BlastAssist: a deep learning pipeline to measure interpretable features of human embryos.

STUDY QUESTION: Can the BlastAssist deep learning pipeline perform comparably to or outperform human experts and embryologists at measuring interpretable, clinically relevant features of human embryos in IVF? SUMMARY ANSWER: The BlastAssist pipeline can measure a comprehensive set of interpretable features of human embryos and either outperform or perform comparably to embryologists and human experts in measuring these features. WHAT IS KNOWN ALREADY: Some studies have applied deep learning and developed 'black-box' algorithms to predict embryo viability directly from microscope images and videos but these lack interpretability and generalizability. Other studies have developed deep learning networks to measure individual features of embryos but fail to conduct careful comparisons to embryologists' performance, which are fundamental to demonstrate the network's effectiveness. STUDY DESIGN, SIZE, DURATION: We applied the BlastAssist pipeline to 67 043 973 images (32 939 embryos) recorded in the IVF lab from 2012 to 2017 in Tel Aviv Sourasky Medical Center. We first compared the pipeline measurements of individual images/embryos to manual measurements by human experts for sets of features, including: (i) fertilization status (n = 207 embryos), (ii) cell symmetry (n = 109 embryos), (iii) degree of fragmentation (n = 6664 images), and (iv) developmental timing (n = 21 036 images). We then conducted detailed comparisons between pipeline outputs and annotations made by embryologists during routine treatments for features, including: (i) fertilization status (n = 18 922 embryos), (ii) pronuclei (PN) fade time (n = 13 781 embryos), (iii) degree of fragmentation on Day 2 (n = 11 582 embryos), and (iv) time of blastulation (n = 3266 embryos). In addition, we compared the pipeline outputs to the implantation results of 723 single embryo transfer (SET) cycles, and to the live birth results of 3421 embryos transferred in 1801 cycles. PARTICIPANTS/MATERIALS, SETTING, METHODS: In addition to EmbryoScope™ image data, manual embryo grading and annotations, and electronic health record (EHR) data on treatment outcomes were also included. We integrated the deep learning networks we developed for individual features to construct the BlastAssist pipeline. Pearson's χ2 test was used to evaluate the statistical independence of individual features and implantation success. Bayesian statistics was used to evaluate the association of the probability of an embryo resulting in live birth to BlastAssist inputs. MAIN RESULTS AND THE ROLE OF CHANCE: The BlastAssist pipeline integrates five deep learning networks and measures comprehensive, interpretable, and quantitative features in clinical IVF. The pipeline performs similarly or better than manual measurements. For fertilization status, the network performs with very good parameters of specificity and sensitivity (area under the receiver operating characteristics (AUROC) 0.84-0.94). For symmetry score, the pipeline performs comparably to the human expert at both 2-cell (r = 0.71 ± 0.06) and 4-cell stages (r = 0.77 ± 0.07). For degree of fragmentation, the pipeline (acc = 69.4%) slightly under-performs compared to human experts (acc = 73.8%). For developmental timing, the pipeline (acc = 90.0%) performs similarly to human experts (acc = 91.4%). There is also strong agreement between pipeline outputs and annotations made by embryologists during routine treatments. For fertilization status, the pipeline and embryologists strongly agree (acc = 79.6%), and there is strong correlation between the two measurements (r = 0.683). For degree of fragmentation, the pipeline and embryologists mostly agree (acc = 55.4%), and there is also strong correlation between the two measurements (r = 0.648). For both PN fade time (r = 0.787) and time of blastulation (r = 0.887), there's strong correlation between the pipeline and embryologists. For SET cycles, 2-cell time (P < 0.01) and 2-cell symmetry (P < 0.03) are significantly correlated with implantation success rate, while other features showed correlations with implantation success without statistical significance. In addition, 2-cell time (P < 5 × 10-11), PN fade time (P < 5 × 10-10), degree of fragmentation on Day 3 (P < 5 × 10-4), and 2-cell symmetry (P < 5 × 10-3) showed statistically significant correlation with the probability of the transferred embryo resulting in live birth. LIMITATIONS, REASONS FOR CAUTION: We have not tested the BlastAssist pipeline on data from other clinics or other time-lapse microscopy (TLM) systems. The association study we conducted with live birth results do not take into account confounding variables, which will be necessary to construct an embryo selection algorithm. Randomized controlled trials (RCT) will be necessary to determine whether the pipeline can improve success rates in clinical IVF. WIDER IMPLICATIONS OF THE FINDINGS: BlastAssist provides a comprehensive and holistic means of evaluating human embryos. Instead of using a black-box algorithm, BlastAssist outputs meaningful measurements of embryos that can be interpreted and corroborated by embryologists, which is crucial in clinical decision making. Furthermore, the unprecedentedly large dataset generated by BlastAssist measurements can be used as a powerful resource for further research in human embryology and IVF. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by Harvard Quantitative Biology Initiative, the NSF-Simons Center for Mathematical and Statistical Analysis of Biology at Harvard (award number 1764269), the National Institute of Heath (award number R01HD104969), the Perelson Fund, and the Sagol fund for embryos and stem cells as part of the Sagol Network. The authors declare no competing interests. TRIAL REGISTRATION NUMBER: Not applicable.

Noninvasive metabolic profiling of cumulus cells, oocytes, and embryos via fluorescence lifetime imaging microscopy: a mini-review.

A major challenge in ART is to select high-quality oocytes and embryos. The metabolism of oocytes and embryos has long been linked to their viability, suggesting the potential utility of metabolic measurements to aid in selection. Here, we review recent work on noninvasive metabolic imaging of cumulus cells, oocytes, and embryos. We focus our discussion on fluorescence lifetime imaging microscopy (FLIM) of the autofluorescent coenzymes NAD(P)H and flavine adenine dinucleotide (FAD+), which play central roles in many metabolic pathways. FLIM measurements provide quantitative information on NAD(P)H and FAD+ concentrations and engagement with enzymes, leading to a robust means of characterizing the metabolic state of cells. We argue that FLIM is a promising approach to aid in oocyte and embryo selection.

Fluorescence lifetime imaging microscopy (FLIM) detects differences in metabolic signatures between euploid and aneuploid human blastocysts.

STUDY QUESTION: Can non-invasive imaging with fluorescence lifetime imaging microscopy (FLIM) detect metabolic differences in euploid versus aneuploid human blastocysts? SUMMARY ANSWER: FLIM has identified significant metabolic differences between euploid and aneuploid blastocysts. WHAT IS KNOWN ALREADY: Prior studies have demonstrated that FLIM can detect metabolic differences in mouse oocytes and embryos and in discarded human blastocysts. STUDY DESIGN, SIZE, DURATION: This was a prospective observational study from August 2019 to February 2020. Embryo metabolic state was assessed using FLIM to measure the autofluorescence metabolic factors nicotinamide adenine dinucleotide dehydrogenase together with nicotinamide adenine phosphate dinucleotide dehydrogenase (NAD(P)H) and flavin adenine dinucleotide (FAD). Eight metabolic FLIM parameters were obtained from each blastocyst (four for NAD(P)H and four for FAD): short (T1) and long (T2) fluorescence lifetime, fluorescence intensity (I) and fraction of the molecules engaged with enzymes (F). The redox ratio (NAD(P)H-I)/(FAD-I) was also calculated for each image. PARTICIPANTS/MATERIALS, SETTING, METHODS: This study was performed at a single academically affiliated centre where there were 156 discarded frozen blastocysts (n = 17 euploids; 139 aneuploids) included. Ploidy status was determined by pre-implantation genetic testing for aneuploidy (PGT-A). Discarded human blastocysts were compared using single FLIM parameters. Additionally, inner cell mass (ICM) and trophectoderm (TE) were also evaluated. Multilevel models were used for analysis. A post-hoc correction used Benjamini-Hochberg's false discovery rate, at a q-value of 0.05. MAIN RESULTS AND THE ROLE OF CHANCE: Comparing euploid (n = 17) versus aneuploid (n = 139) embryos, a significant difference was seen in NAD(P)H-F (P < 0.04), FAD-I (P < 0.04) and redox ratio (P < 0.05). Euploid ICM (n = 15) versus aneuploid ICM (n = 119) also demonstrated significantly different signatures in NAD(P)H-F (P < 0.009), FAD-I (P < 0.03) and redox ratio (P < 0.03). Similarly, euploid TE (n = 15) versus aneuploid TE (n = 119) had significant differences in NAD(P)H-F (P < 0.0001) and FAD-I (P < 0.04). LIMITATIONS, REASONS FOR CAUTION: This study utilized discarded human blastocysts, and these embryos may differ metabolically from non-discarded human embryos. The blastocysts analysed were vitrified after PGT-A biopsy and it is unclear how the vitrification process may affect the metabolic profile of blastocysts. Our study was also limited by the small number of rare donated euploid embryos available for analysis. Euploid embryos are very rarely discarded due to their value to patients trying to conceive, which limits their use for research purposes. However, we controlled for the imbalance with the bootstrap resampling analysis. WIDER IMPLICATIONS OF THE FINDINGS: These findings provide preliminary evidence that FLIM may be a useful non-invasive clinical tool to assist in identifying the ploidy status of embryos. STUDY FUNDING/COMPETING INTEREST(S): The study was supported by the Blavatnik Biomedical Accelerator Grant at Harvard University. Becker and Hickl GmbH and Boston Electronics sponsored research with the loaning of equipment for FLIM. D.J.N. is an inventor on patent US20170039415A1. There are no other conflicts of interest to declare. TRIAL REGISTRATION NUMBER: N/A.

Metabolic state of human blastocysts measured by fluorescence lifetime imaging microscopy.

STUDY QUESTION: Can non-invasive metabolic imaging via fluorescence lifetime imaging microscopy (FLIM) detect variations in metabolic profiles between discarded human blastocysts? SUMMARY ANSWER: FLIM revealed extensive variations in the metabolic state of discarded human blastocysts associated with blastocyst development over 36 h, the day after fertilization and blastocyst developmental stage, as well as metabolic heterogeneity within individual blastocysts. WHAT IS KNOWN ALREADY: Mammalian embryos undergo large changes in metabolism over the course of preimplantation development. Embryo metabolism has long been linked to embryo viability, suggesting its potential utility in ART to aid in selecting high quality embryos. However, the metabolism of human embryos remains poorly characterized due to a lack of non-invasive methods to measure their metabolic state. STUDY DESIGN, SIZE, DURATION: We conducted a prospective observational study. We used 215 morphologically normal human embryos from 137 patients that were discarded and donated for research under an approved institutional review board protocol. These embryos were imaged using metabolic imaging via FLIM to measure the autofluorescence of two central coenzymes, nicotinamide adenine (phosphate) dinucleotide (NAD(P)H) and flavine adenine dinucleotide (FAD+), which are essential for cellular respiration and glycolysis. PARTICIPANTS/MATERIALS, SETTING, METHODS: Here, we used non-invasive FLIM to measure the metabolic state of human blastocysts. We first studied spatial patterns in the metabolic state within human blastocysts and the association of the metabolic state of the whole blastocysts with stage of expansion, day of development since fertilization and morphology. We explored the sensitivity of this technique in detecting metabolic variations between blastocysts from the same patient and between patients. Next, we explored whether FLIM can quantitatively measure metabolic changes through human blastocyst expansion and hatching via time-lapse imaging. For all test conditions, the level of significance was set at P < 0.05 after correction for multiple comparisons using Benjamini-Hochberg's false discovery rate. MAIN RESULTS AND THE ROLE OF CHANCE: We found that FLIM is sensitive enough to detect significant metabolic differences between blastocysts. We found that metabolic variations between blastocyst are partially explained by both the time since fertilization and their developmental expansion stage (P < 0.05), but not their morphological grade. Substantial metabolic variations between blastocysts from the same patients remain, even after controlling for these factors. We also observe significant metabolic heterogeneity within individual blastocysts, including between the inner cell mass and the trophectoderm, and between the portions of hatching blastocysts within and without the zona pellucida (P < 0.05). And finally, we observed that the metabolic state of human blastocysts continuously varies over time. LIMITATIONS, REASONS FOR CAUTION: Although we observed significant variations in metabolic parameters, our data are taken from human blastocysts that were discarded and donated for research and we do not know their clinical outcome. Moreover, the embryos used in this study are a mixture of aneuploid, euploid and embryos of unknown ploidy. WIDER IMPLICATIONS OF THE FINDINGS: This work reveals novel aspects of the metabolism of human blastocysts and suggests that FLIM is a promising approach to assess embryo viability through non-invasive, quantitative measurements of their metabolism. These results further demonstrate that FLIM can provide biologically relevant information that may be valuable for the assessment of embryo quality. STUDY FUNDING/COMPETING INTEREST(S): Supported by the Blavatnik Biomedical Accelerator Grant at Harvard University. Becker and Hickl GmbH and Boston Electronics sponsored research with the loaning of equipment for FLIM. D.J.N. is an inventor on patent US20170039415A1. TRIAL REGISTRATION NUMBER: N/A.

Extensive analysis of mitochondrial DNA quantity and sequence variation in human cumulus cells and assisted reproduction outcomes.

STUDY QUESTION: Are relative mitochondrial DNA (mtDNA) content and mitochondrial genome (mtGenome) variants in human cumulus cells (CCs) associated with oocyte reproductive potential and assisted reproductive technology (ART) outcomes? SUMMARY ANSWER: Neither the CC mtDNA quantity nor the presence of specific mtDNA genetic variants was associated with ART outcomes, although associations with patient body mass index (BMI) were detected, and the total number of oocytes retrieved differed between major mitochondrial haplogroups. WHAT IS KNOWN ALREADY: CCs fulfil a vital role in the support of oocyte developmental competence. As with other cell types, appropriate cellular function is likely to rely upon adequate energy production, which in turn depends on the quantity and genetic competence of the mitochondria. mtDNA mutations can be inherited or they can accumulate in somatic cells over time, potentially contributing to aging. Such mutations may be homoplasmic (affecting all mtDNA in a cell) or they may display varying levels of heteroplasmy (affecting a proportion of the mtDNA). Currently, little is known concerning variation in CC mitochondrial genetics and how this might influence the reproductive potential of the associated oocyte. STUDY DESIGN, SIZE, DURATION: This was a prospective observational study involving human CCs collected with 541 oocytes from 177 IVF patients. mtDNA quantity was measured in all the samples with a validated quantitative PCR method and the entire mtGenome was sequenced in a subset of 138 samples using a high-depth massively parallel sequencing approach. Associations between relative mtDNA quantity and mtGenome variants in CCs and patient age, BMI (kg/m2), infertility diagnosis and ART outcomes were investigated. PARTICIPANTS/MATERIALS, SETTING, METHODS: Massively parallel sequencing permitted not only the accurate detection of mutations but also the precise quantification of levels of mutations in cases of heteroplasmy. Sequence variants in the mtDNA were evaluated using Mitomaster and HmtVar to predict their potential impact. MAIN RESULTS AND THE ROLE OF CHANCE: The relative mtDNA CC content was significantly associated with BMI. No significant associations were observed between CC mtDNA quantity and patient age, female infertility diagnosis or any ART outcome variable. mtGenome sequencing revealed 4181 genetic variants with respect to a reference genome. The COXI locus contained the least number of coding sequence variants, whereas ATPase8 had the most. The number of variants predicted to affect the ATP production differed significantly between mitochondrial macrohaplogroups. The total number of retrieved oocytes was different between the H-V and J-T as well as the U-K and J-T macrohaplogroups. There was a non-significant increase in mtDNA levels in CCs with heteroplasmic mitochondrial mutations. LARGE SCALE DATA: N/A. LIMITATIONS, REASONS FOR CAUTION: Although a large number of samples were analysed in this study, it was not possible to analyse all the CCs from every patient. Also, the results obtained with respect to specific clinical outcomes and macrohaplogroups should be interpreted with caution due to the smaller sample sizes when subdividing the dataset. WIDER IMPLICATIONS OF THE FINDINGS: These findings suggest that the analysis of mtDNA in CCs is unlikely to provide an advantage in terms of improved embryo selection during assisted reproduction cycles. Nonetheless, our data raise interesting biological questions, particularly regarding the interplay of metabolism and BMI and the association of mtDNA haplogroup with oocyte yield in ovarian stimulation cycles. STUDY FUNDING/COMPETING INTEREST(S): This study was funded by National Institutes of Health grant 5R01HD092550-02. D.J.N. and C.R. co-hold patent US20150346100A1 and D.J.N. holds US20170039415A1, both for metabolic imaging methods. D.W. receives support from the NIHR Oxford Biomedical Research Centre. The remaining authors have no conflicts of interest to declare.

Metabolic imaging of human cumulus cells reveals associations among metabolic profiles of cumulus cells, patient clinical factors, and oocyte maturity.

OBJECTIVE: To determine whether fluorescence lifetime imaging microscopy (FLIM) detects differences in metabolic state among cumulus cell samples and whether their metabolic state is associated with patient age, body mass index (BMI), and antimüllerian hormone (AMH) level and maturity of the oocyte. DESIGN: Prospective observational study. SETTING: Academic laboratory. PATIENT(S): Cumulus cell (CC) clusters from cumulus-oocyte complexes were collected from patients undergoing assisted reproductive technology treatment after oocyte retrieval and vitrified. INTERVENTION(S): Cumulus cell metabolism was assessed using FLIM to measure autofluorescence of nicotinamide adenine (phosphate) dinucleotide and flavine adenine dinucleotide, endogenous coenzymes essential for cellular respiration and glycolysis. Patient age, BMI, and AMH level and the maturity of the corresponding oocytes were recorded. MAIN OUTCOME MEASURE(S): Quantitative information from FLIM was obtained regarding metabolite concentrations from fluorescence intensity and metabolite enzyme engagement from fluorescence lifetimes. Associations were investigated between each FLIM parameter and oocyte maturity and patient age, BMI, and AMH. Variance between CC clusters within and between patients was determined. RESULT(S): Of 619 CC clusters from 193 patients, 90 were associated with immature oocytes and 505 with metaphase II oocytes. FLIM enabled quantitative measurements of the metabolic state of CC clusters. These parameters were significantly correlated with patient age and AMH independently, but not with BMI. Cumulus cell nicotinamide adenine (phosphate) dinucleotide FLIM parameters and redox ratio were significantly associated with maturity of the enclosed oocyte. CONCLUSION(S): FLIM detects variations in the metabolic state of CCs, showing a greater variance among clusters from each patient than between patients. Fluorescence lifetime imaging microscopy can detect CC metabolic associations with patient age and AMH and variations between mature and immature oocytes, suggesting the potential utility of this technique to help identify superior oocytes.

Combined noninvasive metabolic and spindle imaging as potential tools for embryo and oocyte assessment.

STUDY QUESTION: Is the combined use of fluorescence lifetime imaging microscopy (FLIM)-based metabolic imaging and second harmonic generation (SHG) spindle imaging a feasible and safe approach for noninvasive embryo assessment? SUMMARY ANSWER: Metabolic imaging can sensitively detect meaningful metabolic changes in embryos, SHG produces high-quality images of spindles and the methods do not significantly impair embryo viability. WHAT IS KNOWN ALREADY: Proper metabolism is essential for embryo viability. Metabolic imaging is a well-tested method for measuring metabolism of cells and tissues, but it is unclear if it is sensitive enough and safe enough for use in embryo assessment. STUDY DESIGN, SIZE, DURATION: This study consisted of time-course experiments and control versus treatment experiments. We monitored the metabolism of 25 mouse oocytes with a noninvasive metabolic imaging system while exposing them to oxamate (cytoplasmic lactate dehydrogenase inhibitor) and rotenone (mitochondrial oxidative phosphorylation inhibitor) in series. Mouse embryos (n = 39) were measured every 2 h from the one-cell stage to blastocyst in order to characterize metabolic changes occurring during pre-implantation development. To assess the safety of FLIM illumination, n = 144 illuminated embryos were implanted into n = 12 mice, and n = 108 nonilluminated embryos were implanted into n = 9 mice. PARTICIPANTS/MATERIALS, SETTING, METHODS: Experiments were performed in mouse embryos and oocytes. Samples were monitored with noninvasive, FLIM-based metabolic imaging of nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD) autofluorescence. Between NADH cytoplasm, NADH mitochondria and FAD mitochondria, a single metabolic measurement produces up to 12 quantitative parameters for characterizing the metabolic state of an embryo. For safety experiments, live birth rates and pup weights (mean ± SEM) were used as endpoints. For all test conditions, the level of significance was set at P < 0.05. MAIN RESULTS AND THE ROLE OF CHANCE: Measured FLIM parameters were highly sensitive to metabolic changes due to both metabolic perturbations and embryo development. For oocytes, metabolic parameter values were compared before and after exposure to oxamate and rotenone. The metabolic measurements provided a basis for complete separation of the data sets. For embryos, metabolic parameter values were compared between the first division and morula stages, morula and blastocyst and first division and blastocyst. The metabolic measurements again completely separated the data sets. Exposure of embryos to excessive illumination dosages (24 measurements) had no significant effect on live birth rate (5.1 ± 0.94 pups/mouse for illuminated group; 5.7 ± 1.74 pups/mouse for control group) or pup weights (1.88 ± 0.10 g for illuminated group; 1.89 ± 0.11 g for control group). LIMITATIONS, REASONS FOR CAUTION: The study was performed using a mouse model, so conclusions concerning sensitivity and safety may not generalize to human embryos. A limitation of the live birth data is also that although cages were routinely monitored, we could not preclude that some runt pups may have been eaten. WIDER IMPLICATIONS OF THE FINDINGS: Promising proof-of-concept results demonstrate that FLIM with SHG provide detailed biological information that may be valuable for the assessment of embryo and oocyte quality. Live birth experiments support the method's safety, arguing for further studies of the clinical utility of these techniques. STUDY FUNDING/COMPETING INTEREST(S): Supported by the Blavatnik Biomedical Accelerator Grant at Harvard University and by the Harvard Catalyst/The Harvard Clinical and Translational Science Center (National Institutes of Health Award UL1 TR001102), by NSF grants DMR-0820484 and PFI-TT-1827309 and by NIH grant R01HD092550-01. T.S. was supported by a National Science Foundation Postdoctoral Research Fellowship in Biology grant (1308878). S.F. and S.A. were supported by NSF MRSEC DMR-1420382. Becker and Hickl GmbH sponsored the research with the loaning of equipment for FLIM. T.S. and D.N. are cofounders and shareholders of LuminOva, Inc., and co-hold patents (US20150346100A1 and US20170039415A1) for metabolic imaging methods. D.S. is on the scientific advisory board for Cooper Surgical and has stock options with LuminOva, Inc.

Interstitial 13q14 deletions detected in the karyotype and translocations with concomitant deletion at 13q14 in chronic lymphocytic leukemia: different genetic mechanisms but equivalent poorer clinical outcome.

Deletion of 13q14 as the sole abnormality is a good prognostic marker in chronic lymphocytic leukemia (CLL). Nonetheless, the prognostic value of reciprocal 13q14 translocations [t(13q)] with related 13q losses has not been fully elucidated. We described clinical and biological characteristics of 25 CLL patients with t(13q), and compared with 62 patients carrying interstitial del(13q) by conventional G-banding cytogenetics (CGC) [i-del(13q)] and 295 patients with del(13q) only detected by fluorescence in situ hybridization (FISH) [F-del(13q)]. Besides from the CLL FISH panel (D13S319, CEP12, ATM, TP53), we studied RB1 deletions in all t(13q) cases and a representative group of i-del(13q) and F-del(13q). We analyzed NOTCH1, SF3B1, and MYD88 mutations in t(13q) cases by Sanger sequencing. In all, 25 distinct t(13q) were described. All these cases showed D13S319 deletion while 32% also lost RB1. The median percentage of 13q-deleted nuclei did not differ from i-del(13q) patients (73% vs. 64%), but both were significantly higher than F-del(13q) (52%, P < 0.001). Moreover, t(13q) patients showed an increased incidence of biallelic del(13q) (52% vs. 11.3% and 14.9%, P < 0.001) and higher rates of concomitant 17p deletion (37.5% vs. 8.6% and 7.2%, P < 0.001). RB1 involvement was significantly higher in the i-del(13q) group (79%, P < 0.001). Two t(13q) patients (11.8%) carried NOTCH1 mutations. Time to first treatment in t(13q) and i-del(13q) was shorter than F-del(13q) (67, 44, and 137 months, P = 0.029), and preserved significance in the multivariate analysis. In conclusion, t(13q) and del(13q) patients detected by CGC constitute a subgroup within the 13q-deleted CLL patients associated with a worse clinical outcome.