RESUMEN
RATIONALE: There is an unmet need to improve the description of the state of T-cell exhaustion in patients with sepsis, its reproducibility and correlation with the outcomes before including immunotherapy (like recombinant interleukin-7 or immune checkpoint inhibitors) in the therapeutic armamentarium against sepsis. DESIGN: Observational prospective study. SETTING: Two ICUs in a teaching hospital (France). PATIENTS: Eighty patients with sepsis admitted to the ICU. INTERVENTIONS: Quantification of CD4+ and CD8+ T-cell exhaustion at days 1 and 3. Quantification of the exhaustion markers (programmed death [PD]-1, 2B4, and cluster of differentiation [CD] 160) on T cells, the number of CD4+ regulatory T cells (CD3+ CD4+ CD25hi CD127Lo cells), and the phorbol myristate acetate/ionomycin/ionomycin-induced cytokines production (tumor necrosis factor-α, interleukin-2, and interferon-γ). MEASUREMENTS AND MAIN RESULTS: Using unsupervised clustering analysis, patients could be split in three clusters according to their dominant pattern expression of exhaustion markers on CD8+ T cells (i.e., 2B4lowPD-1lowCD160low, 2B4hiPD-1hiCD160low, and 2B4hiPD-1lowCD160hi) regardless of their underlying morbidities. Only 2B4hiPD-1hiCD160low CD8+ T cells had cytokine production defect, whereas 2B4hi PD-1lowCD160hi pattern correlated with cytokine overproduction. Patients with a predominant "highly activated" 2B4hiPD-1lowCD160hi pattern did not develop secondary bacterial infections. By multivariate analysis, Simplified Acute Physiology Score 2 gravity score at day 1 (p = 0.003) and patterns of exhaustion markers on CD8+ T cells (p = 0.03) were associated with the risk of death. Neither the level of CD4+ regulatory T cells nor the CD4+ exhaustion patterns were associated with the outcomes. CONCLUSIONS: Easy-to-use multicolor flow cytometry assessing 2B4, PD-1, and CD160 expression on CD8+ T cells at day 1 identifies septic patients with poor outcome and discriminates patient subsets in who immunomodulatory drugs should be tested.