RESUMO
OBJECTIVE: To use machine learning methods to develop prediction models of pregnancy complications in women who conceived with assisted reproductive techniques (ART). DESIGN: A nation-wide register-based cohort study with prospectively collected data. SETTING: Swedish national registers and nationwide quality IVF register. PATIENT(S): all nulliparous women who achieved birth within the first 3 ART treatment cycles between 2008 and 2016 in Sweden. INTERVENTION(S): Characteristics before the use of ART, such as demographics and medical history, were considered potential predictors in the development of before treatment prediction models. ART treatment details were further included in after treatment prediction models. MAIN OUTCOME MEASURE(S): Potential diagnoses of preeclampsia, placental complications (previa, accreta, and abruption), and postpartum hemorrhage were identified using the International Classification of Diseases recorded in the Swedish Medical Birth and Patient registers, respectively. Multiple prediction model algorithms were performed and compared for each outcome and treatment cycle, including logistic regression, decision tree model, naïve Bayes classification, support vector machine, random forest, and gradient boosting. The performance of each model was assessed with C statistic, and nested cross-validation was used to aid model selection and hyperparameter tuning. RESULT(S): A total of 14,732 women gave birth after the first (N = 7,302), second (N = 4,688), or third (N = 2,742) ART cycle, representing birth rates of 24.1%, 23.8%, and 22.0%. Overall prediction performance did not vary much across the different methods used. In the first cycle, the before treatment prediction performance was at best 66%, 66%, and 60% for preeclampsia, placental complications, and postpartum hemorrhage, respectively. Inclusion of after treatment characteristics conferred slight improvement (approximately 1%-5%), as did prediction in later cycles (approximately 1%-5%). The top influential and consistent predictors included age, region of residence, infertility diagnosis, and type of embryo transfer (fresh or frozen) in the later (2nd and 3rd) cycles. Body mass index was a top predictor of preeclampsia and was also influential for placental complications but not for postpartum hemorrhage. CONCLUSION(S): The combined use of demographics, medical history, and ART treatment information was not enough to confidently predict serious pregnancy complications in women who conceived with ART. Future studies are needed to assess if additional longitudinal follow-up during pregnancy can improve the prediction to allow clinical protocol development.
RESUMO
STUDY QUESTION: Can use of a commercially available time-lapse algorithm for Day 5 blastocyst selection improve pregnancy rates compared with morphology alone? SUMMARY ANSWER: The use of a time-lapse selection model to choose blastocysts for fresh single embryo transfer on Day 5 did not improve ongoing pregnancy rate compared to morphology alone. WHAT IS KNOWN ALREADY: Evidence from time-lapse monitoring suggests correlations between timing of key developmental events and embryo viability. No good quality evidence exists to support improved pregnancy rates following time-lapse selection. STUDY DESIGN, SIZE, DURATION: A prospective multicenter randomized controlled trial including 776 randomized patients was performed between 2018 and 2021. Patients with at least two good quality blastocysts on Day 5 were allocated by a computer randomization program in a proportion of 1:1 into either the control group, whereby single blastocysts were selected for transfer by morphology alone, or the intervention group whereby final selection was decided by a commercially available time-lapse model. The embryologists at the time of blastocyst morphological scoring were blinded to which study group the patients would be randomized, and the physician and patients were blind to which group they were allocated until after the primary outcome was known. The primary outcome was number of ongoing pregnancies in the two groups. PARTICIPANTS/MATERIALS, SETTING, METHODS: From 10 Nordic IVF clinics, 776 patients with a minimum of two good quality blastocysts on Day 5 (D5) were randomized into one of the two study groups. A commercial time-lapse model decided the final selection of blastocysts for 387 patients in the intervention (time-lapse) group, and blastocysts with the highest morphological score were transferred for 389 patients in the control group. Only single embryo transfers in fresh cycles were performed. MAIN RESULTS AND THE ROLE OF CHANCE: In the full analysis set, the ongoing pregnancy rate for the time-lapse group was 47.4% (175/369) and 48.1% (181/376) in the control group. No statistically significant difference was found between the two groups: mean difference -0.7% (95% CI -8.2, 6.7, P = 0.90). Pregnancy rate (60.2% versus 59.0%, mean difference 1.1%, 95% CI -6.2, 8.4, P = 0.81) and early pregnancy loss (21.2% versus 18.5%, mean difference 2.7%, 95% CI -5.2, 10.6, P = 0.55) were the same for the time-lapse and the control group. Subgroup analyses showed that patient and treatment characteristics did not significantly affect the commercial time-lapse model D5 performance. In the time-lapse group, the choice of best blastocyst changed on 42% of occasions (154/369, 95% CI 36.9, 47.2) after the algorithm was applied, and this rate was similar for most treatment clinics. LIMITATIONS, REASONS FOR CAUTION: During 2020, the patient recruitment rate slowed down at participating clinics owing to coronavirus disease-19 restrictions, so the target sample size was not achieved as planned and it was decided to stop the trial prematurely. The study only investigated embryo selection at the blastocyst stage on D5 in fresh IVF transfer cycles. In addition, only blastocysts of good morphological quality were considered for transfer, limiting the number of embryos for selection in both groups: also, it could be argued that this manual preselection of blastocysts limits the theoretical selection power of time-lapse, as well as restricting the results mainly to a good prognosis patient group. Most patients were aimed for blastocyst stage transfer when a minimum of five zygotes were available for extended culture. Finally, the primary clinical outcome evaluated was pregnancy to only 6-8 weeks. WIDER IMPLICATIONS OF THE FINDINGS: The study suggests that time-lapse selection with a commercially available time-lapse model does not increase chance of ongoing pregnancy after single blastocyst transfer on Day 5 compared to morphology alone. STUDY FUNDING/COMPETING INTEREST(S): The study was financed by a grant from the Swedish state under the ALF-agreement between the Swedish government and the county councils (ALFGBG-723141). Vitrolife supported the study with embryo culture dishes and culture media. During the study period, T.H. changed his employment from Livio AB to Vitrolife AB. All other authors have no conflicts of interests to disclose. TRIAL REGISTRATION NUMBER: ClinicalTrials.gov registration number NCT03445923. TRIAL REGISTRATION DATE: 26 February 2018. DATE OF FIRST PATIENT'S ENROLMENT: 11 June 2018.
