White Matter Integrity Disruptions Associated With Cognitive Impairments in Type 2 Diabetic Patients
Abstract
Type 2 diabetes mellitus (T2DM) is associated with a twofold increased risk of dementia and can affect many cognitive abilities, but its underlying cause is still unclear. In this study, we used a combination of a battery of neuropsychological tests and diffusion tensor imaging (DTI) to explore how T2DM affects white matter (WM) integrity and cognition in 38 T2DM patients and 34 age-, sex-, and education-matched normal control subjects. A battery of neuropsychological tests was used to assess a wide range of cognitive functions. Tract-based spatial statistics combined with region of interest–wise (ROI-wise) analysis of mean values of DTI metrics in ROIs was used to compare group differences of DTI metrics on WM skeletons to identify severely disrupted WM tracts in T2DM. We found that T2DM patients showed 1) various cognitive impairments, including executive function, spatial processing, attention, and working memory deficits; 2) widespread WM disruptions, especially in the whole corpus callosum, the left anterior limb of the internal capsule (ALIC.L), and external capsule (EC); and 3) a positive correlation between executive function and WM integrity in the ALIC.L and the left EC. In conclusion, T2DM patients show various cognitive impairments and widespread WM integrity disruptions, which we attribute to demyelination. Moreover, executive dysfunction closely correlates with WM abnormalities.
Introduction
Type 2 diabetes mellitus (T2DM) in the elderly is a major public health problem. In China, >20.4% of the elderly population has diabetes, among which T2DM accounts for ∼90% (1). As a risk factor for cognitive decline, T2DM is associated with a twofold increased risk of dementia (2) and can affect a wide range of cognitive abilities (3,4). However, its underlying cause is still unclear.
A number of studies were carried out to investigate the mechanisms of T2DM-induced dementia and attributed the cognition decline to cerebral microstructure impairments, which included gray matter (GM) and white matter (WM) destruction (5,6). Despite the importance of GM atrophy in T2DM, WM microstructure abnormalities played a distinct and irreplaceable role in cognition impairments induced by T2DM (7,8). WM has a vital role for transferring information between GM regions, and its efficiency depends on WM microstructural integrity (9). However, the relationship between WM microstructural changes and T2DM is debated and contradictory. Many studies suggested a close correlation between T2DM and WM lesion (WML) severity or progression (5,10), but others did not (11,12). This uncertainty could be partly due to the insufficient sensitivity of conventional MRI modalities in detecting subtle brain WM changes or assessing the severity of WM hyperintensities (WMH) (7).
Diffusion tensor imaging (DTI), a new type of MRI, has been developed as a powerful noninvasive technique to investigate WM microstructures and integrity (13,14). Fractional anisotropy (FA) and mean diffusivity (MD) describe fiber density, axonal diameter, and myelination in WM based on quantitative measure of the degree of diffusion anisotropy (15,16). As reported previously based on the DTI technique, decreased FA or increased MD in varieties of cognitive dysfunction related to Alzheimer disease (AD) (17,18), amnestic mild cognitive impairment (MCI) (19), schizophrenia (20,21), and type 1 diabetes (22), etc. Thus, DTI techniques are important for exploring more sensitive imaging-based biomarkers in prevention and early treatment of cognitive dysfunction induced by T2DM. However, there were only a few DTI-based investigations in WM disruption in T2DM (7,16), and Hsu et al. (7) reported a remarkable FA decrease in the bilateral frontal WM and increased MD in T2DM patients by assessing the DTI with global and voxel-based analyses (VBA).
However, VBA methods that many previous researches performed may encounter atrophy-induced WM tract misregistration, which causes an unexpected decrease in accuracy. Tract-based spatial statistics (TBSS), a corresponding fully automated whole-brain analysis technique based on DTI images, improves the sensitivity, objectivity, and interpretability of analysis of multisubject diffusion imaging and applies voxel-wise statistics to diffusion measures in various studies (23,24).
In our present study, we assessed cognitive functions with a battery of neuropsychological tests, measured WM structural integrity quantitatively using the TBSS method based on DTI data in T2DM subjects, and evaluated the correlation between WM disruptions and cognitive dysfunctions.
Research Design and Methods
Participants
Participants in this study were all from the Beijing Ageing Brain Rejuvenation Initiative (BABRI), which is an ongoing, longitudinal study investigating aging and cognitive impairment in urban elderly people in Beijing, China. The current study included 38 patients with diabetes (18 male) and 34 age-, sex-, and education-matched healthy control subjects (17 male). All of the T2DM patients were diagnosed by a physician and had a history of using oral antidiabetic medications or insulin. Of the 38 T2DM subjects, 13 subjects are under treatment with insulin and 27 subjects control blood glucose using oral hypoglycemic agents. The duration of diabetes was defined as the number of years since diagnosis. All subjects had a medical history and physical examination, during which height, weight, and BMI were recorded. Fasting plasma glucose (FPG), glycosylated hemoglobin (HbA1c), and cholesterol levels were measured with standard laboratory testing. Exclusion criteria for the groups were a previous history of neurological or psychiatric disease including stroke, dementia, or transient ischemic attack and unsuitability for MRI (e.g., due to a pacemaker, prosthetic heart valve, or claustrophobia). All participants provided written informed consent to our protocol that was approved by the ethics committee of the State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University. All subjects were right-handed and native Chinese speakers.
Neuropsychological Testing
All of the participants received a battery of neuropsychological tests assessing general mental status and other cognitive domains, such as episodic memory, tested using the Auditory Verbal Learning Test (AVLT) (25) and Rey-Osterrieth Complex Figure recall test; attention, tested using the Trail Making Test Part a (TMT-a) and the Symbol Digit Modalities Test (SDMT) (26); visuospatial ability, tested using the Rey-Osterrieth Complex Figure copy test and the Clock Drawing Test (CDT); language, tested using the Category Verbal Fluency Test (CVFT) and the Boston Naming Test (BNT); and executive function, tested using the TMT-b and the Stroop Test. Verbal working memory was assessed with the digit span of the Wechsler Adults Intelligence Scale–Chinese revision. Logical reasoning was assessed with the Similarities subtest of the Wechsler Adult Intelligence Scale–Chinese revision.
Image Acquisition
The MRI data were acquired on a 3.0T Siemens Tim MRI scanner in the Imaging Center for Brain Research, Beijing Normal University. Participants lay supine with the head snugly fixed by a belt and foam pads to minimize head motion. Two sets of DTI data scans were acquired for every subject and then averaged during the data preprocessing. DTI images covering the whole brain were acquired by using a single-shot, twice-refocused, diffusion-weighted echo planar imaging sequence with the following scan parameters: TR = 9,500 ms; TE = 92 ms; 30 diffusion-weighted directions with a b value of 1,000 s/mm2, and a single image with a b value of 0 s/mm2; slice thickness = 2 mm; no interslice gap; 70 axial slices; matrix size = 128 × 128; field of view = 256 × 256 mm2; and voxel size = 2 × 2 × 2 mm3.
DTI Image Preprocessing
All the DTI image preprocessing and analyses described below were implemented using a pipeline tool for diffusion MRI (PANDA) (27). First, the DICOM files of all subjects were converted into NIfTI images using the dcm2nii tool embedded in MRIcron. Second, the brain mask was estimated, and this step yielded the brain mask, which was required for the subsequent processing steps. Third, the nonbrain space in the raw images was cut off, leading to a reduced image size, reducing the memory cost, and speeding up the processing in subsequent steps. Fourth, each diffusion-weighted image was coregistered to the b0 image using an affine transformation to correct the eddy current–induced distortions and simple head-motion artifacts. The diffusion gradient directions were adjusted accordingly. Fifth, a voxel-wise calculation of the tensor matrix and the diffusion tensor metrics were yielded for each subject, including FA, MD, axial diffusivity (λ1), and radial diffusivity (λ23).
Normalizing
To compare across subjects, location correspondence has to be established. To this end, registration of the individual images to a standardized template is always applied. Here, PANDA nonlinearly registered all of the individual images in native space to a standardized template in the MNI space (27).
Voxel-Wise TBSS Statistical Analysis
The voxel-wise statistical analysis in TBSS compares group differences only on the WM skeleton, so that it provides better sensitivity, objectivity, and interpretability of analysis for multisubject DTI studies (23). The TBSS analyses of FA, MD, λ1, and λ23 images were carried out using the FMRIB software library (FSL 4.1.4; http://www.fmrib.ox.ac.uk/fsl). In brief, first the following five-step process on the FA images was performed: 1) the FA image of each subject was aligned to a preidentified target FA image (FMRIB58_FA) by nonlinear registration, 2) all of the aligned FA images were transformed onto the MNI152 template by affine registration, 3) a mean FA image and its skeleton (mean FA skeleton) were created from the images of all the subjects, 4) individual subject FA images were projected onto the skeleton, and 5) voxel-wise statistics across subjects were calculated for each point on the common skeleton.
The voxel-wise statistics in TBSS were carried out using a permutation-based inference tool for nonparametric statistical thresholding (the “randomize” tool, part of FSL). In this study, voxel-wise group comparisons were performed using nonparametric, two-sample Student t tests between the T2DM and control groups. The mean FA skeleton was used as a mask (thresholded at a mean FA value of 0.2), and the number of permutations was set to 5,000. The significance threshold for between-group differences was set at P < 0.05 (family-wise error [FWE] corrected for multiple comparisons) using the threshold-free cluster enhancement option in the “randomize” permutation-testing tool in FSL.
Data for MD, λ1, and λ23 were generated by applying the above FA transformations to additional diffusivity maps and projecting them onto the skeleton with projection vectors that were identical to the vectors inferred from the original FA data. The statistical analyses of these diffusion tensor metrics were performed similarly to the FA analysis.
ROI-Wise Statistical Analysis
Recently, a few WM atlases (e.g., the ICBM-DTI-81 WM labels atlas [see http://cmrm.med.jhmi.edu/] and the JHU WM tractography atlas) have been proposed (28). These WM atlases in the standard space allow for parcellation of the entire WM into multiple region of interests (ROIs), each representing a labeled region in the atlas. In our current study, to investigate the diffusion changes in specific tracts, the ICBM-DTI-81 WM labels atlas was used to parcel the entire WM into 48 ROIs, and only the 40 ROIs in cerebral regions (we focused on the 40 WM tracts within the cerebrum and did not consider the other 8 ROIs within the cerebellum and brain stem) were used for the analysis (Fig. 1 and Supplementary Table 1). Then, the regional diffusion metrics (i.e., FA, MD, λ1, and λ23) were calculated by averaging the values within each region of the WM atlas.
The 40 WM tract ROIs based on the ICBM-DTI-81 WM labels atlas within cerebral regions. For the abbreviations of WM tracts, see Supplementary Table 1.
Linear regression analyses were performed to compare the resultant ROI-based data between the T2DM and control groups. Age, sex, and education years were treated as covariates in the regression analysis. A false discovery rate (FDR) was applied to correct for multiple comparisons.
Demographic, clinical, neuropsychological, and behavioral data were analyzed in SPSS 17.0 (SPSS, Inc.). The comparisons of demographic and clinical data between two groups were performed using Student t tests or χ2 test. P < 0.05 was considered significant. In the two groups separately, multiple linear regression analysis was used to calculate the correlation between the resultant between-group different ROI-based data and behavioral performance, with age, sex, and education years as covariates. The threshold value for establishing significance of group effects was set at P < 0.05 (uncorrected for multiple comparisons).
Results
Demographics and Neuropsychological Testing
There were no significant differences in age, sex, years of education, BMI, total cholesterol, triglycerides, HDL cholesterol, or LDL cholesterol between T2DM and control subjects. As expected, HbA1c (P < 0.001) and FPG levels (P < 0.001) were elevated in the T2DM group (Table 1). In T2DM patients, cognitive function in the domains of executive function, spatial processing, attention, and working memory were significantly worse than healthy control subjects. Demographic data and neuropsychological testing are presented in Table 2.
Demographics and clinical characteristics for each group
Neuropsychological test results for T2DM and control groups
WM Skeleton Voxel-Wise TBSS Comparisons
Voxel-wise TBSS statistical analyses revealed significantly decreased FA in widespread WM tracts in T2DM patients compared with control subjects, including the whole corpus callosum (CC), the bilateral corona radiata, internal capsule (IC), posterior thalamic radiation, cingulum (cingulate gyrus), superior longitudinal fasciculus, inferior longitudinal fasciculus, superior fronto-occipital fasciculus, external capsule (EC), fornix/striaterminalis, uncinate fasciculus, tapetum, and left cingulum (hippocampus) (P < 0.05, FWE corrected) (Fig. 2 and Table 3). Meanwhile, significantly increased MD and λ23 were found in only some of the above tracts in T2DM patients compared with control subjects at P < 0.05 (FWE corrected). Increased MD was observed in the whole CC, bilateral corona radiata, IC (except the right posterior limb of it), EC, posterior thalamic radiation, cingulum (cingulate gyrus), superior longitudinal fasciculus, and tapetum (Fig. 2 and Table 3). Increased λ23 was observed in the whole CC, bilateral corona radiata, IC (except the right posterior limb), posterior thalamic radiation, cingulum (cingulate gyrus), superior longitudinal fasciculus, inferior longitudinal fasciculus, EC, fornix/striaterminalis, uncinate fasciculus, tapetum, and right superior fronto-occipital fasciculus (Fig. 2 and Table 3). However, there were no significant tract-specific λ1 differences between the two groups at P < 0.05 (FWE corrected).
Voxel-wise TBSS analysis results of FA, MD, λ1, and λ23 images between T2DM and control groups. Green represents the mean WM skeleton of all subjects. Blue–light blue (thickened for better visibility) represents regions with decreased FA in T2DM group (P < 0.05, FWE corrected). Red-yellow (thickened for better visibility) represents regions with increased MD and λ23 in T2DM patients compared with control subjects (P < 0.05, FWE corrected).
Diffusion changes in the WM tracts in diabetic patients during the voxel-wise and ROI-wise statistical analysis
To identify the more severely impaired WM tracts in T2DM patients, we conducted a further investigation by using more strict correction criterions of P < 0.01 and P < 0.002 (FWE corrected) in the voxel-wise TBSS statistical analysis. This process was only carried out on FA maps. For P < 0.01 (FWE corrected), significantly decreased FA was observed in the whole CC, bilateral corona radiata, IC (except the right posterior limb), posterior thalamic radiation, superior longitudinal fasciculus, inferior longitudinal fasciculus, superior fronto-occipital fasciculus, EC, cerebral peduncle, tapetum, left cingulum (hippocampus), fornix/striaterminalis, and uncinate fasciculus in T2DM patients compared with control subjects (Fig. 3 and Table 3). For P < 0.002 (FWE corrected), significantly decreased FA was found in the CC, bilateral corona radiata, left anterior limb of the IC (ALIC.L), EC, and posterior thalamic radiation (Fig. 3 and Table 3).
Further results of the TBSS analysis using stricter correction criterions (all after FWE correction). Green represents the mean WM skeleton of all subjects. Blue–light blue (thickened for better visibility) represents regions with decreased FA in T2DM group.
Group Differences of Atlas-Based Tract ROIs
Figure 4 illustrates the mean diffusion metrics of each atlas-based tract ROIs with significant between-group differences after FDR correction in the T2DM and control groups. Compared with control subjects, T2DM patients had significantly lower FA in the whole CC, ALIC.L, sagittal stratum, EC, uncinate fasciculus, right superior fronto-occipital fasciculus, and bilateral tapetum (P < 0.05, FDR corrected). However, there were no significant differences in MD, λ1, and λ23 between the two groups after FDR correction.
Mean diffusion metrics and group differences of each atlas-based tract ROIs in T2DM and control groups. All the ROIs shown in the figure were significantly different after FDR correction. For the abbreviations of WM tracts, see Supplementary Table 1.
Correlations Between ROI-Wise Diffusion Metrics and Behaviors
We next examined the relationship between regional diffusion metrics of ROIs with significant group effects (P < 0.05, FDR corrected, i.e., the FA values of those 10 ROIs in ROI-wise analysis) and neuropsychological scores with significant group differences (P < 0.05, i.e., R-O copy, Stroop C-B time, SDMT, and backward recall) in two groups. In the T2DM group, only a negative correlation between the mean FA value of the left EC (EC.L) and Stroop C-B time was found (r = −0.376, P = 0.026) (Fig. 5). In addition, a marginally significant correlation in T2DM patients between the mean FA of the ALIC.L and the Stroop C-B time score (r = −0.323, P = 0.059) was also taken into consideration (Fig. 5).
The significant correlations between ROI-wise diffusion metrics and behaviors in T2DM and control groups, respectively. For the abbreviations of WM tracts, see Supplementary Table 1.
Discussion
In the current study, we found that T2DM patients showed 1) various cognitive impairments, including executive function, spatial processing, attention, and working memory; 2) widespread WM disruptions, especially in the whole CC, ALIC.L, and EC.L; and 3) a positive correlation between executive function and WM integrity in ALIC.L and EC.L. Considering that WM microstructural impairments underlie common mechanisms of cognitive dysfunction and disrupt the large-scale distributed brain cognitive networks, our findings imply that extensive WM destruction plays a distinct role in cognition decline associated with T2DM. Our investigation provides novel insight into the neuropathological effects of WM integrity reduction on cognition in T2DM.
FA and MD are the primary DTI-derived metrics believed to reflect overall WM health, maturation, and organization (15,16). In addition to these primary DTI measures, λ1, which reflects axon integrity, and λ23, which reflects myelin sheath integrity, can be useful in understanding the underlying physiology (8,29). Based on the voxel-wise TBSS statistical analysis, we found decreased FA in almost the whole cerebral WM skeleton in T2DM (P < 0.05, FWE corrected) and increased MD in the whole CC, bilateral corona radiata, IC (except the right posterior limb of it), EC, posterior thalamic radiation, cingulum (cingulate gyrus), superior longitudinal fasciculus, and tapetum. These widespread WM microstructural impairments disrupt the large-scale distributed brain cognitive networks and underlie the various cognitive dysfunctions in T2DM. To date, there are several studies suggesting various WM impairments in diabetic patients. Partly consistent with our results, Kodl et al. (22) suggested that diabetic subjects had significantly lower FA than control subjects in the posterior corona radiata and the optic radiation. Hsu et al. (7) reported a remarkable FA decrease in the bilateral frontal WM in T2DM patients by assessing the DTI with VBA. Hoogenboom et al. (8) found that T2DM patients showed lower FA in the cingulum bundle and the uncinate fasciculus. Despite these findings, disrupted WM in T2DM patients in our study exhibited significantly decreased FA, and increased MD and λ23, with no striking between-group differences in λ1. Accordingly, the alteration of FA and MD might be mainly attributed to the increase in λ23 but not λ1. Given that λ23 is a specific marker of myelin alterations whereas λ1 is more related to axonal injury (30), our data showed that demyelination may be an important contributing factor to WM abnormalities in T2DM.
To identify the more severely impaired WM tracts in T2DM patients, we used a stricter correction criterion (P < 0.002, FWE corrected) in the voxel-wise TBSS statistical analysis and found significant differences between the T2DM and control group in the FA values of eight WM tracts. Furthermore, we measured the mean value of DTI metrics of ROIs by ROI-wise statistical analysis. Significantly reduced mean FA values were detected in 10 WM ROIs in T2DM patients (after FDR correction). Combining the results of the two analyses (both the ROI-wise statistical analysis and the voxel-wise TBSS statistical analysis), T2DM patients showed WM microstructural impairments in WM tracts of whole CC, the ALIC.L, and the EC.L when compared with the control group. The CC plays an important role in interhemispheric functional integration. In AD (31,32) and MCI (32,33), the loss of WM connectivity and regionally specific atrophy of the CC are observed. It follows that CC abnormalities may be related to cognitive deficits. The ALIC.L is a portion of the IC that connects the medial and anterior thalamic nuclei with the prefrontal cortex and the cingulate gyrus. Several studies have found that the ALIC also plays an important role in cognition function. The lower FA value of the ALIC correlates significantly with impaired declarative/episodic memory performance (34) and cognitive levels (35). The EC is a route for cholinergic fibers from the basal forebrain to the cerebral temporal cortex, transmitting auditory and polymodal sensory information. As reported, the EC is crucial for cognition, including memory and executive functioning (36). Some studies (37) suggested that T2DM exacerbates the damage of WM integrity in the IC and EC bilaterally, which can result in executive dysfunction and memory deficits and may be a downstream consequence of poor learning efficiency secondary to the executive dysfunction. Therefore, these three WM tracts may play particularly important roles in T2DM-induced cognitive dysfunction.
To further understand the brain mechanism of cognitive impairment, we performed a correlation analysis between WM impairment and cognition decline. We found FA in the EC.L and ALIC.L positively correlated with Stroop C-B time in the T2DM group, which represents executive function. Executive functions are high-level cognitive functions for the management of a series of cognitive processes (38), including working memory, problem solving, planning, and execution (39). Several studies reported a close relationship between executive dysfunction and WM impairment. Sun et al. (40) found declines in executive function in mild WML subjects. Zheng et al. (41) revealed that executive function directly correlated with FA in frontal WM tracts, especially the ALIC. Smith et al. (42) found that WMH volume in the ALIC was strongly inversely associated with executive function. Similarly, subjects with WM integrity impairments of the EC, reflected by decreased FA, also showed significantly poorer performance on executive function tasks (43). Therefore, we propose that executive function might be partially supported by the WM integrity of the ALIC and EC. On the other hand, considering that there were no significant associations between WM integrity and cognition in the control group, we believe that executive function may be more sensitive to WM integrity variations in T2DM. WM integrity could be more beneficial to neuropsychological performance in T2DM subjects than in control subjects.
As a major risk factor for dementia, diabetes is reported to be closely associated with the pathological processes of various dementias such as AD and frontotemporal dementia (FTD). AD can be regarded as a cortical disconnection syndrome that affects not only the cortical neurons but also the axons and dendrites in the cerebral WM (18). MCI, the prodromal phase of AD, impairs memory function and disrupts widespread WM tracts (19). Although extensive WM alterations were also exhibited in T2DM patients in the current study, some studies suggested that T2DM might not be a driving factor of AD but might add to the damaging effects of AD on cognition. Correspondingly, executive dysfunctions are closely related to WM abnormalities in T2DM. At the same time, many investigations have found that FTD affects executive function through regional disruption of the frontal and temporal lobes (44), where the more severely disrupted WM tracts were found in T2DM. Therefore, we predict that T2DM might be associated with FTD. However, the relationships between T2DM and AD or FTD are not clear to date. An interaction analysis of T2DM and MCI, as well as follow-up investigations, is worthy of further study in our future research.
It is also interesting that we found that there were more abnormal WM tracts in the left hemisphere than in the right. Thompson et al. (45) suggest that the shifting GM deficits in AD were asymmetric (left > right hemisphere) and correlated with progressively declining cognitive status. The HAROLD (hemispheric asymmetry reduction in older adults) model states that age-related hemispheric asymmetry reductions may function to compensate for the lower prefrontal activity lateralization during cognitive performances in older adults than in younger adults (46). Accordingly, the mechanisms of how T2DM affects hemispheric symmetry of WM integrity are worthy of further investigation. Additionally, our study is limited by its cross-sectional design, which should be interpreted cautiously, and longitudinal studies are needed to investigate the conversion of T2DM to dementia and to evaluate clinical values of WM metrics to predict longitudinal changes. Further, we only focused on the WM changes in T2DM patients in this study. The relationship between these functional and structural changes, however, is still unclear. In future studies, the putative connection between functional connectivity and WM degeneration could be examined.
In conclusion, there are widespread WM integrity disruptions, which are likely caused by demyelination, and various cognitive impairments in T2DM patients, among which executive dysfunction closely correlates with WM abnormalities. The current investigation will contribute to a better understanding of the neuropathological process in T2DM and may lead to better imaging-based biomarkers for prevention and early treatment of cognitive dysfunction caused by T2DM.
Article Information
Funding. This work was supported by the Beijing New Medical Discipline Based Group (grant 100270569), the Natural Science Foundation of China (grants 30873458 and 81173460), the Institute of Basic Research in Clinical Medicine, China Academy of Chinese Medical Sciences (grant Z0175), and the program for New Century Excellent Talents in University (grant NCET-10-0249).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. J.Z. performed the experiments and wrote the manuscript. Yu.W. performed the experiments, analyzed the results, and wrote the manuscript. J.W., X.Z., and N.S. performed the experiments. Yo.W. and Z.Z. designed the experiments and interpreted the results. 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.
Footnotes
This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db14-0342/-/DC1.
- Received February 27, 2014.
- Accepted May 22, 2014.
- © 2014 by the American Diabetes Association. 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.
References
In this Issue
