You are here

Article Text

Article menu

Download PDFPDF

Clinical and epidemiological research
Extended report
Prediction of clinical non-response to methotrexate treatment in juvenile idiopathic arthritis
  1. Maja Bulatović1,
  2. Marloes W Heijstek1,
  3. E H Pieter Van Dijkhuizen1,
  4. Nico M Wulffraat1,
  5. Saskia M F Pluijm2,
  6. Robert de Jonge3
  1. 1Department of Pediatric Immunology, University Medical Center Utrecht, Wilhelmina Children's Hospital, Utrecht, The Netherlands
  2. 2Department of Public Health, Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlands
  3. 3Department of Clinical Chemistry, Erasmus University Medical Center, Rotterdam, The Netherlands
  1. Correspondence to Maja Bulatovic, University Medical Center Utrecht, Wilhelmina Children's Hospital, Department of Pediatric Immunology, 3508AB Utrecht, The Netherlands; m.bulatovic{at}umcutrecht.nl

Abstract

Objectives Methotrexate (MTX) is a cheap and efficacious drug in juvenile idiopathic arthritis (JIA) treatment. If JIA patients are unresponsive to MTX, early and effective combination treatment with biologicals is required to prevent joint damage. The authors developed a prediction model to identify JIA patients not responding to MTX.

Methods In a cohort of 183 JIA patients, clinical variables and single nucleotide polymorphisms (SNPs) in genes involved in the mechanism of action of MTX were determined at the start of MTX treatment. These variables were used to construct a prediction model for non-response to MTX treatment during the first year of treatment. Non-response to MTX was defined according the American College of Rheumatology paediatric 70 criteria. The prediction model was validated in a cohort of 104 JIA patients.

Results The prediction model included: erythrocyte sedimentation rate and SNPs in genes coding for methionine synthase reductase, multidrug resistance 1 (MDR-1/ABCB1), multidrug resistance protein 1 (MRP-1/ABCC1) and proton-coupled folate transporter (PCFT). The area under the receiver operating characteristics curve (AUC) was 0.72 (95% CI: 0.63 to 0.81). In the validation cohort, the AUC was 0.65 (95% CI: 0.54 to 0.77). The prediction model was transformed into a total risk score (range 0–11). At a cut-off of ≥3, sensitivity was 78%, specificity 49%, positive predictive value was 83% and negative predictive value 41%.

Conclusions The prediction model that we developed and validated combines clinical and genetic variables to identify JIA patients not responding to MTX treatment. This model could assist clinicians in making individualised treatment decisions.

https://doi.org/10.1136/annrheumdis-2011-200942

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Juvenile idiopathic arthritis (JIA) is one of the most common chronic rheumatic diseases in childhood with a reported prevalence between 16 and 159 per 100,000.1 In the treatment of JIA, methotrexate (MTX) is the cornerstone disease-modifying antirheumatic drug. MTX is efficacious in 30%–70% of patients, depending on the JIA subtype.2 ,3 Patients who do not respond or partially respond to MTX are given biologicals such as tumour necrosis factor α (TNFα) inhibitors, interleukin 1 (IL-1) receptor blockers or IL-6 blockers alone or in combination with MTX. The high efficacy of these combination therapies4,–,8 is leading to a tendency to apply biologicals early in the treatment of JIA, even before knowing the patient's response to MTX monotherapy.9,–,11 This is consistent with the need for early effective treatment of JIA, crucial for preventing irreversible joint destruction and long-term disabilities.1 ,4 ,12 However, combination therapy is unnecessary in those patients who could respond to MTX monotherapy, given that the long-term adverse effects of biologicals, particularly TNFα blockers, are largely unknown and could include development of autoimmune phenomena such as inflammatory bowel disease and malignancies such as leukaemia and lymphoma.12,–,16 To ensure that only patients unresponsive to MTX receive early additional treatment with biologicals and those responsive to MTX are spared costly drugs with potentially serious adverse effects, it is crucial to predict those patients who will be unresponsive to MTX monotherapy.

    A prediction model for MTX efficacy was successfully constructed in rheumatoid arthritis (RA).17 However, to date no model has been constructed to predict MTX non-response in JIA. The aim of this study was to develop and validate such a prediction model, using clinical and genetic predictors.

    Methods

    Study design and patients

    Two observational cohort studies were performed at the Wilhelmina Children's Hospital, University Medical Center Utrecht. The derivation cohort, consisting of retrospectively collected patients who had started MTX monotherapy between 1990 and 2006, was used to develop the prediction model. The validation cohort, consisting of prospectively collected patients who had started MTX monotherapy between January 2007 and June 2010, was used to test the external validity of the model.

    Patients, aged 1–18 years, with a confirmed JIA according to the International League of Associations for Rheumatology criteria18 and an available blood sample were eligible for inclusion. Patients were excluded if longitudinal data after start of MTX treatment could not be retrieved and blood samples could not be used to determine the SNPs. Their clinical data on disease characteristics, disease activity and medication use were collected from medical charts at the moment of MTX start and at 3, 6 and 12 months after MTX start. This study was approved by the University Medical Center Utrecht Medical Ethics Committee.

    Assessment of MTX clinical response

    Clinical response to MTX in the first year of treatment was determined using the American College of Rheumatology paediatric 70 (ACR70) criteria for disease activity.19 The validated core-set criteria20 for disease activity were: (1) Physician's global assessment of disease activity on a 10 cm visual analogue scale; (2) Number of active joints, defined by joint swelling or limitation of movement accompanied by pain and tenderness; (3) Number of joints with limitation of movement; (4) Physical functional ability, measured with the Childhood Health Assessment questionnaire (CHAQ) disability on a 0–3 scale;21 (5) Parent or patient assessment of patient's well-being on a 10 cm visual analogue scale; and 6) Erythrocyte sedimentation rate (ESR). Good clinical response to MTX according to ACR70 criteria means at least 70% improvement in at least three of the six core-set criteria, with no more than 30% worsening in more than one of the remaining criteria.

    MTX non-responders were defined as patients who did not satisfy the ACR70 criteria in at least two out of three visits during the first year of MTX treatment. This definition was used since clinical response to MTX is known to fluctuate in a large proportion of patients between different time points in the first year of treatment.22 MTX non-responders also included patients discontinuing MTX and/or switching to anti-TNFα therapy or other biologicals due to insufficient effect of MTX.

    Clinical and genetic variables

    At baseline, JIA was divided into three subtype categories: oligoarticular JIA, polyarticular JIA and other subtypes including systemic, psoriatic and enthesitis-related JIA (table 1). Other disease characteristics, core-set criteria and information on medication use are shown in table 1.

    View this table:
    Table 1

    Prevalence, univariate OR (95% CI) for potential predictors of ACR70 MTX non-response for derivation and validation cohorts at baseline

    The genetic variables, single nucleotide polymorphism (SNPs), were selected based on their involvement in the MTX metabolic pathways, their high polymorphic allele frequency and documented functional effects. DNA for SNP analysis was obtained from whole blood or isolated peripheral blood mononuclear cells. Genomic DNA was isolated using the QIAmp DNA Mini Blood Kit (Qiagen, Venlo, The Netherlands). The following SNPs were determined using real-time PCR with Taqman technique according to protocols provided by the manufacturer (Taqman, Applied Biosystems, Foster City, California, USA): methylenetetrahydrofolate reductase (MTHFR rs1801133 and rs1801131), reduced folate carrier (RFC/SLC19A1 rs1051266), methionine synthase reductase (MTRR rs1801394), inosine triphosphatase (ITPA rs1127354), adenosine monophosphate deaminase (AMPD1 rs17602729), 5-aminoimidazole-4-carboxamide ribonucleotide transformylase (ATIC rs2372536), adenosine-deaminase (ADA rs73598374), adenosine A2A receptor (ADORA2A rs5751876), multidrug resistance 1 (MDR-1/ABCB1 rs1128503, rs1045642, rs2032582), multidrug resistance protein 1-5 (MRP-1/ABCC1 rs35592, rs3784862; MRP-2/ABCC2 rs4148396, rs717620; MRP-3/ABCC3 rs4793665, rs3785911; MRP-4/ABCC4 rs868853, rs2274407; MRP-5/ABCC5 rs2139560), breast cancer resistance protein (BCRP/ABCG2 rs13120400, rs2231142), γ glutamyl hydrolase (GGH rs10106587, rs3758149) and proton-coupled folate transporter (PCFT rs2239907).

    Statistical analysis

    To construct a risk model to predict non-responders to MTX, backward logistic regression analysis was performed in several stages. First, all continuous clinical variables were dichotomised to facilitate the use of the model in daily clinical practice. Second, univariate ORs with 95% CI were calculated (table 1). If two potential predictors correlated (Spearman's r≥0.40), the clinically more relevant or the more significant variable in the univariate analysis was given preference. Third, to obtain the final prediction model, clinical and genetic variables with a p value of ≤0.20 on the log-likelihood test were combined in the multivariate logistic regression analysis.

    To calculate predicted probabilities of being an MTX non-responder, we used the following formula:

    Embedded Image

    where P is the predicted probability of being an MTX non-responder, β0 is the constant and β1, β2 and βp represent the regression coefficients for each of the predictors x1, x2 and xp.

    To evaluate the predictive power of the model, we used the predicted probabilities for MTX non-response to construct a receiver operating characteristic (ROC) curve. The area under the ROC curve (AUC) measures the concordance of predictive values with actual outcomes, with an AUC of 0.5 reflecting no predictive power and an AUC of 1.0 reflecting perfect prediction. To assess whether the models fit the data well, we employed the Hosmer–Lemeshow test.

    To compute the risk score of being an MTX non-responder for individual patients, the regression coefficients (β) of the predictors in the final model were transformed into simple scores that sum up to a total risk score (table 3). Within the total risk score, sensitivity, specificity, positive predictive value and negative predictive value were calculated for several cut-off scores.

    View this table:
    Table 3

    Prediction model and scores for ACR70 MTX non-response

    The prediction model was externally validated in the validation cohort. To do this, we entered the regression coefficients of the predictors obtained from the derivation cohort into the abovementioned formula. This was used to construct a ROC curve for the validation cohort. All statistical analyses were carried out with SPSS V.15.0.0 (SPSS, Chicago, Illinois, USA).

    Results

    Patient characteristics

    183 Patients were included in the derivation cohort after removal of five patients due to missing longitudinal data. Upon eliminating three patients receiving IL-1 receptor blockers at MTX start, 104 patients were included in the validation cohort.

    Baseline characteristics (table 1) did not differ significantly between the cohorts, besides disease duration before MTX start, which was longer in the derivation (median: 1.9 years, IQR: 0.3–7.6) than in the validation cohort (median: 0.8 years, IQR: 0.3–4.5) (p=0.001). MTX starting dose was comparable between the two cohorts, namely 9.4 mg/m2/week in the derivation and 9.8 mg/m2/week in the validation cohort. Within the cohorts, MTX starting dose was equivalent in future responders and non-responders. However, in the derivation, but not in the validation cohort, MTX dose was significantly higher in non-responders compared with responders at 6 months (12.3 mg/m2/week vs 9.3 mg/m2/week) and at 12 months (10.9 mg/m2/week vs 7.1 mg/m2/week) after MTX start.

    In the derivation cohort, after 1 year of treatment, 149 patients (81.4%) were still on MTX, and 27 patients (14.8%) had stopped MTX due to insufficient effect (n=5), disease remission (n=18), gastrointestinal intolerance (n=3) or hepatotoxicity (n=1). In the validation cohort, 99 (92.5%) patients were still receiving MTX after 1 year and eight patients (7.7%) had stopped MTX due to insufficient effect (n=3), disease remission (n=2) and gastrointestinal intolerance (n=3).

    During the first year of treatment, 143 patients (78.1%) in the derivation and 68 patients (65.4%) in the validation cohort were ACR70 non-responders (table 2), while 114 (62.3%) patients in the derivation and 52 (50%) patients in validation cohort were ACR50 non-responders (data not shown). These frequencies corresponded to the frequencies at 6 months after MTX start, which is a commonly used time point to establish MTX efficacy. The ACR70 non-responder frequencies in the validation cohort were similar to those found earlier;3 however, ACR70 frequencies in the derivation cohort were higher, possibly due to significantly longer disease duration before MTX start23 in this cohort.

    View this table:
    Table 2

    ACR70 MTX non-response frequency (%)

    Prediction model for MTX non-responders according to ACR70

    The following variables, univariately associated (p≤0.20) with MTX non-response, were included in the multivariate logistic regression: disease duration, limited joints, ESR, MTRR rs1801394, MDR-1/ABCB1 rs1045642, MRP-1/ABCC1 rs35592 and PCFT rs2239907 (table 1). Variables of the final prediction model consisted of ESR and MTRR rs1801394, MDR-1/ABCB1 rs1045642, MRP-1/ABCC1 rs35592 and PCFT rs2239907 (table 2). The AUC of the prediction model was 0.72 (95% CI: 0.63 to 0.81), indicating that it classified 72% of patients correctly (table 3). The Hosmer–Lemeshow goodness-of-fit test was not statistically significant (p=0.91), indicating that the model fit the data well.

    These predictors were used to test the model in a validation cohort. The AUC of the validation cohort was 0.65 (95% CI: 0.54 to 0.77), indicating that 65% of patients were classified correctly (table 3).

    To enable healthcare professionals to easily use the model, the regression coefficients (β) of the model's predictors, transformed into simple scores, were used to compute an individual risk score for being an MTX non-responder (table 3). This score ranged from 0 to 11 points with a higher score reflecting a higher probability of non-response. The risk score of a patient that has all predictors of the final model is calculated by adding up the constant to the simple scores, assigned to individual predictors: 11 (the constant)+(−2)+(−3)+(−2)+(−2)+(−2), which results in a risk score of 0. If all predictors are present, the probability of non-response is 0.42. On the other hand, the risk score of a patient having no predictors would be equal to the constant of 11. If no predictors are present, the probability of non-response is 0.98. Within the 0–11 range, the diagnostic accuracy of different cut-offs for predicting the risk of being an MTX non-responder was evaluated by computing the corresponding sensitivity, specificity, positive predictive value and negative predictive value (table 4).

    View this table:
    Table 4

    Diagnostic parameters for various risk score cut-offs predicting ACR70 MTX non-response

    Our goal was to correctly identify as many future MTX non-responders as possible (high sensitivity), while attempting to avoid misidentification of MTX responders as MTX non-responders as much as possible (reasonable specificity). In the derivation cohort, this was reached at the cut-off ≥3, where 98 of 125 (78%) MTX non-responders and 19 of 39 (49%) MTX responders were identified correctly; 27 non-responders were classified as responders (false negatives) and 20 responders were classified as non-responders (false positives) (table 4). Similarly, in the validation cohort, at the cut-off ≥3, 48 (79%) of 61 MTX non-responders were identified correctly, whereas nine (26%) of 34 MTX responders were identified correctly; 13 non-responders were classified as responders (false negatives) and 25 responders were classified as non-responders (false positives).

    Discussion

    We developed and validated a prediction model for clinical non-response in two large JIA cohorts consisting of ESR and four SNPs in the MTRR, MDR-1/ABCB1, MRP-1/ABCC1 and PCFT genes. The model classified 72% of patients correctly in the derivation and 65% in the validation cohort.

    To our knowledge, no previous studies have constructed a model to predict MTX non-response in JIA. Several studies did report associations of MTX non-response in JIA with polyarticular disease, longer disease duration, ANA negativity and a higher level of disability.24 ,25 In our study, longer disease duration and ANA negativity were univariately associated with MTX non-response, although not significantly. Moreover, extended oligoarticular JIA subtype was associated with MTX response.25 However, we and others26 ,27 observed equal MTX response rates among different JIA subtypes. Therefore, in the present study JIA subtype was not a predictor of MTX non-response. Furthermore, no effect modification was detected upon restricting the analysis to the more prevalent oligo and polyarthritis subtypes since the prediction model preserved its predictive power (AUC: 0.72, 95% CI: 0.61 to 0.82).

    An MTX efficacy prediction model was constructed in RA, classifying 85% of patients correctly.17 This model contained four clinical variables and four SNPs encoding AMPD1, ATIC, ITPA and MTHFD1 (methylenetetrahydrofolate dehydrogenase) enzymes. Despite differences in definitions of response and in demographics of RA and JIA patients, inclusion of SNPs was essential for adequate prediction of MTX non-response in both models. Our prediction model with ESR only yielded a poor AUC of 0.59 (95% CI: 0.49 to 0.69), whereas the addition of SNPs raised the AUC to 0.72. Therefore, SNPs were crucial for a good prediction of MTX non-responders in JIA.

    The goal of our model is to correctly identify future non-responders who can be given early additional treatment with biologicals, and simultaneously to keep misidentification of future responders as non-responders to a minimum. This goal stems from the following important changes in treatment mentality of paediatric rheumatologists over the past years, prompted by the need to establish early disease control to prevent irreversible joint damage. First, paediatric rheumatologists no longer consider MTX response according to ACR30 or ACR50 sufficient, but judge it to be good only if patients satisfy the more stringent ACR70 criteria.28,–,30 Furthermore, they consider patients MTX responders if they satisfy these criteria already within 3 months after MTX start. These changes in treatment mentality have resulted in a lower threshold to start early combination treatment with biologicals. Although very effective, biologicals potentially carry a heightened risk of malignancies and inflammatory bowel disease.12,–,16 To address these risks, while considering it crucial to adequately treat MTX non-responders as early as possible with biologicals and at same time restrict their use to those patients who really need them, we selected a cut-off ≥3 as the optimal score. Using this cut-off in the derivation cohort would allow 98 (78%) of 125 non-responders to receive early additional treatment with biologicals, and spare 19 (49%) of 39 patients, identified as responders, from receiving them. In the validation cohort, 79% of non-responders would be given timely biological treatment, whereas 26% of patients identified as responders would be spared from receiving them (table 4). Although the sensitivity at this cut-off was the same for both cohorts, the specificity was considerably lower in the validation cohort (49% vs 26%), which is due to its relatively small size.

    The choice of a cut-off, however, depends on the clinical goal. A cut-off ≥6 could be chosen, if clinicians use the prediction model primarily to select as many responders as possible, while avoiding misidentification of non-responders as responders. At this cut-off, 34 of 39 (87%) MTX responders were identified correctly, while 55 of 125 (44%) MTX non-responders were identified correctly. Similar diagnostic parameters were obtained in the validation cohort (Table 4).

    Our model was constructed for ACR70 non-responders in at least two of three visits during the first year of treatment, due to known fluctuations in MTX (non-)response during the first year.22 Nevertheless, the model had an equally strong predictive power for ACR70 non-responders (AUC=0.71, 95% CI: 0.62 to 0.80) at 6 months after MTX start. Depending on the clinician's preference, the model could also be applied for a less stringent ACR50 non-response, since its predictive power was strong both in the first year of treatment (AUC=0.70, 95% CI: 0.61 to 0.77) and at 6 months after MTX start (AUC=0.72, 95% CI: 0.63 to 0.80).

    Further studies are needed to evaluate the effect of these SNPs on enzyme activity and transporter function. As we and others have shown, the non-synonymous rs1045642 SNP in the MDR-1/ABCB1 efflux transporter gene was associated with a higher probability of good clinical response to MTX.31 The synonymous rs35592 SNP in another MRP-1/ABCC1 efflux transporter gene has been associated with higher risk of MTX non-response in psoriasis patients,32 whereas here this SNP was associated with a lower risk of non-response. The synonymous PCFT rs2239907 SNP, whose protein is an influx transporter, has not been described earlier in relation to MTX efficacy in arthritis. Finally, the non-synonymous MTRR rs1801394 SNP was associated with decreased MTX sensitivity in acute lymphoblastic leukaemia,33 whereas in our JIA cohorts it conferred a decreased risk of MTX non-response.

    A limitation of the model is its moderate predictive power of 65% in the relatively small validation cohort. This can impede its direct clinical use, indicating the need for further refinement. Therefore, to confirm the model's clinical applicability, validation will be performed in a large international cohort prior to its implementation in daily clinical practice. Pharmacogenetic testing may also challenge the model's application in daily clinical practice. Nevertheless, we show that SNPs are indispensable to adequately predict MTX non-responders in our JIA cohorts. Furthermore, such testing is becoming routinely available and less expensive.

    Our model predicted and validated MTX non-response in two JIA cohorts by combining clinical and genetic variables. The model offers the promise of personalised treatment in JIA where patients unresponsive to MTX monotherapy will promptly receive additional treatment with biologicals and those destined to be MTX responders will not. Therefore, we will implement the model in daily clinical practice to establish whether its use will result in reduction of disease activity and better disease control in JIA patients.

    References

    1. ↵
      1. Ravelli A,
      2. Martini A
      . Juvenile idiopathic arthritis. Lancet 2007;369:767–78.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Foell D,
      2. Wulffraat N,
      3. Wedderburn LR,
      4. et al
      . Methotrexate withdrawal at 6 vs 12 months in juvenile idiopathic arthritis in remission: a randomized clinical trial. JAMA 2010;303:1266–73.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Ruperto N,
      2. Murray KJ,
      3. Gerloni V,
      4. et al
      . A randomized trial of parenteral methotrexate comparing an intermediate dose with a higher dose in children with juvenile idiopathic arthritis who failed to respond to standard doses of methotrexate. Arthritis Rheum 2004;50:2191–201.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Hashkes PJ,
      2. Laxer RM
      . Medical treatment of juvenile idiopathic arthritis. JAMA 2005;294:1671–84.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Lovell DJ,
      2. Giannini EH,
      3. Reiff A,
      4. et al
      . Etanercept in children with polyarticular juvenile rheumatoid arthritis. Pediatric Rheumatology Collaborative Study Group. N Engl J Med 2000;342:763–9.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Ruperto N,
      2. Lovell DJ,
      3. Cuttica R,
      4. et al
      . A randomized, placebo-controlled trial of infliximab plus methotrexate for the treatment of polyarticular-course juvenile rheumatoid arthritis. Arthritis Rheum 2007;56:3096–106.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Swart JF,
      2. Barug D,
      3. Möhlmann M,
      4. et al
      . The efficacy and safety of interleukin-1-receptor antagonist anakinra in the treatment of systemic juvenile idiopathic arthritis. Expert Opin Biol Ther 2010;10:1743–52.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Prince FH,
      2. Twilt M,
      3. ten Cate R,
      4. et al
      . Long-term follow-up on effectiveness and safety of etanercept in juvenile idiopathic arthritis: the Dutch national register. Ann Rheum Dis 2009;68:635–41.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Beresford MW,
      2. Baildam EM
      . New advances in the management of juvenile idiopathic arthritis–2: the era of biologicals. Arch Dis Child Educ Pract Ed 2009;94:151–6.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Mircic M,
      2. Kavanaugh A
      . Inhibition of IL6 in rheumatoid arthritis and juvenile idiopathic arthritis. Exp Cell Res 2011;317:1286–92.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Tynjälä P,
      2. Vähäsalo P,
      3. Tarkiainen M,
      4. et al
      . Aggressive combination drug therapy in very early polyarticular juvenile idiopathic arthritis (ACUTE-JIA): a multicentre randomised open-label clinical trial. Ann Rheum Dis 2011;70:1605–12.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Hashkes PJ,
      2. Uziel Y,
      3. Laxer RM
      . The safety profile of biologic therapies for juvenile idiopathic arthritis. Nat Rev Rheumatol 2010;6:561–71.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Dallocchio A,
      2. Canioni D,
      3. Ruemmele F,
      4. et al
      . Occurrence of inflammatory bowel disease during treatment of juvenile idiopathic arthritis with etanercept: a French retrospective study. Rheumatology (Oxford) 2010;49:1694–8.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Diak P,
      2. Siegel J,
      3. La Grenade L,
      4. et al
      . Tumor necrosis factor alpha blockers and malignancy in children: forty-eight cases reported to the Food and Drug Administration. Arthritis Rheum 2010;62:2517–24.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Horneff G,
      2. De Bock F,
      3. Foeldvari I,
      4. et al
      . Safety and efficacy of combination of etanercept and methotrexate compared to treatment with etanercept only in patients with juvenile idiopathic arthritis (JIA): preliminary data from the German JIA Registry. Ann Rheum Dis 2009;68:519–25.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Ruperto N,
      2. Martini A
      . Pediatric rheumatology: JIA, treatment and possible risk of malignancies. Nat Rev Rheumatol 2011;7:6–7.
      OpenUrlCrossRefPubMed
    17. ↵
      1. Wessels JA,
      2. van der Kooij SM,
      3. le Cessie S,
      4. et al
      . A clinical pharmacogenetic model to predict the efficacy of methotrexate monotherapy in recent-onset rheumatoid arthritis. Arthritis Rheum 2007;56:1765–75.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Petty RE,
      2. Southwood TR,
      3. Manners P,
      4. et al
      . International League of Associations for Rheumatology classification of juvenile idiopathic arthritis: second revision, Edmonton, 2001. J Rheumatol 2004;31:390–2.
      OpenUrlFREE Full Text
    19. ↵
      1. Ruperto N,
      2. Ravelli A,
      3. Falcini F,
      4. et al
      . Performance of the preliminary definition of improvement in juvenile chronic arthritis patients treated with methotrexate. Italian Pediatric Rheumatology Study Group. Ann Rheum Dis 1998;57:38–41.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Giannini EH,
      2. Ruperto N,
      3. Ravelli A,
      4. et al
      . Preliminary definition of improvement in juvenile arthritis. Arthritis Rheum 1997;40:1202–9.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Wulffraat N,
      2. van der Net JJ,
      3. Ruperto N,
      4. et al
      . The Dutch version of the Childhood Health Assessment Questionnaire (CHAQ) and the Child Health Questionnaire (CHQ). Clin Exp Rheumatol 2001;19(4 Suppl 23):S111–5.
      OpenUrlPubMedWeb of Science
    22. ↵
      1. de Rotte MC,
      2. Luime JJ,
      3. Bulatovic M,
      4. et al
      . Do snapshot statistics fool us in MTX pharmacogenetic studies in arthritis research? Rheumatology (Oxford) 2010;49:1200–1.
      OpenUrlFREE Full Text
    23. ↵
      1. Albers HM,
      2. Wessels JA,
      3. van der Straaten RJ,
      4. et al
      . Time to treatment as an important factor for the response to methotrexate in juvenile idiopathic arthritis. Arthritis Rheum 2009;61:46–51.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Vilca I,
      2. Munitis PG,
      3. Pistorio A,
      4. et al
      . Predictors of poor response to methotrexate in polyarticular-course juvenile idiopathic arthritis: analysis of the PRINTO methotrexate trial. Ann Rheum Dis 2010;69:1479–83.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Ravelli A,
      2. Viola S,
      3. Migliavacca D,
      4. et al
      . The extended oligoarticular subtype is the best predictor of methotrexate efficacy in juvenile idiopathic arthritis. J Pediatr 1999;135:316–20.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Woo P,
      2. Southwood TR,
      3. Prieur AM,
      4. et al
      . Randomized, placebo-controlled, crossover trial of low-dose oral methotrexate in children with extended oligoarticular or systemic arthritis. Arthritis Rheum 2000;43:1849–57.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Giannini EH,
      2. Brewer EJ,
      3. Kuzmina N,
      4. et al
      . Methotrexate in resistant juvenile rheumatoid arthritis. Results of the U.S.A.-U.S.S.R. double-blind, placebo-controlled trial. The Pediatric Rheumatology Collaborative Study Group and The Cooperative Children's Study Group. N Engl J Med 1992;326:1043–9.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Ravelli A,
      2. Ramenghi B,
      3. Di Fuccia G,
      4. et al
      . Factors associated with response to methotrexate in systemic-onset juvenile chronic arthritis. Acta Paediatr 1994;83:428–32.
      OpenUrlPubMedWeb of Science
    29. ↵
      1. Bartoli M,
      2. Tarò M,
      3. Magni-Manzoni S,
      4. et al
      . The magnitude of early response to methotrexate therapy predicts long-term outcome of patients with juvenile idiopathic arthritis. Ann Rheum Dis 2008;67:370–4.
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Albers HM,
      2. Brinkman DM,
      3. Kamphuis SS,
      4. et al
      . Clinical course and prognostic value of disease activity in the first two years in different subtypes of juvenile idiopathic arthritis. Arthritis Care Res (Hoboken) 2010;62:204–12.
      OpenUrlPubMedWeb of Science
    31. ↵
      1. Pawlik A,
      2. Wrzesniewska J,
      3. Fiedorowicz-Fabrycy I,
      4. et al
      . The MDR1 3435 polymorphism in patients with rheumatoid arthritis. Int J Clin Pharmacol Ther 2004;42:496–503.
      OpenUrlPubMedWeb of Science
    32. ↵
      1. Warren RB,
      2. Smith RL,
      3. Campalani E,
      4. et al
      . Genetic variation in efflux transporters influences outcome to methotrexate therapy in patients with psoriasis. J Invest Dermatol 2008;128:1925–9.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. de Jonge R,
      2. Hooijberg JH,
      3. van Zelst BD,
      4. et al
      . Effect of polymorphisms in folate-related genes on in vitro methotrexate sensitivity in pediatric acute lymphoblastic leukemia. Blood 2005;106:717–20.
      OpenUrlAbstract/FREE Full Text

    Footnotes

    • Funding Funding provided by the Dutch Arthritis Association and MEDAC GmbH, Germany.

    • Correction notice This article has been corrected since it was published Online First. Corrections were made to Tables 1 and 3.

    • Competing interests None.

    • Ethics approval Approval provided by the Medical Ethics Committee of the University Medical Centre Utrecht (UMCU).

    • Provenance and peer review Not commissioned; externally peer reviewed.