An interpretable machine learning model for individualized gonadotrophin starting dose selection during ovarian stimulation.

RESEARCH QUESTION: Can we develop an interpretable machine learning model that optimizes starting gonadotrophin dose selection in terms of mature oocytes (metaphase II [MII]), fertilized oocytes (2 pronuclear [2PN]) and usable blastocysts? DESIGN: This was a retrospective study of patients undergoing autologous IVF cycles from 2014 to 2020 (n = 18,591) in three assisted reproductive technology centres in the USA. For each patient cycle, an individual dose-response curve was generated from the 100 most similar patients identified using a K-nearest neighbours model. Patients were labelled as dose-responsive if their dose-response curve showed a region that maximized MII oocytes, and flat-responsive otherwise. RESULTS: Analysis of the dose-response curves showed that 30% of cycles were dose-responsive and 64% were flat-responsive. After propensity score matching, patients in the dose-responsive group who received an optimal starting dose of FSH had on average 1.5 more MII oocytes, 1.2 more 2PN embryos and 0.6 more usable blastocysts using 10 IU less of starting FSH and 195 IU less of total FSH compared with patients given non-optimal doses. In the flat-responsive group, patients who received a low starting dose of FSH had on average 0.3 more MII oocytes, 0.3 more 2PN embryos and 0.2 more usable blastocysts using 149 IU less of starting FSH and 1375 IU less of total FSH compared with patients with a high starting dose. CONCLUSIONS: This study demonstrates retrospectively that using a machine learning model for selecting starting FSH can achieve optimal laboratory outcomes while reducing the amount of starting and total FSH used.

Clinical experience with carrier screening in a general population: support for a comprehensive pan-ethnic approach.

PURPOSE: To present results from a large cohort of individuals receiving expanded carrier screening (CS) in the United States. METHODS: Single-gene disorder carrier status for 381,014 individuals was determined using next-generation sequencing (NGS) based CS for up to 274 genes. Detection rates were compared with literature-reported values derived from disease prevalence and carrier frequencies. Combined theoretical affected pregnancy rates for the 274 screened disorders were calculated. RESULTS: For Ashkenazi Jewish (AJ) diseases, 81.6% (4434/5435) of carriers identified did not report AJ ancestry. For cystic fibrosis, 44.0% (6260/14,229) of carriers identified had a variant not on the standard genotyping panel. Individuals at risk of being a silent spinal muscular atrophy carrier, not detectable by standard screening, comprised 1/39 (8763/344,407) individuals. For fragile X syndrome, compared with standard premutation screening, AGG interruption analysis modified risk in 83.2% (1128/1356) premutation carriers. Assuming random pairing across the study population, approximately 1/175 pregnancies would be affected by a disorder in the 274-gene screening panel. CONCLUSION: Compared with standard screening, NGS-based CS provides additional information that may impact reproductive choices. Pan-ethnic CS leads to substantially increased identification of at-risk couples. These data support offering NGS-based CS to all reproductive-aged women.

Correction: Clinical experience with carrier screening in a general population: support for a comprehensive pan-ethnic approach.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

High gonadotropin dosage does not affect euploidy and pregnancy rates in IVF PGS cycles with single embryo transfer.

STUDY QUESTION: Does high gonadotropin dosage affect euploidy and pregnancy rates in PGS cycles with single embryo transfer? SUMMARY ANSWER: High gonadotropin dosage does NOT affect euploidy and pregnancy rates in PGS cycles with single embryo transfer. WHAT IS KNOWN ALREADY: PGS has been proven to be the most effective and reliable method for embryo selection in IVF cycles. Euploidy and blastulation rates decrease significantly with advancing maternal age. In order to recruit an adequate number of follicles, the average dosage of gonadotropins administered during controlled ovarian stimulation in IVF cycles often increases significantly with advancing maternal age. STUDY DESIGN, SIZE, AND DURATION: A retrospective study of SNP (Single Nucleotide Polymorphism) PGS outcome data from blastocysts biopsied on day 5 or day 6 was conducted to identify differences in euploidy and clinical pregnancy rates. Seven hundred and ninety four cycles of IVF treatment with PGS between January 2013 and January 2017 followed by 651 frozen embryo transfers were included in the study (506 patients, maternal age (y.o.) - 37.2 ± 4.31). PARTICIPANTS/MATERIALS, SETTING, METHODS: A total of 4034 embryos were analyzed (5.1 ± 3.76 per case) for euploidy status. All embryos were vitrified after biopsy, and selected embryos were subsequently thawed for a hormone replacement frozen embryo transfer cycle. All cycles were analyzed by total gonadotropin dosage (<3000 IU, 3000-5000 IU and >5000 IU), by number of eggs retrieved (1-5, 5-10, 10-15 and >15 eggs) and patient's age (<35, 35-37, 38-40 and ≥41 y.o.). Clinical pregnancy rate was defined by the presence of a fetal heartbeat at 6-7 weeks of gestation. MAIN RESULTS AND THE ROLE OF CHANCE: Euploidy rates within the same age group were not statistically different regardless of the total dosage of gonadotropins used or the number of eggs retrieved. In the youngest group of patients (<35 y.o. - 187 IVF cycles) euploidy rates ranged from 62.3% (<3000 IU were used in the IVF cycle) to 67.5% (>5000 IU were used in the IVF cycle) (OR = 0.862, 95% CI 0.687-1.082, P = 0.2) and from 69.5% (1-5 eggs retrieved) to 60.0% (>15 eggs retrieved) (OR = 0.658, 95% CI 0.405-1.071, P = 0.09). Similar data were obtained in the oldest group of patients (≥41 y.o. - 189 IVF cycles): euploidy rates ranged from 30.7 to 26.4% (OR = 0.811, 95% CI 0.452-1.454, P = 0.481) when analyzed by total dosage of gonadotropins used in the IVF cycle and from 40.0 to 30.7% (OR = 0.531, 95% CI 0.204-1.384, P = 0.19), when assessed by the total number of eggs retrieved. Ongoing pregnancy rates were similar, not only within particular age groups, but also between different age groups regardless of the total dosage of gonadotropins used: ranging from to 63.6% (<3000 IU, < 35 y.o.) to 54.8% (>5000 IU, ≥41 y.o) (OR = 0.696, 95% CI 0.310-1.565, P = 0.38). LIMITATIONS, REASONS FOR CAUTION: Retrospective study and heterogeneity of patients included. WIDER IMPLICATIONS OF THE FINDINGS: These data are reassuring for the common practice of increasing gonadotropin dosages in PGS cycles, particularly in older woman. STUDY FUNDING/COMPETING INTEREST(S): No formal funding has been received for this study. TRIAL REGISTRATION NUMBER: N/A.

Association between growth dynamics, morphological parameters, the chromosomal status of the blastocysts, and clinical outcomes in IVF PGS cycles with single embryo transfer.

PURPOSE: The purpose of the present study is to examine interconnection between speed of embryo development, the genetic status of the blastocysts, and clinical outcomes in IVF preimplantation genetic screening (PGS) cycles with single embryo transfer (SET). METHODS: The retrospective comparative study has been performed between January 2013 and January 2016. Seven hundred thirty-seven cycles of IVF treatment with PGS, followed by 503 SETs, were included in the study. Normally fertilized oocytes were hatched on day 3, were cultured to the blastocyst stage, and were biopsied only when at least three to seven cells were herniating from zona pellucida on the morning of day 5 (≤118 h) or day 6 (≥139 h). A total of 3705 embryos were analyzed for euploidy rates and blastocyst morphology. All embryos were vitrified after the biopsy, and selected embryos were subsequently thawed for a hormone replacement frozen embryo transfer cycle. RESULTS: The euploidy rate was significantly higher among embryos biopsied on day 5 versus day 6: 59.44 ± 4.1 and 48.19 ± 3.8, respectively, p < 0.05. The difference in euploidy rates between embryos biopsied on day 5 versus day 6 in matched age groups increased from 5.83 to 25.46% with advancing maternal age. Our data demonstrated no statistically significant difference in euploidy rates between good-quality embryos biopsied on day 5 in the group of patients <38 years old and embryos in PGS cycles using donor oocytes: 71.12% (336/472) and 75.68% (221/292), respectively, p = 0.174, χ 2 = 1.848. In 270 out of 503 SETs, transferred embryos were biopsied on day 5 (ongoing pregnancy rate was 64.6% in a group of patients <38 years old, and in a group of patients ≥38 years old, ongoing PR was 64.2%). In 233 out of 503 cycles, transferred embryos were biopsied on day 6 (ongoing PR was 46.6% in a group of patients <38 years old, and in a group of patients ≥38 years old, ongoing PR was 50.8%). In all study groups, the ongoing pregnancy rate was higher when the transferred embryo was available for biopsy on day 5. CONCLUSIONS: Good- and fair-quality embryos available for biopsy on day 5 have higher euploidy rates and have a higher chance to result in an ongoing pregnancy. Euploidy rate has significant variations within the same age group depending on the morphology of the blastocysts.

An interpretable machine learning model for predicting the optimal day of trigger during ovarian stimulation.

OBJECTIVE: To develop an interpretable machine learning model for optimizing the day of trigger in terms of mature oocytes (MII), fertilized oocytes (2PNs), and usable blastocysts. DESIGN: Retrospective study. SETTING: A group of three assisted reproductive technology centers in the United States. PATIENT(S): Patients undergoing autologous in vitro fertilization cycles from 2014 to 2020 (n = 30,278). INTERVENTION(S): None. MAIN OUTCOME MEASURE(S): Average number of MII oocytes, 2PNs, and usable blastocysts. RESULT(S): A set of interpretable machine learning models were developed using linear regression with follicle counts and estradiol levels. When using the model to make day-by-day predictions of trigger or continuing stimulation, possible early and late triggers were identified in 48.7% and 13.8% of cycles, respectively. After propensity score matching, patients with early triggers had on average 2.3 fewer MII oocytes, 1.8 fewer 2PNs, and 1.0 fewer usable blastocysts compared with matched patients with on-time triggers, and patients with late triggers had on average 2.7 fewer MII oocytes, 2.0 fewer 2PNs, and 0.7 fewer usable blastocysts compared with matched patients with on-time triggers. CONCLUSION(S): This study demonstrates that it is possible to develop an interpretable machine learning model for optimizing the day of trigger. Using our model has the potential to improve outcomes for many in vitro fertilization patients.