The primary objective of the study is to investigate the relative value of MRI or X-ray over standard of care as tools to guide DMARD pharmacotherapeutic decision-making by rheumatologists for the care of RA and UIA. It is hypothesized that disease progression determined by MRI or X-ray will result in an increased rate of pharmacotherapeutic escalation. Recent evidence suggests that a proportion of RA patients persist with diagnostic imaging evidence of disease progression while in clinical remission .
The potential impact of this study is multifold. First, it will determine if there is value in implementing diagnostic imaging-guided pharmacotherapeutic clinical decision-making at the RA population level or whether further investigation in a targeted subset is warranted. The clinical trial literature is replete with trials demonstrating less than 50% of the population progressing on diagnostic imaging over 2 years [77, 78]. However, predicting erosive progression is challenging [79, 80]. In the current study, a health economic evaluation will be integrated within the trial and will contribute to the discussion on the value for money of this approach in a sample comprised of RA and UIA participants.
Second, the relative merits of MRI and X-ray as prognosticators in RA and UIA will be determined. The increased sensitivity of MRI for erosions that arise from the direct paired-bone comparisons is evident [14–17, 29, 30] but must be tempered by the increased false-positive rate for MRI erosions in healthy controls [47, 81–83]. When the anatomies commonly imaged in practice are compared, the relative merits are less clear [14, 15]. The probability of erosion is uniform neither by patient nor joint. In active disease, substantial erosive changes were noted in many patient joints, while others were spared . These findings set the expectation that the greater number of bones and joints imaged on X-ray may negate the increased sensitivity of MRI in fewer joints. The advantages to imaging a greater number of joints needs to be taken into consideration with the increased demand on patients, cost, and false-positive rate. Further, the features measured for the X-ray and MRI interventions include more than just erosions and differ by modality. In RAMRIS, erosion score accounts for more than 69% of the total potential score . The other MRI features measured are more transient than erosions and account for a minority of the total potential score: synovitis, the most transient, accounts for up to 10% of the total RAMRIS; edema accounts for up to 21% of the total score . In contrast, vdHSS accounts for erosions and JSN. The former accounts 57% of the score for bilateral hands and wrists [45, 46]. There is sufficient uncertainty in the relative prognostic merits of MRI and X-ray to argue the position of clinical equipoise in relation to the primary objective.
The merits of the proposed study should to be balanced with its limitations. To this end, the study population may be less than ideal. Challenges exist to adjusting diagnostic imaging evidence of disease progression for measurement reliability. The data used to estimate the sample size requirements were limiting. Other limitations may exist as well. The implications of these limitations and alternatives are discussed below.
First, the study population includes patients with varying degrees of disease activity. This may cause background noise to the signal of the intervention. The evidence that 19% of RA patients in clinical remission progress on diagnostic imaging over 1 year  may be limited to a small proportion of patients in clinical practice with quiescent disease. For the remainder of patients with active disease, treatment decision-making based on standard of care disease activity findings may be sufficient. In the presence of clinically active disease, attempts to link the effect of diagnostic imaging evidence of disease progression on treatment escalation may be confounded. The signal from the intervention may be dampened by the noise associated with standard of care treatment escalation. Ideally the study population would be characterized by a treat-to-target standard of care and limited to patients with low disease activity or in clinical remission to minimize the effect of standard of care on treatment escalation. In this trial setting, where the clinical practice use of disease activity measures is limited , the potential impact of diagnostic imaging guided care may be magnified.
Second, the application of the SDD has some limitations. In the current study, the inter-rater SDD will be used for simplicity. The SDD was used given that all image sets are to be scored independently. With the well characterized scoring systems used, the independent scoring of individual image sets carries less potential for bias from sources both known and unknown. The SDD will also be assessed at two separate time points to account for changes in inter-rater reliability and hence SDD over time. The approach does not account for intra-rater reliability. In addition, the lack of pairing image sets for the same patient may result in less sensitivity to change compared to alternatives. The smallest detectable change (SDC) is an alternative approach that uses paired images to assess reliability and change. Use of the SDC produces less variability, enabling smaller differences in disease progression to be detected . Implementation of the SDC also carries incremental logistical complexity (for coordinating pairing) and rater scoring time. Acknowledging these tradeoffs, the SDD was proposed for the current investigation.
Third, for the sample size calculation, rates of disease progression were conservatively estimated from literature reports of disease progression between MRI and X-ray. Standard of care treatment escalation and a clinically meaningful effect size were not specifically accounted for in the estimate. Despite this limitation, the authors considered the sample size to be sufficient to resolve any incremental benefit of the interventions in the study population investigated.
A number of factors are expected to impact the relative merits of MRI and X-ray as prognostic tools in IA, including those affecting image resolution (for example, magnet strength to bore diameter ratio for MRI; number of views for radiography) and stage of disease. Measurement considerations including number of raters involved, their qualification and experience, number of readings and raters per image set, independent versus paired image rating, blinding the chronological order of image acquisitions, and current anti-rheumatic therapy are expected to affect the MRI and X-ray interventions equally.
This study represents an initial investigation into the clinical utility of MRI and X-ray in guiding anti-rheumatic pharmacotherapeutic decision-making for IA. The work will contribute to the knowledge of how diagnostic imaging may be leveraged to optimize rheumatologic care. Pending findings, future work will investigate the optimization of this intervention strategy, including but not limited to identifying population subsets at greatest risk, adjusting the interval over which imaging is conducted, testing in conjunction with a specific intensive DMARD clinical management strategy, and automated abnormality detection and quantification.