The Spectrum of Diabetic Neuropathies

Abstract

Diabetes mellitus is associated with many different neuropathic syndromes, ranging from a mild sensory disturbance as can be seen in a diabetic sensorimotor polyneuropathy, to the debilitating pain and weakness of a diabetic lumbosacral radiculoplexus neuropathy. The etiology of these syndromes has been extensively studied, and may vary among metabolic, compressive, and immunological bases for the different disorders, as well as mechanisms yet to be discovered. Many of these disorders of nerve appear to be separate conditions with different underlying mechanisms, and some are directly caused by diabetes mellitus, whereas others are associated with it but not caused by hyperglycemia. We discuss a number of the more common disorders of nerve found with diabetes mellitus. We discuss the symmetrical neuropathies, particularly generalized diabetic polyneuropathy, and then the focal or asymmetrical types of diabetes-associated neuropathy.

Introduction

The association between diabetes mellitus and neuropathy has been recognized for well over 100 years and soon it was realized that different subtypes existed, and so the first classification was proposed by Leyden in 1893, with hyperesthetic (painful), paralytic (motor), and ataxic forms of diabetic neuropathy (1). There are many varieties of neuropathy classified under the term “diabetic neuropathy”, some clearly linked to hyperglycemia and subsequent metabolic and ischemic change, others with compressive etiologies, and still others which are associated with inflammatory/immune processes. Many types of nerves can be affected, including large-fiber sensory, small-fiber sensory, autonomic, and motor, and findings may or may not be symmetric. Distal nerves as well as large nerve trunks, nerve roots, and cranial nerves can be damaged. The most common of these syndromes is the diabetic sensorimotor polyneuropathy, which can produce mild distal sensory abnormalities as well as distal weakness. However, some of the rarer associated conditions, such as diabetic lumbosacral radiculoplexus neuropathy, are important to recognize, as they can produce severe pain and weakness with considerable morbidity. Prognoses for the diabetic neuropathy syndromes are also varied and dependent on the underlying pathology, with some of them being progressive disorders and others being monophasic illnesses. For the purposes of this review, we will first discuss the more diffuse neuropathic processes, and then the focal and asymmetrical forms. Since Leyden’s original classifications, many different classifications of diabetic neuropathy have been made. The two ways of classification preferred here are to divide diabetic neuropathies by the clinical pattern into symmetrical or asymmetrical forms (Table 1) or to divide diabetic neuropathies based on the current understanding of pathophysiology (Table 2).

Table 1.

Neuropathies Associated with Diabetes Mellitus based on Anatomical Pattern

Symmetric

Diabetic polyneuropathy (DPN)

Diabetic autonomic neuropathy (DAN)

Neuropathy with impaired glucose tolerance

Painful sensory neuropathy with weight loss, “diabetic cachexia”

Insulin neuritis

Hypoglycemic neuropathy

Polyneuropathy after Ketoacidosis

CIDP in DM

Asymmetric

Diabetic cranial neuropathy

Diabetic mononeuropathy:

Median neuropathy at the wrist

Ulnar neuropathy at the elbow

Peroneal neuropathy at the fibular head

Radiculoplexus neuropathies (DRPN)

DTRN

DLRPN

DCRPN

CIDP = chronic inflammatory demyelinating polyradiculoneuropathy; DLRPN = diabetic lumbosacral radiculoplexus neuropathy; DTRN=diabetic thoracic radiculoneuropathy; DCRPN=diabetic cervical radiculoplexus neuropathy; DM = diabetes mellitus. (Reprinted with permission from Sinnreich M, Taylor BV, Dyck PJB. Diabetic neuropathies classification, clinical features and pathophysiological basis. The Neurologist, 11(2):63–79, 2005.)

Table 2.

Pathophysiologic Classification of Diabetic Neuropathies

Presumed Underlying Pathophysiology	Subtype of Neuropathy
Metabolic-microvascular-hypoxic	DPN
	DAN
Inflammatory immune	DLRPN
	DTRN
	DCRPN
	Cranial neuropathies
	Painful neuropathy with weight loss, “diabetic cachexia”
	CIDP in DM
Compression and repetitive injury	Median neuropathy at the wrist
	Ulnar neuropathy at the elbow
	Peroneal neuropathy at the fibular head
Complications of diabetes	Neuropathy of ketoacidosis
	Neuropathy of chronic renal failure
	Neuropathy associated with large vessel ischemia
Treatment related	Insulin neuritis
	Hyperinsulin neuropathy

DPN = diabetic polyneuropathy; DAN = diabetic autonomic neuropathy; DLRPN = diabetic lumbosacral radiculoplexus neuropathy; DTRN = diabetic thoracic radiculoneuropathy; DCRPN = diabetic cervical radiculoplexus neuropathy; CIDP = chronic inflammatory demyelinating polyradiculoneuropathy; DM = diabetes mellitus. (Reprinted with permission from Sinnreich M, Taylor BV, Dyck PJB. Diabetic neuropathies classification, clinical features and pathophysiological basis. The Neurologist, 11(2):63–79, 2005.)

Diabetic Sensorimotor Polyneuropathy (DPN)

Diabetic sensorimotor polyneuropathy (DPN) is felt to result from nerve and blood vessel changes due to chronic hyperglycemic exposure, and can occur in either type 1 or type 2 diabetics. This syndrome typically presents as a slowly progressive primarily sensory deficit in a length-dependent fashion, with symptoms starting in the feet, and spreading upwards, evoking the classic “stocking-glove distribution”. In more severe cases, it will involve motor fibers as well, and can produce footdrop as well as other distal lower extremity weakness. It can involve both large and small fibers, and can have associated autonomic symptoms and signs. Upper extremity symptoms/signs can appear as part of the proximal progression of deficits, but in most cases, these are due to superimposed mononeuropathies (median neuropathy at the wrist and ulnar neuropathy at the elbow) rather than spread of the underlying polyneuropathy (2, 3).

Dyck et al. evaluated 380 patients in a cohort of diabetic patients representative of the community (the Rochester Diabetic Neuropathy Study) and found that 66% of the patients with type 1 and 59% of the patients with type 2 diabetes mellitus had some type of neuropathy. In both the type 1 and type 2 patients, the most common neuropathy found was a DPN (the second most common neuropathy found in both of these groups was carpal tunnel syndrome). Despite the finding that DPN occurs commonly, most cases were asymptomatic and detected because of findings on clinical examination and electrophysiologic studies. Only 15% of the type 1 patients and 13% of the type 2 patients had a symptomatic polyneuropathy, and a very small percentage overall (6% of type 1 patients and 1% of type 2 patients) had severe symptomatic polyneuropathy with inability to walk on heels. These observations underscore the fact that most DPN is mild and that severe weakness only rarely occurs. In diabetic patients with significant weakness, careful attention must be paid to other possible etiologies for their weakness, which can include other forms of diabetes-associated neuropathies, such as diabetic lumbosacral radiculoplexus neuropathies (DLRPN) (to be discussed later), and other types of neuromuscular disease which can occur independently of diabetes mellitus.

It has been shown that there are progressive subclinical nerve conduction abnormalities that precede the ultimate clinical diagnosis of DPN (4). Gregerson (1967) found a correlation between slowing of conduction velocities and duration of diabetes mellitus (5). A longitudinal study of risk factors for the severity of DPN found a correlation between severity and multiple other microvascular complications of diabetes such as retinopathy, proteinuria and microalbuminuria, as well as glycosylated hemoglobin. There was a strong correlation between the length of exposure of hyperglycemia and the degree of neuropathy et al., (Figure 1), and between the degree of neuropathy and degree of retinopathy (Figure 2) (6, 7).

Figure 1.

The change in the stage distribution of neuropathy (top) and retinopathy (bottom) with increasing duration of diabetes mellitus. (Reprinted with permission from Dyck PJ, Kratz KM, Karnes JL, et al. The prevalence by staged severity of various types of diabetic neuropathy, retinopathy, and nephropathy in a population-based cohort: The Rochester Diabetic Neuropathy Study. Neurology, 43:817–824, 1993.)

Figure 2.

A diagrammatic representation of Rochester, MN, diabetic patients with insulin-dependent diabetes mellitus (IDDM, left) and non-insulin-dependent diabetes mellitus (NIDDM, right) who had different stages of severity of neuropathy (top) and retinopathy (bottom). For both complications, there is a more severe distribution of stages of the complication in IDDM than NIDDM. (Reprinted with permission from Dyck PJ, Kratz KM, Karnes JL, et al. The prevalence by staged severity of various types of diabetic neuropathy, retinopathy, and nephropathy in a population-based cohort: The Rochester Diabetic Neuropathy Study. Neurology, 43:817–824, 1993.)

The etiology of DPN seems to relate to microvascular damage. The cause of the microvascular damage may be multifactorial but probably relates to chronic hyperglycemia-mediated direct metabolic effect. Sural nerve biopsies of diabetic patients with peripheral neuropathy reveal increased numbers of endothelial nuclei and excess thickness of endoneurial microvessels due to reduplication of the endothelial basement membrane (Figure 3), with a greater range in the thickness of vessel lumens (8). There is also macrovascular disease that is probably the cause of coronary artery disease and stroke, which is found in nerves but is not the cause of DPN (Figure 4).

Figure 3.

This microvessel shows severe basement membrane reduplication and a very high number of cellular debris among the basement membrane leaflets. Basement membrane leaflets are often incomplete and fragmented. (X 14,000 before 25% reduction; inset, X 3,600 before 25% reduction) (Reprinted with permission from Giannini C, Dyck PJ. Ultrastructural morphometric abnormalities of sural nerve endoneurial microvessels in diabetes mellitus. Ann Neurol, 36:408–415, 1994.)

Figure 4.

Epineurial arterioles were more frequently abnormal in nerves of diabetics than in nerves of controls. The arteriole enclosed by a rectangle in A and shown in greater detail in B, shows obliteration of the lumen by fat deposition, intimal proliferation, and thickening of the wall. (Reprinted with permission from Dyck PJ, Lais A, Karnes JL, et al. Fiber loss is primary and multifocal in sural nerves in diabetic polyneuropathy. Ann Neurol, 19:425–439, 1986.)

Because of the significant morbidity that can occur with DPN and its association with hyperglycemia, there has been considerable interest in the possibility of preventing, slowing, or reversing neuropathy by improved control of blood sugar. The Diabetes Control and Complications Trial Research Group (1995) studied 1441 insulin-dependent diabetic patients over a mean duration of 6.5 years and provided them with either conventional treatment, defined as 1–2 daily insulin injections, or intensive treatment, defined as greater or equal to 3 daily insulin injections or continuous insulin infusion with greater than or equal to 4 glucose tests daily. They found that at 5 years, subjects receiving intensive therapy had a 64% less development of confirmed clinical neuropathy; the prevalence of abnormal nerve conduction studies was lower by 44% and abnormal autonomic function testing was lower by 53% in the intensively treated group (9). This is an encouraging finding, and shows that good glycemic control can delay and probably prevent the development of DPN. However, there continue to be efforts to find neuroprotective agents that may be useful for treatment, once DPN has occurred.

Because of growing evidence that oxidative damage may play a significant role in the development of polyneuropathy, anti-oxidant agents such as lipoic acid have been studied. In an experimental rat model of diabetic neuropathy (streptozotocin-induced), the administration of intraperitoneal lipoic acid improved nerve blood flow and markers of oxidative stress in a dose-dependent manner as well as the conduction velocity of the digital nerve (10). Alpha-lipoic acid was evaluated on human patients with symptomatic DPN as part of the SYDNEY trial in which patients were treated with either intravenous alpha-lipoic acid 5 days a week for a total of 14 treatments or placebo. The primary endpoint of change in Total Symptom Score (which assesses positive neuropathic symptoms) was met by showing a significant improvement in positive symptoms in the treated group versus placebo. There was also a significant improvement in the Neuropathy Impairment Score (NIS) (11). However, the results of the ALADIN III study for alpha-lipoic acid were not as encouraging. This study randomized type 2 diabetic subjects with DPN into 3 groups: 1) alpha-lipoic acid IV daily for 3 weeks, then orally for 6 months, 2) alpha-lipoic acid IV daily for 3 weeks, then placebo orally for 6 months, and 3) placebo IV daily for 3 weeks, then placebo orally for 6 months. In this study, which evaluated treatment effect for a longer period of time than the SYDNEY trial, there was no significant difference after 7 months between the groups in Total Symptom Score (TSS) and in change in Neuropathy Impairment Score (NIS) between the alpha-lipoic acid group and the placebo group (12).

There may be a role of increased activity of the polyol pathway and increased glycation end products in the pathogenesis of DPN (13). Diabetic patients have been found to have higher levels of endoneurial glucose, fructose, and sorbitol than in control patients (14). There is greater overall microvessel area in nerve of streptozocin-induced diabetic rats, and microvessel basement membrane thickening has been prevented with the use of an aldose reductase inhibitor (15). There are multiple trials studying the effects of various aldose reductase inhibitors, which block the polyol pathway, in an attempt to decrease or prevent the neuropathic complications of diabetes mellitus. Unfortunately, results have been mixed.

Judzewitsch et al. (1983) utilized the aldose reductase inhibitor sorbinil in 39 diabetic patients, and found that there were very mild improvements in nerve conduction velocities after 9 weeks of treatment; the clinical significance of these slight changes is indeterminate (16). Another study of sorbinil in patients with DPN utilized sorbinil or placebo for a 6 month period, and found no major differences in clinical benefit between the two groups, and no difference on sensory threshold studies. There was significant benefit in the sorbinil-treated group in only a very limited number of electrophysiologic and autonomic parameters, such as conduction velocity in the posterior tibial nerve (17). Another aldose reductase inhibitor, tolrestat, was used to treat patients with DPN over a 52 week period compared with placebo, and significant concordant improvement in paresthesias and motor nerve conduction velocities over placebo was found in the group receiving the highest tolrestat dose, 200 mg daily. However, only 28% of the tolrestat-treated patients (at 200 mg a day) had concordant improvement at 24 week evaluation maintained at the final 52 week evaluation (18). Krentz et al. (1992) treated patients with DPN with either ponalrestat – another adolase reductase inhibitor – or placebo for 52 weeks and found no significant differences between the groups in motor or sensory nerve conduction velocities, vibration thresholds, symptom scores, or autonomic function tests (19).

Recombinant human nerve growth factor (rhNGF), which promotes the survival of small fiber sensory and sympathetic neurons, has been also studied in patients with DPN, with the hopes that it could improve impaired nerve function. Apfel et al. (1998) performed a study in which 250 patients with symptomatic DPN were treated with 6 months of either rhNGF or placebo, and found a trend toward improvement in NIS(LL) in the treatment group (20). A trial evaluating the use of rhNGF in patients with HIV-associated sensory neuropathy found a benefit versus placebo in self-reported neuropathic pain intensity and pinprick sensitivity, suggesting that this may be a useful agent in different types of neuropathy (21). These studies helped bolster enthusiasm for the potential benefit of this agent. A larger randomized, double-blind placebo controlled trial of 1019 patients with either type 1 or type 2 diabetes and DPN administered either rhNGF or placebo over a 48 week period. No significant change between baseline and the 48 week examinations in the NIS(LL) was found (22).

These results are somewhat disheartening and it remains unclear if any of these agents will ultimately be helpful in treating DPN. At the present time, control of blood sugars and prevention are the most important ways of dealing with DPN.

Diabetic Autonomic Neuropathy (DAN)

Autonomic abnormalities can occur in diabetes mellitus with or without the presence of a large-fiber neuropathy. Many experts consider that diabetic autonomic neuropathy (DAN) is really a part of the larger category of DPN. Low et al. (2004) reviewed the autonomic symptoms and standardized autonomic testing of patients with diabetes mellitus (Type 1 and 2) and of control patients, and found that 54% of patients with Type 1 diabetes, and 73% of patients with Type 2 diabetes had objective autonomic impairment, but this was generally in the mild range. Only 14% of the diabetic patients in that study had moderate to severe generalized autonomic failure (23). The numbers were even smaller for the Rochester Diabetic Neuropathy Study (7).

Autonomic failure can result in many troublesome symptoms, including orthostatic hypotension, nausea and constipation from abnormal GI motility, incontinence and erectile dysfunction. The complex management of each of these problems is beyond the scope of this discussion, and often requires the coordinated care of multiple medical specialties.

Autonomic instability in these patients may result in greater surgical risk and morbidity; Burgos et al. prospectively studied 17 diabetic and 21 non-diabetic patients – who had previously been given autonomic screening – during elective ophthalmologic surgery and found that 35% of diabetics required intraoperative vasopressors compared to only 5% of the control group. Furthermore the diabetics who required pressors had significantly greater autonomic impairment than those who did not (24). Investigators have also had concerns that DAN may be associated with sudden cardiac death, possibly related to abnormal lengthening of QT intervals (25). A recent review of cases of sudden cardiac death in diabetic patients, however, has revealed a greater correlation between sudden cardiac death and atherosclerotic heart disease and nephropathy than with DAN – the authors conclude that while there is an association between autonomic neuropathy and sudden cardiac death in diabetic patients, it is probably not the causative factor, and atherosclerotic heart disease is more important (26).

Polyneuropathy associated with Glucose Intolerance

There is increasing interest in an association between impaired fasting glucose or impaired glucose tolerance (IGT), which does not meet the criteria for diabetes mellitus and the development of a chronic axonal polyneuropathy. Current American Diabetes Association (ADA) guidelines, recently revised in 2003, require a fasting plasma glucose from 100 mg/dL to 125 mg/dL, for a diagnosis of impaired fasting glucose, and a 2-hour glucose level from 140 mg/dL to 199 mg/dL (after a 75 gram oral glucose load) for the diagnosis of IGT (27). It is estimated that approximately 33% of adults in the United States over 60 years old have either diabetes mellitus or impaired fasting glucose (diagnosed or undiagnosed) (28). This value was based on the earlier ADA criteria of impaired fasting glucose as a level between 110 mg/dL and 126 mg/dL, so the expectation is a higher overall incidence and prevalence with the new criteria. This percentage also does not take into account the greater number of patients with impaired glucose metabolism which may be detected through 2 hour oral glucose tolerance tests (OGTT), which would be expected to increase the purported population at risk. Studies have shown higher yields for abnormal glucose metabolism in glucose tolerance tests than in fasting plasma glucose measurements (29, 30).

Singleton et al. prospectively evaluated a cohort of 107 patients with idiopathic symmetric distal peripheral neuropathy, and found that 34% had IGT on an OGTT, which they noted is three times the prevalence of an age-matched historical cohort. Furthermore, only 72 of their subjects had an OGTT done, with a yield of 50% of the subjects receiving this test with an ultimate diagnosis of impaired glucose tolerance. The neuropathy pattern was predominantly sensory, as 81% of the patients with IGT had only sensory complaints, with 92% reporting neuropathic pain as a dominant symptom. Electrophysiologic findings revealed over half (61%) of the IGT patients had decreased sural sensory amplitudes, whereas only 21% had a decreased peroneal motor amplitude (31). Sumner et al. reported on 73 patients with a peripheral neuropathy of unknown cause, and found that 56% of them had abnormal findings on an OGTT - 26 of them with IGT, and 15 with diabetes mellitus. The authors found that the patients with IGT had less severe neuropathy and more predominant small fiber involvement than the diabetic patients (32). Hoffman-Snyder et al. retrospectively assessed 100 consecutive patients with an idiopathic chronic axonal neuropathy, who had fasting plasma glucose and a 2-hour OGTT. By the new ADA 2003 criteria, 39% of these patients had an abnormal fasting blood sugar, 36 of whom were in the range of impaired fasting glucose and 3 who were in the diabetic range, whereas using the OGTT even a higher percentage (62%) of the patients had an abnormal glucose metabolism, 38 with IGT and 24 with diabetes mellitus. These rates were higher than previously published age-matched controls (33%). The authors comment that abnormal glucose metabolism was found at similar rates in sensorimotor, sensory, and small fiber neuropathies (33).

There is also some pathologic evidence of increased endoneurial capillary density in sural nerves of patients with IGT progressing to DM compared to patients with stable IGT, indicating that microvascular abnormalities are occurring in a pre-diabetic state (34). Skin biopsies of patients with neuropathy and IGT have demonstrated abnormal intraepidermal nerve fiber density (35, 36). Smith et al. showed that patients with IGT and neuropathy had significantly improved proximal intraepidermal nerve fiber density and improved response on OGTTs after one year of a diet and exercise counseling program compared to their baseline OGTTs. The authors note however, that patients with absent dermal plexus on the initial biopsy were unlikely to have significant reinnervation on repeat biopsy (36).

However, not all evidence has been supportive of the association between IGT and neuropathy. Eriksson et al.(1994) performed nerve conduction studies, heat, cold and vibration threshold testing, and autonomic function testing on patients with IGT, diabetic patients, and non-diabetic controls. They found that aside from significantly more abnormalities in heart rate with inspiration and expiration (suggestive of vagal nerve involvement), there were no other parameters that were significantly different between their patients with IGT and controls (37). Hughes et al. studied 50 patients with chronic idiopathic axonal polyneuropathy and 50 control patients, with OGTT and fasting plasma glucose analysis, and did not find an association with IGT and neuropathy after adjusting for age and sex (38). The lack of a significant association is particularly compelling, given that there was an active control group in the study, as opposed to most of the other studies cited, which used historical control data as a comparative measure of prevalence of IGT in the general population.

At this point in time, the relationship between IGT and peripheral neuropathy needs further clarification. Although a very interesting and potentially important hypothesis, the association has not been definitely proven. A large epidemiology study has been begun looking at IGT and neuropathy in Olmsted County, Minnesota.

Acute Painful Diabetic Neuropathy with Weight Loss

Cachexia and weight loss may be seen in association with diabetes mellitus (39). This acute painful neuropathy with weight loss is considered by some authors to be a clinical entity separate from DPN (40, 41). This syndrome is also known as diabetic cachexia and is not related to the severity or duration of diabetes mellitus and has a monophasic course, usually over months. The illness begins with sudden, profound weight loss followed by severe pain, often burning, and excessive sensitivity to touch (allodynia) of the lower legs and feet. Autonomic features other than impotence are usually absent. Some experts have argued that this entity is part of the spectrum of DPN with the same underlying mechanism and pain fibers being predominantly involved. This monophasic course, the lack of correlation between diabetes mellitus duration, and the neuropathy with associated weight loss, makes it unlikely to be part of DPN. In contrast, the occurrence of neuropathy in early diabetes mellitus, the associated weight loss, and the monophasic course are features that are typically found in diabetic lumbosacral radiculoplexus neuropathy (DLRPN), which is described later in this review. DLRPN has been shown to be due to microvasculitis and ischemic injury. The pathological basis of acute painful diabetic neuropathy with weight loss has not been determined, but an immune mechanism seems likely.

Insulin Neuritis

A rare form of peripheral neuropathy has been described in patients shortly after instituting insulin for diabetes mellitus, referred to as “insulin neuritis”. It was first described in 1933, but there have been multiple case reports of this phenomenon in the literature (42, 43). A typical example was that of a patient who had begun a continuous subcutaneous insulin infusion, and acutely developed a painful neuropathy. Sural nerve biopsy revealed changes of chronic neuropathy, and the authors raised the possibility that the symptoms were due to abnormal “ectopic” sensory responses caused by axon regeneration (43). Kihara et al. evaluated the effect of insulin on nerve oxygenation in both normal nerves and nerves of rats with streptozotocin-induced diabetes. They found that infusion of insulin caused a reduction in endoneurial oxygen tension in the normal nerves studied, and that streptozotocin-induced diabetic nerves were resistant to these changes. However, when the hyperglycemia in the diabetic nerves was controlled, they became more sensitive to this effect of insulin, and nerve blood flow was decreased and blood through arteriovenous shunts was increased (44). Tesfaye et al. reviewed sural nerve epineurial vessel photography and fluorescein angiography of 5 patients with insulin neuritis, and found severely abnormal epineurial vessels, with arterio-venous shunting and tortuosity, and in three patients proliferating “leaky” vessels were found; they suggest that perhaps these changes lead to an ischemic endoneurium (45). Recognition that immune factors are involved in other forms of diabetic neuropathy raises the possibility in insulin neuritis as well.

Hypoglycemic Neuropathy (Hyperinsulinemic Neuropathy)

There are several reports in the literature of polyneuropathy occurring in association with a chronic hyperinsulinemic state, and repeated episodes of hypoglycemia. This can be found in the context of an insulinoma, or insulin-secreting tumor of the pancreas.

Jaspan et al. reviewed 29 of these cases in the literature, including a case of their own, and found a mean age of 38 years old with a slight male predominance. The typical presentation was initially that of prominent distal uncomfortable paresthesias without significant objective sensory findings, followed by a motor-predominant distal symmetric peripheral neuropathy with atrophy. The upper extremities are generally more involved than the lower extremities, but foot drop is also common. Just less than half of the patients for whom detailed central nervous system data was available showed some signs of CNS dysfunction, usually mental status or personality changes, though a small minority had cerebellar findings as well. In their own case that they reported, both a sural nerve biopsy and a gastrocnemius biopsy were performed. A reduction in large myelinated fibers was observed on nerve biopsy, and muscle showed neurogenic atrophy, as well as a few necrotic fibers and some perivascular lymphocytes and histiocytes. Overall, in these patients, there was some improvement of weakness after removal of the insulinoma, as well as significant improvement of the sensory symptoms. They comment that peripheral nerve disease is rare compared to central nervous system disease in the hypoglycemic state, and that this may be related to the decreased dependence of the PNS on glucose metabolism (46).

Westfall et al. studied the effects of insulin in alloxan-induced diabetic rats, giving either daily insulin injections or insulin through a subcutaneous minipump. Controls were untreated diabetic rats and nondiabetic rats. Through teased fiber analysis of tibial nerves, they found evidence of Wallerian-type axonal degeneration only in the treated diabetic rats. Findings were worse in the daily insulin injection group, suggestive of primary damage to the nerves either through hyperinsulinemia or hypoglycemia (47). Sima et al. studied hyperglycemic (diabetic BB) rats, and gave them either small doses of insulin, high insulin doses to the point of hypoglycemia, or left them in a hyperglycemic state without insulin. In the hypoglycemic rats, loss of anterior horn cells, loss of large myelinated fibers and decreased nerve conduction velocities were found – while the other groups had mainly sensory fiber abnormalities (48).

Though there is clearly evidence that hypoglycemia can damage peripheral nerve, the mechanism by which this happens is unclear. Also not clear is whether peripheral nerve damage occurs in diabetic patients with intermittent periods of hypoglycemia, a topic that deserves further study.

Polyneuropathy after Ketoacidosis

Diabetic ketoacidosis can present with profound central nervous system abnormalities, particularly a depressed level of consciousness (49). A pathologic study of 6 patients who died with coma from diabetic ketoacidosis showed ischemic tissue damage both in the brain and other organs, felt to be related to abnormalities of intravascular coagulation (50). There are limited reports of peripheral nervous system damage during periods of diabetic ketoacidosis, including the clinical picture of multiple mononeuropathies (51, 52). It is unclear if these changes result from peripheral nerve ischemia or other hemodynamic and metabolic abnormalities during the course of their acute illness.

Chronic Inflammatory Demyelinating Polyradiculoneuropathy (CIDP) in Diabetes Mellitus

CIDP, as its name suggests, is an immune-mediated disorder of the peripheral nervous system, with primary damage to the myelin sheath, though many patients will develop secondary axon loss over time. It is characterized by symmetric distal and proximal weakness, hyporeflexia or areflexia, which progresses over at least a 2 month time period. Some people have a relapsing-remitting course and others can have a progressive phenotype. It is usually accompanied by elevated CSF protein and evidence of demyelination (slowed conduction velocities, temporal dispersion, conduction block, prolonged distal latencies, and prolonged F-wave latencies) on nerve conduction studies. Treatment is with immunomodulatory therapies, which can include steroids, intravenous immunoglobulin, plasma exchange, azathioprine, and a host of other immunosuppressive agents. There is suggestion in the literature that patients with diabetes mellitus may be at increased risk to develop CIDP, although this has been incompletely studied (53). Laughlin et al. studied the incidence and prevalence of CIDP in the general population and among diabetic patients in Olmsted County, Minnesota, and did not find an increased frequency among diabetic patients (54). The authors noted that since the population they studied was small, minor association between diabetes mellitus and CIDP might be missed but that a large association was unlikely.

Gorson et al. studied 14 patients with diabetes mellitus and clinical and neurophysiologic findings of CIDP compared them to 60 cases of idiopathic CIDP. Overall the diabetic patients were older and had a greater occurrence of imbalance, but otherwise their clinical symptoms were not remarkably different from the patients with idiopathic CIDP. Where they did differ is that on nerve conduction studies, the DM patients had significantly lower amplitude ulnar motor response, sural response was more likely to be absent, and there was a trend toward lower amplitude CMAPs and SNAPs overall. In addition the nerve biopsies of the diabetic patients were more likely to show predominately axonal loss than the CIDP patients. One explanation for these findings is that there is a superimposed DPN. The authors also noted that there was a lesser improvement in strength after immunomodulatory treatment in the DM group than the idiopathic CIDP (55). Stewart et al. evaluated 7 diabetic patients with CIDP, and found that all had distal muscle chronic denervation changes and distal more than proximal weakness and wasting; electrophysiologically the majority had motor conduction block, slow conduction velocities, prolonged distal latencies, and all had F-wave abnormalities. All received immunomodulatory treatment and had improvement of their symptoms; 6 of 7 improved by at least two modified Rankin grades (56).

While there is no question that diabetic patients can develop CIDP, the question that remains is whether diabetes mellitus is a risk factor for developing CIDP. One problem that exists in answering this question is that some authors have relied predominantly on nerve conduction studies in defining CIDP, and most other forms of diabetic neuropathy are well known to have some demyelinating features on nerve conduction studies, and so some cases of DPN (or other diabetic neuropathy) may be mislabeled as CIDP. Another potentially confusing group of conditions are the diabetic radiculoplexus neuropathies, which cause severe pain and weakness, and may be mislabeled as diabetic CIDP. These conditions may also respond to immune-modulating agents. Laughlin et al. (54) suggest, however, that if an association between diabetes and CIDP exists, it is small and probably not very clinically important. A definitive answer can only be obtained through large population based studies. Regardless of an association, these studies provide a cautionary note about assuming that all neuropathies in a diabetic patient are due solely to diabetes, as incomplete evaluation to identify other causes of neuropathy may deprive a patient of effective treatments, such as immunomodulatory therapy for CIDP.

Diabetic Cranial Neuropathy

Cranial neuropathies likely occur at a higher incidence in diabetic patients than in the general population, but this is insufficiently studied. Watanabe (57) reviewed cranial nerve findings in 1961 diabetics and 3841control subjects. He found a significantly higher incidence of cranial neuropathies (0.97%) in diabetic subjects than in the control subjects (0.13%) although the overall incidence was quite low in both groups. The most common cranial neuropathies found were oculomotor nerve palsies and facial nerve palsies. In the Rochester Diabetic Neuropathy Study cohort, no patients presented with cranial neuropathies (7).

Diabetic oculomotor palsy typically presents acutely with severe eye pain, and paresis of the oculomotor-innervated extraocular muscles – superior, inferior, medial rectus and inferior oblique muscles. This is accompanied by ptosis from involvement of the levator palpebrae. In most cases, this is a pupil-sparing lesion, as the parasympathetic fibers travel in the peripheral layers of the nerve, and the putative mechanism in diabetic ophthalmoplegia is ischemic rather than compressive. Particularly when there is pupillary involvement, alternative diagnoses such as enlarging aneurysm or other compressive lesions must be carefully ruled out. The prognosis in diabetic oculomotor palsy is good, and full recovery generally occurs, though it may take months to regain normal function (58). Other cranial nerves affecting extraocular muscles can also be involved in diabetes mellitus—in a study of 2229 patients who presented with an oculomotor palsy, 13.7% of them were found to be associated with diabetes mellitus. Of the diabetic group with cranial nerve involvement, 50% had cranial nerve VI involvement, 43.3% had cranial nerve III involvement, and 6.7% had cranial nerve IV involvement (59).

Facial nerve palsies are common in the general population, and it is unclear if their overall incidence is higher in the context of diabetes mellitus. As expected for a peripheral nerve lesion, this should result in weakness of the ipsilateral forehead as well as the rest of the ipsilateral face. There is no evidence that this should be treated in a different manner from idiopathic facial nerve palsy. Nerve conduction studies and EMG including the blink reflex may be important for documenting nerve integrity and prognostication but are generally not needed for the diagnosis.

Diabetic Mononeuropathies

Several mononeuropathies appear in greater frequency in diabetic patients than in the general population; these include median neuropathy at the wrist (carpal tunnel syndrome), ulnar neuropathy at the elbow, and peroneal neuropathy at the fibular head. As noted previously, the majority of diabetic patients with upper extremity neuropathic symptoms and signs will have a mononeuropathy or multiple mononeuropathies as the etiology rather than an extension of a generalized polyneuropathy (2, 3).

Median neuropathy at the wrist, or carpal tunnel syndrome, can present with dysesthetic symptoms or numbness in the first 3 fingers of the hand and the lateral surface of the 4^th finger, and/or weakness in median-innervated muscles of the hand. Sensory symptoms can extend into the wrist and forearm. There are many risk factors identified in the literature for development of carpal tunnel syndrome; these include rheumatoid arthritis, osteoarthritis of the wrist, previous wrist fracture, obesity, and diabetes (60). Of 414 patients with mild diabetic neuropathy enrolled for the EDIT trial, 23% met criteria for median neuropathy at the wrist (MNW). These patients overall had a longer duration of diabetes than patients without MNW. The authors also found that degree of abnormality of sural or peroneal nerve conduction did not have an effect on the frequency of MNW (61). Stevens et al. (1992) reviewed the medical records of 1,016 patients diagnosed with carpal tunnel syndrome, and found that 56.8% of patients had a documented underlying conduction associated with MNW, with a standardized morbidity ratio of 2.3 for diabetes mellitus (62). In the Rochester Diabetic Neuropathy cohort, electrophysiological evidence of median neuropathy at the wrist was found in 22% of type 1 diabetes mellitus and in 29% of type 2 diabetes mellitus patients without any symptoms. Clinical evidence for median neuropathy at the wrist was found in 9% of type 1 and 4% of type 2 diabetic patients. The mechanism for the increased risk for carpal tunnel syndrome in diabetics is unclear, and may relate to compression or to increased stiffness of connective tissue (58).

Diabetic Radiculoplexus Neuropathies (DRPN)

There are characteristic diabetic syndromes, which present subacutely with pain followed by weakness, that affect primarily patients with mild diabetes called radiculoplexus neuropathies. Three main types can occur, alone or in combination, and include diabetic cervical radiculoplexus neuropathy (DCRPN), diabetic thoracic radiculoneuropathy (DTRN), and diabetic lumbosacral radiculoplexus neuropathy (DLRPN). All forms (especially DLPRN) can be associated with weight loss.

Diabetic Thoracic Radiculopathy

Diabetic thoracic radiculopathies are a rare, but important complication of diabetes mellitus. These typically present with severe pain and dysesthesias along the trunk, chest or abdominal wall, and often prompt extensive workups for underlying chest or abdominal pathology (63). They can be symmetric and can involve multiple dermatomes (64). They can be associated with weakness and outpouching of the abdominal wall, and can be mistaken for an abdominal hernia (65). They can be associated with significant weight loss. They also can occur in patients who already have a significant distal symmetric polyneuropathy of the lower limbs or a DLRPN. EMG may show abnormalities of denervation in the thoracic paraspinal muscles or in abdominal muscles. The association of these with other forms of radiculoplexus neuropathies was noted by Bastron and Thomas and termed diabetic polyradiculopathy (66). In case of uncertain diagnosis, the thermoregulatory sweat test can be useful, and may show pathologic loss of sweating in a thoracic or abdominal distribution. This sweat pattern has been found to correlate well with thoracic paraspinal muscle fibrillation potentials (67). Skin biopsies in affected areas of the chest or trunk show a decrease in epidermal and dermal nerve fibers compared to biopsies from asymptomatic areas (68). Overall the prognosis is good; most patients reported in the literature have good recovery, usually within months to a year after the onset of their symptoms, without any specific treatment (65, 69, 70).

Diabetic Lumbosacral Radiculoplexus Neuropathy

Diabetic lumbosacral radiculoplexus neuropathy (DLRPN) occurs in approximately 1% of diabetic patients (7) and probably is the form of diabetic neuropathy that causes the most morbidity. It has been variably known by different names, including diabetic amyotrophy, Bruns-Garland syndrome, diabetic mononeuritis multiplex, diabetic polyradiculopathy, proximal diabetic neuropathy, and others. We prefer the designation of DLRPN because it more accurately describes the extensive anatomic localization of the abnormalities in this disorder, i.e. involvement of nerves at the root, plexus and peripheral nerve levels. DLRPN more commonly affects patients with type 2 diabetes mellitus; the median age of presentation is 65 years, but the range in age is wide. These patients tend to have better glycemic control and a lower body mass index compared to a population-based study of diabetic patients (the Rochester Diabetic Neuropathy Study, RDNS) (71). DLRPN patients also have a low rate of coexistent end-organ damage related to diabetes mellitus. In particular, cardiovascular disease and retinopathy occurs at a significantly lower rate in these patients than in the diabetic population as a whole, suggesting an etiology other than metabolic derangement (longstanding hyperglycemic exposure) (71). It is unclear what role diabetes mellitus plays in the pathophysiology of this disorder, as a very similar condition is also found in nondiabetic patients (72). It seems likely that elevated blood sugars and diabetes mellitus is a risk factor for DLRPN but not the direct cause of it.

DLRPN usually presents with acute to subacute onset of severe lower extremity pain, either in the thigh or the leg, but ultimately spreading to the entire lower extremity. The presentation is usually unilateral or asymmetric and severe, and the majority of patients will require narcotic pain medication for adequate control of their symptoms. The pain is of differing types and includes aching, burning, sharp stabbing, and contact allodynia (73). Although pain is initially the worse symptom, weakness and atrophy become the main problem. Like the pain, the weakness usually begins focally but over time becomes widespread and bilateral. Weight loss is common; the median loss in our series is 30 pounds. Numbness and tingling are also very common, and about 50% of patients have autonomic symptoms at the time of presentation, which can include orthostatic hypotension, genitourinary, and gastrointestinal symptoms. Most patients will eventually have bilateral lower extremity signs and symptoms. In our study of 33 prospectively evaluated patients with DLRPN, only one did not develop bilateral symptomatology. However, these patients may be more severely involved than the average patients with DLRPN since they were evaluated at a tertiary referral center. The mean time from onset to bilateral disease is 3 months. In the natural history of the disease, recovery is substantial but incomplete, and the vast majority of patients have residual pain and some degree of leg weakness on long-term follow-up. While one-half of our patients required the use of a wheelchair, only 3 needed a wheelchair in long-term follow-up. However, 16 others still required ongoing assistive aids (e.g. walker, cane, brace) at long-term follow-up (2 years). Footdrop tended to be the biggest long-term problem as proximal segments re-innervated earlier and more completely (73).

Cerebrospinal fluid findings are suggestive of involvement of the root level, with elevated protein, the median value being 89 mg/dL, with a normal cell count being the typical pattern. This may suggest an inflammatory etiology but is not specific (73).

Neurophysiologic testing is most consistent with a process of primarily axonal degeneration. Nerve conduction studies show decreased peroneal and tibial compound muscle action potentials, and decreased sural sensory nerve action potentials, with significant side-to-side differences. The findings on needle electromyography implicate a multifocal process, which involves lumbosacral roots, plexus, and peripheral nerve. There is also objective evidence of autonomic neuropathy. In our series, 14 DLRPN patients had autonomic testing performed, 8 of whom had clinical autonomic symptoms. All 14 of these patients had abnormal composite autonomic severity scores (CASS), with 8 (57%) in the severely abnormal range. Of the four patients who had thermoregulatory sweat tests, all four were abnormal, with patchy anhidrosis of the affected areas of the lower limbs (73).

There is significant evidence to suggest that the primary pathologic process in DLRPN is ischemic injury from microvasculitis. As previously mentioned, we studied 33 prospectively evaluated cases with DLRPN who underwent distal cutaneous nerve biopsy (sural or superficial peroneal) (73). The nerve pathology showed evidence of ischemic injury - focal or multifocal fiber degeneration and loss (Figure 5), perineurial degeneration or thickening (Figure 6), injury neuroma, and epineurial neovascularization (Figure 6). The ischemic injury seemed to be due to altered immunity and microvasculitis. All nerves had increased inflammation, 15 of 33 had inflammatory cells in the vessel walls (suggestive of microvasculitis) (Figure 7) and 19 of 33 had evidence of prior bleeding (hemosiderin) (Figure 6). In two cases, changes diagnostic of necrotizing vasculitis were seen (73). The vasculitic changes tended to be those of microvasculitis involving microvessels and small venules with fragmentation of vessel walls and without fibrinoid necrosis.

Figure 5.

Transverse epoxy sections (p-phenylenediamine) of distal sural nerves from patients with diabetic lumbosacral radiculoplexus neuropathy illustrating the dramatic focal fiber loss characteristic of the disorder (A, fascicle on the left) and the abortive microfascicular nerve regeneration (B, as identified by the arrow). Note that the abortive regeneration (injury neuroma) is made up of multiple regenerating fascicles and that they are situated adjacent to a fascicle devoid of myelinated fibers. Most of the fibers in the right fascicle in the lower panel are actively degenerating. As discussed in the text, these changes are indicative of ischemic injury that we attribute to a microscopic vasculitis. (Reprinted with permission from Dyck PJB, Norell JE, Dyck PJ. Microvasculitis and ischemia in diabetic lumbosacral radiculoplexus neuropathy. Neurology, 53:2113–2121, 1999.)

Figure 6.

Transverse paraffin sections of sural nerves showing changes seen in DLRPN and LRPN. Upper panel, Masson’s trichrome stain, showing inflammation in the wall of an epineurial microvessel (right upper), probably fibrinoid degeneration of the perineurium (long arrow), and a region of neovascularization (arrowhead). Middle panel, Luxol fast blue-periodic acid Schiff stain, showing several fascicles surrounded by normal thickness perineurium (right middle, between the arrowheads) and one fascicle with extremely thick perineurium (left middle, between the arrowheads). We attribute the latter finding to scarring and repair after ischemic injury (note all fascicles are devoid of myelinated fibers). Lower panel, Turnbull blue stain, showing accumulation of hemosiderin (iron stains bright blue, arrow) deposited along the inner aspects of the perineurium. All of these pathological features are frequently seen together and are best explained by ischemic injury. (Reprinted with permission from Dyck PJB, Windebank AJ. Diabetic and nondiabetic lumbosacral radiculoplexus neuropathies: New insights into pathophysiology and treatment. Muscle Nerve, 25:477–491, 2002.)

Figure 7.

Serial skip paraffin sections of a microvessel above (upper panel), at (middle panel), and below (lower panel) a region of microvasculitis in the sural nerve of a patient with diabetic lumbosacral radiculoplexus neuropathy. The sections on the left column are stained with hematoxylin and eosin, the sections in the middle column are reacted to antihuman smooth muscle actin (Dako), and the sections on the right column are reacted to leukocytes (CD 45). The smooth muscle of the tunica media in the region of microvasculitis (middle panel) is separated by mononuclear cells, fragmented and decreased in amount. The changes are those of a focal microvasculitis. (Reprinted with permission from Sinnreich M, Taylor BV, Dyck PJB. Diabetic neuropathies classification, clinical features and pathophysiological basis. The Neurologist, 11(2):63–79, 2005.)

These pathologic findings suggestive of a microvasculitic etiology for DLRPN have also been noted by other authors. Llewelyn et al. performed nerve biopsies (intermediate femoral cutaneous or sural) on 15 patients with findings consistent with DLRPN, and found inflammatory changes in 5 of these, 4 of which had microvasculitis and decreased numbers of myelinated fibers in all specimens (74). Kelkar et al. studied 15 patients with DLRPN by clinical, EMG, and laboratory evaluation, who had nerve and muscle biopsy. In 4 of these patients, there was inflammatory infiltration of vessel walls with PMNs as well as IgM deposits. In 6 of the patients there was perivascular inflammation around epineurial blood vessels (75).

The findings of microvasculitis provide hope for effective treatment, particularly with immunomodulatory therapy. A retrospective study by Pascoe et al (1997) looked at the outcomes of patients with DLRPN either treated or untreated with immunosuppressive agents (either steroids, intravenous immune globulin or plasma exchange). They found a higher rate of improvement (9 of 12 patients) in the treated group compared to the untreated group (17 of the 29 patients), although this was not statistically significant (76).

A randomized double-blinded placebo-controlled prospective treatment trial for DLRPN was recently completed by one of the authors (77). Seventy-five patients with a mean age of 65.3 years were included in this trial, which randomized them to either placebo or IV methylprednisolone, starting at a dose of 1 gram three times a week, with tapering over a 12 week period. Subjects had neuropathy impairment scores (NIS), and neuropathy impairment scores-lower limbs (NIS(LL)), as well as the Neuropathy Symptoms and Change (NSC) taken at baseline, and at weeks 1, 6, 12, 24, 36, 52, and 104. The primary outcome measure for this study was time to improvement in NIS(LL) by four points. Both the steroid and the placebo group had significant improvement in the NIS and NIS(LL) by the end of the study. There was not a significant difference between the two groups in time to improvement in the NIS(LL) by four points even though the treated group reached this endpoint over 30 days sooner. However, the patients treated with methylprednisolone reported a significantly greater symptom improvement, particularly in regards to pain and positive neuropathic symptoms. (77).

While the primary outcome measure was not met for methylprednisolone treatment of DLRPN, there was a statistically significant improvement in secondary endpoints of pain and positive sensory symptoms in the treated group over the placebo group, which is encouraging. As previously noted, the natural history of DLRPN is one of improvement of pain symptoms over time, and marked but often incomplete recovery of motor function. It is unclear if the lack of significant improvement in NIS(LL) in the steroid-treated group compared to the placebo group may have been related to timing of treatment. The study subjects were evaluated at a tertiary referral center, and so typically were not seen early in their disease course. It may be that, if prompt diagnosis and steroid treatment occurred early in the disease, it may have prevented some of the disease and have had a large impact on their neurological impairment.

Diabetic Cervical Radiculoplexus Neuropathy

While DLRPN is a much more familiar branch of the DRPN spectrum, the cervical segment can also be involved. In our study of patients with DLRPN, 3 out of 33 patients (10%) had coexistent bilateral but asymmetric cervical radiculoplexus neuropathies (73). Katz et al. (78) reviewed medical records of 60 consecutive patients with DLRPN and found that nine of them (15%) also had upper extremity involvement. The average age of these patients with cervico-brachial involvement was 59 years old, and all had type 2 diabetes mellitus. Four of these nine patients had bilateral upper extremity involvement. In eight of the nine patients, weakness was restricted to or most prominent in distal muscles, while only one patient had isolated proximal arm weakness. Only one patient was free of sensory loss and the majority also had pain. All of these patients had electrodiagnostic testing, and seven had evaluation of an upper extremity; these showed evidence of denervation in affected muscles. Two of the patients were lost to follow-up but the remaining seven had improvement in their arm symptoms and signs over a period of up to nine months; three of those patients received treatment with some type of immunomodulatory regimen (78).

Conclusions

The spectrum of diabetes mellitus-associated neuropathies is large, and our knowledge of these entities continues to evolve. There can be nerve damage from metabolic injury, compressive injury, ischemic injury, and from altered immunity, and these varied pathologies can present in many different ways. Identifying the pathogenesis of these entities is valuable, as there are clear differences in treatment strategies, which may range from lifestyle modification and strict glucose control to immunosuppressant medications, and can make significant impact in the quality of life of patients suffering from diabetes mellitus. In order to do this effectively, the caring physician needs to be able to make the correct diagnosis first, before starting the appropriate therapy.