Dr. Feldman holds the Canada Research Chair in childhood arthritis.
Predictors of disease course and remission in systemic juvenile idiopathic arthritis: Significance of early clinical and laboratory features
Article first published online: 27 APR 2006
Copyright © 2006 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 54, Issue 5, pages 1595–1601, May 2006
How to Cite
Singh-Grewal, D., Schneider, R., Bayer, N. and Feldman, B. M. (2006), Predictors of disease course and remission in systemic juvenile idiopathic arthritis: Significance of early clinical and laboratory features. Arthritis & Rheumatism, 54: 1595–1601. doi: 10.1002/art.21774
- Issue published online: 27 APR 2006
- Article first published online: 27 APR 2006
- Manuscript Accepted: 19 JAN 2006
- Manuscript Received: 29 AUG 2005
To determine whether the disease course in systemic juvenile idiopathic arthritis (JIA) can be characterized as monophasic, polycyclic, or persistent, and to determine whether early clinical and laboratory characteristics can be used to predict the disease course and time to remission.
Forty-five children with systemic JIA diagnosed between 1996 and 2000 were followed up with a standardized data collection protocol, including data on clinical and laboratory features (mean followup 4.9 years). Disease was considered inactive if the clinical and laboratory features were normal. Three definitions of remission were applied to classify disease course. Predictors of disease course were evaluated using multiple logistic regression. Predictors of time to remission were evaluated using Cox proportional hazards regression.
When applying a definition of remission requiring inactive disease while not receiving any medications for a period of 3 months, 42.2%, 6.7%, and 51.1% of the patients were classified as having monophasic, polycyclic, and persistent disease, respectively. Fever and active arthritis at 3 months (R2 = 0.42, area under the receiver operating characteristics curve [AUC] = 0.76) and an erythrocyte sedimentation rate (ESR) >26 mm/hour and corticosteroid use at 6 months (R2 = 0.49, AUC = 0.92) were predictive of a nonmonophasic course. Absence of active arthritis, an ESR of <26 mm/hour, and no requirement for corticosteroid therapy at 3 and 6 months were predictors of an earlier time to remission.
The disease course in systemic JIA can be characterized as monophasic, polycyclic, or persistent using a definition of remission requiring 3 months of inactive disease while not receiving any therapy. Features at 3 and 6 months are predictive of the disease course and time to remission.
Systemic juvenile idiopathic arthritis (JIA) is a chronic disease that results in significant morbidity and mortality in children. Approximately 10% of children with JIA have the systemic form (1). Children with systemic JIA are known to develop chronic disability and significant functional impairment (2–8). Packham and Hall (3) found that after almost 30 years of followup, 75% of patients with systemic JIA had undergone joint replacement, and almost two-thirds were classified as having Steinbrocker stage III or IV disease and had Health Assessment Questionnaire scores in the severe disability range (9, 10). Bowyer and colleagues (2) showed that after 5 years of followup, 18% of children with systemic JIA were growth impaired and 75% had radiographic evidence of joint space narrowing.
Few studies have adequately evaluated the disease course in systemic JIA; however, the labels monophasic, polycyclic, and persistent are frequently used to describe the course of the disease. Lomater et al (11), in a retrospective review of 80 patients, found that 11% had a monophasic disease course, 34% had polycyclic disease, and 55% had persistent disease. Fantini et al (12), in a retrospective study of 88 patients, found a very low rate of polycyclic disease (only 2.3% of patients), while 65.9% had persistent disease and 31.8% were in remission at followup. In a smaller series of patients reported by Minden et al (13), one-third of the patients were described as having a monophasic disease course. These studies differed significantly in their definition of remission, which may explain the wide variability in results, and only Fantini et al (12) included duration of disease inactivity as part of the remission criteria. Lomater and colleagues also showed that persistent systemic JIA was associated with a worse clinical and radiologic outcome when compared with that in patients with monophasic disease (11). The outcome was directly related to the cumulative duration of disease activity.
We have previously shown that clinical and laboratory features at 6 months after disease onset were predictive of outcome in systemic JIA (4, 14). Persistent systemic symptoms and thrombocytosis 6 months after disease onset were highly predictive of the development of destructive arthritis within 2 years of disease onset (14). Moreover, fever, rash, the need for corticosteroids, and thrombocytosis 6 months after disease onset were predictive of poor functional outcome (4).
The ability to accurately define and predict the course of systemic JIA would allow physicians to inform patients and their families of the pattern of disease activity that might be expected in the future, and would allow rational approaches to therapy, with early aggressive therapy recommended for those likely to have persistent disease and a more conservative approach taken in the others. Furthermore, this information may be useful in determining classification criteria to be applied in future treatment and outcome studies.
By prospectively following up the disease patterns observed in a cohort of children with systemic JIA, we aimed, primarily, to determine whether monophasic, polycyclic, and persistent disease courses exist and, if so, what remission criteria might best define these entities. Our secondary aim was to determine whether the clinical and laboratory characteristics of patients with systemic JIA in the first 12 months after diagnosis could be predictive of the disease course and time to remission.
PATIENTS AND METHODS
Patients who were diagnosed as having systemic JIA between January 1996 and December 2000, using the criteria of the International League of Associations for Rheumatology (15), and who were evaluated at The Hospital for Sick Children in Toronto within 3 months of disease onset were eligible. The study was approved by the hospital's research ethics board.
Data were collected from the Division of Rheumatology patient database. Available data were crossreferenced with the patients' charts and laboratory data. Data collected included age, sex, active joint count (i.e., the numbers of swollen/effused joints, or joints having at least 2 of the following features: heat, limited range of movement, and tenderness/pain on range of movement), presence of fever, rash, lymphadenopathy, hepatomegaly, splenomegaly, or serositis, and laboratory results (antinuclear antibody [ANA] status, rheumatoid factor [RF] status, levels of immunoglobulin and hemoglobin [Hgb], white blood cell [WBC] count, platelet count, and erythrocyte sedimentation rate [ESR]), along with medication use at every clinic or hospital visit. Patients were followed up at regular intervals of 3–6 months while medications were still being administered, and at least yearly after medications were ceased.
Data pertaining to history and examination findings were collected using a standardized data collection form that was completed at clinic visits or using the hospital charts for inpatient visits during the study period. Laboratory data were obtained from the Division of Rheumatology clinical database, including all in-hospital and outpatient results. Laboratory results were considered normal if they were within the testing laboratory's reference range.
At any point in time, disease was considered to be active if the patient was determined to have fever, rash, serositis, hepatomegaly, splenomegaly, or active arthritis on review of the medical history, examination, or imaging, or had abnormal laboratory results (low Hgb and/or an elevated ESR, WBC count, or platelet count). If these features were all absent and laboratory results were normal, the patient was considered to have inactive disease. Three possible definitions of remission were then applied directly to the data, to formally assess the disease course. Remission I was defined as having inactive disease for a period of 6 months while still receiving medications (16). Remission II was defined as having inactive disease while not receiving any medications for a period of 3 months. Remission III was defined as having inactive disease while not receiving any medications for a period of 12 months (16). Definitions for the remission I and remission III criteria were derived from recently published preliminary guidelines for the definition of remission in JIA (16). The definition for the remission II criteria has been used informally within our clinical unit for some time and was considered by the authors to warrant inclusion on the basis of anecdotal experience.
Differences in clinical and laboratory characteristics between the 3 disease courses were analyzed using analysis of variance (ANOVA), and potential predictors of disease course were evaluated at diagnosis and at 3 and 6 months, by univariate and multiple logistic regression analyses, using the Data Desk 6.2 software package (17). Predictors of time to remission were identified through Cox proportional hazards regression, using the JMP 126.96.36.199 statistical package (SAS Institute, Cary, NC).
Forty-five patients (19 male, 26 female) with a median age at diagnosis of 7.1 years (age range 1.3–15.3 years) were included. Seven patients were excluded. Of these 7 patients, 5 were excluded because they were seen once in consultation but not followed up at the hospital, and 2 were excluded because of inadequate data. The mean followup was 4.88 years (range 1.12–8.18 years), and all but 2 patients were followed up for more than 2 years. Of the 2 patients who did not complete followup, 1 died from meningitis after 1.93 years of followup, and the other died from cerebral edema secondary to the syndrome of inappropriate antidiuretic hormone secretion after 1.21 years of followup. Neither of these patients achieved disease remission and therefore, for the purposes of the present analysis, both were categorized as having had persistent disease. With these 2 deceased patients excluded, the mean followup time was in excess of 5 years, with a minimum followup of 2.98 years.
Twenty-nine patients were treated with corticosteroids at any stage of their illness, and 21 of these patients received a second-line agent (methotrexate [n = 21], intravenous immunoglobulin [n = 6], etanercept [n = 3], infliximab [n = 2], and anakinra [n = 1]). Seven patients received >1 second-line agent during the study period (3 patients received 2 second-line agents, and 4 patients received 3 second-line agents).
Monophasic disease was best characterized, on the basis of our observations, as a single episode of active disease not lasting more than 24 months, followed by inactive disease and without recurrence of active disease during the followup period. The duration of 24 months was selected because all patients characterized as having monophasic disease in this cohort had inactive systemic JIA by this time point. Persistent disease was defined as active disease for more than 24 months' duration. Polycyclic disease was defined as active disease for any period of time, followed by inactive disease for any period of time and then recurrence of active disease.
Nine patients (20%) received medication for <6 months once their disease became inactive. Therefore, technically, they could not achieve the criteria defined in remission I; nevertheless, all 9 patients had monophasic disease in accordance with the second and third remission criteria, and it was therefore deemed appropriate to consider them as having achieved remission I for the purposes of this analysis.
Figure 1 shows the distribution of disease course according to each of the 3 formal definitions of remission (remission I, remission II, and remission III). Rates of persistent and polycyclic disease differed in the remission I group when compared with those in the remission II and remission III groups; the only statistically significant difference observed was in the rate of polycyclic disease between those meeting remission I and those meeting remission III (20% versus 4.4%, respectively, difference of 15.6%, 95% confidence interval 2.4–28.7%; P < 0.05).
Remission I was the least stringent definition applied and resulted in 7 patients who achieved remission by these criteria. Among these 7 patients, 5 experienced a relapse prior to ceasing medication (all 5 were receiving nonsteroidal antiinflammatory drugs [NSAIDs] and 2 were receiving prednisone) and the remaining 2 experienced a relapse within 3 months of ceasing medication (both had been treated with NSAIDs and 1 with prednisone). Therefore, these patients were classified as having polycyclic disease according to the criteria of remission I and as having persistent disease according to the remission II and remission III criteria, thus explaining the lower percentage of patients defined as having persistent disease by remission I (Figure 1).
There was a small difference in the rates of persistent and polycyclic disease between patients meeting remission II and those meeting remission III, since 1 of the 3 patients classified as having polycyclic disease by the remission II criteria experienced a flare, with development of arthritis, systemic features, and abnormal laboratory findings, after 11.8 months of inactive disease while not receiving any medications, thus failing to achieve the remission III criteria. The remaining 2 patients with polycyclic disease had a disease flare after remission III was achieved, 1 of whom had a flare, characterized by arthritis, systemic features, and laboratory results, at 1.15 years after the beginning of remission, and the other had a flare, with arthritis and suggestive laboratory results but without any systemic features, at 5.36 years after the beginning of remission.
All of those patients who did not achieve the requirements of remission I progressed, over the followup, to have persistent disease in accordance with all 3 remission definitions. Of the 28 patients who achieved the requirements of remission I, 22 (78.6%) were weaned from medication and thus met the definition of remission II. Of the 22 patients who achieved the requirements of remission II, 21 (95.5%) met the definition of remission III. There was no statistically significant difference between the proportion of patients who relapsed after achieving remission II (3 of 26 [11.5%]) and those who relapsed after achieving remission III (2 of 25 [8.0%]).
Achieving remission by any of the criteria had a sensitivity of 100%, while a specificity of 81.3%, 89.7%, and 92.9% was observed when applying the definitions of remission I, remission II, and remission III, respectively, in predicting the presence of monophasic disease among these patients.
Predictors of disease course and time to remission.
Because the study aimed to identify early predictors of outcome, the remission II criteria were considered more suitable than the remission III criteria in defining the disease course in subsequent analyses, even though the remission II criteria had a slightly lower specificity for remission as compared with the remission III criteria. The features of the patients who were characterized by the remission II criteria are shown in Table 1. There were no significant differences in age at diagnosis, sex, duration of followup, presence of RF and ANAs, or levels of IgA (results not shown) between the 3 disease course groups. When comparing by ANOVA the characteristics of the patients classified as having a monophasic, polycyclic, or persistent disease course, only treatment with corticosteroids (36.8% versus 0% versus 100%, respectively; P < 0.0001) and treatment with any second-line agents (10.5% versus 0% versus 87.0%, respectively; P < 0.0001) were significantly different between the groups.
|Monophasic disease||Polycyclic disease||Persistent disease||All patients|
|Total no. of patients||19||3||23||45|
|Mean (range) age at diagnosis, years||6.4 (1.3–15.3)||10.9 (7.0–14.9)||7.1 (1.3–12.9)||7.1 (1.3–15.3)|
|Male, no. (%)||9 (47.4)||1 (33.3)||9 (39.1)||19 (42.2)|
|Steroids ever, no. (%)||7 (36.8)||0 (0)||23 (100)||30 (66.7)|
|Ceased steroids, if ever started, no. (%)||7 (100)||NA||11 (47.8)||18 (60.0)|
|Second-line therapy ever, no. (%)||2 (10.5)||0 (0)||20 (87.0)||22 (48.9)|
|Mean (range) followup, years||4.4 (2.6–7.9)||5.45 (4.4–7.4)||5.1 (1.2–8.2)||4.88 (1.2–8.2)|
The only difference between the monophasic and nonmonophasic (i.e., polycyclic and persistent disease course) groups at diagnosis, by univariate analysis, was that the nonmonophasic group was more likely to have disease of polyarticular onset. Significant differences in many of the laboratory features were seen, by univariate analysis, between the disease course groups at 3 and 6 months, as shown in Table 2. Among the clinical features at 3 months, lymphadenopathy, hepatomegaly, fever, and rash were significantly more frequent in the nonmonophasic group. Three of the 19 patients classified as having monophasic disease had at least 1 joint with active arthritis at 3 months, whereas 21 of 26 patients with nonmonophasic disease had joints with active arthritis (1 patient had 52 joints with active disease). At 6 months, a nonmonophasic disease course was associated with significantly higher rates of lymphadenopathy, hepatomegaly, fever, rash, and active arthritis. No patient with active arthritis at 6 months achieved remission while not receiving medications, and therefore all had a persistent disease course in accordance with the remission II and remission III criteria. The rate of treatment with corticosteroids was significantly higher in patients with a nonmonophasic disease course at 3 and 6 months, as was the use of second-line therapy.
|Feature||At 3 months||At 6 months|
|Monophasic (n = 19)||Nonmonophasic (n = 26)||Monophasic (n = 19)||Nonmonophasic (n = 26)|
|Mean (range) Hgb, gm/liter||111 (92–130)||99 (71–134)†||119 (102–140)||108 (72–123)|
|Mean (range) WBC count, × 109/liter||10.3 (4.8–20.9)||17.9 (6.8–46.0)†||8.9 (3.2–14.1)||16.4 (5.7–33.4)†|
|Mean (range) Plt count, × 109/liter||342 (209–494)||507 (250–982)†||307 (201–415)||464 (300–734)†|
|Mean (range) ESR, mm/hour||17 (1–32)||38 (5–110)†||13 (1–69)||45 (4–104)†|
|Lymphadenopathy||1 (5.3)||8 (30.8)‡||1 (5.3)||6 (23.1)‡|
|Splenomegaly||1 (5.3)||6 (23.1)||1 (5.3)||6 (23.1)|
|Hepatomegaly||1 (5.3)||8 (30.8)‡||0 (0)||6 (23.1)†|
|Serositis||0 (0)||3 (11.5)||0 (0)||1 (3.8)|
|Fever||1 (5.3)||11 (42.3)†||1 (5.3)||9 (34.6)†|
|Rash||1 (5.3)||13 (50.0)†||0 (0)||16 (61.5)†|
|Arthritis||3 (15.8)||21 (80.8)†||0 (0)||16 (61.5)†|
|Mean (range) joint count§||0.16 (0–1)||8 (0–52)||0||7 (0–35)|
|Steroid therapy||3 (15.8)||17 (65.4)†||1 (5.3)||10 (38.5)†|
Results of the multiple regression analyses are shown in Table 3. Only onset of polyarticular arthritis was predictive of a nonmonophasic disease course at the time of diagnosis; addition of other variables did not improve the predictive value of the model. However, the R2 and the area under the receiver operating characteristics curve (AUC) were both low, indicating that polyarticular arthritis at onset is a poor predictor of the disease course. At 3 months' followup, the multiple regression model was more robust, with an R2 of 0.42 and an AUC of 0.76. Fever and the presence of active arthritis increased the likelihood of a nonmonophasic course. At 6 months, the model was again more strongly predictive of a nonmonophasic disease course, with an R2 of 0.49 and an AUC of 0.92. An ESR of >26 mm/hour and use of corticosteroid therapy at 6 months were predictive of a higher likelihood of a nonmonophasic course.
|Feature||OR for nonmonophasic course (95% CI)||P||R2||AUC|
|Polyarticular onset||3.5 (1.03–12.8)||0.051|
|At 3 months||0.42||0.76|
|Fever present||13.8 (1.6–323.3)||0.037|
|Active arthritis||21.0 (4.3–145.7)||0.005|
|At 6 months||0.49||0.92|
|Corticosteroid therapy||14.4 (2.37–137.7)||0.008|
|ESR >26 mm/hour||54.0 (6.7–1,302.5)||0.0015|
The results of the Cox proportional hazards analysis, which examined the predictors of an earlier time to remission, are shown in Table 4. Oligoarticular onset appeared to be predictive of an earlier time to remission. Moreover, earlier time to remission was strongly associated with the absence of active arthritis, an ESR of <26 mm/hour, and no requirement for corticosteroid therapy at 3 and 6 months.
|Feature||Risk ratio for earlier time to remission (95% CI)||P|
|Oligoarticular onset||1.60 (1.0–2.6)||NS|
|At 3 months|
|No active arthritis||3.39 (1.90–7.16)||<0.0001|
|ESR <26 mm/hour||2.29 (1.29–4.48)||0.004|
|No corticosteroid therapy||1.95 (1.14–3.54)||0.014|
|At 6 months|
|No active arthritis||1,568||0.016|
|No corticosteroid therapy||2.73 (1.53–5.76)||0.0004|
|ESR <26 mm/hour||2.20 (1.15–5.58)||0.015|
This study demonstrates that it is possible to classify the disease course of systemic JIA as monophasic, polycyclic, or persistent. A majority of patients have a persistent or monophasic course, with only a very small proportion having polycyclic disease. The definition of remission as the absence of clinical or laboratory evidence of active disease for a period of 3 months while not receiving medications was an accurate predictor of remission to the end of the followup period and, as such, was useful in helping to define the disease course.
Polyarticular arthritis at onset and evidence of ongoing disease activity at 3 months (fever, arthritis, an elevated ESR, and corticosteroid use) and at 6 months (arthritis, corticosteroid use, and an elevated ESR) were associated with having a persistent disease course or a longer time to remission.
The disease course in systemic JIA has been described as monophasic, polycyclic, and persistent by a number of authors. Classification relies on the definition of remission applied to the study population. This definition needs to encompass 3 things: disease activity (active versus inactive), duration of disease inactivity, and impact of concurrent treatment. Lomater et al (11) used the European League Against Rheumatism criteria for disease activity, which takes into account the influence of drug therapy, but did not use duration of disease inactivity to define remission. This may be reflected in the higher rates of polycyclic disease (33.8%) and lower rates of monophasic disease (11.3%) reported in that study. Fantini et al (12) used a definition of remission that required having inactive disease while not receiving medications for 6 months, and showed that only 2.3% of patients had a temporary remission followed by recurrence of disease, and 65.9% had persistent disease. These proportions are closer to those seen in our study and may reflect the stricter definition of remission used. Other authors have suggested different time frames in defining remission, which range from 2 months to 2 years of inactive disease (13, 18, 19). Thus, there is a need for a uniform definition of remission and disease course in order to make these useful concepts in future studies.
Preliminary consensus criteria for remission in JIA were recently presented by Wallace et al (16) and incorporate clinical features (arthritis, systemic features, and physician's global assessment of health) and laboratory features (the C-reactive protein level and ESR) to define clinical remission as having inactive disease while receiving medications for 6 months and not receiving medications for 12 months. These criteria are preliminary and are currently undergoing prospective validation. The authors of these preliminary criteria have highlighted concerns about the arbitrariness of the time durations selected, the lack of biologic markers to reliably identify active disease, and the absence of any radiologic measures in these guidelines (16).
The results of this study suggest that in systemic JIA, if clinical and laboratory parameters similar to those presented by Wallace et al (16) (excluding the physician's global assessment) are used, a definition of remission requiring inactive disease while receiving medications (remission I) may not be useful in defining the disease course, but a definition of remission requiring inactive disease while not receiving medications for 12 months (remission III) can help define the disease course in systemic JIA. Furthermore, remission while not receiving medications for 3 months (remission II) is as sensitive (100%) and almost as specific (89.7% versus 92.9%) as the definition of remission requiring 12 months of not receiving medications (remission III) and has the advantage of being able to be applied significantly earlier in the disease course.
There are no previous studies examining the predictors of disease course in systemic JIA. This may be due, in part, to the fact that disease course has not been well defined in the past. We have previously reported that persistent systemic features or the need for corticosteroid therapy and thrombocytosis at 6 months were associated with a poor outcome (4, 14). Many of these characteristics were also identified as early predictors of disease course in the present study, which included a different cohort of patients studied in a different time frame as compared with that in our previous studies.
Our study needs to be interpreted in light of its limitations. The mean followup period was ∼5 years in this study. This may mean that patients currently classified as having monophasic disease could be reclassified as having polycyclic disease if late relapses occur with a longer followup period. However, the existing literature reports no evidence that this is likely. The strengths of our study lie in the careful and systematic collection of data from an inception cohort of patients who were evaluated soon after the onset of disease. These patients were followed up regularly, with collection of accurate clinical and laboratory data in a standardized manner.
Our study suggests that the disease course in systemic JIA can indeed be described as monophasic, polycyclic, or persistent; polycyclic disease is quite unusual. These categories can be characterized using a definition of remission requiring the absence of clinical and laboratory features of disease activity while the patient has ceased all medication for a period of 3 months. Clinical features at 3 and 6 months can be accurate predictors of a patient's disease course, particularly the presence of active arthritis and persistent fever at 3 months, along with an elevated ESR and need for corticosteroid therapy at 6 months. This information may be valuable in identifying patients at risk of persistent disease, and therefore having a higher likelihood of poor functional outcome, which would facilitate timely therapeutic intervention to limit the joint damage and subsequent disability in these patients.
- 1Juvenile rheumatoid arthritis. In: CassidyJT, PettyRE, editors. Textbook of pediatric rheumatology. Philadelphia: WB Saunders Company; 2001. p. 218–321., .
- 17Data desk. 6.2 ed. Ithaca (NY): Data Description Inc; 1996., , .