Objective: To investigate the postoperative prognostic factors of non-metastatic colorectal cancer (non-mCRC), and construct a prognostic prediction model. Methods: A total of 846 patients with colorectal cancer who were admitted to the Cancer Hospital, Chinese Academy of Medical Sciences from July 1, 2014 to December 31, 2016 were included in the study. There were 314 patients in the metastatic colorectal cancer (mCRC) group and 532 patients in the non-mCRC group. The data of clinical characteristics, preoperative blood routine and common serum tumor markers for CRC tests were collected retrospectively. The disease-free survival time (DFS) data of patients in non-mCRC group were obtained by follow-up. Univariate and multivariate Cox regression analyses were used to clarify the independent risk factors of DFS, and then these factors were included to construct a nomogram prediction model. The concordance index (C index), receiver operating characteristic curve (ROC) and calibration curve were used to evaluate the performance of the model. Results: Platelet/lymphocyte ratio (PLR), neutrophil/lymphocyte ratio (NLR), carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9) and carbohydrate antigen 242 (CA242) in the mCRC group were higher than those of the non-mCRC group, while the lymphocyte/monocyte ratio (LMR) was lower than that of the non-mCRC group (P<0.05). ROC analysis showed that the area under curve (AUC) of CEA, CA19-9, CA242, NLR, LMR and PLR for the diagnosis of mCRC were 0.775, 0.716, 0.712, 0.607, 0.591 and 0.556, respectively. Multivariate Cox regression analysis demonstrated that age, perineural invasion, pN stage and preoperative CA242 level were independent risk factors for DFS of non-mCRC patients (P<0.05). Based on this, a nomogram prediction model predicting 3 years of DFS for non-mCRC patients was constructed, its C index and AUC for non-CRC prognostic prediction were 0.710 and 0.733, respectively, higher than 0.696 and 0.701 of AJCC 7th edition TNM staging system. The calibration curve of nomogram showed that the predicted DFS rate was consistent with the actual DFS rate. Conclusions: Age, perineural invasion, pN stage and preoperative CA242 level are independent risk factors for 3-year DFS of non-mCRC patients. The nomogram prediction model constructed based on these four indictors has a good predictive performance and may provide prognosis evaluation reference for the patients with non-mCRC.
目的： 探讨非转移性结直肠癌(non-mCRC)手术治疗后的预后影响因素，并构建预后预测模型。 方法： 纳入2014年7月1日至2016年12月31日中国医学科学院肿瘤医院收治的结直肠癌患者846例，其中转移性结直肠癌(mCRC)组314例，non-mCRC组532例。回顾性收集患者的临床资料、术前血液常规和结直肠癌常见肿瘤标志物检查结果。通过随访获得non-mCRC组患者的无病生存时间(DFS)数据。采用单因素和多因素Cox比例风险模型回归分析明确non-mCRC术后DFS的独立影响因素，将这些独立影响因素纳入构建列线图预测模型。采用C指数、受试者工作特征曲线(ROC)及校准曲线评价该模型的性能。 结果： mCRC组的血小板/淋巴细胞比值(PLR)、中性粒细胞/淋巴细胞比值(NLR)、癌胚抗原(CEA)、糖类抗原19-9(CA19-9)和糖类抗原242(CA242)均高于non-mCRC组，淋巴细胞/单核细胞比值(LMR)低于non-mCRC组(均P<0.05)。ROC曲线分析显示，上述6项指标诊断结直肠癌转移的曲下面积(AUC)从高到低依次为CEA(AUC＝0.775)、CA19-9(AUC＝0.716)、CA242(AUC＝0.712)、NLR(AUC＝0.607)、LMR(AUC＝0.591)和PLR(AUC＝0.556)。多因素Cox回归分析显示，年龄、神经侵犯、pN分期和CA242是non-mCRC患者术后DFS的独立影响因素(均P<0.05)，据此构建预测non-mCRC患者术后3年DFS的列线图预测模型。内部验证结果显示，该模型预测non-mCRC患者术后3年DFS的C指数为0.710，ROC曲线的AUC为0.733，优于美国癌症联合委员会第7版TNM分期系统(C指数为0.696，AUC为0.701)，具有较好的区分度。校准曲线显示，该模型的预测结果与实际观测结果具有较好的一致性。 结论： 年龄、神经侵犯、pN分期和CA242是non-mCRC患者术后DFS的独立影响因素。基于这4个指标建立的non-mCRC患者术后DFS列线图预测模型具有较好的预测性能，可为结直肠癌患者的预后评估提供参考。.
Keywords: Colorectal neoplasms; Disease-free survival; Nomogram; Tumor marker.