Contrast-enhanced ultrasound for sentinel lymph node mapping in the routine staging of canine mast cell tumours: a feasibility study.

Canine mast cell tumours (MCTs) typically spread to lymph nodes (LNs) before reaching distant sites, and LN assessment is an important part of MCT staging. Sentinel LN (SLN) mapping techniques to identify draining LNs are being developed and could improve the accuracy of MCT staging. The primary objective of this feasibility study was to determine the safety and effectiveness of contrast-enhanced ultrasound (CEUS) to identify SLNs. Secondary objectives were to determine if the SLNs identified by CEUS coincided with the regional LN predicted by the anatomical lymphosomes, if previous MCT excision altered CEUS SLN findings, and if CEUS could identify MCT nodal metastases. Between June 2017 and March 2019, 59 dogs with 62 MCTs were enrolled. No adverse events related to CEUS were reported. CEUS detected at least 1 SLN in 59/62 MCTs (95.2%, 95% CI: 86.5-99.0%). In only 32/59 (54.2%) MCTs, clinicians would have correctly predicted the SLN(s) identified by CEUS. Among the 35 MCTs that had histological examination of the SLN(s), the prevalence of metastasis was 60% (95% CI: 42.1-76.1%). Additional staging procedures did not reveal any metastases in dogs with histologically non-metastatic SLNs. Integration of CEUS SLN mapping into the routine staging of MCTs is promising, but future studies are required to refine this procedure and to investigate if it would translate into a clinical benefit. This article is protected by copyright. All rights reserved.

proposed, 1,3,4 but there is no consensus defining when and how LNs should be assessed. Currently, the choice of the LN(s) to be sampled relies mainly on our knowledge of anatomical lymphosomes in healthy dogs, 5 but several factors may complicate this initial approach: interindividual variation in the localisation and number of LNs, 6,7 and alterations in the lymphatic drainage pattern associated with neoplasia and surgery. 5,8 Another difficulty may be the occurrence of multiple nodal metastases, 9 and some authors have recommended the systematic extirpation of 4 to 6 LNs for the staging of head and neck canine tumours. 10,11 With a reported 75% sensitivity of fine-needle aspiration cytology (FNAC) to detect MCT nodal metastasis, LN extirpation has also been recommended to complete MCT staging. 12,13 Lymphadenectomy is an invasive procedure that often necessitates additional surgical preparation and time, extending time under anaesthesia for dogs and costs for owners. It is therefore sometimes difficult to justify when the LN draining status is uncertain and/or when the LN is particularly difficult to remove.
Sentinel lymph node (SLN) mapping techniques have been developed in human oncology to identify the first LN(s) receiving the lymph from tumours that predominantly spread through the lymphatic system. 14 Regional lymphoscintigraphy combined with intra-operative lymphoscintigraphy and blue dye, the gold standard SLN mapping technique in human oncology, has been successfully incorporated into canine MCT staging in previous studies. 15,16 Lymphoscintigraphy is however, only available in very few veterinary centres. The use of computed-tomography (CT) SLN mapping is mainly restricted to Japan in human oncology, 17 but is becoming a very popular technique among veterinarians. [18][19][20][21][22] The main advantages being the wide availability of CT, the ability to visualize an entire area at the same time, and to perform 3D-reconstructions. Disadvantages are the necessity for deep sedation or general anaesthetic, cost, risk of adverse reactions to the iodinated contrast and its possible accumulation within the injected tissues and associated LNs for a prolonged period of time, 23 difficulty in determining the ideal time to scan, and in complying with the As Low As Reasonably Achievable radiation safety principle. 19 Other SLN mapping techniques have been developed in human oncology, including contrast-enhanced ultrasound (CEUS) or lymphosonography, a non-ionizing and non-invasive technique. 24 The potential of this technique was validated in canine models, [25][26][27][28] prior to application in human patients diagnosed with breast cancer and cutaneous melanoma. [29][30][31][32][33][34] The main advantages are the wide availability of CEUS, low cost, safety and rapid clearance of the contrast, 35 no requirement for general anaesthetic, quick contrast diffusion to the LNs in real-time, 27 the ability to view the fine details of the lymphatics and LNs, 36 and minimal spillover of the contrast to secondary nodes. 28,37 The main disadvantages are the need to select which lymphatic basins to scan, and inter-operator variability. This technique seems however particularly advantageous for dogs with MCTs as abdominal ultrasound (US) and liver/spleen US-guided FNAC are already often performed for their staging. 3,[38][39][40] Our hypothesis was that CEUS SLN mapping could be integrated into the routine staging of canine MCTs, from both practical and technical aspects. The primary objective was to determine the safety and effectiveness of CEUS to identify SLNs. Secondary objectives were to determine if the SLNs identified by CEUS coincided with the regional LN predicted by the anatomical lymphosomes, if previous excision of the MCT would alter CEUS SLN findings, and if CEUS uptake patterns could identify MCT nodal metastasis.

| Case selection
All dogs included in this feasibility study were presented to the Specialist Oncology service of the University of Edinburgh between June 2017 and March 2019. Dogs were eligible if they were undergoing staging of a cytologically or histologically diagnosed MCT. Dogs with scars of already excised MCTs were also eligible, providing no regional LN was extirpated.
Owners signed an informed consent form before enrolment of their dog.
The study design, procedure protocol, and informed owner consent form were approved by the Institutional Veterinary Ethical Review Committee.

| Study design
Clinical information was collected, including: location of the primary MCT recorded on a body map, presence of negative prognostic factors (location, recurrence, clinical behaviour, histological grade, proliferation marker, multinucleation and infiltrative pattern for subcutaneous MCTs), size of the mass and/or scar, subjective palpation of the LNs, closest LN to the MCT and the regional LN predicted by the anatomical lymphosomes. A standard list of lymphatic basins to be scanned per body area was suggested and adapted to each dog (Supplementary form 1). Dogs underwent routine procedures for staging of their MCT including serum biochemistry and haematology, thoracic radiographs, abdominal US, liver/spleen FNAC, and regional LN(s) FNAC (including the regional LN predicted by the anatomical lymphosomes, the SLN(s) identified by CEUS, and any other LNs of interest identified by the clinician). Dogs received an intramuscular injection of chlorpheniramine (4 mg for dogs <15 kg, 8 mg for dogs ≥15 kg) before lymphosonography, which was performed at the end of the staging workup. Fine-needle aspiration of the identified SLN(s) or any other LN(s) of interest identified by the clinician was performed after lymphosonography. Dogs were checked for any potential adverse events by a clinician immediately after lymphosonography and at least 1 hour later and by owners thereafter. Any identified SLNs were advised to be extirpated, but this was not necessary for study inclusion. When it was performed, SLN(s) extirpation was planned after lymphosonography, together with MCT resection or scar re-excision if indicated. Information on the staging procedures and their results, and histological characteristics of the MCTs were collected. Cytological samples and histological sections from identified SLNs were reviewed and classified according to previously published standardized criteria by a board-certified clinical pathologist (P.C.) and by a board-certified anatomical pathologist (A.M.), respectively. 41,42 Prior to review, histological sections were stained with toluidine blue.

| Lymphosonography
The contrast agent SonoVue (Bracco Imaging, Milan, Italy), based on stabilized sulphur hexafluoride microbubbles surrounded by a phospholipid shell with a mean size of 2.5 μm, was used for this study. 43 A commercially available US machine (MyLab Twice Esaote, Genova, Italy) was used with an electric microconvex probe. In all examinations, the transmission frequency ranged from 3 to 10 Mhz for B-mode imaging, and the manufacturer's preset for contrast imaging was used with a low mechanical index of 0.1, a frequency of 3 Mhz, gain 50%, and a single focus on the lymph node. Images were registered as still images and video clips.

| Data analysis
The safety of lymphosonography was assessed based on the documentation of adverse events reported by the clinicians and owners. The effectiveness of lymphosonography was assessed by determining: the SLN detection rate, the prevalence of nodal metastasis detected, and any discrepancy recorded during staging (ie, distant or regional MCT metastasis in the absence of SLN metastasis). The SLN(s) identified by CEUS were compared with the regional LN predicted by the anatomical lymphosomes and with the closest LN to the MCT, and were used to

| Cell line validation statement
No cell line was used in this study.

| RESULTS
Fifty-nine dogs with 62 MCTs were enrolled in this study (Table 1) and had CEUS SLN mapping integrated into their routine staging procedures ( Figure 1   and aberrant skin lymphatic drainage were also noted ( Figure 2).

| SLN sonographic characteristics
The median total amount of contrast administered was 2 mL (range,   been confirmed with iodinated contrast in one study investigating CT lymphangiography in dogs, since second and third order LNs were commonly identified. 19 It is assumed that the large particle size of the contrast agent used with lymphosonography prevents its uptake by second order LNs. In addition, it has also been noted that the nodal macrophages can trap (phagocytose) the contrast agent. 27,37,49 Nonetheless, efferent lymphatic tracts were detected in a recent study 50 Tumour size (mm) .039 Remarkably, after completion of this feasibility study, another of our canine patients with a large MCT on the thigh also had a colic SLN identified. We also report an accessory axillary SLN as an interval node to the axillary SLN. Both axillary and accessory axillary nodes had histological early metastasis, which highlights the importance of identifying interval nodes even if they probably occur less frequently in dogs than in human. 53 The prevalence of histological nodal metastasis in dogs with

Median
MCTs was 60%, which is similar to what was previously reported using regional lymphoscintigraphy combined with intra-operative lymphoscintigraphy and blue dye. 15,16 It is however higher than the 42% previously reported in our institution when SLN mapping was not part of the routine staging, 12 and the 45.9% reported in another study. 13 This supports the effectiveness of CEUS SLN mapping; but unlike in these two retrospective studies, in this current study all histological sections were stained with toluidine blue, which may also have affected these results.
Remarkably, the prevalence of nodal metastasis for subcutaneous MCTs was 60%, whilst it was previously reported to be only 4%. 54 The prevalence of nodal metastasis for cutaneous MCTs was also higher in our study compared with another large retrospective study, 55 in particular for Patnaik grade 2 MCTs (16.2% in previous publication 55 and 62.5% in this study). It is difficult to draw any conclusion on the impact of SLN mapping from this, since the numbers in this feasibility study were small and the quality of LN assessment was variable in these retrospective studies. Nonetheless, this highlights the marked difference in the reported prevalence of nodal metastasis depending on the regional LN(s) staging procedures performed. Following the use of CEUS SLN mapping and standard histological assessment, none of the grading systems nor any other prognostic factors assessed were significantly associated with nodal metastasis.
Only the size of the primary tumour was associated with LN metastasis, but tumours as small as 3 mm could still be metastatic. In two previous studies using SLN mapping 15

Data presented in part at the European College of Veterinary Internal
Medicine-Companion Animal Annual Congress, Milan, Italy, 2019.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.