If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Can mid-infrared attenuated total reflection (MIR ATR) spectroscopy combined with machine learning methods be used as an additional tool to predict embryo quality and IVF treatment outcomes?
Design
Spent culture media was collected and analysed. MIR ATR absorbance spectra were measured using an ALPHA II spectrometer equipped with an attenuated total reflection (ATR) spectrometry accessory. Patient and treatment data and results were collected and analysed in combination with machine learning techniques to identify possible correlations. The main outcome measures were to define the characteristics of absorbance spectra of spent culture media and to distinguish the difference in absorbance between top- and low-quality embryos, day 3 and day 5 embryos and implanting embryos versus non-implanting embryos.
Results
Spent culture media of 227 embryos was collected and analysed. Absorbance peaks in the culture media were different between day 3 and day 5 embryos. Moreover, significant differences in P-values, spanning from 0.014 to 0.044 in absorbance peaks for day 3 embryos and 0.024 up to 0.04 for day 5 embryos, were seen between implanting and non-implanting embryos. Machine learning techniques offered a pregnancy prediction value of 84.6% for day 3 embryos.
Conclusions
MIR ATR may offer an additional parameter for better selection of embryos based on the spectrometric absorbance and secretions of metabolites in the culture media.
As part of the IVF process, fertilized oocytes (zygotes) are incubated in liquid medium for several days before being transferred to the uterus. Various culture media are used in IVF laboratories, differing in terms of their ingredients and concentrations of amino acids, organic acids (lactate, pyruvate, citrate), trace elements, glucose, insulin, oxygen and proteins (
). During embryonic development, various metabolites are secreted into the medium. These metabolites may contain important information regarding embryo quality, which could help determine the best embryo to transfer to achieve pregnancy.
Previous studies on human embryos have demonstrated a correlation between metabolism process rates and embryo quality. Gardner and colleagues reported that nutrient use is correlated with the highest morphological grade in human blastocysts (
Noninvasive metabolomic profiling of embryo culture media using Raman and near-infrared spectroscopy correlates with reproductive potential of embryos in women undergoing in vitro fertilization.
Noninvasive metabolomic profiling of human embryo culture media using a simple spectroscopy adjunct to morphology for embryo assessment in in vitro fertilization (IVF).
). Although these methods seemed promising, clinical applications have been limited. High cost, the need for qualified staff and long intervals from retrieval of samples until results have made these techniques less efficient as an additional tool when selecting embryos for transfer. Several methods have been used to investigate metabolites in biological samples (
Spectroscopy is the study of the absorption and emission of radiation by matter. Different types of spectroscopy are characterized by the wavelengths of radiated energy, such as UV, visible and infrared. Spectroscopy in the infrared spectrum, also known as vibrational spectroscopy, is considered one of the key techniques for characterizing and analysing matter. As a result, these spectra are considered a material's molecular ‘fingerprint’ (
Absorption spectra for strongly-absorbing aqueous solutions can be measured using the attenuated total reflection (ATR) technique in conjunction with mid-infrared (MIR) spectroscopy. The ATR enables the sample to be placed on an ATR optical element, where the infrared beam propagates through it and undergoes several internal reflections. The internally reflected beam penetrates the sample, which contributes to the signal attenuation and corresponds to the sample absorbance (
). In addition, computational methods like principal component analysis–linear discriminant analysis (PCA–LDA) and partial least-squares regression (PLSR) are used to enhance the chemical profile of a specific sample (
demonstrated specific infrared spectral signatures in saliva samples from oral cancer patients compared with healthy individuals. Differences were observed in the MIR ATR spectra of saliva-derived exosomes (
demonstrated that the fingerprint spectral region of malignant tumours was consistently different from that of benign tumours. A simple tissue preparation followed by MIR ATR spectroscopy provides accurate means for very rapid classification of malignant and benign gynaecological tumours (
This study explored the use of MIR ATR spectroscopy, combined with machine learning methods, for evaluating the characteristic spectra of the liquid medium in which fertilized oocytes were incubated, to predict embryo quality and IVF treatment outcomes.
Materials and methods
This prospective observational study was conducted from 1 January 2018 to 31 December 2020 in a single reproductive centre. Spent culture media of embryos in cleavage or blastocyst stage, cultured in a time-lapse incubator, was collected and analysed.
Patient selection, sample collection and sorting
The study was approved by the Ethics Committee of Hillel Yaffe Medical Center (approval no. HYMC-0071-17, 30 August 2021) and was conducted in accordance with the Declaration of Helsinki. The study was registered to the NIH on 23 October 2017 (registration number NCT03317418, https://clinicaltrials.gov/ct2/show/NCT03317418). All patients provided written informed consent to participate in the study.
To reflect the broad range of patients typically encountered in clinical practice, all IVF/intracytoplasmic sperm injection (ICSI) cycles were included in the study. No inclusion/exclusion criteria were applied regarding baseline characteristics. Each patient was included only once in the study.
Treatment protocol and the type and dose of gonadotrophins were prescribed individually according to patient characteristics and clinician preferences and judgement. The initial dose of gonadotrophin for each patient was determined according to age, basal FSH concentrations, antral follicle count, body mass index (BMI) and previous response to ovarian stimulation. All treatments were conducted as previously described (
Assisted reproduction-in vitro fertilization success is improved by ovarian stimulation with exogenous gonadotropins and pituitary suppression with gonadotropin-releasing hormone analogues.
The Ganirelix Dose-Finding Study Group A double-blind, randomized, dose-finding study to assess the efficacy of the gonadotrophin-releasing hormone antagonist ganirelix (Org 37462) to prevent premature luteinizing hormone surges in women undergoing ovarian stimulation with recombinant follicle stimulating hormone (Puregon).
Data collection included baseline parameters (age, BMI, type and cause of infertility, basal FSH), cycle characteristics (duration of stimulation, quantity of gonadotrophins used, and oestradiol concentrations on day of human chorionic gonadotrophin [HCG] administration), and cycle outcomes (number of oocytes retrieved, fertilization and cleavage rates, top-quality embryos, clinical pregnancy rates [per transfer]).
ICSI was performed after oocyte retrieval. After ICSI, the injected oocytes were placed on EmbryoSlides in culture medium (Sage 1-StepTM, CooperSurgical, CT, USA; global®, CooperSurgical; G-TLTM, Vitrolife, Göteborg, Sweden). The medium was covered with 1.4 ml of paraffin oil (OVOILTM, Vitrolife). All embryos were incubated in the EmbryoScope™ (Unisense FertiliTech, Aarhus, Denmark) for up to 5 days at 37°C in 5.8% CO2 and 5% O2. Embryo quality was scored by known implantation data (KID) (
ALPHA Scientists In Reproductive MedicineESHRE Special Interest Group Embryology Istanbul consensus workshop on embryo assessment: proceedings of an expert meeting.
National collection of embryo morphology data into Society for Assisted Reproductive Technology Clinic Outcomes Reporting System: associations among day 3 cell number, fragmentation and blastomere asymmetry, and live birth rate.
). All three scoring systems were used for day 3 embryos. Day 5 embryos were evaluated using the KID scoring system and morphology grading system.
Spent culture media samples were categorized based on the quality of the embryos that had been incubated in them. As different scoring systems were used, parameters were set for judging each embryo as either top or low quality. Achieving a top score in one system was sufficient to place the embryo in the top-quality group (Table 1).
ALPHA Scientists In Reproductive MedicineESHRE Special Interest Group Embryology Istanbul consensus workshop on embryo assessment: proceedings of an expert meeting.
National collection of embryo morphology data into Society for Assisted Reproductive Technology Clinic Outcomes Reporting System: associations among day 3 cell number, fragmentation and blastomere asymmetry, and live birth rate.
A maximum of two good-quality embryos were transferred on day 3 or one on day 5 of embryonic development. Clinical pregnancy was defined as an intrauterine pregnancy with fetal heartbeat confirmed by ultrasound 5 weeks after oocyte retrieval.
The remaining top-quality embryos were vitrified and used in the next embryo transfer if no clinical pregnancy was achieved in the fresh cycle. Low-quality embryos were discarded. Figure 1 illustrates the procedures involved in sample treatment and processing.
Figure 1Step-by-step process of an IVF cycle incorporating spectroscopy. ATR = attenuated total reflectance; FTIR = Fourier transform infrared spectroscopy.
For the analysis the following samples were collected: the spent culture media after each embryo arrived at its final destination (fresh transfer, cryopreservation or discarding), culture media as a control (no embryo), and samples of the culture oil (used to cover the culture media to avoid evaporation). Samples were collected from day 3 or day 5 embryos. The data on cycle outcomes of cryopreserved embryos were collected after the embryos were warmed and transferred.
Each liquid was analysed separately. The sample of the control media served as a baseline for the analysis on the spent culture media.
The oil samples were also analysed, in order to identify the sample contaminated by oil in the process of collection. When such samples were identified, they were excluded from the study.
Immediately after the removal of an embryo to its destination, 20 μl samples were collected using 10–100 μl pipettes and Biopur™ pipette tips (Eppendorf, Hamburg, Germany). All samples were registered, refrigerated at 2–8°C, and stored for spectral analysis in the MIR ATR laboratory during the following 48 h. Spectra were collected using an ALPHA II spectrometer (Bruker, Karlsruhe, Germany) equipped with a deuterated and L-alanine-doped triglycine sulfate detector operating in the spectral range 600–6000 cm–1. The spectra were collected in ATR mode using a diamond crystal.
A 2 μl drop from each sample was pipetted onto the diamond ATR crystal. The samples were slowly dehydrated using a low pressure flow of pure nitrogen gas. Spectra were measured throughout the dehydration process, which was concluded only when the measured spectra stopped changing. The last measured spectrum was saved and used for further analysis.
Measurements were performed at room temperature, immediately after drying the sample (within approximately 2 min), using OPUS software (Bruker, Karlsruhe, Germany). The nominal resolution was 4 cm–1 and the wavenumber ranged from 600 cm–1 to 5000 cm–1. To increase the signal-to-noise ratio, 45 scans were co-added for each measurement. A Savitsky–Golay smoothing algorithm with a window width of 5 points was applied to each spectrum, to reduce random noise in the data. Then, each absorbance spectrum was normalized to the height of the amide I absorbance band at ∼1650 cm–1.
Before measuring each new sample, the ATR crystal was cleaned with ethanol and a new baseline transmission measurement of the crystal was obtained. In addition, a suitable reference absorbance spectrum of the control culture media was measured with each batch of samples. A 2 μl drop of the culture medium fluid (global or G-TL or 1-Step) was pipetted onto the diamond ATR crystal and a thin dry film was obtained by slowly evaporating the water from the medium. The absorption spectrum of the dry film was measured and saved for further analysis. The spectrum of the oil used to reduce evaporation during incubation was also obtained for each batch and saved.
Spectral processing
The infrared absorption spectra were pre-processed to account for culture medium content, oil content, noise and variable drop size. The size of the drops and the surface tension between the crystal and the drop may influence the results. To eliminate these differences, the results were normalized by dividing the measured spectrum by the highest peak (i.e. amide I at ∼1650 cm–1).
The reference absorption spectrum was subtracted from the measured absorption spectrum of each sample to eliminate the contribution of the culture medium and thus retain only the spectral change that occurred during incubation. If pronounced oil absorption bands were visible in the resulting spectra, these spectra were eliminated from further analysis. Each spectrum was truncated to include only the fingerprint spectral region of 600–1700 cm–1.
It is important to emphasize that three different culture media were used in the study, to calculate absorbance per sample. The calculation was performed by the following classic equation:
The samples may be different but appear in the numerator and denominator of the equation and the effect of different culture media is balanced as a constant.
Statistical analysis
Data were analysed using Student's t-test and Pearson's correlation coefficients using commercially available software (Microsoft Excel) and with The Unscrambler (Camo, Oslo, Norway). P-values <0.05 were considered significant. Pearson's correlation and regression analysis were used to demonstrate the correlations between each embryo quality scoring method and prediction of pregnancy.
Machine learning techniques were used to build discrimination models for the absorbance data. Results were correlated with embryo quality and treatment cycle results. Data processing was carried out using MATLAB (Mathworks, Natick, MA, USA) and The Unscrambler (Camo) packages.
The ability to discriminate between positive and negative pregnancy spectra on day 3 was evaluated using multivariate classification analysis. PCA and LDA techniques were used.
Results
Spent culture media of 227 embryos, from 45 patients, was collected and analysed, of which 88 were cleavage-stage embryos and 139 blastocyst-stage embryos. A total of 45 transfers were conducted to 45 patients, of which four cases were of double embryo transfer (DET). Patient characteristics, infertility cause and treatment details are presented in Table 2.
Table 2Patient characteristics, infertility cause and treatment details
Characteristic
Patients (n = 45)
Age (years)
31.3 ± 8.8
BMI (kg/m2)
24.4 ± 4.4
Infertility
Primary
23/45 (51.1)
Secondary
22/45 (48.9)
Cause of infertility
Anovulation
0/45 (0)
Male
15/45 (33.0)
Mechanical
6/45 (13.3)
Unexplained + POR
17/45 (37.8)
Combined
7/45 (15.5)
Basal FSH concentration (IU/l)
6.85 ± 2.8
Average duration of stimulation (days)
10 ± 2.7
Total gonadotrophin used (IU)
2369.9 ± 1442.0
Oestradiol concentrations on day of trigger (pg/ml)
1569.8 ± 820.3
Number of oocytes aspirated
11.9 ± 6.4
Total number of embryos per patient
7.4 ± 4.1
Data are presented as mean ± SD or n (%).
BMI = body mass index; POR = poor ovarian reserve.
All three embryo quality parameters demonstrated very weak correlations with clinical pregnancy rates.
Thirty-seven samples of day 3 spent culture media were eventually analysed, of which eight embryos were transferred in four cycles of DET and 29 were single embryo transfers (SET). Fourteen transfers resulted in a clinical pregnancy (42.4%), all of which were singletons. None of the DET resulted in clinical pregnancy.
On day 3, the correlations between ALPHA/ESHRE, KID score and morphology to clinical pregnancy rates were 0.24, 0.22 and 0.21, respectively.
Twelve samples of day 5 embryos were analysed; four transfers resulted in clinical pregnancy (33.3%), all of which were singletons. On day 5, the correlations between KID score and morphology to clinical pregnancy rates were 0.08 and 0.11, respectively. Moreover, a regression analysis model including all parameters of embryo quality (KID score, ALPHA/ESHRE and morphology), with clinical pregnancy as the dependent value, resulted in a regression coefficient of R = 0.28 for day 3 samples. For day 5 samples, day 5 KID scores were included and the R value was 0.36.
The study also evaluated whether there were any differences between day 3 and day 5 absorbance spectra. The average absorbance spectrum for samples from each day were calculated. Oil spectral lines were removed before averaging. Due to technical difficulties and contamination of some of the samples by oil, among 227 samples collected and analysed, only 69 day 3 samples and 104 day 5 samples were included.
Figure 2 presents the spectral lines of all cleaved embryos (including those that were discarded or vitrified) compared with samples of all blastocysts. A significant difference with P-values in the range of 0.00001 up to 0.0001 in the wavenumber range of 600 cm–1 to 1300 cm–1 and P-values in the range of 0.0001 up to 0.002 in the wavenumber range of 1420 cm–1 to 1600 cm–1, was seen between the two groups according to the ages of the embryos.
Figure 2Mean absorbance spectra of all day 3 and day 5 samples in the fingerprint region (600–1700 cm–1) with P-values in the range of 0.00001 up to 0.0001 in the wavenumber range of 600–1300 cm–1 and P-values in the range of 0.0001 up to 0.002 in the wavenumber range of 1420–1600 cm–1.
Based on these results, the differences between day 3 and day 5 absorbance spectra of positive and negative pregnancy results were evaluated by calculating the mean absorbance spectra of the media of embryos from the two groups. Absorbance peaks differed significantly between the two groups, especially in the amino acid and glucose molecules. The average absorbance spectra in the fingerprint region are illustrated in Figure 3.
Figure 3(A) Comparison of the mean absorbance spectra of media from 33 day 3 embryo transfers (29 SET, four DET) that were positive (all from SET) or negative for clinical pregnancy. Significant P-values are marked with arrows at the following wavenumbers: 624 cm–1, 852 cm–1, 1046 cm–1, 1124 cm–1, 1596 cm–1, 1419 cm–1 with P-values of 0.014 up to 0.044. (B) Mean absorbance spectra of media from 12 day 5 embryo transfers (all SET) and positive versus negative pregnancy outcomes. Significant P-values are marked with arrows at the following wavenumbers: 624 cm–1, 852 cm–1, 934 cm–1, 1046 cm–1, with P-values of 0.024 up to 0.04. DET = double embryo transfer; SET = single embryo transfer.
The wavenumbers were evaluated and associated with different substances and amino acids (Table 3). It was found that the following wavenumbers were significantly different between the groups that achieved clinical pregnancy (fresh or cryopreserved) and those that did not: 624 cm–1, 852 cm–1, 1046 cm–1, 1124 cm–1, 1596 cm–1 and 1419 cm–1 in the cleavage-stage cohort with P-values of 0.014 up to 0.044, and 624 cm–1, 852 cm–1, 934 cm–1 and 1046 cm–1 in the blastocyst cohort with P-values in the range 0.024 to 0.04. These wavenumbers, their associated P-values and the correlated amino acids are shown in Table 3.
Table 3Wavenumber bands and the associated amino acids for absorbance peaks with significant differences between embryo transfers positive or negative for clinical pregnancy
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Petibois, C., Melin, A.M., Perromat, A., Cazorla, G., Déléris, G., 2000. Glucose and lactate concentration determination on single microsamples by Fourier-transform infrared spectroscopy. J. Lab. Clin. Med. 135, 210–215.
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Following these significant results, the aim was to observe differences between the media from day 3 and day 5 absorbance spectra based on the ‘binary quality score’ of top-quality and low-quality embryos, as defined in Table 1. The mean absorbance spectra were processed according to each embryo's age and for each quality score. For day 3, spent media samples from six low-quality embryos and 63 spent media samples from top-quality embryos were examined. For day 5, there were 70 samples from low-quality and 34 samples from top-quality culture media.
Only one significant difference in spectral lines was observed between top- and low-quality scores for day 3 spectra at 651 cm–1 with a P-value of 0.037, and for day 5 at 1285 cm–1 with a P-value of 0.029. Average absorbance spectra of the ‘quality score’ for day 3 and day 5 samples are shown in Figure 4.
Figure 4Mean absorbance spectra of incubation medium fluid from day 3 (A) and day 5 (B) top- and low-quality embryos. A significant P-value (marked with arrow) for day 3 is at 651 cm–1 with a P-value of 0.037, and for day 5 at 1285 cm–1 with a P-value of 0.029.
Correlation between spectral signature and clinical data
The spectra of 37 embryos transferred in 33 transfer cycles (27 fresh embryo transfers and six vitrified–warmed embryo transfers) were evaluated; 14 embryos resulted in clinical pregnancies and 23 embryos did not implant and resulted in negative pregnancy results. Twelve cycles of blastocyst transfers were also evaluated (six fresh embryo transfers and six vitrified–warmed embryo transfers); four cycles resulted in clinical pregnancy and eight cycles in negative pregnancy results.
Using wavenumbers at peak locations at which maximum differences between groups were found, the accuracy for day 3 samples was 84.62%. Figure 5 presents results of the PCA–LDA discrimination model. Due to the small number of positive pregnancy samples on day 5, PCA–LDA analysis was not calculated.
Figure 5Results of the PCA–LDA discrimination model for day 3 samples from 37 embryos in 29 SET and four DET. All clinical pregnancies resulted from SET. Wavenumbers used were: 600–700 cm–1, 800–1300 cm–1, 1400–1420 cm–1 and 1500–1700 cm–1. DET = double embryo transfer; PCA–LDA = principal component analysis–linear discriminant analysis; SET = single embryo transfer.
To enable better prediction of the likelihood of pregnancy, this study explored an alternative method for non-invasive embryo selection in IVF treatments, based on the embryo's culture media. ATR spectroscopy was used to examine the characteristic spectra of the spent culture medium of fertilized oocytes combined with machine learning methods, and a possible correlation with IVF cycle outcomes was evaluated.
Currently, there is no conclusive evidence about the effect of time-lapse technology on pregnancy outcomes in IVF cycles. This system provides additional morphokinetic parameters to evaluate embryos; however, its ability to improve embryo selection or IVF success rates is questionable (
Does time-lapse imaging have favorable results for embryo incubation and selection compared with conventional methods in clinical in vitro fertilization? A meta-analysis and systematic review of randomized controlled trials.
), the current findings demonstrated a weak or no association between conventional morphologic and morphokinetic scoring systems and pregnancy rates.
The metabolism of the developing embryo is reflected in the culture medium by absorption and secretion of amino acids. The synthesis of proteins required for embryonic growth and development is closely related to amino acid metabolism; among them, glutamic acid is formed by amino acid transferase, which further decomposes to produce α-ketoglutaric acid, which forms adenosine triphosphate under the action of a citric acid cycle. In addition, glutamate provides a secondary source of carbon and nitrogen for the re-synthesis of pyrimidines and hydrazines, and acts as a reducing agent to protect cells from oxidative stress. Arginine forms nitric oxide through the catalytic action of nitric oxide synthase and is involved in the signal transduction pathway of embryos, which is an essential metabolic pathway for embryonic growth and development (
Non-invasive amino acid profiling of embryo culture medium using HPLC correlates with embryo implantation potential in women undergoing in vitro fertilization.
found a correlation between amino acid fingerprint in spent culture media and embryo implantation potential. Other reports demonstrated correlations between different amino acid concentrations in spent culture media and the chances of transferred embryos for implantation (
Noninvasive metabolomic profiling of embryo culture media using proton nuclear magnetic resonance correlates with reproductive potential of embryos in women undergoing in vitro fertilization.
analysed bovine embryo culture medium and recipient blood plasma using MIR metabolomics to predict pregnancy outcomes. This method identified embryos and recipients with improved pregnancy viability.
The current results, using MIR spectral analysis of spent culture media from embryos, demonstrated a remarkable difference in absorbance between day 3 and day 5 embryos, implying the media had different metabolic profiles. Some of the pronounced differences in the heights of peaks in Figure 2 are associated with known molecular groups: the glucose band at 1124 cm–1, alanine at 1419 cm–1 and arginine at 1596 cm–1. These differences could reflect variances in consumption and secretion of different substances in day 3 compared with day 5 embryos. The growing embryo may secrete substances that contribute to the increased absorbance at these wavelengths.
Subsequently, differences were observed between MIR ATR spectra of the medium fluid between day 3 and day 5 embryos according to positive and negative pregnancy outcomes. These findings indicated a difference in metabolic processes between embryos that implanted and those that did not.
Figure 3 shows that the relative absorbance intensity of negative pregnancies is higher than in positive pregnancies, which indicates lower amino acid consumption, in agreement with other publications (
Using MIR spectral analysis, top-quality embryos could be distinguished from low-quality embryos only by significant differences in peaks at 651 cm–1 for day 3 embryos and at 1285 cm–1 for day 5 embryos.
Using a PCA–LDA discrimination model of the data collected, the current method resulted in 84.6% accuracy in predicting implantation of day 3 embryos. More importantly, based on spectral footprinting, it was possible to distinguish between embryos that had positive pregnancy results regardless of the spectral lines of embryo quality.
The current results suggest that MIR ATR spectrometry of spent culture media can offer an additional method to help embryologists and physicians choose the embryo carrying the best chances of achieving pregnancy. This tool may offer relatively better accuracy compared with the inconsistency of the morphokinetic evaluation tool.
The strengths of this study lie in offering a simple, available, non-invasive tool that can assist embryologists and clinicians to choose the embryo with the highest chances of implantation. Limitations include the small number of samples available for analysis because several samples were contaminated by oil. Additional larger studies are required to validate this method and to establish its role in clinical practice.
This is a pioneering study in the field of MIR and non-invasive embryo evaluation, which may offer an option for better selection of embryos based on the spectrometric absorbance and secretions of metabolites in the culture media.
Data availability
Data will be made available on request.
Acknowledgements
This paper and the research behind it would not have been possible without the exceptional work and effort of the embryology staff.
Author contributions
All authors contributed substantially to this work. The authors collectively developed the original concept of this study. Nardin Aslih, Ben Zion Dekel, Dov Malonek and Einat Shalom-Paz analysed and interpreted the data, and wrote and revised the paper. Medeai Michaeli and Diana Paltov collected the samples and performed the laboratory work. All authors contributed to critical discussion and reviewed and approved the final version of the manuscript for submission and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy of integrity of any part of the work are appropriately investigated and resolved.
References
Albano C.
Smitz J.
Camus M.
Riethmüller-Winzen H.
Van Steirteghem A.
Devroey P.
Comparison of different doses of gonadotropin-releasing hormone antagonist Cetrorelix during controlled ovarian hyperstimulation.
Assisted reproduction-in vitro fertilization success is improved by ovarian stimulation with exogenous gonadotropins and pituitary suppression with gonadotropin-releasing hormone analogues.
Does time-lapse imaging have favorable results for embryo incubation and selection compared with conventional methods in clinical in vitro fertilization? A meta-analysis and systematic review of randomized controlled trials.
Non-invasive amino acid profiling of embryo culture medium using HPLC correlates with embryo implantation potential in women undergoing in vitro fertilization.
Petibois, C., Melin, A.M., Perromat, A., Cazorla, G., Déléris, G., 2000. Glucose and lactate concentration determination on single microsamples by Fourier-transform infrared spectroscopy. J. Lab. Clin. Med. 135, 210–215.
National collection of embryo morphology data into Society for Assisted Reproductive Technology Clinic Outcomes Reporting System: associations among day 3 cell number, fragmentation and blastomere asymmetry, and live birth rate.
Noninvasive metabolomic profiling of embryo culture media using proton nuclear magnetic resonance correlates with reproductive potential of embryos in women undergoing in vitro fertilization.
Noninvasive metabolomic profiling of embryo culture media using Raman and near-infrared spectroscopy correlates with reproductive potential of embryos in women undergoing in vitro fertilization.
A double-blind, randomized, dose-finding study to assess the efficacy of the gonadotrophin-releasing hormone antagonist ganirelix (Org 37462) to prevent premature luteinizing hormone surges in women undergoing ovarian stimulation with recombinant follicle stimulating hormone (Puregon).
Wolpert, M., Hellwig, P., 2006. Infrared spectra and molar absorption coefficients of the 20 alpha amino acids in aqueous solutions in the spectral range from 1800 to 500cm-1.
Noninvasive metabolomic profiling of human embryo culture media using a simple spectroscopy adjunct to morphology for embryo assessment in in vitro fertilization (IVF).
Dr Nardin Aslih MD is a senior doctor at the IVF Unit at Hillel Yaffe Medical Center in Hadera, Israel. She graduated from the Technion Institute of Israel in 2006, finished her internship in the Obstetric and Gynecology Department in 2013 and since then has been a senior physician at the IVF Unit and has taken an active role in academic activities.
Key message
This study explored the use of Fourier-transform infrared spectroscopy combined with machine learning methods as an additional way to predict embryo quality and IVF treatment outcomes. The results demonstrated that this technique may offer an additional parameter for better selection of embryos based on the spectrometric characteristics of culture media.
Article info
Publication history
Published online: December 20, 2022
Accepted:
December 13,
2022
Received in revised form:
November 27,
2022
Received:
July 25,
2022
Declaration: The authors report no financial or commercial conflicts of interest.
00862-8&id=gr1.jpg){kind=link}
00862-8&id=gr2.jpg){kind=link}
00862-8&id=gr3.jpg){kind=link}
00862-8&id=gr4.jpg){kind=link}
00862-8&id=gr5.jpg){kind=link}