Chronic nonbacterial osteomyelitis and inflammatory bowel disease in children: A French national cohort

Abstract

Objectives

Inflammatory bowel disease (IBD) and chronic nonbacterial osteomyelitis (CNO) is a rare association. Its prevalence is probably underestimated due to low awareness among clinicians. The aim of this study was to describe the French pediatric national cohort and to estimate the incidence of CNO among IBD.

Methods

We retrospectively collected all reports of IBD‐CNO association from pediatric gastroenterological and rheumatologic units between 2000 and 2024. We analyzed demographic data; age of the onset; IBD type; CNO localizations; dermatological lesions; explorations results; medications; follow‐up. Incidence was estimated using French regional inception population‐based cohort EPIMAD data.

Results

Thirty‐seven patients were identified. Median age at the onset was 13.5 years (interquartile range [IQR] 11.5–14.5). CNO diagnosis preceded IBD in 11 cases, followed it in 13 cases, while 13 IBD‐CNO cases were diagnosed simultaneously (median interval 1 year). IBD included 25 Crohn's disease, 10 ulcerative colitis, 2 IBD‐unclassified. Bone localizations predominated in lower limbs and pelvis (75.3%). Fourteen children (37.8%) had dermatological lesions. At last follow‐up (median 3 years, IQR 1.25–4.75), 28 patients (75.7%) were in IBD‐CNO remission, having mostly anti‐tumor necrosis factor (TNF)‐α (p = 0.041, 82.1% remission, 44.4% relapse) in monotherapy (60.7%). We estimate the incidence of this association among French pediatric IBD at 1 case for 20,000 patients‐year.

Conclusion

This study describes a large national cohort of pediatric IBD‐CNO association. Gastroenterologist should request a whole body‐magnetic resonance imaging when IBD patients present chronic bone pain. Multidisciplinary management involving rheumatologists and dermatologists is recommended. Most patients responded to anti‐TNF‐α antibodies, which could avoid bone complications.

What is Known

• Chronic nonbacterial osteomyelitis (CNO) is a rare pediatric disease, presenting with bone pain, often multifocal, and diagnosed with whole‐body magnetic resonance imaging.

• Inflammatory bowel disease (IBD) and CNO is a rare pediatric association, which incidence is probably underestimated.

What is New

• This study describes a new large national cohort of pediatric IBD‐CNO association.

• These data confirm previous data that most of patients are in remission with anti‐tumor necrosis factor‐α treatment in monotherapy.

• Because dermatological lesions could be associated, we recommend a multidisciplinary care management with rheumatologist and dermatologist.

1. INTRODUCTION

Inflammatory bowel disease (IBD) and chronic nonbacterial osteomyelitis (CNO) is a rare association. In literature, first case was described in 1990. ¹ To date, approximately 150 cases have been reported in children, the Porto Group of the European Society for Gastroenterology, Hepatology, and Nutrition (ESPGHAN) recently published the largest cohort of 45 patients, and pediatric teams have begun to assess its prevalence. ² , ³ , ⁴ , ⁵ , ⁶ However, it is probably underestimated due to low awareness among clinicians.

Indeed, CNO is a rare disease, which can present as unifocal or multifocal lesions (previously known as Chronic Recurrent Multifocal Osteomyelitis, CRMO), and was first described in 1972. ⁷ As it begins in childhood, it is mostly known by pediatric rheumatologist. Its prevalence is estimated to range from 1:160,000 to 1:2,000,000. ⁸ CNO can induce chronic bone pain, fractures, bones deformations, and growth disturbance. ⁹ IBD is more frequent, with an increasing incidence, having a pediatric onset in 9%–10% of cases. ¹⁰ , ¹¹ , ¹² The frequency of extra‐intestinal manifestation (EIM) in IBD pediatric population‐based studies varies from 5% to 37%, involving hepatobiliary, ophthalmologic, dermatologic, and musculoskeletal systems. ¹³ , ¹⁴ , ¹⁵ Joint involvement (sacroiliitis, peripheral arthropathy, and enthesitis) is reported in 2.4% to 18.5% of pediatric IBD cases, ¹⁵ , ¹⁶ however, CNO was less known among gastroenterologist pediatricians.

Pathophysiology of IBD‐CNO association remains unknown but may involve shared pathways. ¹⁷ IBD is a multifactorial disease involving genetic and environmental factors, immune dysregulation, and dysbiosis. ¹⁸ CNO is a chronic aseptic osteomyelitis, for which an autoinflammatory mechanism has been proposed, involving inappropriate activation of the innate immune system (reduced interleukin [IL]‐10 and IL‐19 expression, increased levels of IL‐1β, tumor necrosis factor alpha [TNF‐α], and IL6), ¹⁹ , ²⁰ increased osteoclast activation through the Receptor Activator of Nuclear factor Kappa‐Β (RANK)/RANK Ligand (RANKL) pathway, and microbiome alteration. ¹⁷ CNO may improve with non‐steroidal anti‐inflammatory drugs (NSAIDs) (about half of patients responding), and second‐line treatments are used for non‐responders or patients with spinal lesions (anti‐TNF‐α and bisphosphonates). ²¹ , ²² , ²³ Efficacy data and comparative studies are limited, and no trials have compared NSAIDs, bisphosphonates and anti‐TNF‐α. Bisphosphonates are preferred for spinal lesions and anti‐TNF‐α for extra‐osseous manifestations (EOM) with arthritis. ²⁴ Beyond bone involvement, classical EOM include skin involvement, arthritis, while IBD is the most common inflammatory disease associated. ²¹ , ²⁴

As IBD‐CNO association is rare, the aim of this study was to describe our French national cohort, estimate the incidence of this association in the national pediatric IBD cohort, and compare our data with literature data.

2. METHODS

2.1. Ethics statement

The protocol was approved by the national GFHGNP (Groupe Francophone d'Hépato‐Gastro‐Nutrition Pédiatrie) ethics committee (N° RCB 2023‐49). An information letter and a non‐opposition form were provided to patients and their parents/legal guardians.

2.2. Study design and data collection

We retrospectively collected and analyzed all reported cases of IBD‐CNO association from French pediatric gastroenterological and rheumatologic units between 2000 and 2024. All members of the Groupe d'Etude thérapeutique des Affections Inflammatoires du Tube Digestif (GETAID) pédiatrique and Société Francophone de Rhumatologie Et de Maladie Inflammatoire Pédiatrique (SOFREMIP) groups who had IBD‐CNO cases submitted their data to a central secure digital site during the 2024–2025.

All children had confirmed diagnosis of IBD by pediatric gastroenterologist, and CNO by pediatric rheumatologist, according to the revised ESPGHAN Porto criteria for IBD, ²⁵ and Jansson and/or Bristol criteria for CNO (Table S1). ²⁶ , ²⁷ Exclusion criteria were patients older than 18 years at diagnosis and IBD and/or CNO diagnosed before year 2000.

For all patients, we recorded anonymously patient demographic data (age, sex); age at onset of IBD and CNO; IBD type and localization (Paris Classification); CNO localizations; dermatological lesions described in these two conditions (included psoriasis, pustulosis, neutrophilic dermatosis as pyoderma gangrenosum and severe acne conglobata, hidradenitis suppurativa); explorations results (blood tests, fecal calprotectin, entero‐magnetic resonance imaging [MRI], upper endoscopy and ileo‐coloscopy, whole body [WB]‐MRI, bone biopsies); medications; follow‐up duration; clinical evolution. This chart review was performed using a standardized data collection form, by the two investigators of the study (pediatrician gastroenterologist and rheumatologist). WB‐MRI protocols used T2 STIR or T2 DIXON and T1 sequences.

2.3. Definitions

Biologic inflammatory syndrome was defined by C‐reactive protein (CRP) level >5 mg/L and/or erythrocyte sedimentation rate (ESR) > 20 mm. CNO localizations were revealed by hyperintensity on STIR or DIXON‐T2 and/or hypointensity on T1‐weighted WB‐MRI. For treatment analysis, as salicylates was never used in monotherapy, immunosuppressor or biotherapy with or without salicylates association were called “monotherapy,” “combination therapy” associated biotherapy and immunosuppressor, and “dual‐targeted therapy” (DTT) combined two biologic agents or biologics with small molecules. Remission was defined by both IBD‐CNO remission: for IBD by normal clinical score (Pediatric Crohn's Disease Activity Index for Crohn disease, CD, ≤10; Pediatric Ulcerative Colitis Activity Index for ulcerative colitis, UC, <10), normal biologic parameters (CRP < 5 mg/L, ESR < 20 mm/h, fecal calprotectin ≤50 mcg/g) and normal MRI (no active lesions) or ileo‐coloscopy findings (mucosal healing). For CNO, remission was defined by the absence of bone pain, normal inflammatory markers (CRP, ESR), and improvement on MRI. Relapse was defined as relapse of IBD and/or CNO based on abnormal clinical, biologic, MRI, and/or ileo‐coloscopy findings.

2.4. Incidence evaluation

To estimate the national incidence of the pediatric IBD‐CNO association in France, we used the French hospital discharge database and the EPIMAD data for extrapolation. EPIMAD is an exhaustive prospective registry recording all incident IBD cases in Northern France since 1988, representing 9% of the whole French population. ²⁸ Geographical distribution of this registry is quite similar to the prevalence rate given by the French hospital discharge database. ²⁹ From 2000 to 2024, authors of the EPIMAD registry group estimate that there are 2800 incident cases of pediatric IBD in this registry, representing approximately 31,110 children with IBD in France. ¹¹

2.5. Statistical analysis

A descriptive analysis was performed for the whole cohort. Qualitative variables were described in percentages, and quantitative variables were described with median (interquartile range [IQR]). Fisher's test was used to compare categorical data from small groups. Medians were compared using Mann–Whitney test. Pearson correlation coefficient was used to measure the linear correlation between two sets of data. Statistical analysis was done with the software XlstatBiomed 2020.3.1 (Addinsoft).

3. RESULTS

3.1. Description of our cohort

Thirty‐seven patients were diagnosed with IBD‐CNO between 2011 and 2024, treated in 13 pediatrics centers. The sex ratio was 1.17 (20 males). Median age at IBD onset was 12.5 years (IQR 9–14), at CNO onset 12.5 years (10–14), and at IBD‐CNO diagnosis 13.5 years (11.5–14.5).

IBD and CNO were diagnosed simultaneously in 13 cases. In 11 cases, CNO preceded IBD diagnosis, with a median interval of 2 years (1–3), and in 13 cases, IBD preceded CNO diagnosis, with a median interval of 2 years (1–5). For the entire cohort, the median interval was 1 year (0–3).

Treatments ongoing and blood inflammatory markers at the time of diagnosis of both IBD and CNO are presented in Table 1. Patients with simultaneous IBD‐CNO diagnosis were treatment‐naïve.

Table 1.

Current treatments and blood inflammatory markers at the time of diagnosis of both IBD and CNO.

	CNO first diagnosis n = 11	IBD first diagnosis n = 13	p‐value	IBD‐CNO diagnosis simultaneously done n = 13
Treatments	Treatments before IBD diagnosis	Treatments before CNO diagnosis		None
NSAIDs	10 (90.9%)	0	<0.001	/
Bisphosphonates (pamidronate, zoledronate)	5 (45.5%)	0	0.011	/
Steroids (or Modulen® diet)	0	13 (100%)	<0.001	/
Salicylates	0	4 (30.7%)	0.098	/
Immunosuppressives	0 (0%)	8 (61.5%)	<0.01	/
Azathioprine	0	4
Methotrexate	0	4
Biotherapies	5 (45.4%)	11 (84.6%)	0.082	/
Anti‐TNF‐α	4 (36.3%)	9 (69.2%)	0.11
Infliximab	1	5
Adalimumab	2	3
Golimumab	0	1
Etanercept	1	0
Anti‐IL‐17 (secukinumab)	1	0	/	/
Anti‐IL‐12–23 (ustekunimab)	0	1	/	/
JAK inhibitors (tofacitinib, upadacitinib)	0	0	/	/
Anti‐integrins (vedolizumab)	0	1	/	/
Immunosuppressives and/or biotherapies (patients)	5 (45.4%)	12 (92.3%)	0.023	/
(Since a median duration of)	(10 months, IQR 9–12)	(14 months, IQR 8.5–21.8)	0.28
Blood inflammatory markers at both diagnoses
Biological inflammatory syndrome	8 (88.9%) (on 9 data available)	5 (41.7%) (on 12 data available)	0.067	12 (92.3%) (on 13 data available)
CRP median (IQR) (mg/L)	25 (IQR 21.5–68.5)	10 (IQR 2.5–21.2)		25.5 (IQR 12–53.8)
ESR median (IQR) (mm/h)	28 (IQR 20.2–31.5)	27 (IQR 5–45)		64 (IQR 35–87)
Platelets median (IQR) (G/L)	513 (IQR 458–552)	330 (IQR 288–349)		470 (IQR 426–576)

Abbreviations: CNO, chronic nonbacterial osteomyelitis; CRP, C‐reactive protein; ESR, erythrocyte sedimentation rate; IBD, inflammatory bowel disease; IQR, interquartile range; JAK, Janus kinase; NSAIDs, non‐steroidal anti‐inflammatory drugs; TNF‐α, tumor necrosis factor alpha.

IBD types included 25 CD (67.6%), 10 UC (27%), and 2 IBD‐unclassified (IBD‐U, 5.4%). For CD, most patients were 10–17 years old at diagnosis (A1b n = 16, 64%, A1a n = 9, 36%), 15 had impaired growth (G1, 60%), localizations were mostly ileocolonic (L3 n = 17, 68%, L1 n = 6, 24%, L2 n = 2, 8%), with 12 upper disease (48%) (L4a n = 10, 40%, L4ab n = 2, 8%) and 4 (16%) perianal disease (P1). During evolution 24 patients (96%) were B1, one patient was B2. For UC, seven children (70%) were 10–17 years old, localizations were mostly pancolonic (E4 n = 7, 70%, including three severe cases S1, E2S0 n = 2, 20%, E3S0 n = 1, 10%).

At CNO diagnosis, five patients (13.5%) had moderate fever (38°C–40°C). Patients reported arthralgia and chronic bone pain at various sites, most frequently in the lower limbs and pelvis. Clinical examination exhibited bone pain in all patients. Bone biopsies were performed in five cases, showing nonspecific inflammatory chronic osteomyelitis. Considering a five‐skeleton classification (n = 37 patients): most of CNO lesions identified on WB‐MRI were located in the lower limbs (n = 34, 91.9%), then pelvis skeleton (n = 19, 51.3%), upper limbs (n = 14, 37.8%), axial skeleton (n = 13, 35.1%) and scapula skeleton (n = 5, 13.5%), with a median of two localizations (1–3). Considering all bones localizations on WB‐MRI (n = 304, bilateral counted as 2), patients had a median of eight localizations (3–11), mainly in lower limbs (n = 168, 55.2%) (Figure 1). Long bones were affected at proximal or distal epiphysis and/or metaphysis, rarely in diaphysis, and lesions were frequently bilateral for limbs (73.4%) and pelvis (63.1%).

Figure 1.

Bone CNO localizations in the cohort (% of total of bone lesions on whole body‐MRI), prepared with the help of https://smart.servier.com/. CNO, chronic nonbacterial osteomyelitis; MRI, magnetic resonance imaging.

In addition to the inflammatory markers presented in Table 1, Antinuclear antibodies were positive in four patients (21%, of 19 available data), anti‐neutrophil cytoplasmic and anti‐Saccharomyces cerevisiae antibodies were positive in eight patients (26.7%, 80% of UC) and 11 patients (36,7%, 44% of CD), respectively, of 30 available data. Two HLA B27 were positive (15.3%, on 13 available data).

Extra‐intestinal/osseous associated manifestations: Fourteen children (37.8%) had 15 dermatological lesions described as: diffuse pustulosis (n = 2), pyoderma gangrenosum (n = 1), eczema or psoriasis possibly induced by biotherapies (n = 12). Other auto‐immune diseases included uveitis (n = 1), celiac disease (n = 1).

Treatments received during the follow‐up for the whole cohort (Table 2). After both diagnosis, n = 34 patients (91.8%) received anti‐TNF‐α, followed by azathioprine (n = 14, 37.8%), methotrexate (n = 10, 27%), salicylates (n = 10, 27%), ustekinumab (n = 7, 18.9%), vedolizumab (n = 3, 8.1%), and JAK inhibitors (n = 2, 5.4%), with a majority receiving combination therapy (n = 21, 56.7%). At last follow‐up, n = 29 patients (78.3%) had a monotherapy (combination therapy n = 5, 13.5%; DTT n = 3, 8.1%). Specifically for CNO, n = 8 patients (21.6%) received NSAIDs, and n = 9 patients (24.3%) received bisphosphonates with a median of 4.5 cures (3.25–5.75), on six available data. Bisphosphonates were used to relieve bone pain when NSAIDs were ineffective or contraindicated due to IBD, or in patients with spinal lesions: they were effective in six cases, resulting in regression of lesions on MRI; three patients were non‐responders and required anti‐TNF‐α therapy, with led to clinical and radiological improvement.

Table 2.

Treatments received during the follow‐up for the whole cohort.

Treatments received by n = 37 patients	Before both diagnosis, n (%)	During all follow‐up, n (%)	At the end of the follow‐up, n (%)
NSAIDs	10 (27%)	8 (21.6%)	0
Bisphosphonates (pamidronate, zoledronate)	5 (13.5%)	6 (16.2%)	0
Corticosteroids (and/or diet with modulen®)	13 (35.1%) for IBD 1st diagnosis	33 (89.2%)	0
Salicylates	4 (10.8%)	10 (27%)	6 (16.2%)
Azathioprine	4 (10.8%)	14 (37.8%)	4 (10.8%)
Methotrexate	4 (10.8%)	10 (27%)	3 (8.1%)
Anti‐TNF‐α antibodies	13 (35.1%)	34 (91.8%)	27 (72.9%)
Anti IL12–23 (ustekinumab)	1 (2.7%)	7 (18.9%)	6 (16.2%)
Anti IL17 (secukinumab)	1 (2.7%)	0	0
JAK inhibitors (tofacitinib, upadacitinib)	0	2 (5.4%)	2 (5.4%)
Anti‐integrins (vedolizumab)	1 (2.7%)	3 (8.1%)	3 (8.1%)
Monotherapy	10 (27%)	14 (37.8%)	29 (78.3%)
Azathioprine	0	1	2
Methotrexate	1	1	0
Anti‐TNF‐α (+salicylates)	7	7 (+3)	17 (+2)
Ustekinumab (+salicylates)	1	1 (+1)	4 (+1)
JAK inhibitors	0	0	2
Vedolizumab (+salicylates)	0	1	1
Secukinumab	1	0	0
Combination therapy	7 (18.9%)	21 (56.7%)	5 (13.5%)
Anti‐TNF‐α + azathioprine	4	12	2
Anti‐TNF‐α + methotrexate	2	8	3
JAK inhibitors + methotrexate	0	1	0
vedolizumab + methotrexate	1	0	0
Dual‐targeted therapy	0 (0%)	3 (8.1%)	3 (8.1%)
Anti‐TNF‐α + vedolizumab + salicylates	0	2	2
Anti‐TNF‐α + ustekinumab	0	1	1

Abbreviations: IBD, inflammatory bowel disease; JAK, Janus kinase; NSAIDs, non‐steroidal anti‐inflammatory drugs; TNF‐α, tumor necrosis factor alpha.

Remission/relapse rates at the end of follow‐up. The median follow‐up duration for the entire cohort was 3 years (1.25–4.75), 3 years after IBD diagnosis (1.75–6), and 3.5 years after CNO diagnosis (2.5–5). At the end of follow‐up, 28 patients were in remission for both IBD and CNO (75.7%) and 9 experienced relapse. The degree of correlation between patients in remission for IBD (n = 31) and those in remission for CNO (n = 32) was positive (r = +0.26). Comparison of patients in remission versus relapse is presented in Table 3: there were no significant difference in sex, type of IBD, median dealy between diagnoses, monotherapy/combination therapy/DTT, or initial bones localizations. However, patients in remission received more anti‐TNF‐α (p = 0.041, mostly in monotherapy, n = 17/28, 60.7%), and patients who relapsed had more first IBD diagnosis (p = 0.042), without more extensive and/or penetrating IBD in this group.

Table 3.

Comparison between patients in remission or relapse, and UC/CD at the end of the follow‐up.

	N = 37 patients	Relapse at the end of the follow‐up N = 9 patients	Remission at the end of the follow‐up N = 28 patients	p‐Value	N = 35 patients (2 IBD‐U excluded)		p‐Value
					UC n = 10	CD n = 25
Follow‐up duration: years (IQR)	3 (1.25–4.75)	1.25 (0.5–5)	3 (2–4)	0.51	3.5 (3–5)	2.5 (1.25–4)	0.29
1st diagnosis CNO	11 (29.7%)	2 (22.2%)	9 (32.1%)	0.69	3 (30%)	7 (28%)	1
1st diagnosis IBD	13 (35.1%)	6 (66.7%)	7 (25%)	0.042	3 (30%)	10 (40%)	0.71
IBD‐CNO diagnosis at the same time	13 (35.1%)	1 (11.1%)	12 (42.9%)	0.12	4 (40%)	8 (32%)	0.71
Median interval between both diagnoses: years (IQR)	1 (0–3)	1 (1–4)	1 (0–2.25)	0.23	1.5 (0–2.75)	1 (0–2)	0.91
Sex (M/F)	20/17 (54%)	5/4 (55.5%)	15/13 (53.5%)	1	7	11	0.26
Type of IBD n (%)					At the end of the follow‐up
IBD‐U	2 (5.4%)	0 (0%)	2 (5.4%)	1
UC	10 (27%)	3 (33.3%)	7 (25%)	0.68	Remission 7 (70%)	Remission 19 (76%)	0.69
CD	25 (67.5%)	6 (66.6%)	19 (67.8%)	1	Relapse 3 (30%)	Relapse 6 (24%)
Extensive and/or penetrating IBD diseases (UC: E3 E4; CD: L3+/‐B2 B3)	25 (68%)	7 (78%)	18 (64%)	0.69
	UC 8	UC 3	UC 5		8 (80%)	17 (68%)	0.69
	CD 17 (IC 0)	CD 4	CD 13
Treatments at the end of the follow‐up, n (%)
Monotherapy	29 (78.3%)	7 (77.7%)	22 (78.5%)	1	8 (80%)	19 (76%)	1
Azathioprine	2 (5.4%)	0 (0%)	2 (7.1%)	1	1 (10%)	1 (4%)	0.5
Anti‐TNF‐α (+salicylates)	19 (51.3%)	2 (22.2%)	17 (60.7%)	0.06	5 (50%)	12 (48%)	1
Ustekinumab (+salicylates)	5 (13.5%)	2 (22.2%)	3 (10.7%)	0.58	1 (10%)	4 (16%)	1
JAK inhibitors	2 (5.4%)	2 (22.1%)	0 (0%)	0.054	0 (0%)	2 (8%)	1
Vedolizumab (+salicylates)	1 (2.7%)	1 (11.1%)	0 (0%)	0.24	1 (10%)	0 (0%)	0.29
Combination therapy	5 (13.5%)	2 (22.2%)	3 (10.7%)	0.58	0 (0%)	5 (20%)	0.29
Anti‐TNF‐α + azathioprine	2 (5.4%)	0 (0%)	2 (7.1%)	1	0 (0%)	2 (8%)	1
Anti‐TNF‐α + methotrexate	3 (8.1%)	2 (22.1%)	1 (3.6%)	0.14	0 (0%)	3 (12%)	0.54
Dual‐targeted therapy	3 (8.1%)	0 (0%)	3 (10.7%)	0.56	2 (20%)	1 (4%)	0.19
Anti‐TNF‐α + vedolizumab (+salicylates)	2 (5.4%)	0 (0%)	2 (7.1%)	1	1 (10%)	1 (4%)	0.5
Anti‐TNF‐α + ustekinumab	1 (2.7%)	0 (0%)	1 (3.6%)	1	1 (10%)	0 (0%)	0.29
Molecules at the end of the follow‐up, n (%)	n = 71
Salicylates	6 (16%)	2 (22.2%)	4 (14.8%)	0.62	5 (50%)	1 (4%)	<0.01
Azathioprine	4 (11%)	0 (0%)	4 (14.8%)	0.55	1 (10%)	3 (12%)	1
Methotrexate	3 (8/1%)	2 (22.2%)	1 (3.6%)	0.14	0 (0%)	3 (12%)	0.54
Anti‐TNF‐α	27 (73%)	4 (44.4%)	23 (82.1%)	0.041	7 (70%)	18 (72%)	1
Ustekinumab	6 (16%)	2 (22.2%)	4 (14.8%)	0.62	2 (20%)	4 (16%)	1
JAK inhibitors	2 (5.4%)	2 (22.2%)	0 (0%)	0.054	0 (0%)	2 (8%)	1
Vedolizumab	3 (8.1%)	1 (11.1%)	2 (5.4%)	1	2 (20%)	1 (4%)	0.19
Initial WB‐MRI CNO localizations n (%)	(n = patients)	(n = patients)	(n = patients)		(n = patients)	(n = patients)
Upper limbs	14 (37.8%)	2 (22.2%)	12 (42.8%)	0.43	6 (60%)	7 (28%)	0.12
Lower limbs	34 (91.9%)	7 (77.7%)	27 (96.4%)	0.14	10 (100%)	22 (88%)	0.54
Axial skeleton	13 (35.1%)	5 (55.5%)	8 (28.5%)	0.23	3 (30%)	9 (36%)	1
Scapula skeleton	5 (13.5%)	0 (0%)	5 (17.8%)	0.31	2 (20%)	2 (8%)	0.56
Pelvis skeleton	19 (51.3%)	3 (33.3%)	16 (57.1%)	0.27	4 (40%)	14 (56%)	0.47
Median according 5 skeleton classification (n = skeleton)	2 (IQR 1–3)	2 (IQR 1–2)	2 (IQR 1.75–3)	0.2	2.5 (IQR 2–3)	2 (IQR 1–3)	0.39
Median according to all bone's localizations (n = bones)	8 (IQR 3–11)	7 (IQR 3–10)	9 (IQR 3.75–12.2)	0.11	10 (IQR 8.25–11.8)	6 (IQR 3–11)	0.17

Abbreviations: CD, Crohn disease; CNO, chronic nonbacterial osteomyelitis; F, female; IBD, inflammatory bowel disease; IBD‐U, IBD‐unclassified; IQR, interquartile range; JAK, Janus kinase; M, male; MRI, magnetic resonance imaging; TNF‐α, tumor necrosis factor alpha; UC, ulcerative colitis; WB‐MRI, whole body‐MRI.

Regarding the nine relapses: two patients relapsed for both IBD‐CNO (one monotherapy JAK inhibitors, one anti‐TNF‐α + methotrexate), three patients for CNO (two monotherapies anti‐TNF‐α, ustekinumab, one anti‐TNF‐α + methotrexate), and four patients for IBD (four monotherapies anti‐TNF‐α, ustekinumab, vedolizumab, and JAK inhibitors). Both patients with JAK inhibitors monotherapy had extensive CD: anti‐TNF‐α was withdrawn because anaphylactic shock to infliximab for one, and because of CNO diagnosis under adalimumab for the other. Both CD patients with ustekinumab monotherapy stopped anti‐TNF‐α because of secondary psoriasis. The UC patient with vedolizumab monotherapy had this treatment after he deliberately stopped anti‐TNF‐α, but he recently started this treatment. The four patients with anti‐TNF‐α monotherapy (two extensive UC), or combination therapy (two CD, one extensive) have relapsed, however three had relatively short follow‐up (0.5 year) so treatment adjustment (dose and/or frequency) had not yet been implemented.

Glancing at patients with first IBD diagnosis, 10 were receiving anti‐TNF‐α therapy when CNO was diagnosed: seven improved with same anti‐TNF‐α treatment (with sometimes NSAIDs and/or bisphosphonates, but four will relapse—three for CNO, one for IBD‐CNO), two successfully intensified anti‐TNF‐α treatment, one switched biotherapy (but IBD will relapse after two other biotherapies).

Comparison between patients with UC and CD is presented in Table 3: there were no difference in follow‐up, sex, median age at onset, first diagnosis, median delay between diagnoses, relapse rate, extensive behavior rate, CNO localizations, and treatment at last follow‐up. As excepted, UC patients were more likely to receive salicylate therapy.

3.2. Estimated national incidence

Data from the French hospital discharge database and the EPIMAD registry allowed us to estimate the national incidence of IBD‐CNO association among pediatric IBD. During the study interval (2000–2024), approximately 31,110 children were diagnosed with IBD. Based on these data, the incidence can be estimated at 0.12% over this period, which represents about 1 case for 20,000 patients‐year in France.

4. DISCUSSION

This study describes a large national cohort of pediatric IBD‐CNO association and contributes additional data to the existing knowledge on this rare but clinically relevant association. Patients presented with chronic bone pain and arthralgia and showed a moderate inflammatory syndrome. Most CNO lesions identified on WB‐MRI were located in the lower limbs and pelvic region. Most patients responded to anti‐TNF‐α therapy.

Compared with other IBD‐CNO cohorts (45 cases in the Porto Group study of ESPGHAN, 57 cases in the review by Costi et al., and 18 cases in the Italian study by Scarallo et al.), our cohort included slightly more boys, consistent with the Porto group and Italian cohorts. In contrast, the review of Costi et al. and CNO cohorts included predominantly girls. ² , ⁵ , ⁶ The median age at onset in our cohort was older (13.5 years) compared with 10–11 years in other IBD‐CNO and CNO cohorts. ⁴ , ⁹ , ²¹ , ²⁶ The interval between the two diagnoses was quite similar (0‐few months to 1–2 years), likely reflecting differences in symptom presentation and availability of rheumatology consultation (first CNO diagnosis 22%–59%, first IBD diagnosis 24%–44%, simultaneous to few months 17%–33%). In all cohorts, including ours, the type of IBD was mainly CD (60%–72%), with mostly extensive forms (L3). When specified, our CNO localizations were quite similar to other IBD‐CNO cohorts (mostly lower limbs, with variable involvement of upper limbs, clavicle, spine, and mandible) and to CNO cohorts, except for ribs that we didn't record (5%–20% in literature), and clavicle for which our incidence was lower (2.6% vs. 10%–20% in literature), although it is usually considered, as mandible, as specific location. ⁹ , ²¹ , ²⁶ A French series reported a median of five MRI‐localizations in CNO with EOM. ²¹ Consistent with our findings, Costi et al. observed that patients were more frequently treated with anti‐TNF‐α when IBD was diagnosed before CNO (45% vs. 4% when CNO was diagnosed before). In the Porto group study, CNO could also be diagnosed in patients already receiving anti‐TNF‐α therapy for IBD. Across all cohorts, including ours, most patients achieved remission with anti‐TNF‐α, with a follow‐up duration not always specified (median 2 years in the Porto group study). No complications were observed in our cohort, whereas six complications were reported in the Porto group study (three vertebral collapse, two bone deformities, one bone fracture).

Dermatological features in our cohort and pediatric IBD‐CNO cohorts ² , ⁵ , ⁶ were less frequent than those described in the CNO literature: authors describe skin lesions in approximately 50% of patients with CNO (mainly psoriasis and palmoplantar pustulosis), often preceding bone pain, but also hidradenitis suppurativa, severe acne, and recurrent aphthous stomatitis in 25% of patients. ²⁴ , ²⁶ , ³⁰ , ³¹ In our cohort, psoriasis was frequently described as a paradoxical reaction to anti‐TNF‐α; however, the retrospective design limited our ability to determine whether these lesions represented paradoxical reactions or EOM. More systematic dermatological evaluation could potentially benefit these patients. In the study of Robert and al, skin lesions were the most frequent EOM (57%) and 52% of patients had multiple EOM. ²⁴ Because patients are at increased risk of developing additional EIM/EOM once one is present, physicians should screen for other EIM/EOM when dermatological lesions are identified.

Articular manifestations are common in IBD, ¹⁵ , ¹⁶ , ³² , ³³ and the diagnosis of CNO can be challenging: lesions may clinically mimic arthritis, because long bones are often affected at proximal or distal metaphysis, resulting in pain located around the joints. ³⁴ However, careful clinical examination should distinguish arthralgia from bone pain. When osteoarticular pain persists without clinical or ultrasound evidence of arthritis, additional imaging should be performed. Before the development of MRI, diagnosis was suspected on X‐rays (osteolytic lesions surrounded by a hyper‐sclerotic rim). To avoid bone biopsies (which show nonspecific inflammatory infiltration and hyperostosis), Jansson developed a clinical‐biological‐radiological score. ²⁷ Currently, WB‐MRI is the gold standard for diagnosis, distinguishing CNO from other differential diagnosis (infection, malignancy), detecting early inflammatory changes that are not visible on X‐rays and identifying multifocal lesions. ⁹ , ³⁵ Physicians must rule out other conditions (genetic or metabolic disorders, such as vitamin C deficiency and hypophosphatasia), although the radiologic findings and clinical presentations differ. ²¹ The new European Alliance of Associations for Rheumatology (EULAR)/American College of Rheumatology (ACR) 2025 classification criteria demonstrate high specificity (98%) and sensitivity (82%): in particular, age > 3, bone localizations, absence of fever, moderate inflammatory syndrome, IBD or dermatological lesions support the diagnosis (Table S1). ³⁴ , ³⁶ When these new criteria were retrospectively applied to our cohort, 91.9% had a score ≥55 (mean 66.1, SD 10.9) (it may have been underestimated for three patients because of missing biological data). We recommend that physicians use these useful new criteria.

CNO symptoms appeared even in patients already receiving anti‐TNF‐α therapy, which is surprising because anti‐TNF‐α agents are a second‐line treatment for CNO. ²² Nevertheless, most patients showed favorable outcomes with anti‐TNF‐α therapy, possibly because of dose or frequency adjustments, and/or switching molecules in cases of immunization. In their cohort, Roberts et al. reported that 42% of patients with EOM received several anti‐TNF‐α agents during follow‐up. ²⁴ IBD was mostly extensive in IBD‐CNO cohorts, and the literature suggests that IBD associated with EIM has a more severe disease course. ¹³ Interestingly, patients with CNO and EOM also appear to have more bone lesions and require more intensive and prolonged treatment. ⁹ , ³⁴ , ³⁷ In a French cohort, 60% of patients were still receiving a treatment after 3 years of follow‐up, and children with EOM achieved complete remission in 54% of cases with anti‐TNF‐α therapy. ²⁴ In the Swedish cohort, patients had a higher biologic use compared with IBD patients without CNO, but no more immunomodulators or corticosteroids use. ⁴ In CNO studies, ustekinumab or secukinumab have mainly been used in cases associated with psoriasis, but there are very limited data regarding newer biologic therapies in IBD‐CNO association: in Costi's review, one patient achieved remission with ustekinumab, as did three patients in our cohort. In our cohort, the three patients receiving anti‐JAK and vedolizumab were not in remission; however, these patients had extensive IBD, and vedolizumab had been initiated only recently in one patient, with limited follow‐up. Overall, when physicians recognize this diagnosis, they should avoid NSAIDs, which are contraindicated in IBD, and promptly initiate anti‐TNF‐α therapy to improve both IBD and CNO outcomes. Early treatment may help prevent bone complications (fractures, bones deformations, and growth disturbance). CNO monitoring is mainly clinical, and follow‐up MRI is recommended depending on symptoms. Physical activity is allowed respective of pain.

Future research in CNO will focus on developing large prospective registry to establish consensus therapeutic recommendations and standardized disease activity scoring systems. Some authors have proposed disease activity and radiologic scoring systems, but they are not universally used, and the clinical relevance of MRI‐detected asymptomatic lesions remains uncertain. ³⁸ , ³⁹ , ⁴⁰

Regarding the incidence of CNO among patients with IBD, the recent Swedish cohort study found a rate of 0.13%, similar to our national estimation (0.12%), with a six‐fold increased risk of CNO in childhood‐onset IBD compared to non‐IBD controls. ⁴ Scarallo et al reported a prevalence of 0.4% in their national Italian pediatric IBD cohort. ⁶

This study has several limitations because of its retrospective design. We cannot exclude the possibility of missing or inaccurate data in the medical records. Moreover, because knowledge of this disease is relatively recent (since the 1990s), some cases may have remained undiagnosed, particularly before the 2010s, therefore, the true incidence is likely underestimated.

5. CONCLUSION

In our national cohort from 2000 to 2024, we estimated that at least 0.12% of pediatric patients with IBD developed CNO before, during, or after the diagnosis of IBD, with a median of 8 bone lesions, predominantly affecting the lower limbs. CNO is a rare but possible association with pediatric IBD and should be considered in cases of osteoarticular pain after excluding more common peripheral arthritis. Currently, anti‐TNF‐α biotherapy appears to be the most effective treatment. Multidisciplinary care management is recommended, involving gastroenterologists, rheumatologists, and dermatologists.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Supporting Information.