多囊卵巢综合征患者行拮抗剂方案发生卵巢过度刺激综合征危险因素分析及临床预测模型的构建  

Risk factors analysis and clinical prediction model establishment for ovarian hyper-stimulation syndrome in PCOS patients with GnRH antagonist protocol therapy

作　　者：魏艺 凌秀凤[1] 赵纯[1] 李欣[1] WEI Yi;LING Xiu-feng;ZHAO Chun;LI Xin(Department of Reproductive Medicine,Women’s Hospital of Nanjing Medical University,Nanjing Maternity and Child Health Care Hospital,Nanjing 210004)

机构地区：[1]南京医科大学附属妇产医院/南京市妇幼保健院生殖医学中心,南京210004

出　　处：《生殖医学杂志》2023年第12期1849-1856,共8页Journal of Reproductive Medicine

基　　金：国家自然科学基金资助项目(81971386)。

摘　　要：目的分析多囊卵巢综合征(PCOS)患者行拮抗剂(GnRH-ant)方案促排卵时发生卵巢过度刺激综合征(OHSS)的危险因素,并且构建其预测模型。方法回顾性分析2015年1月至2022年12月于我院生殖中心行GnRH-ant促排卵的PCOS患者的1052个促排卵周期的临床资料,根据是否患有OHSS分为两组:符合全胚冷冻中中重度OHSS标准的患者周期为OHSS组(n=61),未发生OHSS及未因OHSS全胚冷冻的患者周期为非OHSS组(n=991)。分别采用Logistic单因素分析、逐步回归Logistic多因素分析、最小绝对收缩算子(LASSO)回归3种方式筛选OHSS的危险因素;采用列线图构建临床预测模型;采用受试者工作特征(ROC)曲线比较3种模型的区分度,并通过Bootstrap重抽样分别计算C-index指数;采用Hosmer-Lemeshow拟合优度检验来评价模型的校准度并绘制校准曲线;绘制预测模型临床决策曲线临床决策曲线(DCA)评价其有效性。结果Logistic单因素分析结果显示,体质量指数(BMI)、基础黄体生成素(bLH)、基础卵泡刺激素(bFSH)、抗苗勒管激素(AMH)等18个因素为OHSS的危险因素;逐步回归Logistic多因素分析结果显示,窦卵泡数(AFC)、促性腺激素(Gn)启动剂量、Gn天数、扳机日雌二醇(E 2)及扳机日孕酮等5个因素为OHSS的危险因素;LASSO回归结果显示,AFC、扳机日E 2及扳机日孕酮3个变量为OHSS的危险因素;ROC曲线分析结果显示,Logistic单因素模型[0.874,95%CI(0.834,0.914)],逐步回归法模型[0.859,95%CI(0.816,0.902)],LASSO回归模型[0.839,95%CI(0.791,0.887)],预测价值均为中等;在内部验证结果显示,C-index指数分别为0.830、0.848、0.835;Hosmer-Lemeshow拟合优度检验显示,Logistic单因素(χ2=8.473,P=0.389),逐步回归法(χ2=11.412,P=0.179),LASSO回归(χ2=13.361,P=0.100);Brier评分结果显示,Logistic单因素评分为0.047,逐步回归法评分为0.045,LASSO回归评分为0.043;DCA结果显示,3种模型均具有较好的临床适用度。结论经由LASSO回归Objective:To analyze the risk factors of ovarian hyper-stimulation syndrome(OHSS)in patients with polycystic ovary syndrome(PCOS)undergoing GnRH antagonist(GnRH-ant)protocol,and establish predictive model.Methods:The clinical data of 1052 cycles of PCOS patients who underwent ovulation induction with GnRH-ant protocol in the Reproductive Center of Nanjing Maternity and Child Health Hospital from January 2015 to December 2022 were retrospective analyzed.The cycles were divided into the two groups based on the presence or absence of OHSS:the patients’cycles that met the criteria for moderate to severe OHSS for whole embryo freezing were in the OHSS group(n=61),while the cycle that did not experience OHSS and did not undergo whole embryo freezing due to OHSS were in the non-OHSS group(n=991).The risk factors of OHSS were screened by logistic univariate analysis,stepwise logistic multifactor regression analysis and least absolute shrinkage operator(LASSO)regression,and the clinical prediction model was established by using column charts.The discrimination of three models was compared by using the receiver operating characteristic(ROC)curve,and the C-index index was calculated separately through Bootstrap resampling.The calibration of the model was evaluated by using Hosmer-Limeshow goodness of fit test,and the calibration curve was drawn.The clinical decision curve(DCA)of the predictive model was drawn to evaluate its effectiveness.Results:The results of logistic univariate analysis show that 18 factors,including body mass index(BMI),basal LH,basal FSH,anti-Mullerian hormone(AMH),etc.were the risk factors for OHSS.The results of stepwise regression logistic multivariate analysis showed that antral follicle count(AFC),initiation dose of gonadotropin(Gn),Gn days,estrogen(E 2)and progesterone levels on trigger day were the five risk factors for OHSS.LASSO regression results showed that three variables including AFC,E 2 and progesterone levels on trigger day,were the risk factors for OHSS.ROC curve analysis results sho

关 键 词：卵巢过度刺激综合征 多囊卵巢综合征 拮抗剂方案 体外受精-胚胎移植 临床预测模型 

分 类 号：R711.6[医药卫生—妇产科学]