Discussion

RPL has been recognized as a major contributor to first-trimester pregnancy loss (Colley et al., Reference Colley, Hamilton, Smith, Morgan, Coomarasamy and Allen2019; Blue et al., Reference Blue, Page and Silver2019; Ozer et al., Reference Ozer, Akca, Yuksel, Duzguner, Pehlivanli and Kahraman2023; Park et al., Reference Park, Min, Kang, Yang, Hwang and Han2022; Shahine and Lathi, Reference Shahine and Lathi2015; Turesheva et al., Reference Turesheva, Aimagambetova, Ukybassova, Marat, Kanabekova, Kaldygulova, Amanzholkyzy, Ryzhkova, Nogay, Khamidullina, Ilmaliyeva, Almawi and Atageldiyeva2023; Yuksel et al., Reference Yuksel, Ozer, Duzguner, Akca, Kumtepe-Colakoglu, Yelke, Kahraman, Liperis and Serdarogullari2024). In the present study, clinical and embryo-related parameters were comparatively analyzed in PGT-A cycles of patients with UEI and RPL to identify factors associated with embryo aneuploidy and to clarify the relationship between morphokinetic behaviour and chromosomal status.

Baseline patient characteristics, ovarian reserve markers and cycle-related parameters were largely comparable between the RPL and UEI groups, including AMH levels, basal FSH concentrations and rates of fertilization and blastulation. Although some studies have reported lower AMH levels in RPL cases (Bunnewell et al. Reference Bunnewell, Honess, Karia, Keay, Al Wattar and Quenby2020; Shahine et al. Reference Shahine, Marshall, Lamb and Hickok2016), others have found no association between AMH and pregnancy loss (Bliddal et al., Reference Bliddal, Feldt-Rasmussen, Forman, Hilsted, Larsen and Christiansen2023; Yuksel et al., Reference Yuksel, Ozer, Duzguner, Akca, Kumtepe-Colakoglu, Yelke, Kahraman, Liperis and Serdarogullari2024), supporting the heterogeneity of RPL pathophysiology. Similarly, previous studies evaluating euploid embryo transfer cycles have reported inconsistent associations between female age and pregnancy loss (Del Carmen Nogales et al., Reference Del Carmen Nogales, Cruz, de Frutos, Martínez, Gaytán and Ariza2021; Nielsen et al., Reference Nielsen, Kolte, Bliddal, Jørgensen, Johnsen, Krog, Westergaard and Nielsen2024; Yuksel et al., Reference Yuksel, Ozer, Duzguner, Akca, Kumtepe-Colakoglu, Yelke, Kahraman, Liperis and Serdarogullari2024), underscoring the multifactorial nature of RPL.

In the present cohort, the duration of infertility and the number of previous ART cycles were significantly higher in the RPL group. While some studies have reported no association between infertility duration and pregnancy loss (Cimadomo et al., Reference Cimadomo, Capalbo, Dovere, Tacconi, Soscia and Giancani2021; Del Carmen Nogales et al., Reference Del Carmen Nogales, Cruz, de Frutos, Martínez, Gaytán and Ariza2021), others have suggested that repeated ART attempts and prior biopsy exposure may reflect cumulative reproductive burden rather than direct causality (Ozer et al., Reference Ozer, Akca, Yuksel, Duzguner, Pehlivanli and Kahraman2023; Yuksel et al., Reference Yuksel, Ozer, Duzguner, Akca, Kumtepe-Colakoglu, Yelke, Kahraman, Liperis and Serdarogullari2024). Differences in study design, inclusion criteria, age thresholds and outcome-based grouping across studies likely contribute to these discrepancies. BMI was modestly but significantly higher in the RPL group (RPL: 24.48 ± 4.50 vs. UEI: 23.04 ± 3.80, p = 0.009). Increased BMI has been associated with adverse ART outcomes and pregnancy loss in previous reports (Del Carmen Nogales et al., Reference Del Carmen Nogales, Cruz, de Frutos, Martínez, Gaytán and Ariza2021; Ozer et al., Reference Ozer, Akca, Yuksel, Duzguner, Pehlivanli and Kahraman2023; Yuksel et al., Reference Yuksel, Ozer, Duzguner, Akca, Kumtepe-Colakoglu, Yelke, Kahraman, Liperis and Serdarogullari2024; Nielsen et al., Reference Nielsen, Kolte, Bliddal, Jørgensen, Johnsen, Krog, Westergaard and Nielsen2024), particularly in women with obesity (Metwally et al., Reference Metwally, Saravelos, Ledger and Li2010). However, BMI was not identified as an independent predictor of embryo aneuploidy in the current analysis, suggesting that its adverse effects may be mediated through non-embryonic mechanisms.

When morphokinetic parameters were evaluated independently of ploidy status, embryos derived from RPL patients demonstrated accelerated development, particularly at t9 and tSC. However, these differences were no longer evident when only euploid embryos were analyzed, indicating that faster developmental timing alone was not associated with chromosomal competence. In the UEI group, aneuploid embryos demonstrated significantly delayed cleavage and blastulation timings compared with euploid embryos, consistent with prior time-lapse studies linking chromosomal imbalance to altered developmental kinetics (Campbell et al., Reference Campbell, Fishel, Bowman, Duffy, Sedler and Hickman2013; Chavez et al., Reference Chavez, Loewke, Han, Moussavi, Colls and Munne2012; Bamford et al., Reference Bamford, Barrie, Montgomery, Dhillon-Smith, Campbell, Easter and Coomarasamy2022). Nevertheless, existing evidence regarding the predictive value of morphokinetics for embryo ploidy remains inconsistent. In a study by Campbell et al., morphokinetic parameters of 98 blastocysts were examined and blinded to ploidy, and it was found that there were delays in the tSC stage of aneuploid embryos compared to euploid embryos (Campbell et al., Reference Campbell, Fishel, Bowman, Duffy, Sedler and Hickman2013). In addition, a meta-analysis examining morphokinetic timing variables found that delayed t8, t9, tB, tEB were associated with aneuploidy which is consistent with our results in UEI group (Bamford et al., Reference Bamford, Barrie, Montgomery, Dhillon-Smith, Campbell, Easter and Coomarasamy2022). However, Del Carmen Nogales et al. (Reference Del Carmen Nogales, Cruz, de Frutos, Martínez, Gaytán and Ariza2021) questioned the reliability of morphokinetics alone, stating that clinical and laboratory parameters are not always discriminative in cases of miscarriage after euploid transfer (Del Carmen Nogales et al., Reference Del Carmen Nogales, Cruz, de Frutos, Martínez, Gaytán and Ariza2021). A systematic review has reported that no single or combined morphokinetic parameter consistently predicts embryo ploidy status across different studies, underscoring significant methodological variability in design, outcome measures and patient populations (Reignier et al., Reference Reignier, Lammers, Barriere and Freour2018). Moreover, individual cohort studies have reported that certain abnormal morphokinetic behaviours may correlate with higher aneuploidy rates but not necessarily with different clinical outcomes when euploid embryos are transferred, further illustrating the complexity and inconsistency of the current evidence base (Yen et al., Reference Yen, Son, Giang, Quyen, Tho, Thuy and Hoi2025). In the study of Yuksel et al., no significance was obtained in terms of embryo morphokinetics between pregnancy loss and live birth groups after single euploid transfers as well (Yuksel et al., Reference Yuksel, Ozer, Duzguner, Akca, Kumtepe-Colakoglu, Yelke, Kahraman, Liperis and Serdarogullari2024). In the study of McQueen et al., it was suggested that the miscarriage of a euploid embryo is independent from its morphokinetic behaviour (McQueen et al., Reference McQueen, Mazur, Kimelman, Confino, Robins, Bernardi, Yeh, Zhang and Pavone2021). These findings support the established association between chromosomal imbalance and delayed embryo development. In contrast to the UEI group, no significant morphokinetic differences were observed between euploid and aneuploid embryos in the RPL group, suggesting that embryo developmental kinetics in RPL patients may be influenced by factors beyond chromosomal status alone. In the present study, accelerated t9 and tSC observed between study groups, independent of PGT-A results, were not associated with higher aneuploidy rates, particularly in the RPL cohort. This pattern suggests that embryo developmental kinetics in RPL patients may follow a distinct trajectory, in which chromosomal imbalance does not translate into the expected morphokinetic delay. Interestingly, euploid embryos from the RPL group exhibited earlier t9, tSC, tSB, tB and tEB values compared with euploid embryos from the UEI group. Although embryos in the RPL group exhibited earlier tEB, the interval between t9 and tEB did not show a consistent discriminatory pattern between euploid and aneuploid embryos, particularly within the RPL cohort. Importantly, the present study was not powered to assess the t9-tEB interval as an independent predictor of embryo ploidy, and this limitation should be addressed in adequately powered prospective studies. Nonetheless, the accelerated blastulation observed in euploid embryos derived from RPL patients may reflect a compensatory developmental response or altered regulatory mechanisms during the later cleavage stages. Although laser-assisted hatching was uniformly applied on day 3 across all embryos, minimizing its impact on comparative analyses, its potential effect on tEB annotation cannot be excluded. Therefore, tEB-related findings may not be directly extrapolated to embryos cultured without artificial hatching. Together, these findings indicate that morphokinetic behaviour of euploid embryos in RPL patients differs from that observed in infertile controls, and that apparently normal developmental speed does not necessarily reflect implantation competence in this population. Thus, our findings support that TLM alone may not be sufficient for embryo selection in RPL patients and that the use of PGT-A plays a more critical role in this group.

DC (from 1 to 3 cells) was associated with a 3.4-fold increased risk of aneuploidy in embryos from RPL cases, supporting the findings of Zhan et al., who reported that DC is linked to lower blastocyst formation, poorer embryo quality and indirectly reduced euploidy rates (Zhan et al., Reference Zhan, Ye, Clarke, Rosenwaks and Zaninovic2016). Abnormal morphokinetic events such as DC have previously been associated with compromised chromosomal integrity in preimplantation embryos, providing a mechanistic basis for the association observed in our RPL cohort (Bamford et al., Reference Bamford, Barrie, Montgomery, Dhillon-Smith, Campbell, Easter and Coomarasamy2022). DC may represent visible manifestations of underlying mitotic errors, which are known to increase aneuploidy risk during the early cleavage stages due to error-prone embryonic mitosis and limited cell cycle checkpoint control (McCoy et al., Reference McCoy, Summers, McCollin, Ottolini, Ahuja and Handyside2023). Although the mechanisms underlying RPL are multifactorial, accumulating evidence indicates a high prevalence of chromosomal abnormalities and genomic alterations in RPL cases, suggesting impaired genomic integrity as a contributing factor to recurrent loss (Tise and Byers, Reference Tise and Byers2021). The observation that DC was associated with aneuploidy exclusively in the RPL group may therefore reflect intrinsic differences in embryo vulnerability rather than uterine or endometrial factors, which were carefully excluded in both cohorts. Embryos derived from RPL patients may harbour subtle defects in oocyte cytoplasmic competence or chromosomal segregation machinery, predisposing them to mitotic errors manifested as DC, whereas UEI embryos, despite comparable euploid morphokinetic profiles, may exhibit greater tolerance to isolated abnormal cleavage events. From a clinical perspective, these findings indicate that DC represents a biologically meaningful marker of aneuploidy risk in RPL patients and support the use of TLM as a complementary tool alongside PGT-A, rather than as a standalone method for embryo selection in this population.

Regression analysis identified, embryo morphological quality as the strongest predictor of aneuploidy in both groups. Although embryo quality was not different between our study groups, the overall prevalence of aneuploidy in RPL cases was higher than in the UEI group. Our results were consistent with previous research, showing increased aneuploidy rates in RPL patients (Liu et al., Reference Liu, Fan, Wang, Li, Xu, Guo, Wang, Zeng, Ding, Cai, Zhou and Xu2020; Kort et al., Reference Kort, McCoy, Demko and Lathi2018). Capalbo et al. demonstrated that blastocysts with suboptimal morphology had significantly increased aneuploidy rates, reinforcing the predictive value of morphological grading in embryo selection (Capalbo et al., Reference Capalbo, Rienzi, Cimadomo, Maggiulli, Elliott, Wright, Nagy and Ubaldi2014). The risk of aneuploidy increased with decreasing embryo quality in our study, which is compatible with previous research (Capalbo et al., Reference Capalbo, Rienzi, Cimadomo, Maggiulli, Elliott, Wright, Nagy and Ubaldi2014). Other clinical factors such as maternal and paternal ages, female BMI and EB were not found to be statistically significant for the aneuploidy increase in the current study. Our results are similar to the findings of Bamford et al., which included 8147 embryos for ploidy prediction by machine learning in their study and stated that blastocyst expansion and trophectoderm morphology are significantly associated with the euploidy, while male age did not show a correlation with euploidy when stratified for female age (Bamford et al, Reference Bamford, Easter, Montgomery, Smith, Dhillon-Smith, Barrie, Campbell and Coomarasamy2023). Regarding female BMI and aneuploidy risk, our findings support the previous research in which BMI was not found to be associated with blastocyst euploidy (Goldman et al., Reference Goldman, Hodes-Wertz, McCulloh, Flom and Grifo2015; Hallisey et al., Reference Hallisey, Makhijani, Thorne, Godiwala, Nulsen, Benadiva, Grow and Engmann2022).

In the current study, although the distribution of EB and DC in all embryos was similar between RPL and UEI groups, DC was found associated with aneuploidy in the RPL group. These findings contradict the results of Watanabe et al. suggesting that DC and EB could not be associated with blastocyst euploidy since blastomeres showing DC were mostly excluded from blastocyst (Watanabe et al., Reference Watanabe, Yoshikai, Matsuda, Miyai, Sawada, Kurahashi and Sawada2023). Our results are also consistent with Zhan’s study, which found that the presence of DC in RPL patients increases the risk of aneuploidy (Herrero and Meseguer, Reference Herrero and Meseguer2013; Zhan et al., Reference Zhan, Ye, Clarke, Rosenwaks and Zaninovic2016). On the other hand, in the meta-analysis study of Bamford et al., reverse cleavage was reported to be associated with euploidy while DC was not (Bamford et al., Reference Bamford, Barrie, Montgomery, Dhillon-Smith, Campbell, Easter and Coomarasamy2022). In this context, the present study adds nuance by specifically evaluating morphokinetic patterns in a well-defined RPL cohort using combined TLM and PGT-A, thereby providing a more integrated assessment of embryo developmental dynamics and chromosomal status.

The strength of our study lies in the fact that all analyses were conducted in cases with a confirmed diagnosis of RPL defined as ≥3 miscarriages and potential confounding factors such as male factor infertility, AMA, genetic causes and endometrial pathologies were excluded. In addition, all embryo culture procedures and morphokinetic annotations were performed in a single laboratory using standardized protocols, and all time-lapse annotations were conducted by the same experienced senior embryologist who was blinded to embryo ploidy and clinical outcomes, thereby minimizing inter-observer variability.

The major limitation of the study is ART outcomes of the cases were not evaluated. The use of patients with UEI as the control group was a deliberate methodological choice. In ART-based studies, direct comparison with fertile couples is often not feasible due to ethical and practical constraints, particularly the inability to obtain comprehensive embryo morphokinetic and ploidy data from fertile populations. UEI patients represent a clinically relevant comparison group because they lack identifiable female or male factor infertility, genetic abnormalities, or known uterine pathology that could independently affect embryo development or chromosomal status. Alternative control groups such as tubal factor or male factor infertility were considered; however, these groups may introduce additional biological confounders that could independently influence embryo morphokinetics and chromosomal outcomes. Tubal factor infertility often requires surgical or radiological confirmation and may overlap with underlying conditions such as endometriosis or adenomyosis, which have been shown to adversely affect ovarian reserve, oocyte competence and embryo development. Likewise, male factor infertility has been associated with altered embryo developmental dynamics and compromised chromosomal integrity, potentially mediated by increased sperm DNA fragmentation or abnormal sperm parameters (Kahraman et al., Reference Kahraman, Sahin, Yelke, Kumtepe, Tufekci, Yapan, Yesil and Cetinkaya2020; Pellegrini et al., Reference Pellegrini, Gatti, Navarro, Hervas, Marcos, Viviana, Toschi, Galliano and Cozzolino2024). Therefore, neither tubal factor nor male factor infertility can be assumed to represent a biologically neutral baseline for embryo morphokinetic or PGT-A analyses. Although the UEI group had previous ART attempts, this study was not designed to formally classify recurrent implantation failure (RIF). Contemporary ESHRE recommendations emphasize that RIF should not be defined solely by the number of IVF cycles, but rather by individualized estimates of implantation probability and repeated failure after transfers with an acceptable predicted chance of success (ESHRE Working Group on Recurrent Implantation Failure et al., Reference Cimadomo, de Los Santos, Griesinger, Lainas, Le Clef, McLernon, Montjean, Toth, Vermeulen and Macklon2023). Detailed implantation histories and cumulative implantation probability data-particularly from ART treatments performed outside our centre-were not uniformly available to allow standardized RIF classification. Therefore, the UEI cohort was not categorized according to RIF status but was treated as infertile controls without predefined implantation failure. Thus, the use of UEI controls allows for a more controlled evaluation of embryo morphokinetic behaviour and ploidy outcomes while minimizing confounding factors unrelated to RPL. Nevertheless, the use of UEI patients as controls may limit the generalizability of the findings. The results primarily apply to women undergoing ART treatment and may not be directly extrapolated to fertile populations or to patients with other infertility aetiologies. Therefore, the conclusions should be interpreted within the context of ART-treated RPL and UEI cohorts. In addition to these limitations, the retrospective design of the study introduces a potential risk of selection bias. Specifically, only embryos meeting predefined morphological and technical criteria were selected for PGT-A analysis, while PQ embryos or those not suitable for trophectoderm biopsy were excluded. This selection process may have resulted in an underestimation of the true prevalence of aneuploidy, particularly among lower-quality embryos that are more likely to harbour chromosomal abnormalities. Consequently, the observed associations between morphokinetic parameters and chromosomal status primarily reflect embryos eligible for biopsy and may not be fully generalizable to the entire embryo population in RPL patients. Therefore, the findings should be interpreted with caution when extrapolating to all embryos generated in RPL cycles, although this approach reflects routine clinical practice in ART settings.

In conclusion, although the morphokinetic characteristics of euploid embryos in RPL patients were not discriminative, our findings indicate that the use of TLM may be useful for the identification of cleavage anomalies such as DC. Importantly, accelerated morphokinetic parameters,including t9 and tSC, were not associated with an increased risk of aneuploidy in embryos from RPL patients, underscoring that faster developmental kinetics should not be interpreted as a surrogate marker of chromosomal competence in this population. In addition, despite the exclusion of factors such as AMA, male factor and known genetic causes that could independently increase aneuploidy risk in RPL patients, the strong association between embryo quality and aneuploidy represent important findings.Within this context, PGT-A may be considered a complementary tool for embryo selection in selected RPL cases, particularly to reduce the transfer of aneuploid embryos, although its impact on clinical outcomes requires further prospective evaluation.

In routine ART workflows, these findings underline the need for a cautious interpretation of morphokinetic data in patients with RPL. While TLM provides continuous and non-invasive developmental information, morphokinetic speed alone should not be used to infer embryo chromosomal status in this population. Instead, abnormal cleavage behaviours, particularly DC, may serve as an early warning sign during embryo assessment. In clinical decision-making, identifying such cleavage anomalies may assist embryologists in embryo ranking and in guiding discussions with patients regarding embryo selection strategies. When combined with PGT-A, time-lapse observations can contribute to a more structured and transparent embryo evaluation process, potentially improving counselling and expectation management in RPL cases.