The change in terminology from Still’s disease to sJIA described in the ILAR criteria leads to situations where children clearly have this disease, but are unable to be classified as sJIA due to the absence of arthritis. Since the ILAR classification criteria were published in 2004, there has been extensive research on the pathogenic mechanisms involved in sJIA. The 2 most important findings, i.e. , the role of interleukin-1 beta and interleukin-6 in sJIA, has translated not only into a better understanding of the disease but has also lead to the use and approval of newer and possibly more effective therapeutic agents (anakinra and tocilizumab respectively)[12, 13].
Basic research has continued to uncover other cytokines and biomarkers which are characteristically increased in sJIA. These include myeloid-related proteins 8 and 14 (MRP 8/14)[14–16], interleukin-18 (IL-18)[17–19], macrophage migration inhibitory factor (MIF)[20, 21], S100A12 protein, soluble IL-2 receptor, NK cell function, and others. These biomarkers have been studied regarding their utility in diagnosing patients with sJIA, both individually[16, 18, 19, 21, 22] and as part of a panel[18, 25]. Initial results have been quite promising.These biomarkers may also be helpful in distinguishing children with systemic JIA from patients suffering from either other subtypes of JIA, other inflammatory diseases (like Kawasaki disease) or from infections. The baseline clinical characteristics, the mean value of many of the above biomarkers , and the response to therapy of 5 “suspected” sJIA patients without arthritis were described in the work of Vastert et al.. These “suspected” sJIA patients had baseline clinical characteristics and mean value of many of the above mentioned biomarkers similar to a comparison group of 15 sJIA patients fulfilling the ILAR criteria. In addition, the “suspected” sJIA also had a comparably robust improvement on anakinra. Unfortunately, these biomarkers are not routinely available in clinical practice. On the other hand, the Yamaguchi criteria for AOSD are a relatively simple tool with clinical and basic laboratory features.
Including the seminal publication of Cabane et al in 1990, there have been 6 publications[6–9, 26] comparing children and adults with Still’s disease. Most of these studies have revealed a lack of any significant difference between the two groups with respect to the systemic features[5–9], the articular manifestations, or sequelae[5, 7–9]. However, a few significant differences between pediatric and adult patients have been pointed out by Pay et al. with respect to clinical and laboratory features. Lin et al. has also noted differences between AOSD adults and sJIA children in articular outcomes.
Pay et al. showed significant differences with respect to the frequency of fever, rash, myalgia, weight loss, sore throat, LFT abnormalities and neutrophilia which were all higher in patients with AOSD. The pattern of joint involvement was also slightly different. Patients with sJIA had a higher rate of involvement of knee, ankle, elbow, metatarsophalangeal joints, hip joints as well as the cervical spine. However, in spite of these differences, given the large number of similarities the authors conclude that AOSD and sJIA may still be the same disease, and children may simply be reacting differently. This may be true given the fact that, with the exception of this one publication (Pay et al.), the other five publications revealed a lack of any significant difference between the two groups with respect to the systemic features or the articular manifestations at presentation[5–9].
Thus, it appears to be a widely-held belief that sJIA and AOSD are the same disease, albeit with slight differences. Given the similarities, it would be reasonable to expect that sJIA patients would fulfill diagnostic criteria for AOSD. In fact, in the study by Luthi et al as well, all 9 of their sJIA patients fulfilled the Yamaguchi criteria. Thus the fact that the Yamaguchi criteria were fulfilled in 74% of patients in our cohort was not as surprising as the fact that only 58% of patients fulfilled the ILAR criteria for sJIA. The performance of the ILAR criteria in the present cohort was comparable to that reported in a previous cohort from a multicenter registry in the state of Pennsylvania, USA (PASOJAR cohort). In the PASOJAR cohort, only 31% of patients (42/136 sJIA patients) fulfilled the ILAR criteria. The failure of ILAR criteria in our cohort was mainly due to the absence of arthritis.
Data pertaining to the delay in (or absence of) arthritic manifestations of sJIA is scarce. A cohort of 46 patients with “probable” Still’s disease (as defined by the Taplow group criteria) was published in 1962. These children were followed for a mean period of 5.9 years. Ten of these patients who did not evolve into any disease had unique features and were labeled as “benign systemic disease” due to the fact that all had fever with rash but none had persistent arthritis. Arthralgia was present in all of them and 7 of them had arthritis lasting only 2 to 3 days. In the PASOJAR cohort, it was reported that 12% of their patients did not have arthritis. Even in the study by Pay et al, in which all sJIA patients ultimately fulfilled the ILAR criteria, 5/25 (20%) sJIA patients did not have arthritis at initial presentation. The frequency of arthritis or arthralgia was not significantly different between sJIA and AOSD patient groups in the same cohort. Uppal et al reported that 30% of their patients with juvenile onset Still’s disease did not have arthritis and of these only 17% of patients had arthralgia. These numbers are lower than what was found in our cohort. This could possibly be presumed to be due to the fact that patients without arthritis are more likely to be referred to a tertiary level hospital.
As part of the effort to develop consensus based treatment plans and a standardized assessment schedule for sJIA, the Childhood Arthritis and Rheumatology Research Alliance (CARRA) in the US conducted a case-based online survey prior to a consensus meeting. Among the 63 pediatric rheumatologists who completed the survey the majority (87.9%) found the initiation of treatment in sJIA acceptable in the absence of arthritis based on fever and systemic features such as characteristic rash, serositis, and adenopathy, provided that infection and malignancy had been excluded. The final operational definition of sJIA developed by CARRA for initiating treatment in patients was different from the ILAR classification criteria for sJIA. The opinion of the consensus group was that the ILAR criteria were “too stringent” and might often lead to exclusion of patients who need to be treated as sJIA. Despite fulfilling the apparently more stringent ILAR criteria, 7 patients in our cohort failed to fulfill the Yamaguchi criteria. The components of the two criteria fulfilled by these patients are shown in Figure 3. In comparison to the entire cohort, the ILAR criteria appeared to be more sensitive in patients with arthritis. The ILAR criteria require the presence of any one extra feature (apart from fever and arthritis) for fulfillment of the classification criteria. However, in the same patients, three additional features (apart from fever and arthralgia due to arthritis) were required to fulfill the Yamaguchi classification criteria. It appears that in patients with arthritis the "ILAR criteria are more commonly fulfilled (100%) than the Yamaguchi criteria (61%). The Yamaguchi criteria seem to be more efficacious in the absence of arthritis (12/13) compared to patients with arthritis (11/18) (Table 4). It is also interesting to note that one patient was diagnosed as “suspected” sJIA based on splenomegaly, typical fever along with the typical sJIA rash.
One of the features typical of sJIA is the characteristic evanescent rash[2, 29]. The prevalence of rash in previous sJIA cohorts reported from Europe or North America has been quite high. (~80%)[27, 30]. Previous JIA cohorts from India have demonstrated that the prevalence of rash, in Indian sJIA patients, vary between 27 to 57%[31, 32]. Similarly, in our cohort only 18 patients (58%) had the typical rash associated with sJIA. The lower incidence of rash in studies in India may be due to darker skin color that makes the rash harder to see. This difficulty may represent a further limitation of the ILAR criteria in ethnicities associated with darker skin colors.
The ILAR criteria for JIA have been criticized on various counts. One of the criticisms has been the lack of consideration or integration of adult criteria for diseases, like Still’s disease, psoriatic arthritis and spondyloarthropathy, which occur in both children and adults. The utility of such integration was tested in a single cohort of 683 JIA patients in whom 157 patients had been grouped under undifferentiated as per the ILAR (Durban) criteria. A hierarchal tree approach with use of validated adult criteria reduced the number of patients in the undifferentiated category to zero. Analogous to this, in our cohort, when both criteria (ILAR criteria and Yamaguchi criteria) were used in an “either/or” combination, nearly all the patients (30 out of 31) could be classified as sJIA.
During development of the CARRA consensus treatment plans for new-onset sJIA, there was extensive discussion over the fact that patients in the early part of sJIA do not fulfill the ILAR criteria (probably due to lack of arthritis) and “yet need treatment”. The response of early pre-arthritic or “suspected” sJIA patients to anakinra in the study by Vastert et al., suggests that treatment of these “suspected” sJIA patients, may help us to utilize a window of opportunity, and possibly prevent some of the morbidity associated with sJIA.
There are many limitations to this study. The data was retrospectively obtained from a small cohort of patients diagnosed at a single centre. Being a tertiary level referral hospital, the cohort may not be representative of patients in the community. Telephone interviewing to recall the gap between arthritis and systemic features is a crude method at best. The lack of controls (other inflammatory diseases like Kawasaki disease or inflammatory bowel disease; malignancies and infections) prevents a determination of the specificity of the two criteria. Parts of the Yamaguchi criteria such as sore throat could probably have reduced specificity in children. The peak incidence of sJIA is between 1 and 5 years of age. Children at this age are less likely to complain of “arthralgia” or “sore throat”. The patients in the cohort were investigated at length and infections and malignancies were ruled out as possible causes of their symptoms . However, none of the patients were investigated for autoinflammatory diseases which are known to mimic sJIA. Since the prevalence of autoinflammatory syndromes is quite low with only case reports from India, it is unlikely that patients with these disorders could have formed a considerable number in the cohort. Nevertheless, it is possible that patients fulfilling the Yamaguchi criteria alone may have had a distinct disease which is similar to sJIA but not the same. Another limitation was the fact that long-term followup data was not part of the present study. It is possible that patients in our cohort with suspected sJIA may evolve into other rheumatologic disorders on long-term follow up. In the cohort of “probable” sJIA by Ansell et al., 20 out of 43 patients evolved into another disease on followup. Thus it is prudent to point out that given all these above limitations, the results pertaining to the primary objective of the study should only be viewed as hypothesis-generating.
As is evident from the present cohort as well as previous cohorts, patients with sJIA are a heterogeneous group. A subset of patients may not have arthritis for a significant period at the onset of the disease, but may still have elevated biomarkers associated with sJIA. Some of these new biomarkers have been found to specific to sJIA and could represent future diagnostic markers[16, 18, 19, 21, 22, 25]. Initial evidence seems to suggest that the subset of patients without arthritis, or with minimal arthritis, may respond to therapy equally well, or better, than patients with prominent joint involvement[4, 35]. Modification of the ILAR criteria with inclusions of these biomarkers, would help in classification of sJIA patients without arthritis, enable them to qualify for enrollment in trials as well as become eligible for treatment. However, these tests are not yet available outside the research setting and are likely to be expensive when they do become available. In the meantime, the use of the Yamaguchi criteria, with suitable modifications, in a hierarchal fashion, along with the ILAR criteria (analogous to that used in a previous Italian cohort), might prove to be a useful strategy to help classify “suspected” sJIA patients in resource poor settings.