TRAINED dogs can identify lung cancer more effectively from human breath than from saliva, according to new research exploring the diagnostic potential of canine scent detection.
While previous studies have shown that dogs can detect cancer-related volatile organic compounds in exhaled breath, the relative utility of other biological samples has been less clear. To address this, researchers conducted a comparative study of breath and saliva specimens collected from 154 patients attending a general respiratory clinic. Seven dogs were trained using an automated apparatus to discriminate between samples taken from patients who were subsequently diagnosed with lung cancer and those from individuals without cancer. The study then evaluated the dogs’ performance using sensitivity and specificity measures, applying mixed-methods logistic analysis to compare the two sample types.
The findings demonstrated a significant difference in detection accuracy between breath and saliva. Dogs correctly identified lung cancer-positive breath samples with a mean accuracy of 0.78 (95% CI: 0.71–0.83), compared with only 0.42 (95% CI: 0.34–0.50; p<0.001) for saliva samples. Non-target samples were identified at similar rates for both breath and saliva, averaging 0.68, with no significant difference (p=0.854). These results indicate that breath contains more consistent or detectable volatile markers associated with lung cancer than saliva, making it a more reliable medium for canine detection.
The study confirms that breath-based sampling remains the most effective biological substrate for dogs trained to detect lung cancer, while saliva appears less informative at present. However, researchers noted that further methodological refinements, including sample handling, storage, and presentation, may enhance the utility of saliva in the future, offering a potentially more accessible medium for non-invasive screening.
Reference
Crawford MA et al. Breath versus saliva for lung cancer detection with dogs. ERJ Open Res. 2025;11(5):00914-2024.
