摘要
目的:探讨卵巢癌(OC)患者神经元特异性烯醇化酶(NSE)、细胞角蛋白19片段(CYFRA21-1)、铁蛋白(Fer)表达及与淋巴结转移(LNM)的关系。方法:选取2019年1月1日至2021年12月31日收治的264例OC患者为研究对象,其中176例为训练组创建模型,88例为验证组评判模型,另选取同期卵巢良性肿瘤患者88例为对照Ⅰ组,健康体检者88例为对照Ⅱ组。对比训练组、对照Ⅰ组、对照Ⅱ组血清NSE、CYFRA21-1、Fer水平,多因素非条件logistic回归方程筛选危险因素,并建立方程,通过受试者工作特征(ROC)曲线完成自身验证,K折交叉进行组外验证。结果:训练组血清NSE、CYFRA21-1、Fer水平高于对照Ⅰ组、对照Ⅱ组,对照Ⅰ组血清NSE、CYFRA21-1、Fer水平高于对照Ⅱ组(P<0.05);发生LNM患者病灶直径大于无LNM患者,分化程度低于LNM患者,血清NSE、CYFRA21-1、Fer水平高于无LNM患者(P<0.05);logistic回归模型校正病灶直径、分化程度后,血清NSE、CYFRA21-1、Fer水平升高仍是发生LNM的独立危险因素(P<0.05);三者联合评估LNM的ROC曲线下的面积大于单一指标评估值(P<0.05);利用K折交叉验证进行组外验证,以检验模型的稳定性,结果显示,10组训练准确性为0.837±0.030,预测准确性为0.871±0.029。结论:NSE、CYFRA21-1、Fer与OC患者LNM密切相关,三者联合可提高LNM诊断效能,经验证,诊断模型具有良好的准确性和稳定性,有助于临床诊断及治疗方案的决策。
Objective: To investigate the expression of neuron-specific enolase(NSE), cytokeratin 19 fragment(CYFRA21-1), ferritin(Fer) and their relationship with lymph node metastasis(LNM) in patients with ovarian cancer(OC). Methods: A total of 264 OC patients in our hospital from January 1, 2019 to December 31, 2021 were selected as the research objects, of which 176 were used as the training group to create the model, and the other 88 were used as the validation group to judge the model. During the same period, 88 patients with benign ovarian tumors were selected as the control group I, and 88 healthy subjects were selected as the control group II. The levels of serum NSE, CYFRA21-1 and Fer in the training group, control group I and control group II were compared. Multivariate unconditional logistic regression equation was used to screen risk factors, and the equation was established. Self-validation was done by receiver operating characteristic(ROC) curve, and out-of-group validation was done by K-fold crossover. Results: The serum levels of NSE, CYFRA21-1 and Fer in the training group were higher than those in the control group I and the control group II, and the serum NSE, CYFRA21-1 and Fer levels in the control group I were higher than those in the control group II(P<0.05). The diameter of lesions in patients with LNM was larger than that in patients without LNM, the degree of differentiation was lower than that in patients with LNM, and the levels of serum NSE, CYFRA21-1 and Fer were higher than those in patients without LNM(P<0.05). The AUC of the three combined evaluation of LNM was greater than the evaluation value of a single index(P<0.05). K-fold cross-validation was used for out-of-group validation to test the stability of the model. The results showed that the training accuracy of 10 groups was 0.837 ± 0.030, and the prediction accuracy was 0.871 ± 0.029. Conclusion: NSE, CYFRA21-1 and Fer were closely related to LNM in OC patients. The combination of the three could improve the diagnostic efficiency of LNM. It might have been verified that the diagnostic model has good accuracy and stability, and it might be helpful for clinical diagnosis and decision-making of treatment plans.
作者
马建新
严丽花
甘志忠
许玉珍
胡桂华
梁声强
MA Jianxin;YAN Lihua;GAN Zhizhong;XU Yuzhen;HU Guihua;LIANG Shengqiang(Department of Clinical Laboratory,909th Hospital of Joint Logistic Support Force Dongnan Hospital of Xiamen University,Zhangzhou,363000,China)
出处
《临床血液学杂志》
CAS
2023年第2期98-103,共6页
Journal of Clinical Hematology
