Effects of cognitive reserve proxies on cognitive function and frontoparietal control network in subjects with white matter hyperintensities: A cross‐sectional functional magnetic resonance imaging study

Abstract Aims This study aimed to analyze the potential association between cognition reserve (CR) components, including education, working activity, and leisure time activity, and cognitive function in subjects with white matter hyperintensities (WMH). The study also explored the role of the frontoparietal control network (FPCN) in such association. Methods White matter hyperintensities subjects with and without cognitive impairment (CI) were evaluated with multimodal magnetic resonance imaging, neuropsychological testing, and CR survey. FPCN patterns were assessed with dorsolateral prefrontal cortex seed‐based functional connectivity analysis. Results Education was positively associated with cognitive function in WMH subjects with or without CI, whereas working activity and leisure time activity were positively associated with cognitive function only in those without CI. Similarly, education was associated with bilateral FPCN in both WMH groups, whereas working activity and leisure time activity were associated with bilateral FPCN mainly in the group without CI. Furthermore, FPCN partially mediated the association between education and cognitive function in both WMH groups. Conclusion Education showed a positive impact on cognitive function in WMH subjects regardless of their cognitive status, whereas working activity and leisure time activity exhibited beneficial effects only in those without CI. The FPCN mediated the beneficial effect of education on cognitive function.


| INTRODUC TI ON
White matter hyperintensities (WMH), the most common morphologic feature on brain magnetic resonance imaging (MRI) in cerebral small vessel disease, is common among older people. [1][2][3] There is increasing evidence suggesting that WMH can cause cognitive decline and plays a significant role in the etiology of vascular cognitive impairment (CI). 4,5 However, individuals with WMH exhibit high levels of heterogeneity in cognitive performance, and a portion of these individuals even maintain normal cognitive function. 6,7 This high heterogeneity may be related to the effect of cognitive reserve (CR) activities.
Cognitive reserve is a theoretical concept that explains the individual differences in maintaining cognitive function in the face of brain pathology. First, greater CR appears to be associated with better cognitive performance in WMH subjects. 6,8,9 Second, CR may act as a moderator between the burden of WMH and cognitive performance. A more significant burden of WMH links to poorer cognitive performance in subjects with low CR. In contrast, this association tends to be weakened or disappear in subjects with high CR. [10][11][12] CR is commonly measured with proxies, and the Cognitive Reserve Index questionnaire (CRIq) evaluates CR from three aspects: education, working activity, and leisure time activity. 13 The three aspects contribute independently and differentially to CR. 14 Whether these proxies play distinct roles in affecting cognitive function in WMH subjects remains elusive.
Recent studies on Alzheimer's disease (AD) have revealed stagedependent effects of CR on cognitive function across the AD spectrum. CR's positive effects on cognition are stronger in predementia stages than in dementia stages. 15,16 High CR attenuates the cognitive decline in predementia stages, but accelerates cognitive decline in dementia stages. 17 For WMH subjects, Zahodne et al. 18 showed that education mitigated the effect of WMH on cognitive function in subjects at lower risk for dementia, but exacerbated the effect in those at higher risk for dementia. To the best of our knowledge, whether the effects of CR on cognition differ between WMH subjects with or without CI remains unclear.
The link between CR and brain activities has been widely explored using task-based functional magnetic resonance imaging (fMRI) techniques. Colangeli et al. 19 performed a meta-analysis of 17 fMRI studies and found that higher CR was associated with greater activation in frontal, parietal, and anterior cingulate regions in healthy elderly subjects. The frontoparietal control network (FPCN) is thought to flexibly support multiple resting-state networks, e.g., the default mode network and dorsal attention network, to complete cognitive tasks, thus serving as a "regulating" role. 20,21 Alterations in FPCN patterns are related to executive function and attention, commonly affected in vascular CI. [22][23][24] Our recent restingstate fMRI study found that an altered FPCN pattern, i.e., increased within-network functional connectivity (FC) of the FPCN and decreased FC between the FPCN and the default mode network, was related to CI in subjects with WMH. 25 However, the role of CR in such a relationship remains to be defined.
In this study, WMH subjects without CI, WMH subjects with CI, and healthy control (HC) subjects underwent multimodal MRI scans, neuropsychological testing, and CR assessment. We aimed to (1) determine the association of each CR aspect with cognitive function across the three groups and (2) explore the role of the FPCN in the association between CR and cognitive function in WMH subjects.
We hypothesized that the FPCN could mediate a positive effect of CR on cognitive function in WMH subjects. Committee. All subjects provided written informed consent.

| Participants
The inclusion criteria and exclusion criteria were previously described. 26 The inclusion criteria for subjects with WMH were as follows: (1)  To quantify CR, we conducted a survey with the CRIq and generated the composite index of CR, i.e., the Cognitive Reserve Index (CRI). 13 All participants with normal cognitive function were asked questions about three indicators of CR: education, working time activity, and leisure time activity. For those with CI, the questions were asked of a family caregiver who was familiar with the present and past habits of the subject. After collecting the self-reported information about CR, we obtained the scores for each of the three aspects of the CRIq: CRI-education, CRI-working activity, and CRI-leisure time activity, as well as the total CRI for each subject. Detailed information is shown in the Supplemental Material.

| MRI acquisition
The procedure for MRI scanning was described previously. 25,30 All subjects underwent MRI scanning on a 3.0-Tesla MRI scanner (Ingenia 3.0T, Philips Medical Systems, Eindhoven, Netherlands) with a 32-channel head coil. Detailed procedure is shown in the Supplemental Material.

| Volume assessment of grey matter, whole brain, and WMH
As mentioned in our previous research, 30  Detailed procedure is shown in the Supplemental Material. The volumes of grey matter, white matter, and cerebrospinal fluid were obtained, and the whole brain volume was calculated as the sum of these three values. Grey matter atrophy is a calculation of grey matter volume divided by the brain volume.
The volume of WMH lesions was evaluated on T1 and T2-FLAIR images using the Lesion Segmentation Tool (LST) toolbox version 2.0.151 (http://www.stati stica l-model ling.de/lst.html) for SPM12. 31 Detailed procedure is shown in the Supplemental Material.

| FMRI preprocessing and network mapping
The resting-state fMRI data were preprocessed using Data Processing and Analysis of Brain Imaging (DPABI 2.

| Statistical analysis
The Kolmogorov-Smirnov test was used to assess the data normality of continuous variables. Normally distributed data were presented as the mean ± standard deviation (SD) and analysed using one-way analysis of variance (ANOVA). Non-normally distributed data were presented as medians (interquartile range) and analysed using a Kruskal-Wallis test. Chi-square tests were applied to compare the sex ratio among the three groups. We examined the relationship between the total CRI or each aspect of CRI (as independent variables) and the MoCA and MMSE scores (as dependent variables) using multiple linear regression analysis with adjustment for age, sex, WMH volume, whole brain volume, and grey matter atrophy rate in each group. While each aspect of CR served as an independent variable, the other two aspects of CR were additionally treated as covariates. To explore the relationship among the three aspects of CR, Pearson's correlation analyses were performed between any two aspects in each group. These analyses were performed using Statistical Package

| Demographic, neuropsychological, and CR data
As shown in Table 1, no significant differences in sex, years of education, whole-brain volume, grey matter atrophy rate, CRI, CRIeducation, or CRI-working activity were found among the three groups. WMH groups were significantly older than the HC group.
The WMH with CI group had significantly greater WMH volume, poorer performance in all cognitive domains, and lower CRI-leisure time activity than the other two groups. Correlative analyses showed that the three aspects of CR were significantly and positively correlated in both the WMH with CI group and the HC group, not in the WMH without CI group (Table S1 in Supplemental Material).

| Association of CR with cognitive function
As shown in Table 2, CRI and CRI-education were positively associ-

| Association of each aspect of CR with the right FPCN
Both CRI-working activity and CRI-leisure time activity were significantly associated with the FC of the right FPCN in frontal, parietal, and cingulate regions only in the WMH without CI group ( Figure 1A,C), not in the WMH with CI group (Figure 1B,D). In contrast, CRI-education was significantly associated with the FC in the left DLPFC in both WMH groups (Figure 2A-B  Note: Values are presented as mean ± stand deviation (SD) or median (IQR, interquartile range). Grey matter atrophy is a calculation of grey matter volumes divided by the brain volume; lower values indicate more grey matter atrophy. One-way ANOVA was applied in the analyses of age, education, brain volume, brain atrophy rate, memory, and processing speed. χ 2 test was applied in the analysis of gender. The Kruskal-Wallis test was applied in the analyses of WMH volume, MMSE, MoCA, executive function, visual-spatial ability, and cognitive reserve data. Significance is highlighted in bold (p < 0.05). a p < 0.05, differs from the control group. b p < 0.05, differs from the WMH without CI group. Previous studies investigating the relationship between the increased FC in the frontal lobe and cognitive function have yielded conflicting results. Some studies showed that increased frontal FC was associated with better cognitive performance in cognitively normal older adults, suggesting a compensatory neural process. [45][46][47] According to the famous model named the "scaffolding theory of aging and cognition (STAC)", with the neuronal declines, compensatory scaffolding, i.e., compensatory recruitment or reallocation of cognitive resources, could be induced to maintain cognitive function and life-course factors (including CR) could regulate the process. 48,49 However, other studies showed that increased frontal FC was associated with worse cognitive performance in healthy elderly or subjects with mild cognitive impairment, 50,51 suggesting that the increased FC might reflect pathology-or age-related dedifferentiation of brain activities and could be harmful. In the present study, higher FC between bilateral DLPFC was associated with better cognitive performance in WMH subjects without CI and poorer cognitive performance in those with CI. The increased frontal FC suggests a compensatory process in subjects with WMH before the onset of CI but pathology-related dedifferentiation of brain activities with the onset of CI.

ACK N OWLED G EM ENTS
This work was supported by the National Natural Science