In summary, there were significant regional differences in cortical thickness between depressed diabetic subjects and healthy controls. The regional thinning in subjects with diabetes only was noted to a lesser degree compared to healthy subjects. The less extensive cortical thinning in diabetic subjects without depression was suggestive of a greater influence of major depression over type 2 diabetes in subjects with both diseases. Clinical variables representing stroke risk (CVRF) were significantly correlated with cortical gray matter thinning across all subjects in our study population. Performance on attention and executive function tasks were significantly correlated with cortical gray matter thickness across all subjects.
In our study, the regions of significant cortical thinning associated with major depression and type 2 diabetes included the medial superior frontal gyrus and the anterior cingulate. Our previous volumetric study also identified gray matter decreases in the anterior cingulate of diabetic subjects (Kumar et al. 2008
). This study used more conventional techniques measuring gray matter volumes, in contrast to the cortical thickness measurement employed in the present study. Cortical thickness measures have been used in a number of morphometric studies and it represents the combination of neurons, glial cells, axons and dendritic arborization. Reductions in any of these elements may account for the decreases in cortical thickness observed. In the existing literature, the anterior cingulate has been implicated in a number of studies examining neuroanatomical differences associated with major depression. Gray matter volume losses in the anterior cingulate have been shown to occur in late-life depression and depression in adults (Ballmaier et al. 2004b
; Caetano et al. 2006
; Tang et al. 2007
). The anterior cingulate is also the locus of neurochemical differences associated with depression such as lower glutamate/glutamine as measured by magnetic resonance spectroscopy (Auer et al. 2000
). While there is less data on the importance of the superior frontal gyrus in mood regulation, a recent intriguing functional MRI study included this region as a part of cortical network associated with the feeling of regret (Chandrasekhar et al. 2007
). In another functional study, decreased perfusion was noted in the left superior frontal gyrus in depressed patients performing the Tower of London task (Goethals et al. 2005
The complications associated with diabetes are primarily due to vascular compromise which can lead to renal disease, retinal disease, and stroke. Therefore, we wanted to examine the relationship of medical co-morbidities and vascular risk to cortical gray matter thickness. There was no a priori
hypothesis as to which regions would be affected by our measure of medical co-morbidities or stroke risk; however it is notable that regions of cortical thinning in depressed diabetic subjects overlapped with areas of correlative significance. Numerous previous studies have examined the correlation of white matter changes to vascular risk factors and medical co-morbidities (Jeerakathil et al. 2004
; Hickie et al. 2005
). These correlations have also demonstrated implications for the progression of mood disorders. For example, vascular risk has also been associated with cognitive decline and dementia in elderly depressed patients (Steffens et al. 2007
). While we have recently reported gray matter volumetric differences associated with vascular risk factors (Kumar et al. 2008
), to our knowledge, this is the first report of cortical gray matter thickness associations with clinical variables such as stroke risk.
Previous studies from our group have shown that diabetic subjects (both with and without depression) had performance deficits in executive function compared with healthy control subjects and depressed diabetic subjects perform significantly worse on tasks of attention processing compared to healthy control subjects (Watari et al. 2006
; Watari et al. 2007
). When correlating cortical gray matter thickness with cognitive performance in tasks of attention processing and executive function, significant regional differences were noted predominately in the left hemisphere. The strong correlation between simple attention and left parietal tissue around the post-central sulcus, particularly the posterior bank, suggests a relationship between the attentional network and cortical thinning. Mesulam identified three regions that, if compromised by lesions, would lead to attentional neglect, and one region was the dorsolateral parietal cortex just below the intraparietal sulcus (Mesulam 1990
). Functional MRI has shown the areas within this region to be involved in attention to explicit speech (Sabri et al. 2008
), implicit speech (Geiser et al. 2008
), and even the relationship between verbal and nonverbal stimuli (Noordzij et al. 2008
). The cortical thickness differences that correlated with executive function expanded to include the more of the occipital cortex. Some of the executive tests were nonverbal tests that required spatial analysis, so different areas within the inferior parietal may be dedicated to different types of stimuli.
The frontopolar region exhibited significant cortical thickness differences associated with performance on executive function. This highly developed area is well-known to coordinate executive functions involved in complex decision-making (Koechlin and Hyafil 2007
). Another area of correlation between cortical thickness differences and cognitive function occurred in the temporal pole. A recent theory posits that the temporal pole is critical in the binding of highly processed perceptual inputs to visceral emotional responses (Olson et al. 2007
). Intuitively, it seems appropriate that the temporal pole would be associated with cognitive testing. However, until more definitive evidence is available, the possibility must be considered that the correlations with temporal pole and parietal cortex are more associated with general testing ability and confidence in a testing situation than with the specific types of tests administered.
The results discussed in this study are presented within the context of a few limitations. For example, the lack of a significant difference between subjects with diabetes only and healthy control subjects may reflect a limitation of our study involving the recruitment of diabetic subjects from outpatient clinics. These patients were well-monitored as outpatients and subsequently represented a “healthier” population of diabetic subjects than would be found in a more general community sample. This is reflected in the relatively low Hgb A1c values seen in our population. In addition, we did not obtain lifetime hypoglycemic episodes in our diabetic subjects. A recent review suggests that recurrent hypoglycemic episodes have an unclear relationship to cognitive function (McNay and Cotero 2010
), thus exploring the role of hypoglycemia with our neuroanatomical analysis may contribute to our understanding of this controversial issue.
The significant cortical thinning seen in depressed diabetic subjects compared to controls may reflect a synergistic relationship between major depression and type 2 diabetes. The lack of a significant effect of type 2 diabetes alone suggests that the effect of cortical thinning may be due solely to major depression. A limitation of the study involved the lack of a depression only subject group to address this particular issue. However, the focus of the present study was to investigate the impact of depression in the context of type 2 diabetes. The inclusion of a depressed, non-diabetic comparator group, as well as lifetime hypoglycemic episodes, will be addressed in future studies.
In conclusion, microvascular compromise from diabetes may contribute as a risk factor for significant cognitive and behavioral sequelae. The results from our study contributes to the notion that medical co-morbidities such as stroke risk associated with type 2 diabetes are related to structural gray matter alterations reflected in pathophysiology of major depression.