P-383 Development and validation of a visualized prediction model for early miscarriage risk in patients undergoing IVF/ICSI procedures: a real-world multi-center study

Abstract

Abstract Study question What are the key predictors of early miscarriage in patients undergoing IVF/ICSI treatment? Summary answer This study identified risk factors for early miscarriage in IVF/ICSI patients, and developed a validated predictive nomogram, highlighting age-dependent factor variations. What is known already Early miscarriage, occurring in 15-20% of confirmed pregnancies, can lead to complications, psychological distress, and fertility impairment. It is especially prevalent during the first trimester, with up to 80% of all miscarriages happening during this phase. Women who conceive via Assisted Reproductive Technology (ART) face higher stress levels and increased risk of miscarriage. Although many contributing factors to early miscarriage have been identified, there is a scarcity of integrated predictive models for IVF/ICSI patients. This paper aims to develop a validated, user-friendly nomogram to predict first-trimester miscarriage risk in IVF/ICSI patients and help clinicians and patients make informed decisions. Study design, size, duration A retrospective analysis was conducted on 20,322 first cycles out of 31,307 for IVF/ICSI at a reproductive centre in Southern China, After excluding ineligible cycles, 6,724 first fresh cycles were included and randomly divided into a training dataset (n = 4,516) and an internal validation dataset (n = 2,208). An external validation dataset (n = 1,179) was obtained from a reproductive center in Northern China, further ensuring the robustness and generalizability of the findings. Participants/materials, setting, methods This retrospective study analysed 6,724 out of 31,307 IVF/ICSI treatments from 2011 to 2020 at a Southern China reproductive center and 1,179 cycles from a Northern China center (2013-2018) for validation. Early miscarriage was defined as a loss before 12 weeks’ gestation. Patients were divided into two age groups: ≥35 and <35 years. The study followed the TRIPOD statement and received Institutional Review Board approval. Main results and the role of chance The study found statistically significant differences across the training, internal validation, and external validation sets in terms of patient age, BMI, number of good quality embryos, antral follicle count (AFC), basal testosterone (T), luteinizing hormone (LH) levels, among other variables. The early miscarriage rates were 9.43%, 10.55%, and 13.50% across these sets. Elder women (age ≥35 years) had a higher duration of infertility and early miscarriage rate, but lower AFC and numbers of retrieved oocytes, 2PN, and MII oocytes compared to younger women (age <35 years). A nomogram prediction model was developed using variables such as female age, BMI, number of spontaneous abortions, number of induced abortions, and basal FSH, which were found to be independent prognostic factors for early miscarriage. The model’s performance was validated using AUC, calibration plots, and DCA, indicating acceptable performance and good agreement between prediction and observation. The model was more beneficial for predicting miscarriage risk than the treat-all or treat-none schemes. Subgroup analysis for elder and younger women showed different risk factors for miscarriage, notably female age, basal FSH for elder women, and female age, BMI, and the number of spontaneous abortions for younger women. The subgroup models also showed consistent results and good clinical relevance. Limitations, reasons for caution The retrospective design of our study may have introduced certain limitations and potential biases. To overcome these limitations, it is advisable that future research endeavors adopt a prospective clinical study design. Wider implications of the findings This study uses a diverse dataset and rigorous methods to create a user-friendly prediction model for early miscarriage in ART. By considering regional differences, it ensures the model’s broader applicability and accuracy, enabling personalized treatment strategies and improving infertility treatment outcomes. Trial registration number The study was a retrospective study, and it was approved by the Institutional Review Boards of Sun Yat-Sen Memorial Hospital (SYSEC-KY-KS-2021-121) and the First Affiliated Hospital of Xinjiang Medical University (K202106-17) and is reported in accordance with the TRIPOD statement