Development and validation of a preliminary definition of minimal disease activity in patients with juvenile idiopathic arthritis

Authors


Abstract

Objective

To develop and validate a definition of minimal disease activity (MDA) in patients with juvenile idiopathic arthritis (JIA).

Methods

The clinical charts of JIA patients followed over a 16-year period were reviewed to identify visits with high disease activity and MDA, defined on the basis of therapeutic decisions made by the attending physician. For each JIA activity measure recorded at the time of the visit, the cutoff value that best identified states of MDA was calculated by means of the area under the receiver operating characteristic curve analysis. A definition of MDA for oligoarthritis and polyarthritis was set up after testing the relative power of each variable in a multivariate analysis. Validation procedures included assessment of discriminant and construct validity.

Results

The definition that resulted from the analyses led to establish that a state of MDA could be defined as the presence of a physician global assessment ≤2.5 cm and a swollen joint count of 0 in patients with oligoarthritis; and as the presence of a physician global assessment ≤3.4 cm, a parent global assessment ≤2.1 cm, and a swollen joint count ≤1 in patients with polyarthritis. Validation procedures demonstrated that the MDA definition had good discriminant and construct validity in the context of both observational studies and controlled trials.

Conclusion

We developed a preliminary definition of MDA in patients with JIA that represents a useful treatment target state and is proposed for inclusion as an outcome measure in future observational studies and clinical trials in patients with JIA.

INTRODUCTION

Juvenile idiopathic arthritis (JIA) is a broad term that describes a clinically heterogeneous group of arthritides of unknown cause that begins before age 16 years. The International League of Associations for Rheumatology (ILAR) has provided the most recent classification of JIA, which recognizes the following 7 disease categories on the basis of the features present in the first 6 months of illness: systemic arthritis, oligoarthritis (persistent and extended), rheumatoid factor–positive polyarthritis, rheumatoid factor–negative polyarthritis, enthesitis-related arthritis, psoriatic arthritis, and undifferentiated arthritis (1).

In the past decade, there have been major advances in the treatment of JIA, which include the widespread use of methotrexate (MTX) and intraarticular corticosteroids (IACs), the earlier introduction of these medications, and in recent years, the availability of biologic agents (2–4). These advances have increased the expectations of treatment benefit, with disease remission now being the goal for many pediatric rheumatologists (5–7). Although this objective has been easier to achieve in children with oligoarthritis who are treated with only IACs, it is still problematic in the subset of patients who have polyarticular disease. No medication or combination of medications has yet been demonstrated to be effective for inducing remission in the majority of these patients. Few randomized controlled trials in patients with JIA report remission data, which is indirect evidence of the infrequency of remission.

For adult patients with rheumatoid arthritis (RA), remission is not a frequent occurrence in daily clinical care. Furthermore, it has been argued that achieving (and maintaining) a satisfactory state of disease activity is probably more important in the long term than the improvement from a high level of disease activity documented in trials (i.e., that provided by the application of American College of Rheumatology [ACR] criteria for a 20%, 50%, or 70% response) (8). These issues have raised the need for defining a state of low disease activity (i.e., minimal disease activity [MDA]) as an intermediate state between high disease activity (HDA) and remission (9–11). As a result, a preliminary definition of MDA for use in RA clinical trials was recently developed through a consensus process (10). According to this definition, patients with no tender joints, no swollen joints, and an erythrocyte sedimentation rate (ESR) ≤10 mm/hour are considered to have MDA. If patients do not meet this criterion, they are still considered to have MDA if either their Disease Activity Score in 28 joints is ≤2.85 or they meet 5 of the following 7 criteria: pain ≤2, swollen joint count ≤1, tender joint count ≤1, Health Assessment Questionnaire score ≤0.5, physician global assessment ≤1.5, parent global assessment ≤2, and ESR ≤20 mm/hour. It has been suggested that describing the proportion of patients achieving and maintaining such a state for a specific period of time would add useful information for the practicing physician and aid in the interpretation of trials and longitudinal results (12).

The aim of the present study is to devise a preliminary definition of MDA in patients with JIA and to provide evidence of the clinical validity of the definition.

PATIENTS AND METHODS

Study design and patient selection.

The clinical charts of all consecutive patients who were seen in the study units between January 1988 and December 2003, had JIA according to ILAR criteria (1), and had received ≥1 clinic visit were retrospectively reviewed. Patients seen before the publication of ILAR criteria were reclassified according to such criteria. The study protocol was approved by the Ethics Committee of the Istituto G. Gaslini, Genoa, Italy.

Criteria for states of HDA and MDA.

All patients' visits were examined to identify those who met the criteria for HDA or MDA (Table 1). These criteria were based on the therapeutic decision made by the attending pediatric rheumatologist at the time of the visit. The criteria were based on consensus of study investigators and literature review. For a patient to be placed in an activity state, the patient had to meet any 1 of the criteria for that state. If a patient met the criteria in more than 1 visit, only the first visit (or, in case the first visit had insufficient clinical data, the subsequent visit with more clinical information available) meeting the criteria for HDA or MDA was selected. Therefore, each patient contributed to the analysis with a maximum of 1 HDA visit and 1 MDA visit. Owing to their peculiar clinical features, patients with enthesitis-related arthritis deserved distinctive definitions of MDA. For this reason and owing to the relatively small number of available observations, this patient subgroup was excluded from the analysis.

Table 1. Criteria for the definition of states of high and minimal disease activity and their frequency in 529 visits
 Frequency, %
  • *

    Assessed only in patients with oligoarthritis.

  • Assessed only in patients with polyarthritis.

High disease activity (n = 321) 
 New start or restart of nonsteroidal antiinflammatory drug therapy*16.4
 Administration of intraarticular corticosteroid therapy51.6
 New start or restart of prednisone therapy or increase in prednisone dose by >0.2 mg/kg9.2
 New start or restart of a second-line or biologic medication25.6
 Increase in methotrexate dose by >50%15.1
 Addition of another second-line medication or biologic agent to methotrexate6.3
Minimal disease activity (n = 208) 
 Termination of nonsteroidal antiinflammatory drug therapy because of disease remission*10.0
 Not administering intraarticular corticosteroid therapy in a patient receiving no systemic medication*31.3
 Discontinuation of prednisone therapy because of satisfactory disease control2.4
 Termination of methotrexate therapy because of disease remission10.0
 Not changing a second-line or biologic medication for ≥1 year38.5
 Not receiving treatment with a second-line or biologic medication for ≥1 year7.8

Clinical assessment.

General patient information included age at onset, sex, and ILAR category. The following clinical measures of JIA activity were recorded, whenever available, for each instance of HDA and MDA: physician's global assessment of disease activity, measured on a 10-cm visual analog scale (VAS; where 0 = no activity and 10 = maximum activity); parent's global assessment of the child's well-being, measured on a 10-cm VAS (where 0 = very good and 10 = very poor); parent's rating of the intensity of the child's pain, measured on a 10-cm VAS (where 0 = no pain and 10 = maximum pain); swollen joint count, pain on motion/tenderness, restricted motion, and active disease (13); functional ability assessment through the Italian version of the Childhood Health Assessment Questionnaire (C-HAQ; where 0 = best and 3 = worst) (14), or before 1994, through the Modified Lee Index (15) or the Juvenile Arthritis Functional Assessment Report (to ensure uniformity, scores of the 2 latter instruments were converted to the 0–3 score of the C-HAQ, as reported) (13, 16); duration of morning stiffness; ESR (Westergren method); C-reactive protein level (nephelometry); white blood cell count; hemoglobin level; and platelet count.

Development of the definition of MDA.

In the first phase of the study, we considered all measures of JIA activity to be of potentially equal importance to define the states of HDA and MDA, and calculated for each of them the cutoff value that best identified the states of MDA by means of the area under the receiver operating characteristic curve (AUC-ROC) analysis. Because JIA patients with oligoarthritis or polyarthritis have a different severity potential, deserve diverse treatment strategies, and are usually included in distinct clinical trials, we believed each subgroup required a distinct definition of MDA. For this reason, the calculation was performed separately for patients with oligoarthritis (i.e., patients with the ILAR category of persistent oligoarthritis) or polyarthritis (i.e., patients with the ILAR categories of extended oligoarthritis, polyarthritis, and systemic arthritis whose disease followed a polyarticular course). For purposes of the analysis, patients with psoriatic arthritis and undifferentiated arthritis were classified within polyarthritis or oligoarthritis categories based on the number of joints affected throughout the disease course. In the subsequent step, the relative discriminating power of each variable was further scrutinized by means of multiple logistic regression analysis. This analyis was carried out by entering JIA activity measures as explanatory variables and the presence of MDA as an outcome variable. Cases with missing variables were excluded from the analysis.

All JIA activity measures were dichotomized into binary variables, using the thresholds yielded by the AUC-ROC analysis as cut points. Using a backward selection procedure, the most significant independent variables were identified, using a P value >0.05 as the removal criterion. The effect was expressed in odd ratios and 95% confidence intervals were calculated; statistical significance was tested by means of the likelihood ratio test. The predictive ability of multivariate models was evaluated by calculating their AUC-ROC.

Validation procedures.

Validation of the definition of MDA was based on the assessment of discriminant and construct validity. For evaluation of discriminant validity, 3 different data sources were used. The first data sets were composed of 2 controlled trials, including 1 aimed to compare intermediate and higher doses of MTX in 633 patients with polyarthritis (17), and 1 aimed to compare meloxicam and naproxen in 225 patients with polyarthritis or oligoarthritis (18). Data from the trials were provided by the Paediatric Rheumatology International Trials Organisation and by Boehringer Ingelheim, respectively. Patients in the 2 trials were divided into 4 mutually exclusive groups (nonresponders, responders at 30%, responders at 50%, and responders at 70%) by their degree of improvement at 6 and 3 months after treatment start, respectively, and according to ACR Pediatric 30 (Pedi 30) criteria for improvement (19). Patients were classified as responders at 30%, 50%, and 70% if they had an improvement of ≥30%, 50%, or 70%, respectively, with respect to baseline in ≥3 of the 6 core set items, with no more than 1 of the remaining items worsened by >30%. Patients were classified as nonresponders if they did not have an improvement of ≥30%. The proportions of patients with polyarthritis or oligoarthritis in each response group who met the respective MDA definition were then compared. It was considered that the definition would have demonstrated good discriminant ability if the percentage of patients meeting MDA criteria was significantly greater in the group that exhibited the greater degree of therapeutic response (i.e., the ACR Pedi 70 responders).

The third dataset consisted of 42 patients with oligoarthritis who received an IAC injection and were included in a previous study of responsiveness of JIA outcome measures (20). At 6 months after the injection, patients were classified by the physician and the parent as improved, stable, or worsened, compared with the preinjection status. The proportion of patients meeting the MDA definition in the groups judged as improved or not improved (stable to worsened) by the physician or the parent was compared. We anticipated that the definition would demonstrate good discriminant ability if the proportion of patients meeting MDA criteria was significantly greater in the improved group than in the not improved group.

Percentages of patients meeting MDA criteria in each response group were compared by means of the chi-square test or by Fisher's exact test in case of expected frequencies <5. Bonferroni adjustment was applied as a correction for multiple comparisons to explore post hoc differences between pairs of patient groups.

Construct validity was examined using a clinical database that included 270 consecutive patients with oligoarthritis or polyarthritis who had long-standing disease and had received a cross-sectional assessment of JIA outcome measures at last followup visit, 5–21 years after disease onset. The following measures of disease damage and health-related quality of life (HRQOL) were compared in patients who met or did not meet the MDA definition: restricted joint count, C-HAQ score (14), Steinbrocker functional class (21), Juvenile Arthritis Damage Index (JADI) score (22), Poznanski score of radiographic damage (23, 24), and the Child Health Questionnaire of HRQOL (14). We anticipated that the definition would demonstrate good construct validity if the amount of damage was significantly lower in patients who had reached an MDA state at last followup visit than in those who did not. Comparison of quantitative variables was made by means of the Mann-Whitney U test, whereas comparison of qualitative variables was performed by means of the chi-square test or by Fisher's exact test in case of expected frequencies <5.

All statistical tests were 2-sided; P values less than 0.05 were considered statistically significant. The statistical packages used were Statistica (StatSoft, Tulsa, OK) and Stata, release 7 (StataCorp, College Station, TX).

RESULTS

A total of 1,080 visits (made in 414 patients) that met the criteria for HDA or MDA were identified. According to the above criteria, 1 HDA visit and/or 1 MDA visit for each patient was chosen. This led us to select 529 visits, 321 of which met the criteria for HDA and 208 of which met the criteria for MDA. The values of the clinical measures of JIA activity in visits included and excluded were comparable (data not shown). Table 2 shows the main demographic and clinical features of the visits/patients with HDA or MDA included in the analyses.

Table 2. Comparison of demographic and clinical features between patients/visits with minimal disease activity or high disease activity*
 Minimal disease activityHigh disease activity
Oligoarthritis (n = 89)Polyarthritis (n = 119)Oligoarthritis (n = 159)Polyarthritis (n = 162)
  • *

    Values are the median (interquartile range) unless otherwise indicated. ILAR = International League of Associations for Rheumatology.

  • Range 0 (best) to 10 (worst).

  • Range 0 (best) to 3 (worst).

  • §

    Normal value <15 mm/hour.

  • Normal value <0.4 mg/dl.

Men, no. (%)18 (20.2)35 (29.4)33 (20.7)45 (27.8)
Women, no. (%)71 (79.8)84 (70.6)126 (79.3)117 (72.2)
ILAR category, no. (%)    
 Oligoarthritis persistent75 (84.2)0 (0)141 (88.7)0 (0)
 Oligoarthritis extended0 (0)38 (31.9)0 (0)51 (31.5)
 Polyarthritis0 (0)37 (31.1)0 (0)54 (33.3)
 Systemic arthritis0 (0)35 (29.4)0 (0)40 (24.7)
 Psoriatic arthritis7 (7.9)4 (3.4)8 (5.0)7 (4.3)
 Undifferentiated arthritis7 (7.9)5 (4.2)10 (6.3)10 (6.2)
Physician's global assessment of disease  activity0 (0, 0.5)1 (0, 2.5)7.4 (4.5, 8.7)7.1 (5.3, 10)
Parent's global assessment of well-being0 (0, 1.1)0.5 (0, 1.7)3 (1, 5)4.2 (2.5, 5.9)
Parent's pain assessment0 (0, 0.9)0.4 (0, 1.6)2.8 (1.2, 5.0)4 (2, 6.5)
Childhood Health Assessment  Questionnaire score0 (0, 0)0.1 (0, 0.6)0.3 (0, 0.8)0.8 (0.3, 1.4)
Swollen joint count0 (0, 0)1 (0, 3)2 (1, 3)5 (3, 9)
Tender joint count0 (0, 0)0 (0, 2)1 (1, 2)5 (2, 9)
Restricted joint count0 (0, 0)1 (0, 4)1 (1, 2)5 (2, 8)
Active joint count0 (0, 1)1 (0, 4)2 (1, 3)7 (4, 12)
Morning stiffness, minutes0 (0, 0)0 (0, 6)15 (0, 60)20 (0, 60)
Erythrocyte sedimentation rate, mm/hour§12 (8, 19)12.5 (8, 25)30 (15, 54)40 (20, 63)
C-reactive protein level, mg/dl0.1 (0.1, 0.3)0.1 (0.1, 0.9)0.6 (0.1, 1.8)2.3 (0.4, 5.9)
White blood cells, ×109/liter6.8 (5.7, 8)7.5 (6.0, 9.7)8.6 (6.7, 10.3)9.1 (7.3, 12.2)
Hemoglobin, gm/liter129 (122, 137)125 (116, 133)121 (112, 129)114 (103, 124)
Platelets, ×109/liter333 (270, 371)337 (241, 389)374 (307, 447)400 (326, 517)

Calculation of the best thresholds through AUC-ROC analysis.

The cutoff values of clinical measures of JIA activity, obtained through the AUC-ROC analysis, that best identified states of MDA in patients with oligoarthritis and polyarthritis are shown in Table 3. As expected, most cutoffs were higher in patients with polyarthritis than in those with oligoarthritis. That the cutoff for the physician global assessment was greater than that of the parent global assessment is consistent with our previous observations that parents of children with JIA tend to provide lower global ratings than physicians (25, 26). All joint counts should equal 0 for patients with oligoarthritis to be considered as having MDA, whereas in patients with polyarthritis the presence of 1 swollen or tender joint or 3 active joints would still enable classification of a patient in MDA status. Patients with either oligoarthritis or polyarthritis should have no morning stiffness for being considered as having MDA. Cutoffs for laboratory measures were in their normal range or very close to it.

Table 3. Threshold values for juvenile idiopathic arthritis activity measures that best discriminated between states of high and minimal disease activity obtained through the area under the receiver operating characteristic curve analysis
 Oligoarthritis (n = 248)Polyarthritis (n = 281)
  • *

    Range 0 (best) to 10 (worst).

  • Range 0 (best) to 3 (worst).

  • Normal value <15 mm/hour.

  • §

    Normal value <0.4 mg/dl.

Physician's global assessment of disease activity*≤2.5≤3.4
Parent's global assessment of well-being*≤0.3≤2.1
Parent's pain assessment*≤0.9≤2
Childhood Health Assessment Questionnaire score≤0.25≤0.25
Number of swollen joints0≤1
Number of tender joints0≤1
Number of joints with restricted motion00
Number of active joints0≤3
Morning stiffness, minutes00
Erythrocyte sedimentation rate, mm/hour≤16≤20
C-reactive protein level, mg/dl§≤0.41≤0.9
White blood cells, ×109/liter≤8.2≤7.1
Hemoglobin, gm/liter>121>119
Platelets, ×109/liter≤392≤379

Data for the development of the definition of MDA.

For the multivariate analysis, complete data were available for 388 patients. The best-fitting model obtained through logistic regression procedures, which had the presence of MDA as its dependent variable and JIA activity measures as explanatory variables, is shown in Table 4. The physician global assessment and the swollen joint count were the independent predictive variables for MDA in oligoarthritis, whereas the physician global assessment, the swollen joint count, and the parent global assessment were the independent predictive variables for MDA in polyarthritis. The AUC-ROC of the model was 0.95 for oligoarthritis and 0.93 for polyarthritis. Based on the results of the multivariate analysis, a preliminary definition of MDA in oligoarthritis and polyarthritis was set up (Table 5).

Table 4. Best-fitting model obtained through logistic regression procedures with the presence of minimal disease activity as the outcome*
Explanatory variableβSEOR95% CIP
  • *

    The area under the receiver operating characteristic curve of the model was 0.95 for oligoarthritis and 0.93 for polyarthritis. OR = odds ratio; 95% CI = 95% confidence interval.

Oligoarthritis (n = 195)     
 Physician's global assessment of disease activity ≤2.53.80.745.312.6–163.3< 0.0001
 Number of swollen joints 03.21.125.32.9–224.6< 0.0001
Polyarthritis (n = 193)     
 Physician's global assessment of disease activity ≤3.43.20.524.99.4–66.0< 0.0001
 Parent's global assessment of well-being ≤2.11.60.54.81.9–12.40.0013
 Number of swollen joints ≤11.60.74.91.3–18.10.015
Table 5. Definition of minimal disease activity in patients with oligoarticular and polyarticular juvenile idiopathic arthritis (JIA)*
 Oligoarticular JIAPolyarticular JIA
  • *

    To be classified as being in a minimal disease activity state, patients should meet all criteria for their disease subtype simultaneously. In addition to the above criteria, patients with systemic-onset disease should have no fever, rash, serositis, splenomegaly, or generalized lymphadenopathy attributable to JIA.

  • Measured on a 10-cm visual analog scale, where 0 = best and 10 = worst.

Physician's global assessment of disease activity, cm≤2.5≤3.4
Parent's global assessment of well-being, cm ≤2.1
No. swollen joints0≤1

Assessment of discriminant validity.

In the MTX trial (17), the percentage of patients meeting the MDA definition among the nonresponders, 30% responders, 50% responders, and 70% responders was 2.3%, 8.6%, 17.8%, and 62.2%, respectively (P = 0.0002 for comparison of 50% responders versus nonresponders, P < 0.0001 for comparison of 70% responders versus nonresponders, 30% responders, and 50% responders). In the meloxicam trial (18), the percentage of patients meeting the MDA definition among the nonresponders, 30% responders, 50% responders, and 70% responders was 5.1%, 0%, 15%, and 61.7%, respectively, among 137 patients with oligoarthritis (P < 0.0001 for comparison of 70% responders versus nonresponders, P = 0.001 for comparison of 70% responders versus 30% responders, P = 0.003 for comparison of 70% responders versus 50% responders), and 7.7%, 5.5%, 14.3%, and 60.9%, respectively, among 88 patients with polyarthritis (P = 0.0004 for comparison of 70% responders versus nonresponders, P = 0.002 for comparison of 70% responders versus 30% responders, P = 0.009 for comparison of 70% responders versus 50% responders). No patient enrolled in the 2 trials was in an MDA state at baseline, providing face validity to the definitions. Among patients with oligoarthritis who received an IAC injection, the percentage of patients who met the MDA definition at 6 months after the procedure was 50% and 0%, respectively (P < 0.0001), among patients who were judged as improved or not improved by the physician, and 40% and 6%, respectively (P = 0.02), among patients who were judged as improved or not improved by the parent. Altogether, these results demonstrate that the MDA criteria had a strong capability to discriminate between different levels of clinical response to therapy.

Assessment of construct validity.

Table 6 shows the comparison of the measures of damage and HRQOL between patients with oligoarthritis or polyarthritis who met or did not meet the MDA criteria at last followup visit, 5–21 years after disease onset. Compared with patients who had not achieved an MDA state, patients who had reached this state had significantly less functional impairment, as measured through the restricted joint count, the C-HAQ, or the Steinbrocker classification, and better HRQOL. Likewise, in the polyarthritis group, patients with MDA had a smaller amount of articular and extraarticular damage as measured through the JADI, and tended to have a smaller amount of radiographic damage, as assessed through the Poznanski score. All of these findings were consistent with a priori predictions, thereby demonstrating that the MDA definition has good construct validity.

Table 6. Measures of disease damage and health-related quality of life at last followup visits in 270 patients with juvenile idiopathic arthritis who met or did not meet the definition of MDA*
 Oligoarthritis (n = 106)Polyarthritis (n = 164)
MDA (n = 39)No MDA (n = 67)PMDA (n = 60)No MDA (n = 104)P
  • *

    Values are the median (interquartile range) unless otherwise indicated. All patients had ≥5 years of disease duration. MDA = minimal disease activity; C-HAQ = Childhood Health Assessment Questionnaire; JADI = Juvenile Arthritis Damage Index; CHQ = Child Health Questionnaire.

  • Range 0 (best) to 3 (worst).

  • No patient was in Steinbrocker class IV.

  • §

    Abnormal score less than −2.0.

  • Mean ± SD norm-based score (for both physical and psychosocial scores) 50 ± 10, where a higher score means better quality of life.

Age at onset, years2.7 (1.8, 3.8)3.0 (1.9, 5.3)0.352.9 (1.8, 5.9)3.0 (1.7, 5.9)0.83
Disease duration, years6.4 (5.2, 9.7)7.2 (5.7, 10.4)0.146.7 (5.5, 8.6)6.4 (5.5, 8.9)0.96
No. joints with restricted motion0 (0, 0)1.0 (0.0, 1.0)< 0.00010.0 (0.0, 1.5)4.0 (1.0, 9.0)< 0.0001
C-HAQ score0 (0, 0)0 (0, 0.25)0.00080 (0, 0.25)0.38 (0, 0.75)< 0.0001
Steinbrocker class II–III, no. (%)1 (2.6)9 (13.4)0.094 (6.7)50 (48.1)< 0.0001
JADI articular score0 (0, 0)0 (0, 0)0.290 (0, 0)1.0 (0, 3.0)< 0.0001
JADI extraarticular score0 (0, 0)0 (0, 0.5)0.130 (0, 0)0 (0, 1.0)0.0004
Poznanski score, units (n = 116)§---−0.58 (−1.29, 0.38)−1.18 (−3.12, 0.03)0.05
CHQ physical summary score54.1 (51.9, 57.5)51.4 (44.8, 54.2)0.00753.4 (51.2, 55.6)47.2 (38.2, 52.8)< 0.0001
CHQ psychosocial summary score46.9 (43.6, 53.8)52.2 (43.9, 57.4)0.0654.2 (46.0, 59.0)48.3 (42.7, 53.2)0.007

DISCUSSION

In recent years, it has been increasingly recognized that in chronic arthritis clinical trials, it is not only important to know the magnitude of clinical improvement from baseline, but also to understand whether the observed change leads to an acceptable state according to the physician and/or the patient (12, 27). Although the best condition is obviously the absence of symptoms, that is, a state of disease remission, this objective may prove difficult to achieve for many patients. Therefore, alongside the concept of remission, the concept of an MDA state, perceived as a more realistic goal of treatment, has recently emerged (28). In adult patients with RA, MDA has been defined as that state of disease activity deemed a useful target of treatment by both the patient and the physician, given current treatment possibilities and limitations (12). Recent studies have shown that patient outcomes are improved if RA activity is regularly assessed and if low levels of disease activity are aimed for by adjustments of therapy (29–31).

Because the state of MDA has not yet been characterized in patients with JIA, we aimed to create such a definition through the retrospective analysis of a large set of clinical data. We reasoned that any index of disease activity should reflect clinical practice, and therefore, we focused on real decision points in patient management. We defined as high the level of disease activity demonstrated at a clinic visit where the physician decided to initiate or change disease-specific treatment, and as minimal the level at a visit where the physician did not change treatment policy or discontinued therapy because of disease remission. The MDA definition then resulted from a discriminant function that optimally distinguished between these 2 states.

Because JIA patients with oligoarticular or polyarticular course have different severity potential, deserve diverse treatment strategies, and are usually included in distinct clinical trials, we believed the 2 subgroups required a separate definition of MDA. As expected, the cutoff levels of disease activity variables that best identified MDA states in AUC-ROC analysis were, in general, lower for oligoarthritis than for polyarthritis. The definition of MDA that emerged from the multivariate analysis was based on 2 variables (physician global assessment and swollen joint count) for oligoarthritis and on 3 variables (physician global assessment, parent global assessment, and swollen joint count) for polyarthritis. Notably, the physician global assessment appears to be quite dominant, particularly for polyarticular disease. This adds to the previous demonstration of the good psychometric properties of this measure in patients with JIA (21, 32, 33). The inclusion of the parent's global rating in the definition of MDA for polyarthritis, which is the most severe subset of JIA, ensures that it incorporates the parent's perception of the impact of disease activity on the child. The fact that the swollen joint count was the only articular index that entered the best-fitting model of multivariate analyses is consistent with the notion that it represents the purest indicator of joint synovitis (34). As expected, neither the functional ability assessment nor the restricted joint count, which largely incorporate the level of disease damage (34), proved useful in discriminating between disease activity states. None of the laboratory parameters of systemic inflammation entered the best fit model of multivariate analyses. This offers the advantage of avoiding misclassification due to extraneously elevated acute phase proteins.

The definition of MDA revealed good discriminant and construct validity. As expected, none of the patients included in 2 controlled clinical trials in oligoarticular and polyarticular JIA met the criteria for MDA at the time of enrollment, when all patients should have, by definition, a high level of disease activity. Also as expected, the frequency of patients meeting the MDA definition at 6 or 3 months after baseline was significantly greater in the groups who experienced the greatest level of improvement (i.e., an ACR Pedi 70 response) in both trials. In a group of JIA patients with oligoarthritis who underwent an IAC injection, the frequency of MDA was significantly greater among patients who were judged as improved by the physician or parent at 6 months after the procedure. In a cross-sectional data set, we determined that patients with long-standing oligoarthritis or polyarthritis who were in a state of MDA at the last followup visit had better outcomes in terms of functional disability, damage, and HRQOL than patients who had not achieved such a state. This finding suggests that entering a state of MDA during the disease course is associated with a more favorable long-term outcome.

Some limitations to this study need mentioning. Our analysis was retrospective, which means that it was subject to missing and possibly erroneous data, and was based on visits made by only 2 pediatric rheumatologists. We used an observation-based statistical approach, in which MDA was inferred from a proxy variable (e.g., clinician's decisions to reduce/not to increase drug treatment) depending on analysis of clinical visit data on patients with JIA. A consensus definition or a judgmental approach (that is, explicitly asking physicians and/or parents their opinion on what they would consider a useful target in daily practice) could have led to a definition with high face validity and relevance in practice. However, the retrospective design of the study allowed us to describe the real process of decision making in practice because the physicians were unaware that their decisions were part of an investigation. We did not ask for parent opinion, which would be important to establish whether the MDA definition is meaningful to patients and their families.

Additional measures potentially useful to define MDA, such as HRQOL assessment, could not be evaluated. However, HRQOL was thought to be a different dimension of the burden of disease that was relevant to treatment, but only loosely bound to the concept of disease activity. We could not consider the time dimension that makes the definition of MDA most useful, for example, the duration of time spent in the state; however, this aspect is complex and requires longitudinal studies that record disease activity repeatedly and in sufficient detail. The definition of the time component will be addressed at a later stage. We considered MDA as a state, which is deemed a useful target of treatment by the physician, given current treatment possibilities and limitations. However, few visits involved biologics, which means that the definition may change in the biologic era. Like any definition, it should be regularly updated as treatment options and knowledge about them evolves. These limitations do not decrease the usefulness of the definition itself, which represents an absolute measure of treatment efficacy that can be used to set stricter treatment targets. Another potential limitation of our analysis is that clinical perceptions and treatment choices vary between physicians from different regions and according to their particular expertise. Nevertheless, physicians who made therapeutic decisions in our study came from tertiary care pediatric rheumatology centers and likely incorporated new treatment strategies and treatment aspects.

In conclusion, we developed a preliminary definition of MDA in children with oligoarticular and polyarticular JIA, which was found to possess good discriminant and construct validity. This definition identifies an intermediate state between active disease and remission that represents a useful treatment target state for both physicians and patients/parents, and is proposed for inclusion as an outcome measure in future observational studies and clinical trials in patients with JIA. The proportion of patients achieving a satisfactory state of MDA is likely to become an important measure with which to compare different treatment strategies and provides a benchmark for comparison with future cohorts.

AUTHOR CONTRIBUTIONS

Dr. Ravelli 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. Magni-Manzoni, Ruperto, Martini, Ravelli.

Acquisition of data. Magni-Manzoni, Sala, Solari, Palmisani, Cugno, Bozzola.

Analysis and interpretation of data. Magni-Manzoni, Ruperto, Ravelli.

Manuscript preparation. Ruperto, Martini, Ravelli.

Statistical analysis. Pistorio.

Acknowledgements

The authors thank Boehringer Ingelheim Pharma KG, Biberach, Germany, for providing access to the data of the meloxicam trial.

Ancillary