Recognition of breathprints of lung cancer and chronic obstructive pulmonary disease using the Aeonose electronic nose.

OBJECTIVES

There is a high unmet need in a non-invasive screening of lung cancer (LC). We conducted this single-center trial to evaluate the effectiveness of the electronic nose Aeonose in LC recognition.

MATERIALS AND METHODS

Exhaled volatile organic compound (VOC) signatures were collected by Aeonose in 42 incident and 78 prevalent LC patients, of them 29 LC patients in complete remission (LC CR), 33 healthy controls (HC) and 23 COPD patients. By dichotomous comparison of VOC's between incident LC and HC, a discriminating algorithm was established and also applied to LC CR and COPD subjects. Area under Curve (AUC), sensitivity, specificity and Matthews's correlation coefficient (MC) were used to interpret the data.

RESULTS

The established algorithm of Aeonose signature allowed safe separation of LC and HC, showing an AUC of 0.92, sensitivity of 0.84 and a specificity of 0.97. When tested in a blinded fashion, the device recognized 19 out of 29 LC CR patients (=65.5%) as LC-positive, of which only five developed recurrent LC later on (after 18.6 months [Formula: see text]; mean value [Formula: see text]). Unfortunately, the algorithm also recognized 11 of 24 COPD patients as being LC positive (with only one of the 24 COPD patients developing LC 56 months after the measurement).

CONCLUSION

The Aeonose revealed some potential in distinguishing LC from HC, however, with low specificity when applying the algorithm in a blinded fashion to other disease cohorts. We conclude that relevant VOC signals originating from comorbidities in LC such as COPD may have erroneously led to the separation between LC and controls.

CLINICAL TRIAL REGISTRATION

(NCT02951416).