Classifying and characterizing spondyloarthritis in Crohn’s disease
Even early descriptions of CD recognized the co-occurrence of intestinal inflammation with arthritis. These clinical findings of joint inflammation reflect a spectrum of SpA characterized based on the involvement of peripheral or axial joints. Peripheral spondyloarthritis can be pauciarticular (previously called Type I) or polyarticular (previously called Type II) joint inflammation [
4,
9,
10]. Axial spondyloarthritis, traditionally referred to as ankylosing spondylitis (AS), includes inflammation of the spine and/or sacroiliac joint. Although earlier literature distinguished clinical SpA based on concordance or discordance with intestinal disease activity, histologic and molecular evidence of subclinical intestinal inflammation support a more inclusive spectrum of disease linking gut and systemic immunity [
11,
12]. Reflecting this underlying state of systemic inflammation, spondyloarthritis frequently co-occur with other EIMs including erythema nodosum and uveitis. While this overall description serves as a framework for a spectrum of musculoskeletal symptoms associated with CD, challenges exist in defining both clinical disease burden and underlying pathogenic mechanisms that are needed to guide treatment strategies for CD-SpA.
One challenge in assessing the overall burden of CD-SpA is the lack of uniformity in applying conventional clinical definitions of SpA. Strict characterization of point prevalence of peripheral arthritis in an IBD cohort on physical exam can be very low (< 1%), however inclusion of patient reported self-limited episodes increases the prevalence considerably (up to 12%) [
13]. Including patient reported episodes introduces the risk of reporting non-inflammatory arthropathies in these prevalence estimates. As such, the wide variability in prevalence of IBD associated peripheral arthritis (5–44%) and axial inflammation (1–25%) reflect this lack of uniformity in clinical disease characterization [
3-
5,
13], likely arising from the use of non-standardized disease definitions, not distinguishing non-inflammatory arthralgia from arthritis, and the lack of rheumatologic evaluation. The recent European Crohn’s and Colitis Organization (ECCO) guidelines have also addressed this issue, highlighting a need to ensure accurate identification of true inflammatory arthritis [
10]. Criteria established by the Assessment of SpondyloArthritis international Society (ASAS) defines both axial and peripheral SpA. In the context of CD, ASAS criteria define axial SpA as sacroiliitis on magnetic resonance imaging (MRI) or HLA-B27 positivity with one other SpA feature such as inflammatory back pain [
14]. Similarly in the context of CD, ASAS criteria define peripheral SpA as the presence of peripheral arthritis, or enthesitis or dactylitis [
15] (Table
1). These clinical criteria demonstrate strong sensitivity and specificity for correlation with clinical assessment by rheumatologists [
16], but have been applied variably in IBD cohorts.
Table 1
Classification criteria for CD-SpA adapted from the Assessment of SpondyloArthritis international Society (ASAS) criteria [
14,
15]
Inflammatory back paina in a patient with CD AND Sacroiliitis on imagingb OR HLA B-27 antigen positivity | Arthritis and/or dactylitis and/or enthesitis in a patient with CD AND Exclusion of other specific forms of inflammatory joint disease |
Retrospective analysis of longitudinal follow up studies using ASAS criteria to characterize SpA in IBD cohorts provided estimates of axial SpA (7.7–12.3%) and peripheral SpA (9.7–27.9%) [
17,
18]. These studies serve as a strong basis for validating the use of modified ASAS guidelines in defining CD-SpA in future research.
In addition, there is a significant unmet need for the uniform application of joint disease activity indices in CD-SpA to establish validity, reliability, and responsiveness for clinical evaluation as well as endpoint assessment in research studies. The Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) is a patient-reported tool that has been clinically validated for assessment of inflammatory activity and response to therapy in both axial and peripheral SpA [
19-
21]. Ankylosing Spondylitis Disease Activity (ASDAS) includes patient-reported individual and a global activity score and either c-reactive protein (CRP) or erythrocyte sedimentations rate (ESR) [
22]. While the inclusion of CRP or ESR provides an objective measurement of inflammatory burden, there are limitations to assessments based on the subjective patient-reported symptom scores. Although BASDAI and ASDAS have also been used to assess arthritis activity and response to treatment in IBD-related SpA [
23,
24], these scores do not always correlate with joint inflammatory activity in IBD [
25,
26]. Additionally, these scores are validated mostly in axial disease and while they may similarly provide an accurate measure of peripheral disease, patients with predominantly peripheral SpA may benefit from a more focused evaluation [
27]. Peripheral joint characterization included in more extensive exams including Peripheral SpA Response Criteria (PSpARC40) may more accurately assess response, but the required joint examinations by an expert rheumatologist make the broader use of these instruments in gastroenterology practices less practical [
28]. Finally, MRI has revolutionized assessment of SpA over the last two decades, but no validated criteria have yet been developed to assess disease severity or response to therapy in IBD. Thus, there remains a need for studies to validate these indices in CD-SpA and to correlate with pathogenic biomarkers to help guide therapy.
Treatment paradigms for CD-SpA
The progress reviewed above in establishing clinical criteria and elucidating a mechanistic understanding of the pathogenesis of CD-SpA may allow for rational selection of treatment. Besides the gold standard tumor necrosis factor α (TNF) antagonist therapy for SpA, recent commercialization of other biologic therapies targeting α4β7-mediated trafficking of lymphocytes (vedolizumab) and the p40 subunit of IL12/23 (ustekinumab) may help personalize treatment of CD-SpA. However, since there is little direct evidence to support the efficacy of these agents in co-existing IBD and SpA, current treatment strategies are extrapolated from independent studies in SpA [such as AS (axial) and PsA (primarily peripheral SpA)] and IBD or from limited data from post hoc analysis of therapeutic drug trials in IBD (Table
2).
Table 2
Pharmacologic treatments for CD-SpA
Currently approved therapies |
Short term NSAIDs if CD is in remissiona | Short term NSAIDs if CD is in remissiona |
Anti-TNFα therapy: Infliximab Adalimumab Certolizumab pegol | Local steroid injection for pauciarticular or short-term oral steroid for polyarticular peripheral SpA |
Sulfasalazine |
Methotrexate |
Anti-TNFα therapy: Infliximab, adalimumab, certolizumab pegol |
No IL12/23 inhibitor therapy is proven beneficial in axial SpA | Anti IL12/23: Ustekinumab |
Other therapies under investigation/development |
Selective JAK-1 inhibitor: Upadacitinib and filgotinib (successful phase 2 trials in AS, PsA and CD; positive results in phase 3 trial of upadacitinib in PsA. Ongoing phase 3 trial in CD) |
α4β7 anti-integrin: vedolizumab Approved for treatment of moderate-to-severe CD. Based on recent a systematic review, may be effective in preventing onset of arthritis in CD, however, may not be effective in improving co-existing arthritis |
Therapies under investigation: S1P1 receptor modulator—ozanimod, etrasimod Anti-TNF like cytokine 1A (anti-TL1A) therapy Fecal microbiota transplant Combination biologic therapy (combining two or more biologics with different mechanism of action) |
Non-selective anti-inflammatory drugs (NSAIDs) are the first line of therapy in patients with SpA. While non-selective cyclo-oxygenase (COX) inhibitors are generally contraindicated in patients with IBD due the risk of disease exacerbation [
72], two clinical trials found selective COX-2 inhibitors to be safe and effective in the short-term treatment of inactive IBD [
73,
74]. While routine NSAID therapy is generally avoided in IBD patients due to concern for potential deleterious effects on the gut mucosa [
75], there is growing evidence that COX-2 inhibitors, and specifically celecoxib, are safe and effective.
Sulfasalazine (SAS) is one of the disease modifying drugs effective in treating peripheral arthritis symptoms of IBD, however is not effective in axial-SpA [
76]. Although one of the oldest medicines used for the treatment of CD, the mechanism of SAS’s effect is not clear. SAS is a pro-drug that is cleaved by azo-reductase produced by the colonic microbiota into sulfapyridine and 5-aminosalicylate (5-ASA) released into the distal intestine. It remains unclear if there is a role of one or more gut microbe in this activation process of SAS. While 5-ASA may offer the mechanistic benefit to mucosal healing acting through PPARγ (peroxisome proliferator activated receptor gamma) [
77,
78], sulfapyridine inhibits bacterial folic acid synthesis. The specific target of sulfapyridine’s anti-bacterial effect in the microbiome is not known. In addition to understanding the therapeutic mechanism of its impact, another consideration is variability in dosing. While rheumatologists frequently treat with 2 g daily, the gastroenterology literature supports the use of 4–6 g daily for active symptoms [
79].
TNF antagonist therapy has been a mainstay of therapy both for SpA and CD. The efficacy of TNFα blockade underlies the likely pleiotropic effect of TNFα on many shared pathways. Indeed, the efficacy of anti-TNFα therapy in AS correlates with the reduction of Th17 cells in the peripheral blood following treatment [
34]. Despite this response in the peripheral blood, it is not clear that this reflects tissue activity of TNF antagonists in IBD-associated SpA. Systematic reviews of TNF antagonist therapy support the general efficacy of TNFα blockade in IBD-associated EIMs [
80,
81], including axial and peripheral SpA. However, the interventional studies that reported efficacy of infliximab, adalimumab and certolizumab pegol in the treatment of CD-associated arthritis did not utilize objective criteria for diagnosis (e.g. ASAS) or to define response/ remission of joint disease (e.g. ASDAS) [
80,
82]. While limited study data are available to characterize the impact of joint disease activity [
23,
83], these biologics are considered first line therapy for IBD-associated SpA.
Vedolizumab is approved for the induction and maintenance of remission in UC and CD. Although initially discovered and marketed for gut selectivity, vedolizumab (anti-α4β7) may also impact systemic immunity. One theory is that blocking the homing of α4β7 into the tissue increases the abundance of these effectors in the peripheral blood, thereby enhancing systemic immune activation. In one study, HLA-B27 negative de novo or flare of inactive SpA occurred in 11 patients achieving remission of luminal symptoms with vedolizumab [
84]. Moreover, compared to patients receiving anti-TNF therapies, IBD patients receiving vedolizumab were more likely to develop EIMs including arthropathy [
85]. Despite these findings, post-hoc analysis of the GEMINI trials revealed reduced incidence of arthritis in CD (HR 0.14, 95% CI 0.05–0.35) and no increase in the incidence of arthritis in UC patients treated with vedolizumab versus placebo [
86]. Similarly, in the OBSERV-IBD cohort, extra-intestinal inflammation including joint inflammation improved in parallel with bowel disease, however 13% of patients without EIMs at baseline developed non-inflammatory arthralgia [
87]. More recently, a systematic review of 11 studies where vedolizumab was used for EIMs in IBD, demonstrated no effect on pre-existing arthralgia and arthritis, and lower incidence of new rheumatic symptoms among vedolizumab users compared to placebo [
88]. While there is a signal for benefit of vedolizumab in CD-SpA, a significant limitation of these analyses is the lack of uniform clinical phenotyping of inflammatory arthritis and prospective assignment of disease activity indices, which will need to be addressed in future analyses.
The strong genetic association of both SpA and IBD with IL23R genetic variants and overlapping mechanisms involving the IL23/IL17 axis highlighted the potential role of specific blockade of this pathway in drug development. In addition, elevation of IL-17, but not TNFα, has been reported in CD-SpA compared to active CD alone [
20]. However, the translatability of these findings into therapeutic strategies have so far been puzzling. Despite the efficacy of IL-17A inhibitor therapy, secukinumab, for the treatment of AS [
89], it was not effective in the treatment of CD and may lead to a higher rate of adverse events, including infections [
90,
91]. Recent results in psoriatic arthritis reveal that IL-17A blockade correlated with expansion of
Candida albicans with features of subclinical gut inflammation [
92]. Another potential explanation of this paradox comes from murine models where the suppression of IL-17F, but not of IL-17A, provided protection against colitis by inducing Treg cells through modification of the intestinal microbiota [
93,
94]. It is postulated that low levels of IL-17 can be produced from ILCs, Tγδ, independent of IL-23 [
95]. The homeostatic role for this IL-23 independent IL-17 may reflect the unexpected results of IL-17 blockade in CD.
In contrast to IL-17 blockade, anti-IL12/23 therapy, ustekinumab is effective in the treatment of patients with Crohn’s disease [
96]. Ustekinumab is effective in patients that have failed TNF antagonist therapy suggesting the possibility of a distinct pathway of disease. It is not known if this distinct pathway enriched in patients with CD-SpA. Similarly, selective IL23 antagonist, risankizumab is in development (ongoing phase 3 trials after successful phase 2 trials) for treatment of moderately-to-severely active CD [
97,
98]. Although ustekinumab is approved for psoriatic arthritis and was found to be effective in treating concomitant PsA in IBD patients, suggesting its benefit for CD-related peripheral SpA [
99,
100], both ustekinumab and risankizumab failed to demonstrate efficacy for AS [
101,
102]. Moreover, the switch from TNFα blockade to anti-IL-12/23 blockade has been associated with the unmasking of psoriatic arthritis in a small case series within the psoriasis population [
103]. Further studies are needed to assess the impact of these pathways specifically in patients with CD-SpA and tracked with meaningful disease activity markers.
In addition to biologic therapy, small molecule inhibitors of immune pathway are emerging as therapy for both SpA and CD. In particular, the non-selective Janus kinase (JAK) inhibitor, tofacitinib is approved for the treatment of UC, RA and PsA [
104-
106] and a phase 3 trial in AS is ongoing. While tofacitinib did not meet efficacy endpoints in the phase 2 trials of Crohn’s disease [
107], there was a significant biochemical response and evidence from post-hoc analyses suggest benefit in CD as well [
108,
109]. In initial trials, selective JAK-1 (filgotinib and upadacitinib) and tyrosine kinase (TYK) inhibitors have shown efficacy in Crohn’s disease [
110]. Moreover, both filgotinib and upadacitinib were found to be safe and effective in AS and PsA in their respective phase 2 trials, while upadacitinib met efficacy endpoints in a phase 3 study in PsA [
111-
113]. These highly encouraging results support enthusiasm for specifically evaluating their efficacy in CD-SpA.
Despite similarities in pathogenesis, CD and SpA maintain unique aspects of disease, which may need to be therapeutically targeted independently. While both entities may co-exist in a patient, it is possible that they do not respond to common therapies and therefore strategies for combining therapies need to be evaluated. Anecdotal cases have raised the possibility of combining anti-α4β7 and anti-TNFα blockade [
108,
114]. Therapies directed towards novel mechanistic targets are being studied which also offer great potential. Sphingosine 1-phosphate (S1P) modulation is one such strategy to block T cell egress from lymph nodes into the circulation. In contrast to α4β7 blockade, this strategy may be able to block both tissue and systemic inflammatory symptoms that depend on circulating lymphocytes [
115]. In addition, anti-TL1A therapy is in development for treatment of UC. Given the potential role for TL1A in joint inflammation discussed above [
46], it remains to be evaluated whether this mechanism would be effective in SpA as well. Finally, fecal microbiota transplant (FMT) is under active investigation for the treatment of IBD with preliminary success in the treatment of UC. The impact of FMT on inflammatory arthritis is unknown. Given the emerging knowledge of the potential role of strain specific pathobionts in IBD-associated SpA, microbial-based therapies may play an important role in modifying clinical outcomes.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.