Biomarkers in axial spondyloarthritis diagnosis: from clinical signs to multi-omics integration

Axial spondyloarthritis (axSpA) is a chronic inflammatory rheumatic disease primarily affecting the sacroiliac joints and spine, with an estimated global prevalence of 0.3-1.5%. Despite advances in classification criteria and imaging techniques, diagnostic delay remains a persistent clinical challenge, with patients waiting a median of 2-6 years from symptom onset to confirmed diagnosis. Biomarkers-broadly categorized into clinical, molecular, and imaging modalities-offer the potential to shorten this diagnostic gap by providing objective, quantifiable, and earlier indicators of disease. This narrative review maps the current landscape of biomarkers for the diagnosis of axSpA across all three modalities, with particular attention to how machine learning and multi-omics integration are reshaping the diagnostic paradigm. We examine established clinical markers (HLA-B27, inflammatory back pain criteria, C-reactive protein), survey the molecular biomarker literature spanning genomics, proteomics, metabolomics, and liquid biopsy, and trace the imaging trajectory from conventional radiography and MRI through to radiomics and deep learning models that now approach the performance of expert radiologists. We further explore multimodal fusion-the integration of clinical, molecular, and imaging biomarkers into unified diagnostic models-and identify barriers to clinical translation: the lack of prospective external validation, inconsistent standardization, and unresolved questions about algorithmic fairness across diverse populations. We conclude that biomarker-driven precision diagnosis in axSpA requires continued biomarker discovery alongside the construction of shared infrastructure-multi-modal datasets, standardized pipelines, and prospective validation cohorts-to integrate existing and emerging biomarkers into clinically deployable tools.