Factors associated with brain white matter damage in type 2 diabetes mellitus: a tract-based spatial statistics study

Abstract

Background

The pathogenesis and related factors of central nervous system abnormality in patients with type 2 diabetes mellitus (T2DM) have always been the focus of clinical research.

Purpose

To compare and analyze the area of white matter (WM) damage in patients with T2DM based on their level of hemoglobin A_1C (HBA_1c) and discuss any related factors.

Material and Methods

Based on their levels of HBA_1c, 87 patients with T2DM were divided into three groups (Group B, C, or D), of which 29 non-diabetic volunteers served as the control group (Group A). DTI data analysis was based on tract-based spatial statistics (TBSS). The obtained parameters were compared among each group and the relevant clinical factors were analyzed.

Results

For age, sex, mini-mental state examination (MMSE), and Montreal Cognitive Assessment (MoCA) scores, there were no statistically significant differences among groups. For fractional anisotropy (FA) and radial diffusivity (RD) of WM, there were statistically significant differences (P < 0.05, two-tailed, FWE corrected) in the local area of corpus callosum, corona radiate, superior longitudinal fasciculus, etc. Most of these were significantly correlated with body mass index (BMI), left systolic blood pressure (SBP_L), and β₂ microglobulin.

Conclusion

Before the cognitive function was obviously impaired, abnormalities of FA and RD had been found in the corpus callosum, corona radiate, and upper fasciculus in patients with T2DM, which suggested that the damage mainly occurred in the myelin sheath of WM and may be related to systemic vascular damage.

Keywords

Type 2 diabetes mellitus magnetic resonance diffusion tensor imaging tract-based spatial statistics white matter

Introduction

The worldwide incidence of type 2 diabetes mellitus (T2DM) is very high and continues to grow rapidly, accounting for 9.1% of the adult population, which has aroused wide concern globally (1). Although T2DM is different from Alzheimer's disease (AD) in its pathological mechanism, it also has a higher risk of cognitive impairment, and a large amount of evidence indicates that it may be related to vascular pathogenesis (2). At the same time, the incidence and mortality of stroke in patients with T2DM are also significantly increased (3). Therefore, the pathogenesis and related factors of central nervous system abnormality caused by T2DM have always been the focus of clinical and research. As an important part of the central nervous system, white matter (WM) is an important pathway connecting various functional structures of the brain. The correct interpretation of WM abnormalities has become one of the key problems in the pathogenesis of T2DM central nervous system (4). Diffusion tensor imaging (DTI) is the non-invasive imaging method that can show the integrity and functional connection of nerve fibers in vivo, and can be used to study quantitatively the damage degree of WM fibers (5). Trace-based spatial statistics (TBSS) is the most stable and mature quantitative analysis technique of DTI data at present (6). Hemoglobin A_1C (HBA_1c) is a commonly used clinical parameter to evaluate whether blood glucose is well controlled (7). At present, there are few studies evaluating WM changes in T2DM using TBSS, and HBA_1c is not used as a grouping basis in any of them (8). The aim of the present study was to compare and analyze areas of WM damage in patients with T2DM, grouped by level of HBA_1c, and then explore their related factors.

Material and Methods

General information

From December 2019 to August 2020, 87 patients with T2DM (61 men, 26 women; mean age = 59.59 ± 9.35 years) and 29 non-diabetic volunteers (20 men, 9 women; mean age = 59.79 ± 9.12 years) were included.

Inclusion criteria were as follows: (i) all individuals were right-handed; (ii) all patients with T2DM met the American Diabetes Association's criteria for clinical diagnosis of diabetes; and (iii) none of the non-diabetic volunteers had a family history of diabetes and fasting blood glucose was <6.1 mmol/L, and HBA_1c was <6.0% (42 mmol/mol). The study was approved by the medical ethics committee. We complied with the patient consent guidelines and obtained informed consent from all participants.

Exclusion criteria were as follows: (i) traumatic brain injury, craniocerebral surgery, cerebral infarction, cerebral hemorrhage, brain tumor, or other neurological and psychiatric diseases that are not related to diabetes; (ii) non-diabetic macrovascular disease, malignant tumor, or other systemic multisystem disease; (iii) contraindications to magnetic resonance imaging (MRI) examination; (iv) pregnant and lactating women; and (v) unable to cooperate in collecting complete research data.

Group setting of participants

Based on their levels of HBA_1C, 87 patients with T2DM were divided into three groups (Group B, C, or D), of which 29 non-diabetic volunteers served as the control group (Group A). The level of HBA_1c was used as the basis for grouping diabetics. Group B was <7%, Group D was ≥8%, and group C was between the two groups.

Examination equipment and scanning parameters

Images were acquired on a 3.0-T MRI system (SIGNA Pioneer; General Electric, Milwaukee, WI, USA) equipped with a 29-element DST Head Neck Unit coil. Tight but comfortable foam padding was used to minimize head motion, and earplugs were used to reduce scanner noise. T2-weighted (T2W) images were collected using fast spin echo sequence (TR = 3400 ms, TE = 85 ms, flip angle = 90°, field of view [FOV] = 256 × 256 mm, matrix = 256 × 256, layer thickness = 3 mm, interval = 1 mm. T1-weighted (T1W) imaging adopted the BRAVO sequence (TR/TE/TI = 8.15/3.17/450 ms, number of excitations [NEX] = 1, flip angle = 12°, FOV = 256 × 256 mm, matrix = 256 × 256, layer thickness = 1 mm (without gap), layer number = 188. DTI scanning parameters were as follows: TR/TE = 8000/102 ms; FOV = 256×256 mm; matrix = 128 × 128; layer thickness = 3 mm; layer spacing = 0 mm; layer number = 48; B value = 1000 s/mm²; and 64 dispersion gradient directions.

Data processing

Fazekas scores were used to assess WM lesions on T2W imaging (9). Periventricular hyperintensity (PVH) was graded as follows: 0 = absence; 1 = “caps” or pencil-thin lining; 2 = smooth “halo”; and 3 = irregular PVH extending into the deep WM. Separate deep white matter hyperintensities (DWMH) signals were rated as follows: 0 = absence; 1 = punctate foci; 2 = beginning confluence of foci; and 3 = large confluent areas.

DTI data analysis was based on tract-based spatial statistics (TBSS) (10), and the main parameters included fractional anisotropy (FA), axial diffusivity (AD), radial diffusivity (RD), and mean diffusivity (MD). DTI data were preprocessed according to the pipelines of the FMRIB's Diffusion Toolbox (FDT) implemented in FSL 5.0.10 (FDT tool, FSL 5.0.10, http://www.fmrib.ox.ac.uk). First, the synb0-Disco Program with volumetric T1 image was used to correct the image (11). Then, brain tissue was extracted using brain extract toolbox (version 2.1). The diffusion tensor was fitted using a linear least square algorithm and eigenvalues were decomposed from the tensor to calculate diffusion metrics of the FA, AD, RD, and MD (12).

The FA images of all individuals were aligned to a template of averaged FA images (FMRIB-58) in the Montreal Neurological Institute (MNI) space using a non-linear registration algorithm implemented in FNIRT (FMRIB's Non-linear Registration Tool). After transforming into the MNI space, a mean FA image was created and thinned to generate a mean FA skeleton of the WM tracts. The FA image of each individual was then projected onto the skeleton via filling the mean FA skeleton with FA values from the nearest relevant tract center by searching perpendicular to the local skeleton structure for maximum FA value. The registration and projection information derived from the FA analysis were then applied to the other three parametric images (AD, RD, and MD) of each individual to ensure an exact spatial correspondence of the different parameters.

To facilitate visualization, results were thickened using the tbss_fill script implemented in FSL. Referring to the Johns Hopkins University ICBM-DTI-81 White-Matter Labels provided in the FSL toolbox, fiber tracts corresponding to the clusters were identified. Using the standard template of JHU-ICBM-labels for WM fiber tracts, the parameter values of each diffusion tensor in the differential brain area were extracted.

Statistical analysis

The clinical data and DTI data of these four groups were compared and analyzed correlatively. All quantitative values are presented as the mean ± standard deviation (SD). Analysis of variance (ANOVA) was used to test the parameters among multiple groups, followed by the LSD test (Adjusted Tamhane t-test was made if there is heterogeneity of variance of interclass) for multiple comparisons. The Kruskal–Wallis H test was used for the multi-group comparison of non-parametric tests. Pearson correlation analysis was used to evaluate the diffusion tensor indicators and quantitative clinical parameters in the brain regions with differences between groups, and Spearman correlation analysis was used to evaluate the correlation between them and qualitative clinical parameters. The above data were analyzed using the SPSS 19.0 C (SPSS Inc., Chicago, IL, USA) statistical software package. P < 0.05 indicated that differences were statistically significant.

Voxel-wise statistical analysis across individuals on the skeleton space was carried out using a permutation-based inference tool for nonparametric statistic (“randomize,” part of FSL). Comparisons between groups were performed using the F test in the general linear model, with age and gender as unrelated covariables. The mean FA skeleton was used as a mask, and the number of permutations was set to 5000. To simultaneously control for both type I and type II errors, the significance threshold was determined to be a P < 0.05 (two-tailed) after correcting the family-wise error (FWE) using the threshold-free cluster enhancement (TFCE) option in FSL (13). After using post hoc Bonferroni correction, the t-test in the general linear model was carried out between each pair of groups, and the WM regions with differences between each group were then extracted by the above processes and displayed.

Results

Comparison and correlation analysis of clinical characteristic parameters

We selected 25 clinical characteristic parameters that may be associated with T2DM, including Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scores, and were compared between groups ( Table 1 ). For age, sex, and MMSE and MoCA scores, there were no statistically significant differences between groups. Among the nine clinical characteristic parameters with differences between groups, except for blood glucose (BG) and HBA1c, only the P value of the body mass index (BMI) was <0.01. However, these differences only existed between the control group and all or part of the T2DM groups, but the differences between the T2DM groups were not significant. In addition to the two parameters of BG, weight, BMI, and left systolic blood pressure (SBP_L) are correlated with HBA1c, respectively. In addition, there was a correlation between Smoking and Fasting BG (P < 0.05), and β₂ microglobulin was significantly correlated with BMI and SBP_L, respectively (P < 0.01) ( Table 2 ).

Table 1.

Comparison of clinical characteristic parameters of diabetes mellitus between groups.

	A group (n = 29)	B group (n = 26)	C group (n = 28)	D group ( n =33)	ANOVA (P value)
Age (years)	59.79 ± 9.12	60.96 ± 10.36	61.89 ± 10.06	56.55 ± 7.06	0.116
Gender, male/female (n)	20/9	17/9	20/8	24/9	0.937
Weight (kg)	66.14 ± 9.53^b,c,d	72.85 ± 12.29^a	74.50 ± 12.75^a	74.48 ± 11.15^a	0.017*
BMI (kg/m²)	23.25 ± 2.61^b,c,d	25.85 ± 3.33^a	25.66 ± 3.34^a	25.91 ± 3.17^a	0.003^†
Diabetes duration (years)	N/A	13.60 ± 7.82	13.61 ± 7.58	13.79 ± 6.58	0.993
SBP_R (mmHg)	131.69 ± 17.77	136.62 ± 22.88	146.25 ± 24.13	139.12 ± 16.95	0.064
DBP_R (mmHg)	74.52 ± 11.42	70.38 ± 10.58	73.29 ± 9.97	74.61 ± 9.08	0.389
SBP_L (mmHg)	130.24 ± 15.86^c	137.35 ± 23.89	147.89 ± 25.44^a	138.64 ± 18.34	0.021*
DBP_L (mmHg)	72.97 ± 9.77	70.38 ± 11.54	73.18 ± 10.21	73.82 ± 8.91	0.603
Fasting BG (mmol/L)	5.06 ± 0.37^b,c,d	6.74 ± 1.34^a,c,d	8.65 ± 2.55^a,b,d	11.15 ± 4.12^a,b,c	<0.001^†‡
BG before MRI (mmol/L)	6.38 ± 1.09^b,c,d	8.23 ± 3.13^a,c,d	11.16 ± 4.10^a,b	14.12 ± 4.72^a,b	<0.001^†‡
HBA1c (%)	5.58 ± 0.19^b,c,d	6.40 ± 0.27^a,c,d	7.41 ± 0.28^a,b,d	9.44 ± 1.45^a,b,c	<0.001^†‡
Fasting insulin (μIU/mL)	8.53 ± 4.06	11.19 ± 7.93	19.38 ± 24.84	10.45 ± 11.67	0.027*^‡
Fasting C-Peptide (ng/mL)	2.12 ± 0.77	2.23 ± 1.37	2.34 ± 1.15	1.83 ± 0.98	0.281
Total cholesterol (mmol/L)	4.95 ± 0.72	4.91 ± 0.90	4.98 ± 1.33	5.27 ± 1.00	0.476
LDL-cholesterol (mmol/L)	2.76 ± 0.62	2.44 ± 0.59	2.65 ± 0.99	2.75 ± 0.69	0.346
HDL-cholesterol (mmol/L)	1.57 ± 0.26	1.49 ± 0.33	1.40 ± 0.31	1.41 ± 0.31	0.129
Triglycerides (mmol/L)	1.23 ± 0.54	1.95 ± 2.56	1.67 ± 0.98	2.33 ± 2.70	0.176
Current smoking	5 (17.24)^b,d	12 (46.15)^a	9 (32.14)	16 (48.48)^a	0.048*
Current drinking	11 (37.93)	12 (46.15)	11 (39.29)	20 (60.60)	0.256
Blood urea nitrogen (mmol/L)	5.83 ± 1.38	6.01 ± 1.07	6.42 ± 1.33	5.91 ± 1.48	0.354
Creatinine (μmol/L)	72.71 ± 12.57	73.00 ± 15.49	74.23 ± 15.69	67.44 ± 14.31	0.263
β2-MG (μg/mL)	1.46 ± 0.58^b,c	1.86 ± 0.50^a	1.88 ± 0.66^a	1.62 ± 0.60	0.023*
MMSE	27.10 ± 2.85	27.54 ± 1.86	27.61 ± 1.81	26.67 ± 2.50	0.364
MoCA	24.21 ± 4.15	24.42 ± 3.68	24.57 ± 3.53	23.27 ± 3.90	0.541

Values are given as n (%) or mean ± SD. The data marked ^a, ^b, ^c, or ^d indicate significant differences between the corresponding groups, P < 0.05.

The confidence standard is P < 0.05 and the correlation is significant.

^†

The confidence standard is P < 0.01 and the correlation is significant.

^†

Adjusted Tamhane t-test was used due to variances being not aligned.

^β

2-MG, β2-microglobulin; BMI, body mass index; DBP, diastolic blood pressure; MMSE, mini mental state examination; MoCA, Montreal Cognitive Assessment; N/A, not applicable; SBP, systolic blood pressure.

Table 2.

Correlation analysis between clinical characteristic parameters.

	Weight		BMI		SBP_L		Fasting BG		BG before MRI		HBA1c (%)		Fasting insulin		Current smoking*		β₂-MG
	r	P	r	P	r	P	r	P	r	P	r	P	r	P	r	P	r	P
Weight	1	N/A	0.830	<0.001**	0.158	0.090	0.271	0.003^†	0.220	0.018^‡	0.270	0.003^†	0.125	0.181	0.339	<0.001**	0.037	0.692
BMI	0.830	<0.001**	1	N/A	0.265	0.004^†	0.272	0.003^†	0.222	0.017^‡	0.328	<0.001**	0.239	<0.001**	0.169	0.069	0.192	0.039^‡
SBP_L	0.158	0.090	0.265	0.004^‡	1	N/A	0.193	0.038^‡	0.180	0.053	0.196	0.035^‡	0.201	0.030^‡	0.030	0.747	0.370	<0.001**
Fasting BG	0.271	0.003^†	0.272	0.003^†	0.193	0.038^‡	1	N/A	0.674	<0.001**	0.780	<0.001**	0.157	0.092	0.191	0.040^‡	0.102	0.274
BG before MRI	0.220	0.018^‡	0.180	0.053	0.180	0.053	0.674	<0.001**	1	N/A	0.712	<0.001**	0.134	0.151	0.092	0.327	0.140	0.135
HBA1c (%)	0.270	0.003^†	0.196	0.035^‡	0.196	0.035^‡	0.780	<0.001**	0.712	<0.001**	1	N/A	0.076	0.417	0.161	0.084	0.084	0.368
Fasting insulin	0.125	0.181	0.201	0.030^‡	0.201	0.030^‡	0.157	0.092	0.134	0.151	0.076	0.417	1	N/A	−0.101	0.280	0.201	0.030^‡
Current smoking^#	0.339	<0.001**	0.169	0.069	0.030	0.747	0.191	0.040^‡	0.092	0.327	0.161	0.084	−0.101	0.280	1	N/A	0.017	0.860
β₂ microglobulin	0.037	0.692	0.370	<0.001**	0.370	<0.001**	0.102	0.274	0.140	0.135	0.084	0.368	0.201	0.030^‡	0.017	0.860	1	N/A

“r” is the correlation coefficient.

Spearman non-parametric test.

^†

The confidence standard is P < 0.01 and the correlation is significant.

^‡

The confidence standard is P < 0.05 and the correlation is significant.

^β

2-MG, β2-microglobulin; BG, blood glucose; BMI, body mass index; N/A, not applicable; SBP, systolic blood pressure.

Comparison and correlation analysis of white matter parameters

Fazekas scores were used to assess WM lesions on T2W imaging, and there was no significant difference among groups ( Table 3 ).

Table 3.

Fazekas scores of white matter were compared and analyzed among each group.

Fazekas scores	0	1	2	3	4	5	Total cases
Group A	0	5	19	3	2	0	29
Group B	0	4	16	3	3	0	26
Group C	0	2	16	6	3	1	28
Group D	1	5	21	3	2	1	33
Total cases	1	16	72	15	10	2	116

Values are given as n. Kruskal–Wallis was adopted to compare analysis among groups, and the confidence standard is P < 0.05 (X² = 4.127; P = 0.248).

After TBSS analysis of DTI images, some major parameters showed significant differences among groups in the WM regions ( Table 4 ). For the FA and RD of WM, there were statistically significant differences (P < 0.05, two-tailed, FWE corrected) in the body of the corpus callosum, bilateral corona radiate, and right superior longitudinal fasciculus. Most of these were significantly correlated with BMI, SBPL), and β₂ microglobulin ( Table 5 ). The post hoc analysis showed that the FA values of local WM in groups B, C, and D were lower than those in group A, mainly in the right superior longitudinal fasciculus and body of corpus callosum. The FA value in the local area of the right superior longitudinal fasciculus was lower in group C and D than in group B, respectively ( Fig. 1 and Table 6 ). For RD value, groups B, C and D were larger than that in group A in the local area of right superior longitudinal fasciculus, body of corpus callosum, and anterior corona radiata, and group B was larger than C and D in the local area of right superior longitudinal fasciculus ( Fig. 2 and Table 6 ).

Fig. 1.

After using post hoc Bonferroni correction, t-tests were used to compare FA values between each pair of groups: A-B means group A is greater than group B, A-C means group A is greater than group C, A-D means group A is greater than group D, C-B means group C is greater than group B, and D-B means group D is greater than group B. These positive white matter areas are shown as red-yellow. Below the image are the x, y, and z coordinates of the MIN template. FA, fractional anisotropy.

Fig. 2.

After using post hoc Bonferroni correction, t-tests were used to compare RD values between each pair of groups: B-A means group B is greater than group A, B-C means group B is greater than group C, B-D means group B is greater than group D, C-A means group C is greater than group A, and D-A means group D is greater than group A. These positive white matter areas are shown as red-yellow. Below the image are the x, y, and z coordinates of the MIN template. RD, radial diffusivity.

Table 4.

Index parameters of DTI were compared and analyzed among each group.

DTI Index	Anatomical locations	Group A (n = 29)	Group B (n = 26)	Group C (n = 28)	Group D (n = 33)	ANOVA (P value)
FA	04	0.60 ± 0.06^b,c,d	0.52 ± 0.07^a,d	0.54 ± 0.06^a	0.55 ± 0.06^a,b	<0.001*
	24	0.50 ± 0.05^b,c,d	0.44 ± 0.06^a	0.44 ± 0.06^a	0.46 ± 0.06^a	<0.001*
	26	0.52 ± 0.05^b,c,d	0.46 ± 0.07^a	0.46 ± 0.06^a	0.48 ± 0.06^a	<0.001*
	41	0.59 ± 0.04^b,d	0.53 ± 0.05^a,c,d	0.58 ± 0.05^b	0.56 ± 0.04^a,b	<0.001*
RD	04	4.52E-04 ± 6.11E-05^b,c,d	5.30E-04 ± 7.82E-05^a,d	5.06E-04 ± 6.24E-05^a	4.95E-04 ± 6.32E-05^a,b	<0.001*
	24	4.92E-04 ± 6.15E-05^b,c	5.49E-04 ± 6.59E-05^a	5.55E-04 ± 5.85E-05^a,d	5.18E-04 ± 5.60E-05^c	<0.001*
	25	4.57E-04 ± 6.32E-05^b,c,d	5.28E-04 ± 6.76E-05^a	5.49E-04 ± 7.70E-05^a,d	4.97E-04 ± 7.44E-05^a,c	<0.001*
	26	4.85E-04 ± 5.56E-05^b,c,d	5.48E-04 ± 6.69E-05^a	5.41E-04 ± 6.13E-05^a	5.20E-04 ± 5.82E-05^a	0.001*
	41	4.20E-04 ± 2.80E-05^b,c,d	4.83E-04 ± 4.79E-05^a,c,d	4.42E-04 ± 4.13E-05^a,b	4.50E-04 ± 4.54E-05^a,b	<0.001*

The data marked ^a, ^b, ^c, or ^d indicate significant differences between the corresponding groups, P < 0.05. Anatomical locations are as follows: 04 = body of corpus callosum; 24 = anterior corona radiate L; 25 = superior corona radiate R; 26 = superior corona radiate L; 41 = superior longitudinal fasciculus R.

The confidence standard is P < 0.01 and the correlation is significant.

^β

2-MG, β2-microglobulin; BG, blood glucose; BMI, body mass index; FA, fractional anisotropy; N/A, not applicable; RD, radial diffusivity; SBP, systolic blood pressure.

Table 5.

Correlation analysis between index parameters of DTI and clinical characteristic parameters.

DTI Index	Anatomical locations	Weight		BMI		SBP_L		Fasting BG		BG before MRI		HBA1c (%)		Fasting insulin		Current smoking*		β₂-MG
DTI Index	Anatomical locations	r	P	r	P	r	P	r	P	r	P	r	P	r	P	r	P	r	P
FA	04	−0.117	0.209	−0.190	0.041^†	−0.276	0.003^‡	−0.121	0.194	−0.057	0.544	−0.126	0.178	−0.061	0.513	−0.053	0.572	−0.264	0.004^‡
	24	−0.066	0.482	−0.158	0.091	−0.427	<0.001^‡	−0.104	0.268	−0.137	0.142	−0.167	0.073	−0.061	0.514	−0.077	0.410	−0.215	0.020^†
	26	−0.057	0.547	−0.140	0.135	−0.334	<0.001^‡	−0.173	0.064	−0.098	0.297	−0.144	0.124	−0.027	0.774	−0.073	0.434	−0.294	0.001^‡
	41	0.081	0.386	−0.132	0.159	−0.060	0.524	−0.029	0.775	−0.039	0.677	−0.124	0.184	−0.081	0.386	0.062	0.511	−0.164	0.078
RD	04	0.162	0.082	0.238	0.010^†	0.354	<0.001^‡	0.154	0.099	0.059	0.528	0.141	0.132	0.111	0.235	0.140	0.133	0.289	0.002^‡
	24	0.147	0.116	0.221	0.017^†	0.496	<0.001^‡	0.044	0.639	0.079	0.398	0.097	0.299	0.144	0.124	0.142	0.128	0.212	0.022^†
	25	0.103	0.272	0.230	0.013^†	0.371	<0.001^‡	0.082	0.379	−0.007	0.939	0.120	0.201	0.056	0.553	0.072	0.444	0.258	0.005^‡
	26	0.083	0.377	0.187	0.045^†	0.398	<0.001^‡	0.137	0.142	0.056	0.548	0.114	0.224	0.118	0.206	0.102	0.274	0.290	0.002^‡
	41	−0.003	0.974	0.199	0.032^†	0.169	0.070	0.066	0.484	0.058	0.534	0.153	0.100	0.059	0.528	0.051	0.587	0.211	0.023^†

“r” is the correlation coefficient. Anatomical locations are as follows: 04 = body of corpus callosum; 24 = anterior corona radiate L; 25 = superior corona radiate R; 26 = superior corona radiate L; 41 = superior longitudinal fasciculus R.

Spearman non-parametric test.

^†

The confidence standard is P < 0.05 and the correlation is significant.

^‡

The confidence standard is P < 0.01 and the correlation is significant.

^β

2-MG, β2-microglobulin; BG, blood glucose; BMI, body mass index; FA, fractional anisotropy; N/A, not applicable; RD, radial diffusivity; SBP, systolic blood pressure.

Table 6.

The location and size of the different brain regions among groups.

DTI parameters	Intergroup contrast	Anatomical locations (JHU ICBM-DTI-81 White-Matter Labels)	X	Y	Z	P value	Voxel size
FA	A-B	Superior longitudinal fasciculus R	37	–25	30	<0.001	114
		Body of corpus callosum	–16	11	30	<0.001	63
		Body of corpus callosum	–13	–1	32	0.014	5
	A-C	Anterior corona radiata L	–16	18	30	0.001	63
		Superior longitudinal fasciculus R	35	–24	36	0.006	39
	A-D	Superior longitudinal fasciculus R	34	–25	35	0.001	75
		Body of corpus callosum	–16	11	30	0.001	62
	C-B	Superior longitudinal fasciculus R	37	–21	33	<0.001	89
	D-B	Superior longitudinal fasciculus R	39	–22	34	0.006	57
RD	B-C	Superior longitudinal fasciculus R	37	–21	34	<0.001	118
	B-D	Superior longitudinal fasciculus R	37	–23	30	0.001	113
		Superior longitudinal fasciculus R	31	–32	38	0.038	11
		Unclassified	31	–37	37	0.048	4
	B-A	Superior longitudinal fasciculus R	37	–23	29	<0.001	245
		Body of corpus callosum	–15	10	30	<0.001	78
	C-A	Anterior corona radiata L	–16	19	30	0.004	69
		Unclassified	42	–21	34	0.009	69
		Superior corona radiata R	18	–9	39	0.033	4
		Unclassified	34	–12	39	0.049	2
	D-A	Body of corpus callosum	–16	11	30	0.003	66
		Superior longitudinal fasciculus R	34	–25	36	0.002	38
		Superior longitudinal fasciculus R	35	–35	33	0.012	34
		Superior longitudinal fasciculus R	35	–15	35	0.025	29
		Unclassified	30	–40	36	0.039	8

The significance threshold was determined with P < 0.05 (two-tailed) after correcting for family-wise error (FWE) using the threshold-free cluster enhancement (TFCE). Voxel 1 is 1 mm in size.

FA, fractional anisotropy; RD, radial diffusivity.

Discussion

T2DM is a common metabolic disease characterized by chronic hyperglycemia, which can lead to complications of multiple organ systems, and brain tissue is the target tissue of diabetic organ damage (14). The chronic damage of T2DM to the brain function of patients has been a highly concerned problem (15). As WM is an important part of the central nervous system, the correct interpretation of the pathogenesis of WM lesions in patients with T2DM has become one of the key issues (16,17). The WMH is considered a sign of the change of cerebral small vessel disease (CSVD), which is related to the disease of diabetic microvascular disease. Many studies have supported the expansion of WMH in patients with T2DM and insulin resistance, HBA_1c and fasting insulin are significantly related to WMH (18). This study was based on the level of HBA_1c, using Fazekas scores to evaluate the WM of the deep brain, and the difference among each group was not statistically significant. However, using the TBSS analysis method, the FA and RD values of WM in the corpus callosum, corona radiate, and superior longitudinal fascicles were found to be statistically significant among the groups, and were significantly correlated with BMI, SBP_L, and β₂ microglobulin.

Glycosylated hemoglobin (GHb) is the combination of hemoglobin in red blood cells and sugars in serum (19). It is formed by slow continuous and irreversible saccharification. The amount depends on the blood sugar concentration and the contact time between blood sugar and hemoglobin and has no relation with the time of drawing blood, fasting status, insulin use, and other factors. Therefore, GHb can effectively reflect the average blood glucose level of patients with T2DM in the past 8–12 weeks. GHb is composed of HBA_1a, HBA_1b, and HBA_1c, of which HBA_1c accounts for about 70% and has a stable structure. Therefore, HBA1c has always been recognized as the core index for evaluating the treatment effect of T2DM (7,20). In clinical practice, HbA_1c <7.0% (53 mmol/mol) is generally considered the control index for many non-pregnant adults with T2DM, but when patients with T2DM have a history of severe hypoglycemia or complications of vascular damage, the control index of HBA_1c is relaxed to <8.0% (64 mmol/mol) (7). Therefore, in this study, the HBA_1c levels of 7.0% (53 mmol/mol) and 8.0% (64 mmol/mol) were used as the boundary values for grouping diabetics, respectively.

A comparison of 25 clinical characteristic parameters that may be related to T2DM was conducted among each group. The results showed that, except for two indicators of blood glucose, some indicators showed significant differences between the T2DM groups and the non-diabetic control group, but no significant differences between the T2DM groups. In the correlation analysis, in addition to the two indicators of blood glucose, HBA_1c was also correlated with weight, BMI, and SBP_L. Although the correlation between β₂-BG and HBA_1c was not obvious, it was correlated with BMI and SBP_L, respectively. In the study, some obvious abnormal areas of WM were found in patients with T2DM compared with those without diabetes, but few were found between the T2DM groups. FA and RD values of WM regions with differences between groups were extracted, most of which were not significantly correlated with HBA_1c, but were significantly correlated with BMI, SBP_L, andβ₂-BG. There were no significant correlations between these parameters and other blood glucose indicators (fasting BG and BG before MRI) and fasting insulin, respectively. First, an increase in BMI often indicates abnormal lipid metabolism, increasing the risk of atherosclerosis (21). Second, the increase of blood pressure is closely related to the vascular damage caused by arteriosclerosis (22). In addition, β₂-BG is an independent risk factor associated with the severity of peripheral artery disease (PAD) (23,24). All these abnormalities indicated that patients with T2DM have developed vascular damage complications to a certain extent. Therefore, the main cause of the damage of T2DM WM may be the damage of vascular system, rather than the direct cause of hyperglycemia. There is evidence that microvascular dysfunction is one of the key underlying mechanisms that is common in T2DM and began before the onset of T2DM (25). Microvessels are involved in the regulation of many brain functions and are mainly driven by a combination of factors such as high blood sugar, obesity, insulin resistance, and hypertension (26,27). A growing body of research has confirmed that diabetes-related microvascular dysfunction is associated with a higher risk of stroke, cognitive dysfunction, and depression. The prognosis of T2DM brain changes can be improved by treating T2DM disruption of microcirculation (15).

DTI is a highly sensitive, non-invasive magnetic resonance imaging technique that detects abnormalities in the microstructure of WM (28). FA is the most commonly used parameter for anisotropy analysis. Abnormal FA values in cerebral WM usually indicate changes in myelin integrity, nerve fiber density, and consistency of travel, but the specificity of distinguishing different pathological changes is not high (29). MD reflects the overall situation of molecular dispersion level and resistance. MD only indicates the size of the diffusion and is independent for the direction of diffusion. The abnormality of MD generally indicates the enlargement of the pericellular space (30). Abnormal AD indicates possible axonal damage of nerve fibers, and abnormal RD indicates possible demyelination changes (31,32). According to literature reports, the areas of WM lesions in patients with T2DM are mainly in the corpus callosum, radiative corona, cingulate gyrus, supoccipital frontal tract, and visual radiation, with the decrease of FA and the increase of AD, RD, and MD (8). In the results of this study, abnormal WM regions were found in the FA and RD parametric images, with similar locations and sizes. No obvious abnormalities were found in the WM regions of the AD and MD images. Therefore, FA and RD are sensitive to the abnormal change of WM in diabetic patients, and the change of WM was only demyelinating change, but the change of neuraxons was not obvious.

The damage of T2DM to cognitive function of patients has been widely concerned, and elucidating its pathogenesis is still a major problem to be solved, which may be multifaceted (33). MMSE and MoCA are the two most commonly used methods to evaluate cognitive impairment in clinical or related studies (34,35). In order to ensure the accuracy of image registration, participants with a history of stroke or cerebrovascular lesions were excluded from this study, which may be the reason why there were no significant differences in MMSE and MoCA scores among the groups. Nevertheless, it was observed that MMSE and MoCA scores tended to decrease when HBA_1c was >8.0% (64 mmol/mol). At the same time, it was found that the WM damage in patients with T2DM mainly occurred in the corpus callosum, corona radiate, and superior longitudinal fasciculus, which were all closely related to cognitive function. The corpus callosum (CC) connects hemispheres of the brain and plays an important role in their coordination. Different parts of CC are involved in different functions, among which the back part of CC is more involved in visual processing, while the front part is related to cognitive function (36). Corona radiata is made up of ascending and descending nerve fibers that transmit information to and from the cerebral cortex, which is closely related to the work of cognitive control and attention (37). Superior longitudinal fasciculus (SLF) is the main associated fiber pathway connecting the frontal, parietal, and temporal regions, and is closely related to core cognitive functions such as language, attention, memory, and visuospatial skills (38). It can be seen from the results of this study that when there is no obvious impairment of cognitive function, abnormalities have been found in the WM of the corpus callosum, corona radialis, and superior longitudinal fasciculus. Studies have shown that damage to the WM of the brain occurs in pre-diabetes but does not occur due to the compensatory effect of the brain tissue itself (4).

In conclusion, before the obvious impairment of cognitive ability, abnormalities of the WM of patients with T2DM were found, mainly located in the corpus callosum, corona radiata, and superior longitudinal fasciculus, which were closely related to cognitive function. These changes were mainly demyelination changes, but the axonal itself did not change significantly. In T2DM, the WM damage was not significantly associated with blood glucose indicators such as HBA_1c, but was significantly correlated with BMI, SBP_L, andβ₂-BG, which reflected that the main cause of the damage of T2DM WM may be the damage of vascular system, rather than the direct cause of hyperglycemia.

Footnotes

Acknowledgements

The authors thank the Medical Imaging Laboratory of Tianjin Medical University General Hospital for the technical support of image processing, and especially thank Quan Zhang and Qin Wen for her help in the research process. They also thank their radiology colleagues at at Huabei Petroleum General Hospital for their research efforts.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Zhi-Jun Guo

References

Jia

. Diabetes in China: epidemiology and genetic risk factors and their clinical utility in personalized medication. Diabetes 2018;67:3–11.

Chornenkyy

Wang

Wei

, et al. Alzheimer's disease and type 2 diabetes mellitus are distinct diseases with potential overlapping metabolic dysfunction upstream of observed cognitive decline. Brain Pathol 2019;29:3–17.

Bell

Goncalves

. Stroke in the patient with diabetes (part 1) - epidemiology, etiology, therapy and prognosis. Diabetes Res Clin Pract 2020;164:108193.

Vergoossen L

Schram

de Jong

, et al. White matter connectivity abnormalities in prediabetes and type 2 diabetes: the Maastricht study. Diabetes Care 2020;43:201–208.

De Stefano

Giorgio

. Advanced MRI measures like DTI or fMRI should be outcome measures in future clinical trials - commentary. Mult Scler 2017;23:1458–1460.

Bach

Laun

Leemans

, et al. Methodological considerations on tract-based spatial statistics (TBSS). Neuroimage 2014;100:358–369.

Glycemic targets: standards of medical care in diabetes-2020. Diabetes Care 2020;43:S66–S76.

Alotaibi

Tench

Stevenson

, et al. Investigating brain microstructural alterations in type 1 and type 2 diabetes using diffusion tensor imaging: a systematic review. Brain Sci 2021;11:140.

Fazekas

Chawluk

Alavi

, et al. MR Signal abnormalities at 1.5 T in Alzheimer's Dementia and normal aging. Am J Neuroradiol 1987;8:421–426.

10.

, et al. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. NeuroImage 2006;31:1487–1505.

11.

Schilling

Blaber

Huo

, et al. Synthesized b0 for diffusion distortion correction (Synb0-DisCo). Magn Reson Imaging 2019;64:62–70.

12.

Basser

Mattiello

LeBihan

. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 1994;103:247–254.

13.

Smith

Nichols

. Threshold-free cluster enhancement: addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage 2009;44:83–98.

14.

Yao

Yang

Zhang

, et al. A multimodal meta-analysis of regional structural and functional brain alterations in type 2 diabetes. Front Neuroendocrinol 2021;62:100915.

15.

van Sloten

Sedaghat

Carnethon

, et al. Cerebral microvascular complications of type 2 diabetes: stroke, cognitive dysfunction, and depression. Lancet Diabetes Endocrinol 2020;8:325–336.

16.

Wang

, et al. Diabetes mellitus impairs white matter repair and long-term functional deficits after cerebral ischemia. Stroke 2018;49:2453–2463.

17.

Zhuo

Fang

, et al. White matter impairment in type 2 diabetes mellitus with and without microvascular disease. Neuroimage Clin 2019;24:101945.

18.

Wang

Wei

, et al. Relationship between type 2 diabetes and white matter hyperintensity: a systematic review. Front Endocrinol (Lausanne 2020;11:595962.

19.

Weykamp

. Hba1c: a review of analytical and clinical aspects. Ann Lab Med 2013;33:393–400.

20.

Martos-Cabrera

Velando-Soriano

Pradas-Hernandez

, et al. Smartphones and apps to control glycosylated hemoglobin (HbA1c) level in diabetes: a systematic review and meta-analysis. J Clin Med 2020;9:693.

21.

Sundfor

Svendsen

Heggen

, et al. BMI Modifies the effect of dietary fat on atherogenic lipids: a randomized clinical trial. Am J Clin Nutr 2019;110:832–841.

22.

Clark

Nicholls

, et al. Visit-to-visit blood pressure variability, coronary atheroma progression, and clinical outcomes. JAMA Cardiol 2019;4:437–443.

23.

Wilson

Kimura

Harada

, et al. Beta2-microglobulin as a biomarker in peripheral arterial disease: proteomic profiling and clinical studies. Circulation 2007;116:1396–1403.

24.

You

Xie

, et al. High levels of serum beta2-microglobulin predict severity of coronary artery disease. BMC Cardiovasc Disord 2017;17:71.

25.

Horton

Barrett

. Microvascular dysfunction in diabetes mellitus and cardiometabolic disease. Endocr Rev 2020;42:29–55.

26.

Henning R

. Type-2 diabetes mellitus and cardiovascular disease. Future Cardiol 2018;14:491–509.

27.

Ruiz

Lopez

Arivazahagan

, et al. Metabolism, obesity, and diabetes mellitus. Arterioscler Thromb Vasc Biol 2019;39:e166–e174.

28.

Tan

Fang

, et al. Micro-structural white matter abnormalities in type 2 diabetic patients: a DTI study using TBSS analysis. Neuroradiology 2016;58:1209–1216.

29.

Inglese

Bester

. Diffusion imaging in multiple sclerosis: research and clinical implications. NMR Biomed 2010;23:865–872.

30.

Liang

Cai

Tang

, et al. Diffusion tensor imaging of white matter in patients with prediabetes by trace-based spatial statistics. J Magn Reson Imaging 2019;49:1105–1112.

31.

Alexander

Lee

Lazar

, et al. Diffusion tensor imaging of the brain. Neurotherapeutics 2007;4:316–329.

32.

Sun

Liang

Cross

, et al. Evolving Wallerian degeneration after transient retinal ischemia in mice characterized by diffusion tensor imaging. Neuroimage 2008;40:1–10.

33.

Biessels

Nobili

Teunissen

, et al. Understanding multifactorial brain changes in type 2 diabetes: a biomarker perspective. Lancet Neurol 2020;19:699–710.

34.

Folstein

McHugh

. Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189–198.

35.

Nasreddine

Phillips

Bedirian

, et al. The Montreal cognitive assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005;53:695–699.

36.

Frederiksen

. Corpus callosum in aging and dementia. Dan Med J 2013;60:B4721.

37.

Stave

De Bellis

Hooper

, et al. Dimensions of attention associated with the microstructure of Corona Radiata white matter. J Child Neurol 2017;32:458–466.

38.

Komaitis

Skandalakis

Kalyvas

, et al. Dorsal component of the superior longitudinal fasciculus revisited: novel insights from a focused fiber dissection study. J Neurosurg 2019;132:1265–1278.