宫颈癌术后患者列线图预测模型及危险分层系统构建

目的 构建宫颈癌术后患者列线图预测模型，基于列线图个体得分建立危险分层系统。方法 通过搜索美国SEER （Surveillance，Epidemiology，and End Results）数据库中1973—2015年的6 835例宫颈癌术后患者数据构建预测模型，同时选取120例于苏州大学附属第二医院接受宫颈癌手术的患者作为外部验证队列。通过单因素和多因素的Cox回归筛选预后因子并构建列线图，基于列线图模型建立危险分层系统。结果 Cox回归分析显示诊断年龄、人种、组织学分级、T分期、N分期、淋巴结清扫状况、肿瘤大小、肿瘤浸润深度是宫颈癌术后患者的独立预后指标。由此构建的列线图模型的一致性指数在建模队列、内部验证队列和外部验证队列分别为0.824、0.814、0.730，校准曲线显示模型预测效果与实际生存情况基本相符，危险分层系统能区分不同FIGO分期患者的生存情况（均P<0.05）。结论 本研究所建立的列线图模型能有效预测宫颈癌术后患者预后，基于该列线图预测模型的危险分层系统对区分高危患者具有一定临床价值。

Development of a prognostic nomogram and risk stratification system for cervical carcinoma after operation

Objective To construct a nomogram predictive model for postoperative cervical cancer patients, and establish a risk strat-ification system based on the individual scores of the nomogram. Methods The data of 6, 835 postoperative cervical cancer patients in the US SEER (Surveillance, Epidemiology, and End Results) database from 1973 to 2015 was collected to construct a predictive model, and 120 patients who underwent cervical cancer surgery at the Second Affiliated Hospital of Soochow University were selected as an external validation cohort. Through univariable and multivarible Cox regression analysis, prognostic factors were screened out to draw a nomogram, and a risk stratification system was established based on the prognostic risk score of the nomogram. Results The Cox regression analysis showed that the age at diagnosis, race, histological grade, T stage, N stage, lymph node dissection status, tumor size and depth of tumor invasion were independent of prognostic factors for patients with cervical cancer after surgery. The C-index of nomogram constructed in the training cohort, internal verification cohort and external cohort were 0.824, 0.814, and 0.730, respectively. The calibration curve showed that the model prediction was basically consistent with the actual survival situation, and the risk stratification system could clearly distinguish the survivals of patients with different FIGO stages (all P<0.05). Conclusions The nomogram can effectively predict the postoperative prognosis of patients with cervical cancer after surgery. The risk stratification system based on the nomogram prediction model has clinical value in distinguishing high-risk patients.