Abstract Study question Can a classification algorithm based on automatic annotations of early morphokinetic timings improve embryo selection in ICSI cycles? Summary answer The automatic scoring resulted significantly predictive of blastulation, implantation and live birth, especially when combined with morphological evaluation, but not of euploidy status. What is known already The Eeva Test® is an imaging system that automatically annotates morphokinetic parameters of the early embryo development. The latest version features the Xtend® algorithm, which classifies embryos on day 3 of development from highest (1) to lowest (5) potential to reach blastocyst stage, based on four parameters: the automatically recorded parameters P2 (t3-t2) and P3 (t4-t3), plus oocyte age and number of cells. Previous bibliography suggests that the Eeva scoring, in combination with morphologic evaluation, might also be predictive of implantation and live birth potential. However, the efficacy and reproducibility of this algorithm need to be validated with external data. Study design, size, duration Retrospective, observational cohort study, performed over 3736 embryos from oocyte donation cycles and 1291 embryos from autologous cycles with preimplantation genetic testing for aneuploidies (PGT-A) performed in a single IVF clinic over 3 years. All embryos were scored by the Eeva Test on day 3. The performance of the Eeva-Xtend scoring system as predictor of blastocyst development, euploidy, implantation and live birth was assessed. Participants/materials, setting, methods Embryo development was monitored by a time-lapse system. Embryo selection was performed by morphological (ASEBIR) criteria, resulting in the transfer of 959 embryos. The correlation between the Eeva-Xtend scoring, individually and combined with morphological classification, and the study outcomes was quantified by generalized estimating equations (GEE) including main possible confounders, and expressed in terms of odds ratio (OR). The performance of the GEEs was quantified and compared through the Area Under the ROC Curves (AUCs). Main results and the role of chance A positive association (P<0.001) was confirmed between lower Eeva scores and higher odds of reaching blastocyst stage (OR (1 vs 5) = 15.849, 95%CI 12.510-20.078; OR (2 vs 5) = 11.592, 95%CI 9.229-14.560; AUC = 0.768 (95%CI 0.754−0.783)), and consistent in both oocyte donation and PGT-A cycles. Lower morphokinetic embryo scores also resulted significantly associated with higher chances of implantation (OR (1 vs 5) = 3.385, 95%CI 1.507-7.605; P = 0.003) and live birth (OR (1 vs 5) = 5.132, 95%CI 2.089-12.605; P<0.001) in oocyte donation cycles, but not in embryos subjected to PGT-A: OR (1 vs 5) = 2.089, 95%CI 0.321-13.614, P = 0.441 for implantation and OR (1 vs 5) = 0.916, 95%CI 0.161-5.205, P = 0.921 for live birth. In embryos subjected to PGT-A, the automatic scoring system did not result significantly associated with the euploidy status: OR (1 vs 5) = 0.755 (95%CI 0.255−0.981; P = 0.489). In global, the Eeva Test had similar performance than morphological evaluation for implantation (AUC = 0.610 (95%CI 0.572−0.648) vs 0.606 (95%CI 0.568−0.644), respectively) and live birth (AUC = 0.622 (95%CI0.584−0.660) vs 0.609 (95%CI 0.571−0.647)) prediction. However, the highest performance was achieved when combining both scoring systems: AUC for implantation = 0.629 (95%CI 0.592−0.666), AUC for live birth = 0.636 (95%CI 0.598−0.673). Limitations, reasons for caution The retrospective nature of the study introduces some variability in the characteristics of the sample. However, the large sample size partially overcomes said limitation. The automatic annotations of the morphokinetic parameters were not validated in this study. Wider implications of the findings Our results support the applicability and reproducibility of this automatic early embryo evaluation system. Furthermore, our findings support that the use of this automatic embryo evaluation system in combination with morphological evaluation can potentially reduce subjectivity, improve embryo evaluation and increase success rates in IVF treatments. Trial registration number not applicable
Read full abstract