Clinical and CT sialography findings in 22 dogs with surgically confirmed sialoceles

Abstract Sialoceles are an uncommon canine salivary gland disease, and complete surgical resection is important for a positive outcome. Radiographic sialography has been described as a diagnostic test for presurgical planning; however, superimposition artifacts may limit the diagnosis and detection of all affected glands. Computed tomographic (CT) sialography is a promising technique for delineating the salivary gland apparatus. The aims of this retrospective, observational study were to describe clinical and CT sialographic findings in a group of dogs with confirmed sialoceles, to determine the sensitivity of CT sialography for detecting affected salivary glands using surgery as the reference standard and to determine interobserver agreement for CT sialographic assessments. Dogs were included if they underwent a CT sialography study followed by surgical resection of the diseased gland(s) and histopathological analysis. Computed tomography sialography studies of dogs with surgically confirmed sialoceles (n = 22) were reviewed by a European College of Veterinary Diagnostic Imaging (ECVDI)‐certified radiologist and an ECVDI resident. Interobserver agreement was calculated using Cohen's kappa statistics. CT sialography results were compared to surgical findings to determine sensitivity. Contrast leakage was detected in 12 of 22 dogs (54.5%), with intrasialocele leakage being most frequently observed (7/12, 58.3%). There was substantial agreement (κ = 0.70) between reviewers identifying diseased glands, substantial agreement (κ = 0.62) on the diagnostic quality, and no to slight agreement (к = 0.13) in the detection of contrast leakage. The overall sensitivity of CT sialography to detect surgically confirmed diseased glands was 66.7% (95% confidence interval: 48.8‐80.8). In conclusion, these findings support the use of CT sialography as an adjunct diagnostic test for treatment planning in dogs with sialoceles.

Idiopathic sialocele formation, where an underlying cause has not been identified, has been more commonly reported. 12,13,16,[22][23][24][25] The salivary apparatus of the dog is complex, and surgical management of salivary gland pathology can be challenging due to the close proximity to neurovascular bundles, lymph nodes and lymphatic channels. 23 Presurgical identification of the correct diseased gland and respective ducts is important as curative treatment for sialoceles entails sialoadenectomy, i.e., complete surgical resection of the affected salivary gland and ligation of the respective salivary duct. 12,13,16 Recurrence has been reported 27 if complete resection is not achieved. 15,16,23,27 Various imaging modalities and techniques have been described to characterise the salivary gland apparatus and aid in the diagnosis of salivary gland disease. This includes conventional radiography, 28,29 contrast radiography or "sialography", 13,26 ultrasound 30,31 and computed tomography (CT). [31][32][33][34][35] Radiographic sialography has been described to aid presurgical planning in canine sialoceles. 13,26 However, with radiographic examinations of the head and skull, diagnosis is challenging due to organ superimposition. Use of CT as part of a diagnostic work-up has been reported for zygomatic mucoceles, [31][32][33] zygomatic sialolithiasis 34 and parotid duct foreign bodies, 35 reporting advantages of cross-sectional imaging in the assessment of the salivary gland apparatus. A previous cadaver CT sialography study that described the anatomy of the parotid, mandibular and zygomatic salivary glands demonstrated the valuable information CT sialography provides in the assessment of the salivary gland. 36 A recent study described the CT features of sialoceles; 37 however, sialography was not performed in these cases. Based on a current literature search, there is no study describing CT sialography features of surgically confirmed sialoceles or assessing the sensitivity of CT sialography in determining the diseased gland of origin.
Therefore, the aims of the current study were threefold: to first describe the CT sialography features in dogs with surgically confirmed sialoceles; to evaluate the diagnostic quality of the sialography studies and recognition of contrast leakage; and last, to determine the sensitivity of CT sialography in the identification of the diseased salivary gland of origin compared to surgery, taken to be the reference standard test. We hypothesized that there would be strong agreement between reviewers in the identification of diseased glands on CT sialography, evaluation of the diagnostic quality of the study, and recognition of contrast leakage. Additionally, we hypothesized that CT sialography would be a sensitive technique to identify the diseased salivary gland of origin.

Case selection
This was a retrospective single-institutional observational study.

Population
A total of 51 confirmed cases of sialoceles were retrieved from the HfSA database, with 22 dogs meeting the inclusion criteria. Figure 1 summarizes the selection process that was used for the study popu- displayed a combination of two or more of the above clinical signs.   Table 1 and were used in combination with the clinical features as follows:

CT sialography protocol
First, for submandibular or cervical lateralized swellings (cases 1-2, salivary gland 1-5 ml, sublingual salivary gland 1-5 ml, and parotid salivary gland 1-5 ml. Images were immediately acquired following this procedure under the same protocol as for the precontrast series.

Surgical findings and histopathology report analysis
Surgical exploration of the suspected diseased gland was performed.
A ventral approach with tunnelling under the digastricus muscle 12 was the approach used for sublingual or mandibular salivary glands ("submandibular complex"), a lateral approach without zygomatic arch ostectomy was adopted for the zygomatic gland, and a lateral approach with a longitudinal incision ventral to the external acoustic meatus was used for the parotid salivary glands. 23 Final histopathological diagnoses were extracted from the final diagnoses made in the report. If present, additional comments were also reviewed to confirm the final histopathological diagnoses. Sialoceles were defined as the accumulation of salivary secretions in single or multiloculated cavities not lined by secretory epithelium, and sialadenitis was defined as inflammation of the salivary glands. 40 (Figure 2A-B).

Sensitivity of CT sialography
The overall sensitivity of CT sialography to detect surgically confirmed In the cases where a single diseased gland was identified at surgery, CT sialography findings differed from the surgical findings in two cases: incorrectly identifying the mandibular gland when the sublingual was affected and vice versa.
Ten cases displayed no contrast medium leakage. In four of these cases, a sublingual sialocele was diagnosed based on a normal mandibular sialogram in combination with previously specified clinical and precontrast CT features. Figure

Diagnostic utility of CT sialography
Thirteen dogs had CT sialography performed prior to the postcontrast series. One case did not have a postcontrast study. Of the remaining eight cases where CT sialography was performed after the postcontrast study, findings from the CT sialography study differed from those

DISCUSSION
This study is the first published report describing CT sialography features as well as the sensitivity of CT sialography in surgically confirmed canine sialoceles. The findings presented in this study supported our hypothesis that there would be strong agreement between reviewers in the identification of diseased glands on CT sialography and evaluation of the diagnostic quality of the study. However, this study did not support our hypothesis that there would be strong agreement for the recognition of contrast leakage, which was detected neously contrast-enhancing walls. 37 As it was not the primary aim of the present study to describe CT features of sialoceles, not all characteristics highlighted in the previous study, such as the shape and size of the sialoceles, were investigated. Additionally, as our study focused on the identification of the diseased gland of origin, we did not classify the sialoceles according to the CT and surgical characteristics described in the previous study. 37 Interestingly, in our study, one case (case 5) also demonstrated similar nodular intrasialocele protrusions (Figure 2A,B); however, this was less common than in the previous study, and "frond-like" protrusions were not observed in the current study. 37 The results from our study are also in agreement with previous reports, with unilateral involvement more commonly reported than bilateral involvement 12,13,24 . However, in our findings, left-sided involvement was more frequent than right-sided involvement, which differed from a previous study. 37 Adjacent fat stranding was present in nine of the 23 sialoceles (39.1%), which is higher than the two out of the 13 sialoceles previously reported. 37 Our results demonstrated the presence of fat stranding in cases with or without infectious or inflammatory pathologies. Although not part of the objectives of this study, fat stranding does not appear to be a reliable feature in cases with infectious or inflammatory pathologies.
The present study also only identified the presence or absence of contrast enhancement but did not quantify the degree of enhancement. Between the cases with infectious or inflammatory pathologies and those without, the same proportion of cases (75%) displayed rim enhancement. It would be interesting to investigate whether the degree of enhancement is increased in cases with infectious or inflammatory pathologies.
The present study did not identify lymphadenomegaly even when an infectious or inflammatory pathology was present. Multiple regional lymph nodes of the head were assessed, each with individual reference ranges and scoring systems, which vary according to the size of the dog.
As a result, the authors opted to assess the lymph node size qualitatively. Additionally, other features of lymphadenopathy, such as shape and attenuation, were not evaluated. We acknowledge that this is a limitation to our study and that any incidences of lymphadenopathy may be underreported. However, in the evaluation of all 22 cases, no significant lymph node enlargement was identified between the two reviewers, one of whom was an experienced ECVDI-boarded radiologist.
As it was outside the scope of our study, the correlation between concurrent CT characteristics and identified pathologies was not cal- Our study found an overall moderate sensitivity of CT sialography in detecting surgically confirmed diseased glands. Individual sensitivities for the mandibular and sublingual glands were also moderate to high.
Due to the small numbers involving the zygomatic and parotid glands, meaningful statistical analysis could not be performed to evaluate the sensitivity of CT sialography for these glands, which is a limitation of our study. Inclusion of more cases would provide greater confidence in the sensitivity of CT sialography, or the sublingual and mandibular glands could be considered a singular unit (the submandibular complex) given that they are resected en bloc. Establishment of a standardized, optimal CT sialography protocol could also improve the overall sensitivity of the technique. Our study population consisted mainly of medium-to large-breed dogs, with the lowest weight of 8 kg. Salivary duct cannulation, especially for the submandibular gland complex, is challenging in very small dogs, which precluded their inclusion in our study. This is an area that would require further evaluation, and our findings may not rep- with routine studies of the head and neck are needed to establish the diagnostic utility of CT sialography as well as to determine the optimal CT sialography protocol. Last, limited statistics were performed in our study due to the low number of cases. Due to the selection of the study population, cases without salivary gland disease were not included; therefore, the specificity of CT sialography could not be established.
Due to these limitations, future, prospective studies with a larger cohort and population size are needed to confirm our preliminary findings. Further comparison of CT sialography with routine studies of the head and neck is needed to establish the diagnostic utility of CT sialography.

CONCLUSION
In conclusion, salivary gland pathology is uncommon and rarely reported in dogs. Common CT and CT sialography findings were identified for this sample of dogs with surgically confirmed sialoceles. Computed tomography sialography identified the diseased salivary gland or duct with an overall sensitivity of 66.7%. Between the two reviewers, there was substantial agreement (κ = 0.70) on the identification of diseased gland(s) on CT sialography, substantial agreement (κ = 0.62) on the diagnostic quality evaluation, and no to slight agreement (к = 0.13) in the detection of contrast leakage.
Future studies with larger cohorts and population sizes are needed to confirm our preliminary findings and to more definitively determine the utility of CT sialography in the surgical management of sialoceles.

Jordan Mitchell's PHD is supported by a Biotechnology and Biological
Sciences Research Council studentship (BB/M010996/1).