Abstract
PURPOSE: The aim of this study was to determine the predictive value of the morphokinetic parameter S2 (the duration between cleavage from the 3-cell to 4-cell stage in the absence of direct cleavage) for early embryo developmental potential, establish its optimal cutoff value, and explore associated metabolic pathways. METHODS: This prospective study analyzed 662 embryos from 65 patients (September 2023-October 2024) which were cultured in time-lapse incubators. S2's predictive value for blastocyst formation/quality was assessed, with a cutoff value calculated through receiver operating characteristic (ROC) analysis. Subsequently, metabolomic profiling of spent culture media (SCM) from 20 patients was stratified by S2 cutoff using liquid chromatography-tandem mass spectrometry (LC-MS). RESULTS: Among 662 embryos cultured in time-lapse incubators, 474 were analyzed to evaluate the association between S2 and developmental potential. Significant differences in S2 values were observed between the blastocyst formation vs. non-formation groups and high- vs. low-quality blastocyst groups (both median 0.5 vs. 0.8, P < 0.05). Logistic regression analysis showed that S2 predicted available blastocyst formation with an area under the curve (AUC) of 0.622 (95%CI = 0.559-0.685, P < 0.001) and blastocyst quality with an AUC of 0.645 (95% CI = 0.569-0.722), with an optimal cutoff of 0.7 h. Based on this cutoff, the day 3 (D3) SCM from 20 patients was divided into two groups (S2 < 0.7 and S2 ≥ 0.7) for metabolomic analysis. The results revealed 98 differentially expressed metabolites (59 upregulated and 39 downregulated), enriched in protein digestion/ absorption, aminoacyl-tRNA biosynthesis, fatty acid metabolism, central carbon metabolism in cancer, and glutamatergic synapse. CONCLUSIONS: The S2 parameter (optimal cutoff = 0.70 h) shows moderate predictive capacity for embryo developmental potential (AUC = 0.64), with its regulation by specific metabolic pathways, supporting its inclusion in future multifactorial embryo evaluation models.