Percentage of participants with ≥30% improvement in ACR core set consisting of 6 components: 1) Physician's global assessment of disease activity Visual Analog Scale (VAS), 2) Parent/Patient global assessment of overall well-being VAS, 3) Maximum number of joints with active arthritis, 4) Number of joints with limitation of movement, 5) Erythrocyte Sedimentation Rate, and 6) Childhood Health Assessment Questionnaire- Disability Index (CHAQ-DI) consisting of 30 questions in 8 domains.

Absence of fever was defined as no diary temperature recording ≥37.5° Celsius in the preceding seven days.

Intent-to-treat population includes all participants who had at least one dose of study drug.

Percentage of participants with ≥20 percent, ≥50 percent, ≥75 percent and ≥90 percent decreases in oral corticosteroid dose (mg/kg/day) from baseline.

Includes only participants on oral corticosteroids at baseline.

The six JIA ACR components consist of: 1) Physician's global assessment of disease activity, 2) Parent/Patient global assessment of overall well-being, 3) Maximum number of joints with active arthritis, 4) Number of joints with limitation of movement, 5) Erythrocyte Sedimentation Rate, and 6) CHAQ-DI.

At an assessment visit a JIA ACR30/50/70/90 response in comparison to Baseline is defined as: At least three of the six JIA ACR core components improving by at least 30%/50%/70%/90% and no more than one of the remaining JIA ACR core components worsening by more than 30%.

Intent-to-treat population includes all randomized participants who received at least one dose of study drug.

Physician's Global Assessment of disease activity is a Visual Analog Scale. The scale is 0 to 100 mm horizontal scale, the extreme left end of the line represents 'arthritis inactive' (i.e. symptom-free and no arthritis symptoms) and the extreme right end represents 'arthritis very active'. This item is completed by the treating physician.

Intent-to-treat population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. Last observation carried forward (LOCF) rule applied to missing JIA ACR core set components at Week 12.

The Parent/Patient global assessment of overall well-being is a VAS. The scale is a 0 to 100 mm horizontal scale, the extreme left end of the line represents 'very well' (i.e. symptom-free and no arthritis disease activity) and the extreme right end represents 'very poor' (i.e. maximum arthritis disease activity). This item is completed by the patient or parent/guardian as appropriate.

Intent-to-treat population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. LOCF rule applied to missing JIA ACR core set components at Week 12.

The maximum number of joints with active arthritis is 71 and these are defined as those in the joint assessment with: swelling present or pain present and limitation of motion.

The joint assessment is performed by an independent assessor, who is not the treating physician, blinded to all other aspects of the patient's efficacy and safety data.

Intent-to-treat population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. LOCF rule applied to missing JIA ACR core set components at Week 12.

The maximum number of joints with limitation of movement is 67 and these are defined as those in the joint assessment with 'limitation of motion'.

Intent-to-treat population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. LOCF rule applied to missing JIA ACR core set components at Week 12.

Erythrocyte Sedimentation Rate (ESR) is an acute phase reactant measured in mm/hour.

Intent-to-treat population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. LOCF rule applied to missing JIA ACR core set components at Week 12.

Functional ability is assessed using the CHAQ-DI. The questionnaire consists of 30 questions referring to eight domains; dressing/grooming, arising, eating, walking, hygiene, reach, grip and activities. Each domain has at least two component questions and if applicable to the patient there are four possible responses (0 = without any difficulty, 1 = with some difficulty, 2 = with much difficulty, 3 = unable to do).

The CHAQ-DI score is the sum of the domain scores divided by the number of domains that have a non-missing score. This overall score ranges from 0 (best) to 3 (worst).

Intent-to-treat population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. LOCF rule applied to missing JIA ACR core set components at Week 12.

Fever free was defined as no diary temperature recording ≥37.5° Celsius in the preceding fourteen days.

Participants from the Intent-to-treat population (all randomized participants who received at least one dose of study drug) who had a fever due to Systemic Juvenile Idiopathic Arthritis at baseline.

Percentage of participants with a change from an elevated hsCRP value at baseline to a normal hsCRP value at week 12; a change from anemia (low Hemoglobin) at baseline to a normal hemoglobin value at week 12; a change from thrombocytosis (elevated platelets) at baseline to a normal platelet value at week 12; a change from leukocytosis (elevated white blood cell count) at baseline to a normal white blood cell count at week 12.

Intent-to-treat population includes all randomized participants who received at least one dose of study drug. 'n' in each of the categories is the number of participants with data available at baseline and week 12 for analyses.

The percentage of participants receiving oral corticosteroids(CS) with a JIA ACR70 response at week 6 or Week 8 who reduced their oral CS dose by at least 20% without subsequent JIA ACR30 flare or occurrence of systemic symptoms at week 12.

At an assessment visit a JIA ACR70 response is defined as: At least three of the six JIA ACR core components improving by at least 70% and no more than one of the remaining JIA ACR core components worsening by more than 30%.

Participants from the Intent-to-treat population (all randomized participants who received at least one dose of study drug) who were taking oral corticosteroids.

Participants rated their pain by placing a horizontal line on a Visual Analog Scale on a scale of 0 (no pain)- 100 mm (severe pain). The score at 12 weeks minus the score at baseline. A negative number indicates improvement.

Participants from the Intent-to-treat population who had Pain VAS data available at baseline and week 12. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are excluded. LOCF rule applied to missing pain VAS at Week 12.

Percentage of patients who had at least a 0.13 improvement in CHAQ-DI score from Baseline to Week 12.

The CHAQ-DI questionnaire consists of 30 questions referring to eight domains; dressing/grooming, arising, eating, walking, hygiene, reach, grip and activities. Each domain has at least two component questions and if applicable to the patient there are four possible responses (0 = without any difficulty, 1 = with some difficulty, 2 = with much difficulty, 3 = unable to do).

Intent-to-treat Population. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are classified as non-responders.

LOCF rule applied to missing CHAQ-DI Scores at Week 12.

Percentage of participants who had a rash characteristic of sJIA in the 14 days prior to the baseline visit but no rash characteristic of sJIA in the 14 days preceding the Week 12 visit day.

Participants from the Intent-to-treat population for whom data was available. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are classified as non-responders.

Part I: Percentage of patients who had anemia (hemoglobin \

Participants from the Intent-to-treat population for whom hemoglobin data available. Patients who withdrew, received escape medication, or for whom the endpoint cannot be determined are classified as non-responders.

LOCF rule applied to missing hemoglobin values at Week 6 and Week 12.

The six JIA ACR components consist of: 1)Physician's global assessment of disease activity, 2)Parent/Patient global assessment of overall well-being, 3) Maximum number of joints with active arthritis, 4) Number of joints with limitation of movement, 5) Erythrocyte Sedimentation Rate, and 6) CHAQ-DI.

At an assessment visit a JIA ACR70/90 response in comparison to Baseline is defined as: At least three of the six JIA ACR core components improving by at least 70%/90% and no more than one of the remaining JIA ACR core components worsening by more than 30%.

Participants from the Intent to Treat population who reached the time point plus patients who withdrew because of insufficient therapeutic response and are assumed to have been non-responders.

Seventy-one joints were assessed for signs of active arthritis. The mean number of joints with signs of active arthritis is reported.

Participants from the Intent to Treat population who reached this time point. No data imputation is applied and patients with missing data are excluded.

Seventy-one joints were assessed for signs of active arthritis. The percentage of participants with no signs of active arthritis is reported.

The Intent to Treat population in Part II includes 112 participants who received at least one dose of study drug. Only those participants who reached this time point are included in the analyses.

Criteria for Inactive Disease:

1) No joints with active arthritis, 2) No fever, rash, serositis, splenomegaly, hepatomegaly (by physical exam) or generalized lymphadenopathy attributable to systemic juvenile idiopathic arthritis (sJIA), 3) Normal Erythrocyte Sedimentation Rate (<20 mm/hour), 4) Physician's global assessment of disease activity Visual Analog Scale (VAS) indicates no disease activity (where no disease activity is considered to be a score ≤10 mm on a 100 mm VAS).

Participants from the Intent to Treat population who reached time point plus patients who withdrew because of insufficient therapeutic response and are assumed to have been nonresponders.

Functional ability is assessed using the CHAQ-DI. The questionnaire consists of 30 questions referring to eight domains; dressing/grooming, arising, eating, walking, hygiene, reach, grip and activities. Each domain has at least two component questions and if applicable to the patient there are four possible responses (0 = without any difficulty, 1 = with some difficulty, 2 = with much difficulty, 3 = unable to do).

The CHAQ-DI score is the sum of the domain scores divided by the number of domains that have a non-missing score. This overall score ranges from 0 (best) to 3 (worst).

Participants from the Intent to Treat population who withdrew have been excluded at post withdrawal visits.

Percentage is based on only those participants who were on oral corticosteroid at baseline and reached a nominal visit day on which dose was calculated.

Participants from the Intent to Treat population who withdrew have been excluded at post withdrawal visits.

Rate of SAEs, Rate of Serious Infection AEs, Rate of Related SAEs (remotely, possibly, probably) to Tocilizumab (TCZ), Rate of Macrophage Activation Syndrome, Rate of AEs leading to withdrawal and Rate of deaths per 100 patient years (PY) were calculated using the formula:

Number of Patient Events / Duration in study (years) * 100.

Multiple occurrences of the same AE in one individual are counted.

Safety Population- all participants who received at least one dose of study drug and had 1 post-baseline safety assessment. Includes all safety data in the database up to the week 104 infusion based on the date of randomization for each patient. (Last date was 31 May 2011)

Percentage of participants with ≥30%, 50%, 70%, and 90% improvement in ACR core set consisting of 6 components: 1) Physician's global assessment of disease activity VAS, 2) Parent/Patient global assessment of overall well-being VAS, 3) Maximum number of joints with active arthritis, 4) Number of joints with limitation of movement, 5) Erythrocyte Sedimentation Rate, and 6) CHAQ-DI consisting of 30 questions in 8 domains.

The Part III intent-to-treat (ITT3) population consists of all participants who entered into Part III of the study and received at least one administration of tocilizumab during Part III.

JIA ACR core set consisting of 6 components: 1) Physician's global assessment of disease activity VAS, 2) Parent/Patient global assessment of overall well-being VAS, 3) Maximum number of joints with active arthritis, 4) Number of joints with limitation of movement, 5) Erythrocyte Sedimentation Rate, and 6) CHAQ-DI consisting of 30 questions in 8 domains.

The Part III ITT3 population

Oral corticosteroid values summarized are based on average daily dose on the nominal study day. The prednisone equivalent is used in calculation of oral corticosteroid dose. Participants who withdrew are excluded at the the visit of withdrawal and all subsequent visits.

ITT3 population; n=number of participants contributing to the specific statistic

Values summarized are based on average daily dose on the nominal study day. The prednisone equivalent is used in calculation of oral corticosteroid dose. Participants who withdrew are excluded at the timepoint of this event and at all subsequent visits.

Baseline considered first dose of study treatment. Data presented up to entry into the Alternative Dosing Schedule.

ITT3 population; n=number of participants contributing to the specific statistic

Values summarized are based on average daily dose on the nominal study day. The prednisone equivalent is used in calculation of oral corticosteroid dose. Participants who withdrew are excluded at the timepoint of this event and at all subsequent visits.

Baseline considered first dose of study treatment.

ITT3 population; n=number of participants contributing to the specific statistic

Participants who previously withdrew are excluded.

Responders are participants who met all of the following criteria for inactive disease at the visit assessment day:

i. Number of active joints = 0. ii. Absence of lymphadenopathy, hepatomegaly or splenomegaly in the nearest non-missing physical examination prior to or after the week assessment day. This could include results outside of the time window.

iii. Absence of symptomatic serositis adverse event. iv. In the 14 days preceding the week assessment day no fever (temperature >=37.5 C) or rash characteristic of sJIA.

v. Normal ESR as defined by an ESR <20 mm/hr regardless of age and sex. vi. Physician global assessment VAS <=10. LOCF rule applied to missing number of active joints, ESR and Physician global assessment VAS.

Data presented up to the point of entry into the Alternative Dosing Schedule.

ITT3 population; n=number of participants contributing to the specific statistic

Patients who previously withdrew are excluded Responders are patients who met all of the following criteria for inactive disease at all visits in the 6 months (180 days) prior to and including the visit assessment day: i. Number of active joints = 0. ii. Absence of lymphadenopathy, hepatomegaly or splenomegaly in the nearest non-missing physical examination prior to or after the week assessment day. This could include results outside of the time window.

iii. Absence of symptomatic serositis adverse event. iv. In the 14 days preceding the week assessment day no fever (temperature >=37.5 C) or rash characteristic of sJIA. v. Normal ESR as defined by an ESR <20 mm/hr regardless of age and sex.

vi. iv. Physician global assessment VAS <=10. LOCF rule applied to missing number of active joints, ESR and Physician global assessment VAS. ESR = Erythrocyte Sedimentation Rate. VAS = Visual Analogue Scale. Data presented up to the point of entry into the Alternative Dosing Schedule.

ITT3 population; n=number of participants contributing to the specific statistic

There were 4 levels of clinical remission defined while the patient remained on tocilizumab as defined below.. After level 1, each successive level required that all the previous level criteria be met:

Level 1: inactive disease criteria have been met at all assessments in the last 6 months (180 days) preceding the timepoint assessment day Level 2: level 1 criteria and no oral corticosteroids received in the last 6 months (180 days) preceding the timepoint assessment day Level 3: level 2 criteria and no methotrexate received in the last 6 months (180 days) preceding the timepoint assessment day Level 4: level 3 criteria and no NSAIDs received for sJIA in the last 6 months (180 days) preceding the timepoint assessment day

ITT3 population; n=number of participants contributing to the specific statistic

There were 4 levels of clinical remission defined while the patient remained on tocilizumab as defined below.. After level 1, each successive level required that all the previous level criteria be met:

Level 1: inactive disease criteria have been met at all assessments in the last 6 months (180 days) preceding the timepoint assessment day Level 2: level 1 criteria and no oral corticosteroids received in the last 6 months (180 days) preceding the timepoint assessment day Level 3: level 2 criteria and no methotrexate received in the last 6 months (180 days) preceding the timepoint assessment day Level 4: level 3 criteria and no NSAIDs received for sJIA in the last 6 months (180 days) preceding the timepoint assessment day

ITT3 population; n=number of participants contributing to the specific statistic

There were 4 levels of clinical remission defined while the patient remained on tocilizumab as defined below.. After level 1, each successive level required that all the previous level criteria be met:

Level 1: inactive disease criteria have been met at all assessments in the last 6 months (180 days) preceding the timepoint assessment day Level 2: level 1 criteria and no oral corticosteroids received in the last 6 months (180 days) preceding the timepoint assessment day Level 3: level 2 criteria and no methotrexate received in the last 6 months (180 days) preceding the timepoint assessment day Level 4: level 3 criteria and no NSAIDs received for sJIA in the last 6 months (180 days) preceding the timepoint assessment day

ITT3 population; n=number of participants contributing to the specific statistic

Human antibodies against human antibodies (HAHA), anti-tocilizumab antibodies were assessed by immunogenicity techniques from blood samples drawn every two weeks during Part III of the study.

ITT3 population; n=number of participants contributing to the specific statistic

Human antibodies against human antibodies (HAHA), anti-tocilizumab antibodies were assessed by immunogenicity techniques from blood samples drawn every two weeks during Part III of the study.

ITT3 population; n=number of participants contributing to the specific statistic