MRI bright stripes after bisphosphonates mimic disease flare in childhood bone disease

Children with chronic nonbacterial osteomyelitis who receive bisphosphonate therapy develop distinctive bright stripes in the metaphyses of their long bones on whole-body MRI, and a new study shows these stripes are a predictable consequence of treatment rather than a sign that the inflammatory disease has flared. The finding matters because bisphosphonates are given precisely when standard anti-inflammatory drugs have failed, so radiologists and rheumatologists need to know whether marrow changes on a follow-up scan represent worsening disease or the pharmacologic footprint of the drug itself.

Chronic nonbacterial osteomyelitis, or CNO, is a rare autoinflammatory bone disorder in which sterile bone lesions most commonly involve the metaphyses of the long bones, the clavicle, and the spine in children and adolescents. The multifocal form is known as chronic recurrent multifocal osteomyelitis. Whole-body MRI has become the imaging modality of choice for diagnosis and monitoring because of its high sensitivity for clinically silent lesions. Bisphosphonates, particularly pamidronate and zoledronic acid, serve as second-line therapy for patients with an inadequate response to nonsteroidal anti-inflammatory drugs or those with spinal involvement. On MRI after treatment, inflammatory lesions typically resolve substantially, with vertebral lesions showing the highest rates of complete resolution.

On plain radiographs, the effect of bisphosphonate therapy has long been visible as metaphyseal sclerotic bands produced by cyclical treatment in a growing child, the so-called zebra lines. These bands represent areas of transiently increased bone mineralization caused by intermittent inhibition of osteoclastic activity. They run parallel to the physis, and their spacing reflects both the interval between treatment cycles and the rate of skeletal growth. Zebra lines are benign, do not interfere with bone growth, and indicate the pharmacologic action of the drug rather than disease activity. What had been lacking, until now, was a systematic characterization of the corresponding marrow changes on MRI and their relationship to the dose and timing of treatment.

A team at a tertiary children’s hospital conducted a retrospective study covering January 2015 through December 2025, identifying children with CNO who had received bisphosphonate therapy through the Division of Pediatric Rheumatology and comparing them with age- and sex-matched untreated controls. The final population comprised 21 treated patients and 21 controls, with identical diagnostic distributions in both groups: 17 with chronic recurrent multifocal osteomyelitis, 3 with unifocal CNO, and 1 with SAPHO syndrome. Two blinded pediatric radiologists independently reviewed all whole-body MRI studies, and a third experienced radiologist adjudicated the 35 of 129 studies in which the primary readers disagreed.

The treated children, whose median age was 10.7 years and of whom 12 were female, underwent 30 pretreatment and 51 post-treatment whole-body MRI examinations. The 21 controls, with a median age of 10.0 years, underwent 48 examinations. After exclusions for technical problems and one incomplete examination, 129 studies were available for analysis. Readers classified any bisphosphonate effect into one of three symmetric patterns: sclerotic bands without increased signal, alternating high-signal-intensity and sclerotic bands, or predominantly bright signal bands. Marrow abnormalities were considered more consistent with CNO when they showed a patchy or geographic distribution of increased signal rather than a linear or striped pattern, were not completely symmetric, or involved the epiphysis.

The results were striking. Bone marrow changes of any type appeared in 18 of 21 treated patients, or 85.7 percent, compared with 3 of 21 controls, or 14.3 percent, a difference that was highly significant. Sclerotic bands alone occurred in both groups, at 23.8 percent of treated patients versus 14.3 percent of controls, underscoring that this pattern is not specific for treatment and may reflect Park-Harris growth recovery lines, remnants of slowed longitudinal growth during recovery from systemic or local insults. By contrast, the two bright patterns, alternating high-signal and sclerotic bands in 38.1 percent and predominantly bright signal bands in 23.8 percent, occurred exclusively after treatment. Overall, high-signal bands were seen in 61.9 percent of treated patients and in none of the controls. Accounting for repeated measures, the odds of bisphosphonate-related marrow changes were 76.9 times higher in the treated group, with a 95 percent confidence interval of 20.9 to 283.6.

The bright signal also followed a predictable dose-and-time relationship. High-signal patterns were associated with a greater number of infusions and with shorter intervals between imaging and both the first and the most recent infusion. Among studies showing changes, high-signal patterns appeared at mean intervals of 94.4 to 116.0 days from the most recent infusion and after a mean of 7.2 to 7.9 infusions, whereas sclerotic bands alone were seen considerably later, at a mean of 286.7 days after the last infusion and after a mean of 4.8 infusions. In regression models accounting for repeated measurements within patients, shorter intervals since the initial and most recent infusions and a greater number of infusions were each significantly associated with high-signal changes compared with sclerotic bands alone.

The anatomical distribution added further diagnostic clues. Bisphosphonate-related marrow changes concentrated heavily in the lower extremities, with the distal femur, proximal tibia, and distal tibia the most involved sites for both readers. This pattern makes biological sense: the physes about the knee account for the majority of lower-limb longitudinal growth, followed by the distal tibia, and bisphosphonate-related changes would be expected to be most pronounced in the fastest-growing bones. Mechanistically, both pamidronate and zoledronic acid bind hydroxyapatite at sites of active bone remodeling and inhibit farnesyl pyrophosphate synthase in osteoclasts. In a growing child, the most metabolically active mineralizing front lies in the metaphysis immediately beneath the growth plate, so each treatment cycle lays down a mineral-dense band at that site, and successive infusions produce successive bands whose number and spacing track with treatment courses and growth rate.

Why the metaphysis also lights up bright on fluid-sensitive sequences remains uncertain, because the signal has not been correlated with histology. One possibility is that it represents an expanded zone of pre-ossified endochondral tissue, which contains chondrocytes and an extracellular matrix rich in type II collagen and proteoglycan, with abundant bound water that appears bright on water-sensitive MRI. Distinguishing this signal from active CNO is clinically important, and the study offers practical criteria: symmetric bright stripes of similar thickness with sharp margins running parallel to the growth plate characterize the treatment effect, whereas CNO-related metaphyseal changes were asymmetric in all cases according to one radiologist and in 85.0 percent according to the other. Interobserver agreement on the presence of a bisphosphonate effect was substantial, with a kappa of 0.71, though separating the two bright patterns from each other proved more subjective, prompting the authors to recommend combining them into a single high-signal category.

The authors caution that the study was retrospective and that bisphosphonate dosing and intervals varied widely at the treating rheumatologist’s discretion, so a prospective study with imaging at predefined intervals after infusion would be needed to define the time course precisely. The scan-level analysis also means patients contributing several studies are represented more than once, and some upper-extremity metaphyses lay outside the imaging field of view. Even so, the central message is clear: symmetric metaphyseal bright stripes on whole-body MRI in a bisphosphonate-treated child with CNO are a recognizable, dose- and time-related treatment effect that should not be misread as active disease, while sclerotic bands alone, seen in a meaningful minority of untreated children as well, cannot on their own distinguish treatment from background.

Subject of Research: Metaphyseal bone marrow changes on whole-body MRI after bisphosphonate therapy in children with chronic nonbacterial osteomyelitis

Article Title: Metaphyseal changes on whole-body magnetic resonance imaging following bisphosphonate therapy in children with chronic nonbacterial osteomyelitis: distinguishing treatment effect from disease flare

Article References: Karmazyn, B., Kleppel, H., Chilukuri, S. R., Forbes-Amrhein, M. M., Starkey, I. A., Newman, C. L., Wanner, M. R., Imel, E. A., Hays, M. E., & Oliver, M. S. (2026). Metaphyseal changes on whole-body magnetic resonance imaging following bisphosphonate therapy in children with chronic nonbacterial osteomyelitis: distinguishing treatment effect from disease flare. Pediatric Radiology. https://doi.org/10.1007/s00247-026-06819-4

Image Credits: AI Generated

DOI: 10.1007/s00247-026-06819-4

Keywords: chronic nonbacterial osteomyelitis, CRMO, bisphosphonates, whole-body MRI, pamidronate, zoledronic acid, metaphysis, zebra lines, pediatric radiology, bone marrow edema, treatment effect, differential diagnosis