Writing in the journal New Microbes and New Infections, a team led by researchers at the CSIR Institute of Genomics and Integrative Biology and Max Hospital in New Delhi reports a whole-blood transcriptomic analysis of 112 dengue-positive patients recruited at their first hospital visit. The participants were classified according to the revised 2009 WHO guidelines into those without warning signs and those with warning signs, the latter defined by coagulation abnormalities such as thrombocytopenia, clinical fluid accumulation reflected in hemoconcentration, and increased vascular fragility. Within the group without warning signs, 46 patients also presented with leukopenia, allowing the investigators to stratify the cohort into mild, moderate and relatively severe subgroups. Although none of the patients met the formal criteria for severe dengue, the 21 patients carrying warning signs together with thrombocytopenia were treated as demonstrating a propensity for clinically severe outcomes because of their depressed platelet counts.

The technical pipeline behind the study was considerable. The team isolated RNA from one milliliter of whole blood per patient, depleted globin and ribosomal RNA, and generated stranded sequencing libraries that were run as paired-end 151-base-pair reads on an Illumina NextSeq 2000 platform. The effort produced more than two billion reads, averaging about 18.5 million reads per sample. After quality trimming and quasi-mapping to the human transcriptome, differential expression was computed with DESeq2, and confounder analyses for age and sex showed no significant differences between the clinical groups. All 112 patients experienced primary dengue infection, and serotype analysis based on viral reads recovered from the sequencing data indicated that the majority were infected with DENV-2, consistent with the serotype circulating in Delhi during the study period. Samples were collected during the acute febrile phase, with a mean of 3.69 days of fever at enrollment.

The clinical chemistry told a coherent story before any sequencing was considered. Total leukocyte counts, platelets, lymphocytes and neutrophils all differed significantly between the groups, with p-values below 0.001. The two leukopenic groups showed lower neutrophil and higher lymphocyte values than the mild group, a pattern the authors interpret as a subdued first-line defense, since neutrophils are the earliest active killers of pathogens during viral and bacterial infections. Notably, hematocrit was not elevated in the patients with warning signs, suggesting that capillary leak had not yet begun at the time of sampling. Liver function tests available for a subset of patients showed higher SGOT and SGPT values in the warning-sign group, along with significantly lower total protein and elevated bilirubin, although all values remained within normal ranges, hinting at hepatic derangement that was only beginning to emerge.

Differential expression analysis between the moderate and mild groups yielded 15 significantly downregulated genes, most of them involved in immune response and inflammation, including the chemokine CXCL2, the neutrophil marker CD177, the acute-phase protein ORM1 and the inflammation regulator PI3. The authors read this as evidence of a subdued immune response in moderate patients compared with mild ones. By contrast, comparisons involving the severe group were far more dramatic: severe versus mild produced 215 upregulated and 92 downregulated genes, while severe versus moderate produced 142 upregulated and only 2 downregulated genes. Pathway enrichment against the Reactome database showed that upregulated functions in severe patients centered on glycosylation and the nervous system, whereas downregulated pathways encompassed immune and inflammatory response, GPCR signaling, platelet activation and aggregation, and prostaglandin synthesis in platelets and endothelial cells.

One of the most intriguing findings was that, although many immune response genes were downregulated in severe patients, a diverse repertoire of B cell receptor genes, including immunoglobulin light and heavy chain variables, was upregulated, indicating an acute antibody response associated with severity. Upregulation of KIR3DL3, LILRB5 and IRAK1BP1 appeared to divert innate immune pathways toward resolution of inflammation. Taken together, the differential expression profile suggests an attenuation of immune and inflammatory activity in severe patients precisely at the time when platelet counts fall below normal thresholds due to platelet dysfunction, a counterintuitive pattern that complicates the traditional picture of dengue severity as a cytokine-driven hyperinflammatory state.

To move beyond lists of individual genes, the team applied Weighted Gene Co-expression Network Analysis, or WGCNA, to 14,071 genes across all 112 samples. After selecting an optimal soft-threshold power to ensure scale-free network topology, the algorithm organized the genes into 22 modules, each representing a set of genes with highly similar expression patterns. The severe subgroup split into two major clades with starkly different co-expression patterns: clade 1 modules showed higher expression in severe patients and were enriched for metabolism, glycosylation, nervous system processes and DNA repair, while clade 2 modules showed decreased expression and carried functions related to blood coagulation, platelet activation, cell adhesion, immune response, regulation of virus replication, interferon response and cytokine signaling. Correlating module eigengenes with clinical parameters revealed that clade 2 modules correlated positively with platelet counts, while a single module, turquoise, correlated negatively, reinforcing the link between immune response modules and platelet dysregulation. Total leukocyte count, by contrast, showed no significant correlation with any module.

Within the platelet-associated modules, the investigators identified a set of genes that collectively define the severe state. The greenyellow module contained interferon pathway genes such as IFI44L, IFI35, OAS1, SAMD9L and KAT2B, while the darkred module harbored genes governing platelet function, coagulation and endothelial integrity, including ARHGEF12, CTTN, SELP, PPBP, PRKAR2B and F13A1. Other modules captured vesicular trafficking genes such as GOLGA7 and RAB27B, thought to be required for viral replication and autophagy, and innate antiviral genes including IFIH1, which encodes the cytosolic RNA sensor MDA5, and HNRNPA2B1. When the team dichotomized the severe patients by platelet count, comparing those below 100 times 10 to the ninth per liter with those between 100 and 140, 16 genes showed significant differences in baseline transcript abundance, with IFI44L standing out at a difference of more than 800 counts per million. Regression analysis found no genes significantly correlated with SGOT, implying that modest transaminase elevations do not substantially shape disease outcome.

The findings converge with transcriptomic studies of dengue patients in Brazil, where interferon-stimulated genes such as IFI27 and ISG15 emerged as biomarkers capable of distinguishing dengue hemorrhagic fever from uncomplicated dengue during the late acute phase. Although IFI27 was not captured in the Indian dataset, ISG15 was consistently upregulated in patients with warning signs and low platelets, aligning with its proposed role as a severity marker. The Indian study also detected CTTN, a gene tied to endothelial barrier function that was not a focus of the Brazilian datasets, hinting at population-specific elements of the immune-endothelial axis. The authors propose a sequential pathogenic cascade for patients with warning signs: a suboptimal early interferon and innate immune response weakens host defenses and permits viral replication, heightened viral stress drives platelet dysfunction and clearance, and, independently of cytokine-mediated hyperinflammation, upregulation of O-linked glycosylation genes such as GALNT13, B4GALT1 and B3GNT4 alongside downregulation of structural genes like CTTN alters the endothelial glycocalyx, producing the vascular permeability that characterizes the onset of clinical warning signs.

The study carries the inherent limitation of a cross-sectional design, since blood was drawn at a single acute-phase time point, and whole-blood RNA means the observed signatures may partly reflect shifts in circulating cell composition rather than purely transcriptional changes within individual cell types. The authors note that single-cell transcriptomics or computational deconvolution in future longitudinal cohorts would help resolve cell-type-specific contributions. Even so, the work suggests practical avenues: monitoring glycocalyx-related gene signatures could enable early identification of patients at risk of severe disease, and therapeutic strategies aimed at preserving glycocalyx integrity may offer a novel way to reduce vascular complications. If validated across larger and more geographically diverse cohorts, the interferon and glycosylation signatures identified here could become molecular biomarkers for predicting, and perhaps mitigating, severe dengue before the first warning signs appear.

Subject of Research: Host transcriptomic signatures associated with severe dengue infection in Indian patients

Article Title: Interferon gamma and inflammatory response genes by weighted gene co-expression network analysis reveal significant associations with severe dengue infection in India

Article References: Shamim, U., Arora, S., Maurya, R., Soni, J., Shukla, R., Mehta, P., Tarai, B., Budhiraja, S., & Pandey, R. (2026). Interferon gamma and inflammatory response genes by weighted gene co-expression network analysis reveal significant associations with severe dengue infection in India. New Microbes and New Infections, 74, Article 101855. https://doi.org/10.1016/j.nmni.2026.101855

Image Credits: AI Generated

DOI: 10.1016/j.nmni.2026.101855

Keywords: dengue, DENV-2, interferon, WGCNA, transcriptomics, thrombocytopenia, endothelial glycocalyx, glycosylation, innate immunity, biomarkers, India, vascular leakage