Twenty-six patients in our hospital were recruited for this prospective, nonrandomized study from May 31, 2018, to December 30, 2019, with the approval of the institutional ethics committee and written informed patient consent.

The inclusion criteria were as follows: (1) core-needle biopsy-proven unifocal breast cancer by ultrasound; (2) largest breast tumor diameter of 2.0 cm or less and tumors with a lump-like appearance, unifocal as confirmed by MR enhanced imaging; (3) at least 1 cm between the tumor and the skin surface and pectoralis muscle; (4) no enlarged lymph nodes or distant metastases in ultrasonic axillary examination and computed tomography (CT) staging.

The exclusion criteria included the following (1) pregnancy or lactation; (2) the presence of more than one lesion in the breast; and (3) patient’s preference to undergo breast-conserving surgery.

Immunohistochemical analysis was performed for these patients, including hormone receptor and human epidermal growth factor receptor 2 (HER2) status, prior to percutaneous microwave coagulation (PMC) using tissue from a core-needle biopsy specimen. Meanwhile, the histological subtype of breast malignancies was also analyzed.

The microwave delivery system consisted of a microwave generator, a flexible coaxial cable, and an internal water-cooled shaft antenna. The microwave radiation frequency was 2450 MHz, and the size of the probe was 1.8 mm × 180 mm (Vison-China Medical Devices R&D Center, Nanjing, China). The output power ranged from 10 to 100 W, and the ablation pattern of microwave generator was shown in the Supplementary material (Figure S1), 40 W for 4 min was selected in this study according to previous PMC studies and our own experience [14].

The disposable ablation needle was connected to three tubes: the microwave cable, which projected to the microwave generator, and both the entrance and exit of the water-cooling system with circulation in a 500-mL bottle of normal saline (Fig. 1a).

a Ablation setup: The disposable ablation needle is connected to a cooling water circulation system (two blue coils). b Needle setup for guided ablation. c Temperature-measuring electrode. d Setup for use of the temperature-measuring electrode or for preoperative biopsy

DynaCAD (Version 2.0, Invivo Corporation, FL, USA) has an extensive set of computer-aided detection (CAD) tools for performing real-time analysis and interventional procedure planning using MR imaging (MRI) patient exam data.

The grid or the post-and-pillar targeting approach was used to precisely locate the target lesion multidimensionally (horizontal distance, vertical distance, needle depth, angulation). Both the breast coil and biopsy immobilization/grid plates were properly oriented and secured on the MR scanner patient table.

Preoperative MR evaluation and real-time MR-guided percutaneous treatment of breast cancer were performed by one radiologist with over 21 years of experience in interventional breast MR. Both preoperative MR and real-time MR images were performed in the prone position in all women, using an MRI scanner (3.0 T; Achieva; Philips Healthcare) with a breast coil of 7-channel phased array. Parameters of T1WI, T2WI, and T1W contrast-enhanced sequences are in Table 1.

MR was used to identify the lesion, and an image in the maximum plane of the lesion was acquired (the long and short axes of the image can be measured as a1 and b1, respectively). All procedures were carried out in the MR examination room with the patient prone on the operation table and following the subcutaneous injection of local anesthesia (covering the subcutaneous tissue, target tumor, and surrounding tissue with 0.5–2% lidocaine, according to the lesion location with MR images), which was less likely to produce pain.

With the help of the ablation needle guide set (Fig. 1b), after precisely locating the lesion with real-time unenhanced T1-weighted (T1W) and T2-weighted (T2W) MR images (Fig. 2), the tip of the disposable ablation needle was advanced into the tumor to the opposite edge so that the microwave emission area could precisely cover the entire tumor (Fig. 3). After testing the cold-water cycling system, under real-time MR monitoring, when the ablation zone fully covered the target tumor with a sufficient safety margin of > 5 mm on MR images, the PMC process was terminated. If the ablation zone (manifesting as a hyperintense area on unenhanced T1WI and a hypointense area on T2WI) was considered insufficient, the procedure was repeated with proper adjustment such as extending the ablation time or redirecting the ablation needle.

A 58-year-old woman with invasive ductal carcinoma (medial-upper quadrant, 1.7 × 1.6 cm), preablation. a Unenhanced T1WI of the breast lesion, angulated axial view. b T2WI of the breast lesion, angulated sagittal view

a The white marker indicates the entry point of the antenna in DynaCAD. b Locating the lesion in clinical practice according to DynaCAD software. c The tip of the disposable ablation needle gradually passed through the largest diameter of the lesion. d Sagittal view of the needle path, indicated by the white arrow

Just when the procedure was finished, postprocedural T1W contrast-enhanced images (use contrast media only this time) were acquired to evaluate technical results and the creation of a satisfactory coagulation zone: hypointensity in the treatment zone and enhancement of nodular and circular shape (a fine enhancement) in the peripheral tissue due to congestion after treatment. A special antenna with a temperature sensor (Vison-China Medical Devices R&D Center, Nanjing, China) was used to consistently measure the temperature 5 mm from the site of radiation to minimize geothermal damage to the surrounding normal tissue (Fig. 1c, d). The method of application was similar to the placement of an ablation needle.

After PMC, each patient was sent to undergo prescheduled mastectomy (Fig. 4a) to obtain pathological specimens, which were used to confirm the complete loss of tumor biological activity. After surgery, the specimen was cut open along the needle tract (Fig. 4b), and two sections were taken to obtain the full maximum cross-section of the ablated tumor (Fig. 4c). In addition to standard histological assessment with hematoxylin–eosin (H&E) staining, nicotinamide adenine dinucleotide, reduced (NADH)-diaphorase staining of the tumor tissue was performed to assess tumor viability; staining of the breast tumor tissue obtained by preoperative biopsy was also performed for comparison. First, one section of the breast tissue specimen was fixed in formalin and embedded in paraffin. The other section was snap-frozen in liquid nitrogen (- 80 °C) and stained with NADH-diaphorase. Tumor cells that lost their vitality due to ablation would not retain staining with a dark blue color; however, their outlines could still be depicted on H&E staining. The long and short axes of the tumor in the H&E staining and NADH-diaphorase negative staining regions were defined as a2 and b2 and a3 and b3, respectively. The formula π × A × B/4 was used to calculate the tumor area of each slice; if the area of the NADH-diaphorase-negative region fully covered the H&E staining region in the tumor, the ablation was considered complete. Breast tumor tissue obtained by preoperative biopsy was used to demonstrate the reliability of pathological staining. The complications and self-reported sensations of all patients during and after percutaneous treatment were also assessed. After the endpoint of evaluation after breast surgery, sentinel lymph node biopsy (SLNB) was performed with ultrasound guidance and the surgeon carried out corresponding treatment.

a Preoperative positioning and preparation. b Macroscopic evaluation of the treatment zone before surgical resection (white arrow = tumoral ablation zone; black arrow = breast fat around the ablation zone). c Macroscopic evaluation of the treatment zone on the specimen (white arrow = tumoral necrosis zone; black arrow = ablation zone of peritumoral breast fat, approximately 0.5–1.0 cm in size). The blue arrow indicates the hyperemic ring around the ablated area

Numerical data are reported as the mean ± standard deviation. The Wilcoxon rank-sum test was performed to assess the difference between the maximum area of the tumor in H&E-stained sections and on MRI, and the paired t test was used to assess the difference between the tumor area in H&E and NADH-diaphorase negative staining area. All statistical analyses were performed by using SPSS 22.0 software, and p < 0.05 was considered to signify statistical significance.