Brain and Body: A Review of Central Nervous System Contributions to Movement Impairments in Diabetes

Diabetes is associated with microvascular complications in the nervous system. The link between diabetes and degenerative disease of the brain is well established, and an intensive research effort has linked cognitive decline in individuals with diabetes to regional degeneration in the brain (for review see Biessels and Reijmer, 2014 [1]). Multiple studies have employed hypothesis-free, data-driven analyses (i.e., voxel-based morphometry or tract-based spatial statistics) to explore regional impacts of diabetes on the brain, and in these studies movement-related centers in the brain frequently emerge as impacted by diabetes status (2,3). However, the impact of degeneration in movement centers of the brain on motor behavior remains largely overlooked, despite the high prevalence of motor impairments in individuals with diabetes (4).

While the impact of diabetes on central nervous system (CNS) sensorimotor regions is understudied, disease complications in peripheral sensorimotor neurons are well recognized. Diabetic peripheral neuropathy (DPN) is caused by degeneration of peripheral somatic nerves and affects 30–50% of individuals with diabetes (5). Movement impairments in individuals with diabetes have historically been attributed to DPN; however, motor impairments also occur in individuals with diabetes who do not have DPN, including poor balance (6), altered gait (7), and compromised grip control (8). These findings suggest that pathology beyond the peripheral somatic nervous system contributes to diabetes-related sensorimotor impairments, and disease complications in CNS sensorimotor regions are consequential for physical function.

Control of motor behavior relies on reciprocal interactions between the peripheral nervous system (PNS) and CNS. Since there is evidence of both peripheral and central neurodegeneration in individuals with diabetes, both systems must be considered in the study of sensorimotor impairments. Sensorimotor research in individuals with diabetes largely focuses on the PNS, whereas research on CNS function in diabetes has mostly considered cognitive brain regions. Here, we attempt to bridge this gap by contextualizing recent neuroimaging findings of the effects of diabetes on the brain within a framework of neural control of movement. For the purposes of this review, we consider data from both populations with type 1 diabetes and populations with type 2 diabetes. We will use the term “diabetes” to refer to both forms of diabetes and will differentiate between diabetes types where appropriate.

Diabetes-related microvascular complications affect multiple tissue classes, including the retinas, kidneys, peripheral somatic nerves, and the brain. Microvascular disease in diabetes has similar mechanisms regardless of the tissue site; chronic hyperglycemia and loss of insulin signaling cause a cascade of inflammatory pathway activation, oxidative stress, and endothelial dysfunction (9). Inflammatory endothelial dysfunction and subsequent loss of blood-brain barrier integrity cause the development of cerebral microvascular lesions, resulting in an increased load of white matter hyperintensities, lacunar infarcts, and microbleeds (for review see Wardlaw et al., 2013 [10]). Individuals with diabetes also show indicators of gross neurodegenerative pathology, such as accelerated cortical atrophy (11).

The etiology of cerebral microvascular damage is complex, and diabetes is one of many cardiometabolic risk factors that have been associated with an increased load of cerebral microvascular complications; other risk factors include hypertension, smoking, and hyperlipidemia (12). The presence of multiple risk factors will increase the incidence of cerebral microvascular complications (10), a consideration that deserves careful examination in this population. However, diabetes has independent negative impacts on CNS tissues. Specific to diabetes is the frequent cooccurrence of microvascular pathology in nerves of the peripheral somatic nervous system (5). Additionally, animal research has revealed a role of insulin signaling in neuroplasticity and consequently a loss of neuroplasticity in insulin-resistant animals (13). These processes, in addition to gross neurodegeneration, have implications for motor function in individuals with diabetes.

Recent advances in neuroimaging allow for highly region-specific investigations of cerebral structure and function, and these have provided emerging evidence that diabetes impacts areas of the brain involved in perceiving ascending somatosensory information and generating descending motor output. The study of regional brain networks may yield insights into the behavioral declines seen in this population. To guide this review of CNS contributions to movement impairment in diabetes, we focused our discussion on three critical regions for voluntary sensorimotor control and their associated white matter projections: 1) the cerebral cortex, specifically, the motor and somatosensory cortices; 2) the cerebellum; and 3) the basal ganglia (Fig. 1).

The primary motor cortex (M1) and associated secondary motor cortices play a critical role in the initiation and planning of voluntary movements. Descending outputs from the cortex travel via the corticospinal tract (CST) to synapse on efferent neurons of the spinal cord and initiate muscle contractions. The primary somatosensory cortex (S1) contributes to the conscious awareness of somatosensory information. Ascending somatosensory inputs from spinal afferents synapse in the thalamus onto thalamocortical neurons projecting to S1. The sensorimotor cortex operates as a functional unit due to substantial integration of processing between cortical regions, and between the cortex and thalamus, reflecting the importance of sensorimotor integration for motor function (14).

Diabetes is associated with atrophy and altered activity of the somatosensory and motor cortices and their associated white matter projections. At the level of the cortex, gray matter volumes decrease in primary motor cortex (11,15–17), the secondary motor cortices (17), and primary somatosensory cortex (17). Cortical volume loss in sensorimotor cortices is independently associated with diabetes after correction for comorbid cardiometabolic risk factors (17), and sensorimotor cortex atrophy is slowed in individuals undergoing intensive glycemic control (11), suggestive of an independent effect of diabetes status on cortical atrophy. Cortical activity is also impacted by diabetes status. Resting state functional MRI (fMRI) studies of local spontaneous activity consistently report lower activity in S1 (18–21), and S1 activity relates negatively to fasting glucose levels (20). Primary (22) and secondary (23) motor cortex activity is decreased in individuals with diabetes, including reduced excitability in M1 regions specific to control of the upper extremity (24). The time course of changes in volume and activity of the cortical gray matter is unclear; cortical activity may decrease as a direct consequence of cortical atrophy, or conversely, changes in the metabolic activity of the cortex may be an early indicator of regions vulnerable to neuronal death.

Several lines of evidence indicate that diabetes is related to degeneration of white matter projections between the sensorimotor cortices and subcortical structures. At the level of the spinal cord, individuals with diabetes show gross atrophy in the cervical spine that is most severe in individuals with DPN but is also present in individuals with diabetes without DPN (25,26). This suggests that loss of spinal white matter results from a dual contribution of degeneration of peripheral afferents in DPN and degeneration of ascending and descending CNS spinal pathways. For descending corticospinal projections from M1, diffusion tensor imaging studies report decreases in microstructural integrity in the descending white matter of the CST (27–31), which relates to cortical atrophy in M1 (2). The conduction velocity of upper motor neurons of the CST is delayed in humans (32,33) and in rodent models of diabetes (34). Loss of microstructural integrity and delays in conduction velocity are indicative of neuronal loss or demyelination in this critical motor pathway. White matter tracts between the cortex and thalamus are also impacted by diabetes status. Functional connectivity is decreased between the thalamus and M1 (35), suggesting reduced communication between the thalamus and motor cortex. In terms of somatosensory inputs to the thalamus, there is a decrease in the conduction velocity of ascending afferent signals specific to the thalamus-S1 relay (36,37), indicating that thalamocortical projections are impacted by diabetes independently from delays in peripheral afferent conduction velocity caused by peripheral neuropathy.

Evidence from multiple modalities indicates that diabetes impacts both the structure and function of sensorimotor cortical gray matter and projection fibers to associated subcortical and spinal structures. Diabetes-related degeneration to sensorimotor cortex may impact the behaviors supported by these regions during sensorimotor function. Investigations of cortical function in relation to motor function in individuals with diabetes have been performed with pegboard assessments of manual dexterity. Manual dexterity relates to M1 thickness (2), and white matter microstructure of the CST (27), with markers of decreased structural integrity relating to poorer manual dexterity in individuals with diabetes. In addition, neuroplasticity within the sensorimotor cortices is known to be an important mechanism in the acquisition and consolidation of skilled movements (38). Although there is evidence that diabetes decreases capacity for long-term potentiation-like plasticity in the human motor cortex (39), the relationship of neuroplasticity with motor function has not been examined in individuals with diabetes. In summary, while there exists some evidence that diabetes-related motor impairments relate to cortical neurodegeneration, many behavioral metrics that are known to rely on sensorimotor cortical function remain unexplored in individuals with diabetes.

The cerebellum is involved in motor coordination and unconscious proprioception and is organized into anatomically and functionally distinct regions. The cerebellum receives sensory inputs from the spinal cord and projects output onto descending motor pathways. Furthermore, the cerebellum has extensive bidirectional connectivity with the cerebral cortex via thalamic relays. The cerebellum is responsible for maintaining an internal representation of the body, predicting the sensory consequences of movement, and updating motor plans generated by the cortex in response to movement errors (40). Corticocerebellar networks play an important role in both motor coordination and motor learning (41).

Glucose metabolism is more efficient in the cerebellum compared with the cerebrum, meaning the cerebellum is relatively protected from hypoglycemic damage (42); however, the cerebellum is vulnerable to hyperglycemia-related toxicity over the course of diabetes progression (43). Total cerebellar volume is reduced in individuals with diabetes (44,45), and there is a negative linear relationship between cerebellar volume and fasting plasma glucose (44). Decreased cerebellar volume is accompanied by changes to white matter microstructure within the anterior and posterior cerebellar lobes (46,47) and the vermis (3,47). White matter microstructure in the cerebellar lobes decreases with greater disease duration (46,47), indicating a negative cumulative impact of hyperglycemia exposure to cerebellar structure. Tracts between the cerebellum and the cortex are also broadly affected by diabetes status; notably, there is decreased white matter integrity in tracts traveling from the cerebellum to the thalamus and M1 (47).

The impact of diabetes on cerebellar activity is unclear. Resting state studies of local spontaneous brain activity report both increased (18,23,48) and decreased (19,23,49) spontaneous activity within anterior and posterior lobes of the cerebellum. However, these studies all employed data-driven whole-brain approaches, which involve spatial smoothing of brain regions across subjects for alignment to an atlas space. While these approaches allow for exploratory analyses of brain activity, inconsistencies in previous activations studies may be a result of the high interindividual variability in corticocerebellar anatomy, which makes alignment to a common template space problematic (50). Connectivity-based resting state analyses comparing networks of brain activity show reduced cerebellar connectivity in multiple cerebral-cerebellar brain networks (51,52). Cerebellar connectivity negatively relates both to diabetes disease duration and HbA_1c levels (51). To resolve incongruencies in previous findings, future fMRI studies should perform regional analyses of the cerebellum that are robust to individual differences in cerebellar anatomy.

Cerebellar damage is associated with abnormal control of movement, and movement abnormalities vary depending on which specialized region of the cerebellum is impacted. For example, diabetes affects the vermis and intermediate hemispheres of the cerebellar lobes (3,47). These cerebellar regions receive ascending inputs from spinal cord and brainstem centers and are principally involved in the control of proximal muscles and coordination of movement during gait, and individuals with diabetes show gait abnormalities that relate to decreased blood flow in the vermis and intermediate lobe of the cerebellum (53). However, the majority of diabetes cerebellar research has detected alterations in the lateral hemispheres of the cerebellum. Specifically, posterior regions of the lateral cerebellar hemispheres have the most evidence for diabetes-related disruptions (18,46,47). These regions of the cerebellum have connectivity with the cerebral cortex and are involved in voluntary control of distal muscles, as in the coordination of reaching movements. The implications of changes to lateral cerebellar lobes for motor function have been neglected, despite the rich literature demonstrating the importance of these regions in error-based motor learning, anticipatory control of movement, and spatial and temporal patterning of motor coordination (54).

The basal ganglia contribute to the initiation and execution of voluntary movement, as well as the affective components of movement. The basal ganglia are a group of subcortical nuclei comprising the caudate, putamen, globus pallidus, and subthalamic nucleus. These nuclei receive projections from the cortex, thalamus, and brainstem, and their major output returns to the cortex via the thalamus. The best studied basal ganglia–cortical loop is a motor circuit formed with the primary and association motor cortices. This loop is important for the selection and initiation of motor actions, guided by environmental reinforcement (55). The basal ganglia are involved in learning and performance of discrete sequences of movements—in contrast to the cerebellum, which is involved in smoothing and coordinating continuous movements (56).

There are very few investigations of the basal ganglia in individuals with diabetes. Gray matter volumes are reduced in the caudate (57,58) and the putamen (58–60), and there is lower cerebral blood perfusion in the caudate (53,61). Additionally, there is some evidence for decreased connectivity between basal ganglia and cortical networks (52). Consistent with the basal ganglia’s role in response selection, basal ganglia atrophy (60) and decreased basal ganglia blood flow (61) relate to reduced psychomotor speed in individuals with diabetes. Given the known functions of the basal ganglia, alterations in these structures in individuals with diabetes may contribute to diabetes-related delays in reaction time (62) or slowed gait speed (63).

Diabetes-related disease complications exist in both peripheral and central sensorimotor nervous tissues. An open question is the degree to which loss of peripheral signaling caused by DPN impacts CNS somatosensory and motor function. It is possible that the loss of afferent information from the periphery directly causes remodeling of central sensory circuits, as observed in individuals with loss of afferent input after limb amputation (64). Reduced primary somatosensory cortex activity in individuals with diabetes (18–21) may occur as a direct result of loss of peripheral afferent signal or, conversely, may reflect an independent process of cortical atrophy occurring due to central complications of diabetes. Selvarajah et al. (65) (2014) reported decreased S1 volumes in individuals with DPN. However, the comparison group consisted of a mix of individuals with diabetes and no DPN and healthy control subjects without diabetes; therefore, this study did not consider the impact that diabetes alone may have on S1 volumes. Conversely, resting state fMRI studies report no differences in S1 activity between patients with DPN and those without (18,19). These data are suggestive of an independent process of central neurodegeneration caused by diabetes, but future research must better control for confounding effects of loss of peripheral signaling from DPN.

An important caveat to this assertion is the presence of chronic neuropathic pain. There is evidence that, in contrast to insensate forms of neuropathy, painful subtypes of peripheral neuropathy have a direct impact on CNS function (for a detailed review see Fischer and Waxman, 2010 [66]). Pain in diabetic neuropathy is partially neuropathic in origin and relates to altered somatosensory gating and hyperexcitability of the thalamus (67). Research into the effects of diabetic neuropathy on the CNS should therefore consider painful and nonpainful neuropathy subtypes separately, as chronic neuropathic pain may produce a central sensitization that results in a different neurological phenotype.

In summary, neuroimaging evidence from individuals with DPN indicates that diabetes damages central sensorimotor regions in a process concurrent with but separate from peripheral microvascular complications (Table 1). More research is required to establish the typical progression of microvascular complications in the PNS and CNS. If diabetes disease complications in the CNS precede PNS complications, sensorimotor impairments may be present even in individuals who do not show diagnostic indicators of peripheral neuropathy. On the other hand, if peripheral neuropathy typically precedes CNS degeneration there may be a progressive decline in the ability of the CNS to compensate for loss of sensorimotor control in the periphery, creating an additive burden on sensorimotor impairments.

Modern theories of motor control and motor learning emphasize reciprocal relationships between peripheral (feedback) and central (feedforward) control of movement (for review see Scott et al., 2015 [68]). Damage to either the PNS or CNS will lead to characteristic impairments in movement abilities, and these theoretical frameworks can inform our understanding of movement impairments in individuals with diabetes.

Feedback from peripheral afferents provides information about the current state of the body and the success of ongoing goal-oriented movements. Peripheral neuropathy causes loss of afferent inputs and thus a loss of feedback motor control, which manifests in multiple behavioral metrics. A simple example of loss of feedback motor control is an increase in body sway during quiet stance, which is a result of decreased tactile and proprioceptive inputs from the feet and ankles (69). More complex feedback motor control occurs in conditions of unexpected environmental changes, which necessitate rapid correction of ongoing movements. Peripheral neuropathy causes decreased muscle responses to unexpected lower-extremity perturbations (70), indicating an impaired ability to adapt motor patterns in response to somatosensory feedback. A loss of afferent feedback results in decreased movement stability, which causes compensatory increases in feedforward motor control strategies. In individuals with peripheral neuropathy, this may be seen as an increase in postural anticipation of surface changes during gait (71) or a higher grip force being applied when manually manipulating objects (72). Impairments in feedback motor control are clinically significant, as individuals with peripheral neuropathy are at highest risk of falls after unexpected gait perturbances (73).

Feedforward control from the CNS provides descending commands for voluntary movement. Feedforward motor control is the initiation and anticipatory scaling of movements that occurs before sensory feedback on the movement is received by peripheral receptors (74). Movement impairments in individuals with diabetes have primarily been interpreted as a result of loss of feedback signaling from peripheral neuropathy, neglecting potential contributions of feedforward mechanisms. For instance, gait is under relatively greater feedforward control than quiet stance (75), and thus gait abnormalities observed in individuals with diabetes who do not have peripheral neuropathy (7) may result from a loss of feedforward control rather than feedback errors. This also may contribute to dual-task gait impairments in individuals with diabetes (76). Moreover, considering feedforward contributions to motor control may help to resolve contradictory findings in previous research. For example, the counterintuitive finding of reduced grip force applied during object manipulation in individuals with diabetes (77) might be explained by a loss of feedforward grip control. Finally, there are possibly unrecognized motor deficits in individuals with diabetes in view of evidence for degeneration to brain networks involved in feedforward motor control; notably, the impact of diabetes on motor adaptation and motor learning is currently unknown.

Although the prevalence and progression of central complications relative to peripheral complications are not well characterized, individuals with diabetes could present with central microvascular disease, peripheral neuropathy, or a combination of both. An interesting question is how both the loss of feedforward control and the loss of feedback control would interact to influence sensorimotor function in individuals with diabetes. Very few studies have considered both peripheral and central diabetes complications in the study of motor function. Manor et al. (45) (2012) reported decreased cerebellar volumes in individuals with diabetes related to slowed gait speed and decreased stability during gait; however, this relationship was stronger in individuals with peripheral neuropathy. Nunley et al. (60) (2017) reported that putamen volumes and peripheral neuropathy related to psychomotor slowing in individuals with type 1 diabetes, but putamen volumes did not relate to psychomotor speed in control subjects without diabetes. These data suggest an increased reliance on central feedforward control in individuals with loss of afferent feedback from DPN. In both of these studies, individuals with diabetes and DPN had poorer motor function than individuals without DPN (45,60); thus, adequate feedforward compensation for loss of somatosensory inputs may not be possible due to CNS degeneration. There is likely an additive burden of central and peripheral sensorimotor changes on motor behavior; our understanding of the neurological sources of motor impairments is incomplete without interrogation of feedforward motor deficits in individuals with diabetes (Fig. 2).

Diabetes is a major risk factor for cognitive decline and dementia (1). This topic has received considerable research attention; indeed, the primary aim of many of the studies presented in this review was to identify relationships between brain metrics and cognitive decline in individuals with diabetes. Cognition and mobility are inextricably linked, and therefore impairments in attention or executive functions could impact motor performance. For example, in dual-task paradigms, older adults show decreased motor performance with increasing attentional load (78). Conversely, motor performance also impacts cognitive function, as in reports showing that slowed gait speed is an early predictor of cognitive impairment in older adults (79). The purpose of this review is not to suggest that cognitive function is separate from, or less important than, sensorimotor function, particularly in individuals with diabetes who are expected to show both cognitive and sensorimotor symptoms. Instead, our goal is to draw attention to existing neuroimaging evidence for CNS contributions to sensorimotor disability. Future work on diabetes disease complications must consider the complex interactions between cognitive and sensorimotor impairment.

The current review highlights evidence that diabetes-related CNS degeneration may contribute to impairments in motor control, motor performance, and motor learning. We argue that the central contributions to motor deficits in individuals with diabetes are more profound than previously recognized. Existing data suggest that changes in central sensorimotor signaling in diabetes are not simply a passive response to loss of afferent signaling from the PNS but, rather, reflect an independent and additive process of regional neurodegeneration. There is a critical need for controlled behavioral experiments linking cerebral markers of sensorimotor degeneration with movement impairments in individuals with diabetes. We identify several areas in need of more research (Table 2) including 1) identifying novel motor control and motor learning deficits in individuals with diabetes, 2) evaluating the extent to which CNS complications relate to sensorimotor impairments in this population, and 3) delineating the interactions between progression of diabetic neuropathy and CNS sensorimotor complications.

The research outlined in this review has implications for the clinical management of diabetes complications. Most importantly, diabetes-related CNS complications may have a significant and unrecognized contribution to the high rates of physical disability and dependency in activities of daily living in this population (4). However, current clinical screening batteries are not designed to identify individuals with sensorimotor impairments originating in the CNS, and the prevalence of CNS complications is unknown. Exploring relationships between central sensorimotor circuits and impaired behavioral function could therefore lead to the identification of novel markers of sensorimotor decline in individuals with diabetes. In conclusion, the impact of diabetes on central sensorimotor function is a promising, but still underdeveloped, area of research. Future work delineating the nature and extent of sensorimotor deficits in this population is required for the effective management of physical disability in individuals with diabetes.

Funding. This work was funded by the Canadian Institutes of Health Research (GD-146283).

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