Cytokine and endothelial injury signatures associated with severe dengue: a systematic review and meta-analysis integrating viral burden and host-response markers
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Severe dengue reflects a temporally regulated interaction between viral burden, NS1 antigenaemia, cytokine and chemokine amplification, endothelial activation, glycocalyx injury, and organ stress. Although individual cytokines, endothelial markers, viral-burden measures, and clinical markers have been widely studied, the integrated pathogen-host evidence base remains fragmented. We synthesised evidence for cytokine, endothelial, and viral-burden signatures associated with severe dengue and assessed whether paired pathogen-host measurement provides a biologically coherent framework for severity assessment.
Methods
We searched MEDLINE, Embase, Scopus, Web of Science, Cochrane Library, Global Health, WHO Global Index Medicus, and medRxiv from database inception to 30 April 2026, without language restriction, for studies reporting viral burden, NS1 antigenaemia, cytokine, chemokine, endothelial, glycocalyx, inflammatory, or routine host-response markers in laboratory-confirmed dengue with severity outcomes. Eligible designs were prognostic-factor association studies, cross-sectional biomarker studies, and multivariable prediction-model studies. Risk of bias was assessed using QUIPS for prognostic-factor studies, PROBAST for prediction-model studies, and the relevant JBI critical appraisal checklist for cross-sectional biomarker studies, with the Newcastle-Ottawa Scale used selectively for cohort or case-control designs not amenable to QUIPS. Random-effects meta-analysis pooled standardised mean differences using restricted maximum likelihood with Hartung-Knapp adjustment. The protocol was registered with PROSPERO (CRD420261396923) before final extraction and synthesis.
Findings
Of 4,180 records identified, 79 studies including 47,612 participants met eligibility criteria. Forty-nine studies evaluated paired pathogen-host markers, 14 evaluated viral burden or NS1 antigenaemia alone, nine evaluated host biomarkers alone, and seven reported multivariable prediction models. Pathogen-side markers showed modest pooled severity associations whose magnitude depended on day of illness, immune status, and infecting serotype. Cytokine and chemokine markers, particularly IL-10, IL-6, IL-8, and CXCL10/IP-10, showed larger pooled effects favouring severe disease, while endothelial and glycocalyx markers, including angiopoietin-2 and syndecan-1, provided the most direct mechanistic link to plasma leakage. Routine clinical markers, especially platelet count, AST, ferritin, ALT, and lactate, retained substantial discriminatory value. Prediction models reported areas under the curve of up to 0·96 in internal validation and 0·97 in discovery analyses, but three had been externally validated, calibration was reported in two, and decision-curve analysis in none.
Interpretation
Current evidence supports severe dengue as an integrated pathogen-host injury syndrome in which viral burden and NS1 antigenaemia interact with cytokine amplification, endothelial dysfunction, glycocalyx injury, and routine markers of organ stress. The strongest translational direction is not a single biomarker but a parsimonious cytokine-endothelial-pathogen panel requiring prospective external validation across age groups, serotypes, immune-status strata, and endemic regions. Existing evidence supports candidate marker prioritisation and mechanistic synthesis, but not immediate routine clinical deployment.
