In vitro fertilization increases the risk of placenta accreta spectrum in second-trimester pregnancy loss: A retrospective cohort study.

Placenta accreta spectrum (PAS) is among the most morbid etiologies of postpartum hemorrhage (PPH). The current rapid increase in cesarean delivery rates has led to PAS no longer being considered a rare condition but rather a relatively prevalent obstetric challenge. [ 1 , 2 ] Since the 2019 standardization by the International Federation of Gynecology and Obstetrics, PAS encompasses the pathological spectrum of placenta accreta, increta, and percreta. [ 1 ] The prevalence of PAS reported from 1982 to 2018 varied across studies and ranged from 1/100 to 1/1000 births. [ 1 ] In China, the incidence of PAS has increased substantially in parallel with increasing cesarean delivery rates, from 0.18% to 2.2%. [ 3 , 4 ] Crucially, the occurrence of PAS during second-trimester pregnancy loss remains clinically underexplored. The second trimester is a period from 14 +0 to 27 +6 weeks of gestation. However, second-trimester abortions accounted for <15% of all induced abortions but were responsible for two-thirds of all major complications, including PPH and retained placental tissue. The risk of retained products of conception (RPOC) following second-trimester termination has been reported in the literature to be influenced by gestational age, treatment of choice, and follow-up protocol, [ 5 ] with reported rates ranging from 1.5% to 6.0%. [ 6 , 7 ] These tissues are classified as type 0 to type 3 based on Doppler vascularity. Highly vascularized RPOC (type 2–3) might be more indicative of impending PAS. Women undergoing assisted reproductive technology, particularly in vitro fertilization and embryo transfer (IVF-ET), represent a unique population. They often have underlying infertility factors and incur higher conception costs, and the presence of underlying infertility factors may increase the risk of complications during second-trimester pregnancy loss. Although IVF-ET is an established risk factor for PAS in live births, [ 8 ] its association with PAS during second-trimester pregnancy remains unexplored. This knowledge gap is clinically significant, as complications during the management of second-trimester pregnancy loss can severely impact future fertility. Therefore, this study aimed to investigate the characteristics and outcomes of second-trimester pregnancy loss in the IVF-ET population, with a specific focus on determining the effect of IVF-ET on the incidence of PAS and identifying associated risk factors.

Conceptualization: Si Li, Man-Cheng Liu. Formal analysis: Si Li, Man-Cheng Liu, Li Zhou, Chen-Yu Gou. Funding acquisition: Si Li, Chen-Yu Gou. Methodology: Si Li. Project administration: Si Li. Software: Si Li. Data curation: Man-Cheng Liu, Li Zhou, Yun-Xu Zhao, Fei Guo. Writing – original draft: Si Li, Man-Cheng Liu. Investigation: Yun-Xu Zhao, Fei Guo. Supervision: Yu Gao, Chen-Yu Gou. Writing – review & editing: Yu Gao, Chen-Yu Gou.

This retrospective observational cohort study was conducted at the Sixth Affiliated Hospital of Sun Yat-sen University. The study protocol was approved by the Institutional Ethics Committee of the Sixth Affiliated Hospital of Sun Yat-sen University (protocol code 2022ZSLYEC-5558) and complied with the Declaration of Helsinki. This study included women who required medical management during the second trimester following a diagnosis of fetal demise or inevitable miscarriage between 14 +0 and 27 +6 weeks of gestation from January 2013 to October 2023. Women who underwent elective termination for fetal anomalies or maternal indications were excluded. We divided the women into 2 groups based on the conception method: the IVF-ET group, which included those who conceived via in vitro fertilization or intracytoplasmic sperm injection, and the non-IVF-ET group, which included those who conceived naturally or via ovulation induction or artificial insemination. In this cohort, data were extracted from electronic medical records and included the following: demographic characteristics; obstetric and gynecological history, including uterine surgical history, placenta accreta history, presence of placenta previa, and history of hysteroscopic surgery; specifics of the management process, such as the induction methods, estimated blood loss, and management of PAS; and follow-up findings. Parity was defined as a pregnancy lasting ≥28 weeks, irrespective of the outcome. Placenta previa was diagnosed ultrasonographically and defined as the presence of placental tissue covering the internal os partially or completely. Uterine abnormalities included a uterine myoma with a diameter >5 cm, adenomyosis, and a septate uterus. Uterine surgical history included myomectomy, cesarean section, adenomyomectomy, hysteroscopic myomectomy, and laparoscopic cornual wedge resection due to corneal pregnancy. Hysteroscopic surgical history included endometrial polypectomy, adhesiolysis, and uterine septum resection. The management of second-trimester pregnancy loss was tailored to the clinical diagnosis. For patients presenting with inevitable miscarriage, the primary approach involves uterotonics, with oxytocin serving as the cornerstone agent, often augmented by prostaglandin analogs such as misoprostol to expedite the process. In contrast, the management of intrauterine fetal demise utilized a broader spectrum of options. These included various pharmacologic regimens, exemplified by oxytocin, the combination of mifepristone with misoprostol, or ethacridine, as well as mechanical methods like the balloon catheter. Adjuvant oxytocin was employed as needed to ensure complete uterine evacuation and mitigate complications. All the recruited women underwent a pelvic ultrasound within 7 days following the process to access the RPOC. The women with RPOC underwent further outpatient ultrasound and/or serum human chorionic gonadotropin monitoring. Patients were followed for 1 month. Uterine involution refers to the physiological process whereby the uterus decreases in size following abortion and then returns to its prepregnancy state. The time to uterine involution was also collected from patients’ medical records. For suspected morbid PAS, magnetic resonance imaging was performed. Intrauterine adhesion (IUA) was identified hysteroscopically in symptomatic patients or on the basis of ultrasound findings. PAS management strategies included observation, oral administration of mifepristone, percutaneous uterine artery embolization with or without methotrexate, hysteroscopic resection of RPOC, intrauterine balloon tamponade, and methotrexate intramuscular injection. Retention of the placenta in situ was also used in some extreme cases. The management of RPOC was individualized, including expectant management, medical therapy (e.g., misoprostol), dilation and curettage, or hysteroscopic resection. The choice depended on clinical presentation, ultrasound findings (size and vascularity of the RPOC mass), and hemodynamic stability. The primary outcome was the occurrence of PAS, defined as the clinical diagnosis of an abnormal adherent or invasive full range or part of the placenta during abortion or histological confirmation of chorionic villi directly attached to or invading the myometrium. Cases suspected of accreta solely based on imaging findings without clinical or pathological confirmation were not diagnosed as PAS. The condition of the placenta was divided into the following categories according to the International Federation of Gynecology and Obstetrics guidelines in 2019: no accreta (grade 0), placenta accreta (grade 1), placenta increta (grade 2), and placenta percreta (grade 3). [ 1 ] The secondary outcome was the occurrence of morbid PAS, which was defined as PAS necessitating any of the following interventions: nonroutine procedures to stop excess blood loss, such as placement of an intrauterine tamponade balloon, oversewing of the placental bed, ligation of the uterine or utero-ovarian arteries, percutaneous transcatheter embolization or hysterectomy, methotrexate intramuscular injection, or hysteroscopic resection of RPOC. Patients lost to follow-up were excluded from the final follow-up analysis. Statistical analysis was performed with SPSS, version 27.0 (IBM Corp., Armonk). Quantitative variables with a normal distribution were expressed as means with their standard deviation, while nonnormally distributed variables were expressed as medians with the 25 th and 75 th percentiles. Categorical variables were presented as counts (percentages). A priori sample size calculation was performed. On the basis of the estimated incidence of PAS from prior data, a minimum sample size of 158 was calculated to achieve a 90% power with a 2-sided type I error of 0.05. Chi-square tests or Fisher exact tests were used for distribution-based comparisons between groups. T tests or Wilcoxon rank sum tests were conducted to compare group differences. Logistic regression analysis revealed independent risk factors for PAS in the second trimester, which were presented as adjusted odds ratios (aORs) with 95% confidence intervals (CIs). Subgroup analyses were performed to assess the robustness of the results for the primary outcome across different patient characteristics, including advanced maternal age, multiparity, and multiple pregnancy. The results were presented as P values, and a 2-tailed P value <.05 was considered to indicate statistical significance.

Between January 2013 and October 2023, 238 women were initially identified from the electronic medical records as having undergone management for second-trimester pregnancy loss. After excluding 11 cases due to incomplete IVF-ET records from external centers, 227 women were included in the final analysis: 93 in the IVF-ET group and 134 in the non-IVF-ET group (Fig. 1 ). The total incidence of PAS was 26.8% (61/227), and none of the patients were diagnosed with PAS prenatally. Three patients had a history of PAS. All the cases required medical management following a diagnosis of pregnancy failure, specifically fetal demise (n = 70) and inevitable miscarriage (n = 157). Flow chart of study participant inclusion and exclusion. IVF-ET = in vitro fertilization and embryo transfer. The baseline characteristics of the 2 groups are presented in Table 1 . Compared to the non-IVF-ET group, the IVF-ET group had a significantly higher average maternal age (32.9 ± 4.6 vs 29.8 ± 5.6, P  < .001). Gravidity (2.0 [1.0–3.0] vs 2.0 [1.0–3.0], P  = .07), placenta previa rate (7.5% vs 9.7%, P  = .55), and terminal gestation weeks (21.1 [18.6–24.0] vs 20.4 [18.1–23.9], P  = .37) did not significantly differ between the 2 groups. However, parity (1 [0–1] vs 0 [0–0], P  < .001) was significantly lower in the IVF-ET group. In addition, the IVF-ET group had fewer patients with a history of cesarean delivery ( P  = .004) but more patients with uterine complications ( P  = .004), a history of hysteroscopy ( P  < .001), and multiple pregnancy ( P  < .001). The number of patients with a history of uterine surgery did not differ between the 2 groups ( P  = .06). Demographics and baseline uterine characteristics between spontaneous conception group and IVF-ET group. Bold values indicate statistical significance. IQR = interquartile range, IVF-ET = in vitro fertilization and embryo transfer, PAS = placenta accreta spectrum, SD = standard deviation. The IVF-ET group had a significantly higher incidence of PAS (39.8% vs 17.9%, P  < .001) and morbid PAS (17.2% vs 3.7%, P  < .001), compared to the non-IVF-ET group. PPH volume (230.0 [115.0–575.0] mL vs 120.0 [70.0–200.0] mL, P  < .001), incidence of PPH (28.0% vs 11.2%, P  < .001), and the number of postpartum sharp curettage (1.0 [0–1.0] vs 0 [0–1.0], P  = .002) were also significantly greater in the IVF-ET group than in the non-IVF-ET group. Although the distribution of PAS classification was not different between the 2 groups, all cases of placenta percreta were in the IVF-ET group. Regarding the management of morbid PAS, more than half of the affected patients (13/16) in the IVF-ET group required procedures such as an intrauterine tamponade balloon and/or percutaneous transcatheter embolization. No hysterectomy was performed in this cohort. Notably, in the most severe case of placenta percreta, placenta in situ retention was successfully employed as a conservative approach to preserve the uterus. Moreover, the incidences of postpartum complications, including RPOC for more than 4 weeks (38.7% vs 19.4%, P  < .001), and IUA after abortion (9.7% vs 0.8%, P  = .002) were significantly greater in the IVF-ET group. The results are shown in Table 2 . The outcomes between the IVF-ET group and the control group. Bold values indicate statistical significance. IQR = interquartile range, IVF-ET = in vitro fertilization and embryo transfer, PAS = placenta accreta spectrum, RPOC = retained products of conception. The chi-square test was conducted in the patients with a detailed time of RPOC follow-up (26/108 vs 36/66). The chi-square test was conducted in the patients who found RPOC by ultrasound within 7 days postpartum follow-up (17/91 vs 28/82). Given the significant difference in maternal age between the non-IVF-ET group and the IVF-ET group, we adjusted for the effect of maternal age in the logistic analysis. The aORs of the univariate and multivariate analyses are shown in Table 3 . The final multivariate logistic analysis included gravidity, IVF-ET, uterine abnormalities, and a history of hysteroscopy. Gravidity (aOR = 1.28, 95% CI = 1.00–1.64, P  = .048), IVF-ET (aOR = 2.94, 95% CI = 1.43–6.0, P  = .003), and history of hysteroscopy (aOR = 3.51, 95% CI = 1.26–9.81, P  = .02) were independent risk factors for PAS (Table 3 ). We also conducted a subgroup analysis to confirm the effects of IVF-ET. The subgroup analysis revealed a significantly higher incidence of PAS in the IVF-ET group than in the non-IVF-ET group, especially in the young maternal age, primipara, and singleton pregnancy subgroups, but there was no difference in the advanced maternal age, multipara, and multiple pregnancy subgroups. These results confirmed the stability of the risk effect of IVF-ET in patients with PAS, even among the low-risk population (Table 4 ). Logistic regression analysis of the risk factors for the placenta accreta spectrum. Bold values indicate statistical significance. aOR = adjusted odds ratio, CI = confidence interval, IVF-ET = in vitro fertilization and embryo transfer, PAS = placenta accreta spectrum. Data are presented as medians (IQR). Multivariate analysis included gravidity, IVF-ET, uterine abnormalities, and hysteroscopy history, with odds ratios adjusted for age. Subgroup analysis of the incidence of the placenta accreta spectrum. Bold values indicate statistical significance. CI = confidence interval, IVF-ET = in vitro fertilization and embryo transfer, OR = odds ratio. To further determine the potential effect of IVF-ET, we divided the IVF-ET group into 2 groups: a fresh embryo transfer group (n = 36) and a cryopreserved embryo transfer group (n = 57). We compared the different outcomes according to the type of implanted embryo. The number of patients who had a history of hysteroscopy and the number of patients with uterine abnormalities did not differ between these 2 groups. The women who underwent cryopreserved embryo transfer had a significantly higher incidence of PAS (50.9%) than those who underwent fresh ET (8.3%; P  = .006), as well as a higher incidence of morbid PAS (24.6% vs 5.6%, P  = .018). The distribution of PAS classifications was not significantly different between these 2 groups ( P  = .93). Volume of postpartum blood loss ( P  = .19) and incidence of IUA after abortion ( P  = .71) were not significantly different. Postpartum sharp curettage times were significantly greater in the cryopreserved ET group (1 [0–1] vs 0 [0–1], P  = .003). The results are shown in Table 5 . In the IVF-ET group, after controlling for factors such as uterine abnormalities and history of hysteroscopy, cryopreserved ET (aOR = 3.52, 95% CI = 1.29–20.70, P  = .01) was still an independent risk factor for PAS according to logistic analysis (Table 6 ). The analysis between the fresh embryo transfer group and the cryopreserved embryo transfer group. Bold values indicate statistical significance. ET = embryo transfer, IQR = interquartile range, PAS = placenta accreta spectrum. Logistic regression analysis of the placenta accreta spectrum risk factors among the IVF-ET group patients after adjusting for maternal age. Bold values indicate statistical significance. aOR = adjusted odds ratio, CI = confidence interval, IVF-ET = in vitro fertilization and embryo transfer, PAS = placenta accreta spectrum. Multivariate analysis included uterine abnormalities, hysteroscopy history, and cryopreserved ET, odds ratios after adjusting for age. There were 108 patients in the non-IVF-ET group and 66 patients in the IVF-ET group who completed follow-up examinations after abortion. Ultrasound revealed that the proportion of patients without RPOC in the uterus for more than 4 weeks was significantly greater in the IVF-ET group than in the non-IVF-ET group (38.7% vs 19.4%, P  = .03; Table 2 ). Ultrasound revealed RPOC in 173 patients: 91 in the non-IVF-ET group and 82 in the IVF-ET group. In terms of the type of RPOC, types 2 to 3 are highly vascularized masses confined to the cavity and endometrium. [ 9 ] RPOC types 2 to 3 were significantly more common in the IVF-ET group (34.1% vs 18.7%, P  = .02). Ultrasound revealed an unclear boundary between the myometrium and the intrauterine residues in 23 patients, 18 of whom developed PAS during the miscarriage course, and 14 of whom underwent cryopreserved embryo transfer. Only 7 of these 18 patients underwent magnetic resonance imaging to assess the invasion depth of the residue.

Our study revealed that compared with patients in the non-IVF-ET group, those in the IVF-ET group had a significantly higher incidence of PAS, and higher risks of PPH, more uterine curettages, a longer duration of RPOC, and IUAs, although the non-IVF-ET group had more patients with a history of uterine surgery. The persistence of RPOC can cause long-term complications, such as infection and IUAs, potentially leading to secondary infertility. [ 7 ] PAS is classified according to the adhesion or invasion of villi. In our study, most PAS cases were placenta accreta (26/227, 11.4%) and placenta increta (27/227, 11.9%), but all cases of placenta percreta were found in the IVF-ET group. These results indicated that the depth of placental invasion might be greater in those who undergo IVF-ET. Moreover, we reported significantly longer persistence and greater vascularization of RPOC (types 2–3) after pregnancy loss in the IVF-ET group. These results also indicate the high risk of underestimated PAS in this population. In the baseline comparison, maternal age was significantly higher in the IVF-ET group. However, after we adjusted for the effect of maternal age, logistic regression analysis revealed that IVF-ET, especially cryopreserved embryo transfer, was still an independent risk factor for pregnancy loss in the second trimester. Almost no cases of PAS were found prenatally in our cohort. Most cases of PAS were diagnosed by abnormal placental separation or by the presence of RPOC for more than 4 weeks, and follow-up ultrasound revealing an unclear myometrium boundary, which was ultimately confirmed by pathology. The literature indicates that IVF-ET and a history of hysteroscopic surgery constitute significant risk factors for prenatally undiagnosed PAS. [ 10 ] In our study, a history of hysteroscopic surgery and uterine abnormalities were also independent risk factors for PAS in the second trimester. After hysteroscopic surgical procedures, endometrial–myometrial interface defects contribute to the development of PAS. Uterine abnormalities, which include a uterine myoma with a diameter >5 cm, adenomyosis, and a septate uterus, also contribute to infertility and a high risk of PAS. Large uterine myomas limit the cavity of the uterus, resulting in an abnormal, low-lying placenta and increasing the risk of PAS. Similarly, the patient with a septate uterus exhibited cavity constriction. A septate uterus was associated with diminished endometrial vascularization, increased susceptibility to endometrial defects, and impaired decidualization. This pathological process was similar in patients with adenomyosis, who also had a high risk of endometrial–myometrial interface defects. The decidua that develops into the placenta is called the decidua basalis. Many extravillous trophoblast cells and fibrin-like substances in the decidua basalis act as barriers to prevent the invasion of placental villi. If the endometrium at the implantation site of the placenta is defective or poorly developed, the villi are directly implanted into the myometrium or even deep into the myometrium, resulting in pathological PAS. [ 11 ] This pathophysiological process presents as hyperplastic blood vessels in the placenta accreta region on ultrasound. [ 12 ] During the separation of the placenta, the patency of hyperplastic blood vessels might lead to PPH. Prenatal diagnosis of PAS primarily depends on ultrasound findings. For asymptomatic PAS patients, the American College of Obstetricians and Gynecologists guidelines recommend ultrasound screening for placenta accreta at 18 to 20 weeks, 28 to 30 weeks, and 32 to 34 weeks of gestation. [ 13 , 14 ] Established ultrasound markers of PAS in high-risk patients include placental lacunae, abnormal uteroplacental interfaces, bridging vessels, and abnormal uterine contours (placental bulges). [ 15 ] However, the diagnostic challenge is markedly greater in the specific context of our study population. During the second trimester of pregnancy, placental echoes are weak and close to the uterine muscle, which easily obscures the detection of the interface between the uterus and placenta. [ 16 ] This diagnostic challenge is compounded by the fact that many IVF patients are primiparous and lack canonical risk factors, which lowers the index of suspicion and may lead to suboptimal detection rates in standard screening. Consequently, atypical sonographic signs are frequently disregarded, thereby contributing to diagnostic underestimation. This challenge is directly evidenced by the absence of prenatal PAS detection in our cohort, despite the elevated risk profile of the IVF-ET group. This complete lack of antenatal detection underscores a significant gap in current screening paradigms when applied to second-trimester pregnancy loss, particularly among IVF-conceived pregnancies. It highlights the imperative for increased sonographic vigilance and the development of more sensitive diagnostic criteria for this unique high-risk subgroup in the second trimester. In the IVF-ET group, 61.3% (57/93) of the women underwent cryopreserved ET. After the effects of uterine condition and history of hysteroscopy were balanced, patients with cryopreserved ET had a high risk of PAS. A study revealed that during the second and third trimesters, pregnancies with frozen-thawed ET had substantially lower uterine arterial-pulsatility index and uterine arterial resistance index values than pregnancies with fresh ET did, with lower distal vascular resistance and better placental vascular development. [ 17 ] Low vascular resistance in the uterus might provide a chance for aggressive placental development. This evidence indicates greater invasion of the placenta in the second trimester, suggesting that women with frozen-thawed ET have a greater risk of PAS during this period. Recent studies have indicated that patients undergoing frozen embryo transfer (FET) via hormone replacement cycles have significantly higher risks of PPH and PAS than those in natural cycles. This is primarily attributed to higher E2 levels during the implantation period but lower P4 levels during the ovulation, implantation, and pregnancy test periods, as well as a thinner endometrium in FET patients. [ 18 ] Among patients receiving hormone replacements, a history of ≥2 uterine surgeries and a history of blastocyst transfer were identified as independent risk factors for PAS. [ 19 ] Findings from our cohort study demonstrate that the risks of cryopreserved ET-associated placental implantation apply not only to late-trimester pregnancies but also to mid-trimester miscarriage. FET requires preparatory interventions such as hormone therapy and endometrial scratching, which might aggravate preexisting endometrial damage. Hormone replacement cycle regulation disrupts decidualization. The pathological basis of the effects of cryopreserved ET on PAS remains a defective endometrium–myometrium interface, ultimately leading to placenta accreta. Pregnancy loss during the second trimester is an adverse outcome, causing both physiological and psychological trauma. PAS can lead to severe postpartum bleeding and postpartum curettage, exacerbating endometrial damage and affecting subsequent fertility. Patients who underwent IVF-ET, especially those who underwent cryopreserved ET, were at increased risk of PAS and PPH during medical mid-trimester abortion; thus, more attention needs to be given to postpartum endometrial protection and severe complication prevention. Blood products should be prepared before the management process. These results prompted further study of placental and placental vascular development to elucidate the mechanism of PAS during the second trimester. Since ultrasound is the most economical examination for PAS, a scoring system for PAS prediction during the second trimester is needed. Our study has several limitations. The incidence of PAS was relatively high in our center. One of the reasons is the bias of the population. As a territorial tertiary obstetrics and gynecological center, more intractable cases were selected in our cohort, which indicated that these patients had more complications, a greater likelihood of uterine surgery history, and more complicated infertility factors. Given that our study was retrospective, information on postnatal follow-up was incomplete. Some patients prefer their community clinic for return visits. A larger sample of patients is needed for further study.

Among the women who experienced pregnancy loss during the second trimester, those who underwent IVF-ET, particularly cryopreserved embryo transfer, were at high risk of PAS; thus, more attention needs to be given to avoid severe complications during labor. After such events, this population had a greater likelihood of long-term RPOC and IUA, increasing the risk of infertility in the future. Patients with a history of hysteroscopy and uterine abnormalities, including myoma, adenomyosis, and a uterine septum, had a high risk of PAS, even in the second trimester. Clinicians must remain vigilant in learning to identify the markers of PAS on ultrasound during the second trimester. When an abnormal placenta is found during labor, fertility-preserving placental treatments should be used as much as possible.