Abstract Study question Is it possible to score the mosaic embryos in order to prioritise them according to their pregnancy potencial? Summary answer The machine-learning score underscores the crucial role of embryo quality over mosaicism, prioritizing high-quality, day-5 biopsied embryos for enhanced live birth outcomes in ART. What is known already After embryo biopsy and PGT-A analysis, embryos can be classified as euploid, aneuploid, or mosaic, wherein a combination of chromosomally normal and altered cells is detected. The transfer of mosaic embryos has been controversial, but currently, the recommendations of international scientific societies, including ESHRE, advocate for their transfer. In case of pregnancy, a prenatal test is advised to rule out any chromosomal aberration. The non-transfer of these embryos reduces ART success chances, as chromosomally normal children may be born from them. Study design, size, duration The study design is observational and retrospective. A total of 8346 embryos from 2583 PGT-A cycles were considered (January-2017 to January-2023). Only transferred mosaic embryos (n = 263) were included in the study. The trophectoderm biopsies of D5, D6, or D7 blastocysts were analyzed by NGS using the Illumina platform (VeriSeq Illumina®, San Diego, CA, USA). The biopsied embryos were vitrified and transferred in a subsequent cycle. Participants/materials, setting, methods The primary indications for PGT-A included advanced maternal age, abnormal sperm FISH, and recurrent miscarriage or implantation failure. Clinical outcomes, along with variables related to progenitors, embryos, biopsy, ovarian stimulation, and adjuvant treatments, were recorded in a database. Machine learning models were conducted using R statistical software (v.4.3.1), and the ‘Caret’ library was utilized to implement different algorithms. ‘Autoscore,’ as an interpretable scoring tool, combined machine learning and regression modeling. Main results and the role of chance The mean maternal and paternal ages were 34.95±6.38 and 39.13±6.74 years. Biopsies were predominantly performed on day 5 (61%). Additionally, 44% of embryos were of A quality, and 49% were of B quality. Regarding clinical outcomes for mosaic embryos, biochemical, clinical pregnancy, and live birth rates were 51%, 41%, and 36%, respectively. The features used to generate the score included those related to embryo quality, biopsy, and mosaicism. Through 10-fold cross-validation, the study identified top-ranking predictors maximizing the AUC of the predictive model. The highest weight in the score was assigned to embryo quality (42%) and day of biopsy (34%), with lower weights for factors related to mosaicism: monosomy and/or trisomy (18%) and mosaicism level (6%). The machine-learning score indicates that, in transferring embryos with mosaicism, priority should be given to high-quality embryos biopsied on day 5. The type and level of mosaicism have a minor impact on the embryo’s gestational capacity, becoming relevant only when prioritizing embryos of identical quality. In such cases, embryos with lower mosaicism degrees and no monosomy demonstrate higher implantation potential. Limitations, reasons for caution The inherent limitations of a retrospective analysis highlight the need for prospective randomized studies to confirm the validity and usefulness of the mosaic embryo transfer score. Wider implications of the findings A scoring system has been developed to prioritize the transfer of mosaic embryos. The crucial variables are associated with embryo quality and the day of biopsy, reflecting the progression of embryonic development. Conversely, variables related to mosaicism play a secondary role. Trial registration number Not applicable
Read full abstract