Juvenile Arthritis Quantitative Imaging

This body of research will comprise a number of sub-studies, each designed to evaluate the use of quantitative MRI in JIA. Although patients will receive treatment during the study, this will form part of their routine clinical care and therefore this study is best regarded as observational.

1. Subjects and recruitment Patients will be recruited by clinical rheumatologists working in a specialist adolescent and young adult inflammatory arthritis tertiary referral centre, on a person-to-person basis. Patients will be aged 12-24 and have a diagnosis of JIA according to established clinical criteria.8,9 Individuals with enthesitis-related arthropathy and those with other forms of JIA (extended oligorticular and polyarthritis (rheumatoid factor positive and negative)) will be divided into separate groups and asked to have an MRI scan. Control subjects will be recruited from the same clinic, and will consist of individuals with mechanical back pain (non-inflammatory - thought to arise from vertebrae, discs or ligaments) and normal inflammatory markers. Such patients still require an MRI scan as part of their routine care to confirm the mechanical cause for their back pain and definitively rule out inflammation as a cause, but serve as non-inflammatory controls. All patients will be asked to provide written consent at the time of enrolment. Patients will include both patients with a new diagnosis of JIA and those with established disease who are already on treatment. For obvious ethical reasons, patients will not be divided into different treatment groups.
2. Study Design Patients will be divided according to diagnostic group - group A will consist of patients with ERA (and also controls who are subsequently diagnosed with mechanical back pain), while group B will consist of patients with ERA, extended oligo- and polyarthritis. As detailed below, the scan protocol will be different for these two groups. JIA patients in both groups will be scanned at enrolment and will have repeat scans every 6-12 months, depending on clinical need, for up to five years in total. Thereafter, scans will be performed as defined by the requirements of their clinical care up to year 10 of study. The control group in group A will only be scanned once. They will be asked to provide written consent prior to enrolling. The imaging component of the study will run for up to five years in total, and approximately 200 patients will be recruited in total. In group A, DWI and FF measurements will be performed in the SIJs. These measurements will be correlated (see statistics, below) with conventional scores of inflammation so that DWI and FF can be validated as biomarkers of inflammation in the SIJ (see below for more detail on these techniques). In group B, whole body MRI scans will be performed with a view to developing a quantitative score of inflammation in the whole body. This score will be validated by correlation with validated clinical measures of disease activity such as physician global assessment of disease activity, juvenile arthritis disease activity score (JADAS), Health Assessment Questionnaire (HAQ - measure of function) and and biochemical (blood) markers of inflammation (see below for further detail). This technique will be piloted in approximately 10 patients with JIA (ERA, extended oligo and polyarthritis) over the course of 1-3 months before beginning full-scale recruitment.
3. Scan Protocol The scan protocol will be tailored according to diagnostic group. Group A (ERA)

• T1-weighted axial and coronal pelvis, sagittal spine
• Post-contrast T1 axial and coronal pelvis, sagittal spine
• STIR, axial and coronal pelvis
• Diffusion-weighted images (b values 0, 50, 100, 300 and 600) of the pelvis
• mDixon images (in and out of phase images will be acquired and used to generate fat only and water only images; protocol design as per Dr Alan Bainbridge) Group B (oligoarthritis and polyarthritis) patients will undergo whole-body MRI including the following sequences:
• mDixon
• mDixon with contrast
• Diffusion weighted imaging The protocol will requires some technical development and optimisation at the start of the study. This will be undertaken in collaboration with Dr Alan Bainbridge. 2-3 b values (e.g. 50, 100, 300) will be used for the DWI sequences. Anatomical stations will include feet, knees, hips/pelvis and spine (TMJ to pelvis); the shaft of the femur and tibia will be omitted. Image quality is likely to be poor in the arms so these may not be imaged using dedicated sequences. STIR (short tau inversion recovery) methods will be used for fat suppression in the feet, knees and hips.

1. Group A (ERA) - Development of Scoring Systems
2. DWI DWI is already in use as a clinical scan in the ERA cohort. We aim to validate the use of ADC measurements as a quantitative biomarker in the ERA cohort. Apparent diffusion coefficient (ADC) maps will be generated with vendor software using a monoexponential fit. ADC maps will be analysed using a linear region-of-interest (ROI) technique, as previously described.10 Two or three ROIs are drawn across both joints on each image slice. Each ROI measuring 14-16mm in length, but is subsequently 'cut' to a standard 14mm length using dedicated Matlab software. This procedure is repeated on the central four axial slices (each of which measures 8mm in thickness). A further ROI will be drawn on normal sacral bone to use as a reference. Normalised ADC values (nADC) will be calculated for each ROI as the mean ADC divided by the mean reference ADC. In addition, the images will be analysed using histographic analysis of joint ADC values (using 'area' ROIs). Analysis will include thresholding methods and measurement of skew/kurtosis in inflamed compared to normal joints. It is hypothesised that there will be a significant, positive correlation between ADC and conventional STIR scores of inflammation (as shown by pilot data).11 The study will assess the correlation between the various ADC scores and the STIR score in order to determine which correlates most closely. ii. Fat fraction FF measurements are a novel approach to quantifying inflammation; our aim is to develop and ultimately validate FF as a biomarker. Inflamed joints will be analysed using region-of-interest (ROI) analysis on mDixon images, to determine the fat fraction (FF) in inflamed bone compared to normal bone. Polygonal ROIs will be hand-drawn across large joints in a standardized fashion. The pixel values derived from these ROIs will be analysed using multiple methods, including simple averaging and histographic analysis. Again, thresholding methods and analysis of skew/kurtosis will be explored as methods of assessing inflamed compared to normal joints. It is hypothesized that fat fraction will be decrease in inflamed bone compared to normal bone, due to infiltration by inflammatory cells and increased extracellular water. iii. Contrast uptake Post-contrast and perfusion images are already in clinical use in ERA. In this study, post-contrast mDixon images will be developed and validated. Patients will be injected with intravenous contrast and scanned after a fixed delay. Inflamed joints will be analysed on post-contrast images, which will be compared with pre-contrast images to determine contrast uptake. Inflamed joints will again be analysed using region-of-interest (ROI) analysis. Again, quantitative scores will be correlated with clinical data and STIR scores for biological validation.
3. Group B (ERA, extended oligoarticular and polyarthritis) - Whole Body Scoring Whole-body images will be scored using a selection of methods, and each will be validated against conventional measures to determine the optimal approach. Techniques will include a qualitative 'joint count' and also quantitative analysis of affected joints using histographic techniques.
4. Clinical Data Clinical data will be collected as part of routine care, in conjunction with specialist adolescent and young adult rheumatologists in the Arthritis UK Centre for Adolescent Rheumatology at UCLH. Validated clinical measures of disease activity will be documented at each clinic visit such as physician global assessment of disease activity, juvenile arthritis disease activity score (JADAS), Health Assessment Questionnaire (HAQ - measure of function), biochemical (blood) markers of inflammation such as C-reactive protein and erythrocyte sedimentation rate and for ERA validated measure of disease activity for spondyloarthritis will be collected (Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), Bath Ankylosing Spondylitis Functional Index (BASFI) and Ankylosing Spondylitis Disease Activity Index (ASDAS). All of the above measures are collected as part of routine clinical care. Patients' medication use will be documented at each clinic attendance again as part of routine care. This data will also be analysed as part of the study so that clinical measures of disease activity before and after treatment can then be correlated to radiological measures of inflammation (e.g. apparent diffusion coefficient and fat fraction) over time. Clinical data will be stored on an electronic medical record (as for routine data collection at UCLH). For analysis, the data will be downloaded onto secure hospital or university computers in encrypted form.
5. Statistical analysis

Statistical approaches have been chosen to biologically validate these quantitative markers of inflammation (ADC, fat fraction and contrast uptake) in JIA. Analyses will include:

• Analysis of correlation between clinical scores listed above to radiological measures.
• Comparison of ADC, FF and contrast uptake in affected joints compared to normal joints
• Comparison of ADC, FF and contrast uptake in affected joints before treatment compared to after treatment

In addition, patients will undergo serial scans during treatment. The study aims to determine whether ADC, fat fraction and measures of contrast uptake correlate with other measures of disease response (including clinical and biochemical measurements).

Statistical analyses have been chosen in collaboration with a university statistician from the Cancer Trials Centre (Dr Andre Lopes). Correlation analyses for Group A will be performed using Pearson product-moment correlation. In Group B, a mixed regression model will be used to evaluate the association between whole body inflammation scores and clinical scores. Intra- and inter-observer variability will be calculated for each parameter (in both groups) using interclass correlation coefficient (ICC) analysis and Bland-Altman plots.

Sample size calculations have also been performed in collaboration with Dr Andre Lopes. It is expected that approximately 100 patients will be recruited for both groups A and B, which will provide sufficient statistical power. nquery advisor was used to calculate the sample size.

For group A, our assumptions are as follows:

1-sided alpha 5%, power at 80%, null correlation of 0.50 and alternative correlation of 0.70 This suggests a required sample size of 65 patients. A sample size of 88 patients will provide a power of 90%.

For group B, as pilot data are not available, a comparable effect size and therefore sample size has been assumed (approximately 100 patients).

1. Long term goals (beyond the scope of this study)
2. The study aims to determine whether measures of inflammation have an impact on clinical management. This could be achieved by comparing treatment decisions made by clinicians when ADC, FF and contrast uptake information is available compared to when it is not in a 'virtual clinic'. Whether the scans are available will be determined at random (NB: As this this in a virtual clinic environment it will have no impact on patients' actual management). WB-MRI may be of particular value in this regard, as it has the potential to reveal inflammation in joints which the patient or clinician is not aware of.
3. A further, long term goal is to determine whether inflammatory burden correlates with health outcomes. If patients with greater inflammatory burden have poorer long term outcome, this will be an argument for earlier, more aggressive biologic therapy in these patients. Data will also be collected beyond five years (up to ten years) allowing the prospective identification of radiological predictors of poor outcomes in adults with JIA from analysing scan data in their adolescents years.
4. Introduction into Clinical Practice If quantitative measures of inflammation are shown to significantly change clinical practice, these scanning techniques will introduced into routine clinical care. These techniques also have the potential to be useful in other inflammatory disorders, such as rheumatoid arthritis in adults.