Time to treatment as an important factor for the response to methotrexate in juvenile idiopathic arthritis




Methotrexate (MTX) is the most commonly used disease-modifying antirheumatic drug in juvenile idiopathic arthritis (JIA). Currently, individual response to MTX cannot be reliably predicted. Identification of clinical and genetic factors that influence the response to MTX could be helpful in realizing the optimal treatment for individual patients.


A cohort of 128 JIA patients treated with MTX were studied retrospectively. Eleven clinical parameters and genotypes of 6 single nucleotide polymorphisms in 5 genes related to the mechanism of action of MTX were compared between MTX responders and nonresponders using a multivariate regression analysis.


The time from diagnosis to start of MTX treatment, physician's global assessment at baseline, and the starting dose were significantly associated with the response to MTX at 6 months after initiation. Patients with a shorter time from diagnosis to start of MTX and a higher disease activity according to the physician but with a lower MTX dose showed an increased response. The effect of the starting dose on MTX response seemed to be mainly due to the influence of the systemic JIA subtype. The time from diagnosis to start of MTX treatment and physician's global assessment at baseline were highly correlated. Therefore, the precise effect size of each independent variable could not be determined.


In children with JIA, the time from diagnosis to start of MTX appears to be an important factor for MTX response. Our results suggest that an earlier start of MTX treatment will lead to an increased response.


Juvenile idiopathic arthritis (JIA) is the most common chronic rheumatic inflammatory disorder in childhood, with an incidence of approximately 10 per 100,000 children (1). JIA is defined as arthritis of unknown etiology that persists for more than 6 weeks with an onset before the age of 16 years. Seven different subtypes have been defined according to the criteria of the International League of Associations for Rheumatology (ILAR) (2).

Methotrexate (MTX) is the most commonly used disease-modifying antirheumatic drug (DMARD) in JIA, especially in the treatment of polyarticular JIA (3). The efficacy of MTX has been demonstrated in randomized, placebo-controlled trials and in subsequent clinical use (4, 5). The response rate of MTX, prescribed in a weekly standard dosage of 8.5–12.5 mg/m2, is ∼65% at 4–6 months after initiation of therapy (4, 6–8).

The precise mechanism of action of MTX remains unclear, although it is thought that MTX inhibits the de novo synthesis of purine and pyrimidine, essential components of DNA and RNA (9, 10). In this way, it inhibits the proliferation of cells, including T lymphocytes. Additionally, it has been shown that the antiinflammatory effects of MTX are mediated by an increased adenosine release. More recently, research groups have reported that genetically based differences contribute to MTX efficacy because polymorphisms in genes involved in the purine and pyrimidine synthesis have been associated with response to MTX in JIA and rheumatoid arthritis (RA) (11–13).

However, in JIA, reliable predictors for the response to MTX are yet unknown. Factors identifying JIA patients with a high likelihood of responding to MTX therapy would be very helpful for achieving the optimal treatment for individual patients in an early stage of the disease, thereby preventing damage to the joints in the long term. Therefore, the aim of this study was to identify clinical and genetic factors that are associated with the response to MTX in patients with JIA.


Patient population.

The patients in this analysis were a retrospectively observed cohort of children diagnosed with JIA who were recruited from 4 pediatric rheumatology referral centers in The Netherlands, Belgium, and Germany. Clinical data were collected from 347 patients of whom DNA was available. Of these patients, 152 (44%) were treated with MTX and 128 fulfilled the inclusion criteria for the study. Twenty-four patients were excluded for the following reasons: age >18 years at the start of MTX (n = 5), started MTX <6 months earlier (n = 5), use of MTX because of JIA-associated uveitis (n = 3), and missing followup data (n = 11). No statistically significant differences were found regarding subtype, age at onset, and sex between the 128 genotyped patients and the total group of patients receiving MTX (P > 0.05). Patients with undifferentiated JIA (n = 5), JIA with enthesitis (n = 4), and psoriatic arthritis (n = 1) were grouped together in the subgroup “other JIA.” A total of 97% of the patients were of European white ethnicity based on self-report. Written informed consent was obtained from all patients and/or parents together with approval from each institution's medical ethics board.

Clinical data.

Demographic and clinical data together with detailed information about the use of MTX and comedication were collected from the patients' charts. At the time of start of MTX and at 6 months after initiation of MTX, the following parameters were scored: the physician's global assessment of disease activity, the amount of joints with arthritis (defined by swelling not due to bony enlargement, or if no swelling was present, limitation of motion accompanied by either pain on motion and/or tenderness; 32-joint count), and erythrocyte sedimentation rate (ESR). The physician's global assessment was scored on a 5-point scale (1 = no, 2 = mild, 3 = moderate, 4 = severe, and 5 = very severe activity). In addition, the joint score was divided into the following categories: 0 = no arthritis; 1 = monarthritis (1 joint); 2 = oligoarthritis (2–4 joints); 3 = polyarthritis (5–10 joints); 4 = severe polyarthritis (>10 joints); and in systemic JIA patients, an additional category 5 was used when systemic features were present.

Definition of response.

The response to MTX was defined as follows: improvement in physician's global assessment of 1 or more categories together with an equal or improved joint score measured from baseline to 6 months after the start of MTX. ESR was not incorporated in the definition of response because of the large number of missing data. Patients were considered nonresponders to MTX if they did not fulfill the response criteria.


Six single-nucleotide polymorphisms (SNPs) in 5 candidate genes, related to the mechanism of action of MTX, were selected, taking the following criteria into consideration: validated SNP, SNP preferably causing nonsynonymous amino acid change, indications for clinical relevance from previous publications (11–13), and a preferred minimal genotype frequency of ∼10%. These SNPs were located in the genes adenosine monophosphate deaminase (AMPD1) (34C>T; rs17602729), aminoimidazole carboxamide ribonucleotide transformylase (ATIC) (347C>G; rs2372536), inosine triphosphate pyrophosphatase (ITPA) (94C>A; rs41320251), methylenetetrahydrofolate reductase (MTHFR) (677C>T; rs1801133 and 1298A>C; rs1801131), and methylenetetrahydrofolate dehydrogenase (MTHFD1) (1958G>A; rs2236225). Genotyping was performed using real-time polymerase chain reaction with TaqMan technique according to protocols provided by the manufacturer (TaqMan, Applied Biosystems, Foster City, CA). Five to ten percent of samples were genotyped in duplicate. The mean for overall success rate was 96%. All 6 SNP genotype frequencies showed Hardy-Weinberg equilibrium.

Statistical analysis.

Clinical variables considered relevant for the response to MTX at 6 months after initiation were as follows: subtype of JIA, age at start of MTX, time from diagnosis to start of MTX (time-to-start MTX, in months), disease activity at start of MTX (physician's global score, joint score), ESR, starting dose of MTX (mg per body surface area per week), use of intraarticular steroids and/or sulfasalazine (SSZ) and/or other DMARDs before MTX (yes/no), use of systemic steroids before MTX (yes/no), use of systemic steroids during MTX (yes/no), and use of SSZ during MTX (yes/no). These variables were compared between responders and nonresponders by Student's t-test, Mann-Whitney U test, or chi-square test, depending on the tested variable. Differences in genotype distribution between responders and nonresponders were tested in a 2 × 2 cross-tabulation by carrier analysis with a 2-sided chi-square test. MTHFR 677C>T and MTHFR 1298A>C were only tested as number of copies of the MTHFR1298A-677C haplotype. With sample sizes of 55 nonresponders and 73 responders, an increase in frequency of 2 haplotype copies (MTHFR1298A-677C) from 12% to 34% could be detected with 80% power and 95% confidence. Variables with a P value <0.1 between responders and nonresponders were considered relevant for influencing the response to MTX either by a true effect or by confounding. Therefore, variables with a P value <0.1 were included in the multiple binary logistic regression analysis with response as the dependent variable. Additionally, the univariate odds ratios (ORs; with 95% confidence intervals [95% CIs]) of these variables were calculated to illustrate the confounding effect of the different variables. ESR was not included in the multivariate analysis because of the large number of missing data. All statistical analyses were performed using SPSS 14.0 (SPSS, Chicago. IL). Variables with a P value <0.05 in the multivariate regression analysis were considered statistically significant.


Description of the patient population.

The clinical and demographic characteristics for responders and nonresponders are presented in Table 1. In our cohort (n = 128), the response rate at 6 months after initiation of MTX was 44% in patients with persistent oligoarthritis, 69% in patients with extended oligoarthritis, 61% in patients with rheumatoid factor (RF)–negative polyarthritis, 82% in patients with RF-positive polyarthritis, and 32% in patients with systemic JIA, whereas the response rate in the overall JIA population with these subtypes combined was 57% (Table 1). In addition, the comparison of the clinical and demographic characteristics of MTX responders and MTX nonresponders is shown in Table 1.

Table 1. Clinical and demographic characteristics of 128 JIA patients, according to their response to MTX*
 MTX nonresponderMTX responderP
  • *

    Values are the number of patients (percentage) unless otherwise indicated. JIA = juvenile idiopathic arthritis; MTX = methotrexate; RF = rheumatoid factor; SSZ = sulfasalazine; IAS = intraarticular steroid; DMARD = disease-modifying antirheumatic drug; ESR = erythrocyte sedimentation rate.

  • P value of different statistical tests comparing the clinical variables between the nonresponders and responders.

  • Diagnosis of JIA according to the revised International League of Associations for Rheumatology criteria (2).

  • §

    Pearson's chi-square test.

  • ¶, §

    Other JIA consists of undifferentiated JIA (n = 5), JIA with enthesitis (n = 4), and psoriatic arthritis (n = 1).

  • #

    Student's t-test.

  • **

    Mann-Whitney U test.

  • ††

    Not included in further analysis because of the large number of missing data.

  • ‡‡

    Linear-by-linear association.

Total group of JIA patients55 (43)73 (57) 
Diagnosis  0.050§
 Persistent oligoarthritis10 (56)8 (44) 
 Extended oligoarthritis8 (31)18 (69) 
 RF-negative polyarthritis17 (39)27 (61) 
 RF-positive polyarthritis2 (18)9 (82) 
 Systemic JIA13 (68)6 (32) 
 Other JIA5 (50)5 (50) 
Female:male41 (74): 14 (26)58 (79):15 (21)0.51§
Age at start of MTX, mean ± SD years (range)7.9 ± 3.7 (1.9–15.9)7.7 ± 4.3 (1.3–17.4)0.82#
Time-to-start MTX, median (range) months16.3 (0.0–150)9.5 (0.0–94.8)0.074**
 Starting MTX dose, mean ± SD mg/m2 (range)10.4 ± 4.7 (4.5–23.7)8.6 ± 3.3 (1.8–16.5)0.018#
 SSZ, IAS, or other DMARDs used before start of MTX  (yes/no)32 (58)41 (56)0.82§
 Systemic steroids used before start of MTX (yes/no)17 (31)12 (16)0.053§
 SSZ used during MTX (yes/no)18 (33)23 (32)0.88§
 Steroids used during MTX (yes/no)19 (35)38 (52)0.048§
Disease activity at start of MTX   
 ESR, median (range) mm/hour (n = 67)††30 (7–137)25 (2–107)0.040**
 Physician's global assessment  0.000‡‡
  1 = inactive4 (7)0 (0) 
  2 = mild9 (16)3 (4) 
  3 = moderate30 (55)41 (56) 
  4 = severe12 (22)28 (48) 
  5 = very severe0 (0)1 (1) 
Joint score  0.40‡‡
  0 = none4 (7)0 (0) 
  1 = monarthritis1 (2)3 (4) 
  2 = oligoarthritis19 (35)23 (31) 
  3 = polyarthritis25 (45)37 (51) 
  4 = severe polyarthritis0 (0)8 (11) 
  5 = systemic features6 (11)2 (3) 

Genetic analysis.

The genotype frequencies and the MTHFR haplotype frequencies in this population are presented in Table 2. When comparing MTX responders with nonresponders in a haplotype-carrier analysis, a statistically significant difference in number of MTHFR1298A-677C haplotype copies was found. In nonresponders, 0 or 1 copy of the MTHFR1298A-677C haplotype was more frequently observed than in responders (P = 0.039). All other pharmacogenetic association analyses showed no significant differences.

Table 2. Genotype and MTHFR 1298A-677C haplotype frequencies in MTX nonresponders and MTX responders*
 AMPD1 34 C-TATIC 347 C-GITPA 94 C-AMTHFD1 1958 G-AMTHFR 1298A-677C
  • *

    Values are the number (percentage) unless otherwise indicated. MTHFR = methylenetetrahydrofolate reductase; MTX = methotrexate; AMPD1 = adenosine monophosphate deaminase; ATIC = aminoimidazole carboxamide ribonucleotide transformylase; ITPA = inosine triphosphate pyrophosphatase; MTHFD1 = methylenetetrahydrofolate dehydrogenase.

  • The number of MTHFR1298A-677C haplotype copies.

  • Linear-by-linear association.

MTX nonresponder40 (77)12 (23)0 (0)23 (42)27 (49)5 (9)50 (91)5 (9)0 (0)13 (24)29 (53)13 (24)47 (89)6 (11)
MTX responder44 (66)21 (32)1 (2)35 (49)27 (38)9 (13)58 (85)9 (13)1 (2)21 (30)35 (49)15 (21)50 (74)18 (26)
P 0.183  0.748  0.279  0.503  0.039

Univariate and multivariate analysis of variables in relation to MTX response.

Variables with a P value <0.1 (Tables 1 and 2) were considered to influence the response to MTX and were analyzed univariately and thereafter included in a multivariate regression analysis to correct for confounding effects (Table 3). In the multivariate regression analysis, the time-to-start MTX, the baseline physician's global assessment, and the starting dose of MTX were significantly associated with response to MTX at 6 months after initiation. No confounding effect of the included variables on the effect of time-to-start MTX on response was observed. Briefly, responders started MTX earlier and had a higher disease activity at baseline based on the physician's global assessment. What is more remarkable is that responders received a lower starting dose. However, the starting dose of MTX was highly influenced by the subtype of JIA (P < 0.001 by analysis of variance), especially by the systemic JIA patients who received a higher starting dose and had a decreased response. Repeating the multivariate analysis without the systemic JIA subtype resulted in a significant association of the time-to-start MTX and baseline physician's global assessment with MTX response (data not shown), and no effect of the starting dose on MTX response was observed (OR 0.89, 95% CI 0.76–1.05, P = 0.166).

Table 3. Univariate and multivariate regression analysis of clinical and genetic factors with MTX response in JIA patients (n = 118) as the dependent variable*
 Univariate analysisMultivariate analysis
OR (95% CI)POR (95% CI)P
  • *

    OR = odds ratio; 95% CI = 95% confidence interval; MTHFR = methylenetetrahydrofolate reductase; see Table 1 for additional definitions.

  • Total R2 = 0.27 (Cox and Snell).

  • JIA subtype as a category with persistent oligoarthritis as the reference subtype.

  • §

    0 or 1 haplotype (MTHFR1298A-677C) copies versus 2 haplotype copies.

JIA subtype 0.069 0.441
 Persistent oligoarthritis1.00 (reference)   
 Extended oligoarthritis2.81 (0.81–9.80)0.1043.41 (0.62–18.7)0.159
 RF-negative polyarthritis1.99 (0.65–6.03)0.2260.76 (0.18–3.15)0.700
 RF-positive polyarthritis5.63 (0.94–33.8)0.0591.41 (0.16–12.4)0.757
 Systemic JIA0.58 (0.15–2.21)0.4220.68 (0.09–5.12)0.712
 Other JIA1.25 (0.27–5.89)0.7780.92 (0.14–6.2)0.933
Time-to-start MTX (months)0.98 (0.97–0.996)0.0130.97 (0.95–0.996)0.021
Starting dose MTX (mg/m2)0.89 (0.81–0.98)0.0230.84 (0.73–0.97)0.021
Steroid use during MTX (yes/no)2.06 (1.00–4.23)0.0502.34 (0.86–6.34)0.095
Steroid use before MTX (yes/no)0.44 (0.19–1.02)0.0560.43 (0.12–1.6)0.205
Physician's global assessment at start2.6 (1.5–4.7)0.0012.4 (1.1–5.1)0.026
MTHFR haplotype§2.8 (1.03–7.7)0.0432.3 (0.68–7.7)0.178

Regarding the time-to-start MTX (Table 3), it is evident that the use of intraarticular steroids, SSZ, and/or other DMARDs prior to MTX influenced time-to-start MTX. Although the use of prior treatment was not related to MTX response (Table 1), we analyzed whether the association of time-to-start MTX was independent of the use of prior treatment and a true effect for MTX response. Therefore, the effect of time-to-start MTX on response was assessed including only patients with prior treatments. This repeated multivariate analysis showed that the time-to-start MTX was still significantly associated with the response to MTX (OR 0.96, 95% CI 0.94–0.99, P = 0.017).

Finally, the time-to-start MTX was strongly correlated with the physician's global assessment at baseline, meaning that patients with an increased disease activity were treated earlier. Interestingly, our data showed that the physician's global assessment at baseline also reflected the disease activity as measured from diagnosis to start of MTX (data not shown). Although the precise effects size of time-to-start MTX and the physician's global assessment at baseline cannot be individually determined, from a clinical point of view it is important to note that earlier treatment is related to increased response rates.


In this retrospective cohort of JIA patients, the time-to-start MTX, physician's global assessment at baseline, and starting dose are significantly associated with the response to MTX at 6 months. Patients with an earlier start of MTX and increased disease activity show an increased response. Our finding that a lower starting dose of MTX is associated with increased response is mainly due to systemic JIA patients, who receive a higher starting dose and show decreased response.

Although treatment with intraarticular steroids, SSZ, and/or other DMARDs prior to MTX is an important determinant for the delay in starting MTX, only including patients who had previously been treated with MTX therapy still showed that early MTX treatment was significantly associated with an increased response. This indicates that our data were not biased by population selection.

Our data show that patients with an increased disease activity receive MTX earlier after diagnosis, which may partially reflect confounding by indication. Therefore, the independent effects of a higher physician's global assessment at baseline and decreased time-to-start MTX on response cannot be determined in this analysis. The best evidence for these associations remains the replication in controlled trials with JIA patients.

Because of the different treatment strategies and small numbers of patients in the different JIA subtypes, only associations with MTX response in the general JIA population were observed and no conclusion about the influences of response in individual subtypes can be drawn.

With regard to the genetic parameters, only the MTHFR 1298A-677C haplotype showed a significant decrease in lower copy number in nonresponders. This finding is consistent with recent-onset RA in which a higher number of haplotype copies was related to increased response to MTX (13). Remarkably, no other significant genetic associations with response to MTX were detected. This may be due the small sample size resulting in an increased probability of obtaining false-negative findings (Type II error).

Our data analysis used a composite measure for response including the physician's global assessment and joint score. Frequent parent/patient global assessment of overall well-being and information about limitation of joints could not be retrospectively obtained from the patients' charts. Although the definition of improvement as developed by Giannini et al (14) includes 6 core set variables, the 2 variables included in our definition of improvement are sensitive instruments for measuring change, with the subjective assessment of disease activity by the physician as the most responsive instrument (15). The independent factors are able to measure an improvement of 53–60% (16). These factors were combined in our definition of improvement and generated a 57% response. Because improvement in the physician's global assessment is part of our response definition and a higher physician's global assessment at baseline is associated with response, the response rates in our cohort might be partly due to regression to the mean and an effect of unequal distance between the categories and do not fully reflect the true effect size of MTX.

According to the distribution of subtypes, this retrospective cohort is comparable with the clinical practice described by Brunner et al, indicating that no nonrandom exclusion of subtypes has occurred (3). In this study, the response rates of the different subtypes are comparable with the previously reported efficacy of MTX, except for persistent oligoarthritis (4–8). In the persistent oligoarthritis cohort of Brik et al (17), a response rate of 90% was reported in patients receiving early treatment (maximum of 4.3 months), whereas our patients with persistent oligoarthritis had a response rate of 44% and were treated after a mean ± SD of 22.2 ± 27.4 months (range 0–104 months). This difference in response rate underlines the fact that also in persistent oligoarthritis, early treatment with MTX may be an important factor for efficacy.

In summary, in this JIA population, the time-to-start MTX is an important and independent factor for response to treatment. It has already been shown in RA that, early treatment during the “widow of opportunity” is associated with an improved clinical outcome and less radiographic damage (18). Although it is thought that JIA patients may show similar favorable effects of early treatment, our study is the first to illustrate that JIA patients have increased responses to MTX when treated earlier. Patients with a good clinical response to DMARDs other than MTX were not included in this study. Therefore, no conclusion about the effectiveness of MTX compared with other DMARDs or the association between time-to-start MTX and other DMARDs in JIA can be drawn.

In RA, it has been shown that the onset of disease and immunologic events predate the symptoms by many years. The activation of RA is believed to be a multifactorial process that is followed by an ongoing progression of inflammation leading to bone damage already in the first year after diagnosis (19). Chronic arthritis in JIA patients probably has a similar early onset preceding the symptoms and a subsequent progression of inflammation. The increased response to early treatment with MTX that has been described in RA and demonstrated in JIA in this study might reflect the fact that MTX can suppress early stages of inflammation, and that the mechanism of action is less sufficient to control well-established chronic inflammation.

It is highly clinically relevant that reducing the time-to-start MTX may lead to an increased response. Future prospective studies are needed to replicate these findings and reveal the exact window of opportunity for JIA. More importantly, future studies are needed to determine if an increased early response leads to less joint damage in the long term.

In conclusion, our study demonstrates that time-to-start MTX appears to be an important factor for MTX response. Our results suggest that earlier initiation of MTX treatment will lead to an increased response.


Dr. ten Cate had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Albers, Wessels, Wouters, Schilham, Huizinga, ten Cate, Guchelaar.

Acquisition of data. Albers, van der Straaten, Suijlekom-Smit, Kamphuis, Girschick Wouters, ten Cate.

Analysis and interpretation of data. Albers, Wessels, Suijlekom-Smit, Kamphuis, Girschick, Huizinga, ten Cate, Guchelaar.

Manuscript preparation. Albers, Wessels, van der Straaten, Brinkman, Kamphuis, Suijlekom-Smit, Girschick, Wouters, le Cessie, Huizinga, ten Cate, Guchelaar.

Statistical analysis. Albers, Wessels, le Cessie.


The authors are grateful for the hard work of E. L. Breevaart and C. Ehrsam.