Exhaled breath olfactory signature of pulmonary arterial hypertension (PAH) and chronic thromboembolic pulmonary hypertension (CTEPH)

Background: PH is diagnosed by right heart catheterization, an invasive procedure not suitable for large-scale screening. In a small proof-of-concept study we have shown that a bio-inspired intelligent artificial olfactory system (e-nose based on cross-reactive nanoarray and pattern recognition) could distinguish PAH from controls (Cohen-Kaminsky, AJRCCM 2013). Aims: The SNOOPY2 study (NCT02782026) main objective was to evaluate the performance of a novel e-nose for the detection of PAH. We also tested whether the e-nose could detect CTEPH, BMPR2 mutation carriers and high-risk PAH. Methods: In 273 subjects, e-nose signals were analyzed with discriminant factor analysis and a K-fold cross-validation method, to test the accuracy of the models. A subset of samples was also analyzed through GC-MS, to identify discriminant volatile organic compounds (VOCs). Results: E-nose discriminated between breaths from PAH (n=73) and controls (n=121) with an accuracy of 72.1 %, a specificity of 81.8 % and a NPV of 75.6 %. CTEPH (n=41) were distinguished from controls (accuracy 80.3 %, sensitivity 89.4%, NPV 86.9%). PH patients (CTEPH and PAH) were discriminated from controls (NPV 73.4%). E-nose was able to classify patients with and without BMPR2 mutation (accuracy 75.6%). It was also possible to distinguish ESC/ERS guidelines “low-risk” and “high-risk” PAH (sensitivity 92%, NPV 88%). GC-MS revealed 10 discriminant VOCs between PAH and controls and 1 discriminant VOC between CTEPH and controls, that might contribute to the e-nose signature. Conclusions: E-nose is a promising tool for the diagnosis and management of PH. *equal contribution first 3 and last 3 authors