基于Cox回归联合风险评分构建宫颈癌患者术后复发转移进展预测模型  

Prediction model of postoperative recurrence and metastasis in cervical cancer patients based on Cox regression combined risk score

作　　者：杜鸣宇 朱必清[1] 何丹[1] 王海蓉[1] 武渊[2] 李倩[1] DU Mingyu;ZHU Biqing;HE Dan;WANG Hairong;WU Yuan;LI Qian(Jiangsu Cancer Hospital&Jiangsu Institute of Cancer Prevention&Cancer Hospital Affiliated to Nanjing Medical University,Nanjing,Jiangsu210009,China)

机构地区：[1]江苏省肿瘤医院·江苏省肿瘤防治研究所·南京医科大学附属肿瘤医院放疗科,江苏南京210009 [2]江苏省肿瘤医院·江苏省肿瘤防治研究所·南京医科大学附属肿瘤医院内科,江苏南京210009

出　　处：《中华肿瘤防治杂志》2024年第17期1080-1086,共7页Chinese Journal of Cancer Prevention and Treatment

摘　　要：目的分析影响宫颈癌患者术后复发转移进展的危险因素,并基于Cox回归联合风险评分构建预测模型。方法回顾性分析2014-06-01-2015-12-31就诊于江苏省肿瘤医院初次确诊为宫颈癌且接受宫颈癌根治术的152例患者临床资料,随机数字表法以2∶1的比例分为建模组(102例)和验证组(50例)。多因素Cox回归分析影响建模组患者术后复发转移进展的危险因素;构建风险评分模型;受试者工作特征(ROC)曲线和校准曲线对模型进行内外部评价。结果诊断分期4期(HR=12.972,95%CI:3.621~46.472)、低分化(HR=3.535,95%CI:1.258~4.012)、腺癌(HR=2.366,95%CI:1.587~5.236)和淋巴结转移(HR=5.851,95%CI:3.524~7.689)是影响宫颈癌患者术后5年复发转移进展的独立危险因素(P<0.05);风险评分模型结果显示,低危组(45例)评分为0~5分,中危组(36例)评分为6~10分,高危组(21例)评分为11~14分,3组患者复发进展率比较,差异有统计学意义,χ^(2)=10.338,P=0.006;3组患者5年复发转移进展率差异有统计学意义,χ^(2)=5.294,P=0.021。宫颈癌患者术后复发转移进展预测模型建模组和验证组ROC曲线下面积分别为0.886(95%CI:0.813~0.905)和0.884(95%CI:0.811~0.901),模型准确度和区分度良好。结论诊断分期4期、低分化、腺癌和淋巴结转移与宫颈癌患者术后复发转移进展密切相关,Cox回归联合风险评分模型对预测患者术后复发转移进展预测具有良好效果。Objective The risk factors affecting the progression of postoperative recurrence and metastasis of cervical cancer patients were analyzed and the prediction model was constructed based on Cox regression combined risk score.Methods A total of 152 patients with cervical cancer who received radical cervical cancer surgery in Jiangsu Cancer Hospital from June 1,2014 to December 31,2015 were retrospectively selected and divided into modeling group(n=102)and verification group(n=50)by random number table method in a 2:1 ratio.Multivariate Cox regression analysis was performed to an-alyze the risk factors of postoperative recurrence and metastasis in the modeling group.A risk score model was developed;Receiver operating characteristic curve(ROC)and calibration curve were used to evaluate the model internally and exter-nally.Results Diagnostic stage 4(HR=12.972,95%CI:3.621-46.472)poorly differentiated(HR=3.535,95%CI:1.258-4.012),adenocarcinoma(HR=2.366,95%CI:1.587-5.236),and lymph node metastasis(HR=5.851,95%CI:3.524-7.689)were independent risk factors for recurrence and metastasis 5 years after surgery in cervical cancer pa-tients(P<0.05).The results of the risk score model showed that the low-risk group(45 cases)scored 0-5 points,the medium-risk group(36 cases)scored 6-1o points,and the high-risk group(21 cases)scored 11-14 points.The recur-rence and progression rates of the three groups were significantly different(χ^(2)=10.338,P=0.006).There was signifi-cant difference in the 5-year recurrence and metastasis progression rate among the 3 groups.χ^(2)=5.294,P=0.021.The ROC curve AUC values of the model group and the verification group were 0.886(95%CI:0.813-0.905)and 0.884(95%CI:0.811-0.901),respectively,indicating good accuracy and differentiation of the model.Conclusion Diagnostic stage 4,low differentiation,adenocarcinoma and lymph node metastasis are closely related to postoperative recurrence and metastasis of cervical cancer patients.Cox regression combined risk score model has a good effect on predicting posto

关 键 词：宫颈癌 风险评分模型 进展期 复发转移 COX回归分析 

分 类 号：R737.3[医药卫生—肿瘤]