结肠癌切除术患者预后列线图模型及危险分层系统构建与验证

  目的  基于大样本量，构建个体化预测模型及危险分层系统。  方法  从美国SEER临床数据库中，筛选结肠癌术后患者，进行模型构建，并筛选一组独立的中国人群，用于外部验证。经过单因素与多因素Cox回归分析，筛选出独立预后指标，并全部纳入用于构建列线图预测模型。通过计算一致性指数（C-index）及绘制校准曲线，检验模型准确性。  结果  列线图模型共纳入11个独立预后因子，C-index在训练组、内部验证组及外部验证组分别为0.768，0.761和0.759，均>0.7，且优于第7版美国癌症联合委员会（AJCC）-TNM分期系统（0.729，0.720，0.735）。校准曲线显示，模型预测效果与实际生存相吻合，进一步验证了模型的区分及校准能力。通过决策树分析，依据模型预测个体风险评分，进行危险分层，模型的实际应用价值得到确定。  结论  该列线图预测模型能够较准确预测结肠癌术后患者预后状态，并较传统TNM分期系统有所改善，基于预测模型的危险分层系统，能够更好地区分高危患者，并指导选择临床治疗措施。 

Development of a prognostic nomogram and risk stratification system for colon cancer after resection

  Objective  The mainstay treatment for colon cancer is surgical resection. However, factors associated with post-surgical mortality are poorly understood. We sought to characterize the determinants of survival in a large cohort of patients with colon cancer.  Methods  Patients with colon adenocarcinoma who underwent surgery were selected from the Surveillance, Epidemiology, and End Results (SEER) database for model construction, and an independent Chinese cohort was selected for external validation.  Results  Significant factors identified through multivariate Cox regression analysis were incorporated into the nomogram. Discrimination and calibration were evaluated using a relative index. Based on the nomogram, we performed a recursive partitioning analysis (RPA) for risk stratification.  Conclusions  The nomogram provided clear prognostic superiority over the traditional tumor- node- metastasis (TNM) system. RPA based on a clinical nomogram is suitable for risk stratification of long-term survival. This system could help clinicians make individual survival predictions for patients with colon cancer and help provide necessary treatment recommendations.