卵巢储备功能预测的探讨

Study of predictive value of multiple predictors for ovarian reserve function

摘要:

目的 探讨多个临床指标预测卵巢储备功能的效果。方法 回顾性分析350 例行体外受精-胚胎移植患者的临床资料，应用SPSS 软件作ROC 曲线，分别计算年龄、基础卵泡刺激素（bFSH）、基础卵泡刺激素/ 基础黄体生成素（bFSH/bLH）、窦卵泡计数（AFC）4 个指标曲线下面积（AUC），并用Logistic 回归分析及ROC 曲线，综合多个指标对卵巢储备功能预测。结果 ①年龄、bFSH、bFSH/bLH 及AFC 的AUC 分别为0.687、0.643、0.815 及0.867，均大于机会参考下面积，差异有统计学意义（P <0.05）；② Logistic 回归结果显示，综合年龄、bFSH、bFSH/bLH 及AFC 4 个指标的ROC 曲线下AUC 是0.918，差异有统计学意义（P <0.05）；排除年龄的影响，综合bFSH、bFSH/bLH 及AFC 3 个指标的ROC 曲线下AUC是0.913，差异有统计学意义。结论 AFC 是预测卵巢储备功能价值较高的单个指标。排除年龄因素的干扰，综合bFSH、bFSH/bLH 和AFC 3 个指标较单一指标更能准确地预测卵巢储备功能。

Abstract:

Objective To compare the predictive value of multiple predictors and single one on ovarian reserve function. Methods Totally 350 females who had undergone in-vitro fertilization (IVF) were retrospectively analyzed.receiver operating characteristic curve (ROC) was applied to calculate the area under curves (AUC) of age, serum basal follicle-stimulating hormone (bFSH), bFSH/bLH ratio and antral follicle count (AFC). As well as, the AUC of comprehensive multiple predictors. Logistic regression analysis and ROC curve were used to analyze multiple indexes for prediction of ovarian reserve function. Results The AUC of age, bFSH, bFSH/bLH and AFC was 0.687, 0.643, 0.815 and 0.867 respectively, which was statistically larger than the reference area (P < 0.05). Logistic regression model showed that the combination of age, AFC, bFSH and bFSH/bLH improved the predictive function for ovarian reserve and had the AUC of 0.918 (P < 0.05). The AUC of the combination of AFC, bFSH and bFSH/bLH was 0.913 after elimination of the affect of age (P < 0.05). Conclusions AFC is the strongest single predictor for ovarian reserve function. A multivariable modle has higher value than a single predictor on the predictive power and improve the veracity.