Objectives: Primary objective was to identify leukocyte subsets that could predict the early evolution of sepsis at 48 hours (i.e., deterioration or stability/improvement). Secondary objectives were to evaluate the prognostic value of leukocyte subsets on mortality and immunosuppressive properties of immature granulocytes.
Design: Twenty-three peripheral blood leukocyte subsets were analyzed using a new-generation 10-color flow cytometry. T-cell killing activity of immature granulocytes was explored using a sorting method specifically developed.
Setting: ICUs and emergency departments.
Patients: All patients admitted to emergency department and ICU for sepsis ongoing for less than 24 hours were eligible. Exclusion criteria were pregnancy, age less than 18 years, solid tumors, HIV infection, hematological or inflammatory conditions, and immunosuppressive drugs. Finally, 177 patients were included.
Interventions: None.
Measurements and main results: The two most salient features of sepsis were decreased CD10 (CD10) and CD16 (CD16) expressions on granulocytes. With a threshold of 90% of CD10 and 15% of CD16 granulocytes, these immunophenotypic features, which are those of immature granulocytes, predicted sepsis deterioration at 48 hours with a sensitivity of 57% and 70% and a specificity of 78% and 82%, respectively. Survival rate at day 30 was 99% for patients without CD10 and CD16, 85% for patients with increased CD16 only, and 63% for patients with increased CD16 and CD10 granulocytes (p < 0.001). Among CD16 immature granulocytes, we identified a CD14/CD24 myeloid-derived suppressor cell subset with the capability of killing activated T cells. Consistently, an excess of CD16 immature granulocytes was associated with both CD3 and CD4 T-cell lymphopenia in deteriorating patients.
Conclusions: Circulating immature granulocytes predicted early sepsis deterioration and were enriched in myeloid-derived suppressor cells which could be responsible for immunosuppression through the induction of T-cell lymphopenia.