Thoughts and reflections from ESHRE 2020

The 36th annual meeting of ESHRE took place online this year, with more than 12,000 registered participants. Like last year, the Vitrolife team has put together our thoughts and reflections on some of the scientific content from the meeting. We also offer the opportunity to watch our recorded scientific symposium again.

In the light of Covid-19

Split team working has been implemented to meet the requirements of social distancing to minimise the risk of viral transmission at the clinic (Bui et al., P-817). Technology and virtual consultations have been used where possible to continue treatment, but obviously patients have still been negatively impacted by the pandemic (P-819, 821-22, 824 and 826). The importance of the psychological impact of trying to overcome infertility during a pandemic was highlighted by Barra et al., (O-279) suggesting that clinics need to be more aware that there will be a need to strengthen psychological counseling for patients. To be able to confidently perform ART treatments, it will be crucial to know if and how the virus affects pregnancy and the health of the fetus/baby. Information is still scarce; however, infected pregnant women do not seem to be more severely affected by the disease compared to the general population. Recent systematic reviews suggest vertical transmission of the virus is low, but possible (Bahadur et al, O-316 and Adriana P-818). Registries in Europe and U.S.A are attempting to collect as much data as possible.

Comprehensive range of time-lapse and AI presentations

As in previous years, research using EmbryoScope and EmbryoScope+ time-lapse systems were well represented at ESHRE. This included 17 oral presentations and 35 posters.

The use of Artificial Intelligence based technology to improve IVF continues to be a hot topic. In all, there were 12 oral and 14 poster presentations which detailed use of AI for a range of applications including oocyte and embryo evaluation, patient treatment guidance, treatment outcome prediction, and euploidy prediction.

Chambost et al. (P-771) highlighted the need for standardisation in description of methodology and performance reporting in this rapidly growing field. An analysis of abstracts in the past two years found that most studies had insufficient data for the type of AI used and either lacked or used incorrect performance measures based on the claims.

Rimestad et al. (P-163) entitled “Robust embryo scoring model based on artificial intelligence (AI) applied to a large time-lapse dataset” was submitted by members of our own AI team, and was nominated for a basic science award. This data was also presented by our deep learning specialist Jens Rimestad, in our company symposium entitled “AI based embryo evaluation: Empower your decisions”. Our symposium also included a moving memorial to Michelle Lane presented by Prof. David Gardner, as well as a presentation about use of AI in the medical field by Dr. Aengus Tran of Harrison AI. 

Lorena Bori from IVIRMA, presented her oral presentation (O-114) “A universal algorithm is available in last generation time-lapse incubators: embryo score provided by the KIDScoreD5 is strongly correlated with chromosomal status and clinical outcomes”. The data validated the use of KIDScore D5 as an algorithm to prioritise transferring embryos which lead to implantation and live birth. A dataset of 14,603 embryos were annotated using Guided Annotation, followed by application of KIDScore D5 version 3 (the current available version). Embryos were split into 4 score categories with roughly equal data set sizes and known implantation and live birth were compared between score groups. Statistically significant differences were found in implantation, live birth and proportion of euploid, as KIDScore values categories increased.

There were several presentations which examined the ability to identify embryos with a higher probability of being euploid using a variety of methods.

Stefanie De Gheselle from UZ Ghent (O-113), presented data which made use of machine learning to optimise analysis of morphokinetic variables for prediction of euploidy. Machine learning techniques improved AUC of euploid prediction from 61% when using standards statistical methods, up to 72% with a machine learning approach.

Zaninoivic et al. (P-537), combined morphokinetic, morphological and patient meta data from a database of 10,491 blastocysts with known ploidy status with embryo morphokinetic and morphology annotations and associated patient clinical profiles. An algorithm developed using machine learning and deep learning, showed an overall accuracy of 70.4%. A defined threshold was able to identify chromosomal status with accuracy of 94% (for positive) and 95% (for negative) in 25% of the embryos.

Sopaboon et al., (P-185), analysed KIDScore values of embryos with known PGT results. As has been demonstrated previously, embryos in the highest KIDScore range (8-9.9), had a higher chance of being chromosomally normal. This study was based on a rather small dataset but seems to confirm previous data including the data presented by Lorena Bori (O-114).

In another study (P-165) Pelligrini et al., a retrospective analysis of embryos with known PGT results, and implantation outcome showed that embryos with a higher KIDScore had a higher chance of being euploid. Furthermore, implantation rate among euploid embryos was significantly higher in embryos with score 6 and above versus those with lower scores. They conclude that blastocyst selection through time-lapse technology alone should not be considered as a replacement for PGT, and that “KIDScore seems to be a potentially valuable tool for identifying the euploid blastocysts having a higher probability of implantation”. Of note, this study used a previous version of KIDScore. According to the study by Bori et al (O-114), the current version of KIDScore seems to have better ability to segregate embryos by quality.

While it is clear that morphology, and morphokinetics cannot replace PGT, it has been suggested that applying appropriate algorithms may be useful in prioritising for transfer in cases where PGT is not selected by the patient.

In a poster entitled “Embryo morphokinetics and static morphology in the prediction of live birth: evidences that speed is more important than beauty” (P-209), Bartolacci et al. showed that embryo morphokinetics was more accurate than conventional embryo morphology in ranking embryos in terms of live birth potential following a day 3 fresh embryo transfer.

Many clinics have asked if it is necessary to change culture media if performing assisted hatching in the EmbryoSlide Culture dish. Delgive et al. (P-539), showed that embryo development and euploidy rate was unaffected when performing assisted hatching in the EmbryoSlide dish using an Octax laser, without a culture media change.

Montjean. (O-062) presented the results of the publication of the ESHRE working group on time-lapse technology (Human Reproduction Open, Volume 2020, Issue 2, 2020). The presentation recapped the considerations for implementing the technology in the IVF clinic. It highlighted the need for each clinic to decide how the technology should be implemented as well as the practical and the non-clinical/biological interests and benefits of having time-lapse technology in the IVF laboratory, including training/teaching, quality control and the management of staff time and work-flow.

Genomics continues to be in the spotlight

One of the outstanding presentations (O-079) was given by Jenny Gruhn from Eva Hoffman’s lab at the University of Copenhagen, Denmark, on how chromosome segregation errors resulting in aneuploidy in human eggs shape natural fertility, which was published last year in Science (Gruhn et al., 2019)¹. Through collaboration with Claus Yding Andersen, they were able to obtain ovarian cortex from girls having fertility preservation treatment for medical reasons, and isolate small antral follicles, which were then matured in vitro. Together with immature and mature eggs from women undergoing IVF, they were able to study chromosome errors using both next generation sequencing (NGS) based copy number and single nucleotide polymorphism (SNP) analysis throughout the reproductive lifespan of women, from 9-43 years.

Interestingly, they found that the chromosome error rate was high in girls in their teenage years, so the error rate by female age is most likely a U-shaped curve. More importantly, the mechanism causing the errors in girls and older women was different. In older women the types of segregation errors are thought to be related to the functional degradation of cohesin proteins over decades, which play a vital role in holding sister chromosomes together. Whereas in girls and younger women, the errors were mainly failure of the bivalent chromosomes to segregate in the first meiotic division.

PGT-A by trophectoderm biopsy and NGS-based chromosome copy number analysis is now well established for identifying aneuploid embryos at the blastocyst stage. In some cases, however, the sensitivity and high resolution of NGS-based methods detects intermediate copy number changes, possibly indicating chromosome mosaicism among the biopsied cells, or segmental copy number changes only affecting a part of a chromosome. Over recent years, it has become increasingly clear that transferring embryos with only these types of copy number changes can result in healthy live births. To advise patients with these types of embryos appropriately, therefore, clinical outcome data is vitally important. Francesca Spinella, Genoma, Rome, Italy (O-081) and collaborators, presented an update on the clinical outcome following transfer of 822 blastocysts identified as having only mosaic or segmental copy number changes for single or multiple chromosomes. Compared to control embryos with normal copy number (euploid), implantation rates and ongoing pregnancy/live birth rates following transfer of these embryos were significantly lower. However, grouped according to the type of change and the numbers of chromosomes affected, embryos with mosaic segmental changes in one or two chromosomes had the best outcomes. Among those embryos with mosaic changes for whole chromosomes, there was an inverse relationship with the number of chromosomes affected, with single mosaic changes having significantly better outcomes than those with multiple mosaic changes.

The clinical science award for a poster was awarded to Antonio Capalbo, Igenomix, Marostica, Italy, and colleagues for their report of the interim results from the first multicentre prospective non-selection study on the transfer of mosaic embryos. In this non-selection study, the NGS-based PGT-A results indicating only low-level mosaicism (from 20-50% of the expected full copy number change) in women aged 35-44 years were reported blindly as euploid. Following frozen single blastocyst transfers, the clinical outcomes were unblinded to perform the interim analysis and there were no significant differences between uniformly euploid and either very low (20-30%) or low (30-50%) mosaic blastocyst transfers for implantation rates, biochemical pregnancies, miscarriage rates or ongoing pregnancy and live birth rates.

The overall proportion of IVF cycles with PGT-A in which the only embryos that can be considered for transfer have intermediate mosaic or segmental copy number changes is small but increases with advanced maternal age and in poor prognosis patients. For these patients, therefore, accurate information about clinical outcomes is very important to prevent discarding potentially viable embryos. The ‘take home messages’ from the meeting this year, from these and other presentations, are that intermediate, particularly high level, copy number changes for multiple whole chromosomes are more likely to affect clinical outcomes than segmental changes. As ever, however, more research is needed and there is no doubt that the topic will still feature in meetings for some time to come.

Oocyte and embryo vitrification

Cobo et al., (O-306) compared oocyte vitrification outcomes of patients with endometriosis to elective fertility preservation patients. It was found that endometriosis patients ≤35 years old had significantly lower oocyte survival, implantation, pregnancy and cumulative live birth rates. This was not observed with older patients (>35 years old) where, for example, the oocyte survival rates for endometriosis and fertility preservation patients were 80.8 and 82.1%, respectively. Nogueira et al., (P-449) presented data on the quality of embryos derived from vitrified oocytes from low ovarian reserve patients. Although fertilisation rates were similar between fresh and vitrified oocytes, subsequent day 3 and day 5 embryo development was significantly higher for the fresh group. Vitrification did not, however, appear to impact clinical outcomes or birth weight.

Murakami et al., (P-118) demonstrated that following warming, the ongoing pregnancy rates decreased in blastocysts that underwent spontaneous collapsing in the equilibration solution during vitrification. Reignier et al., (P-189) showed that fresh unexpanded day 5 blastocysts have higher birth rates than fresh or vitrified fully expanded day 6 transfers. Gunst et al., (P-135) using the Vitrolife vitrification system, presented results challenging current practices in blastocyst selection criteria. They showed that live birth rates from vitrified slow developing (compacting) embryos on day 5 as well as poor-quality blastocysts contribute substantially to overall success rates.

There were many posters that discussed the time to transfer after warming. Srivastava et al., (P-261) presented a retrospective evaluation of the live birth rate after SET of vitrified-warmed top-quality blastocysts that were either expanded or not at the time of transfer (<2, 2-3 or >3 hours after warming). At each time point the live birth rate was significantly higher following the transfer of an expanded blastocyst (37.4/21.3, 41.7/30.1 and 44.2/ 22.2%). The authors suggest that if there is not an expanded blastocyst within 4 hours such data could be used to discuss with the patient whether to warm another blastocyst. Shin et al., (P-216) reported on poor-quality blastocyst single embryo transfers and found that biochemical, implantation, clinical pregnancy rates were significantly higher if the transfer was performed within 3 to 5 hrs post warming compared to extended culture for 21 to 23 hrs. The ongoing pregnancy rate was also numerically higher but did not quite reach statistical significance. As there was no difference between groups in age and survival rate, the authors propose that the additional culture time could lead to a stress that has a negative effect on the viability of the blastocyst. The results from Nguyen et al., (P-265), however, suggests that for good and fair quality blastocysts that a prolonged (16 hrs) culture prior to transfer can give significantly increase biochemical, implantation and clinical pregnancy rates compared to 2hrs of culture post warming. Premannandan et al., (P-260) reported the live birth rates of 839 top-quality blastocysts that had been vitrified/warmed and cultured for <2, 2-3 and >3 hrs were similar. These are all retrospective studies, so the findings should be validated with prospective randomised controlled studies.

Tatsi et al., (P-139) showed data indicating that double vitrification (of donor oocyte and then embryo) decreases biochemical pregnancy, clinical pregnancy and live birth rates when compared to fresh or single vitrification (oocyte or embryo). It will be important to understand if this is repeated with other patient groups and clinics.

Westvik-Johari et al., (O-029) and Terho et al., (O-030) presented data on the health of children born after cryopreservation. As with other studies there appears to be an increase in birth weight compared to spontaneously conceived children. Compared to fresh transfers, however, there were fewer small for gestational age offspring.

Increased use of piezo-ICSI technique

The word “piezo” is derived from the Greek piezein, which means to squeeze, press, or push. The phenomenon itself is the ability of a material to generate movement when an electric field is applied. In ART this has been used to advance a glass pipette through the oolemma for nuclear transfer or ICSI. The utilisation of the piezo-ICSI technique has increased, especially in Japan. Kameda IVF is one of the clinics with most piezo-ICSI experience and a key opinion leader for the application (Dr Hiraoka) and colleagues presented two posters comparing both laboratory and postnatal outcomes following conventional ICSI and piezo-ICSI. With more than 4000 piezo-ICSI procedures performed it was concluded that piezo-ICSI can generate significantly higher survival, fertilisation and blastocyst rates without increasing the risk of malformation as compared with the conventional-ICSI (P-156, Kuga et al.). In the second retrospective study (P-154, Hiraoka et al.,) piezo-ICSI was compared to insemination. From the 3500 piezo-ICSI procedures performed the study showed significantly increased fertilisation and blastocyst rates without increasing the risk of birth defects as compared to IVF. Piezo-ICSI has the potential to be automated. Mizuta et al., (P-125) used image analysis to identify positions that result in lower risk of oocyte degeneration following piezo-ICSI.

More information

If you would like to read more about some of the research presented at ESHRE, you can download 'Vitrolife abstracts book ESHRE 2020' below.

Watch our symposium 'Development of robust ai based embryo evaluation'

If you missed our symposium or would like to watch it again, then you can find it here.

1) Gruhn JR, Zielinska AP, Shukla V, Blanshard R, Capalbo A, Cimadomo D, Nikiforov D, Chan ACH, Newnham LJ, Vogel I et al. (2019) Chromosome errors in human eggs shape natural fertility over reproductive life span. Science 365 1466–1469.