Predictive value of blastocyst morphology beyond aneuploidy testing: evidences from a randomized non-selection study

Study question: Does blastocyst stage standard morphology assessment enhance embryo selection beyond aneuploidy testing? Summary answer: Blastocyst morphology based selection does not significantly enhance live-birth rate in pre-implantation genetic diagnosis cycles for aneuploidy testing (PGD-A) nor predict adverse pregnancy outcomes. What is known already: Trophectoderm (TE) biopsy and aneuploidy testing is increasingly used to enhance embryo selection in IVF cycles. However, beyond euploidy, no other parameters have been assessed to be used in combination with aneuploidy testing to further enhance the clinical management of PGD-A cycles. Blastocyst morphological evaluation has been extensively validated to predict implantation in IVF cycles and this association has been recently linked to its moderate predictive value on blastocyst aneuploidies. However, it is still unknown whether euploid blastocysts with different morphologies and development implant at a different rate or provide a different risk for biochemical losses, miscarriages or gestational complications. Study design, size, duration: Randomized non-selection study preformed between December 2013 and December 2014 including all consecutive blastocyst stage PGD-A cycles performed in infertile patients of advanced female age (35 years). In the randomization period, whenever more than one euploid blastocyst of different morphology was obtained, the embryo selection was subjected to randomization to rule out patient’s specific factors (Study group). A subsequent follow-up period (Genuary 2015) where the selection between euploid embryos was based on blastocyst morphology was included (control group). Participants/materials, setting, methods: ICSI and embryo culture in 5% O2/6% CO2. Prior to trophectoderm biopsy, morphology was assessed and categorized [excellent (≥3AA; group A)/good (3, 4, 5, 6, AB and BA; group B)/ average (3, 4, 5, 6 BB, AC and CA; group C)/poor quality (≤3BB)] according to Gardner/Schoolcraft’s criteria. Developmental rate was defined according to the day of biopsy post-fertilization (day 5, 6 or 7). Primary outcome was live-birth rate/transfer. Biochemical-losses, miscarriages and gestational outcomes were also compared between morphological classes using logistic regression models adjusted for potential confounding factors. Main results and the role of chance: In the randomization period 171 cycles were excluded from the analysis since only one euploid blastocyst was obtained. 339 single euploid embryo transfers in 196 patients (female age 37.2 3.1) were subjected to randomization since more than one euploid blastocyst (mean per cycle 2.8 1.1) with different morphology (N 199; 104 excellent, 40 good, 41 average, 20 poor) or with different time of development (N 159; 73 day 5, 72 day 6, 14 day 7) was obtained for that cycle. Logistic regression analysis showed that poor morphology class (p 0.01; OR 0.09, 95%CI 0.01–0.77; live-birth rate: 45.2%, 42.5%, 48.8% and 7.1%, for excellent, good average and poor, respectively) and day 7 embryos (p 0.01; OR 0.16; 95%CI 0.03–0.85; live birth rate: 47.9%, 33.3% and 14.3% for day 5, 6 and 7, respectively) were significantly associated with reduced live-birth rate. Female age, sperm parameters, previous IVF failures and miscarriages showed no association. Logistic models showed that morphology and day of development were not related to biochemical losses, miscarriages and gestational outcomes. In the follow-up period, 151 euploid SET were performed where morphological selection was possible. Even though higher percentage of excellent/good quality blastocysts were transferred in the selection group (76% vs 64.5%, respectively, p 0.01), this did not translate in improved live-birth rate (46.3% vs 41.9% respectively; p 0.2). Limitations, reasons for caution: Not all morphological classes were represented in each cycle when the randomization was performed. Not all euploid embryos from a stimulation cycle were transferred. Even if based on a large dataset, this study is still not sufficiently powered to exclude moderate associations between blastocyst morphology and biochemical losses, miscarriages and neonatal outcomes. Wider implications of the findings: This study provides class-one data about the limited value of blastocyst morphology selection beyond euploidy. Only very poor morphology and day-7 blastocysts showed lower potentiality. These embryos are usually discarded. However, when euploid they still provide live-births with no increase adverse outcomes strongly suggesting their use in PGD-A cycles. Trial registration number: ISRCTN81216689.