White matter microstructure alterations in type 2 diabetes mellitus and its correlation with cerebral small vessel disease and cognitive performance

Microstructural abnormalities of white matter fiber tracts are considered as one of the etiology of diabetes-induced neurological disorders. We explored the cerebral white matter microstructure alteration accurately, and to analyze its correlation between cerebral small vessel disease (CSVD) burden and cognitive performance in type 2 diabetes mellitus (T2DM). The clinical-laboratory data, cognitive scores [including mini-mental state examination (MMSE), Montreal cognitive assessment (MoCA), California verbal learning test (CVLT), and symbol digit modalities test (SDMT)], CSVD burden scores of the T2DM group (n = 34) and healthy control (HC) group (n = 21) were collected prospectively. Automatic fiber quantification (AFQ) was applied to generate bundle profiles along primary white matter fiber tracts. Diffusion tensor images (DTI) metrics and 100 nodes of white matter fiber tracts between groups were compared. Multiple regression analysis was used to analyze the relationship between DTI metrics and cognitive scores and CSVD burden scores. For fiber-wise and node-wise, DTI metrics in some commissural and association fibers were increased in T2DM. Some white matter fiber tracts DTI metrics were independent predictors of cognitive scores and CSVD burden scores. White matter fiber tracts damage in patients with T2DM may be characterized in specific location, especially commissural and association fibers. Aberrational specific white matter fiber tracts are associated with visuospatial function and CSVD burden.

In patients with type 2 diabetes mellitus (T2DM), cognitive impairments tend to affect verbal and visual memory, information processing speed, and executive functioning 1 , particularly among people above 65 years 2 .Cerebral vascular injury is one of the possible mechanisms for the increased risk of cognitive impairment in T2DM patients 3,4 , especially microvascular.In addition to the classic target organs of microvascular complications such as the retina and kidney, the brain is also considered as the target organ of microvascular complication in patients with T2DM 5 .
Cerebral small vessel disease (CSVD) is a neuroradiological diagnosis that refers to a group of pathological processes affecting small arteries, arterioles, capillaries, and small veins in the brain 6 .Cerebral small vessels form a small vessel network, which is embedded in the neurovascular unit (NVU) and participates in the formation of the blood-brain barrier, regulating substances from the blood to the brain parenchyma, regulating cerebral blood flow (CBF), maintaining and repairing myelin sheath, clearing metabolites, and other functions.NVU is linked to neurodegenerative diseases and cognitive impairments 7 .

CSVD scores
Two radiologists with 10 years and 6 years of neuroradiology experience (Y.L., Y.J.) provided the CSVD scores of all participants independently, including WMH grade, EPVS score, lacune number, CMB number, and CSVD total burden score.The detailed definition and criteria for WMH grade, EPVS score, and CSVD total burden score were shown in the Supplementary materials.

Sample size calculation
The sample size was calculated using Power Analysis and Sample size (PASS) software version 15.0.5.The calculation was done based on two-tailed test, α of 0.05, and power of 0.8, the ratio of sample size between T2DM group and HC group was 1:1.Referring to previous studies 15,[32][33][34] , the diffusion metrics values of bilateral IFOF, CF_Major, CF_Minor were used as the main observation indicators.The data of our preliminary experiment were used for sample size calculation.
Clinical data, laboratory data, cognitive scores, and CSVD scores Data analyses were performed using the Statistical Package for Social Science (SPSS) version 22.0.Independent samples t-test or Mann-Whitney U test was used to compare the difference in measurement data between T2DM and HC group.Chi-square test was used to compare the difference in enumeration data between T2DM and HC groups.p < 0.05 was considered statistically significant.

Diffusion metric
The mean values of FA, MD, AD, and RD of 16 fibers was compared betweenT2DM and HC groups.Controlling gender, age, BMI, and education as covariates, node-wise analyses were applied based on the general linear model (GLM) permutation test (5000 permutations) in the FSL to compare the differences in 100 nodes diffusion metrics of 16 fibers between the T2DM group and the HC group.Only significant differences observed at ≥ 3 adjacent nodes were reported 35 .
We then performed multiple linear regression analysis to find independent predictors for explaining the relationship between DTI metrics and cognitive and CSVD scores, age, gender, education years were inserted as control variables 36 .Statistical tests were two-tailed, and p < 0.05 was considered statistically significant.

Sample size calculation
The estimated sample size was at least 40.While the total sample size in our study was 55 participants.Therefore, the total sample size for this study was sufficient to meet the accuracy requirements of the study.The details of sample size calculation were shown in Supplementary Table S1.

Clinical and laboratory data
The clinical and laboratory data of T2DM and HC groups were summarized in Table 1.Age, gender, education years, DBP, TCHOL, LDL, and HCY between groups showed no statistical difference (p > 0.05).The SBP, BMI, WHR, FG, HbA1C and TG were increased, and the HDL was decreased in T2DM patients compared with HC group (p < 0.05).

Cognitive scores
The cognitive scores of T2DM and HC groups were shown in

CSVD scores
The CSVD scores of T2DM and HC groups were shown in Table 3.The PVH grade, DWMH grade, CSO-EPVS score, BG-EPVS score, and CSVD total burden score of T2DM group were higher than HC group (p < 0.05).

Diffusion metric
For fiber-wise, the mean MD value in CF_Major, CF_Minor, right IFOF, right ILF, the mean value of AD in CF_Major, right IFOF, right ILF, right SLF, the mean RD value of RD in CF_Major, bilateral IFOF, were increased in T2DM compared to HC group (Table 4, Fig. 1).The nodes 11-16 of MD and RD in right IFOF, nodes 86-100 of AD in right SLF, and nodes 13-28 of AD in right ILF in T2DM patients were increased in the node-wise comparison between groups (Fig. 2) (FWE corrected).

Multiple linear regression analysis
The multiple regression analysis between DTI metrics and cognitive scores revealed that the mean AD value of right ILF was independent predictor of MoCA visuospatial score (Table 5, Fig. 3).
The multiple regression analysis between DTI metrics and CSVD scores revealed that the mean MD value of right IFOF was independent predictor of PVH grade, CSO-EPVS grade and CSVD total burden score, mean MD value of CF_Minor was independent predictor of DWMH grade, BG-EPVS grade and CSVD total burden score, mean AD value of right ILF was independent predictor of CSO-EPVS grade (Table 6, Fig. 3).

Discussion
This study used diffusion metrics, including FA, MD, RD, and AD, to assess the white matter fiber tracts microstructure alterations in T2DM patients.In this study, compared with the HCs, there was extensive damage to the microstructure of the association fibers and CF in T2DM patients, consistent with previous study 37 .Hyperglycemia and broad impairment of microvascular function in T2DM patients will contribute to inflammatory and hypoperfusion, resulting in insufficient brain blood supply 38,39 .Brain is very sensitive to ischemia which could damage white matter fibers microstructural integrity 40 .
In our study, compared with HCs, T2DM patients showed an extensive increase in the mean MD, AD, and RD values of association fibers and CF.Specifically, the MD values of CF_Major, CF_Minor, right IFOF, and right ILF increased, consistent with previous studies 32,41,42 increased.In the node-wise comparison, the nodes 11-16 of MD in right IFOF in T2DM patients were increased.MD is independent of direction, quantifying the diffusion in extent, and it can sensitively reflect the cell number, edema, and necrosis 43 .Indicating that the number cells of CF_Major, CF_Minor, right IFOF, and right ILF in T2DM patients is reduced, particularly the posterior part of the IFOF.Previous studies 44 suggest that in the DTI framework, MD account for the diffusion along all three axes of the diffusion tensor with equally weight, i.e. the average value.Therefore, MD is theoretically more robust in evaluating complex fibers, may be the best DTI metrics.In the current study, the AD values of CF_Major, right IFOF, right ILF, and right SLF of T2DM patients were increased compared to the HCs.In the node-wise comparison, the nodes 86-100 of AD in right SLF in T2DM patients were increased.Suggest that the alteration of AD value is mainly located in the posterior part of SLF.The AD value alterations of white matter fiber tracts in www.nature.com/scientificreports/T2DM patients in this study were partially consistent with previous research findings 45,46 .However, other studies report opposite results, namely AD values in T2DM patients are decreased for CF_Major, right IFOF and right SLF 32,42,47 .It is suggested that AD changes are more subtle.AD reflects the diffusivity parallel to the white matter fibers, which reflects axon integrity 46,48 .Chronic ischemia and hypoxia caused by chronic hyperglycemia result in axonal loss and fiber degeneration, affecting the distribution of water molecules, expansion of extracellular space 46,49 , leading to an increase in AD.Moreover, our research has also found that the RD values of CF_Minor and bilateral IFOF of T2DM patients were increased compared to the HCs.RD reflects diffusivity perpendicular to axonal fibers, and increased RD is associated with the disruption of the myelin sheath.The IFOF connects the extensive cortex of the frontal, temporal, parietal, and occipital lobes through the external capsule 50 , and the particularity of this anatomical site makes it more vulnerable to be damaged than other fibers.Uniformly increased MD, AD, and RD may reflect a decline in tissue structures that normally prevent the free diffusion of fluid and ensuing changes in fluid volume 51 .However, in this study, no significant differences were found in the FA values between groups.FA has long been one of the most commonly reported quantitative diffusion indicators, usually considered as the most sensitive DTI metrics.Previous studies 15,[32][33][34] have been reported the white matter FA value abnormalities in patients with T2DM, mainly focusing on corpus callosum, IFOF, SLF and ILF.Nevertheless, there is some evidence supporting the opposite, suggesting that MD and RD values are more sensitive than FA in discovering the white matter microstructure vulnerability 52,53 .This might be due to FA is a relative measure, the quantitative information might be inaccurate in the presence of crossing fibers 44 .However, cerebral white matter voxels containing complex crossing fibers, kissing fibers etc.Therefore, we should be exercise caution when interpreting FA value.
The multiple regression analysis revealed that the mean AD value of ILF_R was independent predictor of MoCA visuospatial score, indicating that axonal loss and fiber degeneration of the right ILF leads to cognitive www.nature.com/scientificreports/impairment in T2DM patients, particularly in terms of visuospatial function.The ILF is a fiber connecting the temporal and occipital lobes, closely related to visual and cognitive functions.This is consistent with the results of Gao et al. 54 , i.e.T2DM impairs cognition by attacking ILF.The mean RD value of right IFOF was independent predictor of PVH grade and CSO-EPVS grade, mean MD value of CF_Minor was independent predictor of DWMH grade, BG-EPVS grade and CSVD total burden score, reflected that the damage of CF_Minor and right IFOF would lead to CSVD burden increased in T2DM patients.This study also has some limitations: (1) the sample size of this study is relatively small, and the participants are recruited from volunteers, which may be biased to some extent; (2) the confounding factors have not been fully accounted for, gut microbiota 55 , age at T2DM onset 56 , lifestyle 57,58 , and treatment 59 can also have an effect on the cognition of T2DM.The control ability of these confounding factors is insufficient, and more detailed research should be conducted in the future; (3) the CSVD scores are still subjective.In the future, it should be combined with artificial intelligence technology to quantify the burden of CSVD more objectively.(4) some fiber bundles (CH and AF) were adjacent to gray matter and cannot be entirely quantified by AFQ, we cannot know their microstructure changes;(5) due to the cerebral white matter voxel contains multiple fiber populations and complex fiber geometries, DTI metrics should be cautiously interpreted, as they are indirect measures of real biological structures.We would encourage future studies to use other MRI approach that are more robust to the presence of crossing fibers such as diffusion spectrum imaging (DSI) 60 , q-space imaging (QSI) 61 , high-angularresolution diffusion imaging (HARDI) 62 , and diffusion kurtosis imaging (DKI) 63 , etc.
To conclusion, there is extensive white matter fiber tracts damage in T2DM patients, mainly the damage to commissural and association fibers.It reveals that the mechanism of visuospatial function decline in T2DM patients is related to the deterioration of right ILF, and the damage of CF_Minor and right IFOF is closely related to the CSVD burden increase, which provides a new idea for further understanding the mechanism of cognitive function decline and CSVD burden increase in T2DM.

Figure 2 .
Figure 2. Plots of significantly altered locations in point-wise comparison between T2DM and HC group (p < 0.05, FWE corrected).The orange and blue lines represent the HC and T2DM groups; respectively (solid lines for mean value and shaded regions for confidence interval).The pink bar represents fiber nodes with significant differences between the two groups.T2DM type 2 diabetes mellitus; HC healthy control; MD mean diffusivity; AD axial diffusivity; RD radial diffusivity; IFOF inferior fronto-occipital fasciculus; ILF inferior longitudinal fasciculus; SLF superior longitudinal fasciculus.

Table 2 .
The total MMSE score, MoCA visuospatial score, total MoCA score, and SDMT score were decreased in T2DM group compared with HC group (p < 0.05).

Table 2 .
Cognitive scores of T2DM and HC groups.T2DM type 2 diabetes mellitus; HC healthy control; MMSE mini-mental state examination; MoCA Montreal cognitive assessment; CVLT California verbal learning test; SDFR short delay free recall; SDCR short delay cued recall; LDFR long delay free recall; LDCR long delayed cued recall; SDMT symbol digit modalities test.*p < 0.05.a data are expressed as mean value ± standard deviations.b data are expressed as median (inter-quartile range).