2023 Impact Factor
2023 Impact Factor
Alzheimer’s disease (AD) is a neurodegenerative disorder that is characterized by the progressive deterioration of cognition, affecting memory, language, and executive function [1]. It is often accompanied by the extracellular amyloid-beta (Aβ) plaque deposition and intracellular aggregation of misfolded tau proteins [1,2]. However, the extent of age-related brain pathophysiology does not necessarily reflect the degree of cognitive decline, as cognitive reserve acts as a moderator [3]. The concept of cognitive reserve suggests that the brain compensates for the pathology by adapting and reorganizing its neural networks [4]; individuals with greater cognitive reserve tolerate more AD pathology, and the onset of cognitive symptoms of dementia is often delayed [3].
The level of education is commonly used as a proxy for cognitive reserve in the cognitively normal (CN) older adults [3,5], as it is linked to mature brain development and increased synaptic density [6] and is a readily available and easy-to-measure variable [7]. Additionally, numerous studies have shown a positive correlation between the years of education and cognitive performance across different domains such as memory, executive functioning, and language skills [8]. This suggests that more education leads to a stronger cognitive reserve, which may delay the onset of cognitive decline [5] and dementia [9,10] .
Resting-state functional magnetic resonance imaging (MRI) reveals spontaneous neuronal activity of human brain in resting-state [11] and longitudinal studies have demonstrated that changes in resting-state functional connectivity (rFC) may serve as an early marker of AD progression [12]. A decline in rFC has been observed in individuals with mild cognitive impairment (MCI) [13], which often precedes AD, as well as in CN older adults with AD risk factors [14,15]. These changes in rFC over the course of AD are not uniform. Both impairment and compensation of rFC can occur [16,17], and changes have been found to differ depending on the stage of AD [18]. In particular, there is a growing interest in rFC indicators specific to the preclinical stage [19], where the underlying pathology may be latent but not yet manifested in clinical symptoms, making it an opportune time for preventive and therapeutic intervention [20]. In this regard, deriving preclinical stage-specific biomarkers is of clinical im-portance. In prior research, CN older adults with Aβ deposition have displayed remote rFC disconnection [21], but increased local rFC in cortical hub regions [22]. Additionally, an increased lateralized FC has been found in CN older adults with APOE ε4 carrier compared with those with APOE ε4 non-carrier [23].
Additionally, several studies have shown that FC within intrinsic functional networks such as the default-mode network (DMN), central-executive network (CEN), salience network (SN), and dorsal attentional network (DAN) are selectively disrupted in the trajectory of AD compared to CN group [24]. In addition, differential alterations in various intrinsic functional networks have been reported in the trajectory of AD [25] and aging population [26]. In the cognitively normal group, which is the focus of this study, rFC decreased with age in most networks except the visual network [26]. In addition, another previous study has shown that the increased FC of SN, but decreased FC of other networks between CDR 0 and 0.5 [18]. Furthermore, studies using multimodal imaging have found that rFC changes in AD are associated with the accumulation of Aβ and tau proteins, the primary pathological hallmarks of AD [27-30]. The findings support the use of rFC as an intermediate phenotype to reflect AD progression [31,32]; however, it should be noted that its efficacy as a reliable intermediate phenotype hinges on the careful consideration of various factors, including the stage of dementia progression and the presence of specific AD risk factors.
Several studies have investigated the association between education years and rFC in the progression of AD. In the AD patients, the FC of posterior cingulate cortex within DMN has displayed a strong correlation with education years compared to those with MCI and cognitively normal older adults [33]. Additionally, global FC changes within the cognitive control network has predicted education years in MCI patients due to AD [34]. However, limited research has explored the association between years of education and rFC in CN individuals. A previous study has shown that education years are positively associated with the rFC between the anterior cingulate cortex and the hippocampus as well as the inferior frontal lobe, posterior cingulate cortex and angular gyrus in CN older adults [35]. In addition, in a prior study evaluating task-based FC, higher cognitive reserve proxy including education years has shown an association with reduced brain functional activity during cognitive processing, implicating more effective use of cerebral networks in CN participants [36].
As a representative risk factor for AD, APOE ε4 allele is not only related to the risk of dementia independently, but also interactively in association with the level of education [37]. Additionally, higher education has been demonstrated to be related to increased fluorodeoxyglucose positron emission tomography (PET) uptake within brain regions with positive Aβ deposition in CN group [38]. Furthermore, APOE ε4 genotype and Aβ accumulation has also been reported to affect the rFC in the preclinical phase of AD [27,28,39]. However, previous studies on the association between years of education and functional intermediate phenotype have a limitation in that they do not fully account for representative risk factors for AD, such as Aβ deposition and APOE ε4 genotype. Moreover, given the preceding disruption of FC before the onset of clinical phenotype in the trajectory of AD [12], identifying how education influences rFC in the preclinical phase may provide insights into the role of cognitive reserve against the early progression of AD.
In this regard, we aimed to evaluate whether the relationship between years of education and rFC in CN older adults differs depending on Aβ deposition and APOE ε4 carrier status as effect modifiers. In the current study, we evaluated the rFC in the AD vulnerable neural networks including DMN, CEN, SN, and DAN. Additionally, to further characterize the rFC that was significantly associated with years of education, we also assessed how the association with cognitive function varied by effect modifiers.
One hundred twenty-one subjects between 60 and 85 years old were recruited from volunteers registered in the Catholic Aging Brain Imaging database, which contains brain scans of patients who visited the outpatient clinic at the Catholic Brain Health Center, Yeouido St. Mary’s Hospital, The Catholic University of Korea, from 2018 to 2020. The cognitive function of all subjects was assessed using the Korean version of the Consortium to Establish a Registry for AD (CERAD-K) [40]. The measurements included assessments regarding the Korean version of the verbal fluency test, the 15-item Boston Naming Test, Mini-Mental State Examination (MMSE-K) [41], word list memory (WLM), word list recall (WLR), word list recognition (WLRc), constructional praxis, and constructional recall (CR). In addition, total memory (TM) scores were obtained by summing the respective scores from the WLM, WLR, and WLRc tests. The total CERAD-K scores were calculated by summing all subcategory scores, excluding the MMSE-K and CR scores. The inclusion criteria were as follows: (1) unimpaired delayed memory function, quantified by scoring above age-, sex-, and education-adjusted cut-offs on the WLR domains; (2) MMSE-K score between 24 and 30; (3) Clinical Dementia Rating score of 0; (4) Memory Box score of 0; (5) normal cognitive function based on the absence of significant impairment in cognitive function or activities of daily living; and (6) no family history of AD. We excluded participants with a history of alcoholism, drug abuse, head trauma, or psychiatric disorders and those taking any psychotropic medications (e.g., cholinesterase inhibitors, antidepressants, benzodiazepines, and antipsychotics), those with uncontrolled multiple cardiovascular risk factors (e.g., uncontrolled arterial hypertension, diabetes mellitus, dyslipidemia, cardiac disease including coronary heart disease, arrhythmia, etc.), and those with evidence of subcortical ischemic changes corresponding to a score ≥ 2 on the Fazeka’s scale [42]. T2-weighted fluid-attenuated inversion recovery data were acquired to objectively exclude vascular lesions or other diseases. Participants underwent [18F] flutemetamol (FMM) PET-computed tomography (CT) within 3 months of the MRI scan. The study was conducted under the ethical and safety guidelines set forth by the Institutional Review Board of The Catholic University of Korea, which approved all research activities (SC18TESI0143). Written informed consent was obtained from all participants.
Imaging data were collected by the Department of Radiology of Yeouido Saint Mary’s Hospital at the Catholic University of Korea using a 3-T Siemens Skyra MRI machine and a 32-channel Siemens head coil (Siemens Medical Solutions). The scanning parameters of the T1- weighted three-dimensional magnetization-prepared rapid gradient echo sequences were as follows; echo time (TE) = 2.6 ms, repetition time (TR) = 1,940 ms, inversion time (TI) = 979 ms, flip angle (FA) = 9°, field of view (FOV) = 250 × 250 mm, matrix = 256 × 256, and voxel size = 1.0 × 1.0 × 1.0 mm3. Fluid attenuated inversion recovery MRI sequences were as follows: TE = 135 ms; TR = 9,000 ms; TI = 2,200 ms; FA = 90°; FOV = 220 × 220 mm; matrix = 356 × 231; and voxel size = 1.0 × 1.0 × 1.0 mm3. Resting-state fMR images were collected using a T2* weighting gradient echo sequence with TR = 2,000 ms, TE = 30 ms, matrix = 128 × 128 × 29, and voxel size = 1 × 1 × 2 mm3. We acquired 150 volumes in 5 minutes, with the instruction, “keep your eyes closed and think of nothing in particular”.
The CONN Toolbox version 20.b. was utilized to preprocess and analyze functional imaging data and calculate the FC of the resting-state brain networks [43]. The default CONN preprocessing pipeline was utilized, with a sequential process of realignment, unwarping, slice-time correction, scrubbing with Artifact Detection and Removal Tool (ART)-based identification for outlier scans, segmentation into gray matter, white matter, and cerebro-spinal fluid (CSF), normalization to the Montreal Neuro-logical Institute (MNI) template, smoothing using an 8-mm Gaussian kernel. After preprocessing, the CONN denoising pipeline was implemented to remove possible confounders in the blood oxygen level dependent signal. Through the adoption of linear regression and band-pass filter of (0.008−0.09) Hz, removal of white matter, CSF noise components, unwanted subject motion, and physiological noises were conducted. The fMRI data were then parcellated using the functional atlas included in the CONN toolbox. The functional regions of interest (ROIs) considered here included DMN, SN, CEN, and DAN.
We evaluated functional properties from preselected seed regions defined in CONN Toolbox version 20.b. using seed-based connectivity analysis. As seed regions, the same functional atlas-based regions were used as described above focusing on DMN, SN, CEN, and DAN. The given coordinates of each resting-state network were as follows; 1) DMN: posterior cingulate cortex (MNI space: 1, −61, 38); 2) SN: left anterior insula (MNI space: −44, 13, 1); 3) CEN: right dorsolateral prefrontal cortex (MNI space: −27, −9, 64), and 4) DAN: left frontal eye field (MNI space: −27, −9, 64). For each subject, the mean time series of each seed region was computed as the reference time course for each network. Pearson cross-correlation analysis was performed between the seed time course and the time course of the whole-brain voxels. A Fisher’s z-transformation was applied to improve the normality of the correlation coefficients [44]. Finally, the individual maps of each network were obtained. In first-level analyses, computation of seed-to-voxel connectivity maps was implemented in each subject, and these measures were adopted in group-level analysis.
DNA was isolated from blood using the QIAmp Blood DNA Maxi Kit protocol (Qiagen). Genotypes for two APOE SNPs, rs429358 (E*4) and rs7412 (E*2) were determined using TaqMan SNP genotyping assays (Applied Biosystems). Considering the protective effect of APOE e2 allele [45], we excluded participants with the APOE e2 allele. If a participant had at least one APOE ε4 allele, they were categorized as an APOE ε4 carrier; if they had no APOE ε4 allele, they were categorized as an APOE ε4 non-carrier.
FMM was manufactured, and FMM-PET data were collected and analyzed as described previously [46]. Static PET scans were acquired from 90 to 110 minutes after 185 MBq of FMM injection. MRI for each participant was used to co-register and define the ROIs and correct partial volume effects that arose from expansion of the cerebrospinal spaces accompanying cerebral atrophy using a geometric transfer matrix.
The semi-quantification of FMM uptake on PET-CT scan was performed by obtaining the standardized uptake value ratios (SUVRs). The SUVRs were measured using SCALE PET from Neurophet Inc. [47]. The T1 MRI was utilized along with PET imaging for the structural infor-mation. The volumes of interest (VOIs) were restricted to gray matter, covering the frontal, superior parietal, lateral temporal, anterior, and posterior cingulate cortex/precuneus regions. These VOIs were also considered in a previous study [46]. The reference region for SUVR calculations was pons. The mean uptake counts of each VOIs and reference region were measured on the preprocessed image. A regional SUVR was calculated as the ratio of each cortical regional mean count to the pons mean count (SUVRPONS). The global cortical average (com-posite SUVR) was calculated by averaging regional cortical SUVRs weighted for size. The detailed description is appeared in our previous study [47]. We used a cut-off of 0.62 for “positive (Aβ+)” versus ‘negative (Aβ−)’ neocortical SUVR, consistent with the cut-off values used in a previous FMM PET study [46]. PET scans classified with negative Aβ accumulation also exhibited normal visual reading.
Statistical analyses were performed using R software (version 4.0.5), jamovi (version 1.6.23) (https://www.jamovi.org), and SPM 12. Assumptions of normality were tested for continuous variables using the Kolmogorov–Smirnov test in R software; all data demonstrated a normal distribution and are standardized by z score transfor-mation for the analysis. The two-sample ttest and chi- square (c2) tests were used to probe for differences in demographic variables, clinical data, global FMM SUVRPONS, and neuropsychological performance scores between Aβ+ and Aβ− groups. All statistical analyses were conducted considering a two-tailed p value < 0.05 to define statistical significance.
The general linear model was used to evaluate the education years-by-effect modifiers (Aβ deposition and APOE ε4 carrier status) interaction for the functional connectivity from the seed regions of each network, adjusting for age, sex, and effect modifier not included in each interaction evaluation. Statistical significance was deter-mined at p < 0.05, corrected for multiple comparisons using permutation test as implemented in the CONN Toolbox version 20.b. using 5,000 permutations and threshold-free cluster enhancement (TFCE) [48]. TFCE is a technique that does not need pre-statistical smoothing of images and does not depend on an initial cluster-forming threshold such as t-statistic thresholding. However, TFCE needs several parameters to be set. Specifically, these relate to the cluster extent (E) and peak height (H) of the statistic at a given voxel. As suggested by prior research [48], these were set to E = 0.5 and H = 2.0 to provide a statistic that is sensitive to all levels of the signal. After TFCE voxel clusters were deemed significant at p < 0.05. If no significant results were obtained using the permutation test, we performed correction for multiple comparisons using the family-wise-error correction based on Gaussian Random Field Theory (GRFT) [49] at cluster level (p < 0.05) combined with a primary uncorrected voxel-level threshold of p < 0.001. GRFT finds right threshold for a smooth statistical map which gives the required family-wise-error correction.
Regarding the FC displaying a significant education years-by-effect modifiers interaction, we examined the FC-by-effect modifiers interaction for the neuropsycho-logical performance scores by multiple regression analysis, adjusting for age, sex, education years, and effect modifier not included in the interaction evaluation. We applied a threshold of α = 0.05 to consider regression weights significant, and we additionally accounted for multiple testing using the Bonferroni correction for each hypothesis.
Table 1A illustrates the demographic characteristics of cognitively sound older adults in the Aβ+ and Aβ− categories. The average age was 68.7 years for the Aβ− group and 70.1 years for the Aβ+ group. Females were predominantly more than males in both groups, with respective proportions of 67.8% and 61.3%. The average duration of education was noted as 13.7 years for the Aβ− group and 13.9 years for the Aβ+ group. There was no remarkable difference in the aspects of age, sex, and education yeasrs between the two groups. Moreover, the Aβ+ group had a noticeably higher percentage of APOE ε4 carriers at 41.9%, compared to 20% in the Aβ− group. The Global FMM SUVRPONS value, utilized to segregate the Aβ+ and Aβ− groups, was considerably higher in the Aβ+ group. Nevertheless, there were no significant variations in the scores of neuropsychological assessments between the two groups.
Table 1B delineates the demographic details for CN individuals, distinguishing between APOE ε4 carriers and non-carriers. The APOE ε4 non-carriers exhibited a mean age of 69.1 years, slightly older than the carriers who averaged 68.8 years. In both subsets, a higher percentage were females, making up 65.6% and 67.7% respectively. Education years averaged at 13.7 for non-carriers and 13.8 for APOE ε4 carriers, indicating no significant variations in age brackets, sex distribution, or education years between the groups. Moreover, a higher incidence of Aβ positivity was observed in APOE ε4 carriers, recording a significant 41.9% compared to a 20% rate in non-carriers. However, no significant differences were noted in the Global FMM SUVRPONS values or in the neuropsycho-logical battery scores between carriers and non-carriers.
For the rFC from the seed region of the CEN, we found a significant education years-by-APOE ε4 carrier status interaction, after adjusting for age, sex, and Aβ deposition in cognitively normal older adults (Fig. 1A, p < 0.05, corrected for multiple comparisons using permutation test). As expected from the visual inspection of the results shown in Figure 1B, this result can be attributed to APOE ε4 carrier displaying higher education years with higher FC of the CEN. Table 2 shows the anatomical locations and their corresponding MNI coordinates of brain regions showing a significant interaction for the FC from the seed region of CEN.
For the FC from the seed region of the DAN, we found a significant interaction of education years by APOE ε4 carrier status, after adjusting for age, sex, and Aβ deposition in cognitively normal older adults (Fig. 1A, GRFT correction at a p value of < 0.05, voxel p < 0.001). As shown in Figure 1B, higher education years were associated with higher FC from the seed region of the DAN in the APOE ε4 carriers of cognitively normal older adults. Table 2 shows the anatomical locations and their corresponding MNI coordinates of brain regions showing a significant interaction for the FC from the seed region of DAN.
There was not an education years-by-APOE ε4 carrier status interaction for the FC from the seed regions of other networks including DMN and SN. Regarding the interaction of education years-by-Aβ deposition, there was no significant interaction for the FC of predefined resting- state brain networks.
For the FC between the regions of interest and seed regions of CEN and DAN, we evaluated a distinctive association with neuropsychological performance scores modu-lated by the effect modifier (APOE ε4 carrier status). Regarding the FC displaying a significant education years-by-APOE ε4 carrier status interaction, we found a significantly differential relationship with the CERAD-K TM and Total scores according to APOE ε4 carrier status (Fig. 1C). Specifically, there was a significant interaction of the FC between right superior occipital gyrus and seed region of CEN by APOE ε4 carrier status for the CERAD-K TM and Total scores (Fig. 1C, Bonferroni-corrected p < 0.05). This result can be attributed to APOE ε4 carrier showing higher FC of the CEN with lower CERAD-K TM and total scores in cognitively normal older adults.
The current study aimed to evaluate how whether the relationship between education years and rFC varied according to Aβ deposition and APOE ε4 carrier status in the preclinical phase of AD. In this study, among the AD risk factors, only the APOE ε4 carrier status showed a significant interaction with the years of education in relation to the rFC of the CEN and the DAN. Moreover, we found a significant interaction of rFC between right superior occipital gyrus and the CEN seed region by APOE ε4 carrier status for memory performances and overall cognitive function.
Regarding the CEN, this study found a significant interaction between education years and APOE ε4 carrier status for the rFC from the seed region of the CEN. Additio-nally, this interaction contributed to the higher rFC from the CEN seed region with the higher education years in the APOE ε4 carriers of the cognitively intact older adults. Regions of interest that showed significant FC with the seed region of CEN, the right dorsolateral prefrontal cortex, were the superior and middle occipital gyri and the superior temporal gyrus. It has been reported that FC between the CEN and other networks increases in older adults compared with younger adults [50]. Therefore, it is conceivable that the neural network changes observed in older adults may be more pronounced in individuals with high-risk genes for sporadic AD and high cognitive reserve proxy. While previous studies have shown that CEN integration is positively correlated with memory performance [50], the present study showed a negative correlation between rFC from the seed region of the CEN and memory function in the APOE ε4 carriers. These findings should be considered in relation to the fact that Aβ deposition tends to occur at an earlier age and at a faster pace in individuals with the APOE ε4 gene variant [51,52]. In our study, the proportion of APOE ε4 carriers was more than twice as high in the Aβ+ group compared to the Aβ− group. Given this context, it is possible that the impact of higher levels of education in raising the threshold for AD pathology is more pronounced among individuals with the APOE ε4 carriers. In this regard, although the APOE ε4 allele and its induced Aβ accumulation contribute to brain activation in the early phase of AD [53,54], the debilitating effect appears to be more pronounced in states of high cognitive reserve, rather than acting as a compensatory mechanism. Among the regions of interest that showed a significant FC with the CEN seed region, brain activation in the superior occipital gyrus of the cognitive control network was observed in both cognitively intact controls and AD patients, but only in AD patients was a significant compensatory effect of activity identified [55]. In addition, functional activation of the superior temporal gyrus in mild AD patients showed a positive association with cognitive reserve, suggesting that it is driven by an active compensatory mechanism. However, in the cognitively preserved controls, reduced functional activity was observed in those with high cognitive reserve, presumably due to more efficient use of the brain network [36]. In previous studies, increased FC in the CEN was not clearly observed in normal cognition, or there was insufficient evidence to support whether the increased FC was caused by compensatory mechanisms. Given our findings, it is worth considering the possibility of hyperactivation due to Aβ-induced neurotoxicity [56] rather than a compensatory response in the cognitively normal group with higher years of education.
Regarding the DAN, we also found a significant interaction between education years and APOE ε4 carrier status for the FC from the seed region of the DAN. Similar with the CEN, this interaction contributed to the higher FC from the DAN seed region with the higher education years in the APOE ε4 carriers of the cognitively intact older adults. Among the regions of interest from the seed region of the DAN, the left middle frontal gyrus is one of the nodes consisting of the CEN, while the left inferior parietal gyrus makes up the DAN [18]. Therefore, the results of this study could be described as higher DAN-CEN inter-network and DAN intra-network FC in APOE ε4 carriers with higher education years. Considering previous findings that DAN-CEN inter-network FC and DAN intra-network FC did not change significantly during the progression from normal cognition to MCI [18], we may assume that high cognitive reserve in AD vulnerable groups induces hyperactivation of inter- and intra- networks. As previously stated, it is possible that the higher years of education raised the threshold for AD pathology [57]. This could have induced a more intense AD pathogenesis, which in turn caused neurotoxicity resulting in brain functional hyperactivation. Conversely, another research has suggested that functional hyperactivation may be part of a compensatory mechanism [58], therefore long-term follow-up studies are needed to characterize the nature of the hyperactivation. In this study, the interaction between rFC from DAN seed region and AD risk factors did not have a significant effect on cognitive function. Additionally, although DMN and SN are also brain networks vulnerable to AD [24], there was no significant interaction between years of education and AD risk factors for these networks. It is worth considering that the relatively small variation in rFC in DMN and SN compared to other brain networks in the preclinical phase may have influenced these results [25]. In this regard, further longitudinal studies with actual changes in DMN and SN as outcome measures are needed to address these negative findings.
One interesting discovery from this study was that the association between years of education and AD-specific neural network connectivity during the preclinical phase was affected by the APOE ε4 carrier status, but not by the presence of Aβ deposition. In the present study, the average age of subjects was approximately 70 years old. At this age, the Aβ deposition reaches a plateau, resulting in insignificant changes [59]. However, the APOE ε4 carrier status exhibit significant differences in the rate and intensity of Aβ accumulation from middle age, compared to non-carriers [59]. It appears that being an APOE ε4 carrier may have a greater impact on long-term outcomes as an effect modifier, as opposed to the accumulation of Aβ. Additionally, APOE ε4 allele affect tau and TDP-43 pathology, as well as Aβ deposition, which may impact neurological inflammatory responses, blood-brain barrier integrity, and synaptic function [60]. These factors could potentially mediate the association between years of education and rFC of AD-vulnerable neural networks.
The first of the limitations in the current study is the relatively small sample size, which may have jeopardized its statistical strength. Furthermore, a skewed sample size can potentially impair the external reliability of the conclusions, limiting their generalizability to a more extensive population group. As a result, it is recommended to pursue additional studies with augmented sample sizes, particularly increasing the number of participants in the Aβ+ category. Another limitation of this study is that we only used years of education as a proxy for cognitive reserve. It is important to note that years of education only reflects the number of years of formal education during school age, and there may be differences in the quantity and quality of education experienced by individuals. Therefore, it is necessary to conduct additional studies using proxies for cognitive reserve other than years of education that reflect a combination of occupational history, lifestyle, and hobbies during adulthood and old age [61,62]. Additionally, a significant drawback of the present research is its cross-sectional framework. A more accurate analysis could potentially be realized if key elements central to cognitive and brain reserve—encompassing aspects like brain deterioration, functionality, pathology, and alterations in cognition—were examined over a longer period [63]. Consequently, gathering data over extended durations will be vital in enhancing our comprehension of the elements that influence the advancement of AD, and the interconnected dynamics between them. Finally, given that rFC varies by age even in the preclinical phase [26], it is important to investigate whether the association between years of education and rFC varies significantly by AD risk factors across a wider range of ages.
This study set out to explore the differential effects of sporadic AD risk factors on the associations between education years and rFC of AD-vulnerable neural networks during the preclinical phase of AD. We found a significant interaction contributing to the higher rFC from the CEN and DAN seed region with the higher education years in the APOE ε4 carriers. However, higher rFC from the CEN seed region had a debilitating rather than compensatory effect on cognitive function. These results contribute to a deeper understanding of the impact of cognitive reserve on sensitive intermediate phenotypic markers in the preclinical phase of AD. Additionally, the impact of cognitive reserve on AD progression needs to be further investigated across a wider range of ages and stages of AD progression, using proxies that can capture various aspects of cognitive reserve.
Author Hyun Kook Lim, Donghyeon Kim, Regina EY Kim, Yeong Sim Choe and Jiyeon Lee are employed by NEUROPHET Inc. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The data processing services provided by NEUROPHET Inc. were utilized to enhance the quality and analysis of the brain imaging data collected during the study. Authors declare that the research outcomes and conclusions remain unbiased and are not influenced by any commercial interests associated with the NEUROPHET Inc.’s products or services.
Conceptualization: Jiwon Kim, Hyun Kook Lim, Chang Uk Lee, Dong Woo Kang. Methodology: Jiwon Kim, Sheng-Min Wang, Regina EY Kim, Yeong Sim Choe, Jiyeon Lee, Donghyeon Kim, Dong Woo Kang. Software: Yoo Hyun Um. Investigation: Yoo Hyun Um. Data curation: Jiwon Kim, Sheng-Min Wang, Regina EY Kim, Yeong Sim Choe, Jiyeon Lee, Donghyeon Kim. Visualization: Jiwon Kim, Sunghwan Kim. Formal analysis: Jiwon Kim, Sunghwan Kim. Writing—original draft: Jiwon Kim. Writing—review & editing: Sheng-Min Wang, Dong Woo Kang. Supervision: Hyun Kook Lim, Chang Uk Lee, Dong Woo Kang. Project Administration: Dong Woo Kang.
Fig. 1. (A) Brain areas with a significant impact of an interaction between education years and 
CrossRef (1) Scopus (1) Download (349)
Author ORCID Information Jiwon Kim Sunghwan Kim Yoo Hyun Um Sheng-Min Wang Regina EY Kim Yeong Sim Choe Jiyeon Lee Donghyeon Kim Hyun Kook Lim Chang Uk Lee Dong Woo Kang
Funding Information Korea Health Industry Development Institute (KHIDI) Korea Dementia Research Center (KDRC) Ministry of Health & Welfare Ministry of Science and ICT, Republic of Korea |
 |
