The diagnostic criteria for CSP are as follows: Amenorrhea, complicated with or without irregular vaginal bleeding; Elevated serum β-human chorionic gonadotropin (β-hCG) levels; A history of previous cesarean section or uterine surgery. Typical findings on magnetic resonance imaging (MRI) or ultrasonography: (1) The uterine cavity and cervical canal are empty, with no gestational sac detected; (2) The gestational sac is implanted in the myometrium of the lower anterior uterine segment (corresponding to the site of the previous cesarean section incision), and some gestational sacs may contain fetal buds or exhibit fetal cardiac activity; (3) The continuity of the anterior uterine myometrium is disrupted, and the myometrium between the gestational sac and the bladder is significantly thinned or even absent; (4) MRI or color Doppler flow imaging (CDFI) reveals blood flow signals around the gestational sac.

This retrospective clinical study was approved by the Quzhou People’s Hospital Ethics Committee. A total of 98 patients with cesarean scar pregnancy (CSP) who underwent MRI examination at our hospital from May 2016 to September 2023 were included. Exclusion criteria were: history of contrast agent allergy, uterine rupture, hemorrhagic shock, severe coagulation disorders, and severe comorbidities involving major organs (heart, liver, brain, lungs, kidneys). According to MRI findings, patients were categorized into three types (Figure 1): Type I (Hypo vascular type): Gestational sac showed no obvious enhancement or non-enhancement. Type II: Gestational sac was enhanced, with a small amount of flow-void vascular shadows around it, but no obvious vascular lakes, arteriovenous fistulas, pseudoaneurysms, or aneurysms. Type III (High-risk type): Gestational sac showed significant enhancement, with surrounding flow-void vascular shadows, accompanied by one or more of the following: obvious vascular lakes, arteriovenous fistulas, pseudoaneurysms, or aneurysms. All patient MRI images were retrospectively classified independently by two senior radiologists, each possessing over 10 years of experience in abdominal and gynecological imaging diagnosis, thus ensuring a high level of expertise in the initial classification. In the initial classification phase, any cases where the two readers had a disagreement regarding the Type (e.g., borderline Type II/III cases concerning vascularity) were resolved through mutual discussion and consensus.

Patients were required to fast for more than 4 h before the examination and drink an appropriate amount of water to achieve moderate bladder filling. If necessary, 10 mg of anisodamine was intramuscularly injected to reduce motion artifacts caused by intestinal peristalsis. Gadopentetate dimeglumine injection (0.1 mmol/kg) was administered via the cubital vein at a rate of 2 ml/s. After the injection, examinations were performed using a GE Signa Voyager 1.5T magnetic resonance scanner. First, sagittal scans were performed. Based on the sagittal images, axial and coronal scan planes were determined according to the relationship between the lesion and the scar. The main scanning sequences included: sagittal T2WI, sagittal T1WI, axial LAVA-Flex, coronal T2WI FSE with fat saturation, and axial DWI.

Verify the patient’s identity, confirm the puncture site, and establish intravenous access. The patient was placed in a supine position, and local anesthesia was administered. Puncture and catheterization were performed via the right/left femoral artery or left/right radial artery. Angiography with a pigtail catheter was performed at a level slightly above the renal artery plane of the abdominal aorta to observe the visualization of relevant responsible arteries and the lesion. Under fluoroscopic guidance, super-selective catheterization into the bleeding artery was performed, respectively and confirmed by angiography. Gelatin sponge particles were continuously injected until blood flow to the lesion ceased and the blood flow in the main trunk of the uterine artery slowed significantly. The puncture site was then dressed, and the operation was completed (Supplementary Figure 1).

The patient is carefully placed in the lithotomy position. The surgical area is disinfected, sterile drapes are applied, and an ultrasound transducer is placed transabdominally to visualize the uterus. The physician gently inserts a curette into the uterine cavity and, under ultrasound guidance, removes residual tissue without damaging the uterus. Postoperatively, instruments are removed, an ultrasound recheck is performed to ensure complete evacuation, and appropriate postoperative care is provided.

The perineal area is disinfected, and under ultrasound guidance, a puncture needle is inserted transvaginally into the gestational sac to administer methotrexate (MTX). After observing no abnormalities, ultrasound-guided curettage is performed to remove the affected gestational tissue, with real-time monitoring of instrument position and uterine wall status to ensure thorough evacuation and prevent uterine injury. Postoperatively, instruments are removed, a check for residual tissue is performed, and the patient undergoes postoperative observation and care.

The patient is carefully placed in the lithotomy position. The vulva and vagina are disinfected, sterile drapes are applied, and the uterus is distended before inserting a hysteroscope to examine the uterine cavity. After locating the gestational tissue, micro-instruments are introduced via the hysteroscope’s channel to precisely excise the tissue and its attachments, avoiding damage to normal tissues. Post-resection, the uterine cavity is checked for no residual tissue or bleeding, the hysteroscope is withdrawn, and the patient is monitored postoperatively for vital signs and vaginal bleeding, with appropriate nursing and treatment guidance provided.

General anesthesia is administered, and the patient is placed in the supine position. The area from the xiphoid process to the pubic symphysis and both sides to the mid-axillary lines is disinfected and draped. A transverse incision is made two fingerbreadths above the pubic symphysis, and the abdomen is entered. The uterus is examined to determine the location of the gestational tissue within the uterine scar. An incision is made on the uterine scar, avoiding major vessels, and the gestational tissue is carefully removed. The scar is trimmed, necrotic tissue is cleared, and the uterine muscle layers are sutured in layers with absorbable thread to achieve hemostasis and restore uterine anatomy. After confirming no intra-abdominal bleeding and verifying instrument counts, the abdomen is closed in layers. Postoperatively, the removed tissue is sent for pathological examination, and the patient is monitored for vital signs, vaginal bleeding, and wound condition.

During hospitalization, dynamically monitor serum β-hCG levels, hepatorenal function, complete blood count, and CSP lesion size. Record surgical blood loss and complications (e.g., fever, vomiting, dizziness, pain). Vaginal bleeding is graded as none, minimal, moderate, or heavy, relative to menstrual flow. Within 3 months post-discharge, follow up every month. Follow-up includes serum β-hCG, hepatorenal function, complete blood count, abdominal ultrasound, and menstrual resumption. Grade complications using the NCI-CTCAE V3.0 criteria.

SPSS 27.0 statistical software was used for data analysis. Among the data of patients with three magnetic resonance imaging (MRI) types, those who did not undergo uterine artery embolization (UAE) before termination of pregnancy were selected for multiple linear regression analysis. Taking intraoperative blood loss as the dependent variable, and “age, number of cesarean sections, number of previous uterine curettages, interval between cesarean section and pregnancy, duration of amenorrhea, amount of vaginal bleeding before admission, duration of vaginal bleeding, β-human chorionic gonadotropin (β-HCG) level, presence or absence of fetal heart activity, whether the gestational sac morphology protrudes toward the bladder on MRI images, thickness of the thinnest part of the scar, gestational sac area, gestational sac being cystic or cystic-solid, and surgical termination of pregnancy” as independent variables. Prior to conducting the multiple linear regression analysis, the dependent variable, intraoperative blood loss, underwent Log-Transformation. This log-transformation effectively converted the highly skewed distribution into a distribution that is sufficiently close to normal, thereby satisfying the residual normality assumption of the linear regression model. We have further generated and reported the results of a simplified model developed using stepwise regression. In patient data, within each of the three types, blood loss was compared and analyzed between patients who did not undergo preoperative UAE and those who had undergone preoperative UAE. For measurement data with non-normal distribution, the two independent samples rank sum test (Mann-Whitney U test) was used. P-value < 0.05 was considered statistically significant.

A total of 98 patients were included (Table 1). CSP Type I: 35 patients. MRI mainly showed simple cystic images, such as ovoid, teardrop - shaped or dumbbell - shaped. The gestational sac was implanted in the uterine scar, growing toward the isthmus or cavity. Few vascular flow - voids were seen on MRI. All gestational sacs were cystic and none bulged toward the bladder. CSP Type II: 35 patients. The gestational sac could be cystic or mixed cystic - solid. Its area was larger than that of Type I, while the scar thickness was similar. The gestational sac was enhanced, with a few flow - void vessels around. However, there were no obvious vascular lakes, arteriovenous fistulas, pseudoaneurysms or aneurysms. CSP Type III: 28 patients. The time since last menstrual period (LMP) duration and vaginal bleeding duration were longer than the first two types. A higher proportion of patients had mixed cystic - solid gestational sacs, which were larger than the first two types and significantly enhanced. Flow - void vessels were visible around, along with one or more of vascular lakes, arteriovenous fistulas, pseudoaneurysms or aneurysms. Type III patients had more bleeding. All surgeries were successful, and all patients preserved their uterus. One week post - operation, β - HCG levels dropped by over 50% among all patients. Menstruation resumed in 24–47 days post - operation, with no significant difference among Type 3 patients.

A total of 78 patients who did not undergo uterine artery embolization (UAE) before pregnancy termination surgery were included, as shown in Supplementary Table 1. We conducted a multiple linear regression analysis, with surgical blood loss as the dependent variable, and the following as independent variables to calculate risk factors associated with surgical blood loss: age, number of cesarean sections, number of previous curettages, interval between cesarean section and current pregnancy, duration of amenorrhea, amount of vaginal bleeding before admission, duration of vaginal bleeding, β-human chorionic gonadotropin (β-HCG) level, presence or absence of fetal cardiac activity, whether the gestational sac protrudes toward the bladder on magnetic resonance imaging (MRI), thickness of the thinnest part of the scar, gestational sac area, whether the gestational sac is cystic or cystic-solid, and surgical method for pregnancy termination. Supplementary Table 2 shows that the R² is 0.845, indicating that the statistical results can explain 84.5% of the factors associated with surgical blood loss during pregnancy termination. The analysis results are statistically significant, as shown in Supplementary Table 3. According to the results of the multiple linear analysis, the significance was p < 0.001 when the gestational sac was classified as Type III (Supplementary Table 4). MRI classification of the gestational sac is an indicator for predicting surgical blood loss during pregnancy termination, and Type III on MRI is the main factor contributing to bleeding. Due to the small sample size and large number of independent variables, we generated and reported the results of a simplified model developed using stepwise regression to avoid overfitting (Table 2), and the R² is 0.733. The results of the stepwise linear regression analysis demonstrated that, among all risk factors, Type III (high-risk type) of the MRI-based classification was the strongest independent predictor of intraoperative blood loss (β = 0.911, p < 0.001), followed by Type II (β = 0.279, p < 0.001), suggesting that the MRI-based classification system can reliably predict the risk of hemorrhage.

Intra-group comparisons were performed for patients with three types of CSP, and the Mann-Whitney U test (two independent samples rank sum test) was used to analyze the impact of UAE on intraoperative blood loss during pregnancy termination surgery in patients with the three types (Figure 2). Among them, there were 35 patients with MRI-classified CSP Type I, 3 of whom underwent preoperative UAE, with p = 0.8376. For 35 patients with CSP Type II, 6 underwent preoperative UAE, with p = 0.0629. Given the limited dataset of this research, preoperative UAE did not demonstrate any statistically significant advantages in either Type I or Type II patients. Due to the small number of patients who received preoperative UAE in these two types, this conclusion is unreliable and requires further statistical analysis with an expanded sample size. There were 28 patients with MRI-classified CSP Type III, including eleven who underwent UAE before pregnancy termination surgery and 17 who did not. Statistical analysis showed p < 0.001, indicating that preoperative UAE can significantly reduce intraoperative blood loss in patients with MRI-classified CSP Type III. For patients with Type II, due to the small sample size, no statistically significant difference was observed, and further research with an expanded sample size is required.

A case in our center is as follows: a 34-year-old female was admitted due to “58 days of amenorrhea, and B-ultrasound suggesting incisional scar pregnancy for 1 day.” She had a history of two cesarean sections. Color Doppler ultrasound 2 days before admission showed “anterior isthmic pregnancy without cardiac activity (measuring approximately 36 mm × 35 mm × 30 mm, with abundant blood flow signals, 0.7 mm from the serosal layer).” She was admitted with a tentative diagnosis of “cesarean scar pregnancy.” Upon examination, her β-HCG was 154552.88 IU/L. Magnetic resonance imaging (MRI) showed that the gestational sac was cystic-solid on T2-weighted images, with a large number of vascular flow voids around the gestational sac; partial enhancement within the gestational sac was observed on T1-weighted contrast-enhanced images. Angiography under digital subtraction angiography (DSA) revealed retention of contrast agent in the pseudoaneurysm within the gestational sac. On the second day after UAE, the patient underwent local injection of methotrexate (MTX) under ultrasound guidance combined with uterine curettage, with intraoperative blood loss of approximately 20 mL. The operation went smoothly, and the patient returned to the ward safely after surgery. One week after surgery, her β-HCG was 1993.41 IU/L, and menstruation resumed on the 34th day after surgery.