Patients with diabetes mellitus are at risk for developing mild to moderate cognitive impairment.Reference McCrimmon, Ryan and Frier 1 , Reference Munshi, Grande and Hayes 2 The incidence of dementia, including Alzheimer’s disease (AD) and vascular dementia (VD), is higher in individuals with diabetes mellitus than in those without.Reference Biessels, Deary and Ryan 3 , Reference Biessels, Staekenborg, Brunner, Brayne and Scheltens 4 The mechanisms underlying this cognitive dysfunction are complex. In diabetic cognitive dysfunction patients, changes in cerebral perfusion, microstructure of white matter (WM), and increased cortical and especially hippocampal atrophy have been demonstrated.Reference Ryan, Fine and Rosano 5 Diffusion tensor imaging (DTI) is a tool for investigating the local properties of brain WM and detecting early abnormal microstructural changes, including demyelination and axonal injury.Reference Bozzali and Cherubini 6 Hsu et alReference Hsu, Chen and Leu 7 identified structural changes in WM associated with cognitive dysfunction. Another important factor for the pathogenesis of chronic complications due to diabetes are the formation of advanced glycation end-products (AGEs),Reference Brownlee 8 and they have been linked to both VDReference Southern, Williams and Esiri 9 and AD.Reference Horie, Miyata and Yasuda 10 Nε-(Carboxymethyl)lysine (CML) is a predominant and well-studied form of AGEReference Ikeda, Higashi and Sano 11 that is significantly elevated in the cerebrospinal fluid of patients suffering from AD and VD relative to controls.Reference Bar, Franke and Wenda 12 In the current study, we examined the most commonly used DTI metric, fractional anisotropy (FA), to determine if WM microstructure is altered in older community-dwelling adults with type 2 diabetes mellitus (T2DM). In addition, the relationship among plasma CML level, FA value, and cognitive function scores was investigated to further understand potential mechanisms underlying cognitive impairment T2DM patients.
Materials and Methods
Participants
The study was approved by the Medical Ethics Committee of the Second Affiliated Hospital of Dalian Medical University, and written informed consent was obtained from each participant before magnetic resonance imaging (MRI). For the patient group, subjects were selected from outpatients and inpatients of the Department of Neurology, Second Affiliated Hospital of Dalian Medical University, from March 2012 to March 2014. Patient ages ranged from 65 to 75 years with T2DM spanning at least 5 years. Patients were right hand–dominant and Chinese-speaking. There was no history of other glucose-lowering drugs besides biguanide and alpha glycosidase inhibitor. For the control group, subjects were recruited from patient families and the community. They were sex-, age-, and education-matched controls and Chinese-speaking.
Patients and controls were excluded based on the following criteria: (1) history of cardiovascular disease (e.g. hypertension, coronary heart disease, myocardial infarction), (2) history of central nervous system disease (e.g. transient ischemic attack or stroke, Parkinson’s disease, severe brain injury, seizure, multiple sclerosis, infectious diseases of central nervous system), (3) standard contraindications to MRI, (4) hepatic dysfunction, (5) renal dysfunction, (6) a psychiatric or neurological disorder (unrelated to diabetes) that could influence cognitive functioning (such as AD and VD), (7) substance abuse (alcohol, drugs, or others), and (8) diabetic peripheral neuropathy. For all participants, HbA1c and blood glucose were measured in the overnight fasting state.
All participants underwent a detailed and normative cognitive neuropsychology assessment, including Montreal cognitive assessment (MoCA),Reference Nasreddine, Phillips and Bedirian 13 Mini-Mental State Examination (MMSE),Reference Folstein, Folstein and McHugh 14 and clinical dementia rating.Reference Morris 15 All tests were conducted in Mandarin by the same trained licensed neurologist with experience in a fixed order who was blind to the two groups.
CML
Following an overnight fast (12-14 hours), a 5-mL blood sample was collected from each patient, and the level of CML was measured using an enzyme-linked immunosorbent assay kit (Westang Bio-tech, Shanghai, China) for quantitative analysis according to the manufacturer’s protocol, as described previously.Reference Hu, He, Liu, Xu and Ma 16
DTI Acquisition
DTI was performed on a 3.0T MRI scanner (Siemens, Malvern, PA) using an eight-channel phased-array head coil. First, conventional sagittal T1-weightedand fluid-attenuated inversion recovery (FLAIR) images were scanned for routine MRI screening. The parameters for T1-weighted imaging were as follows: repetition time/echo time (TR/TE)=2550/3.6 ms, TI=1100 ms, flip angle=7°, field of view (FOV)=230×180 mm, and matrix size=512×512. The parameters for FLAIR imaging were as follows: TR/TE=10800/120 ms, TI=2700 ms, section thickness=5 mm, FOV=230×180 mm, matrix size=512×512, and modulation of flip angle for refocusing pulses. A single refocusing spin-echo echoplanar imaging sequence was used to acquire DTI data. The sequence parameters were as follows: TR/TE=9800/92 ms; b value: 0 and 1000 s/mm2, respectively; FOV=240 mm×240 mm; matrix size=256×256, Thirty-one isotropically distributed diffusion weighted directions with b=1000 s/mm2 and one additional image without diffusion gradients (b=0 s/mm2) were also acquired.
Image Processing and Analysis
DTI images were processed using the diffusion toolbox implemented in the functional MRI of the brain (FMRIB) software library 5.0 (http://fsl.fmrib.ox.ac.uk/fsl). First, each subject volume was affine registered to the first b0 volume using the image registration tool FLIRT, and brain voxels of DTI data were extracted using the Brain Extraction Tool.Reference Hu, He, Liu, Xu and Ma 16 , Reference Jenkinson, Bannister, Brady and Smith 17 Upon removal of nonbrain tissue, FA maps were calculated using the DTI-FIT tool.
A voxel wise statistical analysis of the FA values between different groups with the tract-based spatial statistics method was performed.Reference Smith 18 First, images from all individuals were aligned into a 1×1×1 mm3 Montreal Neurological Institute152 space. Next, a mean FA image of all aligned FA images was calculated and thinned to create the mean FA skeleton, which represented the centers of all tracts common for the group. The skeleton’s FA threshold was ≥0.20 to ensure that areas of low FA and/or high inter-subject variability were excluded from the analyses. Each subject’s aligned FA data were then projected onto the skeleton. Group statistical analysis was conducted only on voxels within the WM skeleton mask. For each subject, the mean FA of brain tissue was calculated by averaging the WM across voxels. Group differences of FA were evaluated with a permutation-based randomized test and inference by using the threshold-free cluster enhancement method implemented in FSL. The statistical threshold for all image analyses was set to a cluster p<0.05.
Data Analysis
All data are expressed as means ± standard deviation and were analyzed by SPSS 19.0 software (Armonk, NY). Comparisons between groups in demographics, clinical data, cognitive data, and WM FA value were performed using a two-tailed t-test. Pearson correlation coefficient was used to evaluate the relationship of FA of the WM with clinical parameters and plasma CML level; p<0.05 was considered statistically significant.
Results
Demographic and Clinical Data
Demographic and clinical characteristics of the patient and control groups are shown in Table 1. There were no significant differences between groups in age, gender, and education. The patient group had higher levels of fasting glucose, HbA1c, and plasma CML than the control group (p<0.05). The MMSE scores (p<0.05) (25.72±3.13 and 28.16±2.45) and MoCA scores (p<0.01) (22.15±3.56 and 25.63±4.12) were significantly different between patient and control groups.
Table 1 Demographic and clinical data of the patients group and control
group (
$$\bar{x}$$
± S)
CDR scores are described as (minimum value, maximum value).
* Compared with the controls p<0.05.
† Compared with the controls p<0.01.
CDR=clinical dementia rating; CML=Nε-(carboxymethyl)-lysine; MMSE=Mini-Mental State Examination; MoCA=Montreal cognitive assessment; T2DM, type 2 diabetes mellitus.
DTI Image Analysis Results
All participants completed the MRI scan. No obvious abnormal findings were seen in conventional MRI structural images (T1-weighted, T2-weighted, and FLAIR). The Fazekas scale scores in the control and patient groups ranged from 0 to 1 and 0 to 2, respectively. The tract-based spatial statistics whole-brain analysis results revealed that there were no significant differences in FA values between the left- and right-sided regions in either patient or control groups (Figure 1). Relative to controls, FA values in the patient group were reduced in multiple regions, including the hippocampus, parietal and temporal lobes, corpus callosum, and its association fibers (Table 2). FA values in regions of frontal lobe, occipital lobe, and internal capsule were not significantly different.
Figure 1 Tract-based spatial statistics whole-brain fractional anisotropy (FA) analysis. The regions of significantly reduced FA values are red and the mean FA skeleton is in green.
Table 2 Comparison of FA values of brain regions between patient and control groups
The Correlation Between CML and FA and MoCA
Pearson correlation coefficient showed that there was a significant negative correlation between plasma CML level and average FA values in whole-brain WM of T2DM patients (r=−0.58, p<0.01) (Figure 2). In addition, there was a negative correlation between plasma CML level and MoCA scores (r=−0.47, p<0.05; Figure 3). There was no correlation between plasma CML level and MMSE scores.
Figure 2 The correlation between plasma Nε-(carboxymethyl)-lysine (CML) level and average fractional anisotropy (FA) values in global brain in the patient group.
Figure 3 The correlation between plasma Nε-(carboxymethyl)-lysine (CML) level and Montreal cognitive assessment (MoCA) scores in the patient group.
Discussion
We found that the level of plasma CML in T2DM patients was significantly higher than controls. In addition, CML level was negatively correlated with cognitive function score, supporting our hypothesis that CML plays an important regulatory role in the manifestation of diabetic cognitive impairment. Deutekoms et al found CML in brain astrocytes and vascular endothelial cells and that brain CML levels in diabetic patients and a diabetic animal model are much higher than controls.Reference van Deutekom, Niessen, Schalkwijk, Heine and Simsek 19 Other studies demonstrated that CML levels were correlated with the severity of diabetic complications.Reference Genuth, Sun and Cleary 20 , Reference McCance, Dyer and Dunn 21 Diabetes mellitus patients have an increased risk for mild to moderate cognitive impairment, likely from vascular dysfunction.Reference McCrimmon, Ryan and Frier 1 , Reference Munshi, Grande and Hayes 2 A series of studies have shown using arterial spin labeling that cerebral blood flow is significantly reduced in elderly T2DM patients.Reference Smith, Jenkinson and Johansen-Berg 22 A meta-analysis and a series of studies suggested that microangiopathy and vasculopathy play an important role in diabetic cognitive impairment. Nonenzymatic glycation reaction plays an important role in vascular dysfunction.Reference Last, Alsop and Abduljalil 23 , Reference van Harten, de Leeuw, Weinstein, Scheltens and Biessels 24 CML, one of the irreversible end-products of the nonenzymatic glycation reaction, is the main form of AGEs. The deposition of AGEs occurs during the normal aging process, and diabetes mellitus accelerates this process.Reference Brownlee 25 - Reference Yaffe, Lindquist and Schwartz 27 Studies have shown that the interaction of AGEs and their specific receptor may accelerate aging of the brain.Reference Takeuchi and Yamagishi 28
In this study, we found no obvious structural changes on conventional MRI (both T1- and T2-weighted MRI) in T2DM patients, but we found a significant decrease in FA value in T2DM patients in the hippocampus, corpus callosum, and its association fibers, including superior longitudinal fasciculus, cingulate fasciculus, and inferior fronto-occipital fasciculus. These findings identified destruction of the WM of elderly T2DM patients, especially in the WM regions associated with memory and recognition, but not in WM of the frontal lobe, occipital lobe, and internal capsule. Other studies have shown that T2DM patients often demonstrate cognitive impairment, such as processing information slowly, and it remains unclear whether the change is caused by WM microstructural abnormalities.Reference Reijmer, Brundel and de Bresser 29 Kodl et al identified microstructural WM abnormalities in a patient with long-standing type 1 diabetes patient,Reference Kodl, Franc and Rao 30 The FA values in the posterior corona radiate and the optic radiation were decreased, and these WM abnormalities were correlated with poor performance on some neurocognitive tests in the type 1 diabetes patient.Reference Kodl, Franc and Rao 30 Yau et al showed in middle-aged and elderly patients with T2DM in the absence of any obvious vascular pathology or psychiatric disorder that FA values in the frontal and temporal WM, which is associated with emotional and declarative memory, were significantly lower in T2DM patients.Reference Yau, Javier and Tsui 31 When DTI was used to investigate microstructural WM abnormalities among T2DM patients, these patients had significantly decreased FA value in bilateral frontal and total WM relative to controls.Reference Hsu, Chen and Leu 7 , Reference Falvey, Rosano and Simonsick 32 Meanwhile, there was a negative correlation between plasma CML level and average FA values in whole-brain WM in T2DM patients, suggesting that the accumulation of AGEs may be associated with microstructural WM changes.
Other factors also play an important role in diabetic cognitive impairment. Gold et al found hippocampal volume is negatively correlated with HbA1 in diabetes mellitus patients,Reference Gold, Dziobek and Sweat 33 prompting the importance of glucose control. The FA value in genu of corpus callosum was negatively correlated with disease course.Reference Kodl, Franc and Rao 30 Evidence from DTI based on voxel analysis found that the longer the disease course of T2DM, the more serious the WM matter injury.Reference Hsu, Chen and Leu 7 However, this study did not determine the relationship between HbA1c and fasting blood-glucose with cerebral injury. In addition, other studies have found a relationship between disease course of T2DM peripheral neuropathy and autonomic neuropathy.Reference Chen, Lin and Won 34 , Reference Hsu, Chiu, Chen, Chiu, Liou and Chen 35 Because the conclusions are contradictory, the mechanism underlying occurrence and development of diabetic cognitive impairment needs further study.
Our study had several limitations. First, the sample was restricted to old men, thereby limiting the generalizability of our results. In addition, the number of participants was relatively small, which may reduce the power of the statistical significance. Although this study discussed the relationship between the reduction of FA value and plasma CML level, the levels of CML in cerebrospinal fluid may be more relevant.Reference Bar, Franke and Wenda 12 Last, this study used only FA values to study the microstructure changes of brain white matter, and use of a multiple parameter comprehensive analysis may be more informative.
Conclusions
In summary, we have shown that there is a negative correlation between plasma CML and cognitive function score and that the FA value in WM of T2DM patients was significantly decreased. These findings suggest that elevations in CML level and changes in WM microstructure in older T2DM patients may contribute to the development of cognitive impairment in these patients.
Acknowledgments
The study was supported by the National Natural Science Foundation of China (no. 81571237), National Science and Technology Major Special Project on Major New Drug Innovation (No.2012ZX09503001-003), Natural Science Foundation of Liaoning Province of China (No. 2013023022), and Project of Liaoning Province Department of Education of China (No. L2014344).
Disclosures
The authors have nothing to disclose.
