Metabolite-Specific Reproducibility of Cerebral ³¹P-MRS at 3 T: Implications for Clinical Research

Phosphorus magnetic resonance spectroscopy (³¹P-MRS) enables noninvasive measurement of brain metabolism, yet its reproducibility in clinical settings remains unclear. We systematically assessed intrasession and intersession variability as well as interindividual differences of key phosphorus metabolites at 3 T in healthy individuals and persons with Parkinson's disease under various experimental conditions. Intersession variability, as measured by coefficients of variation (CoVs) increased notably for longer scan intervals (~1 year), and metabolite ratios from well-resolved spectral signals (i.e., adenosine triphosphate [ATP], phosphocreatine [PCr], and intracellular inorganic phosphate [P_i]) exhibited consistently higher stability compared with ratios calculated from metabolite signals overlapping on the spectrum (e.g., total nicotinamide adenine dinucleotide [tNAD], as well as phosphate monoesters [PMEs] and phosphate diesters [PDEs]). Test-retest variability ranged from ~5 to 25 CoV%, where PCr, ATP-α, and ATP-γ were the most stable while glycerophosphocholine (GPC), glycerophosphoethanolamine (GPE), phosphoethanolamine (PE), and tNAD varied considerably. Interindividual variability was found to be higher than intraindividual variability for all metabolite ratios, ranging from ~9 to 33 CoV%. By systematically quantifying intraindividual and interindividual variability, as well as providing explicit sample size recommendations, this study facilitates more reliable longitudinal and cross-sectional clinical trials and translational studies of brain metabolism featuring ³¹P-MRS.