P-194 Developing a classification system for euploid embryos: leveraging morphology grades and biopsy day to accurately rank them based on ongoing pregnancy potential

Abstract

Abstract Study question Can morphology grades and day of trophectoderm (TE) biopsy be utilized to accurately rank euploid blastocysts according to their potential for ongoing pregnancy (OP)? Summary answer The proposed ranking system categorized euploid blastocysts accurately in terms of their ongoing pregnancy potential. What is known already Implementing a ranking for embryo selection within a pool is a valuable optimization strategy. Many studies have documented the importance of inner cell mass (ICM) and trophectoderm (TE) grading, as well as the day (D) at which blastocysts are biopsied on clinical outcomes. However, literature sparks controversy on clinical findings in relation to blastocyst quality. This is primarily because it reports a diversity in blastocyst grade-ranking systems, and none of them accounting for the day of biopsy. Enhancing consistency and comparability across various research studies could be achieved by establishing a standardized approach for qualitatively classifying and ranking euploid blastocysts. Study design, size, duration This retrospective cohort study included 2209 single euploid frozen embryo transfers (SeFET) between 2017 and 2023. Blastocyst (BL) expansion, TE and ICM were assessed before biopsy according to Gardner’s criteria. Only blastocysts graded ≥BL3CC were biopsied on D5, D6 or D7. Classification models (in-house, logistic regression, gradient boost) were assessed for their calibration, i.e., correct prediction of ongoing pregnancy (OP) probability. Participants/materials, setting, methods In-house ranking system categorized euploid transferred blastocysts (BL3-BL6) as ‘Top-quality’ for blastocysts AA/AB/BA on D5; ‘Good-quality’ for blastocysts BB on D5 and AA/AB/BA/BB on D6; ‘Fair-quality’ for AC/BC/CA/CB on D5, BC on D6 and AA/AB/BA/BB on D7; and ‘Poor-quality’ for blastocysts CC on D5, AC/CA/CB/CC on D6 and CA/CB/AC/BC/CC on D7. The in-house classification ranking system was compared to other developed models to evaluate whether performance could be increased. Main results and the role of chance A total of 406 blastocysts were ranked as Top-quality (100% D5), 1323 as Good-quality (51.3% D5 and 48.7% D6), 190 Fair-quality (57.9% D5, 36.3% D6 and 5.8% D7) and 290 as Poor-quality (12.8% D5, 73.8% D6 and 13.4% D7). The ongoing pregnancy rate was significantly higher for blastocysts ranked as Top compared to Good, Fair and Poor-quality (58.9%, 52.2%, 35.8% and 26.6%, respectively). Miscarriage rates were significantly associated with ranking quality (20.9%, 24.3%, 30.6% and 39.4% for Top, Good, Fair and Poor-quality blastocysts, respectively). We further developed two new classification ranking systems using logistic regression model and XGBoost machine learning model. But the new models provided similar AUC and calibration slope (CS) values as the in-house ranking system (AUC [CS]: 0.604 ± 0.0167 [0.962 ± 0.213] and 0.601 ± 0.0177 [0.949 ± 0.219] vs.0.597 ± 0.015 [0.980 ± 0.230] for logistic regression, XGBoost versus in-house model, respectively). Limitations, reasons for caution The present embryo classification ranking system may not be extrapolated to non-PGT-A cycles. Wider implications of the findings Different grade-ranking systems of blastocysts keep being reported in the different publications. Reaching a consensus to rank euploid blastocysts qualitatively, as the hereby proposed classification, may improve consistency and comparability across different research studies. Crucially, the present ranking-system aids embryo selection for transfer from available pool. Trial registration number not applicable