Introduction: Creatine (Cr) dynamics within the human brain is a research area with promising clinical value, particularly for cohorts exposed to metabolically demanding periods (e.g., ageing, neurological diseases, sleep deprivation). Consequently, increasing reliance will be placed upon proton magnetic resonance spectroscopy (1H-MRS) for determining total creatine (tCr) concentrations in the brain. Importantly, however, before data from experimental studies can be accurately interpreted, more consideration should be given to the inherent sources of error in 1H-MRS, and their effect on the reliability of repeated measurements. This need is underscored by weaknesses in the design of existing studies, and ultimately, by large margins of error in repeated 1H-MRS reported to date. Herein, we examined the intra- and inter-session reliability and repeatability of 1H-MRS for quantifying tCr concentrations in multiple brain regions (midbrain: MB; visual cortex: VC; frontal cortex FC).
Methods: Eighteen healthy adults aged between 20-32 years were recruited for this study [mean age=25.8±3.0years; 50% female; n=14 intra-session analysis; n=15 inter-session analysis (n=11 both)]. MR imaging and 1H-MRS (PRESS) were performed on a 3T Siemens scanner using a 20-channel head coil. Intra-session analyses involved repeated measurements of the MB, VC and FC without removing the participant from the scanner, while inter-session analyses involved repeated measurements from the same regions, but with a brief break between measurements (involving repositioning of the participant and voxels). TARQUIN was used to analyse 1H-MRS data, and water unsuppressed data were used to determine absolute tCr concentrations. Paired t-tests, minimum detectable change (MDC), Pearson’s correlation coefficient (r), coefficient of variation (CV), intra-class correlation coefficient (ICC) were calculated. Bland-Altman plots were generated to visually assess the data.
Results: A total of 174 spectra were acquired, including 84 for intra-session analyses and 90 for inter-session analyses. No significant differences in absolute tCr concentrations between repeated intra- or inter-session measurements were shown in any region (mean differences=0.1-1.2%). Intra- and inter-session r values were between 0.909-0.985 and 0.836-0.858 (all p<0.001), depending upon region, and no trends in measurement bias were found. For the MB, VC and FC, intra-session CVs were 1.7%, 0.8% and 2.1%, ICCs were 0.903 (95%CI=0.727-0.968), 0.979 (95%CI=0.935-0.993) and 0.921 (95%CI=0.772-0.974) (all p<0.001), and MDCs were 1.2%, 0.6% and 1.5%, while inter-session CVs were 2.7%, 1.7% and 2.7%, ICCs were 0.835 (95%CI=0.578-0.941), 0.854 (95%CI=0.619-0.948) and 0.847 (95%CI=0.603–0.946) (all p<0.001), and MDCs were 1.9%, 1.2% and 1.9%. Inter-region differences in tCr concentration of up to 20.7% were shown.
Conclusions: Our findings indicate that 1H-MRS at 3T can reliably and repeatably quantify absolute tCr concentrations in multiple human brain regions, when appropriate consideration is given to potential sources of error. Changes in tCr concentration as small as 2% may be discernible from measurement error, however, centre-specific margins of error should be established prior to experimental investigation. More studies are required to determine whether similar findings are shown in other populations of interest, such as people suffering from neurological diseases or movement disorders.
Ethical statement: Manchester Metropolitan University’s research ethics committee granted ethical approval and all participants provided written informed consent.