We read with interest the report by Gärtner et al in which the authors argue that, based on comparison of clinical swelling and ultrasound findings in patients with rheumatoid arthritis (RA) in remission, detection of low-grade sonographic findings may be oversensitive (). We would like to suggest that the data can be interpreted differently, particularly with additional background information.
Gärtner and colleagues emphasize the high overall agreement between swelling and high gray-scale (GS) and power Doppler (PD) ultrasound grades; however, we were surprised that only 25% of the joints that fulfilled the most stringent criteria for sonographic joint activity (i.e., both GS grade 3 and PD grade 3) were counted as swollen. This contrasts with the findings of others: in a remission cohort reported by us in 2008 (), 100% of joints that fulfilled the same stringent ultrasound criteria were found to be clinically swollen (Brown AK, et al: unpublished observations), and in a recently reported early arthritis cohort (), 93% of such joints were swollen (Nakagomi D, et al: unpublished observations) Gärtner et al may have been stricter than was the case in other studies in excluding joints for which there was some doubt about whether they should be classified as swollen, but such a large discrepancy with data from experienced units nevertheless raises concerns. These concerns are reinforced by the report of several joints that were clinically counted as swollen but did not show any synovitis by GS or PD ultrasound in their study. Although the proportion is very small, the presence of such joints suggests that clinical swelling is not absolutely specific even by their very conservative assessment. The diagnostic value of clinical assessment against sonographic findings is dependent not only on the cutoff point of ultrasound grades, but also on that of clinical assessment itself, which is clearly subjective.
Second, the authors report that the Health Assessment Questionnaire (HAQ) disability index (DI) () was associated with clinical but not sonographic findings. However, detailed results on the correlation of global sonographic findings (e.g., total scores or scores on a sonography-based clinical disease activity index) with the HAQ DI or change in the HAQ DI are needed in order to assess the superiority of clinical measures to sonographic ones, especially given that the sample size for individual patients was limited (n = 60) and the subanalysis (Table 3 in Gärtner and colleagues' report) may not represent the true associations. In addition, treatment information is necessary for accurate interpretation of the association between synovitis and functional outcome, given the effect of nonsteroidal antiinflammatory drugs (), corticosteroids, and tumor necrosis factor antagonists ().
Third, more detailed information on the methods used for the longitudinal analysis (e.g., interval or treatment modification) would be helpful for accurate interpretation. We are particularly interested in whether the physicians were blinded with regard to the baseline ultrasound findings since if they were not, the presence of inflammation seen on sonography could have influenced clinical decision making and resulted in treatment escalation, weakening the association of sonographic findings with relapse.
As Gärtner et al discuss, a more comprehensive sonographic assessment including other joints (e.g., large joints), other views (e.g., palmar aspect), and other synovial tissues (e.g., tenosynovium) could have demonstrated a closer association with disability, and furthermore, radiographic data would have been informative, as previous studies have shown ([2, 6, 7]). We completely agree with the authors that more detailed assessment of sonographic data is needed to fully understand the value of ultrasound in the management of RA.