The potential role of dynamic MRI in assessing the effectiveness of high-intensity focused ultrasound ablation of breast cancer

Abstract

Purpose: To retrospectively evaluate magnetic resonance imaging (MRI) features of breast cancer after high-intensity focused ultrasound (HIFU) ablation.

Materials and methods: Six patients with invasive ductal carcinomas underwent HIFU ablation. In all patients, dynamic MRI was performed prior to and two weeks after HIFU. Serial follow-up studies were performed. Changes in signal intensity and size of the index tumour in addition to peripheral enhancement patterns were evaluated. Histopathological results were compared with MRI findings.

Results: All patients had a single index tumour with a mean size 25.6 mm (range 12 to 37 mm) at the ablation time. In three of six patients, thin rim enhancement around the ablation zone was seen on the subtraction image after first ablation, which showed no change on follow-up MRI. Complete ablation was confirmed by the histopathology (biopsy in two and surgery in one). In the remaining three patients, nodular or irregular thick enhancement was shown on the subtraction image and viable tumour was confirmed by surgery and biopsy in two patients.

Conclusion: The MR characteristics of successfully ablated breast cancers included central dark signal intensities with thin rim enhancement on subtraction images. Nodular or irregular thick enhancements should raise concern of partial ablation. We propose MRI plays a critical role in assessing the effectiveness of HIFU treatment.

• breast cancer
• focused ultrasound ablation
• high intensity focused ultrasound
• magnetic resonance image
• neoplasms

Introduction

High-intensity focused ultrasound (HIFU) is a non-invasive technique used for thermal ablation of solid tumours. Based on real-time imaging modalities, it can be divided into two types: MR-guided HIFU and US-guided HIFU [1–4]. With this HIFU technique, an US beam that passes through the solid tumour within the body can be focused within a few mm³ of targeted tissue. Once focused, energy from the US beam elevates the temperature of the targeted tissue volume within a few seconds without causing damage to adjacent tissues, thereby leading to protein denaturation and tissue necrosis [5]. Histopathologically, HIFU triggers coagulation, which is characterised by the presence of epithelial necrosis featuring lack of nuclei within cells, and severely damages the tumour vessel. The first HIFU attempts were made in 2001 to treat breast cancer using MR-guided focused ultrasound ablation [3], [6]. In 2003, Wu et al. conducted a prospective, randomised clinical trial to assess the treatment effect of extracorporeal high intensity focused ultrasound systems, a type of US-guided focused ultrasound ablation, in 48 patients with invasive breast cancers [1].

Several studies have shown that HIFU is a safe and feasible modality for the treatment of breast cancer [3], [4]. However, range of successful ablation has been reported to be from 20% to 100%, depending on focused ultrasound ablation (FUA) system type, imaging technique, ablation protocol, and patient selection [4]. To confirm that HIFU ablation can be used as an alternative treatment option for breast cancer, further long-term follow-up studies on safety, technical effectiveness and survival rate are required.

Dynamic MRI is well known as the most sensitive and accurate modality for detecting invasive breast cancer, with a sensitivity range between 95% and 100% [7], [8]. MRI sensitivities for detecting residual disease after excision biopsy and breast conserving surgery were reported to be 61.2% and 77%, respectively [9], [10]. A subtraction image is both a post-processing method to highlight enhancing features, as well as a passive method for fat suppression. Dynamic MRI is a routine protocol that constitutes more than four post-contrast acquisitions for seven minutes following infusion of contrast media as well as one pre-contrast acquisition. A subtraction image is defined as subtraction of the pre-contrast image from the post-contrast images, allowing clear visualisation of the brightest regions of enhancement depending on early and delayed phases [11]. Irregular shape, spiculated margin and heterogeneous or rim internal enhancement are typical morphologic features of breast cancer [12]. Contrast agents were used in dynamic MRI to evaluate the tumour neovascularisation and the analysis of contrast enhancement curves over time was known to be useful for the differentiation of malignant and benign tumours [13]. An early rapid enhancement with a delayed washout pattern is the typical malignant kinetic curve pattern [12], [13].

Only a small number of studies have investigated imaging features including signal intensity change and enhancement pattern of breast cancer in patients who have undergone HIFU ablation [1], [14–16]. In 2003, Wu et al. described the findings of dynamic enhanced MRI which was performed following US-guided HIFU in three patients. According to this, no contrast enhancement was seen in the ablation zone containing the tumour and its adjacent normal tissue, and these lesions corresponded to coagulative necrosis on histopathology [1]. In 2005, Wu et al. performed dynamic MRI in five patients and reported that peripheral thin rim enhancement was present around the ablation zone without any contrast-enhancement [15]. In 2003, Gianfelice et al. described findings of MRI following an MR-guided HIFU treatment [14]. In the fatty breast group, the tumour was noted to have hypointense signal in both T1- and T2-weighted images. Additionally, there were peri-lesional hypointense halos featuring the necrosis of the adjacent tissue. Of these, five patients underwent dynamic MRI. Residual tumour was present in three cases with persistent enhancement, while complete destruction of the tumour was achieved in two cases without any enhancement. These above studies agreed with findings on contrast-enhanced T1-weighted image following HIFU treatment. In association with HIFU treatment, only one study had reported subtraction images, but dynamic images were obtained prior to and following MR-guided HIFU to plan proper treatment strategies [17].

To our knowledge, no reported studies have examined the time-dependent changes of MRI findings following HIFU treatment using subtraction images. Here we defined the MRI features of dynamic MRI including subtraction images in patients who had undergone HIFU ablation for breast cancer and compared them with the histopathological findings. We also evaluated the interval changes in these patients during the follow up.

Materials and methods

Subjects

This study was approved by the institutional review board, and informed consents were obtained from all patients. Between March and December 2006, breast cancer patients who underwent HIFU were included in this study. The selection criteria for HIFU were as follows: a histologically proven invasive cancer of small or intermediate size (less than 30 mm or 30–50 mm, T1 to T2); a single cancer without multifocality or multicentricity; a location at least 10 mm from the skin or rib cage, and more than 20 mm from the nipple; a lesion boundary visualised with an US image; no evidence of distant metastases. Chemotherapy was performed in cases of large tumour sizes greater than 50 mm, and the MRI taken after the completion of three cycles was re-evaluated as to whether the index tumour met the selection criteria. Patients with distant metastases or those otherwise not fitting the selection criteria were excluded.

HIFU therapy

Patients enrolled in this protocol were heated using an extracorporeal, ultrasound-guided focused Model-JC tumour therapy system (Haifu Technology, Chongqing, China). The breast lesions were ablated under real-time US guidance using a 3.5 MHz diagnostic ultrasound transducer (Xario, Toshiba Medical System, Otawara, Japan). Therapeutic US energy was produced by a 155-mm diameter transducer with a central hole of 60 mm for the coaxial imaging transducer.

All patients were under general anaesthesia to ensure immobilisation during the lengthy procedure and to prevent superficial skin pain. Patients were laid prone on a table, placing both breasts in degassed water.

MRI protocol

Dynamic MRI was performed in all patients prior to and two weeks after HIFU treatment. In cases with complete ablation, follow-up dynamic MRIs were recommended. The follow-up was performed between the second and fourth months following HIFU treatment, then at 4- to 6-month intervals within the first year and at 5- to 8-month intervals after 1 year. The term ‘MRI post-HIFU’ refers to the MRI done 2 weeks after the ablation. The term ‘FU MRI’ collectively refers to all other follow-up MRIs done at the earliest 2 months and at the latest a few years after the ablation. Breast MRI was obtained with 1.5 T (Signa Excite; GE Medical Systems, Milwaukee, WI) using a dedicated breast coil (OBC-63, MRI Devices (now Invivo), Orlando, FL). MRI was performed using the following sequences: axial, fat-suppressed, fast spin-echo T2-weighted imaging (TR/TE = 4500/81 ms, a flip angle of 90°, 34 slices with FOV (300 mm), matrix (320 × 224), 0 NEX and 4 mm section, acquisition time of 3 min 5 s); pre- and post-contrast, sagittal T1-weighted three-dimensional, fat-suppressed, fat-spoiled gradient-echo sequence (TR/TE = 7.4/4.2 ms, flip angle 10°, 3 mm section thickness, acquisition time of 8 min 14 s) was obtained before and 99, 198, 297, 396, 495 sec after rapid bolus injection of 0.2 mmol/kg body weight of Gd-DPTA (Magnevist, Schering, Berlin). Subtraction images were obtained by subtracting the pre-contrast image from the serial post-contrast images on a pixel-by-pixel basis and subtraction images at the first and fifth acquisitions were used as early and delayed images, respectively. To obtain the kinetic curve, regions of interest (ROI) were placed on the most enhancing pixels.

MRI analysis

The MR images were retrospectively reviewed by the consensus of two radiologists who specialise in breast imaging. MRI prior to HIFU and follow-up MRIs were reviewed with regard to changes in signal intensity in T1- and T2-weighted images of the index tumour, changes in tumour size, and dynamic enhancement patterns. The tumour signal intensity was compared to that of the breast parenchyma, while tumour size was measured according to the long diameter. Rim enhancement patterns were classified as thin rim, and nodular or irregular thick enhancements on the early and delayed subtraction images. Kinetic curve patterns were classified as rapid, intermediate and slow enhancement on early phase and wash-out, plateau and persistent type on delayed phase [12]. Additional MR imaging findings, including HIFU-related mammary oedema and complications, were also evaluated.

Response evaluation of HIFU

The index tumour was defined as the initially identified tumour prior to ablation [18]. We classified the tumour response according to the criteria described by Goldberg et al. [18], [19] (Figure 1): complete ablation comprised those cases with no enhancement of the index tumour and thin rim enhancement present upon subtraction imaging of MRI post HIFU. In these cases, a follow-up MRI was recommended. Partial ablation comprised those cases with nodular or irregular thick enhancement at the ablated tumour periphery on subtraction imaging of MRI post HIFU. In these cases, an US-guided biopsy and a second HIFU session were recommended.

Figure 1. Diagrams of technical effectiveness. (A) Illustration of HIFU therapy. (B) Illustration of the technical effectiveness as divided into complete versus partial ablation. (C) Illustration of the technical effectiveness as depicted on a MR subtraction image: thin rim enhancement for complete ablation and nodular or irregular thick enhancement for partial ablation.

Pathologic correlation

To evaluate whether dynamic MRI findings are indicative of HIFU ablation effectiveness, the MRI findings were correlated with the pathology for cases in which a pathological review was available following an US-guided core needle biopsy or surgical mastectomy. Pathologic changes of HIFU-treated tumour that were not enhanced on dynamic MR imaging were also examined. Tumour pathology was reviewed, focusing on the presence of viable tumours based on rim-enhancement patterns.

Results

Patients and HIFU therapy

Six patients (mean age, 62.1 years; range, 46–68 years) opted for minimally invasive therapy rather than surgery. Disease TNM stages included two cases each of stage I, stage II, and stage IIIA with tumour sizes ranging from 12 to 70 mm. The location of the tumour was the right upper outer quadrant (n = 1), left upper inner quadrant (n = 2) or the left upper outer quadrant (n = 3). In four patients neoadjuvant chemotherapy was performed prior to HIFU, at which time the mean size of the index tumour was 25.6 mm (range 12 to 37 mm).

The main parameters during HIFU treatment are summarised in Table I. The HIFU treatments were configured as follows: therapy frequency of 0.8 MHz, a mean focal field diameter of 1.1 mm, focal field length of 9.8 mm, focus distance of 135 mm, and therapy power of maximal 350 W. The entire target mass was ablated in multiple 5 mm sections and included a 10 mm minimum ablation margin. Treatment times ranged between 1 h 20 min and 4 h 45 min, with a mean time of 2 h 51 min.

Table I.  Ablation parameters of HIFU.

Pt.No.	Age	Size (mm)^#	First HIFU					Second HIFU
			Date	Power (W)	No. of 5-mm sections*	Procedure time	Sonification time (s)	Date	Power (W)	No of 5-mm sections*	Procedure time	Sonification time (s)
1	49	12	Apr	120–182	8	2 h 35 min	2038	–
2	47	37	Jun	100–240	13	2 h 50 min	1942	–
3	52	33	Aug	80–220	12	4 h 45 min	3677	–
4	68	15	Mar	100–240	9	1 h 20 min	1942	–
5	60	35	Aug	120–180	13	3 h 50 min	2845	Dec	100–200	11	3 h 15 min	2074
6	46	22	Sep	200–220	10	2 h 15 min	2088	Dec	140–180	10	2 h 25 min	3831

Pt no., patient number.

^#Mass size was the longest diameter of the tumour on dynamic MRI at the time of ablation.

*Number of 5–mm sections needed to ablate the index tumour and margin.

MRI findings and pathology were summarised in Table II. Four patients had one session of HIFU treatment and two patients underwent two sessions of HIFU at three-month intervals. A total of eight HIFU sessions were performed. In all patients, dynamic MRI was performed prior to and two weeks after the HIFU. Serial follow up was recommended when a complete ablation was suggested by the MRI post-HIFU at two weeks. In four patients (patients 1, 2, 3 and 6), the mean MRI follow-up duration was 67 weeks (range of 8–120 weeks), and the mean frequency of MR scanning was four times (range of 1–7 times). In three of the six patients (patients 1, 2 and 3), the complete ablation of the index tumour was achieved by the first HIFU session. The first two patients (Table II) were clinically stable, and there was no evidence of viable tumours in the US-guided biopsies or presence of remnant tumours until the 24- or 30-month MRI follow ups. Patient 3 had complete ablation of the tumour, which was confirmed by surgical pathology, although a multifocal malignancy was detected in an area far from the index tumour at the 11-month follow up. In three patients (patients 4, 5 and 6), partial ablation left a viable tumour after the first HIFU session. The remaining viable tumour was confirmed by US-guided core needle biopsies in patients 4 and 5, and second sessions of HIFU ablation were recommended 3 months later. However, patient 4, who had severe mammary oedema, refused the second HIFU session, in which case surgery was performed. Patient 5 underwent surgery due to the presence of nipple depression during the second HIFU session, and a viable tumour was observed upon histopathological examination. In patient 6, surgery was performed due to the occurrence of a skin defect, although thin rim enhancement was present on the MRI scans taken 2 weeks and 8 weeks following the second HIFU session. A histopathological examination showed no viable tumour mass in this instance. Therefore, four out of a total of six patients (patients 1, 2, 3 and 6) displayed complete ablation, while patients 4 and 5 had partial ablation after HIFU treatment.

Table II.  Summary of MRI findings and pathology after HIFU procedure.

Pt. No.	1st HIFU/2nd HIFU			FU MRI
	MRI post-HIFU	Bx or Op Pathology	2nd HIFU	FU period (months)	Enhancement	Index tumour size	Pathology
1	Thin rim	–	N/A	3,5,7,10,18,25,30	Thin rim	12 → 5 mm (12 M)	No viable tumour at Bx (20 M)
2	Thin rim	–	N/A	2,6,12,18,24	Thin rim	37 → 15 mm (24 M)	No viable tumour at Bx (8 M)
3	Thin rim	–	N/A	2,4,11	Thin rim	33 → 26 mm (11 M)	No viable tumour at ablated lesion and multifocal IDC apart from index mass at Op (11 M)
4	Nodular	Viable tumour at Bx	N/A	–			Viable tumour at Op (3 M)
5	Nodular/*	Viable tumour at Bx/–	Yes	–			Viable tumour at Op (1 M after 2nd HIFU)
6	Nodular/thin rim	–/–	Yes	2 after 2nd HIFU			No viable tumour at Op (3 M after 2nd HIFU)

Pt, patient; nodular-nodular or irregular thick enhancement; Bx, US-guided core needle biopsy; Op, mastectomy; IDC, invasive ductal carcinoma; M, month; N/A, not applicable.

The MRI post-HIFU was performed at two weeks after the HIFU, and the FU MRI was performed serially after 2 or 3 months following the HIFU.

*MRI after a second HIFU session was not performed.

MRI findings prior to HIFU

All patients had a single index tumour (mean size, 25.6 mm; range, 12–37 mm) at the ablation time and did not have a concurrent multifocal or multicentric malignancy. Index tumours were classified based on the BI-RADS lexicon [12], all of which were mass lesions (Figures 2A and 3A). The BI-RADS classifications were all Category 6, known biopsy-proven malignancies. The shapes of the index tumours were irregular (n = 3), lobular (n = 2) and oval (n = 1) with borders that were spiculated (n = 4) and irregular (n = 2). Internal enhancements included inhomogeneous enhancement (n = 3), internal septal enhancement (n = 1), rim enhancement (n = 1) and homogeneous enhancement (n = 1).

Figure 2. Dynamic MRI findings of patient 1 mentioned in Table II. (A) An early post-contrast sagittal T1-weighted image prior to HIFU showed a 12 mm, irregular enhancing mass (arrow). (B) A pre-contrast axial T1-weighted image of MRI post-HIFU revealed an isointense index tumour (arrow). (C) An early post-contrast sagittal T1-weighted image of MRI post-HIFU revealed an index tumour, isointense to adjacent parenchyma (arrow). It was surrounded by an ablation zone of decreased signal intensity (arrowheads). (D) The ablation zone on a delayed subtraction image of MRI post-HIFU was demarcated by dark signal intensity. Although no enhancement of the index tumour was observed, an incomplete thin rim enhancement was present (arrowheads). (E) An early post-contrast sagittal T1 weighted image of 12 months FU MRI, the size of the index tumour (arrow) had decreased to 5 mm. (F) The size of the ablation zone surrounded by a thin rim enhancement (arrowheads) had also decreased on a delayed subtraction image of 30 months FU MRI.

Figure 3. Partial ablation in a 68-year-old woman (patient 4 mentioned in Table II). (A) An early post-contrast sagittal T1-weighted image prior to HIFU showed a 15 mm irregular enhancing mass (arrow). (B) An axial fat suppressed T2-weighted image prior to HIFU revealed an isointense index tumour (arrow). (C) An axial fat suppressed T2-weighted image of MRI post-HIFU showed an isointense index tumour (arrow) with mammary oedema. (D) An early post-contrast sagittal T1-weighted image of MRI post-HIFU revealed an index tumour, isointense to adjacent parenchyma (arrow) surrounded by an ablation zone of decreased signal intensity (arrowheads). Irregular enhancement was seen at the periphery of the ablation zone (black arrow). (E) A central dark signal intensity and a nodular or irregular thick enhancement (arrow) was observed on the early and delayed subtraction image of MRI post HIFU. (F) Kinetic curve pattern showed early rapid enhancement and delayed persistent pattern. (G) Photomicrograph of a mastectomy specimen showing prevalent coagulative necrosis and a small portion of viable tumour (arrows) (H and E, original magnification × 40).

MRI Post-HIFU findings (two weeks after the first HIFU ablation)

Index tumour. The index tumour on T1- and T2-weighted images showed isointense signals relative to the breast parenchyma, and no changes in signal intensity were observed when compared to pre-HIFU MRI. Furthermore, no changes in the size of the index tumour were observed on T1- and T2-weighted images (Figures 2B, 3B and 3C). Delayed contrast-enhanced sagittal T1-weighted images displayed isointense signals in the index tumour relative to breast parenchyma, obscuring the degree of ablation (Figures 2C and 3D). On the subtraction image, however, the ablation zone containing the index tumour had noted dark signal intensity distinct from the adjacent breast parenchyma (Figures 2D and 3E). The long diameter of the ablation zone was an average of 47.5 mm (range, 35–62 mm). When HIFU was performed, the size of the index tumour ranged from 12 mm to 37 mm, and the ablation zone was larger than the index tumour by an average of 21.5 mm.

Peripheral enhancement. On the subtraction image, enhancement was found only in the ablation zone periphery containing the index tumour. This was noted to be a thin rim enhancement and nodular or irregular thick enhancement pattern (Table II).

Complete or incomplete thin rim enhancement less than 2 mm in thickness was seen on subtraction images in three cases, in which follow-up MRI was recommended (Figures 2 and 4). Thin rim enhancement was faint on the early and delayed subtraction image (Figure 2D). Nodular or irregular thick enhancement was seen in three cases (5, 8, and 14 mm in patients 4, 5 and 6, respectively), all of which had an enhancement size >5 mm. These nodular or irregular thick enhancements appeared clearly on the early subtraction images, and all showed early rapid enhancement in the kinetic curve analysis (Figures 3 and 5). One patient had a delayed persistent pattern (Figure 3F), and two had delayed washout kinetic patterns (Figure 5B).

Figure 4. Complete ablation and mammary oedema in a 52-year-old woman (patient 3 mentioned in Table II). (A) On an axial fat suppressed T2-weighted image of MRI post-HIFU, mammary oedema (arrow) was accompanied by skin and trabecular thickening and thermal injury was observed as irregular hyperintensity within the chest wall muscle (arrowhead). (B) On an axial fat suppressed T2-weighted image of 6 months FU MRI, mammary oedema and chest wall muscle change almost disappeared. (C) The subtraction image of 11 months FU MRI showed a central dark signal intensity and a thin rim enhancement (arrow). Thin rim enhancement was faint on early subtraction image, but was more prominent on the delayed subtraction image. (D) Photomicrograph of mastectomy specimen shows coagulative necrosis (arrows) and peripheral foreign body reaction (arrowheads) (H and E, original magnification × 1). No viable tumour was present in the ablation zone

Figure 5. Partial ablation in a 60-year-old woman (patient 5 mentioned in Table II). (A) A central dark signal intensity and a nodular or irregular thick enhancement (arrow) was observed on the early subtraction image of MRI post HIFU. (B) Kinetic curve pattern showed early rapid enhancement and delayed washout pattern.

Additional findings. Mammary oedema was present in all HIFU-treated patients. On T2-weighted images, skin thickening and trabecular thickening of reticular hyperintensity were shown within the breast parenchyma (Figure 4A). In all patients, injury to the pectoralis major muscle occurred upon ultrasound beam passage. On T2-weighted images, heterogeneous hyperintense signals were observed (Figure 4A).

Sequential FU MRI findings (more than 2 or 3 months after the ablation)

Follow-up MRI more than 6 months after HIFU treatment was conducted in three patients. The size of the index tumour had decreased since the 6 months following the HIFU treatment. In patient 1, the size of the index tumour had decreased from 12 mm to 5 mm at 12 months (Figure 2E) and the index tumour was not visible thereafter, as it could not be differentiated within the ablation zone. In patient 3, it had decreased from 33 mm to 26 mm at 11 months and in patient 2, an index tumour of 37 mm had decreased to 15 mm at 24 months.

The size of the ablation zone had also decreased in the 6 months following HIFU treatment. The long diameter of the ablation zone had decreased by 66% in patient 1 (from 35 to 12 mm) at 30 months (Figure 2F), 55% in patient 2 (from 55 to 25 mm) at 24 months and 33% in patient 3 (from 58 to 39 mm) at 11 months.

Three cases showed no change in thin rim enhancement of the index tumour from 11 to 30 months. Thin rim enhancement was more clearly visualised on the subtraction image as compared with the MRI post-HIFU. Thin rim enhancement was more prominent on the delayed subtraction image as compared with the early subtraction image (Figure 4C). The size of the index tumour had decreased with time, but the thickness of the enhancing rim either had not changed in patient 2, 3 or had increased in a concentric manner in patient 1. An US-guided biopsy was performed in patient 1 at 20 months and histopathology showed no viable tumour in the index tumour and peripheral rim. There was little change in rim enhancement to 30-month follow-up point.

A signal change on T2-weighted images occurred as a result of mammary oedema, and injury to the pectoralis muscle disappeared 6 months following HIFU treatment (Figure 4B).

MRI and pathologic correlation

Four of six patients underwent surgery. Two (patients 3 and 6 mentioned in Table II) displayed thin rim enhancement upon MRI, and the index tumour was completely ablated with no viable tumour remaining. Upon surgical histopathology, the index tumour was characterised by the presence of coagulative necrosis featuring the epithelial necrosis where no nuclei were present within the cells. An enhancing thin rim corresponded to fibrosis with foreign body reaction (Figure 4D).

Patients 4 and 5 displayed nodular or irregular thick enhancements on MRI. The viable tumour was left in the index tumour periphery, which was noted to have coagulative necrosis upon histopathological examination (Figure 3G).

Patient 1 underwent US-guided core biopsy to evaluate the increased rim thickness and histopathology showed a chronic inflammation with marked fibrosis and fat necrosis.

Discussion

Surgical management represents a conventional treatment regimen for breast cancer, ranging from local excision to total mastectomy depending on the size of mass and its location [16]. Various attempts have been made to replace the surgery by inducing local tumour necrosis with minimally invasive or non-invasive techniques, including cryotherapy, radiofrequency ablation, laser interstitial therapy, microwave and focused ultrasound ablation [2].

In the current study we showed that subtraction images clearly visualised subtle patterns of peripheral enhancement of coagulated lesions, as well as the ablation zone containing the index tumour when compared to contrast-enhanced T1-weighted images. We determined whether follow-up imaging or US-guided biopsies should be performed based on peripheral enhancement. In addition, we determined whether tissue biopsies or a second HIFU session should be employed based on the location of nodular or irregular thick enhancements. In cases of nodular or irregular thick enhancement, a viable tumour was still present due to partial ablation in the periphery of the index tumour. Other study has also reported that residual tumours are often located in the tumour periphery [14], [20].

Based on our histopathological findings, thin rim enhancement corresponded to fibroblast proliferation surrounding regions of coagulative necrosis, and foreign body reaction provoked by macrophages. No viable tumour was observed. Wu et al. also noted that rim enhancement was an inflammatory reaction in response to thermal ablation [14], [21]. We found that the size of the index tumour undergoing coagulative necrosis as well as mammary oedema had decreased with time. However, thin rim enhancement did not change or showed a concentric thickening pattern over time. Treated volumes were decreased over time because of the replacement of the necrotic region with fibrous scar tissue [22].

In the present study, the range of complete ablation including first and second HIFU ablation was 66.6% and complications rates were high. These results were probably due to patient selection and inherent properties of US guidance. There is no inclusion guideline of focused ultrasound ablation, but single breast cancer at least 15 mm apart from the overlying skin and chest wall and not larger than 50 mm has been recommended in previous reports [1], [23]. To consistently achieve complete radiofrequency ablation of breast cancer, van der Ploeg et al. reported that small cancers less than 20 mm were selected [24]. But breast cancers larger than 20 mm were included in our studies (4/6, 66.7%).

Due to mammary oedema associated with ablation, the target lesion may have moved during the procedure. But, inherent shortcoming of US guidance disturbed the accurate delineation of the mass margin because the gas bubbles formed in the ablated tissue reflect the ultrasound waves and make posterior acoustic shadowing during the ablation [4]. Because of these factors, the focused US beam may slightly miss the target in case of partial ablation even if the target area included the index mass and a 10 mm minimum ablation margin.

The current study has several limitations including the small number of enrolled patients, which prevents generalisation of conclusions. MRI evaluation of the ablated lesion was done qualitatively and so lacked an objective description such as signal intensity value, or perfusion parameters. For the guidance we used a 3.5-MHz transducer, which only has about 1/3 of the spatial resolution of a high-resolution breast ultrasound. Additionally, we examined an MRI follow-up period of only 30 months, which is not long enough to predict the outcome of HIFU treatment or to determine the time period required for complete disappearance of the ablated tumour. Furthermore, no authorised consensus has been reached regarding whether HIFU ablation is an alternative therapeutic modality to surgery.

Although our study did not discuss treatment outcomes, we were able to describe MRI findings following HIFU treatment which had not previously been reported.

Conclusion

MRI findings from successful HIFU-treated patients featured subtraction images that included dark signal intensity with thin rim enhancement. Partial ablation was indicated by the presence of nodular or irregular thick rim enhancement. Follow-up MRI findings included the serial decrease of the size of the index tumour and the ablation zone and disappearance of mammary oedema. Thus, dynamic MRI including subtraction images provides critical information for the assessment of treatment effectiveness.