Multi-Factor Analysis And Prediction Model Establishment Of Ectopic Pregnancy After In Vitro Fertilization Embryo Transfer Technology

In recent years,the rapid development of in vitro fertilization-embryo transfer(IVF-ET)technology has brought hope to the families of countless infertility patients.However,ectopic pregnancy is as one of the most common complications after IVF,which seriously threatens the reproductive health of women of childbearing age.Ectopic pregnancy(EP)refers to the implantation of fertilized eggs outside the uterine cavity.In recent years,with the continuous improvement and upgrade of IVF technology,the risk of EP after IVF has been reduced,but it is still higher than that of natural pregnancy.How to more effectively reduce the incidence of EP after IVF has become the focus of attention of scholars.At present,the mechanism of EP after IVF is unclear.It may be related to the super-physiological hormone environment,tubal factors,embryonic factors,embryonic development potential and IVF operation.At present,clinical research still lacks quantitative tools for risk prediction of patients with EP after IVF assisted pregnancy.Therefore,this project comprehensively evaluates the common risk factors for the occurrence of EP after IVF in the clinic,and constructs a clinical prediction model for the occurrence of EP after IVF based on logistic regression,which is of great significance for preventing and reducing the occurrence of EP.ObjectivesTo explore the risk factors of EP after IVF technology assisted pregnancy,and construct a clinical prediction model of EP after IVF assisted pregnancy.Materials and Methods1.Research objects:a retrospective study of the clinical data of 12,766 cycles of patients who received IVF treatment at the Reproductive Center of the Third Affiliated Hospital of Zhengzhou University(Henan Maternal and Child Health Hospital)and obtained clinical pregnancy from January 2016 to April 2020.2.Inclusion criteria and exclusion criteria:inclusion criteria:(1)clinical pregnancy cycle of fresh embryo transfer;(2)clinical pregnancy cycle of frozen resuscitation embryo transfer;exclusion criteria:(1)fresh untransplanted cycle;(2)non-pregnancy or biochemical pregnancy cycle after embryo transfer;(3)cycle with donor oocytes;(4)incomplete records cycle.3.Grouping:according to the above inclusion and exclusion criteria,12766 cycles of clinical pregnancy were finally obtained,including 214 cycles in EP group and 12552cycles of intrauterine pregnancy(IUP).Collect and process the clinical data of the two groups of patients,including the woman’s age,years of infertility,endometrial thickness on the day of transfer,body mass index(BMI),type of infertility(primary infertility or secondary infertility),year of operation,past EP history,history of uterine evacuation,infertility factors(whether tubal infertility),type of embryo transfer(fresh embryo transfer cycle or frozen embryo transfer cycle),embryo stage(cleavage embryo transfer and blastocyst transfer),embryo quality(non-high-quality embryos and high-quality embryos),number of embryos transferred(single embryo transfer or≥2 embryo transfer).4.All statistical management and analysis and processing were performed using Empower Stats and R language(version 3.4.3).The t test or Kruskal-Wallis rank sum test was used to assess between-group differences in continuous variables,and these variables were expressed as the mean±SD.Categorical variables were represented as the rate(%).The means from chi-square analyses or Fisher’s test ware used to assess the differences between groups.Single factor analysis was used to screen out the factors of EP.At the same time,a smooth curve fitting diagram and threshold effect analysis of endometrial thickness and EP on the day of transfer were generated.Multivariate logistic regression analysis was used to analyze the variables of P<0.05in the single-variable results,and the independent influencing factors of the occurrence of EP were screened out.Based on the statistically significant variable regression coefficients in the multi-factor logistic regression model,a predictive model was established.The area under the curve(AUC)of the receiver operating characteristic curve(ROC)and Hosmer-Lemeshow detection were used to evaluate the discrimination and calibration of the model.Bootstrap sampling method was used to internally verify the prediction model.Statistical significance was set at P<0.05.ResultsAccording to the above inclusion and exclusion criteria,12766 cycles of clinical pregnancy patients were finally obtained after IVF,12552 cycles in the IUP group,and 214 cycles in the EP group.The incidence of EP was 1.7%.1.Analysis of risk factors for EP after IVF(1)Comparison of general basic information and treatment process of patients in IUP group and EP group:The average age of female patients in IUP group and EP group was 30.77±4.61 years old and 31.10±4.77 years old,respectively,and the difference was not statistically significant.Compared with IUP group,EP group patients,the endometrium was thinner(10.33±2.16 vs.9.72±2.18),the proportion of patients with tubal infertility increased(37.75%vs.46.73%),and the proportion of blastocysts transfer decreased significantly(46.89%vs.24.77%).The proportion of single embryo transfer patients decreased(38.29%vs.24.77%),and the differences were significant(P<0.05).The rest of the basic information and treatment process were not statistically different between the two groups(P>0.05).(2)The relationship between endometrial thickness on the day of transfer and EP:the smooth curve fitting graph suggested that there was a nonlinear association between endometrial thickness and EP.Threshold analysis of endometrial thickness on the day of transfer to EP(divided into three groups:<7.6mm;7.6-12.1mm;>12.1mm).Single-line linear regression model(model I):OR=0.86,95%CI:0.79-0.93 and three-segment linear regression model(model II):<7.6mm:(OR=0.92,95%CI:0.52-1.64),7.6~12.1mm EMT(OR=0.76,95%CI:0.67-0.87),>12.1mm EMT(OR=1.01,95%CI:0.77-1.32).The log-likelihood ratio test results in P=0.012<0.05,indicating that the model I and model II were significantly different,and it was suitable to fit the segmented model.(3)Univariate analysis:compared with patients with non-fallopian tube factors,the incidence of EP in patients with tubal factors increased by 45%(OR=1.45,95%CI:1.10-1.90,P=0.0076);with the increase in endometrial thickness on the day of transfer,the incidence of EP decreased(OR=0.86,95%CI:0.80-0.93,P<0.0001);compared with the cleavage embryo transfer,blastocyst transfer significantly reduced the incidence of EP(OR=0.37,95%CI:0.27-0.51,P<0.0001);compared with single embryo transfer,EP risk of multiple embryo transfer was significantly increased(OR=1.88,95%CI:1.38-2.58,P<0.0001).(4)Multivariate Logistic regression analysis:the four variables obtained from single factor analysis were subjected to multivariate logistic regression analysis.The risk of EP in patients with tubal factors was 1.48 times that of patients with non-fallopian tubal factors(AOR=1.48,95%CI:1.11 to 1.96,P=0.0072);each increase of one unit(1mm)in the endometrial thickness on the day of transfer would reduce the incidence of EP by 15%(AOR=0.85,95%CI:0.79-0.92,P<0.0001).The stratified analysis of the endometrial thickness showed that compared with women with an EMT of<7.6mm,the EMT between 7.6~12.1mm(AOR=0.57;95%CI0.36-0.90)and the endometrial thickness>12.1mm(AOR=0.42;95%CI 0.24-0.74)were significantly associated with a reduction in the risk of EP(P<0.05);compared with cleavage embryo transfer,blastocyst transfer could significantly reduce the risk of EP(AOR=0.26,95%CI:0.17-0.42,P<0.0001);compared with single embryo transfer,EP risk significantly increased after multiple embryo transfer(AOR=2.04,95%CI:1.46-2.83,P<0.0001).2.Prediction model establishment,evaluation and verificationThe four risk factors obtained from the multivariate analysis,tubal factors,endometrial thickness on the day of transfer,stage of embryo transfer and number of embryos transferred were set as independent variables to predict the occurrence of EP,and the prediction model was established as follows:-1.632+0.357×(non-fallopian tube factors=0;tubal factors=1)-1.245×(cleavage stage embryo=0;blastocyst=1)-0.216×(single embryo transfer=1;≥2 embryo transfer=0)-0.178×endometrial thickness.The model AUC=0.68,95%CI:(0.640-0.711),at the optimal threshold(when the Youden index was the largest,sensitivity+maximum specificity cut-off value)sensitivity is 0.67,specificity is 0.64,model predictive ability medium.The Hosmer-Lemeshow test was performed on the model:χ~2=15.61468,P=0.0754;because P>0.05,it indicated that the predicted observations of the model were in good agreement with the actual observations,indicating that the prediction model had good accuracy.The prediction model was internally verified by self-sampling method(bootstrap sampling resampling times=500),AUC=0.68,95%CI:(0.640-0.712),the sensitivity was 0.67,and the specificity was 0.64 at the optimal threshold.After internal verification,AUC was basically unchanged,and the model was relatively stable.Conclusions1.Tubal factors,endometrial thickness on the day of transfer,embryo transfer stage and number of embryos transferred were independent risk factors for ectopic pregnancy.2.When the endometrial thickness was≥7.6 mm,whether it was a fresh embryo or a freeze-thaw transfer cycle,single blastocyst transfer could significantly reduce the incidence of EP.3.The clinical prediction model based on logistic regression had a certain predictive value for the occurrence of EP after IVF.