Introduction
The treatment of rheumatoid arthritis (RA) has improved dramatically over past decades with the early and intensive use of conventional disease modifying antirheumatic drug (DMARD) strategies [
1] and the introduction of biological agents [
2]. Treatment strategies with dose and medication adjustments tailored to the individual patient (tight control) to achieve a predefined level of low disease activity, or preferentially, remission within a certain limited period of time (treat to target) [
3] are nowadays used for early RA [
4]. It is widely accepted that early after the onset of RA, there is a period of time (window of opportunity) during which effective treatment can beneficially alter the outcome of the disease in the long term [
5,
6]. This requires prompt referral and recognition of RA. Recently, new classification criteria [
7] and new remission criteria [
8] have been published. In the new classification criteria, it is suggested that imaging techniques such as ultrasonography (US) may be used for additional information in joints clinically suspected of arthritis [
7]. Regarding remission criteria, a considerable number of patients in clinical remission according to several clinical criteria shows signs of inflammation on US [
9‐
11]. These findings imply that US may have added value to clinical examination when diagnosing RA, or evaluating remission in RA. For this purpose we would have to make a selection of joints to evaluate by US, because assessment of all joints would be very time consuming.
The aim of this systematic review is to clarify if US, when used for diagnosing RA and for evaluating remission in RA, would give additional information to clinical examination, to elucidate which minimal set of joints should be assessed by US for these two purposes, and by which modality, that is, power Doppler US (PDUS) and/or greyscale US (GSUS).
Material and methods
A systematic literature search was performed in PubMed, Embase and the Cochrane library for articles published up to October 4, 2011. A list of relevant keywords and synonyms for disease (RA, arthritis) and imaging (ultrasonography) was compiled. Keywords, including words of the title and abstract, and medical subject headings (Mesh) were combined using Boolean operators (AND, OR) (see Additional file
1). Included studies were those on adult humans, published in the English or Dutch language, either on diagnosing RA or evaluating US signs of synovitis in RA patients who were clinically in remission. We limited our study to the signs of US inflammation and did not assess structural joint damage. In the domain of diagnosis, assessing structural change probably would not increase the additive value of US very much, given the already very sensitive current 2010 American College of Rheumatology (ACR) classification criteria. Second, there would have been the problem of how to apply the US finding of structural damage to the 2010 ACR classification criteria. In the domain of remission, structural changes are a sign of past inflammation only. Disagreements about study inclusion were resolved by discussion; results are based on full consensus. Excluded were reviews, editorials, case reports and letters to the editor. One reviewer (DTC) screened titles and abstracts. Relevant articles were obtained and their reference lists were screened to find additional studies. Data were extracted by one reviewer (DTC) on year of publication, study population, study design and duration, treatment, possible follow up, number and type of joints under investigation, statistical methods and US parameters.
We used an adaptation of the phases (levels) in diagnostic studies proposed by Sackett and Haynes to reflect the clinical relevance of research data (Additional file
2, box 1) [
12]. All results were summarized descriptively. Heterogeneity in study design and methods precluded pooling the results. Methodological quality of studies on diagnosing RA was assessed by the instrument Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) [
13], with an extra question on sample size. For the studies evaluating remission a quality assessment tool was not available; we created a quality items list (Additional file
3). The quality assessments were performed to check for possible flaws in study design and analyses.
Discussion
The results of our systematic search indicate that when diagnosing RA a greater number of inflamed joints per patient was detected by US compared to clinical examination in populations ranging from aCCP/RF-positive patients with arthralgia, to patients with clinically observed arthritis. The presence of US signs of inflammation seems to increase the risk of progression to persistent arthritis or RA, implying clinical relevance. Regarding assessment of remission, our review shows that in many patients with low disease activity or in clinical remission, US signs of inflammation were detected, even in those who met stringent clinical remission criteria. These findings are relevant, because the results of these studies suggest that PDUS signs of synovitis predict progression of radiographic joint damage and flare. We limited our study to signs of US inflammation and did not assess structural joint damage. The reason for this is that in the publication of the new classification criteria, in which erosions are not included, it was suggested that US may be used to confirm clinical findings, i.e. swelling of the joint. Erosions typical of RA would imply the classification of RA in patients who met the new classification criteria in the past. However, the new classification criteria are very sensitive: the diagnosis of RA can be made on the basis of one swollen joint, so one could argue that for these new criteria, the finding of erosions would not add much to the sensitivity in early RA, in contrast to the situation with the 1987 criteria. Second, there would have been the problem of how to apply the finding of structural damage, assessed by US, to the 2010 ACR classification criteria, that is, what would be the contribution of structural damage assessed by US to the diagnosis, applying the 2010 criteria? In the domain of remission, structural damage reflects inflammation in the past, not the current inflammatory state.
An important question is, which joints should be scanned? Scanning only the joints that are painful or clinically show arthritis does not seem to be a valid strategy, and scanning all joints is not feasible in daily practice. Based on the spectrum of joints most frequently involved in early RA and the results of this review, a recommendation when scanning for diagnosis of early RA could be to scan at the minimum the wrists, MCP and MTP bilaterally using PDUS; PIP joints could be included based on the results of one study. Also in the domain of remission of RA it is important to identify which joints to scan. Although more signs of arthritis are found when scanning a larger number of joints, a clear relation between the number of joints scanned and the number of patients clinically in remission with US signs of synovitis seems lacking. Therefore, it might be sufficient to scan a limited set of joints for this purpose. In eleven studies the wrist and MCP joints of the dominant hand had been scanned as the minimum.
Based on the results of this review it seems that it is not necessary to scan large joints when diagnosing RA or evaluating the remission of RA. In general, the more joints that are scanned, the higher the chance of finding US signs of arthritis in a patient. An earlier diagnosis leads to earlier initiation of adequate therapy, more often within the window of opportunity. This not only improves the prognosis in the short term, for example, by inducing remission at an earlier stage and more frequently, but possibly it also favorably alters the long-term course of the disease.
Another important question is which modality to use, PDUS or GSUS? Our systematic search indicates that PDUS in particular may have an added value in the diagnosis of early RA and evaluation of the remission of RA: the predictive value of PDUS was higher than that of GSUS. This is in line with the findings that GSUS signs of inflammation also occur in non-arthritic individuals [
20]. In a study in an osteoporosis outpatient clinic, GSUS signs of synovitis were detected in up to 88% of 16 individuals who were without clinical symptoms or signs of joint disease (controls), based on scanning 42 joints with a cutoff of at least one joint with a score of 1 according to the OMERACT criteria for synovial hypertrophy [
20]. Of all 672 joints scanned, 76 joints showed GSUS signs of synovitis, 64 of them with grade 1, 12 with grade 2, and none with grade 3 signs. In another study, in which a total of 84 joints was scanned among nine healthy individuals, 23 joints showed GSUS grade 1 signs of inflammation and only one joint was scored grade 2; no joints had a grade 3 score [
18]. It seems that for the purpose of discriminating arthritis patients from non-arthritic patients, the use of the GSUS grade 1 score is debatable. Also in RA patients it is not clear what the significance of GSUS grade 1 is. One study states that in longstanding RA, GSUS might depict chronically thickened tissue without inflammation [
23]. At the patient level, a cumulative GSUS score for discriminating arthritis patients from non-arthritic patients has yet to be determined. A cutoff of 8 when scanning 22 joints has been proposed [
28].
Although the predictive value of PDUS is higher than that of GSUS to predict early RA, flare of RA and radiographic progression, PDUS has limitations as well. It is a technique that is particularly operator-, machine- and setting-dependent [
29]. It is important to avoid pressure on the transducer, and motion artifacts, and to use the correct US settings, for example, wall filter and pulse repetition frequency should be low when assessing joints.
Although the study results in our systematic review generally were not conflicting for either diagnosis or remission, some considerations need to be made. For instance, the number of diagnostic studies is currently limited, and only one study has focused on the ACR/European League Against Rheumatism (EULAR) 2010 criteria. Furthermore, regarding the quality of studies, the longitudinal studies looking at events (flares or radiographic progression) are small, causing a wide variation in the US risk estimates. Also, the variables that have been shown to be predictors of the diagnosis of RA or of remission, such as radiographic joint data and aCCP test results, have not all been taken into account. This might have inflated the added value of US. In addition, clear definitions for US signs of inflammation were not always given.
Something else to consider is that some of the papers reviewed are from the same group [
10,
11,
19,
21,
24,
25,
27]. Data presented in these papers might not be independent of each other, with correlated results being biased in one or the other direction. However, the results from the studies in our review are based on different patient populations. Also, we did not find signs that this group may be evidently pro or contra US, such that it would affect their scientific integrity, especially since one of the papers from this group shows a lower predictive value of US compared to those in other reviewed papers. Large prospective longitudinal studies are necessary to evaluate the additional value of US in diagnosing RA, scanning joints and evaluating the predictive validity of other signs such as US-detected tenosynovitis.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors participated in the conception and design of this study and interpretation of the data. DTC, JJL and JWGJ performed the data extraction and the quality assessment. All authors were involved in drafting the article or revising it critically for important intellectual content. All authors read and approved the final manuscript.