Problem: Unexplained infertility (UI) represents about 25%-40% of all infertility and is a formidable obstacle for successful pregnancy for child-bearing aged women. However, up to now, there is no reliable method to predict this condition with high accuracy, thereby hindering early management of this condition.
Method of study: Our prospective study consists of 84 child-bearing aged women that were clinically diagnosed UI. Forty-four matched healthy fertility (HF) women were served as controls. We examined the profiles of 25 hormones and cytokines that were likely related to pathogeneses and molecular pathways involved in UI with the technique of protein array. The samples were randomly stratified 7:3 into a training set and a testing set. We used the SMOTEboost model with 10 serum proteins in a clinical verification study to identify UI cases.
Results: The predictor had an area under the receiver operating characteristic curve (AUC) of 0.788 with 24 serum protein features. The predictive performance in terms of AUC of the model with the top 10 important serum proteins in the clinical verification study to classify UI cases was 0.809. Three most significantly differentially expressed proteins (DEPs) were prolactin, monocyte chemotactic protein-1 (MCP-1), and leptin.
Conclusion: Examination of serum-based protein profile changes could help to identify child-bearing aged women at risk of UI. This would enable early detection and facilitate development of clinical strategies to treat UI and guide their planned parenthood. It may also give clues to pathogeneses of the condition of test subjects.
Keywords: SMOTEboost model; antibody array-based proteomic screening; child-bearing aged; unexplained infertility.
© 2019 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.