Hidden immune cell states may reveal which sepsis patients will survive

Sepsis remains one of the most feared conditions in modern medicine, a runaway response to infection that can shut down organs within hours. Each year roughly 49 million people develop the syndrome worldwide, and about 11 million of them die, accounting for nearly a fifth of all global deaths. Even among patients who reach the hospital alive, septic shock, the most severe form, carries a mortality rate that can exceed 30 percent. Now a team of researchers in China has taken a major step toward explaining why some patients succumb while others recover, by dissecting the behavior of a mysterious family of immune cells that has long divided the field.

The cells in question are monocytic myeloid-derived suppressor cells, or M-MDSCs. These cells arise from monocytes, the white blood cells that normally serve as first responders and antigen presenters in the immune system. Early in sepsis, monocytes pump out inflammatory signals such as IL-1β, IL-6, and TNF-α, fueling the systemic inflammation that damages tissues. As the disease progresses, however, a subset of monocytes loses its HLA-DR surface molecules, loses its ability to present antigen, and instead begins actively suppressing other immune cells through molecules like arginase 1, indoleamine 2,3-dioxygenase, and TGF-β. The paradox has been that although this suppression might seem protective against the initial inflammatory storm, its persistence is now considered harmful, leaving patients vulnerable to secondary infections and chronic critical illness.

The central obstacle has been heterogeneity. The markers most commonly used to identify M-MDSCs, including CD33, CD11b, CD14, and low or absent HLA-DR, do not cleanly separate suppressive M-MDSCs from ordinary resting monocytes. Worse, studies of MDSCs in animal models and in humans have produced contradictory conclusions about whether these cells help or harm. In the new study, published in the Journal of Advanced Research, investigators led by Feng Wang, Fan He, and Liming Cheng enrolled 195 patients with sepsis at Tongji Hospital, dividing them into those with non-septic shock, survivors of septic shock, and non-survivors of septic shock, and then applied single-cell RNA sequencing to resolve the confusion.

The first surprise came from conventional flow cytometry. Although the percentage of HLA-DR-positive monocytes declined progressively with disease severity, and although M-MDSC frequency correlated positively with SOFA organ dysfunction scores overall, the raw abundance of M-MDSCs performed poorly as a prognostic tool, with an area under the curve of only 0.564 for distinguishing survivors from non-survivors. Even more striking, the correlation between M-MDSC frequency and SOFA score reversed direction in the patients who ultimately died. Something about the cells themselves, rather than their numbers, appeared to hold the prognostic information.

To capture that information, the team purified CD14-positive monocytes from patients classified as having either high or low HLA-DR expression and sequenced them individually on the 10x Genomics platform, integrating the data with a healthy donor control. After quality filtering, 29,545 cells yielded nine distinct transcriptional clusters, ranging from naïve and interferon-stimulated monocytes to pro-inflammatory monocytes and true M-MDSCs marked by high RETN and low HLA gene expression. Crucially, the vast majority of HLA-DR-low CD14-positive monocytes from sepsis patients proved to be authentic M-MDSCs, confirming that this clinically accessible fraction is the right place to look for the suppressive cells.

Zooming in on 11,858 M-MDSCs, the researchers identified five subsets, named for their signature genes. S100A_M-MDSCs expressed S100A12, GAPDH, and LGALS1 and were enriched for mitochondrial respiration and inflammatory response. IL1R2_M-MDSCs expressed IL1R2, FOSB, and DACH1 and showed a striking dual profile, combining positive regulation of cytokine production with negative regulation of the immune response. THBS1_M-MDSCs expressed THBS1, MTSS1, and ID2 and were dominated by TGF-β receptor signaling. Two smaller populations, an interferon-stimulated subset and a PPBP-expressing subset corresponding to monocyte-platelet aggregates, rounded out the picture. Together, the three dominant subsets accounted for roughly 90 percent of all M-MDSCs.

Pseudotime trajectory analysis then revealed that these subsets represent developmental stages rather than separate lineages. Early-stage S100A_M-MDSCs sit at the origin of the trajectory, brimming with pro-inflammatory genes. Middle-stage IL1R2_M-MDSCs and THBS1_M-MDSCs occupy a transitional zone, the former producing inflammatory mediators such as CXCL8 and IFITM2 while the latter ramps up TGF-β signaling. By the late stage, both subsets converge on a strongly immunosuppressive program, co-expressing genes such as ZEB2, FMN1, and ADAM17 that dampen immune responses. Notably, sepsis-derived M-MDSCs lacked the canonical suppressive machinery seen in cancer, including ARG1, NOS2, PD-L1, and CD73, underscoring that these cells behave differently depending on disease context.

The team also traced the transcriptional and metabolic rewiring that accompanies this maturation. Transcription factor analysis showed CEBPD and ETS1 active early, FOSB and CREB5 peaking in the middle stage, and EZH2 and FOXN3 accumulating to maximal levels late. Metabolically, early cells relied on oxidative phosphorylation and arginine metabolism, while later subsets shifted toward fatty acid biosynthesis and branched-chain amino acid metabolism. Seahorse assays confirmed that HLA-DR-low monocytes from sepsis patients had reduced basal and maximal mitochondrial respiration but elevated fatty acid synthase, a metabolic signature consistent with the transition from inflammatory monocyte to suppressive M-MDSC.

Among the markers identified, VSIG4 stood out. This surface molecule was barely detectable on monocytes from healthy donors but significantly elevated in sepsis patients, and it was specifically expressed on the suppressive THBS1_M-MDSC subset. In laboratory experiments, silencing VSIG4 in THP-1 monocyte cells boosted their production of IL-1β, IL-6, and TNF-α, while blocking VSIG4 on HLA-DR-low monocytes from sepsis patients restored the proliferation and HLA-DR expression of co-cultured CD4-positive T cells. These results establish VSIG4 as a functional marker of the suppressive state, not merely a passive correlate.

The translational payoff came when the researchers combined flow cytometric measurements of VSIG4 and IL1R2 on HLA-DR-low CD14-positive monocytes. Neither marker alone reliably separated survivors from non-survivors, but together they achieved a sensitivity of 92.3 percent and a specificity of 90 percent in the tested cohort. Using the maximum of the two values, the combined metric reached an area under the curve of 0.935, with 84.62 percent sensitivity and 93.33 percent specificity, and it outperformed SOFA score, lactate, and age in a multivariable model. A cutoff of 33 percent stratified 28-day survival in Kaplan-Meier analysis. The authors caution that the sample sizes for the sequencing and validation arms were small, that all patients came from a single center, and that longitudinal validation is still needed. Even so, the study delivers something the field has lacked: a coherent map of M-MDSC development in human sepsis, from pro-inflammatory beginnings to immunosuppressive maturity, with surface markers that can be measured in a clinical flow cytometry lab. If larger trials confirm the findings, a simple two-marker blood test could one day help clinicians identify, within days of admission, which septic patients are sliding toward the immunoparalysis that kills so many, and direct immunorestorative therapies to those who need them most.

Subject of Research: Functional heterogeneity of monocytic myeloid-derived suppressor cells in sepsis prognosis

Article Title: Illustrating the functional heterogeneity of M-MDSCs to predict sepsis outcomes

Article References: Tang, G., Xing, W., Zhu, L., Lu, Y., Jiang, K., Wu, S., Sun, Z., Hou, H., Cheng, L., He, F., & Wang, F. (2026). Illustrating the functional heterogeneity of M-MDSCs to predict sepsis outcomes. Journal of Advanced Research, 88, 807-824. https://doi.org/10.1016/j.jare.2026.01.008

Image Credits: AI Generated

DOI: 10.1016/j.jare.2026.01.008

Keywords: sepsis, septic shock, M-MDSCs, single-cell RNA sequencing, VSIG4, IL1R2, TGF-β signaling, immunosuppression, biomarkers, flow cytometry, monocytes, immunometabolism