Ultra-High Performance Liquid Chromatography-Mass Spectrometry-based Serum Metabolomics for Early Diagnosis of Refractory Tumor-induced Osteomalacia: A Case-control Study

Abstract

Abstract Context Nearly 20% patients with tumor-induced osteomalacia (TIO) experienced recurrence or nonrecovery after surgery. Serum fibroblast growth factor 23 and phosphate concentrations are not sufficient for prognosis in such cases. Despite its importance for understanding of prognosis and underlying pathogenesis, the alteration of systemic metabolism in refractory TIO remains unclear. Objective We aimed to find the metabolomic characteristics of refractory TIO and establish a novel predictive model for early discriminating refractory TIO based on their serum metabolomics. Design and setting Cross-section study for comparison of metabolomic profile between TIO and normal control and longitudinal study for identifying prognostic model. Methods Based on liquid chromatography-tandem mass spectrometry, we analyzed the global metabolomes of preoperative sera from 86 samples (32 TIO recovery patients, 11 nonremission patients, and 43 matched controls). Statistical analyses, pathway enrichment, and receiver operating characteristic analysis were performed to identified and evaluate potential markers. Results Sparse partial least squares discriminant analysis indicated a clear separation of metabolomic profiles between healthy controls (HC) and TIO patients. The serum metabolites altered in different prognostic groups. L-pipecolic acid, 2-dodecylbenzenesulfonic acid, and 2-deoxygalactopyranose were the top 3 metabolites that were significantly perturbed. A combination of L-pipecolic acid and 2-dodecylbenzenesulfonic acid demonstrated a high-performance panel for TIO prognosis evaluated by random forest algorithm (area under the curve = 0.921, 95% CI, 0.787-0.995). Conclusions We investigate the global metabolomes of refractory TIO and identify potential prognostic biomarkers preliminarily. A high sensitivity and specificity panel were identified as promising discriminating predictors, which need to be verified in more patients. This work may demonstrate novel insights into TIO prognosis and pathogenesis.