The use of morphokinetics as a predictor of implantation: a multicentric study to define and validate an algorithm for embryo selection

Abstract

STUDY QUESTION: Can we use morphokinetic markers to select the embryos most likely to implant and are the results likely to be consistent across different clinics? SUMMARY ANSWER: Yes, morphokinetic markers can be used to select the embryos most likely to implant and the results were similar in different IVF clinics that share methods and organization to some extent. WHAT IS KNOWN ALREADY: With the introduction of time-lapse technology several authors have proposed the use of kinetic markers to improve embryo selection. The majority of these markers can be detected as early as Day 2 of development. Morphology remains the gold standard but kinetic markers have been proven as excellent tools to complement our decisions. Nevertheless, the majority of time-lapse studies are based on small data sets deriving from one single clinic. STUDY DESIGN, SIZE, DURATION: Retrospective multicentric study of 1664 cycles of which 799 were used to develop an algorithm (Phase 1 of the study) and 865 to test its predictive power (Phase 2 of the study). PARTICIPANTS/MATERIALS, SETTING, METHODS: University-affiliated infertility centres patients undergoing first or second ICSI cycle using their own or donated oocytes. Embryo development was analysed with a time-lapse imaging system. Variables studied included the timing to two cells (t2), three cells (t3), four cells (t4) and five cells (t5) as well as the length of the second cell cycle (cc2 = t3 - t2) and the synchrony in the division from two to four cells (s2 = t4 - t3). Implantation (IR) and clinical pregnancy (CPR) rates were also analysed. MAIN RESULTS AND THE ROLE OF CHANCE: During Phase 1 of the study we identified three variables most closely related to implantation: t3 (34-40 h), followed by cc2 (9-12 h) and t5 (45-55 h). Based on these results we elaborated an algorithm that classified embryos from A to D according to implantation potential. During Phase 2 of the study the algorithm was validated in a different group of patients that included 865 cycles and 1620 embryos transferred. In this phase of the study, embryos were categorized based on the algorithm and significant differences in IR were observed between the different categories ('A' 32%, 'B' 28%, 'C' 26%, 'D' 20% and 'E' 17%, P < 0.001). In addition we identified three quality criteria: direct cleavage from one to three cells, uneven blastomere size in second cell cycle and multinucleation in third cell cycle. LIMITATIONS, REASONS FOR CAUTION: The retrospective nature of the study limits its potential value, although the use of one database to generate the algorithm (embryos from this database were not selected by any morphokinetic criteria) and one database to validate it reinforces our conclusions. WIDER IMPLICATIONS OF THE FINDINGS: The elaboration of an algorithm based on a larger database derived from different (albeit related) clinics raises the possibility that such algorithms could be applied in different clinical settings.