直肠癌术前放疗敏感性相关分子标志物多因素logistic分析及疗效预测模型建立

Establishment of a logistic regression model for the predicting of radio-sensitivity of preoperative radiotherapy in rectal cancer

目的：研究影响直肠癌患者术前放疗敏感性的相关分子标志物，建立logistic回归模型，通过直肠癌术前分子标志物表达水平预测直肠癌术前放疗的敏感性。方法回顾性分析2010年1月至2015年1月直肠癌术前放疗患者33例，收集所有患者放疗前血清癌胚抗原（CEA）、肿瘤血管内皮生长因子（VEGF）、表皮生长因子受体（EGFR）、Ki-67及胸苷酸合成酶（TS）免疫组织化学表达情况以及放疗前影像资料[CT或磁共振成像（MRI）]、术前临床分期、术后病理分期。根据术后病理评价术前放疗疗效，分为有效（完全缓解及部分缓解）、无效（稳定及疾病进展）。运用SPSS 17.0软件将上述分子标志物表达情况与术前放疗效果进行logistic二元回归分析，建立疗效预测模型。结果经单因素及多因素logistic二元回归分析显示，CEA、VEGF、Ki-67为直肠癌术前放疗敏感性的相关因素。建立的分子标志物预测模型为log P＝1.700-0.276×CEA-0.238×VEGF-0.135×EGFR＋1.377×TS＋0.080×Ki-67。血清CEA水平越高、VEGF的表达越高，患者对放疗的敏感性可能越低；Ki-67表达越高，患者对放疗可能越敏感。结论检测CEA、VEGF、Ki-67的表达水平可能为预测直肠癌术前放疗敏感性相对有效的手段。

Objective To investigate the correlation between the biomarkers related to radio-sensitivity and preoperative radiotherapy in rectal cancer patients, and to establish a logistic regression model to predict the effect of the preoperative radiotherapy through detecting the expression levels of the molecular markers. Methods 33 patients with rectal cancer who received preoperative radiotherapy from January 2010 to January 2015 were retrospectively analyzed. Patients＇ information was also collected including the serum level of carcino-embryonic antigen （CEA）, the immune-histochemical expression levels of vascular endothelial growth factor （VEGF）, epidermal growth factor receptor （EGFR）, thymidylate synthase （TS） and Ki-67, and image data （CT or magnetic resonance imaging） before radiotherapy, preoperative clinical staging and the postoperative pathologic staging. According to the postoperative pathological remission, the treatment effects of preoperative radiotherapy included effective （CR＋PR） and ineffective （PD＋SD） were evaluated. The relationship between these molecular markers and the curative effect of preoperative radiotherapy was analyzed by logistic regression analysis using SPSS v17.0 software, and a logistic curative effect prediction model was established. Results As a result of single factor and multiple factors logistic binary regression analysis, CEA, VEGF and Ki-67 were recognized as the interested factors for the radio-sensitivity predicting in patients with rectal cancer who received preoperative radiotherapy. A molecular markers predictive model for radio-sensitivity in preoperative radiotherapy in rectal cancer is as follow： log P=1.700-0.276×CEA-0.238×VEGF-0.135 ×EGFR＋1.377 ×TS＋0.080 ×Ki-67. Serum CEA level and the expression of VEGF might associate with radio-resistant, and the expression of Ki-67 might associate with better reaction to preoperative radiotherapy. Conclusion The levels of serum CEA, VEGF and Ki-67 may be the predictors of radio-sensitivity in rectal cancer patients who received preoperative radiotherapy.