Inflammatory disease assessments

The use of MRI allows a precise assessment of the bony and surrounding soft tissue structures within a targeted joint. MRI has been used increasingly in RA clinical trials primarily to assess peripheral joints (hands and wrists). The main findings of inflammatory “activity” detected by MRI include synovitis, tenosynovitis, and bone marrow edema (BME), while the “damage” findings include bony erosions and joint space narrowing (JSN). The Outcome Measures in Rheumatology (OMERACT) MRI group has devised a validated method for scoring joint abnormalities that has both good inter-/intra-reader reliability (reproducibility) and high sensitivity to change over time . This is known as the RA-MRI Score (RAMRIS) and represents an international collaborative effort to standardize evaluation and reporting of inflammatory and destructive joint abnormalities in RA ( Table 2 ). The RAMRIS uses a semi-quantitative assessment to grade pathology severity, while quantitative measures using dynamic contrast-enhanced MRI (for synovitis) and statistical shape modeling (for all pathologies) have also been studied . In clinical studies, the use of all these quantification techniques has thus far been restricted. In 2013, a task force of the American College of Rheumatology investigated the existing data and concluded that MRI measures met the conditions of the OMERACT filter for truth, discrimination, and feasibility in measuring relevant structural outcomes for randomized clinical trials of RA . In other words, after a critical review of existing data, MRI was found to provide a credible and comprehensive representation of both true inflammation and structural joint damage in RA and to perform well as a clinical outcome measure.

The use of MRI for synovial assessment offers a valuable potential strategy as it is sensitive to early pathology and change . Synovial inflammation may be detected on MRI as hypertrophy, effusion, and/or tissue enhancement post intravenous (IV) contrast (e.g., gadolinium) . Use of contrast allows both easier recognition of inflamed synovial tissue and a more reliable assessment of synovial inflammation . While this is advantageous in clinical trials, in routine practice, this improvement in visualization must be weighed against the increase in cost, invasiveness, and potential side effects of contrast agents. MRI measures of synovitis demonstrate excellent construct validity. Synovial hypertrophy detected on MRI correlates with overall synovial inflammatory activity on histological assessment . In addition, the extent of synovitis in the metacarpophalangeal joints (MCPJs) of RA subjects correlates with both macroscopic findings on mini-arthroscopy and clinical disease activity .

BME (osteitis) refers to an MRI feature on fat-suppressed sequences that can occur in various traumatic, degenerative, and inflammatory joint disorders including RA. As BME cannot be visualized by other imaging modalities, this finding is unique to MRI. BME on MRI is defined as a lesion within the trabecular bone with ill-defined margins and signal characteristics consistent with increased water content . Histopathologic studies on BME lesions in RA have demonstrated areas of cellular infiltrates in the subchondral bone, or osteitis . This lesion has been shown to be a precursor of bone damage/erosion corroborated by its strong ability to predict subsequent bone erosion .

Tendon sheath inflammation is also common in RA, and an MRI scoring system for tenosynovitis with good reliability has been introduced as a proposed adjunct to the current RAMRIS . The OPERA study (OPtimised treat-ment algorithm in Early Rheumatoid Arthritis) recently incorporated this outcome and demonstrated highly significant improvements in tenosynovitis from baseline in both treatment groups over the course of the trial . Therefore, MRI not only provides a useful tool for identifying a pathology that is difficult to assess clinically but also provides another responsive measure of inflammation.

MRI is consequently well placed to provide unique information and accurately quantify the inflammatory disease burden over time. Several studies have demonstrated that the burden of synovitis and BME are predictive of radiographic changes on X-ray. In fact, previous studies have demonstrated MRI BME to be the strongest predictor of subsequent CR erosions . Thus, MRI provides prognostic information that may help to inform risk stratification and early treatment decisions.

Structural damage assessment

Bone erosions are a reflection of bone damage in RA. Because of its three-dimensional acquisition (in comparison to the projectional nature of CR), MRI is more sensitive in detecting bone erosions; they are much more likely to be identified among patients with RA using MRI compared to CR . MRI identifies a substantially greater number of erosions among the finger joints compared to CR . MRI has been shown to have a high concordance rate in comparison to computerized tomography, arguably the gold standard for detecting bony damage .

Due to poor inter-reader reliability, cartilage assessment (via JSN assessment) was initially excluded from RAMRIS. However, an MRI JSN scoring system has been revisited with advances in everyday MRI technology and consequent improved image resolution, including that from higher magnet strengths (e.g., 3-T machines). The OMERACT MRI group has developed a JSN MRI score as a RAMRIS addendum for cartilage assessment . This scoring method has demonstrated good construct validity and intra-/inter-reader agreement With improving image quality, it is possible that scores of cartilage itself, rather than the JSN surrogate, will emerge.

Monitoring RA disease activity

MRI is a highly responsive imaging modality in terms of measuring changes in joint inflammation. In one study, an MRI composite measure of synovitis, tenosynovitis and BME was the most responsive measure of joint inflammation . With these attributes, MRI is being used increasingly as an outcome measure in RA clinical trials. A recent example of the effective use of MRI measures in a clinical trial is the RA-SCORE study . This study demonstrated highly significant improvements in synovitis and BME at 12 and 24 weeks, respectively, among 60 patients receiving rituximab compared to 60 patients who received placebo. By contrast, at 24 weeks, the Disease Activity Score 28 (DAS28) was not consistently different between the active treatment and placebo groups. This implies that MRI measures of inflammation are sensitive to change and more discriminative of the effective treatment. More studies are warranted, however, to confirm whether early improvements in synovitis seen on MRI with treatment could be used in a clinical setting to affect a physician׳s decision making in the setting of questionable improvements in clinical measures of disease activity.

A recent study also demonstrated that these early changes in measures of synovitis and BME at 12 and 24 weeks, respectively, are predictive of subsequent CR progression at 1 and 2 years, suggesting that changes identified on MRI with treatment indeed have clinical relevance . This study further demonstrated that prediction of CR progression with MRI was superior compared to standard clinical measures of disease activity (DAS28) alone.

These observations support the hypothesis that MRI measures add important information to traditional clinical assessments and therefore positively influence decision making, at least in terms of interpreting results of clinical trials. However, it remains unclear how physicians might apply MRI measures in routine clinical care, either alone or in combination with clinical disease activity measures or patient-reported outcomes. Data regarding MRI in real-world settings and its impact on clinical decision making are clearly needed before more widespread use can be recommended. Furthermore, standardization or automation of reading techniques and dissemination of such quantification techniques would be paramount.

In addition to identifying a greater number of erosions than CR , evidence also supports MRI as a more sensitive tool in following change in erosions over time. In the RA-SCORE study, there were significant differences between the rituximab-active treatment group and the placebo group at 24 weeks, while there were no significant differences seen in the group using CR . Other studies have demonstrated that MRI measures of bone erosion may identify differences between treatment groups in roughly half the time and with half the number of subjects in comparison to CR . A recent study comprehensively demonstrated that MRI measures of bone erosion would significantly reduce sample sizes and follow-up times if used as a primary outcome .

Studies have also demonstrated that JSN and cartilage loss as measured by MRI may demonstrate greater sensitivity and discrimination compared to JSN as measured by CR . In the Immunosuppressive Therapy for the Prevention of Restenosis after Coronary Artery Stent (IMPRESS) study, significant changes were seen over 24 weeks, while no significant change was demonstrated measuring JSN using the CR. The RA-SCORE study also demonstrated large differences in a measure of cartilage loss between treatment groups, while there were no significant differences in JSN when performing CR, suggesting greater discrimination using MRI .

In addition to being sensitive to change, a previous study also demonstrated that early MRI erosion progression at 12 or 24 weeks was significantly associated with later CR progression, suggesting that these changes are important and predictive of later damage. Therefore, documentation of early MRI progression at 12 weeks might potentially be used to aid in risk stratification and guide decisions to escalate therapy to prevent long-term damage. As already noted, greater sense of the real-world implications of incorporating these studies in clinical management is needed before clinical use can be recommended. Furthermore, despite evidence for the superior sensitivity to change with regard to structural damage using MRI, it is unclear whether use of MRI would be cost-effective for clinical practice where CR is likely cheaper and more accessible. Clinical trials assessing the impact of regular use of MRI on quality and cost of care are warranted.

Remission and MRI

Recent evidence has demonstrated that MRI inflammatory features are a frequent finding in RA clinical remission and low disease activity states . MRI evidence of inflammation among patients judged by rheumatologists to be in clinical remission is predictive of progressive CR changes . Therefore, in some patients who appear to be in clinical remission, MRI may provide important information on persistent disease activity; MRI may also play a role in identifying patients with elevated clinical assessments of disease activity who are at low risk of structural progression. Although there is little information from MRI studies at present to understand how to use this capability, both decisions are highly relevant to treatment changes.

It is likely that low levels of objectively measured activity are not associated with erosion progression, so it is important to explore “safe” levels of inflammation. This question is further complicated by the common co-presence of osteoarthritis-related inflammation in joints. The OMERACT MRI group has demonstrated that synovitis scores below an inflammatory activity “acceptable state” are at low risk of CR progression . Therefore, MRI may also be used to identify those patients who are unlikely to progress, and this may limit unnecessary treatment with potentially expensive biologic drugs. Studies that directly assess how MRI used in clinical care might influence decision making, quality of care, and cost-effective delivery of that care are a critical next step before MRI can be recommended for more routine use.

Inflammatory disease assessments

Due to greater accessibility to high-resolution US machines and the relative low cost (when compared to MRI) in setting up US capabilities, US is increasingly used by rheumatologists worldwide . Conceptually, US can be applied to diagnostic, prognostic, and outcome assessments in RA. It is used in both the clinical and research settings. Given the increasing use of US, clinicians will need to have a good understanding of the strengths and weaknesses that it can offer. US can directly visualize the inflamed synovium/tendon synovial sheath as well as damaged joints and tendon structures in RA . Greyscale (GS) US is particularly useful to delineate the thickened joint capsule (synovial hypertrophy) and tendon pathologies as well as bony structural changes (erosions and JSN). In addition, the use of power Doppler (PD) US can provide information on synovial and tendon synovial sheath vascularity. However, US cannot assess osteitis and lacks an acoustic window at certain joint sites (e.g., mid-carpal bones of the wrist) which precludes detailed assessment at these joint sites. The EULAR Outcome Measure in Rheumatology (OMERACT) US workgroup has provided consensus US definitions for common pathological lesions in inflammatory arthritis .

US can be used to quantify these RA joint pathologies. At present, the most widely accepted scoring method for synovitis assessment is semi-quantitative (e.g., grading on a 0–3 scale ranging from normal, mild, moderate to severe) ( Table 2 ) . This has been validated and shown to have good inter-/intra-observer reliability when US is performed using standardized acquisition protocols as well as the OMERACT consensus definitions . An atlas providing images to help standardize semi-quantitative US scoring of synovitis is also available . As synovitis has various imaging-detected components (synovial hypertrophy, effusion and PD vascularity), there have been efforts towards developing a global semi-quantitative score that combines these subcomponents . In a large RA cohort on biologics therapy, the OMERACT US task force has tested the use of a person-level US Global Synovitis Score (US-GLOSS), which combines both synovial hypertrophy and PD vascularity. Although further validation work is required for the US-GLOSS, there was early improvement of synovitis as well as sensitivity to change using the proposed grading . Recently, a multiplanar joint scoring method for synovial pathology has been devised and was evaluated in 30 patients with RA using US assessment in eight predefined planes at the second MCPJs . A visual analog scale was used to score US pathology severity. Although the standard midline plane was not identified as the most influential plane, the multiplanar joint assessment significantly correlated with the most severely affected region. Apart from semi-quantitative scoring, quantitative scoring of PD vascularity (by counting flow pixels using automated system within the region of interest) has also been developed . This has been shown to significantly correlate with MRI enhancement rate as well as radiographic structural progression in the finger joints of patients with RA, supporting good construct validity .

There are as yet limited data on the reliability of US for assessing tendon pathologies . In a small study cohort involving nine patients with RA and one healthy subject, a qualitative dichotomous US assessment (absent/present) was performed to detect tendon inflammatory and damage pathologies . In this study, the intra-observer reliability ranged from moderate for GS to good for PD assessment, while the inter-observer reliability was excellent for both GS and PD assessment. Recently, the OMERACT group has devised scoring methods for both tenosynovitis and tendon damage . A consensus-based semi-quantitative scoring systems (grading on a 0–3 scale) for both GS and PD tenosynovitis assessment was tested out in 10 patients with RA . The intra-observer reliability for both the GS and PD assessment was good, while the inter-observer reliability was good for PD assessment and moderate for GS assessment. The first US scoring system for tendon damage in RA (using a semi-quantitative three-grade (0–2) scoring system) was investigated in a small cohort of 12 patients with RA . The mean intra-observer and inter-observer reliability for this was shown to be excellent and good, respectively, among 12 rheumatologists who performed the US.

Structural damage assessment

One advantage of using US in RA is that it can detect joint erosions more sensitively than CR, reflecting its multiplanar acquisition . However, unlike CR with established scoring methods , there is still a need to reach a consensus on how best to quantify joint erosions using US. Existing methods quantifying erosion are quite varied and include qualitative dichotomous (yes/no) assessment , semi-quantitative assessment based on defined categories of erosion sizes , or the extent/appearance of erosions on the bony surfaces .

There are also limited data on the use of US for JSN or cartilage assessment in RA. Qualitative morphological cartilage damage assessment (using a 0–4-point grading system) was performed at bilateral second and third MCPJs of 20 patients with RA by two examiners, and the inter-observer reproducibility was reportedly moderate to good . US can also be used to measure metacarpal cartilage thickness (MCT). In a study involving 35 patients with RA, MCT has been shown to be significantly correlated with radiographic JSN . In the same study, US cartilage assessment at the MCPJs was shown to be closely related with cartilage anatomical measurement on cadaveric specimens. Further validation work on US cartilage assessment, especially in longitudinal cohort, is required before it can be more widely applied as a measure of structural joint progression.

Monitoring RA disease activity

US seems promising as an imaging tool for disease activity monitoring in the RA trial and practice settings . It is non-invasive and has no ionizing radiation which makes it an ideal imaging tool for repeated joint scanning. It permits multiple-site joint scanning and is becoming more feasible for use as US machines are now increasingly available to the rheumatologists in their routine clinical practice .

There are now several longitudinal observational cohorts that have used US to monitor patients with RA on conventional as well as biological disease-modifying antirheumatic drugs (DMARDs) (such as antitumor necrosis factor (TNF) therapies) , tocilizumab , and rituximab . In a study including 25 patients with RA whose active disease was treated with either tocilizumab or adalimumab, the improvement in PD synovial vascularity at the finger joints led to the suppression of radiographic progression at these joints . In another study, 20 patients with RA receiving rituximab were monitored using US, clinical and laboratory parameters . Overall, there was significant improvement in US-detected synovitis. Nine patients subsequently required retreatment with rituximab (between 6 and 9 months from the beginning of therapy). PD synovitis was able to detect disease activity onset before the occurrence of worsening clinical symptoms. In addition, recent randomized trials have used US within their study designs as outcome measures to effectively evaluate drug therapeutic effects .

As extended US scanning of multiple joint sites can be time consuming, there is a need to select a reduced number of joint sites for US monitoring . Reduced joint site scanning has been shown to be representative of extended joint site scanning and is considered a more feasible approach to adopt in clinical trials and practice , in comparison to using reduced joint counts in composite disease activity indices. At present, there is no consensus on the minimal number of joint sites to be scanned. Existing methods use various selection criteria such as frequency of involvement, logistic regression technique, feasibility, and representativeness of selected joints . A 2011 systematic review on US joint counts has highlighted two reduced joint count methods (a seven-joint US score and a 12-joint US score as having good validity). The seven-joint US score described by Backhaus et al. was used in daily rheumatology practice and reported significant improvement in 120 patients (91% RA)with joint inflammation over a 6-month period on various DMARD therapies. The joints selected were chosen a priori, and they included the wrist, MCPJs and proximal interphalangeal joints (PIPJs) of the index and middle finger as well as the second and fifth MTPJ on the clinically dominant side. The seven-joint US score was shown to be sensitive to change in a large RA cohort ( n = 432) on DMARDs over a 12-month treatment period . The 12-joint US score described by Naredo et al. included the bilateral elbows, wrists, MCPJs of the index and middle fingers, knees, and ankles. The selected joints were derived from a process of data reduction performed based on the frequency of synovitis and PD signal involvement at 44 joint sites from 160 active RA patients who were started on biologic therapies. Using US inflammatory results derived from the 44 joint sites, the 12-joint US was able to detect synovitis in all the study patients as well as PD signals in 91% of the study patients. At baseline and at 6 months, the 12-joint US count and index for synovitis and PD signal correlated well with the extended 44-joint US count and index for synovitis and PD signal .

Subsequent to the above-mentioned systematic review , there have been reports on the development of a six-joint US score as well as validation work on a 22-joint US score . The six-joint score (which includes bilateral wrists, MCPJs of the index finger and knee joints) was derived using data reduction from inflammatory findings at the 12-joint sites (formerly described by Naredo et al. ). The six-joint US was quick and easy to perform, had high correlation with DAS28 and was shown to be sensitive to change when used to monitor disease activity in a cohort of 45 patients with RA started on etanercept. In the 22-joint US score, the joints analyzed were the same joints as the DAS though with the thumbs and shoulders excluded (as these sites could be affected by other common rheumatologic disorders not related to RA such as osteoarthritis and rotator cuff pathologies ). The 22-joint US score was shown to be sensitive to change in comparison to the change in disease activity measured by DAS . Its reproducibility was shown to be affected by US machine quality as well as previous sonographer׳s experience with the score .

In the EULAR recommendations on the use of joint imaging, US (as well as MRI) is described as superior to clinical examination; both can be considered for more accurate assessment of joint inflammation and may also be useful for disease activity monitoring . Musculoskeletal US has been shown to influence treatment decisions in the rheumatology outpatient setting. In one study, of the 165 patients with rheumatologic conditions, US affected treatment decision in approximately 45 patients (about 27%) . This figure was about 52% (31 patients) among 60 patients with RA. In another study, involving rheumatology outpatients, DMARDs were altered in 13 out of 100 patients referred for US, and the vast majority (10 out of 13 patients) had extensive subclinical joint inflammation .

Recently, a targeted strategy using both US and DAS28 disease activity assessment to guide DMARD escalation has been tested out in a randomized controlled trial (The Targeting Synovitis in Early Rheumatoid Arthritis (TASER) study) among 111 treatment-naïve early arthritis/RA patients. Data from 53 early RA patients randomized to the US assessment arm of the TASER study showed that US has additional value over DAS28 in guiding modification of treatment decisions for a significant though small group of RA patients . The majority (71%) of DAS28-based treatment decisions were not changed by the additional use of US, although in the remaining 29%, US findings resulted in alteration in treatment decisions.

Remission and ultrasound

The EULAR RA imaging recommendations also support the fact that US (as well as MRI) can detect inflammation, which is predictive of subsequent joint damage, even in the state of clinical remission . Remission is an important target in RA drug therapy . It is now well established that US inflammatory findings are commonly visualized, despite RA patients being considered in clinical remission or low disease activity state . In a study involving 29 RA patients in clinical remission, the occurrence of US-detected joint erosions was significantly higher among those with PD synovitis versus those without PD synovitis . Studies have also shown that US inflammatory findings are predictive of subsequent radiographic structural progression (for GS synovial hypertrophy and PD vascularity) as well as disease relapse (for PD vascularity).

Currently, there are no established guidelines for the tapering or cessation of DMARDs in RA. Within a cohort of 47 patients with RA in disease remission (DAS28 < 2.6) who had cessation of their anti-TNF therapy, 20 patients received US imaging. In these 20 subjects, 11 had sustained remission, while nine had RA flares. Imaging features (including PD activity) were not associated with sustained clinical remission . In a separate study, including 42 patients with RA in clinical remission (DAS28 < 2.6), US assessment was shown to predict disease relapse within a short period after cessation of anti-TNF or tocilizumab treatment . In another recent study, PD activity was found to be the strongest predictor of biologic therapy tapering failure in RA patients who were in sustained clinical remission . As the numbers of subjects in these studies are relatively small, larger studies are needed to determine whether US has a role in guiding, tapering, or cessation of DMARDs.

At present, it is not clear whether there is a need to escalate DMARD therapy when US PD abnormalities are detected in patients who are apparently in remission (based on clinical criteria). Recently, an international group of ultrasonographers and rheumatologists belonging to a research network (known as the Targeted Ultrasound Initiative (TUI) group ) has provided a consensus statement, which proposes that targeting treatment to PD vascularity provides superior outcomes in comparison to treatment based on clinical targets solely. This presented the rationale for a randomized study using US for targeted therapy in RA . In particular, patients with RA in sustained clinical remission will be randomized to either a study arm, whereby their DMARD treatment will be escalated based on PD abnormality findings (even in the state of clinical remission) or another study arm without US, and changes in therapy will solely rely on clinical measures. It is expected that findings from the study will help understanding of the true clinical significance of US-detected PD abnormalities.