New Insights into Adult and Paediatric Chronic Non-bacterial Osteomyelitis CNO

In 1972, Gideon et al. first described an inflammatory bone disease of subacute or chronic nature that affects bones with a symmetrical and multifocal pattern [1]. Since then, a multitude of descriptive terms has been used throughout different medical subspecialties (orthopaedics, infectious diseases, paediatric and adult rheumatology) to describe the disease. From a current standpoint, chronic non-bacterial osteomyelitis (CNO) seems to have become the consistent denomination describing chronic bone inflammation independent of its distribution (unifocal vs. multifocal) or course (self-limited, chronically active vs. recurrent) [2]. Subsequent considerations in the 1980s identified chronic recurrent multifocal osteomyelitis (CRMO) as a severe form of CNO [3, 4].

In the past, a significant fraction of the patients received antibiotic therapy in the initial treatment, when the diagnosis was unclear, but also in part based on the reported occasional isolation of Propionibacterium acnes, currently named “Cutibacterium acnes”. However, starting in the 1990s, the infectious nature of the disease was questioned, because of increasing reports on sterile lesional biopsies and failure to detect bacterial pathogens using molecular techniques, including PCR [5].

More recently, the molecular basis of chronic inflammation has been linked to a pro-inflammatory phenotype of monocytes from CNO patients with increased expression of inflammatory interleukin (IL)-1β, IL-6 and TNF-α [6, 7], and reduced expression of immune regulatory IL-10 and IL-19 [8, 9••]. The resulting dysbalance between pro- and anti-inflammatory pathways contributes to chronic tissue inflammation, osteoclast activation, bone destruction, and in some cases hyperostosis and sclerosis of bone. In addition, mast cells have been identified to contribute to sterile inflammation [10]. The molecular basis of CNO has also been analysed in mouse models of chronic bone inflammation and in patients with familial forms of CNO or (monogenic) autoinflammatory bone disease with bone involvement. Significant knowledge has been gained by the identification of gene mutations associated with bone inflammation in mice (Pstpip2) and humans (IL1RN), predominantly leading to NLRP3 inflammasome activation, the production of pro-inflammatory cytokines IL-1β, TNF-α and IL-18, and/or their reduced regulation [9••, 11, 12••].

Several authors have linked CNO to the concept of spondyloarthropathies (SPA) [13]. However, in most cohorts, HLA-B27 positivity ranges around frequencies in the normal healthy population or is mildly elevated, and male predominance is not present [14, 15••]. Further classical features of SPA, such as uveitis or urethritis, are also not present [15••]. Clinical and pathophysiologic proximity of CNO to the so-called SAPHO syndrome, that usually affects adults, and overlap with psoriasis syndromes have also been considered. The acronym SAPHO stands for synovitis, acne, pustulosis, hyperostosis and osteitis syndrome and was first described in 1987. Since then, a considerable overlap between both entities, SAPHO and CNO/CRMO, has been discussed, which is mostly based on inflammatory cutaneous manifestations in a subset of CNO patients [16‐18, 19••, 20]. Furthermore, also in adult SAPHO syndrome patients, features of spondyloarthropathies are present. Of note, the possible connection to SPA dominates the SAPHO literature much more than in paediatric CNO [21].

Following the historic definition of systemic or organ specific inflammation in the absence of self-reactive lymphocytes and high-titre autoantibodies, and the aforementioned central involvement of innate immune mechanisms, CNO and SAPHO have been classified as autoinflammatory diseases [22, 23].

Inflammatory bone lesions may be uni- or multifocal. With the exception of the neurocranium, which is almost never affected in CNO, inflammation can affect all sites of the skeleton [2]. Typical sites of inflammation include the metaphyses/and epiphyses [24] of long bones of the extremities, the shoulder girdle including the clavicle and the sternum, vertebral bodies, and (in rare cases) the mandible. From a patients’ view, recurrent clinical symptoms, including local swelling, pain and impairment of motion, are common complaints. Imaging techniques suggest a primary chronic course of periosteal and/or endosteal inflammation, osteitis and osteomyelitis. Clinically and also radiologically, arthritis of adjacent or remote joints may be present. In a few cohorts, a strong clinical link towards SPA has been noted [13]. However, this does not seem to be a general feature in children and adolescents [15••].

Other organ systems may be involved as well. Up to 20% of CNO patients develop skin manifestations, including palmoplantar pustulosis, cystic acne and psoriasis [15••]. Chronic inflammatory bowel disease may be present in up to 10% of patients [25, 26]. Occasionally, hepatosplenomegaly and lymph node enlargement are noted (up to 3% of patients). Ocular or cardiac manifestations are generally rare. Uveitis or episcleritis are no typical features of CNO [15••].

Laboratory tests usually show mild to moderate elevation of inflammatory parameters, including CRP and ESR. Preliminary reports promise potential for a set of serum proteins to act as biomarkers for CNO. These include the pro-inflammatory cytokine IL-6, the chemokine CCL-11/eotaxin and others, and may help to differentiate CNO patients from JIA and osteoarticular infections [27]. However, findings need to be validated using alternative technologies in unrelated multi-ethnic cohorts. Thus, and due to the limited general availability, the clinical relevance of these findings is currently limited. Based on register data from Europe, antinuclear antibodies (ANA) are present in up to 39% of patients [15••]. However, the proportion of ANA-positive CNO patients in large individual cohorts is usually much lower, ranging around 10%. The number of lesions, the clinical picture, inflammatory laboratory parameters and the response to treatment did not differ between ANA-positive and ANA-negative patients [15••]. Thus, the relevance of ANA positivity in CNO requires to be challenged and elucidated further. For now, no conclusive clinical or laboratory autoimmune features have been reported.

Taken together, in the absence of reliable and widely accepted laboratory tests, disease biomarkers or known genetic causes, the definition and classification of CNO have to rely on descriptive parameters. Several attempts have been made to develop classification or diagnostic scores for CNO and to separate its clinical picture/symptoms from acute bacterial osteomyelitis or malignant bone disease. All include CNO typical patterns of bone inflammation or monofocal bone disease with histological features resembling chronic inflammation in the absence of malignancy, the absence of other signs severe systemic inflammation (such as fever, high CRP and/or ESR), and chronic disease courses. As such, scores stay descriptive and have not been validated prospectively and/or in unrelated cohorts [28‐30].

In patients who do not exhibit clinical symptoms and/or radiographic findings conclusive for CNO, a biopsy of the/a representative lesion is necessary. Tissue sections should be read by pathologists experienced with bone malignancies and CNO. First and foremost, malignancy (primary bone tumours, metastases, lymphoma, etc.) and systemic disease (Langerhans cell histiocytosis) require to be excluded. While some features, such as coexistence of “acute” inflammatory infiltrates (neutrophils, macrophages) with chronic inflammation (characterised by lymphocytes, plasma cells and monocytes), and/or bone sclerosis, are more common in CNO than other conditions, they are not specific. Chronic infections e.g. may deliver similar histopathological patterns [5, 31, 32].

Thus, fresh (not paraffin embedded) bone tissue requires to be tested for infectious agents. Microbial analyses should include long-term standard cultures, including the search for mycobacteria. In addition, molecular microbial analysis of biopsies, including universal microbial rRNA amplification, mycobacteria search PCRs and possibly next-generation sequencing (NGS, where available and of interest), may be performed. In most patients, non-suspicious microbial results will be noted.

However, in individual cases, Propionibacterium/Cutibacterium acnes may be present, which complicates the diagnosis as they may represent contamination [5, 33]. Thus, a clinical dilemma exists as acute or chronic bacterial osteomyelitis may not exhibit high inflammatory parameters, especially when low-virulent strains of bacteria are present, such as Propionibacteriae [34, 35].

The better the individual physician or clinical division is acquainted with diagnosing CNO, the fewer biopsies will be performed. During clinical work-up until infectious osteomyelitis is excluded, an initial antibiotic therapy may be reasonable. However, if the clinical symptoms resemble those “typical” for CNO, antibiotic therapy may be omitted. Of note, throughout international cohorts, antibiotic therapy has been reported in as many as 38% of patients [15••].

In SAPHO, as in CNO, the fundamental clinical component is inflammatory osteitis, which may result in hyperostosis. Most frequently affected regions include the rib cage (ribs and sternum), the spine and long bones of the extremities. This largely resembles the pattern in CNO [36]. Since arthritis/synovitis and acne are included in the acronym, it appears that SAPHO is closely related to childhood CNO, but covers bone inflammation in the context of associated cutaneous manifestations in a single individual. This association is certainly present, but less common in classical paediatric CNO. However, since in the overall adult population beyond SAPHO patients, acne and pustulotic skin lesions are more prevalent as compared with children and young adolescents, a confounding factor may be present. Of note, one study reported that up to 67% of bone biopsies from adult SAPHO patients were positive for Propionibacterium acnes [37]. In this context, it is interesting to note that Propionibacterium acnes can trigger increased plasma levels of the chemokine IL-8 and pro-inflammatory cytokines IL-18 and TNF-α. This may be caused by stimulation of the Toll-like receptors (TLR) 2 and 4 by Propionibacterium acnes [38, 39]. However, primary antibiotic therapy for SAPHO syndrome seems only effective as long as it is administered [37]. This led to the conclusion that the presence of this bacterium at the site of the lesion or in the skin might not be the only causative trigger of the disease, but of relevance as it may amplify inflammation in otherwise predisposed individuals. In addition, observed effects of antimicrobials may also partially be explained by anti-inflammatory effect of the drug studied (azithromycin). With regard to HLA-B27, no consistent presence above the expected regional frequencies was noted also in SAPHO [40]. In this respect, SAPHO mimics CNO. Though SAPHO syndrome is usually described in late adolescents and adults; some cases of paediatric manifestations have been reported [41]. Studies including both children and adults are rare. Where comparisons are possible, adult patients sometimes may have more skin involvement, but show a comparable distribution of bone lesions. Lastly, treatment available appears less effective in adults as compared with children [42•, 15••].

The molecular pathophysiology of “sporadic” CNO/CRMO (not following Mendelian inheritance) is incompletely understood. There is a significant need to analyse pathophysiological pathways, since not only inflammatory components but also potentially post-infectious reactive features have been observed.

Monocytes isolated from peripheral blood of CNO/CRMO patients fail to produce the immune regulatory cytokine IL-10 (and its homologue IL-19) in response to stimulation with lipopolysaccharide (LPS). This has been linked with reduced activation of mitogen-activated protein kinases (MAPK) ERK1 and 2, which results in reduced activation of the transcription factor signalling protein (SP-)1 and reduced phosphorylation of histone H3 at serine position 10 (H3S10P). Reduced H3S10P results in epigenetic “closure” of the IL10 promoter, which, in context with reduced availability of SP-1, translates to reduced IL-10 expression. At the same time, pro-inflammatory IL-6, IL-20 and TNF-α are expressed at increased levels. This may be secondary to increased activation of the NLRP3 as absence of IL-10 primes inflammasome assembly and subsequently may lead to release of IL-1β, which triggers TNF-α and IL-6 expression [7, 20, 43]. Of note, in monocytes from CNO patients, DNA methylation of inflammasome associated genes IL1, PYCARD (encoding for the adaptor molecule ASC) and NLRP3 is reduced, resulting in increased gene expression [32]. However, the exact chain of events causing the significant cytokine imbalance in CNO has yet to be determined. In line with these observations, tissue analyses of bone biopsies revealed a significant deregulation/activation of the IL-1β axis [6, 9••]. Finally, these altered cytokine expression patterns may lead to increased activation of osteoclasts via the RANK/RANK ligand signalling pathways, directly affecting bone remodelling [8]. Interestingly, comparable pathophysiological findings have been seen in the serum of SAPHO patients [44].

Monogenetic systemic autoinflammatory disorders with bone involvement partially resemble “sporadic” CNO, demonstrating bone inflammation, osteolysis and hyperostosis. CRMO, the severe form of CNO, shares clinical features with chronic auto-inflammatory diseases, including deficiency of the IL-1 receptor antagonist (DIRA), and pyogenic arthritis, pyoderma gangrenosum and acne (PAPA) syndrome. Both syndromes may show lesions of cartilage and osteolysis [19]. In Majeed syndrome [45], an autosomal recessive inheritance has been documented with mutations in the lipin-2 gene (LPIN2) [46]. In addition to multifocal bone lesions resembling CNO, dyserythropoietic anaemia and pustulous dermatosis may be present.

Monogenic dominant mutations in the SH3 domain binding protein 2 (SH3BP2) can lead to a granulomatous bone lesion of the lower jaw, named hereditary cherubism, partially resembling paediatric CNO of the mandibles [19, 47].

Recently, Cox et al. reported homozygous mutations in the filamin-binding domain of FBLIM1 as well as other types of mutations in about 1% of their cohort patients, leading to bone inflammation [48]. Of note, no other paediatric cohort has been analysed genetically in a comparable detail. FBLIM1 analysis has been performed in an adult SAPHO patient cohort, but did not provide evidence of relevance in a German cohort (Assmann G et al., 2020 submitted, in press, permission given in personal communication).

A point mutation in the FGR gene, encoding for the Gardner-Rasheed feline sarcoma protein kinase, a member of SFK family (src family tyrosine kinase), has been demonstrated in Ali18 mice that resemble a model for sterile osteomyelitis. In a large cohort, approximately 13% of CNO patients also carried rare exonic variants in this particular SFK gene [49].

Further monogenic diseases resemble CNO in part: Bone inflammation has been described in familial hyperphosphatemic familial tumoral calcinosis affecting the DALNT3 gene [50], in addition to primary hypertrophic osteoarthropathy, revealing gene mutations in the prostaglandine metabolism and hypophosphatasia with mutations in the tissue non-specific alkaline phosphatase gene (reviewed in [19]).

Besides the already mentioned Ali18 mice, further mouse models affecting the proline-serine-threonine phosphatase interacting protein 2 (Pstpip2) gene have been described. The phenotype of these mice shows sterile inflammation of bone, cartilage and skin. Inflammation is mediated by chemokines and cytokines, including M-CSF, MIP-1α and IL-1β. Particularly the Pstpip2 gene is tempting as a potential target for CNO to be analysed, since on the other side mutations in the PSTPIP1 gene are implicated in the pathogenesis of another autoinflammatory syndrome, the PAPA syndrome. However, no conclusive mutations in PSTPIP1 or 2 have been revealed in sporadic CNO patient cohorts [51‐53].

Though only few patients with suspected “sporadic” CNO cohorts may indeed be affected by rare mutations in single genes, keeping these differential diagnoses in mind is of relevance, especially in very young children with CNO. Target-directed treatments aiming for the correction of the biochemical sequels may already be available for rare conditions, e.g. enzyme replacement therapy in hypophosphatasia or recombinant IL-1 receptor antagonist replacement in DIRA. For a (potentially incomplete) summary of relevant differential diagnoses, see Fig. 1.

The diagnoses of exclusion CNO and SAPHO rely predominantly on imaging techniques. A number of differential diagnoses remains to be considered as long as generally accepted and validated diagnostic criteria including genetics and/or biomarkers are missing. Over the past years, an international expert group has been working towards the identification of CNO-related symptoms and their classification, including imaging findings. This exercise aims at the harmonization of diagnostic approaches and treatment in CNO [54, 55].

Diagnostic imaging frequently includes conventional X-rays and/or regional magnetic resonance imaging (MRI) at sites of pain. Occasionally, computed tomography (CT) is used if MRI is unavailable or does not deliver sufficient detail in depicting the bony structure. In children, CT imaging should be avoided. Whole body imaging is a central imaging tool for the diagnosis of CNO and is performed to monitor for additional sites of bone inflammation, especially in the vertebral column. Whole body (WB-)MRI is the gold standard to screen for additional, potentially clinically silent lesions. In the past, ^99mtechnetium bone scintigraphy was used, but should be considered obsolete and only be used if whole body MRI/MRI is not available [24, 56].

Initially during clinical care, X-ray images are frequently taken to exclude differential diagnoses e.g. bone fractures and may (but certainly do not have to) show CNO-associated changes. Conventional X-rays may already show osteolytic lesions or hyperostotic features. Especially long bones of the extremities are prone to such changes, which appear later in the course of disease. Fractures of long bones are very rare. However, fractures can occur in vertebrae of up to 10% of CNO patients [57]. Conventional X-rays may also be normal and usually cannot distinguish CNO from the features of malignant bone diseases, such as osteosarcoma or Ewing’s sarcoma.

Currently, MRI studies are the imaging gold standard in CNO. Usually, at the time of diagnosis, regional MRIs are performed and can help to diagnose CNO and exclude differential diagnosis [58]. In the authors’ institutions, regional MRIs using TIRM (or STIR) sequences are used to localize and assess the extent of inflammation, followed by native T1 sequences and (not always) contrast-enhanced T1-weighted sequences with fat saturation to assess regional changes and surrounding tissue involvement. At diagnosis, WB-MRI provides information on the distribution of bone lesions, soft tissue involvement and/or possible additional organ involvement which may be present in malignant disease (such as neuroblastoma). Noteworthy, based on growing experience, the use of MRI contrast media can be omitted during follow-up investigations or if initial WB-MRIs at diagnosis show CNO typical multifocal patterns [59] (Fig. 2). Especially during long-term follow-up, repetitive (WB or regional) MRI can be used for assessing the extent and distribution of lesions and, thus, severity of CNO in order to guide treatment.

Though it should not be used for the differential diagnosis of CNO in children, because of radiation and general superiority of MRI, CT may be helpful in rare occasions to exclude differential diagnoses (e.g. detection of the nidus in osteoid osteoma or osteoblastoma) [60]. Based on reports in the literature, for diagnosing and management of SAPHO syndrome in adults, CT appears to be of higher diagnostic relevance in clinical practise when compared with paediatric CNO [39, 42•].

CNO/CRMO and SAPHO are autoinflammatory conditions with a strong involvement of cytokine dysregulation. While paediatric CNO is primarily associated with increased activation of innate immune mechanisms and responds to blockade of these, SAPHO patients respond to blockade of IL-17A, a primarily T lymphocyte-derived cytokine. This may support variable pathomechanisms in adult SAPHO patients as compared with paediatric CNO with adaptive immune cells playing a more pronounced role in SAPHO syndrome. While this is not scientifically proven, T cell involvement in SAPHO may reflect secondary activation of adaptive immune mechanisms as (at least some) SAPHO patients may have experienced “isolated” CNO before the onset of skin symptoms. Prospective collection of treatment response data alongside biosamples for molecular studies may result in patient stratification and individualised treatments. This requires national and international collaborations such as initiated by the PRES Eurofever initiative [15••] or the North American CARRA group in association with many international partners [54].