To the Editor:
We read with interest the article by Hermann et al (1), reporting on a study in which they compared 3 imaging methods in rheumatoid arthritis (RA) patients with shoulder pain. They detected erosions of the glenohumeral joint by radiography in 60% of patients, by ultrasonography (US) in 70%, and by magnetic resonance imaging (MRI) in 91%. Although US was superior to conventional radiography in detecting erosions, significantly more erosions of the humeral head were detected by MRI than by the other 2 imaging methods. MRI was also superior to US in identifying synovitis of the glenohumeral joint (28% versus 63%), tenosynovitis of the biceps tendon (35% versus 65%); and bursitis (30% versus 42%). We would like to raise some concerns about these results.
First, the finding of the low rate of erosions by US in the RA group is difficult to understand. The quality of an US examination depends not only on the experience of the examiner, but also on the equipment. Even in healthy adults, high-resolution US detects erosions (defined as a pit in the bone surface of >1 mm diameter in all 3 diameters) in 23% of shoulders (2). The low resolution of the rather old US equipment used by Hermann and colleagues could explain the low frequency of erosions detected in the RA group. Furthermore, comparison of the rates of erosions identified by MRI and US is a perilous undertaking, since there is no international consensus regarding the definition of erosions of the shoulder as detected by US. Because there is no gold standard, it remains uncertain whether the authors' classification of erosions for US and MRI pertains to the same pathology. For example, the observed cortical defects with hypointense signal on T1-weighted spin-echo images could represent osteoarthritic changes rather than RA-related pathology.
Second, gray-scale US is an excellent method for detecting (para)articular abnormalities such as synovial membrane thickening and proliferation, joint effusion, bursal effusion, peritendinous effusion, and rotator cuff tears (3). The authors stated that their US findings of glenohumeral joint effusion (28%), biceps sheath effusion (35%), and fluid within the bursa (30%) were consistent with those in a study by Alasaarela et al (4), but in fact they were not. The latter group reported these abnormalities in 92%, 83%, and 89%, respectively, of cases. This large discrepancy between the results of Hermann et al and those of Alasaarela et al again raises concerns about the quality of the US equipment. It is also a pity that the authors did not compare clinical examination for soft tissue changes with US and MRI findings. Although US may be less sensitive than MRI, it is likely to be more sensitive and specific than clinical examination.
Third, Hermann et al limited their examination to the glenohumeral joint. As control subjects, they examined individuals with shoulder pain. However, it is well known that shoulder pain may originate from the acromioclavicular joint, which appears to be involved more often than the humeroscapular joint in patients with RA (5). This raises questions as to whether the authors examined the correct joint. Examining both the glenohumeral and the acromioclavicular joint could have settled this issue.
We would like to make a final point. The report by Hermann and colleagues focuses on erosions in the shoulder joint. However, considering the fact that erosions in the glenoid fossa cannot be visualized by US it is questionable whether US should be used in the first place for this purpose. We believe the future of US lies in the direction of visualization of synovitis and bursitis, and making the distinction between inflammatory and noninflammatory disease.