Lyso‐phosphatidylcholine: A potential metabolomic biomarker for alcoholic liver disease?

Potential conflict of interest: Nothing to report. AS was funded by European Social Fund through the Romanian Human Resources Development Operational Programme 2007‐2013 (grant number: POSDRU/159/1.5/S/138776). AS was funded by European Social Fund through the Romanian Human Resources Development Operational Programme 2007‐2013 (grant number: POSDRU/159/1.5/S/138776). To the Editor: We read with great interest the work of Michelena et al.,1 which demonstrated that in patients with alcoholic hepatitis the presence of a systemic inflammatory response, with or without infection, precipitates development of complications and death. By identifying lipopolysaccharide (LPS) as an important player in the inflammatory response, the authors open a novel gateway for identifying the inflammatory mediators that cause this systemic response. Historic reports2 found an inverse association between lyso‐phosphatidylcholine (LYPC) and the amount of liver fibrosis in an experimental model of alcoholic liver disease (ALD). More recently3 lower LYPC levels were found in nonalcoholic fatty liver disease/nonalcoholic steatohepatitis patients compared with controls. The link between LPS and LYPC is considered to be phosphatidylcholine‐specific phospholipase C, the levels and activity of which are increased by LPS.4 Therefore, we hypothesized that LYPC may be a biomarker for severity and short‐term prognosis in patients with ALD. Thirty consecutive patients admitted for ALD were included. Of them, 17 were known to have cirrhosis and 7 had previous clinical liver‐related decompensation (LRD) events. Patients were followed for 30 days for the development of LRD and/or death. At inclusion 16 patients (12 of whom were known to have cirrhosis) had an alcoholic steatohepatitis test score (BioPredictive, France) ≥0.18, indicating at least minimal activity; the test was used in addition to clinical and biological criteria for the diagnosis of alcoholic hepatitis: recent alcohol abuse, leukocytosis, aspartate aminotransferase/alanine aminotransferase >1. High‐performance liquid chromatography‐mass spectrometry was used to determine baseline levels of two LYPC serum metabolites: 16:1 and 20:4. Twelve patients decompensated during follow‐up, 6 of them having multiple LRDs; 2 patients died. Both LYPCs had lower levels in patients with LRD compared to those without (0.277 [0.195−0.737] versus 0.517 [0.166−0.987], P = 0.0001, and 0.564 [0.312−2.116] versus 1.302 [0.795‐3.492], P = 0.008, respectively). The performance of the two LYPC metabolites to predict LRD is summarized in Table 1. They also had a good inverse correlation with Model for End‐Stage Liver Disease (−0.602; −0.619) (see Supporting Fig. S1); Maddrey (−0.632; −0.656); and age, serum bilirubin, international normalized ratio, and serum creatinine (ABIC) (−0.404; −0.529) scores and with liver stiffness by transient elastography (−0.608 and −0.645). No correlation was found with daily alcohol intake (−0.384 and −0.321; P = 0.07 and 0.135, respectively). Table 1 - Area Under the Receiver Operating Characteristic Curve of LYPC 16:1 and LYPC 20:4 to Predict LRD Events in a Cohort of 30 Patients With ALD, During a 30‐Day Follow‐Up LYPC 16:1 LYPC 20:4 Ascites (n = 8) 0.830 (95% CI 0.60‐1), P = 0.007 0.858 (95% CI 0.66‐1), P = 0.003 Bleeding (n = 5) 0.760 (95% CI 0.59‐0.92), P = 0.05 0.816 (95% CI 0.62‐1), P = 0.02 Infections (including SBP) (n = 5 [1]) 0.827 (95% CI 0.68‐0.97), P = 0.03 0.758 (95% CI 0.57‐1), P = 0.05 Encephalopathy (n = 4) 0.952 (95% CI 0.87‐1), P = 0.004 0.933 (95% CI 0.83‐1), P = 0.006 Death (n = 2) 0.911 (95% CI 0.79‐1), P = 0.05 0.946 (95% CI 0.86‐1), P = 0.03 Multiple LRD (n = 6) 0.866 (95% CI 0.69‐1) 0.894 (95% CI 0.75‐1) Cut‐off <0.374 MU <0.747 MU Sensitivity 0.839 1 Specificity 0.917 0.75 Abbreviations: CI, confidence interval; MU, mega‐units; SBP, spontaneous bacterial peritonitis. Using a combined approach (both metabolites below the specified cutoff value), LYPC showed a good accuracy to predict liver decompensation (27/30 patients correctly classified, 100% positive predictive value, and 85.7% negative predictive value) and to exclude 30‐day mortality (23/30 patients correctly classified, 22.2% positive predictive value, and 100% negative predictive value). In conclusion, our pilot data indicate that LYPC is a biomarker for severe ALD, useful to predict the development of liver related complications and to exclude short‐term mortality. The connection with the LPS pathway needs to be further validated.