The impact of in vitro fertilization and intrauterine surgery on the incidence of placenta accreta spectrum and placenta previa: a 15-year retrospective cohort study

Abstract

Background

To compare the incidence of placenta accreta spectrum (PAS) and placenta previa (PP) among unassisted conceptions and those conceived via in vitro fertilization (IVF) using fresh or frozen embryo transfer, and to evaluate the influence of prior intrauterine surgery—particularly surgical correction of a T-shaped uterus—on these outcomes.

Methods

This retrospective cohort study was conducted at a single tertiary center and included 31,905 pregnancies managed between 2008 and 2023. The exposures were mode of conception (unassisted, fresh embryo transfer, frozen embryo transfer) and history of intrauterine surgery, particularly T-shaped uterine correction. The primary outcomes were the incidence of PAS and PP confirmed at delivery. Cesarean hysterectomy was assessed as a secondary outcome reflecting maternal morbidity.

Results

PAS was identified in 4.2% of pregnancies conceived via frozen embryo transfer—significantly higher than fresh embryo transfer (0.2%) and unassisted conception (0.09%) (P < 0.001). PP followed a similar trend, observed in 7.3% of frozen embryo transfer pregnancies versus 0.7% with fresh transfer and 0.3% with unassisted conception (P < 0.001). Prior intrauterine surgery, especially T-shaped uterine correction, showed a strong association with PAS (P < 0.001). Among women with such a surgical history, 18.7% developed limited-surface PAS and 10.5% had placenta accreta. In our clinic, hysteroscopic metroplasty is selectively performed for severe T-shaped anomalies with narrowed uterine cavities prior to IVF. This study is the first to demonstrate a significantly increased incidence of PAS in IVF pregnancies following correction of a T-shaped uterus. Cesarean hysterectomy was notably more common in pregnancies complicated by PAS, especially in the frozen embryo group.

Conclusion

Frozen embryo transfer pregnancies showed a disproportionately high risk of abnormal placentation. Prior correction of a T-shaped uterus emerged as a novel and significant risk factor for PAS. These findings underscore the importance of tailored fertility strategies, cautious surgical interventions, and preconception counseling in at-risk patients.

Trial registration

This study was retrospectively registered. Ethical approval was granted by the institutional ethics committee (Approval No: 2023 − 848). The study was conducted in accordance with the Declaration of Helsinki.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-026-08861-4.

Background

Placental invasion abnormalities, with their increasing prevalence, have become one of the most critical problems in obstetrics due to their potential to cause massive blood loss during delivery, resulting in high morbidity and mortality [1]. The clinical presentations of placenta accreta spectrum (PAS) are caused by the absence or damage to the Nitabuch layer and inadequate or abnormal remodeling of spiral arteries [2].

The leading risk factors for PAS include a history of placenta previa (PP), multiple cesarean sections, or hypoxic scar tissue formation induced by trophoblast cell proliferation due to uterine surgeries performed for structural or developmental reasons. However, recent studies have shown that in vitro fertilization (IVF) is a risk factor in more than half of the cases without a history of uterine surgery or PP [3].

Some studies investigating IVF pregnancy outcomes have suggested that IVF is an independent risk factor for the development of PAS [4, 5]. The possibility that IVF treatment affects epigenetic programming events during gametogenesis and early embryo development may lead to pathological placental outcomes in the long term [6]. PAS has been found to occur more frequently in patients undergoing frozen embryo transfer (FET) with hormone replacement therapy compared to other IVF methods, which is thought to be due to lower endometrial thickness compared to other techniques [3].

The supraphysiological hormonal environment during implantation and placentation in IVF has been suggested to modulate trophoblast invasion and lead to abnormal placentation [7, 8]. Whether PAS can be attributed to the underlying type of infertility, the personal characteristics of the infertile patient, the endogenous hormonal environment associated with controlled ovarian hyperstimulation, the use of various supportive reproductive techniques in IVF, or a combination of these factors remains unclear in the literature [7, 9].

Our study aimed to assess whether the incidence of PP and PAS varies significantly based on IVF preparation and treatment protocols, including FET or fresh embryo transfers (ET), prior surgical interventions before IVF, and patient demographic characteristics.

Methods

The study was designed as a single-center, retrospective cohort study. It included pregnant women aged 18–50 years who presented to a tertiary university hospital in Ankara between January 2008 and October 2023 and reached a gestational age of at least 20 weeks. Since this was a retrospective observational study, sample size calculation was not performed.

Demographic data, obstetric and gynecological history (gravidity, parity, pregnancy losses, delivery mode, prior uterine surgeries such as curettage, hysteroscopic uterine septum resection, endometrial resection, and lateral wall incision for dysmorphic uterus), infertility diagnoses, IVF protocols (fresh vs. frozen embryo transfer), and pregnancy outcomes were retrospectively extracted from the institutional electronic database.

Patients with a history of hysteroscopic metroplasty for severe T-shaped uterus (classified as U1 dysmorphic uterus according to the ESHRE-ESGE classification, 2016) were identified. In our center, hysteroscopic metroplasty is selectively performed in cases with significantly narrowed uterine cavities prior to in vitro fertilization attempts (Fig. 1).

Fig. 1.

Severely dysmorphic uterus and a markedly narrow uterine cavity

Patients were categorized based on their mode of conception into three groups: unassisted conception, fresh ET, and FET. Pregnancies were excluded if they resulted in miscarriage before 20 weeks, if follow-up was incomplete, or if delivery occurred in another institution. Patients lacking complete data were also excluded. Plurality (singleton vs. multiple gestation) was recorded and reported descriptively (Table 1). Although multiple gestations were more common within the IVF arms, the absolute number and PAS events among multiples were small; analyses were therefore conducted at the pregnancy level without stratification or adjustment for plurality.

Table 1.

Comparison of age and obstetric history according to mode of conception

Variable	Unassisted conception n = 30.089	IVF		p-Valuea	p-Valueb	p-Valuec
		Fresh ET n = 1,155	FET n = 622
Age (years) (mean ± SD)	32.4 ± 4.8	31.6 ± 4.8	33.0 ± 5.1	< 0.001	< 0.01	< 0.001
Gravidity (n) (mean ± SD)	2.5 ± 0.9	1.1 ± 0.5	1.4 ± 0.1	< 0.001	< 0.001	< 0.001
Parity (median, min-max)	1 (0–4)	0 (0–3)	0 (0–4)	< 0.001	< 0.001	< 0.001
Twin pregnancy (%)	1.4	20.46	16.4	< 0.001	< 0.001	< 0.001
History of previous curettage (n)(median, min-max)	0 (0–6)	0 (0–6)	0 (0–6)	0.071	< 0.001	< 0.001
Previous cesarean sections (n)(median, min-max)	0 (0–3)	0 (0–2)	0 (0–2)	< 0.001	< 0.001	< 0.001
History of uterine septum surgery (n,%)	36 (0.1)	13 (1.1%)	7(%1.1)	< 0.001	< 0.001	0.999
Hysteroscopic adhesiolysis (n,%)	3 (0.1)	5 (0.4)	13 (2.1)	< 0.001	< 0.001	< 0.001
Hysteroscopic longitudinal uterine-lateral wall incision of dysmorphic uterus (n,%)	0	2 (0.2)	15 (2.4)	< 0.001	< 0.001	0.025
Frequency of Cesarean section in Current pregnancy (n,%)	13,960 (46.5)	1043 (90.0)	501 (80.5)	< 0.001	< 0.001	< 0.001
Frequency of Caesarean Hysterectomy in current pregnancy	24 (0.08)	1 (0.09)	(1.13)	0.938	< 0.001	0.002

p−values are for pairwise comparisons (^aSC vs. fresh ET; ^bSC vs. FET; ^cfresh ET vs. FET)

Abbreviations: SC Spontaneous conception, ET Embryo transfer, FET Frozen embryo transfer; CS Cesarean section

For IVF conceptions, treatment protocols included both fresh ET and FET cycles. In fresh cycles, ovarian stimulation was performed using either a GnRH antagonist or agonist protocol, followed by oocyte retrieval and immediate embryo transfer [10]. In FET cycles, endometrial preparation was performed using either natural or hormone replacement therapy (HRT) cycles. During the study period, both approaches were used; approximately 18% of FET cycles were natural-cycle FET and ~ 82% were HRT-FET (≈ 112/622 vs. ≈ 510/622). Because subtype recording was inconsistent across the 15-year period and events per stratum were sparse, outcomes were not stratified by FET subtype. Luteal phase support was routinely administered in both protocols. Recent studies emphasize that in HRT-FET cycles, luteal phase support is highly dependent on adequate progesterone levels, and individualized strategies such as doubling the vaginal dose or combining vaginal with subcutaneous administration in patients with low serum progesterone (< 10 ng/mL) have been shown to normalize pregnancy outcomes [11].

All patients undergoing IVF in our study cohort were treated using intracytoplasmic sperm injection (ICSI). The number of IVF cycles until successful pregnancy was recorded.

PP was defined according to the standardized ultrasonographic classification, including complete and partial previa as well as low-lying placenta (placental edge within 20 mm of the internal cervical os) on transabdominal or transvaginal ultrasound performed at or after 20 weeks of gestation [12]. In cases with a suspected low-lying placenta at the mid-trimester scan, transvaginal ultrasound (TVS) was performed for confirmation. PP was reassessed between 28 and 32 weeks of gestation to confirm persistence, as most marginal cases regress by the third trimester. Placental location (anterior, posterior, or central) was documented for all confirmed cases. Ultrasound was the primary screening tool; color Doppler was routinely used to assess lacunar flow, subplacental hypervascularity, and bridging vessels. MRI was obtained selectively when sonographic findings were inconclusive or to delineate suspected parametrial/bladder involvement. In all cases, imaging was used to estimate the probability of PAS, and the final diagnosis was confirmed intraoperatively and/or by histopathology when available [13].

For pregnancies without a prior lower-uterine-segment cesarean scar—such as those following hysteroscopic metroplasty for a T-shaped uterus—suspicious sonographic features included multiple grade 2–3 placental lacunae with turbulent flow, focal myometrial thinning, placental bulge, and subplacental hypervascularity/bridging vessels directed toward the serosa. Features specific to LUS scar disease (e.g., uterovesical interface interruption) were not applied to fundal/lateral placentations [14].

The extent of accreta at surgery was documented from operative notes as limited-surface (≤ 10% of the basal plate) or wider-surface (> 10%); when tissue was available, histopathology corroborated villous attachment into/onto myometrium. Intraoperative confirmation of PAS followed the FIGO clinical classification. At cesarean section or postpartum laparotomy, PAS was confirmed if (i) the placenta failed to separate with oxytocin and gentle controlled cord traction and attempts at manual removal caused brisk bleeding from the implantation site (no clear plane of cleavage), (ii) there were abnormal macroscopic findings over the placental bed—bluish–purple discoloration with a placental bulge and dense neovascularity—with a “dimple” on gentle cord traction and no separation, or (iii) placental tissue breached the uterine serosa and/or invaded adjacent organs with loss of the surgical plane between uterus and bladder. When hysterectomy or partial myometrial resection was performed, histopathology confirmed the diagnosis by demonstrating absent decidua basalis with villous attachment to superficial myometrium (accreta), villi within myometrial fibers (increta), or villi breaching the serosa ± invasion of neighboring organs (accreta + uterine dehiscence; ≈ FIGO grade 3) [14, 15]. In some cases, limited-surface accreta was not identified antenatally and was diagnosed intraoperatively by the attending surgeon upon encountering abnormal placental adherence during attempted manual separation at cesarean section.

The primary outcome was the incidence of PP and PAS in the FET and fresh ET groups compared to unassisted conception. The secondary outcome was to assess whether a history of prior intrauterine surgery, particularly hysteroscopic metroplasty for T-shaped uterus, was associated with increased PAS incidence.

Data were analyzed using SPSS version 27. Categorical variables were expressed as counts and percentages, and continuous variables as mean ± standard deviation or median (interquartile range), based on distribution. The chi-square test was used for comparing categorical variables. For numerical data, parametric tests (T-test, ANOVA) were used for normally distributed variables, and non-parametric tests (Mann–Whitney U test) were used otherwise.

We fitted a multivariable logistic regression (logit link) with the dependent outcome coded as PAS = 1 and no PAS = 0. For each covariate we report β (SE), the adjusted odds ratio (OR = exp[β]), 95% confidence interval, and two-sided p value. Covariates were selected a priori based on clinical relevance and prior literature; two-sided p < 0.05 was considered statistically significant. A priori covariates in the multivariable logistic model were prior cesarean section (yes/no), prior intrauterine surgery (yes/no), history of curettage (yes/no), and the mode of conception (fresh ET and FET, with unassisted conception as the reference). We did not adjust for PP to avoid conditioning on a potential mediator on the causal pathway.

Results

A total of 31,905 patients were included in the study, comprising those who achieved pregnancy through unassisted conception (n = 30,128) and those who conceived via assisted reproductive techniques (n = 1,777; FET: n = 622, fresh ET: n = 1,155). The median number of IVF cycles was 2 in the FET group and 1 in the fresh ET group (p < 0.001).

A comparison of demographic characteristics is presented in Table 1. Patients in the FET and unassisted conception groups were significantly older than those in the fresh ET group (p < 0.001). Gravidity and parity were lower in IVF-conceived pregnancies compared to unassisted conception pregnancies (p < 0.001). A higher number of prior cesarean sections were observed in the unassisted conception group, whereas previous endometrial curettages were more common in the FET group (p < 0.001). The incidence of twin pregnancies was higher in both IVF groups compared to the unassisted conception group (p < 0.001). Multiple gestations were more common within the IVF arms (Table 1), but their absolute number in the full cohort remained small. Cesarean section was significantly less common in the unassisted conception group, and the fresh ET group exhibited higher cesarean rates compared to the FET group (p < 0.001).

A greater proportion of patients in the IVF groups, particularly the FET group, had undergone prior intrauterine surgery (p < 0.001). In the unassisted conception group, 24 of 30,128 patients (0.07%) required cesarean hysterectomy, whereas 1/1,155 (0.09%) occurred after fresh ET and 7/622 (1.13%) after FET (Table 1).

PAS and PP Incidence Comparison

The incidence of PP was significantly higher in the FET group (7.3%) compared to the fresh ET (0.7%) and unassisted conception groups (0.3%) (p < 0.001). Placental location among confirmed PP cases was predominantly anterior (n = 92/136, 68%), followed by posterior (n = 37/136, 27%) and central (n = 7/136, 5%). At cesarean section, abnormal placental adherence required cesarean hysterectomy in 24/30,128 (0.08%) after spontaneous conception, 1/1,155 (0.09%) after fresh ET, and 7/622 (1.13%) after FET; in selected cases, uterus-preserving partial myometrial resection was performed when feasible. Relative to spontaneous conception, the risk was 1.09 (95% CI 0.15–8.04) for fresh ET and 14.12 (95% CI 6.10–32.52) for FET. Similarly, PAS was more prevalent in the FET group (4.2%) than in the fresh ET (0.2%) and unassisted conception groups (0.09%) (p < 0.001). The incidence of limited-surface accreta was also highest in the FET group (2.4%) compared to 0.1% and 0.003% in the fresh ET and unassisted conception groups, respectively (p < 0.001). Notably, in this cohort, no PAS case co-occurred with PP (i.e., no placenta previa accreta was identified) (Table 2).

Table 2.

Frequency of PAS and its subgroups according to the mode of conception

PP–PAS outcome	Unassisted conception (SC), n (%)	IVF		p-Valuesa		p-Valuesb		p-Valuesc
		Fresh ET, n (%)	FET, n (%)		RR (a) Fresh vs. SC (95% CI)		RR (b) FET vs. SC (95% CI)		RR (c) FET vs. Fresh (95% CI)
PP	83 (0.3)	8(0.7)	45 (7.3)	0.014	2.51 (95% CI 1.22–5.18)	< 0.001	26.26 (95% CI 18.43–37.42)	< 0.001	10.45 (95% CI 4.96–22.02)
Limited-surface accreta	1 (0.003)	1 (0.1)	15 (2.4)	< 0.001	26.08 (95% CI 1.63–416.79)	< 0.001	726.56 (95% CI 96.12–5491.97)	< 0.001	27.85 (95% CI 3.69–210.38)
Wider-surface accreta	11 (0.04)	0	8 (1.3)	0.545	NE	< 0.001	35.23 (95% CI 14.22–87.28)	< 0.001	NE
Increta	4 (0.01)	0	1 (0.2)	0.715	NE	< 0.001	12.11 (95% CI 1.36–108.19)	0.317	NE
Accreta + uterine dehiscence	12 (0.04)	1 (0.1)	2 (0.3)	0.362	2.17 (95% CI 0.28–16.70)	0.019	8.07 (95% CI 1.81–35.99)	0.317	3.71 (95% CI 0.34–40.88)
Total PAS	28 (0.09)	2 (0.2)	26 (4.2)	0.285	1.86 (95% CI 0.44–7.81)	< 0.001	44.98 (95% CI 26.53–76.25)	< 0.001	24.14 (95% CI 5.75–101.37)

^aSC vs. fresh ET; ^bSC vs. FET; ^cfresh ET vs. FET. p-values refer to the corresponding pairwise tests (χ² or Fisher’s exact, as appropriate). RRs were computed from raw counts; 95% CIs by the log method. Comparisons with zero events in one arm are denoted. NE (not estimable); to avoid unstable continuity-corrected estimates, RR (95% CI) is not reported for these cells

Abbreviations: SC Spontaneous conception, ET Embryo transfer, FET Frozen embryo transfer, RR Risk ratio, CI Confidence interval, PP Placenta previa, PAS Placenta accreta spectrum

Topographic terminology: accreta + uterine dehiscence (≈ FIGO grade 3), comparable to cases historically reported as “percreta”

Among 17 patients who underwent hysteroscopic metroplasty for a T-shaped uterus, six PAS cases were identified—all fundal/lateral accreta: four limited-surface accreta lesions (≤ 10% of the basal plate) and two wider-surface accreta lesions (> 10%); no increta or accreta + uterine dehiscence occurred. In this subgroup, antenatal ultrasound suggested PAS in 6/17 based on multiple grade 2–3 placental lacunae with turbulent flow, focal myometrial thinning, placental bulge, and subplacental hypervascularity/bridging vessels directed toward the serosa; lower-uterine-segment–specific signs (e.g., uterovesical interface interruption) were not applied to these fundal/lateral placentations. At delivery, these six cases demonstrated failed placental separation with brisk hemorrhage at the fundal/lateral placental bed. All six were accreta confined to the fundal/lateral bed (four limited-surface ≤ 10%, two wider-surface > 10%); no increta or percreta occurred. The remaining 10/17 post-metroplasty pregnancies had normal placental separation without PAS; one had PP without PAS. In the metroplasty subgroup, PAS occurred in 6/17 (35.3%), 95% CI 17.3–58.7; limited-surface accreta in 4/17 (23.5%), 95% CI 9.6–47.3; wider-surface accreta in 2/17 (11.8%), 95% CI 3.3–34.3; and PP without PAS in 1/17 (5.9%), 95% CI 1.0–27.0. The rates of limited-surface and total accreta were significantly elevated in this group (limited-surface accreta: 23.5% [4/17]; total accreta: 35.3% [6/17]; p < 0.001). Importantly, these patients represented a high-risk, anatomically severe subset, as hysteroscopic metroplasty in our center is reserved only for patients with markedly narrowed uterine cavities. Therefore, the elevated PAS and PP rates observed in the FET group likely reflect the selection bias toward anatomically complex cases, rather than an inherent risk associated with FET alone. The incidence and types of hysteroscopic uterine interventions associated with PAS are presented in Table 3.

Table 3.

Distribution of hysteroscopic surgical operations according to PAS findings

n (%)	Septum Surgery	T-Shape correction	Intra-uterine adhesiolysis	p-Value
Placenta Previa	2 (3.6)	1 (0.7)	0	0.003
Total PAS	1 (12.5)	6 (35.3)	1 (33,3)	< 0.001
Limited-surface accreta	0	4 (18.7)	0	< 0.001
Wider-surface accreta	0	2 (10.5)	0	< 0 0.001
Increta	0	0	1 (33.3)	0.358
Accreta + uterine dehiscence	1 (12.5)	0	0	-

*Topographic terminology: accreta + uterine dehiscence (≈ FIGO grade 3); comparable to cases historically reported as “percreta

Multivariable logistic regression showed that FET, prior intrauterine surgery, and previous curettage were independently associated with PAS, whereas prior cesarean section and fresh ET were not significantly associated. All odds ratios correspond to exp(β) from the multivariable logistic regression (PAS coded 1 vs. 0; Table 4).

Table 4.

Multivariable logistic regression analysis of factors associated with PAS

Predictor	β (SE)	Adjusted OR (95% CI)	p-Value
History of previous C-section	0.16 (0.09)	1.17 (0.98–1.40)	0.067
Intrauterine surgery (Yes vs. No)	-0.36 (0.09)	0.70 (0.58–0.83)	< 0.001
Fresh ET (vs. SC)	-0.12 (0.10)	0.89 (0.73–1.08)	0.237
FET (vs. SC)	0.77 (0.11)	2.16 (1.74–2.68)	< 0.001
History of curettage (Yes vs. No)	0.115 (0.01)	1.12 (1.10–1.14)	< 0.001

Model: binary logistic regression (logit). Coding: PAS = 1 (event), no PAS = 0. Adjusted odds ratios (aOR) = exp(β). Values are β (SE), aOR (95% CI), and p. Reference categories: mode of conception = unassisted (SC); previous cesarean = No; intrauterine surgery = No; history of curettage = No. Adjusted for prior cesarean, prior intrauterine surgery, and history of curettage; exposure = mode of conception (fresh ET, FET; Ref = unassisted). Placenta previa was not included as a covariate

Discussion

IVF has been associated in some studies with a higher incidence of PAS, despite its role in significantly improving pregnancy rates for many women, highlighting the need for further etiological investigation [3, 16, 17]. Several studies have proposed IVF as an independent risk factor for PAS development [4, 5, 18]. A meta-analysis reported an odds ratio (OR) of 5.03 for PAS in pregnancies achieved through IVF compared to those resulting from unassisted conception [3]. In our study, PAS was significantly more common in FET pregnancies, while no statistically significant increase was observed in the fresh ET group compared to unassisted conceptions—possibly due to the smaller sample size in the fresh ET group.

PAS is frequently associated with PP; a meta-analysis reported that 11% of pregnant women with PP also developed PAS [13]. Some studies have demonstrated that PP occurs more frequently in IVF pregnancies than in unassisted conceptions, with a 3- to 6-fold increase in risk [19–21]. Our study supports this association, particularly in the FET group, where the incidence of PP was markedly elevated. Additionally, IVF has been linked to PAS even in the absence of PP, as noted in another meta-analysis [22], reinforcing its role as an independent risk factor for PAS [18]. In our cohort, none of the patients diagnosed with PAS had a history of PP, which highlights the importance of other contributing factors.

There is limited literature evaluating PAS occurrence across different IVF protocols [4, 23, 24]. One meta-analysis comparing IVF and unassisted pregnancies showed a significant increase in placental abnormalities in IVF pregnancies. However, only two studies in that analysis categorized IVF procedures, and they did not specifically assess PAS incidence [25]. A more recent and comprehensive meta-analysis, which included a sensitivity analysis based on embryo transfer type, found that FET was associated with a significantly higher PAS risk compared to fresh ET’s [3]. Our findings are in line with this analysis, as only the FET group demonstrated a significantly elevated PAS risk.

Biological interpretation (FET-related endometrial environment). Hormonally prepared FET cycles may modify decidualization and placental bed vascular remodeling, providing a biologically plausible pathway for increased PAS risk.

Methodological interpretation (selection/confounding by indication). Patients with prior intrauterine surgery are more likely to follow FET pathways or differ in baseline risk; in our adjusted model (controlling for prior intrauterine surgery and curettage) the FET–PAS association persisted, though residual confounding cannot be excluded.

This cohort study compared women who achieved pregnancy through unassisted conception, ovulation induction without IVF, and assisted reproductive technologies including conventional IVF and ICSI. Compared to unassisted conceptions, the risk of PP and PAS was higher in pregnancies conceived via conventional IVF, with further elevation in those conceived through ICSI [21].

An additional strength of our study lies in the evaluation of the number of IVF attempts. The median number of IVF cycles was 2 in the FET group and 1 in the fresh ET group (p < 0.001). The higher number of IVF cycles required to achieve pregnancy in the FET group may suggest a potential link between repeated endometrial manipulation or hormonal exposure and the risk of PAS.

A limitation of our study is that all IVF patients underwent ICSI, preventing a comparative analysis between ICSI and traditional IVF protocols. FET preparation subtype (natural vs. HRT) was not consistently available across all years; consequently, subtype-stratified analyses were underpowered and not reported. Event counts constrained the precision of subgroup contrasts; to avoid over-interpretation, we did not report stratum-specific estimates, and—as with any observational analysis—some residual confounding may remain. We did not perform plurality-specific analyses owing to sparse PAS events among multiple gestations; inferences therefore reflect pregnancy-level rather than plurality-specific risks.

In our study, limited-surface accreta and wider-surface accreta were significantly more prevalent in IVF pregnancies, especially among FET pregnancies. Diffuse or focal pathologic placental invasion into the myometrium is associated with severe postpartum hemorrhage and remains a leading indication for emergency peripartum hysterectomy [26, 27]. In our study, the cesarean hysterectomy rate was significantly higher in IVF pregnancies than in unassisted ones.

Numerous studies have investigated pregnancy outcomes following hysteroscopic metroplasty in patients with T-shaped uterus—a Müllerian anomaly often associated with infertility and recurrent pregnancy loss [28–32]. Favorable reproductive outcomes have been reported in such cases [30, 33, 34]. However, these surgical interventions may lead to endometrial scarring, particularly concerning in IVF contexts. The interplay between IVF-associated inflammatory processes and scar tissue formation could alter the endometrial and decidual immune microenvironment, although this mechanism remains poorly understood. In our clinic, hysteroscopic metroplasty is not routinely performed for every T-shaped uterus diagnosed via hysterosalpingography (HSG). Surgical correction is selectively offered only in cases of severe uterine dysmorphism with markedly narrowed uterine cavities prior to IVF. This selective clinical strategy explains the higher rates of PAS observed in this subgroup, reflecting a concentration of anatomically complex cases. While the metroplasty subgroup is small (n = 17), the observed patterns are consistent and biologically plausible; estimates are therefore presented with 95% CIs and should be interpreted in context pending confirmation in larger cohorts.

To our knowledge, no previous study has evaluated PAS incidence specifically in IVF pregnancies following T-shaped uterine correction surgery. The observed increase in PAS among FET pregnancies may reflect two complementary mechanisms. These two explanations—one biological and one methodological—likely operate in parallel but through distinct pathways. The biological hypothesis relates to the altered decidualization in hormonally prepared FET cycles, whereas the methodological explanation involves selection of patients with prior uterine surgery, who inherently carry higher baseline risk. Biologically, hormonally prepared FET cycles can alter decidualization and vascular remodeling, creating a suboptimal endometrial environment that facilitates abnormal trophoblastic invasion. Methodologically, patients undergoing FET in our cohort more frequently had a history of intrauterine surgery, including hysteroscopic metroplasty, which may have contributed to a higher baseline risk of abnormal placentation. After adjustment for prior intrauterine surgery and curettage, the association between FET and PAS remained significant, suggesting that while confounding by indication cannot be fully excluded, it does not solely explain the observed relationship.

Importantly, in this cohort, no PAS case co-occurred with placenta previa. All accreta lesions were confined to the fundal or lateral uterine walls, consistent with the postoperative topography in patients who had undergone metroplasty. This finding differs from the classical “placenta previa accreta” pattern described in anterior lower-segment scars and may reflect anatomical selection bias in our population.

The most distinctive contribution of our study is that it is the first to report an increased PAS incidence in IVF-conceived pregnancies following hysteroscopic correction of T-shaped uterus, thereby filling a critical gap in the literature.

Conclusion

In this retrospective cohort study, pregnancies achieved through assisted reproductive techniques—particularly frozen embryo transfer (FET)—were associated with significantly higher rates of PP and PAS compared to unassisted conception. Furthermore, patients who underwent hysteroscopic correction for T-shaped uterus prior to IVF demonstrated an elevated risk of abnormal placentation, suggesting a compounded risk related to both surgical history and ART protocols. These findings underscore the importance of personalized reproductive planning and careful preconception evaluation, especially for individuals with a history of intrauterine surgery. Future multicenter studies with larger and more diverse populations are warranted to validate these results and better elucidate the mechanisms linking IVF modalities, uterine anomalies, and placental disorders.

Abbreviations

PAS

Placenta Accreta Spectrum

PP

Placenta Previa

IVF

In Vitro Fertilization

ICSI

Intracytoplasmic Sperm Injection

FET

Frozen Embryo Transfer

ET

Embryo Transfer

ESHRE

European Society of Human Reproduction and Embryology

ESGE

European Society for Gynecological Endoscopy

Authors’ contributions

All authors read and approved the final manuscript.- Conceptualization: Suheda Bahadir, Nuray Bozkurt- Methodology: Suheda Bahadir, Mehmet Erdem, Nuray Bozkurt- Data curation: Suheda Bahadir, Münire Funda Cevher Akdulum- Formal analysis: Suheda Bahadir, Ahmet Erdem- Writing – original draft preparation: Suheda Bahadir- Writing – review and editing: Münire Funda Cevher Akdulum, Mehmet Erdem, Ahmet Erdem, Nuray Bozkurt- Supervision: Mehmet Erdem, Ahmet Erdem, Nuray Bozkurt- Project administration: Nuray Bozkurt.

Funding

None.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the local ethics committee of Gazi University Faculty of Medicine (approval number: 2023 − 848) and conducted in accordance with the Declaration of Helsinki. For retrospectively evaluated patients, the requirement for informed consent to participate was waived by the institutional ethics committee due to the anonymized use of previously recorded clinical data.

Consent for publication

Not applicable, as this manuscript does not contain any individual person’s data (including images or other identifying information).

Competing interests

The authors declare no competing interests.