ORIGINAL REPORT

Peripheral neuropathy associated with leflunomide:is there a risk patient profile?y

Karin Martin pharmD, PhD1,2*, Fabrice Bentaberry MD4, Chantal Dumoulin MD4,Ghada Miremont-Salame MD, MPH1,2,3, Francoise Haramburu1,2,3,Joel Dehais4 and Thierry Schaeverbeke4

1Departement de Pharmacologie, Universite Victor Segalen, Bordeaux cedex, France2Inserm Bordeaux cedex, France3Centre Regional de Pharmacovigilance, Bordeaux cedex, France4Service de rhumatologie, CHU, Bordeaux, France

SUMMARY

Purpose (i) To monitor the potential clinical neurotoxic symptoms in patients treated with leflunomide in daily practiceand (ii) to describe the characteristics of patients presenting with this peripheral nervous system symptoms.Method All patients treated with leflunomide between May 2000 and April 2003 and followed in the rheumatologydepartment of the University Hospital participated in the study. Data concerning treatment patterns with leflunomide,demographic and disease characteristics were obtained from clinical charts. Neuropathy was diagnosed with nerveconduction study (NCS). Cases of neuropathy were described and then compared to other patients using univariate analyses.Results One hundred and thirteen patients were included in the study.M/F sex ratiowas 0.45.Mean age at start of treatmentwas 55.6 years (range¼ 27–81). During the study period, eight incident cases of peripheral neuropathy and two cases ofworsening of preexisting neuropathy were reported (incidence: 9.8%). Compared with other patients, neuropathy cases wereolder (69 vs. 54 years, p¼ 0.0006), more often diabetic (30% vs. 2.9%, p¼ 0.009) and more often treated with potentiallyneurotoxic drugs (20% vs. 1.9%, p¼ 0.039). At least one risk factor (potentially neurotoxic drug or diabetes) was found in50% of patients with neuropathy versus 4% of patients without neuropathy (56% PPV, 96% NPV).Conclusion Cases of toxic neuropathy have been observed during treatment of rheumatoid arthritis with leflunomide.Their occurrence seems to be associated with known risk factors. Careful monitoring of the patient’s neurological statusduring leflunomide treatment is therefore mandatory. Copyright # 2006 John Wiley & Sons, Ltd.

key words—Leflunomide; peripheral neuropathy; adverse drug reactionReceived 15 September 2005; Revised 13 March 2006; Accepted 29 May 2006

BACKGROUND

Leflunomide is a new disease-modifying drug(DMARD) recently licensed for treatment of rheuma-toid arthritis. This isoxazole derivative acts as a

prodrug for the active metabolite A77 1726 which hasimmunosuppressive properties. Several clinical trials(3 randomized clinical trials1–3 and 4 extended openclinical trials4–7 have shown that its efficacy and safetyfor treating rheumatoid arthritis (RA) are similar to thegold standard in this pathology, i.e., methotrexate. Ithas therefore become part of the therapeutic arsenalpresently available for rheumatic disease. Never-theless, since its launch, several case reports havesuggested that its safety profile could be less favorablethan that of methotrexate, and that adverse effects are

pharmacoepidemiology and drug safety 2007; 16: 74–78Published online 14 July 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/pds.1282

*Correspondence to: Dr K. Martin, Departement de Pharmacologie,Inserm U657, IFR 99 Universite Victor Segalen Bordeaux 2, 33076Bordeaux cedex, France.E-mail: karin.martin@pharmaco.u-bordeaux2.fryNo conflict of interest was declared.

Copyright # 2006 John Wiley & Sons, Ltd.

the first cause of treatment withdrawal.8–11 Duringclinical trials, paresthesia has been the only peripheralnervous event reported (difference statistically sig-nificant with methotrexate). Since launching of thedrug, several cases of neuropathy have been pub-lished,12–17 nearly 80 cases reported to the US Foodand Drug Administration,16 and to date 19 cases to theFrench Pharmacovigilance system.18

OBJECTIVE

The aim of this study was (i) to monitor the potentialclinical neurotoxic symptoms in patients treated withleflunomide in daily practice and (ii) to describe thecharacteristics of patients presenting with thisperipheral nervous system symptoms.

METHOD

Inclusion criteria

In this follow-up, i.e., cohort study, all patients whosetreatment with leflunomide was started between May2000 and April 2003 and followed in the Departmentof Rheumatology of the University Hospital wereincluded in the study. At inclusion, data of somepatients were already registered (historical charts).Therefore, the follow-up is based both on historicaland prospective data.

There were no exclusion criteria. The followingdata were obtained from the clinical charts: (i)demographic and disease characteristics, (ii) medicalhistory, (iii) previous rheumatoid treatment, (iv)coprescribed drugs, (v) daily dose and duration oftreatment with leflunomide. Known risk factors suchas neurotoxic drugs, diabetes mellitus, alcohol, renalimpairment, dysthyroidy and vasculitis were alsosought.

ANALYSIS PLAN

Descriptive study

First, we sought for the cases of neuropathy anddescribed it. During the follow-up, patients presentingwith symptoms suggesting neurotoxicity such asparesthesia were systematically investigated or inves-tigation were look for in the historical clinical charts ifneeded. Nerve conduction study (NCS) was used toascertain the diagnosis of peripheral neuropathy.Patients who were positive for this examination weredefined as the cases. Results were given as the mean orthe median for continuous variables and percentagefor categorical variables.

Comparative study

A comparison analysis was performed betweenpatients presenting with peripheral neuropathy ‘cases’to the remaining patients of the cohort. All the patientsconsidered were followed for at least 3 weeks, whichcorresponds to the minimal time to onset of peripheralneuropathy in cases.For the univariate analysis, the Fisher exact test was

used to compare the distribution of categoricalvariables and the Student t or Mann Whitney testsfor continuous variables with a significance level at0.05. Statistical analysis was performed using theSTATA statistical package (STATA version 8.0 forMacintosh; StataCorp LP,Texas, USA).

RESULTS

Description of the cohort

One hundred thirteen patients treated with leflunomideduring the study period, mostly with a daily dose of20mg, were included in the study. All patients hadpreviously received a 100mg loading dose for 3 days.Mean age at the start of treatment was 55.7 years(range: 27–81). M/F sex ratiowas 0.45. Mean durationof treatment was 9.7 months (median: 6 months,range: 3 weeks - 31 months). Mean duration of diseaseat treatment onset was 10.1 years (median¼ 8 years,range: 0.5–55). All patients had a Disease ActivityScore (DAS) score higher than 3.2 at the time ofleflunomide treatment.

Description of the cases (Table 1)

During the study period, 8 cases of neuropathy and twocases of worsening of a preexisting neuropathy wereidentified. Paresthesia led to discontinuing the treatmentin one patient, but the case was not retained becauseNCS was not performed. Mean duration of disease attreatment onset was 8.7 years (median¼ 5 years, range:0.5–20). Rheumatoid factor was positive in nine cases.Median age at start of treatment was 69 years (range:57–78). Mean duration of treatment before theneuropathy occurred was 9.8 months (median: 6.6months, range: 3 weeks–29 months). Table 1 Patientswere previously treated with methotrexate in sevencases, gold salts in five, sulfazalazine in four,chloroquine in four and tiopronine in three. The averagenumber of prior DMARDs was 3.4 (range: 0–7).However, none of the 10 cases were treated con-comitantly with neurotoxic DMARDs. Among fourpatients who were diabetics, retinal angiographyperformed in two did not show any diabetic micro-

Copyright # 2006 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2007; 16: 74–78DOI: 10.1002/pds

peripheral neuropathy associated with leflunomide 75

angiopathy. Vasculitis was diagnosed in two patients.However, in all cases, the routine causes of neuropathy(metabolic or immunologic disorders, vitamindeficiency, neoplasia, viral and bacterial serology) wereruled out. In five patients, the neuropathy improved orrecovered clinically, while it stabilized in the otherswhen leflunomide was stopped.

Comparison between neuropathy and non-neuropathy cases (Table 2)

Compared with other patients, neuropathy cases wereolder (69 vs. 54 years, p¼ 0.0006), more often diabetic(30 vs. 2.9%, p¼ 0.009), and more frequently weretaking neurotoxic drugs previously or concomitantly(50% vs. 1.9%, p¼ 0.039). There were no statisticaldifferences for gender, mean duration of disease,median duration of treatment, daily dose of leflunomideor daily dose of corticosteroids (in mg prednisoneequivalent). None of the patients presented withdysthyroidy, renal insufficiency or vasculitis beforetreatment with leflunomide. Possible risk factors(neurotoxic drug or diabetes) were found in 50% ofpatients with neuropathy versus 4% of patients withoutneuropathy (positive predictive value (PPV)¼ 56%;negative predictive value (NPV)¼ 96%). Table 2

DISCUSSION

In this cohort, 10 cases of peripheral neuropathy wereidentified and prescribed, corresponding to 9% of thewithdrawal of leflunomide due to adverse effects.CCompared to the entire cohort, these patients wereolder (69 vs. 54 years), more often diabetics andseemed to have more known risk factors for neuropathysuch as concomitant or previous treatment with apotential neurotoxic drug. To date, six publications12–17

have described cases of peripheral neuropathy associ-ated with leflunomide (almost 100 cases reported), andthere was no report during clinical trials1–3 or duringlong-term follow-up studies of those trials.4–7 One ofthe publication described the nearly 90 cases reported tothe US Food and Drug Administration until 2004.16

Table

1.

Descriptionofneuropathycases(M

ay2000–May

2003)

Patientnumber

12

34

56

78

910

Ageat

treatm

ent

63

66

72

76

78

69

61

78

57

69

Gender

MF

FM

MM

FF

FM

Durationofdisease

(years)

517

16

110

15

35

20

Durationoftreatm

entat

onsetofneuropathy(m

onths)

10

19.1

1.7

19

3.2

112

0.6

29

2

Neurologic

statebefore

leflunomidetreatm

ent

Norm

alnorm

alnorm

alnorm

aldysesthesia

norm

alnorm

alhypoesthesia

norm

alnorm

al

Diabetes

mellitus

no

no

yes

yes

yes

no

yes

no

no

no

Previous/concomitant

neurotoxic

drug(s)

almitrine

atorvastatine

NCS

SAN

MSAN

SAN

MSAN

MSAN

SAN

SAN

MSAN

SAN

SAN

Washout�

yes

no

no

no

no

yes

yes

yes

no

no

Evolution

Stabilization

recovery

improvem

ent

stabilization

stabilization

improvem

ent

improvem

ent

stabilization

improvem

ent

stabilization

M:male,

F:female;

MSAN:motorandsensitiveaxonal

neuropathy;SAN:sensitiveaxonal

neuropathy.

� perform

edwithcolestyramine;

Table 2. Comparison between neuropathy cases and non-neuropathy cases

Neuropathy Non-neuropathy p

Median age 69 54 0.0006Diabetes % 30.0 2.9 0.009Neurotoxic treatment 20.0 1.9 0.039Previous/concomitantrisk factor %

50.0 3.9 <10�3

Copyright # 2006 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2007; 16: 74–78DOI: 10.1002/pds

76 k. martin ET AL.

Differences between trial populations and those treatedin real practice may account for this. For example,several studies have shown that the characteristics ofthe patients treated in real life, their concomitantpathologies, treatment patterns and the safety profile ofthe drug may differ from those in clinical trialsconducted for approval.19–21 Another explanation couldbe studies to date have not mentioned whether NCSwas systematically performed in patients includedin clinical trials and presenting with paresthesia. Ifnot, the prevalence of neuropathy may have beenunderestimated.

Nevertheless, since the launch of leflunomide,several studies of its utilization in real practice havebeen also performed8–11 but none has identified anycases of neuropathy. Our patients appear to be similarto those described in these cases reports: most areolder than 60 years and the duration of the RA diseaseis quite similar (5.8 vs. 7.8 years). One of the casereport patients was concomitantly treated with apotentially neurotoxic drug (statin) and in one casethere was a preexisting neuropathy. The adverse eventresolved after leflunomide withdrawal.12 Carulliet al12 hypothesized that the neuropathy could bedue to neurologic vasculitis induced by leflunomide.However, since no neuromuscular biopsy wasperformed, this hypothesis could not be confirmed.Other evidence for this hypothesis is (i) case reports ofcutaneous vasculitis induced by leflunomide withoutneurologic disorder,22–24 and (ii) the fact that twocases in our cohort presented with aspecific vasculitisdiagnosed by neuromuscular biopsy when the etiologyof the neuropathy was explored.

In our 10 cases, the causal role of leflunomide isdifficult to establish categorically. In five cases theneuropathy clinically resolved or improved after thetreatment was stopped, a finding which tends toincriminate leflunomide. Nevertheless, this improve-ment was only based on clinical evaluation since NCSwas not performed. In two patients, the neuropathywas already present but worsened after leflunomidewas started. In six patients, another potential cause ofneuropathy was present such as diabetes mellitus orconcomitant/previous use of a neurotoxic drug.However, this does not rule out the responsibility ofleflunomide which could have precipitated the clinicalexpression of an infraclinical neuropathy. Even if thecauses of neuropathy were explored, an abdominalscan was not systematically performed in search of aneoplasia and only tumoral markers were assessed.Nevertheless, after a median follow-up of 21 months(range: 17–30 months) none of the subjects has anyevidence of neoplasia. Moreover, since another cause

of neuropathy was possible in a few cases, wecompared patients with or without diagnosis ofperipheral neuropathy in order to establish whether‘these cases of neuropathy’ were different from theother patients. ‘The cases’ were older, more oftendiabetic and had risk factors for neuropathy such asconcomitant or previous treatment with a potentiallyneurotoxic drug.The present findings should be interpreted in the

light of potential methodological limitations. Asalready mentioned, the design of this study allowedcomparisons between patient with and withoutneurological symptoms and, to establish a possiblepattern of patients at risk, but not to conclude to acausal association between leflunomide use andperipheral neuropathy. Only a comparison withanother cohort of unexposed rheumatic patients or agroup treated with another DMARD prescribed in RAcould solve this issue. Such a cohort would have beendifficult to set up since ideally requiring randomiz-ation to control for confounding factors such asprecription channeling and confounding by indication.Another limitation is that neuropathy was confirmedwith NCS. Since there was no systematic NCS inpatients without clinical signs of neuropathy, some ofthe control group patients may have had a subclinicalneuropathy that was not identified, so the rate ofsubclinical neuropathy in this cohort cannot beconclusively established but only the rate of clinicalperipheral neuropathy. Only systematic nerve con-duction study could solve this problem but we do notthink this is feasible in practice due to ethicalconsiderations.

CONCLUSION

Several cases of neuropathy were observed in patientswith RA treated with leflunomide. Patients taking

KEY POINTS

(1) Incidence of peripheral neuropathy (PN) wasalmost 10% in the cohort treated with lefluno-mide.

(2) Treatment duration ranged from 3 weeks to 29months. PN improved or recovered clinicallyin 50% of the patients.

(3) Patients presenting with PN were older, moreoften diabetics and more often treated withpotentially neurotoxic drugs.

(4) The use of leflunomide in such patients shouldbe carefully considered.

Copyright # 2006 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2007; 16: 74–78DOI: 10.1002/pds

peripheral neuropathy associated with leflunomide 77

leflunomide and older than 60 years, presenting withdiabetes mellitus, preexisting neuropathy and previousor concomitant use of neurotoxic drugs should beclosely monitored for the emergence or worsening of aperipheral neuropathy. The use of leflunomide in suchpatients should therefore be carefully considered.

REFERENCES

1. Strand V, Cohen S, Schiff M, et al. Treatment of activerheumatoid arthritis with leflunomide compared with placeboand methotrexate. Leflunomide Rheumatoid Arthritis Investi-gators Group. Arch Intern Med 1999; 159 (21): 2542–2550.

2. Emery P, Breedveld FC, Lemmel EM, et al. A comparison of theefficacy and safety of leflunomide and methotrexate for thetreatment of rheumatoid arthritis. Rheumatology (Oxford) 2000;39 (6): 655–665.

3. Smolen JS, Kalden JR, Scott DL, et al. Efficacy and safety ofleflunomide compared with placebo and sulphasalazine inactive rheumatoid arthritis: a double-blind, randomised, multi-centre trial. European Leflunomide Study Group. Lancet 1999;353 (9149): 259–266.

4. Kalden JR, Scott DL, Smolen JS, et al. Improved functionalability in patients with rheumatoid arthritis—longterm treat-ment with leflunomide versus sulfasalazine. European Leflu-nomide Study Group. J Rheumatol 2001; 28 (9): 1983–1991.

5. Cohen S, Cannon GW, Schiff M, et al. Two-year, blinded,randomized, controlled trial of treatment of active rheumatoidarthritis with leflunomide compared with methotrexate. Utiliz-ation of leflunomide in the treatment of rheumatoid arthritis trialinvestigator group. Arthritis Rheum 2001; 44 (9): 1984–1992.

6. Scott DL, Smolen JS, Kalden JR, et al. Treatment of activerheumatoid arthritis with leflunomide: two year follow up of adouble blind, placebo controlled trial versus sulfasalazine. AnnRheum Dis 2001; 60 (10): 913–923.

7. Kalden JR, Schattenkirchner M, Sorensen H, et al. The efficacyand safety of leflunomide in patients with active rheumatoidarthritis: a five-year followup study. Arthritis Rheum 2003; 48(6): 1513–1520.

8. Brault I, Gossec L, Pharm T, Dougados M. Leflunomidetermination rates in comparison with others DMARDs inRheumatoid Arthritis. Arthritis Rheum 2002; 46: S540.

9. Van Pelt P, Van Schaardenburg D, Bultink I, Dijkmans B. Highdiscontinuation rate of treatment with leflunomide in dailypractice [abstract]. Arthritis Rheum 2001; 44 (suppl 9): S143.

10. MillerM,Morton G, Anderson L, Keroach B, Radis C. One yearcommunity based experience with leflunomide in the treatmentof rheumatoid arthritis [abstract]. Arthritis Rheum 2000;43(Suppl 9): S343.

11. Siva C, Eisen SA, Shepherd R, et al. Leflunomide use during thefirst 33 months after food and drug administration approval:experience with a national cohort of 3,325 patients. ArthritisRheum 2003; 49 (6): 745–751.

12. Carulli MT, Davies UM. Peripheral neuropathy: an unwantedeffect of leflunomide? Rheumatology (Oxford) 2002; 41 (8):952–953.

13. Lormeau C, Vandecandelaere MNG, Schmidt J, Houvenagel E.Neuropathies peripheriques au cours d’un traitement par leflu-nomide: deux nouvelles observations. [French]. Rev Rhum2003; 70: 1001.

14. Martin K, Bentaberry F, Dumoulin C, et al. Neuropathy associ-ated with leflunomide: a case series. Ann Rheum Dis 2005; 64(4): 649–650.

15. Bharadwaj A, Haroon N. Peripheral neuropathy in patients onleflunomide. Rheumatology (Oxford) 2004; 43 (7): 934.

16. Bonnel RA, Graham DJ. Peripheral neuropathy in patientstreated with leflunomide. Clin Pharmacol Ther 2004; 75 (6):580–585.

17. Gabelle A, Antoine JC, Hillaire-Buys D, Coudeyre E, CamuW.[Leflunomide-related severe axonal neuropathy]. Rev Neurol(Paris) 2005; 161 (11): 1106–1109.

18. Moore N, Noblet C, Kreft-Jais C, Lagier G, Ollagnier M, ImbsJL. French pharmacovigilance database system: examples ofutilisation. Therapie 1995; 50 (6): 557–562.

19. Martin K, Begaud B, Latry P, Miremont-Salame G, Fourrier A,Moore N. Differences between clinical trials and postmarketinguse. Br J Clin Pharmacol 2004; 57 (1): 86–92.

20. Wieringa NF, de Graeff PA, van der Werf GT, Vos R. Cardi-ovascular drugs: discrepancies in demographics between pre-and post-registration use. Eur J Clin Pharmacol 1999; 55 (7):537–544.

21. Wieringa NF, Vos R, van der Werf GT, van der Veen WJ, deGraeff PA. Co-morbidity of ‘clinical trial’ versus ‘real-world’patients using cardiovascular drugs. Pharmacoepidemiol DrugSaf 2000; 9 (7): 569–579.

22. Holm EA, Balslev E, Jemec GB. Vasculitis occurring duringleflunomide therapy. Dermatology 2001; 203 (3): 258–259.

23. Chan AT, Bradlow A, McNally J. Leflunomide induced vascu-litis—a dose-response relationship. Rheumatology (Oxford)2003; 42 (3): 492–493.

24. Macdonald J, Zhong T, Lazarescu A, Gan BS, Harth M.Vasculitis associated with the use of leflunomide. J Rheumatol2004; 31 (10): 2076–2078.

Copyright # 2006 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2007; 16: 74–78DOI: 10.1002/pds

78 k. martin ET AL.