This Week in Rheumatology — Sep 10, 2026

Welcome to This Week in Rheumatology. This week we're covering 10 notable papers spanning three broad themes: how we define and manage so-called difficult-to-manage disease in spondyloarthritis and psoriatic arthritis, new technology and triage strategies reshaping how patients reach us and how we measure their disease, and a cluster of studies on treating the sickest patients — refractory Behçet's, macrophage activation syndrome, scleroderma-related pulmonary hypertension, and paediatric vasculitis. Let's dive in.

We start with vaccination, because it remains the single most modifiable risk in our immunosuppressed patients. In Arthritis and Rheumatology, Jackson and colleagues analysed National Clinical Cohort Collaborative data on nearly 61,000 adults with autoimmune rheumatic conditions who developed incident COVID-19 between December 2020 and August 2024 [1]. Only about one in ten patients had received the updated 2023 to 2024 vaccine. Compared with no vaccination at all, the initial series cut the adjusted odds of COVID-related hospitalisation by roughly 40 percent, while a booster or an updated vaccine cut those odds by about seventy percent. Crucially, moving from the initial series alone to a booster or updated dose translated into an absolute reduction in hospitalisation risk of just under six percent — a number worth quoting directly to a hesitant patient in clinic. The message for practice is simple: staying current, not merely being vaccinated once, is what protects this population.

The second theme is the newly fashionable concept of difficult-to-manage disease, and two papers this week test how well the EULAR and ASAS frameworks survive contact with real-world data. In Annals of the Rheumatic Diseases, Nielung and colleagues pooled 5 Nordic biologics registries covering more than 14,000 patients with psoriatic arthritis starting their first biologic or targeted synthetic DMARD [2]. Discontinuing at least two such drugs was common — it happened in around a third of patients over a median follow-up of just over 6 years. But once you layer on the full framework, requiring problematic symptoms and objectively persistent disease activity, only about 2 percent of patients qualified as difficult-to-manage, and just over 1 percent as genuinely treatment-refractory. The patients who churned through the most drugs were more often women, and had more depression and more opioid use. In RMD Open, Fakih and colleagues asked the same question in axial spondyloarthritis using the French DESIR inception cohort, following 177 patients exposed to at least one biologic over ten years [3]. Cumulative incidence of difficult-to-manage disease was around 7 percent by the strict ASAS definition and about 14 percent by a broader definition, while true treatment refractoriness was rare — under one percent strictly defined. And here is the clinically important part: the difficult-to-manage group was not distinguished by structural progression. What set them apart was fibromyalgia, present in about 60 percent of them versus roughly a quarter of the others, nearly double the number of medical visits, and universal opioid use. Read together, these two papers from different journals and different diseases converge on the same conclusion. When your patient has failed multiple biologics, true refractory inflammation is uncommon; central pain, mood, and psychosocial burden are far more often the driver. The action point is to stop cycling mechanisms of action reflexively and start assessing for fibromyalgia, depression, and opioid dependence.

The third theme is measurement and access — how patients get to us and how we quantify what we find. Also in Annals of the Rheumatic Diseases, Redeker and colleagues prospectively triaged 1,180 consecutive referrals to a tertiary centre using a telephone interview followed by a brief 10-minute rheumatologist consultation, then sent essentially everyone on for comprehensive assessment so the triage could be scored honestly [4]. Inflammatory disease was ultimately diagnosed in about a third of those assessed. The short consultation picked up around two thirds of the true inflammatory cases, with specificity of about 75 percent, and it did meaningfully compress waiting times — patients with inflammatory disease reached final assessment in about 33 days versus 55 days for those without. Machine learning models applied retrospectively performed modestly, with areas under the curve of 0.73 and 0.78. The honest interpretation is that triage accelerates care for most patients with inflammatory disease but misses roughly one in three, so a negative triage cannot be treated as a rule-out. On the imaging side, Jamaludin and colleagues, in the same journal, trained a fully automated deep learning pipeline on sacroiliac joint magnetic resonance imaging from the MEASURE 1 trial and validated it externally in the PREVENT and SURPASS trials, together nearly a thousand patients [5]. Detection of structural lesions was strong, with ankylosis reaching areas under the curve of 0.97 to 0.99, and erosions and fat lesions also robust. Bone marrow oedema was the weaker spot — good discrimination but balanced accuracy of only around 0.72. Performance held across datasets without retraining, and overall matched expert inter-reader agreement, which positions this as a trial and cohort tool rather than something to hand a clinical report to yet. Rounding out the measurement theme, Moysidou and colleagues validated the paediatric systemic Juvenile Arthritis Disease Activity Score 10 in adults with Still's disease across 129 clinic visits, and it outperformed four adult-specific instruments, with the best agreement with the reference standard and excellent discrimination of inactive disease [9]. That is a practical step toward one disease activity language spanning paediatric and adult Still's disease.

Our final theme is the severe and refractory end of the spectrum, where the evidence is necessarily small and observational. In Rheumatology, Wu and colleagues report 17 consecutive patients with anti-TNF-refractory intestinal Behçet's disease treated with upadacitinib, starting at 15 milligrams daily with escalation to 30 in nine patients [6]. At 6 months, half of those who underwent endoscopy achieved complete remission on the composite index, symptom remission was reached in about three quarters of patients, and the median disease activity index fell from 55 to essentially zero, with a glucocorticoid-sparing effect and acceptable tolerability. Uncontrolled and single-centre, but a signal worth knowing when anti-TNF fails. In Arthritis and Rheumatology, the AMETHYST cohort described by Grom and colleagues assembled 64 patients across eight sites with glucocorticoid-refractory macrophage activation syndrome complicating Still's disease, mostly children with systemic juvenile idiopathic arthritis [7]. Three quarters of patients received anakinra and about half received ciclosporin, yet complete laboratory remission occurred in only about one in nine, complete clinical response in under 40 percent, MAS recurred in roughly a third of patients, and there were 7 deaths, with one-year survival just under 94 percent. That is a sobering benchmark and a clear argument that our current salvage regimens are inadequate. Staying with severe disease, Stano and colleagues in Rheumatology followed 51 Italian patients with systemic sclerosis-associated pulmonary arterial hypertension on selexipag for a median of nearly two years [8]. Survival from the time of pulmonary hypertension diagnosis was about 88 percent at three years and 70 percent at five, persistence on drug was 84 percent at one year, and preserved right ventricular function was the key determinant of outcome. Notably, achieving a low one-year mortality risk category was far more common among patients who started selexipag within a year of pulmonary hypertension onset — roughly 38 percent versus 8 percent — supporting early combination therapy. Finally, Menentoğlu and colleagues report a multicentre paediatric ANCA-associated vasculitis series of 26 children, 11 of whom received therapeutic plasma exchange [10]. Those patients were sicker at baseline, with nearly double the vasculitis activity score and worse renal function, and while activity scores fell substantially after exchange, persistent renal abnormalities remained common, and one child died of catheter-related sepsis. Plasma exchange here looks like selective rescue rather than a routine addition.

If you only have time for one paper this week, make it the Nordic registry analysis of difficult-to-manage psoriatic arthritis in Annals of the Rheumatic Diseases [2]. Together with the DESIR axial spondyloarthritis data, it reframes the patient in front of you who has failed three biologics: the problem is usually not that you have run out of mechanisms of action.

Here are the key takeaways from this week in Rheumatology. First, boosters and updated COVID vaccines, not just an initial series, are what reduce hospitalisation in autoimmune rheumatic disease — and uptake is only about one in ten. Second, genuine treatment-refractory psoriatic arthritis and axial spondyloarthritis are rare, on the order of one percent; multiple biologic failures far more often signal fibromyalgia, depression, or opioid use that deserve direct attention. Third, telephone-plus-brief-consultation triage speeds access for most patients with inflammatory disease but misses about a third, so it cannot be used to rule out disease. Fourth, automated magnetic resonance scoring of sacroiliac joints now matches expert reader agreement for structural lesions, though bone marrow oedema remains harder. Fifth, in glucocorticoid-refractory macrophage activation syndrome, current anakinra and ciclosporin-based regimens leave a third of patients relapsing and roughly one in ten dying — safer, more effective options are needed. And sixth, in scleroderma-associated pulmonary arterial hypertension, starting selexipag early and preserving right ventricular function are what track with better risk profiles.

That's your roundup for This Week in Rheumatology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if you have a paper you have been meaning to read, upload the PDF, or paste any link, at audioscholar dot C C. We will turn it into audio like this one, in any of thirty-one languages.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.