Susceptibility-weighted imaging has become an important tool for the detection of hemorrhages and iron accumulation in the brain^1,2. In its quantitative form, R2*, MR frequency shift imaging and quantitative susceptibility mapping (QSM) have been used for the assessment of tissue damage in multiple sclerosis (MS) lesions^3,4,5. Increases in the frequency and susceptibility signal have been attributed to iron accumulation^5,6,7, with demyelination as a potential confounder and not always supported by corresponding histology⁸. A recent study utilized non-local magnetic field changes around bulk susceptibility inclusions as a measure of iron content in MS lesions⁹ and concluded that iron is a rare source of contrast in agreement with histological studies¹⁰. While these results are in qualitative agreement, MR-histology correlations showed little overlap between non-local effects and actual iron content¹¹. We hypothesize that this discrepancy may be resolved by considering susceptibility changes in the surrounding normal appearing white matter (NAWM).Single Gradient-echo MR images of 4 fixed post-mortem samples were acquired at 3T at TR/TE=40/20ms, flip angle=30 and acquisition voxel size=0.5x0.5x1mm³, reconstructed to 0.42x0.42x0.5mm³. Subsequently, the tissue blocks were embedded in paraffin and cut according to MRI planes. Luxol-fast blue-periodic-acid Schiff myelin and turnbull blue iron staining was performed on adjacent slices. We estimated average iron and myelin densities in the center of MS lesions and their vicinity on the digitalized histological sections. Additionally, we performed simulations of MR frequency shifts using a simple spherical model in a box with changes in the tissue's magnetic susceptibility affecting independently only the sphere ('lesion' , upper row in each part of the simulation ) or only the surrounding box (NAWM , bottom row of each simulation ). All susceptibility maps were forward field calculated with independent addition of gaussian noise to the real and imaginary parts of the signal. All MR images and simulated field maps were unwrapped using a Laplacian unwrapping algorithm¹², followed by background field removal using PDF¹³. Susceptibility maps were computed by solving of the inversion problem using an LSQR approach¹².Our simulations demonstrated that with changes in the surrounding NAWM alone (Fig. 1, susceptibility model), the lesional tissue will appear more paramagnetic, creating the same dipolar field modification as observed for actual iron accumulation (Fig. 1, PDF frequency), which is in turn reflected as falsely elevated magnetic susceptibility on QSM. Such a diamagnetic shift of the surrounding tissue can occur through strong demyelination of the lesion and maintained myelin of the WM (Δ_myelin↑=Myelin_WM-Myelin_Lesion). Accounting for iron, the lesion will appear less paramagnetic the more iron is lost (Δ_iron↑=Iron_WM-Iron_Lesion), while iron loss of the NAWM will strengthen the dipolar appearance (Δ_iron↓). However, demyelination of the NAWM is expected to a certain degree. Our post mortem study showed little correlation between frequency shifts observed in MS lesions with either diffuse or cellular iron content (R= -0.17 & R= 0.37). In contrast, 13/15 lesions demonstrated iron loss compared to the surrounding NAWM. Figure 2 shows two lesions with comparable iron and myelin densities in their centers, but in different scenarios regarding their NAWM state. The upper row (A/B) corresponds to a lesions that showed dipolar field modifications, while the lesion in the bottom row (C/D) did not show any non-local features. Lesion 1 (A/B) shows high Δ_myelin, while Δ_iron is low compared to most other lesions. In contrast, Lesion 2 (C/D) shows complete demyelination and iron loss, with partial myelin and iron loss of the NAWM causing Δmyelin to be lower than in Lesion 1, while Δiron was approximately the double compared to Lesion 1.While increases in resonance frequency within MS lesions are also concordant with the theory of microstructural changes during myelin debris formation and during the presence of myelin debris^3,14, debris formation is mainly expected in younger, more active lesions. Local frequency shifts due to iron-rich marcophages can occur, however we observed that local accumulation of such macrophages causes the MR signal to change in only parts of the lesions, rather than throughout.MR frequency shift data and QSM of MS lesions need to be interpreted with caution. Apparent increases in the MR susceptibility signal in MS lesions rarely relate to actual increases in magnetic susceptibility, and may in fact be the result of changes of non-local origin, such as demyelination of NAWM tissue.