Clinical manifestations

Presentation and course

Diabetic amyotrophy is a disabling neuropathy that usually occurs in patients with type 2 diabetes mellitus in middle or later age. Concomitant weight loss is frequent. Patients usually have not had diabetes for a long time, and glycemic dysregulation is often not severe. Interestingly, a diagnosis of diabetic amyotrophy leads to the discovery of underlying diabetes mellitus in one quarter to one third of cases. Long-term diabetic complications such as diabetic retinopathy and nephropathy are often absent at the time of diagnosis. The clinical presentation is characterized by sudden, sharp, and asymmetric pain that usually starts in one hip and thigh and subsequently spreads to the other side within weeks to months (04; 03; 09). Cases of worsening over 18 months are described (03). Weakness starts distally in one third of cases, leading to unilateral or bilateral foot drop. As the pain improves, weakness becomes the major symptom; it affects both proximal and distal leg muscles.

In about one third of cases, weakness spreads to proximal arm muscles and is attributed to cervicobrachial radiculoplexopathy (39; 13; 27). However, some of these upper extremity deficits result from compression neuropathy from reliance on the arms to support body weight. Approximately 12% of patients develop thoracic radiculopathy, leading to radiating belt-like chest or abdominal pain and intercostal muscle weakness causing pseudohernia (53; 30; 51; 50).

Prominent bulbar symptoms have been described (10). Respiratory weakness may also be the initial manifestation of diabetic proximal neuropathy (06).

Diabetic amyotrophy can be painless. Graces-Sanchez and colleagues studied 23 patients (22 had type 2 diabetes mellitus) who developed painless, lower-limb, motor-predominant neuropathy. The clinical features were similar to diabetic amyotrophy, except painless patients had more symmetrical and upper limb involvement with slower evolution. They compared the nerve biopsy findings of these patients with the pathological features of 33 painful diabetic amyotrophy biopsies and 25 chronic inflammatory demyelinating polyradiculoneuropathy biopsies. The nerve biopsy findings of these patients were similar to the findings in classic painful diabetic amyotrophy but not CIDP (23).

Prognosis and complications

The weakness initially worsens over weeks or sometimes months, although the pain resolves earlier. With disease progression, half of the patients become wheelchair-bound, and almost all require assistance with ambulation. Progression may continue up to 18 months from onset. Recovery usually occurs spontaneously but is slow and often incomplete. Although only 9% of patients need a wheelchair after 2 years, only 10% of patients achieve a full recovery in the same time period. Relapses on the same or other side occasionally occur.

In a population-based study comparing patients with diabetic and nondiabetic lumbosacral radiculoplexus neuropathy (LRPN) to age and sex-matched control group, patients with lumbosacral radiculoplexus neuropathy had lower median survival (12.2 years for the lumbosacral radiculoplexus neuropathy group vs. greater than or equal to 17 years for the control group), which translated to 76% increased risk of death. This increased risk of death was related to the comorbidities found in the lumbosacral radiculoplexus neuropathy group and not the lumbosacral radiculoplexus neuropathy itself (40).

Clinical vignette

A 58-year-old diabetic man on an oral hypoglycemic treatment developed sudden severe pain that started in the lower back and radiated to the right leg. After 3 days, he experienced weakness in the same leg. After 45 days, similar pain and weakness occurred in the left leg. Symptoms persisted for 4 months, during which time the patient had a 10 kg unintentional weight loss. Neurologic examination revealed asymmetric proximal and distal weakness in both legs. There was atrophy of the quadriceps and hamstring muscles. A stocking pattern of loss of pinprick, light touch, and vibration sense was found. Tendon reflexes were absent at the ankles and decreased at the knees bilaterally. Upper extremities had normal motor and sensory function.

At the initial evaluation, fasting glucose was mildly elevated (136 mgs/dl), with a normal glycosylated hemoglobin of 6.8%. Peroneal motor conduction velocity was reduced at 36 m/s, and compound motor action potential was reduced at 300µV. Tibial nerve conduction was 38 m/s, and sural potentials were unobtainable. Median and ulnar nerve conduction velocities were normal. An EMG revealed denervation potentials in proximal (quadriceps and hamstrings) and distal (gastrocnemius and anterior tibial) muscles bilaterally. A sural nerve biopsy revealed an asymmetric loss of myelinated fibers and microvasculitis with endoneurial mononuclear cell infiltration.

His recovery was protracted. He required a knee brace for ambulation and was treated with tricyclic antidepressants for pain. Leg weakness progressed for 3 additional months before subsequent gradual improvement of strength. At the 6-month follow-up (10 months into the disease), the pain had almost resolved. However, he still had considerable residual weakness.

Biological basis

Etiology and pathogenesis

As the term “diabetic amyotrophy” indicates, this disorder was classically linked to diabetes mellitus. In an epidemiology study, diabetes was found to be the most common risk factor for radiculoplexus neuropathy, followed by autoimmune disorder, stroke, and higher BMI (41). The most common associated autoimmune disorders in this study were autoimmune thyroiditis, inflammatory bowel disease, and type 1 diabetes mellitus, which could be metabolic risk factors as well. The study conclusion was that “altered metabolism and immune dysfunction seem to be the most influential factors in the development of lumbosacral radiculoplexus neuropathy.” Other risk factors and triggers may also be at play. For example, a case of radiculoplexus neuropathy after COVID-19 infection (42) and another one after mRNA vaccine (08) have been reported, suggesting an immune-mediated process.

Although metabolic derangement is considered the main mechanism of the generalized diabetic neuropathies, a vascular mechanism is suggested to explain the acute onset and focal nature of the focal and multifocal diabetic neuropathies.

Clinical differences between diabetic polyneuropathy and diabetic amyotrophy suggest that they have different mechanisms and pathogenesis.

Table 1. Clinical Differences Between Diabetic Amyotrophy and Diabetic Polyneuropathy

A vascular process is suggested to explain the clinical picture of subacute, painful, asymmetric neuropathy with weight loss and the multifocal axonal neuropathy evident from the EMG. Raff and colleagues documented multiple areas of nerve ischemia within the lumbosacral plexus in a postmortem study (43). They also reported perivascular inflammation, but there was no pathological indication of vasculitis. Barohn and colleagues confirmed the ischemic damage by studying 10 nerve biopsies showing multifocal nerve loss (03); there was no comment on the presence of inflammation. Said reported the result of 10 cutaneous femoral nerve biopsies from patients with diabetic amyotrophy; ischemic lesions were found in three patients and vasculitis in two (45).

In another series, vasculitis was found in three of 14 femoral cutaneous nerve biopsies from patients with diabetic amyotrophy (34). Dyck studied 33 patients with diabetic amyotrophy and found multifocal fiber loss in 19 biopsies, perineurial thickening in 24, and neovasacularization in 21 (13).

These studies suggested that the mechanism of injury was ischemic vasculopathy.

What is the cause of ischemic vasculopathy in diabetic amyotrophy? Chronic hyperglycemia and the resulting advanced tissue glycation end-products alone do not fully explain the pathogenesis because most patients with diabetic polyneuropathy do not develop this syndrome despite worse underlying glycemic regulation than that of diabetic amyotrophy patients. Population-based studies show that diabetic amyotrophy groups have better glycemic control, use less insulin, are more likely to have type 2 diabetes mellitus, have been diabetic for a shorter period of time, and have less vascular complications than the diabetic polyneuropathy group (12).

Alternatively, the primary event may be an inflammatory (probably immune) attack against the wall of the endoneurial blood vessels.

A lesson from non-diabetic lumbosacral radiculoplexus neuropathy. Comparison between diabetic and non-diabetic lumbosacral radiculoplexus neuropathy reveals striking similarities, suggesting that diabetes mellitus is a risk factor rather than a cause (11; 14).

The similarities also support the vascular and inflammatory nature of the disease as opposed to the metabolic factors.

Table 2. Comparison Between Diabetic and Non-Diabetic Lumbosacral Radiculoplexus Neuropathy

Evidence of inflammation in pathological studies. In a study that included 12 patients with diabetic amyotrophy, the sural nerve demonstrated a mononuclear cell infiltrate in four patients and a perivascular infiltrate of activated T-cells expressing both interleukin-2 and major histocompatibility complex class II antigens in six patients (33).

However, there was no evidence of B-cells or polymorphonuclear cells. The majority of nerves from patients also showed staining for tumor necrosis factor, interleukin-6, and interleukin-1beta. Furthermore, C3d and C5b-9 complement proteins were found within the endoneurial and epineurial blood vessel walls in all patients (33). In a biopsy study, Kelkar and colleagues described small-vessel neutrophilic vasculitis with polymorphonuclear infiltration in postcapillary venules along with IgM deposition and complement deposition in endoneurium and in affected vessel walls (29).

These studies support that proximal diabetic neuropathy has an immune-mediated and inflammatory basis and suggest that polymorphonuclear vasculitis with immune complex and complement deposition may be the primary event in the acute phase of proximal diabetic neuropathy. Up-regulated specific inflammatory mediators, such as intercellular adhesion molecule-1 in vessels and tumor necrosis factor-alpha in Schwann cells and some macrophages, may be involved in an immune-mediated inflammatory process that is shared by both the diabetic and non-diabetic variants (28).

Dyck studied 33 sensory nerve biopsies from diabetic amyotrophy patients and compared them to biopsies from patients with diabetic polyneuropathy and to biopsies from healthy controls.

Table 3. Pathological Analysis of Nerve Biopsies From Patients With Diabetic Amyotrophy, Diabetic Polyneuropathy, and Healthy Controls

He found perivascular inflammation in all nerves sampled, mural invasion (microvasculitis) in half, and hemosiderin-laden macrophages in half. The latter finding indicates previous bleeding (13).

The notion of ischemic injury from microvasculitis was further supported by Said’s findings of necrotizing vasculitis of perineurial and endoneurial blood vessels in six of 22 and perivascular inflammation in 21 of 22 nerve biopsies from diabetic amyotrophy patients. He also found endoneurial red blood cells in 11, endoneurial hemorrhage in five, and ferric deposits (a sign of previous bleeding) in seven of these 22 biopsies. In contrast, nerve biopsies from 30 patients with severe distal diabetic polyneuropathy demonstrated only mild epineurial mononuclear cell infiltration in one biopsy and endoneurial red blood cells in one other (46).

Interestingly, muscle biopsies from Said’s 22 patients displayed inflammatory infiltrate in three cases and vasculitis in one case. The lack of fibrinoid necrosis of the vessel wall in all the mentioned series as opposed to the systemic vasculitis may be due to the very small size of the involved vessels in the former.

Diabetic amyotrophy, thus, may be a form of nonsystemic vasculitis isolated to peripheral nerves. However, it remains elusive why small vessels are more susceptible in this condition.

The offensive antigen. The cause of inflammation in diabetic amyotrophy remains intriguing. An immune reaction against the altered walls of small endoneurial blood vessels is possible. The cause of such alteration may be diverse. The association of inflammation with atherosclerosis is increasingly recognized.

After initiation of an atherogenic diet, patches of arterial endothelial cells express selective adhesion molecules on their surface that bind to various classes of leukocytes. The types of vascular cell adhesion molecules (CAM-1) that bind lymphocytes and monocytes in experimental atheroma have been shown in experimental rats (37; 32; 47).

Another possibility is that hyperglycemia modifies macromolecules by forming advanced glycation end-products, and that is shown to augment the production of cytokines (26; 36; 35); but again, patients with diabetic amyotrophy have less vascular complications and better diabetes control than patients with diabetic polyneuropathy, yet microvasculitis is not a feature of the latter. Furthermore, the presence of the same pathology in a non-diabetic suggests an alternative mechanism rather than hyperglycemia and atherosclerosis.

Epidemiology

The prevalence of diabetes mellitus in the United States is 6.5% and rising. The epidemiology of diabetic neuropathies has been studied widely with variable results. The wide range is partly due to different criteria used for diagnosis and partly due to the failure of recognition of different types of diabetic neuropathies. Hence, the reported prevalence of diabetic neuropathy has varied from 10% to 100% (12). The two major types of diabetic neuropathies are generalized neuropathies and focal and multifocal neuropathies (Table 4). In the population-based Rochester Diabetic Neuropathy Study, 60.8% of the subjects had some form of diabetic neuropathy, with sensorimotor polyneuropathy being most common (overall prevalence 47.6%). The prevalence of asymmetrical proximal neuropathy was 1% each in the type 1 diabetes and type 2 diabetes groups (12).

Table 4. Classification of Diabetic Neuropathies

Differential diagnosis

Chronic inflammatory demyelinating polyradiculoneuropathy in diabetes mellitus. The diagnosis of CIDP in diabetic patients represents a dilemma because of many shared features between diabetic polyneuropathy and CIDP, making the application of the CIDP diagnostic criteria less reliable.

Some diabetics present with progressive, painless, symmetrical, proximal, and distal weakness of all extremities evolving over weeks to months and associated with diffuse areflexia or hyporeflexia. Physical examination also shows stocking and glove sensory impairment, which can be explained by diabetic polyneuropathy. CSF protein is frequently elevated, and electrodiagnostic studies meet the American Academy of Neurology research criteria of CIDP, except for some axonal features that are attributed to diabetic polyneuropathy (01). Although the magnitude of response to treatment is not as robust as that seen in idiopathic-CIDP, these cases share more features with CIDP than diabetic polyneuropathy.

Ongoing issues about CIDP development in diabetic patients.

Sharma and colleagues studied the frequency of CIDP in patients seen in an EMG laboratory over a 14-month period: 1127 patients were seen; 189 of them were diabetic with various neuromuscular disorders, of whom 32 (16.9%) met the American Academy of Neurology criteria for CIDP. Among the remaining 938 patients without diabetes mellitus, 17 (1.8%) had idiopathic CIDP (49). The odds of occurrence of diabetes mellitus and CIDP were 11 times higher among diabetics than non-diabetics.

Does diabetes mellitus CIDP differ from idiopathic CIDP in response to therapy? Sharma and colleagues treated 26 diabetics who met the American Academy of Neurology criteria for CIDP with intravenous immunoglobulins; 21 of them had significant improvement in their neurologic deficit at the end of 4 weeks of therapy (48).

Haq compared 10 nondiabetic patients who met the American Academy of Neurology criteria for CIDP to nine patients with diabetes mellitus and CIDP. The clinical, neurophysiological, and histological features of both groups were similar. Six patients with diabetes mellitus-CIDP and eight patients with idiopathic-CIDP were treated with prednisone with or without intravenous immunoglobulins, plasmapheresis, or azathioprine. All patients responded favorably to the treatment in a similar fashion (24).

Gorson compared the clinical, electrophysiological, and pathological features of 14 patients with diabetes mellitus-CIDP with 60 patients with CIDP. The two groups were similar, except that the sural nerve action potentials were more often absent in the former group, and axonal loss in the nerve biopsy was more frequent (22).

The response rate to treatment with different modalities was also similar (Table 5), except that the magnitude of functional recovery was greater in idiopathic-CIDP as measured by mean strength score, medical research council score, and Rankin disability score.

Table 5. Response of DM-CIDP and I-CIDP to Treatments

Caution is recommended in diagnosing CIDP in diabetic patients. Mandatory American Academy of Neurology research criteria for the diagnosis of CIDP is not ideal or practical (Table 6).

Table 6. American Academy of Neurology Research Criteria for the Diagnosis of CIDP

Conclusions about CIDP in diabetes mellitus.

Therefore, better and more specific criteria are needed and better consideration should be given to the following features that are more consistent with CIDP than diabetic polyneuropathy:

A therapeutic trial with steroids, intravenous immunoglobulin, or plasmapheresis may be necessary to distinguish between the two in more diagnostically challenging cases.

Diagnostic workup

The history and physical examination, supplemented by laboratory and neurophysiological findings, usually lay the groundwork for a diagnosis of diabetic amyotrophy.

Laboratory findings. In any case of lumbosacral plexopathy, laboratory studies should include a complete blood count, chemistries, coagulation profile, erythrocyte sedimentation rate, hemoglobin A1c, and fasting blood sugar; if necessary, a 2-hour glucose tolerance test is advocated by many. In diabetic amyotrophy, cerebrospinal fluid protein is frequently elevated (44 to 214 mg/dl), suggesting extension of the disease process to the nerve roots. The erythrocyte sedimentation rate may be raised in patients with proximal diabetic neuropathy (05; 54). Appropriate imaging of the lumbosacral spine, lumbosacral plexus, and pelvis is indicated to exclude compressive etiologies. However, results must be interpreted with caution because incidental abnormalities are common, in particular lumbosacral spondylosis.

MRI has been utilized in many neuromuscular disorders, including lumbosacral radiculoplexus neuropathy. Different findings have been reported, including T2 hyperintensity and enhancement of the lumbosacral roots, plexus, and nerves. T2 hyperintensity of the affected muscles, as can be seen with denervation, is also reported (19). In addition, compared to controls, cross-sectional area of the lumbosacral roots and the femoral nerves is significantly enlarged in diabetic amyotrophy (25). These changes are not specific to lumbosacral radiculoplexus neuropathy, including diabetic amyotrophy, but some or all of these changes can be seen with other disorders (demyelinating polyneuropathies, radiculitis secondary to viral illness, sarcoid, etc.). In the authors’ opinion, the main benefit of the MRI is to exclude certain possibilities, such as severe degenerative spinal disease, benign or malignant compressive lesions, and others, but MRI findings are not specific enough to be helpful in the diagnosis of diabetic amyotrophy.

Neurophysiologic findings. Nerve conduction studies usually reveal marked reduction of compound muscle action potential amplitudes of affected motor nerves in a very asymmetrical fashion. Sensory nerve action potentials show low amplitude or, more frequently, absent responses in the affected regions (13). Mild slowing of nerve conduction velocities is seen (52). These neurophysiologic findings indicate multifocal axonal degeneration rather than demyelination. Needle examination shows spontaneous activity, reduced recruitment of motor unit potentials, and long duration and high amplitude motor unit potentials in muscles supplied by multiple nerve roots and different peripheral nerves, further supporting the occurrence of axonal pathology (44). Paraspinal muscles are usually affected. The EMG abnormalities tend to be much more prevalent and widespread than the clinical picture suggests. In patients with underlying diabetic sensory polyneuropathy, in addition to the above findings, sural sensory nerve action potentials are usually absent, and peroneal and tibial compound motor action potential amplitudes are reduced.

Histologic findings. Biopsies are rarely indicated. Early in the disease course, epineurial and perivascular inflammation around the small vessels may be caused by infiltration by mononuclear cells, with or without polymorphonuclear cells (13). Endoneurial and subperineurial IgM deposition has been reported (29). Activated complement (C5b-9) deposition in the endothelium of small vessels was demonstrated by Krendel (33). A decrease in the number of myelinated and unmyelinated axons may be observed. Differential fascicular loss of axons is also characteristic.

Management

Treatment is centered initially around pain control. Agents that are commonly used for pain control include non-steroidal anti-inflammatory agents (eg, ibuprofen, naproxen), tricyclic antidepressants (eg, amitriptyline, imipramine), antiepileptics (eg, phenytoin, carbamazepine, gabapentin), and opioids (eg, codeine phosphate, morphine elixir). Mexiletine, an antiarrhythmic, has been used effectively in treating neuropathic pain in several clinical trials. Other allied pain treatments include transcutaneous electronic nerve stimulation, hypnosis, relaxation training, biofeedback, and acupuncture.

Good glycemic control is of paramount importance. Oral hypoglycemic agents may need to be changed to insulin therapy. Meal planning and exercise may be beneficial.

Neurologic recovery is slow. Walking regularly or using elastic stockings may help with leg pain. The physical therapist can assist in improving functional mobility (eg, transfers, ambulation). Assistive devices may be employed when necessary. An exercise and range-of-motion program supervised by the physical therapist is also helpful in maintaining and improving function and avoiding contractures. The occupational therapist can recommend appropriate adaptive equipment (eg, reacher, elevated toilet seat, tub bench), depending on the amount of weakness, so that the patient can be independent in activities of daily living and perform self-care tasks in a seated position.

The pathological findings discussed prompted several investigators to treat diabetic amyotrophy as an inflammatory syndrome. In addition, a favorable response of the non-diabetic lumbosacral radiculoplexus neuropathy to intravenous methylprednisolone in one trial suggests that its diabetic counterpart may respond similarly (14; 15; 17). In this clinical trial, intravenous methylprednisolone was administered to 11 patients with non-diabetic lumbosacral radiculoplexus neuropathy over 8 to 16 weeks. All treated patients had marked improvement in pain and weakness, and pain resolved completely in many. The median neuropathy impairment score (NIS) and neuropathy symptoms and change (NSC) statistically improved after treatment.

Before treatment, half of the patients were wheelchair-bound, and all patients but one used walking aids. After treatment, only one was in a wheelchair, and six walked independently.

The lack of a control arm makes conclusions at best preliminary; the following observations suggest benefit from treatment:

Therapeutic trials in diabetic amyotrophy. Said treated two patients with diabetic amyotrophy and biopsy-proven vasculitis with prednisone; both had a favorable outcome. Krendel treated 15 patients: seven with intravenous immunoglobulin and prednisone, five with intravenous immunoglobulin alone, two with prednisone and cyclophosphamide, and one with prednisone alone; all improved, five markedly (33). Kilfoyle reviewed the records of nine patients with diabetic amyotrophy treated with steroids and compared them with untreated patients; of the treated group, 75% improved within 1 month, whereas only 59% of the untreated patients eventually improved and none before 6 months. In seven of 10 episodes, pain resolved within 3 months of onset, in contrast with natural history in which pain typically persists for more than 6 months (31). Pascoe treated 12 diabetic amyotrophy patients with prednisone, intravenous immunoglobulin, or plasmapheresis and compared the outcome with 29 untreated patients. The treated group improved faster and to a greater extent (39). Rapid recovery of diabetic lumbosacral radiculoplexus neuropathy after administration of intravenous immunoglobulin has been reported (18).

In a prospective, randomized, double-blind, multicenter controlled trial by Dyck and associates, two thirds of patients received intravenous methylprednisolone (initially 1 gram several times per week, with decreasing doses and frequency over 12 weeks), and the other one third received placebo in a dose of 1 gram three times a week followed by less frequent doses over 12 weeks. The patients were followed for 2 years. Preliminary results from the first year of this trial reveal a significant improvement of mean NIS in both groups. The primary outcome measure (improvement of NIS in lower limbs by 4 points) showed no difference between groups. However, a secondary neuropathic pain symptom outcome measure showed greater improvement that lasted 6 months in the methylprednisolone group (16).

Anecdotal reports of dramatic improvement in neurologic function (38) have resulted in the initiation of a monocenter clinical trial of intravenous immunoglobulin; the study had three arms: high dose, low dose, and control groups. The study was put on hold because of a lack of intravenous immunoglobulin supply.