Electronic Nose for Distinguishing Chronic Lung Allograft Dysfunction Phenotypes

Abstract

<h3>Purpose</h3> Chronic lung allograft dysfunction (CLAD) increases risk of death of lung transplant recipients (LTR). Although the main CLAD phenotypes bronchiolitis obliterans syndrome (BOS) and restrictive allograft syndrome (RAS) differ in spirometric and radiological characteristics, establishing the final diagnosis can be challenging. Yet, differentiation between BOS and RAS is crucial, since prognosis of the phenotypes considerably differs. Timely diagnosis of CLAD and CLAD phenotypes remains problematic due to a lack of accurate markers of CLAD. Promising evidence suggests that electronic nose (eNose) technology has 86% accuracy for the detection of CLAD and may even identify CLAD phenotype. Therefore, we aimed to assess the diagnostic accuracy of exhaled breath analysis using an eNose to distinguish between CLAD phenotypes. <h3>Methods</h3> In `this cross-sectional study, exhaled breath of consecutive LTR with ISHLT criteria based CLAD was collected using an eNose (SpiroNose). Patients with mixed or undefined phenotype were excluded (n = 13). Statistical analyses were conducted using partial least square discriminant analysis and receiver operating characteristics (AUC) analysis to assess differences in breathprint between CLAD phenotypes. <h3>Results</h3> A total of 25 LTR with CLAD were included during outpatient follow-up. 56% were male, median age was 63 (range 32 - 77) years, time after LTx was 9.7 (2.5 - 18.8) years, and time till onset of CLAD was 5.8 (1.7 - 18.8) years. Based on ISHLT criteria, 20 LTR were diagnosed with BOS, and 5 with RAS. The eNose accurately discriminated between BOS and RAS with an AUC of 0.94 (CI 0.85-1.00, Figure 1), a sensitivity of 100%, specificity of 90%, and an accuracy of 92%. <h3>Conclusion</h3> The BOS and RAS phenotype differ in breathprint. Exhaled breath analysis using an eNose is a promising tool to distinguish between CLAD phenotypes. Validation of results is needed in a larger dataset, as well as assessment whether this technique allows earlier diagnosis of CLAD and its phenotype.