Cognitive implications and associated transcriptomic signatures of distinct regional iron depositions in cerebral small vessel disease

Abstract

Introduction: Regional brain iron dyshomeostasis is observed in cerebral small vessel disease (cSVD) and other neurodegeneration processes. However, its spatial patterns, cognitive impact, and underlying pathological mechanisms remain unclear.

Methods: Voxel-based analysis of quantitative susceptibility mapping (QSM) was used to detect regional susceptibility changes, and their correlations with cognitive function were assessed using linear regression. We combined the microarray dataset from the Allen Human Brain Atlas (AHBA) to explore the pathological mechanisms of iron deposition patterns.

Results: A total of 87 cSVD patients and 80 controls were included in the study. Increased QSM values in the bilateral putamen and caudate were associated with cognitive decline in cSVD. Gene set enrichment analysis revealed the enrichment of gene sets related to central nervous system integrity.

Discussion: Iron deposition in deep gray matter may indicate cognitive changes in cSVD and could be linked to the disruption of brain structural and functional integrity.

Highlights: Increased susceptibility values, indicating focal iron deposition, were observed in the deep gray matter of patients with cerebral small vessel disease (cSVD). Regional iron concentration in the deep gray nuclei was associated with cognitive impairment in cSVD patients. Imaging transcriptomics suggests that cSVD-related iron deposition is linked to the structural and functional integrity of the brain. An open-source script for imaging transcriptomics focusing on regional gene expression was developed and proposed.

Keywords: Allen Human Brain Atlas; cerebral small vessel disease; gene expression; iron deposition; quantitative susceptibility mapping.

FIGURE 1

Schematic summary of the study design. A, QSM preprocessing and reconstruction. The magnitude and phase imaging were used to generate the susceptibility map in native space. The QSM imaging was then normalized to the MNI space using the affine parameters derived from registering T1‐weighted imaging in the space of the first echo magnitude imaging to the standard MNI template. B, Voxel‐wise QSM analysis. We compared the QSM value in gray matter between cSVD patients and healthy controls, resulting in a TFCE map. C, Correlation between regional QSM value and cognitive function. We extracted the mean susceptibility value in the statistically significant clusters and explored the relationship between iron deposition and cognitive function. D, Imaging transcriptomics analysis. In the main analysis, the volumetric DK atlas was used to generate a regional TFCE map and gene expression matrix. We performed PLS regression to acquire ranked gene lists representing their correlation with the parcellated TFCE value, which was consequently used to conduct the GSEA, seeking enriched gene sets, to explain the iron deposition pattern in cSVD. cSVD, cerebral small vessel disease; DK, Desikan–Killiany; FANSI, fast non‐linear susceptibility inversion; GM, gray matter; GSEA, gene set enrichment analysis; HC, healthy controls; MNI, Montreal Neurological Institute; MoCA, Montreal Cognitive Assessment; PLS, partial least squares; PT, patients; QSM, quantitative susceptibility mapping; RAVLT, Rey Auditory Verbal Learning Test; ROMEO, rapid opensource minimum spanning tree algorithm; STROOP, Stroop Color and Word Test; STT, Shape Trail Test; TFCE, threshold‐free cluster enhancement; VBA, voxel‐based analysis; VSHARP, variable kernel sophisticated harmonic artifact reduction for phase data.

FIGURE 2

The quantification of cSVD imaging markers. Ai and Aii, The subject's T1w images. Aiii and Aiv, The corresponding PVS segmentation results (in red) along with the basal ganglia and deep white matter region masks (in transparent blue), which were obtained using the SHiVAi pipeline. Bi, Representative axial slice from a T2 FLAIR image. Bii, Corresponding WMH segmentation result obtained using WMH‐SynthSeg. PWMH (Biv, marked in yellow) and DWMH (Biv, marked in green) were distinguished based on whether any voxel of a given WMH cluster was located within 3 mm of the lateral ventricle (Biii: the lateral ventricles are highlighted in transparent red, and the periventricular mask within 3 mm is shown in transparent blue). Ci and Cii, A representative CMB on QSM and SWI images, respectively. Ciii, Skull‐stripped SWI image, registered to the T1w image, which serves as the input for the SHIVA‐CMB model. Civ, Detected CMB lesion (in red), classified as a lobar CMB because it falls within the right parietal region (in transparent yellow) given by SHiVAi. CMB, cerebral microbleed; DWMH, deep white matter hyperintensity; FLAIR, fluid‐attenuated inversion recovery; PVS, perivascular space; PWMH, periventricular WMH; QSM, quantitative susceptibility mapping; SWI, susceptibility‐weighted imaging; WMH, white matter hyperintensity.

FIGURE 3

Iron deposition in cSVD and the correlation with cognitive performance. A and B, Results of the voxel‐based analysis revealed focal iron deposition in bilateral putamen and caudate. The log‐transformed P values were overlayed on the study‐wise susceptibility maps in the MNI space. C, Scatter plots showed the relationship between increased QSM value and declined cognitive performance in cSVD patients. cSVD, cerebral small vessel disease; FWE, family‐wise error; HC, healthy controls; MNI, Montreal Neurological Institute; MoCA, Montreal Cognitive Assessment; QSM, quantitative susceptibility mapping; STROOP, Stroop Color and Word Test.

FIGURE 4

Results of the imaging transcriptomics analysis. A, The parcellated TFCE map. B, The explained variance in the first 10 PLSR components. The correlation coefficient between PLS scores and TFCE value was also annotated and reflected by the bar color. C, Enriched gene sets and their NES, log‐transformed P value from GSEA. For GO terms, the top five significant gene sets of BP, CC, and MF were demonstrated with the first five core enrichment gene symbols. Results of cell type–specific gene sets were displayed as the bubble plot. BP, biological process; CC, cellular component; GO, Gene Ontology; GSEA, gene set enrichment analysis; MF, molecular function; OPC, oligodendrocyte precursor cell; PLS, partial least squares; PLSR, partial least squares regression; TFCE, threshold‐free cluster enhancement.