O24 Unbiased clustering of breath signature in NAFLD identifies disease progression high-risk patient phenotype- 5 year study

Abstract

<h3>Background and Aims</h3> Human breath contains numerous volatile compounds which reflect metabolic activity. Electronic nose (eNose) technology can distinguish non-alcoholic fatty liver disease (NAFLD) from healthy with reasonable confidence.^{1 2} We hypothesised that breath prints obtained from eNose could identify patient subgroups at higher risk for progression within the NAFLD spectrum. <h3>Method</h3> The study was a prospective single-centre cohort study (ClinicalTrials.gov: NCT02950610). eNose is a custom-made device made up of five-sensor arrays, each containing four sensors, previously validated in respiratory and liver disease.^{1 3} eNose on the exhaled breath was performed on well characterised NAFLD patients (n=60) – Child’s A NAFLD cirrhotics (n=30) and non-cirrhotic NAFLD (n=30), previously described.¹ Data were analysed using R studio (v 2.3.2) and SPSS 21. An unbiased machine learning clustering technique was applied. First hierarchical ward clustering, combined with similarity profile analysis, was used to assess the number of significant cluster groups. A 5-year cohort longitudinal data was collected with endpoints of disease progression, liver disease-related complications and all-cause mortality. Logistic regression and univariate analysis were performed to identify risk factors for endpoints. <h3>Results</h3> Data reduction to 3 principal components (PCs) explained 97.8% of the total variance. Three groups of patients with NAFLD disease were delineated solely based on their exhaled breath profiles. Three clusters were identified; cluster 1 consists of 23 patients, cluster 2 consists of 24 patients, and cluster 3 consists of 13 patients. The clusters were comparable in clinical phenotyping. Cluster 2 was identified as a higher risk group with significant differences in serum hyaluronic acid levels (<i>p</i>=0.001), endoscopic evidence for portal hypertension(<i>p</i>=0.003) and exhaled breath concentration of dimethyl sulphide (<i>p</i>=0.041) and D-limonene (<i>p</i>=0.015). Cluster 2 was associated with significant 5-year odds risk of 8.5 [95%CI 1.8 – 39.7] for disease progression and remained significant despite adjusting for age and gender. Serum hyaluronic acid, HbA1c and Neutrophil-to-Lymphocyte (NLR) ratio were other independent predictors for 5-year disease progression. 25% decompensation rate, 12.5% variceal bleeding and 12.5% all-cause mortality was noted in cluster 2 compared with 4.3% and 0% for decompensation: 0% each for variceal bleeding and 1% each for all-cause mortality in cluster 1 and 3 respectively. <h3>Conclusion</h3> This study shows that unbiased clustering of exhaled breath profiles captured using eNose technology identifies three phenotypes within the NAFLD spectrum. One of the three clusters included patients with more advanced liver disease who had significant disease progression and a higher proportion of decompensation, variceal bleeding, and all-cause mortality. These results warrant prospective studies on the potential of exhaled breath fingerprinting using eNose technology as point-of-care diagnostics and identifying high-risk disease progressors. <h3>References</h3> Sinha R, <i>et.al</i>. Electronic-nose breath print distinguishes non-alcoholic fatty liver disease from healthy lean control: A pilot study. <i>Journal of Hepatology</i> 2018;<b>68</b>:S556–S557. de Vries R, Brinkman P, van der Schee MP, et al. Integration of electronic nose technology with spirometry: validation of a new approach for exhaled breath analysis. <i>J Breath Res</i> 2015;<b>9</b>(4):046001. McDonald NSR, <i>et.al</i>. Exhaled breath profiling by electronic nose as a novel non-invasive method for assessment of chronic liver disease: proof of principle study. <i>Journal of Hepatology</i> 2016;<b>64</b>:S734–S735.