Magnetic Resonance guided High Intensity Focused Ultrasound (MRgHIFU) is emerging as an important tool in the cancer treatment armamentarium, where delivery of ablative thermal energy can induce precise tissue necrosis at the focus [1]. Delivery under MR guidance allows anatomical visualization of the treatment volume and MR thermometry can estimate temperature change using the proton resonance frequency shift (PRFS) method [2]. However, temperature as measured by PRFS thermometry is unreliable in fat or bone and may be confounded by patient motion [3]. This makes it crucial to determine tissue effects during the procedure and relate their extent to post-procedural imaging changes.This study aims to describe tissue changes observed during MRgHIFU to bone metastases and relate them to post-procedural and 30-day imaging appearances.

Patients: Nine patients (7 female, 2 male, mean age 55 years) with painful bone metastases underwent MRgHIFU using a 3T Achieva MR/Sonalleve HIFU clinical system (Philips Healthcare, Vantaa, Finland). Patients were positioned in acoustic contact with the HIFU device using a dampened Aquaflex gel-pad (Parker Laboratories Inc, USA) placed over the HIFU window. Once positioning was confirmed as adequate, patients were sedated and MRgHIFU treatment was delivered as a series of sonications, planned on a patient and lesion specific basis using cell sizes 4-12mm in diameter.

Imaging: Images were acquired using HIFU window and pelvis coils in combination. 3D T1-W imaging was exported to the Sonalleve for treatment planning. Diffusion-weighted images (DWI) were acquired before sonications in 8 patients (sequence not acquired in Patient 1). PRFS thermometry was performed after each sonication to evaluate temperature changes. In 7 patients, further DWI series were re-acquired during intra-procedural post-sonication cooling periods, and immediately post treatments (in Patient 4, baseline DWI was degraded by artefacts and subsequent series were not acquired). In patients 2-5, the DW sequence was still being optimised, before it was fixed for patients 6-9. In all cases however, the sequence was acquired using 3 b-values of 0, 100 and 700 s/mm² and a voxel size of 3.5x3.5x5.0mm. Final sequence parameters were: single-shot EPI, TR 9000ms, TE 72ms, SPIR and gradient reversal off-resonance fat suppression, 24 slices, 0.1mm gap, phase R-L, 3NSA, Δ=35.5ms, δ=13.4ms, scan time 4:48 minutes. On completion of treatments 3D T1-W THRIVE images were obtained after Gadolinium (Gd) contrast agent (Dotarem). Sequences were repeated in the 8 patients who reached the Day 30 post-treatment assessment to date.

Analysis: b=0,100,700 s/mm² and Gd-THRIVE images were reviewed by 2 observers in conjunction. A consensus score which described the degree of change seen in intra- and extra-osseous treated tissue regions on the DWI and Gd-THRIVE images mid-procedure, post-procedure and at Day 30 was assigned as follows: 0=no change, 1=mild change, 2=moderate change, 3=striking change. Spearman’s correlation coefficient described any relationship between changes seen across timepoints and techniques.

Lesions were treated at the following sites: shoulder (1), rib (1), ilium (4) sacrum (1), femur (2). 7/9 cases had intact bone cortex around the tumour; in 2 cortical breach was present. An increase in high signal intensity in an extra-osseous distribution along the bone surface was evident on DWI in 4/7 cases where DWI was available mid-procedure. Three of these cases went on to have clear changes on the Gd-enhanced images obtained at 30 days, which showed a non-enhancing band of extra-osseous necrosis with an enhancing rim (Figure 1); in the 4th case, 30 day images have not yet been obtained (Table 1). Correlation between mid-procedure DWI changes and immediate post-procedure DWI and Gd-THRIVE appearances was significant (r²= 0.90, p=0.006; r²= 0.87, p=0.01 respectively), as were correlations between mid-procedure DWI and 30-day DWI and Gd-THRIVE images (r²= 0.89, p=0.02 and r²=0.87, p=0.03). Changes were more difficult to visualize in the intra-osseous component; only 1 patient had mid-procedural DWI changes, which were also present on post-procedure and 30 day scans. Another patient with striking intra-osseous changes at 30 days did not have recognizable DWI changes mid-procedure. There was no correlation between post-procedural imaging changes and total treatment energy delivered.Evaluation of MRgHIFU treatment has relied on Gd-enhanced imaging, which cannot be administered during procedures. Intra-procedural monitoring with DWI provides a good indication of subsequent extra-osseous tissue change 30-days post palliative MRgHIFU for bone metastases. Further ADC analysis will establish whether this is mainly due to change in ADC or T2, which might be better assessed by higher-resolution T2-W imaging. Intra-osseous changes are difficult to recognize during treatment. Immediately post-procedure and at subsequent follow-up, Gd-enhanced images are recommended for delineating the extent of tissue damage.