In the present study we investigated the agreement between ultrasonography, radiography and clinical evaluation in the assessment of RA and healthy finger joints, with MRI as the reference method. It showed high agreement between ultrasonography and MRI in assessing RA bone erosions in finger joints. Using MRI as the reference method, ultrasonography exhibited markedly higher sensitivity in detecting RA bone erosions than did radiography, without loss of specificity.
In agreement with a study of RA MCP joints conducted by Wakefield and coworkers [12
], we found that ultrasonography of those MCP joints with good accessibility by this modality (such as second and fifth) exhibited better correlations with MRI than did ultrasonography of joints only accessible in two planes (third and fourth). We found ultrasonographic bone erosions in many PIP joints in which MRI and radiography were unable to detect any destructive bone changes. This finding is probably explained by the use of 3 mm thick MRI slices, which must be considered suboptimal for the small PIP joints. In a heterogeneous group of patients with joint complaints, Backhaus and coworkers [14
] did not find any advantage of ultrasonography over MRI in assessing bone destruction in PIP joints. This may be due to use of a 7.5 MHz transducer with a distance pad that is inferior to the high-frequency transducers employed in the present study. Furthermore, Backhaus and coworkers employed thinner MRI slices than were used in our study (1 mm versus 3 mm), favouring MRI over ultrasonography. Another possible reason for greater frequency of detection of erosions in PIP joints with ultrasonography may be its higher resolution in relation to MRI.
Preliminary data were reported by Alarcon and coworkers [15
] and Lopez-Ben and colleagues [16
] on the detection of bone erosions with ultrasonography in the second and fifth MCP joints of RA patients. They reported that ultrasonography had high accuracy, with MRI as the reference method, in the second and fifth MCP joints. In a group of patients with nonerosive RA on conventional radiography, Magnani and coworkers [17
] visualized significantly more erosions in patients' MCP joints with ultrasonography than with MRI. Similar to our study, they used 3 mm thick MRI slices. Optimal technique would probably have improved the sensitivity of MRI.
Unlike in metatarsophalangeal joints [18
], we found no ultrasonographic erosive changes in the examined finger joints of control persons; this is in contrast to MRI, which showed several single erosive changes in these joints. Erosive changes in control persons were detected with MRI with a frequency twice that reported in another study from our group (4.2% in the present study versus 2.2% in the study by Ejbjerg and coworkers [19
]), but all except one were small. A possible reason for the MRI finding of erosions in the finger joints is that the visualized changes were subchondral cysts, which are not detected with ultrasonography because the employed ultrasound frequencies do not penetrate cortical bone. The less efficient/optimal blinding of the ultrasonographer as compared with the MRI evaluator might have caused bias toward finding fewer healthy control joints with erosions and synovitis by ultrasonography than by MRI.
The rate of detection of ultrasonographic destructive changes by MRI and radiography increased with the extent of erosion, as defined by its ultrasonographic grading. Correspondingly, the gradings of ultrasonographic inflammatory changes correlated with the volume-based MRI scoring of synovitis. Our results suggest that MRI and ultrasonography both allow assessment of abnormalities of the bone structures, and that performance differences are probably caused by technical aspects such as accessibility for ultrasonographic examination, high resolution of ultrasonographic assessment, or thickness of the MRI slices, rather than the physical principles of the examinations.
Ultrasonography had higher sensitivity for detecting signs of inflammation in the examined finger joints than did clinical examination, when MRI was considered the reference method, without considerable loss of specificity. Likewise, regarding the correlation of detection of synovitis between the methods, the moderate-to-good ICCs suggest that both ultrasonography and MRI were able to detect signs of inflammation. However, incomplete agreement between the methods suggested a margin of difference, probably due to ultrasonographic visualization of both 'active' and fibrotic pannus in the joints. The results are in agreement with those reported by Backhaus and coworkers [14
], who showed greater frequency of detecting synovitis with ultrasonography than with MRI. A large proportion of 'disagreement', in which ultrasonography alone showed signs of synovitis, was found in patients with early RA. This suggests that fibrotic changes, which are probably less frequent in the early stages of the disease, are not the only changes identified by B-mode ultrasonography and not visualized on MRI. Current knowledge does not allow definite conclusions to be drawn regarding the cause of the discrepancy between ultrasonographic and MRI findings.
The difficulty associated with recognizing both 'active' and 'inactive' synovial tissue may be alleviated by the addition of Doppler ultrasonography. The growing number of reports comparing Doppler ultrasonography with MRI [20
]. and histology of joints [22
],. and describing the advantages of supporting ultrasonography with Doppler evaluation suggests that it will soon become a routine aspect of the joint assessment. However, many methodological and technical aspects of the use of Doppler ultrasonography remain to be clarified [24
MRI did not permit visualization of joint effusion in RA finger joints, probably because of the minimal amount of fluid in the examined joints, whereas ultrasonography detected effusion in a considerable number of finger joints. This may be explained by the higher magnification of joints with ultrasonography than with MRI and the better resolution with ultrasonography. In our study, magnetic resonance images were read on hard-copy films. Evaluation on a computer screen, allowing magnification, would probably increase the sensitivity of MRI in detecting joint effusions. Additionally, MRI contrast diffusion into the joint cavity may contribute to making the detection of joint effusions with MRI more difficult [2
]. In contrast to MRI, ultrasonography is a dynamic, real-time examination method, which permits evaluation of the findings in motion and under compression. The latter is a distinct feature of joint effusion on ultrasonography, which may explain the apparent advantage of this modality over MRI in detecting it.
In the present study the detection of joint effusion on ultrasonography did not improve its sensitivity in comparison with MRI on detecting signs of inflammation because it most often accompanied synovial thickening. However, in joints in which accessibility may be problematic, joint effusion could be used as indirect proof of an ongoing inflammatory process. Other researchers reported difficulty in differentiating between synovitis and joint effusion [14
]. Standardization and precise definitions, as suggested in our earlier study [9
], may be helpful in this respect.
Localization of signs of inflammation showed the dominance of the palmar aspect in PIP joints and a slight dominance of the dorsal aspect in MCP joints. In our opinion, the uneven distribution of signs of inflammation warrants examination of the joints from all possible aspects in order to avoid losing important information on the extent of inflammation [27
With MRI as the reference method, ultrasonography almost doubled the sensitivity of assessing RA small joints for signs of inflammation compared with clinical assessment, without loss of specificity. The low sensitivity of clinical examination may account for the deterioration of RA patients despite clinically adequate control of the disease, as reported by Mulherin and coworkers [28
]. Accordingly, a longitudinal study conducted by Backhaus and coworkers [29
] showed progression of erosive changes with both ultrasonography and MRI, despite limited signs of clinical activity. The present study strongly suggests that clinical examination is far from optimal for assessing signs of inflammation in RA finger joints, and that the use of ultrasonography can considerably improve the detection of signs of synovial inflammation.