This Week in Rheumatology — Sep 3, 2026

Welcome to This Week in Rheumatology. This week we're covering 10 notable papers spanning new frontiers in targeted and cellular therapy, the practical questions of steroid route and drug survival, and risk stratification in Still's disease, scleroderma and lupus overlap. Let's dive in.

We'll start with the therapeutic frontier. In Nature Medicine, Albach and colleagues report the phase 1 portion of the COMPARE trial, testing a fully human CD19 CAR T cell product, mivocabtagene autoleucel, in six patients with severe, treatment-refractory, ACPA-positive rheumatoid arthritis. All disease-modifying drugs were stopped, patients underwent standard lymphodepletion, and each received a single infusion, with follow-up out to between 36 and 52 weeks. Every patient developed cytokine release syndrome, but all events were grade 1 or 2; there was no neurotoxicity and no serious adverse event, and the single dose-limiting toxicity was a grade 3 transaminase rise that resolved. B cells were depleted in both blood and tissue, autoantibodies declined continuously, and four of the six patients seroconverted for anti-citrullinated vimentin antibodies while five of six seroconverted for IgM rheumatoid factor. Disease activity improved in all six despite no ongoing immunosuppression, with a median DAS28-CRP reduction of about a third, and three of the six reaching both remission and an ACR70 response. This is six patients and a safety endpoint, so the honest read is that the primary endpoint was met and the trial may proceed to phase 2 — not that CAR T therapy has arrived in rheumatoid arthritis.

Staying with targeted therapy, the Annals of the Rheumatic Diseases publishes a pharmacodynamic analysis from the phase 2 PAISLEY trial of deucravacitinib, the selective TYK2 inhibitor, in 363 patients with active lupus. At screening, dozens of immune genes and serum proteins separated patients from healthy volunteers. With treatment there were rapid and sustained falls in interferon levels and interferon-responsive gene and protein expression, reductions in B-cell pathway markers, rises in complement C3 and C4, and lower anti-double-stranded DNA titres in the interferon-high subgroup. Importantly, these analyses were purely descriptive, and the abstract notes similar kinetics compared with placebo, so this is mechanistic support for the ongoing phase 3 POETYK lupus programme rather than evidence of clinical efficacy. Alongside that, RMD Open carries two papers on the oral small molecules we already use. Coates and colleagues report a post hoc analysis of SELECT-PsA 2 restricted to the 195 patients with psoriatic arthritis who had failed exactly one TNF inhibitor. At week 24, upadacitinib 15 milligrams daily produced an ACR20 response in close to six in ten patients versus roughly two in ten on placebo, with minimal disease activity in about a quarter of those on upadacitinib versus almost none on placebo, and responses were maintained or improved through 152 weeks. Efficacy did not depend on how long the previous TNF inhibitor had been used, and lower pain scores as early as week 2 predicted minimal disease activity three years later — a genuinely useful bedside signal that early pain response is worth tracking. The second RMD Open paper, from Martinez-Molina and colleagues in the Spanish JAGUAR registry, follows 435 JAK inhibitor treatment episodes in rheumatoid arthritis over eight years. By four years, about a third of episodes had been stopped for lack of effectiveness and about one in five for adverse drug reactions. Two modifiable-adjacent factors stood out: patients who needed glucocorticoids for more than six months were markedly more likely to discontinue for inefficacy, with about half stopping by four years versus a third of those with shorter steroid exposure, and patients aged 65 and over were roughly twice as likely to stop because of adverse events. Chronic steroid dependence should be read as a marker of a JAK inhibitor that is not doing its job, and age remains the dominant safety consideration.

That brings us neatly to steroids themselves, and to what may be the most immediately actionable paper of the week. In the Annals of the Rheumatic Diseases, Fernández-Codina and colleagues used the Canadian Early Arthritis Cohort — 2,222 adults with newly diagnosed early rheumatoid arthritis enrolled between 2007 and 2023 — to ask whether the route of the initial glucocorticoid matters. Three quarters of the cohort got no steroid at all, about one in five received oral steroid, and one in twenty received parenteral steroid, either intra-articular or intramuscular. At twelve months, close to half OF THE PATIENTS given oral steroid were still taking it, compared with about a quarter of those given parenteral steroid. Put another way, oral steroid raised the odds of still being on steroid at one year roughly tenfold compared with no steroid, while the parenteral route roughly quadrupled them — so patients started on a parenteral bridge were about half as likely to remain steroid-dependent. Crucially, escalation to advanced therapy was the same across groups at around 14 percent, so the parenteral route did not buy steroid sparing at the cost of worse disease control. This is observational and channel bias is inevitable — the patients chosen for an intramuscular or intra-articular injection are not the same patients handed a prednisone taper — but it supports what many of us already suspect: an oral steroid bridge is far easier to start than to stop.

The companion measurement paper comes from the Lancet Rheumatology, where Wood and colleagues derive and validate the Lupus Arthritis and Musculoskeletal Disease Activity instrument, or LAMDA. Using 133 participants from the multicentre USEFUL study of intramuscular glucocorticoid for lupus arthritis, they regressed American College of Rheumatology core set variables against ultrasound grey scale and power Doppler synovitis scores, and ended up with a four-item score: swollen joint count, erythrocyte sedimentation rate, physician musculoskeletal disease activity on a visual analogue scale, and patient musculoskeletal pain. Reassuringly for feasibility, the 28-joint swollen count could substitute for the 66-joint count almost perfectly. LAMDA correlated with conventional activity measures, ultrasound findings and patient-reported outcomes, was responsive to glucocorticoid with a medium effect size, and held up in a separate 44-patient external validation cohort in Leeds, including provisional thresholds for trial entry, minimal disease activity, acceptable symptom state and clinically important improvement. Given how often lupus trials have foundered on insensitive musculoskeletal endpoints, this is a measurement problem worth fixing.

Our third theme is risk stratification, and it opens with Still's disease and macrophage activation syndrome, where two papers dovetail. In Rheumatology, Shiga and colleagues studied 86 adults with active Still's disease across three Japanese centres, 25 of whom had macrophage activation syndrome adjudicated by an expert panel. The HScore discriminated almost perfectly, with a cut-off of 192 giving sensitivity and specificity both around 96 percent, outperforming the modified HLH-2004 criteria and the 2016 systemic juvenile idiopathic arthritis MAS criteria. A modified HScore that drops bone marrow haemophagocytosis, with a cut-off of 171, still performed well — sensitivity around 92 percent and specificity around 84 percent — which matters when the patient is too unstable for an aspirate. Complementing that, Arthritis and Rheumatology publishes work from Rogani and colleagues showing that the activated CD38-positive, HLA-DR-positive CD8 T cell population, previously thought of as a MAS signature, is already expanded at treatment-naïve Still's disease onset at around 5 percent of CD8 cells versus under 1 percent in healthy donors, rising to roughly a third of CD8 cells at MAS. Frequencies tracked serum interleukin-18, and sustained interleukin-18 exposure in vitro reproduced the phenotype — placing persistent interleukin-18 upstream of this cytotoxic T cell expansion and offering a candidate biomarker and target.

Rounding out risk stratification are two cohort studies on long-term outcomes. Also in Arthritis and Rheumatology, Tonutti and colleagues report a nested case-control study within the EUSTAR systemic sclerosis registry, matching 454 patients with cancer to 454 cancer-free controls. Timing mattered: cancers synchronous with scleroderma onset were associated with anti-RNA polymerase III antibodies, anti-U1RNP and smoking, and inversely with digital ulcers, while lung cancer occurred mainly later and was associated with interstitial lung disease, anti-topoisomerase antibodies and smoking — each roughly doubling to tripling the odds. Cancers were more frequent among cyclophosphamide-treated patients, malignancy worsened overall survival particularly when it developed during follow-up, and reassuringly radiation therapy did not increase mortality or new-onset interstitial lung disease. Finally, in Rheumatology, Uzun and Isenberg analysed 855 lupus patients followed at a single centre since 1978. Myositis overlap was present in about 4 percent, and that group had far more interstitial lung disease — roughly 14 percent versus 1 percent — far more anti-RNP positivity, and a higher proportion of Afro-Caribbean patients. Over a median follow-up of more than two decades, mortality was more than twice as high in the overlap group, about 30 percent versus 14 percent, with myositis an independent predictor after adjustment. Myositis in lupus is a prognostic flag, not just a phenotypic curiosity.

If you only have time for one paper this week, make it the Canadian Early Arthritis Cohort analysis of glucocorticoid route in early rheumatoid arthritis. It addresses a decision you make several times a week, and it suggests that how you give the bridging steroid strongly predicts whether your patient ever gets off it.

Here are the key takeaways from this week in Rheumatology. First, in early rheumatoid arthritis, a parenteral rather than oral steroid bridge was associated with roughly half the likelihood of persistent steroid use at one year, with no increase in escalation to advanced therapy. Second, chronic steroid dependence alongside a JAK inhibitor signals inadequate disease control and predicts discontinuation for inefficacy, while age 65 and over predicts discontinuation for adverse events. Third, in psoriatic arthritis after one TNF inhibitor failure, upadacitinib gave durable responses out to three years, and early pain reduction at two to twelve weeks predicted long-term minimal disease activity. Fourth, in adults with Still's disease, the HScore performed excellently for macrophage activation syndrome, and a version omitting bone marrow findings remains usable when aspiration is not feasible. Fifth, cancer risk in systemic sclerosis is patterned by antibody and timing, and myositis overlap in lupus marks a group with more interstitial lung disease and more than double the long-term mortality. And finally, CD19 CAR T therapy in refractory seropositive rheumatoid arthritis cleared its phase 1 safety hurdle in six patients with encouraging serological and clinical signals — promising, but early.

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 it saves you time, a quick rating in your podcast app helps other physicians find the show.

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