While there are earlier reports of spinal cord lesions associated with diabetes mellitus (refs. 5-7), the publication of a series of key (8-12) and other (13-16) neuropathologi-cal studies in the latter half of the 20th century established the histological nature of this injury. These reports helped promote the concept that myelopathy is a part of the diabetic process and remain the definitive neuropathological studies. Unlike myelopathy, the existence of peripheral neuropathy and radiculopathy is not disputed and the pathology has been well documented. The following section will first consider morphological evidence from autopsy material for diabetes-induced injury to the spinal cord, associated ganglia and roots, and then available evidence from studies using experimental models of diabetes.
Among the earliest observations of spinal cord lesions in diabetes mellitus are those of Williamson (5,6), who described macroscopic changes evident in the posterior columns of the cervical and thoracic regions after hardening in Muller's fluid. In sections from the affected regions of celloidin-embedded spinal cords of three patients, swollen axis cylinders surrounded by distended or thin myelin sheaths and scattered degenerated fibers were evident. These early observations suggest that both demyelina-tion and axonal degeneration occur in lesions of the posterior column white matter. Later, distension and/or thinning of myelin sheaths and axonal loss in the white matter of paraffin-embedded autopsy material from patients with diabetes mellitus were also described (8,11), whereas others have simply noted degeneration of the long tracts (9,10,13). The most comprehensive autopsy investigation of diabetic myelopathy is that reported by Slager (12), who expanded an earlier series of 37 patients (16) to 75 unselected diabetic patients and reported posterior column demyelination with relative sparing of axons in 27% of spinal cords. A subsequent report quantitatively documented a decrease in myelinated fiber density in the fasciculus gracilis of the cervical posterior columns (17). Although the posterior columns are the most commonly reported site of injury in diabetes mellitus, lesions have also been documented in the lateral (spinocerebellar tracts) and ventral columns (8,9), with the lower segments of all regions affected more than the upper segments.
Diabetes-associated injury to the gray matter of the spinal cord is not as commonly reported as that in the white matter and, where reported, involves neuron shrinkage and loss, chromatolysis, and gliosis. Dolman described a slight loss of anterior horn cells in three patients and chromatolysis in four others (8). Similarly, neuron loss in the anterior horn was reported in a single patient in each of two studies (9,13) where surviving cells appeared shrunken. In contrast, Reske-Nielsen and Lunbaek documented mild-to-moderate loss of neurons in both the posterior and anterior horns that was associated with mild gliosis in their series of 15 long-term juvenile diabetic patients (11). In this study, the remaining neurons had swollen or displaced nuclei and cytoplasmic PAS- and fat-positive accumulations. Similar to lesions in the white matter, neuronal loss in the gray matter has been reported to be more severe in the lower spinal cord segments (13).
As with motor neurons in the anterior horn, diabetes-associated injury to primary sensory neurons in spinal ganglia has only been sporadically documented. Dolman (8) and Greenbaum et al. (13) observed some loss of neurons, particularly in lumbar ganglia, accompanied by proliferation of satellite cells. In a clinically uniform series of nine patients dying after early-onset diabetes of long duration, Olsson and colleagues (9) reported heavy loss of neurons and formation of noduli of Nageotte in ganglia from all levels, with chromatolysis in some of the remaining neurons. In a morphometric analysis of posterior spinal ganglia, Ohnishi et al. (17) documented a decrease in the relative frequency of large size (type A) cell bodies, but no loss of neurons. More recently, neuroaxonal dystrophy has been described in the posterior spinal ganglia (18). The swollen axon terminals and enlarged initial segments were usually located within the satellite cell sheath, where they compressed and distorted adjacent sensory neuronal cell bodies. These dystrophic accumulations, consisting of hyperphosphorylated neurofilaments or collections of tubulovesicular bodies with intermingled neurotransmitter granules, occurred in calcitonin gene related peptide (CGRP)-containing axons, but not in sympathetic noradrenergic axons, which appeared earlier and with greater frequency in diabetic patients than in aged human subjects, and are also a structural hallmark of diabetic autonomic neuropathy (reviewed in ref. 19).
Similar to peripheral neuropathy, radiculopathy is a frequently described manifestation of diabetes mellitus and is well documented in the literature. Williamson described degenerated nerve fibers in the intramedullary course of posterior roots in Lissauer's fasciculus of lumbar and cervical spinal segments and noted that generally roots external to the spinal cord in these regions lacked degenerated fibers (5). Others (9,11,13) have also documented varying degrees of loss and degeneration of myelin sheaths and axons in spinal roots. In these studies, the extramedullary portions of both posterior and anterior roots were affected, although injury to the anterior roots was usually less advanced. In contrast to these observations are reports of widespread demyelination without extensive axonal degeneration in spinal roots (8,12,17). The presence of segmental demyelination and remyelination in both posterior and anterior roots with and without myelinated fiber loss, respectively (17), argues that this Schwann cell response is not secondary to axonal atrophy and/or degeneration.
In summary, there is unequivocal histological evidence of radiculopathy and degeneration of the long tracts in studies with diabetic patients that range from clinically uniform to unselected. Lesions at both sites are more frequent in the posterior roots and columns, and with the exception of microinfarcts, may well be independent of vascular lesions in these sites. Neuronal degeneration is less commonly observed in sensory ganglia and spinal cord gray matter. Much of the debate related to diabetes-induced spinal cord injury has centered on whether myelopathy is an independent lesion or a secondary consequence of peripheral neuropathy and/or radiculopathy, the most frequently described neurological manifestations of diabetes mellitus. Indeed, many symptoms and clinical signs considered indicative of myelopathy can be attributed to involvement of the peripheral neuraxis. For example, abnormal cutaneous sensation, paresthesia, and ataxia clearly involve lesions in peripheral nerves and roots. However, alterations in proprioceptive sensation and lack of muscle coordination without profound changes in cutaneous sensation point to an independent lesion in the ascending tracts of the posterior column (20). Further, the dissociation of histological changes in the posterior columns from posterior radiculopathy or clinical signs of sensory neuropathy in some of the cases reported by Slager (12) argue that myelopathy is an independent lesion.
Although there is structural evidence for spinal lesions in human diabetes mellitus, little support is present in the literature for comparable injury in experimental animal studies. There is better documentation of radiculopathy in animal studies and some reports of neuronal degeneration and loss in the dorsal root ganglia. Lack of evidence for myelopathy in animal, particularly rodent, studies is perhaps not surprising, given that there are relatively few pathological abnormalities in peripheral nerves from diabetic animals. The majority of animal studies concerned with structural injury to the spinal cord, dorsal root ganglia, and roots involve rats and mice with streptozotocin-induced diabetes, with an occasional report concerned with alloxan-induced diabetes or genetic models. Unlike the paraffin-embedded autopsy material of human studies noted earlier, most animal studies are based on plastic-embedded material and appear in the literature after the definitive human reports.
The earliest report of diabetes-induced structural abnormalities in the spinal cord is a morphometric study of lower motor and primary sensory neurons in rats with streptozotocin-induced diabetes of 4 weeks duration (21). Although no histological signs of neuronal degeneration or loss were identified, diabetes was associated with a significant reduction in perikaryal volume of both motor and sensory neurons of the same magnitude as the atrophy previously reported for axons in the peroneal nerve (22,23). Aside from preliminary observations of a motor neuron loss throughout the anterior horn in streptozotocin- and alloxan-diabetic rats and db/db mice that have not been subsequently published (24) there are no reports of neuronal loss in spinal cord gray matter. Regarding structural injury in the spinal cord, Yagihashi et al. (25) observed neuroaxonal dystrophy in neurons of the sensory ganglia and in the myelinated axons constituting the tracts of the posterior columns in spontaneously diabetic BB/Wor rats. As described by Schmidt et al. (18) in human studies, the dystrophic accumulations consisted of tubulovesicles, tubular rings, layered and electron-dense membranes and neurofilaments, and increased with increasing duration of diabetes.
Unlike the spinal cord, the literature concerned with structural injury to sensory ganglia and roots in experimental diabetes is more extensive but also more controversial, particularly with respect to the ganglia. Points of contention include whether there is neuronal loss in sensory ganglia and whether there is histological evidence of neuronal degeneration. As noted above, Sidenius and Jakobsen (21) reported no cell loss in the lumbar dorsal root ganglia of streptozotocin-diabetic rats, with abnormalities restricted to decreases in perikaryal volume and the relative number of large type-A neurons. Using systematic dissector counting approaches, Russell et al. (26) and Zochodne et al. (27) report neuronal atrophy but with a relative preservation of neurons in lumbar dorsal root ganglia from rats with streptozotocin-induced diabetes of 3 or 12 months duration, respectively. Using less rigorous profile counts of lumbar dorsal root ganglia from streptozotocin-diabetic rats after 12 months of diabetes, Kishi et al. (28) were also unable to detect neuronal loss but did confirm an increase in the ratio of small type-B neurons to large type-A neurons reported earlier (21,29). At this point, overt sensory neuron loss in experimental diabetes has only been observed in a recent study (30) using long-term streptozotocin-diabetic mice, which do not show the neuronal atrophy reported in rats except when overexpressing human aldose reductase (31).
In spite of morphometric evidence documenting preservation of neuron number, neuronal apoptosis has been claimed in several experimental reports using streptozotocin-diabetic rats with diabetes ranging from 1 to 12 months duration (26,32). TUNEL-positive neurons in the dorsal root ganglia, considered indicative of apoptosis, consisted of 7-34% of the total and was concomitant with markers of oxidative stress and activation of caspase-3. It is worth noting that apoptosis in experimental diabetic neuropathy is not supported by morphometric determination of neuronal or peripheral myelinated fiber loss, suggesting DNA strand damage does not necessarily equate to neuronal loss. Further, sensory neurons with activated caspase-3 survive long-term streptozotocin-induced diabetes (33), again indicating that expression of apoptotic markers does not always correspond to cell death.
Regarding structural injury in the dorsal root ganglia in experimental diabetes, Sidenius and Jakobsen (21) report that neuronal degeneration was not evident in rats with short-term streptozotocin-induced diabetes. As noted earlier, neuroaxonal dystrophy has been observed in sensory ganglia from diabetic BB rats (25). Several studies have documented degenerative changes in the dorsal root ganglia neurons from short-term (26) and long-term (28,34) streptozotocin-diabetic rats. Neuronal vacuolation is the most prominent change described, with vacuoles evenly distributed throughout the perikaryon in diabetic animals (28). In some instances, vacuoles appear to contain remnants of mitochondrial cristae or to be associated with lipofuscin granules (34). Vacuoles, condensed chromatin, and ballooned mitochondria with disrupted cristae were observed in neurons and dorsal root Schwann cells in short-term experimental diabetes by Russell et al. (26). Although neuronal vacuolation and mitochondrial disruption have been suggested to result from hyperglycemia-induced oxidative stress (26,28,34), these changes may be a feature of poor fixation. Moreover, aging-associated oxidative mitochondrial damage involves effaced cristae in an otherwise intact, not ballooned, organelle (35).
Radicular pathology in experimental diabetes was first reported by Tamura and Parry (36) and has subsequently been confirmed with remarkable agreement by others in both streptozotocin-diabetic and galactose-fed rats (28,34,37-39). The structural abnormality is focused on the myelin sheath and occurs in the context of marked interstitial oedema in both roots, although it is more frequent in the dorsal root. The earliest change consists of myelin splitting at the intraperiod line progressing to often-spectacular myelin ballooning. At this stage, strands of tubulovesicular myelin debris span the intramyelinic space and intratubal macrophages are sometimes observed stripping away myelin lamellae. There is minimal axonal degeneration associated with this myelin defect, suggesting that this lesion is a primary Schwann cell defect. Similar radicular pathology has been described in aged rodents (40), leading to the suggestion that its earlier appearance in experimental diabetes represents an acceleration of the aging process in this disease (41). However, myelin splitting and ballooning is prevented by aldose reductase inhibition and is present in several toxic neuropathies, pointing to other aetiologies (38).
In summary, whether there is neuronal loss and degeneration in dorsal root ganglia remains an unresolved issue in experimental diabetes. Presently, neuronal loss has only been documented in a murine model of long-term streptozotocin diabetes. In spite of claims of programmed cell death based on various markers of apoptosis, pathological evidence of neuronal loss and degenerative changes has not yet been convincingly demonstrated in the streptozotocin-diabetic rat. Radiculopathy is the most striking and consistently observed pathology in experimental diabetes and provides a stark contrast to the more modest injury present in peripheral nerves in rodent models of diabetes. As for spinal cord injury, there is a distinct paucity of evidence, with a single mention in the literature of neuroaxonal dystrophy in the posterior columns. With the possible exception of radiculopathy, there is discordance between the neuropathology present in diabetic animals and humans. The short life-span and small size of rats and mice, the most frequently used species in animal studies, may preclude the full development of the diabetic neuropathy that develops over time with a "stocking and glove" distribution in the longer nerves of humans. Also, particularly with reference to the spinal cord, it is possible that lesions have been overlooked because this tissue has not been as routinely studied as the peripheral nerves and spinal roots. Careful evaluation of longer-term rat and mouse models, while controlling for age-related confounds, might help reconcile the differences in structural injury between human and experimental diabetic neuropathy, as might the development and characterization of new animal models.
Was this article helpful?