Sequencing of uterine fluid is a promising tool for predicting endometrial receptivity: a proof-of-concept study

Background: Endometrial receptivity is crucial to a successful pregnancy. Although molecular markers of endometrial receptivity can be obtained through endometrium biopsy, the prediction method is invasive and delayed. Hence, it is important and necessary to establish a less invasive and more reliable method for the determination of endometrial receptivity to increase the efficiency and effectiveness of assisted reproductive technology (ART) treatments. Methods: This study was performed in the Centre for Assisted Reproduction of Shanghai First Maternity and Infant Hospital between June 2021 and March 2022. Samples left on the transfer catheter after cleavage stage embryos transfer (day 3, D3, presumable pre-receptivity, n = 96) and blastocysts transfer (day 5, D5, presumable receptivity, n = 127) in 223 ART patients who achieved successful clinical pregnancies subsequently were collected. The collected samples were further divided into a training set (n = 111 samples) and a test set (n = 112 samples) for endometrial receptivity prediction model training and validation, respectively. Results: All the collected samples contained sufficient total RNA for further analyses. Ultimately, 57 signature genes specifically expressed in the epithelial cells and the macrophages were identified. The endometrial receptivity predictability of our model in the training set exhibited an area under the curve (AUC) of 1.00, indicating complete stratification of pre-receptive and receptive samples. Likewise, in the test set, our model achieved an AUC of 0.91 with accuracy of 0.83, specificity of 0.86 and sensitivity of 0.79. Conclusions: Our study demonstrates a non-invasive method of endometrial sample collection for accurate determination of endometrial receptivity.