Background
Ankylosing spondylitis (AS) is a debilitating spondyloarthropathy which predominantly affects the spine and pelvis. The disease initially presents as inflammation followed by osteoproliferation that can result in axial joint fusion and bone formation at inflamed entheses [
1]. In many countries, first-line treatment recommendations involve the use of non-steroidal anti-inflammatory drugs (NSAIDs) and physiotherapy. Use of TNF-inhibitors is restricted to patients fulfilling the modified New York classification criteria for AS (i.e. presence of osteoproliferative changes in the sacroiliac joints required) [
2] and who have failed NSAID therapy [
3]. Both treatment approaches are effective at relieving inflammatory symptoms and suppressing objective measures of joint inflammation. However, it is not clear whether anti-inflammatory treatments retard progression of syndesmophyte formation, indicative of disease-associated osteoproliferation.
Recent reports have suggested that anti-TNF treatment can retard late stage disease [
4,
5]. However, delayed diagnosis [
6] and/or delayed treatment delivery [
4], which frequently occur in AS, are associated with higher radiographic progression rates. Additionally, presence of syndesmophytes at anti-TNF treatment commencement is prognostic of radiographic progression [
7] indicating that anti-TNF treatment may be less effective if structural damage is established. Hence, a “window of opportunity” [
8] might exist early in AS during which optimal long-term benefit will be gained from initiating TNF-inhibitors or alternative anti-inflammatory therapies.
There is conflicting evidence regarding osteoproliferation dependence on inflammatory mechanism in AS. Anti-TNF treatment in the human leukocyte antigen-B27/human β2-microglobulin (HLA-B27/hβ2m) transgenic rat model decreased inflammatory symptoms but failed to inhibit activation of signalling pathways responsible for chondroproliferation [
9]. Similarly, induction of osteophyte formation in the human TNF transgenic mouse arthritis model required the additional hit of Dickkopf-1 blockade [
10]. The most compelling data comes from human magnetic resonance imaging (MRI) studies suggesting that whilst syndesmophytes are more likely to develop at vertebral corners with prior evidence of inflammatory or fatty lesions, many syndesmophytes occur without evidence of such changes [
11]. However, the insufficient sensitivity of MRI to low grade inflammatory changes that may be the precursor of syndesmophyte formation, or the possibility that inflammation at vertebral corners is episodic in nature, limits the specificity confidence of this technical approach [
12]. No superior imaging modalities are available and human histopathological studies are limited by the difficulties associated with accessing involved joints. Therefore, mouse models must be utilized to investigate the relationship between inflammation and osteoproliferation.
The proteoglycan-induced spondylitis (PGISp), a mouse model of AS, is accompanied with peripheral arthritis allowing a visual diagnosis of early inflammatory events [
13]. An advantage of this mouse model is that both inflammation and osteoproliferation are evident in the spine [
14]. Slow progression of structural changes and high heterogeneity between individual animals and between joints within the same individual, while experimentally challenging, are similar to disease progression and characteristics observed in AS patients [
15]. Using this model, we recently reported that the axial disease initiated as a intervertebral joint associated inflammatory response leading to intervertebral disc (IVD) destruction. We postulated that IVD destruction subsequently increased axial mechanical stress, joint damage and ultimately excessive tissue formation, as an aberrant repair response [
16]. Consequently, suppressing inflammation prior to irreparable IVD damage might be an effective therapeutic approach for preventing or ameliorating progression of AS [
16].
The objective of this study was to interrogate whether early and aggressive anti-inflammatory treatment is effective at preventing inflammation and whether this ultimately prevented/reduced progression to osteoproliferation. To achieve this we used an aggressive combination anti-inflammatory therapy that was initiated in the early, predominantly inflammatory phase of disease in the PGISp mice AS model [
16]. The combination anti-inflammatory therapy was supra-clinical dosing of etanercept (ENT, a soluble TNF decoy receptor) plus prednisolone (PRD, a potent glucocorticoid). Supra-clinical doses were utilised to achieve robust and broad spectrum immune-suppression. Impacts of this combination treatment over short- and long-term interventions were assessed using clinical scoring and semi-quantitative histopathological approaches.
Discussion
The PGISp mouse model of AS was used to address whether early intervention of an aggressive anti-inflammatory drug regimen could alter broad aspects of the disease course, particularly retardation of irreversible joint changes. The study was also designed to inform on the dependence between early inflammatory processes and ongoing destructive changes in this AS model [
16]. We observed delayed progression and reduced severity of peripheral disease in the first 2–3 weeks of ETN + PRD treatment. Histological analysis demonstrated that this transient systemic suppressive effect of anti-inflammatory treatment was associated with longer term significant reductions in axial inflammation, number of affected axial joints and vertebral joint destruction as well as declining trends in excessive tissue formation. Notably, the chosen anti-inflammatory treatment also clearly decreased disease activity compared to vehicle even in mice that exhibited a more severe disease course.
In the present study, early intervention with ETN + PRD clearly decreased bone erosion and weakened its association with IVD destruction. These observations provide further support that this destructive disease outcome is a direct result of the inflammatory process [
16]. While IVD destruction was significantly ameliorated by the anti-inflammatory therapy it was not completely abated, and in joints with compromised IVD, progressive excessive tissue formation was observed. This provides support in favour of the hypothesis that IVD destruction-induced structural instability plays an important role in driving osteoproliferative outcomes. Increased intervertebral body mechanical forces as a consequence of IVD damage/loss may promote local bone formation through differentiation of osteoprogenitor cells as a consequence of elevated mechanosensing signals [
18]. This is in agreement with clinical studies showing that presence of syndesmophytes when treatment commenced strongly predicted radiographic progression regardless of treatment [
7].
The PGISp model recapitulates many key features of AS, including variability in disease onset and heterogeneity in disease severity both between and within individuals [
15,
16]; while this variability makes it a good AS model, it confounded the current study outcomes by reducing statistical power. Peripheral disease is not an accurate predictor of axial disease and consequently cannot be used to reliably inform appropriate treatment initiation. In the absence of a sufficiently sensitive in vivo longitudinal imaging approach to detect spinal inflammation, therapy initiation could not be uniformly initiated at a defined and consistent threshold of axial disease development. Therefore treatment initiation was based on a fixed time post-priming that was selected based on our previously reported axial disease time course data [
16]. Study outcomes were further limited by the fact that the length of study, which was dictated by anti-inflammatory therapy delivery limitations, did not extend to permit peak attainment of osteoproliferative tissue formation in the vehicle control group [
16]. Consequently the current study was underpowered for robust assessment of some disease features. Nevertheless, the data provide clear evidence of beneficial impact of an early and aggressive anti-inflammatory intervention across a broad range of AS-like disease characteristics. Extended time course studies in larger cohorts will be required to definitively dissect the effects of early anti-inflammatory interventions on late-stage disease outcomes.
The results of the current study provide further support to the concept that inflammation and osteoproliferation in AS are dependent but sequential events. Accordingly the results advocate that early intervention with effective anti-inflammatory treatment in AS has a high potential to prevent both primary joint destruction and secondary osteoproliferative responses. However this study does not preclude usefulness of anti-inflammatory treatments in established disease. Specifically, aggressive anti-inflammatory therapy did ameliorate disease in mice that had a more progressed and severe disease course.
Acknowledgements
We acknowledge QFAB Bioinformatics for providing biostatistical advice and Pfizer for providing Etanercept.