Correlation between computed tomography and histological evaluation of nodal metastasis in dogs with mast cell tumours

Abstract Early diagnosis of nodal metastasis has been shown to impact prognosis for dogs with mast cell tumours (MCT). The objective of this retrospective study was to determine the correlation between computed tomographic characteristics of lymph nodes and histologic nodal metastasis using the HN classification system, in dogs with cutaneous or subcutaneous MCT and regional lymph node(s) removal. Dogs that had removal of MCT and regional lymphadenectomy within 31 days of the initial staging computed tomography (CT) were enrolled. Subjective lymph node characteristics used included margination, loss of fat at hilus, shape of margin, perinodal fat pattern, increase in number of nodes, and pre‐ and post‐contrast heterogeneity. Enhancement, heterogeneity, and short‐long axis ratio were calculated. Seventy‐one lymph nodes from 37 dogs were included. Generalised linear mixed model of assessment of lymph node was performed twice, with binary outcome [non‐metastatic (HN0/1) versus metastatic (HN2/3)] and 4‐point scales (HN0–HN3). After blind assessment of 7 characteristics described above, a final subjective interpretation of each lymph node as non‐metastatic or metastatic was assigned. A significant correlation was found between final interpretation and prediction of metastasis. Higher HN classification was also significantly correlated with the increased number of nodes and pre‐ and post‐contrast heterogeneity. No correlation was found in short‐long axis ratio, calculated heterogeneity, or degree of enhancement. Sensitivity of CT was 35.7%, specificity was 96.6%, and accuracy was 60.5% for nodal metastasis. CT alone cannot be recommended for assessment of metastasis. The use of multiple computed tomographic characteristics may increase accuracy of nodal metastasis detection.