局部晚期宫颈癌治疗决策评分系统的构建和评价

The establishment and evaluation of scoring system for treatment of locally advanced cervical cancer

目的:筛选局部晚期宫颈癌术后辅助治疗的相关影响因素,并以此为基础构建宫颈癌治疗决策评分系统。方法:回顾分析2004年1月至2017年12月安徽医科大学附属省立医院收治的行手术治疗的671例FIGO分期为Ib2、IIa2和IIb期的宫颈鳞癌患者的临床病理资料,其中术后未辅助化疗/放疗组152例(22.7%),术后辅助治疗组519例(77.3%)。采用单因素和多因素逐步Logistic回归(前进法)分析,根据回归分析纳入变量的回归系数和OR值构建Logistic评分系统和Additive评分系统,评分系统的区分能力采用受试者工作特征曲线评估,评分系统校正能力通过Hosmer-Lemeshow拟合优度检验。结果:术后辅助治疗组和未辅助治疗组宫颈癌患者的婚姻状况、职业、组织学分化程度、肿瘤最大径、血红蛋白浓度、术前影像学、高危HPV及术前治疗方式比较,差异均有统计学意义(P<0.05)。多因素逐步Logistic回归(前进法)分析显示,高危HPV阳性、术前影像学、血红蛋白浓度、肿瘤最大径以及组织分化程度进入回归方程。定义简易Logistic评分:高危HPV阳性11分,术前影像学有危险因素21分,血红蛋白(<110g/L) 8分,组织学分级中分化4分,低分化12分,肿瘤最大径(>5cm) 24分,分值范围0～76分。Logistic评分为19分时,预测术后需辅助治疗的敏感度和特异度分别为62.0%、78.9%,AUC为0.757,校正能力Hosmer-Lemeshow拟合优度检验满意(P> 0.05)。Additive评分:高危HPV阳性3分,术前影像学有危险因素8分,血红蛋白浓度(<110g/L) 2分,组织学分级高分化1分,中分化2分,低分化3分,肿瘤最大径(>5cm) 11分,分值范围1～27分。Additive评分为5分时,预测术后需辅助治疗的敏感度和特异度分别为62.0%、76.3%,AUC为0.755,校正能力Hosmer-Lemeshow拟合优度检验满意(P>0.05)。结论:构建宫颈癌术后辅助治疗的简化Logistic评分系统和Additive评分系统,可预测宫颈癌术后需追加辅助治疗的概率,为临床选择最佳治疗方案提供依据。

Objective:To screen the factors related to adjuvant treatment of locally advanced cervical cancer after surgery,and to construct a decision scoring system for cervical cancer treatment based on these factors.Methods:The clinical and pathological characteristics of 671 patients with cervical squamous cell carcinoma staged FIGO stage Ib2,IIa2 and IIb from Jan.2004 to Dec.2017 were retrospectively analyzed.152 cases(22.7%)in the postoperative adjuvant chemotherapy/radiotherapy group and 519 cases(77.3%)in the postoperative adjuvant treatment group.Univariate analysis and multivariate stepwise logistic regression analysis was used.Logistic scoring system and additive scoring system were constructed based on the regression coefficients and OR values of variables included in the regression analysis.The distinguishing ability of the scoring system was evaluated by receiver operating characteristic curve,and the correction ability of the scoring system passed the hosmer-lemeshow goodness-of-fit test.Results:The differences between the two groups were statistically significant(P<0.05)in marital status,occupation,histological differentiation degree,maximum tumor diameter,hemoglobin concentration,preoperative imaging,high-risk HPV and preoperative treatment methods.Multivariate stepwise logistic regression analysis showed that high-risk HPV was positive,preoperative imaging,hemoglobin concentration,tumor maximum diameter,and tissue differentiation entered the regression equation(P<0.05).Definition of simple logistic score:11 high-risk HPV positive points,21 preoperative imaging risk factors,8 hemoglobin concentration(<110g/L),4 differentiation points in histological grade,12 low differentiation points,and 24 tumor maximum diameter(>5cm).The score ranged from 0 to 76.When the logistic score was 19,the sensitivity and specificity of predicting the need for adjuvant therapy after surgery were 62.0%and 78.9%respectively,and the AUC was 0.757.The goodness of fit test of the correction ability hoster-lemeshow was satisfactory(P>0.05).Additive score:high-risk HPV positive 3 points,preoperative imaging with risk factors 8 points,hemoglobin concentration(<110g/L)2 points,histological grade with high differentiation 1 point,medium differentiation 2 points,low differentiation 3 points,and tumor maximum diameter(>5cm)11 points.The score ranged from 1 to 27 points.When the additive score was 5 points,the sensitivity and specificity of predicting postoperative adjuvant therapy were 62.0%and 76.3%respectively,and AUC was 0.755.The goodness of fit test of the correction ability hoster-leme show was satisfactory(P>0.05).Conclusion:The constructed simplified logistic scoring system and additive scoring system for postoperative adjuvant treatment of cervical cancer can predict the probability of additional adjuvant treatment after cervical cancer surgery and provide the basis for clinical selection of the best treatment plan.