O-071 Stringent statistical analysis of 108,684 oocyte donation cycles identifies 2.6% of donors with specific oocyte/embryo developmental arrest phenotypes for genome association studies

Abstract

Abstract Study question Can statistical modelling accurately identify an outlier group of presumed fertile donors with specific reduced embryological outcomes suggestive of an underlying infertility-related genomic condition? Summary answer Statistical modelling identified a highly specific group of donors as embryologically under-performing outliers across multiple treatment cycles for the optimisation of downstream genomic association studies. What is known already Female infertility has a wide range of clinical presentations. While some causes of infertility are well defined, the aetiology of most infertility remains unexplained. Since IVF allows for in vitro monitoring of oocyte and embryo, it is now possible to identify subtle infertility endophenotypes, characterized by specific developmental failure that would otherwise go undetected in natural conception. Currently, the lack of large and well-categorized datasets of infertility endophenotypes has limited research in the area. Therefore, a clear and methodological definition of infertility endophenotypes in IVF is the first crucial step towards a better understanding of their mechanism and biological/genetic aetiology. Study design, size, duration A retrospective analysis was performed on clinical and embryological data from 40497 egg donors, undergoing 108.684 stimulations, with 4.235.036 oocytes retrieved. The data also included information on the 85482 recipients and their 190996 oocyte reception cycles, spanning from 1999 to 2023. Infertility endophenotypes were able to be well categorized as oocyte donors are presumed fertile based on their young age and on normal ranges displayed for the parameters assessed during typical oocyte donation work-up. Participants/materials, setting, methods Donor performance was calculated according to: (I) oocyte maturation rate (MR); (ii) survival rate of cryopreserved oocytes (SR); (iii) oocyte fertilization rate (FR;2PN/Injected MII); and (iii) blastulation rate (BR;blastocyst/2PN). Donors with significantly impaired rates, considering ≥2 stimulations and/or recipients, were identified as outliers using probabilistic modelling (Binomial test, Bonferroni corrected P < 0.05). Outliers were computed for each rate on data subsets after applying this stringent filtering criteria and excluding inconsistent or missing data. Main results and the role of chance 2.6% of donors were identified as underperforming outliers when summing the rate observed for each embryological outcome using the stringent statistical modelling. The average MR was 79.8% (95%CI: 79.6-80). Taking into account the correlation of follicle size during ultrasound tracking and the number of mature oocytes collected, a total of 168/29768 (0.56%) donors with statistically lower MR (mean MR = 43% 95%CI:40.1-45.7) across ≥2 egg collection cycles were identified as outliers. The average SR was 88% (95%CI:87.9-88.2). A total of 176/18544 (0.95%) donors with ≥2 stimulations and ≥2 recipient cycles were deemed outliers, showing a statistically reduced SR (mean SR = 51% 95%CI:48.9-53.1). The average FR was 74% (95%CI: 73.8-74.2). 164/31657 (0.52%) donors were found to have statistically low FR (mean FR = 37% 95%CI:34.6-38.7) and deemed outliers when assessed over ≥2 stimulation cycles with ≥2 recipient cycles. The average BR was 53.4% (95%CI: 53.1-53.7). After considering only donors with ≥2 recipients and ≥2 stimulations, a total of 142/24716 (0.57%) donors with statistically low BR (mean BR = 14% 95%CI:12.4-15.3) were considered outliers. Limitations, reasons for caution The stringent criteria applied to our statistical modelling has likely resulted in underestimating the true level of outlier donors. This is also likely compounded by the selection bias associated with donor recruitment and the retention policies after the first stimulation cycle preventing the evaluation of recurrent patterns in some cases. Wider implications of the findings The analytical tools developed on this unprecedentedly large dataset are fundamental to define a methodological workflow for accurately identifying unexpectedly infertile individuals. This is crucial for investigating the genetic and biological mechanisms underlying infertility endophenotypes, leading to improvements in treatment and developing screening tool to predict infertility and related pathologies. Trial registration number not applicable