The different blood supplies of residual fibroid (RF) can affect treatment outcomes and long-term prognosis. Currently, few studies focus on the predictability of CE-MRI RF-myometrium SI ratio for the RF after HIFU treatment.

Univariate analysis showed differences in fibroid diameter, volume, RFV, NPVR, and fibroid position within the uterus between the regrowth and non-regrowth groups. Our finding that the cutoff value of NPVR was 76% is consistent with the results of previous research [23, 24]. The higher the NPVR is, the less chance of RF to regrow and require re-intervention after HIFU ablation [11]. Previous studies have shown that patient age is an essential factor affecting fibroid regeneration [25]. Younger patients are more likely to need re-intervention, but it is also believed that NPVR ≥ 80% is an effective measure to reduce re-intervention in young patients [26]. In our study, there was no difference in age between the two groups. A possible reason for this is that the median NPVR of the RF non-regrowth group reached 90%, so this factor was effectively eliminated.

In FIGO classification, significant differences in RF regrowth were found between the groups of type 2–5 and cervical fibroids. Their RFs, with SI ratios of 1.1 (0.949–1.22) for type 2–5 and 1.184 (0.944–1.28) for cervical, were more prone to regrowth, as indicated by higher CE-MRI SI ratios. Therefore, we speculated that these two FIGO types have a richer blood supply compared to others.

RF thickness and RF-myometrium SI ratio of CE-MRI are independent factors affecting RF regrowth and are related to multiple indicators. Post-treatment RF thickness affects fibroid regrowth [7, 8]. It was positively correlated with CE-MRI RF-myometrium SI ratio. Therefore, we speculated that certain fibroids with rich blood supply can be difficult to ablate and prone to regrowth. In the regrowth group, the increase in RFV of the fibroids was negatively correlated with the reduction in NPV, while in the non-regrowth group, RFV was positively correlated with the reduction in NPV. This may suggest that the fibroids in the regrowth group are surrounded by a relatively richer blood supply.

The CE-MRI of uterine fibroids is closely related to their histopathological blood supply. When the RF SI is higher than that of the myometrium SI during the arterial phase, it means that RF has more blood supply than the myometrium, indicating maybe more than one supplying artery or even uncommonly vascular variations [13]. DSA shows that the blood supply to the uterine fibroids is different [10, 14, 27, 28]. We speculate that variations in uterine fibroid blood vessels affect the order of perfusion; therefore, RFs may also have different or multiple blood supplies. The RF exhibited a higher SI than the myometrium, indicating more abundant blood supply. If the ratio of RF-myometrium is higher than the cutoff value of 0.968, the RF is likely to regrow, while those below this value may shrink or even disappear.

Preoperative T2WI and CE-MRI in the treatment of fibroid regrowth have been widely explored in previous studies [8, 11, 18, 28]. The T2WI was thought to reflect the composition of the fibroid [20, 21, 29,30,31,32,33]. It is generally believed that hyperintense fibroids on the T2WI and high enhancement on CE-MRI are prone to regrowth, with the latter being an independent factor for fibroid regrowth, as confirmed by our research [16, 34, 35]. In our study, there were 123 cases (75%) of isointense fibroids before treatment, of which 59 cases (64.8%) had fibroids regrowth. However, the Funaki classification could not predict the regrowth of isointense fibroids owing to its visual inspection. Our results showed that the CE-MRI SI ratio 1 day after treatment could be a better predictor than the Funaki classification. We comprehensively compared the Funaki classification and fibroid-myometrium SI ratio of the T2WI before and after HIFU treatment with the CE-MRI RF-myometrium SI ratio. Observing the RFs through CE-MRI other than the entire fibroid through T2WI after HIFU treatment can more accurately indicate the regrowth of the RF.

Radiomics is a technique for extracting high-throughput data [36]. Previous studies have used radiomics to predict the RF regrowth through T2WI [7, 37, 38]. However, extracting image data requires complex post-processing support [39]. Our study found that the blood supply of RFs with higher ratios has multiple blood supplies and is more prone to regrowth through the CE-MRI RF-myometrium SI ratio, and the long-term prognosis can be evaluated 1 day after treatment. This may also help solve the problem that follow-up usually depends on MRI examination, which is more convenient and economical.

It has been reported that fibroid initially decreases in volume significantly after HIFU treatment [17]. We found that in the RF regrowth group, only 14 (15.4%) of 91 patients had enlarged fibroids after 1 year. However, the remaining 77 (84.6%) patients experienced a significant increase in RF volume, and the RF regrowth group exhibited a higher SI ratio. Long-term follow-up revealed a significant positive correlation among RF regrowth, the SI ratio of RF-myometrium, and the number of re-interventions, with 45 (52.9%) of patients in the RF regrowth group requiring additional treatment. Therefore, the CE-MRI RF-myometrium SI ratio can reflect the re-intervention for a long term.

To validate the predictive ability of the CE RF-myometrium SI ratio, we conducted an external validation. Then we found results showed that the higher the RF-myometrium ratio is, the more likely it is to regrow quickly. By using ROC curve analysis, the CE-MRI RF-myometrium SI ratio can be a better predictor in the external validation group than in the internal validation group. However, it generally reflects the trend of the RF-myometrium ratio in predicting RF regrowth. Because of the small sample size of the external validation group and the complex factors for RF regrowth, this finding still needs to be further confirmed. Nevertheless, in predicting regrowth, the RF-myometrium ratio was more accurate and reliable than the preoperative and postoperative T2WI and the Funaki classification.

Our study had limitations. All the data of this retrospective study were obtained only from two medical centers. More data from multicenter studies with larger sample size are still needed to confirm our findings. In addition, other quantitative parameters, such as the time-intensity curve of DCE-MRI, could be introduced in this analysis for better results. So far, the relationship between the CE-MRI SI ratio, which reflects RF blood supply characteristics, and its regrowth provides a more preferable reference for clinical practice.