Abstract
Objective
To develop a pre-percutaneous microwave ablation (PMWA) model combining conventional ultrasound (CUS) and contrast-enhanced ultrasound (CEUS) for predicting non-perfused volume ratio (NPVR) ≥ 80% in patients with uterine fibroids.
Methods
This multicentre, retrospective study enrolled 276 patients with pathologically confirmed uterine fibroids (training/validation = 140/136). CUS features (maximum diameter, pre-operative volume, location, echo), CEUS parameters (maximum intensity: IMAX), and ablation parameters (energy, time, energy per unit volume [EPV], and time per unit volume [TPV]) were analysed. Multivariate logistic regression identified predictors, and model performance was assessed using the area under the receiver operating characteristic curve (AUC) and clinical decision curve analysis.
Results
Multivariate analysis identified four independent predictors: maximum diameter (odds ratio [OR] = 0.47, p < 0.001), IMAX (OR = 0.34, p = 0.013), energy (OR = 1.01, p = 0.002), and TPV (OR = 0.69, p < 0.001). The prediction model performed well in both the training (AUC = 0.79, 95% confidence interval [CI] = 0.71–0.87) and validation (AUC = 0.70, 95%CI = 0.61–0.79) datasets. P values for the Hosmer–Lemeshow test were 0.263 and 0.980, respectively, and the Brier Score was 0.184.
Conclusion
This study established a prediction model based on CUS features, CEUS perfusion parameters, and ablation parameters to evaluate the efficacy of PMWA for uterine fibroids; NPVR ≥ 80% can be predicted using the nomogram. This model enables accurate prediction of the efficacy of PMWA for uterine fibroids, providing profound guidance for pre-operative assessment and intraoperative management in clinical practice and facilitating personalised, precision treatment.
Keywords: Uterine fibroids, Percutaneous microwave ablation, Conventional ultrasound, Contrast-enhanced ultrasound, Predictive model
Introduction
Uterine fibroids (UFs) are benign, oestrogen- and progesterone-sensitive smooth muscle tumours, with a prevalence of 70%–80% in women of reproductive age [1]. Most patients with uterine fibroids (UFs) are asymptomatic and require no intervention. However, over one-third of affected individuals have distressing symptoms, mainly heavy menstrual bleeding (HMB)—which may lead to iron deficiency anemia—and bulk symptoms (e.g., pelvic discomfort, impaired bladder function) that disrupt daily activities [2] [3]. And approximately 25% of UF patients experience symptoms severe enough to need surgical intervention, such as myomectomy or minimally invasive ablation[4] The most common symptom is abnormal uterine bleeding, which can lead to anaemia, fatigue, and dysmenorrhoea. Other symptoms include increased abdominal size, dyspareunia, and compression-related issues such as urinary and bowel dysfunction [5]. Conventional treatments include hysterectomy and myomectomy, which, although effective, are invasive and require extended recovery [6]. Minimally invasive alternatives such as embolisation and ablation have gained prominence with advancements in imaging technologies. Among ablation therapies, percutaneous microwave ablation (PMWA) has emerged as a promising option because of its high efficacy, minimal invasiveness, and favourable safety profile [7]. PMWA offers significant fibroid volume reduction, sustained shrinkage, and improved quality of life with low complication rates [8–10]. Recent studies have highlighted the safety, efficacy, and reliability of ablation therapies as a treatment option for UFs [11].
Contrast-enhanced ultrasound (CEUS), which uses intravascular contrast agents for microvascular imaging, is widely adopted in clinical practice. CEUS not only provides clear visualisation of tumour size, margins, morphology, and real-time perfusion patterns but also enables quantitative assessment of haemodynamic parameters, including time-intensity curves (TICs), peak enhancement time, and peak intensity [12, 13]. In addition to pre-operative CEUS evaluation, postprocedural CEUS can accurately delineate the ablation zone to evaluate the fibroid ablation effect. The ablation effect, represented by the non-perfusion volume ratio (NPVR) of ablation, is a key indicator of the therapeutic effect and closely related to treatment outcomes and symptom alleviation. At the 4th Radiological-Gynecological Expert Meeting, it was mentioned that NPVR of ablation should be as large as possible; this is positively correlated with symptom relief. Moreover, the NPVR is considered a technical parameter for treatment success [14]. Several studies showed that treatment of uterine fibroids with HIFU was safer and more effective when the NPVR was higher than 70%[15, 16].Therefore, accurate prediction of NPVR before and during surgery is crucial. Optimisation of pre-operative evaluation criteria is essential for selecting suitable patients for PMWA and accurately determining the required ablation energy.
The aim of this study was to establish a model for predicting NPVR ≥ 80% by investigating the relationship between conventional ultrasound (CUS) characteristics, CEUS perfusion parameters, and ablation indicators in patients with UFs treated by PMWA. The clinical utility of this model was further assessed through external validation, providing both theoretical guidance and evidence-based support for PMWA therapy for UFs.
Methods
Patients
This study included 353 patients with UFs who received ultrasound-guided PMWA treatment at our hospital and four other medical centres from June 2024 to December 2024. Patients were aged between 23 and 60 years. This retrospective study was approved by the ethics committee, and the requirement for informed consent was waived. Written informed consent was obtained from each patient before PMWA treatment. Figure 1 shows the detailed patient selection process.
Fig. 1.
Study flowchart. This flowchart shows the patient selection process and the schema for statistical analysis
The inclusion criteria were as follows: pathologically confirmed uterine fibroids (UFs); International Federation of Gynecology and Obstetrics (FIGO) type 1 to 6 determined by magnetic resonance imaging (MRI); symptoms such as menorrhagia, secondary anaemia, and lower abdominal discomfort; availability of a safe percutaneous puncture route; absence of perimenopausal signs; and completion of pre- and post-PMWA ultrasound and CEUS assessments.
The exclusion criteria were as follows: suspected uterine malignancy; UFs combined with adenomyosis; menstruation, pregnancy, or lactation; FIGO type 0 and 7 UFs; uncontrolled acute pelvic inflammatory disease; dysfunction of any vital organ; severe coagulation disorder (platelets < 50 × 109/L, prothrombin time > 25 s, prothrombin activity < 40%); incomplete ultrasound or CEUS data; ablation rate > 100%; and TIC fitting quality < 75%.
Measures for ensuring data consistency across multiple centres
To ensure the consistency and comparability of microwave ablation data among the five centres, strict quality control measures were implemented. First, we developed a detailed, standardised operating protocol and provided unified training to all participating physicians. The ablation procedures at the five centres were performed by senior physicians with > 10 years of experience, including rich experience in ablation treatment. Second, all five centres used the same ablation equipment and accurately recorded key parameters such as power, time, and energy. All operations were conducted under real-time ultrasound guidance, and image data were saved for subsequent review. Finally, to ensure objectivity of CEUS parameters, TIC analyses for all UFs were performed by a single investigator not involved in the ablation procedures.
Pre-treatment assessment
All patients underwent routine pre-operative examinations, including laboratory tests, chest radiography, electrocardiography, and MRI to rule out contraindications. CUS, colour Doppler ultrasound, and CEUS were performed at three medical centres using an RS80A system (Samsung Medison Co., Ltd.) equipped with a CA1-7A convex array probe. Ultrasound systems (ACUSON Sequoia, Siemens Medical Solutions USA, Inc.) equipped with a 5C1 MHz convex array transducer were used at two medical centres. Conventional ultrasound was used to record the size, location, echogenicity, and pre-treatment volume of fibroids, which was calculated using the following formula: pre-V = abc*0.523 (V: volume, a: anteroposterior diameter, b: transverse diameter, c: longitudinal diameter). Sonazoid (0.015 mL/kg body weight; Daiichi-Sankyo, Tokyo, Japan) was used as the ultrasound contrast agent [17]. Continuous scanning for 120 s was performed to evaluate blood perfusion in the fibroids, and DICOM video clips were stored for analysis.
CEUS video clips were evaluated using Tom Tec’s SonoLiver software (version 1.0) to generate TICs. An ellipsoid region of interest (ROI) was placed within each target lesion. First, we defined ROI, including the lesion and surrounding normal myometrial tissues. Next, for ROI analysis, we selected both fibroid tissue and normal myometrial tissue as areas of interest, using normal myometrial tissue as a reference. Only TICs with a fitting quality of ≥ 75% were included. SonoLiver output data were recorded as follows: maximum intensity (IMAX), rise time (RT), time to peak (TTP), and mean transit time (MTT).
PMWA treatment and evaluation
Preoperatively, the skin was disinfected, local anesthesia was administered, and the bladder was emptied. Under ultrasound guidance, a safe puncture path was selected to avoid critical structures. The microwave ablation needle was inserted into the fibroid, and ablation was conducted at 50–60 W using a "mobile ablation" technique to achieve conformal treatment [11]. The procedure was terminated when hyperechoic signals from vaporization covered most of the lesion. Intraoperative CEUS was performed to assess the ablation effect, with supplemental ablation applied to any enhancing areas. The number of needles and ablation time were recorded. Final NPVR was confirmed by CEUS four hours post-procedure. NPVR was calculated as follows: NPVR = non-perfused volume of the target lesion after ablation/volume of the target lesion before ablation ×100% [11]. The effect was classified into four grades according to NPVR: completely, NPVR ≥ 80%; majority, NPVR of 60% to 79%; partially, NPVR of 20% to 59%; ineffectively, NPVR ≤ 19% [18]. The energy consumed for each 1 ml of nodular volume was defined as the energy required per unit volume (EPV) and calculated as follows: EPV = power (W) × time (s)/non-perfused volume (cm3) [19, 20]. The time consumed for each 1 ml of nodular volume was defined as the time required per unit volume (TPV) and calculated as follows: TPV = time (s)/non-perfused volume (cm3). According to a previous study, the RCS regression model showed a U-shaped, non-linear dose–response relationship between IMAX and both EPV and TPV (all pnonlinear < 0.001) (Fig. 2). The inflection point corresponding to the lowest risk was 203.56%. Therefore, we defined IMAX < 203.56% as non-high-enhancement = 0, and IMAX ≥ 203.56% as high-enhancement = 1.
Fig. 2.
RCS curves for IMAX, EPV, and TPV. The x-axis represents the maximum enhancement intensity (IMAX) of CEUS, whereas the y- axis represents EPV and TPV. The shaded area indicates the 95% CI (pnonlinear < 0.001)
Statistical analysis
All analyses were performed using SPSS (version 24.0; IBM Corp., Chicago, IL, USA) and R software (version 4.3.0; https://www.R-project.org/). Continuous variables are expressed as mean ± standard deviation (SD) and were analysed using Student’s t-test or the Mann–Whitney U test as appropriate. Categorical variables were compared using the χ2 test or Fisher’s exact test. Based on patient recruitment sources, cases were stratified into an internal training cohort and an external validation cohort. Multivariate logistic regression with stepwise selection was used to identify independent predictors in the training dataset. Significant features were evaluated using odds ratios (ORs) with corresponding 95% confidence intervals (CIs). Statistically significant predictors of ablation efficacy were incorporated into a predictive nomogram model. R software was used to build a prediction model based on logistic regression in the training cohort. Predictive models for NPVR ≥ 80% were established in the training cohort based on CUS and CEUS features. The model was comprehensively validated and evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), Hosmer–Lemeshow test, clinical decision curve analysis (DCA), and Brier Score. P < 0.05 indicated statistical significance.
Results
Baseline data
In total, 276 patients with UFs were included. From these, 177 fibroids underwent complete ablation, while 99 underwent incomplete ablation. The average age of patients was 44.5 ± 5.6 years. The average maximum diameter of fibroids was 7.82 ± 2.35 cm, and the average maximum volume of fibroids before the operation was 209.91 ± 187.46 cm3.
For the training dataset, we selected 140 cases of UFs from our centre. Complete and incomplete ablation were achieved in 89 (63.57%) and 51 (36.43%) cases, respectively. These 140 lesions were used to construct the prediction model. For the validation dataset, we selected 136 patients with UFs from the other three centres. Complete and incomplete ablation were achieved in 88 (64.71%) and 48 (35.29%) cases, respectively. These 136 lesions were used to evaluate the performance of the predictive model. The demographic characteristics of patients and ultrasound characteristics of UFs in the training and validation datasets are shown in Tables 1 and 2, respectively.
Table 1.
Comparison of demographic characteristics of patients and ultrasound characteristics of uterine fibroids between the training and validation datasets
| Variables | Total (n = 276) | test (n = 136) | train (n = 140) | p value |
|---|---|---|---|---|
| Age(years), Mean ± SD | 44.53 ± 5.56 | 44.25 ± 5.11 | 44.81 ± 5.97 | 0.406 |
| MAX-D(cm), Mean ± SD | 7.82 ± 2.35 | 7.88 ± 2.53 | 7.76 ± 2.17 | 0.673 |
| Pre-V(ml), Mean ± SD | 209.91 ± 187.46 | 219.74 ± 208.27 | 200.37 ± 164.95 | 0.392 |
| TIME(s), Mean ± SD | 783.24 ± 488.71 | 775.48 ± 483.07 | 790.79 ± 495.75 | 0.795 |
| Enger(J), Mean ± SD | 45,247.43 ± 32,556.67 | 44,354.78 ± 32,644.99 | 46,114.57 ± 32,564.27 | 0.654 |
| TPV, Mean ± SD | 7.94 ± 3.93 | 7.60 ± 3.67 | 8.28 ± 4.16 | 0.151 |
| EPV, Mean ± SD | 405.02 ± 185.88 | 389.48 ± 181.55 | 420.11 ± 189.42 | 0.171 |
| RT(s), Mean ± SD | 12.33 ± 6.86 | 12.26 ± 7.46 | 12.38 ± 6.25 | 0.885 |
| TTP(s), Mean ± SD | 15.64 ± 15.42 | 16.28 ± 19.78 | 15.03 ± 9.47 | 0.500 |
| Mtt(s), Mean ± SD | 65.58 ± 53.60 | 64.71 ± 57.56 | 66.42 ± 49.64 | 0.792 |
| Location, n (%) | 0.443 | |||
| Intramural | 142 (51.64) | 66 (48.53) | 76 (54.68) | |
| Subserosal | 95 (34.55) | 52 (38.24) | 43 (30.94) | |
| Submucosal | 38 (13.82) | 18 (13.24) | 20 (14.39) | |
| Eco, n (%) | 0.746 | |||
| Hypoechoic | 228 (91.57) | 111 (90.98) | 117 (92.13) | |
| Hyper/ isoechoic | 21 (8.43) | 11 (9.02) | 10 (7.87) | |
| AT group, n (%) | 0.844 | |||
| NPVR ≥ 80% | 99 (35.87) | 48 (35.29) | 51 (36.43) | |
| NPVR < 80% | 177 (64.13) | 88 (64.71) | 89 (63.57) | |
| IMAX, n (%) | 0.903 | |||
| non-high-enhancement | 137 (49.64) | 67 (49.26) | 70 (50.00) | |
| high-enhancement | 139 (50.36) | 69 (50.74) | 70 (50.00) |
Table 2.
Intergroup comparison of demographic characteristics of patients and ultrasound characteristics of uterine fibroids between the training and validation datasets
| Training dataset | Validation dataset | |||||||
|---|---|---|---|---|---|---|---|---|
| Variables | Total (n = 140) | NPVR<80% (n = 51) | NPVR≥80% (n = 89) | p | Total (n = 136) | NPVR<80% (n = 48) | NPVR≥80% (n = 88) | p |
| Age(years), Mean ± SD | 44.81 ± 5.97 | 44.92 ± 7.39 | 44.74 ± 5.02 | 0.877 | 44.25 ± 5.11 | 44.21 ± 4.87 | 44.27 ± 5.26 | 0.944 |
| MAX-D(cm), Mean ± SD | 7.76 ± 2.17 | 7.85 ± 2.30 | 7.71 ± 2.11 | 0.705 | 7.88 ± 2.53 | 8.22 ± 2.65 | 7.70 ± 2.46 | 0.249 |
| Pre-V(ml), Mean ± SD | 200.37 ± 164.95 | 197.22 ± 152.58 | 202.17 ± 172.45 | 0.865 | 219.74 ± 208.27 | 246.00 ± 235.06 | 205.41 ± 192.01 | 0.279 |
| TIME(s), Mean ± SD | 790.79 ± 495.75 | 764.90 ± 490.08 | 805.62 ± 501.12 | 0.642 | 775.48 ± 483.07 | 775.10 ± 489.09 | 775.68 ± 482.58 | 0.995 |
| Energy (J), Mean ± SD | 46114.57 ± 32564.27 | 45631.37 ± 33976.63 | 46391.46 ± 31919.77 | 0.895 | 44354.78 ± 32644.99 | 44604.17 ± 31969.83 | 44218.75 ± 33188.25 | 0.948 |
| TPV, Mean ± SD | 8.28 ± 4.16 | 10.09 ± 5.08 | 7.24 ± 3.10 | <.001 | 7.60 ± 3.67 | 8.55 ± 4.21 | 7.07 ± 3.25 | 0.024 |
| EPV, Mean ± SD | 420.11 ± 189.42 | 496.15 ± 223.35 | 376.54 ± 151.86 | 0.001 | 389.48 ± 181.55 | 451.48 ± 196.69 | 355.66 ± 164.22 | 0.003 |
| RT(s), Mean ± SD | 12.38 ± 6.25 | 11.68 ± 4.98 | 12.79 ± 6.86 | 0.314 | 12.26 ± 7.46 | 11.74 ± 6.44 | 12.55 ± 7.99 | 0.548 |
| TTP(s), Mean ± SD | 15.03 ± 9.47 | 13.48 ± 5.78 | 15.91 ± 10.98 | 0.144 | 16.28 ± 19.78 | 14.16 ± 8.36 | 17.44 ± 23.78 | 0.357 |
| Mtt(s), Mean ± SD | 66.42 ± 49.64 | 68.69 ± 55.66 | 65.11 ± 46.12 | 0.683 | 64.71 ± 57.56 | 61.77 ± 46.31 | 66.31 ± 63.05 | 0.661 |
| Location, n (%) | 0.497 | 0.206 | ||||||
| Intramural | 76 (54.68) | 29 (56.86) | 47 (53.41) | 66 (48.53) | 19 (39.58) | 47 (53.41) | ||
| Subserosal | 43 (30.94) | 17 (33.33) | 26 (29.55) | 52 (38.24) | 20 (41.67) | 32 (36.36) | ||
| Submucosal | 20 (14.39) | 5 (9.80) | 15 (17.05) | 18 (13.24) | 9 (18.75) | 9 (10.23) | ||
| Eco, n (%) | 1.000 | 0.632 | ||||||
| Hypoechoic | 117 (92.13) | 45 (91.84) | 72 (92.31) | 111 (90.98) | 44 (93.62) | 67 (89.33) | ||
| Hyper/ isoechoic | 10 (7.87) | 4 (8.16) | 6 (7.69) | 11 (9.02) | 3 (6.38) | 8 (10.67) | ||
| IMAX, n (%) | 0.008 | 0.191 | ||||||
| non-high-enhancement | 70 (50.00) | 18 (35.29) | 52 (58.43) | 67 (49.26) | 20 (41.67) | 47 (53.41) | ||
| high-enhancement | 70 (50.00) | 33 (64.71) | 37 (41.57) | 69 (50.74) | 28 (58.33) | 41 (46.59) |
Multivariate logistic stepwise regression analysis revealed that IMAX (OR = 0.34, 95%CI = 0.14–0.79, p = 0.013), maximum diameter (OR = 0.47, 95%CI = 0.31–0.71, p < 0.001), energy (OR = 1.01, 95%CI = 1.01–1.01, p = 0.002), and TPV (OR = 0.69, 95%CI = 0.58–0.81, p < 0.001) were independent predictors for achieving NPVR ≥ 80% (Table 3).
Table 3.
Univariate and multivariate logistic stepwise regression analyses of predictive factors for achieving NPVR ≥ 80%
| Variables | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
| β | p | OR (95%CI) | β | p | OR (95%CI) | |
| Location | ||||||
| Intramural | ||||||
| Subserosal | −0.06 | 0.882 | 0.94 (0.44 ~ 2.03) | |||
| Submucosal | 0.62 | 0.278 | 1.85 (0.61 ~ 5.63) | |||
| Eco | ||||||
| Hypoechoic | ||||||
| Hyper/ isoechoic | −0.06 | 0.924 | 0.94 (0.25 ~ 3.51) | |||
| IMAX | ||||||
| non-high-enhancement | ||||||
| high-enhancement | −0.95 | 0.009 | 0.39 (0.19 ~ 0.79) | −1.09 | 0.013 | 0.34 (0.14 ~ 0.79) |
| MAX-D(cm) | −0.03 | 0.703 | 0.97 (0.83 ~ 1.14) | −0.76 | <.001 | 0.47 (0.31 ~ 0.71) |
| Pre-V(ml) | 0.00 | 0.864 | 1.00 (1.00 ~ 1.00) | |||
| TIME(s) | 0.00 | 0.639 | 1.00 (1.00 ~ 1.00) | |||
| Energy (J) | 0.00 | 0.894 | 1.00 (1.00 ~ 1.00) | 0.01 | 0.002 | 1.01 (1.01 ~ 1.01) |
| TPV | −0.18 | <.001 | 0.84 (0.76 ~ 0.92) | −0.38 | <.001 | 0.69 (0.58 ~ 0.81) |
| EPV | −0.01 | <.001 | 0.99 (0.99 ~ 0.99) | |||
| RT(s) | 0.03 | 0.313 | 1.03 (0.97 ~ 1.09) | |||
| TTP(s) | 0.03 | 0.152 | 1.03 (0.99 ~ 1.08) | |||
| Mtt(s) | −0.00 | 0.681 | 1.00 (0.99 ~ 1.01) | |||
The prediction model from the training dataset showed good discriminatory ability (AUC = 0.79, 95%CI = 0.71–0.87) (Fig. 3A) and was appropriately calibrated (Hosmer–Lemeshow goodness of fit test, X2 = 10.029, p = 0.263; Fig. 3B). Application of the prediction model in the external validation dataset also yielded good discrimination (AUC = 0.70, 95%CI = 0.61–0.79; Fig. 3A) and calibration (Hosmer–Lemeshow goodness of fit test, χ2 = 2.030, p = 0.980; Fig. 3C). Clinical DCA showed that significant net clinical benefits could be obtained (Fig. 3D, E). The Brier Score was 0.184.
Fig. 3.
Performance evaluation of the model for predicting NPVR ≥ 80% to evaluate the efficacy of PMWA in patients with uterine fibroids. A ROC curves for the training and validation datasets show AUCs of 0.79 and 0.70, respectively. B, C Calibration abilities for the training and validation datasets. The dashed line represents the reference line where the ideal diagnostic model lies. The dotted line indicates the calibration ability of the prediction model, whereas the solid line corrects for bias. D, E DCA curves for the training and validation datasets
The optimal cut-off value for the prediction model was 0.551, obtained by analysing AUC and calculating the maximum Youden index. In the training dataset, the prediction model showed 0.75 accuracy, 0.63 sensitivity, 0.82 specificity, 0.67 positive predictive value (PPV), and 0.79 negative predictive value (NPV); these values were 0.66, 0.54, 0.73, 0.52, and 0.74, respectively, in the validation dataset (Table 4).
Table 4.
Estimation of the prediction model by using a cut-off value of 0.551
| Data | AUC (95%CI) | Accuracy (95%CI) | Sensitivity (95%CI) | Specificity (95%CI) | PPV (95%CI) | NPV (95%CI) | cut off |
|---|---|---|---|---|---|---|---|
| Train | 0.79 (0.71–0.87) | 0.75 (0.67–0.82) | 0.63 (0.49—0.76) | 0.82 (0.74—0.90) | 0.67 (0.53—0.80) | 0.79 (0.71—0.88) | 0.551 |
| Test | 0.70 (0.61–0.79) | 0.66 (0.58–0.74) | 0.54 (0.40—0.68) | 0.73 (0.63—0.82) | 0.52 (0.38—0.66) | 0.74 (0.65—0.84) | 0.551 |
The independent predictors obtained from multivariate logistic regression analysis were used to construct the nomogram (Fig. 4). The points assigned to each factor were summed to generate a total score, with each total score corresponding to a probability of achieving NPVR ≥ 80%. Specifically, this risk value represents the likelihood that the UFs will attain an NPVR exceeding 80% following PWMA—a threshold that denotes the complete ablation of the lesion[18]. A higher risk value means a greater probability of achieving complete ablation of the leiomyoma after surgery, as well as a higher likelihood of symptom relief for the patient [14]. Conversely, a lower probability value suggests an increased probability of suboptimal ablation outcomes, and thus the need to adjust the treatment strategy accordingly. Figure 5 shows an example of the nomogram used in this study.
Fig. 4.
Nomogram for predicting the NPVR ≥ 80% in patients with uterine fibroids
Fig. 5.
Example of nomogram usage. A This is a representative case of a 47-year-old patient with a uterine fibroid. The fibroid measures 8.9 cm ×7.4 cm ×7.4 cm, with a volume of approximately 254.1 cm3. B MRI (T2) demonstrated an intramural uterine fibroid with multiple smaller fibroids. C CEUS indicates an IMAX of approximately 34.0%. D Real-time ultrasound-guided microwave ablation therapy; the arrow indicates vaporized hyperechoic. E, F The ablation energy was 73,400 J, and a TPV of 7.65 J/s. The post-operative NPVR was 85%, which was calculated by CEUS and contrast-enhanced T1-weighted magnetic resonance images (MRI). G: This patient showed a total score of approximately 158 and the predicted NPVR ≥ 80% was 0.74
Discussion
We successfully constructed and validated a prediction model based on CUS features and CEUS perfusion and ablation parameters. This model was used to predict the possibility of complete ablation after PMWA for UFs (NPVR ≥ 80%). In both the training and validation sets, the model demonstrated good discrimination ability (AUC: 0.79 in the training set and 0.70 in the validation set) and calibration. Using multivariate analysis, we identified four independent predictors: maximum fibroid diameter, IMAX on CEUS, ablation energy, and TPV. Furthermore, the model correctly identified 73% of patients who failed to achieve complete ablation (NPVR < 80%), which helps minimize the risk of misclassifying ineligible patients as suitable candidates for complete PMWA. Additionally, NPV was 74%, indicating that 74% of patients categorized by the model as "unlikely to achieve complete ablation" indeed did not meet the NPVR ≥ 80% threshold. Consequently, for patients identified by the model as having a low likelihood of complete ablation, proactive adjustments to treatment strategies—such as increasing ablation duration or energy delivery—may be implemented to enhance ablation efficacy. These findings provide an objective, evidence-based basis for clinical treatment decision-making in patients with UFs.
Morphological characteristics of UFs
Previous studies on high-intensity focused ultrasound (HIFU) ablation of UFs have shown that fibroid volume is an important factor for achieving complete fibroid ablation [21, 22]. Larger lesions demonstrate greater residual fibroid tissue owing to uneven thermal conduction and distribution resulting from their substantial volume. Notably, contrary to previous research findings, the present results indicate that the maximum fibroid diameter is a better predictor of the ablation effect than is volume. There are three potential explanations for this finding. Firstly, for irregular fibroids, maximum diameter better captures critical tumor extension. As the thermal ablation radius is limited (2–3 cm)[23], dimensions exceeding this limit risk incomplete necrosis, unlike spherical fibroids which allow uniform heating. Secondly, including both maximum diameter and volume introduces multicollinearity, destabilizing regression models. Thirdly, volume estimation relies on inherently inaccurate geometric approximations for irregular lesions [24]. Thus, maximum diameter offers superior clinical practicality and reproducibility.
Hemodynamic characteristics of UFs
TICs derived from CEUS enable the acquisition of blood perfusion parameters for UFs, which are crucial for monitoring and evaluating the fibroid ablation efficacy [12].In the assessment of UF perfusion following HIFU therapy, the enhancement rate on CEUS emerged as the optimal parameter, demonstrating the highest predictive value for ablation efficacy in ROC analysis[16]. This finding is consistent with the results of our study, which demonstrated that IMAX obtained from CEUS can serve as an effective predictor of fibroid ablation outcomes. The ability of CEUS to quantify blood perfusion and assess treatment response further highlights its potential for guiding minimally invasive therapeutic interventions. Previous studies have indicated that high blood perfusion reduces heat accumulation during ablation, likely because of perfusion-mediated tissue cooling within the targeted treatment area [25, 26]. Therefore, fibroids with an abundant blood supply or those surrounded by more blood vessels may experience reduced ablation efficiency due to heat dissipation, resulting in a smaller microwave ablation zone [20, 27]. Both contrast-enhanced MRI and CEUS can reflect the vascular characteristics of fibroids, with higher enhancement intensities indicating increased blood supply [28]. CEUS enables pre-operative prediction of ablation difficulty in UFs. For hypervascular fibroids, combined chemical ablation or intraoperative energy adjustment may be required. Conversely, hypovascular fibroids with predominant fibrous tissue components present unique thermal conduction challenges, necessitating meticulous intraoperative monitoring to ensure complete ablation zone coverage.
Ablation parameters for UFs
Ablation duration and energy output are the most crucial intraoperative parameters demonstrating a strong correlation with complete ablation. Sufficient energy delivery is essential to ensure that the tumour tissue reaches the threshold for irreversible thermal damage [29]. Insufficient energy or time may lead to incomplete fibroid ablation owing to uneven thermal diffusion or excessive heat dissipation, particularly in areas with abundant blood supply or fibrous content [30]. Complete ablation requires a sufficient thermal dose, achievable through either extended time or higher power [31]. Ultrasound-guided PMWA allows real-time monitoring of fibroid location and vascularity while enabling dynamic adjustments of ablation power and duration during the procedure. Special attention must be paid to fibroids located near the intestinal tract or uterine serosal layer, where precise control of power and exposure time is crucial to minimise thermal damage to adjacent tissues. For deeper fibroids, longer ablation times or higher power settings are often required to ensure complete tissue necrosis.
To our knowledge, this is the first study to jointly establish an ablation prediction model using multimodal parameters, including conventional ultrasound features and CEUS perfusion and ablation parameters. Second, we conducted a more accurate quantitative assessment of CEUS parameters. Finally, we developed a practical nomogram model for clinical use to facilitate rapid and accurate decision-making before and during surgery. This study also had several limitations. First, the retrospective design may have introduced a selection bias. Second, CEUS quantification depends on semi-automated software, which can lead to inter-observer variability. Future multicentre, prospective studies should implement standardised imaging protocols alongside machine learning-based perfusion analysis. Further validation across diverse populations and investigation of combined predictive factors may enhance the accuracy of the fibroid ablation model. Lastly, the model demonstrates limited efficacy in accurately identifying patients who will ultimately achieve complete ablation (NPVR ≥ 80%). This limitation may be attributed to the inherent complexity and heterogeneity of uterine fibroids, as well as potential unmeasured confounding factors that modulate ablation outcomes—factors not fully integrated into the current predictive framework. Future studies should prioritize model optimization, such as the incorporation of additional clinical, imaging, or biological markers, to enhance its performance in predicting ablation success.
Conclusions
The prediction model developed in this study demonstrated robust performance in both the training and validation datasets, exhibiting excellent discriminative ability and calibration, and NPVR ≥ 80% could be predicted using the developed nomogram. This model enables accurate and effective prediction of the efficacy of PMWA for UFs, providing profound guidance for pre-operative assessment and intra-operative management in clinical practice and facilitating personalised and precision treatment. Additionally, it offers parameter adjustment recommendations for less-experienced physicians, helping to minimise efficacy variations caused by insufficient experience.
Acknowledgements
We thank Qian Zhang (Clinical Epidemiology and EBM Unit, Beijing Friendship Hospital, Capital Medical University) for her guidance on statistics. We would like to thank Editage (www.editage.cn) for English language editing.
Author contribution
Yuqing Huang, Yahui Ma and Lei Liang conceived designed this study. Yahui Ma, Yufeng Wang, Bin Wang, Ya Sun, Nan Zhou, Xuedi Han, Ruyue Tian,Hao Yu, and Lei Liang completed the operation. Xin Zhang, Yankun Zhao, Xiaohong Sun and Yandong Deng came from different centers and completed the operation. Yuqing Huang acquired, analyzed the data as well as wrote the original manuscript. Lei Liang review the manuscript. Yuqing Huang and Yufeng Wang contributed to the work equally and should be regarded as co-first authors. Yandong Deng and Lei Liang regarded as corresponding authors.
Funding
National Natural Science Foundation of China,62371010,Sponsored by Science Foundation of AMHT,2024YK07
Data availability
No datasets were generated or analysed during the current study.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Yandong Deng, Email: dyd1977@126.com.
Lei Liang, Email: lianglei_csk@126.com.
References
- 1.Giuliani E, As‐Sanie S, Marsh EE (2020) Epidemiology and management of uterine fibroids. Int J Gynecol Obstet 149:3–9. 10.1002/ijgo.13102 [DOI] [PubMed] [Google Scholar]
- 2.Zimmermann A, Bernuit D, Gerlinger C, et al (2012) Prevalence, symptoms and management of uterine fibroids: An international internet-based survey of 21,746 women. BMC Womens Health 12:6. 10.1186/1472-6874-12-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Miller CE, Osman KM (2019) Transcervical Radiofrequency Ablation of Symptomatic Uterine Fibroids: 2-Year Results of the SONATA Pivotal Trial. J Gynecol Surg 35:345–349. 10.1089/gyn.2019.0012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Stewart EA, Cookson CL, Gandolfo RA, Schulze-Rath R (2017) Epidemiology of uterine fibroids: A systematic review. BJOG Int J Obstet Gynaecol 124:1501–1512. 10.1111/1471-0528.14640 [DOI] [PubMed] [Google Scholar]
- 5.Drayer SM, Catherino WH (2015) Prevalence, morbidity, and current medical management of uterine leiomyomas. Int J Gynecol Obstet 131:117–122. 10.1016/j.ijgo.2015.04.051 [DOI] [PubMed] [Google Scholar]
- 6.Rakotomahenina H, Rajaonarison J, Wong L, Brun J-L (2017) Myomectomy: technique and current indications. Minerva Obstet Gynecol 69:. 10.23736/S0026-4784.17.04073-4 [DOI] [PubMed]
- 7.Zhang H-L, Deng E-Y, Zhu J-E, et al (2024) Intrauterine chilled saline instillation reduces endometrial impairment on MRI after ultrasound-guided percutaneous microwave ablation of uterine adenomyosis. Insights Imaging 15:134. 10.1186/s13244-024-01707-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Liu L, Wang T, Lei B (2021) Ultrasound-guided Microwave Ablation in the Management of Symptomatic Uterine Myomas: A Systematic Review and Meta-analysis. J Minim Invasive Gynecol 28:1982–1992. 10.1016/j.jmig.2021.06.020 [DOI] [PubMed] [Google Scholar]
- 9.Beermann M, Jonsdottir G, Cronisoe A, et al (2022) Long term follow-up of uterine fibroids treated with microwave ablation: an up to 3-year observational study of volume, regrowth, and symptoms. Int J Hyperth Off J Eur Soc Hyperthermic Oncol North Am Hyperth Group 39:1158–1163. 10.1080/02656736.2022.2109764 [DOI] [PubMed] [Google Scholar]
- 10.Ierardi AM, Savasi V, Angileri SA, et al (2018) Percutaneous High Frequency Microwave Ablation of Uterine Fibroids: Systematic Review. BioMed Res Int 2018:2360107. 10.1155/2018/2360107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Xia J, Teng D, Sui G, et al (2023) Effectiveness and Safety of Ultrasound-guided Percutaneous Microwave Ablation for a Single Uterine Fibroid Greater than 300 cm3. J Minim Invasive Gynecol 30:290–299. 10.1016/j.jmig.2022.12.013 [DOI] [PubMed] [Google Scholar]
- 12.Stoelinga B, Dooper AMC, Juffermans LJM, et al (2018) Use of Contrast-Enhanced Ultrasound in the Assessment of Uterine Fibroids: A Feasibility Study. Ultrasound Med Biol 44:1901–1909. 10.1016/j.ultrasmedbio.2018.03.030 [DOI] [PubMed] [Google Scholar]
- 13.Stoelinga B, Juffermans L, Dooper A, et al (2021) Contrast-Enhanced Ultrasound Imaging of Uterine Disorders: A Systematic Review. Ultrason Imaging 43:239–252. 10.1177/01617346211017462 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kröncke T, David M (2019) MR-Guided Focused Ultrasound in Fibroid Treatment – Results of the 4th Radiological-Gynecological Expert Meeting. Geburtshilfe Frauenheilkd 79:693–696. 10.1055/a-0893-4752 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wei C, Fang X, Wang C-B, et al (2018) The predictive value of quantitative DCE metrics for immediate therapeutic response of high-intensity focused ultrasound ablation (HIFU) of symptomatic uterine fibroids. Abdom Radiol N Y 43:2169–2175. 10.1007/s00261-017-1426-7 [DOI] [PubMed] [Google Scholar]
- 16.Wang Y, Zhang P, Zhang R, An P (2019) Predictive Value of Quantitative Uterine Fibroid Perfusion Parameters From Contrast‐Enhanced Ultrasound for the Therapeutic Effect of High‐Intensity Focused Ultrasound Ablation. J Ultrasound Med 38:1511–1517. 10.1002/jum.14838 [DOI] [PubMed] [Google Scholar]
- 17.Maruyama H, Sekimoto T, Yokosuka O (2016) Role of contrast-enhanced ultrasonography with Sonazoid for hepatocellular carcinoma: evidence from a 10-year experience. J Gastroenterol 51:421–433. 10.1007/s00535-015-1151-3 [DOI] [PubMed] [Google Scholar]
- 18.(2018) Clinical application guidelines for ultrasound-guided percutaneous microwave (radiofrequency) ablation for uterine fibroids (2017). Chin J Med Ultrasound Electron Ed 15:90–94
- 19.Xia M, Jing Z, Zhi-Yu H, et al (2014) Feasibility study on energy prediction of microwave ablation upon uterine adenomyosis and leiomyomas by MRI. Br J Radiol 87:20130770. 10.1259/bjr.20130770 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Xia M, Jing Z, Zhi-yu H, et al (2014) Research of dose-effect relationship parameters of percutaneous microwave ablation for uterine leiomyomas--a quantitative study. Sci Rep 4:6469. 10.1038/srep06469 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Xue H, Wu S, Xiao K, et al (2024) Sonographic features of uterine fibroids that predict the ablation rate and efficacy of high-intensity focused ultrasound. Front Oncol 14:1451626. 10.3389/fonc.2024.1451626 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Gong X, Zhang X, Liu D et al (2022) Physician Experience in Technical Success of Achieving NPVR ≥ 80% of High-Intensity Focused Ultrasound Ablation for Uterine Fibroids: A Multicenter Study. Front Med Technol 3:790956. 10.3389/fmedt.2021.790956 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Goldberg SN, Grassi CJ, Cardella JF, et al (2005) Image-guided tumor ablation: Standardization of terminology and reporting criteria. Radiology 235:728–739. 10.1148/radiol.2353042205 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Binar M, Serindere M, Bozlar U, et al (2019) Determining the thyroid gland volume causing tracheal compression: A semiautomated 3D CT volumetry study. Med Kaunas Lith 55:143. 10.3390/medicina55050143 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Shitzer A, Kleiner MK (1980) On the Relationship Between Blood Perfusion, Metabolism and Temperature in Biological Tissue Heat Balance. J Biomech Eng 102:162–169. 10.1115/1.3138217 [DOI] [PubMed] [Google Scholar]
- 26.Poch FG, Geyer B, Neizert CA, et al (2021) Periportal fields cause stronger cooling effects than veins in hepatic microwave ablation: an in vivo porcine study. Acta Radiol 62:322–328. 10.1177/0284185120928929 [DOI] [PubMed] [Google Scholar]
- 27.28. Mathy RM, Giannakis A, Franke M, et al (2024) Factors Impacting Microwave Ablation Zone Sizes: A Retrospective Analysis. Cancers 16:1279. 10.3390/cancers16071279 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.29. Zhang M, Yin C, Jiang J, et al (2023) Application value of contrast‐enhanced ultrasonography in the treatment of uterine fibroids by high‐intensity focused ultrasound ablation: A retrospective study. J Clin Ultrasound 51:113–120. 10.1002/jcu.23294 [DOI] [PubMed] [Google Scholar]
- 29.30. Sun Y-D, Zhang H, Liu J-Z, et al (2019) Efficacy of radiofrequency ablation and microwave ablation in the treatment of thoracic cancer: A systematic review and meta-analysis. Thorac Cancer 10:543–550. 10.1111/1759-7714.12973 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.31. Jin CS, Wada H, Anayama T, et al (2016) An integrated nanotechnology-enabled transbronchial image-guided intervention strategy for peripheral lung cancer. Cancer Res 76:5870–5880. 10.1158/0008-5472.CAN-15-3196 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Dou J-P, J Yu, Yang X-H et al (2017) Outcomes of microwave ablation for hepatocellular carcinoma adjacent to large vessels: A propensity score analysis. Oncotarget 8:28758–28768. 10.18632/oncotarget.15672 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
No datasets were generated or analysed during the current study.