Stringer et al. have recently reported on the results of a large survey of treatment in JDM [3]. In that study, a survey was sent to members of the Childhood Arthritis and Rheumatology Research Alliance (CARRA). As part of the survey, 11 clinical case scenarios were developed, based on actual patients. Each clinical case was chosen to represent a prototypic JDM presentation or course. All respondents received the case chosen to reflect moderately severe, typical JDM, as well as 3 of the remaining 10 cases. Respondents were asked to answer 3 open-ended questions about the cases they received: 1. What investigations would you order? 2. What medication therapy would you start? and 3. What non-medication therapy would you start? One hundred and forty-one CARRA members completed the survey (response rate 84%).

Stringer et al. found that there was considerable variation in the treatment of all cases, including the moderately severe, “typical” case. While there was general agreement on the need to treat with corticosteroids, different doses, durations and routes were used. Most respondents also used second-line medications, but there was marked variation in choice of medication and doses used. These observations are not surprising, given that there are little data on which to base treatment decisions.

The CARRA survey characterized the range of current treatment for JDM in North America. The goal of the present study was to use these data, supplemented by expert opinion, to develop a small number of consensus treatment protocols which reflect current initial treatment of moderately severe JDM, and can be studied and refined to improve treatment of JDM.

The paper case representing moderately severe JDM described a 5 year old girl with a 5 month history of progressive weakness and classic heliotrope and Gottron's rashes. She had difficulty walking up hills or stairs, and needed help to dress or brush her hair. She found it somewhat difficult to swallow, but there was no choking. Strength of the shoulder and hip girdle muscles was assessed as 3/5. She was unable to lift her head off the bed, and had a marked head lag when assisted to the seated position. She had a positive Gower's sign and when walking had a Trendelenburg gait with exaggerated lumbar lordosis. Blood work showed elevations in all muscle enzymes. Muscle magnetic resonance imaging and electromyography were both abnormal and consistent with myositis. She had a Childhood Health Assessment Questionnaire score of 2.125 (potential range 0-3, higher scores denote worse physical function) and Childhood Myositis Assessment Scale score of 21 (potential score 0-52, higher scores denote better strength and function).

The questions which were considered at this consensus meeting are listed in Table 1. Nominal group technique was used to reach consensus [4]. The process followed for each question is shown in Figure 1. First, there was an item generation phase. After 5-10 minutes of individual contemplation, a list of items was created without discussion. There was no limit on how many items could be contributed by an individual. Next, each participant had 1-2 uninterrupted minutes to speak to the group about which items they felt were most relevant to the question. There was then a vote. Voting was done using either stickers (questions 1, 3-6) or score cards (questions 2, 7, 8). In the sticker voting, each conference participant was given 7 coloured stickers. The items for the question were written on flip-charts. Participants then physically placed their stickers on the flip charts to indicate their support for an item. Stickers could be distributed in any manner the participant desired (including putting multiple stickers on 1 item). The total number of stickers associated with each item was then counted. In the score card voting, participants voted to support or reject each item individually. This meant that that an individual could vote in support of multiple items (for example, investigations to be done as part of initial work-up). There was no limit to how many items an individual could support. After this vote, each participant once again had 1-2 minutes to speak to the group to present their viewpoint on the issue; this was followed by a short, general discussion. There followed a second round of sticker-voting or score card-voting. For the sticker-voting, “natural” cut-offs became apparent for all questions (a relatively small number of items had most of the stickers/votes), and agreed to by ≥ 75% of participants. Items above each cut-off were retained while the remainder were discarded. For the score cards, a clear majority (≥ 2/3 of participants) was required to reflect consensus on each item. Where additional clarification was required, consensus was reached using either additional rounds of sticker-voting as above or through a show of hands (clear majority, ≥ 75% of participants required).

Conference participants agreed on the initial use of IVMP (30 mg/kg, daily for 3 days). The majority of participants also recommended the continued weekly use of single doses of IVMP, but this was not unanimous. This was retained as an option in protocol A and B. Participants also agreed on the dose and frequency of methotrexate (the lesser of 15 mg/m² or 1 mg/kg, maximum 40 mg). Although all participants agreed that the subcutaneous route for methotrexate was preferable, the option of oral methotrexate was left for circumstances where the subcutaneous route was not considered possible. The participants agreed on a starting dose of oral prednisone of 2 mg/kg, but there was also discussion of lower doses (0.5-1.5 mg/kg/day). There was considerable variation in suggested initial tapering of prednisone, but a consensus was reached that prednisone would continue at 2 mg/kg/day until 4 weeks, and then would be reduced by 20% if the patient was stable and doing well. There was some disagreement regarding the recommended dose of IVIG (2 g/kg given monthly (maximum 100-120 g) vs. 2 g/kg given every 2 weeks for 3 doses and then monthly (maximum 70 g)). A third round of voting resulted in decision for the latter approach (8 votes to 4).

Inclusion and exclusion criteria (Table 3), baseline investigations (Table 4), data collection (Table 4) and concomitant medications were common to all protocols. There was clear agreement on inclusion criteria, with all retained criteria having ≥ 10 votes and no rejected criteria having more than 3 votes. It was agreed that all items receiving any votes as exclusion criteria would be kept (unanimous decision). For baseline investigations, all items which received votes from ≥ 2/3 of participants as “definitely indicated” were retained. There was considerable discussion about the inclusion of muscle biopsy. However, in the end, it received only 4 of 12 votes as being “definitely indicated”, largely related to concerns about it not being done routinely in many centres. There was general agreement (11/12) on frequency of clinical and research assessments for children being treated with these protocols, as well as for which assessments should be included as part of a research evaluation. However, there was disagreement about the inclusion of the Pediatric Rheumatology International Trials Organization [5] and/or International Myositis Assessment Collaborative Study Group [6] core sets. These received the largest number of votes in the first round. However, through the discussion, concerns were raised that completion of the core sets would be burdensome to treating physicians. In the end, the complete core sets were included as optional assessments.

Concomitant non-medical treatments were considered required if agreed to by ≥ 2/3 of participants, with the remaining items being recommended. Dietary modifications and dietary consultation if possible (to minimize prednisone toxicity), sunscreen counselling and folic acid supplementation (if on methotrexate) were considered required while calcium/vitamin D supplementation and referral to physiotherapy and/or occupational therapy were recommended. There was some disagreement on whether supplementation with calcium and vitamin D should be required or recommended. After an additional round of discussion and voting, it was agreed that it should be recommended (11 in favour).

Despite its acknowledged value and ubiquitous use, the use of corticosteroids in the treatment of JDM has never been studied in a randomized controlled trial (RCT). In fact, there have been almost no RCT's of any medications in JDM. The explanation for this is largely related to the rarity of JDM (incidence of 2-4 cases per million per year [7]), but also to the difficulties in studying this complex disease and the lack of tools with which to measure outcome. Recent efforts have largely eliminated the latter issue. There are now a variety of tools that have been shown to be valid assessments of muscle function [8-10], skin disease activity [11, 12], extra-muscular disease activity [13] and overall disease activity [13-16]. Unfortunately, researchers are still left with the challenges presented when studying a rare illness.

In order to improve outcomes in JDM and minimize both morbidity and mortality, it is necessary to optimize currently available therapeutic regimens, and to evaluate new medications which may become available. It will be difficult for traditional RCT's to achieve these goals because of the small number of patients available for study and the costs associated with conducting clinical trials which must include large numbers of centres and encompass vast geographic areas.

Development of these protocols is a first step in beginning to study treatments in JDM using a new approach. We envision a study where treating physicians could choose the protocol which most closely reflected their typical treatment of patients with moderately severe JDM. By collecting data in a standardized fashion, data could be pooled from multiple physicians, who are all engaging in routine care decisions. These data could then be analyzed using statistical methods which account for variations in illness severity and other factors measured at baseline which may influence outcomes (confounding by indication) [17, 18]. In this way, these protocols could be compared without the expense and complexity of infrastructure that would be required for an RCT. However, this assumes that it is possible to ensure that similar patients are being compared, and that differences between patients are accounted for. All important disease and patient characteristics would need to be measured. Advances in statistical methods (such as propensity scoring and other techniques) and development of a variety of validated measurement tools in JDM make this less problematic. However, it remains to be seen whether this type of analysis will ultimately be successful.

Our work should be interpreted in the light of potential limitations. Although these protocols should be similar to treatment decisions made by most pediatric rheumatologists, they cannot represent all possible options. For example, the initial use of hydroxychloroquine or cyclosporine is not addressed, as these medications were not as widely used in the CARRA survey. As well, data and expert opinion used in this project were primarily derived from North American pediatric rheumatologists. This may mean that not all pediatric rheumatologists will be able to identify a protocol which is similar to their usual practice. However, the protocols should have wide enough applicability to allow their evaluation. Finally, new data is emerging that there may be more sensitive, immunologically-based indicators of continued disease. These were not included in our consensus discussions, but may one day play a role in determining therapy.

It should be emphasized that these treatment protocols are not intended as treatment recommendations. They have been chosen to reflect care provided by clinicians. Given the near complete lack of clinical trial data, it is not clear if one of these protocols is the optimal treatment for children with moderately severe JDM. Future research will need to investigate this question.