Demyelination occurring during anti–tumor necrosis factor α therapy for inflammatory arthritides


  • The opinions expressed herein are those of the authors and do not reflect the official policy of the institutions that they are affiliated with (the FDA, the Armed Forces Institute of Pathology, and Georgetown University).



To review the occurrence of neurologic events suggestive of demyelination during anti–tumor necrosis factor α (anti-TNFα) therapy for inflammatory arthritides.


The Adverse Events Reporting System of the Food and Drug Administration (FDA) was queried following a report of a patient with refractory rheumatoid arthritis who developed confusion and difficulty with walking after receiving etanercept for 4 months.


Nineteen patients with similar neurologic events were identified from the FDA database, 17 following etanercept administration and 2 following infliximab administration for inflammatory arthritis. All neurologic events were temporally related to anti-TNFα therapy, with partial or complete resolution on discontinuation. One patient exhibited a positive rechallenge phenomenon.


Further surveillance and studies are required to better define risk factors for and frequency of adverse events and their relationship to anti-TNFα therapies. Until more long-term safety data are available, consideration should be given to avoiding anti-TNFα therapy in patients with preexisting multiple sclerosis and to discontinuing anti-TNFα therapy immediately when new neurologic signs and symptoms occur, pending an appropriate evaluation.

Although the tumor necrosis factor (TNF)/TNF receptor (TNFR) system is ubiquitous in the human body, its various roles in normal physiologic conditions and disease pathogenesis have not been fully elucidated. TNFα is produced in response to infection or immunologic injury and effects multiple responses that extend beyond its well-characterized proinflammatory properties (1) to include diverse signals for cellular differentiation, proliferation, and death (2). Some studies have indicated that the TNF/TNFR system is an important mediator of inflammation in both rheumatoid arthritis (RA) and multiple sclerosis (MS). Its role in mediating the arthritogenic response has been demonstrated by the improvement of arthritis lesions in anti-TNFα–treated animal models of arthritis (3), and anti-TNFα agents have been reported to be safe and effective in the treatment of RA (4, 5). TNFα is also thought to play a significant role in the pathogenesis of inflammatory demyelinating disease of the central nervous system (CNS), as has been demonstrated in experimental autoimmune encephalomyelitis (EAE) (6), an established animal model for human MS, and in studies of MS in humans (7, 8).

We report a series of patients who developed new-onset neurologic signs and symptoms, in most cases associated with demyelinating lesions of the CNS, while undergoing therapy with anti-TNFα agents, namely, etanercept (Enbrel; Immunex, Seattle, WA) or infliximab (Remicade; Centocor, Malvern, PA). Etanercept is a p75TNFR fusion protein and infliximab is a chimeric monoclonal antibody against TNFα. These cases underscore 1) the importance of clarifying further the interactions and functions of TNF/TNFR systems in vivo, both in the periphery and within the CNS, and 2) the need for further surveillance and followup of patients being treated with anti-TNFα agents, in order to better define potential adverse events and prevent neurologic damage.


The index patient was a 48-year-old man with RA who developed neurologic signs and symptoms while receiving etanercept. We conducted a search of the medical literature and did not find any reports of neurologic events associated with etanercept. This case was then reported to the MedWatch program (for the voluntary reporting by health professionals of adverse events and product problems as well as mandatory reporting of the same by drug manufacturers) of the Food and Drug Administration. The Adverse Events Reporting System (AERS), which is used to categorize reports of adverse events related to drugs, was queried using the following terms from the Medical Dictionary for Drug Regulatory Affairs: central nervous system infections and inflammatory disorders, cranial nerve disorders (excluding neoplasms), demyelinating disorders, encephalopathies (not otherwise specified), Guillain-Barré syndrome, mental impairment, MRI (magnetic resonance imaging) changes, neurologic disorders of the eye, neurologic signs and symptoms, peripheral neuropathies, and spinal cord and nerve root disorders. We reviewed the information available in the MedWatch forms as well as available MRI scans.


A search of the AERS database identified 17 patients with neurologic events suggestive of demyelination following etanercept administration and 2 patients (patients 19 and 20) with such events following infliximab administration (Table 1). Reports to the AERS do not necessarily represent causal relationships between adverse events and drugs. In addition, underreporting of adverse events occurs, and the reports may be missing data. Therefore, these reports should be interpreted cautiously.

Table 1. Patients with demyelination occurring during anti–tumor necrosis factor α therapy*
Patient/ age/sexDiagnosisTime receiving drugConcomitant medicationsClinical signs and symptomsMRI findingsTreatmentFindings at followup
  • *

    From the database of the Adverse Events Reporting System of the Food and Drug Administration. MRI = magnetic resonance imaging; RA = rheumatoid arthritis; PsA = psoriatic arthritis; MTX = methotrexate; MS = multiple sclerosis; IFNβ = interferon-β; IVIG = intravenous immunoglobulin; JRA = juvenile rheumatoid arthritis; C2 = second cervical vertebra; C-spine = cervical spine; IV = intravenous.

  • Index patient.

 1/48/MRA4 monthsAcyclovir, metronidazole, ceftriaxone, ranitidineConfusion, difficulty walkingWhite matter changesPulse methylprednisolonePartial resolution at 6 weeks
 2/37/MRA3 monthsNaproxen sodiumConfusion, apraxiaParietal and occipital demyelinationPulse methylprednisolonePartial resolution and positive rechallenge at 4 months
 3/37/MPsA2 monthsMTXAscending dysesthesiaDemyelination of cervical cordOral dexamethasoneComplete resolution with recurrence at 7 months
 4/43/MPsA4 monthsPrednisoneParesthesiaDemyelinationUnknownContinued symptoms while continuing etanercept
 5/48/FUnknown15 monthsUnknownSpasticity and lower trunk paresthesiaThoracic and brain MSUnknownUnknown
 6/43/MUnknown2.5 monthsAllopurinol, omeprazole, prednisolone, probenecid, propoxypheneOptic neuritis, papilledemaNormalUnknownUnknown
 7/40/FRA10 monthsPrednisone, nabumetone, alendronate, hydrocodoneOptic neuritis, paresthesiaDemyelinationIFNβ, IVIG, glatiramerPartial resolution at 4 months
 8/56/FUnknown12 monthsPrednisone, ranitidineParesthesiaDemyelinationPrednisoneComplete resolution at 3 months
 9/53/MRA2.5 monthsAlprazolam, atenolol, fluoxetine, fentanyl patchProgressive weakness and paresthesiaNormalPlasmapheresisPartial resolution
10/48/FPsA5 monthsMTX, piroxicam, bupropionLhermitte's sign, leg numbnessOptic neuritisUnknownContinued symptoms at 6 months
11/21/FJRA9 monthsMTX, prednisone, piroxicamParesthesiaCervical myelitis at C2UnknownUnknown
12/41/FRAUnknownUnknownParesthesia, optic neuritis, hemiparesisMultiple plaquesPulse methyl-prednisoloneComplete resolution at 4 weeks
13/47/MRA4 monthsMTX, dexamethasone, epoetin, calcium, iron, nandroloneAltered mental status, personality and visual changesFingerlike brain edema, barrier damageAcyclovir, ampicillin, high-dose cortisonePartial clinical and MRI resolution at 7 months
14/44/FRA7 dosesPrednisoneTransverse myelitisPosterior conus medullaris lesionMethylprednisoloneComplete resolution at 2 weeks
15/51/FRA4 monthsEstradiol, zolpidemParesthesias; speech, gait, visual, and cognitive disturbances; headache; back painDemyelinationFluoxetineContinued symptoms while continuing etanercept
16/21/FJRA7 monthsLansoprazole, cyclobenzaprineVisual loss due to optic neuritisT2 abnormality, brain and C-spine enhancementIV steroids with steroid taperContinued symptoms with 2 new MRI lesions at 2 months
17/46/FRA1 monthPrednisone, glatiramer, sulfasalazine, rofecoxib, hydrocodoneIncontinence; visual, balance, and cognitive difficulties; paresthesiasUnknownUnknownPartial resolution at 3 months
18/50/FPsA5 monthsMTXGait and bladder difficultiesMSUnknownUnknown
19/53/FRA2.5 monthsUnknownDiplopia, nystagmus, weakness, neurogenic bladder, paresthesiaDemyelination in tectum and spinal cordMethylprednisolone, IVIGPartial resolution
20/42/MRA1 doseLeflunomide, prednisolone, amlodipine, bendroflumethiazide, tramadol, lansoprazole, co-proxamol (paracetanol + dextropropoxyphene)DysarthriaDemyelinationUnknownUnknown

The index patient (patient 1) was a 48-year-old man with refractory RA for 10 years who had developed fever, confusion, and gait disturbance after receiving etanercept (25 mg subcutaneously twice a week) for 4 months. Findings on physical examination were notable for altered mental status, mild left facial palsy (central), and ataxia. Etanercept was discontinued and the patient was started on a regimen of ceftriaxone, metronidazole, acyclovir, and methylprednisolone for a presumed infection at a community hospital. Cerebrospinal fluid (CSF) analysis yielded normal findings. MRI revealed large areas of increased signal intensity on T2-weighted images throughout the periventricular and subcortical white matter without abnormal contrast enhancement. At this juncture, the patient was transferred to Georgetown University Medical Center.

On admission, the patient was febrile (with a temperature of 38.4°C) and responsive only to noxious stimuli; by the second day of admission, he became unresponsive. A repeat CSF analysis revealed an elevated protein level of 79 mg/dl (normal range 15–45 mg/dl). An electroencephalogram showed diffuse slowing. Findings on a cerebral angiogram were normal. A repeat MRI revealed progression of the white matter signal abnormality such that it became confluent (Figures 1A and B). Brain biopsy revealed spongiotic changes in the white matter with gross macrophage infiltration on hematoxylin and eosin staining (Figure 2), raising the possibility of a demyelinating process. This prompted an investigation for demyelination as mentioned above. However, special stains (luxol fast blue for myelin) revealed preservation of myelin. The final reading of the biopsy sample was most consistent with a leukoencephalopathy. Differential diagnosis included toxic and metabolic injuries, cerebral edema, and spongy degeneration.

Figure 1.

Magnetic resonance imaging scan of brain of index patient. A, Axial fluid-attenuated inversion recovery image shows confluent signal abnormality in the periventricular and deep white matter. B, Axial postcontrast T1-weighted image demonstrates no abnormal enhancement.

Figure 2.

Brain biopsy sample (index patient) showing spongiotic changes in the white matter with gross macrophage infiltration. (Hematoxylin and eosin stained; original magnification × 20.)

The patient improved gradually following 5 days of pulse methylprednisolone (1 gm/day). At discharge, he had residual left-sided weakness and dysphagia. His neurologic status continued to improve, and at the followup visit 6 weeks later, he had normal mentation, 4/5 strength on the left side, and normal speech and swallowing. He was also amnesic for the events of his hospital stay. A repeat MRI of the brain was planned for the 3-month visit. The patient has since been lost to followup.

Clinical presentation. The most common presenting clinical symptoms among the 20 patients identified in the AERS search were paresthesias (13 of 20) followed by visual disturbances secondary to optic neuritis (8 of 20). Other signs and symptoms included confusion (5 of 20), gait disturbance, apraxia, facial palsy, and Guillain-Barré syndrome. One of the atypical features that occurred in 25% of patients was confusion, which is uncommon in MS. Four patients had a prior history of MS or an MS-like syndrome with flares of their previous symptoms. Patient 2 exhibited a positive rechallenge phenomenon (i.e., his neurologic symptoms improved on discontinuation of the drug and were exacerbated on reexposure to etanercept 2 months later). His steroid treatment was being tapered off at the same time. Etanercept was again discontinued, and the patient currently has a permanent residual neurologic deficit. Five of 20 patients were reported to be receiving methotrexate (MTX), and 1 of 20 patients was reported to be receiving leflunomide in addition to anti-TNFα therapy at the time of the neurologic event. Most patients responded either partially or completely, with clinical resolution of their neurologic symptoms, on discontinuation of anti-TNFα therapy. Other therapies that were tried with varying degrees of success included corticosteroids (pulse or rapidly tapering), plasmapheresis, interferon-β1A (IFNβ1A), and intravenous immunoglobulin (IVIG).

Tissue diagnosis and imaging. Lumbar puncture was reported for only 1 patient in the series, and findings were normal. Brain biopsy was reportedly performed in only 2 patients; one patient's biopsy result was suggestive of toxic encephalopathy, and the other patient had demyelination. Imaging studies were reported for 19 of 20 patients. Among them, 16 disclosed a pattern compatible with demyelination in different areas of the CNS, while the index patient had an extensive, confluent, bilaterally symmetric pattern of signal abnormality in the periventricular, deep, and subcortical white matter. We were able to independently review MRI scans of the index patient (patient 1) and of 4 other patients (patients 3, 4, 5, and 7). Findings in these 4 patients were similar to each other, but different from those in the index patient. These 4 patients had small white matter lesions in the brain (3 patients) and spinal cord (2 patients) that were most compatible with a demyelinating process (Figure 3).

Figure 3.

Magnetic resonance imaging scan of the brain of patient 7. A, Axial fluid-attenuated inversion recovery image demonstrates multiple small white matter lesions typical of a demyelinating process. B, Axial postcontrast T1-weighted image demonstrates that some of the lesions abnormally enhance, suggesting an active demyelinating process.


Assessing the likelihood of a causal connection between an environmental exposure and an adverse event is referred to as attribution analysis. Miller et al (9) reviewed the many methods that have been proposed to substantiate such associations. The criteria proposed by these investigators (Table 2) were used to review the association of anti-TNFα therapy with the neurologic events in our series of patients. According to the criteria of Miller et al, the specified primary attribution elements are more critical than secondary attribution elements and include temporal association, the lack of likely alternative explanations, dechallenge (improvement in symptoms following discontinuation of the agent), rechallenge (reappearance or worsening of symptoms on reexposure to the agent), and biologic plausibility (the likelihood of the agent causing the signs and symptoms, based on its known in vivo and/or in vitro effects). Secondary attribution elements include analogy (prior published or unpublished reports of a similar disorder developing after the exposure), dose responsiveness (dose of agent related to the likelihood of developing the disorder), and specificity (similar symptoms, signs, and laboratory features in previous patients after exposure to the same agent).

Table 2. Elements of attribution analysis*
Primary elementsSecondary elements
  • *

    To publish findings of a possible causal relationship between an environmental (drug) exposure and a clinical syndrome, at least 4 of the 8 attribution elements and at least 3 of the 5 primary elements should be present. The 3 primary attribution elements should include temporal association, lack of likely alternative explanations, and at least 1 of the other primary elements.

Temporal associationAnalogy
Lack of likely alternative explanationsDose responsiveness
Biologic plausibility

Three factors argue against a true association of events and anti-TNFα therapy. First, a total of 77,152 patients were exposed to etanercept from November 1998 through May 2000, representing 55,313 person-years of exposure (Immunex: personal communication). Among these patients, only 9 were identified as having symptoms suggestive of a demyelinating disorder; the other patients, including the 2 patients receiving infliximab, were reported after May 2000. This compares with the natural incidence of MS of 4–6 cases per population of 100,000 per year (10). Second, these events could possibly represent the unrelated, chance occurrence of a demyelinating disorder (such as MS) in the setting of inflammatory arthritis and anti-TNFα therapy. Finally, only 1 of 20 patients had a positive rechallenge phenomenon.

The 5 primary elements of the attribution analysis suggest a relationship between TNFα blockade and demyelinating syndromes and are addressed herewith. Regarding the temporal relationship, the average time between the beginning of therapy and the onset of symptoms was 5 months, with exposure times ranging between 1 week and 15 months. When we searched for likely alternative explanations, we found that MTX has been associated with a variety of neurologic sequelae in the brain (such as demyelination and even necrosis) when used in combination with high doses of irradiation (11). However, these changes have been reported only with the use of high-dose MTX (intravenous, intrathecal, intraventricular) in the treatment protocols for CNS prophylaxis of lympho- and myeloproliferative malignancies. Similar events have not been reported with the use of leflunomide. A literature search for reports of demyelination associated with other concomitant medications (Table 1) yielded negative results.

Dechallenge, or discontinuation of anti-TNFα therapy, resulted in complete or partial improvement of symptoms in all patients. In the case of rechallenge, reexposure to etanercept in patient 2 caused worsening of his neurologic status on MRI, although steroid taper at that point may have contributed to his clinical deterioration. Two patients continued to receive etanercept, and their symptoms continued. None of the other patients were reported to have been reexposed to the drug. Rechallenge will be difficult to assess in the future due to the risk to patients.

Upon reviewing biologic plausibility, several studies suggest that a perturbation of the TNF/TNFR system by anti-TNFα agents can potentially precipitate or worsen an underlying demyelinating process. Elevation of TNFα levels is a recognized component of the pathophysiology of both RA and MS. TNFα was found to be a key coordinator of proinflammatory cytokines in RA (11). Similarly, TNFα is overproduced in the serum and CSF of MS patients and by resident and infiltrating cells at sites of CNS injury (8). TNFα also induces selective cytotoxicity of oligodendrocytes in vitro and myelin damage in brain slices (12).

Due to the success of anti-TNFα therapies in animal models of MS, lenercept, a soluble dimeric p55TNFR-immunoglobulin fusion protein (Hoffmann-La Roche, Basel, Switzerland), was examined in a double-blind, placebo-controlled study of 168 MS patients. The number of lenercept-treated patients experiencing exacerbations was significantly increased compared with patients receiving placebo, and their exacerbations occurred earlier. Neurologic deficits tended to be more severe in the lenercept treatment groups, although this was not statistically significant. This finding resulted in the sponsor's decision to terminate the study and to release the treatment code halfway through the study (13). Two patients with rapidly progressing MS who were treated with infliximab (Remicade; Centocor) developed increases in the number of gadolinium-enhancing lesions on MR scan, the CSF IgG index, and the CSF lymphocyte count after each infusion, although there was no reported clinical worsening of disease (14).

Little is known of the prevalence of other autoimmune diseases among population-based, unselected MS patients. Even less is known about the incidence and prevalence of MS in autoimmune diseases like RA. Two recent studies suggest a possible association between autoimmune diseases and MS. A hospital-based, case–control study of 155 MS patients and 200 controls revealed a statistically significant more frequent coexistence of RA, psoriasis, and goiter compared with controls (odds ratio 2.96, 95% confidence interval 1.23–7.66) (15). Another study of autoimmune diseases in families of French patients with MS suggested a possible familial association of MS and autoimmune disease (16). There are also conflicting case reports of improvement or worsening of inflammatory arthritis when patients were treated for MS with IFNβ (17–19). The finding of common loci for genes associated with EAE and pristane-induced arthritis suggests that common genes are likely to be involved in different autoimmune diseases (20). It is therefore conceivable that the patients in our series could have had a genetic propensity to develop MS, which may have been exacerbated by the administration of an anti-TNFα agent.

Recent studies in transgenic and knockout animals, where the pathogenic influence of genetically perturbed TNFα expression has been evaluated, indicate that several pathways of TNF/TNFR action may contribute independently or in concert to initiate, promote, or down-regulate disease pathogenesis. Overexpression of TNF in the CNS of transgenic mice causes spontaneous inflammatory demyelination, which is prevented if the p55TNFR is knocked out (21, 22). In EAE, anti-TNFα treatment completely prevents initiation of pathology and ameliorates the progression of established disease (23). In contrast, when EAE was induced in mice lacking TNFα, they developed severe neurologic impairment with a high mortality rate and extensive inflammation and demyelination. Treatment with TNFα dramatically reduced disease severity, suggesting that TNFα is not essential for the induction and expression of inflammatory and demyelinating lesions and may have a neuroprotective function in the CNS (24). In transgenic mice with overexpression of p75TNFR, the high p75TNFR level has been shown to mediate its own inflammatory pathologic changes, independent of TNFα, lymphotoxin, and p55TNFR (25). Since the TNF/TNFR system acts as a complex network, mechanisms may exist by which TNFα blockade could augment those immune responses that contribute to demyelination.

Soluble receptors influence TNFα activity in vitro and in vivo and maintain the balance between active, free TNFα and the inactive form bound to its soluble receptors. The soluble forms in high concentrations act as inhibitors by competing with TNFα cell surface receptors; however, in lower concentrations, soluble receptors can prolong the biologic half-life of TNFα by functioning as a carrier protein and protecting TNFα from degradation, and therefore stabilizing its activity (26).

A dominant role of p55TNFR has been shown in the induction phase of several TNFα-mediated pathologies, including endotoxemic shock and several transgenic mouse models of disease for arthritis (27) and demyelination (28). Elevated levels of p55TNFR have been detected in the serum of patients with MS (29). Levels of soluble TNFRs correlated with levels of the circulating adhesion molecules L-selectin and vascular cell adhesion molecule 1, which reflect lymphocyte/monocyte and endothelial cell activation occurring in the setting of acute inflammatory processes affecting the CNS (29). In contrast, there is ample in vitro evidence for a cooperative role for p75TNFR in p55TNFR-mediated responses, leading to the concept that p75TNFR plays an accessory role in p55TNFR signaling (30). Hence, increasing levels of soluble p75TNFR by administering etanercept (p75 receptor fusion protein) or decreasing levels of circulating TNFα by administering infliximab may interfere with immune homeostasis and disease pathogenesis by as-yet-undefined mechanisms to potentially exacerbate a patient's underlying tendency to develop MS or cause a relapse in a patient already diagnosed as having MS.

Analysis of the secondary attribution elements (Table 2) shows a positive analogous exposure occurring in the 2 MS patients treated with infliximab (14). The data on dose responsiveness and specificity of these symptoms to etanercept and infliximab are currently unavailable.

Despite the small number of patients in our series, the occurrence of these neurologic events in the setting of TNFα blockade raises the possibility of a true association between demyelination and anti-TNFα therapy. Clinicians should consider avoiding anti-TNFα therapy in those patients who have a preexisting diagnosis of MS and should be cautious with its use in those with a strong family history of MS. If a patient receiving anti-TNFα therapy develops new neurologic signs and/or symptoms suggestive of demyelination, the following steps are warranted:

  • 1Immediate discontinuation of anti-TNFα therapy pending further investigations.
  • 2A thorough neurologic examination, including a fundus examination for papilledema and optic neuritis, coupled with a neurology consultation.
  • 3MRI of the brain with and without gadolinium.
  • 4Lumbar puncture for CSF studies, including oligoclonal bands and IgG levels.
  • 5When the above investigations are inconclusive or the patient has rapid clinical and/or radiographic progression, a brain biopsy may be helpful to delineate the differential diagnosis.
  • 6Therapies used in MS, such as glucocorticoids, IFNβ, or IVIG, should be considered.

This case series suggests the possible association of a demyelinating syndrome or other forms of white matter injury (leukoencephalopathy) with the use of anti-TNF agents in inflammatory arthritides. Due to the relatively small number of patients who have been exposed to anti-TNFα agents, further epidemiologic, clinical, and laboratory studies are necessary to test this hypothesis. What occurs in vivo in the CNS when the TNF/TNFR systems are perturbed in the periphery will remain a matter of speculation until more is learned about the complex regulation of these cytokines across the blood–brain barrier. Therefore, it is critical to monitor patients receiving anti-TNFα therapy for the development of new neurologic signs and symptoms, and to discontinue the use of anti-TNF agents when clinical signs of white matter injury appear, in order to prevent neurologic damage.


We are extremely grateful to Rosemary Neuner, MD, MPH, for her help and advice with the review of the case series and preparation of the manuscript.