After Hospices Civils de Lyon (Lyon, France) ethics committee review and approval, we enrolled 64 patients with septic shock in this observational clinical study (from 2007 to 2009). Septic shock was diagnosed according to the diagnostic criteria of the American College of Chest Physicians/Society of Critical Care Medicine 16 . Patients were admitted to one of the two intensive care units (ICUs) (one medical, the other surgical) of the Lyon-Sud University Hospital (France).

Septic shock was defined by an identifiable site of infection, which was evidence of a systemic inflammatory response manifested by at least two of the following criteria: (a) temperature of greater than 38°C or less than 36°C, (b) heart rate of greater than 90 beats per minute, (c) respiratory rate of greater than 20 breaths per minute, and (d) white blood cell count of greater than 12,000 or less than 4,000/mm³ and hypotension persisting despite fluid resuscitation and requiring vasopressor therapy. The beginning of vasopressive therapy was considered the time of diagnosis of septic shock. Exclusion criteria were age of less than 18 years and the absence of circulating leukocytes for flow cytometry phenotyping. No patients with HIV were included. Patients with cancer were excluded from our study if they presented with an aplasia (defined by a polymorphonuclear neutrophil count of less than 0.5 G/L) or were treated with a high dose of corticoids (estimated as treatment superior to 10 mg equivalent prednisolone/day or more than 700 mg equivalent prednisolone accrued the first day of inclusion) or both.

The following clinical and biological data were collected: demographic characteristics (age and gender), admission category (elective or emergency surgery and medicine), referral pattern (community-, hospital-, or ICU-acquired septic shock), microbiological findings, clinical scores (Simplified Acute Physiology Score II (SAPS II) and sepsis-related organ failure assessment (SOFA) score), incidence of secondary nosocomial infections (defined as microbiologically documented pulmonary infection, urinary tract infection, bloodstream infection, and catheter-related infection that occurred 48 hours after ICU admission and up to ICU discharge 17 ), and the outcome after 28 days (death or survival).

The protocol was reviewed by the institutional ethics committee, which waived the need for informed consent because the study was observational and involved sampling of very small quantities of blood. The purpose of the study was explained to the patients or members of their families. Samples were collected from residual blood after completion of routine follow-up. Ethylenediaminetetraacetic acid (EDTA)-anti-coagulated blood was collected from patients at different time points: day (D) 1-2, D3-5, and D6-10 after diagnosis of septic shock. Additionally, 13 trauma patients were included in the study within the first 48 hours of admission. Inclusion criteria were trauma, age of at least 18 years, and an initial injury severity score (ISS) of at least 25. Finally, 49 healthy volunteers from laboratory staff of our hospital were included as controls.

The following antibodies were used: PC5-labeled anti-CD4, PC5-labeled anti-CD8, PC5-labeled anti-CD14, PC5-labeled anti-CD25, PE-labeled anti-CD127, FITC-labeled anti-CD14, ECD-labeled anti-CD4 (Beckman Coulter, Miami, FL, USA), and PE-labeled anti-HLA-DR or its isotype PE-labeled IgG2a (Becton-Dickinson Biosciences, San Jose, CA, USA), PE-labeled anti-human CD249 (PD-1, clone MIH4), FITC-labeled anti-human CD274 (PD-L1, clone MIH1), or PE-labeled anti-human CD273 (PD-L2, clone MIH18) (BD Biosciences). Red blood cells were lysed using the automated TQ-Prep (Beckman Coulter) or using FACS-lysing solution (BD Biosciences). Samples were run on FC500 (Beckman Coulter) and analyzed using CXP software (Beckman Coulter).

IL-10 concentration in patients' plasma samples was measured by Bio-Plex Pro Assays (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Unknown sample values presented as picograms per milliliter were determined against human standards as described by the manufacturer.

In brief, peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density gradient centrifugation (PAA Laboratories, Pasching, Austria). PBMCs were washed three times in phosphate-buffered saline (bioMérieux, Marcy-l'Etoile, France) and resuspended in complete medium - that is, RPMI supplemented with HEPES (25 mM), sodium bicarbonate (2 g/L) (Eurobio Laboratories, Les Ulis, France), 10% human serum AB (obtained from a pool of healthy volunteers), 2 mM L-glutamine (Lonza, Verviers, Belgium), 20 UI/mL penicillin, 20 μg/mL streptomycin (Sigma-Aldrich, St. Louis, MO, USA), and 2.5 μg/mL Amphotericin B (Bristol-Myers Squibb Company, Princeton, NJ, USA). Cells were kept on ice until stainings or cell cultures were performed.

PBMCs were seeded at a density of 1 × 10⁶ cells/mL (50,000 cells/well, 100 μL) in flat-bottom 96-well microtiter plates and were stimulated with 5 μg/mL phytohemagglutinin (PHA) (Remel, part of Thermo Fisher Scientific, Lenexa, KS, USA). Cells were incubated 48 hours at 37°C in a humidified 5% CO₂ atmosphere.

[methyl-³H]-Thymidine (20 μCi/mL) (PerkinElmer, Waltham, MA, USA) was added 24 hours before harvesting cells on fiberglass filters by means of an automated cell harvester (PerkinElmer). Incorporated radioactivity was measured in a direct beta counter (PerkinElmer). Assays were carried out in triplicate.

Patients' clinical and biological parameters were presented as frequencies, percentages, medians, and interquartile ranges (IQRs). Differences in expression levels were calculated using the Mann-Whitney U test or, when multiple comparisons were performed, the Friedman test. Correlations were calculated using the Spearman rank test. P values of not more than 0.05 were considered statistically significant; if necessary, correction for the number of tests was performed. Statistical analysis was performed using SPSS software (version 12.0; SPSS Inc., Chicago, IL, USA).

Sixty-four patients with septic shock (20 women and 44 men) were included in the study. Their clinical characteristics are shown in Table 1. Median age at admission was 63 years (IQR 54 to 73). Median values for SAPS II and SOFA score at diagnosis of shock were 53 (IQR 39 to 64) and 10 (IQR 8 to 12), respectively, indicating a high level of severity. Approximately 30% of patients developed secondary nosocomial infections, and 28-day mortality was 17%.

Septic patients presented with typical features of sepsis-induced immunosuppression and displayed a reduced monocyte HLA-DR expression at D3-5 (median value 45.5%, IQR 29.5 to 69.5) in comparison with control values (>90% 18 ). Median CD4⁺ T-cell count was also decreased in patients in comparison with healthy volunteers (319 cells/μL (IQR 226 to 681) versus 822 cells/μL (IQR 679 to 1,075), respectively; P < 0.001), whereas percentage of circulating regulatory T cells (CD4⁺CD25⁺CD127^- T lymphocytes) was augmented (8.5% (IQR 6.1% to 11.2%) versus 6.2% (IQR 5.2% to 7.6%), respectively; P = 0.001).

Thirteen trauma patients (9 men and 4 women) were also included in the study. Median age at admission was 34 years (IQR 24 to 56). In the first 24 hours of admission, they presented a median ISS of 32 (IQR 26 to 34) and a median SAPS II of 39 (IQR 22 to 52).

PD-1, PD-L1, and PD-L2 expressions were measured on circulating CD4⁺ lymphocytes, CD8⁺ lymphocytes (PD-1 only), and monocytes at D1-2 and 3-5 after the onset of septic shock. Results for CD4⁺ lymphocytes and monocytes are shown in Figure 1.

The percentages of circulating monocytes expressing PD-1, PD-L1, or PD-L2 were markedly increased in patients with septic shock in comparison with healthy volunteers during the overall monitoring (Figure 1a). This augmentation was present for PD-1 (median control values: 5.0% versus 18.6% (D1-2) and 17.8% (D3-5) in patients; P < 0.001), for PD-L1 (control values: 10.2% versus 46.6% (D1-2) and 34.9% (D3-5) in patients; P < 0.001), and for PD-L2 (control values: 2.6% versus 8.7% (D1-2) and 8.5% (D3-5) in patients; P < 0.001). Similar results were observed when flow cytometry data were expressed as mean fluorescence intensity (MFI) (Table 2). In trauma patients, PD-1-related molecule expressions on monocytes were significantly increased in comparison with healthy individuals (for PD-1: control value: 5.0% versus 9.6%, P = 0.005; for PD-L1: control value: 10.2% versus 40.1%, P < 0.001; and for PD-L2: control value: 2.6% versus 7.2%, P < 0.001). However, PD-1 expression on monocytes was significantly lower in trauma than in septic shock patients at D1-2 (9.6% versus 18.6%, respectively; P = 0.008) (data not shown).

Likewise, the percentages of circulating CD4⁺ lymphocytes expressing PD-1 or PD-L1 were notably increased in patients with septic shock in comparison with healthy volunteers during the overall monitoring (for PD-1: control values: 5.4% versus 15.0% (D1-2) and 13.6% (D3-5), P < 0.001; for PD-L1: control values: 2.5% versus 3.9% (D1-2; P = 0.002) and 3.6% (D3-5; P = 0.016) in patients) (Figure 1b). Alternatively, no significant differences were observed between patients and healthy volunteers for percentages of CD4⁺ cells expressing PD-L2 (Figure 1b) or of CD8⁺ lymphocytes positive for PD-1 (Table 2). Once again, similar results were observed when flow cytometry results were expressed as MFI (Table 2). No difference in PD-1-related molecule expressions was observed between trauma patients and healthy individuals. However, the percentage of PD-1 expressing CD4⁺ cells was significantly lower in trauma than in septic shock patients at D1-2 (5.2% versus 15.0%, respectively; P < 0.001) (data not shown).

Of note, there was no variation of PD-1-related molecule expressions in regard to age or gender either in healthy subjects or in patients with septic shock. Indeed, we did not observe significant correlations between PD-1-related molecule expressions and the age of septic shock patients (r = 0.21, P = 0.12 for PD-1 expression on CD4⁺ lymphocytes; r = 0.04, P = 0.78 for PD-L1 expression on monocytes) or of healthy volunteers (r = 0.10, P = 0.49 for PD-1 expression on CD4⁺ lymphocytes; r = -0.15, P = 0.30 for PD-L1 expression on monocytes).

Finally, in 10 patients with septic shock, sequential blood samples were obtained at D1-2, D3-5, and D6-10 after the onset of shock. During this period, no significant variations over time in regard to PD-1 molecule expressions either on monocytes or on lymphocytes were observed (Figure 2).

To assess the clinical relevance of the increase in PD-1-related molecule expressions after septic shock, flow cytometric measurements were correlated with clinical parameters and usual biomarkers of sepsis-induced immunosuppression. No significant correlations were found between PD-1-related molecule expressions and percentages of HLA-DR expressing monocytes, CD4⁺ lymphocyte count, percentage of circulating regulatory T cells, or severity scores calculated at the onset of shock (SAPS II or SOFA score) (data not shown). However, at D1-2, we observed that PD-L1 expression on monocytes was significantly higher in non-survivors in comparison with survivors (Figure 3a). Moreover, at D3-5, patients who went on to develop a secondary nosocomial infection presented with higher PD-1 (Figure 3b) and PD-L2 (Figure 3c) expressions on their blood monocytes in comparison with those who remained free of any secondary nosocomial episode.

Increased circulating IL-10 concentration has been linked with mortality after septic shock 19 and recently with enhanced PD-1 expression in HIV-infected patients 20 . We thus measured circulating IL-10 levels in 29 septic shock patients for whom plasma samples were available and we correlated this parameter with leukocyte PD-1/PD-L expressions. Not surprisingly, we observed that non-survivors exhibited higher plasma IL-10 concentration than survivors at D1-2 and D3-5 (P = 0.01 for both) (Figure 4a). Interestingly, a significant positive correlation was measured between PD-1 monocyte expression and plasma IL-10 concentration in patients at D1-2 (r = 0.49; P = 0.007) (Figure 4b) but not at D3-5 (data not shown). In addition, significant correlations were observed between both PD-L1 or PD-L2 monocyte expressions and increased plasma IL-10 concentration at D1-2 (r = 0.58; P = 0.001 and r = 0.45; P = 0.014, respectively) and D3-5 (r = 0.45; P = 0.015 and r = 0.53; P = 0.003, respectively) (Figure 4c, d). Of note, no correlations were found between PD-1/PD-L-related molecule expressions on CD4⁺ lymphocytes and changes in plasma IL-10 concentration (data not shown). Also, for all of these observations made for percentage of positive cells, similar correlations were obtained when flow cytometry results were expressed as MFI (data not shown).

In an attempt to begin to address the biological significance of these changes in PD-1 expression to the development of sepsis-induced lymphocyte dysfunction, freshly isolated PBMCs from septic shock patients and healthy volunteers were assessed for their capacity to respond to PHA. As expected, we observed that lymphocyte proliferation was significantly reduced in patients in comparison with healthy volunteers (P < 0.001) (Figure 5a). Interestingly, in patients, a significant negative correlation was observed between this reduced proliferation and PD-1 (r = -0.81 with P = 0.003) (Figure 5b) or PD-L1 (r = -0.63 with P = 0.039) (data not shown) overexpression on circulating CD4⁺ lymphocytes. Similar results were obtained when PD-1 and PD-L1 staining was expressed as MFI (r = -0.80 with P = 0.003 and r = -0.63 with P = 0.038, respectively).