P-193 Using blastocyst re-expansion rate in the decision-making when to warm a new blastocyst for optimal single vitrified-warmed blastocyst transfer

Abstract

Abstract Study question Can post-warmed blastocyst re-expansion rate be used to identify embryos with low likelihood of live birth (LB) in single vitrified blastocyst transfer (SVBT)? Summary answer A statistical model using blastocyst re-expansion rate and Gardner score can be used to identify blastocysts that do not result in live birth. What is known already The ability to identify the most viable embryo with greatest chance of implantation and LB is crucial. As embryo development is a dynamic process, research has recently focused on finding the most predictive morphokinetic parameters to incorporate in the assessment of which embryos to select for transfer. Time-lapse systems allow continuous monitoring of the embryo while cultured uninterruptedly. Several studies have found that the degree of blastocoele expansion and re-expansion are significant parameters for prediction of LB. Furthermore, re-expansion rate has been demonstrated to correlate with pregnancy outcomes in SVBT. Study design, size, duration This study is a retrospective cohort study including a total of 931 SVBT´s at the Fertility Clinic at Horsens Regional Hospital in a 2-year period from 2019-2020. Prior to vitrification, laser assisted collapsing was performed on all blastocysts with an expansion grade ≤ 4, and grade 5 after individual assessment. Only blastocysts with data on both pre-vitrification Gardner score, LB outcome and at least two hours of post-warmed time-lapse data were included in the analysis. Participants/materials, setting, methods Using a time lapse system, both fresh and post-vitrification embryo developmental parameters were evaluated during culturing in an EmbryoScope (Vitrolife, Denmark). Re-expansion rate was calculated as the change in blastocyst area within the first two hours. Logistic regression models were used to estimate the relation between LB and pre- and post-vitrification variables. Based on model predictions, the diagnostic performance was evaluated at thresholds corresponding to a 1%, 2% and 5% false negative rate (FNR). Main results and the role of chance Of the 931 SVBT´s, a total of 388 resulted in a LB (41.7%). Using the initial blastocyst size and the two-hour re-expansion rate yielded a model with an area under the ROC curve (AUC) of 0.61 (95% bootstrap CI: 0.57-0.65) Adding pre-vitrification Gardner score to the model increased the AUC to 0.62 (95% bootstrap CI: 0.59-0.66). This model was used to calculate the true negative rate (TNR), being the proportion of transferred blastocysts that would not result in LB at each threshold. At a FNR threshold of 2%, the model predicts that out of 50 discarded blastocysts, 7 would have resulted in a LB and 43 would not have resulted in LB. In summary, the ratio of erroneously discarded embryos vs. correctly discarded embryos would be approximately 1:6 using a 2% FNR. Limitations, reasons for caution The statistical model was developed and tested on the same dataset with a risk of overfitting. Any potential clinical application of the model requires testing and validation on an independent dataset. Furthermore, clinical procedures and in particular laser collapse may bias the results. Wider implications of the findings Being able to identify blastocysts with a low likelihood of a LB allows for the possibility to thaw a new blastocyst. In a clinical setting, this could lead to a reduced time to pregnancy and LB in women who have multiple vitrified blastocysts. Trial registration number 1-16-02-583-20