TNFα-inhibitors for ankylosing spondylitis and non-radiographic axial spondyloarthritis

Introduction

Spondyloarthropathy (SpA) refers to a group of rheumatic disorders — ankylosing spondylitis (AS), undifferentiated SpA, psoriatic arthritis, arthritis related to inflammatory bowel disease, reactive arthritis, a juvenile form and SAPHO (synovitis, acne, pustulosis, hyperostosis and osteitis) — sharing common clinical features, extra-articular manifestations and a genetic association with the type 1 major histocompatibility complex HLA-B27^{[1]
,[2]
,[3]}
. Amor et al. and the European Spondyloarthropathy Study Group (ESSG) in the 1990s and, more recently, the Assessment of SpondyloArthritis International Society (ASAS), estabilished a classification criteria that culminated in classifying two large SpA subtypes, axial (axSpA) and peripheral SpA (phSpA)^{[4],[5],[6]
,[7]}
.

SpA subtype classification and diagnosis

axSpA is characterised by inflammation of the sacroiliac joints and spine^[6],[8]
, and patients with axSpA typically present with chronic back pain and spinal stiffness, as well as a reduction in mobility and quality of life (QoL). Over time, permanent damage to spinal mobility and function can occur owing to new bone formation in the spine^[9]
. According to current evidence, active inflammation in the sacroiliac joints and the spine is followed by bone repair, which leads to excessive new bone formation — a morphological substrate of structural damage in axSpA^[10]
. New bone formation/ankylosis in the sacroiliac joints has a diagnostic value without a signiï¬cant impact on the patient’s functional status, whereas new bone formation in the spine (syndesmophytes, bridging syndesmophytes and ankylosis) has a clear correlation with reduction of spinal mobility and functional status in patients with axSpA^[11]
.

SpA are categorised into axSpA and phSpA on the basis of primary symptom localisation. According to the most recent ASAS definition, axSpA predominantly includes non-radiographic axSpA (nr-axSpA), radiographic axSpA and ankylosing spondylitis (AS), which is regarded as the most commonly known SpA type. Whereas phSpA predominantly includes psoriatic arthritis, reactive arthritis, enteropathic arthritis and undifferentiated SpA^[12]
.

Several criteria are available for diagnosis of SpA, the earliest of which was the modified New York (mNY) criteria for AS (1984) that combined clinical and radiological criteria. In particular this classification distinguishes sacroiliitis using plain radiographs: grade 0: normal; grade I: some blurring of the joint margins — suspicious; grade II: minimal sclerosis with some erosion; grade III: definite sclerosis on both sides of the joint or severe erosions with widening of the joint space with or without ankylosis; and grade IV: complete ankyloses^[13]
.

The Amor criteria for SpA (1990) and ESSG criteria (1991) were designed to capture the entire spectrum of SpA earlier in its disease course; this criteria has sensitivity and specificity of 87% and 90%, respectively, but their more general nature means they are unable to specifically identify early axSpA^[14]
.

The introduction of tumour necrosis factor (TNF) inhibitors as an effective treatment resulted in a new classiï¬cation criteria being proposed by ASAS that aimed to capture both early and established disease^[6]
. This highlighted the concept of axSpA that includes two groups: patients with mNY criteria for AS, and patients who do not meet the mNY criteria but have classiï¬able nr-axSpA (where definite structural changes are not evident in the sacroiliac joints on plain radiographs). When the ASAS axSpA criteria are compared with physician gold standard, they have a sensitivity and speciï¬city of 83% and 84%, respectively. The imaging arm alone has a sensitivity of 66% and a speciï¬city of 97%. This demonstrates that positive imaging with plain radiograph or magnetic resonance imaging (MRI) has high speciï¬city but lacks sensitivity when assessed against physician gold standard^[15]
.

Several biomarkers have demonstrated promise in differentiating axSpA from nr-axSpA, and identifying poor prognosis in patients with axSpA. These biomarkers include vascular endothelial growth factor, matrix metalloprotein 3, sclerostin, citrullinated vimentin, dikkopf-1 and antibodies to MHC class II-associated invariant chain^[16]
.

MRI enables the diagnosis of axSpA at an early stage of disease and in patients who do not have erosive disease^{[17],[18],[19]}
. Research has demonstrated that T1 and short tau inversion recovery (STIR) sequences are the best indicators of axSpA and that gadolinium contrast does not add anything to the diagnosis^[20],[21]
.

Deï¬nitions of a positive sacroiliac joint scan have been proposed as either two discrete STIR lesions on the same slice or one STIR lesion that is observed on more than one slice^[22]
. A positive spinal MRI is currently deï¬ned as three or more corner inflammatory lesions (osteitis), with each lesion being present on at least two slices^[23]
. Finally, research examining the value of scanning the entire spine, in addition to the sacroiliac joints for the diagnosis of nr-axSpA, indicates that there is little additional value in imaging the spine^[24]
.

Current goals for SpA treatment are: reduction/remission of inflammatory signs and symptoms; improvement of functional performance; prevention, slowing or arrest of disease progression; avoidance of toxicities; and minimisation of co-morbidities^[25]
.

Existing treatments for SpA

The first-line drug treatment of AS and other SpA are based on non-steroidal anti-inflammatory drugs (NSAIDs), selective inhibitors of COX-2 (COXIBs) and physical therapy, which have demonstrated efficacy, especially in the control of spinal symptoms, pain and stiffness.

Continuous treatment with NSAIDs can be considered for patients with active disease; research suggests that a prolonged treatment (two years) with NSAIDs can retard radiographic progression in AS. In nr-axSpA, however, this retardation appears to be less evident; maybe only for a low grade of new bone formation in the spine at the earliest stage of the disease^[26],[27]
.

There is only a little evidence of the efficacy of disease-modifying anti-rheumatic drugs (DMARDs) in SpA. In controlled studies, sulfasalazine, methotrexate and leflunomide have demonstrated modest efficacy on the peripheral manifestations of AS, but their use is not included among the therapeutic alternatives in patients with axial manifestations refractory to NSAIDs^{[26],[28],[29]}
.

Corticosteroids (CS) are frequently used to treat patients who have a poor response to NSAIDs or DMARDs, especially for peripheral localisations of PsA^[30],[31]
. However, the long-term use of systemic CS is not recommended in axSpA because there is nearly no evidence of their efï¬cacy^[32]
. However, in a small open-label clinical trial, a short-term treatment with intravenous pulse of methylprednisolone was successful in reducing symptoms and improving spinal mobility in active AS^[33]
. Intra-articular and soft-tissue CS injections are widely recommended for monoarthritis and oligoarthritis^[34]
.

The ASAS recommends that patients fulï¬lling the ASAS classiï¬cation criteria for axSpA (including patients fulï¬lling the mNY criteria for AS), whose disease activity remains high despite the use of NSAIDs, are candidates for TNFα blocking therapy^[35]
. In these patients, the clinical benefit of TNFα blocking therapy has been demonstrated in various clinical trials^[36]
. Five different TNFα-inhibitors (i.e. etanercept, infliximab, adalimumab, golimumab and certolizumab pegol [CZP]) differ in their molecular structure and administration route^{[37],[38],[39]}
. Only infliximab is administered intravenously and has an induction regimen at zero, two and six weeks, followed by a maintenance regimen for every six to eight weeks. The other TNFα-inhibitors are administered subcutaneously^[40],[41]
.

The aim of this review is to summarise the current evidence for the treatment of AS and nr-axSpA with TNFα-inhibitors.

Sources and selection criteria

PubMed (National Library of Medicine) was the main electronic database that was searched, using the search terms ‘ankylosing spondylitis’, in combination with ‘treatment’ and/or ‘TNF alpha inhibitors’. Similar separate searches were made with ‘non-radiographic axial spondyloarthritis’ in combination with ‘treatment’ and/or ‘TNF alpha inhibitors’ to ensure that no articles were missed. Searches were limited to links to full-text only, humans, English language articles, males and females, and all adult ages. The ‘Related Articles’ function of PubMed was used to cross-check for additional relevant studies. There was no restriction on the age of the literature, but the studies generated were primarily from the 1980s to the present. Titles and abstracts identified in the broad search were examined, and the studies included in the review if they were directly relevant to treatment with TNFα-inhibitors and one of the ‘trigger’ areas. Retrieved articles were also reviewed for relevant citations. Research studies published only in abstract form were excluded. Finally, each article was reviewed for quality and clinical relevance.

TNFα-inhibitors for the treatment of AS

AS is a chronic, progressive inflammatory disease that primarily affects the spine and sacroiliac joints. The onset of AS is typically in the third decade of life. AS has a standardised prevalence rate of 0.55% among white patients^[42]
, and it is frequently associated with HLA-B27 positivity, until 90%^[43]
. Patients with AS can experience significant long-term functional impairment and disability, with reduced QoL and an increased risk of comorbid conditions^[44]
. An extensive report about the sustained durability and tolerability of etanercept demonstrated that patients who switched from placebo had an improved Bath Ankylosing Spondylitis Functional Index (BASFI) and all levels of ASAS response (20%, 50% and 70%). After receiving etanercept for 24 weeks, 70% of patients achieved an ASAS 20 response, and after 96 weeks almost half of these patients (46%) had attained an ASAS 70 response. In addition, at the end of 72 weeks of etanercept treatment, all mean measures of spinal mobility (e.g. chest expansion, occiput to wall measurement and modified Schober’s index) improved in this group^[45]
.

ATLAS (the Adalimumab Trial Evaluating Long-Term Efficacy and Safety for AS) is a long-term study of a TNF antagonist for the treatment of active AS, with 255 patients continuing adalimumab treatment during a two-year analysis. During this time, the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) 50, ASAS 20, ASAS 40, ASAS 5/6 and ASAS partial remission responses, which were achieved during the initial 24-week, double-blind period of ATLAS, were sustained for up to two years of treatment. Improvements in individual components of the ASAS response (e.g. the patient’s global assessment of disease activity, pain, function and inflammation) were also sustained for up to two years of adalimumab treatment. The treatment demonstrated sustained improvements in physical function and mobility during long-term treatment, with a sustained improvement in Bath Ankylosing Spondylitis Metrology Index (BASMI) in 75% of patients^[46]
.

In 2015, van der Heijde et al. published data for a five-year follow-up of spinal mobility in patients treated with adalimumab. Using a composite BASMI_lin score, patients reported improvements in lumbar side flexion, intermalleolar distance and cervical rotation components. The analysis demonstrated that the progress in physical function was mantained over five years, however, progression of structural damage (as determinated by radiographs) does not appear to be inhibited by the treatment^[47]
. The significance of radiographic damage in AS has recently been studied and reported that anti-TNFα does not seem to have a significant effect on radiographic progression in these patients^[47]
. Baraliakos et al. reported that patients with a slight increase in BASFI scores after eight years of infliximab therapy had more radiographic damage^[48]
. However, in 2014, the same authors reported that long-term treatment (over six to eight years) might be associated with reduced radiographic spinal progression^[49]
.

In 2015, a retrospective study that evaluated the effectiveness of infliximab in AS patients over a five-year treatment period found a rapid, sustained decrease in disease activity within six months of treatment onset. The same authors demonstrated an opposite trend in the non-responders: after an initial improvement, the BASFI and BASDAI began to rise, while the ESR and Câ€reactive protein (CRP) values remained at higher, but stable, levels when compared with the responders group. As reported by other studies, these data support the hypothesis that disease activity indices and inflammatory markers could be predictors of a patient’s clinical response to anti-TNFα treatment and could be useful when patients with a negative response are being evaluated for discontinuation. In addition, it has been suggested that female gender is a negative predictor of drug response and disease outcome. Other potential negative predictors were the global disease activity score and elevate steroid intake (> 5mg/day)^[50],[51]
.

These results were confirmed by the ATTRA registry, which found a longer adherence to anti-TNFα therapy owing to clinical improvement and a prolonged, satisfactory response, in particular to infliximab, in AS compared with RA patients^[52]
. Intensive rehabilitation seems to have a synergistic effect in AS when it is associated with TNFα, producing positive effects on pain, function, QoL and disability^[53],[54]
. In several reports, TNFα-inhibitors were discontinued because of the development of antidrug antibodies, which occurs more often with infliximab than with adalimumab. This difference can be explained by the structures of these inhibitors: infliximab, a chimeric monoclonal antibody, has a murine component and, therefore, a greater likelihood of inducing antidrug antibodies than adalimumab, which is a humanised monoclonal antibody^[55]
. Antidrug antibodies neutralise TNFα-inhibitors, therefore, their presence is associated with lower trough concentrations of these drugs^[56]
and, consequently, with a diminished clinical response^[57]
.

This would account for the more frequent switching from infliximab to adalimumab, followed by etanercept to adalimumab. The use of adalimumab as the first TNFα-inhibitor was less likely to lead to switching, whereas complete ankylosis evident on radiographs of the sacroiliac joint was more likely to lead to switching^[58]
.

In the BIOBADASER registry, the use of adalimumab was a protective factor for drug discontinuation related to an adverse event in patients with RA and SpA^[59]
.

TNFα-inhibitors for nr-axSpA

The distinction between AS and nr-axSpA has recently gained attention because TNFα-inhibitors approved for AS are effective in treating symptoms of axSpA in patients without any evidence of related sacroiliitis^[60]
.

The SPEED-2 (Spondyloarthropathy Epidemiology and Burden) study, conducted in the United States, demonstrated that a small proportion of patients diagnosed with nr-axSpA (28%) progressed to AS over 11 years^[61]
.

In multivariate analyses, predictors of radiographic progression were smoking, the presence of HLA-B27 and inflammatory changes on MRI^[62]
. In 2012 the European Medicines Agency (EMA) recommended that adalimumab^[63]
, etanercept^[64]
, CZP^[65]
and golimumab should be approved for treatment of nr-axSpA in addition to AS^[66]
. It has been demonstrated that patients with nr-axSpA and AS respond similarly to TNFα-inhibitor treatment. Despite baseline characteristics indicating more longstanding and severe disease in the AS group, no between-group differences were observed regarding clinical developments or anti-TNF adherence. Compared with AS patients, those with nr-axSpA were significantly younger, had a shorter mean disease duration, and fewer previous and ongoing conventional DMARDs (cDMARDs). Furthermore, patients with nr-axSpA were more often female and had less elevation of acute phase reactants at baseline, with CRP levels remaining significantly lower than in the AS group throughout follow-up. Male sex and higher baseline CRP levels were significantly associated with better treatment adherence and could predict the development of radiographic changes^[67]
. The same results were observed in other studies^[68],[69]
.

In the randomised, placebo-controlled trial ABILITY-1, the efï¬cacy and safety of adalimumab in patients with nr-axSpA was assessed. In the adalimumab group, 36% of patients achieved ASAS 40 at week 12 compared with 15% in the placebo group. Signiï¬cant clinical improvements based on other ASAS responses (i.e. ASDAS and BASDAI), as well as improvements in physical function and QoL measures, were also detected at week 12 with adalimumab treatment. Inflammation in the spine and sacroiliac joints on MRI, scored using the Spondyloarthritis Research Consortium of Canada (SPARCC) index, signiï¬cantly decreased after 12 weeks of adalimumab treatment. Shorter disease duration (<5 years), younger age (<40 years), elevated baseline CRP or higher SPARCC MRI sacroiliac joint scores were associated with better week 12 responses to adalimumab^[63]
. The same results were achieved in the study of Dougados et al., whose objective was the assessment of the efficacy of etanercept in the treatment of early active NSAID–refractory nr-axSpA. As in ABILITY-1, at 12 weeks the proportion of patients with improved ASAS 40 was significantly higher in the etanercept group (34 of 105 [32%]) than in the placebo group (17 of 108 [16%]). In addition, patients who received etanercept had a greater reduction in MRI-based scores for sacroiliac joint inflammation (46.9% versus 10.9%) and spinal inflammation (45.4% versus 33.4%) compared with placebo-treated patients at week 12. Furthermore, at 24 weeks, it was demonstrated that patients in the placebo group who had switched to etanercept at 12 weeks exhibited improvement similar to patients who had received etanercept for 24 weeks^[64]
. A trial testing the efficacy and safety of CZP in patients with nr-axSpA and patients with AS demonstrated similar rates of response to treatment in both groups^[65]
.

The ESTHER trial demonstrated that, after one year of etanercept treatment, the response rate for the nr-axSpA and AS groups was similar, except for a non-signiï¬cant advantage in favour of the nr-axSpA group in some of the outcome variables (e.g. BASDAI 50 [75% versus 50%], ASAS partial remission [60% versus 40%] or ASDAS clinically important improvement [75% versus 55%])^[70]
.

Patients with an elevated CRP level and HLA-B27-positive/MRI-negative (clinical arm of ASAS criteria) responded to treatment with golimumab slightly better than the group of MRI-positive, while patients with CRP levels below the upper limit of normal from the clinical arm did not benefit from treatment with golimimab. These data indicated that the treatment response to golimumab in patients with nr-axSpA was similar to the response observed in AS patients: in total, 71.1% of patients treated achieved the primary end point, an ASAS 20 response, at week 16^[66]
.

The same results were reported by Paccou and Flipo, who demonstrated that golimumab treatment led to a rapid reduction in the signs and symptoms of nr-axSpA, and this effect was sustained through to week 16. Thus, objective evidence of active inflammation at baseline (e.g. a positive MRI scan and/or an elevated CRP level) could be a good predictor of treatment response to TNFα-inhibitors^[71]
.

RAPID-axSpA was the first, large, randomised controlled trial that demonstrated CZP efficacy in reducing the signs and symptoms of axSpA in AS and nr-axSpA patients. The primary end point was met with significantly more CZP-treated patients (57.7–63.6%) achieving ASAS 20 response at week 12 versus placebo (38.3%). Similar efficacy was demonstrated in AS and nr-axSpA subpopulations. The onset of CZP action was rapid: improved response rates for ASAS20, ASAS40 and ASDAS were observed as early as week 1 and the effects were still apparent at week 24^[65]
.

An Italian multicentre study investigated partial remission (PR) in patients with nr-axSpA versus AS treated with TNFα antagonists in a real clinical practice setting. A total of 321 patients with axSpA were treated, 62 of whom were nr-axSpA, while the remaining 259 were AS. A log-rank test to compare survival curves demonstrated that the probability of obtaining PR in nr-axSpA and AS during treatment with anti-TNFα was not significantly different. At 12 weeks of exposure to the first anti-TNFα drug, PR was achieved in seven patients with nr-axSpA (11.3%) and in 68 patients with AS (26.2%)^[72]
.

It is much more important to distinguish between early and established disease based on the duration of symptoms, rather than on the absence or presence of structural changes in the sacroiliac joints. After two years of inflammatory back pain, around 20% of patients with axSpA already have radiographic change in the sacroiliac joints and progress to develop AS^[73]
,[74]
. These data strengthen the hypothesis that the two entities represent different aspects/phases of a common pathology^{[75]
,[76],[77]}
.

In a recent meta-analysis, it was demonstrated that in AS and nr-axSpA patients, anti-TNFα drugs produced a significant reduction in disease activity and improvement in function. For AS, the relative risks for ASAS 40 ranged from 2.53 to 3.42. The efficacy estimates were consistently slightly smaller for nr-axSpA than for AS. In AS, anti-TNFα drugs are approximately equally effective and the effectiveness appears to be maintained over time, with around 50% of patients still responding at two years. Sequential treatment with anti-TNFα may be worthwhile, but the drug survival response rates and benefits are reduced with second and third anti-TNFα drugs^[78]
.

Conclusion

AS and nr-axSpA treatment aims to reduce disease activity, improve QoL and preserve the functional abilities of patients.

First-line therapy is often represented by NSAIDs and low-dosage CS (oral and intra-articular administration). cDMARDs (MTX and SSZ) have shown poor efficacy in the treatment of patients with peripheral joint disease. TNFα-inhibitor treatment is currently the only effective therapy for patients with axial SpA in whom conventional therapy with NSAIDs and cDMARDs has failed. Even though TNFα-inhibitors have proven effective in the treatment of SpA, there is an unmet clinical need for new therapies that have alternative mechanisms of action for these diseases, owing to an increasing number of non-responder patients or patients with contraindications to TNFα-inhibitor therapy. Among the newer therapies studied to date, targeting IL-17, IL-12/IL-23 and PDE4, appear to show more promise than therapies targeting T-cell co-stimulation, B-cell surface antigens and IL-6 (’Table 1: Biological treatment for ankylosing spondilitis and non-radiographic axial spondyloarthritis’).