Comprehensive artificial intelligence-powered investigation of blastocyst expansion dynamics: associations with competence

Abstract Study question Are blastocyst expansion dynamics from time of starting blastulation (tSB) to time of biopsy (t-biopsy) indicative of embryo competence? Summary answer Early expansion dynamics across 5hours after tB, embryo-proper area (emb-A), zona-pellucida-area (zp-A), ZP-thickness (zp-T) at t-biopsy and their t-biopsy/tB ratios are associated with competence. What is known already Blastocyst expansion is the very first morphogenetic event common to several species. Time-lapse-technology (TLT) implementation in IVF allowed deeper understanding of blastocyst expansion process. Some studies leveraged TLT and Artificial-Intelligence to investigate blastocyst expansion timings and dynamics for their association with embryo competence. Huang’s group, in particular, designed a quantitative standard expansion assay (qSEA) showing promising results. However, data about qSEA reproducibility are missing. Here we comprehensively investigated the expansion processes between tSB and t-biopsy through Artificial-Intelligence, and adapted the qSEA to our setting that encompasses PGT-A without day3-hatching and single-euploid-blastocyst-transfer. Study design, size, duration Retrospective study including 2184 blastocysts cultured in EmbryoScope during 786 PGT-A cycles conducted across 2013-2020. Videos were analyzed through an Artificial-Intelligence-powered tool (CHLOE™, Fairtility). The software automatically extracted timings in hours-post-insemination and measures as proportions of video frames occupied by each feature under investigation (single pixel = 300µm; wells’ area = 90,000µm2) recorded every 30min from tSB. These data were tested for their association with euploidy and live-birth after 548 euploid transfers via multivariate regressions. Participants/materials, setting, methods ICSI, trophectoderm biopsy on fully-expanded blastocysts without day3 zp drilling, and qPCR/NGS to assess full-chromosome non-mosaic aneuploidies were performed. The timings assessed were tSB, tB, tEB and t-biopsy ( = end of video). At these timings and every 30min across 5hours after tSB the software recorded the following measures emb-A, zp-A, zp-T, inner-cell-mass area (ICM-A), and ICM-to-trophectoderm ratio. Also increase/decrease ratios for all measures between timings were assessed. Putative confounders (e.g., maternal age, blastocyst quality) were considered. Main results and the role of chance Larger emb-A and zp-A at t-biopsy and zp-T at both tEB and t-biopsy were associated with euploidy. Similarly, the ratios zp-A at t-biopsy/tB and t-biopsy/tEB highlighted a larger expansion among euploid versus aneuploid blastocysts. All these differences were confirmed when adjusting for maternal age, morphological quality and tB (p < 0.01). zp-A t-biopsy/tB ratio (aneuploid:+68.8% versus euploid:+79.9%) showed a more relevant association than final zp-A at t-biopsy per se (24082 ± 5763 versus 25438 ± 5969µm2). The ratios zp-T at t-biopsy/tB and t-biopsy/tEB were also significantly associated with euploidy, even when adjusting for confounders (p < 0.01). In this case, zp-T at t-biopsy (8.1 ± 3.2 versus 7.1 ± 2.7µm) per se showed a stronger association than zp-T t-biopsy/tB ratio (-50% versus -55%). ICM-A and ICM-to-trophectoderm ratio showed no association with euploidy. All features showed no association with LBs (N = 233/548 euploid transfers). The qSEA every 30min across 5hours after tB outlined different early expansion dynamics between euploid and aneuploid blastocysts, with the former expanding more (larger areas and thinner zp) and sooner. The differences became significant already after 2.5-3 hours, due to rather constant expansion rates in both groups, but faster among euploid. The same significant trend was reported for euploid blastocysts resulting in a LB versus not. Limitations, reasons for caution Retrospective single-center study. Previous studies on qSEA were based on 10 measurements every hour from tB, instead of 10 measurements every 30min. To properly assess the association between expansion dynamics and timings with LB, more transfers are required. To outline a predictive power, instead, a prospective randomized design is warranted. Wider implications of the findings Blastocyst expansion dynamics, timings and ratios measured through Artificial-Intelligence, already during the 5hours following tB, provide objective quantitative data associated with embryo competence. qSEA is a promising clinical strategy, user-friendly and easily applicable, that deserves further appraisal. Basic research on the mechanisms that govern blastocyst expansion processes is warranted. Trial registration number None