Cesarean scar pregnancy (CSP) occurs when an embryo implants in the lower uterus at the site of a previous cesarean section (CS) scar, affecting women with a history of CS [1]. It represents a specific type of ectopic pregnancy and is considered a long-term complication of CS. Globally, CS rates have been increasing, with over 30% of deliveries conducted via CS in 15 countries as of 2008. In China, the CS rate has consistently remained high, reaching 34.9% according to the 2014 National Maternal and Child Health Annual Report [2]. Due to persistently high CS rates, resulting in the likelihood of experiencing CSP in subsequent pregnancies for women with a history of CS estimated at 1 in 531 [3], and factors such as the relaxation of birth policies, the incidence of CSP in China is on a clear upward trend.

Due to the lack of large-scale studies on the outcomes of continuing pregnancies in patients with CSP, accurately classifying these patients remains challenging. As the pregnancy progresses, the risks of spontaneous bleeding, placental implantation, and uterine rupture significantly increase, especially during the mid to late stages. These complications can require a hysterectomy, eliminating the chance of future pregnancies and potentially endangering the patient’s life. Therefore, it is widely recommended to terminate a CSP as soon as it is diagnosed [4].

To date, no standardized treatment protocol exists for CSP. Treatment plans are typically tailored based on the patient’s overall condition, including clinical symptoms, gestational age (GA), serum $\beta$-human chorionic gonadotropin ($\beta$-hCG) levels, mass size, CSP classification, and other factors [5, 6]. This retrospective study reports on the management of 220 patients diagnosed with CSP at our institution, involving various treatment methods including dilation and curettage (D&C), curettage after uterine artery embolization (UAE+C), hysteroscopy plus curettage (H/S+C), H/S+C following UAE (UAE+H/S+C), and hysteroscopy combined with laparoscopic resection (L/S+H/S). The primary aim of this study is to analyze the efficacy of different treatment modalities, to identify key determinants in the selection of CSP treatment methods, and explore the feasibility of establishing standardized treatment protocols for CSP based on these findings. Through this approach, we aim to provide clearer and more scientifically grounded guidance for the effective management of CSP, thereby improving patient outcomes.

The baseline clinical characteristics of patients undergoing different strategies for treating CSP are detailed in Table 1. The L/S+H/S group, which consisted of only one individual, was not included in the statistical analysis. Significant differences were observed in CS frequency (p 0.05), GA (p 0.001), GSD (p 0.001), EL (p 0.001), fetal heartbeat (p = 0.001), myometrial thickness (p 0.001), CSP subtype (p 0.001), pre-treatment serum $\beta$-hCG levels (p 0.001), reduction in $\beta$-hCG levels post-treatment (p 0.001), blood loss (p 0.001), duration of hospitalization post-treatment (p 0.001), and cost among the other four treatment strategies (p 0.001) (Table 1).

Patients undergoing UAE, more specifically UAE+C and UAE+H/S+S, exhibited a greater reduction in post-operative serum $\beta$-hCG levels compared to patients not undergoing UAE (D&C, H/S+C, L/S+H/S), and incurred significantly higher hospitalization costs. The D&C group experienced less bleeding compared to other groups (p 0.05, Table 1).

There was no significant difference in treatment success rates among the groups (p $>$ 0.05, Table 1). Two patients with treatment failure were diagnosed with type Ⅱ CSP. In the first case, post-operative detection of residual pregnancy tissue followed an initial D&C. A successful resolution was achieved four weeks later with a combined approach of UAE+H/S+C. The second case involved UAE+C, where intra-operative ultrasound exposed complications in removing residual placental tissue due to thinning of the anterior isthmic wall and placental implantation. Significant active bleeding ensued, not controlled by intramuscular carbetocin or compression with iodine-soaked gauze, leading to 1500 ml of blood loss. The patient required a transfusion of 1.5 units of red blood cells and 400 mL of plasma. Subsequently, a procedure was performed to remove the CSP lesion and re-suture the uterus.

This study employed binary logistic regression to assess the relationship between GA, pre-treatment serum $\beta$-hCG levels, GSD, EL, and myometrial thickness in relation to UAE. Among the five variables included, GSD (odds ratio (OR): 1.651, 95% confidence interval (95% CI): 1.041–2.616, p = 0.033) and myometrial thickness (OR: 0.295, 95% CI: 0.175–0.499, p 0.001) were found to be statistically significant (Table 2). The area under the ROC curve (AUC) for the GSD was 0.856, with a standard error of 0.026, and its 95% CI ranged from 0.804 to 0.908. The cut-off value was 3.05 cm, corresponding to a sensitivity of 81.5% and a specificity of 76.0% (Fig. 2). The AUC for the myometrial thickness was 0.814, with a standard error of 0.032, and its 95% CI ranged from 0.751 to 0.878. The cut-off value was 1.55 mm, corresponding to a sensitivity of 83.8% and a specificity of 67.7% (Fig. 3).

The Pearson Chi-square test and Fisher’s exact test were used to assess the association of fetal heartbeat and CSP subtype with UAE. Both variables were significantly related to UAE, with p 0.005 (Table 3).

The relationship of GA, pre-treatment serum $\beta$-hCG levels, GSD, EL, and myometrial thickness on surgical approach was evaluated using binary logistic regression. Among the five variables included, myometrial thickness demonstrated statistical significance (OR: 0.412, 95% CI: 0.291–0.581, p 0.001) (Table 2). The AUC for the myometrial thickness was 0.748, with a standard error of 0.032, and its 95% CI ranged from 0.684 to 0.811. The cut-off value was 1.65 mm, corresponding to a sensitivity of 80.7% and a specificity of 57.0% (Fig. 4).

The association of fetal heartbeat and CSP subtype with the surgical approach was evaluated using the Pearson Chi-Square test and Fisher’s exact test. Among the two variables included, fetal heartbeat was not associated (p = 0.898), whereas CSP subtype was significantly associated with the surgical approach (p 0.001) (Table 3).

As the rate of CS increases and ultrasound diagnostic techniques advance [10], CSP has become increasingly common. To date, there is no unified standard for the treatment of CSP, both domestically and internationally. Treatment plans are primarily dependent on the expertise and judgment of specialist physicians [4], whereby the professional proficiency and depth of disease understanding of the attending physician significantly influence the treatment outcomes.

The primary treatment methods for CSP include local and/or systemic administration of MTX [11], UAE combined with MTX infusion chemotherapy, D&C, hysteroscopic lesion removal, as well as laparoscopic, abdominal, or transvaginal lesion removal [4]. Each of these diagnostic and therapeutic approaches present its own set of advantages and disadvantages.

The efficacy of MTX treatment is limited and depends upon the patient’s hemodynamic stability. Studies [10, 12, 13] suggest that initiating treatment early and encountering lower initial serum $\beta$-hCG levels are associated with a higher success rate. However, no consensus exists regarding the specific $\beta$-hCG levels and GSD thresholds for predicting the success of MTX therapy [14, 15]. UAE combined with MTX offers a less invasive option for clinically stable CSP patients, but may lead to delayed $\beta$-hCG recovery and potential liver function damage [16, 17, 18, 19, 20]. Additionally, drawbacks of MTX treatment encompass prolonged hospitalization and higher risks of bleeding and uterine rupture [21]. Considering the limitations associated with medical treatment [22], our medical institution generally favors surgical interventions.

UAE stands out as the most effective and rapid treatment for controlling bleeding [18]. Nevertheless, it is associated with high costs and post-operative complications [23], including spasmodic pain, fever (post-embolization syndrome), ovarian failure, uterine-rectal fistulas, uterine-bladder fistulas, and uterine necrosis. These complications have been reported to potentially affect the outcomes of subsequent pregnancies [18, 24]. Previous studies have indicated that adjunctive UAE may not be necessary for all CSP patients [25, 26], and therefore, it should be carefully considered. A meta-analysis of 3101 patients with CSP found that larger GSD, advanced GA, higher serum $\beta$-hCG levels, and abundant blood flow around the lesion are risk factors for significant bleeding [27]. The study by De Braud et al. [28] indicated that in cases involving a viable fetus, the risk of severe bleeding during surgery increases with GA, particularly in the presence of placental lakes. In women with a GA of $\ge$9 weeks (crown-rump length $\ge$23 mm), one-third required blood transfusions [28]. However, in our study, GA and EL were not significantly associated with the need for UAE. The logistic regression analysis in this study identified larger GSD and thinner myometrial thickness as risk factors for requiring UAE. In addition, fetal heartbeat and CSP subtypes are also associated with the need for UAE. Our study shows that the likelihood of requiring UAE increases when GSD exceeds 3.05 cm and myometrial thickness is less than 1.55 mm. This inconsistency may arise from certain CSP patients in the study experiencing halted embryonic development, where the enlargement of the GS size did not align with embryonic development and serum $\beta$-hCG levels. Our findings also suggest that CSP subtypes are associated with UAE. Considering the high correlation between the Chinese classification of CSP and myometrial thickness, a combined analysis of myometrium thickness and CSP subtypes concluded that type I CSP is less likely to require UAE, which is consistent with findings from prior literature [9]. In our study, the presence of a fetal heartbeat emerged as a risk factor for UAE treatment. However, in our study, the majority of type I CSP patients (13/15) with a fetal heartbeat did not undergo UAE treatment in our study and still achieved successful outcomes. Therefore, our study has not yet reach a conclusion regarding whether GSD, myometrial thickness, or fetal heartbeat is the more critical risk factor for UAE.

In our study, there was no significant difference in the success rates between D&C and H/S+C. D&C stands as the simplest and least invasive surgical method. However, its non-selective and blind uterine evacuation may potentially result in uncontrolled massive bleeding or uterine rupture, thereby posing a risk of uterine resection [29]. Various studies recommend varied criteria for determining the suitability of D&C, and as of yet, there is no unified standard [26, 30, 31]. In contrast to D&C, hysteroscopy provides direct visualization of the uterine cavity, enabling a clear diagnosis and thorough removal of pregnancy tissue from the cesarean scar site [3, 32, 33].

Our study found that when myometrial thickness was less than 1.65 mm, H/S+C was more likely to be chosen. This finding differs from previous studies [34], which considered hysteroscopy a better choice for women with a myometrial thickness $>$3 mm. Thinner myometrial thickness is often associated with deep cesarean scar diverticulum (CSD), and the deeper the CSD, the more challenging it becomes for D&C due to the lack of direct visualization, leading to a higher risk of initial treatment failure. Even in cases where the myometrial thickness is extremely thin, careful operation can mitigate the risk of uterine perforation, as this thin layer is closely attached to the posterior wall of the bladder. In our study, no cases of uterine perforation or organ trauma occurred.

Compared to D&C, hysteroscopy reduces the chances of initial treatment failure. However, hysteroscopy significantly increases the cost of treatment, and thus is not recommended for all types of CSP patients. Consistent with this study, another retrospective study conducted by our medical institution also showed that there is no difference in the failure rate of CSP treatment between ultrasound-guided D&C and hysteroscopy, regardless of whether UAE is performed [26].

In our study, a type III CSP patient underwent L/S+H/S, a comprehensive approach that excises the CSP lesion and simultaneously repairs the original uterine scar defect. This method is considered the preferred and optimal approach for cases of conservative treatment failure, uterine rupture, or suspected uterine rupture [35]. While UAE combined with surgical excision exhibits more favorable outcomes compared to other treatments regarding the duration of vaginal bleeding, abnormal serum $\beta$-hCG levels, and the presence of abnormal pregnancy masses [36], concerns persist regarding the significant trauma associated with CSP lesion excision surgery, the relatively prolonged post-operative recovery period, and the ongoing debate surrounding the improvement of subsequent pregnancy outcomes by uterine scar repair surgery [37, 38]. Therefore, excision of the CSP lesion should not be considered the first-line treatment method [37].

The strengths of this study primarily lie in its large sample size and rigorous research design, which encompassed well-defined inclusion and exclusion criteria, detailed treatment procedures, and comprehensive follow-up. However, due to the retrospective nature of the study, the treatment approach was non-randomized and primarily based on expert opinion. Therefore, to provide a nuanced analysis, patients were divided into subgroups, allowing for a separate examination of the risk factors associated with each treatment strategy. The limitations of the study stem from its retrospective cohort nature, wherein the allocation of patients was subject to many constraints, precluding strict randomization. Although initial risk factors might influence the decision on treatment strategy, the final treatment plan was chosen based on expert opinion. Moreover, the study focused solely on severe complications, leading to the omission of recording some relatively minor post-operative complications. Lastly, as a gynecological teaching and treatment center, the surgeons’ operative experience and diagnostic expertise were superior, which likely contributed to better diagnostic and surgical outcomes. However, this study’s failure to consider the surgeons’ experience as a potential risk factor represents a missed opportunity to explore its impact on outcomes. Therefore, future research will be designed with more rigorous methodologies to reduce allocation bias.

Overall, this study demonstrated that there is no significant difference in the success rate for treating CSP between D&C and H/S+C alone, or in combination with UAE. However, the incorporation of UAE or H/S+C into the treatment regimen significantly increases both the cost of treatment and the duration of hospital stay. Factors such as a larger GSD, reduced myometrial thickness, the presence of a fetal heartbeat, and a higher CSP subtype were identified as risk factors for choosing UAE, while a reduced myometrial thickness and a higher CSP subtype were identified as risk factors for opting for H/S+C. For different patients, it is essential to carefully select the treatment option based on the relevant risk factors.

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request with the permission of the Ethics Committee of the Women’s Hospital, School of Medicine, Zhejiang University.

XN and YZ conceived and designed the study. SN, JY and SL collected the data. XN and JY analyzed the data. XN and SN wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors have read and approved the final manuscript.

Given Considering the nature of the retrospective approach and the anonymization of patient data, approval (IRB-20220061-R) for the study protocol has been granted by the Ethics Committee of Women’s Hospital, School of Medicine, Zhejiang University. Consequently, the necessity of obtaining informed consent from individual patients has been waived. This exemption is due to the low risk posed to participants by the retrospective analysis of anonymized data, which ensures that patient confidentiality and privacy are maintained.

We would like to express our gratitude to all those who helped us during the writing of this manuscript.

This study was funded by the General Research Project of Department of Education of Zhejiang Province, P.R. China (Grant No. Y202146817).

The authors declare no conflict of interest.