ARTHRITIS & RHEUMATISMVol. 63, No. 10, October 2011, pp 31633168DOI 10.1002/art.30502 2011, American College of Rheumatology
Clinical Outcomes After Withdrawal of AntiTumor Necrosis Factor Therapy inPatients With Juvenile Idiopathic Arthritis: A Twelve-Year Experience
Kevin Baszis, Jane Garbutt, Dana Toib, Jingnan Mao, Allison King, Andrew White, and Anthony French
Objective. To estimate the length of time to dis-ease flare and the likelihood of achieving clinical remis-sion after discontinuation of treatment with tumornecrosis factor (TNF) blockers in patients withjuvenile idiopathic arthritis (JIA).
Methods. We conducted a retrospective chartreview in a cohort of patients with JIA treated withTNF inhibitors between January 1, 1998 and Novem-ber 1, 2009. Demographic information, laboratory data,and medication exposure were extracted using a stan-dardized tool. Outcomes of interest were based onpreliminary criteria for remission in JIA.
Results. One hundred seventy-one patients with255 discrete episodes of anti-TNF treatment werereviewed. The median duration of patient observationwas 59.7 months (range 5.8211.2 months). Amongpatients in whom disease was inactive after discontinu-ation of anti-TNF therapy, 50% had persistently inac-tive disease at 6 months, and 33% had clinical remissionat 12 months. The median duration of anti-TNFtherapy after inactive disease was obtained was 6.1months (range 067.9 months). No significant associa-tion was observed between the time to disease flare aftercessation of treatment with TNF antagonists and thelength of time from the diagnosis of JIA to the initiationof anti-TNF therapy, the duration of therapy followingthe onset of inactive disease, or the total duration oftreatment with TNF antagonists prior to discontinua-tion. The category of JIA, sex, and age at diagnosis werenot associated with the risk of relapse.
Conclusion. One-third of patients with JIA can
successfully undergo withdrawal of treatment withTNF antagonists and be spared the cost and potentialmorbidity of treatment for at least 12 months. Furtherstudies are needed to identify factors to accuratelyidentify these patients.
Antitumor necrosis factor (anti-TNF) drugs,including etanercept, infliximab, and adalimumab, haverevolutionized the treatment of juvenile idiopathic ar-thritis (JIA), and disease remission has become anattainable goal (1). To realize this goal, treatmentproviders must balance the potential of adverse effectsfrom prolonged treatment with the risk of disease flareafter withdrawal of therapy.
The risk of immunosuppression and infections,particularly mycobacterial and fungal infections, associ-ated with TNF blockade is well established (2). Arecent US Food and Drug Administration analysis of 10years of postmarketing surveillance highlighted the po-tential risk of secondary malignancy, resulting in a boxedwarning for these medications (3). That report inducedanxiety in patients, parents, and healthcare providersand offered additional incentive to explore withdrawingpatients from anti-TNF therapy once inactive disease isachieved. Minimizing exposure to TNF antagonistscould lead to decreased adverse effects and costs.
Currently, no guidelines exist to outline appro-priate withdrawal of TNF inhibitors once inactivedisease status has been achieved. Treatment practicesvary and remain provider-dependent, and many patientswho experience disease remission are treated indefi-nitely, often for many years, unless adverse outcomesoccur (4,5). Several studies have investigated factors forsuccessful withdrawal of TNF inhibitors in adult pa-tients with rheumatoid arthritis (6,7), but the pediatricliterature remains limited (8,9). The goal of the currentstudy was to examine the effect of withdrawing TNFantagonists while continuing treatment with disease-modifying antirheumatic drugs (DMARDs) such asmethotrexate, leflunomide, or sulfasalazine. The pri-
Kevin Baszis, MD, Jane Garbutt, MB ChB, Dana Toib, MD,Jingnan Mao, MS, Allison King, MD, MPH, Andrew White, MD,Anthony French, MD, PhD: Washington University School of Medi-cine, St. Louis, Missouri.
Address correspondence to Kevin Baszis, MD, WashingtonUniversity School of Medicine, Department of Pediatrics, Division ofRheumatology, Campus Box 8116, One Childrens Place, St. Louis,MO 63110. E-mail: firstname.lastname@example.org.
Submitted for publication September 29, 2010; accepted inrevised form June 9, 2011.
mary objective was to inform the clinical management ofpatients with JIA receiving anti-TNF therapy. Specifi-cally, we sought to determine the length of time todisease flare and the likelihood of clinical remissionafter discontinuation of TNF blockers in patients withJIA.
PATIENTS AND METHODS
This retrospective cohort study was approved by theWashington University Institutional Review Board. The studypopulation included all patients with JIA who were treated atthe Washington University division of Pediatric Rheumatology(2 clinical sites) between January 1, 1998 and November 1,2009.
Our general practice is to initiate treatment with TNFantagonists in JIA patients in whom disease is refractory to atleast 3 months of therapy with methotrexate and nonsteroidalantiinflammatory drugs and who are not candidates for treat-ment with intraarticular steroids. Patients in whom diseaseremained inactive for 612 months while they were receivinganti-TNF therapy are considered candidates for discontinu-ation of TNF antagonists. These agents are typically stoppedabruptly rather than by decreasing the dose or dosing interval,and the majority continue to receive DMARDs.
A total of 241 patients who were treated with TNFantagonists were screened for inclusion in this retrospectivestudy. Exclusion criteria were as follows: noncompliance withtherapy, limited followup or observation deemed inadequatefor extraction of the appropriate data, and concomitant treat-ment with other biologic agents during anti-TNF therapy.Noncompliance was defined as cessation of therapy beforephysician-recommended cessation (i.e., before an adequatetherapeutic trial [23 months in most cases]) or cessation oftherapy during active disease. Limited followup was defined asan insufficient number of encounters deemed necessary forobservation of disease activity and response to therapy (en-counters less often than every 34 months or transfer toanother center). The percentage of missing data in the in-cluded patient records was minimal.
Seventy patients were excluded due to incompletemedical records (the diagnosis was made at another center orthe patients care was transferred to another site, with noopportunity for followup [n 42]), diagnosis/symptom onsetafter the patients sixteenth birthday (n 13), noncompliancelimiting adequate data points (n 11), development ofmacrophage activation syndrome (n 2), ambiguity in diag-nosis (n 1), and limited drug exposure due to parentalconcerns (n 1).
Inclusion criteria were the following: fulfillment of the2001 International League of Associations for Rheumatologycriteria for the classification of JIA (10), treatment with TNFantagonists (etanercept, infliximab, adalimumab) during thestudy period, and onset of arthritis symptoms before age 16years. Patients continuing treatment with methotrexate, le-flunomide, sulfasalazine, or other DMARDs after cessation ofTNF antagonists were included. After stopping anti-TNFtherapy during inactive disease, no patient received cortico-
steroids or biologic therapies. Chart review was performed bya pediatric rheumatologist, using a standardized tool.
Data extracted by chart review included demographicinformation, baseline serology and acute-phase reactant levels,concurrent medication use, and dates of diagnosis, therapy,and remission/flare. The category of JIA was recorded asoligoarticular, extended oligoarticular, rheumatoid factor(RF)negative polyarticular, RF-positive polyarticular, system-ic, psoriatic, enthesitis-related, or undifferentiated. To verifythe reproducibility of data extraction, 10% of the charts werereviewed by a second pediatric rheumatologist.
The criteria described by Wallace et al were used forthe study definitions (11). Inactive disease was defined as noactive arthritis or uveitis; a physicians global assessmentindicating no disease activity; no fever, rash, serositis, spleno-megaly, or lymphadenopathy; and no elevated erythrocytesedimentation rate or C-reactive protein level attributable toJIA. Clinical remission on medication was defined as 6 contin-uous months of inactive disease on anti-TNF therapy. Forthis study, clinical remission off medication was defined as
Table 1. Characteristics of the 171 JIA patients treated with anti-TNF therapy*
CharacteristicFemale sex 24 (14)Age at disease onset, median (range) years 7.1 (0.818.5)JIA category
Persistent oligoarticular 20 (12)Extended oligoarticular 19 (11)RF-negative polyarticular 56 (33)RF-positive polyarticular 24 (14)Systemic-onset 21 (12)Psoriatic 6 (4)Enthesitis-related 21 (12)Undifferentiated 4 (2)
ANA statusPositive 77 (45)Negative 86 (50)Not determined 8 (5)
Uveitis present 27 (16)ESR at diagnosis, median (range) mm/hour 42 (3111)CRP at diagnosis, median (range) mg/liter 18.3 (0.1250)Anti-TNF agent used during treatment course
Etanercept 129 (75)Infliximab 68 (40)Adalimumab 13 (8)
Other therapies used during disease courseMethotrexate 155 (91)Sulfasalazine 25 (15)Leflunomide 24 (14)Intraarticular steroid injection 73 (43)
No. of injections per patient, median(range)
Time from diagnosis to initiation of anti-TNFtherapy, median (range) months
Time from diagnosis to last recorded visit,median (range) months
* Except where indicated otherwise, values are