Avacopan and severe infection: the steroid-sparing argument, measured

Design

Multicentre retrospective cohort with time-varying exposure, stabilised IPTW and Andersen-Gill time-dependent Cox models

Population

387 patients with ANCA-associated vasculitis; 52 exposed to avacopan, 335 not

Primary outcome

Recurrent relapse and severe infection requiring hospitalisation

Effect

Severe infection adjusted HR 0.22 (95% CI 0.07-0.68, P=0.008); no significant difference in recurrent relapse; lower prednisolone exposure over time

Avacopan's case has always been that sparing glucocorticoid should spare infection. This multicentre retrospective cohort tried to measure that rather than assume it, in 387 patients with ANCA-associated vasculitis of whom 52 received avacopan and 335 did not.

The analytic care is what distinguishes it. Avacopan use was modelled as a time-varying exposure rather than a baseline attribute — important, because patients start it at different points and a baseline classification misattributes person-time. Stabilised inverse probability of treatment weighting adjusted for baseline differences, and recurrent relapse and severe infection (defined as requiring hospitalisation) were analysed with time-dependent Cox models on the Andersen-Gill formulation. Exploratory models added time-varying and cumulative prednisolone exposure.

Avacopan exposure was associated with a lower risk of severe infection (adjusted HR 0.22, 95% CI 0.07-0.68, P=0.008) and with lower prednisolone exposure over time. There was no significant difference in recurrent relapse. The infection association stayed directionally consistent in glucocorticoid-adjusted models, though its size moved depending on how the model was specified.

That last sentence is the honest heart of it. If the benefit runs entirely through steroid sparing, adjusting for steroid should abolish it; it did not, but the magnitude was unstable — which is what you would expect when the mediator and the exposure are this tangled.

• Severe infection here means requiring hospitalisation — a hard endpoint, not a reported symptom
• No relapse difference; the case for avacopan in this cohort rests on infection, not disease control
• Fifty-two exposed patients is a small group for a hazard ratio this extreme
• Avacopan availability and cost are the binding constraints in most Indian units
• Time-varying exposure modelling matters here — a baseline classification would have given a different answer

Why it matters

It tests the mechanism avacopan is sold on — sparing steroid to spare infection — against real-world outcomes.

Don't overread it

Observational with 52 exposed patients; the authors state explicitly that causal inference is not available here.

The statistics, in plain English

A hazard ratio of 0.22 sounds like a 78% reduction, but the confidence interval runs from 0.07 to 0.68 — so the real effect could be anywhere from near-abolition to a modest third. With 52 exposed patients and hospitalised infections as the event, the number of events driving this is small. And in an observational cohort, the patients started on avacopan may have been those judged able to tolerate steroid reduction, which is exactly the group at lower infection risk to begin with.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.