EULAR 2018 took place from 13-16thJune in the beautiful city of Amsterdam. Dr Elizabeth Reilly from the Royal Bath Hospitals NHS Trust and junior member of BRITSpA was there and has written an insightful summary of the main abstracts and sessions on axSpA from the meeting.
The first day of EULAR 2018 brought an interesting mix of clinical cases and management challenges during the session ‘Little arthritis in PsA and axSpA with newly onset arthritis –does it matter?’ where Alexandre Sepriano(abstract SP0003 )and Maxime Dougados (SP0004), discussed the importance of extra spinal manifestation in axSpA. Professor Dougados categorised these manifestations (EAMs) into 3 groups:
- extra spinal rheumatic features (enthesitis, arthritis, dactylitis),
- extra rheumatic manifestations (uveitis, psoriasis, inflammatory bowel disease), and
- concomitant widespread pain syndromes such as fibromyalgia.
And summarised data from the DESIR and ASAS-COMOSpA cohorts where 26% of patients had a peripheral EAM beforethe onset of axial symptoms (20% enthesitis, with Achilles or plantar fasciitis being the commonest sites). He warned about presence of concomitant Fibromyalgia, as one of the main differential diagnosis to consider in the presence of entheseal pain and symptoms, which he described as challenging and with significant ramifications for subsequent management.
Loukianos Couvaras (OP0022) reported on whether the response to NSAIDs can act as a predictor for subsequent response to TNF inhibition (TNFi). Exploring data from the DESIR cohort, lack of NSAID response was defined as no response at 48 or rapid recurrence of symptoms on discontinuation. All patients had IBP <3 years and were TNF naïve at baseline, and follow up was 5 years. Of 236 patients, responders to NSAIDs were more likely to be HLA B27 positive, have lower BASDAI at baseline, less likely to have psoriasis and had higher NSAID scores. Using multivariate analysis and propensity scoring, no relationship was shown between response to NSAIDs and subsequent TNFi response however.
Ulf Lindstrom (OP0023) presented his work evaluating EAMs (uveitis, psoriasis, IBD) on drug survival for TNFi, from Swedish register data of 2577 biologic naïve AS patients between 2006-2015. Co morbid uveitis was found to be associated with better TNFi drug survival but psoriasis with reduced drug survival, although this effect was reduced once confounders were adjusted for. When reviewing persistence with sequential TNFi, drug retention for the 1stand 2ndagents were similar, but poorer for the 3rd.
Valerie Rodriguez (OP0025) discussed her work on predictors of structural progression at the sacroiliac joints in early axSpA on longterm TNFi, as most work so far on structural progression has centred on spinal changes. 55 patients were evaluated over 6 years, including 36 with nr-axSpA. Patients were randomised to Etanercept or Sulphasalazine for one year. Baseline, 2 and 4 year SIJ xrays and MRI were carried out. Using 4 definitions of progressive imaging features and longitudinal mixed model analysis, raised CRP and the presence of osteitis were predictors for progression at the SIJ at 6 years. Maximal radiographic progression was seen in the first 2 years of follow up, but decelerated between years 2 – 4, and no progression was seen after year 4.
Philip Mease (OP0026) outlined whether trough serum Adalimumab levels could predict sustained remission or the absence of flares in nr-axSpA patients, from the ABILITY-3 study. Patients had a diagnosis of nr-axSpA, satisfied ASAS classification criteria, had raised CRP and MRI evidence of inflammation at either the spine or SIJ, and achieved criteria for remission by week 28 (n=305). All patients achieved steady trough level concentrations on Adalimumab treatment. Receiver operator curves for serum trough levels (<0.6) however were not consistent with prediction of either sustained remission at week 28 or the absence of flares at week 68, contrasting to that seen within the Gastroenterology literature interestingly.
Desirée van der Heijde (LB0001) reported on the results of the BE AGILE study evaluating Bimekizumab (dual IL17-A and F neutralisation, inhibiting neutrophil migration and matrix mineralisation). This double blind study included 303 patients (mNYc, BASDAI and back pain score >4, 11.2% one previous TNFi). 4 doses of Bimekizumab were used (16mg, 64mg, 160mg 320mg every 4 weeks) and placebo. A dose dependant response in ASAS20 and ASAS40 was seen at week 12, with response seen as early as week 1. ASDAS, BASDAI and BASFI were also seen to improve in all treatment groups even by week 1. The most common adverse events were nasopharyngitis and headache, but there was no difference in serious AE across the groups. One unrelated death occurred in the treatment group. A subgroup analysis of MRI outcomes from this cohort is planned.
On the second day of conference, Desirée van der Heijde (SP0043) reviewed radiographic scoring systems in SpA utilising data from the OASIS cohort. Although the BASRI performed well for the cervical spine, the lumbar spine was not as well assessed, although the converse was seen for the SASSS. Overall, the mSASSS was the most reliable and inclusive scoring system. Specifically, the mSASSS identified more patients with progression over 1 year. Of note, 10 unit change in the mSASSS was associated with a reduction in BASMI, when adjusted for BASDAI. The RASSS does include a more comprehensive assessment of the thoracic spine, but had greater variability in scoring. The mSASSS again performed best in the DESIR cohort (at 2 and 5 years).
Of interest, the CT syndesmophyte score (De Bruin et al. ARD 2018;77:371-7) was outlined. This uses 3D low dose (4mSv) whole spine scanning, and scores bone proliferation using a 0-3 system based upon mSASSS (0 – no changes, 1 – syndesmophytes <50% intervertebral disc height (IVDH), 2 – syndesmophytes >50% IVDH but non bridging, 4 – bridging syndesmophytes). This method is sensitive to change over 2 years, but the clinical relevance of a change in score is as yet unknown.
Xenophon Baraliakos (SP0044) nicely summarised what we learn from RCTs regarding treatment effect on structural progression in axSpA. Although we know that bDMARDs have good effect on inflammation, there has been little to evidence their modulatory effect on radiographic progression. This may be due to historic studies’ follow-up being too short to identify radiographic change (<2 years), lack of controls, and ethical concerns with regard to withholding bDMARDs in active SpA. In addition, scoring methods only capture a limited assessment of the radiographic process. Suggestions included longer duration studies, the role of head to head studies (H2H), and the use of newer imaging and scoring techniques (low dose CT, or more sophisticated image analysis techniques).
Robert Landewé (14/6/18, 2.15-2.35pm, Hall 3) then followed with an insightful summary of the limitations of observational studies to demonstrate effect on disease progression. Progression signal in trials is low and yet the measurement error high, so power and precision are poor. In addition there is right censorship – physically more able patients are more likely to persist through the study period and less well patients are more likely to drop out. These ‘well’ patients are less likely to show such disease progression and therefore give a falsely positive view of radiographic progression.
In addition, patient selection and treatment intensity, as chosen by the physician for the more severe patients, introduced confounders. Cases of SpA are less severe than in the past, meaning that there is a relative scarcity of ‘severe’ cases for study inclusion. This makes historic study comparisons impossible. Further, we make indirect comparisons between treatments due to the lack of H2H trials, and use indirect analyses (network meta analyses, Bayesian approaches). Therefore our view of the utility of an intervention and outcomes is coloured by the beliefs of the stakeholders. Propensity modelling, baseline adjustment for intermediate variables, and longitudinal data analysis do help to adjust for this.
Lianne Gensler (OP0198) then eloquently set out to demonstrate the utility of observational studies by presenting her work exploring the effect of TNFi and NSAID usage (type and dose) on radiographic progression. 519 study participants were reviewed every 6 months, with xrays every 2 years and mSASSS scoring at years 2 and 4. TNFi users and non TNFi users were stratified for NSAID usage (none/ low / high doses), and then by their use of Celecoxib vs other NSAIDs. Causal inference modelling (longitudinal targeted maximum likelihood estimation) accounted for time varying confounders and censoring. Combined treatment of TNFi and NSAID was seen to slow radiographic progression in AS, with additional dose related effects at higher doses of NSAIDs at 4 years. Greatest effect was seen with Celecoxib over other NSAIDs at years 2 and 4.
Maxime Dougados (OP0199) presented evidence towards a Treat to Target strategy, evaluating the effect of sustained remission on structural damage at the SIJs. This post hoc analysis comprised 161 patients treated for 2 years with Etanercept (EMBARK cohort), and 76 controls not on TNFi (DESIR cohort), comparing those in sustained clinical remission to those who were not. Baseline and 104 week T1W MRI of SIJ were scored using SPARCC. Those with an ASDAS<1.3 has decreased erosions compared to those with ASDAS>1.3, in both cohorts. A corresponding increase in backfill was noted in the same subset of both cohorts, suggested as being a regenerative process although further work is required to fully understand the meaning of this imaging feature.
Janneke de Winter (OP0244) presented findings from MRI SIJ in chronic back pain patients without axSpa (CBP, n=47), healthy controls (HC, n=47), known axSpA patients (n=47 from the SPACE cohort), 7 women with post partum back pain and 24 frequent runners. MRIs were scored using SPARCC. 91.5% of axSpA, 23.4% of HC and 6.4% of CBP had positive MRIs according to ASAS. In addition, 12.5% and 57.1% respectively of the post partum back pain patients and runners had a positive MRI. However, ‘deep’ bone marrow oedema (BMO) lesions were only found in axSpA patients (80.9%) and one woman with post partum back pain.
Xenaphon Baraliakos (OP0243) also presented his work from a large population based study of under 45 year olds undergoing MRI spine/SIJ (n=802), scored by two blinded readers. 18% were found to have BMO at the SIJ. Similarly, inflammatory spinal lesions were seen, with the thoracic spine being the commonest vertebral segment for these (18.6%). The presence of fatty lesions was higher in all spinal segments, but again maximally in the thoracic spine (72.4%). From logistic regression, age was the only demographic independently contributing to the presence of BMO or fatty lesions. Overall, 86.5% had >1 positive spinal lesion, and 50.2% had >3 spinal lesions. Hence, caution needs to be used when interpreting the results of imaging in the absence of robust clinical features.
Alexandre Sepriano (OP0247)helpfully reviewed the differing referral strategies for possible SpA, from a large, 3 stage, national health survey (EpiReumaPt). From 10661 initial patients screened, 3877 patients were included in the 2ndstage (who had >1 rheumatic complaint, plus 20% negative screening patients for comparison). All of the information was then reviewed by a Rheumatologist, and each referral strategy tested against SpA diagnosis (any imaging or lab component of a referral strategy was omitted). 1.6% (92) of participants received an SpA diagnosis, and 3107 of another rheumatic condition. Overall, the ASAS criteria were the most sensitive of the 11 referral strategies assessed (sens 85.4%, spec 38.8%, PPV 2.2%, 1-NPV 0.6%), with Brandt III being the most specific but least sensitive (sens 7.9%, spec 98.4%, PPV 7.6%, 1-NPV 1.5%).