BACKGROUND AND PURPOSE: Metachromatic leukodystrophy (MLD) is a lysosomal enzyme deficiency disorder leading to demyelination and subsequently to a progressive decline in cognitive and motor function. It affects mainly white matter where changes during the course of the disease can be visualized on T2-weighted MRI as hyperintense areas. Associated changes in brain metabolism can be quantified by MR spectroscopy (MRS) and may give complementary information as biomarkers for disease characterisation and progression. Our study aimed to further investigate the correlation of MRS with clinical parameters for motor and cognitive function by using a model free MRS analysis approach that would be precise and straightforward to implement.
MATERIALS AND METHODS: 53 MRS datasets derived from 29 patients (10 late-infantile, 19 juvenile) and 12 controls were acquired using a semi-LASER CSI sequence covering a slice through the centrum semiovale above the corpus callosum. We defined four regions of interest in the white matter (frontal white matter [FWM] and the cortico-spinal tract [CST] area, each left and right) and one in cortical grey matter. Spectra were analysed using a model and fitting free approach by calculating the definite integral of 10 intervals which were distributed along the whole spectrum. These 10 intervals were orientated towards the main peaks of the metabolites N-acetylaspartate (NAA), creatine, myo-inositol, choline, glutamine/glutamate and aspartate to approximately attribute changes in the intervals to corresponding metabolites. Their ratios to the main creatine peak integral were correlated with clinical parameters assessing motor and cognitive abilities. Furthermore, in a post-hoc analysis, NAA levels of a subset of 21 MR datasets were correlated to NAA levels in urine measured byH (proton) nuclear magnetic resonance (NMR) spectroscopy. The applied interval integration method was validated in the control cohort against the standard approach, using spectral profile templates of known metabolites (LCModel). Both methods showed good agreement, with coefficients of variance being slightly lower for our approach compared to the related LCModel results. Moreover, the new approach was able to extract information out of the frequency range around the main peaks of aspartate and glutamine where LCModel showed only few usable values for the respective metabolites. 1
RESULTS: MLD spectra clearly differed from controls. The most pronounced differences were found in white matter (much less in grey matter), with larger values corresponding to main peaks of myo-inositol, choline and aspartate, and smaller values associated with NAA and glutamine. Late-infantile patients had more severe changes compared to later-onset patients, especially in intervals corresponding to NAA, aspartate, myo-inositol, choline and glutamine. There was a high correlation of several intervals in the corticospinal tract region with motor function (with the most relevant interval corresponding to NAA peak with a correlation coefficient of -0.75; p < 0.001), while cognitive function, by means of IQ, was found to be most correlating in frontal white matter corresponding to the NAA peak (r = 0.84, p < 0.001). The post-hoc analysis showed that the main NAA peak interval correlated negatively with the NAA in urine (r = -0.584, p < 0.001).
CONCLUSION: The applied model and fitting free interval integration approach to analyse MRS data of a semi-LASER sequence at 3T suits well to detect and quantify pathological changes in MLD patients through the different courses of the disease and correlates well with clinical symptoms while showing smaller dimensions of variation compared to the more sophisticated single metabolite analysis using LCModel. NAA seems the most clinically meaningful biomarker to use in this context. Its correlation with urine measurements further underlines its potential as a clinically and biologically useful parameter of disease progression in MLD.
