This year, Denver was the place to be when participating in the annual meeting of ASRM. Similar to previous years, IVF professionals from all over the world gathered to present or listen to the latest scientific developments in the field of assisted reproduction. We have collected some of our thoughts and reflections of the scientific content of the meeting in this blog post.
Artificial Intelligence for embryo selection
Exciting new research at ASRM included the use of AI to analyse time-lapse images and movies generated by EmbryoScope or EmbryoScope+. Two oral abstracts and four posters showed promising data for the future of AI for embryo assessment with time-lapse imaging.
An oral abstract (O-265) was presented by Aengus Tran from Harrison AI, in collaboration with Virtus Health groups in Australia.They took images of over 8,000 embryos of known outcome, from 8 IVF laboratories over 4 different countries, and used a supercomputer to look at the development of human embryos. By applying deep learning to examine over 7 million images, an algorithm was developed capable of determining the most viable embryo with an AUC of 0.93. This exciting development in AI, known as IVY, is now being validated in a prospective randomised trial in Australia.
In a second oral abstract (O-209), presented by Nikica Zaninovic (Weill Cornell Medicine) an AI convolutional neural network was used to analyse EmbryoScope images of good, fair and poor quality embryos based on embryo grade and implantation outcome data. The AI was able to accurately identify blastocyst quality at 110 hours post insemination.
A poster (P-629) from Weill Cornell Medicine also used an AI convolutional neural network to analyse 661,000 images of 11,898 human embryos to identify division times with a 91% accuracy of cell-stage annotations within 5 frames of embryologists’ annotations.
Two further posters representing an international collaboration between researchers from Weill Cornell Medicine, Sao Paulo State University in Brazil and Imperial College London (P-663) showed that a combination of morphokinetic parameters along with AI analysis of blastocyst images taken at 111 -111.5 hours post insemination had an excellent ability to predict implantation. IVI-RMA Valencia was included in the second collaborative effort (P-662) to show that live birth could also be predicted with high accuracy.
The Women’s Health Institute at Cleveland Clinic teamed up with the Computer Science department at UCSD and the Machine Learning Department at Carnegie Mellon to predict morphokinetic annotations of time-lapse videos using a convolutional neural network (Gingold et al. P-282).The CNN predicted division events from pronuclear to expanded blastocyst stage with high accuracy. However, the model consistently misclassified embryos with cellular fragments or which had undergone chaotic divisions. Future work will refine the models and work towards prediction implantation potential.
Time-lapse variables linked to implantation and pregnancy rate
Other new time-lapse variables linked to implantation potential were presented by Lorena Bori of IVI-RMA (O-205). Migration distance and time of pronuclei, measurement of blastocyst expansion, size of ICM and cell cycle length of trophectoderm cells were all analysed using the EmbryoViewer software and shown to correlate to implantation potential.
Mitsuru Kobayashi et al. from Kanagawa Ladies Clinic showed that the morphokinetic intervals of blastulation start to blastulation and expanded blastulation were correlated with clinical pregnancy rate in good quality frozen blastocysts (P-660).
Minimising your mosaicism rates by taking extra care of your embryos in the laboratory and what should you consider when all embryos are classified as mosaic
Mandy Katz-Jaffe presented evidence that laboratory conditions can increase the level of mosaicism in embryos (O-268). Vitrified zygotes (donated with patient and IRB consent) were warmed and then cultured under 20% O2 and reduced CO2 (effectively increasing the pH of the medium to around 7.9). The incidence of mosaicism was significantly higher (20%) compared to embryos that had been cultured under optimal conditions (2.3%). The majority of mitotic errors present in those embryos that developed under sub-optimal conditions were due to anaphase lag, resulting in a chromosome separation error. This research further supports the importance of culture in a reduced oxygen environment.
Embryonic mosaicism is being actively debated in the field at the moment. The poster from Zouves Fertility Center (P-789) offers some guidance into how to handle those embryos classified as mosaic. According to the poster some mosaic embryos are capable of producing healthy babies, so can be considered for transfer when there is no euploid embryo to transfer. Embryos with a single segmental abnormality should be considered for the first transfer when only embryos classified as mosaic are available.
Intention to treat analysis shows that PGT-A improves clinical outcomes
PGT-A allows for embryo selection to decrease pregnancy loss and increase live birth rates, however, it does mean that some patients do not have an euploid embryo to transfer. An intent to treat analysis performed by the Colorado Center for Reproductive Medicine (O-102) demonstrated that compared to using morphology alone for embryo selection, PGT-A still significantly improved clinical outcomes in maternally age-matched patients. This supports the decision to consider the use of PGT-A with patients undergoing IVF.
Polygenic disorder testing and non-invasive PGT-A and PGT-M could be closer to becoming a reality
Validation of a combined method allowing aneuploidy testing, monogenic and polygenic disorder testing was presented by Nathan Treff (Genomic Prediction) (0-248). A combination of cell line analysis and re-biopsied embryos were used to demonstrate that such disorders can be accurately predicted. Such findings suggest that current PGT methodologies could be modified, enabling a reduction in the risk of inherited genetic disorders.
A poster (P-786) presented by Mandy Katz-Jaffe demonstrated that amplification of the DNA in the media drop was achieved in all samples. Almost 90% of the time there was sufficient DNA to obtain a reliable PGT-A result. A 90% concordance with the TE biopsy was observed for embryo sex. The discordance was explained by the fact that some male embryos were being diagnosed as female due to the presence of maternal DNA contamination from cumulus cells. 80% of the media samples were concordant with the biopsy result for the PGT-A result. The reason for some discordance was the expected reverse aneuploidy to that found in the biopsy sample.
Christine Robinson (RHS) showed that non-invasive pre-implantation genetic testing using spent culture media could also include monogenic disease detection (P-762). Targeted sequence enrichment enabled both detection of BRCA 1 and Haemoglobulin subunit beta as well as aneuploidy information.
Protect your gametes and embryos against reactive oxygen species
A prospective RCT performed in two clinics in Sweden (O-115) was presented by David Gardner. The study demonstrated that the inclusion of antioxidants in the culture medium did not significantly impact embryo development, although there was a trend toward increased clinical pregnancy rate. The same antioxidant strategy was assessed using sperm survival under sub-optimal conditions (O-42). Marius Meintjes used a 5 day sperm survival assay with no oil overlay or poor quality oil. It was observed that the presence of antioxidants maintained significantly higher sperm motility.
Mitochondria could be a specific target for antioxidants (O-174). In a mouse model Jack Becker (Colorado Center for Reproductive Medicine) showed that although embryo development was not altered by mitochondrial targeted antioxidants the number of inner cell mass cells was significantly higher. Furthermore, the mitochondrial DNA copy number was lower in those embryo exposed to antioxidants, which is thought to be an indicator or viability.
Mouse Embryo Assay, when optimised for sensitivity, is a critical tool for identifying toxicity
Grimm et al. (Colorado Center for Reproductive Medicine) presented a poster (P-281) investigating the performance of disposable plastics routinely used in the ART laboratory using a mouse embryo assay (MEA). Various plasticware was exposed to culture media at 37° C for an extended period. The exposed media was then used to culture outbred mouse zygotes. Four of the items tested (40%) did not meet the acceptance criteria indicating the presence of toxicity. This investigation outlines the importance of the mouse embryo assay as a tool for detecting toxicity. The use of sensitive mouse strains, multiple relevant endpoints and specialized assay design can increase the sensitivity of the MEA. Sensitive bioassays for screening are crucial to ensure that only safe, non-toxic disposables are introduced into the culture system.
Vitrification of Macaque ovarian cortex
Mary Zelinski (Oregon Health and Science University) presented very interesting research on vitrification of Macaque ovarian cortex, followed by warming and transplantation into heterotropic sites and production of fertilizable oocytes (O-127).
Fortunately Zelinski and colleagues used ethylene glycol as their cryoprotectant because O-64 demonstrated that DNA methyltransferases (Dnmts) in mouse oocytes are altered by DMSO. Gary Smith (University of Michigan) stated that while DMSO is an efficient membrane permeable cryoprotectant, its safety has been questioned as it has been demonstrated that it can impact DNA methylation and epigenetic profiles in mouse cells. Dnmts are found in all murine oocyte stages. When MII oocytes were exposed to DMSO, Dnmts1 and oocyte specific Dnmts1o expression were altered significantly compared to MII oocytes not exposed to DMSO. The authors indicate that DMSO exposure is a relevant concern as it can impact embryonic methylation epigenetic profiles, offspring health and transgenerational epigenetics.
Fertility preservation on the agenda
In the symposium "Technological Advances in Fertility Preservation: from Spermatogonial Cell Transplants to Bioprosthetic", Monica Laronda (Northwestern Medicine) gave insight into bioprosthetic ovaries, which were produced in her lab. Essentially they created a 3-d scaffold in a 3-D printer that was colonised with mouse primordial follicles and was then transplanted into ovarectomised mice. The structure supported ovary-like tissue growth and the mice became cyclic, pregnant and delivered healthy offspring. The offspring also gave birth, showing that there is no trans-generational negative effect.
The second talk by Hanna Valli-Pulska (University of Pittsburgh) focused on male fertility preservation in prepubertal primates using two different approaches. In one approach they injected spermatogonial stem cells into sterile Macaque, which caused a re-popularisation of the seminiferous tubules in the testis and re-initiation of spermatogenesis. In another approach they transplanted frozen-thawed prepubertal testicular tissue. The sperm cells were isolated and used for ICSI and embryo transfer, resulting finally in the birth of a female Macaque, the first of its kind using such a technology in primates.
Find more information on ASRM's website or download our Selection of Abstracts below. And finally we would like to thank all of you who came to our booth during ASRM. We look forward to seeing you again.
Topics: IVF community insights