Hemangiosarcoma (HSA) is a common, aggressive, and deadly vascular cancer in dogs that is usually diagnosed only at advanced stages. Because treatment options are limited once HSA is advanced, early detection is essential to improving survival and quality of life. Five trained bio-detection dogs were evaluated using double-blinded tests with automated olfactometer line-ups containing blood serum samples from dogs with confirmed HSA, non-cancerous diseases other than HSA (diseased controls), and healthy controls. All test samples were novel to the dogs. Across all 423 blinded trials, accuracy was 70.0 % (range = 57.1-78.6 %). First-trial accuracy, representing each dog's initial response to a novel matched sample set, averaged 70.0 % (range = 58.3-83.3 %). When considering each dog's first encounter with each sample, dogs achieved an overall sensitivity of 70.0 % and specificity of 70.0 %. A mixed-effects logistic regression showed that dogs alerted to HSA samples in 73.4 % of presentations, compared with 21.3 % of diseased controls and 17.1 % of healthy controls. Dogs were over 10 times more likely to alert to HSA than to diseased controls (OR = 10.2, p < .001) and over 13 times more likely than to healthy controls (OR = 13.3, p < .001). This study finds that trained dogs can distinguish serum samples from dogs with HSA from those of healthy and diseased controls, indicating that HSA produces a detectable odor signature. Conclusions are constrained by the limited number of HSA samples. These results suggest a potential feasibility of VOC-based detection for canine HSA.
Keywords: Biodetection; Canine cancer; Hemangiosarcoma; Olfaction; Volatile organic compounds.
Copyright © 2025 The Authors. Published by Elsevier Ltd.. All rights reserved.