Risk of Placenta Accreta Spectrum With Early Pregnancy Subchorionic Hematoma Location and Size in Hormone Replacement Cycles.

This was a retrospective cohort study. Between November 2018 and September 2023, 2593 patients underwent FET with HRC at Tawara IVF Clinic (Figure  1 ). Of these, 993 patients achieved pregnancy, and 968 patients had singleton pregnancies. Among the singleton pregnancies, 685 patients who delivered live births were included in this study. Pregnant patients were referred to delivery facilities at 5–15 weeks of gestation. Information regarding the presence or absence of PAS was obtained from both delivery facilities and individual patients. Delivery outcomes were collected for all HRC pregnancies (100% follow‐up). PAS and other complications were obtained via standardized questionnaires sent to delivery facilities and patients; however, some clinical variables had missing values, as shown in Table  1 . This study was approved by the clinical research committee of Tawara IVF Clinic. Patient information was obtained from electronic medical records after opt‐out consent. Study subjects. Clinical characteristics in the SCH and non‐SCH groups. Note: Data are presented as n/N (%), where n represents the number with the characteristic and N represents the total available cases for that variable, or as mean ± SD ( n  = available cases) for continuous variables. Missing data are reflected in differences in denominators. Variables included in the propensity score model are marked with †. Variables additionally used in sensitivity analyses are marked with ‡. Abbreviations: GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; NRFS, non‐reassuring fetal status; OB, obstetric facility; PAS, placental adherence spectrum; PROM, premature rupture of the membranes. Only blastocyst transfers were included in this analysis. The HRC protocol administered to the study participants was as follows. After menstruation onset in the transfer cycle, estrogen replacement was initiated with three Estrana patches 0.72 mg (Hisamitsu Pharmaceutical Co. Inc., Japan) administered every other day. Progesterone supplementation was started (Day 0) with the goal of achieving an endometrial thickness of ≥ 7 mm and serum E2 concentration of 200 pg/mL. Cleavage‐stage embryo transfer was performed 2 days after ovulation (Day 2), and blastocyst transfer was performed 5 days later (Day 5). Serum P4 concentration was measured on the transfer day. Pregnancy was determined by serum hCG at 4 weeks 0–1 days of gestation, and positive cases were confirmed for the gestational sac by ultrasound examination at 5 weeks of pregnancy. In this study, confirmation of the gestational sac was defined as clinical pregnancy. For progesterone supplementation, Duphaston tablet 5mg (Viatris Pharmaceuticals Japan G.K.) as an oral preparation, and LUTINUS (Ferring Pharmaceuticals Co. Ltd.), UTROGESTAN vaginal capsules 200mg (Fuji Pharma Co. Ltd.), Luteum vaginal suppositories (ASKA Pharmaceutical Co. Ltd), and OneCrinone vaginal gel 90 mg (Merck Biopharma Co. Ltd.) as vaginal preparations were selected and used according to patient preference. SCH was defined as an echo‐free space adjacent to the gestational sac. Ultrasound examinations were first performed at 5 weeks of gestation and then every 1–2 weeks until referral to the obstetric facility. SCH size and location were assessed by physicians specialized in reproductive medicine using all ultrasound scans obtained between 5 and 15 weeks of gestation, following a standardized assessment protocol. (i) SCH location: SCH location was classified into three categories: cervical‐side, mid‐cavity, and fundal‐side (Figure  2 ). These categories were not mutually exclusive; if SCH was observed at multiple locations—either by extending across regions within a single scan or by appearing at different locations across serial scans during early pregnancy—the pregnancy was counted in each applicable location category. Specifically, a pregnancy was classified as cervical‐side SCH if cervical involvement was observed at least once; mid‐cavity SCH if mid‐cavity involvement was observed at least once; and fundal‐side SCH if fundal‐side involvement was observed at least once. Sonogram of subchorionic hemorrhage. Blue, Purple and Red lines indicate the boundary of the uterus, GS, SCH, respectively. The positions are indicated in the image as cervical‐side, mid‐cavity, and fundal‐side. This ultrasound image is at 8 weeks of pregnancy, where the SCH is located on the Fundal‐side and evaluated to be more than 50% relative to the gestational sac in size. (ii) SCH size: There are two methods for evaluating the size of SCH: one is to assess its absolute size by calculating the volume, and the other is to evaluate its relative size as a proportion of the gestational sac [ 13 ]. We adopted the latter approach in this study. SCH size was determined by the maximum size recorded across all early pregnancy scans, expressed as a percentage of the gestational sac. Pregnancies were categorized into three groups: small SCH ( 50%). The median number of ultrasound examinations performed during the observation period was 5 (range: 1–13). SCH location was assessable in 676/685 pregnancies (98.7%) and SCH size in 681/685 pregnancies (99.4%). Pregnancies with missing values were excluded from the corresponding analyses. RRs and 95% CIs were estimated using modified Poisson regression with robust variance. Propensity scores were calculated using JMP 14 (SAS Institute Inc., Cary, NC) based on clinically relevant, pre‐exposure variables selected a priori: maternal age, body mass index, obstetric history, cause of infertility, number of embryos transferred, embryo developmental stage at transfer, intracytoplasmic sperm injection use, and gestational age at referral to the obstetric facility (Table  1 and Figure  S1 ). Crude RRs represent unadjusted estimates, and adjusted RRs represent estimates adjusted for the propensity score. To evaluate robustness to potential mediator adjustment, a sensitivity analysis was performed by additionally adjusting for post‐exposure variables (placenta previa, aspirin use, and oral hormone supplementation) (Table  1 ). Categorical variables were compared using the chi‐square test or Fisher's exact test, as appropriate. Two‐sided p ‐values < 0.05 were considered statistically significant. All regression analyses were performed in RStudio (R version 4.1.1).

Analysis was limited to singleton pregnancies resulting in live birth (Figure  1 ). Among 685 HRC pregnancies, PAS was diagnosed in 41 cases (6.0%). Baseline characteristics of the HRC cohort are presented in Table  1 . Overall SCH presence showed a trend toward increased PAS frequency compared to non‐SCH pregnancies (8.0% vs. 4.6%, adjusted RR 1.56, 95% CI 0.87–2.80, Table  2 ). We then evaluated whether specific SCH characteristics, including location and size, were associated with PAS development. Association between SCH characteristics and PAS in HRC. Note: SCH characteristics and PAS risk were compared between pregnancies with and without SCH. The “Overall SCH effect” compares any SCH (regardless of location or size) versus non‐SCH pregnancies. Location‐specific analyses compare each location (cervical‐side, mid‐cavity, fundal‐side) versus pregnancies without SCH at that specific location. SCH size is expressed as the percentage of the gestational sac occupied by the hematoma, determined by the maximum recorded size across early pregnancy scans. Risk ratios (RR) with 95% confidence intervals were calculated using modified Poisson regression with robust variance estimation. The propensity score included pre‐exposure confounders†: maternal age, BMI, obstetric history, cause of infertility, number of embryos transferred, embryo developmental stage at transfer, ICSI use, and gestational age at referral to obstetric facility. Crude RR represents unadjusted estimates. Adjusted RR represents estimates adjusted for propensity score. Sensitivity Adjusted RR represents estimates with additional adjustment for post‐exposure variables‡ (placenta previa, aspirin use, and oral hormone supplementation) to assess robustness to potential over‐adjustment for mediators. 95% confidence intervals are shown in brackets. †See Table  1 for variables marked with †. ‡See Table  1 for variables marked with ‡. Regarding SCH location, cervical‐side SCH conferred significantly elevated PAS risk compared to pregnancies without cervical‐side SCH (10.5% vs. 4.9%, adjusted RR 2.21, 95% CI 1.18–4.15), whereas mid‐cavity SCH (9.8% vs. 5.1%, adjusted RR 1.74, 95% CI 0.94–3.22) and fundal‐side SCH (5.6% vs. 6.1%, adjusted RR 0.59, 95% CI 0.20–1.76) showed no significant associations (Table  2 ). For SCH size, we compared large SCH (> 50% of gestational sac) versus small SCH (< 10%) to assess whether larger hematomas conferred higher PAS risk. However, large SCH was not associated with increased PAS risk compared to small SCH (9.7% vs. 9.6%, adjusted RR 0.59, 95% CI 0.16–2.13, Table  2 ). In further analysis comparing each size category against non‐SCH pregnancies, only small SCH ( 50%) SCH showed no significant associations (Table  3 ). Association between SCH size categories and PAS compared to non‐SCH pregnancies. Note: PAS risk was compared between non‐SCH pregnancies and each SCH size category. For SCH size definition, risk ratio calculation, propensity score adjustment, and post‐exposure variables used in sensitivity analyses, see Table  2 legend. Crude RR represents unadjusted estimates. Adjusted RR represents estimates adjusted for propensity score. Sensitivity Adjusted RR represents estimates with additional adjustment for post‐exposure variables to assess robustness to potential over‐adjustment for mediators. 95% confidence intervals are shown in brackets. We conducted sensitivity analysis, additionally adjusting for post‐exposure variables: oral hormone supplementation, aspirin use, and placenta previa. The association between cervical‐side SCH and increased PAS risk remained consistent (adjusted RR 2.27, 95% CI 1.22–4.24, Table  2 ), confirming the robustness of our primary findings. To contextualize these HRC‐specific findings, we compared SCH characteristics between HRC and NC pregnancies (Figure  S2 and Table  S1 ). HRC showed a trend toward higher overall SCH frequency (40.3% vs. 36.1%, p  = 0.0628, Table  S1 ). Notably, cervical‐side SCH was significantly more prevalent in HRC compared to NC (21.2% vs. 16.1%, p  = 0.0043, Table  S1 ), whereas SCH size distribution showed no clinically meaningful difference between cycle types (Table  S1 ). In contrast to HRC, NC pregnancies showed no elevated PAS risk with cervical‐side SCH (1.3% vs. 1.1%, p  = 0.7326, Table  S2 ), highlighting that this association is specific to HRC.

In terms of the relationship between SCH location and pregnancy outcomes, Kurjak et al. reported that SCH located in the uterine corpus or fundus was associated with an increased risk of both spontaneous miscarriage and preterm delivery [ 21 ]. This location‐specific pattern suggests that the anatomical site of SCH may play a crucial role in determining pregnancy outcomes, potentially reflecting regional differences in uterine vasculature and placentation. In our study, we found that cervical‐side SCH showed a significant association with the development of PAS. The cervical‐side region possesses a complex vascular network consisting of uterine, cervical, and vaginal arteries with numerous anastomoses [ 22 , 23 ]. Furthermore, this region has a high collagen content and is predominantly composed of fibrous tissue, making the vessels distributed in this area structurally more fragile compared to those in the uterine fundus and more susceptible to damage from physical stress [ 24 , 25 ]. Additionally, in HRC, no corpus luteum is formed due to the absence of ovulation, resulting in a deficiency of luteal‐derived factors [ 26 , 27 , 28 ]. Consequently, the maternal circulatory system fails to undergo adequate vascular dilatation and adaptive vascular remodeling required in early pregnancy. This makes the system unable to withstand the increased blood flow and pressure changes associated with pregnancy, thereby predisposing to vascular rupture and microhemorrhage. Moreover, in HRC, exogenous estrogen induces vascular dilatation relatively early, which differs from the gradual and locally adaptive process observed in natural pregnancy [ 29 ]. In this way, the anatomical features of the uterus combined with the vascular physiological characteristics of HRC may contribute to making cervical‐side segment SCH a risk factor for PAS in HRC. This hypothesis proposes a novel pathophysiological concept in which the anatomical location of SCH represents crucial prognostic information and may influence the development of PAS. In addition to location, SCH size has also been considered an important prognostic factor for pregnancy complications. Abu‐Yousef et al. reported that larger SCH was associated with higher rates of miscarriage or preterm delivery compared to smaller hematomas [ 30 ], and Lou et al. reported that large SCH increased the risk of placental abruption [ 31 ]. In our study, when directly comparing small versus large SCH (> 50%), larger hematomas did not confer higher PAS risk, suggesting that hematoma size itself may not be the primary determinant of PAS risk. In further analysis comparing each size category against non‐SCH pregnancies, we found that small SCH (< 10%) was significantly associated with increased PAS frequency. These findings collectively indicate that while large SCH does not confer additional risk, even small SCH—which are often overlooked clinically—may serve as a potential risk factor for PAS development. To ensure the robustness of our findings, we conducted sensitivity analyses, additionally adjusting for post‐exposure variables that could potentially confound or mediate the SCH‐PAS association: oral hormone supplementation (reflecting HRC protocol variation), aspirin use (a known SCH risk factor [ 17 ]), and placenta previa (a known PAS risk factor [ 11 ]). The association between cervical‐side SCH and increased PAS risk remained consistent after this additional adjustment, confirming that our primary findings are robust to potential over‐adjustment for mediators. Furthermore, when comparing SCH characteristics between HRC and NC pregnancies, cervical‐side SCH—which was associated with a significantly increased risk of PAS in this study—was clearly more frequent in HRC pregnancies. In addition, among NC pregnancies, the presence of cervical‐side SCH was not associated with an increased incidence of PAS. These findings support the hypothesis that the unique physiological environment of HRC—characterized by the absence of corpus luteum function and reliance on exogenous hormones—may predispose the anatomically vulnerable cervical region to abnormal placentation when SCH occurs. In the facility participating in this study, progesterone supplementation is routinely administered even in NC; therefore, differences in exogenous progesterone administration alone are unlikely to account for the cycle‐specific PAS risk pattern observed. Moreover, in sensitivity analyses adjusting for oral hormone supplementation, the association between cervical‐side SCH and PAS risk in HRC pregnancies remained materially unchanged, suggesting that this finding was not driven by variations in luteal support protocols. As a plausible physiological explanation, in NC, luteal‐derived vasoactive factors promote physiological vascular remodeling, which may help preserve vascular integrity in the cervico–lower uterine segment and prevent progression to PAS even when SCH develops. Taken together, our findings suggest that SCH location, rather than hematoma size, represents a clinically meaningful determinant of PAS risk in hormonally programmed cycles. Recognition of cervical‐side SCH as a high‐risk marker may support individualized surveillance strategies and inform future preventive approaches in perinatal care. The limitations of this study include that the observation period for SCH was limited to early pregnancy. Furthermore, PAS diagnosis was made based on clinical criteria at multiple delivery facilities and reported by both facilities and patients via standardized questionnaires, and inter‐facility variation in diagnostic thresholds cannot be excluded. Among 10 facilities with ≥ 20 deliveries (accounting for 51.2% [21/41] of PAS cases), PAS diagnosis rates ranged from 0% to 11.4% (mean: 6.0%). While our standardized survey methodology ensured consistent data collection, potential variation in clinical judgment across facilities represents a limitation of this study. In the future, in addition to addressing these limitations, examination of the effects of different HRC protocols on SCH development is necessary.

This study demonstrates that specific characteristics of SCH in early pregnancy are associated with PAS risk in HRC. Cervical‐side SCH was identified as a significant risk factor, whereas SCH size alone did not predict PAS; notably, even small hematomas (< 10% of the gestational sac) were associated with increased PAS risk. These findings highlight the importance of SCH location in risk assessment and suggest that careful monitoring may be warranted in HRC pregnancies presenting with cervical‐side SCH.

Frozen–thawed embryo transfer (FET) is one of the important treatment options in fertility treatment. There are two approaches for endometrial preparation during embryo transfer: natural cycle (NC) and hormone replacement cycle (HRC). NC relies on the woman's natural ovulation and endogenous hormone production for endometrial preparation, while HRC is widely used as a method of preparing the endometrium using exogenous estrogen and progesterone [ 1 , 2 ]. However, multiple studies have revealed that HRC significantly increases the risk of obstetric complications such as hypertensive disorders of pregnancy (HDP) and placenta accreta compared to NC [ 1 , 3 , 4 ]. Placenta accreta is known to be a condition where part or all of the placenta abnormally adheres to the myometrium, and the resulting failure of placental separation increases postpartum hemorrhage [ 5 ]. In recent years, a series of spectrum‐like conditions including placenta accreta and placenta adherens have been collectively referred to as Placenta Accreta Spectrum (PAS) [ 5 ]. Saito et al. reported an increase in PAS (referred to as placenta accreta in the original paper, but considered to be similar to PAS) in HRC compared to NC (Adjusted Odds Ratio (AOR): 6.91 [95% Confidence Interval (CI) 2.87–16.66]) [ 3 ]. Takeshima et al. similarly demonstrated an increase in PAS in HRC compared to TrueNC (AOR: 4.14 [95% CI 1.64–10.44]) [ 4 ]. Furthermore, Li et al. also demonstrated that the frequency of placenta adherence was significantly higher in HRC compared to NC (15.30% vs. 9.24%, p  = 0.004) [ 6 ]. Subchorionic hematoma (SCH) is a hematoma formed between the placenta or chorion and the uterine wall, and is observed as a hypoechoic area around the gestational sac on ultrasound examination [ 7 , 8 ]. A systematic review by Tuuli et al. reported that the presence of SCH is associated with increased rates of miscarriage, abruption, and preterm premature rupture of membranes [ 9 ]. In our previous retrospective cohort study, we found that SCH in early pregnancy was significantly associated with the development of PAS (referred to as abnormal placental adhesion in the original paper), with frequencies of 9.4% versus 3.0% in groups with and without SCH, respectively (AOR 7.01 [95% CI 2.96–18.00]) [ 10 ]. Although placenta previa is known to be a risk factor for placenta accreta, Matsuo et al. reported an association between SCH and the development of PAS in non‐previa placentas [ 11 , 12 ]. These results indicate that the presence of SCH in early pregnancy serves as a risk factor for PAS. The incidence of SCH is significantly higher in infertility treatment patients compared to natural pregnancies [ 9 , 13 ]. While natural pregnancies show SCH incidence ranging from 1.3% to 11% [ 14 , 15 , 16 , 17 ], infertility treatment including assisted reproductive technology (ART) demonstrates markedly elevated rates of 22.4%–40.2% [ 17 , 18 , 19 ]. Notably, Asato et al. identified FET as a specific risk factor for SCH development (OR 6.18 [1.7–22.4]) [ 19 ]. Furthermore, Reich et al. reported that SCH in early pregnancy occurs more frequently in HRC compared to NC [ 20 ]. Since HRC predisposes to SCH development, and SCH is associated with increased PAS risk, the increased SCH incidence in HRC may contribute to the higher rates of PAS observed in HRC. However, the specific characteristics of SCH associated with PAS in HRC remain unclear. This study aimed to investigate the relationship between SCH characteristics (location and size) in early pregnancy and PAS development in HRC. By identifying specific SCH features that predict PAS risk, this research aims to provide evidence for individualized risk stratification and improved clinical management in FET.

S. S. and N. M. are affiliated with the funded laboratory of TawaraIVF clinic.

Figure S1: rmb270005‐sup‐0001‐FigureS1.pptx. Table S1: rmb270005‐sup‐0002‐TableS1.docx.