1. Image analysis is an excellent tool for quantifying Ki-67 to predict the prognosis of gastrointestinal stromal tumor patients
- Author
-
Shintaro Sugita, Hiroshi Hirano, Terufumi Kubo, Hiroyuki Hisai, Noriaki Kikuchi, Keiko Segawa, Takayuki Nobuoka, Kentaro Yamashita, Tadashi Hasegawa, Yutaka Hatanaka, Taro Sugawara, Hiromi Fujita, Yoshihiro Matsuno, and Yumika Ito
- Subjects
medicine.medical_specialty ,Pathology ,Mitotic index ,Multivariate analysis ,biology ,Receiver operating characteristic ,GiST ,business.industry ,General Medicine ,medicine.disease ,Primary tumor ,Pathology and Forensic Medicine ,03 medical and health sciences ,0302 clinical medicine ,030220 oncology & carcinogenesis ,Ki-67 ,biology.protein ,Medicine ,030211 gastroenterology & hepatology ,Radiology ,Stromal tumor ,business ,Survival analysis - Abstract
We investigated the quantification of Ki-67 staining using digital image analysis (IA) as a complementary prognostic factor to the modified National Institutes of Health (NIH) classification in patients with gastrointestinal stromal tumor (GIST). We examined 92 patients, focusing on the correlation between age, sex, primary tumor site, tumor size, predominant histologic type, mitotic index, modified NIH classification (low/intermediate vs high), Ki-67 quantitation, and recurrence-free survival (RFS). We compared two IA processes for whole slide imaging (WSI) and manually captured image (MCI) methods. A Ki-67 quantitation cutoff was determined by receiver operator characteristics curve analysis. In the survival analysis, the high-risk group of a modified NIH classification, a mitotic count >5 per 20 high-powered fields, and Ki-67 cutoffs of ≥6% and ≥8% obtained by IA of the WSI and MCI methods, respectively, had an adverse impact on RFS. On multivariate analysis, each Ki-67 quantitation method strongly predicted prognosis, more strongly than the modified NIH classification. In addition, Ki-67 quantitation using IA of the MCI method could stratify low or intermediate risk and high risk GIST patients. Thus, IA is an excellent tool for quantifying Ki-67 to predict the prognosis of GIST patients, and this semiautomated approach may be preferable for patient care.
- Published
- 2017
- Full Text
- View/download PDF