多囊卵巢综合征患者早期妊娠丢失的风险预测模型构建与验证

Establishment and Validation of Risk Prediction Model for Early Pregnancy Loss in Polycystic Ovary Syndrome

目的 探讨多囊卵巢综合征(PCOS)患者早期妊娠丢失的危险因素,并根据此构建风险预测列线图模型。方法 回顾性分析河南科技大学第一附属医院于2019年1月至2021年12月常规产检的228例PCOS患者的临床资料。按照3∶2比例,将其随机分配为建模组(137例)和验证组(91例)。根据早期妊娠是否发生胚胎丢失将建模组分为妊娠丢失组(35例)和妊娠未丢失组(102例),对比分析两组患者的一般资料。采用多因素logistic回归分析法分析PCOS患者早期妊娠丢失的危险因素,并采用R3.4.7软件包绘制列线图模型,并绘制受试者工作特征(ROC)曲线对列线图模型的预测效能进行评估,绘制校准度曲线图,并采用Bootstrap法检验预测列线图模型的校准度。结果 建模组与验证组患者年龄、体重指数(BMI)、自然流产史、促排卵治疗、妊娠前稀发排卵或无排卵、雌二醇(E2)、睾酮(T)、促卵泡成熟激素(FSH)、促黄体生成素(LH)、空腹胰岛素水平(FINS)、空腹血糖水平(FPG)、叶酸、血清25羟基维生素D3[25-(OH)D3]、口服葡萄糖耐量试验(OGTT)2 h葡萄糖水平比较,差异无统计学意义(P>0.05);两组自然流产史、促排卵治疗、妊娠前稀发排卵或无排卵、E2、FSH、FPG、叶酸、OGTT 2 h葡萄糖水平比较,差异无统计学意义(P>0.05);妊娠丢失组年龄、BMI、T、LH、FINS水平均高于妊娠未丢失组(P<0.05),血清25-(OH)D3水平低于妊娠未丢失组(P<0.05);经logistic回归分析可知,年龄、BMI、T、LH、FINS、血清25-(OH)D3均为影响建模组PCOS患者早期妊娠丢失的危险因素(P<0.05);ROC曲线分析结果显示:列线图预测PCOS患者早期妊娠丢失的曲线下面积(AUC)为0.836(95%CI:0.764~0.894),灵敏度为77.14%,特异度为88.24%。运用Bootstrap法对模型进行验证,计算建模组的一致性指数(C-index)值为0.836,验证组的C-index为0.827。结论 年龄、BMI、T、LH、FINS、血清25-(OH)D3均为影响PCOS患者早期妊娠丢失的独立危险因素,在此基础上构建列线图模型能够更直观、有效预测PCOS患者早期妊娠丢失的风险,具有良好的预测效能。

Objective To explore the risk factors of early pregnancy loss in patients with polycystic ovary syndrome(PCOS),and to construct a risk prediction nomogram model.Methods The clinical data of 228 PCOS patients who underwent routine prenatal examination in the First Affiliated Hospital of Henan University of Science and Technology from January 2019 to December 2021 were retrospectively analyzed.They were randomly assigned to the modeling group(137 cases)and the validation group(91 cases)in a 3∶2 ratio.According to whether embryo loss occurs in early pregnancy,the modeling group was divided into pregnancy loss group(35 cases)and pregnancy non loss group(102 cases).The general data of the two groups were compared and analyzed.Multivariate logistic regression analysis was used to analyze the risk factors of early pregnancy loss in patients with PCOS,and the R3.4.7 software package was used to draw the nomogram model.The receiver operating characteristic(ROC)curve was drawn to evaluate the prediction efficiency of the nomogram model,and the calibration curve was drawn.Bootstrap method was used to test the calibration of the predictive nomogram model.Results There were no significant differences in age,body mass index(BMI),history of spontaneous abortion,ovulation induction therapy,rare ovulation or anovulation before pregnancy,estradiol(E 2),testosterone(T),follicle-stimulating hormone(FSH),luteinizing hormone(LH),fasting insulin(FINS),fasting plasma glucose(FPG),folic acid,serum 25-hydroxyvitamin D 3[25-(OH)D 3]and oral glucose tolerance test(OGTT)2-hour glucose levels between the modeling group and the verification group(P>0.05).There were no significant differences in history of spontaneous abortion,ovulation induction therapy,rare ovulation or anovulation before pregnancy,E 2,FSH,FPG,folic acid,serum 25-(OH)D 3 and OGTT 2-hour glucose levels between the two groups(P>0.05).The age,BMI,T,LH,FINS levels in the pregnancy loss group were higher than those the non-pregnancy loss group(P<0.05),and the serum 25-(OH)D 3 levels were lower than those the non-pregnancy loss group(P<0.05).Logistic regression analysis showed that BMI,T,LH and FINS were risk factors for early pregnancy loss in PCOS patients in the modeling group(P<0.05).ROC curve analysis showed that the area under curve(AUC)for predicting early pregnancy loss in patients with PCOS was 0.836(95%CI:0.764-0.894),with a sensitivity of 77.14%and specificity of 88.24%.Bootstrap method was used to verify the model.The consistency index(C-index)of the modeling group was 0.836,and the C-index value of the verification group was 0.827.Conclusion Age,BMI,T,LH,FINS and the serum 25-(OH)D 3 are all independent risk factors for early pregnancy loss in patients with PCOS.On this basis,the nomogram model can more intuitively and effectively predict the risk of early pregnancy loss in patients with PCOS,and has good predictive efficiency.