Introduction

Neurodegeneration with brain iron accumulation (NBIA) is a set of highly debilitating genetic disorders with short life expectancy affecting children and adults. MR’s role is crucial in initial diagnosis as the first suspicion is drawn from increased iron deposition observed in clinical MRIs¹. Qualitative special features, such as “eye of the tiger”² sign in T2W images, are in clinical use but are insufficient especially for understanding mechanisms behind selective iron deposition in globus pallidus and substantia nigra. In contrast, recently popularized quantitative susceptibility mapping³ (QSM) methods might provide significant objective information for overcoming the broad phenotypic NBIA range and variability.

Methods

In this IRB approved study, MR data were collected from 16 NBIA patients (11 with anesthesia) 7, 2, 3, 1 male and 2, 0, 0, 1, female PKAN, PLAN, MPAN and Kufor-Rakeb patients respectively with $$$21.37\pm 13.98\mathrm{yrs.}$$$ and range $$$[5\mathrm{yrs.}- 35\mathrm{yrs.}]\cup \{56\mathrm{yrs. (male, PKAN)}\} $$$ who were referred from the collaborating institutions. MRIs were also collected without anesthesia from 11 (5 females) healthy volunteers of $$$ 29.54 \pm 13.76\mathrm{yrs.} $$$ with $$$ [7\mathrm{yrs.} - 49\mathrm{yrs.}] $$$ range. Prior to imaging, informed written consent was obtained from the volunteers, patients or their legal guardians.
MRI data were collected on a Siemens™ Magnetom Prisma® 3 Tesla scanner equipped with a 60 cm gantry, a 80 mT/m gradient with 200 T/m/s slew rate and a 64 channel head/neck coil. A 3D FLASH sequence with (single echo) TE=30ms, TR=45ms was used for QSM data collection. Axial 256x256 images with a resolution of 0.86x0.86x2 mm/voxel were obtained.
All images were converted from DICOM to NIfTI format MIPAV⁴ software which was also used for brain surface extraction (BSE) algorithm. BSE brain masks were then fed to the QSM software package STI Suite^5-7 v2.2 along with the phase images for obtaining susceptibility maps. Therein, phase scaling (to $$$2\pi$$$), phase unwrapping and background phase removal using iHARPERELLA (integrated HARmonic (background) PhasE REmovaL using the Laplacian operator) routine⁷ were performed respectively prior to applying iLSQR (improved sparse linear equation and least-squares) algorithm.
Basal ganglia regions, CN, PT, GP, RN and SN were manually segmented independently from QSM images using MIPAV⁴’s draw polygon segmentation tool on either unprocessed phase images or frequency shift maps. region of interest (ROI) based data analysis and presentation were realized with in-house developed MATLAB™ 2018b (Mathworks™, Natick, MA, USA) code.
By ferromagnetism, iron increases magnetic susceptibility of brain tissue⁸. However, heterogeneous iron distributions, paramagnetic and diamagnetic tissue with negative susceptibility might impair accumulation assessment when using ROI mean values as metric. Accordingly, for assessing iron accumulation within ROIs more precisely than mean ROI-susceptibility we propose herein a novel metric. First, the volume of high susceptibility ($$$\chi_t$$$ ) values above a chosen threshold is calculated as
$$V(\chi_t)=\int\limits_{\chi(x)>\chi_t} dx=\sum_i I(\chi(x_i)>\chi_t) \quad (x,x_i \in R^3)$$
where $$$I$$$ denotes the indicator function. Subsequently, high susceptibility density $$$\rho(\chi_t)$$$ defined as
$$\rho(\chi_t)=\frac{1}{V(\chi_t)}=\int\limits_{\chi(x)>\chi_t} dx=\frac{1}{V(\chi_t)}\sum\limits_{i, s.t. \chi(x_i)>\chi_t} \chi(x_i) \quad (x,x_i \in R^3)$$
is calculated for each patient’s ROIs. The threshold was chosen by inspecting the data for high iron concentration while providing best subject clusters in ROI-scatter plots (see figure).

Results

Whereas TH, PT and CN demonstrate cluster heterogeneity with various levels of healthy subject clustering (not shown), GP plays a key role in separating subtypes and healthy volunteers. RN and SN display similar properties however with a lesser prominence.
With the exception of one 27Y, male PKAN patients demonstrate age dependent simultaneous increase in GP-RN (see figure), while a 5Y and 8Y fall apart from the linear trend with low RN values. In contrast, older healthy volunteers have higher accumulation in RN but not in GP.
Two Kufor-Rakeb patients remain clustered in GP-RN-SN combinations with accumulations lower, similar and higher in GP, SN and RN respectively compared to healthy cohort (RN-SN not shown).
PLAN-28Y-M patient have consistently a large accumulation in all ROI combinations similarly for PKAN-56Y-M patient which is expected with age related accumulation.
MPAN patients remain close to each other but are separated from the patients and volunteers with values in mid-high range in GP and SN.

Conclusion

The mechanisms of iron accumulation, including the discovery of new NBIA genes and pathways is an ongoing research topic. In this preliminary study, to the best of our knowledge QSM data collected from a rare cohort of patients provided for the first time means of describing and differentiating basal ganglia regions affected by NBIA subtypes. The method herein might potentially be effective in supporting future NBIA molecular and genetic research.

Acknowledgements

Special thanks to Arzu Karabay for providing genomic information.