TOPLINE:
A multicenter prospective study validated that an electronic nose (eNose) can accurately detect lung cancer in individuals with suspected disease. In the validation cohort, a new eNose model achieved a sensitivity of 94% (true positives) and specificity of 63% (true negatives) for detecting lung cancer.
METHODOLOGY:
- The eNose breath analysis offers promise as a noninvasive, cost-effective tool for diagnosing lung cancer, but robust external validation studies have been limited.
- A multicenter prospective study evaluated 364 adults with clinical or radiologic signs of lung cancer between March 2019 and November 2023. Overall, 59% of these patients had lung cancer diagnoses confirmed, and the rest had either no cancer or a different type of cancer.
- Exhaled breath analysis was performed using a cloud-connected eNose device (SpiroNose) after participants underwent blood tests, spirometry, and PET-CT.
- An original eNose model was validated in all participants (n = 364) and in those with chronic obstructive pulmonary disease (COPD; n = 116). A new eNose model specifically tailored to a more general population with suspected lung cancer was developed and tested using training (n = 242) and validation (n = 121) cohorts.
- Diagnostic accuracy for both models was assessed using receiver operating characteristic area under the curve (ROC-AUC), specificity, positive predictive value (PPV), and negative predictive value (NPV), with sensitivity targeted at 95%.
TAKEAWAY:
- In patients with COPD, the original eNose model detected lung cancer with an ROC-AUC of 0.92. The sensitivity was 95%, the specificity was 72%, PPV was 95%, and NPV was 72%.
- In the overall population, the original model had an ROC-AUC of 0.80, with a sensitivity of 95%, a specificity of 51%, a PPV of 74%, and an NPV of 88%, accurately identifying most lung cancer cases but performing less well in correctly identifying those without lung cancer.
- In the validation cohort, the new model achieved an ROC-AUC of 0.83 for detecting lung cancer across all participants, with a sensitivity of 94%, a specificity of 63%, a PPV of 79%, and an NPV of 89%.
- The new model also identified 23% of those with other cancers as well as 80% of those with benign diagnoses and no cancer. Performance was consistent across tumor types, disease stages, and clinical sites.
IN PRACTICE:
This “multicenter prospective external validation study confirms that eNose analysis of exhaled breath enables reliable, noninvasive, point-of-care detection of lung cancer in individuals suspected of having lung cancer,” the authors wrote. “Results suggest that implementation of eNose analysis could avoid performing unnecessary invasive diagnostic procedures in approximately 75% of individuals with a benign diagnosis, at a 5%-6% risk of withholding a lung cancer diagnosis at thoracic oncology outpatient clinics.”
SOURCE:
This study, led by Alessandra I.G. Buma, MD, Radboud University Medical Center, Nijmegen, the Netherlands, was published online in the Annals of Oncology.
LIMITATIONS:
Diagnostic subgroup classification relied partly on radiologic and expert judgment, which could have led to misclassification. In addition, the new eNose model was only internally validated.
DISCLOSURES:
This study did not receive any specific funding. Several authors reported receiving personal fees or grants from or having other ties with various sources.
This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.