Objectives Mitochondrial diseases are clinical variable and difficult to diagnose due to a lack of reliable frontline indicator tests. We aimed to improve on our previous single serum diagnostic indicator studies by investigating a serum metabolomics approach.

Methods Serum small metabolite profiles were measured using nuclear magnetic resonance spectroscopy. Twenty-eight identifiable metabolite peaks were used to create a multivariate disease discriminant test from a discovery cohort of control, non-mitochondrial neuromuscular disease control and mitochondrial disease subjects. The model was then tested on a validation cohort of control and mitochondrial disease subjects to determine the diagnostic sensitivity of the model.

Results Our previous studies of single serum biomarkers, Fibroblast Growth Factor 21 (FGF–21) and Growth Differentiation Factor 15 (GDF–15), provided 69% and 78% diagnostic sensitivity respectively. Our metabolomics approach improved upon this with a diagnostic sensitivity of 89% and Q2 of 0.0883. Adding the single serum biomarkers to the metabolomics model improved the predictability of the model further (Q2=0.1651) and could indicate ~95% of mitochondrial disease patients.

Conclusions Metabolomic phenotyping of serum presents as a useful clinical indicator of mitochondrial diseases for triaging patients to subsequent comprehensive and definitive genetic investigations.