The impact of CHLOE-EQ and embryologist seniority on the ability and confidence to predict ploidy

Study question Can embryologists and CHLOE-EQ predict ploidy? Does their confidence and ability to predict vary with embryologist seniority? Summary answer High inter-observer variability between embryologists on prediction of ploidy. CHLOE-EQ provided a consistent prediction of ploidy. What is known already Previous studies have demonstrated the value of using Artificial Intelligence (AI)-based tools, such as CHLOE-EQ (Fairtility), to support, quantify and standardize embryo assessment. CHLOE-EQ uses AI-based algorithms to predict implantation. Recent studies have demonstrated that the algorithms also have ploidy predictive capabilities. Little is known about the ability of human embryologists to predict ploidy of blastocysts deemed suitable for biopsy, and whether this ability to predict varies with seniority or confidence level. Study design, size, duration Cohort study including 141 patients treated in a Memorial Sisli Hospital ART and Reproductive Genetics Center between January 2020-August 2022, with at least 4 blastocysts with different PGT-A results per cycle, leading to a total of 734 embryos. The same blastocysts were blindly assessed by CHLOE-EQ and by a senior and a junior embryologists working in the same clinic. Intraobserver variance of senior embryologist was also evaluated. Participants/materials, setting, methods Embryologists were asked to predict whether a blastocyst was euploid or aneuploid, their confidence of this determination (confident, neutral, not confident), the rank of the embryos based on chance of being euploid. The same embryos were assessed using CHLOE-EQ, scoring embryos from 0 to 10. The extent of mosaicism (from 35-70%) was quantified. Trophectoderm biopsy was performed to embryos at least 3BB, four low quality blastocysts (Aneuploid:5AC,4AC,6CB; Euploid:2AA) were also included into the study. Main results and the role of chance The average patient age was 33 ± 4 years (ranging from 22 to 39 years). Embryologists agreed on euploid prediction in a minority of blastocysts(48%,324/670). Agreement of the senior embryologist on same embryos in a different time was higher (Kappa = 0.42, moderate) than with the junior embryologist [Kappa = 0.19(slight)/0.26(fair)], suggesting seniority affected prediction consistency. CHLOE-EQ ranking had fair agreement with the embryologists (Kappa = 0.22,0.22,0.23), bringing consistency to prediction irrespective of seniority. Confidence was not affected by seniority (senior vs junior: ‘Confident’/’Neutral’/’Not confident’:60/26/14% vs 57/27/16% vs 66/24/10%,NS). Efficacy of prediction of ploidy reduced with junior embryologist (senior: AUC = 0.58,0.57 vs junior:AUC = 0.55). The senior embryologist was able to predict ploidy with a greater accuracy when ‘confident’ (AUC = 0.60, n = 441, p<0.001) compared with ‘not confident’ (AUC = 0.50, n = 107, NS). This was not the case with the junior embryologist (‘Confident’ AUC = 0.56,NS vs ‘Not Confident’ AUC = 0.54,NS). Euploidy rate was greater in high scoring embryos(CHLOE-EQ 5.1-10) than low(0-5) scoring (60%, 272/457 vs 48%,135/280;p<0.005,AUC = 0.58). This was maintained in blastocysts where the senior embryologist was ‘confident’(64% vs 54%) and ‘not confident’ (57% vs 36%), showing consistency in assessment irrespective of confidence. CHLOE-EQ score reduced with increasing degree of mosaicism (Euploid 6.3 ± 2; Mosaic<35% 5.98 ± 2.87; Mosaic>50% 4.31 ± 2.80; Mosaic>70% 2.85 ± 4;p = 0.04). Euploid embryos had a higher CHLOE-EQ score than aneuploid/mosaic embryos (6.3 ± 2, n = 396 vs 5.78 ± 2,n = 319;p = 0.008). Limitations, reasons for caution Mostly optimal quality blastocysts were included in this study. There is a need to better understand the role of Artificial Intelligence in improving consistency of selection of blastocysts for biopsy, and to extend this study for viable lower quality blastocysts to be included rather than discarded. Wider implications of the findings Effective prediction of ploidy can (i) improve biopsy criteria so that viable embryos are not discarded, (ii) reduce the cost of PGT-A by prioritizing embryos with increased chances of being euploid. CHLOE-EQ’s ploidy prediction improved consistency can support both PGT-A programs, and cycles where PGT-A is not an option. Trial registration number Not applicable