Machine learning for prediction of euploidy in human embryos: in search of the best-performing model and predictive features.

OBJECTIVE: To assess the best-performing machine learning (ML) model and features to predict euploidy in human embryos. DESIGN: Retrospective cohort analysis. SETTING: Department for reproductive medicine in a university hospital. PATIENT(S): One hundred twenty-eight infertile couples treated between January 2016 and December 2019. Demographic and clinical data and embryonic developmental and morphokinetic data from 539 embryos (45% euploid, 55% aneuploid) were analyzed. INTERVENTION(S): Random forest classifier (RFC), scikit-learn gradient boosting classifier, support vector machine, multivariate logistic regression, and naïve Bayes ML models were trained and used in 9 databases containing either 26 morphokinetic features (as absolute [A1] or interim [A2] times or combined [A3]) alone or plus 19 standard development features [B1, B2, and B3] with and without 40 demographic and clinical characteristics [C1, C2, and C3]. Feature selection and model retraining were executed for the best-performing combination of model and dataset. MAIN OUTCOME MEASURE(S): The main outcome measures were overall accuracy, precision, recall or sensitivity, F1 score (the weighted average of precision and recall), and area under the receiver operating characteristic curve (AUC) of ML models for each dataset. The secondary outcome measure was ranking of feature importance for the best-performing combination of model and dataset. RESULT(S): The RFC model had the highest accuracy (71%) and AUC (0.75) when trained and used on dataset C1. The precision, recall or sensitivity, F1 score, and AUC were 66%, 86%, 75%, and 0.75, respectively. The accuracy, recall or sensitivity, and F1 score increased to 72%, 88%, and 76%, respectively, after feature selection and retraining. Morphokinetic features had the highest relative predictive weight. CONCLUSION(S): The RFC model can predict euploidy with an acceptable accuracy (>70%) using a dataset including embryos' morphokinetics and standard embryonic development and subjects' demographic and clinical features.

Low feasibility of in vitro matured oocytes originating from cumulus complexes found during ovarian tissue preparation at the moment of gender confirmation surgery and during testosterone treatment for fertility preservation in transgender men.

OBJECTIVE: To study the feasibility of in vitro maturation of ovarian tissue oocytes for fertility preservation in transgender men on testosterone treatment. DESIGN: Cross-sectional study SETTING: University hospital PATIENT(S): Eighty-three transgender men enrolled from November 2015 to January 2019 INTERVENTION(S): In vitro maturation of cumulus-oocyte complexes (COCs) harvested at the time of gender confirmation surgery, and fertilization through intracytoplasmic sperm injection. MAIN OUTCOME MEASURE(S): In vitro maturation, fertilization, and blastulation rates; comparison of morphokinetics with vitrified-warmed oocytes; and analysis of the genetic profiles of embryos. SECONDARY OUTCOMES: association between serum hormone levels; COCs' morphologic characteristics, and vitrification rate. RESULT(S): All participants were on testosterone treatment for a median of 83 (64[Quartile 1]; 113.2[Quartile 2]) weeks. A total of 1,903 COCs (mean per participant, 23 ± 15.8) were collected. The in vitro maturation rate was 23.8%, vitrification rate was 21.5%, and survival rate after warming was 72.6% (n = 151). Intracytoplasmic sperm injection was performed in 139 oocytes. The rate of normal fertilized oocytes was 34.5%, and 25 (52.1%) embryos reached day 3. One blastocyst was achieved on day 5. Aberrant cleavage patterns and early embryo arrest were observed in 22 (45.8%) and 44 (91.7%) zygotes, respectively. Compared with vitrified-warmed donor oocytes, a delay was observed in pronuclei disappearance, t2 (time to reach 2 cell stage) timings, and CC1 (the duration of the 1st cell cycle) and SS3 (synchronization of cleavage pattern (calculated as t8-t5) time intervals. A normal genetic pattern was seen in 42% embryos. The proportion of vitrified oocytes was negatively associated with progesterone (odds ratio, 0.76) and positively associated with antimüllerian hormone serum levels (odds ratio, 1.23). The highest vitrification rate was achieved by the morphologic characteristic 344 at day 0 and by 433 at day 2. CONCLUSION(S): Ovarian tissue oocytes matured in vitro show low developmental capacity in transgender men, when collected under testosterone treatment.

A methodological validation of an easy one-step swimout semen preparation procedure for selecting DNA fragmentation-free spermatozoa for ICSI.

The main purpose of this methodological paper was to describe a recently designed one-step ICSI semen preparation swim-out method (called swim-ICSI) and to compare its efficacy with our conventional two-step swim-out method for the selection of motile spermatozoa for ICSI with minimal DNA damage. In this observational cohort study, 42 fresh ejaculate sperm samples for ICSI were included to compare the new swim-ICSI with the conventional swim-out. In a sub-analysis (n = 20), both in-house designed ICSI preparation methods were compared with a commercial magnetic-activated cell sorting test (MACS® ). Sperm DNA fragmentation (SDF), using Halosperm® , was determined at different time points during sperm preparation: on the native sample (a), after density gradient centrifugation (DG) (b), on the motile (A + B) spermatozoa selected with conventional swim-out post-DG (c) and selected with swim-ICSI method post-DG (d). For a subgroup (n = 20), SDF was also calculated after MACS (e). The mean SDF significantly reduced after EACH preparation step and reduced to almost zero in the recovered A + B spermatozoa when the semen prepared with DG was further processed for ICSI (swim-ICSI vs. swim-out, p = .001). In conclusion, the optimised one-step and fine-tuned swim-ICSI technique shows the possibility to select a population of spermatozoa with almost zero SDF to be used in ICSI treatments.

In-vitro development of embryos derived from vitrified-warmed oocytes is delayed compared with embryos derived from fresh oocytes: a time-lapse sibling oocyte study.

RESEARCH QUESTION: Is there a difference in blastocyst formation between fresh and vitrified-warmed sibling oocytes and can this difference be attributed to changes in embryo morphokinetics? DESIGN: Between February 2016 and December 2017, 472 metaphase II (MII) oocytes in 67 donor-recipient cycles from 27 different healthy anonymous oocyte donors were allocated for fresh transfer (FSHO) (n = 220) to a synchronous recipient (n = 36) or vitrified (VITO) (n = 252) to be warmed and transferred to another recipient (n = 31). Embryos derived from the FSHO and their sibling VITO were analysed for morphokinetic development using time-lapse imaging, blastocyst formation and clinical outcome. RESULTS: Time-lapse analysis showed an overall delay in cleavage rate from the time of pronuclei disappearance up to the time of blastulation in the VITO compared with their sibling FSHO. Twelve morphokinetic variables were significantly different between the groups. On Day 5 significantly more FSHO embryos developed to blastocyst (expansion 1-6) and reached the full blastocyst stage (expansion 3-6) compared with the VITO embryos [53.2% (84/158) versus 40.0% (64/160); P = 0.0244 and 48.1% (76/158) versus 31.3% (50/160); P = 0.0028, respectively]. The embryo utilization rate was similar in both groups at the time of cryopreservation; 51.3% (FSHO) versus 45.0% (VITO) (P = 0.3124). The pregnancy rate per cycle was 47.2% (17/36) in FSHO patients and 48.4% (15/31) in VITO patients (P = 1). Limitations in this study: non-randomized, small study size and not powered to detect differences in clinical outcomes. CONCLUSIONS: Timing of development is altered and blastocyst formation is delayed in embryos derived from vitrified-warmed donor oocytes compared with their fresh sibling counterparts. Although preliminary results suggest that the clinical impact of this delay may be limited, this needs further investigation in larger randomized studies.

Assisted oocyte activation significantly increases fertilization and pregnancy outcome in patients with low and total failed fertilization after intracytoplasmic sperm injection: a 17-year retrospective study.

OBJECTIVE: To investigate the extent to which assisted oocyte activation (AOA) improves clinical outcomes in patients diagnosed with oocyte activation deficiencies (OADs). DESIGN: Retrospective cohort study comparing AOA cycles and previous intracytoplasmic sperm injection (ICSI) cycles in couples experiencing low or total failed fertilization after ICSI. Importantly, the sperm-related oocyte-activating capacity was examined in all patients before AOA with the use of the mouse oocyte activation test (MOAT). SETTING: Infertility center at a university hospital. PATIENT(S): A total of 122 couples with a history of low or total failed fertilization after ICSI. INTERVENTION(S): ICSI, MOAT, AOA, and embryo transfer. MAIN OUTCOME MEASURE(S): Fertilization, pregnancy, and live birth rates. RESULT(S): MOAT revealed 19 patients with a sperm-related OAD (MOAT group 1), 56 patients with a diminished sperm-related oocyte-activating capacity (MOAT group 2), and 47 patients with a suspected oocyte-related OAD (MOAT group 3). AOA (191 cycles) significantly improved fertilization, pregnancy, and live birth rates in all MOAT groups compared with previous ICSI attempts (243 cycles). Fertilization rates after AOA were significantly different among MOAT groups 1 (70.1%), 2 (63.0%), and 3 (57.3%). Between MOAT group 1 and 3, significant differences in pregnancy (49.0% vs. 29.4%) and live birth (41.2% vs. 22.1%) rates were observed. In total, 225 embryo transfers resulted in 60 healthy live births following AOA. CONCLUSION(S): Patients undergoing diagnostic testing before AOA show a significant improvement in clinical outcomes compared with previous cycles. Our findings highlight that AOA should be reserved for patients with clear OADs.

Embryo selection using time-lapse analysis (Early Embryo Viability Assessment) in conjunction with standard morphology: a prospective two-center pilot study.

STUDY QUESTION: Does prospective embryo selection using the results from the Eava Test (Early Embryo Viability Assessment) in combination with standard morphology increase the pregnancy rate of IVF and ICSI patients compared to embryo selection based on morphology only? SUMMARY ANSWER: Embryo selection using the Eeva Test plus standard morphology on Day 3 results in comparable pregnancy rates as conventional morphological embryo selection. WHAT IS KNOWN ALREADY: Time-lapse monitoring of embryo development may represent a superior way to culture and select embryos in vitro. The Eeva Test records the development of each embryo with a cell-tracking system and predicts the likelihood (High, Medium or Low) that an embryo will form a blastocyst based on an automated analysis of early cell division timings. STUDY DESIGN, SIZE, DURATION: This trial was designed as a prospective, observational, two-center pilot study with a propensity matched control group. The analysis involved 280 of 302 enrolled patients who were included in the Eeva Test group in 2013 and 560 control patients who were treated in the years 2011-2013. The majority of transfers (98%) were single embryo transfers. PARTICIPANTS/MATERIALS, SETTING, METHODS: Two academic hospitals (VUmc Amsterdam and UZ Gent) enrolled patients <41 years old, with <3 previous attempts and ≥5 normally fertilized eggs. Propensity matching was used to identify a propensity matched control group from a cohort of 1777 patients based on age, cycle number, oocyte number and number of fertilized oocytes. MAIN RESULTS AND THE ROLE OF CHANCE: There was no difference in patient baseline characteristics between the two groups. The ongoing pregnancy rate (OPR) of patients enrolled in the Eeva Test group (34.3%; 96/280) did not differ significantly from the OPR in the propensity matched control group (34.6%, 194/560; P = 0.92). However, significantly less top quality embryos (eight-cell embryos with ≤25% fragmentation) were transferred in the Eeva Test group compared to the propensity matched control group (70.4% vs. 82.3%; P < 0.001). The transfer of Eeva High and Medium embryos resulted in a significantly higher OPR of 36.8% (89/242) compared to 18.4% (7/38) for Eeva Low embryos (P = 0.02). LIMITATION, REASONS FOR CAUTION: This pilot study is limited by its nonrandomized design with a concurrent and historical control. WIDER IMPLICATIONS OF THE FINDINGS: Our pilot data did not reveal significant differences between time-lapse based and conventional embryo selection. Interestingly, the pregnancy rates were comparable in both groups even though the morphological quality of the transferred embryos was significantly lower in the Eeva Test group compared to the propensity matched control group. A sufficiently powered three-armed randomized controlled trial (RCT) with a solid design should be performed to generate decisive evidence in the future. STUDY FINDING/COMPETING INTERESTS: Progyny Inc., formerly Auxogyn provided the Eeva scopes, software and technical support for this study. The funding sources did neither influence data collection, management, analysis and interpretation of the data, nor the preparation of the manuscript. TRIAL REGISTRATION NUMBER: ClinicalTrials.gov: NCT01671644.

Sperm involved in recurrent partial hydatidiform moles cannot induce the normal pattern of calcium oscillations.

OBJECTIVE: To assess the Ca2+-releasing ability of sperm involved in partial hydatidiform moles. DESIGN: Analysis of the activating and Ca2+-releasing ability of human sperm. SETTING: University hospital research laboratory. PATIENT(S): Patients undergoing intracytoplasmic sperm injection (ICSI) treatment. INTERVENTION(S): Microinjection of mouse and human oocytes with sperm. MAIN OUTCOME MEASURE(S): Measurement of the fertilizing and Ca2+-releasing ability of human sperm. RESULT(S): The mouse oocyte Ca2+ analysis showed that only 19.0% (4/21) of the mouse oocytes injected with sperm involved in molar pregnancies exhibited a normal pattern of Ca2+ oscillations versus 63.2% (36/57) of those injected with control sperm. Further, 83.3% (15/18) of donated in vitro-matured human oocytes injected with deficient sperm did not exhibit any Ca2+ release, while 76.9% (10/13) failed to show normal pronuclear development. Yet the sperm oocyte activation factor phospholipase C zeta (PLCζ) was present in the majority (96.6%, n=113) of the analyzed sperm at a normal expression level. Eventually, fertilization failure was overcome with assisted oocyte activation in subsequent therapeutic ICSI cycles, which led to normal deliveries. CONCLUSION(S): Sperm that previously provoked recurrent partial hydatidiform mole pregnancies due to dispermic fertilization is not able to activate human oocytes or trigger the normal pattern of Ca2+ oscillations in mouse and human oocytes in vitro.

Efficiency of assisted oocyte activation as a solution for failed intracytoplasmic sperm injection.

Failed fertilization after intracytoplasmic sperm injection (ICSI) can occur due to an oocyte activation defect. In these cases assisted oocyte activation (AOA) may help but efficiency is still unknown. Prior to AOA, the mouse oocyte activation test (MOAT) can be carried out by injecting human spermatozoa into mouse oocytes to evaluate their activating capacity. According to the MOAT activation percentage achieved, patients were classified into three groups: 0-20% (16 patients); 20-85% (seven patients); 85-100% (seven patients). For AOA, CaCl(2) was injected together with spermatozoa followed by a double Ca(2+) ionophore treatment. The fertilization rates before application of AOA in 50 cycles were 6%, 22% and 14% in, respectively, groups 1, 2 and 3 without any pregnancy. Fertilization and pregnancy rates after AOA in 61 cycles were significantly increased to 75% and 34% for group 1, 73% and 43% for group 2, and 75% and 17% for group 3 (P < 0.0001 and P < 0.004, respectively). Application of AOA results in normal fertilization and pregnancy rates in patients whose spermatozoa show deficient activation. When MOAT reveals no activation deficiency in spermatozoa, AOA still allows for high fertilization and acceptable pregnancy rates. The obstetric and neonatal outcomes after AOA were normal as no malformations were observed.