Brain iron deposition and cognitive decline in patients with cerebral small vessel disease : a quantitative susceptibility mapping study

Abstract

Background

Quantitative susceptibility mapping (QSM) can study the susceptibility values of brain tissue which allows for noninvasive examination of local brain iron levels in both normal and pathological conditions.

Purpose

Our study compares brain iron deposition in gray matter (GM) nuclei between cerebral small vessel disease (CSVD) patients and healthy controls (HCs), exploring factors that affect iron deposition and cognitive function.

Materials and methods

A total of 321 subjects were enrolled in this study. All subjects had cognitive examination including the Stroop color word test (SCWT) and MRI including multiecho gradient echo (mGRE) sequence. The patients with CSVD were divided into mild to moderate group (CSVD-M, total CSVD score ≤ 1) and severe group (CSVD-S, total CSVD score > 1). Morphology-enabled dipole inversion with an automated uniform cerebrospinal fluid zero reference algorithm (MEDI + 0) was used to generate brain QSM maps from mGRE data. Deep gray regional susceptibility values and cognitive function were compared among three groups (CSVD-S, CSVD-M, and HC) using multiple linear regression analysis and mediation effect analysis.

Results

There were significant differences in the SCWT scores and mean susceptibility values of the globus pallidus (GP), putamen (Put), and caudate nucleus (CN) among the three groups (P < 0.05, FDR correction). Age had a significant positive impact on the susceptibility values of GP (p = 0.018), Put (p < 0.001), and CN (p < 0.001). A history of diabetes had a significant positive influence on the susceptibility values of Put (p = 0.011) and CN (p < 0.001). A smoking history had a significant positive association with the susceptibility values of CN (p = 0.019). Mediation effect analysis demonstrated that iron deposition in the neostriatum partially mediated the relationship between hypertension and cognitive function. Age, diabetes, and smoking may increase iron deposition in the basal ganglia, associated with cognitive decline. The mean susceptibility values of the neostriatum played a mediating role in the association between hypertension and cognitive scores.

Conclusions

Age, diabetes, and smoking are associated with increased iron deposition in the basal ganglia and also linked to cognitive decline. This can help with understanding CSVD and its prevention and treatment.

Introduction

Cerebral small vessel disease (CSVD) is a common clinical disease affecting middle-aged and older individuals. It is characterized by damage to the brain’s white matter (WM) and deep gray matter (GM) due to degeneration of various small blood vessels within the brain [1]. CSVD accounts for approximately 20% of all strokes, including 25% of ischemic strokes and 45% of vascular dementias [2]. In addition, CSVD can also cause cognitive impairment, affective changes, and neurological disorders in affected individuals, and CSVD is widely recognized as a predominant factor contributing to cognitive decline in the elderly [3]. The neuroimaging markers of CSVD include recent small subcortical infarct, lacune, white matter hyperintensities (WMHs), cerebral microbleeds (CMBs), enlarged perivascular spaces (PVSs), cortical superficial siderosis, and cortical cerebral microinfarct [4]. The neuroimaging markers mentioned above frequently co-occur with varying degrees of severity. To determine the overall burden of CSVD, total CSVD score can be calculated based on the presence of lacunes of presumed vascular origin, WMH of presumed vascular origin, PVSs, and CMBs [5]. This score is represented by an ordinal scale ranging from 0 to 4, which indicates the severity of CSVD [6] and has recently become widely accepted to reflect the total brain injury of patients. Studies have shown that the total CSVD score is closely related not only to extensive impairment of cognitive ability and language function [7] but also to cortical atrophy and disruption of structural networks [8].

The accumulation of iron in the deep GM of the brain is a well-recognized pathological hallmark of several neurodegenerative disorders associated with advanced age [9, 10]. It is widely acknowledged that iron accumulation in the brain plays a crucial role in generating reactive oxygen species, thereby triggering oxidative stress and ultimately contributing to neuronal death, which in turn impairs cognitive function [11]. Caudate nucleus (CN), putamen (Put), and globus pallidus (GP) are the primary structures in the basal ganglia, representing a major information processing center and playing a vital role in the cognitive of individuals [12]. Cognitive functions can be assessed using the Stroop Color and Word Test (SCWT), a neuropsychological test extensively used to assess the ability to inhibit cognitive interference [13].

Quantitative susceptibility mapping (QSM) of magnetic susceptibility properties of brain tissue is sensitive to paramagnetic tissue iron [14] and allows noninvasive study of local brain iron under both normal and pathological conditions [15]. This study utilized QSM technology to quantify differences in susceptibility of the basal ganglia between CSVD subjects with varying total CSVD scores and healthy controls. Additionally, we investigated the effects of different levels of susceptibility values on cognitive function and further analyzed which factors could influence susceptibility values in CSVD subjects.

Materials and methods

Subjects

From December 2019 to December 2022, a total of 218 patients with CSVD and 103 healthy controls (HCs) were included at Shandong Provincial Hospital in this study. All subjects signed informed consent prior to the study, which was approved by the Shandong Provincial Hospital Affiliated to Shandong First Medical University Subcommittee on Human Studies Institutional Review Board.

In this study, the clinical diagnosis of CSVD was based on a comprehensive evaluation of imaging evidence, clinical symptoms, and related risk factors [6]. Patients diagnosed with lacunes, WMHs, CMBs, PVSs based on the current MRI consensus standards were included [16]. Clinically, CSVD presents with a range of symptoms, including cognitive impairment, gait abnormalities, and emotional or behavioral changes. Additionally, risk factors associated with CSVD include hypertension, diabetes, hyperlipidemia, and lifestyle factors such as smoking and alcohol consumption [6]. For this study, the comprehensive diagnosis of CSVD required at least one typical imaging marker on MRI, along with clinical symptoms associated with cognitive impairments, and the presence of risk factors such as hypertension. Subjects included in this study was scored based on CSVD and grouped based on the presence of lacunes, WMH, CMBs, and PVSs 4 rating imaging characteristics to reflect the total MRI CSVD burden [5]. Figure 1 presents a visual representation of the total CSVD scores for three groups of participants. The total CSVD score from 0 to 1 was divided into the mild to moderate group (CSVD-M), and from 2 to 4 was divided into the severe group (CSVD-S). A comprehensive cognitive examination was conducted on the study participants using standardized protocols, including the Montreal Cognitive Assessment (MoCA) and the Stroop color-word test (SCWT).

Fig. 1

A visual representation of the total CSVD scores for three groups of participants representatively illustrated in three rows: CSVD-S (row A), CSVD-M (row B) and healthy control (row C). Column a shows QSM images, column b shows T2FlAIR images, column c shows T2WI images, and column c shows SWI images. Row A: Male, 74 years old CSVD-S with WMH grade 3, PVS grade 1, and CMBs score 1. Row B: Male, 67 years old CSVD-M patient with CMBs score 1. Row C: Female, 58 years old, healthy subjects

Grouping method of CSVD subjects are as follows

The CSVD total score used in our study, based on Staals et al. [5], assigns up to 1 point for each of the four CSVD markers. Specifically, a score of 1 is given for the presence of any lacunes, 1 point for any CMB, 1 point for moderate-to-severe PVS in the basal ganglia (grades 2–4), and 1 point for significant WMH burden (defined as either confluent deep WMH with Fazekas grades 2–3 or irregular periventricular WMH extending into deep white matter with Fazekas grade 3). This results in a possible total score ranging from 0 to 4 points.

Exclusion and inclusion criteria are as follows

Individuals who had any evidence of organic brain lesions, such as cerebral apoplexy, brain tumors, or trauma, as well as those with a history of psychiatric or neurological conditions that could potentially impact cognitive functioning was excluded. Furthermore, individuals with a history of alcohol or substance abuse, acute diabetes complications, major organ damage, severe hypertension, or significant visual or auditory impairments were excluded from the study. The HC group consisted of healthy elderly volunteers aged 40 to 80, recruited from the local community. All participants had over 7 years of education and completed a thorough cognitive function assessment. Figure 2 shows the enrollment and exclusion flowchart of the participants.

Fig. 2

Flowchart of participant inclusion and exclusion criteria

MRI acquisition

All participants underwent brain imaging on a Siemens Healthcare MAGNETOM Skyra 3.0 T MR scanner using a 32-channel head coil for optimal signal reception. The imaging protocol included the following sequences: a 3D T1-weighted (T1W) magnetization-prepared rapid gradient echo (MPRAGE) sequence with TR = 7.3 msec, TE = 2.4 msec, TI = 900 msec, flip angle = 9°, and voxel size of 1 × 1 × 1 mm³; a 3D multi-echo gradient echo (mGRE) sequence for QSM with TR = 50 msec, first TE = 6.8 msec, TE interval = 4.1 msec, echo number = 10, flip angle = 15°, and voxel size of 1 × 1 × 2 mm³. Four other imaging sequences were added to detect any abnormal brain structures, including T2-weighted (T2W) turbo spin echo imaging, T2W fluid attenuated inversion recovery (FLAIR) imaging, diffusion-weighted imaging, and susceptibility-weighted images (SWI). This comprehensive approach to brain imaging allowed for accurate detection and characterization of brain abnormalities in study participants. (The parameters of these sequence are detailed in the supplementary materials).

Image postprocessing

Brain QSM is a powerful imaging technique that provides highly sensitive and specific maps of the susceptibility distribution within brain tissue. In this study, Quantitative susceptibility mapping (QSM) was derived from multigradient echo (mGRE) complex image data using a morphology-enabled dipole inversion (MEDI + 0) with an automatic uniform cerebrospinal fluid (CSF) zero reference algorithm [17, 18]. First, a nonlinear fitting process estimated the total field from the multiecho data [19]. Then, the projection onto dipole fields algorithm was used for spatial field unwrapping and background field removal, resulting in the computation of the local field [20]. This local field was subsequently inverted to produce the final susceptibility map. To enhance the quality of the QSM and automatically reference CSF susceptibility, structural priors (edges) from the magnitude image and a regularization term ensuring a uniform susceptibility distribution within the lateral ventricles’ CSF were applied in the numerical inversion process [21]. The process of determining the CSF mask was achieved by thresholding the R2* map computed from the mGRE magnitude data. This approach effectively removes the susceptibility artifacts associated with CSF, which helps to ensure that the resulting QSM maps accurately reflect the true susceptibility properties of the brain tissue. In summary, the MEDI + 0 algorithm is highly effective in generating high-quality QSM maps from mGRE complex image data. Its ability to automatically provide a uniform zero reference of CSF susceptibility and its use of structural priors make it an ideal choice for quantitative susceptibility mapping of the brain [18, 21]. The conventional images (T1W, T2W, FLAIR) were processed through an automated pipeline using the FMRIB Software Library. This pipeline included brain extraction with the BET algorithm, bias field correction via the FAST algorithm, and linear co-registration with the echo-combined mGRE magnitude image (aligned to the QSM space) using the FLIRT algorithm with six degrees of freedom.

Brain iron deposition burden of the region of interest (ROI)

In this study, specific subcortical gray matter structures closely associated with cognitive function were designated as regions of interest (ROIs), including the thalamus, GP, CN, Put, red nucleus, and substantia nigra (Fig. 3). The FIRST algorithm in FSL was utilized to segment these structures from T1W images, followed by linear co-registration of the segmentation masks to QSM images to ensure accurate alignment. An experienced neuroradiologist (LF.G., with 20 years of experience) visually inspected the segmentation masks on QSM images using ITK-SNAP v3.8 software, performing manual edits to exclude regions that could compromise measurement accuracy, such as veins, CMBs, and WMHs. During the ROI selection process, the neuroradiologist carefully traced the boundaries of these structures based on normative brain anatomy while avoiding adjacent tissues. To enhance measurement precision, susceptibility values within the ROIs were averaged across full slices, and distinct left and right ROIs were delineated. This methodology ensured the reliability of susceptibility data within each ROI for subsequent analyses.

Fig. 3

ROI sketch diagram. The 3D T1WI (bottom row) and QSM (top row) images were coregistered to a magnitude image of the first echo acquired from the 3D GRE sequence of the same subject by using FSL software. The gray matter nuclei (ROIs larger than 150 voxels) were drawn entirely by hand. The average QSM value in each ROI was then computed from all voxels overlapping with the corresponding label

Data analysis

We utilized the Statistical Package for the Social Sciences (SPSS) to conduct data analysis (Version 21.0 for Windows; SPSS, Chicago, Ill). First, 218 patients with CSVD and 103 healthy controls underwent a descriptive analysis. The measurement data were shown as mean ± standard deviation. The counting information was displayed as n (%). We used a one-way ANOVA test to compare the mean susceptibility values within a ROI and cognitive scores across the three groups. We employed the false discovery rate (FDR) method for correction to prevent type I error because numerous hypotheses were evaluated. Through the use of Pearson correlation analysis, the relationships between mean susceptibility values and cognitive function scores were identified. The multiple linear regression analysis was used to investigate the independent influencing factors of the mean susceptibility values of all ROIs. The independent variables included gender, age, education level, BMI, alcohol history, smoking history, hypertension history, diabetes history and hyperlipidemia history. Additionally, this study utilized mediation effect analysis to explore whether CSVD total score could mediate the relationship between age and the susceptibility values of striatum. Following the mediation effect analysis, this study employed the process script in SPSS24.0 analysis software and utilized a bootstrap-based analysis method (bootstrap repetitions = 5000) to examine the standardized age as the independent variable, the standardized CSVD total score as the mediator, and the standardized susceptibility values of striatum as the dependent variable.

Results

Clinical characteristics

Table 1 presents a summary of the clinical characteristics of all subjects. It includes information on the gender, age, education level, medical history, and cognitive level of the participants. Significant differences in gender and age were found between the CSVD-S and CSVD-M groups, CSVD-S and HC group, and CSVD-M and HC group. Significant differences in education level were observed only between the CSVD-S and CSVD-M, as well as the CSVD-S and HC group. Significant differences were observed in the MoCA and SCWT scores among the three groups. Compared to the CSVD-S group, the CSVD-M and HC group showed better MoCA and SCWT scores. Compared to the HC group, the CSVD-M group had lower MoCA score but no significantly different SWCT score. Considering the significant age and differences among the three groups, we performed corresponding demographic analyses in age-matched subgroups. Please refer to Supplemental Table S1 for detailed subgroup demographic information.

Table 1 Clinical characteristics of the participant groups

Susceptibility value analysis across ROIs

The results of the one-way ANOVA indicated a significant difference in the mean susceptibility values of the GP (P < 0.028), the Put (P < 0.001) and the CN (P = 0.009) among the three groups. The post-hoc tests revealed that the mean susceptibility values in the GP, the Put, and the CN were higher in patients with CSVD-S than in those with CSVD-M and HC group, and these differences were statistically significant (for CSVD-S vs. CSVD-M and for CSVD-M vs. HC, all the P values < 0.05). Notably, no significant differences were detected in the mean susceptibility levels of these brain regions between the CSVD-M and HC group (the P values were 0.447, 0.344, and 0.356, respectively) (Table 2; Fig. 4). One-way ANOVA of the three subgroups showed that the mean susceptibility values in the GP, Put, and CN still displayed significant differences among the groups, please refer to Supplementary Table S2 for further details.

Table 2 Susceptibility value (ppb [×10⁻⁹]) differences among the three groups

Fig. 4

Susceptibility value differences (in ppb) among the CSVD-S, CSVD-M, and HC groups. Among the three groups, the mean susceptibility values of the globus pallidus, the putamen, and the caudate nucleus were significantly different (b, c, d), while the difference in the mean susceptibility values in the thalamus, the red nucleus, and the substantia nigra did not reach the significant level (a, e, f). *P < 0.05; **P < 0.01; ***P < 0.001. CSVD-S = patients with CSVD with severe CSVD burden scores; CSVD-M = patients with CSVD with mild or moderate CSVD burden scores; HCs = healthy controls

Comparison of cognitive performance

The correlation analysis showed a significant positive association between the mean susceptibility values of CSVD subjects with their scores on the SCWT assessment in the following regions: the GP, the Put, and the CN (Fig. 5). We also conducted a correlation analysis between the mean susceptibility values of the differentiated nuclei (GP/Put/CN) and SCWT scores across the three subgroups, and the results were consistent with those in the current analysis. Please refer to Supplementary Figure S2 for further details.

Fig. 5

Correlations between mean susceptibility values of altered brain regions and SCWT in patients with CSVD

The multiple linear regression analysis (Table 3) indicated that age (P<0.001) was independently associated with an increase in the mean susceptibility values of the GP. In contrast, age (P<0.001), diabetes history (P = 0.003), and smoking history (P = 0.003) were independently associated with an increase in the mean susceptibility values of the Put. Furthermore, age (P<0.001) and diabetes history (P<0.001) were independently associated with an increase in the mean susceptibility values of the CN. Multiple comparison correction was applied using the Bonferroni correction, with a significance threshold set at P < 0.0056. Although age and diabetes history were found to influence the mean susceptibility values of the thalamus, these effects did not reach statistical significance. Similarly, while age, diabetes history, and hyperlipidemia demonstrated associations with the mean susceptibility values of the red nucleus, as well as age, hyperlipidemia, and BMI affecting the mean susceptibility values of the substantia nigra, none of these associations achieved statistical significance. Similarly, these results from these subgroup analyses were consistent with the original group comparisons, demonstrating that the relationships between the mean susceptibility values of ROIs and age and diabetes remained stable even when controlling for age differences.

Table 3 Determinants of the mean susceptibility values of altered brain regions: results of multiple linear regression analysis

Mediation analysis

Given the widespread and significant influence of age on the mean susceptibility values of gray matter nuclei across the three groups, particularly the stable effect observed in the striatum, we conducted an in-depth analysis to explore the relationships among these factors. We set age as the independent variable, striatum susceptibility as the dependent variable, and CSVD total score as the mediating variable, with gender and education level included as covariates (Table 4; Fig. 6). The mediation analysis demonstrated that age was positively associated with CSVD total score (estimated effect = 0.233, 95% CI: 0.0928 to 0.373, P = 0.001). Additionally, CSVD total score were significantly associated with the striatum susceptibility value (estimated effect = 0.137, 95% CI: 0.005 to 0.269, P = 0.041). These findings indicate a significant indirect effect of age on striatum susceptibility value mediated through CSVD total scores. Furthermore, the direct effect of age on striatum susceptibility value through CSVD total score was also significant (estimated effect = 0.337, 95% CI: 0.197 to 0.477, P < 0.001). This suggests that both direct and indirect pathways contribute to the association between age and striatum susceptibility value, highlighting the potential mediating role of CSVD total score in this relationship. Similarly, we conducted a mediation analysis of age, CSVD total score, and striatal susceptibility values in the subgroup analysis, and the mediation trend was consistent with that of the original grouping. For further details, please refer to Supplementary Table S4 and Figure S1 in the supplementary materials.

Table 4 Mediation analysis results

Fig. 6

Mediation analysis model. The CSVD-total score is an mediating variable between age and The mean susceptibility values of the striatum, suggesting that CSVD-total score is associated with Age and that the effect of age on striatum is mediated by the CSVD-total score. Striatum: including the caudate nucleus, putamen and globus pallidus

Discussion

Our results indicate a significantly higher iron deposition in GP, Put, and CN in the CSVD-S group, and a significant decline in cognitive function among subjects in this group, suggesting that iron overload may lead to more severe cognitive impairment in patients with higher total CSVD scores. As the most metabolically active organ in the body, the brain has a high demand for iron, an essential neurochemical catalyst or co-factor [22]. Nonheme brain iron is primarily stored in ferritin and plays a crucial role in influencing MRI [15]. As a paramagnetic substance, nonheme iron exhibits high susceptibility values on QSM images [23]. The distribution of these values closely corresponds to the distribution of iron accumulation in postmortem brain examinations [24], further confirming the role of nonheme iron in the QSM results [23]. Therefore, QSM may be a highly reliable tool for diagnosing and monitoring the severity of CSVD.

CSVD can be attributed to two main pathologies: arteriolosclerosis, linked to traditional vascular risk factors, and cerebral amyloid angiopathy, driven by β-amyloid protein [3]. Research on CSVD progression suggests that early endothelial dysfunction disrupts the blood-brain barrier (BBB), causing fluid and toxic plasma proteins to leak into surrounding tissues and the vascular media [2]. This leakage can adversely affect vascular reactivity, pericyte function, oligodendrocyte proliferation, and the drainage pathways of perivascular fluid. For instance, damage to oligodendrocytes, myelin-producing cells, can result in the release of iron-rich proteins, leading to the abnormal deposition of iron in other regions [2]. Additionally, the inflammatory reaction surrounding blood vessels can also lead to damage to blood vessel walls and changes in permeability [25]. Our findings reveal a significant increase in iron deposition in the GP, Put, and CN regions of the brains of CSVD-S patients, indicating a possibly more severe disruption of the BBB in local brain regions in these individuals.

The striatum comprising the neostriatum (CN and Put) and the paleostriatum (GP) plays a crucial role as a relay nucleus in the cortico-striatal-thalamo-cortical circuit, contributing to cognitive functions and the regulation of motor movements. Furthermore, the neurotoxicity of iron overload and its release in the form of free radicals within these brain areas may contribute to neuronal death and neural function impairment [26]. A high total CSVD score indicates more severe and widespread cognitive impairment, especially in information processing speed and overall cognitive function [27]. Our study showed that patients with higher total CSVD scores had more serious impaired cognitive function, and the severity of impaired cognitive function increased with the increase in iron deposition.

A higher total score in patients with CSVD is associated with more severe impairment of the cerebrovascular system and disruption of iron homeostasis in the brain, which have risk factors of age, gender, hypertension, diabetes, hyperlipidemia, smoking, and alcohol consumption are recognized as traditional risk factors for CSVD. However, local brain iron deposition in patients with CSVD is not the result of a single factor but rather the joint effect of multiple factors. Our study reveals that age is the primary factor affecting iron overload in the GP, Put, and CN of patients with CSVD, indicating a close relationship between age and the iron content increase in these brain regions. Various factors gradually manifest with age, such as decreased oxidative phosphorylation, functional decline of oligodendrocytes, and abnormal BBB permeability. Our data show that the brain iron content in the GP, Put, and CN increases with age, which is in line with previous research [28].

Another finding was that diabetes also affected the iron deposition in CN and Put in the brains of patients with CSVD. A study on the characteristics of iron deposition in deep gray matter in the elderly hypothesized that the iron deposition in T2DM patients was due to the changes in BBB permeability caused by hyperglycemia induced neuronal damage and insulin resistance [29]. Our study of patients with CSVD suggests the need for increased testing of blood glucose levels in patients with CSVD. Furthermore, there has been research indicating that smoking is a crucial determining factor for the accumulation of brain iron in normally aging individuals. Smoking has been shown to be associated with iron deposition in the basal ganglia [30], and our study has demonstrated that smoking can affect the brain iron content of patients with CSVD in the Put. These findings suggest that we need to take into account the impact of smoking when considering the management and treatment of patients with CSVD.

Mediation analysis further revealed that age is a potential pathway to influence the mean susceptibility values of striatum through CSVD total score. It is found that age not only has an indirect effect on the mean susceptibility value of striatum through CSVD total score, but also has a significant direct effect on it, suggesting that iron deposition in striatum increases with age. Specifically, age has a significant positive effect on the CSVD total score, which in turn is positively associated with striatal mean susceptibility values. This suggests that with advancing age and a higher CSVD score, the severity of CSVD correlates with elevated mean susceptibility values in the striatum. This mediating effect implies that worsening CSVD with age contributes to pathological changes in the striatum region. The striatum plays a crucial role in cognitive and executive functions, especially in learning, decision-making, reward processing, inhibitory control, task switching, working memory, and error monitoring [31] Our study further supports this connection by finding significant positive correlations between the mean susceptibility values of the GP, Put, and CN with SCWT scores. Our study further supports this connection by finding significant positive correlations between the mean susceptibility values of the GP, Put, and CN with SCWT scores. Therefore, greater severity of CSVD results in reduced blood flow, and the striatum, when subjected to chronic ischemic conditions, may experience neuronal energy deficits, affecting its structure and function. The combined effects of aging and CSVD progression further exacerbate degenerative changes in the striatum.

The mediation analysis results in our study support the critical role of CSVD in age-related functional decline of the striatum, suggesting that interventions to curb CSVD progression could help protect the structure and function of the striatum, potentially slowing age-associated declines in cognitive and executive functions. This provides a theoretical basis for future intervention and mechanistic studies. Additionally, a mediation analysis conducted on age-matched subgroups also indicated a near-significant trend in the mediating role of the CSVD total score between age and striatal iron deposition, further details can be found in Supplementary Table S4 and Fig. 1.

Research has indicated that iron deposition in the GP is independently and positively correlated with the severity of WMHs, highlighting a significant link between tissue iron accumulation and the development of WMHs [6]. This aligns with our study’s findings, which similarly confirm the association between brain iron levels and WMH severity. WMHs are known to be associated with endothelial dysfunction and increased blood-brain barrier permeability [16], factors that are also implicated in the mechanisms underlying brain iron deposition [5]. Therefore, this evidence supports the idea that WMHs contribute to overall susceptibility changes seen in CSVD by influencing iron accumulation. Additionally, PVS, an important component of the brain’s glymphatic clearance system, play a key role in interstitial fluid and waste drainage, which is particularly relevant in the context of CSVD [32, 33]. Studies suggest that, in the healthy aging brain, the glymphatic system participates in iron clearance, while its dysfunction may lead to increased iron deposition [34]. In our study, patients with prominent PVS in specific brain regions, such as the Put, presented elevated susceptibility, supporting this theory and linking PVS to iron-related changes in CSVD. Further, lacunes and CMBs, which are common CSVD markers, share similar pathophysiological mechanisms. Lacunes are small, round or oval cavities in the subcortical white matter that often contain CSF-like fluid, surrounded by various levels of gliosis, axonal damage, and hemosiderin deposition. CMBs, on the other hand, are characterized by erythrocyte leakage due to small vessel damage, followed by local hemosiderin accumulation through macrophage phagocytosis [35]. Both lesions are closely related, with CMBs shown in previous studies to cause ischemic microvascular damage and subsequent blood-brain barrier disruption or inflammatory responses that may ultimately result in iron deposition [36].

Given the diagnostic limitations of using any single imaging marker, our study used the composite CSVD total score to reflect CSVD’s total burden on brain tissue, as it better captures the multi-faceted impact of CSVD on brain iron deposition. This approach accounts for both gray matter (e.g., lacunes) and white matter (e.g., WMHs) alterations [37, 38]. Our findings indicate that using the composite CSVD total score can more comprehensively depict CSVD’s cumulative influence on brain susceptibility measures, while also allowing us to recognize distinct contributions from individual CSVD features, such as WMHs and PVS, to susceptibility values. In summary, while individual CSVD components like WMHs, PVS, and CMBs contribute independently to susceptibility changes, the composite CSVD total score serves as a robust marker to reflect the overall burden of CSVD, especially as it relates to cumulative iron deposition across different brain regions.

This study has the following limitations that may be addressed in future investigations. QSM values were analyzed using ROI, which is a vast reduction of imaging information. QSM texture analyses [39] and radiomics [40] may reveal more specific information on about CSVD. QSM is sensitive to both deoxyheme iron in small veins and capillaries and nonheme iron in tissue mostly stored in ferritin, which can be separated by combining quantitative blood oxygenation level dependent modeling of mGRE magnitude signal with QSM processing of mGRE phase signal [41]. This leads to mapping of nonheme iron and oxygen extraction fraction without additional data acquisition. These further analyses may improve the accuracy in distinguishing CSVD-S patients from the HC group.

Conclusion

Our investigation revealed a correlation between increased iron deposition in the basal ganglia region and greater cognitive dysfunction in patients with CSVD with high CSVD total scores. Factors such as age, diabetes history, and smoking history have been found to exacerbate local brain iron deposition in patients with CSVD, which in turn mediates cognitive dysfunction in individuals with comorbid hypertension and CSVD. These findings underscore the critical role of iron deposition in the pathogenesis of CSVD, particularly in regard to its deleterious effects on cognitive function in high-risk patient populations. As such, future research should focus on developing targeted interventions aimed at reducing iron accumulation in the brain as a potential strategy for improving cognitive outcomes in individuals with CSVD.