Cytokine storm dynamics in hantavirus pulmonary syndrome: a multiscale ODE model with Wasserstein early-warning score and application to the 2026 Andes virus outbreak
des Rochettes, B.
Show abstract
Background. The ongoing Andes hantavirus outbreak linked to the cruise ship MV Hondius (April-May 2026, seven confirmed cases, three deaths, patients hospitalised across six countries including France) highlights the urgent need for mechanistic tools to predict which hantavirus pulmonary syndrome (HPS) patients will progress to fatal cytokine storm. Methods. We present a 14-variable antigen-gated ordinary differential equation (ODE) model integrating viral dynamics, CD8+ cytotoxic T lymphocyte (CTL) expansion, four cytokines (TNF-alpha, IFN-gamma, IL-6, IL-10), VEGF-mediated vascular permeability, and platelets. We derive two reproduction numbers: the viral invasion number R0 and the immunopathological loop gain Rip. We apply Villani's hypocoercivity theory and the HWI optimal transport inequality to prove that the spectral gap of the CTL-IFN-gamma feedback loop collapses to zero at a critical infected endothelial cell count Ic*, providing a computable early-warning threshold. We define a Wasserstein patient stratification score from six clinically observable variables. Results. At default parameters (R0 = 0.396, Rip = 1.875): (1) the disease-free equilibrium is locally asymptotically stable - the virus self-limits - but the CTL-IFN-gamma loop has sufficient gain to amplify autonomously once established; (2) the storm-block spectral gap collapses exactly to zero at Ic* = 2.23 cells/uL, a threshold attained within hours of infection onset, confirming that immunopathological amplification is essentially unavoidable; (3) the Wasserstein score rises 1-2 days before vascular permeability reaches clinical threshold, providing an early-warning window; (4) exogenous IL-10 supplementation is the single most effective intervention (predicted 40% reduction in peak permeability), outperforming corticosteroid immunosuppression and ECMO; the combination of all three applied at day 7 reduces peak permeability below the fatal threshold. Conclusions. This framework predicts that HPS cytokine storm is a structural consequence of Rip > 1 rather than excessive viral load, explaining death after viral clearance. For clinicians managing MV Hondius Andes virus patients, the model identifies a six-variable triage score and a day-7 IL-10-centred therapeutic window as the highest-priority clinical targets. A live simulator and bedside triage tool are available at xvirus.org.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- The relationship between controllability, optimal testing resource allocation, and incubation-latent period mismatch as revealed by COVID-19 95%
- Risk of COVID-19 variant importation – How useful are travel control measures? 95%
- A Metapopulation Network Model for the Spreading of SARS-CoV-2: Case Study for Ireland 94%
Similar papers in this journal
- Introducing a framework for within-host dynamics and mutations modelling of H5N1 influenza infection in humans 95%
- Vaccination strategies when vaccines are scarce: On conflicts between reducing the burden and avoiding the evolution of escape mutants 95%
- Adaptive vaccination may be needed to eliminate COVID-19: Results from a runtime-alterable strain-drift and waning-immunity model 95%
Similar papers in this journal
Similar papers in this journal
- Heterogeneous immune recovery after viral response through a dynamical model of feedback-driven persistence and clearance 95%
- Data driven model discovery and interpretation for CAR T-cell killing using sparse identification and latent variables 94%
- From infection to immunity: understanding the response to SARS-CoV2 through in-silico modeling 94%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.