Time-series clustering of the migration speed of nucleolus-precursor bodies in human zygotes and live birth following assisted reproductive technology.

PURPOSE: Nucleolus-precursor body (NPB) migration speed in human zygotes predicts live birth after assisted reproduction. However, the impact of changes in NPB migration speed on live births was unclear. This study investigated the relationship between patterns of change in NPB migration speed over time and clinical outcomes.

METHODS

The central coordinates of NPBs were tracked at multiple time points to assess changes in NPB migration speed from 1 to 7 h before the disappearance of the pronucleus. Time-series clustering was performed. Cluster centroids were calculated using the time-series K-means method. A training dataset (N = 210 NPBs) was used to define clusters and determine the optimal number. A test dataset (N = 141 NPBs) was then examined.

RESULTS

When examining the live-birth rate in training data, the standardized residuals of Cluster 7 were significantly higher (100.0%, P < 0.01), while those of Cluster 2 were significantly lower (36.7%, P < 0.05). However, in the test data, the live-birth rate was not significantly different among the clusters. When we placed the clusters into three groups based on the live-birth rates in the training data (low, ≤ 50.0%; middle, 50.1-65.0%; high, > 65.0%), the cluster groups were correlated with live-birth rates in the test data. The live-birth rate in clusters comprising the high subgroup was significantly higher than that in the clusters comprising the low subgroup (80.0% vs. 47.4%, P = 0.034).

CONCLUSION

Time-series analysis of NPB migration speed may serve as a useful indicator for selecting human zygotes with a higher potential for viable pregnancies and live births following assisted reproduction.