tetano
Editor, Senior Moderator
Front Immunol
. 2026 Aug 19:17:1805650.
doi: 10.3389/fimmu.2026.1805650. eCollection 2026.
Kailin Mai # 1 2 3 , Yang Wang # 3 4 5 6 , Wei Qu # 7 , Yunceng Weng 3 , Zhengshi Lin 3 , Liping Chen 3 , Arlindo Oliveira 4 6 8 , Nanshan Zhong 3 9 , Chitin Hon 4 6 10 , Weiqi Pan 3 4 6 , Zifeng Yang 3 4 6 9
Affiliations
Background: Hypercytokinemia is a major contributor of tissue damage and mortality in severe influenza. However, most studies rely on static, single-time-point measurements, providing limited insight into the evolving inflammatory response. The prognostic relevance of longitudinal cytokine trajectories remains poorly defined.
Methods: We conducted a multi-cohort analysis of 186 patients from three public datasets to identify cytokines with influenza-severity-dependent expression patterns. We then analyzed a longitudinal cohort of 33 patients with severe influenza who underwent serial cytokine measurements, using linear mixed-effects models to identify the top 10 severity-associated cytokines. Integrating both analyses, we defined a core cytokine panel and a composite core panel score. Prognostic value was assessed using time-varying Cox regression and landmark analyses, with comparisons against clinical predictors, including LODS scores and CRP, as well as sensitivity analyses adjusting for diabetes.
Results: A core panel of six pro-inflammatory cytokines (IL-8, IL-6, G-CSF, MCP-1, TNF-α, and MIP-1α) was identified. Non-survivors showed distinct longitudinal trajectories of the core panel score, characterized by a progressive increase over the disease course. Paired sputum analyses suggested that these systemic cytokine signatures partially reflected pulmonary inflammation. Baseline core panel scores were not significantly associated with mortality (HR 3.29, 95% CI 0.95-11.38; P = 0.059), whereas time-varying monitoring showed better model fit (AIC 29.34 vs 34.85). Core panel scores remained associated with mortality in analyses accounting for LODS scores, baseline CRP, and diabetes. Landmark analysis further suggested that updated core panel scores may provide increasing prognostic information over time compared with baseline assessment, with delta AUC increasing from near zero at day 5 to 0.24 by day 17 after ICU admission.
Conclusions: Dynamic monitoring of a core cytokine panel may provide greater prognostic information than single-time-point assessment in severe influenza, although these findings require validation in larger independent cohorts. Longitudinal inflammatory profiling may help refine risk stratification and provide a framework for future precision studies.
Keywords: biomarkers; cytokines; longitudinal analysis; risk stratification; severe influenza.
. 2026 Aug 19:17:1805650.
doi: 10.3389/fimmu.2026.1805650. eCollection 2026.
Dynamic monitoring of a core pro-inflammatory cytokine panel improves mortality prediction in severe influenza
Kailin Mai # 1 2 3 , Yang Wang # 3 4 5 6 , Wei Qu # 7 , Yunceng Weng 3 , Zhengshi Lin 3 , Liping Chen 3 , Arlindo Oliveira 4 6 8 , Nanshan Zhong 3 9 , Chitin Hon 4 6 10 , Weiqi Pan 3 4 6 , Zifeng Yang 3 4 6 9
Affiliations
- PMID: 42688046
- PMCID: PMC13533885
- DOI: 10.3389/fimmu.2026.1805650
Abstract
Background: Hypercytokinemia is a major contributor of tissue damage and mortality in severe influenza. However, most studies rely on static, single-time-point measurements, providing limited insight into the evolving inflammatory response. The prognostic relevance of longitudinal cytokine trajectories remains poorly defined.
Methods: We conducted a multi-cohort analysis of 186 patients from three public datasets to identify cytokines with influenza-severity-dependent expression patterns. We then analyzed a longitudinal cohort of 33 patients with severe influenza who underwent serial cytokine measurements, using linear mixed-effects models to identify the top 10 severity-associated cytokines. Integrating both analyses, we defined a core cytokine panel and a composite core panel score. Prognostic value was assessed using time-varying Cox regression and landmark analyses, with comparisons against clinical predictors, including LODS scores and CRP, as well as sensitivity analyses adjusting for diabetes.
Results: A core panel of six pro-inflammatory cytokines (IL-8, IL-6, G-CSF, MCP-1, TNF-α, and MIP-1α) was identified. Non-survivors showed distinct longitudinal trajectories of the core panel score, characterized by a progressive increase over the disease course. Paired sputum analyses suggested that these systemic cytokine signatures partially reflected pulmonary inflammation. Baseline core panel scores were not significantly associated with mortality (HR 3.29, 95% CI 0.95-11.38; P = 0.059), whereas time-varying monitoring showed better model fit (AIC 29.34 vs 34.85). Core panel scores remained associated with mortality in analyses accounting for LODS scores, baseline CRP, and diabetes. Landmark analysis further suggested that updated core panel scores may provide increasing prognostic information over time compared with baseline assessment, with delta AUC increasing from near zero at day 5 to 0.24 by day 17 after ICU admission.
Conclusions: Dynamic monitoring of a core cytokine panel may provide greater prognostic information than single-time-point assessment in severe influenza, although these findings require validation in larger independent cohorts. Longitudinal inflammatory profiling may help refine risk stratification and provide a framework for future precision studies.
Keywords: biomarkers; cytokines; longitudinal analysis; risk stratification; severe influenza.