The Effect of Diabetes on Cortical Function in Stroke: Implications for Poststroke Plasticity

Diabetes is associated with an increased risk of ischemic stroke. Poststroke, patients with diabetes are predisposed to a slower and poorer recovery of function, even after adjusting for factors such as stroke severity and age (1–4). Furthermore, it remains plausible that diabetes may impair the plasticity of neural circuitry after stroke. Successful recovery from stroke requires the brain to remap sensorimotor functions to surviving and functionally homologous regions within the brain network (5,6).

Separately, diabetes may directly impair the central nervous system with impairments in cognitive function, synaptic plasticity, synaptogenesis, and neurogenesis (3), as well as direct damage to neurons (7). Structural and functional imaging has consistently revealed varying degrees of brain atrophy with disturbed white matter connectivity in patients with diabetes relative to control subjects (8–12).

Neuroplasticity is defined as the ability of the nervous system to respond to intrinsic and extrinsic stimuli by reorganizing its structure, function, and connections. More specifically, it is important to recognize that neuroplasticity is defined by change in neuronal structure or function, not the observation of change in behavior. For this reason, it can be viewed as adaptive when associated with a gain in function or as maladaptive when associated with negative consequences such as loss of function (13,14).

Transcranial magnetic stimulation (TMS) is a useful way to noninvasively index neuroplasticity through measures of cortical excitability. The assessment of inhibitory and facilitatory cortical circuits provides an important component of understanding how neuroplastic changes are mediated and how they underlie associated behavioral and functional changes after disease or injury (15,16).

Given this background, the current study hypothesized there may be cortical dysfunction present in patients with diabetes and that these patients may demonstrate an impaired capacity for cortical plasticity after an acute stroke. This may consequently have implications for future developments of neuroprotection and clinical trials aimed at facilitating poststroke recovery. In this regard, the current study used threshold-tracking TMS to assess intracortical excitability and thereby function over both the lesioned and contralesional motor cortices of stroke patients with and without diabetes and compared results to findings obtained from a cohort of nonstroke patients with diabetes and normal control subjects.

Patients diagnosed with acute unilateral ischemic stroke (with brain MRI demonstrating acute infarct on diffusion-weighted imaging) after presenting to a tertiary teaching hospital stroke unit were recruited and studied within the acute period poststroke. Exclusion criteria were 1) previous history of stroke, 2) cognitive impairment or dysphasia sufficient to affect informed consent, 3) drugs or neurological disorders beyond stroke that may affect cortical excitability, and 4) any contraindications to TMS. All subjects provided written informed consent, and study was approved by the local Health and Research Ethics Committee.

Measures of cortical excitability were assessed by applying a 90-mm circular coil and assessing motor cortices ipsilateral and contralateral to the infarct separately, with recordings measured over the contralateral abductor pollicis brevis (APB) muscle (Fig. 1). In nonstroke subjects, the left motor cortex was studied. The side of stimulation was dependent on the side of the coil (A or B) facing upwards. The induced current flow was from the posterior to anterior direction in both hemispheres, such that when stimulating the left motor cortex, side A of the coil was facing up, ensuring a posterior-to-anterior current flow within the left motor cortex, while side B was facing up when stimulating the right motor cortex. The coil was adjusted tangentially over the patient’s scalp until the optimal position for a motor-evoked potential (MEP) was obtained from the APB. To determine the optimal cortical location of each subject’s APB, TMS was first delivered initially over the vertex of the scalp and then moved in the anterior-to-posterior and medial-to-lateral directions until a position that produced the largest MEP amplitude at the lowest TMS output was located. This “hot spot” on the scalp was marked with an “X” using ink.

Currents were generated by two magnetic stimulators connected via a BiStim200² (Magstim Co., Whitland, U.K.), such that both conditioning and test stimuli were independently set and delivered through the one coil. Recordings of MEPs were amplified and filtered (3Hz–3 kHz) using a Grass ICP511 AC amplifier (Grass-Telefactor; Astro-Med, Inc., West Warwick, RI) and sampled at 10 kHz using a 12-bit data acquisition card (National Instruments PCI-MIO-16E-4). Data acquisition and stimulus delivery were controlled by QTRACS software (Hugh Bostock, Institute of Neurology, London, U.K.).

Paired-pulse threshold-tracking TMS techniques were used to assess intracortical neuronal excitability according to a previously reported and validated method using threshold tracking (17–20). The technique was developed to overcome the variability in MEP amplitudes with consecutive stimuli that resulted in limitations using the constant-stimulus method (21). This method was first developed by Fisher et al. (17) in 2002, later validated by Vucic et al. (18) in 2006, and is currently used and published widely in numerous pathological studies (22–27). Importantly, as suggested by Fisher et al. (17), measurement of intracortical excitability at a constant MEP response using threshold-tracking methods limits the contribution of spinal and peripheral elements to the output measurement.

The threshold-tracking strategy used a target response of 0.2 mV (± 20%) located in the middle of the established linear relationship between the logarithm of the MEP amplitude and the stimulus intensity (18). By selecting a target that is located in the steepest portion of the stimulus response (SR) curve, relatively large variations in the MEP amplitude translate to relatively small variations in stimulus intensity or threshold.

Resting motor threshold (RMT) was determined in the initial part of the protocol and was defined as the stimulus intensity required to consistently (average 10 trials) produce and maintain the target MEP response of 0.2 mV peak-to-peak, with the patient seated comfortably and the testing limb relaxed. The SR curve for cortical stimulation was determined by increasing the intensity of the magnetic stimulus to the following levels: 60, 80, 90, 100, 110, 120, 130, 140, and 150% RMT. Three stimuli were delivered at each intensity level, and the maximum MEP amplitudes were determined.

The cortical silent period (CSP) is mediated by both spinal mechanisms in its early part, and cortical inhibitory neurons acting via γ-aminobutyric acid (GABA)-B receptors in the latter part. Since the duration is determined by the latter part, the CSP is a measure of cortical inhibition (28). CSP was evoked by single-pulsed TMS with intensities that varied according to the SR curve and recordings made with participants performing a weak voluntary contraction (10–30% maximum voluntary contraction). This was achieved through the use of a force transducer to measure APB contraction, with output processed and displayed using Spike2 data acquisition software (Cambridge Electronic Design Ltd., Cambridge, U.K.) that plots the maximum force generated by a given patient and sets a “window” between 10% and 30% of this maximum, whereby patients are encouraged to keep their subsequent generated force within this window during the CSP protocol. Maximum CSP duration was measured, according to convention (29), from the beginning of MEP to the return of electromyogram activity at 150% RMT stimulus intensity. The selection for this weak tonic voluntary contraction was based on initial studies of CSP, with subsequent studies demonstrating no effect of differing levels of contraction on the CSP duration (29–32).

Short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF) were measured according to a previously described and published protocol (18). Subthreshold conditioning stimuli at 70% RMT were delivered sequentially at interstimulus intervals of 1, 1.5, 2, 2.5, 3, 3.5, 4, 5, 7, 10, 15, 20, and 30 ms.

SICI was measured as the increase in test stimulus intensity required to evoke the target MEP and calculated as follows (17,18):

Facilitation was measured as the decrease in the conditioned test stimulus intensity required to evoke the target MEP. Averaged SICI was determined over the interstimulus intervals of 1 to 7 ms, and averaged ICF was determined over the intervals of 10 to 30 ms, as described previously (18).

Statistical analysis was performed using SPSS 22 software (IBM Corp, Armonk, NY). To address the cortical functional changes that may be present in patients with diabetes, electrophysiological data were compared between the cohort with diabetes but without stroke and the normal control cohort, with the Student t test or Mann-Whitney U test used depending on normality of data distribution. To address the changes in cortical function that stroke may induce in those with and without diabetes, all electrophysiological data among the four groups were compared using a one-way ANOVA, with the between-group factor being diabetes, stroke patients with diabetes, stroke patients without diabetes, and normal control subjects. Post hoc multiple pairwise comparisons were used for subgroup analyses and corrected for by the Tukey honest significant difference test with the family-wise significance level set at 0.05. Correlations were assessed by Spearman rank test. A P value of <0.05 was regarded as statistically significant. Results are expressed as mean ± SEM.

Nineteen patients diagnosed with acute unilateral ischemic stroke were recruited and studied within the acute period (mean 5.8 days, range 1–18) (Table 1). Of these stroke patients, 7 (3 men) had diabetes (aged 63–86, mean 70.7 years; 6 type 2 and 1 type 1) and 12 (7 men) did not have diabetes (aged 29–76, mean 63.3 years). The patients in groups with and without diabetes were reasonably well matched for age (P = 0.22), duration from stroke onset to assessment (with diabetes: 4.7 ± 0.92 days, without diabetes: 6.6 ± 1.33 days; P = 0.35) and size of stroke (with diabetes: 29.01 ± 6.4 mm, without diabetes: 24.1 ± 5.1 mm; P = 0.56). The stroke locations were three cortical and four subcortical in the group with diabetes and six cortical and six subcortical in the group without diabetes. Subcortical infarcts were defined as lesions within the basal ganglia, internal capsule, or corona radiata and sparing the motor cortex, whereas cortical strokes were defined as wedge-shaped, superficial lesions in the territory of the large major cerebral arteries or lesions in a border zone that involved the motor cortex and potentially underlying white matter. Two patients in the stroke group without diabetes and two in the stroke group with diabetes received acute reperfusion therapy.

The functional severity of strokes were similar between the groups with diabetes (Barthel index: 83.6 ± 6.9, Fugl-Meyer: 58.4 ± 0.61, modified Rankin Score: 2.4 ± 0.7) and without diabetes (Barthel index: 73.8 ± 8.3, P = 0.38; Fugl-Meyer: 51.5 ± 4.8, P = 0.18; modified Rankin Score: 2.5 ± 0.4, P = 0.93). The fasting blood glucose levels (BGL) were higher in the stroke patients with diabetes (mean 9.09 ± 1.2 mmol/L) compared with those stroke patients without diabetes (mean 6.38 ± 0.8 mmol/L, P = 0.036). In addition, 8 patients with diabetes (4 males; aged 43–76, mean 65.6 years; 1 type 1 and 7 type 2) and 30 control subjects (aged 37–73, mean 55.8 years) without a history of stroke were recruited for the study. HbA_1c levels were reasonably matched between the stroke (7.8 ± 0.47%) and control groups (7.3 ± 0.22%, P = 0.46) with diabetes, as was the duration of diabetes (18.1 ± 3.3 years and 19.5 ± 3.6 years, respectively; P = 0.79). Six patients from the stroke group with diabetes were taking metformin (one was also on gliclazide), and one was on insulin, whereas five control patients with diabetes were taking metformin (one also on gliclazide and one also on sitagliptin), and three were on insulin.

There was a significant difference present across subgroups when compared over the ipsilesional hemisphere of stroke patients (ANOVA, F = 19.4, P < 0.001). SICI was significantly reduced over the ipsilesional cortex in the stroke patients without diabetes (2.7 ± 1.2%) compared with normal control patients (14.0 ± 0.8%, P < 0.001), but there was no significant change in SICI in stroke patients with diabetes (6.6 ± 2.3%) compared with control subjects with diabetes (8.9 ± 1.4%, P = 0.72) (Figs. 2 and 3). In other words, after a stroke, a change in cortical excitability (reduction in intracortical inhibition) is observed over the stroke hemisphere in patients without diabetes, but no such change was demonstrated in patients with diabetes after a stroke. Furthermore, there were no significant differences in SICI over the affected hemisphere in stroke patients with diabetes when comparing between cortical and subcortical stroke locations (P = 0.40).

When assessing changes over the contralesional cortex, there were also significant differences in SICI across subgroups (ANOVA, F = 11.8, P < 0.001). In particular, SICI was reduced over the contralesional motor cortex in stroke patients without diabetes (2.8 ± 2.7%) compared with normal control subjects (14.0 ± 0.8%, P = 0.001). Of interest, there were also differences in SICI over the contralesional cortices of stroke patients with diabetes (−4.0 ± 6.1%) compared with control patients with diabetes (8.9 ± 1.4%, P = 0.013) (Figs. 2 and 3). In other words, after an acute stroke, a significant change in cortical excitability (reduction in intracortical inhibition) was observed over the unaffected hemisphere in both patients with and without diabetes when compared with their respective control groups. Furthermore, there were no significant differences in SICI over the unaffected hemisphere in stroke patients with diabetes when comparing between cortical and subcortical stroke locations (P = 0.20).

Moreover, averaged SICI was significantly reduced in the control patients with diabetes (8.9 ± 1.4%) compared with normal control subjects (14.0 ± 0.8%, P = 0.004) (Fig. 4). That is, an abnormality in cortical excitability (reduction in intracortical inhibition) is demonstrated in the control group with diabetes compared with the control group without diabetes that may suggest altered cortical function in patients with diabetes that may potentially interfere with adaptive cortical reorganizational processes after an acute neurological insult such as stroke.

After SICI, a period of intracortical facilitation develops. In contrast to SICI findings, there were no significant differences in ICF between the subgroups when comparing the measures over the ipsilesional (ANOVA, F = 1.3, P = 0.28) and contralesional motor cortices (ANOVA, F = 1.8, P = 0.17).

When considered as an entire stroke cohort, no significant correlations were found between BGL and SICI over the lesioned (r = −0.14, P = 0.6) and contralesional motor cortices (r = 0.08, P = 0.75). Similarly, no significant correlations were found when analyzed according to stroke groups with diabetes (lesional: r = −0.26, P = 0.63; contralesional: r = 0.66, P = 0.16) and without diabetes (lesional: r = −0.26, P = 0.45; contralesional: r = 0.19, P = 0.55).

The RMT was comparable across all four subgroups over both ipsilesional (ANOVA, F = 0.76, P = 0.97) and contralesional (ANOVA, F = 0.414, P = 0.74) motor cortices. In addition, the maximum MEP amplitude was also comparable across the four subgroups when measured over the ipsilesional (ANOVA, F = 0.37, P = 0.78) and contralesional (ANOVA, F = 0.11, P = 0.95) motor cortices. Of further relevance, maximum CSP duration was also similar across the groups (ipsilesional: F = 0.37, P = 0.77; contralesional: F = 0.65, P = 0.59).

The current study has provided insight into intracortical excitability and, thereby, functional changes that occur after an ischemic stroke in patients with diabetes and how they differ compared with their counterparts without diabetes. In addition, the current study has also established changes in cortical function in nonstroke patients with diabetes potentially contributing to the changes seen after acute stroke, with implications for poststroke recovery.

The observed reduction in intracortical inhibition in control (nonstroke) patients without diabetes reflects reductions in GABAergic pathways and may be explained by several mechanisms.

Firstly, a loss of cortical GABAergic inhibitory neurons may in part underlie the findings in the current study. Structural changes have been demonstrated in human and animal subjects with diabetes that show reductions in gray matter volume (11,33) with longer disease duration and that increased fasting BGLs appear to be inversely correlated with total gray matter volume (33), and cortical neuronal and axonal degeneration has been reported in animal models (9).

Of further relevance, hyperglycemia may result in neuronal damage through accentuated tissue acidosis and lactate generation (12). Altered glucose metabolism in diabetes has been suggested to result in alterations in neuronal glucose utilization and subsequent neuronal dysfunction. Impaired insulin within the central nervous system in individuals with diabetes may also lead to changes in the metabolic pathways necessary for synaptic maintenance (10). Impairment in brain neurotransmitter systems that include GABA (34) have been demonstrated in diabetic rats and can be a direct result from glucose dysregulation on GABA systems as well as through the effect on other neurotransmitters. Specifically, cholinergic and GAD activity, both of which modulate cortical GABA neurotransmission, are decreased in diabetic animal models that ultimately also result in reduction in GABA production (34).

Taken together, these mechanisms may underlie the observed reduction in SICI observed in the cohort of control subjects with diabetes compared with normal control subjects and, consequently, may affect the brain’s intrinsic ability to reorganize after a stroke.

The reductions in GABA-mediated intracortical inhibition observed in those with diabetes may influence the degree of damage after an acute ischemic stroke as well as negatively affect the cortical reorganization required for subsequent functional recovery.

In the acute phase immediately after a stroke, extracellular levels of glutamate contribute to excitotoxic damage. Consequently, the reduced levels of GABA and resultant loss of inhibitory tone in patients with diabetes may lead to increased neuronal damage (35). After the immediate phase of stroke, however, the role of GABA-mediated inhibition changes during the structural and functional reorganization process mediating recovery.

Assessing these inhibitory and facilitatory circuits (and thereby intracortical excitability) provides an important understanding of how neuroplastic changes may be mediated and how they underlie associated behavioral change and functional improvement after an insult such as a stroke (16). Functional recovery after the acute event will depend on the reorganization of neural networks in the brain that occur over both the ipsilesional and contralesional motor cortices. Previous studies have consistently demonstrated that immediately after an ischemic stroke, GABA-mediated intracortical inhibition in both hemispheres is reduced and is associated with functional improvement over time (19,20,36,37) and that reducing GABA-mediated cortical inhibition promotes functional outcome after stroke (38). Further to this, other studies have also demonstrated that interfering with this intracortical excitability or reduction in SICI by pharmacological or neuromodulatory interventions resulted in deterioration in stroke functional recovery (39,40). Cortical hyperexcitability resulting from the reduced GABA-mediated inhibition is associated with greater long-term potentiation (41) and has been demonstrated to positively influence stroke recovery.

The results of the current study have demonstrated that after an acute stroke in individuals with diabetes, such cortical plasticity changes were observed only over the contralesional hemisphere and not the stroke side. This may facilitate pyramidal tract axonal sprouting originating from the contralesional motor cortex traversing the midline in order to reach neurons denervated by the stroke in the ipsilesional cortex that is less able to participate in the process of reorganization (42). In this regard, diabetes limits the brain’s capacity for repair and rewiring that is critical for stroke recovery, and there is reduced synaptic plasticity and dendritic density in the cerebral cortex (3). In particular, axonal density is decreased in the ipsilesional motor cortex of diabetic stroke rat models compared with stroke rats without diabetes, consequently resulting in impaired neuroplasticity after the ischemic lesion (43). Of further relevance, diabetes in rats prevented the reemergence of forelimb sensory representation onto peri-infarct regions in the stroke hemisphere and limited stroke-induced functional changes (3).

Studies in nondiabetic animals have shown that stroke upregulates the production of growth-associated proteins, dendritic spines, axonal sprouting, microglial activity, and angiogenesis, all of which may need an unmasking effect by cortical disinhibition to occur (44,45), which is not observed over the stroke hemisphere of our stroke patients with diabetes. A potential explanation for why the ipsilesional cortex of subjects with diabetes is unable to undergo such functional change after stroke may be the “double-hit” phenomenon: an acute insult on a system that already had preexisting impairments. Remodeling of ischemic brain tissue involves interactions between neurons, glial, and microvascular cells that create a microenvironment in which neurological recovery may ensue. Specifically, it is related to the interaction of neuroblasts with the microvasculature in the vicinity of the ischemic lesion that creates an environment to nurture and foster brain remodeling (42). It is therefore possible that preexisting microvascular damage from diabetes prevents the release of required neurotrophic factors critical in this coupling process needed for neuronal reorganization in the peri-infarct cortex. Compounding this, patients with diabetes have network disorganization with inefficient connections between brain regions (7) and may not cope after further insults such as a stroke. This then poses limitations on the brain’s ability to reorganize and facilitate functional recovery, with much of the poststroke neuroplasticity dependent on the contralesional hemisphere. However, without a study examining the evolution of these cortical changes in the same population with diabetes pre- and poststroke, this assumption remains speculative.

A large body of literature has shown that stroke outcomes depend on changes in bihemispheric plasticity (46). Such changes can be adaptive or maladaptive that involve alterations in inter- and intrahemispheric connections. The preexisting cortical excitability changes in the nonstroke patients with diabetes may be considered as maladaptive potentially limit the scope for optimal recovery by impairing the ability of the stroked hemisphere in patients with diabetes to make the necessary alterations in cortical reorganization. Previous studies have shown a significant relationship between lesioned and unlesioned hemispheres and how this interaction between hemispheres can be representative of poststroke plasticity and its association with cortical engagement for motor recovery (46).

The current study has several limitations. Firstly the cohorts with diabetes were relatively small in numbers, and future studies using larger numbers will be needed to confirm the current findings. Being a cross-sectional study, only baseline electrophysiological changes and clinical scores were assessed, without longitudinal data on changes in cortical function and clinical outcome measures. As such, longitudinal studies with clinical correlations will also needed to assess whether the baseline changes in the stroke cohort with diabetes represent an adaptive or maladaptive electrophysiological response after an ischemic stroke. Moreover, longitudinal studies will determine whether changes in cortical function between stroke groups with diabetes with different lesion locations (cortical vs. subcortical) evolve over the course of stroke recovery as they relate to outcome measures and how these changes may be related to the baseline clinical and lesion characteristics and their stroke outcome (20). Of further interest, future studies may also specifically examine the effect of acute reperfusion therapies, such as thrombolysis, on cortical electrophysiological function. Only two patients in the current study had undergone such therapy in either group, and unfortunately, such small numbers would not permit a meaningful statistical analysis. In addition, selection of patients with stroke into either the groups with or without diabetes was based on the patients’ background history, and hence, it may be possible that there were patients with previously undiagnosed diabetes.

The current study has demonstrated that after an acute stroke, much of the electrophysiological changes in patients with diabetes occur over the contralesional and not the ipsilesional motor cortex, unlike their counterparts without diabetes where reductions in intracortical inhibition occur equally over both cortices. This potentially provides new insights into the neurophysiological mechanism underlying poor functional outcome after stroke in patients with diabetes. The current study has also demonstrated alterations in cortical electrophysiology in patients with diabetes that may correlate with previously explored changes in cortical structure observed in neuroimaging and may contribute to the impaired cortical plastic changes after stroke. The results of the current study may have implications toward the development of neuroprotective strategies for patients with diabetes, particularly those suffering a stroke, and will guide the use of novel tools, such as noninvasive brain stimulation, at improving poststroke recovery.

Funding. W.H. was supported by the University of Sydney Post-Doctoral Fellowship. A.V.K. was supported by a Career Development Award of the National Health and Medical Research Council of Australia (grant number 1065663). M.C.K. was supported by funding to Forefront, a collaborative research group supported by the National Health and Medical Research Council of Australia Program Grant (#1037746).

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

Author Contributions. W.H. researched data and wrote the manuscript. N.K. researched data. R.A., C.S.-Y.L., and S.V. reviewed and edited the manuscript. A.V.K. and M.C.K. contributed to discussion and reviewed and edited the manuscript. W.H. 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.