BACKGROUND: Idiopathic non-cirrhotic portal hypertension (INCPH) is a rare life-threatening liver disease that lacks a specific diagnostic test being frequently misdiagnosed as cryptogenic cirrhosis. Preliminary data from our group identified a plasma metabolomic profile able to differentiate INCPH from patients with cirrhosis (CH) and healthy volunteers (HV). However, the untargeted methodology applied was unable to identify all the specific metabolites, hampering the possibility of building-up diagnostic models. This study applies a wide-coverage of previously identified metabolites through a high-throughput metabolomics technology, evaluating if there is a metabolomic profile that allows a non-invasive diagnosis of INCPH. METHODS: We included 34 patients with INCPH, 34 with CH and 34 HV. We performed a targeted metabolomic analysis of serum samples using UPLC-MS. The best combination of a set of specific metabolites was obtained using stepwise logistic regression (LR) and recursive partitioning analysis (RPA). RESULTS: After internal cross-validation, LR analysis identified a subset of 5-metabolites that clearly differentiate INCPH patients from CH and HV (average corrected optimism AUROC = 0.8871 [0.838-0.924]). Using high and low cut-off values the model has an excellent capacity to respectively diagnose or exclude INCPH. The RPA analysis strategy used the 3-metabolites signature differentiating INCPH from CH and the 2-metabolites signature differentiating INCPH from HV. A decision tree applying sequentially these metabolic profiles diagnosed 88% of INCPH patients. CONCLUSIONS: Different metabolomic profiles allow the diagnosis of INCPH with high specificity and sensibility and may represent excellent clinical tools for its diagnosis avoiding multiple and invasive tests.
BACKGROUND: Idiopathic non-cirrhotic portal hypertension (INCPH) is a rare life-threatening liver disease that lacks a specific diagnostic test being frequently misdiagnosed as cryptogenic cirrhosis. Preliminary data from our group identified a plasma metabolomic profile able to differentiate INCPH from patients with cirrhosis (CH) and healthy volunteers (HV). However, the untargeted methodology applied was unable to identify all the specific metabolites, hampering the possibility of building-up diagnostic models. This study applies a wide-coverage of previously identified metabolites through a high-throughput metabolomics technology, evaluating if there is a metabolomic profile that allows a non-invasive diagnosis of INCPH. METHODS: We included 34 patients with INCPH, 34 with CH and 34 HV. We performed a targeted metabolomic analysis of serum samples using UPLC-MS. The best combination of a set of specific metabolites was obtained using stepwise logistic regression (LR) and recursive partitioning analysis (RPA). RESULTS: After internal cross-validation, LR analysis identified a subset of 5-metabolites that clearly differentiate INCPH patients from CH and HV (average corrected optimism AUROC = 0.8871 [0.838-0.924]). Using high and low cut-off values the model has an excellent capacity to respectively diagnose or exclude INCPH. The RPA analysis strategy used the 3-metabolites signature differentiating INCPH from CH and the 2-metabolites signature differentiating INCPH from HV. A decision tree applying sequentially these metabolic profiles diagnosed 88% of INCPH patients. CONCLUSIONS: Different metabolomic profiles allow the diagnosis of INCPH with high specificity and sensibility and may represent excellent clinical tools for its diagnosis avoiding multiple and invasive tests.