Abstract

Background

Sepsis is a global health burden characterized by high heterogeneity and uncontrolled immune response, with a notable lack of reliable methods for early prognosis and risk stratification. Epigenetic modifications, particularly lactylation, have recently emerged as key regulators in the early pathophysiology of sepsis. However, their potential for immune-related mortality risk stratification remains largely unexplored. This study aimed to investigate dynamic changes in lactylation during sepsis progression and to develop a rapid, lactylation-based prognostic signature.

Methods

Blood transcriptional profiles and single-cell RNA sequencing data from septic patients were analyzed to assess glycolytic activity and lactylation in relation to patient mortality. Patients were stratified into subgroups using k-means clustering based on lactylation levels. Machine learning algorithms, integrated with pseudotime trajectory reconstruction, were employed to map the temporal dynamics of lactylation. A prognostic model was then constructed using lactylation-associated hub genes and validated in external transcriptomic datasets, a prospective single-center clinical cohort. The underlying mechanism was further explored in vitro using human monocytes.

Results

The study systematically characterized the dynamic alterations in lactylation patterns and immune microenvironment across distinct patient clusters. A lactylation-based prognostic model was developed, comprising eight key genes (CD160, HELB, ING4, PIP5K1C, SRPRA, CDCA7, FAM3A, PPP1R15A), and demonstrated strong predictive performance for sepsis outcomes (AUC = 0.78 in the training cohort; AUC = 0.73 in the validation cohort). Temporal expression patterns of lactylation-related hub genes revealed dynamic immune responses throughout disease progression. In the prospective cohort of septic patients (N = 51), the model showed high predictive accuracy for survival, with AUCs of 0.82 (7-day), 0.80 (14-day), and 0.86 (28-day). Additionally, global lactylation levels were significantly elevated in THP-1 cells following treatment with Sephin1, a selective PPP1R15A inhibitor, suggesting a mechanistic link.

Conclusions

Lactylation is significantly associated with increased mortality risk in sepsis. The proposed individualized prognostic model, based on dysregulated immune cell metabolism, accurately predicts early mortality and may inform optimized clinical management of septic patients.

MeSH

Cells, Genes, Mortality, Immune System Diseases, Sepsis, Immunity