O04 Serum nuclear magnetic resonance metabolomic signature can discriminate immunoglobulin G4-related sclerosing cholangitis and primary sclerosing cholangitis

Abstract

<h3></h3> Immunoglobulin G4-related sclerosing cholangitis (IgG4-SC) is often difficult to distinguish from primary sclerosing cholangitis (PSC) in clinical practice. Accurate, non-invasive biomarkers for discriminating IgG4-SC from PSC are required, but the diagnostic utility of global serum nuclear magnetic resonance (NMR)-based metabolomics is untested. We performed serum metabolomic profiling in patients with IgG4-SC and PSC to assess for evidence of a distinctive signature that could discriminate between these conditions, predict response to therapy and provide a biomarker for disease relapse. Stored serum samples collected prospectively from patients with IgG4-related disease (IgG4-RD; n=39 at diagnosis prior to therapy), PSC (n=100; 81 large duct; 19 small duct) and healthy controls (HC; n=16) were prepared for NMR spectroscopy using a standardised protocol. Principal component analysis (PCA) and orthogonal partial-least squares discriminatory analysis (OPLS-DA) were used to identify discriminatory serum metabolites. Clinical data was obtained from review of electronic databases for correlation with metabolomics data and to adjust for confounders. The median (range) age and gender proportion were 65 (33–83) years and 85% male for IgG4-RD, and 45 (12–84) years and 63% male for PSC. Lactate, glucose, and glutamine were increased in IgG4-RD compared to PSC (p&lt;0.0001), whereas -CH3 lipoprotein and beta-hydroxybutyric acid resonances were decreased (p&lt;0.001). NMR-based metabolomic profiling discriminated IgG4-RD from PSC with greater accuracy (AUC 0.941, 95% CI 0.904–0.977) than upper limit normal of IgG4 titre (AUC 0.865, 95% CI 0.787–0.943). OPLS-DA modelling with an optimised threshold diagnosed IgG4-RD – as opposed to PSC – with an accuracy of 86%, sensitivity of 96%, and specificity of 70% (figure 1), adjusted for age, gender, comorbidities, serum IgG4 level and medications. Both IgG4-RD and PSC were independently distinguishable from HC with an accuracy of 96% and 91%, respectively. When only IgG4-SC patients (n=23) were included with large-duct PSC patients (n=81), the accuracy was 88%. When IBD was excluded as a comorbid condition (IgG4-SC n=20, PSC n=22), the diagnostic AUC was 0.998 (0.991–1.000). The metabolomic signature determined by serum NMR in patients with IgG4-RD and more specifically IgG4-SC, is distinct from PSC and HC in our cohort. Metabolomic profiling has the potential to be incorporated as an additional criterion to improve the diagnosis of IgG4-RD and help distinguish IgG4-SC from PSC.