Predicting oocyte fertilisability in intracytoplasmic sperm injection cycles: a retrospective observational study

Abstract Background Conception assisted by intracytoplasmic sperm injection (ICSI) requires oocyte stripping for morphological evaluation of maturity status. However, this approach prevents further maturation and poorly predicts fertilisability, so more robust assessment strategies are needed. Given that cytokines orchestrate oocyte development, we aimed to assess the association of follicular fluid cytokine profiles with maturation stage and develop predictive machine learning-based methods to identify those with the greatest fertilisation potential. Methods In this retrospective study, follicular fluid was collected at oocyte retrieval from 64 women and linked to oocyte maturity status or fate—namely, germinal vesicle (n=26), metaphase I (51), metaphase II not fertilised (51), and metaphase II fertilised (84). 51 follicular fluid cytokines were profiled by multiplex immunoassay. Machine learning-based classifiers to predict oocyte fertilisability were subjected to iterative feature reduction to a threshold suitable for developing a clinically viable assessment of oocyte maturity. Women gave written, informed consent. Findings Cytokine profiles varied dynamically throughout maturation. When applied to naive samples with known outcome, classifiers developed using tumour necrosis factor-related apoptosis-inducing ligand and interleukin 18 profiles alone correctly discriminated 89% of metaphase II not fertilised oocytes (ie, those with the highest fertilisability) with high confidence from all other maturation stages. Interpretation These classifiers offer the prospect of cost-effective, point-of-care testing, and streamlined ICSI-based workflows. This assessment circumvents stripping such that immature or low fertilisability oocytes could benefit from in-vitro maturation and increase the pool of usable oocytes. Further studies will confirm the robustness of these classifiers in women with a broader morbidity spectrum and their translational value across a range of clinical settings. Funding Infertility Research Trust.