Chronic nonbacterial osteomyelitis in children: a retrospective multicenter study

In this retrospective study we report the disease characteristics of a group of 41
patients with chronic nonbacterial osteomyelitis collected from 6 Swiss pediatric
centers. We were able to show the huge diversity of presentation, disease course and
response to therapy.

Diagnosis

Similar to other pediatric series, mean age at onset of nonbacterial osteomyelitis
was 9.5 years with a predominance of females of 3:1 19], 20]. CNO still remains a difficult diagnosis, so mean diagnostic delay in our cohort
was 8 months. Possible reasons include the fact that plain radiographs are not sensitive
enough to detect osteomyelitis or unifocal lesions were misdiagnosed as acute infectious
osteomyelitis. This may also explain why the diagnostic delay in patients with a unifocal
presentation was significant longer than in multifocal onset. Obviously histological
examination is non-specific, but biopsy was helpful to exclude diseases like Langerhans
cell Histiocytosis X, benign or malignant bone tumors especially in unifocal lesions
21].

Although today we assume that CNO belong to the family of autoinflammatory diseases
with osseous manifestation, many patients with CNO and SAPHO-Syndrome have beside
the osteitis, symptoms from the spectrum of the spondylarthritis as an axial involvement,
the occurrence of IBD and an increased prevalence of the HLA-B27 phenotype 15]. This raises the question, whether CNO is a disease with different subgroups. Based
on accompanying features we divided our patients into 2 subgroups, with and without
features of juvenile spondylarthritis. In our cohort only 6 patients had arthritis,
of which 5 had also sacroiliitis. Arthritis in CNO has been reported in up to 80 %
in one serie 9], but most series report about 30 % (17, 23). Comparing the two subgroups we found
no significant difference regarding disease presentation and course. But interestingly
pelvic osteomyelitis was significantly associated with features of spondylarthritis,
as all patients from the spondylarthritis group had an osteomyelitis in this localization
but only one of the other 32 patients had. The pelvis is a typical site of CNO with
11–34 % of patients affected 8], 22]–24], which is consistent with our findings (25 %). This is a fact which may be helpful
in the future to distinguish subgroups of patients with CNO. On the other hand we
couldn’t find a difference in spinal involvement comparing the two groups. Hence in
our study-population an axial lesion is not a criteria for the evolution of spondylarthritis,
as it can be seen in CNO. The frequency of HLA-B27 is low in CNO compared to patients
with ERA, 21 % in our population.

Treatment

Pathogens such as proprionebacterium acnes are no longer considered relevant in the
pathogenesis of CNO, as today CNO is placed in the category of autoinflammatory diseases.
Nevertheless, half of our patients were treated with antibiotics first, but with a
significant difference across the centers. There was no difference between the use
of antibiotics for unifocal or multifocal presentation.

Fifty-seven percent of our patients responded well to the treatment with nonsteroidal
anti-inflammatory drugs (NSAID), which is in keeping with data published in other
reports 25], 26]. NSAID can control pain, which doesn’t mean there is remission radiologically 26]. In case pain does not respond to NSAIDs, a short course of corticosteroids may be
an alternative. Methotrexat (MTX) is an approved drug in children with rheumatologic
disorders, also several cases of SAPHO-syndrome responsive to methotrexat therapy
have been reported 15]. Nevertheless, MTX therapy was ceased in all but one of our patients because of lack
of improvement. In a cohort of 70 children with CNO reported by Borzutzky et al.24] 20 % had clinical remission treated with methotrexate. They observed the highest
rate of clinical remission with TNF-? inhibitors (46 %). Several other case reports
describe the efficacy of anti-TNF? therapy 27], 28]. In our population we found mixed success with TNF?-agonists as well as with bisphosphonate
therapy.

The good effect of bisphosphonate therapy has been documented in several reports 29]–31]. Rodrick et al. found a good or moderately good response in 8 out of 11 patients (73 %) to pamidronate
therapy, bone lesions resolved or showed significant improvement in the second WB-MRI.
The improvement of bone inflammation after pamidronate therapy was also reported by
Hofmann et al.32]. Although no complete radiological remission could be achieved, bisphosphonates are
an optional treatment for patients with vertebral lesion to prevent fractures and
orthopedic complications as shown by Hospach et al.33], while fractures occurred often (up to 40 %) in vertebral involvement 34]. However in the face of the long half-life time of bisphosphonates and the side effects,
the indication for bisphosphonate therapy has to be made carefully.

The inclusion of patients with this rare disease in a large registry (for example
www.printo.it/eurofever) as initiated by the Paediatric Rheumatology International Trials Organisation PRINTO
will be helpful to determine an effective treatment.

Disease course

In contrast to the previous assumption that nonbacterial osteomyelitis is a recurrent
disease, the majority of our patients (67.5 %) suffered from a chronic persistent
illness. This is a higher proportion than described in other cohorts. Gikas at al
found 49 % with a non-recurrent disease pattern 22].

We saw that disease activity may persist for years or even decades. In our cohort
9 children (22 %) with chronic disease needed treatment for more than 5 years. The
longest course of active illness was 13 years. Only 30 % came into remission, which
is similar to the findings of Catalano-Pons et al.19] in an equally large cohort. In their study 58.6 % of patients had active disease
at follow up (0.5–15 years after diagnosis). More than 25 % of the cohort examined
by Huber et al.23] had persistent CNO activity at the time of evaluation a median of 12 years later
and after a median overall duration of active disease of 5.7 years. In the follow-up
study of Duffy et al.35] the duration of symptoms ranged from 2.5 to 20 years. Huber et al.30] conclude that CRMO usually has a favorable evolution with no major sequelae. This
is in keeping with our findings, where also only 4 patients had orthopedic complications
during the observation period of 52 months (range 6 months to 14 years).

Vittecoq et al. concluded 17] that CRMO usually evolved to spondylarthropathy. Despite special considerations of
these features we can’t agree with this evolution in our cohort, because 32 of our
41 patients still had osteomyelitis at the end of the observation period. As described
by Zibroswska-Bech et al.36] we found the extra-osseous manifestation typically present at the time of diagnosis.

Our study is limited by the retrospective multicenter and multidisciplinary design,
which does not allow for analyses regarding disease details and treatment. However,
despite these shortcomings we think our findings may still be helpful to improve knowledge
and enhance awareness about this unique disease across the involved disciplines.