Reviews/COPD Pharmacotherapy/Eosinophils predict inhaled corticosteroid response
Ics Risk · COPD-ICS-02 · v2
Borderline: within 4 points of the Likely/Uncertain boundary, so a re-grade could land either side.
Several large studies looked at whether a simple blood test measuring eosinophils can tell which COPD patients get the most benefit from inhaled steroids in avoiding flare-ups. Three big independent analyses pointed the same way: patients with higher eosinophil counts tended to gain more from steroid-containing treatment, and in one pooled analysis [2] this relationship was statistically formal. One analysis [1] found that taking steroids away caused more flare-ups specifically in higher-eosinophil patients. However, every one of these findings came from analyses done after the trials ended rather than from studies designed up front to test the question, and one older re-analysis [3] actually found the opposite pattern for flare-ups. A single-arm real-world study [4] saw similar flare-up reductions regardless of eosinophil level. The panel judged the direction of the evidence consistent but limited by its after-the-fact design and one conflicting result.
Three large, sponsor-independent analyses converge directionally: a pooled post-hoc analysis of budesonide-formoterol trials [2] (4528 patients) modelled eosinophils continuously and found a significant treatment effect for exacerbation reduction at counts of 0·10 × 10^9 cells per L or more (rate ratio 0·75, 95% CI 0·57-0·99; pinteraction=0·015); WISDOM [1] (2296 patients) showed exacerbation rate after ICS withdrawal rising with eosinophils (1·63 at ≥4%, 1·82 at ≥5%) with significant interaction at those cutoffs; and the IMPACT cluster analysis [6] (10,355 patients) found a high-eosinophil cluster (median 930/mm³) with a numerically lower exacerbation rate on triple versus dual therapy. Countervailing, the ISOLDE re-analysis [3] (751 patients) found exacerbation reduction was higher in the <2% eosinophil group, and the uncontrolled ORESTES study [4] showed similar exacerbation reductions across strata (29·8% below 100 cells/µL versus 25·8% at 100 cells/µL or more). The signal is consistent in direction but rests on post-hoc and exploratory designs. The contested point: The evidence divides on whether the eosinophil relationship holds for the exacerbation endpoint specifically: [2] and [1] support it while [3] runs the other way and [4] shows no separation. It also divides on design quality, since the supportive results are all post-hoc or exploratory [6] rather than prospectively tested.
All 9 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from COPD Pharmacotherapy, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
Every supportive result is post-hoc or exploratory; no prospectively-designed RCT tested the eosinophil-by-ICS interaction, and IMPACT [6] explicitly frames its finding as numerical and hypothesis-generating.
ISOLDE [3] is directly discordant on the claim's exact endpoint — exacerbation reduction was greater in the <2% eosinophil group — so the FEV1-decline benefit does not transfer to exacerbations.
Where it remains uncertain. The ISOLDE re-analysis [3] found exacerbation reduction higher in the <2% eosinophil group, and single-arm ORESTES [4] showed similar reductions across eosinophil strata, so the endpoint-specific signal is not uniform.
Aleatory
Exacerbation rates are count outcomes with substantial individual variability, so stratum differences such as 29·8% versus 25·8% in [4] can arise partly from chance.
The load-bearing papers from a corpus of 9, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Shows exacerbation harm on ICS withdrawal scales with eosinophils, a mirror-image of ICS exacerbation benefit, with significant interaction at the higher cutoffs.
“Our data suggest that counts of 4% or greater or 300 cells per μL or more might identify a deleterious effect of ICS withdrawal, an effect not seen in most patients with eosinophil counts below these thresholds.”
Panel auditInternal validity: medium · scope of inference: Patients with severe to very severe COPD and a history of exacerbations, assessing ICS withdrawal rather than ICS initiation.Key limitations: post-hoc subgroup analysis · withdrawal design not initiation · interaction significant only at higher cutoffs · industry funded
Pooled primary-data analysis with a significant eosinophil-by-treatment interaction for exacerbation reduction (RR 0.75, 95% CI 0.57-0.99; pinteraction=0.015) — the most direct support for the claim.
“In patients with COPD treated with formoterol, blood eosinophil count predicts exacerbation risk and the clinical response to ICS.”
Panel auditInternal validity: medium · scope of inference: COPD patients with a history of exacerbations and no asthma history, treated with budesonide-formoterol versus formoterol.Key limitations: post-hoc analysis · comparator is formoterol not placebo · industry funded
Uncontrolled real-world cohort showing no eosinophil-dependent separation in exacerbation reduction; weak counter-signal lacking a non-ICS comparator.
“According to blood eosinophil count, decreases were 29.8% (< 100 cells/µL, N = 110) and 25.8% (≥ 100 cells/µL, N = 208).”
Panel auditInternal validity: low · scope of inference: Routine-practice Spanish COPD patients initiating BGF triple therapy; speaks to broad benefit, not to a biomarker-by-ICS interaction.Key limitations: no comparator arm · retrospective observational · before-after only
Discordant primary evidence: on the exacerbation endpoint the predictive direction reversed, tempering the claim.
“Exacerbation rate reduction on ICS for fluticasone propionate versus placebo was higher in the eosinophil <2% group compared with the ≥2% group; time-to-first moderate/severe exacerbation was not different between treatments in either group.”
Panel auditInternal validity: medium · scope of inference: COPD patients with mean FEV1 50% predicted; the eosinophil finding that supports prediction is for FEV1 decline, not for exacerbations.Key limitations: post-hoc re-analysis · no significance reported for exacerbation subgroup difference · eosinophil-predictive signal was on FEV1 not exacerbations
4 of 9 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
A consistent, independently-replicated directional signal from large datasets supports the predictive relationship, but reliance on post-hoc designs plus one discordant endpoint keeps it short of high certainty.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
Prospective eosinophil-stratified ICS trial for exacerbations. Every supportive result in the corpus is post-hoc or exploratory ([2], [1], [6]), so the core predictive claim has never been tested by design. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A randomised controlled trial prospectively stratifying COPD patients by baseline blood eosinophils, with exacerbation rate as the pre-specified primary endpoint and ICS versus non-ICS treatment arms.
Open question: Does a prospectively tested eosinophil-by-ICS interaction predict exacerbation reduction? Why it is still open: The significant interaction in [2] and the signals in [1] and [6] are all post-hoc or exploratory, and [6] is explicitly hypothesis-generating.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19