Altered Brain Activation Patterns Under Different Working Memory Loads in Patients With Type 2 Diabetes

Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder that is primarily characterized by insulin resistance. According to the International Diabetes Federation, the number of people with diabetes worldwide will rise to 552 million by 2030 from the 366 million people who had diabetes in 2011 (1). The prevalence of T2DM has been estimated to be 9.7% in China and has been growing over recent years (2). In addition to the known insulin signaling dysfunction, patients with T2DM can have cognitive deficits that involve many domains, such as executive functions, memory, attention, and visuospatial abilities (3). T2DM may accelerate the aging process of the brain and increase the risk for Alzheimer disease (AD) (4). Conversely, a convincing view suggested that many AD patients have comorbid T2DM (5). Therefore, an important contribution to early AD intervention would be to focus on the intrinsic brain mechanisms associated with cognitive function in T2DM.

Many cognitive performances, such as working memory (WM), deteriorate in older adults compared with younger adults and teenagers (6). Given that AD is a high-incidence disease of elderly people, WM impairment is generally a considerable cognitive deficit of AD throughout the development of the disease. As proposed by Baddeley (7), WM provides the temporary storage of and operations on information necessary for advanced cognition. Some studies have shown that WM deficits furtively emerge with the early pathology that involves the accumulation of neuritic plaques and neurofibrillary tangles in the frontal cortex (8). There has been growing interest in detecting how WM is influenced during aging and in AD and mild cognitive impairment (MCI). In older adults, the behavioral testing of WM has been associated with extent of frontal activation, which may suggest that the decreased frontal cortex contributions to cognitive experience ultimately result in age-associated decreases in memory ability (9). A previous functional magnetic resonance imaging (fMRI) study demonstrated that during an n-back task assessing WM, patients with MCI exhibited reduced activation in the frontoparietal regions compared with control subjects (10). However, AD and MCI patients may exhibit more activation than control subjects in the right frontal and bilateral middle frontal gyri in another visual WM task, suggesting compensation for decreased neural efficiency (11). Thus, it is conceivable that cognitive functions such as WM are particularly susceptible to the effects of aging and some diseases, including AD (12).

Previous studies of T2DM using magnetic resonance imaging demonstrated that brain atrophy appears in the hippocampal, amygdalar (13), and prefrontal (14) areas of the brain. Some researchers have emphasized changes in the white matter and identified decreases in fractional anisotropy in the bilateral frontal white matter (15) and extensive white matter abnormalities in the superior longitudinal fasciculus, uncinate fasciculus, inferior longitudinal fasciculus, and splenium of the corpus callosum of both hemispheres (16) in T2DM patients compared with healthy control subjects. Coincidently, a growing body of literature has been revealing WM deficits in T2DM patients (17).

To date, however, few studies have focused on how and the extent to which T2DM affects WM tasks and related brain function. The identification of brain activation patterns in T2DM could enhance our understanding of cognitive dysfunctions such as in WM. The correlations between brain activation and behavioral performance may delineate neural mechanisms during neurodegeneration in T2DM.

We conducted an fMRI study of subjects performing a visual n-back WM task to determine brain activation differences in older T2DM patients without dementia. To our knowledge, there are no publications of data about brain activity alterations in older T2DM patients under different WM loads. We hypothesized that patients with T2DM would exhibit WM deficits and altered patterns of brain activation when performing the WM task.

The participants in this study were all from the Beijing Aging Brain Rejuvenation Initiative (BABRI), which is an ongoing, longitudinal study investigating aging and cognitive impairment in urban elderly people in Beijing, China. In this cross-sectional study, all participants were selected according to the following criteria: 1) no less than 6 years of education; 2) scores ≥24 on the mini-mental state examination (MMSE), Chinese version; 3) no history of coronary disease, nephritis, tumors, gastrointestinal disease, or psychiatric illness; and 4) ability to meet the physical demands of the imaging procedure. Sixty-seven participants met these criteria, including 30 patients with diabetes and 37 healthy control participants. T2DM was diagnosed using established criteria on the basis of patients’ medical histories, medication use, or fasting plasma glucose (FPG) levels (≥7 mmol/L) (18). All participants underwent a medical history and physical examination during which height, weight, and body mass index (BMI) were recorded. FPG, glycosylated hemoglobin (HbA_1c), and cholesterol levels were measured by standard laboratory testing. Of the 30 T2DM patients, 6 were under treatment with insulin and 24 controlled blood glucose with oral hypoglycemic agents (17 with glucosidase inhibitors, 4 with gliquidone, 3 with biguanides). All participants self-reported to have never experienced severe hypoglycemia and to have a low frequency of minor hypoglycemia.

All participants were subjected to a battery of neuropsychological tests that assessed their general mental status and other cognitive domains, such as memory, attention, spatial processing, executive function, and language abilities. As previously mentioned, general mental status was assessed with the MMSE (19), and scores ≤23 were a reason for exclusion because subjects with such low scores were considered to have possible dementia. The comprehensive neuropsychological battery comprised the following five cognition domains and associated tests: 1) memory [Auditory Verbal Learning Test (20), Rey-Osterrieth Complex Figure test (ROCF)-delay recall (21), and the digit span test (22)]; 2) attention [Trail Making Test (TMT) A (23), Symbol Digit Modalities Test (24), and Stroop Color and Word Test (SCWT) B (25)]; 3) visuospatial ability [ROCF-copy (21) and Clock Drawing Test (26)]; 4) language [Category Verbal Fluency Test (CVFT) and Boston Naming Test (27)]; and 5) executive function [TMT-B (23) and SCWT-C].

A blocked periodic design that incorporated alternating 0-back, 1-back, and 2-back tasks was used during the n-back WM task. Participants viewed single digits (0–9, white on a black background) in pseudorandom order. During the 0-back task, the participants were asked to decide whether the target digit (e.g., 1) on the screen was the preassigned digit. During the n-back (n = 1 or 2) task, the participants were required to press a button when a digit appeared that matched the one presented n positions ago in the sequence. The stimuli were presented on a personal computer using E-Prime version 1.0 software (Psychology Software Tools, Inc., Pittsburgh, PA). To ensure that each participant understood the instructions and performed the task correctly, the participants were asked to practice all blocks of the task for 10–15 min before the experiment.

The experiment comprised three blocks of each of the three WM load conditions presented in a random order, with 20 stimulus digits presented per block. At the beginning of each block, the instruction for the condition was presented on the screen for 10 s. Each digit was presented for 1,000 ms, with a blank display for 1,000 ms. The entire fMRI session took ∼8 min.

The fMRI data were acquired with a Siemens Trio 3T scanner in the Imaging Center for Brain Research, Beijing Normal University. WM task images were acquired using an echo planar imaging sequence with the following parameters: 33 axial sections, repetition time of 2,000 ms, echo time of 30 ms, section thickness of 3.5 mm, flip angle of 90°, field of view of 200 × 200 mm, and acquisition matrix of 64 × 64. For each participant, 235 image volumes were obtained.

Functional data were preprocessed and statistically analyzed with the SPM5 software package (http://www.fil.ion.ucl.ac.uk/spm/software/spm5). The preprocessing procedures included slice timing correction, within-subject interscan realignment to correct for possible movement, spatial normalization to a standard brain template in the Montreal Neurological Institute coordinate space, resampling to 3 × 3 × 3 mm, and smoothing with an 8-mm full-width half-maximum Gaussian kernel. On the single-subject level, the data were analyzed according to the fixed-effects model (SPM5). The six head movement parameters were included in the model as confounding factors. Contrast images were created by subtracting the 0-back images from the n-back (n = 1 or 2) images. On the second level, activation differences between groups were computed by two-sample t tests using these single-subject contrasts. An AlphaSim-corrected threshold of P < 0.01 was considered significantly activated.

In addition to the voxel-wise analyses, several regions of interest were identified based on the significant results from the comparisons between groups and were used to probe the relationship between regional activation and behavior in the T2DM group. Spherical regions (6-mm radius) were defined around each peak activation. Signal intensities (β weights) from the significantly activated voxels for each region of interest were then calculated separately for each task comparison (2-back vs. 0-back and 1-back vs. 0-back). Furthermore, Pearson correlation analyses were used for all participants to assess the relationships between the β weights and cognitive performance.

Independent two-sample t tests were used to assess the between-group differences in age and education. The χ² test was used to compare the sex ratios of the groups. For the neuropsychological assessment and lipid levels, an ANCOVA was used to test for between-group differences (age, sex, and education were included as covariates). Pearson correlation analyses were performed to explore the relationship between the activities of the brain areas with significant intergroup differences and the neuropsychological performance of the T2DM group after controlling for the influences of age, sex, and education. All statistical analyses were performed using SPSS version 17.0 for Windows software.

The groups did not differ significantly in age, sex, or years of education. The neuropsychological and biochemical measures were analyzed with ANCOVA after adjusting for age, sex, and education. Compared with the control participants, the patients with T2DM performed significantly worse on the WM tasks (backward digit span, P = 0.027; digit span, P = 0.05) and executive function task (SCWT-C-B time, P = 0.045). The patients with T2DM had significantly higher HbA_1c and glucose levels and BMIs than the healthy control participants (Table 1). The response times and accuracies in the 1-back task were significantly different between the T2DM and control participants but were no different in the 2-back or 0-back tasks, but there was a trend toward an increase in the T2DM participants (Table 2).

Both groups exhibited activation in regions associated with WM tasks, such as the middle frontal gyrus (MFG), inferior frontal gyrus (IFG), medial frontal gyrus/supplementary motor areas, and inferior parietal gyrus, in both the 1-back versus 0-back and 2-back versus 0-back contrasts (Supplementary Fig. 1).

We examined the main effects of condition and group and their interaction under the framework of ANOVA. No areas exhibited a group (patients and control participants) by condition (1-back vs. 0-back or 2-back vs. 0-back) interaction. More central to our research interests, we found several frontal areas (including the right superior frontal gyrus [SFG], bilateral MFG, and IFG) that exhibited significant differences in activity between the groups. Patients with T2DM showed less brain activation than the control participants in these areas. Additionally, the postcentral gyrus, superior temporal gyrus, medial frontal gyrus, and precuneus exhibited main effects of condition and showed higher activation in the high-load condition (i.e., the 2-back vs. 0-back task) than in the low-load condition (i.e., the 1-back vs. 0-back task) (Supplementary Fig. 2).

We also examined possible group differences in the two conditions. Comparisons between the groups for the 1-back versus 0-back condition revealed that the T2DM patients exhibited diminished activation in the left IFG (Brodmann area 13) compared with the control participants. The T2DM patients exhibited less activation in the left MFG and left SFG than the control participants in the 2-back versus 0-back contrast (Fig. 1).

Pearson correlation analyses indicated that higher SFG activity was associated with better performances on the MMSE (r = 0.57, P = 0.002), ROCF-delay recall (r = 0.50, P = 0.009), digit span (r = 0.48, P = 0.011), CVFT (r = 0.43, P = 0.025), ROCF-copy (r = 0.57, P = 0.002), and SCWT-B time (r = −0.42, P = 0.031) tests after controlling for the effects of age, sex, and education (Fig. 2).

For the first time in our knowledge, the current study found altered patterns of brain activation under different WM loads in T2DM patients. First, T2DM patients exhibited cognitive deficits in the domains of memory and executive function. Second, they showed reduced activation in the left IFG in the easy WM task as well as in the left MFG and SFG under the more difficult WM task compared with the healthy control participants, demonstrating that with increasing task difficulty, activation in the frontal areas of the brain also increases. Third, the study revealed that greater activation in the left SFG correlates with better cognitive performance in multiple tasks, including the MMSE, ROCF-delay recall, digit span, CVFT, ROCF-copy, and SCWT-B tests.

The T2DM patients exhibited diminished activation in the left IFG compared with the healthy control participants in the low WM load condition (i.e., the 1-back vs. 0-back task contrast). Some studies using positron emission tomography imaging and fMRI revealed that the IFG plays a critical role in verbal WM (28,29). It has been documented that the IFG is specifically involved in memory impairments during aging (30) and that obese older adults exhibit significantly smaller gray matter volume in the left IFG relative to their healthy counterparts (31). Evidence from separate lines of investigation suggests that AD and MCI patients exhibit reduced activation in the IFG during several memory tasks compared with control subjects (32). Although the mechanisms of WM impairments in T2DM are not yet understood, these findings may indicate that the IFG is particularly damaged or exhibits decreased connectivity with other susceptible regions that adversely affects memory during aging and disease processes and that this region is targeted in diabetes.

More importantly, the T2DM patients in the current study exhibited decreased activation in the left MFG and SFG relative to the control participants in the high WM load condition (i.e., the 2-back vs. 0-back task). These findings indicate that with increasing task difficulty, additional frontal brain regions might have been recruited to play compensatory roles to complete the harder task. In general, the MFG plays a key role in maintaining information and supporting executive function during WM (33), and the SFG is recruited during object location in WM (34). In particular, results from a previous report indicated that gray matter density loss in the bilateral medial frontal gyrus and SFG of type 1 diabetic patients is significantly related to the severity of retinopathy, a common diabetes complication (35). In further support of the current findings, one study found that reduced activation in the MFG also appears in mild to moderate AD during a verbal short-term memory task (32). In summary, we suggest that impairments of MFG and SFG are quantitatively phenotypic of diabetes severity and deficits in the manipulation of stored information, a finding that helps us to better understand the effects of T2DM on the brain. Moreover, the current results are somewhat similar to those of the Kochan et al. (36) study, which found a greater number of affected regions under greater loads relative to lower loads in MCI patients. However, some contradictory studies have found enhanced activation in the left MFG in middle-aged T2DM patients during the 2-back task, which suggests that the MFG potentially serves a compensatory role in the memory network (37). This discrepancy may have appeared due to variations in task selection and processing methods. Clinically, insulin resistance was negatively correlated with neural activity in the frontal regions in T2DM patients (38). Other abnormal levels of brain metabolites, including reduced N-acetylaspartate and increased myoinositol and choline in the frontal lobe, were found in diabetic patients compared with healthy control subjects (39), which is consistent with the report that impairments of neuronal density or viability, glial proliferation, and demyelination emerge in diabetes (40).

The current examination of the relationship between brain activation and behavioral performance in patients with T2DM revealed that activation in the left SFG was positively related to MMSE, ROCF-delay recall, digit span, CVFT, and ROCF-copy scores and negatively related to the SCWT-B time score. These findings support the idea that lower SFG activation was associated with worse cognitive function, including general mental status, WM, language, visuospatial ability, and attention. Consistent with these findings, emerging data suggest that the blood oxygen level–dependent signals from the SFG and MFG are positively correlated with major cognitive functions, such as short-term memory, attention, and language, in older subjects (41). Studies on middle-aged adults with cognitive complaints also have shown that lower activation in the SFG is related to poorer WM performance (42). Moreover, in light of recent white matter studies, increased impairments in the frontal fiber tracts were associated with reduced executive function (43) and information processing speed (16) in T2DM patients. Demyelination of axons, loss of glial cells, expanded microvascular lesions, enlarged perivascular spaces, and even blood-brain barrier disruptions possibly induce neuropathological changes in brain structure and function in T2DM patients (44). As a corollary, the current findings of significant correlations could partly explain T2DM as a risk factor for cognitive impairment.

Nevertheless, we did not find any relationships between the activated regions during low WM load, the IFG and the cognitive performance of patients with T2DM possibly because these patients’ brains were able to recruit additional processing resources to compensate for cognitive limitations during the low-load task. Therefore, the relationship between IFG activation and cognitive behavior was weakened. In contrast, the T2DM patients had few additional resources available in the high-load conditions and, thus, exhibited a significant correlation between the SFG impairments and cognitive processes. We have previously found some evidence from an examination of the effect of APOE4, the strongest genetic risk factor for AD, on brain activation in response to different WM loads (45).

We also examined the group (patients vs. control participants) × condition (1-back vs. 0-back and 2-back vs. 0-back) interactions to make our analysis more thorough, but no regions exhibited a significant group × condition interaction. Some of the areas in the frontal, parietal, and temporal regions exhibited main effects of group or condition. We conjecture that the lack of significance of the interactions suggests that the regions with altered activation between groups did not survive when the activation differences between conditions were examined.

This study had several limitations. First, participants with different social class or level of educational achievement may have experienced different effects. Second, we only performed testing of patients with current cognitive function, but assessments of baseline/premorbid intelligence are important for exploring the mechanism and consequences of T2DM. Therefore, the effects of social class, level of educational achievement, and baseline/premorbid intelligence should be studied in future work.

In conclusion, this study provides insight into the alterations of brain activation patterns under different WM loads in T2DM. Specifically, a greater number of frontal regions exhibited diminished activation under the high WM load relative to the low WM load. To our knowledge, this study is the first to detect the specific brain mechanisms related to diabetes-induced WM dysfunction. This study provides evidence about the neural mechanisms that underlie cognitive impairment in T2DM and may lead to establishing potential imaging-based biomarkers for the prevention and early treatment of cognitive dysfunction caused by T2DM. Further longitudinal studies are needed to verify these findings and to evaluate the clinical value of brain imaging methods in the prediction of disease progress.

Acknowledgments. The authors thank all the volunteers and participants for participating in the study and Lu Zheng, Lei Qiu, Xiaotang Zhu, Yao Zhang, and Tianjiao Feng, State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, for community contact and data acquisition.

Funding. This work was supported by the Beijing New Medical Discipline Based Group (grant number 100270569), the Natural Science Foundation of China (grant numbers 30873458 and 81173460), project of Institute of Basic Research in Clinical Medicine, China Academy of Chinese Medical Sciences (grant number Z0175), and program for New Century Excellent Talents in University (grant number NCET-10-0249).

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

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