Page 99 - 南京医科大学自然版

第45卷第10期 南京医科大学学报（自然科学版）
2025年10月 Journal of Nanjing Medical University（Natural Sciences） ·1467 ·

·临床研究·

早期宫颈鳞癌盆腔淋巴结转移的预测模型构建

李俊玫 ，王 露 ，毛熙光 ，廖思静 3
2*
1
1
西南医科大学附属医院妇科，四川 泸州 646000；西南医科大学附属中医医院妇科，四川 泸州 646000；西南医科大学
1 2 3
附属天府医院妇产科，四川 眉山 620000

［摘 要］ 目的：构建早期宫颈鳞癌盆腔淋巴结转移的预测模型。方法：收集2019年1月—2023年10月西南医科大学附属医
院妇科早期宫颈鳞癌患者的临床资料。以盆腔淋巴结病理检查结果为结局指标，进行单因素Logistic分析，向前和向后逐步回
归分析得到 2 个模型。应用赤池信息量准则（Akaike information criterion，AIC）、连续净重新分类改进指数（net reclassification
improvement，NRI）和综合判别改善指数（integrated discriminant improvement，IDI）确定最优模型，并将最优模型转换为列线
图。使用受试者工作特征曲线下面积（area under the curve，AUC）、Hosmer⁃Lemeshow 检验、校准曲线、临床决策曲线（decision
curve analysis，DCA）对模型进行评价。使用Bootstrap 自助法对模型进行内部验证。结果：本研究共纳入221例患者，建立模型
1 和模型 2。根据 AIC、连续 NRI 与 IDI 结果对模型进行评价，确定模型 2 为最优模型［纳入指标：年龄、鳞状上皮细胞癌抗原
（squamous cell carcinoma antigen，SCCA）、糖类抗原（carbohydrate antigen，CA）125、磁共振成像淋巴结状态］。模型 AUC 为
0.818；Hosmer⁃Lemeshow检验，χ =0.942，P=0.332；校正后AUC为0.800；DCA显示阈值概率在0.03~0.50时，临床净受益值较高；
2
Bootstrap内部验证的AUC为0.784。结论：基于年龄、SCCA、CA125以及磁共振成像淋巴结状态构建的模型有助于术前预测早
期宫颈鳞癌盆腔淋巴结状态。
［关键词］ 宫颈鳞癌；淋巴结转移；磁共振成像；生物标志物；列线图
［中图分类号］ R737.33 ［文献标志码］ A ［文章编号］ 1007⁃4368（2025）10⁃1467⁃09
doi：10.7655/NYDXBNSN241453

Construction of a prediction model for pelvic lymph node metastasis in early ⁃ stage
cervical squamous cell carcinoma

1
1
2*
LI Junmei ，WANG Lu ，MAO Xiguang ，LIAO Sijing 3
1 Department of Gynecology，the Affiliated Hospital of Southwest Medical University，Luzhou 646000；Department of
2
Gynecology，the Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University，Luzhou 646000；
3
Department of Obstetrics and Gynecology，the Affiliated Tianfu Hospital of Southwest Medical University，Meishan
620000，China
［Abstract］ Objective：To construct a model for predicting pelvic lymph node metastasis in early ⁃ stage cervical squamous cell
carcinoma. Methods：The clinical data of patients in the department of gynecology，Affiliated Hospital of Southwest Medical University
from January 2019 to October 2023 were retrospectively analyzed. A univariate logistic analysis was performed based on the
pathological examination results of pelvic lymph nodes as the outcome indicator. Two models were developed by univariate logistic
analysis as well as forward and backward stepwise regression analysis. The Akaike information criterion（AIC），continuous net
reclassification improvement index（NRI），and integrated discriminant improvement index（IDI）were used to determine the optimal
model. The optimal model was then converted into a nomogram，and its efficacy was evaluated by the area under the receiver operating
characteristic curve（AUC），Hosmer⁃Lemeshow test，calibration curve and decision curve analysis（DCA）. Bootstrap method was used
for internal validation. Results：A total of 221 patients were enrolled. Forward and backward stepwise regression methods were used to
establish model 1 and model 2，respectively. According to the results of AIC，serial NRI and IDI，model 2 was the optimal model

［基金项目］ 四川省中医药管理局科学技术研究专项课题（2023MS524）
通信作者（Corresponding author），E⁃mail：mxg33366@163.com（ORCID：0009⁃0005⁃3881⁃4020）
∗