Objectives: The aim of this pilot study was to compare clinical examination (CE) and ultrasound (US) with high field MRI (as the reference standard) for the detection of rearfoot and midtarsal joint synovitis and secondly tenosynovitis of the ankle tendons in patients with established rheumatoid arthritis (RA).
Methods: Patients with RA (as determined by the modified American College of Rheumatology (ACR) criteria) with symptoms of midfoot and rearfoot disease were recruited. Demographic data were collected. All underwent CE, US and high field MRI (with intravenous gadolinium contrast) of their right foot. Percentage exact agreement (PEA), sensitivity and specificity were calculated for CE and US when compared to MRI. Inter-reader reliability for CE and US was also assessed.
Results: Compared to the gold standard of MRI, for CE (joint synovitis) the ranges for sensitivity, specificity and PEA were 55–83%, 23–46% and 46–60%, and for US were 64–89%, 60–80% and 64–78%, respectively. Compared to the gold standard of MRI, for CE (tenosynovitis) the ranges for sensitivity, specificity and PEA were 0–100%, 20–91% and 55–91%, and for US were 0–67%, 86–100% and 59–86%, respectively.
Conclusion: CE was sensitive but US more specific in identifying hindfoot pathology in RA when compared to the reference standard of MRI. There was poor interobserver variability between ultrasonographers suggesting a need for standardisation of acquisition and interpretation of US images of the hindfoot.
Statistics from Altmetric.com
Foot disease in rheumatoid arthritis (RA) is a common problem, with at least 90% of patients affected during the course of the disease.1 Involvement usually begins in the forefoot and later progresses to involve the midfoot and rearfoot.2 3 Chronic inflammatory changes in the joints and local tendons are known to result in important and irreversible structural changes such as flattening of the medial longitudinal arch, valgus deformity of the calcaneus and tibialis posterior tendon dysfunction.4 5
Clinical decisions are often based on clinical examination (CE) of the ankle alone (ie the presence of tender and swollen joints) but CE can be adversely affected by factors such as deformity, overlying structures, obesity and peripheral oedema. Radiography has limited value in detecting soft tissue changes and lacks tissue specificity. An earlier and more accurate method of identifying and therefore treating joint inflammation would be a clinical advantage.
Recently interest has therefore been directed towards the use of new imaging techniques such as MRI and ultrasound (US) in order to potentially improve our assessment of inflammatory disease of the foot.6 7 US and MRI have both previously been shown to be more sensitive than CE for detecting synovitis in the forefeet in RA8–11 but few data exist for the rearfoot.12 CE has been shown to correlate poorly with metatarsophalangeal (MTP) and tibiotalar joint effusion detected by US.13 Maillefert et al compared CE and high field MRI with respect to synovitis and tenosynovitis in the hindfoot in a mixed inflammatory arthritis cohort and also showed a poor correlation between imaging and CE.14 Lehtinen et al15 demonstrated that in painful RA ankles with normal x rays, US and low field MRI were superior to CE for the detection of synovitis and tenosynovitis, with high concordance between the imaging modalities but poor correlation with CE.
The aim of this pilot study was to compare CE and US with high field MRI (as the reference standard) for the detection of (1) synovitis of the rearfoot and midtarsal joints, and (2) tenosynovitis of the ankle tendons in patients with established RA.
PATIENTS AND METHODS
Patients with RA (according to the 1987 modified American College of Rheumatology (ACR) criteria)16 with symptoms of midfoot and rearfoot disease were recruited consecutively over a 3-month period from the rheumatology outpatient departments of two West Yorkshire, UK hospitals. Local ethics committee approval was obtained. All patients had had foot symptoms for less than 1-month duration. Demographic and disease data were also collected.
Clinical examination was performed by a podiatrist (JW) for the presence or absence of swelling in the right tibiotalar (TTJ), subtalar (STJ), talonavicular (TNJ) and calcaneocuboid (CCJ) joints. The following tendons were examined for the presence of tenosynovitis: tibialis anterior (TA) and posterior (TP), peroneus longus (PL) and brevis (PB) (assessed together), flexor digitorum longus (FDL), flexor hallucis longus (FHL), extensor digitorum longus (EDL) and extensor hallucis longus (EHL). An overall grade for the severity of the combined foot deformities was assigned (grade 1 = mild, grade 2 = moderate, grade 3 = severe), as described previously by our group.17 A second examiner (AB) examined the same joint and tendon areas in five randomly selected subjects in order to assess interobserver reliability.
US was performed by an experienced sonographer (RJW) using an ATL HDI (Advanced Technologies Laboratories, High Definition Imaging, Bothel, Washington, USA) 3000 machine employing a 10–5 MHz linear array “hockey stick” transducer. Examinations were performed on the same joints and tendons as clinical examination, although peroneus longus and brevis were assessed individually. In addition, the STJ was examined from the medial and lateral aspects. The presence or absence of synovitis and tenosynovitis was recorded where appropriate. Synovitis was defined as an abnormal hypoechoic area within the joint and tenosynovitis as an abnormal hypoechoic area around the tendon within the tendon sheath compatible with the Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) ultrasound group definitions.18 Comparison was made with the opposite side. A second experienced sonographer (POC) also examined the same joint and tendon areas in the same five randomly selected subjects (as for clinical examination) in order to assess interobserver reliability.
MRI of the ankle to the midtarsal joints of the right foot was performed using a Gyroscan ACS-NT 1.5 Tesla scanner (Philips, Best, The Netherlands) with gadolinium contrast enhancement. The MR imaging parameters were as follows: a 3D T1-weighted gradient-echo sagittal pulse sequence with repetition time 22 ms, echo time 9.2 ms and flip angle 55°; a 256×256 acquisition matrix; 150–190 mm field of view; 50–60 1.5 mm thick contiguous slices with pixel size 0.78 mm and an acquisition time of 212 s. Sequences were obtained before and after gadolinium contrast injection. The orientation of the right foot in the scanner was standardised using a bespoke non-metallic pronation-supination jig (as previously described by our group).17 To maximise the MRI signal, two elements of a commercial body coil were placed medially and laterally over the ankle regions inside the jig. The MRI images were assessed by a musculoskeletal radiologist (POC) for the presence or absence of synovitis at the TTJ, STJ, TNJ and CCJ joints and for presence or absence of tenosynovitis of the TP, TA, FDL, FHL, EDL and EHL tendons.
Using MRI as the reference standard, percentage exact agreement (PEA), sensitivity, and specificity of CE and US for each joint and tendon area were calculated. Subanalysis was also performed on the data divided into feet with a mild deformity and those with a moderate/severe deformity (as described above). PEA and prevalence and bias-adjusted κ (PABAK) values19 were calculated for the inter-reader reliability analyses.
A total of 22 patients (14 female, 8 male) were assessed. The mean age was 52 years (range 33–70) and mean disease duration was 6.8 years (range 1–20). All patients were on stable doses of non-steroidal anti-inflammatory drugs (NSAIDs). The following number of patients were on disease-modifying antirheumatic drug (DMARD) therapy: 10 taking methotrexate monotherapy, 1 sulfasalazine, 1 hydroxychloroquine, 1 gold and 6 on combination methotrexate and anti-tumour necrosis factor (TNF)α therapy. All patients reported symptoms of pain in the subtalar/midtarsal region with varying combinations of stiffness and instability. With respect to foot deformity grading, 8 patients had mild, 11 moderate and 3 severe deformity.
Frequency of pathology on MRI
In the 22 patients, synovitis was noted in the following joints: TTJ in 11 patients, STJ in 12, TNJ in 9 and CCJ 11. The commonest tendons involved were TP and peroneal tendons involved in 12 and 11 patients respectively. The least common tendons to be affected were TA and the extensor tendons. When dividing the patients into those with mild and those with moderate/severe foot deformity, an increase in all joint synovitis and TP tenosynovitis was seen in the latter group with all three modalities (data not shown).
CE compared to MRI (table 1)
CE detected synovitis in 63 joints compared to 43 with MRI. There was poor correlation between CE and MRI as demonstrated by low PEA (range 45.5–59.1). The sensitivities of CE were generally high except at the CCJ (range 54.5–81.8) but with poor specificity (range 23.1–45.5).
CE detected tenosynovitis at 53 sites compared to 38 with MRI. PEA values were generally higher than for synovitis except for the peroneal tendons (range 54.5–90.9). The sensitivities of CE were variable, ranging from 0–100%. Sensitivity was highest at the TP and EHL and the lowest at EDL and TA. Specificities were generally higher than for joints, ranging from 20–90.5. The highest values were seen for EHL and EDL and the lowest for TP and the peroneal tendons.
US and MRI detected synovitis in 55 joints each. PEA values ranged from 63.6–77.3. Sensitivities ranged from 50–88.9, the lowest being medial STJ and the highest TNJ. Specificities ranged from 60–80, the lowest being lateral STJ and the highest medial STJ.
US detected tenosynovitis in 31 tendons compared to 56 on MRI. PEA values ranged from 59.1–86.4 at the peroneal brevis and FDL respectively. Sensitivities ranged from 0–66.7 (lowest TA/EDL and EHL and highest TP) and specificities from 85.7–100 (lowest TA and highest FDL/FHL).
Interobserver reliability for CE (table 1)
Reader 1 identified synovitis in 14 joints whereas reader 2 identified 7. PEA ranged from 20–60, the lowest value being for CCJ and the highest for TTJ. PABAK ranged from −0.6–0.2 (the lowest for CCJ and the highest for TTJ and TNJ).
In all, 15 tendons were found to be affected by tenosynovitis by reader 1 and 11 by reader 2. PEA ranged from 0.4–0.8, the lowest being for FHL and the highest for TP, EDL and peroneal tendons. PABAK ranged from −0.2–0.6, the lowest for FHL and the highest TP, EDL and peroneal tendons.
Interobserver reliability for US (table 2)
Readers 1 and 2 both identified synovitis in 12 joints each, with PEA results ranging from 60–100, the lowest value for TNJ. PABAK ranged from 0.2–1, the lowest value again being for TNJ.
Reader 1 identified tenosynovitis in two tendons compared to five by reader 2. PEA ranged from 80–100, the lowest value for the peroneal tendons. PABAK ranged from 0.6–1, the lowest being for the peroneal tendons.
In patients with established RA with symptomatic foot disease, CE correlated poorly with MRI for the detection of synovitis and tenosynovitis. Although CE was sensitive for detecting pathology, it crucially lacked specificity especially for synovitis. By contrast, US had better correlation with MRI and was more specific for pathology (especially tenosynovitis) but lacked sensitivity. These data highlight the difficulty of using CE alone to assess the rheumatoid foot as findings may be unreliable. In particular, CE, unlike imaging, can be confounded by external factors such as obesity and subcutaneous oedema. This study was the first to our knowledge to employ high field MRI in a formal comparison of three assessment modalities (CE, MRI and US) in the rheumatoid foot. This was also the first study that assessed any differences in CE and US results related to the degree of foot deformity.
These data are similar to those found by Lehtinen et al15 although their results were qualitative and no formal statistical comparison was made between the three assessment modalities. In addition, low field 0.1T MRI rather than high field was used although the authors did use intravenous gadolinium contrast, which would improve sensitivity and specificity for synovitis. Maillefert et al14 found a poor correlation between high field MRI and CE with κ values of 0.1 for synovitis and 0.27 for tenosynovitis. The authors studied a mixed inflammatory disease cohort including only nine patients with RA so it is difficult to extrapolate further on their detailed results for different joints and tendons. Scheel et al20 compared US with high field MRI in only four patients with inflammatory arthritis demonstrating poor interobserver agreement (k 0.28) for US when compared to MRI; however, this figure was for a combination of ankle and toe joints. Premkumar et al21 examined the TP and FDL tendons in 31 patients with high field MRI and US (using colour Doppler), showing that US had a high sensitivity (80%) and specificity (90%) for diagnosing tendinopathy compared to MRI. None of these studies, however, assessed intraobserver variation.
There may be a number of different reasons why CE appeared poor in some cases. First, anatomical structures lie close together or may be overlying making it difficult to differentiate between adjacent structures; for example, the STJ and TP tendon. Second, the amount of inflammatory tissue present may be small, making it harder to detect particularly in areas predisposed to osteophytes such as the midtarsal region. Third, joints such as the TTJ and STJ lie relatively deep to the skin, making assessment by palpation more difficult. However, in this study we used an experienced podiatrist in order to maximise CE findings. It could be argued, however, that podiatrists are most effective at assessing deformity and have less experience at detecting synovitis.
The limitations of this study included the small number of patients studied, which limits the ability to extrapolate from these results to wider clinical practice. Only five patients were used in the assessment of interobserver variation, which may account for the low PABAK values seen for some joint and tendon areas. US showed a marked interobserver variation despite readers being experienced. This may be artefactual because of small numbers or because there is a lack of standardisation of image acquisition/interpretation.
Dichotomous scoring was used for the MRI and US images but a more quantitative assessment may have allowed us to identify a threshold above which CE proved to be much more specific in its identification of pathology. Radiographs were not performed and bone damage was not evaluated. Technical limitations included examining the TTJ only from the anterior aspect on US as per EULAR guidelines whereas MRI is tomographic and thus gives a potentially more accurate assessment. Due to the perceived low sensitivity of the power Doppler tool on the US machine used for this study, we were not able to employ this assessment in our US examination, which may have increased specificity but decreased sensitivity.
In conclusion, CE was sensitive but US more specific in identifying hindfoot pathology in RA when compared to the reference standard of MRI. Even though the foot is not part of the typical disease activity clinical assessment (such as the DAS28), foot problems are an important cause of disability in RA and imaging offers a means to further investigate and direct therapy (such as corticosteroid injections). This study has highlighted the poor interobserver variability between ultrasonographers suggesting a need for standardisation of acquisition and interpretation of US images of the hindfoot.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.