MRI Evaluation of Axial Spondyloarthritis: From Early Diagnosis to Therapeutic Response Assessment

Yun Wang; Benlu Chen; Jun Chen

doi:10.3791/73057

A subscription to JoVE is required to view this content. Sign in or start your free trial.

Review Article

MRI Evaluation of Axial Spondyloarthritis: From Early Diagnosis to Therapeutic Response Assessment

86 views

⸱

DOI:

10.3791/73057

⸱

August 21st, 2026

Yun Wang¹ , Benlu Chen¹ , Jun Chen¹

¹The Second People's Hospital of Wuhu

Summary

This narrative review provides a practical, evidence-informed MRI framework for adult and pediatric axial spondyloarthritis, from standardized acquisition and probability-based reporting to selective assessment of treatment response and early structural remodeling.

Abstract

Magnetic resonance imaging (MRI) has reshaped the evaluation of axial spondyloarthritis (axSpA), which comprises radiographic axSpA (historically ankylosing spondylitis) and non-radiographic axSpA, by demonstrating active inflammation and structural lesions before definitive radiographic sacroiliitis. Its clinical value is not the isolated detection of bone marrow edema (BME). MRI is most useful when acquisition is standardized, active and structural domains are reported separately, and findings are interpreted against clinical probability and mimics. This narrative review provides a practical framework for radiologists and rheumatologists. We distinguish classification from diagnosis; separate adult and pediatric pathways; summarize the Assessment of SpondyloArthritis International Society–Spondyloarthritis Research and Treatment Network (ASAS–SPARTAN) four-sequence sacroiliac joint protocol; compare the Spondyloarthritis Research Consortium of Canada (SPARCC), Berlin, and Canada–Denmark (CANDEN) scores; and provide four author-proposed reporting categories: “Typical for axSpA,” “Suspicious/equivocal,” “Nonspecific,” and “Alternative diagnosis favored.” Recent TRACE and COAST-V data show that inflammation can decrease markedly within 4 weeks and that erosion reduction, fat metaplasia, and backfill may begin during the same early interval or by 16 weeks; follow-up MRI may therefore depict overlapping suppression of inflammation and early structural remodeling rather than strictly sequential phases. However, MRI improvement is not a validated surrogate for long-term structural or patient-reported outcomes, and current guidelines do not support routine serial imaging. Advanced sequences, synthetic bone imaging, low-dose CT integration, and artificial intelligence remain limited by protocol heterogeneity, gaps in external validation, reader variability, and uncertain clinical utility. MRI should reduce diagnostic delay without assigning disease labels to nonspecific lesions; formal scoring and repeat imaging should be used only when results are likely to change management.

Introduction

Axial spondyloarthritis (axSpA) is a chronic inflammatory disease that predominantly affects the sacroiliac joints (SIJs), spine, and entheses. It comprises non-radiographic axSpA (nr-axSpA) and radiographic axSpA (r-axSpA), the latter historically termed ankylosing spondylitis (AS). Because contemporary imaging recommendations and most treatment studies address this spectrum, this review uses axSpA as the umbrella term and reserves AS or r-axSpA for historical criteria or studies requiring definite radiographic sacroiliitis¹^,²^,³^,⁴. MRI can show active osteitis, capsulitis, enthesitis, and synovitis as well as erosion, fat metaplasia, backfill, sclerosis, and ankylosis before or alongside changes visible on radiographs⁵^,⁶^,⁷^,⁸^,⁹^,¹⁰^,¹¹^,¹²^,¹³^,¹⁴^,¹⁵^,¹⁶.

MRI presents both important clinical opportunities and potential risks. Diagnostic delay remains common and is associated with pain, functional loss, and missed opportunities for timely treatment¹⁷^,¹⁸. MRI can increase confidence when pretest probability is enriched by inflammatory back pain, onset before 45 years, HLA-B27 positivity, elevated C-reactive protein (CRP), family history, or extra-musculoskeletal manifestations. Yet mechanical loading, pregnancy-related change, athletic activity, degeneration, fracture, infection, and neoplasm can produce similar marrow signal¹⁹^,²⁰^,²¹^,²²^,²³^,²⁴^,²⁵^,²⁶^,²⁷^,²⁸. A technically positive scan is therefore neither necessary nor sufficient for an individual diagnosis.

The primary audience is radiologists who acquire and report SIJ or spinal MRI and rheumatologists who use these findings in clinical decisions; trainees are a secondary audience. The review’s distinct contribution is an acquisition-to-action framework: standardized acquisition, separate active and structural assessment, probability-based reporting of equivocal findings, and time-aware interpretation of treatment response. It also makes explicit where statements arise from guidelines or consensus, where they reflect primary studies, and where the authors propose pragmatic reporting categories that have not yet been validated.

Access restricted. Please log in or start a trial to view this content.

Review and Perspective

Scope, terminology, and evidence framework

The modified New York and Assessment of SpondyloArthritis International Society (ASAS) criteria were designed to classify research populations, not to diagnose individuals⁵^,⁶^,⁷^,⁸^,⁹. Classification prioritizes reproducibility within defined entry criteria; diagnosis must incorporate the entire clinical picture, competing explanations, consequences of labeling, and evolution over time. Thus, an ASAS-positive MRI should not be used as shorthand for a confirmed clinical diagnosis, and a negative MRI does not exclude axSpA when inflammation is absent, intermittent, outside the imaged field, or suppressed by treatment¹²^,¹³^,¹⁴^,¹⁵.

Literature search and evidence selection

We performed a structured narrative search of PubMed/MEDLINE, last updated July 22, 2026. Search concepts combined axial spondyloarthritis or ankylosing spondylitis with MRI, sacroiliac joint, spine, BME, structural lesion, differential diagnosis, the Spondyloarthritis Research Consortium of Canada (SPARCC), Berlin, the Canada–Denmark (CANDEN) system, diagnosis, monitoring, treatment response, pediatric imaging, advanced MRI, synthetic CT, and artificial intelligence. Backward and forward citation tracking was performed from key international recommendations, consensus statements, validation studies, and recent treatment-monitoring reports.

Human studies and professional guidance directly relevant to acquisition, lesion definition, diagnostic interpretation, scoring, mimics, monitoring, or emerging applications were considered. Current international guidance, systematic reviews, randomized trials, prospective or multicenter cohorts, and validation studies were prioritized; seminal older studies were retained when they established criteria or instruments. This was a narrative rather than a systematic review, so no PRISMA selection process, formal risk-of-bias assessment, or meta-analysis was undertaken. The reporting examples and workflow algorithms below are author-proposed, pragmatic tools informed by published guidance, unless explicitly stated otherwise.

Standardized MRI acquisition

For diagnostic SIJ evaluation, the 2024 Assessment of SpondyloArthritis International Society–Spondyloarthritis Research and Treatment Network (ASAS–SPARTAN) international consensus recommends a minimum four-sequence, two-plane protocol²⁹. Three semicoronal sequences, aligned parallel to the dorsal cortex of S2, comprise T1-weighted imaging for fat signal and structural damage, a fat-suppressed T2-weighted or short tau inversion recovery (STIR) sequence for active inflammation, and an erosion-sensitive sequence optimized for the bone–cartilage interface. A fourth semiaxial inflammation-sensitive sequence, oriented perpendicular to the semicoronal plane, improves anatomical localization and assessment of mimics. The consensus standardizes acquisition; it should not be interpreted as proof that the protocol itself improves outcomes.

Intravenous gadolinium is not routinely required for typical adult diagnostic or follow-up examinations. It may be added when infection, neoplasm, a postoperative complication, or another atypical process is suspected¹²^,²⁷^,³⁰. Spinal MRI should also be selective: SIJ MRI usually has the highest yield in early disease, whereas vertebral corner and discovertebral lesions are less specific. Sagittal T1 and fluid-sensitive whole-spine imaging is most useful when spinal symptoms predominate, SIJ findings are inconclusive despite high suspicion, a trial requires whole-axial quantification, or fracture, neurologic compromise, infection, or an Andersson lesion is suspected¹¹^,³¹^,³²^,³³.

Longitudinal examinations require technical comparability. Coverage, plane, slice thickness, spatial resolution, field strength, coil, and fat-suppression method should be reproduced as closely as practical; otherwise, apparent lesion change may reflect acquisition rather than biology¹²^,³⁴^,³⁵^,³⁶^,³⁷. The request should identify the prior MRI, treatment start and exposure, symptom trajectory, CRP or erythrocyte sedimentation rate trend, and the decision to repeat the scan. Side-by-side review of baseline and follow-up studies is preferable to isolated reporting.

Protocol harmonization also supports multicenter studies, quantitative imaging, and artificial intelligence (AI), but it does not eliminate domain shift. Differences between 1.5-T and 3-T systems, vendors, reconstruction, erosion-sensitive sequences, and site-specific quality control remain important. Minimum acquisition standards, documented deviations, periodic protocol audits, and reader training should therefore accompany any numerical threshold, longitudinal score, or algorithmic output.

Adult and pediatric imaging pathways

In adults with chronic back pain that begins before 45 years, imaging follows assessment of pretest probability. Radiography can identify established r-axSpA, whereas SIJ MRI is preferred when radiographs are normal or equivocal, and suspicion persists; in an early-disease pathway, MRI may be the first cross-sectional test¹²^,¹⁴^,¹⁵. The report should combine active and structural lesions, distribution, technical adequacy, and competing explanations. Spine MRI is not a routine substitute for an adequate SIJ study, and the ASAS criteria should not be applied as a stand-alone clinical checklist.

Children and adolescents require a separate pathway. Juvenile spondyloarthritis may initially present with peripheral arthritis, enthesitis, or hip disease, and typical inflammatory back pain may be absent. When axial involvement is suspected, or objective evidence would change treatment, a dedicated SIJ MRI is generally more informative than radiography. Interpretation requires awareness of open physes, red marrow, non-ossified cartilage, and maturation-related joint-margin irregularity that can resemble BME or erosion³⁸^,³⁹.

Adult ASAS definitions and adult numerical thresholds should not be transferred automatically to children. In 109 children, the adult ASAS MRI definition had 26% sensitivity and 97% specificity for clinical juvenile spondyloarthritis, whereas global expert assessment was more sensitive but less specific³⁸. Recent pediatric data-driven lesion thresholds show strong agreement with expert image classification, but the reference standard remains expert imaging rather than independent clinical diagnosis⁴⁰. Equivocal studies should therefore be discussed with pediatric rheumatology and pediatric musculoskeletal imaging experts.

Routine contrast is also difficult to justify in pediatric SIJ MRI. Small pediatric studies found that fluid-sensitive BME was present in all contrast-positive cases and that gadolinium did not improve case detection⁴¹. Contrast may still be useful when an infection, tumor, or an unusual synovial process is a realistic alternative. A normal radiograph or a negative MRI cannot independently exclude juvenile axial disease; clinical evolution and the management consequences of additional imaging remain central.

Sacroiliac joint lesions and mimics

BME or osteitis is hyperintense on STIR or T2-weighted fat-suppressed images and usually lies in subchondral or periarticular marrow. Specificity increases with intensity, depth, multiplicity, persistence across slices, and a coherent bilateral or multifocal distribution⁸^,⁹^,¹⁰^,²⁵^,⁴²^,⁴³. Small, shallow anterior or inferior foci are common after mechanical loading and should not be equated with disease. Capsulitis, synovitis, and enthesitis can support activity but are less specific and should be interpreted with BME and structural findings.

Structural lesions modify probability. Multiple erosions, backfill, or ankylosis are more persuasive than isolated low-grade BME; fat metaplasia supports previous inflammation only when its distribution and associated lesions are characteristic¹⁰^,⁴²^,⁴³^,⁴⁴^,⁴⁵^,⁴⁶^,⁴⁷. Sclerosis is depicted less directly by conventional MRI than by radiography or CT. The report should describe each domain separately rather than collapsing them into a binary positive or negative label. Table 1 summarizes lesion appearance, diagnostic contribution, and common pitfalls.

Systematic assessment of mimics is an essential component of image interpretation. Pregnancy and the postpartum period, running and other athletic loading, osteitis condensans ilii, degeneration, stress reaction, fracture, infection, and neoplasm can all produce SIJ abnormalities¹⁹^,²⁰^,²¹^,²²^,²³^,²⁴^,²⁵^,²⁶^,²⁷^,²⁸. Distribution, soft-tissue change, fracture lines, abscess, destructive marrow replacement, and the relationship between active and structural findings help distinguish these entities. When red flags or features of an alternative diagnosis are present, they should be reported explicitly rather than categorized generically as sacroiliitis.

Spinal MRI: added value and specificity limits

Spinal lesions associated with axSpA include anterior and posterior vertebral corner inflammatory lesions, non-corner inflammatory lesions, facet and costovertebral inflammation, and discovertebral abnormalities¹¹^,³¹^,³²^,³³. Structural findings include corner fat lesions, erosion, and new bone formation. Confidence increases when several typical corner lesions occur in a young patient with corroborating SIJ disease; an isolated endplate focus provides substantially weaker diagnostic support.

Degenerative Modic change, Schmorl nodes, trauma, infection, diffuse idiopathic skeletal hyperostosis, and osteoporotic fracture can mimic inflammatory disease. Lesion shape, endplate integrity, disc signal, soft-tissue extension, age, and mechanical context should be considered. MRI is essential when acute pain in an ankylosed spine raises concern for occult fracture or neurologic compromise, but CT is complementary for cortical fracture definition.

Spine MRI provides limited additional classification value when an adequate SIJ MRI is negative in an otherwise low-probability patient, and nonspecific spinal lesions may reduce diagnostic precision³². Its use should therefore be driven by symptoms, complications, or a prespecified quantitative study question. If a whole-spine study is obtained, the report should distinguish vertebral body, posterior element, and complication findings, and avoid implying that all corner edema is inflammatory.

Diagnostic performance, reader reliability, and probability-based reporting

MRI performance depends on the referral spectrum, reference standard, protocol, lesion definition, and reader expertise. In a cohort of 109 patients with suspected axSpA, combined SIJ radiography and MRI, interpreted by three radiologists, yielded 74.1% sensitivity, 90.2% specificity, a positive predictive value of 91.5%, a negative predictive value of 71.2%, and a Fleiss kappa of 0.77 without structured clinical information. With structured clinical information, sensitivity was 70.7%, specificity 97.6%, positive predictive value 97.6%, negative predictive value 70.2%, and Fleiss kappa 0.76⁴⁸. These estimates apply to combined radiography and MRI in an enriched referral cohort, not to MRI used as a screening test.

Reliability also varies by lesion. In the ASAS classification cohort, agreement between local and majority central readings was kappa 0.62 (95% confidence interval 0.53–0.72) for active lesions typical of axSpA but kappa 0.38 (95% confidence interval 0.25–0.50) for structural lesions. Among seven central readers, pairwise reliability for active lesions was higher (median kappa 0.74; range 0.63–0.83); the local false-positive rate was 33.3% against the majority central reading, and discrepant reads changed ASAS classification in 6.4%⁴⁹. Multireader studies likewise report better reproducibility for inflammation and fat scores than for erosion or backfill, and trial-level scoring reliability does not guarantee agreement in routine practice²⁵^,³⁴^,³⁵^,³⁶^,³⁷^,⁵⁰^,⁵¹^,⁵²^,⁵³. Research studies should use trained, blinded independent readers, prespecified rules, and adjudication. Clinical services benefit from protocol templates, periodic calibration, access to prior examinations, and multidisciplinary review of high-impact equivocal cases.

We propose four pragmatic reporting categories; they are not validated diagnostic criteria. “Typical for axSpA” denotes substantial or multifocal subchondral inflammation and/or characteristic structural support. “Suspicious/equivocal” denotes limited but potentially characteristic abnormalities whose confidence is reduced by extent, technique, or context. “Nonspecific” is appropriate for small, mechanically located edema without structural support. “Alternative diagnosis favored” is used when morphology and context support degeneration, postpartum or mechanical stress, fracture, infection, or neoplasm. Each report should state observations, confidence, technical limitations, and the principal alternative. Figure 1 summarizes this acquisition-to-action workflow.

Quantifying MRI abnormalities: SPARCC, Berlin, and CANDEN

The Spondyloarthritis Research Consortium of Canada (SPARCC) SIJ inflammation score evaluates six consecutive semicoronal slices and ranges from 0 to 72; the SPARCC spine score evaluates the six most abnormal discovertebral units and ranges from 0 to 108³⁴^,³⁵^,³⁶. Both are responsive and widely used in trials. Their value depends on trained readers, standardized acquisition, and comparable time points; a numerical change should not be interpreted without image review and the clinical question.

The Berlin modification of the ASspiMRI-a grades BME in 23 vertebral units from C2–C3 to L5–S1 on a 0–3 scale per unit, yielding a total score of 0–69³⁷^,⁵¹. The Canada–Denmark (CANDEN) system is more comprehensive, covering vertebral bodies and posterior elements and scoring inflammation, fat, erosion, and new bone formation in separate domains⁵²^,⁵³. Its anatomical detail supports mechanistic and treatment studies, but increases reading time, and reproducibility is generally lower for some structural than inflammatory domains.

These instruments are research-oriented and not interchangeable. Routine reports should generally include location, extent, active versus structural domain, confidence, and comparison with prior studies rather than a formal score. A score is most useful when a trial or longitudinal assessment uses a prespecified instrument, acquisition is comparable, reader expertise is available, and the result could change interpretation or management. Table 2 compares their scope, structure, validation, and practical role.

Baseline MRI before targeted therapy

Baseline MRI can document objective inflammation when symptoms, examination, and biomarkers are discordant; define whether disease burden lies mainly in the SIJs or spine; and provide a reproducible baseline for subsequent comparisons when future imaging is likely to be clinically useful⁵⁴^,⁵⁵^,⁵⁶^,⁵⁷. It should not be ordered solely because a targeted therapy is contemplated if the diagnosis and objective inflammatory status are already sufficiently established by other clinical data.

The baseline report should state protocol adequacy, active and structural domains, anatomical distribution, important mimics or complications, and—when formally scored—the instrument and reader method. Clinical documentation should include the Ankylosing Spondylitis Disease Activity Score (ASDAS), Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), CRP, treatment history, and exposure. This facilitates a valid comparison with a documented and technically comparable baseline examination.

Objective MRI inflammation can predict greater average imaging improvement in treatment trials, but it does not identify with certainty which individual will improve symptomatically⁵⁸^,⁵⁹^,⁶⁰^,⁶¹^,⁶²^,⁶³^,⁶⁴^,⁶⁵^,⁶⁶^,⁶⁷^,⁶⁸. Conversely, a patient can have substantial symptoms with little measurable inflammation because pain may reflect structural damage, mechanical disease, central sensitization, or another process. Baseline MRI should therefore be interpreted as one biomarker within a broader clinical model and should not serve as the sole basis for treatment eligibility or exclusion.

Treatment response and early structural remodeling

Tumor necrosis factor inhibitors, interleukin-17 inhibitors, and Janus kinase inhibitors reduce MRI inflammation on average, often before structural outcomes can be assessed⁵⁷^,⁵⁸^,⁵⁹^,⁶⁰^,⁶¹^,⁶²^,⁶³^,⁶⁴^,⁶⁵^,⁶⁶^,⁶⁷^,⁶⁸. Repeat MRI may be clinically justified when persistent symptoms, uncertainty regarding treatment adherence or exposure, discordance between CRP and clinical findings, a suspected complication, or a treatment decision with important clinical consequences remain unresolved. Routine serial imaging in every stable patient is not recommended, and no validated MRI treat-to-target endpoint has been shown to improve long-term outcomes⁵⁴^,⁵⁵^,⁶⁹.

Recent TRACE data refine the traditional sequential model. In 90 biologic-naive patients, mean combined SIJ and spine inflammation fell by 14.7 points by week 4; reduced SIJ erosion and increased fat lesions and backfill were also detectable at that time⁷⁰. At the 1-year follow-up, early reductions in inflammation and structural remodeling changes persisted, whereas no early changes in ankylosis or spinal new bone formation were detected⁷¹. The open-label design, attrition, and later mixed treatment pathways preclude attributing all observed changes to a single drug or interpreting them as evidence of beneficial long-term structural modification.

COAST-V provides complementary evidence from a randomized treatment period: ixekizumab was associated with reduced SIJ erosion extent and increased backfill by week 16, with further change through week 52 in continuously treated participants⁷². Because this was a post hoc structural analysis and control groups switched treatment after week 16, it does not establish the prevention of ankylosis. Figure 2, therefore, presents a time-aware framework in which inflammation suppression and early structural remodeling overlap over weeks to months, followed by uncertain long-term structural evolution over years rather than a fixed sequence.

MRI improvement, structural remodeling, and long-term outcomes

MRI improvement is a responsive marker of biological activity but is not a validated surrogate for pain relief, function, quality of life, radiographic progression, or prevention of ankylosis. Residual BME may persist after clinical improvement, and symptoms may remain active despite quiet imaging. Early erosion reduction, fat metaplasia, and backfill may reflect early structural remodeling; however, fat metaplasia and backfill are also implicated in complex pathways linking prior inflammation to new bone formation⁴⁶^,⁴⁷^,⁷³^,⁷⁴^,⁷⁵^,⁷⁶^,⁷⁷^,⁷⁸^,⁷⁹^,⁸⁰.

Long-term structural change occurs on a different timescale and requires methods suited to the lesion. MRI depicts marrow and soft-tissue biology, radiography remains a conventional reference for syndesmophyte progression, and low-dose CT is more sensitive to cortical new bone but involves ionizing radiation. Short therapeutic trials are generally underpowered to demonstrate changes in structural outcomes that evolve over years. Accordingly, a fall in SPARCC or a rise in backfill should not be interpreted as evidence that disability or ankylosis has been prevented.

When clinical and imaging responses diverge, treatment exposure and timing, CRP and ASDAS, mechanical and non-inflammatory pain, imaging technique and mimics, and possible complications should be reassessed. Escalation or switching should be based on the combined evidence and patient goals, not an isolated residual MRI focus. The same caution applies to apparent imaging remission, which does not prove durable drug-free disease control.

Advanced MRI and artificial intelligence: potential and present limitations

Diffusion-weighted and dynamic contrast-enhanced MRI can generate quantitative markers, while three-dimensional gradient-echo, ultrashort- or zero-echo-time imaging, and synthetic CT-like reconstructions may improve structural visualization³⁰^,⁸¹^,⁸²^,⁸³^,⁸⁴. Routine use is limited by acquisition time, vendor dependence, post-processing, incomplete standardization, and uncertain incremental value for clinical decision-making. Improved depiction of erosion is not equivalent to improved diagnostic accuracy or patient outcomes.

AI systems can detect or segment SIJ inflammation and structural lesions, but most evidence comes from enriched retrospective expert-center datasets⁸⁵^,⁸⁶^,⁸⁷^,⁸⁸^,⁸⁹^,⁹⁰. Important limitations include reference-standard and label circularity, underrepresentation of mimics, scanner and protocol domain shift, calibration drift in low-prevalence practice, and performance reporting that emphasizes reader-level accuracy without evaluating effects on clinical decision-making.

Before clinical deployment, advanced methods require multicenter, multi-vendor external validation; reproducibility across field strength and sequence variants; transparent failure analysis; calibration in realistic referral populations; prospective workflow evaluation; and evidence of decision or outcome benefit. Regulatory requirements, cost, interoperability, health equity, and continued radiologist oversight also require consideration. These tools should currently complement rather than replace standardized conventional acquisition and expert interpretation.

Implementation, reproducibility, and evolving definitions

A high-quality service links the referral question, standardized protocol, trained reader, structured report, and multidisciplinary action. Referral forms should record age at onset, symptoms, HLA-B27, CRP, extra-musculoskeletal manifestations, pregnancy or postpartum status, athletic loading, prior imaging, treatment exposure, and the intended clinical decision. Reports should state technical adequacy; active and structural lesions; distribution and extent; mimics, red flags, or alternative diagnosis; reporting category; and comparison with prior examinations. Table 3 maps common clinical questions to minimum coverage, key domains, reporting outputs, and cautions.

Implementation should be standardized alongside terminology. Sites should audit adherence to the four-sequence diagnostic protocol, document any sequence substitutions, calibrate readers using representative mimics, and review discrepant, high-impact cases. Multicenter studies need acquisition manuals, blinded central reading, prespecified adjudication, and scanner-level quality control. Algorithms and quantitative measures should be version-controlled, externally validated, and monitored for performance drift across settings.

MRI definitions and treatment paradigms continue to evolve. Current ASAS criteria remain classification tools, and BME-dominant definitions can generate false positives in mechanically exposed populations. Structural lesions and quantitative thresholds may improve specificity, but proposed updates should not be presented as endorsed standards before formal validation and approval. As treatment access broadens across r-axSpA and nr-axSpA, MRI may confirm objective inflammation, but it should not serve as the sole basis for diagnosis, treatment eligibility, or treatment switching.

Limitations

This narrative review was designed as a practical synthesis rather than a systematic review or meta-analysis. Study selection and emphasis may therefore be influenced by author judgment, publication bias, and the targeted PubMed/MEDLINE search. No formal risk-of-bias assessment was performed, and reported diagnostic or treatment estimates should not be pooled informally across dissimilar populations or reference standards.

Generalizability is limited by heterogeneity in referral prevalence, disease duration, field strength, vendor, protocol, reader training, lesion definition, scoring method, treatment exposure, and follow-up interval. Much of the evidence on scoring systems and AI comes from trials or expert centers. Pediatric validation, routine serial MRI, external AI validation, and the long-term clinical meaning of early structural remodeling remain incomplete. The author-proposed reporting categories and workflow algorithms have not undergone prospective validation.

The evidence base is also changing rapidly. The 2024 acquisition consensus and 2025–2026 treatment-monitoring studies improve standardization and time-aware interpretation but do not settle diagnostic thresholds, surrogate validity, or cost-effectiveness. Future work should use representative low-prevalence cohorts, harmonized acquisition, independent clinical reference standards, transparent central reading, and outcomes that assess whether imaging improves clinical decision-making and yields patient benefits.

Access restricted. Please log in or start a trial to view this content.

Conclusions

MRI is central to modern axSpA evaluation when it is used as a calibrated component of clinical reasoning. An evidence-informed pathway combines the ASAS–SPARTAN four-sequence SIJ protocol, separate active and structural assessment, explicit consideration of mimics, and probability-based reporting using the “Typical for axSpA,” “Suspicious/equivocal,” “Nonspecific,” and “Alternative diagnosis favored” categories. Adult and pediatric studies require different interpretive expectations, and classification criteria must not...

Access restricted. Please log in or start a trial to view this content.

Disclosures

The authors declare no competing interests.

Acknowledgements

None. This work received no specific funding.

Access restricted. Please log in or start a trial to view this content.

References

Bittar M, Deodhar A. Axial spondyloarthritis: a review. JAMA. 2025;333(5):408-420.
Sieper J, Poddubnyy D. Axial spondyloarthritis. Lancet. 2017;390(10089):73-84.
Taurog JD, Chhabra A, Colbert RA. Ankylosing spondylitis and axial spondyloarthritis. N Engl J Med. 2016;374(26):2563-2574.
Robinson PC, van der Linden S, Khan MA, Taylor WJ. Axial spondyloarthritis: concept, construct, classification and implications for therapy. Nat Rev Rheumatol. 2021;17(2):109-118.
van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic criteria for ankylosing spondylitis: a proposal for modification of the New York criteria. Arthritis Rheum. 1984;27(4):361-368.
Rudwaleit M et al. The development of Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis. Part I: Classification of paper patients by expert opinion including uncertainty appraisal. Ann Rheum Dis. 2009;68(6):770-776.
Rudwaleit M et al. The development of Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis. Part II: Validation and final selection. Ann Rheum Dis. 2009;68(6):777-783.
Rudwaleit M et al. Defining active sacroiliitis on magnetic resonance imaging (MRI) for classification of axial spondyloarthritis: a consensual approach by the ASAS/OMERACT MRI group. Ann Rheum Dis. 2009;68(10):1520-1527.
Lambert RGW et al. Defining active sacroiliitis on MRI for classification of axial spondyloarthritis: update by the ASAS MRI Working Group. Ann Rheum Dis. 2016;75(11):1958-1963.
Maksymowych WP et al. MRI lesions in the sacroiliac joints of patients with spondyloarthritis: an update of definitions and validation by the ASAS MRI Working Group. Ann Rheum Dis. 2019;78(11):1550-1558.
Baraliakos X et al. MRI lesions of the spine in patients with axial spondyloarthritis: an update of lesion definitions and validation by the ASAS MRI Working Group. Ann Rheum Dis. 2022;81(9):1243-1251.
Mandl P et al. EULAR recommendations for the use of imaging in the diagnosis and management of spondyloarthritis in clinical practice. Ann Rheum Dis. 2015;74(7):1327-1339.
Maksymowych WP. The role of imaging in the diagnosis and management of axial spondyloarthritis. Nat Rev Rheumatol. 2019;15(11):657-672.
Diekhoff T, Lambert R, Hermann K-G. MRI in axial spondyloarthritis: understanding an "ASAS-positive MRI" and the ASAS classification criteria. Skeletal Radiol. 2022;51(9):1721-1730.
Poddubnyy D, Diekhoff T, Baraliakos X, Hermann KGA, Sieper J. Diagnostic evaluation of the sacroiliac joints for axial spondyloarthritis: should MRI replace radiography? Ann Rheum Dis. 2022;81(11):1486-1490.
Ritchlin C, Adamopoulos IE. Axial spondyloarthritis: new advances in diagnosis and management. BMJ. 2021;372:m4447.
Wang R, Ward MM. Epidemiology of axial spondyloarthritis: an update. Curr Opin Rheumatol. 2018;30(2):137-143.
Zhao SS et al. Diagnostic delay in axial spondyloarthritis: a systematic review and meta-analysis. Rheumatology (Oxford). 2021;60(4):1620-1628.
de Winter J et al. Magnetic resonance imaging of the sacroiliac joints indicating sacroiliitis according to the Assessment of SpondyloArthritis international Society definition in healthy individuals, runners, and women with postpartum back pain. Arthritis Rheumatol. 2018;70(7):1042-1048.
Agten CA et al. Postpartum bone marrow edema at the sacroiliac joints may mimic sacroiliitis of axial spondyloarthritis on MRI. AJR Am J Roentgenol. 2018;211(6):1306-1312.
Weber U et al. MRI of the sacroiliac joints in athletes: recognition of non-specific bone marrow oedema by semi-axial added to standard semi-coronal scans. Rheumatology (Oxford). 2020;59(6):1381-1390.
Eshed I et al. Peripartum changes of the sacroiliac joints on MRI: increasing mechanical load correlating with signs of edema and inflammation kindling spondyloarthropathy in the genetically prone. Clin Rheumatol. 2015;34(8):1419-1426.
Kiil RM et al. Evolution of magnetic resonance imaging lesions at the sacroiliac joints during and after pregnancy by serial magnetic resonance imaging from gestational week twenty to twelve months postpartum. Arthritis Rheumatol. 2023;75(7):1166-1175.
Varkas G et al. Effect of mechanical stress on magnetic resonance imaging of the sacroiliac joints: assessment of military recruits by magnetic resonance imaging study. Rheumatology (Oxford). 2018;57(3):508-513.
Seven S et al. Magnetic resonance imaging of lesions in the sacroiliac joints for differentiation of patients with axial spondyloarthritis from control subjects with or without pelvic or buttock pain: a prospective, cross-sectional study of 204 participants. Arthritis Rheumatol. 2019;71(12):2034-2046.
Seven S et al. Morphological characteristics of sacroiliac joint MRI lesions in axial spondyloarthritis and control subjects. Rheumatology (Oxford). 2022;61(3):1005-1017.
Jurik AG et al. Diagnostics of sacroiliac joint differentials to axial spondyloarthritis changes by magnetic resonance imaging. J Clin Med. 2023;12(3):1039.
Hoballah A et al. MRI of sacroiliac joints for the diagnosis of axial SpA: prevalence of inflammatory and structural lesions in nulliparous, early postpartum and late postpartum women. Ann Rheum Dis. 2020;79(8):1063-1069.
Lambert RGW et al. Development of international consensus on a standardised image acquisition protocol for diagnostic evaluation of the sacroiliac joints by MRI: an ASAS–SPARTAN collaboration. Ann Rheum Dis. 2024;83(12):1628-1635.
Martin-Noguerol T, Casado-Verdugo OL, Beltran LS, Aguilar G, Luna A. Role of advanced MRI techniques for sacroiliitis assessment and quantification. Eur J Radiol. 2023;163:110793.
Althoff CE et al. Active inflammation and structural change in early active axial spondyloarthritis as detected by whole-body MRI. Ann Rheum Dis. 2013;72(6):967-973.
Ez-Zaitouni Z et al. The yield of a positive MRI of the spine as imaging criterion in the ASAS classification criteria for axial spondyloarthritis: results from the SPACE and DESIR cohorts. Ann Rheum Dis. 2017;76(10):1731-1736.
Baraliakos X, Davis J, Tsuji W, Braun J. Magnetic resonance imaging examinations of the spine in patients with ankylosing spondylitis before and after therapy with the tumor necrosis factor alpha receptor fusion protein etanercept. Arthritis Rheum. 2005;52(4):1216-1223.
Maksymowych WP et al. Defining the minimally important change for the Spondyloarthritis Research Consortium of Canada spine and sacroiliac joint magnetic resonance imaging indices for ankylosing spondylitis. J Rheumatol. 2012;39(8):1666-1674.
Maksymowych WP et al. Spondyloarthritis Research Consortium of Canada magnetic resonance imaging index for assessment of sacroiliac joint inflammation in ankylosing spondylitis. Arthritis Rheum. 2005;53(5):703-709.
Maksymowych WP et al. Spondyloarthritis Research Consortium of Canada magnetic resonance imaging index for assessment of spinal inflammation in ankylosing spondylitis. Arthritis Rheum. 2005;53(4):502-509.
Hededal P et al. Development and validation of MRI sacroiliac joint scoring methods for the semiaxial scan plane corresponding to the Berlin and SPARCC MRI scoring methods, and of a new global MRI sacroiliac joint method. J Rheumatol. 2018;45(1):70-77.
Herregods N et al. ASAS definition for sacroiliitis on MRI in SpA: applicable to children? Pediatr Rheumatol Online J. 2017;15(1):24.
Hemke R et al. Imaging assessment of children presenting with suspected or known juvenile idiopathic arthritis: ESSR–ESPR points to consider. Eur Radiol. 2020;30(10):5237-5249.
Weiss PF et al. Data-driven magnetic resonance imaging definitions for active and structural sacroiliac joint lesions in juvenile spondyloarthritis typical of axial disease: a cross-sectional international study. Arthritis Care Res (Hoboken). 2023;75(6):1220-1227.
Weiss PF, Xiao R, Biko DM, Johnson AM, Chauvin NA. Detection of inflammatory sacroiliitis in children with magnetic resonance imaging: is gadolinium contrast enhancement necessary? Arthritis Rheumatol. 2015;67(8):2250-2256.
Maksymowych WP et al. Data-driven definitions for active and structural MRI lesions in the sacroiliac joint in spondyloarthritis and their predictive utility. Rheumatology (Oxford). 2021;60(10):4778-4789.
Baraliakos X et al. Which magnetic resonance imaging lesions in the sacroiliac joints are most relevant for diagnosing axial spondyloarthritis? A prospective study comparing rheumatologists' evaluations with radiologists' findings. Arthritis Rheumatol. 2021;73(5):800-805.
Maksymowych WP et al. MRI evidence of structural changes in the sacroiliac joints of patients with non-radiographic axial spondyloarthritis even in the absence of MRI inflammation. Arthritis Res Ther. 2017;19(1):126.
Braun J, Kiltz U, Baraliakos X. Significance of structural changes in the sacroiliac joints of patients with axial spondyloarthritis detected by MRI related to patients' symptoms and functioning. Ann Rheum Dis. 2022;81(1):11-14.
Sepriano A et al. Inflammation of the sacroiliac joints and spine and structural changes on magnetic resonance imaging in axial spondyloarthritis: five-year data from the DESIR cohort. Arthritis Care Res (Hoboken). 2022;74(2):243-250.
Maksymowych WP et al. Structural changes in the sacroiliac joint on MRI and relationship to ASDAS inactive disease in axial spondyloarthritis: a 2-year study comparing treatment with etanercept in EMBARK to a contemporary control cohort in DESIR. Arthritis Res Ther. 2021;23(1):43.
Pohlner T et al. Diagnostic accuracy in axial spondyloarthritis: a systematic evaluation of the role of clinical information in the interpretation of sacroiliac joint imaging. RMD Open. 2024;10(2):e004044.
Maksymowych WP et al. Central reader evaluation of MRI scans of the sacroiliac joints from the ASAS classification cohort: discrepancies with local readers and impact on the performance of the ASAS criteria. Ann Rheum Dis. 2020;79(7):935-942.
Heuft-Dorenbosch L, Weijers R, Landewé R, van der Linden S, van der Heijde D. Magnetic resonance imaging changes of sacroiliac joints in patients with recent-onset inflammatory back pain: inter-reader reliability and prevalence of abnormalities. Arthritis Res Ther. 2006;8(1):R11.
Lukas C et al. Scoring inflammatory activity of the spine by magnetic resonance imaging in ankylosing spondylitis: a multireader experiment. J Rheumatol. 2007;34(4):862-870.
Krabbe S et al. Canada-Denmark MRI scoring system of the spine in patients with axial spondyloarthritis: updated definitions, scoring rules and inter-reader reliability in a multiple reader setting. RMD Open. 2019;5(2):e001057.
Krabbe S et al. Inflammatory and structural changes in vertebral bodies and posterior elements of the spine in axial spondyloarthritis: construct validity, responsiveness and discriminatory ability of the anatomy-based CANDEN scoring system in a randomised placebo-controlled trial. RMD Open. 2018;4(1):e000624.
Ramiro S et al. ASAS-EULAR recommendations for the management of axial spondyloarthritis: 2022 update. Ann Rheum Dis. 2023;82(1):19-34.
Ward MM et al. 2019 update of the American College of Rheumatology/Spondylitis Association of America/Spondyloarthritis Research and Treatment Network recommendations for the treatment of ankylosing spondylitis and nonradiographic axial spondyloarthritis. Arthritis Rheumatol. 2019;71(10):1599-1613.
Mandl P, et al. EULAR recommendations for the use of imaging in the diagnosis and management of spondyloarthritis in clinical practice. Ann Rheum Dis. 2015;74(7):1327–1339.
Webers C et al. Efficacy and safety of biological DMARDs: a systematic literature review informing the 2022 update of the ASAS-EULAR recommendations for the management of axial spondyloarthritis. Ann Rheum Dis. 2023;82(1):130-141.
Machado P et al. MRI inflammation and its relation with measures of clinical disease activity and different treatment responses in patients with ankylosing spondylitis treated with a tumour necrosis factor inhibitor. Ann Rheum Dis. 2012;71(12):2002-2005.
Deodhar A et al. Upadacitinib for the treatment of active non-radiographic axial spondyloarthritis (SELECT-AXIS 2): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2022;400(10349):369-379.
Lambert RGW et al. Adalimumab significantly reduces both spinal and sacroiliac joint inflammation in patients with ankylosing spondylitis: a multicenter, randomized, double-blind, placebo-controlled study. Arthritis Rheum. 2007;56(12):4005-4014.
Sieper J et al. Efficacy and safety of adalimumab in patients with non-radiographic axial spondyloarthritis: results of a randomised placebo-controlled trial (ABILITY-1). Ann Rheum Dis. 2013;72(6):815-822.
Maksymowych WP et al. Clinical and MRI responses to etanercept in early non-radiographic axial spondyloarthritis: 48-week results from the EMBARK study. Ann Rheum Dis. 2016;75(7):1328-1335.
Braun J et al. Effect of certolizumab pegol over 96 weeks of treatment on inflammation of the spine and sacroiliac joints, as measured by MRI, and the association between clinical and MRI outcomes in patients with axial spondyloarthritis. RMD Open. 2017;3(1):e000430.
van der Heijde D et al. Limited radiographic progression and sustained reductions in MRI inflammation in patients with axial spondyloarthritis: 4-year imaging outcomes from the RAPID-axSpA phase III randomised trial. Ann Rheum Dis. 2018;77(5):699-705.
van der Heijde D et al. Ixekizumab, an interleukin-17A antagonist, in the treatment of ankylosing spondylitis or radiographic axial spondyloarthritis in patients previously untreated with biological disease-modifying anti-rheumatic drugs (COAST-V): 16-week results of a phase 3 randomised, double-blind, active-controlled and placebo-controlled trial. Lancet. 2018;392(10163):2441-2451.
Deodhar A et al. Ixekizumab for patients with non-radiographic axial spondyloarthritis (COAST-X): a randomised, placebo-controlled trial. Lancet. 2020;395(10217):53-64.
Baeten D et al. Secukinumab, an interleukin-17A inhibitor, in ankylosing spondylitis. N Engl J Med. 2015;373(26):2534-2548.
van der Heijde D et al. Tofacitinib in patients with ankylosing spondylitis: a phase II, 16-week, randomised, placebo-controlled, dose-ranging study. Ann Rheum Dis. 2017;76(8):1340-1347.
Smolen JS et al. Treating axial spondyloarthritis and peripheral spondyloarthritis, especially psoriatic arthritis, to target: 2017 update of recommendations by an international task force. Ann Rheum Dis. 2018;77(1):3-17.
Vladimirova N et al. Effect of secukinumab dose escalation on MRI-detected inflammation and structural damage in axial spondyloarthritis: a clinical, investigator-initiated, treat-to-target study (TRACE). RMD Open. 2026;12(2):e006774.
Vladimirova N et al. Clinical and MRI remission in axial spondyloarthritis: one-year outcomes and predictors from the TRACE treat-to-target study with secukinumab. Arthritis Res Ther. 2026;28(1):130.
Maksymowych WP et al. The effect of ixekizumab treatment on MRI sacroiliac joint structural lesions in patients with radiographic axial spondyloarthritis: post-hoc analysis of a 52-week, randomised, placebo-controlled trial with an active reference arm. Lancet Rheumatol. 2025;7(5):e314-e322.
Baraliakos X et al. Which spinal lesions are associated with new bone formation in patients with ankylosing spondylitis treated with anti-TNF agents? A long-term observational study using MRI and conventional radiography. Ann Rheum Dis. 2014;73(10):1819-1825.
van der Heijde D et al. MRI inflammation at the vertebral unit only marginally predicts new syndesmophyte formation: a multilevel analysis in patients with ankylosing spondylitis. Ann Rheum Dis. 2012;71(3):369-373.
Maksymowych WP et al. Inflammatory lesions of the spine on magnetic resonance imaging predict the development of new syndesmophytes in ankylosing spondylitis: evidence of a relationship between inflammation and new bone formation. Arthritis Rheum. 2009;60(1):93-102.
Chiowchanwisawakit P, Lambert RGW, Conner-Spady B, Maksymowych WP. Focal fat lesions at vertebral corners on magnetic resonance imaging predict the development of new syndesmophytes in ankylosing spondylitis. Arthritis Rheum. 2011;63(8):2215-2225.
Pedersen SJ, Chiowchanwisawakit P, Lambert RGW, Ostergaard M, Maksymowych WP. Resolution of inflammation following treatment of ankylosing spondylitis is associated with new bone formation. J Rheumatol. 2011;38(7):1349-1354.
Stal R et al. Role of vertebral corner inflammation and fat deposition on MRI on syndesmophyte development detected on whole spine low-dose CT scan in radiographic axial spondyloarthritis. RMD Open. 2022;8(2):e002250.
Ramiro S et al. Higher disease activity leads to more structural damage in the spine in ankylosing spondylitis: 12-year longitudinal data from the OASIS cohort. Ann Rheum Dis. 2014;73(8):1455-1461.
van Tubergen A et al. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis. 2012;71(4):518-523.
Gensler LS et al. Unmet needs in spondyloarthritis: imaging in axial spondyloarthritis. J Rheumatol. 2024;51(12):1241-1246.
Chen X et al. Multi-b-values-fitting readout-segmentation of long variable echo-trains diffusion-weighted imaging (RESOLVE DWI) in evaluation of disease activity and curative effect of axial spondyloarthritis (axSpA). Front Immunol. 2023;14:1136925.
Gaspersic N, Sersa I, Jevtic V, Tomsic M, Praprotnik S. Monitoring ankylosing spondylitis therapy by dynamic contrast-enhanced and diffusion-weighted magnetic resonance imaging. Skeletal Radiol. 2008;37(2):123-131.
Kucybala I, Urbanik A, Wojciechowski W. Radiologic approach to axial spondyloarthritis: where are we now and where are we heading? Rheumatol Int. 2018;38(10):1753-1762.
Adams LC, Bressem KK, Poddubnyy D. Artificial intelligence and machine learning in axial spondyloarthritis. Curr Opin Rheumatol. 2024;36(4):267-273.
Bressem KK et al. Deep learning detects changes indicative of axial spondyloarthritis at MRI of sacroiliac joints. Radiology. 2022;305(3):655-665.
Lee S et al. Artificial intelligence for the detection of sacroiliitis on magnetic resonance imaging in patients with axial spondyloarthritis. Front Immunol. 2023;14:1278247.
Faleiros MC et al. Machine learning techniques for computer-aided classification of active inflammatory sacroiliitis in magnetic resonance imaging. Adv Rheumatol. 2020;60(1):25.
Xie Z et al. Enhanced diagnosis of axial spondyloarthritis using machine learning with sacroiliac joint MRI: a multicenter study. Insights Imaging. 2025;16(1):91.
Zhang K et al. Use of MRI-based deep learning radiomics to diagnose sacroiliitis related to axial spondyloarthritis. Eur J Radiol. 2024;172:111347.

Access restricted. Please log in or start a trial to view this content.

MRI Evaluation of Axial Spondyloarthritis: From Early Diagnosis to Therapeutic Response Assessment

In This Article

Summary

Abstract

Introduction

Review and Perspective

Conclusions

Disclosures

Acknowledgements

References

Reprints and Permissions

Tags