Functional Connectivity of Insula, Basal Ganglia, and Prefrontal Executive Control Networks during Hypoglycemia in Type 1 Diabetes

Abstract

Human brain networks mediating interoceptive, behavioral, and cognitive aspects of glycemic control are not well studied. Using group independent component analysis with dual-regression approach of functional magnetic resonance imaging data, we examined the functional connectivity changes of large-scale resting state networks during sequential euglycemic-hypoglycemic clamp studies in patients with type 1 diabetes and nondiabetic controls and how these changes during hypoglycemia were related to symptoms of hypoglycemia awareness and to concurrent glycosylated hemoglobin (HbA1c) levels. During hypoglycemia, diabetic patients showed increased functional connectivity of the right anterior insula and the prefrontal cortex within the executive control network, which was associated with higher HbA1c. Controls showed decreased functional connectivity of the right anterior insula with the cerebellum/basal ganglia network and of temporal regions within the temporal pole network and increased functional connectivity in the default mode and sensorimotor networks. Functional connectivity reductions in the right basal ganglia were correlated with increases of self-reported hypoglycemic symptoms in controls but not in patients. Resting state networks that showed different group functional connectivity during hypoglycemia may be most sensitive to glycemic environment, and their connectivity patterns may have adapted to repeated glycemic excursions present in type 1 diabetes. Our results suggest that basal ganglia and insula mediation of interoceptive awareness during hypoglycemia is altered in type 1 diabetes. These changes could be neuroplastic adaptations to frequent hypoglycemic experiences. Functional connectivity changes in the insula and prefrontal cognitive networks could also reflect an adaptation to changes in brain metabolic pathways associated with chronic hyperglycemia.

Significance statement: The major factor limiting improved glucose control in type 1 diabetes is the significant increase in hypoglycemia associated with insulin treatment. Repeated exposure to hypoglycemia alters patients' ability to recognize the autonomic and neuroglycopenic symptoms associated with low plasma glucose levels. We examined brain resting state networks during the induction of hypoglycemia in diabetic and control subjects and found differences in networks involved in sensorimotor function, cognition, and interoceptive awareness that were related to chronic levels of glycemic control. These findings identify brain regions that are sensitive to variations in plasma glucose levels and may also provide a basis for understanding the mechanisms underlying the increased incidence of cognitive impairment and affective disorders seen in patients with diabetes.

Keywords: executive control resting-state network; functional connectivity; functional magnetic resonance imaging; hypoglycemia; independent component analysis; type 1 diabetes.

Copyright © 2015 the authors 0270-6474/15/3511013-12$15.00/0.

Figures

Figure 1.

Experimental protocol: plasma glucose levels. The experiment used the insulin clamp technique with four successive time periods corresponding to different plasma glucose levels: (1) basal (30 min); (2) euglycemic clamp (40 min, target glucose of 5.0 mmol/L); (3) declining glycemia (40 min, plasma glucose reduced from 5.0 to 2.8 mmol/L); and (4) hypoglycemic clamp (30 min, target glucose of 2.8 mmol/L). Anatomical MRI was performed during baseline, and fMRI was performed during the last 5 min of the euglycemic period and the 40 min of the declining glucose period. For dual regression analysis comparing conditions, the first 10 min and the last 10 min were extracted from the full 45 min time series of the resting state fMRI to create the euglycemic (EU fMRI) and hypoglycemic (HY fMRI) periods, respectively. Plasma glucose levels: dashed line and open circles, T1DM; solid line and filled circles, nondiabetic control (CONT). *p < 0.05 between groups.

Figure 2.

Counter-regulatory hormones and hypoglycemia symptom questionnaire scores. A, Time courses of counter-regulatory hormones. Data are presented as mean ± SE at each time point. The gray region indicates the time period of declining glycemia. Open circles and dashed line, T1DM; filled circles and solid line, nondiabetic controls (CONT). Comparison of T1DM versus CONT: *p < 0.05, **p < 0.01, ***p < 0.001. B, Hypoglycemia symptom questionnaire scores. Open bars, T1DM; filled bars, CONT. Data are presented as mean ± SE. Top, Neuroglycopenic symptom score; middle, autonomic symptom score; bottom, total hypoglycemic (neuroglycopenic + autonomic) symptom score. Comparison of T1DM versus CONT: **p < 0.01, ***p < 0.001. Comparison of euglycemia versus hypoglycemia: ‡p < 0.01.

Figure 3.

Spatial maps of RSNs of interest. Each RSN is shown in green overlaid onto orthographic views of the standard MNI-152 T1 brain image [ICBM152 nonlinear 6th generation asymmetric average brain stereotaxic registration brain model (Grabner et al., 2006)] shown in grayscale (2 mm isotropic resolution). Left, Sagittal view; middle, coronal view; right, axial view. Slice coordinates in standard MNI space are noted in the top left of each slice view. P, Posterior; A, anterior; R, right; L, left; ACC, anterior cingulate cortex; PCC, posterior cingulate cortex; FP, frontal pole.

Figure 4.

Group and condition effects. Regions of the RSNs that are common to both subject groups during the glycemic descent are shown in green. Regions showing significant group effects are shown in a color scale from red to yellow for higher FC in T1DM versus controls (red: p < 0.05; orange: p < 0.03; yellow: p < 0.01), condition effects are shown in a color scale either from red to yellow for increased FC during hyperglycemia versus euglycemia (red, p < 0.05; orange, p < 0.03; yellow, p < 0.01) or from dark blue to light blue for decreased FC during hyperglycemia versus euglycemia (dark blue, p < 0.05; blue, p < 0.03; light blue, p < 0.01). Left, Sagittal view; middle, coronal view; right, axial view. Slice coordinates in standard MNI space are noted in the top left of each slice view. P, Posterior; A, anterior; R, right; L, left; SFG, superior frontal gyrus; PCC, posterior cingulate cortex; FP, frontal pole. A, EU: T1DM > CONT, Regions of higher within-network FC of the hypothalamus–basal ganglia RSN (RSN A) in patients with T1DM relative to controls during euglycemia. B, C, CONT: EU > HY, Regions of decreased within-network FC of the cerebellum–basal ganglia RSN (RSN B) and temporal lobe RSN (RSN C) during hypoglycemia relative to euglycemia in controls. D, E, CONT: HY > EU, Regions of increased within-network FC of the default mode RSN (RSN D) and sensorimotor cortex RSN (RSN E) during hypoglycemia relative to euglycemia in controls. F, T1DM: HY > EU, Regions of increased within-network FC of the medial prefrontal lobe RSN (RSN F) during hypoglycemia relative to euglycemia in patients with T1DM.

Figure 5.

Correlation of FC changes with HbA1c. Top, T1DM: ΔFC α HbA1c, Regions of correlation (α) between HbA1c and within-network RSN F FC changes (ΔFC) for T1DM. Regions of positive correlation with HbA1c are shown in red to yellow scale (red, p < 0.05; orange, p < 0.03; yellow, p < 0.01). Regions of RSN F that are common to both subject groups during the glycemic descent are shown in green. Left, Sagittal view; middle, coronal view; right, axial view. Slice coordinates in standard MNI space are noted in the top left of each slice view. P, Posterior; A, anterior; R, right; L, left; SFG, superior frontal gyrus; ACC, anterior cingulate cortex; FP, frontal pole. Bottom, Plot of RSN F FC changes from euglycemia (EU) to hypoglycemia (HY) versus HbA1c. Open circles, T1DM; filled squares, nondiabetic controls (CONT); dashed line, linear fit for T1DM.

Figure 6.

Correlation of FC changes with symptom score changes. Top, CONT: ΔFC α ΔSX, Regions of stronger correlation (α) between symptom score changes (ΔSX) and within-network RSN B FC changes (ΔFC) in controls relative to T1DM (CONT > T1DM). Regions of stronger correlation are shown in red to yellow scale (red, p < 0.05; orange, p < 0.03; yellow, p < 0.01). Regions of RSN B that are common to both subject groups during the glycemic decline are shown in green. P, Posterior; A, anterior; R, right; L, left. Bottom, Plot of RSN B FC changes versus hypoglycemic symptom score changes from euglycemia (EU) to hypoglycemia (HY). The z transforms of FC change (HY − EU) of cerebellum–basal ganglia RSN within-network FC, averaged over the voxels of the right basal ganglia with the strongest between-group difference (p < 0.05), are plotted on the vertical axis for each individual. Open circles, T1DM; filled squares, nondiabetic controls (CONT); dashed line, linear fit for T1DM; solid line, linear fit for controls.