Making IVF more effective through the evolution of prediction models: is prognosis the missing piece of the puzzle?

Does a public online IVF prediction tool help set patient expectations? A mixed methods evaluation study

STUDY QUESTION

Does a public online IVF success prediction calculator based on real-world data help set patient expectations?

SUMMARY ANSWER

The YourIVFSuccess Estimator aided consumer expectations of IVF success: one quarter (24%) of participants were unsure of their estimated IVF success before using the tool; one half changed their prediction of success after using the tool and one quarter (26%) had their expectations of IVF success confirmed.

WHAT IS KNOWN ALREADY

Several web-based IVF prediction tools exist worldwide but have not been evaluated for their impact on patient expectations, nor for patient perceptions of usefulness and trustworthiness.

STUDY DESIGN, SIZE, DURATION

This is a pre-post evaluation of a convenience sample of 780 online users of the Australian YourIVFSuccess Estimatorhttps://yourivfsuccess.com.au/ between 1 July and 31 November 2021.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Participants were eligible if they were over 18 years of age, Australian residents, and considering IVF for themselves or their partner. Participants filled in online surveys before and after using the YourIVFSuccess Estimator.

MAIN RESULTS AND THE ROLE OF CHANCE

The response rate of participants who completed both surveys and the YourIVFSuccess Estimator was 56% (n = 439). The YourIVFSuccess Estimator aided consumer expectations of IVF success: one quarter (24%) of participants were unsure of their estimated IVF success before using the tool; one half changed their prediction of success after using the tool (20% increased, 30% decreased), bringing their predictions in line with the YourIVFSuccess Estimator, and one quarter (26%) had their IVF success expectations confirmed. One in five participants claimed they would change the timing of IVF treatment. The majority of participants found the tool to be at least moderately trustworthy (91%), applicable (82%), and helpful (80%), and would recommend it to others (60%). The main reasons given for the positive responses were that the tool is independent (government funded, academic) and based on real-world data. Those who did not find it applicable or helpful were more likely to have had a worse-than-expected prediction, or to have experienced non-medical infertility (e.g. single women, LGBTQIA+), noting that at the time of evaluation the Estimator did not accommodate these patient groups.

LIMITATIONS, REASONS FOR CAUTION

Those who dropped out between the pre- and post-surveys tended to have a lower education status or have been born outside of Australia or New Zealand, therefore there may be issues with generalizability.

WIDER IMPLICATIONS OF THE FINDINGS

With consumers demanding increasing levels of transparency and participation in decisions around their medical care, public-facing IVF predictor tools based on real-world data are useful for aligning expectations about IVF success rates. Given differences in patient characteristics and IVF practices internationally, national data sources should be used to inform country-specific IVF prediction tools.

STUDY FUNDING/COMPETING INTEREST(S)

The YourIVFSuccess website and evaluation of the YourIVFSuccess Estimator are supported by the Medical Research Future Fund (MRFF) Emerging Priorities and Consumer Driven Research initiative: EPCD000007. BKB, ND, and OF have no conflicts to declare. DM holds a clinical role at Virtus Health. His role did not influence the analysis plan or interpretation of results in this study. GMC is an employee of the UNSW Sydney, and Director of the UNSW NPESU. UNSW receives research funding on behalf of Prof Chambers from the MRFF to develop and manage the Your IVF Success website. Grant ID: MRFF Emerging Priorities and Consumer Driven Research initiative: EPCD000007.

TRIAL REGISTRATION NUMBER

N/A.

Machine Learning Approach to Predict Clinical Pregnancy Potential in Women Undergoing IVF Program

Objective: Hidden knowledge could be discovered within a large practical data of in vitro fertilization (IVF) practice. In this study, Machine learning-based data mining techniques were utilized to construct a reliable prediction model for clinical pregnancy in IVF.

Study Design: A retrospective cohort multicenter study involving 4.570 IVF cycles. All patients underwent fresh embryo transfer at either the cleavage or blastocyst stage between January 2015 and December 2019. The experiment focused on utilizing tree-based classifiers to generate and compare the most effective prediction model that could predict a clinical pregnancy through clinical data. Additionally, each classifier is optimized via a genetic algorithm technique, along with the selection of variables.

Results: Both the decision tree and random forest showed similar performance that was much better than the gradient boost. The two superior classifiers achieved a balanced accuracy of roughly 0.62. Additionally, each prediction model was shown to work optimally with different combinations of variables, with some variables being consistently included, such as female age, and some consistently excluded, which provides an insight into the relationship between the variables and each prediction model.

Conclusion: Machine learning algorithm remains effective for the purpose of data mining and knowledge extraction in IVF clinical datasets through which a relatively reliable prediction system for clinical pregnancy could be constructed, provided the available data is sufficient.

Whole-Genome Methylation Analysis Revealed ART-Specific DNA Methylation Pattern of Neuro- and Immune-System Pathways in Chinese Human Neonates

The DNA methylation of human offspring can change due to the use of assisted reproductive technology (ART). In order to find the differentially methylated regions (DMRs) in ART newborns, cord blood maternal cell contamination and parent DNA methylation background, which will add noise to the real difference, must be removed. We analyzed newborns’ heel blood from six families to identify the DMRs between ART and natural pregnancy newborns, and the genetic model of methylation was explored, meanwhile we analyzed 32 samples of umbilical cord blood of infants born with ART and those of normal pregnancy to confirm which differences are consistent with cord blood data. The DNA methylation level was lower in ART-assisted offspring at the whole genome-wide level. Differentially methylated sites, DMRs, and cord blood differentially expressed genes were enriched in the important pathways of the immune system and nervous system, the genetic patterns of DNA methylation could be changed in the ART group. A total of three imprinted genes and 28 housekeeping genes which were involved in the nervous and immune systems were significant different between the two groups, six of them were detected both in heel blood and cord blood. We concluded that there is an ART-specific DNA methylation pattern involved in neuro- and immune-system pathways of human ART neonates, providing an epigenetic basis for the potential long-term health risks in ART-conceived neonates.

Anti-Müllerian hormone as a predictor of pregnancy in women under 35 years with unexplained infertility undergoing ICSI: a retrospective study

BACKGROUND

Anti-Müllerian hormone (AMH) is a well-established marker for the determination of ovarian reserve. However, its role in the prediction of pregnancy is still under debate. In this retrospective study, we aimed to evaluate the relationship of serum AMH levels with pregnancy rates in patients with unexplained infertility undergoing ICSI. Moreover, we compared the predictive value of AMH with that of antral follicle count (AFC).

METHODS

Records of 76 patients under 35 years of age with AMH levels between 1 and 3.5 ng/ml were examined retrospectively. Participants were divided into groups based on their AMH level and age.

RESULTS

AMH levels in women under 30 years were found significantly higher than those in women over 30 years (P=0.033). 57 of 76 patients (75%) were pregnant. Age did not have a significant effect on the pregnancy rates in the selected study group (P=0.252). On the other hand, despite the poor predictive accuracy, serum AMH was shown to have a predictive value with a cut-off point of 1.95 ng/mL. Logistic regression tests demonstrated a higher pregnancy rate (3.396 fold) with an AMH level 1.95 or above. There was no significant relationship between AFC and pregnancy.

CONCLUSIONS

AMH might have a role in the prediction of pregnancy after ICSI in patients under 35 years with unexplained infertility.

Sperm selection during ICSI treatments reduces single- but not double-strand DNA break values compared to the semen sample

PURPOSE

To detect a possible bias in sperm DNA fragmentation (SDF) testing when performed on semen samples or on those few spermatozoa selected for Intracytoplasmic Sperm Injection (ICSI) treatments.

METHODS

A multimethodological analysis of Single- and Double-Strand DNA Breaks (SSB and DSB, respectively) was performed through the Neutral Comet, the Alkaline Comet, the Sperm Chromatin Dispersion (SCD) and the Terminal deoxynucleotidyl transferase dUTP Nick End Labelling (TUNEL) assays. SDF was evaluated in (i) semen samples from 23 infertile patients (not achieving pregnancy or suffering recurrent miscarriage); (ii) samples after a Swim-up and (iii) spermatozoa microselected for ICSI (ICSI-S).

RESULTS

The analysis of 3217 ICSI-S revealed a significant reduction of SSB values compared to the Ejaculate and the Swim-up samples. On the contrary, DSB values were not reduced after any sperm selection method. The No-pregnancy group presented poorer semen parameters and higher SSB values. The Recurrent miscarriage group presented better semen parameters but also higher DSB values.

CONCLUSION

The analysis of SDF on semen samples may not be fully representative of those few spermatozoa selected for ICSI. Since oxidative stress impairs sperm motility and causes SSB, selecting a motile sperm may intrinsically imply choosing a sperm not affected by this damage. DSB have an enzymatic origin which does not affect motility, making it difficult to select a sperm without this damage. Therefore, ICSI treatments could be effective in patients presenting high SSB values. Patients presenting high DSB values should expect bad ICSI results if this damage is not reduced through other specific methods.

Evaluating the value of day 0 of an ICSI cycle on indicating laboratory outcome

A number of oocyte characteristics have been associated with fertilization, implantation and live-birth rates, albeit without reaching a consensus. This study aims to delineate possible associations between oocyte characteristics, oocyte behavior during intracytoplasmic sperm injection (ICSI), fertilization potential, and laboratory outcomes. Four-hundred and seventy-seven patients, yielding 3452 oocytes, were enrolled in this prospective observational study from 2015 to 2018. Οoplasm granularity was associated with poor embryo quality and higher probabilities of post-ICSI oocytes and embryos discarded in any developmental stage and never selected for embryo transfer or cryopreservation (p < 0.001). Both sudden or difficult ooplasm aspiration, and high or lack of resistance during ICSI were associated with either a poor Zygote-Score or fertilization failure (p < 0.001). Sudden or difficult ooplasm aspiration and high resistance during ICSI penetration were positively associated with resulting to a post-ICSI oocyte or embryo that would be selected for discard. Evaluation of oocyte characteristics and oocyte behavior during ICSI may provide early information regarding laboratory and cycle outcomes. Particularly, ooplasm granularity, and fragmentation of polar body, along with sudden or difficult ooplasm aspiration and high or lack of resistance during ICSI penetration may hinder the outcome of an ICSI cycle. The associations presented herein may contribute towards development of a grading system or a prediction model. Taking into account information on oocytes and ICSI behavior may effectively assist in enhancing IVF outcome rates.

An elective single cleavage embryo transfer strategy to minimize twin live birth rate based on a prediction model from double cleavage embryos transfer patients

OBJECTIVE

To minimize twin birth rate by establishing an elective single cleavage embryo transfer strategy based on a twin live birth prediction model from fresh double cleavage embryos transfer (cleavage DET) patients.

METHODS

A total of 2478 patients underwent fresh cleavage DET in Nanjing Tower Hospital were enrolled to establish the twin live birth prediction model by logistic regression analysis and the cutoff value was calculated by ROC curve. Another 300 fresh cleavage DET patients and 550 cleavage single-embryo transfer (SET) patients were selected to testify the sensitivity, specificity and usefulness of this model.

RESULTS

The twin live birth probability (TLBP) = e^X /(e^X + 1), e is a natural logarithm, X = -1.763 - (0.319 × female age) + (0.329 × endometrial thickness) + (0.282 × the number of transferred top embryos) - (0.314 × previous transfer times), and the cutoff value of TLBP was 24.2%. The sensitivity of this model for predicting twin live birth was 75.6%, while the specificity was 52.5% in the external validation of 300 DET patients. Furthermore, the validation of 550 SET patients showed that the live birth rate of TLBP value positive patients was significantly higher than that in negative patients (54.3% vs. 35.5%, p < .001). When adopted an elective single cleavage embryo transfer strategy, the patients with a positive TLBP value choose SET, while still undergo DET who with a negative TLBP value, the live birth rate would maintain as 56.7%; however, the twin birth rate would significantly decline to 7.4%.

CONCLUSION

Female age, endometrial thickness, the number of transferred top embryos and previous embryo transfer times were critical variables for the twin live birth prediction model. An elective single cleavage embryo transfer strategy according to this model can maintain the relatively high live birth rate, meanwhile get the acceptable low twin birth rate.

Lactoferrin levels in cervical fluid from in vitro fertilization (IVF) patients - correlation with IVF parameters

Since our previous results suggest that lactoferrin (LF) might have roles in the reproductive process and that its levels might change in the female tract as a response to various factors, the aim of this investigation was to assess whether LF levels in cervical secretions correlate with reproductive parameters from in vitro fertilization (IVF) patients. Cervical fluid samples were obtained from 34 women under 40 years old enrolled for assisted reproduction techniques, and LF concentration was measured. The mean total protein concentration in all cervical fluid samples was 842.8 ± 116.9 µg/mL. The mean concentration of LF was 0.73 ± 0.06 ng LF/µg of total proteins. We observed that higher LF levels in cervical fluid correlated with lower IVF rates when all patients were analyzed; this negative correlation was also sustained when only patients ≥35 years were studied. The mean LF concentration in cervical fluid was significantly lower among patients with normal IVF rates than in those with values 50% or less. Using a LF cutoff value of 0.83 ng/μg of total proteins, the study revealed a significant association between the LF levels below 0.83 ng/µg of total proteins and IVF rates above 50%. LF levels in cervical mucus could potentially be used as a marker of fertilization outcome.

Levels of caspase-3 and histidine-rich glycoprotein in the embryo secretome as biomarkers of good-quality day-2 embryos and high-quality blastocysts

Morphological assessment at defined developmental stages is the most important method to select viable embryos for transfer and cryopreservation. Timing of different developmental stages in embryo development has been shown to correlate with its potential to develop into a blastocyst. However, improvements in pregnancy rates by using time-lapse techniques have been difficult to validate scientifically. Therefore, there is a need for new methods, preferably non-invasive methods based on metabolomics, genomics and proteomics, to improve the evaluation of embryo quality even further. The aim of this study was to investigate if different levels of caspase-3 and histidine-rich glycoprotein (HRG), secreted by the embryo into the culture media, can be used as biomarkers of embryo quality. In this study, a total of 334 samples of culture media were collected from in vitro fertilization (IVF) treatments at three different clinics. Protein analysis of the culture media was performed using multiplex proximity extension protein analysis to detect levels of caspase-3 and HRG in the embryo secretome. Protein levels were compared in secretome samples from high- and low-quality blastocysts and embryos that became arrested during development. Correlation between protein levels and time to morula formation was also analyzed. Furthermore, protein levels in secretomes from day-2 cultured embryos were compared on the basis of whether or not pregnancy was achieved. The results showed that caspase-3 levels were lower in secretomes from high-quality vs. low-quality blastocysts and those that became arrested (p ≤ 0.05 for both). In addition, higher HRG levels correlated with a shorter time to morula formation (p ≤ 0.001). Caspase-3 levels were also lower in secretomes from day-2 cultured embryos resulting in a pregnancy vs. those that did not (p ≤ 0.05). Furthermore, it was shown that caspase-3 might be used as a marker for predicting potential success rate after transfer of day-2 cultured embryos, where a caspase-3 cutoff level of 0.02 gave a prediction probability of 68% (p = 0.038). In conclusion, in future prediction models, levels of caspase-3 and HRG might be used as potential markers of embryo quality, and secreted caspase-3 levels could to some extent predict the outcome after transfer of day-2 cultured embryos.

Personalized medicine for in vitro fertilization procedure using modeling and optimal control

In vitro fertilization (IVF) is the most common technique in assisted reproductive technology and in most cases the last resort for infertility treatment. It has four basic stages: superovulation, egg retrieval, fertilization, and embryo transfer. Superovulation is a drug-induced method to enable multiple ovulation per menstrual cycle and key component towards a successful IVF cycle. Although there are the general guidelines for dosage, the dose is not optimized for each patient, and complications, such as overstimulation, can occur. To overcome the shortcomings of this general system, a mathematical procedure is developed which can provide a customized model of this stage regarding the size distribution of eggs (follicles/ oocytes) obtained per cycle as a function of the chemical interactions of the drugs used and the conditions imposed on the patient during the cycle, which provide a basis for predicting the possible outcome. Uncertainty and risk are modeled and included in optimal drug dosage decisions. This paper describes the theory, model, and the optimal control procedure for improving outcomes of IVF treatment for one of the four protocols used in real practice. The validation of the procedure is performed using clinical data from the patients previously undergone IVF cycles. Customized patient-specific model parameters are obtained by using initial two-day data for each patient. Subsequently, this model is used to predict the FSD for the remaining days of the cycle. This procedure was conducted for 49 patients. The results of the customized models are found to be closely matching with the observed FSD. These results thus validate the modeling approach and consequently its use for predicting the customized optimal drug dosage for each patient. Using the customized model and the optimized dosage, the FSD at the end of the cycle was determined. A small double-blind clinical trial was also conducted in India. The results from the trial show that the dosage predicted by using the model is 40% less than the suggestion made by the IVF clinicians. The testing and monitoring requirements for patients using optimized drug dosage is reduced by 72%. Work on the other three protocols and for patients in the USA is started and is showing promising results.

Discarding IVF embryos: reporting on global practices

PURPOSE

To provide a global scale report on a representative sample of the clinical embryology community depicting the practice of discarding supernumerary IVF embryos.

METHODS

A web-based questionnaire titled "Anonymous questionnaire on embryo disposal practices" was designed in order to ensure anonymous participation of practicing clinical embryologists around the world.

RESULTS

During a data collection period of 8 months, 703 filled-in questionnaires from 65 countries were acquired. According to the data acquired, the majority of practitioners, dispose of embryos by placing them directly in a trash can strictly dedicated for embryo disposal for both fresh and frozen cycles (39% and 36.7% respectively). Moreover, 66.4% of practitioners discard the embryos separately-case by case-at different time points during the day. Over half of embryologists (54%) wait until day 6 to discard the surplus embryos, while 65.5% do not implement a specially allocated incubator space as a designated waiting area prior to disposal. The majority of 63.1% reported that this is a witnessed procedure. The vast majority of embryologists (93%) do not employ different protocols for different groups of patients. Nonetheless, 17.8% reported the request to perform a ceremony for these embryos. Assessing the embryologists' perspective, 59.5% of participants stated that the embryology practice would benefit from a universally accepted and practiced protocol.

CONCLUSION(S)

This study uniquely provides insight into global embryo disposal practices and trends. Results highlight the divergence between reported practices, while indicating the significance on standardization of practice, with embryologists acknowledging the need for a universally accepted protocol implementation.