In the absence of validated diagnostic criteria for JoAS, diagnosis of JoAS in this study required the following two conditions (I) bilateral active sacroiliitis confirmed by magnetic resonance imaging OR unilateral active sacroiliitis confirmed by magnetic resonance imaging and active peripheral joint disease restricted to the lower extremities (hip, knee, ankle) AND (II) at least one of three clinical criteria: (A) limitation of lumbar spine motion in all three planes, (B) pain or history of pain at the dorsolumbar junction of the spine and/or (C) limitation of chest expansion to 2.5 cm or less at the level of the fourth intercostal space.

Patients included were at least 12 and up to 17 years of age with a weight of at least 30 kg. Manifestations were refractory to two different non-steroidal anti-inflammatory drugs given at appropriate dosage and for a total of four weeks.

Active disease was defined by a spinal inflammation score of at least three (see below) AND at least two of the following criteria: (1) back pain score of at least 3; (2) patient global assessment of disease activity of at least 3; (3) physical function score as determined by the BASFI of at least 3.

Patients had to have stable doses of NSAIDs. Low doses of corticosteroids of no more than 0.2 mg of prednisone per kilogram body weight per day, with a maximum of 10 mg per day, were allowed. Intra-articular and soft-tissue corticosteroid injections were not permitted for four weeks prior to the Screening Visit. Patients treated with etanercept or infliximab or adalimumab or anakinra at any time for any period or with antimalarials, gold salts, sulfsalazine, azathioprine, penicillamine, leflunomide, cyclosporine A, intravenous immunoglobulin or methotrexate within four weeks prior to the first administration of study medication, or with plans to begin the intake of these drugs were excluded. Additional major exclusion criteria were a history of any chronic disease other than JoAS, JRA/JIA, especially chronic renal disease, liver disease, hematological, gastrointestinal, pulmonary, cardiological or neurological disease, which in the opinion of the investigator might influence the efficacy or safety of the study medication or which in the opinion of the investigator might lead to an unacceptable risk for the patient.

On the basis of pharmacokinetic considerations and recent experience with dosing of adalimumab in children, a dose of 40 mg every other week was given. The study drug was provided as an injection solution in prefilled syringes containing 0.8 ml of placebo or adalimumab 50 mg/ml concentration. Stable doses of NSAIDs and low dose of corticosteroids (≤10 mg per day) were permitted throughout the study period.

This was a multi-center, randomized, double-blind, parallel-group Phase III study. Patients with JoAS received 1:1 40 mg of adalimumab or placebo subcutaneously every other week for a 12-week period (Controlled Phase). Clinical assessments were carried out at baseline and after 4, 8 and 12 weeks. At week 12 all patients who finalized the 12-week double-blind study received adalimumab. Further study visits occurred after 16, 20 and 24 weeks.

Clinical and laboratory evidence of adverse events on a routine basis were completed throughout the study. The investigator assessed and recorded any adverse event in detail on the adverse event form, including the date and time of onset, description, severity, time course, duration and outcome, relationship of the adverse event to study drug and alternative etiology for events not considered 'probably related' to study drug.

The clinical response to adalimumab was assessed using the Assessment of SpondyloArthritis International Society (ASAS) 3031, the pediatric ACR (PedACR) criteria 32, the Bath Ankylosing Spondylitis Disease activity score (BASDAI) and the Bath Ankylosing Spondylitis Functional Index (BASFI) 33. The individual domains contributing to the ASAS are (i) the spinal inflammation, defined as mean of items 5 (overall level of morning stiffness) and 6 (duration of morning stiffness) of the BASDAI; (ii) back pain, defined as mean of total back pain and nocturnal back pain; (iii) patient's global assessment; and (iiii) the physical function, defined as BASFI. The categories contributing to the PedACR 30 Score are the physician's global assessment of subject's disease activity (numeric rating scale, NRS), the parents' global assessment of subject's overall well-being (NRS), number of active joints (swelling not due to deformity or in joints without swelling, limitation of motion (LOM) plus pain and/or tenderness), number of joints with LOM, Childhood Health Assessment Questionnaire (CHAQ) and C-reactive protein (CRP). In addition, the erythrocyte sedimentation rate (ESR) was recorded.

The primary endpoint was the achievement of ASAS40 at week 12. A responder has been defined as a patient who achieved improvement in at least three of four domains contributing to the ASAS, with no worsening in the remaining domain. An improvement of a single domain is defined as a decrease of ≥40% and ≥2 points on the rating scale ranging from 0 to 10. Worsening of a single domain is defined as increase of >20% or >1 point on the rating scale.

The PedACR30 and the PedACR70 scores were calculated as published 32.

The efficacy analyses were performed in the intention-to-treat (ITT) population. The ITT population was defined as all subjects who were randomized and who have received at least one dose of the drug and at least one post-dose efficacy assessment at any dose. Further, to address the possible impact of major protocol violations, an additional 'per-protocol' population was defined excluding all subjects with premature discontinuation or major protocol violations. No substantial differences were observed between the ITT and the per protocol analysis populations. Therefore, the ITT analysis was chosen. Data on patients who prematurely discontinued were included in a last observation carried forward modus. The efficacy variables were analyzed either using Pearson's χ2 test, U-test, independent t-tests, analysis of covariance (ANCOVA) for repeated measures.

The study was conducted in accordance with the protocol ICH GCP, FDA regulations governing clinical study conduct, ethical principles that have their origin in the Declaration of Helsinki, 1996 revision and 2000 revision with subsequent clarifications, and all applicable local regulations. Before the study was initiated, the study protocol, the Informed Consent Form and Subject Information were submitted to the responsible independent ethics committee of the Aerztekammer Nordrhein, Duesseldorf, Germany for review and it was approved on 02 May 2008. Parents/legal guardian signed the Informed Consent form before any study-related procedures occurred.

The target population of 50 patients could not be reached and recruiting was stopped prematurely because of the expiration date of the study drug. Seventeen patients were randomized to receive adalimumab 40 mg/2 wks and 15 patients received placebo. Two patients (one of each group) discontinued prematurely due to lack of efficacy and were labeled as non-responders (Figure 1). Sixteen patients of the adalimumab group and 14 patients of the placebo group remained in the study and reached week 12, the final end point of the controlled part of the study. All these 30 patients entered into the open labeled phase at week 12 and remained in the study till week 24. One patient in the adalimumab group has been diagnosed with Complex Pain Syndrome and turned out to be an ASAS non-responder in both phase 1 and phase 2 of the study.

The detailed baseline patients' characteristics are outlined in Table 1. The median duration of specific complaints was 34.9 months, the median duration since diagnosis was 6.5 months. HLA-B27 was present in 76% of cases. AS in the family was known in 37.5% of patients. Two to six (median 2.5) different previous AS medications were reported; these were, predominantly, conventional NSAIDs (100.0%), COX-2-inhibitoirs (28.1%), oral corticosteroids (40.6%), intra-articular corticosteroids (37.5%), sulfasalazine (34.4%), methotrexate (31.3%) and COX-2-inhibitors (28.1%). Comparability of study groups was analyzed thoroughly at baseline. A higher proportion of patients of the placebo cohort had received oral corticosteroids (U-test, P = 0.04).

In the double-blind part of the study, more patients on adalimumab achieved an ASAS40, the primary outcome parameter, at week 4 (41%), week 8 (53%) and week 12 (53%) than on placebo (20%, 33%, 33%), while differences at week 8 only reached the border significance (P = 0.05). Secondary outcome parameter, ASAS20 and PedACR30 and -70 response rates at 4, 8 and 12 weeks were higher in the adalimumab group (Table 2).

In the adalimumab group at week 12, a decrease of all single disease activity parameters was noted compared to baseline. The mean (+/-SD) BASDAI spinal inflammation score decreased from 4.3 ± 2.1 to 1.5 ± 1.7 (-66%; P <0.001), the back pain score decreased from 5.5 ± 1.6 to 2.8 ± 2.9, (-48%; P <0.005), the BASFI score decreased from 3.8 ± 1.8 to 2.0 ± 2.3 (-47%; P <0.02), the CHAQ-DI score improved from 1.0 ± 0.4 to 0.6 ± 0.7 (-65%; P <0.005), the ESR improved from 23 +/- 27 mm/h to 6 +/- 3 mm/h (- 75%; P <0.05) and finally the CRP improved from 13 +/- 22 mg/l to 4 +/- 0.8 mg/l (- 80%; P = 0.07).

In the placebo group as well, some disease activity parameters improved while only the back pain score decreased significantly from 6.2 ± 1.4 to 4.3 ± 2.4 (-31%; P <0.02). In the placebo group the mean (+/- SD) BASDAI spinal inflammation score decreased from 5.0 ± 2.2 to 3.8 ± 3.0 (-25%; n.s.), the BASFI score decreased from 4.2 ± 2.3 to 3.3 ± 2.6 (-20%; n.s.), the Childhood Health Assessment Questionnaire-Disability Index (CHAQ-DI) score improved from 1.1 ± 0.8 to 0.8 ± 0.6 (-31%; n.s.), the ESR increased from 16 +/- 16 mm/h to 18 +/- 20 mm/h (+ 7%, n.s.) and finally the CRP increased from 5.4 +/- 9.6 mg/l to 10 +/- 26 mg/l (+ 88%) at week 12.

Improvement of the median of the individual disease activity parameters composing the ASAS score is given in Figure 2. The reduction of the "Spinal inflammation Score", the "Pain Score", the "Physical function score" (BASFI) and the "Patient's Global Disease Activity Score" was more pronounced in patients of the adalimumab group than the placebo group, but did not reach significance in the intergroup comparison.

Improvement of the individual disease activity parameters composing the PedACR Score is given in Figure 3. In patients of the adalimumab group, a marked reduction of the median of the "Physician's Global Disease Activity Score", the "Parent's Global Assessment of subject's overall well-being", the "Active Joint Count", the "Physical Function Score" (CHAQ), the "LOM Joint Count" and the CRP are evident, while there was no change in patients of the placebo group for "parent's global" and "active joint count", a weak decrease only for "physician's global", "CHAQ" and "limited motion joint count" or an increase for the CRP level.

Although a randomization has been performed, there were some differences between the patient groups at baseline (Table 1). Therefore, for statistical comparison, baseline adjustment has been performed (Table 3). The course of all single items was analyzed at different time points and in the repeated measurement design to assess change over time in the ITT population. The global treatment difference is outlined in Table 3. The difference for physician's global assessment of subject's disease activity after 8 weeks (P = 0.014) and 12 weeks (P = 0.043) were statistically significant, while the tests after 4 weeks (P = 0.067) showed marked tendencies in favor of adalimumab. The course of parents' global assessment of patient's overall well-being showed significant differences (week 4: P = 0.022, week 8: P = 0.045, week 12: P = 0.013) in the entire course of double-blind treatment phase. The course of the number of active joints showed significant advantages of adalimumab at week 8 (P = 0.002).

The course of CHAQ showed nearly significant differences in favor of adalimumab at week 8 (P = 0.054).

In the open label phase of the study, all patients received active treatment with adalimumab. The improvements obtained in the adalimumab group were maintained or even augmented in the open-label follow-up study phase in those patients of the sequence adalimumab - adalimumab (Figure 4A-D). Due to the improvement reached in the patients treated by the sequence placebo-adalimumab in the open label study phase, the treatment differences observed after 12 weeks between placebo and adalimumab showed decreasing tendencies. After 24 weeks, there were no or only slight differences in favor of the long-term treatment with adalimumab.

In the double-blind study phase 27 adverse events were observed in 10 of 17 patients on adalimumab (58.8%) and 29 adverse events were observed in 10 of 15 patients on placebo (66.7%), predominantly as 'Infections and infestations' (adalimumab: N = 6; placebo: N = 6) and 'General disorders and administration site conditions' (adalimumab N = 3; placebo: N = 4). Adverse events at least possibly related to the study medication (adverse drug reactions) were reported in four patients on adalimumab (23.6%) and five patients on placebo (33.3%), predominantly as 'General disorders and administration site conditions' (adalimumab: N = 3; placebo: N = 3). Drug related infections were reported only once on adalimumab (appendicitis).

In the open study phase, there were 19 adverse events in 10 of 16 patients of the sequence adalimumab-adalimumab (62.5%) and 29 adverse events in 7 of 14 patients in the sequence placebo-adalimumab (50.0%). The most common events were 'Infections and infestations' (adalimumab-adalimumab: N = 8; placebo-adalimumab: N = 6). Deviating from the double-blind study phase, the infections were predominantly classified by the local investigators as related to the study medication (N = 9), while 'General disorders and administration site conditions' occurred in only N = 3 patients.

Serious adverse events occurred in three patients during the double-blind study phase, two on adalimumab (appendicitis, tendonitis) and one on placebo (gastritis) and in four patients during the open study phase upon adalimumab, two in the sequence adalimumab-adalimumab (vertigo, general pain) and two in the sequence placebo-adalimumab (colitis, pyelonephritis). In two patients the events (appendicitis and pyelonephritis) were possibly related to adalimumab, the further serious adverse events were not related or probably not related, respectively, to the study medication (Table 4). No clinically significant laboratory abnormalities were related to the treatment with adalimumab. In all, the tolerability of the study medication was assessed very good or good in the majority of patients (patient's overall assessment: 27/32 (84.4%), investigator's overall assessment: 30/32 (93.8%)).

Shortcomings of the study are the low number of patients attributed to the rarity of the disease and the differences between both the adalimumab patient group and the placebo patient group at baseline, despite randomization. Furthermore, it would have been interesting to have MRI data at week 12 and 24 in the patients; perhaps we would have seen more differences between the groups.