Targeting inflammatory pathways in axial spondyloarthritis

Several lines of evidence implicate TNF in the pathogenesis of axSpA; however, the exact TNF-associated cellular and molecular mechanisms involved are not well understood [2]. TNF, a 233 amino acid protein, is synthesized as a transmembrane protein (tmTNF), in a trimeric form which can be cleaved by a converting enzyme to a soluble molecule (sTNF). Both act by binding to two receptors, tumor necrosis factor receptor (TNFR) 1 and TNFR2, activating multiple cellular pathways ranging from cell homeostasis and proliferation via nuclear factor (NF)-κB and protective immunity from infections and cell death via the caspases, respectively (Fig. 1) [2]. Additionally, tmTNF can act as a ligand by binding cell-to-cell to TNF receptors and as a receptor that transmits reverse or outside-to-inside signals to induce local inflammation within the tmTNF-bearing cells. [43].

Key TNF signaling pathways include TNFR1 or TNFR superfamily type 1A (TNFRSF1A), TNFR1-associated death domain (TRADD), and TNFSF15 [1, 2, 44]. TNFR1 signaling is associated with NF-κB–mediated cell survival and growth and/or apoptosis through the TRADD adaptor protein [2]. TNFR2 is predominantly produced by cells of immunologic and endothelial origin, and is also associated with NF-κB activation [2, 45]. Overexpression of TNF has been documented in the sacroiliac joints of patients with AS, and patients with AS have been shown to have high levels of the circulating soluble TNF receptors (sTNF-R1 and sTNF-R2) [46].

Immunologic studies have strongly implicated the IL-23/IL-17 axis in axSpA [2, 9]. Macrophages, dendritic cells, ILCs, and mucosal-associated invariant T cells (MAITs) contribute to elevated IL-23 levels found in the serum, inflamed gut, synovial fluid, entheses, and bone-derived cells, including the marrow, and fibrous tissues taken from the peripheral joints of patients with AS [47‐52]. In addition, IL-17–positive mast cells and neutrophils have been reported in facet joints [53], and IL-22 levels have been shown to be elevated in the gut of patients with AS [54]. Macrophages from patients with AS stimulated by Toll-like receptor agonists such as LPS produced IL-23; and IL-23 enhanced IL-17 production from pathogenic Th17 cells [55].

In response to IL-23, additional cells, including CD4⁺ and CD8⁺ T cells in the microenvironment of other pro-inflammatory cytokines, chemokines, and transcriptional regulators (e.g., signal transducer and activation of transcription [STAT] 3, RAR-related orphan receptor γt [RORγt]), drive Th17 cells to produce IL-17 and other cytokines, including IL-6, IL-22, IL-26, and TNF in a reciprocating continuum [9, 56, 57]. The gut microbiome may also contribute, as serum amyloid A from the terminal ileum induces Th17 cell differentiation from naïve CD4⁺ T lymphocytes, and microbiota-induced IL-1β stimulates development of Th17 lymphocytes in the intestine [58].

Some have proposed that HLA-B27 misfolding and homodimer formation triggers IL-23 and IL-17 production through unfolded protein response and autophagy, further supporting the link between the IL-23/IL-17 pathway and the development of axSpA [2, 59]. This supports the emerging role of a pathogenic IL-23/IL-17 axis in axSpA [55, 60].

The strongest evidence supporting the role of the TNF pathway in the pathophysiology of axSpA comes from the use of TNF inhibitors in the clinical setting. TNF inhibitors are effective in reducing pain and stiffness and improving function in patients with AS [81]. TNF inhibitors reduce inflammation in most patients, but may not provide long-term remission. Recent studies now demonstrate that TNF blockers slow spinal radiographic progression by reducing disease activity (e.g., based on changes in the modified Stoke Ankylosing Spondylitis Spine Score [mSASSS]) [82]. Complete inhibition of radiographic progression is possible, as evidenced by patients who have reached an inactive disease state, defined as ASDAS < 1.3; significant reduction in disease activity is defined as ASDAS < 2.1 [2, 81, 83, 84]. However, a reported 20% to 30% of patients with axSpA do not respond adequately to TNF inhibitors, resulting in the need for other treatment options [85].

For patients who do show an initial response to TNF inhibition, treatment persistence of 5 years has shown sustained efficacy and sustained benefits in spinal mobility, disease activity, physical function, and health-related quality of life [86, 87]. After 1 year of treatment with adalimumab, ASAS20 and ASAS40 responses were achieved by 82% and 62% of patients, respectively, and after 5 years of treatment with adalimumab, ASAS20 and ASAS40 clinical responses were achieved by 89% and 70% of patients, respectively [86]. Furthermore, adalimumab and etanercept increased spine and femoral neck bone mineral density of patients with active AS with low bone mineral density [88]. Interestingly, TNF blockade did not influence the IL-23/Th17 axis, but TNF blockade did significantly reduce the erythrocyte sedimentation rate and CRP levels [89].

TNF inhibitors approved for the treatment of AS include adalimumab, certolizumab pegol, etanercept, infliximab, and golimumab (Table 2). Adalimumab and etanercept are also approved for nr-axSpA in Europe, and certolizumab pegol is approved for nr-axSpA in the US and Europe. Comparative efficacy analyses indicate that all five of the approved TNF inhibitors provide comparable ASAS responses, with similar safety profiles [90]. Notably, the effects of different TNF inhibitors on the extra-articular manifestations of axSpA are not equivalent. For example, infliximab provides better improvement in acute anterior uveitis than etanercept and adalimumab, and etanercept is not an effective treatment for inflammatory bowel disease [85].

Results of clinical trials aimed at blocking the IL-23/IL-17 axis in patients with AS support involvement of this pathway in the pathogenesis of axSpA [2]. As such, IL-17 has emerged as a novel therapeutic target in AS. Secukinumab is the only IL-17A inhibitor currently indicated for the treatment of AS (Table 2) [41, 91]. Secukinumab demonstrated efficacy comparable to the TNF inhibitors clinically in a patient population that included both individuals with and without prior exposure to TNF inhibitors [41, 42]. Secukinumab (10 mg/kg) given intravenously at weeks 0, 2, and 4, then subcutaneously (150 mg or 75 mg) every 4 weeks was associated with ASAS20/40 response rates of 61%/42% compared with 29%/13% with placebo (P < 0.001 for both) at week 16 [42]. When secukinumab was given only subcutaneously, ASAS20/40 was achieved by 61%/36% and by 41%/26% of patients treated with secukinumab 150 mg and 75 mg, respectively, compared with 28%/11% with placebo [42]; responses were maintained at week 52 [92]. The efficacy of secukinumab has also been demonstrated over 2 years in patients who are TNF naïve (ASAS20/40 response rates at week 104 of 77%/56% with secukinumab 150 mg and 80%/60% with secukinumab 75 mg) [92]. Further, an observational study of treatment with secukinumab (2 × 10 mg/kg intravenous loading doses followed by 3 mg/kg intravenously every 4 weeks for 94 weeks) reported low progression of spinal radiographic changes with 87% of the inflammatory vertebral edges and 30% of vertebral edges with fatty lesions at baseline resolved by week 94 [93].

Ixekizumab, an IL-17 inhibitor in an IgG4 formulation, has been approved for psoriasis and psoriatic arthritis, and is under investigation for treatment of AS (NCT02696785). At 16 weeks, ixekizumab 80 mg every 2 or 4 weeks elicited ASAS 20/40 response rates of 69%/52% and 64%/48%, respectively, compared with 40%/18% with placebo (P < 0.01 for all) and 59%/36% with adalimumab. In addition, there were significant reductions in spinal inflammation measured by MRI Spondylitis Research Consortium of Canada (SPARCC) spine score change from baseline [94]. Studies of ixekizumab (NCT02696798) and the IL-17RA inhibitor brodalumab (NCT02985983) are ongoing in patients with axSpA.

The dual IL-17A and IL-17F inhibitor, bimekizumab, has also demonstrated therapeutic potential for the treatment of AS. In a phase 2b study, 12 weeks of treatment was associated with an ASAS40 response rate of up to 47% compared to 13% with placebo (P < 0.001) [95].

Of great interest would be the use of bispecific biologic therapies that inhibit both TNF and IL-17 [96]. Dual inhibition of these cytokines was more effective at suppressing arthritis in a collagen-induced arthritis model than TNF inhibition alone [97].

Ustekinumab, a human IgG1κ monoclonal antibody that binds to the common p40 subunit of IL-12 and IL-23, failed to meet the primary endpoint of a phase 3 study in patients with AS naïve to TNF inhibitors (NCT02437162). At week 24, ASAS40 was achieved by 31% of patients receiving ustekinumab 45 mg, 28% of patients receiving ustekinumab 90 mg, and 28% of patients receiving placebo. Additional phase 3 studies of ustekinumab in AS refractory to TNF inhibitors (NCT02438787) and nr-axSpA (NCT02407223) were terminated based on this result [98].

Similarly, risankizumab, a humanized IgG1 monoclonal antibody specific for the IL-23 p19 subunit, failed to meet the primary endpoint of a phase 2 study in biologic naïve patients with AS [99]. At week 12, ASAS40 response rates were 25% with risankizumab 18 mg, 21% with risankizumab 90 mg, 15% with risankizumab 180 mg, and 18% with placebo. Tildrakizumab, another IL-23 p19 monoclonal antibody, is in development for axSpA (NCT02980705). There are no currently planned studies of the IL-23 p19 monoclonal antibody, guselkumab, in AS.

The negative results from these studies of ustekinumab and risankizumab were surprising and have led to speculation about the differing biologic mechanisms that occur with IL-23 versus IL-17 inhibition. There may be pathogenic differences between inflammation at spinal and peripheral sites, possibly related to different mechanical load and stress responses of ligament and tendon insertions through PGE2 activation of ILC3s and production of IL-17 via an IL-23–independent pathway [100]. Of particular interest, recent studies using single-cell–based technology found that ILC3s were not a significant source of IL-17 in the joints of patients with peripheral SpA [101, 102]. It has also been hypothesized that IL-23 may play a pathogenic role in only certain stages of axSpA (e.g., during initiation but not established disease), suggesting that higher serum concentrations of IL-23 inhibitors may be necessary to have clinically meaningful effects [100].

The Janus kinase (JAK)/STAT pathway is thought to activate the IL-23/IL-17 cytokine axis, and inhibition of the JAK-STAT pathway has been proposed as a therapeutic strategy in AS [103]. In a phase 2 trial, the JAK 1/3 inhibitor tofacitinib demonstrated clinical efficacy in patients with AS [104]. At 12 weeks, tofacitinib 5 mg BID elicited ASAS20/40 response rates of 81%/46%; however, placebo responses were also high (41%/20%). Subanalysis discovered that the best responders were those with high CRP and higher MRI SPARCC scores. Indeed, the improvement in SPARCC sacroiliac joint and spine scores showed a dose response [104].

While serum IL-6 levels have been shown to be elevated in patients with AS, a recently published phase 2/3 study failed to demonstrate clinical benefit with tocilizumab, an IL-6 receptor–targeted monoclonal antibody, in TNF inhibitor-naïve patients with AS [34]. A related phase 3 trial in patients with AS who had an inadequate response to previous TNF inhibitor therapy was subsequently terminated (NCT01209689). A phase 2 study of sarilumab, an IL-6 inhibitor, in TNF inhibitor-naïve patients with active AS also failed to show a clinical benefit [105].

The IL-1 receptor family–specifically IL-1β–is a therapeutic target for several systemic and local auto-inflammatory conditions [106]. Two open-label studies have been conducted to investigate anakinra (an IL-1–receptor antagonist) in patients with AS. A pilot study suggested that anakinra may be effective in controlling AS symptoms; however, another study demonstrated limited improvement in only a small subgroup of patients with AS [33, 107].

A phase 3 randomized, double-blind study of apremilast, a selective inhibitor of PDE4, did not meet its primary endpoint of ASAS20 compared with placebo (32.5% vs 36.6%; P = 0.4383) in 490 patients with active AS [108].

Imaging techniques, such as conventional radiography, bone scintigraphy, MRI, and PDUS, are used for the diagnosis of axSpA, monitoring disease activity, and assessing structural damage. Initially, the New York criteria for diagnosis of AS required radiographic evidence of sacroiliitis [127, 128]. Sacroiliitis appears only after several years of undiagnosed inflammatory back pain symptoms, while MRI has demonstrated that osteitis or bone marrow edema in the sacroiliac joint is present earlier—before it becomes radiographically detectable [127, 129]. Thus, MRI is increasingly being used to detect sacroiliitis early in patients with axSpA, particularly utilizing coronal and axial sequences and high-resolution erosion-specific sequences [127, 130‐132].

MRI is useful for following therapeutic outcomes. Anti-TNF with infliximab therapy decreased spinal and sacroiliac osteitis scores more effectively than nonsteroidal therapies [133]. Similarly, anti-IL-17A therapy with secukinumab over 2 years decreased spinal osteitis scores and fatty lesions, which eventuate to ankyloses [93].

PDUS to identify increased vascular flow is also a highly sensitive and less costly tool for detecting enthesitis, which is not always detectable by clinical examination [134]. However, this method is limited by operator proficiency [125] and its use in sacroiliitis needs confirmation. In vivo probing of TNF for guidance of therapy using 99m Tc-labeled anti-TNF monoclonal antibodies and specific aptamers are proposed [135].

Investigations into the discovery of circulating and tissue-related biomarkers are ongoing. These biomarkers may help accurately diagnose axSpA, predict disease activity/progression, and improve response to therapy. The most frequently used axSpA marker in the clinical setting for diagnosis is HLA-B27 [136]. Erythrocyte sedimentation rate and CRP are not always dependable biomarkers for monitoring disease activity, but if elevated, predict better responses to biologic therapies and more comorbidities. [136]. Other research has focused on examining the prognostic value of cytokines, particularly IL-17 and IL-23, and downstream matrix metalloproteinase (MMP) markers such as MMP-3 (stromelysin-1), which degrades collagen II, III, IV, IX, X, and the extracellular matrix proteins. Markers of bony metabolism have also been investigated, including adipokines and cartilage/connective tissue degradation products [75]. However, no individual prognostic biomarker for disease activity has demonstrated adequate reproducibility, and an unmet need for robust biomarkers remains [75]. Biomarkers from the specific inflamed sites may be more informative and descriptive than from peripheral blood samples.

Various biomarkers have been investigated for their predictive value to treat axSpA. For example, in patients with axial and peripheral SpA, infliximab treatment response is associated with high-sensitivity CRP and calprotectin levels [137], while response to golimumab in patients with AS is associated with various combinations of markers comprising specific biomarker signatures [138]. Furthermore, secukinumab response after 6 weeks of treatment in patients with AS is associated with a decrease in levels of S100A8 and S100A9, which form calprotectin, the calcium-binding protein used as a marker of gut inflammation [41].