Introduction

High-intensity focused ultrasound (HIFU) ablation under US guidance (USgFUS) has been widely validated for the local treatment of both benign and malignant tumors in abdominal organs [1, 2], becoming an accepted clinical alternative to conventional ablative techniques in patients with liver and pancreatic cancers [3–6]. Recently, MR-guided HIFU devices (MRgFUS) received FDA and CE approval for clinical treatment of uterine fibroids and bone tumors [7] and are under preclinical or clinical trials [8] for other applications. MR guidance offers superior anatomic image quality as well as the capability to quantify in real-time HIFU-induced temperature variations of targeted tissues [9, 10]. On the other hand, whereas USgFUS allows real-time visualization of target motion with adequate compensation of organ movements during the ablation [3], MRgFUS currently is facing decisive challenges in this field of application [11]. The purpose of this preliminary study was to evaluate the feasibility and safety of MRgFUS for treatment of selected pancreatic and liver tumors, as well as to discuss limitations and future perspectives of this technique.

Materials and Methods

Patients and Lesions

This preliminary feasibility study received institutional review board approval. Inclusion criteria were: biopsy-proven unresectable primary liver or pancreatic cancer; failure or refusal to control tumor growth and/or symptoms by conventional treatment. Exclusion criteria were: distant metastases; life expectancy less than 3 months; contraindication to general anesthesia or contrast-enhanced MR. The initial study population included five subjects, with two dropouts: one due to treatment refusal in a patient with pancreatic adenocarcinoma and one due to gastrointestinal bleeding during hospitalization in a patient with hepatocellular carcinoma (HCC). After obtaining written, informed consent, we finally enrolled three consecutive patients (one male, two females, age 61–80 years): one with an atypical, biopsy-proven hypovascular HCC (size: 16 mm) in liver segment VI and two with unresectable adenocarcinomas of the pancreatic isthmus (size: 24 and 37 mm). In all cases, MRgFUS was considered the only feasible focal treatment option, due to lesions unresectability and refusal of celiac plexus alcoholization for pain palliation in the patients with pancreatic adenocarcinoma and refusal of surgery or percutaneous ablation modalities in the patient with HCC. At the time of enrollment and during follow-up, the patients with pancreatic adenocarcinoma were under a standard chemotherapy regimen and underwent unsuccessful radiotherapy for pain palliation. The clinical endpoints were pain control and palliative tumor ablation in the patients with pancreatic adenocarcinoma and complete tumor ablation in the patient with HCC.

Patient Positioning and Treatment Planning

All procedures were performed on 3T MR scanner (GE Medical Systems), featuring a 208 element annular phased-array HIFU transducer embedded into the patient table (ExAblate 2100, InSightec, Haifa, Israel, diameter: 120 mm, radius of curvature: 160 mm, focal distance: 60–200 mm, frequencies: 0.95–1.35 MHz, and energy range 100–7,200 J). Treatment was performed under general anesthesia (balanced administration of sevofluorane and fentanyl), with patients placed in prone position. Before treatment planning, patients observed fast for 12 h and underwent intestinal preparation to reduce at minimum the distension of stomach, small bowel, and colon. A convex gel pad also was used to compress the abdominal wall, reducing at minimum the distance between the HIFU transducer and the target. An expert anesthesiologist decided the adequate amount of air to be inhaled by the patient and the duration of the apnea, using an MR-compatible respiratory monitoring system with a mechanical ventilator (iVent201), similarly to that described previously for USgFUS ablation of liver and pancreatic tumors [12–14]. A combination of T1-w, T2-w, and diffusion-weighted sequences were acquired and repeated three times to confirm identical organ shift and lesion position in all planes. Immediately before starting the procedure, 3D GRE T1-weighted contrast-enhanced sequences (0.5 mL/kg of gadobenate dimeglumine—MultiHance, Bracco SpA, Milan, Italy) were acquired on three-planes to define the target area correctly. Based on these images, the ultrasound beam path was manually drawn into the target area, avoiding ribs and bowel and including a margin of 4–5-mm tumor-free tissue into the ablation area.

Temperature Monitoring and MRgFUS Ablation

Energy level and number of sonications required to ablate the target volume were automatically generated by the system but were modifiable by the operator. Proton-resonance frequency (PRF) shift method was used for real-time monitoring of temperature increase in both target area and surrounding tissues [15]. MR thermometry was performed with a phase-difference fast spoiled gradient-echo sequence providing temperature-dependent images in real-time, which is substantially independent from tissue type and thermally induced tissue changes and it can be used at all stages of the procedure to continuously monitor the location and thermal effects of the sonication. Lesion targeting (Figs. 1, 2) was confirmed with a low energy test sonication (350–400 J, 1.10 MHz) to verify the accuracy of ultrasound beam targeting; system power was then progressively increased to ablation levels until temperature changes could be detected on temperature-sensitive MR images [16], subsequently administering a series of higher powered sonications until the thermal ablation temperature threshold (65 °C) was reached in the target area; following each sonication tissue temperature and estimation of the ablated volume were obtained on the basis of the temperature-sensitive MR images. In order to reduce the chance of tumor movement to a minimum, ablation was performed during controlled deep inspiration, at the same extent at which the morphological imaging sequences had been acquired. Treatment was considered complete once the lesion and 5-mm, tumor-free margins have been completely ablated. A contrast-enhanced, T1-weighted sequence was acquired immediately after treatment completion to verify the effects of the ablation on the target lesion and to exclude thermal damage to surrounding structures.

Fig. 1
figure 1

Patient positioning and lesion targeting for liver ablation. The patient was positioned prone over the transducer surface (A, arrow). The target area was manually defined (B, yellow circle) as well as critical anatomical structures (c, skin red line, liver yellow line) and sonication spot (C, yellow box). Proton resonance frequency sequences (D) were used to monitor temperature increase in the target area and surrounding tissues with a color scale (green-red area), corresponding to a temperature scale

Full size image
Fig. 2
figure 2

Patient positioning and lesion targeting for pancreatic ablation. MRgFUS treatment planning (A) with the patient placed in the supine position over the ultrasound transducer (t): image shows liver (l), spleen (s), right kidney (k), and abdominal aorta (a), as well as the predicted ultrasound beam path (blue triangles) with the focal spot placed over the target lesions (green box). PRF shift-weighted image for MR thermometry (B) demonstrate increase of tissue temperature in the focal spot area (green spot) as well as heat distribution beyond the target zone (red spots), along the abdominal aorta and within the coeliac plexus. Temperature chart (C) shows progressive increase of tissue temperature during sonication in the target area

Full size image

Analysis of Results

Contrast-enhanced-MR follow-up was performed at 30 days, 3 months, and 6 months after MRgFUS using the same sequences acquired in the pretreatment phases. To assess treatment response, contrast-enhanced images were used to quantify the nonperfused volume (NPV), defined as the volume of tumor tissue perfused at baseline that did not show any contrast uptake after treatment. NPV was measured immediately after ablation, at 1, 3, and 6 months follow-up using a semiautomatic software program (Advantage, GE), and assessed as a percentage of the baseline neoplastic volume. In the patients with pancreatic adenocarcinoma, symptom severity was assessed during follow-up with the VAS score (from a minimum of 0 to a maximum of 10) to investigate the potential effects of ablation on pain related to the infiltration of the celiac plexus. Pain interference score also was assessed during follow-up using a validated translation of the Brief Pain Inventory-Quality of Life (BPI-QOL) questionnaire [17]. Eventual treatment-related complications after each treatment were evaluated on the basis of treatment-related clinical events and/or follow-up laboratory and imaging findings.

Results

A total of 43 sonications (energy range 1,538–2,615 J) was necessary for the ablation of HCC, whereas 28–33 sonications (energy range 1,574–2,036 and 1,851–3,289 J) were required for pancreatic ablations (Table 1). The average duration of each ablation cycle (including apnea) was 30 ± 1 s. The average procedure duration was 84 min (±10), with a mean temperature of 71 °C (±2) in the target volume. Patients were monitored for adverse events and discharged uneventfully the day after treatment. In the patient with HCC, immediate posttreatment MR evaluation and 1 month follow-up showed complete ablation of the lesion, without residual viable tissue (100 % NPV) (Fig. 3); 3- and 6-month follow-up imaging showed a small focus (8 mm in maximum diameter) of recurrent tumor tissue along the lateral edge of the ablation zone, with a NPV of 85 %. Sixteen months after MRgFUS, the patient underwent liver transplantation; histopathology demonstrated coagulation necrosis of ablated area with a focus of tumor regrowth along the ablation margins (Fig. 4). In the two patients with pancreatic adenocarcinoma, immediate posttreatment MR evaluation and 1-month follow-up showed almost complete ablation of both pancreatic lesions, with a NPV respectively of 80 and 85 %. Due to accurate planning, no signs of damage to adjacent structures were observed; little regrowth of tumor tissue along the ablation margins was demonstrated at the 3- and 6-month MR follow-up in both cases (NPV of 70 and 80 %) (Fig. 5).

Table 1 Treatment details
Full size table
Fig. 3
figure 3

A 64-year-old female with biopsy-proven hypovascular HCC in liver segment VI. Pretreatment, T1-weighted, contrast-enhanced MR images acquired in the hepatobiliary phase demonstrate hypovascular lesion in segment VI (A, arrow). Posttreatment T1-weighted MR images acquired in the hepatobiliary phase after treatment demonstrate the thermal ablation area (B, arrow), without significant evidence of damage to surrounding structures

Full size image
Fig. 4
figure 4

A 64-year-old female with biopsy-proven hypovascular HCC in liver segment VI. Evaluation of gross specimen of explanted cirrhotic liver demonstrates ablation zone (a, arrow) as a white-yellow area in segment VI, surrounded by regenerative nodules. Histologic slice (hematoxylin–eosin stain, magnification ×20) of the ablated lesions show an extensive area of fibrosis that replaced coagulative necrosis (B, arrow) surrounded by viable-appearing tumor (B, arrowheads)

Full size image
Fig. 5
figure 5

A 61-year-old female with biopsy-proven pancreatic adenocarcinoma. Pretreatment, T1-weighted, contrast-enhanced, MR images acquired in the portal-venous phase demonstrate hypovascular lesion in the pancreatic isthmus (A, arrow). Posttreatment, T1-weighted, MR images acquired in the portal-venous phase after treatment demonstrate the thermal ablation area (B, arrow) and partial residual viable tissue, without significant evidence of damage to surrounding structures

Full size image

VAS score decreased from a mean of 7 ± 1 points to a mean of 3 ± 1 points during the week after treatment in both patients; one of these patients died due to metastatic disease 13 months after MRgFUS procedure (VAS was 2 points at the time of death), while the other is still alive at present (VAS of 3 and 4, respectively at 6 months and 8 months from treatment). In these patients, the mean pain interference score decreased from 6.7 ± 5 at baseline, to 2.0 ± 2 at 1 month, 2.3 ± 2 at 3 months and 3.0 ± 2 at 6 months. No treatment-related complications were observed after each ablation or during the entire follow-up period.

Discussion

The advent of USgFUS represented a major breakthrough to reach completely noninvasive control of solid tumors in abdominal organs [11]. In particular, the real-time capability of US guidance coupled to the high efficacy of HIFU ablation progressively led to a widespread diffusion and clinical consolidation of this technique for the treatment of unresectable liver and pancreatic cancers, with satisfactory results in terms of local tumor control, symptoms palliation, and survival rates [4, 6, 18, 19]. Notwithstanding the above-mentioned strength point of USgFUS, the recent introduction of MRgFUS rapidly raised clinical interest for the expansion and technical improvement of previously explored applications, including treatment of uterine fibroids, breast cancer, and symptomatic bone tumors. In particular, the superiority of anatomic imaging of MR compared with US and the in-procedure quantification of thermal dose and thermal damage to targeted tissue allowed by dedicated MR sequences drove to the clinical approval and to the continuous improvement of MRgFUS technology. On the other hand, several limitations precluded a wide clinical applicability of MRgFUS technique in the upper abdomen. First, because most of the liver, spleen, and kidneys are located behind the ribcage, these sites are currently inaccessible due to the lack of a system capable of focusing the HIFU beam through the bone cortex, avoiding reflection, refraction, and absorption-induced tissue overheating [20]. A technique based on the time-reversal process to focus HIFU through the ribcage is under development [21] and will be probably implemented in future. A second important limitation is due to the fact that MR is only capable of near real-time imaging and anatomic monitoring and that organ movements (up to 20 mm during respiration) [22] may significantly affect lesions targeting and ablation efficacy of MRgFUS. Whereas forced breath-hold allows precise HIFU ablations under MR guidance in our study, in the next future this technique will be probably replaced by more sophisticated methods of targeting: in particular, the adaptive method, in which the HIFU beam and the temperature monitoring automatically follow in real-time the motion of the target organ basing on pre-generated models from navigator echoes [23, 24], seems to be the most promising approach to expand the application of MRgFUS in the upper abdomen.

Notwithstanding these limitations, we demonstrated that MRgFUS may be currently feasible in selected tumors in pancreas and liver, with detailed morphological evaluation of the target lesion and accurate monitoring of thermal damage, without complications during or after the procedure. The breathing-control approach with general anesthesia and forced inspiration was shown to be safe without significant errors in lesion targeting or thermometry. In the patients with pancreatic adenocarcinoma MRgFUS treatment was adequately successful in obtaining both local tumor control and symptoms palliation due to thermal damage to the coeliac plexus. In the patient with HCC, successful lesion ablation was demonstrated both at imaging and histopathology.

Conclusions

Even if performed on a small patient population, our experience demonstrates that MRgFUS has the potential to be a feasible and safe procedure for the ablation of selected liver and pancreatic tumors. Further developments in real-time motion tracking and thermal mapping techniques are indispensable to consolidate these results and expand the field of application of this technique in abdominal organs.