• KSBNS 2024


Original Article

Exp Neurobiol 2024; 33(2): 107-117

Published online April 30, 2024

© The Korean Society for Brain and Neural Sciences

Alterations in Brain Morphometric Networks and Their Relationship with Memory Dysfunction in Patients with Type 2 Diabetes Mellitus

Rye Young Kim1, Yoonji Joo1, Eunji Ha1, Haejin Hong1, Chaewon Suh1, Youngeun Shim1,2, Hyeonji Lee1,2, Yejin Kim1,2, Jae-Hyoung Cho3, Sujung Yoon1,2* and In Kyoon Lyoo1,2,4*

1Ewha Brain Institute, Ewha Womans University, Seoul 03760, 2Department of Brain and Cognitive Sciences, Ewha Womans University, Seoul 03760, 3Department of Internal Medicine, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul 06591, 4Graduate School of Pharmaceutical Sciences, Ewha Womans University, Seoul 03760, Korea

Correspondence to: *To whom correspondence should be addressed.
Sujung Yoon, TEL: 82-2-3277-2478, FAX: 82-2-3277-6562
In Kyoon Lyoo, TEL: 82-2-3277-6554, FAX: 82-2-3277-6562

Received: March 4, 2024; Revised: March 31, 2024; Accepted: April 9, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Cognitive dysfunction, a significant complication of type 2 diabetes mellitus (T2DM), can potentially manifest even from the early stages of the disease. Despite evidence of global brain atrophy and related cognitive dysfunction in early-stage T2DM patients, specific regions vulnerable to these changes have not yet been identified. The study enrolled patients with T2DM of less than five years’ duration and without chronic complications (T2DM group, n=100) and demographically similar healthy controls (control group, n=50). High-resolution T1-weighted magnetic resonance imaging data were subjected to independent component analysis to identify structurally significant components indicative of morphometric networks. Within these networks, the groups’ gray matter volumes were compared, and distinctions in memory performance were assessed. In the T2DM group, the relationship between changes in gray matter volume within these networks and declines in memory performance was examined. Among the identified morphometric networks, the T2DM group exhibited reduced gray matter volumes in both the precuneus (Bonferroni-corrected p=0.003) and insular-opercular (Bonferroni-corrected p=0.024) networks relative to the control group. Patients with T2DM demonstrated significantly lower memory performance than the control group (p=0.001). In the T2DM group, reductions in gray matter volume in both the precuneus (r=0.316, p=0.001) and insular-opercular (r=0.199, p=0.047) networks were correlated with diminished memory performance. Our findings indicate that structural alterations in the precuneus and insular-opercular networks, along with memory dysfunction, can manifest within the first 5 years following a diagnosis of T2DM.

Keywords: Type 2 diabetes mellitus, Cognitive dysfunction, Morphology, Neuroimaging

Type 2 diabetes mellitus (T2DM) is rapidly emerging as a global health crisis due to its increasing prevalence and associated severe complications [1]. Recent advancements in medical care and the extension of life expectancy in patients with T2DM have broadened the understanding of diabetic complications, extending beyond the conventional micro- and macrovascular complications to encompass emerging issues such as cognitive dysfunction [2-4]. While most existing research has focused on detrimental changes in brain and cognitive functions in chronic T2DM, particularly those with vascular complications [5, 6], recent studies on middle-aged patients with a disease duration less than 10 years indicated that changes in the brain associated with T2DM can occur in the early stages of T2DM [7-9]. Furthermore, even subtle alterations in brain structure may contribute to cognitive decline in these patients [7-9].

Global brain atrophy serves as an essential neuroimaging marker, with the degree of atrophy possibly reflecting the severity and progression of brain parenchymal damage among T2DM patients [10-12]. Clinically, this widespread atrophy correlates with memory impairment observed in individuals with T2DM [5, 6]. However, our understanding is still limited regarding the specific brain regions most susceptible in the initial stages of T2DM and how these changes contribute to memory dysfunction [3]. Earlier investigations utilizing voxel-based morphometry (VBM) and volumetric analysis have identified various affected brain regions in T2DM, including the frontal lobes [13, 14], temporal lobes [5, 14-16], hippocampus [8, 13, 15], precuneus [14], and insula [14]. Nevertheless, inconsistencies across studies have arisen, potentially due to differences in clinical characteristics, especially pertaining to the duration of the disease among the study populations. Moreover, the precise relationship between regional brain changes in T2DM and their consequential effects on memory function remains elusive [5, 6, 10].

Recent advancements in neuroscience have enabled the leveraging of methodologies that consider inter-regional dependencies within analyses, resulting in the application of multivariate network-based analyses. These advanced methods have been successful in detecting subtle changes in brain structure across various neurodegenerative diseases [17, 18]. Recognizing that brain regions are interconnected both structurally and functionally, disease-related atrophy often appears within networks of highly correlated regions rather than a single isolated anatomical structure [17, 18]. Imaging the networks of brain regions with structural correlations can provide insights into disease-specific pathologies at the network level [17-24]. Moreover, the multivariate network-based analysis employs a model-free approach, negating the need for pre-selected regions [22, 25]. This particular methodological characteristic provides a unique advantage in identifying early and nuanced changes in brain structure, especially when the information about disease-specific vulnerable brain regions is unclear or undefined [17-19].

In this study, our objective was to explore the structural brain changes specific to T2DM and evaluate how these alterations might influence memory function. To gain deeper insights into regional deficits at the network level, we utilized a data-driven approach, employing independent component analysis (ICA) to extract significant structural networks based on the structural covariance of gray matter volumes [19-24]. By concentrating on the early stage of T2DM, we included middle-aged patients who had been diagnosed with the disease for less than five years and had no chronic complications. In addition to investigating specific brain alterations, we assessed memory dysfunction in these patients, examining whether the T2DM-related structural changes at the network level were correlated with declines in memory.


The study comprised 100 individuals with physician-diagnosed T2DM (constituting the T2DM group), and fifty age- and sex-matched healthy individuals (forming the control group). The mean age of the T2DM group was 49.2 years with a standard deviation (SD) of 7.7, while the control group’s mean age was 49.0 years with an SD of 7.8. The participants were derived from previously published studies [26, 27]. In the T2DM group, all patients had a diagnosis duration of less than five years, with a mean duration of 22.0 months and an SD of 18.4 months.

Healthy participants were ascertained to be free of T2DM, confirmed by fasting plasma glucose levels below 5.55 mmol/L and two-hour postprandial plasma glucose levels under 7.77 mmol/L [28]. Their glycated hemoglobin A1c (HbA1c) levels were additionally found to be less than 5.7% (39 mmol/mol) [28].

Exclusion criteria encompassed major medical, neurological, or psychiatric disorders, and contraindications to magnetic resonance imaging (MRI). Additionally, T2DM patients with chronic diabetic complications such as micro- or macrovascular issues including diabetic nephropathy, symptomatic diabetic neuropathy, proliferative diabetic retinopathy, cerebrovascular disorders, or cardiovascular disorders were excluded. Detailed demographic and clinical characteristics of the participants are presented in Table 1.

The research protocol received approval from the institutional review board at the College of Medicine, Catholic University of Korea. All participants provided written informed consent prior to their involvement in the study.

Image acquisition and preprocessing

Brain MRI scans were performed using a 1.5 Tesla whole-body imaging system (Signa HDx, GE Healthcare, Milwaukee, WI, USA). High-resolution T1-weighted structural images were obtained for all individuals by employing a three-dimensional spoiled gradient echo sequence. The following acquisition parameters were used: repetition time=24 ms, echo time=5 ms, field of view=24 cm, matrix=256×256 mm2, flip angle=45°, number of excitations=2, slice thickness=1.2 mm, with no interslice gap.

The T1-weighted structural images were processed using Statistical Parametric Mapping 12 (, executed on MATLAB version 2017b (MathWorks, Natick, MA, USA). Initially, T1-weighted images from all participants were segmented into gray matter, white matter, and cerebrospinal fluid (CSF) components. This segmentation employed default settings, apart from affine regularization, for which the East Asian Brains International Consortium for Brain Mapping space template was selected.

Subsequently, gray matter images were aligned using the Diffeomorphic Anatomical Registration through Exponentiated Lie Algebra (DARTEL) [29]. This iterative method generated a study-specific gray matter template, which was then spatially registered to the Montreal Neurological Institute (MNI) standard template image. The individual gray matter images were non-linearly mapped to this template and then resampled to a uniform voxel size of 2×2×2 mm3. Any local expansions or contractions arising from this non-linear registration step were corrected through modulation. The final modulated gray matter images were smoothed with an 8 mm3 isotropic Gaussian kernel. Visual inspections were incorporated at each processing stage to ensure the quality and accuracy of the data.

Independent component analysis on gray matter maps

In order to perform a data-driven analysis independent of predefined regions of interest (ROIs), ICA was employed on the gray matter maps of all participants. ICA is a statistical method that separates multivariate data into spatial components that display maximal statistical independence [30]. Applied to structural gray matter images across individuals, ICA produces spatially independent component (IC) maps based on the covariance among gray matter volumes. For this purpose, a four-dimensional (4D) dataset was required to serve as the input for ICA. The modulated and smoothed gray matter images of all participants were concatenated to form this 4D dataset. The gray matter maps were subsequently masked with a maximum probability gray matter mask. This mask included only voxels with the highest likelihood of being gray matter and was specifically tailored to target the cerebral cortex rather than the cerebellar cortex, thereby excluding cerebellar regions in the mask creation. The ICA was carried out using the Multivariate Exploratory Linear Optimized Decomposition into Independent Component (MELODIC) toolbox ( within the FSL analysis package (version 6.0.4) [30].

In alignment with prior research [21, 23], the ICA output was configured to 10 components. Each component was subject to thresholding at z=3.0, preserving only the voxels contributing significantly to the component. These thresholded components were recognized as “morphometric networks” in the study [19]. The anatomical locations of each morphometric network were identified using the Harvard-Oxford structural atlas within the FSL analysis package. The mean gray matter volumes were then extracted from each of the 10 morphometric networks for every individual and were employed in subsequent analyses.

Memory performance assessments

Memory performance was evaluated in all participants using the Rey-Osterrieth complex figure test (ROCF) and California Verbal Learning Test (CVLT). The ROCF, a commonly used neuropsychological test, examines the integrity of visual memory and visuospatial constructional ability [31, 32]. The CVLT serves as another widely accepted neuropsychological assessment, measuring verbal memory and episodic learning [32, 33].

The outcome variables for the ROCF included the scores assessing the precision of drawing during both immediate recall (immediate recall score) and delayed recall (delayed recall score), as well as the scores for correct recognition of elements within the figure (retention score). For the CVLT, the outcome variables consisted of the total number of words recalled in the first five trials for immediate memory, a short-delay free recall score for short-term memory, and a long-delay free recall score for long-term memory. These outcome variables were standardized to z scores using the mean and standard deviation of the control group. The six z scores were subsequently averaged to form a composite score, representing overall memory performance. A higher composite score denotes better performance, and disparities in memory performance between the groups were analyzed by comparing this composite score.

Statistical analysis

The demographic and clinical characteristics between the T2DM and control groups were compared using an independent t-test for continuous variables and a chi-square test for categorical variables, as appropriate.

Differences in the mean gray matter volume of each morphometric network between the T2DM and control groups were analyzed using multiple linear regression, considering age and sex as covariates. To control for multiple comparisons, a Bonferroni correction was applied, rendering a corrected p value of less than 0.05 statistically significant. Specifically, Bonferroni-corrected p values were calculated by multiplying each individual p value by the total number of tests performed (n=10).

Memory performance between the T2DM and control groups was evaluated through multiple linear regression analysis, with adjustments made for age, sex, and educational level.

The associations between the mean gray matter volumes in each morphometric network and memory performance were investigated using Pearson correlation analysis. Furthermore, the relationship between the mean gray matter volumes in each morphometric network and memory performance was examined using partial correlation analysis after adjusting for disease duration in patients with diabetes.

We examined potential metabolic factors contributing to T2DM-related reductions in gray matter volume in specific networks. To accomplish this, Pearson correlation analyses were applied to assess the relationship between metabolic measures and gray matter volumes in networks showing T2DM-related volume reductions. These metabolic measures included hemoglobin A1C (HbA1c), a biomarker of hyperglycemia, and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR), a biomarker of insulin resistance.

A two-tailed p value of less than 0.05 was considered significant for all analyses. Data were analyzed using Stata SE v16.1 (StataCorp LP, College Station, Texas).

Morphometric network alterations in patients with T2DM

By applying ICA to gray matter maps, ten morphometric networks were identified based on the structural covariance of gray matter across all participants (Fig. 1). Notably, the precuneus and insular-opercular networks (Fig. 2) exhibited significant between-group differences following Bonferroni correction. The T2DM group demonstrated reduced gray matter volumes in both these networks in comparison to controls (Fig. 2): precuneus (β=-0.280, Bonferroni-corrected p=0.003) and insular-opercular (β=-0.209, Bonferroni-corrected p=0.024).

In the sensitivity analysis, consistent with previous results, the T2DM group showed smaller gray matter volumes in the insular-opercular network (β=-0.202, Bonferroni-corrected p=0.003) and the second precuneus network (β=-0.203, Bonferroni-corrected p=0.007) when compared to the control group. Detailed information regarding the sensitivity analysis is provided in Supplementary Materials (Supplementary Fig. 1, Supplementary Table 1).

In contrast, between-group differences in other morphometric networks were not statistically significant after Bonferroni corrections. Detailed statistical information regarding the group comparisons of these morphometric networks is provided in Table 2.

As a sensitivity analysis, we derived network masks from the ICA applied to the gray matter maps of the T2DM group. This analysis revealed a similar pattern in the morphometric networks (Supplementary Fig. 1 for details). When comparing the gray matter volumes between the groups, specifically for the precuneus network (β=-0.203, Bonferroni-corrected p=0.071) and insular-opercular network (β=-0.202, Bonferroni-corrected p=0.032), the results remained consistent, as detailed in Supplementary Table 1.

Memory dysfunction in patients with T2DM

Utilizing a global composite score derived from the ROCF and CVLT, the T2DM group was found to have significantly lower memory performance compared to the control group (β=-0.246, p=0.001; Fig. 3).

Associations between morphometric network alterations and memory dysfunction in patients with T2DM

We explored potential associations between gray matter volume reductions in the precuneus and insular-opercular networks, particularly concerning T2DM, and memory dysfunction (Fig. 4). Within the T2DM group, smaller gray matter volumes in the precuneus network were correlated with reduced memory performance (r=0.316, p=0.001). A comparable positive correlation was identified between the gray matter volumes of the insular-opercular network and memory performance (r=0.199, p=0.047).

As an additional analysis, we examined the correlation between mean gray matter volumes and memory performance in the T2DM group, controlling for disease duration as a covariate. While the significance levels were altered upon adjusting for disease duration, the overall results remained consistent (precuneus network: r=0.310, p=0.002; insular-opercular network: r=0.191, p=0.058).

In contrast, the control group did not exhibit any significant association between memory performance and gray matter volumes within these two morphometric networks (r=0.095, p=0.512 for the precuneus network; r=0.134, p=0.352 for the insular-opercular network).

Associations between metabolic measures and morphometric network alterations in patients with T2DM

In the T2DM groups, we found no significant associations between HbA1c levels and gray matter volumes in either the precuneus network (r=0.039, p=0.697) or insular-opercular network (r=-0.013, p=0.896) (Supplementary Fig. 2). Similarly, HOMA-IR levels showed no significant correlation with gray matter volumes in both the the precuneus network (r=0.103, p=0.316) and insular-opercular network (r=-0.020, p=0.838) (Supplementary Fig. 2).

In the present study, we sought to explore T2DM-induced structural changes in the brain at the network level. The investigation focused on morphometric networks, which are derived from the structural covariance of gray matter volumes. Each network’s components represent brain regions that are both structurally and functionally interconnected [17, 18]. We discovered that patients with T2DM showed reductions in gray matter volume within the precuneus and insular-opercular networks, contrasting with the control group. In addition, the T2DM group’s memory performance was not only inferior to that of the control group but also correlated with the observed reductions in gray matter volumes within the aforementioned networks.

Our observations are consistent with previous meta-analysis of VBM studies on T2DM [14], which demonstrated structural alterations in the precuneus and insula regions of T2DM patients. Conversely, changes in the frontal and temporal regions, previously reported in connection with T2DM [13, 15], were not found to be significant in our sample after adjusting for multiple comparisons. Considering that our patient cohort was in the early stage of T2DM, having been diagnosed for less than five years and without severe complications, we cautiously suggest that the precuneus and insular regions might be affected earlier in the course of T2DM in comparison to other known vulnerable brain regions.

Our study focused on investigating the reductions in gray matter volume associated with T2DM within morphometric networks. These networks may underline the complex structural relationships among highly interconnected regions of the brain. Consequently, the insights obtained from examining the entire morphometric network, rather than isolated individual brain regions, provide a more complete understanding of disease-specific changes by assimilating information at the network level [17, 18]. Several studies have already exhibited the enhanced sensitivity of data-driven network analysis for detecting disease-specific brain changes, surpassing the conventional voxel-based method [20-22]. Given the robust association between the decline in gray matter volume within these networks and memory dysfunction, the reductions in gray matter volume in the precuneus and insular-opercular networks appear to be the neural correlates underpinning cognitive dysfunction in T2DM patients. As a part of the sensitivity analysis, we also employed ICA on the gray matter maps of the T2DM group. The sensitivity analysis showed similar results, but the group difference was a lesser degree.

The precuneus, a major constituent of the default mode network (DMN), collaborates with other brain regions to fulfill essential roles in memory retrieval, information integration, and self-awareness [34]. Previous studies have repeatedly identified structural and functional abnormalities in the DMN, including the precuneus, in connection with T2DM [35]. Moreover, an arterial spin labeling MRI study recorded that hypoperfusion in the precuneus corresponds with cognitive dysfunction in T2DM patients [36]. Intriguingly, the precuneus is recognized as one of the earliest affected brain regions in Alzheimer’s dementia, exhibiting changes up to a decade before the manifestation of clinical symptoms [37-39]. Furthermore, the precuneus is emphasized as a vital region contributing to the potential connection between T2DM and Alzheimer’s disease [35, 39]. Consistent with these insights, our correlation analysis supports the hypotheses that T2DM-related structural modifications in the precuneus contribute to memory impairment, even in the disease’s nascent stages.

The insular region is often implicated in T2DM, a connection supported by numerous studies [12, 13, 40]. Our findings suggest that insular deficits might emerge during the early stages of T2DM. Understanding the insula’s essential role in interoception —sensing the body’s internal state [41, 42]—it is evident that the insula is vital for detecting and regulating glucose levels [43, 44]. Consequently, insular deficits could be either a contributing factor to, or a result of, the metabolic disturbances tied to T2DM [43, 44].

Furthermore, our observations concerning the link between insular deficits and memory dysfunction are consistent with recent findings. A particular study emphasized that impaired neurovascular coupling, especially in the insular region, played a substantial role in the memory decline observed in T2DM patients [40]. Previous research has documented T2DM-associated decreases in visuospatial and verbal memory [12], but it is notable that middle-aged T2DM patients, even with a disease duration of less than five years, showed perceptible memory decline in comparison to healthy controls. Historically, most investigations into T2DM’s effects on the brain or cognition have predominantly targeted the elderly [45, 46]. Given that clinically significant cognitive impairments like mild cognitive impairment and dementia typically manifest in individuals 65 years or older, there was a prevailing belief that T2DM’s harmful effects on brain and cognitive function were limited to this age group [47]. However, an evolving body of evidence contradicts this view, revealing that T2DM’s impact on both brain [5, 6] and cognitive function [48-50] also reaches the pre-elderly population. For instance, brain changes attributable to T2DM have been observed not only in older adults but also in the middle-aged population [7-9]. Moreover, longitudinal studies have shown that middle-aged adults with T2DM undergo a more rapid cognitive decline over spans of 10 [48] and 20 years [49] compared to their counterparts without the condition. In alignment with these studies, our findings reinforce the idea that T2DM-induced cognitive alterations might appear earlier in life, even among middle-aged individuals.

The observation of accelerated memory decline in patients with T2DM carries profound clinical implications. This dysfunction hampers an individual’s ability to perform critical self-management tasks, such as glycemic monitoring, adherence to treatment protocols, routine clinic visits, and the coordination of meal timing and content [4, 28]. The ineffective self-management of T2DM is more than concerning; it may result in grave outcomes, including an elevated risk of hypoglycemic and cardiovascular incidents. Such outcomes could further manifest as frequent complications related to T2DM, additional hospital admissions, and even fatality [4, 28]. By identifying specific brain regions exhibiting T2DM-related abnormalities tied to memory dysfunction in middle-aged subjects, our study reveals potential therapeutic targets. This understanding may facilitate the development of strategies to prevent memory deterioration and related complications in patients with T2DM.

Several limitations warrant mention in the interpretation of our findings. Our study employed a cross-sectional design and, consequently, does not shed light on the temporal changes in brain and cognitive indicators in a long-term manner. For example, given the established role of the insula in metabolic sensing [42-44], it is plausible that pre-existing dysfunction within the insular network could influence both the onset and progression of T2DM. To address some of these limitations, we conducted an additional analysis to examine the correlation between cognitive dysfunction and the networks, taking into account the time after the diagnosis of diabetes as a variable. As a result, the findings in the precuneus network were consistent with those of the primary analysis. However, in the insular-opercular network, although the trends were similar, the correlation was not statistically significant.

Our investigation specifically addressed changes in brain function and cognition during the initial phases of T2DM. Consequently, the current sample was confined to patients with disease durations of less than five years and without significant diabetes-related complications. However, since micro- and macrovascular complications such as diabetic retinopathy, stroke, and myocardial infarction have the potential to worsen cognitive decline and heighten dementia risk [2, 3], it is vital that subsequent studies thoroughly explore their effects on brain function and cognition.

It is important to note that metabolic measures, including HbA1c levels and HOMA-IR, were not linked to reductions in gray matter volume among our T2DM patients. While HbA1c levels reflect glycemic control over the previous three months, and insulin resistance is a key aspect of T2DM pathophysiology, a more robust marker for long-term glycemic control, such as lifetime cumulative glycemic control levels [50], might be required to elucidate the exact metabolic mechanisms underlying T2DM-related brain alterations. This suggestion stems from the understanding that the brain’s susceptibility to diabetes-related alterations could result from the cumulative impact of the condition over an extended period, which current HbA1c levels may not adequately represent [51].

The intricate nature of this relationship highlights the necessity for further investigations based on a longitudinal observation. Such studies may clarify the mechanisms, accompanying structural modifications in the precuneus and insular-opercular regions, that could induce cognitive declines in patients diagnosed with T2DM.

In this study, our emphasis was specifically placed on memory function as the primary cognitive domain of interest, a focus motivated by the fact that memory is one of the earliest cognitive domains impacted by T2DM [33, 34]. It is imperative, however, to acknowledge that cognitive dysfunction associated with T2DM may encompass various domains, including executive function, processing speed, and motor function, particularly as the disease advances [4, 52]. To gain a deeper understanding of how structural changes in the precuneus and insular-opercular networks influence cognitive function in T2DM, future studies should include a broader assessment of cognitive domains.

In conclusion, our study has identified significant reductions in gray matter volume at the brain network level and elucidated their correlations with memory dysfunction in middle-aged T2DM patients. Specifically, the brain regions within the precuneus and insular-opercular networks are particularly vulnerable to T2DM and may play a complex role in diabetes-related cognitive dysfunction. These findings are clinically significant, indicating that such changes can begin in the early stages of disease, with cognitive and brain alterations potentially appearing in less than 5 years following a T2DM diagnosis. This underscores the critical need for earlier detection and specialized interventions to address T2DM-related brain and cognitive changes.

This research was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (2020M3E5D9080555), grant funded by the Ministry of Education (2020R1A6A1A03043528), Creative & Basic Technology Research Project funded by the Electronics and Telecommunications Research Institute (Grant No. 23YS1110), and the Health Fellowship Foundation.

Fig. 1. An overview of the ten distinct morphometric networks. Ten networks were identified, reflecting the structural covariance of gray matter across all participants. There include: (A) Frontal-ACC network, (B) Insular-opercular network, (C) Temporal-limbic network, (D) Basal ganglia-insular network, (E) Occipital-precuneus network, (F) Occipital network, (G) Right lateral occipital-temporal network, (H) Temporal-ACC network, (I) Precuneus network, and (J) Thalamus-calcarine network. Details of Montreal Neurological Institute (MNI) coordinates are provided for the axial, coronal, and sagittal slices of each network. ACC, anterior cingulate cortex; L, left; R, right; T2DM, Type 2 diabetes mellitus.
Fig. 2. Comparison of gray matter volumes within two morphometric networks between the T2DM and control groups. (A) Precuneus network. (B) Insular-opercular network. GM, gray matter; L, left; R, right; T2DM, type 2 diabetes mellitus.
Fig. 3. Group difference in memory performance between the T2DM and control groups. T2DM, type 2 diabetes mellitus.
Fig. 4. Relationships between gray matter volumes within the morphometric networks and memory performance in the T2DM group. Relationships between mean GM volume in the morphometric networks and memory composite score in the T2DM group were analyzed using Pearson correlation analysis. Thicker lines represent regression lines, while thinner lines indicate the 95% confidence intervals. GM, gray matter; T2DM, type 2 diabetes mellitus.
Table. 1.

Demographic and clinical characteristics of study participants

CharacteristicsControl (n=50)T2DM (n=100)p value
Demographic characteristics
Age, mean (SD), years49.0 (7.8)49.2 (7.7)0.93
Women, n (%)25 (50.0)50 (50.0)1.00
Right handedness, n (%)48 (96.0)95 (95.0)0.78
Lower education level, n (%)a31 (62.0)77 (77.0)0.05
Clinical characteristics
Time since T2DM diagnosis, mean (SD), months-22.0 (18.4)-
HbA1c, mean (SD), %5.29 (0.15)7.12 (1.43)< 0.001
Fasting plasma glucose, mean (SD), mmol/L5.24 (0.21)7.67 (2.24)< 0.001
BMI, mean (SD), kg/m222.7 (1.8)25.5 (3.4)< 0.001
Systolic blood pressure, mean (SD), mmHg120.5 (9.9)126.1 (11.6)0.004
Diastolic blood pressure, mean (SD), mmHg75.1 (8.5)76.5 (8.7)0.35
Total cholesterol, mean (SD), mmol/L4.87 (0.93)4.77 (0.97)0.54

aHigh school graduate or lower.

BMI, body-mass index; HbA1c, hemoglobin A1c; SD, standard deviation; T2DM, type 2 diabetes mellitus.

Table. 2.

Group comparisons of gray matter volume within the morphometric networks

Morphometric networksMean gray matter volumes of the
morphometric networks
p value
p value
Control (n=50)T2DM (n=100)
[Network A] Frontal-ACC0.465±0.0560.448±0.0470.1570.016
[Network B] Insular-opercular0.474±0.0520.45±0.0510.0240.002
[Network C] Temporal-limbic0.515±0.0520.497±0.0440.1840.018
[Network D] Basal ganglia-insular0.487±0.0470.472±0.0410.1860.019
[Network E] Occipital-precuneus0.486±0.0640.478±0.0481.0000.369
[Network F] Occipital0.453±0.0570.444±0.0421.0000.206
[Network G] Right lateral occipital-temporal0.422±0.0480.412±0.0491.0000.162
[Network H] Temporal-ACC0.479±0.0520.461±0.0480.1650.016
[Network I] Precuneus0.497±0.0630.464±0.0480.003< 0.001
[Network J] Thalamus-calcarine0.448±0.0450.432±0.0460.2760.028

Mean gray matter volumes are expressed as mean±standard deviation. The mean gray matter volumes for each morphometric network were assessed between the T2DM and control groups utilizing multiple regression analysis, with age and sex included as covariates in the model. Bonferroni-corrected p value was calculated as each uncorrected p value was multiplied by the number of comparisons (n=10).

ACC, anterior cingulate cortex; T2DM, type 2 diabetes mellitus.

  1. Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, Stein C, Basit A, Chan JCN, Mbanya JC, Pavkov ME, Ramachandaran A, Wild SH, James S, Herman WH, Zhang P, Bommer C, Kuo S, Boyko EJ, Magliano DJ (2022) IDF diabetes atlas: global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract 183:109119
    Pubmed KoreaMed CrossRef
  2. Tomic D, Shaw JE, Magliano DJ (2022) The burden and risks of emerging complications of diabetes mellitus. Nat Rev Endocrinol 18:525-539
    Pubmed KoreaMed CrossRef
  3. Biessels GJ, Despa F (2018) Cognitive decline and dementia in diabetes mellitus: mechanisms and clinical implications. Nat Rev Endocrinol 14:591-604
    Pubmed KoreaMed CrossRef
  4. Srikanth V, Sinclair AJ, Hill-Briggs F, Moran C, Biessels GJ (2020) Type 2 diabetes and cognitive dysfunction-towards effective management of both comorbidities. Lancet Diabetes Endocrinol 8:535-545
    Pubmed CrossRef
  5. Karvani M, Simos P, Stavrakaki S, Kapoukranidou D (2019) Neurocognitive impairment in type 2 diabetes mellitus. Hormones (Athens) 18:523-534
    Pubmed CrossRef
  6. Roriz-Filho JS, Sá-Roriz TM, Rosset I, Camozzato AL, Santos AC, Chaves ML, Moriguti JC, Roriz-Cruz M (2009) (Pre)diabetes, brain aging, and cognition. Biochim Biophys Acta 1792:432-443
    Pubmed CrossRef
  7. Marder TJ, Flores VL, Bolo NR, Hoogenboom WS, Simonson DC, Jacobson AM, Foote SE, Shenton ME, Sperling RA, Musen G (2014) Task-induced brain activity patterns in type 2 diabetes: a potential biomarker for cognitive decline. Diabetes 63:3112-3119
    Pubmed KoreaMed CrossRef
  8. Gold SM, Dziobek I, Sweat V, Tirsi A, Rogers K, Bruehl H, Tsui W, Richardson S, Javier E, Convit A (2007) Hippocampal damage and memory impairments as possible early brain complications of type 2 diabetes. Diabetologia 50:711-719
    Pubmed CrossRef
  9. Manschot SM, Brands AM, van der Grond J, Kessels RP, Algra A, Kappelle LJ, Biessels GJ; Utrecht Diabetic Encephalopathy Study Group (2006) Brain magnetic resonance imaging correlates of impaired cognition in patients with type 2 diabetes. Diabetes 55:1106-1113
    Pubmed CrossRef
  10. Biessels GJ, Nobili F, Teunissen CE, Simó R, Scheltens P (2020) Understanding multifactorial brain changes in type 2 diabetes: a biomarker perspective. Lancet Neurol 19:699-710
    Pubmed CrossRef
  11. van Harten B, de Leeuw FE, Weinstein HC, Scheltens P, Biessels GJ (2006) Brain imaging in patients with diabetes: a systematic review. Diabetes Care 29:2539-2548
    Pubmed CrossRef
  12. Moran C, Phan TG, Chen J, Blizzard L, Beare R, Venn A, Münch G, Wood AG, Forbes J, Greenaway TM, Pearson S, Srikanth V (2013) Brain atrophy in type 2 diabetes: regional distribution and influence on cognition. Diabetes Care 36:4036-4042
    Pubmed KoreaMed CrossRef
  13. Moulton CD, Costafreda SG, Horton P, Ismail K, Fu CH (2015) Meta-analyses of structural regional cerebral effects in type 1 and type 2 diabetes. Brain Imaging Behav 9:651-662
    Pubmed CrossRef
  14. Liu J, Liu T, Wang W, Ma L, Ma X, Shi S, Gong Q, Wang M (2017) Reduced gray matter volume in patients with type 2 diabetes mellitus. Front Aging Neurosci 9:161
    Pubmed KoreaMed CrossRef
  15. Wu G, Lin L, Zhang Q, Wu J (2017) Brain gray matter changes in type 2 diabetes mellitus: a meta-analysis of whole-brain voxel-based morphometry study. J Diabetes Complications 31:1698-1703
    Pubmed CrossRef
  16. den Heijer T, Vermeer SE, van Dijk EJ, Prins ND, Koudstaal PJ, Hofman A, Breteler MM (2003) Type 2 diabetes and atrophy of medial temporal lobe structures on brain MRI. Diabetologia 46:1604-1610
    Pubmed CrossRef
  17. Alexander-Bloch A, Giedd JN, Bullmore E (2013) Imaging structural co-variance between human brain regions. Nat Rev Neurosci 14:322-336
    Pubmed KoreaMed CrossRef
  18. Seeley WW, Crawford RK, Zhou J, Miller BL, Greicius MD (2009) Neurodegenerative diseases target large-scale human brain networks. Neuron 62:42-52
    Pubmed KoreaMed CrossRef
  19. Pichet Binette A, Gonneaud J, Vogel JW, La Joie R, Rosa-Neto P, Collins DL, Poirier J, Breitner JCS, Villeneuve S, Vachon-Presseau E (2020) Morphometric network differences in ageing versus Alzheimer's disease dementia. Brain 143:635-649
    Pubmed KoreaMed CrossRef
  20. Hafkemeijer A, Möller C, Dopper EG, Jiskoot LC, van den Berg-Huysmans AA, van Swieten JC, van der Flier WM, Vrenken H, Pijnenburg YA, Barkhof F, Scheltens P, van der Grond J, Rombouts SA (2016) Differences in structural covariance brain networks between behavioral variant frontotemporal dementia and Alzheimer's disease. Hum Brain Mapp 37:978-988
    Pubmed KoreaMed CrossRef
  21. Coppen EM, van der Grond J, Hafkemeijer A, Rombouts SA, Roos RA (2016) Early grey matter changes in structural covariance networks in Huntington's disease. Neuroimage Clin 12:806-814
    Pubmed KoreaMed CrossRef
  22. Hafkemeijer A, Altmann-Schneider I, de Craen AJ, Slagboom PE, van der Grond J, Rombouts SA (2014) Associations between age and gray matter volume in anatomical brain networks in middle-aged to older adults. Aging Cell 13:1068-1074
    Pubmed KoreaMed CrossRef
  23. Lee S, Kim J, Kim BJ, Kim RY, Ha E, Kim S, Hong SN, Lyoo IK, Kim DW (2022) Gray matter volume reduction in the emotional brain networks in adults with anosmia. J Neurosci Res 100:1321-1330
    Pubmed CrossRef
  24. Joo Y, Lee S, Hwang J, Kim J, Cheon YH, Lee H, Kim S, Yurgelun-Todd DA, Renshaw PF, Yoon S, Lyoo IK (2023) Differential alterations in brain structural network organization during addiction between adolescents and adults. Psychol Med 53:3805-3816
    Pubmed KoreaMed CrossRef
  25. Damoiseaux JS, Greicius MD (2009) Greater than the sum of its parts: a review of studies combining structural connectivity and resting-state functional connectivity. Brain Struct Funct 213:525-533
    Pubmed CrossRef
  26. Yoon S, Cho H, Kim J, Lee DW, Kim GH, Hong YS, Moon S, Park S, Lee S, Lee S, Bae S, Simonson DC, Lyoo IK (2017) Brain changes in overweight/obese and normal-weight adults with type 2 diabetes mellitus. Diabetologia 60:1207-1217
    Pubmed CrossRef
  27. Lee S, Joo YJ, Kim RY, Hwang J, Lim SM, Yoon S, Kim J (2020) Obesity may connect insulin resistance to decreased neuronal viability in human diabetic brain. Obesity (Silver Spring) 28:1626-1630
    Pubmed CrossRef
  28. ElSayed NA, Aleppo G, Aroda VR, Bannuru RR, Brown FM, Bruemmer D, Collins BS, Hilliard ME, Isaacs D, Johnson EL, Kahan S, Khunti K, Leon J, Lyons SK, Perry ML, Prahalad P, Pratley RE, Seley JJ, Stanton RC, Gabbay RA (2023) 2. Classification and diagnosis of diabetes: standards of care in diabetes-2023. Diabetes Care 46(Suppl 1):S19-S40
    Pubmed KoreaMed CrossRef
  29. Ashburner J (2007) A fast diffeomorphic image registration algorithm. Neuroimage 38:95-113
    Pubmed CrossRef
  30. Beckmann CF, Smith SM (2004) Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans Med Imaging 23:137-152
    Pubmed CrossRef
  31. Caffarra P, Vezzadini G, Dieci F, Zonato F, Venneri A (2002) Rey-Osterrieth complex figure: normative values in an Italian population sample. Neurol Sci 22:443-447
    Pubmed CrossRef
  32. Sowell ER, Delis D, Stiles J, Jernigan TL (2001) Improved memory functioning and frontal lobe maturation between childhood and adolescence: a structural MRI study. J Int Neuropsychol Soc 7:312-322
    Pubmed CrossRef
  33. Namgung E, Kim J, Jeong H, Hong G, Kim M, Kim RY, Kim S, Lyoo IK (2021) Effects of Korean red ginseng on human gray matter volume and cognitive function: a voxel-based morphometry study. Hum Psychopharmacol 36:e2767
    Pubmed CrossRef
  34. Cavanna AE, Trimble MR (2006) The precuneus: a review of its functional anatomy and behavioural correlates. Brain 129(Pt 3):564-583
    Pubmed CrossRef
  35. Musen G, Jacobson AM, Bolo NR, Simonson DC, Shenton ME, McCartney RL, Flores VL, Hoogenboom WS (2012) Resting-state brain functional connectivity is altered in type 2 diabetes. Diabetes 61:2375-2379
    Pubmed KoreaMed CrossRef
  36. Cui Y, Liang X, Gu H, Hu Y, Zhao Z, Yang XY, Qian C, Yang Y, Teng GJ (2017) Cerebral perfusion alterations in type 2 diabetes and its relation to insulin resistance and cognitive dysfunction. Brain Imaging Behav 11:1248-1257
    Pubmed KoreaMed CrossRef
  37. Gordon BA, Blazey TM, Su Y, Hari-Raj A, Dincer A, Flores S, Christensen J, McDade E, Wang G, Xiong C, Cairns NJ, Hassenstab J, Marcus DS, Fagan AM, Jack CR Jr, Hornbeck RC, Paumier KL, Ances BM, Berman SB, Brickman AM, Cash DM, Chhatwal JP, Correia S, Förster S, Fox NC, Graff-Radford NR, la Fougère C, Levin J, Masters CL, Rossor MN, Salloway S, Saykin AJ, Schofield PR, Thompson PM, Weiner MM, Holtzman DM, Raichle ME, Morris JC, Bateman RJ, Benzinger TLS (2018) Spatial patterns of neuroimaging biomarker change in individuals from families with autosomal dominant Alzheimer's disease: a longitudinal study. Lancet Neurol 17:241-250
    Pubmed CrossRef
  38. Sweeney MD, Kisler K, Montagne A, Toga AW, Zlokovic BV (2018) The role of brain vasculature in neurodegenerative disorders. Nat Neurosci 21:1318-1331
    Pubmed KoreaMed CrossRef
  39. Jagust W (2018) Imaging the evolution and pathophysiology of Alzheimer disease. Nat Rev Neurosci 19:687-700
    Pubmed KoreaMed CrossRef
  40. Zhang Y, Zhang X, Ma G, Qin W, Yang J, Lin J, Zhang Q (2021) Neurovascular coupling alterations in type 2 diabetes: a 5-year longitudinal MRI study. BMJ Open Diabetes Res Care 9:e001433
    Pubmed KoreaMed CrossRef
  41. Namkung H, Kim SH, Sawa A (2017) The insula: an underestimated brain area in clinical neuroscience, psychiatry, and neurology. Trends Neurosci 40:200-207
    Pubmed KoreaMed CrossRef
  42. Chen WG, Schloesser D, Arensdorf AM, Simmons JM, Cui C, Valentino R, Gnadt JW, Nielsen L, Hillaire-Clarke CS, Spruance V, Horowitz TS, Vallejo YF, Langevin HM (2021) The emerging science of interoception: sensing, integrating, interpreting, and regulating signals within the self. Trends Neurosci 44:3-16
    Pubmed KoreaMed CrossRef
  43. Al-Zubaidi A, Iglesias S, Stephan KE, Buades-Rotger M, Heldmann M, Nolde JM, Kirchner H, Mertins A, Jauch-Chara K, Münte TF (2020) Effects of hunger, satiety and oral glucose on effective connectivity between hypothalamus and insular cortex. Neuroimage 217:116931
    Pubmed CrossRef
  44. Zanchi D, Depoorter A, Egloff L, Haller S, Mählmann L, Lang UE, Drewe J, Beglinger C, Schmidt A, Borgwardt S (2017) The impact of gut hormones on the neural circuit of appetite and satiety: a systematic review. Neurosci Biobehav Rev 80:457-475
    Pubmed CrossRef
  45. Okereke OI, Kang JH, Cook NR, Gaziano JM, Manson JE, Buring JE, Grodstein F (2008) Type 2 diabetes mellitus and cognitive decline in two large cohorts of community-dwelling older adults. J Am Geriatr Soc 56:1028-1036
    Pubmed CrossRef
  46. Yaffe K, Falvey C, Hamilton N, Schwartz AV, Simonsick EM, Satterfield S, Cauley JA, Rosano C, Launer LJ, Strotmeyer ES, Harris TB (2012) Diabetes, glucose control, and 9-year cognitive decline among older adults without dementia. Arch Neurol 69:1170-1175
    Pubmed KoreaMed CrossRef
  47. Ryan CM, Geckle M (2000) Why is learning and memory dysfunction in type 2 diabetes limited to older adults?. Diabetes Metab Res Rev 16:308-315
    Pubmed CrossRef
  48. Tuligenga RH, Dugravot A, Tabák AG, Elbaz A, Brunner EJ, Kivimäki M, Singh-Manoux A (2014) Midlife type 2 diabetes and poor glycaemic control as risk factors for cognitive decline in early old age: a post-hoc analysis of the Whitehall II cohort study. Lancet Diabetes Endocrinol 2:228-235
    Pubmed CrossRef
  49. Rawlings AM, Sharrett AR, Schneider AL, Coresh J, Albert M, Couper D, Griswold M, Gottesman RF, Wagenknecht LE, Windham BG, Selvin E (2014) Diabetes in midlife and cognitive change over 20 years: a cohort study. Ann Intern Med 161:785-793
    Pubmed KoreaMed CrossRef
  50. Lyoo IK, Yoon S, Jacobson AM, Hwang J, Musen G, Kim JE, Simonson DC, Bae S, Bolo N, Kim DJ, Weinger K, Lee JH, Ryan CM, Renshaw PF (2012) Prefrontal cortical deficits in type 1 diabetes mellitus: brain correlates of comorbid depression. Arch Gen Psychiatry 69:1267-1276
    Pubmed KoreaMed CrossRef
  51. Husain KH, Sarhan SF, AlKhalifa HKAA, Buhasan A, Moin ASM, Butler AE (2023) Dementia in diabetes: the role of hypoglycemia. Int J Mol Sci 24:9846
    Pubmed KoreaMed CrossRef
  52. Palta P, Schneider AL, Biessels GJ, Touradji P, Hill-Briggs F (2014) Magnitude of cognitive dysfunction in adults with type 2 diabetes: a meta-analysis of six cognitive domains and the most frequently reported neuropsychological tests within domains. J Int Neuropsychol Soc 20:278-291
    Pubmed KoreaMed CrossRef