Patients with type 2 diabetes mellitus (T2DM) have a considerably high risk of developing dementia, especially for those with mild cognitive impairment (MCI). The investigation of the microstructural change of white matter (WM) between T2DM with amnesic MCI (T2DM-aMCI) and T2DM with normal cognition (T2DM-NC) and their relationships to cognitive performances can help to understand the brain variations in T2DM-related amnesic cognitive impairment. In the current study, 36 T2DM-aMCI patients, 40 T2DM-NC patients, and 40 healthy control (HC) individuals underwent diffusion tensor image and T1-weighted MRI scans and comprehensive cognition assessments. All of these cognitive functions exhibited intergroup ranking differences in patients. The T2DM-NC patients and HC individuals did not reveal any significant differences in WM integrity. The T2DM-aMCI patients showed disrupted integrity in multiple WM tracts compared with HC and T2DM-NC. Specifically, the damaged WM integrity of the right inferior fronto-occipital fasciculus and the right inferior longitudinal fasciculus exhibited significant correlations with episodic memory and attention function impairment in T2DM patients. Furthermore, cognitive impairment-related WM microstructural damage was associated with the degeneration of cortex connected to the affected WM tract. These findings indicate that degeneration exists extensively in WM tracts in T2DM-aMCI, whereas no brain WM damage is evident in T2DM-NC.

Diabetes is a common metabolic disorder that can lead to chronic complications such as cardiovascular disease and peripheral neuropathy. In diabetes, the type 2 diabetes mellitus (T2DM) caused by insulin resistance accounts for 90–95% of diabetes cases. Previous reports have indicated that T2DM is associated with an increase in the risk of dementia and in the proportion of patients who convert from mild cognitive impairment (MCI) to dementia (1). T2DM is associated with amnestic MCI (aMCI) and nonamnestic MCI (2,3), the former being considered to be a prodromal stage of Alzheimer disease (AD) (4). The evidence of a link between T2DM and aMCI seems relatively less and weak and is the subject of much debate.

Recent studies demonstrated a pathophysiological link between T2DM and cognitive impairment as well as the underlying pathways that connect these two diseases (5). Longitudinal population-based studies have revealed that the relative risk of dementia is 1.2- to 2.8-times higher in individuals with diabetes relative to those without diabetes (6). In addition, different stages of diabetes-associated cognitive dysfunction exist with different cognitive features (7,8), and the rates of conversion to dementia in patients with T2DM with MCI (T2DM-MCI) are significantly higher compared with T2DM patients who were cognitively intact and with MCI patients without diabetes (9). The mean annual conversion rate to dementia of the T2DM-MCI and cognitively intact T2DM cohorts, respectively, was 8.43% and 1.64% per year and was 3.86% per year for the MCI cohort without T2DM (9), which suggests that more brain toxicity may have occurred in T2DM with cognitive impairment than MCI without T2DM or T2DM with normal cognition (T2DM-NC). However, it is still unclear, and limited literature has researched, why a subset of T2DM patients develop MCI and how they are different, brainwise, from non-MCI patients.

As reported previously, many studies have revealed that white matter (WM) integrity measured by a diffusion tensor imaging (DTI) technique is particularly sensitive to T2DM (1013) and cognitive impairment (14). These studies currently showed that regions of WM integrity degeneration caused by T2DM are scattered and inconsistent across the entire brain. Reijmer et al. (12), for example, found WM integrity damage in the superior longitudinal fasciculus (SLF), the inferior longitudinal fasciculus (ILF), and the uncinate fasciculus (UF) in T2DM patients. Hoogenboom et al. (11) demonstrated that patients with T2DM exhibited lower integrity in the cingulum bundle and the UF, whereas patients did not show significantly lower integrity in the SLF. Zhang et al. (10) revealed that T2DM patients with various degrees of cognitive impairment exhibited widespread WM disruptions, especially in the whole corpus callosum, the left anterior limb of the internal capsule, and external capsule. These inconsistent results may occur because T2DM patients were in different stages of diabetes-associated cognitive dysfunction in different studies. Therefore, this study investigated T2DM patients with and without risk of aMCI and the WM integrity mechanism.

In the current study, we enrolled T2DM-aMCI patients, T2DM-NC patients, and healthy control (HC) subjects to conduct clinical assessments, a battery of neuropsychological tests, DTI, and T1-weighted imaging. Voxel-wise analyses were applied to investigate WM differences in the three groups, and then we researched the relationships between WM injuries and cognitive performance to test for cognitive impairment-related WM fibers in T2DM. Finally, we evaluated whether cognitive impairment-related WM microstructural damage was associated with the degeneration of cortex connected to the affected WM tract. We expected that our study could provide insight into the pathogenic mechanisms of T2DM-related cognitive impairment.

Participants

All participants of this study were recruited from the Beijing Aging Brain Rejuvenation Initiative (BABRI) project, an ongoing longitudinal study from an urban population in Beijing, China. There are 36 T2DM-aMCI, 40 T2DM-NC, and 40 HC subjects. The participant inclusion criteria were: 1) no less than 6 years of education; 2) aged between 50 and 80 years; 3) right-handed; 4) Mini-Mental State Examination (MMSE) scores >24 on the Chinese version; 5) no history of coronary disease, nephritis, tumors, gastrointestinal disease, or psychiatric illness; 6) no contraindications to MRI; and 7) no obvious microvascular disease. All participants self-reported to have never experienced severe hypoglycemia and to have a low frequency of minor hypoglycemia. The Beijing Normal University Institutional Review Board approved the study. Each participant provided written informed consent.

All T2DM patients had a history of using oral antidiabetic medications, insulin, or no treatment and were diagnosed using established criteria based on the patients’ medical histories, medication use, or fasting plasma glucose (FPG) levels (>7 mmol/L). Of the 36 T2DM-aMCI patients, 9 were being treated with insulin, 4 were being treated with diet and exercise, and 23 controlled their blood glucose with oral hypoglycemic agents (6 with glucosidase inhibitors, 11 with biguanides, 3 with sulfonylurea, 1 with nateglinide, and 2 with glucosidase inhibitors and biguanides). Of 40 T2DM-NC patients, 5 were receiving treatment with insulin, 11 were being treated with diet and exercise, and 24 controlled their blood glucose with oral hypoglycemic agents (2 with glucosidase inhibitors, 5 with biguanides, 2 with sulfonylurea, 12 with 2 drugs combined, and 3 with 3 drugs combined). BMI, FPG, glycosylated hemoglobin (HbA1c), triglyceride, total cholesterol, HDL, and LDL levels were measured via standard laboratory testing.

Neuropsychological Testing

All participants completed a battery of neuropsychological tests, including the MMSE (15) and the following six cognition domains and associated tests:

  1. Episodic memory: Auditory Verbal Learning Test (AVLT)-Delay Recall and AVLT-Total (16), and Rey-Osterrieth Complex Figure (ROCF)-Delay Recall test (17)

  2. Working memory: the digit span test and digit span backward (18)

  3. Spatial processing: ROCF-Copy (17) and Clock Drawing Test (19)

  4. Executive function: Trail Making Test (TMT)-B and Stroop Color and Word Test–C-B (20)

  5. Language ability: Category Verbal Fluency Test and Boston Naming Test (21)

  6. Attention: TMT-A and Symbol Digit Modalities Test (20)

We diagnosed aMCI subjects according to Petersen (22) criteria for aMCI, including subjective memory complaints, cognitive impairment in memory (scoring >1.5 SDs below the age- and education-adjusted norm on at least two tests with the AVLT-Delay Recall, AVLT-Total or ROCF-Delay Recall), normal general cognitive function (scoring no less than 24 on the MMSE), essentially intact activities of daily living, and lack of dementia. If memory was the only area of impairment, the subject was considered to be single-domain aMCI. If memory plus other cognitive domains assessed with neuropsychological testing were affected, the subject was considered to be multidomain aMCI. The study included single-domain aMCI and multidomain aMCI. The HC subjects had no evidence of cognitive deficits on any of these the neuropsychological tests.

MRI Data Acquisition

Imaging data were collected using a Siemens Trio 3-T MRI system including DTI data and T1-weighted image scans. Participants were positioned supine with their head snugly fixed by straps and foam pads to minimize head movement. Diffusion-weighted imaging was performed using a single-shot, twice-refocused, diffusion-weighted echo planar sequence aligned along the anterior-posterior commissural plane. T1-weighted, sagittal three-dimensional magnetization prepared rapid gradient echo (MP-RAGE) sequences were acquired and covered the entire brain. The parameters of DTI and T1-weighted images refer to our previous study (23).

DTI Data Preprocessing and Analysis

The FMRIB Diffusion Toolbox (FDT, v3.0) in the FMRIB Software Library (FSL, v5.0.10) (24) was used for image preprocessing. Briefly, the preprocessing involves correction of eddy current and head movement, creating a brain mask and fitting the diffusion tensor model. The output yielded voxel-wise maps of fractional anisotropy (FA), mean diffusivity, radial diffusivity, and axial diffusivity. The FA index of DTI is the most sensitive neuroimaging measure of the degeneration observed in T2DM patients and describes overall WM health, maturation, and organization (25). Another index, axial diffusivity, reflects axon integrity and can be useful in understanding the underlying physiology (26). Voxel-wise statistical analysis of the FA data was performed using Tract-Based Spatial Statistics (TBSS) (27), part of FSL (24). TBSS projects all subjects’ FA data onto a mean FA tract skeleton before applying voxel-wise cross-subject statistics.

Voxel-wise statistical analyses were performed using a nonparametric permutation-based inference tool (“randomize,” part of FSL) with the general linear model (GLM) as statistical modeling. First, pairwise group comparisons based on voxels were performed using GLM for T2DM-aMCI versus HC, T2DM-aMCI versus T2DM-NC, and T2DM-NC versus HC. The DTI parameters at each voxel were modeled as a linear combination of predictors (three grouping variables) and covariates (sex, age, education, hypertension, hyperlipidemia, and cerebrovascular disease) stored in the columns of a “design matrix.” For subsequent analyses, we calculated WM volumes in regions with between-group differences, and DTI parameters at voxels within significant intergroup different tracts in all T2DM patients. WM tracts were identified using the JHU_ICBM_tracts_maxprob_thr25 atlas (28).

Second, a voxel-wise linear relationship was determined using GLM in all T2DM. The DTI parameters at each voxel in significant intergroup different tracts are modeled as a linear combination of cognitive scores with significant intergroup difference domains and covariates. The significance threshold for between-group differences and linear relationships were set at P < 0.05 (5,000 permutations, familywise error [FWE] correction for multiple comparisons correction) using the threshold-free cluster enhancement.

Next, we extracted the mean value of axial diffusivity on significantly related fibers for subsequent analyses. FA of significant intergroup different tracts did not show any significant relationships with cognitive measures.

T1 MRI Data Processing and Analysis

T1-weighted images were analyzed with the FSL voxel-based morphometry (FSLVBM) pipeline (24,29). First, structural images were brain extracted and gray matter segmented before being registered to the Montreal Neurological Institute (MNI) 152 standard space using nonlinear registration. The resulting images were averaged and flipped along the x-axis to create a left-right symmetric, study-specific gray matter template.

Second, all native gray matter images were nonlinearly registered to this study-specific template and “modulated” to correct for local expansion due to the nonlinear component of the spatial transformation. The modulated gray matter images were then smoothed with an isotropic Gaussian kernel with a σ of 3 mm. All the modulated registered gray matter images were concatenated into a four-dimensional image, and the file was fed into voxel-wise statistics.

Next, a voxel-wise linear relationship was determined with GLM in all T2DM using a randomize tool to evaluate whether cognitive impairment-related WM microstructural damage was associated with the degeneration of cortex connected to the affected WM tract. The gray matter density in regions of interests (ROIs) at each voxel was modeled as a linear regression of diffusion metrics in cognitive impairment-related WM tracts and covariates (sex, age, education, hypertension, hyperlipidemia, and cerebrovascular disease). The significance threshold for a linear relationship was set at P < 0.05 (5,000 permutations, FWE-correction for multiple comparisons). In this study, the gray matter ROIs were the lingual and frontal middle orbital (Frontal_Mid_Orb) with anatomic connections to inferior fronto-occipital fasciculus (IFOF) as well as the temporal inferior (Temporal_Inf) and occipital middle (Occipital_Mid) with anatomic connections to ILF (30).

Statistical Analysis

Consistency in age, sex, and education level among the T2DM-aMCI, T2DM-NC, and HC groups was determined with one-way ANOVA or χ2 analysis. The neuropsychological assessment and clinical data were analyzed using an ANCOVA, with age, sex, education, hypertension, hyperlipidemia, and cerebrovascular disease included as covariates. Post hoc pairwise t tests with Bonferroni correction were performed to follow the significant main effects yielded by the ANOVA and ANCOVA tests (P < 0.05). To adjust for possible spurious findings due to multiple ANCOVA testing, Bonferroni correction was conducted for all the cognition tests. Next, we assessed the relationship between the mean value of axial diffusivity of the significant region (SR, cognitive scores significantly related to the WM index in this region) and the neuropsychological scores in all T2DM patients, which was performed with the regression correlation analysis. Similarly, we assessed the regression correlation between cognitive impairment–related diffusion metrics in the SRs and gray matter density in all T2DM patients. Regression correlations were performed after controlling for the influences of age, sex, and education (Bonferroni correction for multiple WM fibers or multiple gray matter regions). We present normalized cognitive scores as well as z values. To test which cognitive functions were most affected, we calculated the percentage of patients with a score <1.5 SD across all participants in Supplementary Data (Supplementary Table 1). All statistical analyses were performed using SPSS 22.0 software. Please see Supplementary Data for the technical MRI details.

Data and Resource Availability

The data sets generated during and/or analyzed during the current study are not publicly available because they are under construction but are available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.

Demographics and Neuropsychological Testing

Demographic data and neuropsychological testing are presented in Table 1 and Supplementary Fig. 1. There were no significant differences in age, sex, year of education, BMI, total cholesterol, triglycerides, HDL, or LDL among the T2DM-aMCI, T2DM-NC, and HC groups. There was no difference in T2DM disease duration between T2DM-aMCI and T2DM-NC, but significant differences of HbA1c were observed (P < 0.001) between the HC group and each of the T2DM groups. As expected, the MMSE scores were significantly higher in HC subjects compared with T2DM-aMCI and T2DM-NC patients, and T2DM-NC patients performed better than T2DM-aMCI patients. For various neuropsychological scores, one-way ANOVA showed that there were significant group effects for most of the cognitive domains with the best performance in HC subjects, an intermediate performance in T2DM-NC patients, and the worst performance in T2DM-aMCI patients. Comparing the neuropsychological test scores between the T2DM-aMCI and T2DM-NC groups revealed significant differences in the episodic memory, working memory, spatial processing, language ability, and attention domains; however, there were no differences in executive function.

Table 1

Demographic and neuropsychological test results

T2DM-aMCIT2DM-NCHCF value (t2)P value
 n = 36 n = 40 n = 40 — — 
Sex, n    0.054 0.973# 
 Male 13 15 14   
 Female 23 25 26   
Age (years) 65.97 ± 7.81 65.35 ± 7.77 66.48 ± 7.46 0.215 0.807 
Education (years) 10.89 ± 3.64 11.88 ± 3.16 12.46 ± 2.99 2.219 0.113 
Diabetes duration (years) 11.13 ± 8.05 8.92 ± 6.29 — 1.19 0.238 
BMI (kg/m227.46 ± 7.59 25.51 ± 3.33 24.03 ± 4.87 1.98 0.144 
Hypertension, n (%) 25 (69) 23 (58) 16 (40) 6.78 0.034#† 
Hyperlipidemia, n (%) 13 (9) 22 (55) 11 (27) 6.59 0.037#‡ 
Cerebrovascular disease, n (%) 10 (28) 7 (18) 1 (3) 9.42 0.009#† 
General mental status      
 MMSE 25.26 ± 4.86 27.78 ± 1.56 29.43 ± 0.75 16.23 <0.001†ξ 
Episodic memory      
 AVLT-Delay Recall 1.18 ± 1.07 5.63 ± 2.64 8.55 ± 2.06 89.28 <0.001†‡ξ 
 AVLT-Total 14.49 ± 4.41 29.20 ± 9.62 40.38 ± 7.86 80.12 <0.001†‡ξ 
 ROCF-Delay Recall 8.9 ± 6.63 13.97 ± 7.15 20.48 ± 7.33 18.96 <0.001†‡ξ 
Working memory      
 Digit Span 10.97 ± 2.23 12.63 ± 2.27 13.18 ± 1.59 6.73 0.002†ξ 
 Digit Span Backward 3.77 ± 1.18 4.95 ± 1.28 5.10 ± 0.96 8.42 <0.001†ξ 
Spatial processing      
 ROCF-Copy 31.61 ± 6.42 33.80 ± 2.94 34.08 ± 4.16 1.29 0.279 
 Clock Drawing Test 20.88 ± 5.56 24.97 ± 4.93 26.28 ± 3.30 12.22 <0.001†ξ 
Executive function      
 Stroop Color and Word Test C-B time 41.36 ± 16.83 40.25 ± 22.01 34.63 ± 22.13 1.29 0.279 
 TMT-B time 177.11 ± 52.11 163.35 ± 53.69 145 ± 49.29 3.73 0.028 
Language ability      
 Boston Naming Test 21.71 ± 4.47 24.40 ± 2.88 24.46 ± 2.94 5.08 0.008 
 Category Verbal Fluency Test 37.64 ± 9.69 44.85 ± 9.35 49 ± 7.07 8.62 <0.001†ξ 
Attention      
 Symbol Digit Modalities Test 28.96 ± 9.39 33.89 ± 9.44 38.55 ± 10.92 5.62 0.005† 
 TMT-A time 84.63 ± 78.77 60.88 ± 17.69 53.15 ± 13.88 4.66 0.012 
Biochemical indicator      
 HbA1c (%) 6.49 ± 0.45 6.40 ± 1.31 4.93 ± 0.38 10.74 <0.001†‡ 
 HbA1c (mmol/mol) 47.37 ± 5.3 46.48 ± 4.29 30.31 ± 4.11 10.74 <0.001†‡ 
 Total cholesterol (mmol/L) 5.11 ± 0.73 4.80 ± 0.701 5.67 ± 1.10 0.904 0.427 
 Triglycerides (mmol/L) 2.13 ± 1.47 2.99 ± 1.99 1.55 ± 0.55 2.015 0.17 
 HDL (mmol/L) 1.29 ± 0.18 1.59 ± 0.29 1.38 ± 0.28 0.906 0.427 
 LDL (mmol/L) 3.01 ± 1.03 1.85 ± 0.09 3.52 ± 1.02 2.554 0.113 
T2DM-aMCIT2DM-NCHCF value (t2)P value
 n = 36 n = 40 n = 40 — — 
Sex, n    0.054 0.973# 
 Male 13 15 14   
 Female 23 25 26   
Age (years) 65.97 ± 7.81 65.35 ± 7.77 66.48 ± 7.46 0.215 0.807 
Education (years) 10.89 ± 3.64 11.88 ± 3.16 12.46 ± 2.99 2.219 0.113 
Diabetes duration (years) 11.13 ± 8.05 8.92 ± 6.29 — 1.19 0.238 
BMI (kg/m227.46 ± 7.59 25.51 ± 3.33 24.03 ± 4.87 1.98 0.144 
Hypertension, n (%) 25 (69) 23 (58) 16 (40) 6.78 0.034#† 
Hyperlipidemia, n (%) 13 (9) 22 (55) 11 (27) 6.59 0.037#‡ 
Cerebrovascular disease, n (%) 10 (28) 7 (18) 1 (3) 9.42 0.009#† 
General mental status      
 MMSE 25.26 ± 4.86 27.78 ± 1.56 29.43 ± 0.75 16.23 <0.001†ξ 
Episodic memory      
 AVLT-Delay Recall 1.18 ± 1.07 5.63 ± 2.64 8.55 ± 2.06 89.28 <0.001†‡ξ 
 AVLT-Total 14.49 ± 4.41 29.20 ± 9.62 40.38 ± 7.86 80.12 <0.001†‡ξ 
 ROCF-Delay Recall 8.9 ± 6.63 13.97 ± 7.15 20.48 ± 7.33 18.96 <0.001†‡ξ 
Working memory      
 Digit Span 10.97 ± 2.23 12.63 ± 2.27 13.18 ± 1.59 6.73 0.002†ξ 
 Digit Span Backward 3.77 ± 1.18 4.95 ± 1.28 5.10 ± 0.96 8.42 <0.001†ξ 
Spatial processing      
 ROCF-Copy 31.61 ± 6.42 33.80 ± 2.94 34.08 ± 4.16 1.29 0.279 
 Clock Drawing Test 20.88 ± 5.56 24.97 ± 4.93 26.28 ± 3.30 12.22 <0.001†ξ 
Executive function      
 Stroop Color and Word Test C-B time 41.36 ± 16.83 40.25 ± 22.01 34.63 ± 22.13 1.29 0.279 
 TMT-B time 177.11 ± 52.11 163.35 ± 53.69 145 ± 49.29 3.73 0.028 
Language ability      
 Boston Naming Test 21.71 ± 4.47 24.40 ± 2.88 24.46 ± 2.94 5.08 0.008 
 Category Verbal Fluency Test 37.64 ± 9.69 44.85 ± 9.35 49 ± 7.07 8.62 <0.001†ξ 
Attention      
 Symbol Digit Modalities Test 28.96 ± 9.39 33.89 ± 9.44 38.55 ± 10.92 5.62 0.005† 
 TMT-A time 84.63 ± 78.77 60.88 ± 17.69 53.15 ± 13.88 4.66 0.012 
Biochemical indicator      
 HbA1c (%) 6.49 ± 0.45 6.40 ± 1.31 4.93 ± 0.38 10.74 <0.001†‡ 
 HbA1c (mmol/mol) 47.37 ± 5.3 46.48 ± 4.29 30.31 ± 4.11 10.74 <0.001†‡ 
 Total cholesterol (mmol/L) 5.11 ± 0.73 4.80 ± 0.701 5.67 ± 1.10 0.904 0.427 
 Triglycerides (mmol/L) 2.13 ± 1.47 2.99 ± 1.99 1.55 ± 0.55 2.015 0.17 
 HDL (mmol/L) 1.29 ± 0.18 1.59 ± 0.29 1.38 ± 0.28 0.906 0.427 
 LDL (mmol/L) 3.01 ± 1.03 1.85 ± 0.09 3.52 ± 1.02 2.554 0.113 

All subjects (T2DM-aMCI, T2DM-NC, HC) were matched for age, sex, and education. Values are the mean ± SD except where indicated. The comparisons of each cognition test among three groups were performed with ANCOVA. P < 0.0036 (0.05/14) was considered significant after Bonferroni correction. Post hoc pairwise comparisons were performed using t tests. P < 0.017 (0.05/3) was considered significant after Bonferroni correction. #The P value in the three groups was obtained using a χ2 test. Post hoc paired comparisons showed significant group differences between T2DM-aMCI and HC (†), between T2DM-NC and HC (‡), and between T2DM-aMCI and T2DM-NC (ξ).

Group Differences in WM Diffusion Metrics Among the Groups

We assessed voxel-wise WM integrity differences with DTI parameters between each group pair (Fig. 1 and Supplementary Table 2). Compared with the HC subjects, T2DM-aMCI patients showed a significantly decreased FA value in widespread WM tracts, including the bilateral anterior thalamic radiation (ATR), bilateral corticospinal tract (CST), bilateral cingulum, forceps minor, bilateral IFOF, bilateral SLF, bilateral UF, and left SLF temporal (SLF-temp) part (FWE correction, P < 0.05). Compared with T2DM-NC patients, T2DM-aMCI patients showed significantly reduced FA value decreases in right ATR, right CST, forceps major, forceps minor, right IFOF, right ILF, right SLF, and right SLF-temp (FWE correction, P < 0.05). There were no significant differences between T2DM-NC patients and HC subjects. Volumes in intergroup differences in WM regions were calculated and are reported in Table 2. Axial and mean diffusivity of WM regions showed no significant difference among the three groups.

Figure 1

Voxel-wise Tract-Based Spatial Statistics differences in FA metrics between groups. Green represents the mean WM skeleton of all subjects. Top row: Red-yellow voxels (thickened for better visibility) represent the WM regions with decreased FA in the T2DM-aMCI patients compared with HC subjects (FWE corrected, P < 0.05). Middle row: Red-yellow voxels represent the WM regions with reduced FA in the T2DM-aMCI group compared with T2DM-NC group (FWE corrected, P < 0.05). Bottom row: There are no significant WM microstructural differences between the T2DM-NC patients and HC subjects.

Figure 1

Voxel-wise Tract-Based Spatial Statistics differences in FA metrics between groups. Green represents the mean WM skeleton of all subjects. Top row: Red-yellow voxels (thickened for better visibility) represent the WM regions with decreased FA in the T2DM-aMCI patients compared with HC subjects (FWE corrected, P < 0.05). Middle row: Red-yellow voxels represent the WM regions with reduced FA in the T2DM-aMCI group compared with T2DM-NC group (FWE corrected, P < 0.05). Bottom row: There are no significant WM microstructural differences between the T2DM-NC patients and HC subjects.

Close modal
Table 2

WM integrity differences and its relationship with cognitive scores in all T2DM patients

TractsVolumes (mm3)Axial diffusivity and cognitive correlation
T2DM-aMCI vs. HCT2DM-aMCI vs. T2DM-NCAVLT-TotalTMT-A time
ATR.L 83 — — 
ATR.R 80 — — 
CST.L 102 — — 
CST.R 17 142 — — 
CG.L 28 — — 
CG.R — — 
CH.L — — 
CH.R — — 
Fmaj 119 0.359** — 
10 Fmin 369 13 — — 
11 IFOF.L 86 — — 
12 IFOF.R 111 63 0.441*** −0.409** 
13 ILF.L — — 
14 ILF.R 35 0.431*** −0.512*** 
15 SLF.L 29 — — 
16 SLF.R 73 — — 
17 UF.L 66 — — 
18 UF.R 28 — — 
19 SLF-temp.L — — 
20 SLF-temp.R 20 — — 
TractsVolumes (mm3)Axial diffusivity and cognitive correlation
T2DM-aMCI vs. HCT2DM-aMCI vs. T2DM-NCAVLT-TotalTMT-A time
ATR.L 83 — — 
ATR.R 80 — — 
CST.L 102 — — 
CST.R 17 142 — — 
CG.L 28 — — 
CG.R — — 
CH.L — — 
CH.R — — 
Fmaj 119 0.359** — 
10 Fmin 369 13 — — 
11 IFOF.L 86 — — 
12 IFOF.R 111 63 0.441*** −0.409** 
13 ILF.L — — 
14 ILF.R 35 0.431*** −0.512*** 
15 SLF.L 29 — — 
16 SLF.R 73 — — 
17 UF.L 66 — — 
18 UF.R 28 — — 
19 SLF-temp.L — — 
20 SLF-temp.R 20 — — 

WM volumes were from the significant group differences (FWE corrected, P < 0.05). CG, cingulum (cingulate gyrus); CH, cingulum (hippocampus); Fmaj, forceps major; Fmin, forceps minor.

**

P < 0.01; ***P < 0.001.

WM Integrity in Relation to Cognitive Performance

Based on the WM tracts with significant intergroup differences (FWE correction, P < 0.05), we explored the relationship between diffusion metrics (FA and axial diffusivity) of those tracts as well as the neuropsychological scores in all T2DM patients. The results of correlation analysis based on voxel-wise show that the TMT-A test was significantly correlated with the axial diffusivity of ATR.right hemisphere (R), IFOF.R, ILF.R, SLF.R, SLF-temp.R, UF.R, CST.R, and CST.left hemisphere (L), and AVLT-Total was significantly correlated with the axial diffusivity of forceps major, IFOF.R and ILF.R. The results of regression correlations based on tractwise showed higher axial diffusivity for the forceps major (β = 0.359, P = 0.004), the IFOF.R (β = 0.441, P < 0.001, in SR 1), and the ILF.R (β = 0.431, P < 0.001) in SR 4, which was significantly correlated with better episodic memory performance (AVLT-Total) after Bonferroni correction (Fig. 2 and Table 2). In addition, axial diffusivity of IFOF.R (β = −0.409, P = 0.005 in SR 2) and ILF.R (β = −0.512, P < 0.001) in SR 3 was significantly and negatively correlated with attention performance (TMT-A Time) after Bonferroni correction (Fig. 2 and Table 2).

Figure 2

The relationship between WM integrity, cognitive scores, and gray matter density. A: The correlations between IFOF.R and ILF.R axial diffusivity and cognitive scores in T2DM patients. B: The correlations between IFOF.R and ILF.R axial diffusivity and gray matter density for areas anatomically connected to these two tracks. SR 1/4, the SR of AVLT-Total related to the axial diffusivity index; SR 2/3, the significant region of TMT-A related to the axial diffusivity index.

Figure 2

The relationship between WM integrity, cognitive scores, and gray matter density. A: The correlations between IFOF.R and ILF.R axial diffusivity and cognitive scores in T2DM patients. B: The correlations between IFOF.R and ILF.R axial diffusivity and gray matter density for areas anatomically connected to these two tracks. SR 1/4, the SR of AVLT-Total related to the axial diffusivity index; SR 2/3, the significant region of TMT-A related to the axial diffusivity index.

Close modal

Correlations Between WM Integrity and Gray Matter Density

Based these above results, we found that the IFOF.R and ILF.R are important fiber tracks with differences for patients with T2DM at different cognitive stages and that these regions are closely related to cognitive function. Therefore, we further explored the relationship between these two fiber tracks and the gray matter structures connected to them. The lingual.R and Frontal_Mid_Orb.R were anatomically connected to IFOF.R. Correlation analysis results showed the gray matter density of lingual.R was significantly related to the axial diffusion metrics of IFOF.R in SR 1 (β = −0.519, P < 0.001) and SR 2 (β = −0.327, P = 0.004) in T2DM patients (Fig. 2). The Temporal_Inf.R and Occipital_Mid.R were anatomically connected to the ILF.R. Correlation analysis results showed the gray matter density of right Temporal_Inf.R was significantly related to the axial diffusion metrics of ILF.R in SR 3 (β = 0.376, P = 0.001) and SR 4 (no significance) in T2DM patients (Fig. 2).

In the current study, several neuropsychological scores, including general mental status, episodic memory, working memory, spatial processing, language ability, and attention, exhibited intergroup gradual declines, with the best performance occurring in HC subjects, an intermediate performance in T2DM-NC patients, and the worst performance in T2DM-aMCI patients. Our subsequent analysis revealed that T2DM-aMCI patients had widespread WM disruptions and that T2DM-NC patients did not show any WM damage. In particular, the right IFOF and right ILF were significantly correlated with episodic memory and attention performance in T2DM patients. More importantly, cognitive impairment-related WM microstructural damage was associated with the degeneration of cortex connected to the affected WM tract.

This study examined DTI metrics to characterize comprehensive WM microstructural changes. FA measurements are the primary DTI-derived metrics believed to reflect the degree of alignment of cellular structures within fiber tracts and their structural integrity (31). Although the specific cellular mechanisms underlying changes in FA measurements remain unclear in T2DM, our previous findings of decreased FA may be explained by the loss of axonal fibers, demyelination, or both (10).

Previous review studies have shown that T2DM and AD share several pathogeneses, including neurogenesis, blood-brain barrier, hyperglycemia, insulin resistance, and vascular dysfunction (32). Insulin resistance represents the main feature linking T2DM and the risk to develop AD and also seems to be an important determinant of axon pathologically related tau protein phosphorylation (33,34) and amyloid deposition (35), which are the critical pathology of AD. Given that insulin receptors are widely expressed and involved in the hippocampus and connected limbic brain structures, the most important brain region for learning and memory, it is plausible that decline of insulin signaling leads to cognitive impairment (36,37). A similar mechanism may also explain the observed significant relationships between axial diffusivity and cognitive performance in T2DM patients in Fig. 2. Accordingly, diffuse WM changes with a loss of myelin or axonal abnormalities have been well documented in almost all subtypes of T2DM and cognitive impairment patients (10,14,38), and T2DM-related cognitive impairment is related to axon pathologies. Moreover, previous study suggested that type 2 diabetes may be associated with aMCI through both AD and vascular pathology (2). We therefore speculated that the decreased WM integrity related to TMT-A in the results might be caused by vascular pathological factors of T2DM.

To explore the WM integrity change due to T2DM and T2DM-related cognitive impairment, we compared the difference of WM integrity between three groups. It was not surprising that the widespread decreased integrity in the cerebral WM skeleton in T2DM-aMCI patients compared with HC subjects was observed in the bilateral ATR and CST, right cingulate gyrus, and forceps major and minor, bilateral IFOF, and left UF. Especially, when the T2DM-aMCI group was compared with T2DM-NC patients without clinically confirmed cognitive impairment, the results were still considerably different in most regions, such as the right CST, forceps major and minor, right IFOF, right ILF, and right SLF. In addition, the T2DM-NC patients did not show significant reductions in WM integrity compared with the HC subjects. Thus, these results suggested that the observed changes in WM integrity are strongly coupled with cognitive impairment in T2DM patients with cognitive impairment. But previous studies suggest that T2DM-NC patients were observed to have microstructural WM abnormalities in multiple fibers such as the SLF, UF, ILF (12), and external capsule (39). The inconsistency of these results may be due to sample differences, with all of our patients from a community study and participants reported by Reijmer et al. (12) from a clinical study (part of the second Utrecht Diabetic Encephalopathy Study). In addition, different methodologies—an ROI-based approach may help examine the overall change in fiber bundles, and voxel-based analysis may help identify areas of local variability on long fibers— may lead to different results.

More importantly, although previous studies have suggested that T2DM patients will have increases in microstructural injury in multiple WM fibers connecting distinct cortical and subcortical regions (10,12), the fibers that are related to T2DM-related cognitive disturbances are still unclear. Our findings clarify this relationship. The diffusion measures of the forceps major, IFOF, and ILF were correlated with cognitive test scores among T2DM patients, and these areas are involved performance of episodic memory and attention. It is possible that selective involvement of those tracts may represent the neural substrates that underlie the principal component of episodic memory and attention function due to the important role of these connecting tracts in cognitive functions in the brain. As supported by previous studies, the ILF, which is an occipitotemporal fiber bundle connecting occipital and temporal areas (40), was associated with attention-deficit disorder, slower processing speeds, and episodic memory performance in T2DM-NC patients and cognitively normal older adults (12,41,42). The forceps major, which connects the occipital lobes and crosses the midline via the splenium of the corpus callosum (43,44), is connected to the memory domain (45). The IFOF, which connects the frontal lobe with the posterior portions of the parietal and temporal lobes, mediates a direct communication between the occipital and frontal lobes (41,46) and is important in attention (47) as well as in episodic memory (48,49) in cognitive dysfunction in T2DM patients. In summary, T2DM damaged and accelerated patient cognitive ability mainly by attacking the ILF, IFOF, and forceps major. The aMCI-specific effects of T2DM suggest that T2DM had effects on the WM integrity primarily in patients with MCI and that the WM changes seen in the prodromal stages of dementia may be accelerated by T2DM.

Some limitations to our study should be noted. First, this is a cross-sectional study with a relatively small sample size; therefore, a longitudinal study with a larger sample size is needed in the future to achieve more convincing and accurate results. Second, participants with different levels of educational achievement may have shown different effects, although education was used as a control variable. Third, we only focused on the correlated relationship between the WM injury and gray matter atrophy in T2DM patients. Fourth, there was no aMCI-normoglycemic group. The causal relationship between these changes is still unclear. Fifth, it would be important to monitor the retinal disease information of patients to investigate the effect of retinal disease on the incidence of dementia. These factors should be included in future studies.

In summary, the current study revealed that T2DM-NC patients did not show any WM damage, and T2DM-aMCI patients showed significantly decreased WM integrity in widespread WM tracts, which suggests that significant WM changes occurred in cognitive impairment stages in T2DM patients. In particular, the episodic memory and attention function impairment in T2DM patients has significant correlations with the integrity decline of WM fibers in the right IFOF and right ILF. Moreover, WM microstructural damage related to cognitive impairment was associated with the degeneration of cortex connected to the affected WM tract.

Acknowledgments. The authors thank all of the volunteers and participants for participating in the study and the State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, for community contact and data acquisition.

Funding. This work was supported by National Science Fund for Distinguished Young Scholars (grant number 81625025), Funds for International Cooperation and Exchange of the National Natural Science Foundation of China (grant number 81820108034), National Key Research and Development Project of China (grant number 2018YFC1315200), State Key Program of National Natural Science of China (grant number 81430100), Beijing Municipal Science & Technology Commission (grant number Z161100000216135), National Natural Science Foundation of China (grant number 31700997), and the Fundamental Research Funds for the Central Universities (grant number 2017XTCX04).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.G. contributed to the recruitment of the study population, neuropsychological testing, data analysis, and drafting and final approval of the manuscript. Y.C. contributed to the data analysis and drafting and final approval of the manuscript. F.S., Y.Y., J.X., X.L., and J.Z. contributed to the recruitment of the study population and neuropsychological testing. K.C. contributed to final approval of the manuscript. Z.Z. contributed to the study concept and revision and final approval of the manuscript. Z.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Koekkoek
PS
,
Kappelle
LJ
,
van den Berg
E
,
Rutten
GE
,
Biessels
GJ
.
Cognitive function in patients with diabetes mellitus: guidance for daily care
.
Lancet Neurol
2015
;
14
:
329
340
2.
Luchsinger
JA
,
Reitz
C
,
Patel
B
,
Tang
MX
,
Manly
JJ
,
Mayeux
R
.
Relation of diabetes to mild cognitive impairment
.
Arch Neurol
2007
;
64
:
570
575
3.
Luchsinger
JA
.
Type 2 diabetes and cognitive impairment: linking mechanisms
.
J Alzheimers Dis
2012
;
30
(
Suppl. 2
):
S185
S198
4.
Vos
SJ
,
van Rossum
IA
,
Verhey
F
, et al
.
Prediction of Alzheimer disease in subjects with amnestic and nonamnestic MCI
.
Neurology
2013
;
80
:
1124
1132
5.
Chatterjee
S
,
Mudher
A
.
Alzheimer’s disease and type 2 diabetes: a critical assessment of the shared pathological traits
.
Front Neurosci
2018
;
12
:
383
6.
McCrimmon
RJ
,
Ryan
CM
,
Frier
BM
.
Diabetes and cognitive dysfunction
.
Lancet
2012
;
379
:
2291
2299
7.
Li
C
,
Li
C
,
Yang
Q
, et al
.
Cortical thickness contributes to cognitive heterogeneity in patients with type 2 diabetes mellitus
.
Medicine (Baltimore)
2018
;
97
:
e10858
8.
Li
W
,
Risacher
SL
,
Huang
E
,
Saykin
AJ
;
Alzheimer’s Disease Neuroimaging Initiative
.
Type 2 diabetes mellitus is associated with brain atrophy and hypometabolism in the ADNI cohort
.
Neurology
2016
;
87
:
595
600
9.
Ma
F
,
Wu
T
,
Miao
R
,
Xiao
YY
,
Zhang
W
,
Huang
G
.
Conversion of mild cognitive impairment to dementia among subjects with diabetes: a population-based study of incidence and risk factors with five years of follow-up
.
J Alzheimers Dis
2015
;
43
:
1441
1449
10.
Zhang
J
,
Wang
Y
,
Wang
J
, et al
.
White matter integrity disruptions associated with cognitive impairments in type 2 diabetic patients
.
Diabetes
2014
;
63
:
3596
3605
11.
Hoogenboom
WS
,
Marder
TJ
,
Flores
VL
, et al
.
Cerebral white matter integrity and resting-state functional connectivity in middle-aged patients with type 2 diabetes
.
Diabetes
2014
;
63
:
728
738
12.
Reijmer
YD
,
Brundel
M
,
de Bresser
J
,
Kappelle
LJ
,
Leemans
A
,
Biessels
GJ
;
Utrecht Vascular Cognitive Impairment Study Group
.
Microstructural white matter abnormalities and cognitive functioning in type 2 diabetes: a diffusion tensor imaging study
.
Diabetes Care
2013
;
36
:
137
144
13.
Tan
X
,
Fang
P
,
An
J
, et al
.
Micro-structural white matter abnormalities in type 2 diabetic patients: a DTI study using TBSS analysis
.
Neuroradiology
2016
;
58
:
1209
1216
14.
Li
H
,
Liang
Y
,
Chen
K
, et al
.
Different patterns of white matter disruption among amnestic mild cognitive impairment subtypes: relationship with neuropsychological performance
.
J Alzheimers Dis
2013
;
36
:
365
376
15.
Zhang
MY
,
Katzman
R
,
Salmon
D
, et al
.
The prevalence of dementia and Alzheimer’s disease in Shanghai, China: impact of age, gender, and education
.
Ann Neurol
1990
;
27
:
428
437
16.
Schmidt
M
.
Rey Auditory Verbal Learning Test A Handbook
.
Los Angeles, CA
,
Western Psychological Services
,
1996
17.
Rey
A
.
L’examen Psychologique dans les cas d’encephathie traumatique
.
Arch Psychol
1941
;
28
:
286
340
18.
Dai
XY
,
Ryan
JJ
,
Paolo
AM
,
Harrington
RG
.
Sex differences on the Wechsler Adult Intelligence Scale—revised for China
.
Psychol Assess
1991
;
3
:
282
284
19.
Rouleau
I
,
Salmon
DP
,
Butters
N
,
Kennedy
C
,
McGuire
K
.
Quantitative and qualitative analyses of clock drawings in Alzheimer’s and Huntington’s disease
.
Brain Cogn
1992
;
18
:
70
87
20.
Gordon
NG
.
The Trail Making Test in neuropsychological diagnosis
.
J Clin Psychol
1972
;
28
:
167
169
21.
Guo
Q
,
Hong
Z
,
Shi
WX
,
Lu
CZ
.
Boston naming test using by Chinese elderly, patient with mild cognitive impairment and Alzheimer’s dementia
.
Chin Ment Health J
2006
;
20
:
81
84
22.
Petersen
RC
.
Mild cognitive impairment as a diagnostic entity
.
J Intern Med
2004
;
256
:
183
194
23.
Chang
P
,
Li
X
,
Ma
C
, et al
.
The effects of an APOE promoter polymorphism on human white matter connectivity during non-demented aging
.
J Alzheimers Dis
2017
;
55
:
77
87
24.
Smith
SM
,
Jenkinson
M
,
Woolrich
MW
, et al
.
Advances in functional and structural MR image analysis and implementation as FSL
.
Neuroimage
2004
;
23
(
Suppl. 1
):
S208
S219
25.
Basser
PJ
,
Pierpaoli
C
.
Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI
.
J Magn Reson B
1996
;
111
:
209
219
26.
Budde
MD
,
Kim
JH
,
Liang
HF
, et al
.
Toward accurate diagnosis of white matter pathology using diffusion tensor imaging
.
Magn Reson Med
2007
;
57
:
688
695
27.
Smith
SM
,
Jenkinson
M
,
Johansen-Berg
H
, et al
.
Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data
.
Neuroimage
2006
;
31
:
1487
1505
28.
Mori
S
,
Oishi
K
,
Jiang
H
, et al
.
Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template
.
Neuroimage
2008
;
40
:
570
582
29.
Douaud
G
,
Smith
S
,
Jenkinson
M
, et al
.
Anatomically related grey and white matter abnormalities in adolescent-onset schizophrenia
.
Brain
2007
;
130
:
2375
2386
30.
Gong
G
,
He
Y
,
Concha
L
, et al
.
Mapping anatomical connectivity patterns of human cerebral cortex using in vivo diffusion tensor imaging tractography
.
Cereb Cortex
2009
;
19
:
524
536
31.
Basser
PJ
,
Mattiello
J
,
LeBihan
D
.
MR diffusion tensor spectroscopy and imaging
.
Biophys J
1994
;
66
:
259
267
32.
Umegaki
H
.
Type 2 diabetes as a risk factor for cognitive impairment: current insights
.
Clin Interv Aging
2014
;
9
:
1011
1019
33.
Umegaki
H
.
Pathophysiology of cognitive dysfunction in older people with type 2 diabetes: vascular changes or neurodegeneration
?
Age Ageing
2010
;
39
:
8
10
34.
Schubert
M
,
Brazil
DP
,
Burks
DJ
, et al
.
Insulin receptor substrate-2 deficiency impairs brain growth and promotes tau phosphorylation
.
J Neurosci
2003
;
23
:
7084
7092
35.
Zhao
L
,
Teter
B
,
Morihara
T
, et al
.
Insulin-degrading enzyme as a downstream target of insulin receptor signaling cascade: implications for Alzheimer’s disease intervention
.
J Neurosci
2004
;
24
:
11120
11126
36.
Squire
LR
.
Mechanisms of memory
.
Science
1986
;
232
:
1612
1619
37.
Craft
S
,
Watson
GS
.
Insulin and neurodegenerative disease: shared and specific mechanisms
.
Lancet Neurol
2004
;
3
:
169
178
38.
Qi
D
,
Wang
A
,
Chen
Y
, et al
.
Default mode network connectivity and related white matter disruption in type 2 diabetes mellitus patients concurrent with amnestic mild cognitive impairment
.
Curr Alzheimer Res
2017
;
14
:
1238
1246
39.
Xiong
Y
,
Sui
Y
,
Xu
Z
, et al
.
A diffusion tensor imaging study on white matter abnormalities in patients with type 2 diabetes using tract-based spatial statistics
.
AJNR Am J Neuroradiol
2016
;
37
:
1462
1469
40.
Forkel
SJ
,
Thiebaut de Schotten
M
,
Kawadler
JM
,
Dell’Acqua
F
,
Danek
A
,
Catani
M
.
The anatomy of fronto-occipital connections from early blunt dissections to contemporary tractography
.
Cortex
2014
;
56
:
73
84
41.
Lockhart
SN
,
Mayda
AB
,
Roach
AE
, et al
.
Episodic memory function is associated with multiple measures of white matter integrity in cognitive aging
.
Front Hum Neurosci
2012
;
6
:
56
42.
Sasson
E
,
Doniger
GM
,
Pasternak
O
,
Tarrasch
R
,
Assaf
Y
.
White matter correlates of cognitive domains in normal aging with diffusion tensor imaging
.
Front Neurosci
2013
;
7
:
32
43.
Wakana
S
,
Jiang
H
,
Nagae-Poetscher
LM
,
van Zijl
PC
,
Mori
S
.
Fiber tract-based atlas of human white matter anatomy
.
Radiology
2004
;
230
:
77
87
44.
Vergani
F
,
Mahmood
S
,
Morris
CM
,
Mitchell
P
,
Forkel
SJ
.
Intralobar fibres of the occipital lobe: a post mortem dissection study
.
Cortex
2014
;
56
:
145
156
45.
Kraus
MF
,
Susmaras
T
,
Caughlin
BP
,
Walker
CJ
,
Sweeney
JA
,
Little
DM
.
White matter integrity and cognition in chronic traumatic brain injury: a diffusion tensor imaging study
.
Brain
2007
;
130
:
2508
2519
46.
Kier
EL
,
Staib
LH
,
Davis
LM
,
Bronen
RA
.
MR imaging of the temporal stem: anatomic dissection tractography of the uncinate fasciculus, inferior occipitofrontal fasciculus, and Meyer’s loop of the optic radiation
.
AJNR Am J Neuroradiol
2004
;
25
:
677
691
47.
Owens
JA
,
Spitz
G
,
Ponsford
JL
,
Dymowski
AR
,
Willmott
C
.
An investigation of white matter integrity and attention deficits following traumatic brain injury
.
Brain Inj
2018
;
32
:
776
783
48.
Carter
SF
,
Embleton
KV
,
Anton-Rodriguez
JM
,
Burns
A
,
Ralph
MA
,
Herholz
K
.
Regional neuronal network failure and cognition in late-onset sporadic Alzheimer disease
.
AJNR Am J Neuroradiol
2014
;
35
(
Suppl.
):
S18
S30
49.
Sun
X
,
Liang
Y
,
Wang
J
, et al
.
Early frontal structural and functional changes in mild white matter lesions relevant to cognitive decline
.
J Alzheimers Dis
2014
;
40
:
123
134
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Supplementary data