超越形态学:单囊胚培养液多蛋白分析预测胚胎妊娠潜能
本研究纳入2021年1月至2023年12月行冻融单囊胚移植的不孕症患者,用单分子阵列技术检测单囊胚培养液中12种蛋白浓度(n=1390),在训练集(n=240)建模型、测试集(n=60)验证。IL-8和β-HCG升高与临床妊娠率升高相关,RANTES升高与妊娠率降低相关;三蛋白模型AUC=0.722,联合形态学评分AUC=0.744,独立测试集AUC=0.724。
要点
- 单囊胚培养液中IL-8、β-HCG和RANTES与临床妊娠显著相关。
- 三蛋白模型预测临床妊娠AUC=0.722,优于各单一标志物(P≤0.005)。
- 联合形态学评分后AUC达0.744,独立测试集验证AUC=0.724。
核心结论
- 单囊胚培养液中IL-8、β-HCG和RANTES三蛋白模型可非侵入性预测临床妊娠,联合形态学评估效果更佳。
研究设计与样本
- 回顾性研究纳入2021年1月至2023年12月行冻融单囊胚移植的不孕症患者,检测单囊胚培养液12种蛋白(n=1390)。
- 训练集n=240用于建模,测试集n=60用于验证,以ROC曲线AUC评价预测性能。
主要发现
- IL-8浓度升高与临床妊娠率显著升高相关(P<0.001)。
- β-HCG浓度升高与临床妊娠率显著升高相关(P=0.009)。
- RANTES浓度升高与临床妊娠率显著降低相关(P<0.001)。
- 单一标志物AUC:IL-8为0.600(P=0.03),β-HCG为0.588(P=0.002),RANTES为0.397(P=0.001)。
关键数值与推荐
- 三蛋白联合模型预测临床妊娠AUC=0.722,显著优于各单一蛋白(P≤0.005)。
- 三蛋白联合形态学评分模型AUC=0.744,独立测试集验证AUC=0.724。
临床怎么用
- 可探索检测单囊胚培养液中IL-8、β-HCG和RANTES辅助胚胎选择。
- 多蛋白模型宜与形态学评分联合使用以提高预测准确性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
RESEARCH QUESTION: Can a non-invasive approach predicting embryonic pregnancy potential be developed by analysing proteins in single blastocyst conditioned medium (SBCM)? DESIGN: Patients with infertility who underwent vitrified-warmed single blastocyst transfer (January 2021-December 2023) were included in this study. SBCM corresponding to transferred blastocysts was collected and analysed using highly sensitive single-molecule array technology to quantify the concentrations of 12 candidate proteins (n = 1390). Key proteins were selected for model development for prediction of clinical pregnancy using a training set (n = 240), and subsequently validated with a test set (n = 60). Performance was evaluated by the area under the curve (AUC) of receiver operating characteristic curves. RESULTS: Among the 12 proteins examined, elevated concentrations of interleukin-8 (IL-8) (P < 0.001) and beta-human chorionic gonadotrophin (B-HCG) (P = 0.009) were associated with a significantly higher clinical pregnancy rate, whereas elevated C-C motif chemokine 5 (RANTES, 'regulated on activation, normal T cell expressed and secreted') was associated with a significantly lower pregnancy rate (P < 0.001). Individual biomarkers demonstrated modest discriminatory capacity for predicting clinical pregnancy for IL-8 (AUC = 0.600; P = 0.03), B-HCG (AUC = 0.588; P = 0.002) and RANTES (AUC = 0.397; P = 0.001). A multi-protein prediction model integrating these three biomarkers exhibited superior discriminatory performance (AUC = 0.722), and significantly outperformed each individual protein marker (P ≤ 0.005). The combined model incorporating both the multi-protein signature and morphological grading parameters achieved optimal predictive accuracy (AUC = 0.744). These findings were validated in an independent test set (AUC = 0.724). CONCLUSIONS: This study establishes a novel multi-protein model for predicting clinical pregnancy. The combination of IL-8, B-HCG and RANTES in SBCM provides a promising non-invasive approach for embryo selection, particularly when integrated with morphological assessment.
引用
- Chen P, Huang C, Zhang H, et al. Beyond morphology: a multi-protein analysis of single blastocyst conditioned medium to predict pregnancy potential of embryos. Reprod Biomed Online. 2025. PMID: 41455340. doi:10.1016/j.rbmo.2025.105213
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。