METHODS: Cases with cell blocks containing adequate tumor cells and surgical resections were included. Ki-67 immunostain was retrospectively performed on cell block sections, and Ki-67 index was calculated on the "hot spot" areas. RESULTS: This study included 50 GIST cases from stomach (n = 45; 90%), duodenum (n = 4; 8%), and distal esophagus (n = 1; 2%). The tumor size ranged from 1.5 cm to 21 cm (mean, 5.4 cm). Based on the mitotic count, 37 GISTs (74%) had low mitotic rate (LMR) and 13 GISTs (26%) had high mitotic rate (HMR). The spindle cell, epithelioid, and mixed types accounted for 60%, 14%, and 26% of GIST, respectively. Ki-67 index counted on cell block sections correlated well with mitotic count evaluated in surgical specimens (r = 0.8031). Mean Ki-67 index was higher in HMR than LMR groups (3.5% vs. 1%, p < .001). The receiver operating characteristic curve using Ki-67 index to predict mitotic rate was further analyzed, and area under the curve was 0.839. Using a cutoff of 2.5% yielded a sensitivity of 70% at 92% specificity.
CONCLUSIONS:This study demonstrates good correlations between Ki-67 index and mitotic count or risk stratification, suggesting that Ki-67 index evaluated on cytology specimens may offer a promising approach to preoperatively predict the mitotic rate and risk of GIST.