基于子宫内膜超声参数的列线图模型预测冻融胚胎移植临床妊娠的研究

Prediction of clinical pregnancy in frozen-thawed embryo by a Nomogram model based on endometrial ultrasound parameters

目的结合子宫内膜容受性相关超声参数构建和验证列线图预测模型,评估冻融胚胎移植(FET)临床妊娠成功的个体化概率。方法前瞻性队列研究分析2023年3—12月于解放军总医院第六医学中心生殖中心接受FET治疗的357例患者(共357个FET周期)的临床资料。并根据纳入时间不同分为建模组(219个周期)和验证组(138个周期),建模组又根据是否妊娠分为临床妊娠组(107个周期)和未妊娠组(112个周期)。收集纳入患者的基线资料及超声测量指标,采用单因素和多因素逻辑回归分析筛选不孕患者临床妊娠的独立影响因素,通过最小绝对收缩和选择算子(LASSO)回归及赤池信息量准则(AIC)分析确定构建最优列线图模型的变量组合,通过受试者工作特征(ROC)曲线、校准曲线及临床决策曲线评估列线图的性能和临床价值。结果建模组和验证组的女方年龄、不孕年限、基础性激素水平等基线资料比较均无显著性差异(P>0.05);建模组中临床妊娠组和未妊娠组间女方年龄、不孕类型等基线资料比较均无显著性差异(P>0.05)。单因素分析结果显示,子宫内膜厚度、子宫动脉阻力指数(RI)、子宫动脉搏动指数(PI)、子宫内膜容积、子宫内膜血管化指数(VI)、血流指数(FI)、血管化血流指数(VFI)、血流Applebaum分型均对临床妊娠率有显著影响(P<0.05);进一步多因素逻辑回归分析显示子宫内膜容积[OR=2.769,95%CI(1.571,4.881),P<0.001]、子宫动脉RI[OR=3.091,95%CI(1.332,7.170),P=0.009]、内膜VI[OR=2.641,95%CI(1.439,4.847),P=0.002]是临床妊娠率的独立影响因子;LASSO交叉验证及最小AIC标准筛选变量结果一致:子宫内膜容积、子宫动脉RI、内膜VI联合子宫内膜FI构建的模型性能更优;建模组和验证组的ROC曲线下面积(AUC)分别为0.814[95%CI(0.756,0.872)]和0.797[95%CI(0.720,0.873)],模型区分度良好,校准曲线及决策曲线显示列线图预测临床妊娠率具有较优的准确度及临床净效益。结论本研究建立了预测FET早期妊娠结局的列线图模型,有助于个性化评估FET的临床妊娠概率。

Objective:To construct and validate a Nomogram prediction model incorporating ultrasound parameters related to endometrial tolerance,so as to assess the individualized probability of successful clinical pregnancy in frozen embryo transfer(FET).Methods:A prospective cohort study was conducted to analyze the clinical data of 357 patients(included 357 FET cycles)who underwent FET at the Reproductive Centre of the Sixth Medical Center of PLA General Hospital from March to December 2023.They were divided into a modelling group(219 cycles)and a validation group(138 cycles)according to the time of inclusion,and the modelling group was further divided into a clinical pregnancy group(107 cycles)and a non-pregnant group(112 cycles)according to whether they were pregnant or not.Baseline data and ultrasound measurements of the included patients were collected,single and multifactorial logistic regression analyses were used to screen for independent influences on clinical pregnancy,least absolute shrinkage and selection operator(LASSO)regression and Akaike information criterion(AIC)analyses were used to determine the variable combinations for constructing the optimal Nomogram model,and the performance of the Nomogram was assessed by the receiver operator characteristic(ROC)curves,the calibration curves,and the clinical decision curves.Results:There were no significant differences in female age,infertility years and basic sex hormone level between the modelling and validation groups(P>0.05).And there were no significant differences in female age and infertility type between the clinical pregnancy and non-pregnant groups in the modelling group(P>0.05).The results of univariate analysis showed that endometrial thickness,uterine artery resistance index(RI),uterine artery pulsatility index(PI),endometrial volume,endometrial vascularisation index(VI),flow index(FI),vascularised flow index(VFI),and Applebaums blood typing had significant effects on the clinical pregnancy rate(P<0.05).Furthermore,multifactorial logistic regression analysis showed that endometrial volume[OR=2.769,95%CI(1.571,4.881),P<0.001],uterine artery RI[OR=3.091,95%CI(1.332,7.170),P=0.009],endometrial VI[OR=2.641,95%CI(1.439,4.847),P=0.002]were the independent influencing factors of the clinical pregnancy rate.For screening model variables,the results of LASSO cross-validation agreed with those of the minimum AIC criterion,namely the performance of the model constructed by endometrial volume,uterine artery RI and endometrial VI combined with endometrial FI was better.The area under the ROC curve(AUC)in the modelling group and the validation group were 0.814[95%CI(0.756,0.872)]and 0.797[95%CI(0.720,0.873)],respectively,showing good model discrimination.The calibration curves and the decision curve showed that the predicted clinical pregnancy rates of the Nomogram were with superior accuracy and net clinical benefit.Conclusions:This study developed a Nomogram model for predicting early pregnancy outcomes in FET,which could be useful for personalized assessment of clinical pregnancy probability in FET.