We assessed the spatial patterns of WMH using Fazekas rating scale and then investigated their associations with underlying etiologies in a population-based cohort. The main conclusions are: (1) information provided by DWMH and PWMH are highly correlated and similar across the associations we tested; (2) both PWMH ≥ 2 and DWMH ≥ 2 are often seen in participants greater than 80 years of age, greater than two cardiovascular metabolic conditions, and global z-score < 0; (3) we found a weak positive correlation of PWMH and DWMH burden with amyloid-PET measures, but not with tau-PET measures, and these associations did not differ between PWMH and DWMH; (4) 2D GE vs. 3D Siemens FLAIR did not yield substantial differences in WMH assessments of both visual and automated measurements, supporting the decision to pool data; (5) the automated assessments were able to predict abnormal Fazekas scale on both 2D and 3D FLAIR images with similar AUROC.
Association of Fazekas measures with age and vascular risk
It is well-established that age plays a pivotal role in the prevalence and severity of WMH, with strong evidence for increased WMH presence on MRI with increased age. In alignment with previous literature on WMH [26, 27], we found significant association between age and severity of PWMH and DWMH graded via the Fazekas scale. Furthermore, studies such as the retrospective investigation conducted by Zhuang et al., provided further clarity on the independent role of age as a risk factor for prevalence and severity of WMH, even after controlling for sex, education, hyperlipidemia, and hyperhomocysteinemia [28]. WMH were previously known to be a normal consequence of aging with limited clinical importance [29]. However, in recent years, multiple studies have established the connection between deteriorating vascular health and increased WMH. Notably, WMH tend to develop in areas characterized by low cerebral perfusion [29, 30] and exhibit heightened prevalence and severity in the presence of conditions like hypertension [31,32,33], diabetes [34], and recurrent risk of stroke [35].
Our study shows significant correlations for both PWMH and DWMH with higher age and with CMC burden after adjusting for age, suggesting a pathophysiologic involvement of diffusely increased WMH, rather than localization of WMH changes to periventricular or deep regions as a response to worsening vascular health and age. In addition to age, hypertension, especially elevated systolic blood pressure, is a well-known factor linked with WMH presence, especially in adults > 65 years [36]. This is strengthened by the fact that WMH are formed from endothelial activation, inflammation, and ischemic damage, especially increased central arterial stiffness and pressure and increased cerebral blood flow pulsatility [37, 38], all of which are worsened by hypertension [36]. The duration of hypertension also has an incremental impact on WMH, as suggested by de Leeuw et al. 2002 who demonstrated that periventricular and subcortical WMH are associated with duration of hypertension [5]. While controlling blood pressure with antihypertensives is widely known to reduce cardiovascular morbidity and mortality risk among endless health benefits, their role in preventing WMH is not unanimous. Some studies demonstrate that successful control of hypertension (lifestyle modifications, antihypertensives) reduces the risk of dementia [36] and white matter lesions [39]. However, results from the Cardiovascular Determinants of Dementia (CASCADE) study, a multicenter European collaborative study, revealed that sudden hypoperfusion from aggressive blood pressure management in chronic hypertension cases increases further risk of WMH [32]. This is attributed to vascular remodeling and hyalinosis over time that increases flow requirements to maintain adequate perfusion, which can be disrupted by sudden control. Overall, our study reinforces the established connections between Fazekas WMH grading, age, and CMC, shedding light on the intricate interplay between these factors in the context of white matter integrity in the brain.
Association of Fazekas WMH measures with AD imaging biomarkers
WMH are known to be more severe and prevalent in individuals with AD compared to healthy controls of the same age [40, 41]. Currently, two theories for the involvement of WMH in AD exist: (1) additive theory, where WMH is primarily due to cerebrovascular disease and it lowers the threshold for AD diagnosis independent of AD pathology, and (2) interactive theory, where WMHs interact with amyloid/tau to potentiate their effects on cognitive impairment [19]. There is also a distinct spatial pattern for WMH in AD individuals, with a greater involvement of posterior parieto-occipital and periventricular areas, compared to healthy controls of the same age [42,43,44]. Potential hypothesized mechanisms underlying formation of WMH in AD include Wallerian degeneration secondary to tau neurofibrillary tangles [45], overlap of amyloidosis from CAA with AD [19], and neuroinflammation, although this has mostly been observed in SVD [46].
A previous study by our group as well as others have established a link between WMH and amyloid burden, localized to regions with lobar cerebral microbleeds, indicating the involvement of CAA in increased WMH [18, 36]. The current study further investigates this association by investigating differential involvement of PWMH and DWMH, quantified according to standardized Fazekas scale ratings, with amyloid and tau burden with a large number of participants. Although the association was minimal, our study identified a correlation between increased DWMH and PWMH burden with amyloid burden. Furthermore, neither PWMH nor DWMH had a significant association with tau burden, which contrasts findings in McAleese et al., where they demonstrated an independent influence of cortical tau burden on the severity of WMH [20]. This disparity in results is likely due to differences in the characteristics of study populations. Our cohort primarily comprises of cognitively unimpaired individuals with lower tau burden. However, our population characteristics are reflective of the general population that enabled us to investigate the influence of AD pathology influences on these two common patterns of WMH.
Association of Fazekas WMH measures with cognition
Consistent with prior literature that reported association between worsening cognitive impairment and increased WMH burden [47], we also found significant negative relationship between periventricular and deep WMH with global cognition. Poor vascular health, especially cerebrovascular, significantly contributes to worsening cognitive impairment and rates of dementia [48]. Furthermore, the impact of WMH observed on cognitive function has been evaluated globally [48] as well as differentially on specific cognitive domains, executive dysfunction [48,49,50,51,52,53,54,55] and inconsistent memory impairment [48, 50, 53, 56,57,58].
A more recent study has made efforts to create comprehensive neural mapping of structures that directly correlate WMH to domain-specific cognitive task impairments. The anterior thalamic radiations, forceps major, and left inferior fronto-occipital fasciculus are major drivers of decline in tasks related to executive function, information processing, language, and verbal memory [59]. While Coenen et al. 2023 found that cognitive impairment was independent of total WMH volume loss, another recent study regards WMH volume change as a more dynamic process that can help predict cognitive dysfunction in individuals with minor strokes, which is a common comorbidity in those with WMH [60]. Of note, WMH shape including confluence, rather than volume, may also play a role in increased executive dysfunction and memory impairment, as reported by Zwartbol et al. 2022 [61].
Low distinction between PWMH and DWMH
In the present study, global cognition was associated with PWMH and DWMH burden, demonstrating an equal level of significance for both, indicating the collective impact of diffuse WMH changes across the white matter networks in the brain on cognitive decline, rather than localization to periventricular or subcortical regions. Our findings suggest an absence of differential involvement of DWMH or PWMH in VCID, given both scorings’ significant association with CMC and cognitive impairment. This aligns with established findings of global cognitive impairment in multiple domains rising from equal dysfunction and damage to both WMH from deeper axonal tract and periventricular regions. In contrast, a recent meta-analysis by Botz et al. [10] reported a distinctive spatial correlation between WMH and cognition using ROI, visual rating, and voxel-wise approaches, suggesting the spatial distinction as an early marker of cognitive impairment in younger individuals. The observed contrasting findings might be due to the difference in the studied population including cognitively normal, participants with MCI or AD dementia, while we included a population-based sample in which only 13% were cognitively impaired.
Strengths and limitations
The main strength of this study is its extensive analysis of the relationship of both PWMH and DWMH and cognition, cerebrovascular health, and amyloid and tau burden. While these correlations have previously been investigated and assessed, this study uniquely divides WMH into its subtypes based on localization in the brain and attaching a standardized, quantitative scale to appropriately classify the extent and severity of WMH presence. The characteristics of the population-based sample in this study strengthen the generalizability of the results of this study. In the interest of maximizing our sample size and thus the power of this study, we tested whether we could combine data from our GE 2D FLAIR and Siemens 3D FLAIR subsamples. The comparative sensitivity and specificity of each were determined and compared, both individually and combined, yielding satisfactory results. However, it should be noted that our in-house WMH measurement algorithm contains steps to internally adjust for manufacturer and 2D vs. 3D FLAIR in its assessments, developed based on previous analyses of our head-to-head crossover study. Readers should not incorrectly conclude that these automated measurements would be directly combinable or comparable if they came from other WMH measurement software and/or without steps for scanner harmonization.
Our study has certain limitations. First, the cross-sectional design of the study restricts our ability to establish cause-effect relationship between age, CMC, AD biomarkers and the spatial distribution of WMH. Second, risk factors such as BP, HbA1c levels, smoking and drinking were not considered. Third, although the lobar regional WMH measures may be more relevant in the context of AD pathology, we focused solely on the contribution of periventricular and deep WMH to global AD biomarkers, although the Fazekas scale is a widespread, established measure using the two aforementioned divisions. Lastly, the relatively small number of individuals with a Fazekas grade of 3 may impact the generalizability of our findings. Future longitudinal studies will target how regional distribution of WMHs influences the pathophysiological cascade of AD and dementia, with a specific focus on discerning the differential involvement and association of cognitive domains such as attention and memory that with WMH burden.
Add Comment