The core problem is deceptively simple. Axial spondyloarthritis attacks the sacroiliac joints and the spine, driving chronic inflammation that eventually produces new, abnormal bone formation. But the spines of these patients, many of them still relatively young, also accumulate degenerative lesions of the kind seen in osteoarthritis: worn discs, bony outgrowths called osteophytes, and enlarged facet joints. On conventional X-rays and even magnetic resonance imaging, the two lesion types can look frustratingly similar, with overlapping features such as bone proliferation, bone marrow edema, and fatty infiltration. Chronic inflammation and mechanical stress in spondyloarthritis can even accelerate degenerative change, so the two processes frequently coexist in the same vertebra. Misreading a degenerative lesion as active inflammatory disease, or vice versa, can lead to the wrong treatment decision.

The Amsterdam University Medical Center-led team turned to sodium fluoride labeled with fluorine-18, a radioactive tracer with a strong affinity for hydroxyapatite, the mineral that makes up bone. Wherever osteoblasts are actively laying down new bone, the tracer accumulates, lighting up regions of heightened bone remodeling on PET/CT scans. That makes it a sensitive molecular readout of osteoblastic activity, capable of detecting changes before structural damage becomes visible on plain radiographs, which typically take at least two years to show meaningful change. The catch is that degenerative lesions also remodel bone, so they too can accumulate the tracer. Until now, no study had systematically defined how to tell the two lesion types apart on fluoride PET, or tested whether their longitudinal behavior during treatment could serve as a discriminating signature.

The researchers analyzed data from twenty patients with active radiographic axial spondyloarthritis enrolled in a substudy of the AS0013 trial, a phase 2a study of the biologic drugs bimekizumab and certolizumab pegol conducted across eight countries. All patients fulfilled the modified New York criteria, had high disease activity scores, elevated C-reactive protein, and had previously failed at least two nonsteroidal anti-inflammatory drugs. Each participant underwent fluoride PET/CT scans at three timepoints: before starting biologic therapy, and again at twelve and fifty-two weeks after treatment began. The imaging protocol was unusually rigorous, including dynamic scanning over the heart and thoracic spine with venous blood sampling to validate quantitative measurements of tracer delivery and uptake.

Two blinded readers, a musculoskeletal radiologist and a nuclear medicine specialist, independently reviewed the scans as chronological triplets, using the low-dose CT component as an anatomical reference. They applied predefined classification principles: degenerative lesions typically sit at intervertebral discs and facet joints and show disc degeneration, osteophytes, and joint hypertrophy on CT, while spondyloarthritis lesions favor the sacroiliac joints and vertebral corners, where erosions, sclerosis, and bridging syndesmophytes develop. Only lesions classified identically by both readers were carried forward, a deliberately conservative choice that sacrificed statistical power in exchange for confidence. In total, the readers detected 276 PET-positive lesions across the twenty patients, with individual burdens ranging from two to forty-two lesions per person.

The results were striking. Of the 276 lesions, 98, roughly 36 percent, were classified as spondyloarthritis-related, while 46, about 17 percent, were classified as degenerative. Over the year of biologic treatment, the number of spondyloarthritis lesions fell from 98 at baseline to 67 at week twelve, a 32 percent drop, and then to 55 at week fifty-two, an additional 18 percent decline. The degenerative lesions, by contrast, barely moved, fluctuating only between 40 and 42 throughout the study. The anatomical distribution reinforced the pattern: 84 percent of the spondyloarthritis lesions clustered in the thoracic spine, particularly the costovertebral joints, while nearly half of the degenerative lesions sat in the lumbar spine, the classic home of osteoarthritic wear.

Quantitative analysis told the same story with even greater precision. The team measured standardized uptake values corrected for each patient’s integrated whole-blood tracer activity, a pharmacokinetic refinement that accounts for individual differences in tracer delivery. Between baseline and week twelve, 84 percent of the spondyloarthritis lesions showed a decrease in tracer uptake, with individual declines reaching nearly 70 percent, and the mean decrease across these lesions was 27 percent. By week fifty-two, 88 percent of spondyloarthritis lesions had declined. The degenerative lesions behaved entirely differently: most either increased, in some cases by more than 100 percent, or showed only mild decreases. A repeated measures mixed model confirmed the divergence statistically, with a nominally significant difference in uptake change between the lesion types at both week twelve and week fifty-two.

The logic behind this natural experiment is elegant. Biologic drugs such as bimekizumab, which blocks interleukin-17A, and certolizumab pegol, an anti-TNF agent, are designed to suppress inflammatory bone remodeling but have no proven efficacy on degenerative joint disease. So lesions that quiet down under treatment are almost certainly inflammatory, while lesions that ignore the drug are almost certainly degenerative. The longitudinal behavior of the tracer effectively becomes a functional fingerprint of each lesion’s underlying biology, something no single static scan could provide. Notably, the researchers found no instance in which one reader classified a lesion as spondyloarthritis-related while the other called it degenerative, suggesting the visual criteria are robust when applied by experienced readers.

The study does carry caveats. With only twenty patients, generalizability is limited, and there is no gold standard against which to verify the lesion classifications. The readers, although blinded to clinical data, were not blinded to changes in tracer uptake over time, since the triplet-review approach was chosen to keep closely spaced lesions from being confused across timepoints. The authors also note that fluoride PET does not directly reveal the underlying pathology of a lesion, only its bone-remodeling activity, and that cost and availability currently limit routine clinical use. Future validation studies, including cohorts of pure osteoarthritis patients without inflammatory disease, will be needed to confirm the approach.

Even so, the implications are considerable. As total-body PET scanners with lower radiation burdens enter clinical service, and as artificial intelligence algorithms mature for automated lesion detection and classification, fluoride PET/CT could become a practical tool for separating treatable inflammation from irreversible degeneration in spondyloarthritis patients. The authors also envision hybrid PET/MRI approaches that combine metabolic and soft-tissue information, and standardized imaging protocols that would enable multicenter research. For a disease in which treatment decisions hinge on knowing whether active inflammation is still smoldering in the spine, a tracer that can effectively interrogate each lesion individually, and watch it respond to therapy in real time, represents a genuinely new window into disease biology.

Subject of Research: Differentiating inflammatory from degenerative spinal lesions in axial spondyloarthritis using longitudinal sodium fluoride PET/CT imaging during biologic therapy

Article Title: Differentiation of axial spondyloarthritis-related lesions from degenerative lesions on Na[18F]Fluoride PET/CT in a longitudinal dataset of patients with radiographic axial spondyloarthritis

Article References: de Jongh, J., Groothuizen, S., Zwezerijnen, G. J., Hemke, R., Doorschodt, T., Maguire, R. P., Vaux, T., & van der Laken, C. J. (2026). Differentiation of axial spondyloarthritis-related lesions from degenerative lesions on Na[18F]Fluoride PET/CT in a longitudinal dataset of patients with radiographic axial spondyloarthritis. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08178-6

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08178-6

Keywords: axial spondyloarthritis, Na[18F]Fluoride PET/CT, biologic therapy, bimekizumab, certolizumab pegol, degenerative lesions, osteoblastic activity, sacroiliac joints, syndesmophytes, molecular imaging, rheumatology, SUV quantification

The research team, led by investigators at Tri-Service General Hospital and the National Defense Medical University in Taiwan, enrolled 43 adults with chronic low back pain who were being evaluated for suspected axial spondyloarthritis. Every participant underwent the full diagnostic workup: plain pelvic radiography, magnetic resonance imaging of the sacroiliac joints, and a whole-body positron emission tomography scan using fluorine-18 labeled sodium fluoride. Based on the combined clinical, laboratory, and imaging assessment, 29 patients were classified as having radiographic ankylosing spondylitis, 5 as having non-radiographic axial spondyloarthritis, and 9 as having degenerative joint disease, a non-inflammatory comparator group. The design was exploratory, but it was prospective and carefully standardized, with all three imaging studies completed within a two-month window.

The chemistry behind the tracer is what makes it so revealing. The fluoride ion exchanges with hydroxyl groups on the surface of newly deposited hydroxyapatite, the mineral scaffold of bone, converting it to fluoroapatite. In other words, wherever osteoblasts, the bone-forming cells, are actively laying down mineral, the tracer accumulates. Compared with older technetium-based bone agents, fluorine-18 sodium fluoride binds less to plasma proteins, reaches bone faster, and clears from the blood more quickly, producing sharper images. And because it is imaged on a PET scanner rather than a gamma camera, it delivers far better spatial resolution, standardized quantitative uptake values, and whole-body coverage in a single thirty-minute acquisition. Where MRI shows where marrow is inflamed, the fluoride signal shows how hard bone is actually responding.

The study’s primary question was whether the quantitative uptake in the sacroiliac joints, measured as the maximum standardized uptake value, or SUVmax, tracked with structural damage scored on MRI using the validated SPARCC system. It did. Sacroiliac SUVmax correlated moderately with the SPARCC structural score, with a Spearman coefficient of 0.606, and even more strongly with the SPARCC inflammation score, at 0.629, both highly significant. Because inflammation and structural damage scores were themselves correlated, the team ran partial rank correlations adjusting each association for the other, and both survived. This is the first patient-level demonstration that a single quantitative PET measure runs in parallel with both inflammatory and structural MRI domains in this disease.

The most striking finding concerned patients whose MRI showed no active inflammation at all. Bone marrow edema, the bright signal on fluid-sensitive MRI sequences that defines active sacroiliitis, was present in only 13 of the 29 ankylosing spondylitis patients. Yet visually increased fluoride uptake appeared in 23 of the 29, or 79.3 percent, including 10 of the 16 patients with no edema whatsoever. Not a single patient showed bone marrow edema without increased uptake. That asymmetry matters: it means the tracer was sensitive to disease activity that MRI classified as quiescent, consistent with the idea that bone remodeling continues even when inflammatory water signals fade. The median SUVmax in the ankylosing spondylitis group was 14.7, roughly two and a half times the values seen in either the non-radiographic or degenerative groups, which clustered around 5.5 to 5.7.

The diagnostic implications are tantalizing but come with heavy caveats. In exploratory receiver operating characteristic analyses, sacroiliac SUVmax separated ankylosing spondylitis from degenerative joint disease with an area under the curve of 0.841, and from non-radiographic disease with an area of 0.848. A threshold of 10.8 yielded 79.3 percent sensitivity and 88.9 percent specificity against degenerative disease. After bootstrap optimism correction, performance held up reasonably well, though the authors are emphatic that these thresholds were derived and tested in the same small cohort and must not be used as diagnostic cutoffs until validated externally. The non-radiographic subgroup, with only five patients, was too small for any firm conclusion, and its uptake values resembled the degenerative group’s more than the ankylosing spondylitis group’s.

Not everything the tracer lit up proved useful. Whole-body imaging revealed frequent uptake in peripheral joints and entheses, the attachment points of tendons and ligaments, across all three diagnostic groups, concentrated at mechanically loaded sites such as the feet and knees. These signals did not distinguish ankylosing spondylitis from degenerative disease, and in fact non-sacroiliac joint uptake was numerically more common in the degenerative group. Previous work in psoriatic arthritis has shown that most fluoride-positive peripheral sites are clinically silent, suggesting they reflect subclinical bone turnover rather than active inflammation. The message is clear: the whole-body scan’s real diagnostic value in this disease lies in the sacroiliac joints, not in the periphery.

Equally instructive was what did not correlate. Blood inflammatory markers, erythrocyte sedimentation rate and C-reactive protein, showed no significant relationship with sacroiliac uptake or with either SPARCC score, reinforcing a long-standing observation that systemic blood tests poorly reflect local disease in the axial skeleton. HLA-B27, the classic genetic marker, was far more prevalent in the ankylosing spondylitis group but showed no association with imaging measures within it. A single post hoc correlation between entheseal uptake and sedimentation rate failed to survive correction for multiple testing and was reported only as hypothesis-generating. Taken together, the data portray sacroiliac fluoride uptake as a measure of local osteoblastic burden, a dimension of disease that blood tests, genetics, and even MRI each capture only partially.

The study has honest limitations. The comparator groups were small, the degenerative group was defined by exclusion rather than a positive diagnostic standard, and the PET and MRI studies were not always performed concurrently because imaging occurred during routine care disrupted by the COVID-19 pandemic. Reliability of the MRI readings was not formally re-tested, and fluorine-18 sodium fluoride is fundamentally not inflammation-specific: it marks mineralization and perfusion regardless of trigger, so degenerative remodeling can mimic disease. Still, with excellent inter-reader agreement on the PET measurements, an intraclass correlation of 0.961 for SUVmax, and a biologically coherent pattern of results, the study makes a compelling case that fluoride PET/CT measures something MRI cannot see. Larger multicenter trials with concurrent imaging and validated clinical activity indices will determine whether this bone-seeking tracer earns a place beside MRI in the diagnostic arsenal, offering patients earlier and more complete answers about the disease quietly remodeling their spines.

Subject of Research: Fluorine-18 sodium fluoride PET/CT measurement of osteoblastic disease burden in axial spondyloarthritis

Article Title: ¹⁸F-sodium fluoride PET/CT as a complementary measure of osteoblastic disease burden in axial spondyloarthritis: a prospective exploratory study

Article References: ¹⁸F-sodium fluoride PET/CT as a complementary measure of osteoblastic disease burden in axial spondyloarthritis: a prospective exploratory study. (n.d.). https://doi.org/10.1007/s00259-026-08190-w

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08190-w

Keywords: ankylosing spondylitis, axial spondyloarthritis, PET/CT, fluorine-18 sodium fluoride, sacroiliac joints, MRI, SPARCC score, bone remodeling, osteoblastic activity, medical imaging, rheumatology, biomarkers