Identification of distinct immune subtypes in sepsis through dual immunomarker trajectory.

BACKGROUND

While endotyping approaches have enhanced precision immunotherapy in sepsis, their temporal instability limits consistent clinical application. To overcome this challenge, the current study utilizes longitudinal immune markers to derive clinically relevant classifications.

METHODS

This study included three cohorts of 3,223 adult septic patients with at least three measurements of two readily available longitudinal immune markers, C-reactive protein (CRP) and absolute lymphocyte count (ALC), within four days following ICU admission. Group-based multi-trajectory modeling was used to cluster sepsis patients based on the dynamic changes of CRP and ALC to identify potential immune subgroups. Comparisons of clinical characteristics, hospital mortality, and persistent inflammation immunosuppression catabolism syndrome (PICS) incidence were made across the derived subgroups. We further characterized the immune-specific properties of each subgroup using 28 immune markers.

RESULTS

Based on the optimal classification, 471 sepsis patients in the derivation cohort were categorized into three subgroups: immune homeostasis subgroup (n = 39, 8.3%), immunosuppression subgroup (n = 180, 38.2%), and immune imbalance subgroup (n = 252, 53.5%). This classification remained robust upon validation in both internal (n = 2,527) and external (n = 225) cohorts. Differences in clinical profiles were observed in these subgroups: the immune homeostasis subgroup had the mildest disease severity with the lowest rates of PICS and mortality; the immunosuppression subgroup featured more chronic comorbidities, longer hospital stays, and higher mortality; while the immune imbalance subgroup exhibited the most critical condition, with more complications, longer ICU stays, and the highest incidence of PICS and mortality. Immune features revealed distinct subtype properties: the immune homeostasis subgroup exhibited inflammatory resolution and near-normal immunity; the immunosuppression subgroup characterized by a general reduction in lymphocyte subsets, persistent early T-cell activation, and the highest degree of T cell inhibition; and the immune imbalance subgroup showed a cytokine storm alongside the most severely impaired innate and adaptive immunity, as indicated by markedly low monocyte human leukocyte antigen-DR (mHLA-DR) expression and deficient late T cell activation.

CONCLUSIONS

This study presents a practical immunological classification for sepsis based on the longitudinal dynamics of two immune markers. These immune subtypes complement existing consensus endotypes and provide a simple, clinically actionable approach particularly applicable in resource-limited settings.