Skip to main content

Inflammatory protein trajectories and severe respiratory failure in COVID-19: a secondary analysis of the SAVE-MORE randomised trial.

EBioMedicine

Authors: Nick Keur, Emmanouil Stylianakis, Isis Ricaño-Ponce, Vassiliki Rapti, Haralampos Milionis, Simeon Metallidis, Georgios Adamis, Archontoula Fragkou, Aggeliki Rapti, Pierluigi Del Vecchio, Eleni Florou, Ioannis Kalomenidis, Georgios Chrysos, Andrea Angheben, Ilias Kainis, Konstantina Iliopoulou, Francesco Castelli, Francesco Saverio Serino, Agamemnon Bakakos, Emanuele Nicastri, Maria Dafni, Sofia Ioannou, Lorenzo Dagna, Katerina Dimakou, Glykeria Tzatzagou, Maria Chini, Matteo Bassetti, Vasileios Kotsis, George Tsoukalas, Carlo Selmi, Evangelos Siarkos, Michael Samarkos, Michael Doumas, Aikaterini Masgala, Ilias Papanikolaou, Aikaterini Argyraki, Karolina Akinosoglou, Konstantinos Syrigos, Styliani Sympardi, Periklis Panagopoulos, George N Dalekos, Mihai G Netea, Evangelos J Giamarellos-Bourboulis, Vinod Kumar

BACKGROUND: Severe COVID-19 is characterised by maladaptive host inflammation. In the SAVE-MORE trial (EudraCT no. 2020-005828-11; ClinicalTrials.gov NCT04680949), suPAR-guided anakinra treatment reduced mortality and progression to severe respiratory failure (SRF), but the underlying molecular mechanisms remain unclear.

METHODS: We performed longitudinal proteomic profiling (Olink Inflammation panel, 67 proteins) in 587 SAVE-MORE participants (400 treated with anakinra and 187 with placebo), of whom 453 had samples at all three time points (baseline, Day 4, and Day 7). Analyses aimed to (i) identify baseline predictors of SRF, (ii) define inflammatory trajectories using mixed-effects models, and (iii) characterise proteomic changes at SRF onset. Cox models identified prognostic markers, validated in an independent anakinra-treated cohort from the SAVE trial (n = 153).

FINDINGS: Multivariate Cox regression models adjusted for sex, age, BMI, and dexamethasone treatment showed that baseline IL-6 concentrations predicted SRF across both treatment arms (placebo HR = 1·99, CI: 1·31-3·02; anakinra HR = 2·12, CI: 1·56-2·87), while PD-L1 predicted SRF only in the placebo group (HR = 2·75, CI: 1·79-4·24). Anakinra allocation was associated with changes in inflammatory trajectories related to SRF progression. Patients developing SRF showed sustained activation marked by elevated concentrations of IL-6, MCP-1/3, EN-RAGE, and PD-L1, whereas SRF-free patients had higher levels of proteins linked to regulatory and adaptive immunity, including CD6, CD244, TRAIL, TNF-B, and FLT3L. Event-aligned analyses showed an acute inflammatory surge at SRF onset involving IL-6, CXCL5, OSM, and VEGFA, which was significantly attenuated by anakinra. Results were further confirmed in the SAVE cohort.

INTERPRETATION: Progression to SRF was associated with an inflammatory amplification programme involving cytokine, chemokine, tissue-injury, and immune-checkpoint pathways. Anakinra allocation was associated with smaller changes in proteins associated with inflammatory amplification and acute inflammatory changes accompanying respiratory deterioration. These findings provide biological context for the clinical benefit observed in the SAVE-MORE trial.

FUNDING: Hellenic Institute for the Study of Sepsis (sponsor) and Swedish Orphan Biovitrum (funder).

Copyright © 2026 The Author(s). Published by Elsevier B.V. All rights reserved.

PMID: 43050905

Participating cluster members