CT-based structural phenotypes of thoracic small-joint involvement in psoriatic arthritis

This study sought to develop and evaluate an interpretable computed tomography (CT)-based structural phenotype framework for characterizing thoracic small-joint damage and differentiating CT-defined structural burden profiles in patients with PsA.

Methods: In this retrospective study, 279 patients with PsA who underwent chest CT were included. The sternoclavicular joint (SCJ), costovertebral joint (CVJ), and costotransverse joint (CTJ) were systematically assessed for cortical erosion, joint-space narrowing, osteophyte formation, cortical sclerosis, and partial or complete ankylosis.

A rule-based three-axis framework integrating destructive, proliferative, and locking lesions was constructed to classify patients into five mutually exclusive CT structural phenotypes. Upper-quartile CT structural burden was operationally defined using the cohort-specific 75th percentile as an internal stratification threshold.

Results: Structural lesions were detected more frequently in the CVJ and CTJ than in the SCJ, with involvement rates of 83.2%, 84.6%, and 47.0%, respectively. The framework classified patients as low structural burden, destructive-dominant, proliferative-dominant, destructive–proliferative mixed, or locked phenotypes. Total structural burden differed significantly across phenotypes.

Compared with the proliferative-dominant phenotype, the destructive–proliferative mixed and locked phenotypes were independently associated with upper-quartile CT structural burden after adjustment for age, sex, and disease duration, with adjusted odds ratios of 4.41 and 7.04, respectively.

Conclusion: This CT-based structural phenotype framework offers a structured approach to describing thoracic small-joint damage and differentiating imaging-defined burden profiles.