Imaging has long played an essential role in the diagnosis and management of rheumatoid arthritis (RA). Conventional radiography has been used primarily to document structural damage, whereas musculoskeletal ultrasound (MSUS) and magnetic resonance imaging (MRI) have enabled us to visualize inflammation at a much earlier stage. The challenge now is to determine how these imaging modalities can be best incorporated into clinical practice, not simply to detect abnormalities but to identify patients and joints at risk of future structural damage.
New technologies often follow a familiar trajectory: initial enthusiasm, subsequent disappointment, and eventually, a more balanced appreciation of their true value. In RA, enthusiasm for MSUS-guided treatment strategies was tempered by the results of the Clinical Tight Control Therapy (ARCTIC) and Targeting Synovitis in Early Rheumatoid Arthritis (TaSER) trials, both of which failed to demonstrate superior clinical or structural outcomes with systematic ultrasound-driven treat-to-target strategies.1,2 These findings appropriately cautioned against indiscriminate escalation of treatment based solely on imaging abnormalities. However, they should not be interpreted as evidence that MSUS has little prognostic value. Rather, these findings may indicate that we have been asking the wrong question: not whether ultrasound should guide treatment in every patient, but in which patients, for what purpose, and at what time.
In this issue of The Journal of Rheumatology, Nakagawa et al provide an important contribution to this evolving perspective.3 In treatment-naïve patients with early RA, the presence of intraerosive power Doppler (PD) signals on baseline MSUS was independently associated with radiographic progression at 1 year, whereas conventional clinical and laboratory measures, including disease activity indices, were not. The finding that a specific ultrasound feature was more informative than global measures of disease activity is particularly intriguing. It suggests that the biological processes occurring within individual joints may contain prognostic information that is not captured by patient-level assessments.
The concept itself is not entirely new. Previous studies have shown that residual synovitis detected by ultrasound is associated with subsequent flare and structural progression, even in patients who meet clinical criteria for remission.4,5 In particular, PD signals located in contact with or penetrating the bone cortex have been associated with subsequent radiographic progression in patients in clinical remission.6 Quantitative PD measures have also demonstrated the potential to stratify the risk of structural deterioration in early RA.7 Taken together, these observations suggest that ultrasound may provide more than a snapshot of current inflammation; it may also offer a window into the future behavior of individual joints.
Nevertheless, the findings of Nakagawa et al3 should not be interpreted as evidence that every joint with an intraerosive PD signal requires immediate treatment intensification. Their data also highlight an important limitation of prediction. Among the anatomical units with baseline intraerosive PD signals, only a minority subsequently demonstrated radiographic erosion progression. Indeed, many units with intraerosive PD signals did not develop radiographic progression during follow-up. Conversely, only a proportion of joints that ultimately progressed had shown intraerosive PD signals at baseline. Thus, although this finding that intraerosive PD signal is strongly associated with subsequent damage, its positive predictive value and sensitivity remain insufficient for it to function as a standalone treatment trigger.3
This distinction is critical. Predicting disease does not automatically justify treating every predictive abnormality. The experience with ultrasound-guided treat-to-target strategies has already taught us that detecting more inflammation does not necessarily translate into better outcomes when treatment is escalated indiscriminately.1,2 The clinical value of a prognostic biomarker therefore depends not only on its statistical association with an outcome but also on whether acting on that information improves the balance between the benefits and harms of treatment.
An alternative approach may be to regard intraerosive PD signals as one component of a risk-stratification strategy. Rather than treating a PD signal itself as a therapeutic target, clinicians could use its presence, together with clinical disease activity, serological risk factors, existing structural damage, and longitudinal imaging findings, to identify joints or patients who warrant closer surveillance or reconsideration of treatment adequacy. Importantly, the findings of Nakagawa et al3 also raise the possibility that serial MSUS examinations may be more informative than a single baseline examination. Persistent or recurrent vascular signals may provide a more robust indication of ongoing destructive activity than that of a single observation.
This perspective represents a subtle but important shift in the role of imaging in RA. For decades, imaging primarily documented what had already happened. Radiographs showed accumulated damage, whereas ultrasound and MRI subsequently allowed us to detect inflammation that was clinically occult. The next step is prediction, namely, identifying which apparently vulnerable joints are most likely to become damaged. The goal, however, should not be prediction for its own sake. It should be prevention.
The message is therefore not “ultrasound instead of clinical assessment,” but rather “ultrasound in addition to clinical assessment, when it can answer a clinically meaningful question.” As Caporali and Smolen pointed out several years ago, current evidence does not support routine ultrasound-driven treatment decisions in all patients with RA.8 However, the limitations of a treatment strategy should not be confused with the limitations of the technology itself. The emerging evidence suggests that MSUS may have a valuable role in identifying high-risk joints and refining prognostic assessment.
The future of ultrasound in RA may therefore lie not in detecting more abnormalities but in identifying the abnormalities that matter. From detection to prediction, the next challenge is to determine whether this predictive information can ultimately be translated into better decisions and, most importantly, into prevention of irreversible structural damage.
Footnotes
See Intraerosive PD signals, page 1188
DECLARATION OF AI USE
Generative AI (ChatGPT, GPT-5.6 Luna; OpenAI) was used to assist with translation, language editing, and organization of the manuscript. The author critically reviewed and revised all AI-assisted content, verified the cited references, and takes full responsibility for the accuracy and integrity of the final manuscript.
FUNDING
The author declares no funding or support for this work.
COMPETING INTERESTS
The author declares no conflicts of interest relevant to this article.
- Copyright © 2026 by the Journal of Rheumatology







