中文摘要：

目的：研究胃肠道间质瘤中是否存在血管生成拟态及临床意义.方法：收集84例CIST标本及临床病理资料．CD31／PAS双重染色结合CD117、CD31免疫组化染色证实VM存在，分析VM与患者临床病理指标之间的关系结果：84例GIST中21例具有VM；核分裂数≥5个／50HPF组和〈5个／50HPF组和有无肝转移组间VM阳性率差别有统计学意义（P=0．000，0．008）。极低危／低危组、中危组和高危组三组VM阳性率分别为5．9%，12．5％，39.5％，三组之间VM阳性率差异有统计学意义（P=0．010，0．020）Kaplan—Meier生存分析提示有VM组和无VM组生存时间差异有统计学意义（P=0．0000）.Cox比例风险模型分析表明有VM、肿瘤大小≥10cm和出血是影响GIST患者预后的危险因素（P=-0．000．0．005，0．032）=结论：GIST中存在VM，VM是影响GIST患者预后的不利因素，有VM的患者易发生肝转移．预后比无VM的患者差、

英文摘要：

Objective： To study vasculogenic mimicry （VM） and its role in gastrointestinal stromal tumors （GISTs）. Methods： Specimens of 84 GIST cases were collected along with their clinicopathologic data. VM channels were detected using CD31/PAS double staining and CD117 and CD31 immunohistochemical staining and were then used to study the correlation between VM and clinicopathologic factors. Results： VM channels were found in 21 of the 84 GIST cases. There was a significant difference in the VM-positive rate between the group with a mitotic rate of greater than or equal to 5/50HPF and the group with a mitotic rate of less than 5/50HPF. There was also a significant difference between the cases with and without liver metastasis （P=0.000, 0.008）. The VM-positive rate in the uhra-low/lowrisk, intermediate-risk and high risk groups was 5.9%, 12.5% and 39.5%, respectively. There was a significant difference when comparing the VM-positive rates between the three groups （P=0.010, 0.020）. Kaplan-Meier survival analysis showed the presence of VM channels had a negative influence on prognosis （P=0.0000）. Cox＇s proportional hazards model analysis indicated that the presence of VM channels, tumor size of 10 cm or larger and hemorrhaging were independent predictors of poor prognosis for the GIST patients （P=0.000, 0.005, 0.032）. Conclusion： VM exists in GIST. VM is an unfavorable factor influencing prognosis of GIST patients, and the patients with VM channels in the tumor are prone to suffer liver metastasis. Prognosis for patients with VM channels in GIST is poorer than that for those without VM.