Design: retrospective multicentre cohort from nine tertiary referral centres, with exploratory multivariable profiling and comparison against historical lupus cohorts
Population: adults with macrophage activation syndrome complicating systemic lupus erythematosus or Still's disease; historical comparison cohorts of 1,000, 2,228 and 2,055 lupus patients
Primary outcome: clinical and laboratory features, treatments and outcomes compared between the two forms
Effect: Still's-associated cases showed greater hyperferritinaemia, C-reactive protein, organomegaly and liver injury with 100% meeting 2016 criteria; lupus-associated cases showed more CNS dysfunction, deeper cytopenias, 73% direct antiglobulin positivity and 92% meeting criteria

Macrophage activation syndrome is usually taught through Still's disease, and when it complicates systemic lupus erythematosus the same template gets applied. A retrospective cohort from nine tertiary referral centres compared adults with lupus-associated and Still's-associated macrophage activation syndrome directly, and also against three large historical lupus cohorts of 1,000, 2,228 and 2,055 patients.

The Still's cases behaved as the textbook describes: more marked hyperferritinaemia, very high C-reactive protein, organomegaly and liver injury, with every case meeting the 2016 ACR/EULAR/PRINTO criteria. The lupus cases did not. They more often involved central nervous system dysfunction, showed less liver injury, and carried deeper cytopenias including anaemia — with 73% direct antiglobulin test positivity, pointing at autoimmune haemolysis rather than consumption as a major driver. Only 92% met the classification criteria, which matters, because those criteria were derived in the Still's phenotype. Against the historical lupus cohorts, the macrophage activation group stood out for mucocutaneous disease, haemolysis, central nervous system involvement, a less pronounced female preponderance and much higher direct antiglobulin positivity.

Two consequences. Diagnostically, waiting for the ferritin and the hepatosplenomegaly of the Still's picture will delay recognition in lupus — a lupus patient with new neuropsychiatric features and falling counts should prompt a haemolysis screen and a ferritin, not a lumbar puncture alone. Therapeutically, the extrapolation is the thing being questioned: interleukin-1 antagonism was used in seven lupus cases with mixed results and no rapid resolution, which is not what the Still's experience would predict.

Send a direct antiglobulin test, reticulocytes, lactate dehydrogenase and haptoglobin in any lupus patient with unexplained falling counts.
Take new neuropsychiatric features plus cytopenias in lupus as a reason to check ferritin, triglycerides and fibrinogen.
Do not require the Still's phenotype — organomegaly and marked liver injury were less common here.
Remember the classification criteria were built on the Still's picture; 8% of lupus cases did not meet them.
Do not assume interleukin-1 blockade will work as it does in Still's disease.

Why it matters

The diagnostic template and the treatment approach for this complication are both borrowed from a different disease, and the two phenotypes turn out to diverge.

The statistics, in plain English

This is a retrospective cohort of a rare complication assembled from nine referral centres, so the comparison groups are small and the phenotype differences are descriptive rather than tested with a formal model — the exploratory multivariable profiling is labelled as such. The 73% direct antiglobulin positivity is the most striking single figure and the one most likely to survive replication, because it is a binary laboratory result rather than a clinical judgement. The treatment observation rests on seven patients and cannot establish that interleukin-1 blockade does not work; it establishes only that nobody in this series had a rapid response to it.