Four Symptom Profiles Emerge in Rheumatoid Arthritis When Inflammation Meets Patient-Reported Burden

Rheumatoid arthritis has long been treated as a disease whose misery tracks its inflammation, but a new clustering analysis suggests the relationship is far messier than that. In a study published in Arthritis Research & Therapy, the authors reported applying an unsupervised K-means algorithm to data from 283 participants in the Central Pain in Rheumatoid Arthritis Study, described as having established, active rheumatoid arthritis and initiating treatment changes. By combining objective inflammatory markers with patient-reported measures of pain, fatigue, sleep and psychosocial distress, the team identified four clinically distinct subgroups whose symptom burdens could not be explained by inflammatory activity alone.

According to the study, inflammatory activity was assessed with two measures: the 28-swollen joint count, described as assessing peripheral inflammatory activity, and serum high-sensitivity C-reactive protein, used to assess systemic inflammatory activity. The study reported that participants completed patient-reported measures covering pain, fatigue, sleep and psychosocial factors. The authors reported that clustering was performed on standardized symptom scores using baseline data, and that cluster differences in baseline demographic and clinical characteristics were then examined.

The four clusters split along revealing lines. Cluster 1, the largest at 117 participants, showed low-to-moderate inflammatory activity paired with low-to-moderate symptoms across all domains. Cluster 2, with 77 participants, was striking for its moderate inflammation but minimal symptom burden, a reminder that measurable disease activity does not always translate into felt suffering. Cluster 3, comprising 68 participants, carried the highest symptom burden, with severe pain, fatigue, sleep disturbances and psychological distress alongside only moderate inflammation. The smallest group, Cluster 4 with 21 participants, showed the highest inflammatory activity yet only low-to-moderate symptoms.

Statistical testing confirmed that the separation was not an artifact of the algorithm. Significant overall heterogeneity across the four clusters was observed for high-sensitivity C-reactive protein, the swollen joint count, and every patient-reported measure, with p values below 0.001. In other words, both the biological and the experiential dimensions of the disease varied meaningfully between groups.

The most clinically suggestive finding concerns Cluster 3.

The study also reported analyses of participants at 12 weeks. In exploratory analyses, all four clusters improved in both inflammation and symptom domains at the 12-week follow-up, but Cluster 3 remained the most symptomatic even after treatment changes.

Several caveats deserve emphasis. The analysis used baseline data to define the clusters, so the groupings are descriptive patterns rather than validated diagnostic categories, and the 12-week observations do not establish that any particular treatment works better for any cluster. The sample came from a single study of patients with active disease initiating treatment changes, which may limit generalizability to milder or newly diagnosed disease. The study has been published as an open-access article under Springer Nature’s early-release process, meaning it is citable with a permanent DOI but subject to further edits before the final version of record appears.

For now, the study's central finding is that in active rheumatoid arthritis, symptom burden was only partly related to inflammatory activity, and the authors identified a subgroup with features they described as consistent with centralized pain.

Subject of Research: Data-driven identification of symptom and inflammation profiles in rheumatoid arthritis

Article Title: Distinct symptom profiles in rheumatoid arthritis identified using data-driven clustering of inflammation and patient-reported measures

Article References: Aydemir, B., Muhammad, L. N., Song, J., Campbell, L., Wallace, B. I., Marder, W., Bolster, M. B., Bingham, C. O., Neogi, T., Winter, D. R., & Lee, Y. C. (2026). Distinct symptom profiles in rheumatoid arthritis identified using data-driven clustering of inflammation and patient-reported measures. Arthritis Research & Therapy. https://doi.org/10.1186/s13075-026-03935-2

Image Credits: AI Generated

DOI: 10.1186/s13075-026-03935-2

Keywords: rheumatoid arthritis, clustering, K-means, inflammation, patient-reported outcomes, centralized pain, fatigue, C-reactive protein, symptom burden, pain, Distinct, symptom