Embryologist agreement when assessing blastocyst implantation probability: is data-driven prediction the solution to embryo assessment subjectivity?

STUDY QUESTION: What is the accuracy and agreement of embryologists when assessing the implantation probability of blastocysts using time-lapse imaging (TLI), and can it be improved with a data-driven algorithm? SUMMARY ANSWER: The overall interobserver agreement of a large panel of embryologists was moderate and prediction accuracy was modest, while the purpose-built artificial intelligence model generally resulted in higher performance metrics. WHAT IS KNOWN ALREADY: Previous studies have demonstrated significant interobserver variability amongst embryologists when assessing embryo quality. However, data concerning embryologists' ability to predict implantation probability using TLI is still lacking. Emerging technologies based on data-driven tools have shown great promise for improving embryo selection and predicting clinical outcomes. STUDY DESIGN, SIZE, DURATION: TLI video files of 136 embryos with known implantation data were retrospectively collected from two clinical sites between 2018 and 2019 for the performance assessment of 36 embryologists and comparison with a deep neural network (DNN). PARTICIPANTS/MATERIALS, SETTING, METHODS: We recruited 39 embryologists from 13 different countries. All participants were blinded to clinical outcomes. A total of 136 TLI videos of embryos that reached the blastocyst stage were used for this experiment. Each embryo's likelihood of successfully implanting was assessed by 36 embryologists, providing implantation probability grades (IPGs) from 1 to 5, where 1 indicates a very low likelihood of implantation and 5 indicates a very high likelihood. Subsequently, three embryologists with over 5 years of experience provided Gardner scores. All 136 blastocysts were categorized into three quality groups based on their Gardner scores. Embryologist predictions were then converted into predictions of implantation (IPG ≥ 3) and no implantation (IPG ≤ 2). Embryologists' performance and agreement were assessed using Fleiss kappa coefficient. A 10-fold cross-validation DNN was developed to provide IPGs for TLI video files. The model's performance was compared to that of the embryologists. MAIN RESULTS AND THE ROLE OF CHANCE: Logistic regression was employed for the following confounding variables: country of residence, academic level, embryo scoring system, log years of experience and experience using TLI. None were found to have a statistically significant impact on embryologist performance at α = 0.05. The average implantation prediction accuracy for the embryologists was 51.9% for all embryos (N = 136). The average accuracy of the embryologists when assessing top quality and poor quality embryos (according to the Gardner score categorizations) was 57.5% and 57.4%, respectively, and 44.6% for fair quality embryos. Overall interobserver agreement was moderate (κ = 0.56, N = 136). The best agreement was achieved in the poor + top quality group (κ = 0.65, N = 77), while the agreement in the fair quality group was lower (κ = 0.25, N = 59). The DNN showed an overall accuracy rate of 62.5%, with accuracies of 62.2%, 61% and 65.6% for the poor, fair and top quality groups, respectively. The AUC for the DNN was higher than that of the embryologists overall (0.70 DNN vs 0.61 embryologists) as well as in all of the Gardner groups (DNN vs embryologists-Poor: 0.69 vs 0.62; Fair: 0.67 vs 0.53; Top: 0.77 vs 0.54). LIMITATIONS, REASONS FOR CAUTION: Blastocyst assessment was performed using video files acquired from time-lapse incubators, where each video contained data from a single focal plane. Clinical data regarding the underlying cause of infertility and endometrial thickness before the transfer was not available, yet may explain implantation failure and lower accuracy of IPGs. Implantation was defined as the presence of a gestational sac, whereas the detection of fetal heartbeat is a more robust marker of embryo viability. The raw data were anonymized to the extent that it was not possible to quantify the number of unique patients and cycles included in the study, potentially masking the effect of bias from a limited patient pool. Furthermore, the lack of demographic data makes it difficult to draw conclusions on how representative the dataset was of the wider population. Finally, embryologists were required to assess the implantation potential, not embryo quality. Although this is not the traditional approach to embryo evaluation, morphology/morphokinetics as a means of assessing embryo quality is believed to be strongly correlated with viability and, for some methods, implantation potential. WIDER IMPLICATIONS OF THE FINDINGS: Embryo selection is a key element in IVF success and continues to be a challenge. Improving the predictive ability could assist in optimizing implantation success rates and other clinical outcomes and could minimize the financial and emotional burden on the patient. This study demonstrates moderate agreement rates between embryologists, likely due to the subjective nature of embryo assessment. In particular, we found that average embryologist accuracy and agreement were significantly lower for fair quality embryos when compared with that for top and poor quality embryos. Using data-driven algorithms as an assistive tool may help IVF professionals increase success rates and promote much needed standardization in the IVF clinic. Our results indicate a need for further research regarding technological advancement in this field. STUDY FUNDING/COMPETING INTEREST(S): Embryonics Ltd is an Israel-based company. Funding for the study was partially provided by the Israeli Innovation Authority, grant #74556. TRIAL REGISTRATION NUMBER: N/A.

Adaptive sequencing using nanopores and deep learning of mitochondrial DNA.

Nanopore sequencing is an emerging technology that reads DNA by utilizing a unique method of detecting nucleic acid sequences and identifies the various chemical modifications they carry. Deep learning has increased in popularity as a useful technique to solve many complex computational tasks. 'Adaptive sequencing' is an implementation of selective sequencing, intended for use on the nanopore sequencing platform. In this study, we demonstrated an alternative method of software-based selective sequencing that is performed in real time by combining nanopore sequencing and deep learning. Our results showed the feasibility of using deep learning for classifying signals from only the first 200 nucleotides in a raw nanopore sequencing signal format. This was further demonstrated by comparing the accuracy of our deep learning classification model across data from several human cell lines and other eukaryotic organisms. We used custom deep learning models and a script that utilizes a 'Read Until' framework to target mitochondrial molecules in real time from a human cell line sample. This achieved a significant separation and enrichment ability of 2.3-fold. In a series of very short sequencing experiments (10, 30 and 120 min), we identified genomic and mitochondrial reads with accuracy above 90%, although mitochondrial DNA comprised only 0.1% of the total input material. The uniqueness of our method is the ability to distinguish two groups of DNA even without a labeled reference. This contrasts with studies that required a well-defined reference, whether of a DNA sequence or of another type of representation. Additionally, our method showed higher correlation to the theoretically possible enrichment factor, compared with other published methods. We believe that our results will lay the foundation for rapid and selective sequencing using nanopore technology and will pave the approach for clinical applications that use nanopore sequencing data.

Assessing the involvement of the placental microbiome and virome in preeclampsia using non coding RNA sequencing.

OBJECTIVES: Preeclampsia is a dangerous pregnancy complication. The source of preeclampsia is unknown, though the placenta is believed to have a central role in its pathogenesis. An association between maternal infection and preeclampsia has been demonstrated, yet the involvement of the placental microbiome in the etiology of preeclampsia has not been determined. In this study, we examined whether preeclampsia is associated with an imbalanced microorganism composition in the placenta. METHODS: To this end, we developed a novel method for the identification of bacteria/viruses based on sequencing of small non-coding RNA, which increases the microorganism-to-host ratio, this being a major challenge in microbiome methods. We validated the method on various infected tissues and demonstrated its efficiency in detecting microorganisms in samples with extremely low bacterial/viral biomass. We then applied the method to placenta specimens from preeclamptic and healthy pregnancies. Since the placenta is a remarkably large and heterogeneous organ, we explored the bacterial and viral RNA at each of 15 distinct locations. RESULTS: Bacterial RNA was detected at all locations and was consistent with previous studies of the placental microbiome, though without significant differences between the preeclampsia and control groups. Nevertheless, the bacterial RNA composition differed significantly between various areas of the placenta. Viral RNA was detected in extremely low quantities, below the threshold of significance, thus viral abundance could not be determined. CONCLUSIONS: Our results suggest that the bacterial and viral abundance in the placenta may have only limited involvement in the pathogenesis of preeclampsia. The evidence of a heterogenic bacterial RNA composition in the various placental locations warrants further investigation to capture the true nature of the placental microbiome.

Early diagnosis of gestational diabetes mellitus using circulating microRNAs.

DESIGN: Gestational diabetes mellitus (GDM) is one of the most common pregnancy complications and its prevalence is constantly rising worldwide. Diagnosis is commonly in the late second or early third trimester of pregnancy, though the development of GDM starts early; hence, first-trimester diagnosis is feasible. OBJECTIVE: Our objective was to identify microRNAs that best distinguish GDM samples from those of healthy pregnant women and to evaluate the predictive value of microRNAs for GDM detection in the first trimester. METHODS: We investigated the abundance of circulating microRNAs in the plasma of pregnant women in their first trimester. Two populations were included in the study to enable population-specific as well as cross-population inspection of expression profiles. Each microRNA was tested for differential expression in GDM vs control samples, and their efficiency for GDM detection was evaluated using machine-learning models. RESULTS: Two upregulated microRNAs (miR-223 and miR-23a) were identified in GDM vs the control set, and validated on a new cohort of women. Using both microRNAs in a logistic-regression model, we achieved an AUC value of 0.91. We further demonstrated the overall predictive value of microRNAs using several types of multivariable machine-learning models that included the entire set of expressed microRNAs. All models achieved high accuracy when applied on the dataset (mean AUC = 0.77). The significance of the classification results was established via permutation tests. CONCLUSIONS: Our findings suggest that circulating microRNAs are potential biomarkers for GDM in the first trimester. This warrants further examination and lays the foundation for producing a novel early non-invasive diagnostic tool for GDM.

Dual inhibition of ABCE1 and LCP1 by microRNA-96 results in an additive effect in breast cancer mouse model.

MicroRNAs (miRNAs) are short non-coding RNAs that regulate the expression of target genes at the post-transcriptional level. Each miRNA can modulate multiple genes and, as a result, a single miRNA may have a profound effect on a specific biological pathway consisting of several of its target genes. Recent studies have indicated that specific miRNA signatures are correlated with tumor aggressiveness and clinical outcome in breast cancer. We previously demonstrated that miR-96 has a suppressive effect on breast cancer aggressiveness and that this effect was mediated by ABCE1 gene regulation. In this study we investigated whether other miR-96 regulated genes can enhance ABCE1's anti-cancer effects. We identified one such gene - LCP1 - and proved its negative effect on breast cancer progression. Interestingly, dual inhibition of ABCE1 and LCP1 resulted in an additive effect on cancer cell migration, invasion, and proliferation. Furthermore, in vivo analysis of dual ABCE1 and LCP1 knockdown resulted in significant tumor growth inhibition, decreased metastatic activity, and contributed to survival compared to either gene, separately. This indicates that the combined downregulation of two miR-96 gene targets has an additive effect on reducing cancer aggressiveness. Overall, our work supports seeking more than one target in miRNA-based studies in order to enhance functional effects and better characterize the miRNA wide-spread activity.

Bayesian-based noninvasive prenatal diagnosis of single-gene disorders.

In the last decade, noninvasive prenatal diagnosis (NIPD) has emerged as an effective procedure for early detection of inherited diseases during pregnancy. This technique is based on using cell-free DNA (cfDNA) and fetal cfDNA (cffDNA) in maternal blood, and hence, has minimal risk for the mother and fetus compared with invasive techniques. NIPD is currently used for identifying chromosomal abnormalities (in some instances) and for single-gene disorders (SGDs) of paternal origin. However, for SGDs of maternal origin, sensitivity poses a challenge that limits the testing to one genetic disorder at a time. Here, we present a Bayesian method for the NIPD of monogenic diseases that is independent of the mode of inheritance and parental origin. Furthermore, we show that accounting for differences in the length distribution of fetal- and maternal-derived cfDNA fragments results in increased accuracy. Our model is the first to predict inherited insertions-deletions (indels). The method described can serve as a general framework for the NIPD of SGDs; this will facilitate easy integration of further improvements. One such improvement that is presented in the current study is a machine learning model that corrects errors based on patterns found in previously processed data. Overall, we show that next-generation sequencing (NGS) can be used for the NIPD of a wide range of monogenic diseases, simultaneously. We believe that our study will lead to the achievement of a comprehensive NIPD for monogenic diseases.

Comparison of breast cancer metastasis models reveals a possible mechanism of tumor aggressiveness.

In breast cancer patients, the lungs are among the first sites of cancer metastasis, and in nearly one quarter of metastatic patients, the exclusive first event. Two common mouse models mimic breast cancer lung colonization and distal metastasis: an orthotopic model and intravenous (IV) cell injections. Gene expression analysis of pulmonary lesions from these two methods demonstrated high inter-model resemblance. However, microRNA (miRNA) expression profiles were not compared. In this study, we compared the overall miRNA expression profiles (miRNome) of the orthotopic and IV breast cancer metastasis models and identified significant miRNome changes between the two models. Overexpression of the most significant candidate, miR-96 or downregulation of its validated gene-target, ABCE1 reduced cancer cells 2D/3D cell movement and proliferation in vitro, and abated tumor growth and metastasis formation in vivo. Human data analysis further strengthened miR-96/ABCE1 role in breast cancer tumor aggression. Taken together, our results indicate that IV- and orthotopic models differ by their miRNome. Specifically in our study, breast cancer aggressiveness was dictated by miR-96 regulating ABCE1. Overall, miRNome analysis of various metastatic cancer models may lead to the identification of candidate genes critical to metastasis development.

The gene regulatory program of Acrobeloides nanus reveals conservation of phylum-specific expression.

The evolution of development has been studied through the lens of gene regulation by examining either closely related species or extremely distant animals of different phyla. In nematodes, detailed cell- and stage-specific expression analyses are focused on the model Caenorhabditis elegans, in part leading to the view that the developmental expression of gene cascades in this species is archetypic for the phylum. Here, we compared two species of an intermediate evolutionary distance: the nematodes C. elegans (clade V) and Acrobeloides nanus (clade IV). To examine A. nanus molecularly, we sequenced its genome and identified the expression profiles of all genes throughout embryogenesis. In comparison with C. elegans, A. nanus exhibits a much slower embryonic development and has a capacity for regulative compensation of missing early cells. We detected conserved stages between these species at the transcriptome level, as well as a prominent middevelopmental transition, at which point the two species converge in terms of their gene expression. Interestingly, we found that genes originating at the dawn of the Ecdysozoa supergroup show the least expression divergence between these two species. This led us to detect a correlation between the time of expression of a gene and its phylogenetic age: evolutionarily ancient and young genes are enriched for expression in early and late embryogenesis, respectively, whereas Ecdysozoa-specific genes are enriched for expression during the middevelopmental transition. Our results characterize the developmental constraints operating on each individual embryo in terms of developmental stages and genetic evolutionary history.

Early Detection of Preeclampsia Using Circulating Small non-coding RNA.

Preeclampsia is one of the most dangerous pregnancy complications, and the leading cause of maternal and perinatal mortality and morbidity. Although the clinical symptoms appear late, its origin is early, and hence detection is feasible already at the first trimester. In the current study, we investigated the abundance of circulating small non-coding RNAs in the plasma of pregnant women in their first trimester, seeking transcripts that best separate the preeclampsia samples from those of healthy pregnant women. To this end, we performed small non-coding RNAs sequencing of 75 preeclampsia and control samples, and identified 25 transcripts that were differentially expressed between preeclampsia and the control groups. Furthermore, we utilized those transcripts and created a pipeline for a supervised classification of preeclampsia. Our pipeline generates a logistic regression model using a 5-fold cross validation on numerous random partitions into training and blind test sets. Using this classification procedure, we achieved an average AUC value of 0.86. These findings suggest the predictive value of circulating small non-coding RNA in the first trimester, warranting further examination, and lay the foundation for producing a novel early non-invasive diagnostic tool for preeclampsia, which could reduce the life-threatening risk for both the mother and fetus.

Expansion of human and murine hematopoietic stem and progenitor cells ex vivo without genetic modification using MYC and Bcl-2 fusion proteins.

The long-term repopulating hematopoietic stem cell (HSC) population can self-renew in vivo, support hematopoiesis for the lifetime of the individual, and is of critical importance in the context of bone marrow stem cell transplantation. The mechanisms that regulate the expansion of HSCs in vivo and in vitro remain unclear to date. Since the current set of surface markers only allow for the identification of a population of cells that is highly enriched for HSC activity, we will refer to the population of cells we expand as Hematopoietic Stem and Progenitor cells (HSPCs). We describe here a novel approach to expand a cytokine-dependent Hematopoietic Stem and Progenitor Cell (HSPC) population ex vivo by culturing primary adult human or murine HSPCs with fusion proteins including the protein transduction domain of the HIV-1 transactivation protein (Tat) and either MYC or Bcl-2. HSPCs obtained from either mouse bone marrow, human cord blood, human G-CSF mobilized peripheral blood, or human bone marrow were expanded an average of 87 fold, 16.6 fold, 13.6 fold, or 10 fold, respectively. The expanded cell populations were able to give rise to different types of colonies in methylcellulose assays in vitro, as well as mature hematopoietic populations in vivo upon transplantation into irradiated mice. Importantly, for both the human and murine case, the ex vivo expanded cells also gave rise to a self-renewing cell population in vivo, following initial transplantation, that was able to support hematopoiesis upon serial transplantation. Our results show that a self-renewing cell population, capable of reconstituting the hematopoietic compartment, expanded ex vivo in the presence of Tat-MYC and Tat-Bcl-2 suggesting that this may be an attractive approach to expand human HSPCs ex vivo for clinical use.

TC-PTP is required for the maintenance of MYC-driven B-cell lymphomas.

We sought to determine the contributions of protein tyrosine phosphatases (PTPs) to the pathogenesis of B-cell lymphomas. We found that T-cell PTP (TC-PTP) was overexpressed in transformed B cells. We hypothesized that TC-PTP may be a tumor-promoting gene that is regulated by MYC overexpression in B cells. Knockdown of TC-PTP in murine tumors resulted in decreased cell viability in vitro because of an arrest in the G(1) phase of the cell cycle. Furthermore, cells with reduced TC-PTP expression were unable to either engraft or expand in vivo. Taken together, these data indicate that TC-PTP is required for B-cell tumor maintenance. Our data also suggested a correlation between TC-PTP expression and MYC overexpression. To investigate this further, we used malignant murine B cells that contain a doxycycline-repressible MYC transgene. We found that repression of MYC overexpression with doxycycline reduced TC-PTP expression. Moreover, enforced expression of TC-PTP showed partial rescue of the expansion of tumor cells after suppression of MYC overexpression. These results suggest that MYC overexpression induces TC-PTP overexpression, which in turn promotes tumor proliferation, implicating TC-PTP as an important effector of the MYC-driven proliferation program in B-cell lymphomas. Thus, TC-PTP may be a suitable molecular target for the treatment of B-cell lymphomas.