Contrast-Enhanced Computed Tomography for Identification of Potential Malignancy in Canine Colorectal Mesenchymal Tumours.

Cross-sectional imaging may be used to characterise the location and extent of colorectal mesenchymal tumours (CRMTs). Given the anticipated variation in tumour behaviour and varying morbidity based on surgical margins, a reliable, non-invasive means of predicting malignant potential could facilitate case management. The purpose of this multi-institutional, retrospective study was to determine the diagnostic accuracy of contrast-enhanced CT for distinguishing benign and malignant CRMTs. Twenty-seven dogs with CRMTs were included. Initial diagnoses were reviewed, and slides or blocks were available for 24/27 dogs for further histologic review and immunohistochemical labelling for smooth muscle actin, KIT and vimentin. Two masked radiologists reviewed DICOM images for tumour characteristics, including a final, binary, consensus, subjective interpretation of malignancy. Eighteen tumours (66.7%) were classified as leiomyoma, one (3.7%) as a benign other non-lymphogenic intestinal mesenchymal tumour (benign), one (3.7%) as leiomyosarcoma, and seven (25.9%) as gastrointestinal stromal tumour (malignant). Agreement between radiologists ranged from none to weak for categorical variables, with no agreement (κ = 0.135) for the final assessment of a tumour as benign or malignant. Substantial overlap was noted between groups, with no single categorical variable demonstrating high accuracy as a predictor of malignancy. Consensus final assessment was a sensitive (80.0%) but not specific (29.4%) predictor of malignancy (accuracy: 48.2%). No association was identified between CT determination of malignancy and histologic determination of malignancy (p = 0.678). Non-standardised contrast-enhanced CT was ineffective at distinguishing malignant from benign CRMTs in this study.