MRI for the Assessment of Histotripsy Ablation in a Canine Osteosarcoma Comparative Oncology Model

Abstract

Objective:

To explore the use of MRI for evaluation of mechanical focused ultrasound (histotripsy) ablation in spontaneous canine osteosarcoma (OS) as a comparative oncology model of human OS.

Methods:

A retrospective analysis was conducted on MR images from a veterinary clinical trial, with the purpose of using canine OS as a preclinical model in which to study human OS. Eight dogs received histotripsy treatment to a portion of their tumor, with pre- and post-treatment MRI (pre- and post-contrast T1- and T2-weighted). Dogs then received standard-of-care surgical resection. Qualitative assessments included characterization of ablation on MRI, gross anatomy, and histopathology. Quantitative assessments included measurements of ablation on MRI and gross anatomy, as well as scoring tumor bone production in untreated tumor on MRI and histopathology to compare tumor composition assessments across modalities.

Results:

Contrast-enhanced sequences with and without subtraction reliably visualized histotripsy ablation. For all dogs, the ablation zone was visible as a nonenhancing, hypointense region. There was a significant decrease in normalized mean signal intensity of the targeted tumor region from pre- to post-histotripsy. No significant differences were found between MRI and gross measurements of ablation dimensions and volume. Histopathology confirmed ablation in the targeted regions as identified grossly and on MRI, and tumor bone composition assessments on MRI and histopathology had a significant positive correlation.

Conclusions:

MRI is a valuable tool for the assessment of histotripsy ablation in OS and builds upon the potential of histotripsy as a non-invasive therapy for bone tumors in dogs and humans.

Introduction

Histotripsy is a non-thermal, non-ionizing, and non-invasive focused ultrasound (US) therapy currently under development for the treatment of osteosarcoma (OS) and other primary bone tumors [1−4]. Histotripsy applies short, high amplitude US pulses to generate acoustic cavitation bubble clouds which expand and collapse to mechanically destroy tissue [5−7]. Histotripsy has been investigated preclinically for the ablation of many soft tissue tumors, including prostate [8], liver [9,10], kidney [11,12], pancreas [13,14], brain [15], and soft tissue sarcoma [16] tumors, as well as clinically for liver tumors [17,18]. Recent studies have pioneered the investigation of histotripsy for OS ablation through a series of ex vivo [1] and in vivo experiments [2,3]. While the general findings of these studies suggest that histotripsy can be used for the safe and effective ablation of bone tumors, technical challenges remain prior to histotripsy being adopted clinically for these tumors.

Histotripsy is typically guided and monitored using real-time B-mode US imaging. The cavitation bubble cloud formed during treatment can be visualized sonographically as a dynamic hyperechoic area, while the resulting ablation is typically visualized as a hypoechoic region [3,19]. Bone tumors, however, present a unique challenge for US-guided therapies such as histotripsy. US evaluation of bone tumors is limited by the bone’s significant absorption and reflection of the US waves, resulting in posterior acoustic shadowing [20,21]. Thus, other imaging modalities are needed to characterize the tumor, aid in treatment planning, and assess histotripsy ablation. CT imaging has been used to successfully visualize histotripsy ablation in the liver [17,18]. Previously, pre- and post-histotripsy CT scans were collected for canine patients enrolled in (i) a veterinary clinical trial for canine soft tissue sarcoma (STS) tumors [16], and (ii) two veterinary clinical trials for canine OS [2,3]. These trials showed that CT scans reliably visualized histotripsy ablation in STS but not in OS. For the latter, it was often difficult to distinguish the boundaries of the tumor due to the heterogenous attenuation pattern in the soft tissue regions, the variation in tumor shape, and the contrast enhancement differences between tumors. Likewise, histotripsy ablation zones were difficult to consistently visualize on post-treatment CT scans [2,3]. As a result, an alternative imaging modality to US and CT is necessary to aid in the evaluation of histotripsy ablation of heterogeneous bone tumors.

MRI has been previously used to assess ablation-induced damage following histotripsy ablation of soft tissues, offering improved soft tissue contrast over CT, showing clear distinctions between treated and untreated tissues, and demonstrating an eventual resorption of the ablation zone [10,17,19,22,23]. To date, MRI has not been used to assess histotripsy ablation of bone tumors, and its utility for this purpose remains unknown. MRI is routinely used to stage human bone tumors [24,25] and has been utilized to chronically assess high intensity focused US ablation of metastatic bone tumors [26,27]. MRI has also been explored for the evaluation of bone tumors in canines [28−30]. Specifically, a comparison of US, CT, and MRI for the assessment of canine OS found that MR was the only imaging modality that did not underestimate tumor boundaries [30]. Additionally, studies in canine OS have reported that T1-weighted MR imaging is superior for identifying intramedullary tumor extent and identifying the boundary between normal and abnormal signal intensity in the bone marrow [29,30]. The improved contrast between soft tissues on MRI as compared to CT, combined with its non-ionizing nature, history of use to assess focused US ablation, and prevalence in clinical practice for staging human bone tumors, make it a natural investigative pathway as a modality to assess histotripsy ablation of OS.

Naturally-occurring tumors in canines, such as OS, present a unique model in which to study cancer biology and novel therapeutics [31]. Canine OS shares many biological similarities with human OS, including clinical presentation, histopathological appearance, and genetic abnormalities. These similarities, combined with the dog’s increased incidence of OS and accelerated disease progression, make the canine a valuable comparative oncology model in which to study OS [31,32]. The objective of this study is to evaluate whether MRI can be used to acutely visualize and characterize histotripsy ablation in canine OS as a model for human OS, with the ultimate goal of advancing the clinical translation of histotripsy for human patients.

Materials and methods

Canine patient recruitment

Client-owned dogs with suspected appendicular OS (n = 8) were enrolled in a veterinary clinical trial investigating patient response following tumor ablation achieved by histotripsy. Patients were recruited between August 2022 and May 2023. All animal work was approved by the Institutional Animal Care and Use Committee approval (IACUC protocol #19–229) at Virginia Tech, and signed owner consent was obtained for all canine patients. Eligibility criteria included: (i) suspected appendicular OS based on radiographic and/or cytologic evidence; (ii) no evidence of pulmonary metastases on 3-view thoracic radiographs; and (iii) no tumor-targeted treatments at time of enrollment.

Pre-treatment imaging

Canine patients underwent MR imaging of the tumor-bearing limb 0 −2 d prior to histotripsy using a 1.5T Philips Intera MRI scanner (Konin-klijke Philips N.V., Amsterdam, Netherlands). Positioning and MR imaging took approximately 1 hour per patient. Each patient was imaged using the following sequences in the sagittal plane: dual-echo T2w and PD (T2- and proton density-weighted, respectively), STIR (short T1 inversion recovery), VISTA (variable density incoherent spatiotemporal acquisition, a three-dimensional T2-weighted sequence), T1w TFE (three-dimensional T1-weighted turbo field echo), and T1w FS (T1-weighted fat-saturated). The latter two sequences (T1w TFE and T1w FS) were acquired prior to and following the intravenous administration of a gadolinium-based contrast agent (ProHance^® from BRACCO Diag-nostics, Monroe, NJ, USA). The contrast agent was administered at 0.2 mL per kg for each canine patient with an approximately 3 min delay between contrast injection and image acquisition. Additionally, the three-dimensional sequences (T1w TFE and VISTA) allow for multi-planar reconstruction of MR images in the coronal and axial planes. Imaging protocols for all sequences are detailed in Supplementary Table S1.

Histotripsy system and treatment

The histotripsy system and workflow used in this study have been detailed in prior publications of veterinary clinical trials in canine OS [2,3,16]. Briefly, a custom, hydrophone-calibrated 32-element 500 kHz histotripsy transducer with a coaxially-aligned 3 MHz curvilinear ultrasonic imaging probe (Model C5–2, Analogic Corporation, Peobody, MA, USA) was integrated onto a prototype clinical system (HistoSonics, Ann Arbor, MI, USA) as previously described [2,3,16]. The transducer has a focal length of 78 mm with elevational and transverse apertures of 112 mm and 128 mm, respectively. More details regarding the transducer specifications and pressure calibration methods are outlined in prior publications [1,2].

For canine treatments, patient-specific ablation plans were developed based on pre-treatment MR images. When possible, soft tissue and/or lytic bone regions of the tumor, as visualized on MRI, were targeted to maximize ablative efficacy. All dogs were placed under general anesthesia maintained with inhaled isoflurane for the entirety of the histotripsy treatment. The patient was prepared for histotripsy ablation as previously described [2,3,16].

A 2 cm diameter spherical volume (4.19 cm³) within the primary bone tumor was targeted. The bubble cloud (approximately 6×2×2 mm) [2] was robotically steered through the targeted volume. Following recent OS work [2,3], histotripsy was delivered using single cycle pulses at a pulse repetition frequency (PRF) of 500 Hz and a dosage of approximately 1000 pulses per point, applied in two sequential automated scans through the planned ablation volume. The treatment volume contained a 3D grid of equidistant treatment points (spaced 3.5 mm in the axial direction and 1.5 mm in the lateral and elevational directions).

For each patient, the pressure at the treatment focus was chosen by increasing pressure incrementally until cavitation was identified on US imaging and/or passive cavitation detection (PCD). PCD was achieved by connecting one of the transducer’s therapy elements to a high voltage probe and recording the backscatter of the incident pulse from the focal point using an oscilloscope [2,3,33,34]. Additionally, PCD was used to monitor for prefocal cavitation on the skin and cortical bone surface, which has been encountered when treating appendicular OS [2,3]. Both focal and prefocal cavitation are observable with PCD and distinguishable based on their timing within the PCD waveform. The overall histotripsy set-up, with specific examples of US images and PCD waveforms when targeting in soft versus bony components of OS tumors, is shown in Figure 1.

Figure 1.

Clinical prototype histotripsy system (a) with real-time US image feedback (b, c) and single-element passive cavitation detection (PCD, inserts in b, c). Patient-matched US images and PCD waveforms are shown for two representative patients, one showing ablation targeted in OS soft tissue (Patient #8, b) and one showing ablation targeted in OS bone (Patient #2, c). The PCD waveforms show the signal from the therapy pulse as well as the reflected bubble cloud signal. The red crosshair on the US images (b, c) represents the treatment focus.

Imaging assessment of histotripsy ablation

Canine patients underwent a post-treatment MRI 1−5 d after histotripsy following the same MR sequences and parameters as the pre-treatment imaging. MR images were assessed by a board-certified veterinary radiologist (G.B.D., 35 y of experience), a human diagnostic and interventional radiologist with extensive experience in assessing ablation on imaging (T.J.Z., 18 y of experience), and 2 human musculoskeletal radiologists (S.B.S. and G.B.M., 14 and 3 y of experience, respectively). Specifically, assessments focused on signal intensity and enhancement changes in the targeted region of tumor, as well as on off-target changes or damage from histotripsy visible on post-treatment MR images. OS tumors were scored as minimally, mildly, variably, or highly productive (tumor bone production, in untreated tumor) according to qualitative signal intensity characteristics on T2w and T1w TFE pre-contrast images, based on a consensus read between the first author and 2 human musculoskeletal radiologists.

Special attention was paid to the contrast-enhanced sequences (T1w TFE and T1w FS). For these sequences, digital contrast subtractions (post-contrast image minus pre-contrast image) were performed using the open-source software Horos (www.horosproject.org) to better visualize changes in tumor contrast enhancement after histotripsy treatment. The dimensions of the identified ablation zone were measured on multi-planar reconstructions of both post-contrast (contrast-enhanced) and subtraction T1w TFE images in all 3 planes (sagittal, coronal, axial). The height (proximal-distal), width (cranial-caudal), and depth (medial-lateral) of the ablation zone was measured, such that the sagittal plane measured height and width, the coronal plane measured height and depth, and the axial plane measured width and depth. All 3 dimensions (height, width, depth) were therefore measured twice for each set of patient images. These measurements were performed by a board-certified veterinary radiologist and by one of the primary authors (E.R. V., 3 y of experience), with oversight and training from the veterinary radiologist (G.B.D.). The final measurement for each dimension is reported as the average of the two measurements from different planes, which are then averaged between the two author measurements. Ablation volumes are calculated assuming a uniform ellipsoidal volume from these averaged measurements (estimated ablation volume).

To more accurately measure ablation volume on MRI, the ablation zone was segmented from both contrast-enhanced and subtraction T1w TFE images by both primary authors (E.R.V. and L.N.R., 3 and 5 y of experience, respectively) to yield the segmented ablation volume. The open-source software 3D Slicer (www.slicer.org) was used for volumetric segmentations. Segmentations were performed by comparing pre- and post-treatment images slice-by-slice and highlighting on post-treatment imaging any areas with changes in signal intensity or loss of contrast enhancement as compared to pre-treatment imaging. Specifically, the hypointense and nonenhancing core of ablation was identified and segmented; however, any peripheral contrast enhancement due to edema and inflammation was not included in the ablation volume. The area identified as the ablation zone on post-treatment imaging was then copied onto co-registered pre-treatment images. Co-registration was performed in 3D Slicer using automated composite registration. Volume, mean and median signal intensity, and minimum and maximum signal intensity of the segmented ablation zone on pre- and post-treatment images were calculated and averaged between observers.

To normalize the signal intensity of the targeted tumor region of interest for each imaging session, a spherical volume (100 − 1000 voxels, 0.1 − 1.0 cm³) was segmented in the healthy muscle adjacent to the tumor. The spherical volume was chosen to be as close to the tumor (< 6.5 cm) and the ablation zone (< 10 cm) as possible while selecting a homogenous region of muscle tissue. After normalizing the mean signal intensity of the targeted tumor region to the mean signal intensity of the healthy muscle region on each image, the difference in normalized mean signal intensity between pre-treatment and post-treatment tissue was calculated and compared.

Surgical resection of tumor

Three or five days after histotripsy treatment, the primary bone tumor was surgically resected via standard of care amputation of the affected limb performed by a board-certified veterinary surgical oncologist (J.T.).

Gross and microscopic evaluation of tumor tissue

After surgical amputation, the affected limb was processed and grossly evaluated by a board-certified veterinary pathologist (S.C.O., 10 y of experience) to locate the histotripsy-treated tumor region and the remaining untreated tumor. To ensure accurate identification of the histotripsy ablation zone, the skin overlying the targeted tumor was marked immediately after ablation, and treatment details were sent to the pathologist before processing. To measure the ablation zone, the tumor-bearing bone was cut along the same plane as MRI acquisition (sagittal), and the ablation zone was identified by hemorrhage, tissue softening, and demarcation from adjacent untreated tumor tissues, as previously described [2,3]. The height (proximal-distal) and width (cranial-caudal) were measured, and then the bone was cut one final time in the third dimension to obtain a measurement of ablation depth (medial-lateral). The grossly measured ablation volumes are calculated assuming a uniform ellipsoidal volume from these three measurements (gross ablation volume).

Tissues from the treated and untreated tumor regions were collected for standard-of-care diagnostics and characterization of ablation. All tissues were collected and fixed in 10% formalin. Tumor samples with too much dense bone to be easily trimmed and processed following formalin fixation were placed into 10% EDTA decalcification solution until soft enough to easily cut by hand (3−10 d). All samples were then embedded in paraffin, sectioned at a thickness of 5 μm, and routinely stained with hematoxylin and eosin. Samples were assessed by a board-certified veterinary pathologist with expertise in the evaluation of ablated tumor tissue. Particular focus was dedicated to the overall viability of tumor cells, the integrity of tumor matrix, and the presence or absence of hemorrhage. OS tumors were characterized as minimally, mildly, variably, or highly productive (tumor bone production, in untreated tumor) based on gross and histopathological characteristics.

Statistical analysis

All statistical analyses were performed using GraphPad Prism (version 10). Statistical significance was defined as a p < 0.05. The residuals of all datasets being used for t tests and ANOVAs were tested for normality using the Shapiro-Wilk, Anderson-Darling, Kolmogorov-Smirnov, and D’Agostino-Pearson tests; the residuals were considered to follow a normal distribution if p > 0.05 for all tests. If the residuals did not follow a normal distribution (p < 0.05), the nonparametric statistical test equivalent was used. Summary statistics are reported as mean ± standard deviation.

Paired t tests were used to compare the normalized mean signal intensity of the targeted tumor tissue on pre- versus post-histotripsy MR images. To evaluate how well measurements of ablation on MRI correspond to gross measurements of ablation, repeated measures one-way ANOVAs with Tukey’s multiple comparisons tests were used to compare the differences between the ablation measurements and volumes on MRI (contrast-enhanced and subtraction images) and (i) targeted ablation measurements and volumes, and (ii) gross ablation measurements and volumes. Additionally, Pearson’s correlations were used to test the relationship between individual observer ablation volumes on MRI to grossly measured ablation volumes. Pearson’s correlations were additionally used to test the relationship between individual observer measurements of ablation on MRI and individual segmentations of the ablation zone on MRI. Inter-observer variability was considered significant if there was no significant correlation between measurements [35]. Lastly, Spearman’s rank correlation was used to test the relationship between tumor bone production on histopathology and on MRI. The pathological and radiological assessments of minimally, mildly, variably, and highly productive were converted to a numerical scale (1−4) to score bone production.

Results

Patient population and histotripsy treatment

Three male and five female dogs with OS were enrolled in this study (n = 8). At the time of enrollment, patients averaged 7.3 ± 3.2 y of age and 39 ± 13 kg. Patient and tumor characteristics are summarized in Table 1. All canine patients enrolled in this trial were diagnosed with OS based on histopathological analysis of the amputated tumor-bearing limb.

Table 1.

Characteristics of canine OS patients in the present study, along with the timing of pre- and post-histotripsy MR images and post-histotripsy surgical resection

Patient	Breed	Age (y)	Gender	Weight (kg)	Tumor location	Days pre-histotripsy MRI	Days post-histotripsy MRI	Days post-histotripsy amputation
1	Saint Bernard	3	MN	59	Distal tibia	1	1	3
2	Siberian Husky	9	FS	25	Femur	1	3	5
3	Golden Retriever	11	MN	36	Proximal tibia	2	2	5
4	Pointer	7	FS	35	Distal tibia	1	1	5
5	Standard Poodle	2	MN	30	Distal radius	0	5	5
6	Boxer (mix)	9	FS	25	Proximal tibia	0	5	5
7	Cane Corso	9	FS	50	Proximal humerus	0	5	5
8	Cane Corso	8	FS	52	Proximal humerus	0	5	5

M, male; F, female; N, neutered; S, spayed.

All 8 dogs received histotripsy ablation of a portion of their OS tumor. Histotripsy treatment details are summarized in Supplementary Table S2. Treatments were applied at estimated, free-field peak negative pressures averaging 25.8 ± 5.1 MPa and centered at an average depth of 2.6 ± 0.9 cm from the surface of the skin, as measured on US imaging. The histotripsy bubble cloud was visible on real-time US imaging for the entire treatment duration in 1/8 dogs, intermittently visible (primarily in softer tissues) in 5/8 dogs, and not visible on US for 2/8 dogs. PCD was additionally used to monitor cavitation, and in instances where the bubble cloud was not visible, PCD confirmed cavitation generation (Fig. 1c). For all patients, no obvious changes in the echogenicity of the treated regions were noted on US imaging immediately post-treatment. Overall, histotripsy was well-tolerated in all dogs. No clinically significant perturbations in vital signs were noted during histotripsy treatment, and no adverse events related to histotripsy treatment were observed.

Assessment of osteosarcoma ablation on MRI

Prior to histotripsy treatment, OS tumors were heterogeneously hyperintense in comparison to healthy muscle on dual-echo T2w and PD sequences, as well as on the VISTA sequence. On STIR images, bony proliferation had low signal intensity, while soft tissue and marrow components of the tumor had high signal intensity. This was consistent with mineralized regions appearing T2-hypointense and softer regions appearing T2-hyperintense. OS generally appeared isointense on pre-contrast T1w TFE images but had a mixed to hyperintense signal on pre-contrast T1w FS images. Patients characterized as having more bone production in their tumor had lower signal intensity on T2w and pre-contrast T1w FS images. One dog (Patient #5) had a minor linear defect suspected to be a nondisplaced pathological fracture that was visible on pre-treatment imaging. This fracture was confirmed at gross sectioning.

Overall, all OS tumors had some degree of contrast enhancement pre-treatment. 4/8 bone tumors demonstrated avid enhancement, while 4/8 bone tumors only showed enhancement in the soft tissue portions of the tumor and in red marrow infiltrated by tumor. Contrast-enhanced sequences revealed heterogeneous pockets of pre-existing nonenhancing necrosis within bone tumors. Qualitative descriptions of the pre-treatment tumor and the post-treatment ablation zone across all MRI sequences are detailed in Table 2. None of the noncontrast sequences were able to clearly and consistently visualize the ablation zone across dogs, limiting the utility of these images for ablation characterization. Post-histotripsy MR images are shown on all sequences for two representative dogs in Figure 2, showing histotripsy ablation in OS soft tissue (Patient #8) versus in OS bone (Patient #2). The ablation zone was more identifiable on noncontrast sequences in dogs with softer or less productive bone tumors. Post-treatment STIR images demonstrated hyperintense signal within the ablated zone or along the ablation periphery for 4/8 dogs, particularly for treatments in which the ablation was targeted in softer tissue. No off-target damage or ablation was noted in any post-treatment images.

Table 2.

Bone tumor and ablation zone characteristics on all MR sequences for all canine OS patients in the present study

	Patient	New bone production	T2 and VISTA	PD	STIR	T1w TFE PRE	T1w FS PRE	T1w TFE POST	T1w FS POST
Tumor (pre-histotripsy)	1	High	Hypo	Hypo	Hypo bone, hyper soft tissues	Hypo	Hypo	Enhancing soft tissue	Enhancing soft tissue
	2	High	Mixed	Mixed	Mixed	Hypo	Hyper	Heterogeneously enhancing	Heterogeneously enhancing
	3	Minimal	Hyper	Hyper	Hyper	Iso	Hyper	Enhancing	Enhancing
	4	High	Hypo	Hypo	Hyper	Hypo	Hyper	Mildly enhancing	Mildly enhancing
	5	Variable	Hypo bone, hyper soft tissues	Mixed	Hyper	Hyper	Hyper	Enhancing soft tissue	Enhancing soft tissue
	6	High	Hypo	Hypo	Hyper	Hyppo	Mixed	Enhancing	Enhancing
	7	Mild	Mixed	Iso	Hypo bone, hyper soft tissues	Mixed	Mixed	Heterogeneously enhancing	Heterogeneously enhancing
	8	Minimal	Hyper	Hypo bone, hyper soft tissues	Hypo bone, hyper soft tissues	Hypo	Hypo bone, hyper soft tissues	Heterogeneously enhancing	Heterogeneously enhancing
Ablation (post-histotripsy)	1	High	NC	NC	NC	NC	NC	Nonenhancing	Nonenhancing
	2	High	NC	NC	Hyper around ablation	Hyper around ablation	Hyper around ablation	Nonenhancing core with enhancing rim	Nonenhancing core with enhancing rim
	3	Minimal	Hyper	Hyper	Hyper	Hyper	Hyper	Nonenhancing core with enhancing rim	Nonenhancing core with enhancing rim
	4	High	NC	NC	NC	NC	NC	Nonenhancing	Nonenhancing
	5	Variable	NC	NC	NC	NC	NC	Nonenhancing	Nonenhancing
	6	High	NC	NC	NC	NC	NC	Nonenhancing core with enhancing rim	Nonenhancing core with enhancing rim
	7	Mild	NC	NC	Hyper	NC	NC	Nonenhancing	Nonenhancing
	8	Minimal	Hypo	Hypo	Hypo	Hyper	Hyper	Nonenhancing core with enhancing rim	Nonenhancing core with enhancing rim

Rows in bold indicate post-histotripsy ablation assessments.

Hypo, hypointense relative to muscle; hype, hyperintense relative to muscle; iso, isointense relative to muscle; Mixed, regions of both hyperintensity and hypointensity; NC, no change post-histotripsy.

Figure 2.

Post-histotripsy MR images on all sequences for two canine patients with OS, one showing ablation in OS soft tissue (top row, Patient #8) and one showing ablation in OS bone (bottom row, Patient #2). Red arrows indicate ablation post-histotripsy, and blue arrowheads indicate inflammation and edema around the ablation zone. The righthand panel shows a representative pre-histotripsy MR image (T1w TFE post-contrast) for each patient, with a yellow dashed line outlining the tumor boundaries.

Post-treatment, the ablation zone was most clearly visualized as a nonenhancing, hypointense spherical region on post-contrast T1w TFE and T1w FS images. As T1w TFE images allowed for 3D reconstruction and image analysis, we focused the remainder of our ablation zone analysis on post-contrast T1w TFE images. Contrast-enhanced T1w images are shown from each patient pre- and post-histotripsy in Figure 3a. For improved visualization of the ablation zone, particularly in cases where the targeted tumor itself was nonenhancing and hypointense, contrast subtractions were performed on T1w TFE images (post-contrast minus pre-contrast image, subtraction). For all patients, the histotripsy ablation zone was visible on both contrast-enhanced (Fig. 3a, 3b) and subtraction (Fig. 3c) T1w images. In addition to the hypointense spherical region representing the ablation zone, 4/8 dogs displayed a hyperintense rim of enhancement around the periphery of the ablation zone (Patients #2, 3, 6, and 8). This typically occurred in less productive bone tumors where the histotripsy treatment was targeted in a softer portion of the tumor.

Figure 3.

Contrast-enhanced images (a, pre-histotripsy and b, post-histotripsy) and subtraction (c, post-histotripsy) of canine OS on T1-weighted MRI. Red arrows indicate ablation post-histotripsy, and blue arrowheads arrows indicate inflammation and edema around the ablation zone. Yellow dashed lines on pre-treatment images indicate the tumor visible on the displayed slice of the MR image.

Pre- and post-histotripsy subtraction T1w images are shown in all 3 anatomical planes for two representative dogs, showing histotripsy ablation in OS soft tissue (Patient #8, Supplementary Fig. S1a, S1b) and in OS bone (Patient #2, Supplementary Fig. S2a, S2b). Additionally, volumetric renderings of the ablation zone are shown for these two representative patients in Supplementary Figures S1, S2 (c, d). Pearson’s correlations were used to assess the agreement between individual observer volumetric segmentations and dimensional measurements, showing significant positive correlations for all segmentations from contrast-enhanced and subtraction MRI (Supplementary Table S3). Volu-metric segmentations of the ablation zone were also used to quantify the change in signal intensity in the treated tissue from the pre-treatment image to the post-treatment image (Fig. 4). Across all dogs, the normalized mean signal intensity of the targeted tumor region had a significant decrease from pre- to post-histotripsy on contrast-enhanced MRI (−38.8 ± 23.3 %, p = 0.0121, Fig. 4a) and on subtraction MRI (−85.8 ± 15.6 %, p = 0.0293, Fig. 4b).

Figure 4.

Mean normalized signal intensity of the targeted tumor region before and after histotripsy treatment on (a) contrast-enhanced and (b) subtraction T1w MRI. Signal intensity of the targeted tumor region is normalized to the signal intensity of adjacent healthy muscle for each image. Asterisks indicate statistical significance (p < 0.05) as determined by a paired t test (a) and a Wilcoxon matched-pairs signed rank test (b).

Assessment of osteosarcoma ablation on anatomic and histopathology, with comparisons to radiological ablation assessments

Gross and histopathological evaluations of treated versus untreated tumor samples are summarized in Table 3. Grossly, OS tumors varied in composition including predominantly hard bone tissue, predominantly firm tissue (typically a mixture of bone, cartilage, and soft tissue), predominantly soft tissue, or a variable mixture of all three. These correlated well with histopathologic findings. Untreated tumor samples grossly classified as hard were microscopically identified as highly productive and composed of tumor cells embedded within or separated by dense bone or bone precursor tissue (osteoid), and far lesser cartilage. Untreated tumor samples grossly classified as firm were microscopically composed of tumor cells producing more sporadic bone matrix often with prominent areas of cartilage. Untreated tumor samples grossly classified as soft were microscopically identified as minimally productive and composed of mostly tumor cells with minimal bone or cartilage production. Regardless of pre-existing tumor composition, treated areas exhibited variably demarcated foci of hemorrhage and softening of the tissue. Microscopically, these often correlated to varying degrees of tumor cell damage characterized by classic features of coagulative and lytic necrosis. Additionally, most treated tumor samples also exhibited damage to the involved matrix characterized by a range of degenerative changes, including cartilage and osteoid matrix breakdown.

Table 3.

Gross and microscopic descriptions of canine OS tumors and ablation zones after histotripsy treatment

Patient	Gross tumor characteristics	Microscopic tumor characteristics	Gross ablation characteristics	Microscopic ablation characteristics
				Tumor cell damage	Matrix damage
1	Hard	Highly productive, osteoid	Well demarcated zone of hemorrhage and tissue softening	Yes	Yes
2	Hard	Highly productive, osteoid and cartilage	Poorly demarcated zone of hemorrhage	Yes	Rare
3	Firm	Minimally productive, osteoid and cartilage	Well demarcated zone of hemorrhage and tissue softening	Yes	Yes
4	Hard	Highly productive, osteoid	Well demarcated zone of hemorrhage	Yes	No
5	Hard	Variable	Intermediately demarcated zone of hemorrhage	Yes	Yes
6	Hard	Highly productive, osteoid	Intermediately demarcated zone of hemorrhage	No*	No*
7	Soft	Minimally productive, osteoid	Well demarcated zone of hemorrhage and tissue softening	Yes	Yes
8	Soft	Minimally productive, osteoid	Well demarcated zone of hemorrhage and tissue softening	Yes	Yes

^*For Patient #6, histopathologic analysis was unable to be performed on the ablation site.

The scoring of tumor bone production in untreated OS tissues on histopathology was correlated against the tumor bone production as assessed on MRI (in untreated portions of the tumor), showing a significant positive correlation between these two methods of characterizing tumor composition (Spearman’s r = 0.8747, p = 0.0036, Supplementary Fig. S3). Furthermore, for each patient, the achieved ablation was characterized on MRI (contrast-enhanced and subtraction), on gross examination, and on histopathology. Gross and microscopic characterizations of the ablation zone described in Table 3 are shown on patient-matched MRI, gross examination, and histopathology for two representative canine OS patients in Figure 5, highlighting ablation in soft tissue versus bony components of OS tumors. For all patients, radiological signs of ablation (nonenhancing and nonperfused core of ablation with a potential hyperintense rim, Fig. 5a) matched the location of gross signs of ablation (hemorrhage and tissue softening, Fig. 5b) and histopathological signs of ablation (tumor cell and matrix damage, Fig. 5c).

Figure 5.

Comparison of histotripsy ablation in two representative patients with canine OS on (a) contrast-enhanced T1w MRI, (b) gross anatomy, and (c) histopathology (hematoxylin and eosin staining). White outlines indicate the observed ablation zone on MRI (a) and gross anatomy (b). Key OS features (osteoid matrix and malignant OS cells) are indicated with black arrows on histopathology (c). Red black-lined arrows indicate red blood cells and hemorrhage associated with histotripsy treatment (both patients), and blue white-lined arrows indicate mineralization and mature bone (Patient #2, untreated tumor). Black triangles indicate malignant chondrocytes (Patient #2, untreated tumor), while white triangles indicate necrotic chondrocytes (Patient #2, treated tumor). Similarly, black diamonds indicate intact chondroid matrix (Patient #2, untreated tumor), while white diamonds indicate degenerate chondroid matrix (Patient #2, treated tumor). Asterisks indicate dead bone fragments (Patient #2, treated tumor).

After qualitatively confirming that ablation on MRI matched gross and microscopic ablation characteristics, the grossly measured ablation dimensions (height, width, and depth) and volume were compared to the ablation dimensions and volumes as measured on contrast-enhanced and subtraction MRI. Overall, there were no significant differences between ablation dimensions and volumes measured on MRI as compared to those measured on gross anatomy (Table 4, p > 0.05 for all comparisons). Additionally, there were no significant differences between ablation on MRI and on gross anatomy as compared to the targeted ablation dimension (2 cm) and volume (4.19 cm³). Both methods of calculating ablation volume (estimated and segmented) did not significantly differ from the grossly measured ablation volume or targeted ablation volume (Table 4). Neither contrast-enhanced or subtraction MRI ablation dimensions and volumes significantly differed from each other or from the targeted dimensions and volumes (data not shown). Additionally, the estimated and segmented ablation volumes on MRI from each individual observer were correlated against the grossly measured ablation dimensions and volume and showed a significant positive correlation for all comparisons (Fig. 6, p < 0.05 for all comparisons), further validating that the volumes of ablation measured on MRI were comparable and correlated to ablation volumes measured on gross anatomy.

Table 4.

Observer-averaged measurements and segmentations of the ablation zone on contrast-enhanced and subtraction MRI as compared to gross anatomy, where all comparisons were nonsignificant

Patient	Ablation measurements on contrast-enhanced T1w MRI					Ablation measurements on subtraction T1w MRI					Ablation measurements on gross anatomy
	Height	Width	Depth	Estimated volume	Segmented volume	Height	Width	Depth	Estimated volume	Segmented volume	Height	Width	Depth	Estimated volume
1	3.1	1.9	1.7	5.4	5.7	3.2	1.9	1.7	5.3	8.0	2	2	2	4.2
2	2.8	3.1	2.9	13.1	6.3	2.5	3.1	2.6	10.3	5.5	3	3.5	2.5	13.7
3	1.9	1.9	2.3	4.3	3.3	1.6	1.8	2.1	3.2	4.6	1.9	2	2.3	4.6
4	1.8	1.7	1.6	2.6	1.6	2.0	1.9	1.6	3.2	2.1	1.2	1.5	1.1	1.0
5	2.9	1.3	1.3	2.4	2.7	3.0	1.3	1.9	3.8	4.4	2.5	1.5	1	2.0
6	1.9	1.6	1.3	2.0	2.5	2.1	1.5	1.3	2.2	2.8	2.4	2	1.5	3.8
7	3.2	3.0	2.9	16.4	11.7	3.4	2.9	2.8	15.4	10.5	3	4	3	18.8
8	2.8	2.2	2.4	7.6	6.1	2.7	2.3	2.6	8.5	7.3	3	3	3	14.1
Mean ± Standard Deviation	2.6 ± 0.6	2.1 ± 0.6	2.0 ± 0.7	6.7 ± 5.3	5.0 ± 3.3	2.6 ± 0.6	2.1 ± 0.6	2.1 ± 0.5	6.5 ± 2.8	5.7 ± 4.6	2.4 ± 0.6	2.4 ± 0.9	2.1 ± 0.8	7.8 ± 6.7

Estimated volume = calculated from observer-averaged measurements assuming an ellipsoidal ablation volume. Segmented volume = segmentations averaged between observations. MRI measurements and volumes are compared to (i) targeted measurements (2 cm) and volumes (4.2 cm³), and (ii) gross measurements and volumes, using repeated measures one-way ANOVAs with Tukey’s multiple comparisons. All comparisons between MRI and gross ablation were nonsignificant (p > 0.05, data not shown).

Figure 6.

Estimated (top row, a, b) and segmented (bottom row, c, d) ablation volumes as measured on contrast-enhanced MRI (left column) and subtraction MRI (right column) from each individual observer. Pearson’s correlations were used to compare MRI-measured ablation volumes to those measured on gross anatomy for each observer. All correlations are significant (p < 0.05).

Discussion

This is the first report to use MRI to assess histotripsy ablation in OS. While effective ablation of bone tumors has been shown on histopathology with histotripsy [1−3], this ablation had never been well-visualized on US or CT imaging, unlike histotripsy ablations in softer tissues such as STS [16] and liver tumors [17,18]. Overall, pre-treatment MR imaging of canine OS yielded similar observations as reported in prior literature for human and canine OS. Bone tumors were hypointense to adjacent healthy bone marrow on T1w images and were heterogeneously hypointense to adjacent healthy bone marrow on T2w images, with more mineralized tissues appearing darker due to the absence of free water. Contrast-enhanced T1w images were most useful for identifying the extent and boundaries of OS tumors. This result corroborates prior literature, which reported that T1w images were best for identifying the boundary between healthy and malignant tissue in the medullary cavity [30]. As previously reported, all canine OS tumors demonstrated enhancement but often heterogenous secondary to pre-existing necrosis and fibrosis.

Post-histotripsy changes in the targeted region of the tumor were best and most consistently visualized in contrast-enhanced T1w TFE images. Tumors with homogenous enhancement had the most identifiable ablation zones post-treatment due to the contrast of hyperintense, enhancing tumor tissue versus hypointense ablated tissue. All 8 dogs had identifiable ablation volumes on these contrast-enhanced images; however, for 1 patient (Patient #7) with extensive heterogenous enhancement pre-treatment, the targeted volume was near a pre-existing pocket of nonenhancing tissue and was difficult to identify without knowledge of the targeted treatment location. For heterogeneously enhancing OS tumors such as Patient #7, subtraction of T1w TFE images was helpful in identifying the ablation zone with sharper demarcation between ablated and un-ablated tumor. Comparing pre-treatment and post-treatment multi-planar reconstructions of subtraction images provided the best visualization of changes in enhancement following treatment, allowing for the identification of the nonenhancing ablation zone in all eight dogs without knowledge of the treatment location.

Other additional features on post-treatment images included a hyper-intense rim around the periphery of the ablation zone on contrast-enhanced and subtraction T1w images in 4/8 dogs. This signal has previously been hypothesized to be due to transient edema after ablation and/or an infiltrating immune response, as reported for histotripsy treatments of liver cancer in humans [14]. Edema and increased fluid content, particularly within and around the ablation zone, appeared T2-hyperintense and especially hyperintense on STIR images. 4/8 dogs had hyperintensity within or around the ablated tissue on STIR imaging, again suggesting edema and inflammation around the treated area. Promisingly, no off-target ablation was seen, supporting histotripsy’s precise mechanism of damage.

A major focus of this study was comparing assessments of histotripsy ablation on MRI, gross anatomy, and histopathology. Qualitatively, for each treatment, an area closely matching the targeted region was observed as (i) a nonenhancing and nonperfused volume on MRI, (ii) a region of tissue softening and/or hemorrhage on gross anatomy, and (iii) an area of tumor cell and matrix damage with few to no intact tumor cells remaining on histopathology. Histopathological and radiological assessments of bone production in OS tumors had a significant positive correlation, further validating the use of MRI for the planning and assessment of histotripsy ablation in canine OS. Softer and/or less productive bone tumors may be more amenable to histotripsy ablation or may have more obvious differences on post-treatment MRI.

Quantitatively, there were no significant differences between the targeted and gross histotripsy ablation measurements as compared to ablation measurements on MRI, validating the use of contrast-enhanced and subtraction MR imaging to measure and assess histotripsy ablation in OS. While the mean estimated ablation volume on MRI (6.7 cm³ on contrast-enhanced, 6.5 cm³ on subtraction) was closer to the mean grossly measured ablation volume (7.8 cm³), it is important to remember that the gross ablation volume is also an estimation from gross measurements assuming a spherical or ellipsoidal ablation volume. The segmented ablation volume on MRI (5.0 cm³ on contrast-enhanced, 5.7 cm³ on subtraction) was closer to the targeted ablation volume (4.2 cm³) and may represent a more accurate assessment of ablation volume. Ultimately, whether the ablation volume was estimated from MRI measurements or segmented from MR images, the ablation volume was not significantly different from the grossly measured ablation volume or the targeted ablation volume. While there were some nonsignificant yet large deviations between grossly-measured, MRI-measured, and targeted ablation volumes for individual patient cases (particularly Patients #7 and #8), these were primarily restricted to patients with relatively soft tumors with imaging and amputation at 5 d post-histotripsy. This suggests that the ablation zone, particularly in softer regions of bone tumors, may evolve at further post-treatment timepoints.

The primary limitation of this study is the timing of pre- and post-treatment MRIs, which varied between some patients due to veterinary hospital scheduling capabilities. All dogs except for one were amputated at 5 d post-histotripsy, so the histopathology and gross analysis presented is largely from the same timepoint; however, post-treatment MR images varied from 1−5 d post-histotripsy. Half (4/8) of treated dogs received post-treatment imaging on the same day of surgical resection (5 d post-histotripsy). It is possible and likely that the timing of the post-treatment MRI has subtle effects on the ablation zone appearance. We predict that at later acute post-histotripsy timepoints, after the immune response has a chance to activate, there may be more inflammation in and around the ablation zone. A hyperintense rim suggestive of inflammation and edema was visible in 4/8 dogs in this study, 3 of which were imaged after 3 d post-histotripsy. The immune response and immunogenic cell death noted after histotripsy [36,37] could potentially expand the necrotic ablation zone at later post-treatment timepoints, leading to a larger ablation volume than targeted. This was seen in Patients #7 and #8, with relatively soft tumors that were imaged and amputated at 5 d post-histotripsy. However, it remains unknown how the histotripsy ablation zone will behave and resorb in bone tumors over time, as liver abla-tions have been seen to fully resorb within 1−4 mo posthistotripsy [17,19,22]. Further work is needed to characterize the long-term effects and behavior of histotripsy ablation in bone tumors.

An advantage of this study is its treat-and-resect nature, which allowed for histopathological evaluation of the treated and untreated tumor. This allowed for confirmation and visualization of ablation as a comparison to and validation of MR findings. As the clinical translation of histotripsy progresses towards a limb-sparing approach to OS, MRI has the potential to fill an important role in both guiding treatment planning and providing long-term monitoring of the tumor after treatment and over time. This will be especially important when considering chronic treat-and-leave clinical trials in dogs and in humans, where his-topathology of the treated tumor will not be available for confirmation of ablation after histotripsy. Additionally, only small portions of OS tumors have been treated with histotripsy, but future work can utilize MRI to plan and assess multiple and/or larger treatments covering the majority of the tumor. Lastly, future work to improve PCD capabilities towards full transmit-receive systems [38] rather than single-element detection will allow for enhanced localization of the bubble cloud, allowing for cavitation dose data [39] to potentially be compared to MRI data. Future work should also investigate the use of longitudinal MRI to follow histotripsy ablation and eventual resorption in bone tumors, with the goal of advancing the clinical translation of histotripsy as a therapy for human OS.

Conclusion

This is the first study of its kind using MRI to characterize and assess histotripsy ablation in OS. On pre-treatment imaging, OS tumors showed at least some degree of enhancement on T1w contrast-enhanced and subtraction images. On post-treatment imaging, the histotripsy ablation zone was characterized by a nonperfused and hypointense core, with a peripheral rim of hyperintensity suggestive of inflammation and edema in half of the treated patients. Qualitatively, the nonenhancing area of ablation on MRI matched the gross location of ablation, indicated by hemorrhage and necrosis, for all patients, and the ablation zone was characterized on histopathology to be comprised of necrotic cells, degenerate osteoid matrix, and hemorrhage. Characterization of bone production within OS tumors were significantly positively correlated between MRI and histopathological assessments. Quantitatively, measurements of histotripsy ablation on MRI were not significantly different from and were significantly correlated with gross measurements of ablation, indicating the utility of MRI for quantifying histotripsy ablation in OS. Overall, this study validates the use of MRI for the assessment of histotripsy ablation in bone tumors and provides a tool for monitoring histotripsy ablation over time in this heterogenous tumor type, allowing for future long-term studies without limb amputation to investigate histotripsy as a non-invasive limb salvage therapy for OS in dogs and in humans.

Supplementary Material

Supplementary material associated with this article can be found in the online version at doi:10.1016/j.ultrasmedbio.2025.09.017.

Data availability

Data from this study will be available upon reasonable request to the corresponding author.