P40 1H NMR Spectroscopy can distinguish steatotic liver disease and HCC in steatotic liver disease in a prospective patient cohort

<h3>Introduction</h3> Steatotic Liver Disease (SLD) is the most common cause of Hepatocellular Carcinoma (HCC) in many Western countries. Ultrasound sensitivity to detect lesions in steatotic liver parenchyma is low (33% for early tumours).¹Nuclear Magnetic Resonance (NMR) spectroscopy offers a robust and reproducible method of interrogating the plasma metabolome for early cancer detection. We investigated whether the plasma metabolome could be used to distinguish HCC in SLD from ‘at-risk’ SLD controls. <h3>Methods</h3> Plasma samples from a prospective, predominantly cirrhotic SLD patient cohort n = 45 (n = 10 HCC+SLD (cirrhosis n = 9), n = 35 SLD (cirrhosis n = 27) (cirrhosis p = 0.65) were analysed using untargeted 1H NMR spectroscopy. A semi-supervised OPLS-DA model incorporating a k*l cross-validation scheme with 100-fold external cross-validation was used to discriminate between HCC and SLD control patients. Specific discriminatory metabolites were identified using variable importance projection scores derived from the OPLS-DA model. <h3>Results</h3> The OPLS-DA predictive model (figure 1) achieved sensitivity 89.3% (SD +/- 11.1%), specificity 77.9% (SD +/- 9.6%) and accuracy 83.7 (SD +/- 8.3%) for the diagnosis of HCC in SLD. Increased plasma glucose in combination with decreased mobile lipid groups within lipoproteins were identified as the metabolic changes best discriminating between HCC and SLD control patients. Univariate analysis confirmed significant between-group differences in mobile lipid groups; (=CH-CH2-CH2-) p = 0.0003, beta (-CH2-) p = 0.0009, Triglyceride/VLDL (-CH2-) p = 0.0016, alpha (-CH2-) p = 0.0002, (-CH3) p = 0.0098, unsaturated (-CH=CH-) p = 0.0016; and glucose p = 0.0005. No significant associations were observed between the OPLS-DA model’s discriminatory score and parameters of age, gender, diabetes, cirrhosis, albumin, bilirubin and prothrombin time. The OPLS-DA model’s discriminatory score was independent of tumour size. <h3>Discussion</h3> This exploratory metabolomic model demonstrated promising discriminatory performance in a cohort of SLD patients, for whom the benefit of ultrasound-based HCC surveillance remains contentious. Whether the observed metabolic dysregulation of lipids and carbohydrates represents a direct tumour effect or an HCC-associated change to the background metabolism is an avenue for ongoing research. <h3>Reference</h3> Tzartzeva K, <i>et al. Gastroenterology</i>. 2018;<b>154</b>:1706–1718.e1. doi:10.1053/j.gastro.2018.01.064.