The molecular pathophysiology of chronic non-bacterial osteomyelitis (CNO)—a systematic review

Chronic non-bacterial osteomyelitis (CNO) is an autoinflammatory bone disorder. It primarily affects children and adolescents with a peak onset between 7 and 12 years. However, CNO can generally occur in all age groups. The clinical spectrum is variable and ranges from single asymptomatic bone lesions up to the most severe form of chronic recurrent multifocal osteomyelitis (CRMO) [1‐7]. Due to this variable clinical presentation with sometimes rather mild and unspecific clinical symptoms, diagnosis is frequently delayed or even missed. This is particularly worrying, since untreated CNO may result in bone sclerosis, pathological fractures (mainly of the vertebral bodies) sometimes with subsequent neurological symptoms, growth anomalies, pain amplification, and psychosocial problems [8‐11].

Recent discoveries contributed to a better understanding of underlying molecular mechanism leading to systemic inflammation in CNO. However, the exact molecular pathophysiology remains incompletely understood [9].

To our current understanding, several molecular disturbances contribute to the molecular pathophysiology of CNO. Bone inflammation may be the result of dysbalanced cytokine expression from innate immune cells and subsequent osteoclast differentiation and activation, resulting in bone remodeling and inflammatory bone loss [4‐6, 9, 12, 13].

Genetic predisposition for CNO/CRMO has been suspected by the occurrence of familial monogenic disorders including non-infectious osteomyelitis as one of the main clinical features. Patients with Majeed syndrome (caused by homozygous mutations in the LPIN2 gene) [27], the deficiency of interleukin-1 receptor antagonist (DIRA; caused by autosomal recessive loss-of-function mutations in IL1RN) [28‐30], and pyogenic arthritis, pyoderma gangrenosum and acne syndrome (PAPA; caused by autosomal dominant loss-of-function mutations in PSTPIP1) [31] develop severe aseptic osteomyelitis. Because of the central involvement of increased IL-1 signaling and inflammasome activation in these monogenic disorders involving CNO as a key symptom, inflammasome activation and IL-1 release appeared likely involved in the pathophysiology of “sporadic” CNO. However, screening of IL1RN did not deliver disease-causing mutations in a cohort of CNO patients [32].

Associations of CNO/CRMO with other autoinflammatory disorders, such as psoriasis or inflammatory bowel disease, and the occurrence of familial clusters further suggest a common genetic predisposition for sporadic CNO [1, 7, 33, 34]. Golla et al. performed an association study in CNO patients identifying a potential susceptibility locus on chromosome 18q21.3-22 [35]. However, findings of this study were not confirmed in subsequent efforts in larger and independent cohorts. Recently, we analyzed the polymorphisms in the proximal promoter region of the immune-regulatory cytokine interleukin 10 (IL10) in a cohort of CRMO patients. Three IL10 promoter polymorphisms rs1800896 (c.-1082A>G), rs1800871 (c.-819T>C), and rs1800872 (c.-592A>C) form distinct haplotype blocks (GCC, ACC, and ATA) that influence transcription factor recruitment capacities to regulatory elements within the proximal promoter [16, 17]. Based on promoter haplotypes, IL-10 expression can be stratified from “high” (GCC), over “medium” (ACC), to “low” (ATA). Previously, recruitment of Sp-1 was shown to depend on “permissive” GCC IL10 promoter haplotypes, which may contribute to disease outcomes in infectious disorders [15, 17, 36]. Unexpectedly, we determined an enrichment of IL10 promoter haplotypes encoding for “high” IL-10 expression (GCC), while we did not find a single CNO patient with homozygous ATA haplotypes in our cohort. Thus, IL10 polymorphisms can by themselves certainly not explain impaired IL-10 secretion by monocytes from CRMO patients [15, 17]. Therefore, it appeared likely that additional molecular mechanisms contribute to altered IL-10 activation and secretion in CNO (see above). Indeed, the absence of IL10 promoter haplotypes encoding for “low” IL-10 expression suggests that individuals with the aforementioned molecular defects contributing to reduced IL-10 expression in CNO/CRMO and ATA haplotypes may develop more severe symptoms and will therefore not be diagnosed with CNO but potentially some other inflammatory condition. While this hypothesis sounds promising, it currently remains to be investigated and proven in larger multi-national cohorts.

Recently Cox et al. identified mutations in FBLIM1 in two CNO patients from South Asia using whole-exome sequencing [37]. While the exact function of Filamin Binding LIM Protein 1 (FBLIM1) remains somewhat unclear, it acts as an anti-inflammatory molecule by regulating receptor activator of NF-κB ligand (RANKL) activation [38]. Interactions between RANK receptors on the surface of osteoclast precursor cells and soluble RANKL result in osteoclast generation and activation, thereby contributing to bone remodeling and inflammatory bone loss [13, 25, 26]. Expression of FBLIM1 is regulated by STAT3 [37]. Since IL-10 induces the activation of STAT3, reduced IL-10 expression may result in impaired STAT3 activation and subsequently altered FBLIM1 expression. Indeed, both reported patients carry IL10 promoter haplotypes encoding for “low” IL-10 expression. Therefore, the combination of IL10 promoter haplotypes encoding for “low” gene expression together with FBLIM1 variants may result in CNO [16, 39].

Treatment of CNO is based on personal experience, expert opinion, case reports, and small case series. Nonsteroidal anti-inflammatory drugs (NSAIDs) are generally used as first-line therapy [11, 14, 40‐45]. Anti-inflammatory effects of NSAIDs are achieved through inhibition of cyclooxygenase enzymes. Recently, NSAIDs were demonstrated to exert variable suppressive effects on inflammasomes [46]. Efficacy of NSAIDs in CNO may therefore be explained by the involvement of pro-inflammatory monocytes in the pathophysiology of CNO that likely contribute to the generation and activation of osteoclasts. Prostaglandins are involved in osteoclast activation, and the inflammasome activation is a hallmark of pro-inflammatory monocytes in CNO [6]. As NSAIDs, corticosteroids suppress prostaglandin production, which is achieved through the inhibition of phospholipase A1. Furthermore, corticosteroids reduce the expression of NFκB-mediated pro-inflammatory cytokines, including IL-1, IL-6, and TNF-α [6]. Some authors report successful short-term use of corticosteroids in CNO, usually over a few weeks [6, 40].

Tumor necrosis factor α centrally contributes to bone inflammation in CNO. Thus, blockade of TNF-α appears a promising therapeutic intervention. Indeed, several small case series report induction of clinical and radiological remission in CNO in response to TNF blockade [6, 8, 11, 40, 47, 48].

The bisphosphonate pamidronate inhibits osteoclast activity and exerts incompletely understood effects on inflammatory cytokine expression [6]. Pamidronate was reported to induce rapid and long-lasting remission in most CNO patients [6, 8, 49‐52].

Provided the previously described overactivation of inflammasomes and increased IL-1 release from PBMCs and monocytes from CNO patients, surprisingly, few cases of anti-IL-1 treatment have been reported with mixed response and variable outcomes (Girschick H.J. et al., unpublished data). The potential explanation for the poor response may include low tissue concentrations and pathophysiological heterogeneity in CNO among others.

Dysregulated cytokine expression from monocytes centrally contributes to the inflammatory phenotype of CNO. Recent reports on genetic and molecular alterations in CNO patients indicate complex and variable immune pathology. Response to previously empiric treatment can now be explained, based on molecular patterns in immune cells from CNO patients.