Evidence suggests iron homeostasis is distorted in multiple sclerosis (MS). Increased iron levels have been observed in the deep gray matter and at the center and rim of MS white matter lesions (WML) whilst iron seems to be decreased in the normal appearing white matter (NAWM). ¹ Iron released by demyelination is believed to facilitate neurodegeneration and perpetuate inflammation in NAWM. ¹ MRI holds promise to noninvasively measure regional iron content in the brain. However reliable mapping of iron in the WM is confounded by the diamagnetic contribution of myelin to the acquired indices and its orientational effects. ^{2, 3} We recently proposed a novel MRI technique based on the temperature dependency of the paramagnetic susceptibility which allows reliable estimates of iron in WM. ⁴ Here, we applied this technique to fixed post mortem MS brains and correlated MRI estimates of regional iron concentration with ferritin measured using immunohistochemistry.

MRI was performed on hemispheric 10 mm thick formalin fixed brain slices from 3 subjects with MS and one control subject. T₂ weighted FLAIR images (TE/TR = 78/4000 ms) and 2D gradient echo images (TR = 340 and TE = 4.92, 9.84, 14.76, 19.68, 24.6, 29.52 ms) were acquired at 14 temperature points between 8°C and 22°C on a 3T system (Magnetom TimTrio, Siemens Medical Systems, Erlangen, Germany). All sequences were performed with an in plane resolution of 500x500 µm and a slice thickness of 2 mm. R₂* mapping was performed assuming a mono-exponential decay at each temperature point. The iron map was calculated by pixel wise estimating the temperature coefficient of R₂* (TcR₂*) using a linear regression model ($$$R_2^*=R_{2,0}^* + TcR_2^* \cdot temperature$$$). After MRI, the brain samples were cut in half to allow a good match between histology and the central (MRI) slice. One half was embedded in paraffin, sections cut and stained for ferritin light chain. For ferritin quantification, an overview image of each area was acquired at x4 magnification on myelin basic protein (MBP) and Ferritin stained sections enabling a precise match of homotopic areas on both sections and manual outlining of corresponding lesion core, edge and NAWM. Normal appearing cortex (NAC) and basal ganglia (BG) were also assessed. Within each region of interest (ROI) 8 counting fields were randomly cast and acquired at x40 magnification within which ferritin positive (⁺) cells were manually counted. The relative density of ferritin was expressed as n cells/counting field. Linear regression analysis was performed to study the association between the number of ferritin⁺ cells and TcR₂*.

The TcR₂* map suggested high iron content in the basal ganglia and gray matter bridges of the striatum but also of subcortical U-fibers and at the edge of lesions. Low iron content was observed in lesions and in the cortex (figure 1). As shown in figure 2, these findings were in good agreement with the distribution of ferritin⁺ cells (TcR₂* = -0.7737 - 0.0031 · [Fe], r=-0.64, p<0.0001).We applied for the first time a novel MRI technique to map the iron concentration across all brain structures in post mortem multiple sclerosis brain. The technique is based on the temperature dependency of the magnetic susceptibility allowing to compensate for myelin related diamagnetic effects. As a result, susceptibility shifts due to demyelination or changes in water content do not affect the iron measurement. Our results suggest TcR₂* mapping is a useful technique to measure iron content in brain tissue. While the described method is limited to post mortem tissue, it enables iron mapping in entire brain slices at a reasonable resolution. Due to the elimination of myelin induced susceptibility changes, it is well suited for further research into MS and other neurological diseases with regional distortions of the iron homeostasis.