Feasibility study on training dogs to detect lung cancer: findings of a retrospective evaluation

Early detection is critical for lung cancer patients. One lung cancer detection method under study is using sniffer dogs. This study aimed to evaluate, retrospectively, the sensitivity and specificity of theCancerDetectionDogCollective (CDDC®) method under training conditions. A team of five trained sniffer dogs analyzed breath samples from lung cancer patients and cancer-free volunteers, and a cancer sample is positive if at least three dogs indicate it. Dog handlers and experimental observers were blinded to sample identity, and detection accuracy was assessed. Primary endpoint was sensitivity, and specificity and confounding factors were also assessed. Samples were collected in 2024 from 824 volunteers, including 111 with a confirmed diagnosis of lung cancer (mean age 60, range 34-80, 18% early-stage cancer, 46% not yet oncological treated). A total of 11 900 breath samples were tested with 125 test runs per dog. Each of the five dogs demonstrated a detection performance with a sensitivity between 82% and 89%, a specificity over 95%, and an accuracy over 94%. The CDDC® dog team's corporate decision revealed a sensitivity of at least 95.5%. The cancer-free volunteers were primarily young, healthy individuals. According to the CDDC® decision rules, none of these control samples were identified as false positives by more than two dogs. Analysis of potential confounding factors revealed that weather conditions and supervisor skills were associated with the dogs' performance. The CDDC® method showed high consistency in training scenarios. Further studies should evaluate this method in a controlled clinical study alongside lung cancer screening.