Clinical manifestations

Presentation and course

Symptoms, with some exceptions, are indistinguishable from prototypic toxic neuropathies including initial distal paresthesias and sensory loss in a stocking distribution affecting large or small fiber modalities or both, and lesser and later onset distal motor impairment. Some cases never progress to demonstrable motor changes or have purely or predominantly sensory manifestations. Tendon reflexes are distally depressed or absent. In some cases, the neuropathic manifestations improve but chronically persist after stopping the agent, clouding the association. Also included are presentations similar or identical to Guillain-Barré syndrome, such as with tacrolimus, zimeldine, and lithium overdose as well as a number of other agents discussed in other sections. Some agents may preferentially target dorsal root ganglia or autonomic ganglia neurons, but manifestations usually coalesce into the typical distal predominant pattern. Thus, sensory symptoms, which may be small or large diameter fiber-mediated or both, can occur. Associations are easiest with acute or subacute symptoms beginning soon after exposure to a new medication or dosage increase and subsequent improvement after cessation; however, some progression for several weeks may occur due to “coasting,” when persistent toxicity is present in some cases due to slow toxin clearance. However, in some cases, recovery is delayed even without persistent elevated levels of toxin, such as with pyridoxine neurotoxicity (14). Of interest are cases of possible occult toxic idiopathic neuropathy due to chronic medications. The example of statin drugs is a case in point. In general, although unusual, these entities are important to uncover because, unlike many neuropathy etiologies, removal of a toxic agent may lead to significant recovery (188; 144).

Prognosis and complications

Prognosis depends on the severity of involvement and the degree of axonal injury.

Biological basis

Etiology and pathogenesis

No broad etiology or pathogenic mechanism has been suggested, but isolated cases may be part of an acute hypersensitivity reaction (71). Most of the potentially pathogenic mechanisms in this section are speculative.

Specific agents.

Allopurinol. Allopurinol has been used for the treatment of gout since its approval in 1966. Allopurinol inhibits the enzyme xanthine oxidase, which blocks the metabolism of hypoxanthine and xanthine (oxypurines) to uric acid, interfering with the catabolism of purines. A number of cases of neuropathy have been associated with this agent in reports with various strengths of association (71; 194; 10). The initial description included a hypersensitivity reaction with later drug rechallenge with a subsequent repeat allergic reaction, which included symptoms and signs of peripheral neuropathy. The patient also received colchicine after symptom onset with unclear timing related to neuropathy onset. Symptoms improved but persisted after cessation (71). Fewer than 10 cases have been noted in the literature with at least two cases having complicating issues (uremia). One included some electrophysiologically and pathologically demyelinating features (10). These features were not previously noted, and regression occurred after drug cessation. Most cases occur after several or many years of therapy. No predisposition or ancillary factors are currently known. No experimental evidence supports the association, making this a possible, but not a definite, rare idiosyncratic association. In fact, the agent has been used to preserve nerve and vascular function in streptozotocin-induced diabetic neuropathy in rats (81). Inhibition of xanthine oxidase produced reactive oxygen species is the suspected beneficial effect. Blood flow declines caused by the experimental diabetes were also partially corrected.

Almitrine. Almitrine bismesylate is not FDA approved for use in the United States, but it is available in many other countries for treatment of chronic obstructive pulmonary disease and some vascular disorders including stroke prophylaxis. Almitrine acts as a peripheral chemoreceptor agonist. The component is noteworthy because it appears to commonly induce a predominantly sensory neuropathy. In one small placebo-controlled study of seven controls and five treated chronic obstructive pulmonary disease patients, three of five treated patients and none of the controls developed significant neuropathy (06). Bouche and colleagues reported 46 cases of almitrine-associated neuropathy in one series (25). The range of onset described is nine to 25 months after medication onset. Sensory symptoms restricted to distal legs involving all modalities are typical. Electrophysiology and nerve biopsy findings were consistent with sensory axonopathy (69; 141). Improvement is described in most cases and is usually complete after a year (25). Numerous other reports have been made that support the high incidence of neurotoxicity with this agent (109; 21; 141; 05; 195; 06; 70). Small placebo control studies have reported variable, but significant, percentages of patients stopping trials because of neuropathic symptoms. Gherardi and colleagues have reported nerve biopsy and ultrastructural studies of eight cases. The primary finding is axonal loss of large myelinated fibers with signs of regeneration in one delayed biopsy. In addition, signs including segmental demyelination on teased fiber preparations suggested a demyelinating component in a variable percentage of fibers. No animal studies are available for consideration. No predisposing conditions are known. Moreover, the severity of hypoxemia from the pulmonary disease does not appear to correlate with the appearance of neuropathy or subsequent improvement after cessation. One series did describe a shorter latency to neuropathy onset in chronic obstructive pulmonary disease versus vascular patients, but most were receiving higher doses. However, some additional neuropathological signs seen in isolated cases (microangiopathy) could be secondary to chronic hypoxemia. Weight loss is commonly associated with the appearance of neuropathy. Some studies using a lower dosage (< 100 mg/day) have shown no significant neuropathy, including electrophysiologic changes (189). Series have reported no dropout due to neuropathy of patients who used similar dosages, but these series are without specific methods to detect sensory loss.

Amitriptyline. Amitriptyline is a useful drug in the treatment of painful conditions including peripheral neuropathy, especially conditions with marked small fiber mediated pain involvement. However, several reports have associated this agent and, even more rarely, other tricyclic antidepressants including imipramine with inducing peripheral neuropathy (105; 106). Many of the cases described are in the setting of overdose with other complications including rhabdomyolysis and cholinergic effects on the CNS and periphery; however, a small number have described suspected neuropathy on conventional amitriptyline dosages with improvement after cessation (83; 133; 197). There is limited experimental evidence of ultrastructural lesions in cultured neurons and astrocytes, but this relation to human toxicity is speculative at best and has not altered use in patients with neuropathy.

Chloroquine. Chloroquine is an agent used to treat malaria prophylaxis and some autoimmune conditions. The primary neuromuscular complication is a vacuolar myopathy, which can be fulminant (160); however, rare cases of neuropathy with demyelinating features and axonal loss have been described (186; 167). Onset is typically 1 to 2 years after starting medication and involves both sensory and motor fibers. Severity is not typically marked. Schwann cells have shown dense and laminar cytoplasmic inclusions similar to those seen with amiodarone and perhexiline, notable other causes of demyelinating toxic neuropathy. Sural biopsies have shown axonal loss and segmental demyelination and remyelination (173). Electrodiagnostic studies have also suggested a neurogenic component superimposed on the predominant myopathy. Some have noted the pattern could mimic a polyradiculopathy. This effect has been reproduced in rats. Neuromuscular toxicity from short-term use of hydroxychloroquine or this agent against COVID19 is speculative at present. A review using the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS) compared adverse reports comparing chloroquine and hydroxychloroquine spurred by increased hydroxychloroquine usage for COVID-19 (138). Over 6.6 million reports were assessed. Relative odds ratio of 3.3 was found for chloroquine but no increase (1.0) for hydroxychloroquine.

Cyclosporin. Cyclosporin A has been used as an immunosuppressive agent in numerous conditions including organ transplantation and even some cases of immune mediated neuropathy. Limited information has associated cyclosporin A with otherwise unexplained peripheral neuropathy. The evidence rates this agent at best as a possible, but not probable, causative agent at present (20; 34).

Dichloroacetate. Dichloroacetate is used experimentally to treat chronic lactic acidemia from mitochondrial diseases. Peripheral neuropathy is common with chronic dichloroacetate treatment. Clinical and electrophysiologic signs of sensorimotor neuropathy are found (165; 09; 91). The neuropathy is significant but can be reversible over months if the drug is stopped. Neuropathy also develops in younger children with lactic acidosis but was said to be tolerated by most in one trial of 36 impaired children (166). The neurotoxic mechanism is not fully known, but the agent causes reversible demyelination in cultured rat Schwann cells and dorsal root ganglia neurons exposed to dichloroacetate for up to 12 days (64). The heme precursor delta-aminolevulinate is implicated in neurologic complications associated with porphyria and tyrosinemia type I. The compound is elevated in the urine of animals and humans on dichloroacetate and appears to damage Schwann cells in part by reducing the levels of myelin-associated lipids and proteins, including myelin protein zero and peripheral myelin protein 22 (63). It is currently unclear but suspected that patients with mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes (MELAS) may be at increased risk of developing the toxic neuropathy. The drug is also under investigation in the treatment of glioblastoma multiforme and other oncology effects. Dichloroacetate may help minimize treatment resistance mediated by changes in mitochondria that reduce cancer cell apoptosis in part by a switch from mitochondrial oxidative phosphorylation to cytoplasmic glycolysis. However, studies in humans have been limited by dose-dependent peripheral neuropathy (123). Dichloroacetate has the interesting property of inhibiting its own metabolism on repeat dosing, resulting in alteration of its pharmacokinetics. This dichloroacetate effect on its own metabolism complicates the selection of an effective dose with minimal side effects (85).

Glutethimide. Glutethimide is a sedative-hypnotic agent originally used as an ethanol substitute, but it proved to be highly addictive in its own right. The compound was reclassified as schedule II and withdrawn from general availability in 1991. Chemically, the compound is structurally similar to thalidomide, an agent that more commonly induces sensory neuropathy. Rarely, neuropathy has been associated with chronic glutethimide use at high doses (134; 76). Cerebellar ataxia is also described and may be marked and persistent. Manifestations are predominantly but not exclusively sensory and are supported by limited electrodiagnostic data. Improvement or resolution over ensuing months is described. No experimental evidence or nerve biopsy data are available for correlation.

Ixabepilone. The epothilones are a relatively new class of chemotherapeutic agent with currently low tumor resistance. The class includes the natural agents, epothilone B (patupilone) and epothilone D, which are not yet FDA approved as well as the semisynthetic analog, ixabepilone (16; 170); ixabepilone was FDA approved in October 2007. This agent is also discussed in the separate chemotherapy neuropathy section. Breast cancer is the primary indication, but phase 2 studies are completed or are ongoing for a variety of cancer types. The group binds to tubulin similarly to certain other chemotherapeutic agents, such as vinca alkaloids and taxanes, but at differing binding sites. Similar to taxanes, the drugs promote dysfunctional stabilization of microtubules but with a differing mechanism, in contrast to microtubular destabilizing agents such as vincristine, colchicine, and podophyllotoxin (49). Peripheral neuropathy from ixabepilone is a major dose-limiting side effect. Mild-to-moderate (grade 1 to 2), predominately sensory neuropathy that improves or resolves is most common, but more severe grades (grade 3) occur rarely in monotherapy and at rates of 10% to 15% with combination therapy or in patients previously exposed to taxanes or capecitabine; grade 4 neuropathy appears to be very rare (54; 152; 176). Dose reduction may be adequate in many, but treatment discontinuation occurs as well; more dispersed treatment protocols may have lesser toxicity (175). Twenty-one percent of patients treated with ixabepilone plus capecitabine discontinued treatment because of sensory neuropathy in one large, phase-3 trial (175). Severity increases with cumulative dosing, especially after an average of four treatment cycles. The overall neuropathy incidence varies depending on dose and coincident treatments but is as high as 67%. The reported reversibility of sensory neuropathy is surprising considering the experience with other microtubule targeting agents such as vincristine and taxanes. One patient is reported who developed significant weakness associated with neuropathy after one treatment cycle (24). Review of all phase 2 and phase 3 clinical trials found a 1% incidence of severe neuropathy in patients previously untreated and up to 24% of breast cancer patients treated with other agents (180). Carefully monitoring for neuropathy and timely dose adjustment or treatment discontinuation is advocated depending on the neuropathy severity (170). Neuropathy is increasingly recognized as a dose-limiting side effect and 20% dose reduction is one proposed strategy (181). This entity is also discussed in the section on chemotherapy-induced neuropathy.

Leflunomide and teriflunomide. The immunosuppressive prodrug leflunomide was FDA approved in late 1998 as a disease-modifying rheumatoid arthritis treatment. It was subsequently recognized that an axonal, sometimes painful, sensorimotor polyneuropathy is associated with leflunomide (30; 23; 122). Eighty cases reported to the FDA were uncovered and described (23). After this report, additional series have been reported identifying numerous additional probable cases (15; 116; 93). Bharadwaj and Haroon describe 150 prospectively tracked rheumatoid arthritis patients in India. Fifty received leflunomide either as monotherapy or in combination with other drugs. Five developed new neuropathy (10%) in contrast to two of 100 receiving other treatments (2%). Nerve biopsy in three demonstrated epineural perivascular inflammation around small and medium-sized arterioles patchily affecting large and small myelinated nerve fibers suggesting a predominant axonopathy with features of vasculitis. All showed clinical improvement and were said to become asymptomatic within 3 months, but residual nerve conduction abnormalities remained (15). Kopp and colleagues describe a case and suggest a potential interaction between 5-FU and leflunomide and include the possible mechanism (96). Onset is usually after 3 to 6 months of drug use, although symptoms may appear sooner. Another study compared 16 rheumatoid arthritis patients treated with leflunomide with 16 others receiving alternative disease-modifying therapies. Neuropathy symptoms scores increased in 54% of the leflunomide group compared with 8% of the others; however, electrophysiology did not correlate with clinical symptoms (149). Stopping therapy within 30 days of symptom onset gives a better chance of improvement, though recovery is typically slow. Sural nerve biopsies have shown nonspecific axonal loss in most, but signs of perivascular inflammation have been described. Primary rheumatoid arthritis is an independent neuropathy risk factor often associated with vasculitis, but leflunomide reports have not generally described this type of pattern. Neuropathy incidence is higher than with rheumatoid arthritis alone or with other rheumatoid arthritis medications. One retrospective analysis found increased associated neuropathy risk with increasing age, diabetes, and the use of other potentially neurotoxic medication (117). The mechanism of neurotoxicity is not known; neuropathy cases were not detected in clinical trials. The drug remains an effective treatment and efficacy appears to be similar to methotrexate and better than sulfasalazine. However, withdrawal rates are higher than methotrexate because of toxicity; peripheral neuropathy is one of several forms of toxicity (03). Comparison of 94 rheumatoid arthritis patients treated with either leflunomide or other disease modifying agents found significant differences in quantitative cold but not vibration perception measures; leflunomide-treated patients were roughly twice as likely to have increased cold perception measures (95).

Teriflunomide, an active metabolite of leflunomide, is approved in the United States to treat multiple sclerosis. Paresthesia and peripheral neuropathy based on nerve conduction study data are associated with this agent as well. Clinical trial data suggest an incidence of 1% to 2% depending on dose, but no aftermarket reports of significant neuropathy cases are known. The TOWER extension study of 751 relapsing remitting multiple sclerosis patients treated with teriflunomide for a median of 4.25 years found four patients discontinued treatment because of clinical neuropathy (124). Three cases of small fiber neuropathy attributed to teriflunomide treatment of relapsing remitting multiple sclerosis have been reported (59). Symptom onset began 6 months, 1 year, and 4 years after treatment initiation. All failed other therapies. One had metabolic syndrome. All had a limited search for other neuropathy causes, and all had reduced epidermal nerve fiber density on skin biopsy.

Lipid lowering agents. The statin-class of cholesterol medications acts by inhibiting the rate-limiting step in cholesterol synthesis, hydroxymethylglutaryl coenzyme A. The predominant neuromuscular complication with these agents is a toxic myopathy referred to as cholesterol-lowering agent myopathy, which is well appreciated by physicians and patients. An increasingly recognized acute necrotizing myositis with rhabdomyolysis associated with antibodies against the HMG CoA enzyme can develop. However, a number of cases of peripheral neuropathy temporally associated with conventional doses of simvastatin and other agents in the class have been reported (84; 01; 143; 198; 87; 108). Partial or complete recovery after drug cessation is described. One report described sural biopsy data demonstrating small and large fiber axonal loss (143). Several cases have serial electrophysiological studies showing sensorimotor axonal neuropathy with variable levels of subsequent improvement. No experimental model to support the effect is known. Symptom onset has been described within days to as long as several years after onset. One case described neuropathy onset after several years of treatment with lovastatin; when treatment stopped, the condition improved (198). Rechallenge with pravastatin, simvastatin, and later atorvastatin each caused a subacute recurrence of burning dysesthesias that improved with cessation. Similar rapid worsening with rechallenge has been noted in other reports. One speculative mechanism proposed is that inhibition of mitochondrial hydroxymethylglutaryl coenzyme A reductase causes a subsequent decrease of ubiquinone synthesis, which potentially may disturb neuronal energy utilization (183).

Thus, only the temporal association with the neuropathy development and subsequent improvement was available to support a causative link until a case control study reported by Gaist and colleagues (67). Gaist and colleagues suspected a possible link between these agents and cases of idiopathic neuropathy, despite an earlier negative United Kingdom study (68). They then conducted a much larger population-based study in one Danish county (465,000 inhabitants) and cross referenced a prescription registry to a national patient diagnosis registry from 1994 to 1998, when statin use in Denmark increased from 11,547 to 50,318 nationwide. Gaist and colleagues identified 1084 registered patients with a diagnosis of polyneuropathy. They excluded 492 with onset prior to 1994 or concurrent cause of neuropathy (diabetes, renal failure, monoclonal gammopathy, etc.). Only cases with clinical signs of distal, symmetric neuropathy and an adequate workup including electrodiagnostic studies were analyzed and categorized as definite, probable, or possible idiopathic neuropathy. Twenty-five controls were randomly chosen per index case. Thirty-five definite, 54 probable, and 77 possible neuropathy cases from the registry (166 total) were found. Nine had been exposed to statins including simvastatin, pravastatin, lovastatin, and fluvastatin. Odds ratios were calculated as 4.6% overall with current users of statins compared to controls and 16.1% with definite neuropathy cases compared to controls. The researchers also calculated an interesting number needed to harm measure and found, based on their odds ratios, one excess case of idiopathic peripheral neuropathy for every 2200 person-years of statin use. Considered in this way, neuropathy was suggested as a more important public health concern than myopathy in patients taking statins. However, potential pitfalls complicate the study, such as whether all symptomatic neuropathy causes were in fact excluded. Examples of complicating disorders include conditions associated with statin use, such as occult diabetes, glucose intolerance, or metabolic syndrome. (56); however, not all series found a clear association with statins and idiopathic neuropathy (08). Despite the rarity of the association, the large number of patients who take these medications makes the association potentially clinically relevant. Further uncertainty was raised in 2007 by the announcement at the meeting of the American Diabetes Association of the large 8-year long Australian Fremantle study of nearly 1300 diabetic patients that demonstrated significantly decreased risk of developing neuropathy in patients treated with statins or fibrates compared to untreated patients. The reduction was 35% and 48%, respectively (51). Experimental evidence suggests that the statin rosuvastatin improves a mouse model of diabetic neuropathy through improved microcirculation independent of cholesterol lowering effects (80). The combination of studies and evidence challenges the importance of the earlier Gaist results; statin neuropathy likely occurs but may be much less frequent than recently thought and appears to be neuroprotective in some settings. A large Danish study of over 135,000 type 2 diabetes patients on new or chronic statin treatment found no significant risk of statin induced neuropathy (98). In contrast, a study of a large cohort from Korean health insurance data of type II diabetes patients found a 1.22 neuropathy odds ratio associated with statin use and reduced odds ratio for metformin-treated patients (125). A systematic review of nearly 5000 studies including two metaanalyses examined statin use and neuropathy (184). When pooled, no significant association between statin use and neuropathy was found, though they acknowledge high heterogeneity between the metaanalyses. Hyperlipidemia, notably triglycerides, is increasingly suspected as an important neuropathy factor that complicates the statin link to neuropathy (82; 139). A later Taiwanese study of 2448 patients, of which 21.4% had baseline diabetic neuropathy, has been reported. Multivariate analysis of risk factors found that hyperlipidemia, low-density lipoprotein, and total cholesterol, lipid lowering therapy, statin, or fibrate use were not associated with neuropathy prevalence (38). A systematic literature review of available trials and reports found the evidence for statin induced neuropathy to be minimal (50).

One possible case following initiation of simvastatin rapidly developed into neuropathy mimicking Guillain-Barré syndrome; a pravastatin challenge six months earlier had led to milder symptoms. The combination suggested a possible hypersensitivity reaction (146). In contrast, lovastatin attenuated nerve injury in an experimental model of experimental allergic neuritis. The effect was blocked by mevalonate (156).

There is no supportive experimental model of the potentially toxic effects, but alteration of membrane function though inhibition of cholesterol synthesis, reduction of axon transport, and inhibition of mitochondrial function have been suggested as possible factors. Oddly, some animal models support statin-induced benefit in neuropathic pain despite discussed concerns of increased neuropathic pain in humans (140). A streptozotocin (STZ)-induced diabetic rat model found that atorvastatin significantly reduced the hyperalgesia in diabetic rats, suggesting it is mediated by the nitric oxide pathway (02).

Interference with selenoprotein synthesis, a well-established pathway also implicated in some hereditary muscle disorders, has been postulated to be causative but probably relates better to myotoxicity. Myopathy from severe selenium deficiency shares some features with statin-induced myopathy (128).

Lithium. Lithium is rarely associated with neuropathy. Isolated reports describe the onset of typical toxic neuropathy manifestations after prolonged exposure on a convention dose (177; 127); however, most reports are after acute intoxication or overdose (28; 179; 136; 41; 182; 88; 121; 35). Excessive levels can occur due, in part, to the narrow therapeutic range of the drug; moreover, neuropathic findings may be underrecognized. Some reported cases are complicated by more generalized toxicity including cerebral impairment with the neuropathy becoming evident only with subsequent recovery. A chronically treated 72-year-old bipolar patient was found to have weakness, myoclonus, encephalopathy, and sensorimotor neuropathy attributed to lithium toxicity; both her EMG and EEG were notably abnormal. She improved after lithium cessation (168). Secondary infections are also problematic, raising the issue of critical illness neuromyopathy in some instances. No convincing experimental evidence is known other than an isolated report suggesting a tendency toward reduced nerve fiber area in rats chronically given lithium over control animals (107). In fact, in a small series, lithium has been reported to blunt the symptoms of vincristine-associated neuropathy in both mice and humans (142). Lithium pretreatment was found to attenuate neuropathy in paclitaxel-treated mice possibly by interacting with paclitaxel-related intracellular calcium signaling pathways (126). A double-blind placebo versus lithium 300 mg pretreatment trial of 36 breast cancer patients was reported (129). Treatment or placebo was given for 5 days in each chemotherapy cycle, starting the day prior to chemotherapy. Assessment was done by EMG, clinical symptoms, and examination at 3 and 9 months. Both groups had neuropathic signs and symptoms that improved by month 9. No significant differences between groups was noted (129). A larger multicenter, randomized phase 2 trial (N=84 subjects) including lithium versus placebo pretreatment in paclitaxel-treated patients to assess differences with chemotherapy-induced neuropathy prevention is ongoing (79).

Phenelzine. Phenelzine is a rarely used monoamine oxidase inhibitor for atypical or refractory depression. Side effects such as hypertensive crises and serious reactions with other agents are well known. Rarely, this agent (but not other MAOIs) has been implicated in inducing peripheral neuropathy. Phenelzine has been shown to affect pyridoxine metabolism and reduce measurable active pyridoxal phosphate levels in humans (113). The compound is in the same chemical class as hydralazine and isoniazid, which both reduce pyridoxal phosphate levels and can cause peripheral neuropathy. Whether this effect is clinically relevant remains to be seen. Malcolm and colleagues demonstrated pyridoxal phosphate levels reduced, on average, by half in 19 patients on phenelzine, but none developed clinical symptoms (113). Several reports of neuropathy associated with phenelzine have been published (77; 74). The neuropathy is described as a typical toxic neuropathy with sensorimotor axonal involvement with predominantly sensory manifestations.

Phenytoin. Peripheral neuropathy from chronic phenytoin use has been long recognized and generally accepted. However, despite many reported patient series, the phenomenon is based on relatively few uncomplicated prospective studies. Most likely, there is a probable effect of protracted use, especially with serum levels chronically higher than 20 µg/ml (in excess of the standard therapeutic range). Peripheral neuropathy was more commonly seen early in the history of phenytoin use when doses of 500 mg/day or higher were not uncommon. However, many of the earlier series had relatively few patients on phenytoin monotherapy, and the contributions of acute reversible phenomena were not taken into account. At current dosages with monitored serum levels, peripheral neuropathy is rare and typically produces only asymptomatic examination findings or minimally discernible neuropathy after many years of therapy. The incidence of neuropathy in epileptics on phenytoin varies considerably depending on patient populations and criteria employed (110; 58; 171; 159; 172). Several variables have been proposed as risk factors for neuropathy development, including supra-therapeutic serum levels (greater than 20 µg/ml), protracted use (less than 10 years), and low folate levels (110; 58; 44; 159). Other series have not found any significant association with phenytoin use compared with other anticonvulsants or these risk factors (171; 172). Swift and colleagues found signs of neuropathy in epileptic patients on various therapies and showed a higher incidence among patients on phenobarbital (171). One case with long-term chronically elevated serum levels (31 to 38.5 µg/ml) had clinically symptomatic neuropathy, and sural nerve biopsy demonstrated mild decreases in large diameter axonal number, axonal shrinkage, and secondary demyelination (147). This patient improved clinically and on electrophysiologic studies subsequent to phenytoin cessation. A review of the literature questioned the relevance of this association at currently utilized doses (92).

In addition, there appears to be separate acute effects on nerve function. Acute exposure to high-dose phenytoin causes reversible slowing of nerve conduction velocity. Phenytoin affects sodium permeability across neuronal membranes by stabilizing inactive sodium channels (112). Phenytoin in myelinated nerve preparations produces a voltage-dependent block of sodium channels, a shift of the sodium channel inactivation curve to more negative voltages, and a reduced rate of sodium channel recovery from inactivation (157). However, carbamazepine produced some of these effects as well. Several animal studies have examined the effects of phenytoin on peripheral nerve function. Acute reversible effects have been produced with reduced conduction velocity and compound motor action potentials with acute high dose phenytoin administration in rats (115) and slow velocity after several days in guinea pigs (104). Serum levels were higher than 50 µg/ml. This reversible phenomenon likely represents a physiologic effect but is not a model of long-term toxicity. Some degree of acute reversible effects may have complicated some prior studies that examined chronic toxicity on high dose therapy. A human report has described similar reversible symptomatic effects three hours after a phenytoin loading dose (196). This may represent an additional acute or subacute idiosyncratic syndrome, but a separate syndrome is not well established. The acute reversible effects on nerve function are well established, but the chronic neuropathy is considered a probable association (114).

Podophyllin. Extracts of various plants containing podophyllin resin have been used in Asia for hundreds and even thousands of years for a multitude of purported therapeutic benefits including snake bites, general weakness, lymphadenopathy, tumors, and warts. Similar uses arose in the West over a hundred years ago, commonly as a laxative. The active resin is derived from the roots or rhizomes of the mandrake or mayapple plant (podophyllum peltatum). Additionally, podophyllin is a component of herbs used in Chinese herbal medicines including Bajiaolian (dsyosma pleianthum) and Guijiu (podophyllum emodi wall). The chemotherapeutic agents etoposide and teniposide are semisynthetic derivatives of podophyllotoxin.

Kaplan first introduced the resin as an effective medical treatment for condyloma acuminata, for which it is still commonly used (90). Furthermore, alternative uses are still commonplace in both Asia and the West. Peripheral neuropathy from podophyllin is an unusual complication of exposure to podophyllin resin from a variety of prescription and nonprescription sources. However, despite the numerous clinical reports, this entity is typically omitted from lists of medication or toxic neuropathies. In the West, toxicity from nonprescription sources is known. Dobb and Edis reported a case in Australia of severe podophyllin toxicity with neuropathy from a simple over-the-counter dietary supplement (55). Numerous cases have followed ingestion of Chinese herbal broth containing podophyllin (132; 89; 36; 45) as well as other preparations. Confusion of intended herbs is not rare due to superficial similarities, and toxic reactions have resulted from substitution of Guijiu for other intended ingredients (89).

Conventional medical application of podophyllin is for treatment of condyloma acuminata. Treatment typically involves topical use on the wart or direct intralesional injection. Care is taken not to expose healthy tissue, especially surrounding mucosa, which produces systemic dosage and can lead to toxicity; also, application on friable skin, recent biopsy sites, or over large areas is vulnerable to toxicity. Cases of intentional overdosage have been described as well (31; 22; 135; 103).

Acute manifestations within the first 1 to 2 hours of exposure consist of nausea, vomiting, abdominal pain, and subsequent diarrhea (185; 66; 45). Exposure can be topical (185; 163; 66) oral (47; 118; 190; 55; 22; 75; 78; 37; 42; 132; 40; 89; 45), or intralesional (111; 178). There is a lone report of intramuscular exposure (187). During the first day, central nervous system impairment may ensue, including altered mental status ranging from mild confusion to frank coma. Hallucinations and seizures can also occur. Paresthesias heralding the onset of peripheral neuropathy may begin within the first 2 days but can be delayed for over a week. The neuropathy may progress to marked sensory loss of all modalities, especially large fiber light touch and proprioception. Severe sensory ataxia with wide-based gait and pseudoathetosis may develop (75; 135; 187). Patients generally develop areflexia with a distal stocking-glove pattern of sensory loss.

Autonomic neuropathy has been reported in numerous cases, with findings including sinus tachycardia, urinary retention, paralytic ileus, orthostatic hypotension, diaphoresis, and syncope (66; 187). Peripheral neuropathy has been found in all but a handful of reports; however, central involvement may be absent or transient (75; 187). Non-neurologic involvement may include hepatic dysfunction, pancytopenia, renal insufficiency, and lesser toxicity in other organ systems.

Podophyllin resin contains numerous compounds, some of which have yet to be fully elucidated, but podophyllotoxin, a highly lipid soluble alkaloid, has been shown to be the primary active agent. Other constituents include picropodophyllum, podophyllic acid, alpha and beta peltatin, and the mutagens quercetin and kaempferol (154). Quercetin may have independent neurotoxic effects (46). The resin, as well as purified podophyllotoxin, has been shown to cause parallel toxicity in experimental animals, including marked axonal neuropathy and morphologic changes in the dorsal root ganglia as well as central and peripheral nervous system axons (192; 39). Correspondingly, electrodiagnostic studies have suggested a sensory neuronopathy in addition to the peripheral dying-back form of toxic axonopathy (40).

Podophyllotoxin has a chemical structure similar to colchicine and vinblastine, which are other causes of axonal neuropathy as well as agents that arrest mitosis in metaphase. All of these compounds disrupt microtubular function and inhibit axonal transport. Both podophyllotoxin and colchicine bind to tubulin leading to axonal transport disruption. Dorsal root and autonomic ganglia may also be susceptible to circulating toxin and lead to the neuronopathy. Evidence of both effects is seen in experimental models and clinical reports (75; 39). Toxicity is rare with careful medical application. Manufacturers recommend strict guidelines for medication availability and usage techniques, including application procedures and subsequent drug removal. Added risk from application on friable areas or recent biopsy sites is also critical. Risk from nonmedical sources is less controllable. Additionally, peripheral neuropathy is prevalent in the population at large. Many patients choose to self-medicate with complimentary or alternative medical remedies without informing their treating physicians, most commonly patients with diabetic neuropathy. Some remedies could potentially contain podophyllin resin and have a detrimental effect. In one series, 43% of neuropathy patients tried some form of alternative treatment (27).

Treatment with podophyllin is contraindicated in pregnancy and may lead to miscarriage or stillbirth. One case of intrauterine death at week 32 after mucosal application has been reported (33), as has another with limb malformations (154). However, no controlled human studies are available.

Proton pump inhibitors. A rare effect of commonly used medications can be particularly problematic to resolve and substantiate. One example is the proton pump inhibitors omeprazole and lansoprazole. Rajabally and Jacob reported a 42-year-old woman who developed predominantly sensory neuropathy after three months of lansoprazole use (145). Some partial improvement was noted after later stopping the medication, and no worsening was seen after switching to rabeprazole and then to ranitidine. Three other cases are reported with omeprazole, two of which have adequate electrophysiology and clinical information (62). Additional carefully studied examples are needed to further substantiate this possible link with medication-induced neuropathy in this widely used class of medications. No new cases have been published as of the most recent literature search since these reports despite continued widespread use of these agents. However, these agents and histamine-2 blockers may affect vitamin B12 absorption and lead to secondary neurologic complications (101).

Slaughterhouse workers progressive inflammatory neuropathy. Although not technically a medication-induced neuropathy, this local toxic epidemic at several pork processing plants in Minnesota and surrounding states produced considerable activity and investigation by numerous researchers, mostly at the Centers for Disease Control (32). Twelve workers in a swine slaughterhouse in Minnesota developed a progressive inflammatory neuropathy with symptoms ranging from acute paralysis to gradually progressive symmetric weakness predominantly in the legs from eight to 213 days with varying severity between November 2006 and 2007. Eleven patients had evidence of axonal or demyelinating features by electrodiagnostic testing. Spinal fluid from seven patients showed elevated protein (mean 120 mg/dl) with no or minimal pleocytosis. Ten patients had evidence of inflammation on spinal magnetic resonance imaging (nine patients in peripheral nerves or roots and one patient in the anterior spinal cord). Three patients with sural nerve biopsy showed mild perivascular inflammation. In summary, patients were characterized with a sensory greater than motor polyradiculopathy, predominantly at the root or distal nerve level. The CDC researchers identified that all patients were working in close proximity to swine heads. A compressed air device used to liquefy porcine brain material may have generated aerosolized brain material, which may have induced an immune neurotoxic response. Ultimately work at the Mayo Clinic led by Vanda Lennon found a complex autoantibody profile dominated by neural cation channel IgGs that most significantly affected voltage-gated potassium channels (120).

Tacrolimus. Prograf, previously known as FK-506, is a novel immunosuppressant that is widely used in transplant medicine and for suppression of some inflammatory disorders. The agent is a macrolide antibiotic that suppresses both cellular and humoral mediated immune responses. Neurotoxicity is common in treated patients, in part, because of the relatively high doses usually given. Central toxicity is more common with a variety of findings including leukoencephalopathy, seizures, behavioral changes, headache, or other cortical signs, many of which are dose dependent. Peripheral neuropathy appears to take the form of a severe multifocal demyelinating neuropathy that resembles chronic inflammatory demyelinating neuropathy (191; 26; 100). Patients have responded to IVIG or plasmapheresis as well. Optic neuropathy is also uncommon but described (148). A possible case of cyclosporine-induced erythromelalgia in a patient with myasthenia gravis is reported (29).

Both cyclosporin A and tacrolimus act through inhibition of calcineurin, though by different means (tacrolimus binding protein: FKBP-12) (164). The calcineurin inhibition, through several steps, decreases IL-2 and eventually T-cell proliferation. This pathway is also the likely cause of much of the central neurotoxicity and possibly the peripheral effects. Tacrolimus also has an additional separate function through a different binding protein, FKBP-52, that acts as a nerve stimulator, increases growth associated protein (GAP-43), and is beneficial to nerve regeneration in nerve axotomy and ischemia models (73; 94). FKBP-52 is part of a steroid receptor complex and may represent a target for future regenerative therapies separate from the growth factor and Trk pathways. The mechanism of why, in some patients, an immune attack that resembles chronic inflammatory demyelinating neuropathy or other autoimmune neuropathy is unclear; however, the number of reported examples is small. Interestingly, tacrolimus has also been shown to have significant and potentially therapeutic neuroregenerative activity, possibly derived from a separate pathway from the immunosuppressive calcineurin inhibition--FKBP-52 binding protein (99; 72). Schwann cells may play an important intermediary role (18). A similar agent, sirolimus, appears to have less risk of this reaction but at least one case is reported (17).

Tumor necrosis factor-alpha blockers. Tumor necrosis factor-alpha (TNF-alpha) blockers are used in the treatment of various forms of inflammatory arthritis and inflammatory bowel diseases but are also associated with inducing or worsening other autoimmune disorders including multiple sclerosis (169). One agent (etanercept) has been reported to improve chronic inflammatory demyelinating neuropathy (CIDP) (102; 43). Postmarketing reporting identified 15 patients diagnosed with Guillain-Barré syndrome or Miller Fisher syndrome from 6 weeks to 2 years after starting a TNF-alpha blocker, although associated infection may be a more important risk factor (151; 158). One case developed acute sensorimotor neuropathy and concomitant encephalopathy (61). Richez and colleagues reported two cases that developed a CIDP-like illness (150). One treated with etanercept for rheumatoid arthritis developed a demyelinating neuropathy 17 months later. The other received infliximab for ankylosing spondylitis and developed CIDP 3 months later. Both incompletely improved after drug cessation without specific treatment for CIDP. Infliximab is also associated with several other CIDP-like cases with underlying rheumatoid arthritis (86; 174; 07) and three cases with underlying psoriatic arthritis, a condition that is much less likely to induce spontaneous or vasculitic neuropathy (169; 57). One reported patient on chronic adalimumab for rheumatoid arthritis developed significant demyelinating neuropathy on initiation of steroids that may have unmasked the inflammatory neuropathy (193). Chronic adalimumab therapy was also implicated in a case of distal acquired demyelinating symmetric neuropathy for the treatment of rheumatoid arthritis (119). Numerous cases resembling multifocal motor neuropathy are reported in association with infliximab (162; 48; 153; 137; 65; 155; 11). However, others question whether some of these cases were actually a form of vasculitic mononeuritis multiplex triggered by the infliximab (19). One case of proposed infliximab-associated immune-mediated sensory polyradiculopathy was successfully treated with intravenous gammaglobulin (131). There are additional less-clear associations with mononeuropathy and axonal sensory or sensorimotor neuropathy (86). In any event, it seems that infliximab and etanercept can contribute to or trigger an immune-mediated neuropathy in some possibly susceptible patients (97). Adalimumab is not clearly associated with chronic neuropathy but was associated with one possible Guillain-Barré syndrome case. Ipilimumab, a monoclonal antibody that is not a TNF alpha antagonist but instead blocks a natural inhibitor of cytotoxic T-cell response to cancer cells, is approved to treat melanoma and is undergoing trials for other cancer types. A case of acute neuropathy mimicking Guillain-Barré syndrome is reported; acute enteric neuropathy is also recognized. A case of adalimumab inducing sensory vasculitic neuropathy is reported (161). A very large observational study in five Nordic countries separately tracked patients with rheumatoid arthritis and spondyloarthritis treated with TNF-alpha monoclonal antibody treatments from 2001 to 2020 (53). The primary outcome measure was development of a neuroinflammatory complication using ICD-10 neuropathy codes. They found 314 such events out of about 62,000 total patients. Central demyelinating events and multiple sclerosis were most common, but 86 cases of probable inflammatory neuropathy were reported. No difference in various treatments were detected, though comparing etanercept to other agents was a primary study goal. Unexpectedly, spondyloarthritis patients had a higher complication rate than rheumatoid arthritis patients.

Interestingly, in light of fluoroquinolone patient-derived reports, peripheral neuropathy associated with TNF-alpha agents was the most common adverse neurologic event reported to the Food and Drug Administration Adverse Event Reporting System (296 reports, 38.3%), exceeding central nervous system and/or spinal cord demyelination (153 reports, 19.8%) (52). The majority of reports (71%) were labeled as “possibly associated” and not higher grades of certainty.

In contrast, these agents may have other protective properties. A mouse model of bortezomib neuropathy found that upregulation of TNF-alpha was neuroprotective, possibly by limiting certain inflammatory cytokines (04).

Zimeldine. Zimeldine is another agent never approved for use in the United States but available transiently as an antidepressant in Sweden, functioning as a 5-HT reuptake inhibitor with purported fewer side effects. The drug is best known as a probable precipitating factor of an outbreak of Guillain-Barré syndrome in Sweden in 1983. The drug was withdrawn from the market 18 months after introduction because of this outbreak. A subsequent Bayesian analysis concluded that the association was supported by relevant data (130). No additional cases, however, were identified in a retrospective review of 761 patients on zimeldine reported from the same region (12). Hypersensitivity reactions are relatively common with this agent (1.4% to 13%), raising the question of potential immune-mediated mechanisms in this phenomenon. Zimeldine appears to affect T-cell function and blunt experimental allergic neuritis in a rat model of Guillain-Barré syndrome (13). The risk of developing Guillain-Barré syndrome from zimeldine was estimated as increased 25-fold compared to natural incidence controls (60).

Epidemiology

No predisposition to the toxic manifestations is known for any of these agents. However, possible underlying occult neuropathy has been noted in multiple case reports, especially from diabetes mellitus. Other potential or theoretical underlying factors include genetic susceptibility (hereditary neuropathies) or variations in metabolism or clearance.

Differential diagnosis

Because of the rarity of these entities, a reasonable search for other causes of neuropathy is necessary, except possibly in the case of almitrine. Improvement after drug cessation is necessary to consider the association.

Diagnostic workup

Approach is the same as in other cases of peripheral neuropathy.

Management

Supportive measures as needed are the mainstay of management and are discussed in Peripheral neuropathies: supportive measures and rehabilitation.