Parsing Fabry Disease Metabolic Plasticity Using Metabolomics

Abstract

Background: Fabry disease (FD) is an X-linked lysosomal disease due to a deficiency in the activity of the lysosomal α-galactosidase A (GalA), a key enzyme in the glycosphingolipid degradation pathway. FD is a complex disease with a poor genotype–phenotype correlation. FD could involve kidney, heart or central nervous system impairment that significantly decreases life expectancy. The advent of omics technologies offers the possibility of a global, integrated and systemic approach well-suited for the exploration of this complex disease. Materials and Methods: Sixty-six plasmas of FD patients from the French Fabry cohort (FFABRY) and 60 control plasmas were analyzed using liquid chromatography and mass spectrometry-based targeted metabolomics (188 metabolites) along with the determination of LysoGb3 concentration and GalA enzymatic activity. Conventional univariate analyses as well as systems biology and machine learning methods were used. Results: The analysis allowed for the identification of discriminating metabolic profiles that unambiguously separate FD patients from control subjects. The analysis identified 86 metabolites that are differentially expressed, including 62 Glycerophospholipids, 8 Acylcarnitines, 6 Sphingomyelins, 5 Aminoacids and 5 Biogenic Amines. Thirteen consensus metabolites were identified through network-based analysis, including 1 biogenic amine, 2 lysophosphatidylcholines and 10 glycerophospholipids. A predictive model using these metabolites showed an AUC-ROC of 0.992 (CI: 0.965–1.000). Conclusion: These results highlight deep metabolic remodeling in FD and confirm the potential of omics-based approaches in lysosomal diseases to reveal clinical and biological associations to generate pathophysiological hypotheses.

Keywords

• inborn errors of metabolism
• Fabry disease
• lysosomal storage diseases
• metabolomics
• systems biology
• machine learning