PTH-123 Urinary volatile organic compounds as a biomarker for hepatocellular carcinoma (HCC)

Abstract

<h3>Introduction</h3> Primary liver cancer is the third largest contributor to cancer mortality in the world, with more than half a million people worldwide diagnosed each year. Over the last 3 decades the incidence and mortality of HCC has been steadily increasing. The major factors contributing to survival are the degree of the underlying liver disease and the size of the tumour at diagnosis. Therefore early diagnosis is vital to improve outcomes. Volatile organic compounds (VOCs) have been investigated as biomarkers of a range of disorders. We report the first pilot study of urinary VOCs as biomarkers for HCC. <h3>Method</h3> A 400 ul urine sample was collected from 60 patients with a confirmed diagnosis of cirrhosis: 33 patients without and 27 patients with HCC. Urine samples were subsequently freeze-dried and analysed by Gas Chromatography – Mass Spectrometry. VOCs were identified and quantified using an in-house pipeline involving the Automated Mass Spectral Deconvolution and Identification System (AMDIS), the NIST mass spectral library and the R package Metab. Statistical analyses were performed using the R software and the web-tool Metaboanalyst. <h3>Results</h3> A total of 118 VOCs were identified across all samples. There were significantly higher number of VOCs identified in the cirrhotics without HCC compared to those with HCC (<i>t</i>-test, <i>p-value = 0.034</i>). Categorical comparison for prevalence demonstrated methanethiol to be less prevalent VOC in those with HCC (<i>t</i>-test, <i>p-value = 0.0083</i>). Comparison of VOCs abundances reported methanethiol at significantly higher levels in patients without HCC (<i>t</i>-test, <i>p-value = 0.0088</i>) (Figure 1). PLS-DA of those with HCC showed great separation between those with alcohol and viral hepatitis as the underlying aetiology. A supporting vector machine (SVM) model was then developed using methanethiol and additional selected features of qualitative data. After validation using a repeated 10 fold cross validation scheme, the SVM model returned a mean accuracy of 80% for the diagnosis of HCC. <h3>Conclusion</h3> The results presented here are in agreement with the literature. In particular, the decrease of methanethiol appears to have the most promise as a biomarker for the development of HCC. The model developed here showed great accuracy in classifying urine samples from patients with HCC, which indicates a great potential of VOCs as diagnosis or screening tool. A larger study would be required to investigate it further. <h3>Disclosure of interest</h3> None Declared.