Embryo classification beyond pregnancy: early prediction of first trimester miscarriage using machine learning.

PURPOSE: First trimester miscarriage is a major concern in IVF-ET treatments, accounting for one out of nine clinical pregnancies and for up to one out of three recognized pregnancies. To develop a machine learning classifier for predicting the risk of cleavage-stage embryos to undergo first trimester miscarriage based on time-lapse images of preimplantation development. METHODS: Retrospective study of a 4-year multi-center cohort of 391 women undergoing intra-cytoplasmatic sperm injection (ICSI) and fresh single or double embryo transfers. The study included embryos with positive indication of clinical implantation based on gestational sac visualization either with first trimester miscarriage or live-birth outcome. Miscarriage was determined based on negative fetal heartbeat indication during the first trimester. Data were recorded and obtained in hospital setting and research was performed in university setting. RESULTS: A minimal subset of six non-redundant morphodynamic features were screened that maintained high prediction capacity. Features that account for the distribution of the nucleolus precursor bodies within the small pronucleus and pronuclei dynamics were highly predictive of miscarriage outcome as evaluated using the SHapley Additive exPlanations (SHAP) methodology. Using this feature subset, XGBoost and random forest models were trained following a 100-fold Monte-Carlo cross validation scheme. Miscarriage was predicted with AUC 0.68 to 0.69. CONCLUSION: We report the development of a decision-support tool for identifying the embryos with high risk of miscarriage. Prioritizing embryos for transfer based on their predicted risk of miscarriage in combination with their predicted implantation potential is expected to improve live-birth rates and shorten time-to-pregnancy.

Delineating the heterogeneity of embryo preimplantation development using automated and accurate morphokinetic annotation.

PURPOSE: Our objective was to design an automated deep learning model that extracts the morphokinetic events of embryos that were recorded by time-lapse incubators. Using automated annotation, we set out to characterize the temporal heterogeneity of preimplantation development across a large number of embryos. METHODS: To perform a retrospective study, we used a dataset of video files of 67,707 embryos from four IVF clinics. A convolutional neural network (CNN) model was trained to assess the developmental states that appear in single frames from 20,253 manually-annotated embryos. Probability-weighted superposition of multiple predicted states was permitted, thus accounting for visual uncertainties. Superimposed embryo states were collapsed onto discrete series of morphokinetic events via monotonic regression of whole-embryo profiles. Unsupervised K-means clustering was applied to define subpopulations of embryos of distinctive morphokinetic profiles. RESULTS: We perform automated assessment of single-frame embryo states with 97% accuracy and demonstrate whole-embryo morphokinetic annotation with R-square 0.994. High quality embryos that had been valid candidates for transfer were clustered into nine subpopulations, as characterized by distinctive developmental dynamics. Retrospective comparative analysis of transfer versus implantation rates reveals differences between embryo clusters as marked by poor synchronization of the third mitotic cell-cleavage cycle. CONCLUSIONS: By demonstrating fully automated, accurate, and standardized morphokinetic annotation of time-lapse embryo recordings from IVF clinics, we provide practical means to overcome current limitations that hinder the implementation of morphokinetic decision-support tools within clinical IVF settings due to inter-observer and intra-observer manual annotation variations and workload constrains. Furthermore, our work provides a platform to address embryo heterogeneity using dimensionality-reduced morphokinetic descriptions of preimplantation development.

The effect of hyaluronic acid in embryo transfer media in donor oocyte cycles and autologous oocyte cycles: a systematic review and meta-analysis.

STUDY QUESTION: Does the addition of hyaluronic acid (HA) to embryo transfer medium improve pregnancy outcomes in both autologous and oocyte donation IVF cycles? SUMMARY ANSWER: The best available evidence indicates that the addition of HA to embryo transfer medium is clinically beneficial in cycles with autologous oocytes. WHAT IS KNOWN ALREADY: There is a known clinical benefit of HA addition to embryo transfer media but it is not known if HA affects donor and autologous oocyte cycles differently. STUDY DESIGN, SIZE, DURATION: A systematic review with meta-analysis was performed. The Cochrane Gynaecology and Fertility Group Trials Register, CENTRAL via Cochrane Register of Studies Online (CRSO), MEDLINE, Embase and PsycINFO electronic databases (until 8 January 2020) were searched for randomized controlled trials (RCTs) examining the effect of HA in embryo transfer medium on pregnancy outcomes. PARTICIPANTS/MATERIALS, SETTING, METHODS: RCTs with separate donor and autologous oocyte data that compared embryo transfer medium with functional HA concentrations (0.5 mg/ml) to those containing no or low HA concentrations (0.125 mg/ml) were included. Two review authors independently selected trials for inclusion, extracted data and assessed the included studies using the Cochrane risk of bias assessment tool. Pooled risk ratios and 95% CIs were calculated. A summary of findings table was generated using Grading of Recommendations, Assessment, Development and Evaluation criteria. Judgements about evidence quality were justified and incorporated into the reported results for each outcome. MAIN RESULTS AND THE ROLE OF CHANCE: Fifteen studies, totalling 4686 participants, were analysed. In autologous oocyte cycles, live birth increased from 32% to 39% when embryo transfer media contained functional HA concentrations (risk ratio (RR) 1.22, 95% CI 1.11-1.34; nine studies, 3215 participants, I2 = 39%, moderate-quality evidence (number needed to treat (NNT) 14). HA-enriched media increased clinical pregnancy and multiple pregnancy rates by 5% and 8%, respectively (RR 1.11, 95% CI 1.04-1.18; 13 studies, 4014 participants, I2 = 0%, moderate-quality evidence, NNT 21) and (RR 1.49, 95% CI 1.27-1.76; 5 studies, 2400 participants, I2 = 21%, moderate-quality evidence, number needed to harm 13). Conversely, in donor oocyte cycles, HA addition showed little effect on live birth and clinical pregnancy (RR 1.12 95% CI 0.86-1.44; two studies, 317 participants, I2 = 50%, low-quality evidence) and (RR 1.06, 95% CI 0.97-1.28; three studies, 351 participants, I2 = 23%, low-quality evidence). There was insufficient available information on multiple pregnancy in donor oocyte cycles and on total adverse effects in both groups to draw conclusions. LIMITATIONS, REASONS FOR CAUTION: There were limited studies with separate data on donor oocyte cycles and limited information on oocyte quality. Additionally, one-third of the included studies did not include the main outcome, live birth rate. WIDER IMPLICATIONS OF THE FINDINGS: There is a moderate level of evidence to suggest that functional HA concentration in embryo transfer medium increases clinical pregnancy, live birth and multiple pregnancy rates in IVF cycles using autologous oocytes. This effect was not seen in donor oocyte cycles, indicating either intrinsic differences between donor and autologous oocytes or lack of statistical power. The combination of HA addition to transfer media in cycles using autologous oocytes and a single embryo transfer policy might yield the best combination, with higher clinical pregnancy and live birth rates without increasing the chance of multiple pregnancies. STUDY FUNDING/COMPETING INTEREST(S): No financial assistance was received. The authors have no competing interests. REGISTRATION NUMBER: N/A.

Hyaluronic acid in embryo transfer media for assisted reproductive technologies.

BACKGROUND: This is an update of a Cochrane Review first published in the Cochrane Library (2010, Issue 7). To increase the success rate of assisted reproductive technologies (ARTs), adherence compounds such as hyaluronic acid (HA) have been introduced into subfertility management. Adherence compounds are added to the embryo transfer medium to increase the likelihood of embryo implantation, with the potential for higher clinical pregnancy and live birth rates. OBJECTIVES: To determine whether adding adherence compounds to embryo transfer media could improve pregnancy outcomes, including improving live birth and decreasing miscarriage, in women undergoing assisted reproduction. SEARCH METHODS: We searched the Cochrane Gynaecology and Fertility Group Trials Register, CENTRAL, MEDLINE, Embase, and PsycINFO electronic databases on 7 January 2020 for randomised controlled trials that examined the effects of adherence compounds in embryo transfer media on pregnancy outcomes. Furthermore, we communicated with experts in the field, searched trials registries, checked reference lists of relevant studies, and conference abstracts were handsearched. SELECTION CRITERIA: Only truly randomised controlled trials comparing embryo transfer media containing functional concentrations of adherence compounds to media with no or low adherence compound concentrations were included. DATA COLLECTION AND ANALYSIS: Two review authors selected trials for inclusion according to the above criteria, after which the same two review authors independently extracted data for subsequent analysis. Statistical analysis was performed according to the guidelines developed by Cochrane. We combined data to calculate pooled risk ratios (RRs) and 95% confidence intervals (CIs). We assessed statistical heterogeneity using the I² statistic. We used GRADE methods to assess the overall quality of evidence for the main comparisons. MAIN RESULTS: We analysed 26 studies with a total of 6704 participants. Overall, the certainty of evidence was low to moderate: the main limitations were imprecision and/or heterogeneity. Compared to embryos transferred in media containing no or low (0.125 mg/mL) HA, the addition of functional (0.5 mg/mL) HA concentrations to the transfer media probably increases the live birth rate (RR 1.21, 95% CI 1.1 to 1.31; 10 RCTs, N = 4066; I² = 33%; moderate-quality evidence). This suggests that if the chance of live birth following no HA addition in media is assumed to be 33%, the chance following HA addition would be between 37% and 44%. The addition of HA may slightly decrease miscarriage rates (RR 0.82, 95% CI 0.67 to 1.00; 7 RCTs, N = 3091; I² = 66%; low-quality evidence). Nevertheless, when only studies with low risk of bias were included in the analysis, there was no conclusive evidence of a difference in miscarriage rates (RR 0.96, 95% CI 0.75 to 1.23; N = 2219; I² = 36%). Adding HA to transfer media probably results in an increase in both clinical pregnancy (RR 1.16, 95% CI 1.09 to 1.23; 17 studies, N = 5247; I² = 40%; moderate-quality evidence) and multiple pregnancy rates (RR 1.45, 95% CI 1.24 to 1.70; 7 studies, N = 3337; I² = 36%; moderate-quality evidence). We are uncertain of the effect of HA added to transfer media on the rate of total adverse events (RR 0.86, 95% CI 0.40 to 1.84; 3 studies, N = 1487; I² = 0%; low-quality evidence). AUTHORS' CONCLUSIONS: Moderate-quality evidence shows improved clinical pregnancy and live birth rates with the addition of HA as an adherence compound in embryo transfer media in ART. Low-quality evidence suggests that adding HA may slightly decrease miscarriage rates, but when only studies at low risk of bias were included in the analysis, the results were inconclusive. HA had no clear effect on the rate of total adverse events. The increase in multiple pregnancy rates may be due to combining an adherence compound and transferring more than one embryo. Further studies of adherence compounds with single embryo transfer need to be undertaken.

Is there a pre-retrieval serum human chorionic gonadotrophin (hCG) threshold level for prediction of subsequent retrieval outcomes and pregnancy in fresh ICSI cycles?

Modern ICSI (intracytoplasmic sperm injection) cycles' outcomes are difficult to predict. Whether human chorionic gonadotrophin (hCG) or luteinizing hormone (LH) serum levels 24 h prior to oocyte retrieval are correlated with retrieval and subsequent cycle results is unclear. An observational historic cohort study of 645 fresh ICSI cycles was conducted. After controlled oocyte stimulation, and 10-12 h after a self-administered trigger, serum levels of hCG (hCG trigger n = 563) and LH (GnRHa trigger n = 82) were measured. Correlations between pre-retrieval hormone levels and cycle results were assessed. No correlation (p > .12) was found between serum pre-retrieval hCG levels or LH levels (in GnRHa-triggered cycles) and total oocytes, M2, M1 + M2 or oocyte maturity rates (OMR) for any of the stimulation protocols. ROC (receiver operator curve) analysis for fertilization rates showed a possible cutoff for LH levels. Pregnancy rates (PR) were higher in rising hCG groups; a cutoff of 117 IU/L was associated with an increase in PR (30.9% to 45.6%) and a moderate sensitivity and specificity (60.6% and 55.0%). However, HCG was not predictive of pregnancy in a logistic regression model. We conclude that preretrieval hCG serum levels are not useful for pre-retrieval estimation of aspiration results but might have a role in prediction of pregnancy.

Worldwide live births following the transfer of chromosomally "Abnormal" embryos after PGT/A: results of a worldwide web-based survey.

PURPOSE: Preimplantation genetic testing for aneuploidy (PGT-A) has become increasingly controversial since normal euploid births have been reported following transfer of embryos diagnosed as "abnormal." There is an increasing trend in transferring "abnormal" embryos; but it is still unknown how many IVF centers transfer "abnormal" embryos and with what efficiency. METHODS: We performed a worldwide web-survey of IVF centers to elucidate PGT-A related practice patterns including transfer of human embryos found "abnormal" by PGT-A. Participating centers reflected in vitro fertilization (IVF) cycles in the USA, Canada, Europe, Asia, South America, and Africa. RESULTS: One hundred fifty-one IVF centers completed the survey; 125 (83%) reported utilization of PGT-A. Europe had the highest utilization (32.3%), followed by the USA and Canada combined at 29.1%. The leading indications for PGT-A were advanced maternal age (77%), followed by recurrent implantation failure (70%), unexplained pregnancy loss (65%), and sex determination (25%); 14% of respondents used PGT-A for all of their IVF cycles; 20% of IVF units reported transfers of chromosomally "abnormal" embryos, and 56% of these took place in the USA, followed by Asia in 20%. Remarkably, 106 (49.3%) cycles resulted in ongoing pregnancies (n = 50) or live births (n = 56). Miscarriages were rare (n = 20; 9.3%). CONCLUSIONS: The transfers of "abnormal" embryos by PGT-A offered robust pregnancy and live birth chances with low miscarriage rates. These data further strengthen the argument that PGT-A cannot reliably determine which embryos should or should not be transferred and leads to disposal of many normal embryos with excellent pregnancy potential.

Oncofertility: insights from IVF specialists-a worldwide web-based survey analysis.

PURPOSE: This research sought to understand IVF-physicians' knowledge of, experience with, and attitudes toward fertility preservation for cancer patients. METHODS: A 35-question, self-report survey request was emailed to IVF providers who were registered on the IVF-Worldwide.com network (3826 clinics). Physicians submitted responses on the IVF-Worldwide.com website. Survey results were reported as a proportion of the responding clinics. RESULTS: Survey responses were completed by 321 (8.4%) globally distributed IVF clinics, representing 299,800 IVF cycles. Of these clinics, 86.6% (278) performed fertility preservation, treating approximately 6300 patients annually. However, 18.4% of the centers reported that patients sought advice independently, without an oncologist's referral. Ovarian tissue cryopreservation was performed by 37.7% of the clinics, yet 52.6% considered the procedure experimental. IVM was performed by 16.5% of responding clinics. A majority (63.6%) of the clinics selected treatment protocols based on each patient's malignancy. Most respondents (76.3%) disagreed that fertility preservation was not yet successful enough to make it an available option. However, 44.2% believed that pregnancy rates following oocyte cryopreservation could not be determined because not enough oocyte cryopreservation patients had completed embryo transfer. CONCLUSIONS: Most clinics performed fertility preservation, tailoring protocols to each patient's disease and condition. Almost 20% of patients sought advice independently, indicating that more effort is needed to encourage oncologists to refer patients. Most survey respondents believed that data was not yet available on either live birth outcomes or the best protocol for each disease. Therefore, long-term study must continue, with the establishment of interim milestones and an outcome-tracking registry.

Preimplantation genetic screening: results of a worldwide web-based survey.

Our objective was to evaluate and characterize the extent and patterns of worldwide usage of preimplantation genetic screening (PGS) among the assisted reproductive technique community. A prospective, web-based questionnaire with questions relating to practices of, and views on, PGS was directed to users and non-users of PGS. A total of 386 IVF units from 70 countries conducting 342,600 IVF cycles annually responded to the survey. A total of 77% of respondents routinely carry out PGS in their clinics for a variety of indications: advanced maternal age (27%), recurrent implantation failure (32%) and recurrent pregnancy loss (31%). Few (6%) offer PGS to all their patients. In most cycles (72%), trophectoderm biopsy is carried out and either array-comparative genomic hybridization (59%) or next-generation sequencing (16%) are used for genetic analysis. Only 30% of respondents regard PGS as clearly evidenced-based, and most (84%) believe that more randomized controlled trials are needed to support the use of PGS. Despite ongoing debate and lack of robust evidence, most respondents support the use of PGS, and believe that it may aid in transferring only euploid embryos, thereby reducing miscarriage rates and multiple pregnancies, increasing live birth rates and reducing the risk of aneuploid pregnancies and births.

Use of various gonadotropin and biosimilar formulations for in vitro fertilization cycles: results of a worldwide Web-based survey.

PURPOSE: The purpose of this study was to identify trends in gonadotropin therapy in patients undergoing in vitro fertilization (IVF) treatment worldwide. METHODS: Retrospective evaluation utilizing the results of a Web-based survey, IVF-Worldwide ( www.IVF-worldwide.com ) was performed. RESULTS: Three hundred fourteen centers performing a total of 218,300 annual IVF cycles were evaluated. Respondents representing 62.2% of cycles (n = 135,800) did not believe there was a difference between urinary and recombinant gonadotropins in terms of efficacy and live birth rate. Of the respondents, 67.3% (n = 146,800) reported no difference between recombinant and urinary formulations in terms of short-term safety and risk of ovarian hyperstimulation syndrome. In terms of long-term safety using human urinary gonadotropins, 50.6% (n = 110,400) of respondents believe there are potential long-term risks including prion disease. For 95.3% of units (n = 208,000), the clinician was the decision maker determining which specific gonadotropins are used for IVF. Of the units, 62.6% (n = 136,700) identified efficacy as the most important factor in deciding which gonadotropin to prescribe. While most (67.3%, n = 146,800) were aware of new biosimilar recombinant FSH products entering the market, 92% (n = 201,000) reported they would like more information. A fraction of respondents (25.6%, n = 55,900) reported having experience with these new products, and of these, 80.3% (n = 46,200) reported that they were similar in efficacy as previously used gonadotropins in a similar patient group. CONCLUSIONS: Respondents representing the majority of centers do not believe a difference exists between urinary and recombinant gonadotropins with respect to efficacy and live birth rates. While many are aware of new biosimilar recombinant FSH products entering the market, over 90% desire more information on these products.

How to define, diagnose and treat poor responders? Responses from a worldwide survey of IVF clinics.

Poor responders represent a significant percentage of couples treated in IVF units (10-24%), but the standard definition of poor responders remains uncertain and consequently optimal treatment options remain subjective and not evidence-based. In an attempt to provide uniformity on the definition, diagnosis and treatment of poor responders, a worldwide survey was conducted asking IVF professionals a set of questions on this complex topic. The survey was posted on www.IVF-worldwide.com, the largest and most comprehensive IVF-focused website for physicians and embryologists. A total of 196 centres replied, forming a panel of IVF units with a median of 400 cycles per year. The present study shows that the definition of poor responders is still subjective, and many practices do not use evidence-based treatment for this category of patients. Our hope is that by leveraging the great potential of the internet, future studies may provide immediate large-scale sampling to standardize both poor responder definition and treatment options.