Embryology in the era of proteomics

Leveraging federated learning for boosting data privacy and performance in IVF embryo selection

PURPOSE

To study the effectiveness of federated learning in in vitro fertilization on embryo evaluation tasks.

METHODS

This is a retrospective cohort analysis. Two datasets were used in this study. The ploidy status dataset consisted of 10,065 embryo records, 3760 treatments, and 2479 infertile couples from 5 hospitals. The clinical pregnancy dataset consisted of 4495 embryo records, 4495 treatments, and 3704 infertile couples from 4 hospitals. Federated learning and the gradient boosting decision tree algorithm were utilized for modeling.

RESULTS

On the ploidy status dataset, the areas under the receiver operating characteristic curves of our model trained with federated learning were 71.78%, 73.10%, 69.39%, 69.72%, and 73.46% for 5 hospitals respectively, showing an average increase of 2.5% compared to those of our model trained without federated learning. On the clinical pregnancy dataset, the areas under the receiver operating characteristic curves of our model trained with federated learning were 72.03%, 56.77%, 61.63%, and 58.58% for 4 hospitals respectively, showing an average increase of 3.08%.

CONCLUSIONS

Federated learning can improve data privacy and data security and meanwhile improve the performance of embryo selection tasks by leveraging data from multiple sources. This study demonstrates the effectiveness of federated learning in embryo evaluation, and the results show the promise for future application.

An artificial intelligence model (euploid prediction algorithm) can predict embryo ploidy status based on time-lapse data

BACKGROUND

For the association between time-lapse technology (TLT) and embryo ploidy status, there has not yet been fully understood. TLT has the characteristics of large amount of data and non-invasiveness. If we want to accurately predict embryo ploidy status from TLT, artificial intelligence (AI) technology is a good choice. However, the current work of AI in this field needs to be strengthened.

METHODS

A total of 469 preimplantation genetic testing (PGT) cycles and 1803 blastocysts from April 2018 to November 2019 were included in the study. All embryo images are captured during 5 or 6 days after fertilization before biopsy by time-lapse microscope system. All euploid embryos or aneuploid embryos are used as data sets. The data set is divided into training set, validation set and test set. The training set is mainly used for model training, the validation set is mainly used to adjust the hyperparameters of the model and the preliminary evaluation of the model, and the test set is used to evaluate the generalization ability of the model. For better verification, we used data other than the training data for external verification. A total of 155 PGT cycles from December 2019 to December 2020 and 523 blastocysts were included in the verification process.

RESULTS

The euploid prediction algorithm (EPA) was able to predict euploid on the testing dataset with an area under curve (AUC) of 0.80.

CONCLUSIONS

The TLT incubator has gradually become the choice of reproductive centers. Our AI model named EPA that can predict embryo ploidy well based on TLT data. We hope that this system can serve all in vitro fertilization and embryo transfer (IVF-ET) patients in the future, allowing embryologists to have more non-invasive aids when selecting the best embryo to transfer.

An artificial intelligence model based on the proteomic profile of euploid embryos and blastocyst morphology: a preliminary study

RESEARCH QUESTION

The study aimed to develop an artificial intelligence model based on artificial neural networks (ANNs) to predict the likelihood of achieving a live birth using the proteomic profile of spent culture media and blastocyst morphology.

DESIGN

This retrospective cohort study included 212 patients who underwent single blastocyst transfer at IVI Valencia. A single image of each of 186 embryos was studied, and the protein profile was analysed in 81 samples of spent embryo culture medium from patients included in the preimplantation genetic testing programme. The information extracted from the analyses was used as input data for the ANN. The multilayer perceptron and the back-propagation learning method were used to train the ANN. Finally, predictive power was measured using the area under the curve (AUC) of the receiver operating characteristic curve.

RESULTS

Three ANN architectures classified most of the embryos correctly as leading (LB+) or not leading (LB-) to a live birth: 100.0% for ANN1 (morphological variables and two proteins), 85.7% for ANN2 (morphological variables and seven proteins), and 83.3% for ANN3 (morphological variables and 25 proteins). The artificial intelligence model using information extracted from blastocyst image analysis and concentrations of interleukin-6 and matrix metalloproteinase-1 was able to predict live birth with an AUC of 1.0.

CONCLUSIONS

The model proposed in this preliminary report may provide a promising tool to select the embryo most likely to lead to a live birth in a euploid cohort. The accuracy of prediction demonstrated by this software may improve the efficacy of an assisted reproduction treatment by reducing the number of transfers per patient. Prospective studies are, however, needed.

Expressed proteins and activated pathways in conditioned embryo culture media from IVF patients are diverse according to infertility factors

OBJECTIVE

Given that the embryo culture medium secretome reflects the embryo development, we hypothesize that protein profiles are affected according to infertility factors, which can be responsible for detrimental embryonic developmental competence. The aim of this study was to screen the protein profile of conditioned embryo culture media in patients presenting deep infiltrating endometriosis (ENDO) and polycystic ovarian syndrome (PCOS) undergoing IVF, by proteomics approaches. The control group was constituted by tubal factor patients.

METHODS

Patients underwent in vitro fertilization (IVF) treatment as routine and oocytes were fertilized by ICSI. The embryos were group cultured until day 3 of development, and after transfer the culture media were collected. For the proteomics analysis, two pools of samples were prepared for groups CONTROL and PCOS, and 4 pools of samples for group DIE. Samples were prepared to deplete high abundant proteins and followed evaluated by high throughput proteomics approach.

RESULTS

The embryonic organ and tissue development were physiological functions activated, based on proteins identified in the 3 study groups of samples. The samples coming from DIE patients presented a high calcium activity and on the other hand, embryos coming from PCOS patients showed a decreased calcium action. Other pathways as grow factors through the EGF signaling pathway overexpressed in ENDO culture medium and protein kinase A in PCOS were also observed.

CONCLUSIONS

Proteomic embryonic secretome will advance our knowledge of early embryogenesis and additionally could lead to improved selection of embryos for transfer warrants further investigation.

Analysis and characterisation of bovine oocyte and embryo biomarkers by matrix-assisted desorption ionisation mass spectrometry imaging

In the field of 'single cell analysis', many classical strategies like immunofluorescence and electron microscopy are the primary techniques of choice. However, these methodologies are time consuming and do not permit direct identification of specific molecular classes, such as lipids. In the present study, a novel mass spectrometry-based analytical approach was applied to bovine oocytes and embryos. This new metabolomics-based application uses mass spectrometry imaging (MSI), efficient data processing and multivariate data analysis. Metabolic fingerprinting (MF) was applied to the analysis of unfertilised oocytes, 2-, 4- and 8-cell embryos and blastocysts. A semiquantitative strategy for sphingomyelin [SM (16:0)+Na](+) (m/z 725) and phosphatidylcholine [PC (32:0)+Na](+) (m/z 756) was developed, showing that lipid concentration was useful for selecting the best metabolic biomarkers. This study demonstrates that a combination of MF, MSI features and chemometric analysis can be applied to discriminate cell stages, characterising specific biomarkers and relating them to developmental pathways. This information furthers our understanding of fertilisation and preimplantation events during bovine embryo development.

A Review of The Society for Assisted Reproductive Technology Embryo Grading System and Proposed Modification

The Society for Assisted Reproductive Technology (SART) method of embryo grad- ing is unique, simple, and widely practiced, and its use has been mandatory for SART membership programs since 2010. Developed by SART in 2006, the current embryo grading system categories, "good, fair, and poor," are limited because they do not describe the best 1-2 embryos in the interest of keeping pace with the shift in clinical practice to be more selective and to transfer fewer embryos. This inspired us to conduct a review on the SART embryo grading system. In this retrospective study, the literature on evaluation of human embryo quality in gen- eral, and the SART method of evaluation in particular, were reviewed for the period of 2000 to 2014. A multifaceted search pertaining to methods of embryo grading and trans- fer using a combination of relevant terms [embryo, mammalian, embryo transfer, grade, grading, morphology, biomarkers, SART, and in vitro fertilization (IVF)] was performed. The inclusion and exclusion in this review were dictated by the aim and scope of the study. Two investigators independently assessed the studies and extracted information. A total of 61 articles were reviewed. Very few studies have evaluated the efficacy of the SART embryo grading method. The present study suggests the necessity for revision of the current SART grading system. The system, as it is now, lacks criteria for describing the cohort specific best embryo and thus is of limited use in single embryo transfer. The study foresees heightened descriptive efficiency of the SART system by implementing the proposed changes. Strengths and weaknesses of the SART embryo grading were identified. Ideas for selecting the best cohort-specific embryo have been discussed, which may trigger methodological improvement in SART and other embryo grading systems.

Multiple-embryo transfer for studying very early maternal-embryo interactions in cattle

In the present paper, we highlight the need to study very early maternal-embryo interactions and discuss how these interactions can be addressed. Bovine species normally carry one or, less frequently, two embryos to term; there are very rare cases of triplets or higher-order multiple pregnancies in which all the offspring are born alive. Multiple-embryo transfer (MET) in cattle allows for the detection of endometrial responses in scenarios where single-embryo transfer would not. Although MET is non-physiological, the present study shows that at the very early embryonic stages, a uterus carrying zona-enclosed embryos does not exhibit non-physiological reactions. On the contrary, MET should be considered the sum of multiple individual effects triggered by developing embryos. We provide arguments to support our hypothesis that describe a rationale for current work with MET, and we discuss alternative hypotheses. Using cattle as a model, we describe how technical approaches to analyzing zona-enclosed early embryo-maternal interactions (i.e., transcriptomics, proteomics, and endometrial cell culture) can help identify molecular changes that may be difficult to observe when only a single embryo is present. We conclude that MET can be used for studying very early maternal-embryo interactions in vivo in monotocous species. Free Spanish abstract: A Spanish translation of this abstract is freely available at http://www.reproduction-online.org/content/150/2/R35/suppl/DC1.

Metabolic control of oocyte development: linking maternal nutrition and reproductive outcomes

Obesity, diabetes, and related metabolic disorders are major health issues worldwide. As the epidemic of metabolic disorders continues, the associated medical co-morbidities, including the detrimental impact on reproduction, increase as well. Emerging evidence suggests that the effects of maternal nutrition on reproductive outcomes are likely to be mediated, at least in part, by oocyte metabolism. Well-balanced and timed energy metabolism is critical for optimal development of oocytes. To date, much of our understanding of oocyte metabolism comes from the effects of extrinsic nutrients on oocyte maturation. In contrast, intrinsic regulation of oocyte development by metabolic enzymes, intracellular mediators, and transport systems is less characterized. Specifically, decreased acid transport proteins levels, increased glucose/lipid content and elevated reactive oxygen species in oocytes have been implicated in meiotic defects, organelle dysfunction and epigenetic alteration. Therefore, metabolic disturbances in oocytes may contribute to the diminished reproductive potential experienced by women with metabolic disorders. In-depth research is needed to further explore the underlying mechanisms. This review also discusses several approaches for metabolic analysis. Metabolomic profiling of oocytes, the surrounding granulosa cells, and follicular fluid will uncover the metabolic networks regulating oocyte development, potentially leading to the identification of oocyte quality markers and prevention of reproductive disease and poor outcomes in offspring.

OMICS: Current and future perspectives in reproductive medicine and technology

Many couples present fertility problems at their reproductive age, and although in the last years, the efficiency of assisted reproduction techniques has increased, these are still far from being 100% effective. A key issue in this field is the proper assessment of germ cells, embryos and endometrium quality, in order to determine the actual likelihood to succeed. Currently available analysis is mainly based on morphological features of oocytes, sperm and embryos and although these strategies have improved the results, there is an urgent need of new diagnostic and therapeutic tools. The emergence of the - OMICS technologies (epigenomics, genomics, transcriptomics, proteomics and metabolomics) permitted the improvement on the knowledge in this field, by providing with a huge amount of information regarding the biological processes involved in reproductive success, thereby getting a broader view of complex biological systems with a relatively low cost and effort.

Maternal enzyme masks the phenotype of mouse embryos lacking dihydrolipoamide dehydrogenase

During early embryogenesis, the phenotype reflecting the embryonic genotype emerges only as maternal proteins are replaced by embryonically encoded forms, a process known as the maternal-to-embryonic transition (MET). Little is understood about MET for most proteins. This study investigates how complete deficiency of the murine dihydrolipoamide dehydrogenase gene (Dld), a gene that encodes an enzyme of mitochondrial energy metabolism, affects the phenotype of the early embryo and how the MET of the DLD protein affects the phenotype. Dld-deficient (-/-) embryos were found to develop similarly to wild-type (+/+) or heterozygous (+/-) embryos throughout the preimplantation period. These three genotypic classes also have comparable rates of glucose uptake (4.9-5.0 pmoles/embryo/h) and lactate production (0.97-1.0 pmoles/embryo/h). Dld-deficient embryos at the end of the preimplantation stage have 44% of DLD enzyme present in oocytes, a proportion similar to that found in +/+ or +/- embryos. This study demonstrates that Dld-deficient preimplantation embryos are phenotypically normal, as the MET for the DLD enzyme is only partially complete by the end of the preimplantation period. These findings have implications for phenotype- or enzyme-based approaches to identify mutations in Dld and other genes that encode proteins with similar MET kinetic profiles.

Mass spectrometry fingerprinting of media used for in vitro production of bovine embryos

Using the bovine species as a biological model, direct infusion chip-based nano-electrospray ionization mass spectrometry (nano-ESI-MS) fingerprinting in the positive ion mode is used to obtain fast chemical profiles of media used for in vitro production of bovine embryos. Nano-ESI-MS fingerprinting is useful for characterization and routine quality control requiring no sample pre-separation, being able to differentiate four different media (IVM, IVF, SOF and HSOF) via principal component analysis (PCA). For media stored at +4 degrees C for up to 45 days, no significant (p>0.05) variation was observed in cleavage and blastocyst rate development, as well as in the nano-ESI-MS chemical profiles. For media exposed to a heat shock (60 degrees C for 3 h), no significant decrease (p>0.05) in embryo development rates was observed, but nano-ESI-MS profiles were quite distant from fresh control media in the PCA. For frozen media (-70 degrees C for 2 months), again no significant variation (p>0.05) in embryo development was noticed, but nano-ESI-MS profiles from all media were significantly affected. These results indicate that nano-ESI(+)-MS fingerprinting was able to characterize different media based on their specific chemical profile. The technique seems therefore applicable as a routine quality control assay, detecting, for example, compositional changes after temperature variations that may affect post-transfer embryo viability.

Symposium: innovative techniques in human embryo viability assessment. Can proteomics help to shape the future of human assisted conception?

Proteomics describes the changes in all proteins expressed and translated from a single genome. At present little is known regarding either the genome or proteome of human gametes or the preimplantation embryo. The unravelling of this information is fundamental to understanding the complexity of reproductive physiology, including the dialogue between the developing embryo and its maternal environment. To date, a lack of sensitivity has been the main reason behind the inability to introduce proteomics technology into assisted reproduction techniques. Proteomics alone involves several sophisticated techniques including imaging, mass spectrometry and bioinformatics to identify, quantify and characterize a proteome. The recent increased sensitivity of these techniques has allowed for the development of new protocols that are capable of not only profiling the proteome of individual human oocytes and embryos, but also the proteins produced by the embryo into the surrounding medium (the secretome). Hence, the identification of proteins that are involved in oocyte maturation, embryo development and implantation could lead to further improvements in assisted reproduction techniques as well as the development of new diagnostic tests. Furthermore, proteomics may contribute in the design of a non-invasive viability assay to assist in the selection of embryos for transfer in human assisted reproduction.

Assessment of embryo viability in assisted reproductive technology: shortcomings of current approaches and the emerging role of metabolomics

PURPOSE OF REVIEW

The present article describes the current status of embryo assessment in assisted reproductive technologies and discusses two important issues that derive, at least in part, from our inability to adequately assess the reproductive potential of individual embryos: low implantation rates and high multiple pregnancy rates. After an overview of studies evaluating embryo metabolism as a predictor of embryo viability, use of metabolomics and additional emerging approaches for rapid, noninvasive, embryo assessment are discussed.

RECENT FINDINGS

Morphology and cleavage rate remain the mainstay of embryo assessment. A number of technologies are, however, under investigation. These include the assessment of glucose, lactate, pyruvate, and amino acid metabolism, proteomic profiling, evaluation of oxygen consumption, and most recently, examination of the metabolome.

SUMMARY

As the number of assisted reproductive technology cycles increases worldwide, improvements in the ability to quickly and noninvasively determine the best embryos for transfer remain a critical goal for reproductive medicine. Recent studies suggest that metabolomic profiling of embryo culture media may provide a useful adjunct to the current embryo assessment strategies based on morphology and cleavage rate.

Challenges of functional genomics applied to farm animal gametes and pre-hatching embryos

The genomes of many commercially important farm animals have already been or are in the process of being decrypted. The genomic era is generating an important wave of downstream developments and derived disciplines are also progressing at a very fast pace. The post-genomic era is already ongoing as exemplified by the introduction of new concepts such as phenomics and functional genomics. These new fields are complementary but do not necessarily target similar applications even though they are often used to refer to one another. In an attempt to categorize the fields according to their respective potential applications, a brief comparative description of phenomics and functional genomics has been put together. However, the focus of this paper is mainly directed toward the introduction of functional genomics specifically applied to the study of the molecular mechanisms underlying gamete and early mammalian developments. Many aspects of the peculiar nature of these cells are introducing numerous methodological challenges to the applicability of functional genomics to unravel their molecular physiology. This is particularly true for transcriptomic studies and it is currently of high relevance for the field of reproductive biology to take into consideration these technical hurdles before tackling the implementation of this technology on a large scale. Nonetheless, functional genomics should prove to be up to the expectations in providing sound information to better understand the fascinating window spanning gamete development that leads to the first weeks of life.