Brain iron accumulation or loss may serve as biomarker for neurological diseases including multiple sclerosis as well as for normal aging.^1-4 Consequently, there is high interest in determining the cerebral iron content non-invasively. A unique view into the cerebral iron distribution in vivo is offered by quantitative susceptibility mapping (QSM),^5-7 an MR technique that relies on post-processing of gradient-echo (GRE) phase images.⁸ Although not only paramagnetic iron complexes but also diamagnetic myelin around axons contribute to the magnetic susceptibility,^8,9 a linear correlation of magnetic susceptibility with iron concentration was proven in post mortem experiments by using sophisticated analysis methods (X-ray fluorescence, inductively-coupled plasma – mass spectrometry).^6,7 In this contribution, we present a histochemical-driven approach to quantify iron in post mortem brain tissue and compare the spatial distribution of iron with in situ magnetic susceptibility maps to further validate QSM for iron mapping.

In situ multi-echo GRE data (TE₁/TE₂/TE₃/TE₄=4ms/9ms/15ms/22ms, TR/FA/BW_1-4=26ms/15°/391Hz/px, voxel size = 0.43mm × 0.43mm × 0.65mm) were acquired from a human, female cadaver head with a 32 channel head coil on a 7T-MRI system. The single-channel GRE phase images were combined with the COMPOSER method,¹⁰ spatially unwrapped,^11,12 combined across the different echo times,¹³ and corrected for background variations utilizing sophisticated harmonic artifact removal for phase data (SHARP).⁸ Susceptibility maps were then obtained from SHARP-processed data using homogeneity enabled incremental dipole inversion (HEIDI)¹⁴ and referenced to cerebrospinal fluid.

Following MRI, the brain was extracted from the cadaver head, fixed in 4% neutral-buffered formalin for three weeks, and cut into 6 mm-thick axial slices. Twenty-two cylindrical tissue samples of both gray and white matter tissue, each with a diameter of 8 mm, were harvested from the brain (Fig. 1B) and used for ferrozine iron quantification as suggested by Fish (1988).¹⁵ Subsequently, the tissue slice was immersed in ammonium sulfide solution, washed with distilled water, and stained for di– and trivalent non-heme iron with DAB-enhanced Turnbull blue staining (Fig. 1C).¹⁶ The iron stain was scanned, converted into gray-levels, and then converted to iron concentration (Fig. 1D) using the equation resulting from linear regression between the iron concentration determined using ferrozine assay and the iron stain signal intensity around the corresponding extracted tissue samples (Fig. 1A). Finally, the susceptibility values and distribution were linearly correlated with the iron concentrations determined with the ferrozine assay and visually compared with the local iron distribution, respectively.

A linear correlation between magnetic susceptibility and measured iron concentration across gray and white matter regions was observed (Fig. 2). The histological iron map is opposed with the susceptibility map in Figure 3, indicating excellent correspondence between iron accumulations and elevated susceptibility in deep gray matter. Regions with elevated susceptibility in the thalamus coincide with regions of increased iron content (arrows a and b in Fig. 3). Large fiber bundles (e.g., internal capsule, optic radiation) exhibit lower iron concentrations and susceptibility values. Interestingly, the external capsule exhibits high iron concentration, but the presence of diamagnetic myelin substantially reduces its magnetic susceptibility (arrow c in Fig. 3). The high iron content measured in superficial white matter is not discernible in the susceptibility maps (arrow f in Fig. 3), maybe due to interpolation effects by registering the susceptibility map to the iron stain. The linear regression between iron concentration and magnetic susceptibility revealed a slope of 1.68 ppb kg wet tissue/mg (Figure 2), which is higher to values reported previously for post mortem examinations (0.97 ppb kg/mg⁶ or 0.8 ppb kg/mg⁷) and the theoretical contribution of ferritin to tissue susceptibility (1.27 ppb kg/mg).¹⁷ Our high slope might be due to the fact that susceptibility values and iron content were extracted in unfixed and fixed tissue, respectively, because formalin fixation is known to decrease tissue iron levels by up to 40%.¹⁸ In addition, the calibration slope (Fig. 1A) may be affected by varying iron content in the cylindrical samples used for ferrozine assay and their direct vicinity. In deep GM structures magnetic susceptibility resembles the histological iron map enabling identification of thalamic substructures. In white matter magnetic susceptibility is additionally influenced by diamagnetic myelin, explaining the inconsistency between high iron and low susceptibility in the external capsule.Ferrozine assay calibrated Turnbull blue (FACTB) staining is an elegant method to determine iron concentration and to compare the spatial distribution of iron with MR images. Further investigations are still required to validate iron measurement with FACTB staining across different brains and slices. Direct comparison of FACTB staining and magnetic susceptibility is expected to improve the understanding of biophysical origins of susceptibility variations within brain tissue.